cpplint.py 254.3 KB
Newer Older
Z
zhangjinchao01 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242
#!/usr/bin/env python
#
# Copyright (c) 2009 Google Inc. All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are
# met:
#
#    * Redistributions of source code must retain the above copyright
# notice, this list of conditions and the following disclaimer.
#    * Redistributions in binary form must reproduce the above
# copyright notice, this list of conditions and the following disclaimer
# in the documentation and/or other materials provided with the
# distribution.
#    * Neither the name of Google Inc. nor the names of its
# contributors may be used to endorse or promote products derived from
# this software without specific prior written permission.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
"""Does google-lint on c++ files.

The goal of this script is to identify places in the code that *may*
be in non-compliance with google style.  It does not attempt to fix
up these problems -- the point is to educate.  It does also not
attempt to find all problems, or to ensure that everything it does
find is legitimately a problem.

In particular, we can get very confused by /* and // inside strings!
We do a small hack, which is to ignore //'s with "'s after them on the
same line, but it is far from perfect (in either direction).

EDIT(yuyang18): Add #pragma once as include guard.
EDIT(yuyang18): Add NOLINTNEXTLINES_ to suppress multiline lint.
"""

import codecs
import copy
import getopt
import math  # for log
import os
import re
import sre_compile
import string
import sys
import unicodedata

_USAGE = """
Syntax: cpplint.py [--verbose=#] [--output=vs7] [--filter=-x,+y,...]
                   [--counting=total|toplevel|detailed] [--root=subdir]
                   [--linelength=digits]
        <file> [file] ...

  The style guidelines this tries to follow are those in
    http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml

  Every problem is given a confidence score from 1-5, with 5 meaning we are
  certain of the problem, and 1 meaning it could be a legitimate construct.
  This will miss some errors, and is not a substitute for a code review.

  To suppress false-positive errors of a certain category, add a
  'NOLINT(category)' comment to the line.  NOLINT or NOLINT(*)
  suppresses errors of all categories on that line.

  The files passed in will be linted; at least one file must be provided.
  Default linted extensions are .cc, .cpp, .cu, .cuh and .h.  Change the
  extensions with the --extensions flag.

  Flags:

    output=vs7
      By default, the output is formatted to ease emacs parsing.  Visual Studio
      compatible output (vs7) may also be used.  Other formats are unsupported.

    verbose=#
      Specify a number 0-5 to restrict errors to certain verbosity levels.

    filter=-x,+y,...
      Specify a comma-separated list of category-filters to apply: only
      error messages whose category names pass the filters will be printed.
      (Category names are printed with the message and look like
      "[whitespace/indent]".)  Filters are evaluated left to right.
      "-FOO" and "FOO" means "do not print categories that start with FOO".
      "+FOO" means "do print categories that start with FOO".

      Examples: --filter=-whitespace,+whitespace/braces
                --filter=whitespace,runtime/printf,+runtime/printf_format
                --filter=-,+build/include_what_you_use

      To see a list of all the categories used in cpplint, pass no arg:
         --filter=

    counting=total|toplevel|detailed
      The total number of errors found is always printed. If
      'toplevel' is provided, then the count of errors in each of
      the top-level categories like 'build' and 'whitespace' will
      also be printed. If 'detailed' is provided, then a count
      is provided for each category like 'build/class'.

    root=subdir
      The root directory used for deriving header guard CPP variable.
      By default, the header guard CPP variable is calculated as the relative
      path to the directory that contains .git, .hg, or .svn.  When this flag
      is specified, the relative path is calculated from the specified
      directory. If the specified directory does not exist, this flag is
      ignored.

      Examples:
        Assuming that src/.git exists, the header guard CPP variables for
        src/chrome/browser/ui/browser.h are:

        No flag => CHROME_BROWSER_UI_BROWSER_H_
        --root=chrome => BROWSER_UI_BROWSER_H_
        --root=chrome/browser => UI_BROWSER_H_

    linelength=digits
      This is the allowed line length for the project. The default value is
      80 characters.

      Examples:
        --linelength=120

    extensions=extension,extension,...
      The allowed file extensions that cpplint will check

      Examples:
        --extensions=hpp,cpp

    cpplint.py supports per-directory configurations specified in CPPLINT.cfg
    files. CPPLINT.cfg file can contain a number of key=value pairs.
    Currently the following options are supported:

      set noparent
      filter=+filter1,-filter2,...
      exclude_files=regex
      linelength=80

    "set noparent" option prevents cpplint from traversing directory tree
    upwards looking for more .cfg files in parent directories. This option
    is usually placed in the top-level project directory.

    The "filter" option is similar in function to --filter flag. It specifies
    message filters in addition to the |_DEFAULT_FILTERS| and those specified
    through --filter command-line flag.

    "exclude_files" allows to specify a regular expression to be matched against
    a file name. If the expression matches, the file is skipped and not run
    through liner.

    "linelength" allows to specify the allowed line length for the project.

    CPPLINT.cfg has an effect on files in the same directory and all
    sub-directories, unless overridden by a nested configuration file.

      Example file:
        filter=-build/include_order,+build/include_alpha
        exclude_files=.*\.cc

    The above example disables build/include_order warning and enables
    build/include_alpha as well as excludes all .cc from being
    processed by linter, in the current directory (where the .cfg
    file is located) and all sub-directories.
"""

# We categorize each error message we print.  Here are the categories.
# We want an explicit list so we can list them all in cpplint --filter=.
# If you add a new error message with a new category, add it to the list
# here!  cpplint_unittest.py should tell you if you forget to do this.
_ERROR_CATEGORIES = [
    'build/class',
    'build/c++11',
    'build/deprecated',
    'build/endif_comment',
    'build/explicit_make_pair',
    'build/forward_decl',
    'build/header_guard',
    'build/include',
    'build/include_alpha',
    'build/include_order',
    'build/include_what_you_use',
    'build/namespaces',
    'build/printf_format',
    'build/storage_class',
    'legal/copyright',
    'readability/alt_tokens',
    'readability/braces',
    'readability/casting',
    'readability/check',
    'readability/constructors',
    'readability/fn_size',
    'readability/function',
    'readability/inheritance',
    'readability/multiline_comment',
    'readability/multiline_string',
    'readability/namespace',
    'readability/nolint',
    'readability/nul',
    'readability/strings',
    'readability/todo',
    'readability/utf8',
    'runtime/arrays',
    'runtime/casting',
    'runtime/explicit',
    'runtime/int',
    'runtime/init',
    'runtime/invalid_increment',
    'runtime/member_string_references',
    'runtime/memset',
    'runtime/indentation_namespace',
    'runtime/operator',
    'runtime/printf',
    'runtime/printf_format',
    'runtime/references',
    'runtime/string',
    'runtime/threadsafe_fn',
    'runtime/vlog',
    'whitespace/blank_line',
    'whitespace/braces',
    'whitespace/comma',
    'whitespace/comments',
    'whitespace/empty_conditional_body',
    'whitespace/empty_loop_body',
    'whitespace/end_of_line',
    'whitespace/ending_newline',
    'whitespace/forcolon',
    'whitespace/indent',
    'whitespace/line_length',
    'whitespace/newline',
    'whitespace/operators',
    'whitespace/parens',
    'whitespace/semicolon',
    'whitespace/tab',
    'whitespace/todo',
Y
Yu Yang 已提交
243
]
Z
zhangjinchao01 已提交
244 245 246

# These error categories are no longer enforced by cpplint, but for backwards-
# compatibility they may still appear in NOLINT comments.
Y
Yu Yang 已提交
247
_LEGACY_ERROR_CATEGORIES = ['readability/streams', ]
Z
zhangjinchao01 已提交
248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392

# The default state of the category filter. This is overridden by the --filter=
# flag. By default all errors are on, so only add here categories that should be
# off by default (i.e., categories that must be enabled by the --filter= flags).
# All entries here should start with a '-' or '+', as in the --filter= flag.
_DEFAULT_FILTERS = ['-build/include_alpha']

# We used to check for high-bit characters, but after much discussion we
# decided those were OK, as long as they were in UTF-8 and didn't represent
# hard-coded international strings, which belong in a separate i18n file.

# C++ headers
_CPP_HEADERS = frozenset([
    # Legacy
    'algobase.h',
    'algo.h',
    'alloc.h',
    'builtinbuf.h',
    'bvector.h',
    'complex.h',
    'defalloc.h',
    'deque.h',
    'editbuf.h',
    'fstream.h',
    'function.h',
    'hash_map',
    'hash_map.h',
    'hash_set',
    'hash_set.h',
    'hashtable.h',
    'heap.h',
    'indstream.h',
    'iomanip.h',
    'iostream.h',
    'istream.h',
    'iterator.h',
    'list.h',
    'map.h',
    'multimap.h',
    'multiset.h',
    'ostream.h',
    'pair.h',
    'parsestream.h',
    'pfstream.h',
    'procbuf.h',
    'pthread_alloc',
    'pthread_alloc.h',
    'rope',
    'rope.h',
    'ropeimpl.h',
    'set.h',
    'slist',
    'slist.h',
    'stack.h',
    'stdiostream.h',
    'stl_alloc.h',
    'stl_relops.h',
    'streambuf.h',
    'stream.h',
    'strfile.h',
    'strstream.h',
    'tempbuf.h',
    'tree.h',
    'type_traits.h',
    'vector.h',
    # 17.6.1.2 C++ library headers
    'algorithm',
    'array',
    'atomic',
    'bitset',
    'chrono',
    'codecvt',
    'complex',
    'condition_variable',
    'deque',
    'exception',
    'forward_list',
    'fstream',
    'functional',
    'future',
    'initializer_list',
    'iomanip',
    'ios',
    'iosfwd',
    'iostream',
    'istream',
    'iterator',
    'limits',
    'list',
    'locale',
    'map',
    'memory',
    'mutex',
    'new',
    'numeric',
    'ostream',
    'queue',
    'random',
    'ratio',
    'regex',
    'set',
    'sstream',
    'stack',
    'stdexcept',
    'streambuf',
    'string',
    'strstream',
    'system_error',
    'thread',
    'tuple',
    'typeindex',
    'typeinfo',
    'type_traits',
    'unordered_map',
    'unordered_set',
    'utility',
    'valarray',
    'vector',
    # 17.6.1.2 C++ headers for C library facilities
    'cassert',
    'ccomplex',
    'cctype',
    'cerrno',
    'cfenv',
    'cfloat',
    'cinttypes',
    'ciso646',
    'climits',
    'clocale',
    'cmath',
    'csetjmp',
    'csignal',
    'cstdalign',
    'cstdarg',
    'cstdbool',
    'cstddef',
    'cstdint',
    'cstdio',
    'cstdlib',
    'cstring',
    'ctgmath',
    'ctime',
    'cuchar',
    'cwchar',
    'cwctype',
Y
Yu Yang 已提交
393
])
Z
zhangjinchao01 已提交
394 395 396 397 398 399 400 401 402 403 404 405 406

# These headers are excluded from [build/include] and [build/include_order]
# checks:
# - Anything not following google file name conventions (containing an
#   uppercase character, such as Python.h or nsStringAPI.h, for example).
# - Lua headers.
_THIRD_PARTY_HEADERS_PATTERN = re.compile(
    r'^(?:[^/]*[A-Z][^/]*\.h|lua\.h|lauxlib\.h|lualib\.h)$')

# Assertion macros.  These are defined in base/logging.h and
# testing/base/gunit.h.  Note that the _M versions need to come first
# for substring matching to work.
_CHECK_MACROS = [
Y
Yu Yang 已提交
407 408 409 410 411 412 413 414 415 416 417
    'DCHECK',
    'CHECK',
    'EXPECT_TRUE_M',
    'EXPECT_TRUE',
    'ASSERT_TRUE_M',
    'ASSERT_TRUE',
    'EXPECT_FALSE_M',
    'EXPECT_FALSE',
    'ASSERT_FALSE_M',
    'ASSERT_FALSE',
]
Z
zhangjinchao01 已提交
418 419 420 421

# Replacement macros for CHECK/DCHECK/EXPECT_TRUE/EXPECT_FALSE
_CHECK_REPLACEMENT = dict([(m, {}) for m in _CHECK_MACROS])

Y
Yu Yang 已提交
422
for op, replacement in [('==', 'EQ'), ('!=', 'NE'), ('>=', 'GE'), ('>', 'GT'),
Z
zhangjinchao01 已提交
423
                        ('<=', 'LE'), ('<', 'LT')]:
Y
Yu Yang 已提交
424 425 426 427 428 429 430 431 432 433 434 435 436
    _CHECK_REPLACEMENT['DCHECK'][op] = 'DCHECK_%s' % replacement
    _CHECK_REPLACEMENT['CHECK'][op] = 'CHECK_%s' % replacement
    _CHECK_REPLACEMENT['EXPECT_TRUE'][op] = 'EXPECT_%s' % replacement
    _CHECK_REPLACEMENT['ASSERT_TRUE'][op] = 'ASSERT_%s' % replacement
    _CHECK_REPLACEMENT['EXPECT_TRUE_M'][op] = 'EXPECT_%s_M' % replacement
    _CHECK_REPLACEMENT['ASSERT_TRUE_M'][op] = 'ASSERT_%s_M' % replacement

for op, inv_replacement in [('==', 'NE'), ('!=', 'EQ'), ('>=', 'LT'),
                            ('>', 'LE'), ('<=', 'GT'), ('<', 'GE')]:
    _CHECK_REPLACEMENT['EXPECT_FALSE'][op] = 'EXPECT_%s' % inv_replacement
    _CHECK_REPLACEMENT['ASSERT_FALSE'][op] = 'ASSERT_%s' % inv_replacement
    _CHECK_REPLACEMENT['EXPECT_FALSE_M'][op] = 'EXPECT_%s_M' % inv_replacement
    _CHECK_REPLACEMENT['ASSERT_FALSE_M'][op] = 'ASSERT_%s_M' % inv_replacement
Z
zhangjinchao01 已提交
437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454

# Alternative tokens and their replacements.  For full list, see section 2.5
# Alternative tokens [lex.digraph] in the C++ standard.
#
# Digraphs (such as '%:') are not included here since it's a mess to
# match those on a word boundary.
_ALT_TOKEN_REPLACEMENT = {
    'and': '&&',
    'bitor': '|',
    'or': '||',
    'xor': '^',
    'compl': '~',
    'bitand': '&',
    'and_eq': '&=',
    'or_eq': '|=',
    'xor_eq': '^=',
    'not': '!',
    'not_eq': '!='
Y
Yu Yang 已提交
455
}
Z
zhangjinchao01 已提交
456 457 458 459 460 461

# Compile regular expression that matches all the above keywords.  The "[ =()]"
# bit is meant to avoid matching these keywords outside of boolean expressions.
#
# False positives include C-style multi-line comments and multi-line strings
# but those have always been troublesome for cpplint.
Y
Yu Yang 已提交
462 463
_ALT_TOKEN_REPLACEMENT_PATTERN = re.compile(r'[ =()](' + ('|'.join(
    _ALT_TOKEN_REPLACEMENT.keys())) + r')(?=[ (]|$)')
Z
zhangjinchao01 已提交
464 465 466 467 468 469 470 471 472 473

# These constants define types of headers for use with
# _IncludeState.CheckNextIncludeOrder().
_C_SYS_HEADER = 1
_CPP_SYS_HEADER = 2
_LIKELY_MY_HEADER = 3
_POSSIBLE_MY_HEADER = 4
_OTHER_HEADER = 5

# These constants define the current inline assembly state
Y
Yu Yang 已提交
474 475 476 477
_NO_ASM = 0  # Outside of inline assembly block
_INSIDE_ASM = 1  # Inside inline assembly block
_END_ASM = 2  # Last line of inline assembly block
_BLOCK_ASM = 3  # The whole block is an inline assembly block
Z
zhangjinchao01 已提交
478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501

# Match start of assembly blocks
_MATCH_ASM = re.compile(r'^\s*(?:asm|_asm|__asm|__asm__)'
                        r'(?:\s+(volatile|__volatile__))?'
                        r'\s*[{(]')

_regexp_compile_cache = {}

# {str, set(int)}: a map from error categories to sets of linenumbers
# on which those errors are expected and should be suppressed.
_error_suppressions = {}

# The root directory used for deriving header guard CPP variable.
# This is set by --root flag.
_root = None

# The allowed line length of files.
# This is set by --linelength flag.
_line_length = 80

# The allowed extensions for file names
# This is set by --extensions flag.
_valid_extensions = set(['cc', 'h', 'cpp', 'cu', 'cuh'])

Y
Yu Yang 已提交
502

Z
zhangjinchao01 已提交
503
def ParseNolintSuppressions(filename, raw_line, linenum, error):
Y
Yu Yang 已提交
504
    """Updates the global list of error-suppressions.
Z
zhangjinchao01 已提交
505 506 507 508 509 510 511 512 513 514 515

  Parses any NOLINT comments on the current line, updating the global
  error_suppressions store.  Reports an error if the NOLINT comment
  was malformed.

  Args:
    filename: str, the name of the input file.
    raw_line: str, the line of input text, with comments.
    linenum: int, the number of the current line.
    error: function, an error handler.
  """
Y
Yu Yang 已提交
516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547
    matched = Search(r'\bNOLINT(NEXTLINE(S_\d+)?)?\b(\([^)]+\))?', raw_line)
    if matched:
        if matched.group(1):
            lines = matched.group(2)
            if lines:
                lines = int(lines[2:])
                suppressed_line = [linenum + i for i in xrange(lines)]
            else:
                suppressed_line = linenum + 1
        else:
            suppressed_line = linenum
        category = matched.group(3)
        if category in (None, '(*)'):  # => "suppress all"
            if isinstance(suppressed_line, int):
                _error_suppressions.setdefault(None, set()).add(suppressed_line)
            else:
                for _line in suppressed_line:
                    _error_suppressions.setdefault(None, set()).add(_line)
        else:
            if category.startswith('(') and category.endswith(')'):
                category = category[1:-1]
                if category in _ERROR_CATEGORIES:
                    if isinstance(suppressed_line, int):
                        _error_suppressions.setdefault(
                            category, set()).add(suppressed_line)
                    else:
                        for _line in suppressed_line:
                            _error_suppressions.setdefault(category,
                                                           set()).add(_line)
                elif category not in _LEGACY_ERROR_CATEGORIES:
                    error(filename, linenum, 'readability/nolint', 5,
                          'Unknown NOLINT error category: %s' % category)
Z
zhangjinchao01 已提交
548 549 550


def ResetNolintSuppressions():
Y
Yu Yang 已提交
551 552
    """Resets the set of NOLINT suppressions to empty."""
    _error_suppressions.clear()
Z
zhangjinchao01 已提交
553 554 555


def IsErrorSuppressedByNolint(category, linenum):
Y
Yu Yang 已提交
556
    """Returns true if the specified error category is suppressed on this line.
Z
zhangjinchao01 已提交
557 558 559 560 561 562 563 564 565 566

  Consults the global error_suppressions map populated by
  ParseNolintSuppressions/ResetNolintSuppressions.

  Args:
    category: str, the category of the error.
    linenum: int, the current line number.
  Returns:
    bool, True iff the error should be suppressed due to a NOLINT comment.
  """
Y
Yu Yang 已提交
567 568
    return (linenum in _error_suppressions.get(category, set()) or
            linenum in _error_suppressions.get(None, set()))
Z
zhangjinchao01 已提交
569 570 571


def Match(pattern, s):
Y
Yu Yang 已提交
572 573 574 575 576 577 578
    """Matches the string with the pattern, caching the compiled regexp."""
    # The regexp compilation caching is inlined in both Match and Search for
    # performance reasons; factoring it out into a separate function turns out
    # to be noticeably expensive.
    if pattern not in _regexp_compile_cache:
        _regexp_compile_cache[pattern] = sre_compile.compile(pattern)
    return _regexp_compile_cache[pattern].match(s)
Z
zhangjinchao01 已提交
579 580 581


def ReplaceAll(pattern, rep, s):
Y
Yu Yang 已提交
582
    """Replaces instances of pattern in a string with a replacement.
Z
zhangjinchao01 已提交
583 584 585 586 587 588 589 590 591 592 593

  The compiled regex is kept in a cache shared by Match and Search.

  Args:
    pattern: regex pattern
    rep: replacement text
    s: search string

  Returns:
    string with replacements made (or original string if no replacements)
  """
Y
Yu Yang 已提交
594 595 596
    if pattern not in _regexp_compile_cache:
        _regexp_compile_cache[pattern] = sre_compile.compile(pattern)
    return _regexp_compile_cache[pattern].sub(rep, s)
Z
zhangjinchao01 已提交
597 598 599


def Search(pattern, s):
Y
Yu Yang 已提交
600 601 602 603
    """Searches the string for the pattern, caching the compiled regexp."""
    if pattern not in _regexp_compile_cache:
        _regexp_compile_cache[pattern] = sre_compile.compile(pattern)
    return _regexp_compile_cache[pattern].search(s)
Z
zhangjinchao01 已提交
604 605 606


class _IncludeState(object):
Y
Yu Yang 已提交
607
    """Tracks line numbers for includes, and the order in which includes appear.
Z
zhangjinchao01 已提交
608 609 610 611 612 613 614 615 616 617

  include_list contains list of lists of (header, line number) pairs.
  It's a lists of lists rather than just one flat list to make it
  easier to update across preprocessor boundaries.

  Call CheckNextIncludeOrder() once for each header in the file, passing
  in the type constants defined above. Calls in an illegal order will
  raise an _IncludeError with an appropriate error message.

  """
Y
Yu Yang 已提交
618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646
    # self._section will move monotonically through this set. If it ever
    # needs to move backwards, CheckNextIncludeOrder will raise an error.
    _INITIAL_SECTION = 0
    _MY_H_SECTION = 1
    _C_SECTION = 2
    _CPP_SECTION = 3
    _OTHER_H_SECTION = 4

    _TYPE_NAMES = {
        _C_SYS_HEADER: 'C system header',
        _CPP_SYS_HEADER: 'C++ system header',
        _LIKELY_MY_HEADER: 'header this file implements',
        _POSSIBLE_MY_HEADER: 'header this file may implement',
        _OTHER_HEADER: 'other header',
    }
    _SECTION_NAMES = {
        _INITIAL_SECTION: "... nothing. (This can't be an error.)",
        _MY_H_SECTION: 'a header this file implements',
        _C_SECTION: 'C system header',
        _CPP_SECTION: 'C++ system header',
        _OTHER_H_SECTION: 'other header',
    }

    def __init__(self):
        self.include_list = [[]]
        self.ResetSection('')

    def FindHeader(self, header):
        """Check if a header has already been included.
Z
zhangjinchao01 已提交
647 648 649 650 651 652 653

    Args:
      header: header to check.
    Returns:
      Line number of previous occurrence, or -1 if the header has not
      been seen before.
    """
Y
Yu Yang 已提交
654 655 656 657 658
        for section_list in self.include_list:
            for f in section_list:
                if f[0] == header:
                    return f[1]
        return -1
Z
zhangjinchao01 已提交
659

Y
Yu Yang 已提交
660 661
    def ResetSection(self, directive):
        """Reset section checking for preprocessor directive.
Z
zhangjinchao01 已提交
662 663 664 665

    Args:
      directive: preprocessor directive (e.g. "if", "else").
    """
Y
Yu Yang 已提交
666 667 668 669
        # The name of the current section.
        self._section = self._INITIAL_SECTION
        # The path of last found header.
        self._last_header = ''
Z
zhangjinchao01 已提交
670

Y
Yu Yang 已提交
671 672 673 674 675 676
        # Update list of includes.  Note that we never pop from the
        # include list.
        if directive in ('if', 'ifdef', 'ifndef'):
            self.include_list.append([])
        elif directive in ('else', 'elif'):
            self.include_list[-1] = []
Z
zhangjinchao01 已提交
677

Y
Yu Yang 已提交
678 679
    def SetLastHeader(self, header_path):
        self._last_header = header_path
Z
zhangjinchao01 已提交
680

Y
Yu Yang 已提交
681 682
    def CanonicalizeAlphabeticalOrder(self, header_path):
        """Returns a path canonicalized for alphabetical comparison.
Z
zhangjinchao01 已提交
683 684 685 686 687 688 689 690 691 692 693

    - replaces "-" with "_" so they both cmp the same.
    - removes '-inl' since we don't require them to be after the main header.
    - lowercase everything, just in case.

    Args:
      header_path: Path to be canonicalized.

    Returns:
      Canonicalized path.
    """
Y
Yu Yang 已提交
694
        return header_path.replace('-inl.h', '.h').replace('-', '_').lower()
Z
zhangjinchao01 已提交
695

Y
Yu Yang 已提交
696 697
    def IsInAlphabeticalOrder(self, clean_lines, linenum, header_path):
        """Check if a header is in alphabetical order with the previous header.
Z
zhangjinchao01 已提交
698 699 700 701 702 703 704 705 706

    Args:
      clean_lines: A CleansedLines instance containing the file.
      linenum: The number of the line to check.
      header_path: Canonicalized header to be checked.

    Returns:
      Returns true if the header is in alphabetical order.
    """
Y
Yu Yang 已提交
707 708 709 710 711 712 713 714 715
        # If previous section is different from current section, _last_header will
        # be reset to empty string, so it's always less than current header.
        #
        # If previous line was a blank line, assume that the headers are
        # intentionally sorted the way they are.
        if (self._last_header > header_path and
                Match(r'^\s*#\s*include\b', clean_lines.elided[linenum - 1])):
            return False
        return True
Z
zhangjinchao01 已提交
716

Y
Yu Yang 已提交
717 718
    def CheckNextIncludeOrder(self, header_type):
        """Returns a non-empty error message if the next header is out of order.
Z
zhangjinchao01 已提交
719 720 721 722 723 724 725 726 727 728 729 730

    This function also updates the internal state to be ready to check
    the next include.

    Args:
      header_type: One of the _XXX_HEADER constants defined above.

    Returns:
      The empty string if the header is in the right order, or an
      error message describing what's wrong.

    """
Y
Yu Yang 已提交
731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762
        error_message = ('Found %s after %s' % (
            self._TYPE_NAMES[header_type], self._SECTION_NAMES[self._section]))

        last_section = self._section

        if header_type == _C_SYS_HEADER:
            if self._section <= self._C_SECTION:
                self._section = self._C_SECTION
            else:
                self._last_header = ''
                return error_message
        elif header_type == _CPP_SYS_HEADER:
            if self._section <= self._CPP_SECTION:
                self._section = self._CPP_SECTION
            else:
                self._last_header = ''
                return error_message
        elif header_type == _LIKELY_MY_HEADER:
            if self._section <= self._MY_H_SECTION:
                self._section = self._MY_H_SECTION
            else:
                self._section = self._OTHER_H_SECTION
        elif header_type == _POSSIBLE_MY_HEADER:
            if self._section <= self._MY_H_SECTION:
                self._section = self._MY_H_SECTION
            else:
                # This will always be the fallback because we're not sure
                # enough that the header is associated with this file.
                self._section = self._OTHER_H_SECTION
        else:
            assert header_type == _OTHER_HEADER
            self._section = self._OTHER_H_SECTION
Z
zhangjinchao01 已提交
763

Y
Yu Yang 已提交
764 765
        if last_section != self._section:
            self._last_header = ''
Z
zhangjinchao01 已提交
766

Y
Yu Yang 已提交
767
        return ''
Z
zhangjinchao01 已提交
768 769 770


class _CppLintState(object):
Y
Yu Yang 已提交
771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803
    """Maintains module-wide state.."""

    def __init__(self):
        self.verbose_level = 1  # global setting.
        self.error_count = 0  # global count of reported errors
        # filters to apply when emitting error messages
        self.filters = _DEFAULT_FILTERS[:]
        # backup of filter list. Used to restore the state after each file.
        self._filters_backup = self.filters[:]
        self.counting = 'total'  # In what way are we counting errors?
        self.errors_by_category = {}  # string to int dict storing error counts

        # output format:
        # "emacs" - format that emacs can parse (default)
        # "vs7" - format that Microsoft Visual Studio 7 can parse
        self.output_format = 'emacs'

    def SetOutputFormat(self, output_format):
        """Sets the output format for errors."""
        self.output_format = output_format

    def SetVerboseLevel(self, level):
        """Sets the module's verbosity, and returns the previous setting."""
        last_verbose_level = self.verbose_level
        self.verbose_level = level
        return last_verbose_level

    def SetCountingStyle(self, counting_style):
        """Sets the module's counting options."""
        self.counting = counting_style

    def SetFilters(self, filters):
        """Sets the error-message filters.
Z
zhangjinchao01 已提交
804 805 806 807 808 809 810 811 812 813 814 815

    These filters are applied when deciding whether to emit a given
    error message.

    Args:
      filters: A string of comma-separated filters (eg "+whitespace/indent").
               Each filter should start with + or -; else we die.

    Raises:
      ValueError: The comma-separated filters did not all start with '+' or '-'.
                  E.g. "-,+whitespace,-whitespace/indent,whitespace/badfilter"
    """
Y
Yu Yang 已提交
816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860
        # Default filters always have less priority than the flag ones.
        self.filters = _DEFAULT_FILTERS[:]
        self.AddFilters(filters)

    def AddFilters(self, filters):
        """ Adds more filters to the existing list of error-message filters. """
        for filt in filters.split(','):
            clean_filt = filt.strip()
            if clean_filt:
                self.filters.append(clean_filt)
        for filt in self.filters:
            if not (filt.startswith('+') or filt.startswith('-')):
                raise ValueError(
                    'Every filter in --filters must start with + or -'
                    ' (%s does not)' % filt)

    def BackupFilters(self):
        """ Saves the current filter list to backup storage."""
        self._filters_backup = self.filters[:]

    def RestoreFilters(self):
        """ Restores filters previously backed up."""
        self.filters = self._filters_backup[:]

    def ResetErrorCounts(self):
        """Sets the module's error statistic back to zero."""
        self.error_count = 0
        self.errors_by_category = {}

    def IncrementErrorCount(self, category):
        """Bumps the module's error statistic."""
        self.error_count += 1
        if self.counting in ('toplevel', 'detailed'):
            if self.counting != 'detailed':
                category = category.split('/')[0]
            if category not in self.errors_by_category:
                self.errors_by_category[category] = 0
            self.errors_by_category[category] += 1

    def PrintErrorCounts(self):
        """Print a summary of errors by category, and the total."""
        for category, count in self.errors_by_category.iteritems():
            sys.stdout.write('Category \'%s\' errors found: %d\n' %
                             (category, count))
        sys.stdout.write('Total errors found: %d\n' % self.error_count)
Z
zhangjinchao01 已提交
861 862 863 864 865 866


_cpplint_state = _CppLintState()


def _OutputFormat():
Y
Yu Yang 已提交
867 868
    """Gets the module's output format."""
    return _cpplint_state.output_format
Z
zhangjinchao01 已提交
869 870 871


def _SetOutputFormat(output_format):
Y
Yu Yang 已提交
872 873
    """Sets the module's output format."""
    _cpplint_state.SetOutputFormat(output_format)
Z
zhangjinchao01 已提交
874 875 876


def _VerboseLevel():
Y
Yu Yang 已提交
877 878
    """Returns the module's verbosity setting."""
    return _cpplint_state.verbose_level
Z
zhangjinchao01 已提交
879 880 881


def _SetVerboseLevel(level):
Y
Yu Yang 已提交
882 883
    """Sets the module's verbosity, and returns the previous setting."""
    return _cpplint_state.SetVerboseLevel(level)
Z
zhangjinchao01 已提交
884 885 886


def _SetCountingStyle(level):
Y
Yu Yang 已提交
887 888
    """Sets the module's counting options."""
    _cpplint_state.SetCountingStyle(level)
Z
zhangjinchao01 已提交
889 890 891


def _Filters():
Y
Yu Yang 已提交
892 893
    """Returns the module's list of output filters, as a list."""
    return _cpplint_state.filters
Z
zhangjinchao01 已提交
894 895 896


def _SetFilters(filters):
Y
Yu Yang 已提交
897
    """Sets the module's error-message filters.
Z
zhangjinchao01 已提交
898 899 900 901 902 903 904 905

  These filters are applied when deciding whether to emit a given
  error message.

  Args:
    filters: A string of comma-separated filters (eg "whitespace/indent").
             Each filter should start with + or -; else we die.
  """
Y
Yu Yang 已提交
906 907
    _cpplint_state.SetFilters(filters)

Z
zhangjinchao01 已提交
908 909

def _AddFilters(filters):
Y
Yu Yang 已提交
910
    """Adds more filter overrides.
Z
zhangjinchao01 已提交
911 912 913 914 915 916 917 918

  Unlike _SetFilters, this function does not reset the current list of filters
  available.

  Args:
    filters: A string of comma-separated filters (eg "whitespace/indent").
             Each filter should start with + or -; else we die.
  """
Y
Yu Yang 已提交
919 920
    _cpplint_state.AddFilters(filters)

Z
zhangjinchao01 已提交
921 922

def _BackupFilters():
Y
Yu Yang 已提交
923 924 925
    """ Saves the current filter list to backup storage."""
    _cpplint_state.BackupFilters()

Z
zhangjinchao01 已提交
926 927

def _RestoreFilters():
Y
Yu Yang 已提交
928 929 930
    """ Restores filters previously backed up."""
    _cpplint_state.RestoreFilters()

Z
zhangjinchao01 已提交
931 932

class _FunctionState(object):
Y
Yu Yang 已提交
933
    """Tracks current function name and the number of lines in its body."""
Z
zhangjinchao01 已提交
934

Y
Yu Yang 已提交
935 936
    _NORMAL_TRIGGER = 250  # for --v=0, 500 for --v=1, etc.
    _TEST_TRIGGER = 400  # about 50% more than _NORMAL_TRIGGER.
Z
zhangjinchao01 已提交
937

Y
Yu Yang 已提交
938 939 940 941
    def __init__(self):
        self.in_a_function = False
        self.lines_in_function = 0
        self.current_function = ''
Z
zhangjinchao01 已提交
942

Y
Yu Yang 已提交
943 944
    def Begin(self, function_name):
        """Start analyzing function body.
Z
zhangjinchao01 已提交
945 946 947 948

    Args:
      function_name: The name of the function being tracked.
    """
Y
Yu Yang 已提交
949 950 951
        self.in_a_function = True
        self.lines_in_function = 0
        self.current_function = function_name
Z
zhangjinchao01 已提交
952

Y
Yu Yang 已提交
953 954 955 956
    def Count(self):
        """Count line in current function body."""
        if self.in_a_function:
            self.lines_in_function += 1
Z
zhangjinchao01 已提交
957

Y
Yu Yang 已提交
958 959
    def Check(self, error, filename, linenum):
        """Report if too many lines in function body.
Z
zhangjinchao01 已提交
960 961 962 963 964 965

    Args:
      error: The function to call with any errors found.
      filename: The name of the current file.
      linenum: The number of the line to check.
    """
Y
Yu Yang 已提交
966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986
        if Match(r'T(EST|est)', self.current_function):
            base_trigger = self._TEST_TRIGGER
        else:
            base_trigger = self._NORMAL_TRIGGER
        trigger = base_trigger * 2**_VerboseLevel()

        if self.lines_in_function > trigger:
            error_level = int(
                math.log(self.lines_in_function / base_trigger, 2))
            # 50 => 0, 100 => 1, 200 => 2, 400 => 3, 800 => 4, 1600 => 5, ...
            if error_level > 5:
                error_level = 5
            error(filename, linenum, 'readability/fn_size', error_level,
                  'Small and focused functions are preferred:'
                  ' %s has %d non-comment lines'
                  ' (error triggered by exceeding %d lines).' % (
                      self.current_function, self.lines_in_function, trigger))

    def End(self):
        """Stop analyzing function body."""
        self.in_a_function = False
Z
zhangjinchao01 已提交
987 988 989


class _IncludeError(Exception):
Y
Yu Yang 已提交
990 991
    """Indicates a problem with the include order in a file."""
    pass
Z
zhangjinchao01 已提交
992 993 994


class FileInfo(object):
Y
Yu Yang 已提交
995
    """Provides utility functions for filenames.
Z
zhangjinchao01 已提交
996 997 998 999 1000

  FileInfo provides easy access to the components of a file's path
  relative to the project root.
  """

Y
Yu Yang 已提交
1001 1002
    def __init__(self, filename):
        self._filename = filename
Z
zhangjinchao01 已提交
1003

Y
Yu Yang 已提交
1004 1005 1006
    def FullName(self):
        """Make Windows paths like Unix."""
        return os.path.abspath(self._filename).replace('\\', '/')
Z
zhangjinchao01 已提交
1007

Y
Yu Yang 已提交
1008 1009
    def RepositoryName(self):
        """FullName after removing the local path to the repository.
Z
zhangjinchao01 已提交
1010 1011 1012 1013 1014 1015 1016 1017

    If we have a real absolute path name here we can try to do something smart:
    detecting the root of the checkout and truncating /path/to/checkout from
    the name so that we get header guards that don't include things like
    "C:\Documents and Settings\..." or "/home/username/..." in them and thus
    people on different computers who have checked the source out to different
    locations won't see bogus errors.
    """
Y
Yu Yang 已提交
1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054
        fullname = self.FullName()

        if os.path.exists(fullname):
            project_dir = os.path.dirname(fullname)

            if os.path.exists(os.path.join(project_dir, ".svn")):
                # If there's a .svn file in the current directory, we recursively look
                # up the directory tree for the top of the SVN checkout
                root_dir = project_dir
                one_up_dir = os.path.dirname(root_dir)
                while os.path.exists(os.path.join(one_up_dir, ".svn")):
                    root_dir = os.path.dirname(root_dir)
                    one_up_dir = os.path.dirname(one_up_dir)

                prefix = os.path.commonprefix([root_dir, project_dir])
                return fullname[len(prefix) + 1:]

            # Not SVN <= 1.6? Try to find a git, hg, or svn top level directory by
            # searching up from the current path.
            root_dir = os.path.dirname(fullname)
            while (root_dir != os.path.dirname(root_dir) and
                   not os.path.exists(os.path.join(root_dir, ".git")) and
                   not os.path.exists(os.path.join(root_dir, ".hg")) and
                   not os.path.exists(os.path.join(root_dir, ".svn"))):
                root_dir = os.path.dirname(root_dir)

            if (os.path.exists(os.path.join(root_dir, ".git")) or
                    os.path.exists(os.path.join(root_dir, ".hg")) or
                    os.path.exists(os.path.join(root_dir, ".svn"))):
                prefix = os.path.commonprefix([root_dir, project_dir])
                return fullname[len(prefix) + 1:]

        # Don't know what to do; header guard warnings may be wrong...
        return fullname

    def Split(self):
        """Splits the file into the directory, basename, and extension.
Z
zhangjinchao01 已提交
1055 1056 1057 1058 1059 1060 1061 1062

    For 'chrome/browser/browser.cc', Split() would
    return ('chrome/browser', 'browser', '.cc')

    Returns:
      A tuple of (directory, basename, extension).
    """

Y
Yu Yang 已提交
1063 1064 1065
        googlename = self.RepositoryName()
        project, rest = os.path.split(googlename)
        return (project, ) + os.path.splitext(rest)
Z
zhangjinchao01 已提交
1066

Y
Yu Yang 已提交
1067 1068 1069
    def BaseName(self):
        """File base name - text after the final slash, before the final period."""
        return self.Split()[1]
Z
zhangjinchao01 已提交
1070

Y
Yu Yang 已提交
1071 1072 1073
    def Extension(self):
        """File extension - text following the final period."""
        return self.Split()[2]
Z
zhangjinchao01 已提交
1074

Y
Yu Yang 已提交
1075 1076 1077
    def NoExtension(self):
        """File has no source file extension."""
        return '/'.join(self.Split()[0:2])
Z
zhangjinchao01 已提交
1078

Y
Yu Yang 已提交
1079 1080 1081
    def IsSource(self):
        """File has a source file extension."""
        return self.Extension()[1:] in ('c', 'cc', 'cpp', 'cxx')
Z
zhangjinchao01 已提交
1082 1083 1084


def _ShouldPrintError(category, confidence, linenum):
Y
Yu Yang 已提交
1085
    """If confidence >= verbose, category passes filter and is not suppressed."""
Z
zhangjinchao01 已提交
1086

Y
Yu Yang 已提交
1087 1088 1089 1090 1091
    # There are three ways we might decide not to print an error message:
    # a "NOLINT(category)" comment appears in the source,
    # the verbosity level isn't high enough, or the filters filter it out.
    if IsErrorSuppressedByNolint(category, linenum):
        return False
Z
zhangjinchao01 已提交
1092

Y
Yu Yang 已提交
1093 1094
    if confidence < _cpplint_state.verbose_level:
        return False
Z
zhangjinchao01 已提交
1095

Y
Yu Yang 已提交
1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107
    is_filtered = False
    for one_filter in _Filters():
        if one_filter.startswith('-'):
            if category.startswith(one_filter[1:]):
                is_filtered = True
        elif one_filter.startswith('+'):
            if category.startswith(one_filter[1:]):
                is_filtered = False
        else:
            assert False  # should have been checked for in SetFilter.
    if is_filtered:
        return False
Z
zhangjinchao01 已提交
1108

Y
Yu Yang 已提交
1109
    return True
Z
zhangjinchao01 已提交
1110 1111 1112


def Error(filename, linenum, category, confidence, message):
Y
Yu Yang 已提交
1113
    """Logs the fact we've found a lint error.
Z
zhangjinchao01 已提交
1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133

  We log where the error was found, and also our confidence in the error,
  that is, how certain we are this is a legitimate style regression, and
  not a misidentification or a use that's sometimes justified.

  False positives can be suppressed by the use of
  "cpplint(category)"  comments on the offending line.  These are
  parsed into _error_suppressions.

  Args:
    filename: The name of the file containing the error.
    linenum: The number of the line containing the error.
    category: A string used to describe the "category" this bug
      falls under: "whitespace", say, or "runtime".  Categories
      may have a hierarchy separated by slashes: "whitespace/indent".
    confidence: A number from 1-5 representing a confidence score for
      the error, with 5 meaning that we are certain of the problem,
      and 1 meaning that it could be a legitimate construct.
    message: The error message.
  """
Y
Yu Yang 已提交
1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144
    if _ShouldPrintError(category, confidence, linenum):
        _cpplint_state.IncrementErrorCount(category)
        if _cpplint_state.output_format == 'vs7':
            sys.stderr.write('%s(%s):  %s  [%s] [%d]\n' %
                             (filename, linenum, message, category, confidence))
        elif _cpplint_state.output_format == 'eclipse':
            sys.stderr.write('%s:%s: warning: %s  [%s] [%d]\n' %
                             (filename, linenum, message, category, confidence))
        else:
            sys.stderr.write('%s:%s:  %s  [%s] [%d]\n' %
                             (filename, linenum, message, category, confidence))
Z
zhangjinchao01 已提交
1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160


# Matches standard C++ escape sequences per 2.13.2.3 of the C++ standard.
_RE_PATTERN_CLEANSE_LINE_ESCAPES = re.compile(
    r'\\([abfnrtv?"\\\']|\d+|x[0-9a-fA-F]+)')
# Match a single C style comment on the same line.
_RE_PATTERN_C_COMMENTS = r'/\*(?:[^*]|\*(?!/))*\*/'
# Matches multi-line C style comments.
# This RE is a little bit more complicated than one might expect, because we
# have to take care of space removals tools so we can handle comments inside
# statements better.
# The current rule is: We only clear spaces from both sides when we're at the
# end of the line. Otherwise, we try to remove spaces from the right side,
# if this doesn't work we try on left side but only if there's a non-character
# on the right.
_RE_PATTERN_CLEANSE_LINE_C_COMMENTS = re.compile(
Y
Yu Yang 已提交
1161 1162
    r'(\s*' + _RE_PATTERN_C_COMMENTS + r'\s*$|' + _RE_PATTERN_C_COMMENTS +
    r'\s+|' + r'\s+' + _RE_PATTERN_C_COMMENTS + r'(?=\W)|' +
Z
zhangjinchao01 已提交
1163 1164 1165 1166
    _RE_PATTERN_C_COMMENTS + r')')


def IsCppString(line):
Y
Yu Yang 已提交
1167
    """Does line terminate so, that the next symbol is in string constant.
Z
zhangjinchao01 已提交
1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178

  This function does not consider single-line nor multi-line comments.

  Args:
    line: is a partial line of code starting from the 0..n.

  Returns:
    True, if next character appended to 'line' is inside a
    string constant.
  """

Y
Yu Yang 已提交
1179 1180
    line = line.replace(r'\\', 'XX')  # after this, \\" does not match to \"
    return ((line.count('"') - line.count(r'\"') - line.count("'\"'")) & 1) == 1
Z
zhangjinchao01 已提交
1181 1182 1183


def CleanseRawStrings(raw_lines):
Y
Yu Yang 已提交
1184
    """Removes C++11 raw strings from lines.
Z
zhangjinchao01 已提交
1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202

    Before:
      static const char kData[] = R"(
          multi-line string
          )";

    After:
      static const char kData[] = ""
          (replaced by blank line)
          "";

  Args:
    raw_lines: list of raw lines.

  Returns:
    list of lines with C++11 raw strings replaced by empty strings.
  """

Y
Yu Yang 已提交
1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248
    delimiter = None
    lines_without_raw_strings = []
    for line in raw_lines:
        if delimiter:
            # Inside a raw string, look for the end
            end = line.find(delimiter)
            if end >= 0:
                # Found the end of the string, match leading space for this
                # line and resume copying the original lines, and also insert
                # a "" on the last line.
                leading_space = Match(r'^(\s*)\S', line)
                line = leading_space.group(1) + '""' + line[end + len(
                    delimiter):]
                delimiter = None
            else:
                # Haven't found the end yet, append a blank line.
                line = '""'

        # Look for beginning of a raw string, and replace them with
        # empty strings.  This is done in a loop to handle multiple raw
        # strings on the same line.
        while delimiter is None:
            # Look for beginning of a raw string.
            # See 2.14.15 [lex.string] for syntax.
            matched = Match(r'^(.*)\b(?:R|u8R|uR|UR|LR)"([^\s\\()]*)\((.*)$',
                            line)
            if matched:
                delimiter = ')' + matched.group(2) + '"'

                end = matched.group(3).find(delimiter)
                if end >= 0:
                    # Raw string ended on same line
                    line = (matched.group(1) + '""' +
                            matched.group(3)[end + len(delimiter):])
                    delimiter = None
                else:
                    # Start of a multi-line raw string
                    line = matched.group(1) + '""'
            else:
                break

        lines_without_raw_strings.append(line)

    # TODO(unknown): if delimiter is not None here, we might want to
    # emit a warning for unterminated string.
    return lines_without_raw_strings
Z
zhangjinchao01 已提交
1249 1250 1251


def FindNextMultiLineCommentStart(lines, lineix):
Y
Yu Yang 已提交
1252 1253 1254 1255 1256 1257 1258 1259
    """Find the beginning marker for a multiline comment."""
    while lineix < len(lines):
        if lines[lineix].strip().startswith('/*'):
            # Only return this marker if the comment goes beyond this line
            if lines[lineix].strip().find('*/', 2) < 0:
                return lineix
        lineix += 1
    return len(lines)
Z
zhangjinchao01 已提交
1260 1261 1262


def FindNextMultiLineCommentEnd(lines, lineix):
Y
Yu Yang 已提交
1263 1264 1265 1266 1267 1268
    """We are inside a comment, find the end marker."""
    while lineix < len(lines):
        if lines[lineix].strip().endswith('*/'):
            return lineix
        lineix += 1
    return len(lines)
Z
zhangjinchao01 已提交
1269 1270 1271


def RemoveMultiLineCommentsFromRange(lines, begin, end):
Y
Yu Yang 已提交
1272 1273 1274 1275 1276
    """Clears a range of lines for multi-line comments."""
    # Having // dummy comments makes the lines non-empty, so we will not get
    # unnecessary blank line warnings later in the code.
    for i in range(begin, end):
        lines[i] = '/**/'
Z
zhangjinchao01 已提交
1277 1278 1279


def RemoveMultiLineComments(filename, lines, error):
Y
Yu Yang 已提交
1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292
    """Removes multiline (c-style) comments from lines."""
    lineix = 0
    while lineix < len(lines):
        lineix_begin = FindNextMultiLineCommentStart(lines, lineix)
        if lineix_begin >= len(lines):
            return
        lineix_end = FindNextMultiLineCommentEnd(lines, lineix_begin)
        if lineix_end >= len(lines):
            error(filename, lineix_begin + 1, 'readability/multiline_comment',
                  5, 'Could not find end of multi-line comment')
            return
        RemoveMultiLineCommentsFromRange(lines, lineix_begin, lineix_end + 1)
        lineix = lineix_end + 1
Z
zhangjinchao01 已提交
1293 1294 1295


def CleanseComments(line):
Y
Yu Yang 已提交
1296
    """Removes //-comments and single-line C-style /* */ comments.
Z
zhangjinchao01 已提交
1297 1298 1299 1300 1301 1302 1303

  Args:
    line: A line of C++ source.

  Returns:
    The line with single-line comments removed.
  """
Y
Yu Yang 已提交
1304 1305 1306 1307 1308
    commentpos = line.find('//')
    if commentpos != -1 and not IsCppString(line[:commentpos]):
        line = line[:commentpos].rstrip()
    # get rid of /* ... */
    return _RE_PATTERN_CLEANSE_LINE_C_COMMENTS.sub('', line)
Z
zhangjinchao01 已提交
1309 1310 1311


class CleansedLines(object):
Y
Yu Yang 已提交
1312
    """Holds 4 copies of all lines with different preprocessing applied to them.
Z
zhangjinchao01 已提交
1313 1314 1315 1316 1317 1318 1319 1320 1321

  1) elided member contains lines without strings and comments.
  2) lines member contains lines without comments.
  3) raw_lines member contains all the lines without processing.
  4) lines_without_raw_strings member is same as raw_lines, but with C++11 raw
     strings removed.
  All these members are of <type 'list'>, and of the same length.
  """

Y
Yu Yang 已提交
1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341
    def __init__(self, lines):
        self.elided = []
        self.lines = []
        self.raw_lines = lines
        self.num_lines = len(lines)
        self.lines_without_raw_strings = CleanseRawStrings(lines)
        for linenum in range(len(self.lines_without_raw_strings)):
            self.lines.append(
                CleanseComments(self.lines_without_raw_strings[linenum]))
            elided = self._CollapseStrings(self.lines_without_raw_strings[
                linenum])
            self.elided.append(CleanseComments(elided))

    def NumLines(self):
        """Returns the number of lines represented."""
        return self.num_lines

    @staticmethod
    def _CollapseStrings(elided):
        """Collapses strings and chars on a line to simple "" or '' blocks.
Z
zhangjinchao01 已提交
1342 1343 1344 1345 1346 1347 1348 1349 1350

    We nix strings first so we're not fooled by text like '"http://"'

    Args:
      elided: The line being processed.

    Returns:
      The line with collapsed strings.
    """
Y
Yu Yang 已提交
1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405
        if _RE_PATTERN_INCLUDE.match(elided):
            return elided

        # Remove escaped characters first to make quote/single quote collapsing
        # basic.  Things that look like escaped characters shouldn't occur
        # outside of strings and chars.
        elided = _RE_PATTERN_CLEANSE_LINE_ESCAPES.sub('', elided)

        # Replace quoted strings and digit separators.  Both single quotes
        # and double quotes are processed in the same loop, otherwise
        # nested quotes wouldn't work.
        collapsed = ''
        while True:
            # Find the first quote character
            match = Match(r'^([^\'"]*)([\'"])(.*)$', elided)
            if not match:
                collapsed += elided
                break
            head, quote, tail = match.groups()

            if quote == '"':
                # Collapse double quoted strings
                second_quote = tail.find('"')
                if second_quote >= 0:
                    collapsed += head + '""'
                    elided = tail[second_quote + 1:]
                else:
                    # Unmatched double quote, don't bother processing the rest
                    # of the line since this is probably a multiline string.
                    collapsed += elided
                    break
            else:
                # Found single quote, check nearby text to eliminate digit separators.
                #
                # There is no special handling for floating point here, because
                # the integer/fractional/exponent parts would all be parsed
                # correctly as long as there are digits on both sides of the
                # separator.  So we are fine as long as we don't see something
                # like "0.'3" (gcc 4.9.0 will not allow this literal).
                if Search(r'\b(?:0[bBxX]?|[1-9])[0-9a-fA-F]*$', head):
                    match_literal = Match(r'^((?:\'?[0-9a-zA-Z_])*)(.*)$',
                                          "'" + tail)
                    collapsed += head + match_literal.group(1).replace("'", '')
                    elided = match_literal.group(2)
                else:
                    second_quote = tail.find('\'')
                    if second_quote >= 0:
                        collapsed += head + "''"
                        elided = tail[second_quote + 1:]
                    else:
                        # Unmatched single quote
                        collapsed += elided
                        break

        return collapsed
Z
zhangjinchao01 已提交
1406 1407 1408


def FindEndOfExpressionInLine(line, startpos, stack):
Y
Yu Yang 已提交
1409
    """Find the position just after the end of current parenthesized expression.
Z
zhangjinchao01 已提交
1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420

  Args:
    line: a CleansedLines line.
    startpos: start searching at this position.
    stack: nesting stack at startpos.

  Returns:
    On finding matching end: (index just after matching end, None)
    On finding an unclosed expression: (-1, None)
    Otherwise: (-1, new stack at end of this line)
  """
Y
Yu Yang 已提交
1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483
    for i in xrange(startpos, len(line)):
        char = line[i]
        if char in '([{':
            # Found start of parenthesized expression, push to expression stack
            stack.append(char)
        elif char == '<':
            # Found potential start of template argument list
            if i > 0 and line[i - 1] == '<':
                # Left shift operator
                if stack and stack[-1] == '<':
                    stack.pop()
                    if not stack:
                        return (-1, None)
            elif i > 0 and Search(r'\boperator\s*$', line[0:i]):
                # operator<, don't add to stack
                continue
            else:
                # Tentative start of template argument list
                stack.append('<')
        elif char in ')]}':
            # Found end of parenthesized expression.
            #
            # If we are currently expecting a matching '>', the pending '<'
            # must have been an operator.  Remove them from expression stack.
            while stack and stack[-1] == '<':
                stack.pop()
            if not stack:
                return (-1, None)
            if ((stack[-1] == '(' and char == ')') or
                (stack[-1] == '[' and char == ']') or
                (stack[-1] == '{' and char == '}')):
                stack.pop()
                if not stack:
                    return (i + 1, None)
            else:
                # Mismatched parentheses
                return (-1, None)
        elif char == '>':
            # Found potential end of template argument list.

            # Ignore "->" and operator functions
            if (i > 0 and (line[i - 1] == '-' or Search(r'\boperator\s*$',
                                                        line[0:i - 1]))):
                continue

            # Pop the stack if there is a matching '<'.  Otherwise, ignore
            # this '>' since it must be an operator.
            if stack:
                if stack[-1] == '<':
                    stack.pop()
                    if not stack:
                        return (i + 1, None)
        elif char == ';':
            # Found something that look like end of statements.  If we are currently
            # expecting a '>', the matching '<' must have been an operator, since
            # template argument list should not contain statements.
            while stack and stack[-1] == '<':
                stack.pop()
            if not stack:
                return (-1, None)

    # Did not find end of expression or unbalanced parentheses on this line
    return (-1, stack)
Z
zhangjinchao01 已提交
1484 1485 1486


def CloseExpression(clean_lines, linenum, pos):
Y
Yu Yang 已提交
1487
    """If input points to ( or { or [ or <, finds the position that closes it.
Z
zhangjinchao01 已提交
1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509

  If lines[linenum][pos] points to a '(' or '{' or '[' or '<', finds the
  linenum/pos that correspond to the closing of the expression.

  TODO(unknown): cpplint spends a fair bit of time matching parentheses.
  Ideally we would want to index all opening and closing parentheses once
  and have CloseExpression be just a simple lookup, but due to preprocessor
  tricks, this is not so easy.

  Args:
    clean_lines: A CleansedLines instance containing the file.
    linenum: The number of the line to check.
    pos: A position on the line.

  Returns:
    A tuple (line, linenum, pos) pointer *past* the closing brace, or
    (line, len(lines), -1) if we never find a close.  Note we ignore
    strings and comments when matching; and the line we return is the
    'cleansed' line at linenum.
  """

    line = clean_lines.elided[linenum]
Y
Yu Yang 已提交
1510 1511 1512 1513 1514
    if (line[pos] not in '({[<') or Match(r'<[<=]', line[pos:]):
        return (line, clean_lines.NumLines(), -1)

    # Check first line
    (end_pos, stack) = FindEndOfExpressionInLine(line, pos, [])
Z
zhangjinchao01 已提交
1515
    if end_pos > -1:
Y
Yu Yang 已提交
1516 1517 1518 1519 1520 1521 1522 1523 1524
        return (line, linenum, end_pos)

    # Continue scanning forward
    while stack and linenum < clean_lines.NumLines() - 1:
        linenum += 1
        line = clean_lines.elided[linenum]
        (end_pos, stack) = FindEndOfExpressionInLine(line, 0, stack)
        if end_pos > -1:
            return (line, linenum, end_pos)
Z
zhangjinchao01 已提交
1525

Y
Yu Yang 已提交
1526 1527
    # Did not find end of expression before end of file, give up
    return (line, clean_lines.NumLines(), -1)
Z
zhangjinchao01 已提交
1528 1529 1530


def FindStartOfExpressionInLine(line, endpos, stack):
Y
Yu Yang 已提交
1531
    """Find position at the matching start of current expression.
Z
zhangjinchao01 已提交
1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545

  This is almost the reverse of FindEndOfExpressionInLine, but note
  that the input position and returned position differs by 1.

  Args:
    line: a CleansedLines line.
    endpos: start searching at this position.
    stack: nesting stack at endpos.

  Returns:
    On finding matching start: (index at matching start, None)
    On finding an unclosed expression: (-1, None)
    Otherwise: (-1, new stack at beginning of this line)
  """
Y
Yu Yang 已提交
1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600
    i = endpos
    while i >= 0:
        char = line[i]
        if char in ')]}':
            # Found end of expression, push to expression stack
            stack.append(char)
        elif char == '>':
            # Found potential end of template argument list.
            #
            # Ignore it if it's a "->" or ">=" or "operator>"
            if (i > 0 and
                (line[i - 1] == '-' or Match(r'\s>=\s', line[i - 1:]) or
                 Search(r'\boperator\s*$', line[0:i]))):
                i -= 1
            else:
                stack.append('>')
        elif char == '<':
            # Found potential start of template argument list
            if i > 0 and line[i - 1] == '<':
                # Left shift operator
                i -= 1
            else:
                # If there is a matching '>', we can pop the expression stack.
                # Otherwise, ignore this '<' since it must be an operator.
                if stack and stack[-1] == '>':
                    stack.pop()
                    if not stack:
                        return (i, None)
        elif char in '([{':
            # Found start of expression.
            #
            # If there are any unmatched '>' on the stack, they must be
            # operators.  Remove those.
            while stack and stack[-1] == '>':
                stack.pop()
            if not stack:
                return (-1, None)
            if ((char == '(' and stack[-1] == ')') or
                (char == '[' and stack[-1] == ']') or
                (char == '{' and stack[-1] == '}')):
                stack.pop()
                if not stack:
                    return (i, None)
            else:
                # Mismatched parentheses
                return (-1, None)
        elif char == ';':
            # Found something that look like end of statements.  If we are currently
            # expecting a '<', the matching '>' must have been an operator, since
            # template argument list should not contain statements.
            while stack and stack[-1] == '>':
                stack.pop()
            if not stack:
                return (-1, None)

Z
zhangjinchao01 已提交
1601
        i -= 1
Y
Yu Yang 已提交
1602 1603

    return (-1, stack)
Z
zhangjinchao01 已提交
1604 1605 1606


def ReverseCloseExpression(clean_lines, linenum, pos):
Y
Yu Yang 已提交
1607
    """If input points to ) or } or ] or >, finds the position that opens it.
Z
zhangjinchao01 已提交
1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623

  If lines[linenum][pos] points to a ')' or '}' or ']' or '>', finds the
  linenum/pos that correspond to the opening of the expression.

  Args:
    clean_lines: A CleansedLines instance containing the file.
    linenum: The number of the line to check.
    pos: A position on the line.

  Returns:
    A tuple (line, linenum, pos) pointer *at* the opening brace, or
    (line, 0, -1) if we never find the matching opening brace.  Note
    we ignore strings and comments when matching; and the line we
    return is the 'cleansed' line at linenum.
  """
    line = clean_lines.elided[linenum]
Y
Yu Yang 已提交
1624 1625
    if line[pos] not in ')}]>':
        return (line, 0, -1)
Z
zhangjinchao01 已提交
1626

Y
Yu Yang 已提交
1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642
    # Check last line
    (start_pos, stack) = FindStartOfExpressionInLine(line, pos, [])
    if start_pos > -1:
        return (line, linenum, start_pos)

    # Continue scanning backward
    while stack and linenum > 0:
        linenum -= 1
        line = clean_lines.elided[linenum]
        (start_pos, stack) = FindStartOfExpressionInLine(line,
                                                         len(line) - 1, stack)
        if start_pos > -1:
            return (line, linenum, start_pos)

    # Did not find start of expression before beginning of file, give up
    return (line, 0, -1)
Z
zhangjinchao01 已提交
1643 1644 1645


def CheckForCopyright(filename, lines, error):
Y
Yu Yang 已提交
1646
    """Logs an error if no Copyright message appears at the top of the file."""
Z
zhangjinchao01 已提交
1647

Y
Yu Yang 已提交
1648 1649 1650 1651 1652 1653 1654
    # We'll say it should occur by line 10. Don't forget there's a
    # dummy line at the front.
    for line in xrange(1, min(len(lines), 11)):
        if re.search(r'Copyright', lines[line], re.I): break
    else:  # means no copyright line was found
        error(filename, 0, 'legal/copyright', 5, 'No copyright message found.  '
              'You should have a line: "Copyright [year] <Copyright Owner>"')
Z
zhangjinchao01 已提交
1655 1656 1657


def GetIndentLevel(line):
Y
Yu Yang 已提交
1658
    """Return the number of leading spaces in line.
Z
zhangjinchao01 已提交
1659 1660 1661 1662 1663 1664 1665

  Args:
    line: A string to check.

  Returns:
    An integer count of leading spaces, possibly zero.
  """
Y
Yu Yang 已提交
1666 1667 1668 1669 1670
    indent = Match(r'^( *)\S', line)
    if indent:
        return len(indent.group(1))
    else:
        return 0
Z
zhangjinchao01 已提交
1671 1672 1673


def GetHeaderGuardCPPVariable(filename):
Y
Yu Yang 已提交
1674
    """Returns the CPP variable that should be used as a header guard.
Z
zhangjinchao01 已提交
1675 1676 1677 1678 1679 1680 1681 1682 1683

  Args:
    filename: The name of a C++ header file.

  Returns:
    The CPP variable that should be used as a header guard in the
    named file.

  """
Y
Yu Yang 已提交
1684 1685
    filename = os.path.basename(filename)
    return re.sub(r'[^a-zA-Z0-9]', '_', filename).upper() + '_'
Z
zhangjinchao01 已提交
1686 1687 1688


def CheckForHeaderGuard(filename, clean_lines, error):
Y
Yu Yang 已提交
1689
    """Checks that the file contains a header guard.
Z
zhangjinchao01 已提交
1690 1691 1692 1693 1694 1695 1696 1697 1698 1699

  Logs an error if no #ifndef header guard is present.  For other
  headers, checks that the full pathname is used.

  Args:
    filename: The name of the C++ header file.
    clean_lines: A CleansedLines instance containing the file.
    error: The function to call with any errors found.
  """

Y
Yu Yang 已提交
1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756
    # Don't check for header guards if there are error suppression
    # comments somewhere in this file.
    #
    # Because this is silencing a warning for a nonexistent line, we
    # only support the very specific NOLINT(build/header_guard) syntax,
    # and not the general NOLINT or NOLINT(*) syntax.
    raw_lines = clean_lines.lines_without_raw_strings
    for i in raw_lines:
        if Search(r'//\s*NOLINT\(build/header_guard\)', i):
            return

    cppvar = GetHeaderGuardCPPVariable(filename)

    ifndef = ''
    ifndef_linenum = 0
    define = ''
    endif = ''
    endif_linenum = 0
    pragma_linenum = -1
    for linenum, line in enumerate(raw_lines):
        linesplit = line.split()
        if len(linesplit) >= 2:
            if linesplit[0] == '#pragma' and linesplit[1] == 'once':
                pragma_linenum = linenum
            # find the first occurrence of #ifndef and #define, save arg
            if not ifndef and linesplit[0] == '#ifndef':
                # set ifndef to the header guard presented on the #ifndef line.
                ifndef = linesplit[1]
                ifndef_linenum = linenum
            if not define and linesplit[0] == '#define':
                define = linesplit[1]
        # find the last occurrence of #endif, save entire line
        if line.startswith('#endif'):
            endif = line
            endif_linenum = linenum
    if pragma_linenum != -1:
        return  # short path for pragma once
    if not ifndef or not define or ifndef != define:
        error(filename, 0, 'build/header_guard', 5,
              'No #ifndef header guard found, suggested CPP variable is: %s' %
              cppvar)
        return

    # The guard should be PATH_FILE_H_, but we also allow PATH_FILE_H__
    # for backward compatibility.
    if ifndef != cppvar:
        error_level = 0
        if ifndef != cppvar + '_':
            error_level = 5

        ParseNolintSuppressions(filename, raw_lines[ifndef_linenum],
                                ifndef_linenum, error)
        error(filename, ifndef_linenum, 'build/header_guard', error_level,
              '#ifndef header guard has wrong style, please use: %s' % cppvar)

    # Check for "//" comments on endif line.
    ParseNolintSuppressions(filename, raw_lines[endif_linenum], endif_linenum,
Z
zhangjinchao01 已提交
1757
                            error)
Y
Yu Yang 已提交
1758
    match = Match(r'#endif\s*//\s*' + cppvar + r'(_)?\b', endif)
Z
zhangjinchao01 已提交
1759
    if match:
Y
Yu Yang 已提交
1760 1761 1762 1763 1764
        if match.group(1) == '_':
            # Issue low severity warning for deprecated double trailing underscore
            error(filename, endif_linenum, 'build/header_guard', 0,
                  '#endif line should be "#endif  // %s"' % cppvar)
        return
Z
zhangjinchao01 已提交
1765

Y
Yu Yang 已提交
1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788
    # Didn't find the corresponding "//" comment.  If this file does not
    # contain any "//" comments at all, it could be that the compiler
    # only wants "/**/" comments, look for those instead.
    no_single_line_comments = True
    for i in xrange(1, len(raw_lines) - 1):
        line = raw_lines[i]
        if Match(r'^(?:(?:\'(?:\.|[^\'])*\')|(?:"(?:\.|[^"])*")|[^\'"])*//',
                 line):
            no_single_line_comments = False
            break

    if no_single_line_comments:
        match = Match(r'#endif\s*/\*\s*' + cppvar + r'(_)?\s*\*/', endif)
        if match:
            if match.group(1) == '_':
                # Low severity warning for double trailing underscore
                error(filename, endif_linenum, 'build/header_guard', 0,
                      '#endif line should be "#endif  /* %s */"' % cppvar)
            return

    # Didn't find anything
    error(filename, endif_linenum, 'build/header_guard', 5,
          '#endif line should be "#endif  // %s"' % cppvar)
Z
zhangjinchao01 已提交
1789 1790 1791


def CheckHeaderFileIncluded(filename, include_state, error):
Y
Yu Yang 已提交
1792 1793 1794 1795 1796 1797 1798 1799 1800
    """Logs an error if a .cc file does not include its header."""

    # Do not check test files
    if filename.endswith('_test.cc') or filename.endswith('_unittest.cc'):
        return

    fileinfo = FileInfo(filename)
    headerfile = filename[0:len(filename) - 2] + 'h'
    if not os.path.exists(headerfile):
Z
zhangjinchao01 已提交
1801
        return
Y
Yu Yang 已提交
1802 1803 1804 1805 1806 1807 1808 1809
    headername = FileInfo(headerfile).RepositoryName()
    first_include = 0
    for section_list in include_state.include_list:
        for f in section_list:
            if headername in f[0] or f[0] in headername:
                return
            if not first_include:
                first_include = f[1]
Z
zhangjinchao01 已提交
1810

Y
Yu Yang 已提交
1811 1812 1813
    error(filename, first_include, 'build/include', 5,
          '%s should include its header file %s' % (fileinfo.RepositoryName(),
                                                    headername))
Z
zhangjinchao01 已提交
1814 1815 1816


def CheckForBadCharacters(filename, lines, error):
Y
Yu Yang 已提交
1817
    """Logs an error for each line containing bad characters.
Z
zhangjinchao01 已提交
1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832

  Two kinds of bad characters:

  1. Unicode replacement characters: These indicate that either the file
  contained invalid UTF-8 (likely) or Unicode replacement characters (which
  it shouldn't).  Note that it's possible for this to throw off line
  numbering if the invalid UTF-8 occurred adjacent to a newline.

  2. NUL bytes.  These are problematic for some tools.

  Args:
    filename: The name of the current file.
    lines: An array of strings, each representing a line of the file.
    error: The function to call with any errors found.
  """
Y
Yu Yang 已提交
1833 1834 1835 1836 1837 1838 1839 1840 1841
    for linenum, line in enumerate(lines):
        if u'\ufffd' in line:
            error(
                filename, linenum, 'readability/utf8', 5,
                'Line contains invalid UTF-8 (or Unicode replacement character).'
            )
        if '\0' in line:
            error(filename, linenum, 'readability/nul', 5,
                  'Line contains NUL byte.')
Z
zhangjinchao01 已提交
1842 1843 1844


def CheckForNewlineAtEOF(filename, lines, error):
Y
Yu Yang 已提交
1845
    """Logs an error if there is no newline char at the end of the file.
Z
zhangjinchao01 已提交
1846 1847 1848 1849 1850 1851 1852

  Args:
    filename: The name of the current file.
    lines: An array of strings, each representing a line of the file.
    error: The function to call with any errors found.
  """

Y
Yu Yang 已提交
1853 1854 1855 1856 1857 1858 1859 1860
    # The array lines() was created by adding two newlines to the
    # original file (go figure), then splitting on \n.
    # To verify that the file ends in \n, we just have to make sure the
    # last-but-two element of lines() exists and is empty.
    if len(lines) < 3 or lines[-2]:
        error(filename,
              len(lines) - 2, 'whitespace/ending_newline', 5,
              'Could not find a newline character at the end of the file.')
Z
zhangjinchao01 已提交
1861 1862 1863


def CheckForMultilineCommentsAndStrings(filename, clean_lines, linenum, error):
Y
Yu Yang 已提交
1864
    """Logs an error if we see /* ... */ or "..." that extend past one line.
Z
zhangjinchao01 已提交
1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879

  /* ... */ comments are legit inside macros, for one line.
  Otherwise, we prefer // comments, so it's ok to warn about the
  other.  Likewise, it's ok for strings to extend across multiple
  lines, as long as a line continuation character (backslash)
  terminates each line. Although not currently prohibited by the C++
  style guide, it's ugly and unnecessary. We don't do well with either
  in this lint program, so we warn about both.

  Args:
    filename: The name of the current file.
    clean_lines: A CleansedLines instance containing the file.
    linenum: The number of the line to check.
    error: The function to call with any errors found.
  """
Y
Yu Yang 已提交
1880
    line = clean_lines.elided[linenum]
Z
zhangjinchao01 已提交
1881

Y
Yu Yang 已提交
1882 1883 1884
    # Remove all \\ (escaped backslashes) from the line. They are OK, and the
    # second (escaped) slash may trigger later \" detection erroneously.
    line = line.replace('\\\\', '')
Z
zhangjinchao01 已提交
1885

Y
Yu Yang 已提交
1886 1887 1888 1889 1890 1891 1892
    if line.count('/*') > line.count('*/'):
        error(filename, linenum, 'readability/multiline_comment', 5,
              'Complex multi-line /*...*/-style comment found. '
              'Lint may give bogus warnings.  '
              'Consider replacing these with //-style comments, '
              'with #if 0...#endif, '
              'or with more clearly structured multi-line comments.')
Z
zhangjinchao01 已提交
1893

Y
Yu Yang 已提交
1894 1895 1896 1897 1898
    if (line.count('"') - line.count('\\"')) % 2:
        error(filename, linenum, 'readability/multiline_string', 5,
              'Multi-line string ("...") found.  This lint script doesn\'t '
              'do well with such strings, and may give bogus warnings.  '
              'Use C++11 raw strings or concatenation instead.')
Z
zhangjinchao01 已提交
1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924


# (non-threadsafe name, thread-safe alternative, validation pattern)
#
# The validation pattern is used to eliminate false positives such as:
#  _rand();               // false positive due to substring match.
#  ->rand();              // some member function rand().
#  ACMRandom rand(seed);  // some variable named rand.
#  ISAACRandom rand();    // another variable named rand.
#
# Basically we require the return value of these functions to be used
# in some expression context on the same line by matching on some
# operator before the function name.  This eliminates constructors and
# member function calls.
_UNSAFE_FUNC_PREFIX = r'(?:[-+*/=%^&|(<]\s*|>\s+)'
_THREADING_LIST = (
    ('asctime(', 'asctime_r(', _UNSAFE_FUNC_PREFIX + r'asctime\([^)]+\)'),
    ('ctime(', 'ctime_r(', _UNSAFE_FUNC_PREFIX + r'ctime\([^)]+\)'),
    ('getgrgid(', 'getgrgid_r(', _UNSAFE_FUNC_PREFIX + r'getgrgid\([^)]+\)'),
    ('getgrnam(', 'getgrnam_r(', _UNSAFE_FUNC_PREFIX + r'getgrnam\([^)]+\)'),
    ('getlogin(', 'getlogin_r(', _UNSAFE_FUNC_PREFIX + r'getlogin\(\)'),
    ('getpwnam(', 'getpwnam_r(', _UNSAFE_FUNC_PREFIX + r'getpwnam\([^)]+\)'),
    ('getpwuid(', 'getpwuid_r(', _UNSAFE_FUNC_PREFIX + r'getpwuid\([^)]+\)'),
    ('gmtime(', 'gmtime_r(', _UNSAFE_FUNC_PREFIX + r'gmtime\([^)]+\)'),
    ('localtime(', 'localtime_r(', _UNSAFE_FUNC_PREFIX + r'localtime\([^)]+\)'),
    ('rand(', 'rand_r(', _UNSAFE_FUNC_PREFIX + r'rand\(\)'),
Y
Yu Yang 已提交
1925 1926
    ('strtok(', 'strtok_r(', _UNSAFE_FUNC_PREFIX + r'strtok\([^)]+\)'),
    ('ttyname(', 'ttyname_r(', _UNSAFE_FUNC_PREFIX + r'ttyname\([^)]+\)'), )
Z
zhangjinchao01 已提交
1927 1928 1929


def CheckPosixThreading(filename, clean_lines, linenum, error):
Y
Yu Yang 已提交
1930
    """Checks for calls to thread-unsafe functions.
Z
zhangjinchao01 已提交
1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943

  Much code has been originally written without consideration of
  multi-threading. Also, engineers are relying on their old experience;
  they have learned posix before threading extensions were added. These
  tests guide the engineers to use thread-safe functions (when using
  posix directly).

  Args:
    filename: The name of the current file.
    clean_lines: A CleansedLines instance containing the file.
    linenum: The number of the line to check.
    error: The function to call with any errors found.
  """
Y
Yu Yang 已提交
1944 1945 1946 1947 1948 1949 1950 1951
    line = clean_lines.elided[linenum]
    for single_thread_func, multithread_safe_func, pattern in _THREADING_LIST:
        # Additional pattern matching check to confirm that this is the
        # function we are looking for
        if Search(pattern, line):
            error(filename, linenum, 'runtime/threadsafe_fn', 2,
                  'Consider using ' + multithread_safe_func + '...) instead of '
                  + single_thread_func + '...) for improved thread safety.')
Z
zhangjinchao01 已提交
1952 1953 1954


def CheckVlogArguments(filename, clean_lines, linenum, error):
Y
Yu Yang 已提交
1955
    """Checks that VLOG() is only used for defining a logging level.
Z
zhangjinchao01 已提交
1956 1957 1958 1959 1960 1961 1962 1963 1964 1965

  For example, VLOG(2) is correct. VLOG(INFO), VLOG(WARNING), VLOG(ERROR), and
  VLOG(FATAL) are not.

  Args:
    filename: The name of the current file.
    clean_lines: A CleansedLines instance containing the file.
    linenum: The number of the line to check.
    error: The function to call with any errors found.
  """
Y
Yu Yang 已提交
1966 1967 1968 1969 1970 1971
    line = clean_lines.elided[linenum]
    if Search(r'\bVLOG\((INFO|ERROR|WARNING|DFATAL|FATAL)\)', line):
        error(filename, linenum, 'runtime/vlog', 5,
              'VLOG() should be used with numeric verbosity level.  '
              'Use LOG() if you want symbolic severity levels.')

Z
zhangjinchao01 已提交
1972 1973 1974

# Matches invalid increment: *count++, which moves pointer instead of
# incrementing a value.
Y
Yu Yang 已提交
1975
_RE_PATTERN_INVALID_INCREMENT = re.compile(r'^\s*\*\w+(\+\+|--);')
Z
zhangjinchao01 已提交
1976 1977 1978


def CheckInvalidIncrement(filename, clean_lines, linenum, error):
Y
Yu Yang 已提交
1979
    """Checks for invalid increment *count++.
Z
zhangjinchao01 已提交
1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993

  For example following function:
  void increment_counter(int* count) {
    *count++;
  }
  is invalid, because it effectively does count++, moving pointer, and should
  be replaced with ++*count, (*count)++ or *count += 1.

  Args:
    filename: The name of the current file.
    clean_lines: A CleansedLines instance containing the file.
    linenum: The number of the line to check.
    error: The function to call with any errors found.
  """
Y
Yu Yang 已提交
1994 1995 1996 1997 1998
    line = clean_lines.elided[linenum]
    if _RE_PATTERN_INVALID_INCREMENT.match(line):
        error(
            filename, linenum, 'runtime/invalid_increment', 5,
            'Changing pointer instead of value (or unused value of operator*).')
Z
zhangjinchao01 已提交
1999 2000 2001


def IsMacroDefinition(clean_lines, linenum):
Y
Yu Yang 已提交
2002 2003
    if Search(r'^#define', clean_lines[linenum]):
        return True
Z
zhangjinchao01 已提交
2004

Y
Yu Yang 已提交
2005 2006
    if linenum > 0 and Search(r'\\$', clean_lines[linenum - 1]):
        return True
Z
zhangjinchao01 已提交
2007

Y
Yu Yang 已提交
2008
    return False
Z
zhangjinchao01 已提交
2009 2010 2011


def IsForwardClassDeclaration(clean_lines, linenum):
Y
Yu Yang 已提交
2012
    return Match(r'^\s*(\btemplate\b)*.*class\s+\w+;\s*$', clean_lines[linenum])
Z
zhangjinchao01 已提交
2013 2014 2015


class _BlockInfo(object):
Y
Yu Yang 已提交
2016
    """Stores information about a generic block of code."""
Z
zhangjinchao01 已提交
2017

Y
Yu Yang 已提交
2018 2019 2020 2021 2022
    def __init__(self, seen_open_brace):
        self.seen_open_brace = seen_open_brace
        self.open_parentheses = 0
        self.inline_asm = _NO_ASM
        self.check_namespace_indentation = False
Z
zhangjinchao01 已提交
2023

Y
Yu Yang 已提交
2024 2025
    def CheckBegin(self, filename, clean_lines, linenum, error):
        """Run checks that applies to text up to the opening brace.
Z
zhangjinchao01 已提交
2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036

    This is mostly for checking the text after the class identifier
    and the "{", usually where the base class is specified.  For other
    blocks, there isn't much to check, so we always pass.

    Args:
      filename: The name of the current file.
      clean_lines: A CleansedLines instance containing the file.
      linenum: The number of the line to check.
      error: The function to call with any errors found.
    """
Y
Yu Yang 已提交
2037
        pass
Z
zhangjinchao01 已提交
2038

Y
Yu Yang 已提交
2039 2040
    def CheckEnd(self, filename, clean_lines, linenum, error):
        """Run checks that applies to text after the closing brace.
Z
zhangjinchao01 已提交
2041 2042 2043 2044 2045 2046 2047 2048 2049

    This is mostly used for checking end of namespace comments.

    Args:
      filename: The name of the current file.
      clean_lines: A CleansedLines instance containing the file.
      linenum: The number of the line to check.
      error: The function to call with any errors found.
    """
Y
Yu Yang 已提交
2050
        pass
Z
zhangjinchao01 已提交
2051

Y
Yu Yang 已提交
2052 2053
    def IsBlockInfo(self):
        """Returns true if this block is a _BlockInfo.
Z
zhangjinchao01 已提交
2054 2055 2056 2057 2058 2059 2060

    This is convenient for verifying that an object is an instance of
    a _BlockInfo, but not an instance of any of the derived classes.

    Returns:
      True for this class, False for derived classes.
    """
Y
Yu Yang 已提交
2061
        return self.__class__ == _BlockInfo
Z
zhangjinchao01 已提交
2062 2063 2064


class _ExternCInfo(_BlockInfo):
Y
Yu Yang 已提交
2065
    """Stores information about an 'extern "C"' block."""
Z
zhangjinchao01 已提交
2066

Y
Yu Yang 已提交
2067 2068
    def __init__(self):
        _BlockInfo.__init__(self, True)
Z
zhangjinchao01 已提交
2069 2070 2071


class _ClassInfo(_BlockInfo):
Y
Yu Yang 已提交
2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085
    """Stores information about a class."""

    def __init__(self, name, class_or_struct, clean_lines, linenum):
        _BlockInfo.__init__(self, False)
        self.name = name
        self.starting_linenum = linenum
        self.is_derived = False
        self.check_namespace_indentation = True
        if class_or_struct == 'struct':
            self.access = 'public'
            self.is_struct = True
        else:
            self.access = 'private'
            self.is_struct = False
Z
zhangjinchao01 已提交
2086

Y
Yu Yang 已提交
2087 2088 2089
        # Remember initial indentation level for this class.  Using raw_lines here
        # instead of elided to account for leading comments.
        self.class_indent = GetIndentLevel(clean_lines.raw_lines[linenum])
Z
zhangjinchao01 已提交
2090

Y
Yu Yang 已提交
2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 2136 2137 2138 2139
        # Try to find the end of the class.  This will be confused by things like:
        #   class A {
        #   } *x = { ...
        #
        # But it's still good enough for CheckSectionSpacing.
        self.last_line = 0
        depth = 0
        for i in range(linenum, clean_lines.NumLines()):
            line = clean_lines.elided[i]
            depth += line.count('{') - line.count('}')
            if not depth:
                self.last_line = i
                break

    def CheckBegin(self, filename, clean_lines, linenum, error):
        # Look for a bare ':'
        if Search('(^|[^:]):($|[^:])', clean_lines.elided[linenum]):
            self.is_derived = True

    def CheckEnd(self, filename, clean_lines, linenum, error):
        # If there is a DISALLOW macro, it should appear near the end of
        # the class.
        seen_last_thing_in_class = False
        for i in xrange(linenum - 1, self.starting_linenum, -1):
            match = Search(
                r'\b(DISALLOW_COPY_AND_ASSIGN|DISALLOW_IMPLICIT_CONSTRUCTORS)\('
                + self.name + r'\)', clean_lines.elided[i])
            if match:
                if seen_last_thing_in_class:
                    error(filename, i, 'readability/constructors', 3,
                          match.group(1) +
                          ' should be the last thing in the class')
                break

            if not Match(r'^\s*$', clean_lines.elided[i]):
                seen_last_thing_in_class = True

        # Check that closing brace is aligned with beginning of the class.
        # Only do this if the closing brace is indented by only whitespaces.
        # This means we will not check single-line class definitions.
        indent = Match(r'^( *)\}', clean_lines.elided[linenum])
        if indent and len(indent.group(1)) != self.class_indent:
            if self.is_struct:
                parent = 'struct ' + self.name
            else:
                parent = 'class ' + self.name
            error(filename, linenum, 'whitespace/indent', 3,
                  'Closing brace should be aligned with beginning of %s' %
                  parent)
Z
zhangjinchao01 已提交
2140 2141 2142


class _NamespaceInfo(_BlockInfo):
Y
Yu Yang 已提交
2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188
    """Stores information about a namespace."""

    def __init__(self, name, linenum):
        _BlockInfo.__init__(self, False)
        self.name = name or ''
        self.starting_linenum = linenum
        self.check_namespace_indentation = True

    def CheckEnd(self, filename, clean_lines, linenum, error):
        """Check end of namespace comments."""
        line = clean_lines.raw_lines[linenum]

        # Check how many lines is enclosed in this namespace.  Don't issue
        # warning for missing namespace comments if there aren't enough
        # lines.  However, do apply checks if there is already an end of
        # namespace comment and it's incorrect.
        #
        # TODO(unknown): We always want to check end of namespace comments
        # if a namespace is large, but sometimes we also want to apply the
        # check if a short namespace contained nontrivial things (something
        # other than forward declarations).  There is currently no logic on
        # deciding what these nontrivial things are, so this check is
        # triggered by namespace size only, which works most of the time.
        if (linenum - self.starting_linenum < 10 and
                not Match(r'};*\s*(//|/\*).*\bnamespace\b', line)):
            return

        # Look for matching comment at end of namespace.
        #
        # Note that we accept C style "/* */" comments for terminating
        # namespaces, so that code that terminate namespaces inside
        # preprocessor macros can be cpplint clean.
        #
        # We also accept stuff like "// end of namespace <name>." with the
        # period at the end.
        #
        # Besides these, we don't accept anything else, otherwise we might
        # get false negatives when existing comment is a substring of the
        # expected namespace.
        if self.name:
            # Named namespace
            if not Match((r'};*\s*(//|/\*).*\bnamespace\s+' +
                          re.escape(self.name) + r'[\*/\.\\\s]*$'), line):
                error(filename, linenum, 'readability/namespace', 5,
                      'Namespace should be terminated with "// namespace %s"' %
                      self.name)
Z
zhangjinchao01 已提交
2189
        else:
Y
Yu Yang 已提交
2190 2191 2192 2193 2194 2195 2196 2197 2198 2199 2200 2201 2202 2203 2204
            # Anonymous namespace
            if not Match(r'};*\s*(//|/\*).*\bnamespace[\*/\.\\\s]*$', line):
                # If "// namespace anonymous" or "// anonymous namespace (more text)",
                # mention "// anonymous namespace" as an acceptable form
                if Match(r'}.*\b(namespace anonymous|anonymous namespace)\b',
                         line):
                    error(
                        filename, linenum, 'readability/namespace', 5,
                        'Anonymous namespace should be terminated with "// namespace"'
                        ' or "// anonymous namespace"')
                else:
                    error(
                        filename, linenum, 'readability/namespace', 5,
                        'Anonymous namespace should be terminated with "// namespace"'
                    )
Z
zhangjinchao01 已提交
2205 2206 2207


class _PreprocessorInfo(object):
Y
Yu Yang 已提交
2208
    """Stores checkpoints of nesting stacks when #if/#else is seen."""
Z
zhangjinchao01 已提交
2209

Y
Yu Yang 已提交
2210 2211 2212
    def __init__(self, stack_before_if):
        # The entire nesting stack before #if
        self.stack_before_if = stack_before_if
Z
zhangjinchao01 已提交
2213

Y
Yu Yang 已提交
2214 2215
        # The entire nesting stack up to #else
        self.stack_before_else = []
Z
zhangjinchao01 已提交
2216

Y
Yu Yang 已提交
2217 2218
        # Whether we have already seen #else or #elif
        self.seen_else = False
Z
zhangjinchao01 已提交
2219 2220 2221


class NestingState(object):
Y
Yu Yang 已提交
2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242
    """Holds states related to parsing braces."""

    def __init__(self):
        # Stack for tracking all braces.  An object is pushed whenever we
        # see a "{", and popped when we see a "}".  Only 3 types of
        # objects are possible:
        # - _ClassInfo: a class or struct.
        # - _NamespaceInfo: a namespace.
        # - _BlockInfo: some other type of block.
        self.stack = []

        # Top of the previous stack before each Update().
        #
        # Because the nesting_stack is updated at the end of each line, we
        # had to do some convoluted checks to find out what is the current
        # scope at the beginning of the line.  This check is simplified by
        # saving the previous top of nesting stack.
        #
        # We could save the full stack, but we only need the top.  Copying
        # the full nesting stack would slow down cpplint by ~10%.
        self.previous_stack_top = []
Z
zhangjinchao01 已提交
2243

Y
Yu Yang 已提交
2244 2245
        # Stack of _PreprocessorInfo objects.
        self.pp_stack = []
Z
zhangjinchao01 已提交
2246

Y
Yu Yang 已提交
2247 2248
    def SeenOpenBrace(self):
        """Check if we have seen the opening brace for the innermost block.
Z
zhangjinchao01 已提交
2249 2250 2251 2252 2253

    Returns:
      True if we have seen the opening brace, False if the innermost
      block is still expecting an opening brace.
    """
Y
Yu Yang 已提交
2254
        return (not self.stack) or self.stack[-1].seen_open_brace
Z
zhangjinchao01 已提交
2255

Y
Yu Yang 已提交
2256 2257
    def InNamespaceBody(self):
        """Check if we are currently one level inside a namespace body.
Z
zhangjinchao01 已提交
2258 2259 2260 2261

    Returns:
      True if top of the stack is a namespace block, False otherwise.
    """
Y
Yu Yang 已提交
2262
        return self.stack and isinstance(self.stack[-1], _NamespaceInfo)
Z
zhangjinchao01 已提交
2263

Y
Yu Yang 已提交
2264 2265
    def InExternC(self):
        """Check if we are currently one level inside an 'extern "C"' block.
Z
zhangjinchao01 已提交
2266 2267 2268 2269

    Returns:
      True if top of the stack is an extern block, False otherwise.
    """
Y
Yu Yang 已提交
2270
        return self.stack and isinstance(self.stack[-1], _ExternCInfo)
Z
zhangjinchao01 已提交
2271

Y
Yu Yang 已提交
2272 2273
    def InClassDeclaration(self):
        """Check if we are currently one level inside a class or struct declaration.
Z
zhangjinchao01 已提交
2274 2275 2276 2277

    Returns:
      True if top of the stack is a class/struct, False otherwise.
    """
Y
Yu Yang 已提交
2278
        return self.stack and isinstance(self.stack[-1], _ClassInfo)
Z
zhangjinchao01 已提交
2279

Y
Yu Yang 已提交
2280 2281
    def InAsmBlock(self):
        """Check if we are currently one level inside an inline ASM block.
Z
zhangjinchao01 已提交
2282 2283 2284 2285

    Returns:
      True if the top of the stack is a block containing inline ASM.
    """
Y
Yu Yang 已提交
2286
        return self.stack and self.stack[-1].inline_asm != _NO_ASM
Z
zhangjinchao01 已提交
2287

Y
Yu Yang 已提交
2288 2289
    def InTemplateArgumentList(self, clean_lines, linenum, pos):
        """Check if current position is inside template argument list.
Z
zhangjinchao01 已提交
2290 2291 2292 2293 2294 2295 2296 2297

    Args:
      clean_lines: A CleansedLines instance containing the file.
      linenum: The number of the line to check.
      pos: position just after the suspected template argument.
    Returns:
      True if (linenum, pos) is inside template arguments.
    """
Y
Yu Yang 已提交
2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336 2337 2338
        while linenum < clean_lines.NumLines():
            # Find the earliest character that might indicate a template argument
            line = clean_lines.elided[linenum]
            match = Match(r'^[^{};=\[\]\.<>]*(.)', line[pos:])
            if not match:
                linenum += 1
                pos = 0
                continue
            token = match.group(1)
            pos += len(match.group(0))

            # These things do not look like template argument list:
            #   class Suspect {
            #   class Suspect x; }
            if token in ('{', '}', ';'): return False

            # These things look like template argument list:
            #   template <class Suspect>
            #   template <class Suspect = default_value>
            #   template <class Suspect[]>
            #   template <class Suspect...>
            if token in ('>', '=', '[', ']', '.'): return True

            # Check if token is an unmatched '<'.
            # If not, move on to the next character.
            if token != '<':
                pos += 1
                if pos >= len(line):
                    linenum += 1
                    pos = 0
                continue

            # We can't be sure if we just find a single '<', and need to
            # find the matching '>'.
            (_, end_line, end_pos) = CloseExpression(clean_lines, linenum,
                                                     pos - 1)
            if end_pos < 0:
                # Not sure if template argument list or syntax error in file
                return False
            linenum = end_line
            pos = end_pos
Z
zhangjinchao01 已提交
2339 2340
        return False

Y
Yu Yang 已提交
2341 2342
    def UpdatePreprocessor(self, line):
        """Update preprocessor stack.
Z
zhangjinchao01 已提交
2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361

    We need to handle preprocessors due to classes like this:
      #ifdef SWIG
      struct ResultDetailsPageElementExtensionPoint {
      #else
      struct ResultDetailsPageElementExtensionPoint : public Extension {
      #endif

    We make the following assumptions (good enough for most files):
    - Preprocessor condition evaluates to true from #if up to first
      #else/#elif/#endif.

    - Preprocessor condition evaluates to false from #else/#elif up
      to #endif.  We still perform lint checks on these lines, but
      these do not affect nesting stack.

    Args:
      line: current line to check.
    """
Y
Yu Yang 已提交
2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400
        if Match(r'^\s*#\s*(if|ifdef|ifndef)\b', line):
            # Beginning of #if block, save the nesting stack here.  The saved
            # stack will allow us to restore the parsing state in the #else case.
            self.pp_stack.append(_PreprocessorInfo(copy.deepcopy(self.stack)))
        elif Match(r'^\s*#\s*(else|elif)\b', line):
            # Beginning of #else block
            if self.pp_stack:
                if not self.pp_stack[-1].seen_else:
                    # This is the first #else or #elif block.  Remember the
                    # whole nesting stack up to this point.  This is what we
                    # keep after the #endif.
                    self.pp_stack[-1].seen_else = True
                    self.pp_stack[-1].stack_before_else = copy.deepcopy(
                        self.stack)

                # Restore the stack to how it was before the #if
                self.stack = copy.deepcopy(self.pp_stack[-1].stack_before_if)
            else:
                # TODO(unknown): unexpected #else, issue warning?
                pass
        elif Match(r'^\s*#\s*endif\b', line):
            # End of #if or #else blocks.
            if self.pp_stack:
                # If we saw an #else, we will need to restore the nesting
                # stack to its former state before the #else, otherwise we
                # will just continue from where we left off.
                if self.pp_stack[-1].seen_else:
                    # Here we can just use a shallow copy since we are the last
                    # reference to it.
                    self.stack = self.pp_stack[-1].stack_before_else
                # Drop the corresponding #if
                self.pp_stack.pop()
            else:
                # TODO(unknown): unexpected #endif, issue warning?
                pass

    # TODO(unknown): Update() is too long, but we will refactor later.
    def Update(self, filename, clean_lines, linenum, error):
        """Update nesting state with current line.
Z
zhangjinchao01 已提交
2401 2402 2403 2404 2405 2406 2407

    Args:
      filename: The name of the current file.
      clean_lines: A CleansedLines instance containing the file.
      linenum: The number of the line to check.
      error: The function to call with any errors found.
    """
Y
Yu Yang 已提交
2408
        line = clean_lines.elided[linenum]
Z
zhangjinchao01 已提交
2409

Y
Yu Yang 已提交
2410
        # Remember top of the previous nesting stack.
Z
zhangjinchao01 已提交
2411
        #
Y
Yu Yang 已提交
2412 2413 2414
        # The stack is always pushed/popped and not modified in place, so
        # we can just do a shallow copy instead of copy.deepcopy.  Using
        # deepcopy would slow down cpplint by ~28%.
Z
zhangjinchao01 已提交
2415
        if self.stack:
Y
Yu Yang 已提交
2416 2417 2418 2419 2420 2421
            self.previous_stack_top = self.stack[-1]
        else:
            self.previous_stack_top = None

        # Update pp_stack
        self.UpdatePreprocessor(line)
Z
zhangjinchao01 已提交
2422

Y
Yu Yang 已提交
2423 2424 2425 2426 2427 2428 2429 2430 2431 2432 2433 2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450 2451 2452 2453 2454 2455 2456 2457 2458 2459 2460 2461 2462 2463 2464 2465 2466 2467 2468 2469 2470 2471 2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 2483 2484 2485 2486 2487 2488 2489 2490 2491 2492 2493 2494 2495 2496 2497 2498 2499 2500 2501 2502 2503 2504 2505 2506 2507 2508 2509 2510 2511 2512 2513 2514 2515 2516 2517 2518 2519 2520 2521 2522 2523 2524 2525 2526 2527 2528 2529 2530 2531 2532 2533 2534 2535 2536 2537 2538 2539 2540 2541 2542 2543 2544 2545 2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565 2566 2567
        # Count parentheses.  This is to avoid adding struct arguments to
        # the nesting stack.
        if self.stack:
            inner_block = self.stack[-1]
            depth_change = line.count('(') - line.count(')')
            inner_block.open_parentheses += depth_change

            # Also check if we are starting or ending an inline assembly block.
            if inner_block.inline_asm in (_NO_ASM, _END_ASM):
                if (depth_change != 0 and inner_block.open_parentheses == 1 and
                        _MATCH_ASM.match(line)):
                    # Enter assembly block
                    inner_block.inline_asm = _INSIDE_ASM
                else:
                    # Not entering assembly block.  If previous line was _END_ASM,
                    # we will now shift to _NO_ASM state.
                    inner_block.inline_asm = _NO_ASM
            elif (inner_block.inline_asm == _INSIDE_ASM and
                  inner_block.open_parentheses == 0):
                # Exit assembly block
                inner_block.inline_asm = _END_ASM

        # Consume namespace declaration at the beginning of the line.  Do
        # this in a loop so that we catch same line declarations like this:
        #   namespace proto2 { namespace bridge { class MessageSet; } }
        while True:
            # Match start of namespace.  The "\b\s*" below catches namespace
            # declarations even if it weren't followed by a whitespace, this
            # is so that we don't confuse our namespace checker.  The
            # missing spaces will be flagged by CheckSpacing.
            namespace_decl_match = Match(r'^\s*namespace\b\s*([:\w]+)?(.*)$',
                                         line)
            if not namespace_decl_match:
                break

            new_namespace = _NamespaceInfo(
                namespace_decl_match.group(1), linenum)
            self.stack.append(new_namespace)

            line = namespace_decl_match.group(2)
            if line.find('{') != -1:
                new_namespace.seen_open_brace = True
                line = line[line.find('{') + 1:]

        # Look for a class declaration in whatever is left of the line
        # after parsing namespaces.  The regexp accounts for decorated classes
        # such as in:
        #   class LOCKABLE API Object {
        #   };
        class_decl_match = Match(
            r'^(\s*(?:template\s*<[\w\s<>,:]*>\s*)?'
            r'(class|struct)\s+(?:[A-Z_]+\s+)*(\w+(?:::\w+)*))'
            r'(.*)$', line)
        if (class_decl_match and
            (not self.stack or self.stack[-1].open_parentheses == 0)):
            # We do not want to accept classes that are actually template arguments:
            #   template <class Ignore1,
            #             class Ignore2 = Default<Args>,
            #             template <Args> class Ignore3>
            #   void Function() {};
            #
            # To avoid template argument cases, we scan forward and look for
            # an unmatched '>'.  If we see one, assume we are inside a
            # template argument list.
            end_declaration = len(class_decl_match.group(1))
            if not self.InTemplateArgumentList(clean_lines, linenum,
                                               end_declaration):
                self.stack.append(
                    _ClassInfo(
                        class_decl_match.group(3),
                        class_decl_match.group(2), clean_lines, linenum))
                line = class_decl_match.group(4)

        # If we have not yet seen the opening brace for the innermost block,
        # run checks here.
        if not self.SeenOpenBrace():
            self.stack[-1].CheckBegin(filename, clean_lines, linenum, error)

        # Update access control if we are inside a class/struct
        if self.stack and isinstance(self.stack[-1], _ClassInfo):
            classinfo = self.stack[-1]
            access_match = Match(
                r'^(.*)\b(public|private|protected|signals)(\s+(?:slots\s*)?)?'
                r':(?:[^:]|$)', line)
            if access_match:
                classinfo.access = access_match.group(2)

                # Check that access keywords are indented +1 space.  Skip this
                # check if the keywords are not preceded by whitespaces.
                indent = access_match.group(1)
                if (len(indent) != classinfo.class_indent + 1 and
                        Match(r'^\s*$', indent)):
                    if classinfo.is_struct:
                        parent = 'struct ' + classinfo.name
                    else:
                        parent = 'class ' + classinfo.name
                    slots = ''
                    if access_match.group(3):
                        slots = access_match.group(3)
                    error(filename, linenum, 'whitespace/indent', 3,
                          '%s%s: should be indented +1 space inside %s' % (
                              access_match.group(2), slots, parent))

        # Consume braces or semicolons from what's left of the line
        while True:
            # Match first brace, semicolon, or closed parenthesis.
            matched = Match(r'^[^{;)}]*([{;)}])(.*)$', line)
            if not matched:
                break

            token = matched.group(1)
            if token == '{':
                # If namespace or class hasn't seen a opening brace yet, mark
                # namespace/class head as complete.  Push a new block onto the
                # stack otherwise.
                if not self.SeenOpenBrace():
                    self.stack[-1].seen_open_brace = True
                elif Match(r'^extern\s*"[^"]*"\s*\{', line):
                    self.stack.append(_ExternCInfo())
                else:
                    self.stack.append(_BlockInfo(True))
                    if _MATCH_ASM.match(line):
                        self.stack[-1].inline_asm = _BLOCK_ASM

            elif token == ';' or token == ')':
                # If we haven't seen an opening brace yet, but we already saw
                # a semicolon, this is probably a forward declaration.  Pop
                # the stack for these.
                #
                # Similarly, if we haven't seen an opening brace yet, but we
                # already saw a closing parenthesis, then these are probably
                # function arguments with extra "class" or "struct" keywords.
                # Also pop these stack for these.
                if not self.SeenOpenBrace():
                    self.stack.pop()
            else:  # token == '}'
                # Perform end of block checks and pop the stack.
                if self.stack:
                    self.stack[-1].CheckEnd(filename, clean_lines, linenum,
                                            error)
                    self.stack.pop()
            line = matched.group(2)

    def InnermostClass(self):
        """Get class info on the top of the stack.
Z
zhangjinchao01 已提交
2568 2569 2570 2571

    Returns:
      A _ClassInfo object if we are inside a class, or None otherwise.
    """
Y
Yu Yang 已提交
2572 2573 2574 2575 2576
        for i in range(len(self.stack), 0, -1):
            classinfo = self.stack[i - 1]
            if isinstance(classinfo, _ClassInfo):
                return classinfo
        return None
Z
zhangjinchao01 已提交
2577

Y
Yu Yang 已提交
2578 2579
    def CheckCompletedBlocks(self, filename, error):
        """Checks that all classes and namespaces have been completely parsed.
Z
zhangjinchao01 已提交
2580 2581 2582 2583 2584 2585

    Call this when all lines in a file have been processed.
    Args:
      filename: The name of the current file.
      error: The function to call with any errors found.
    """
Y
Yu Yang 已提交
2586 2587 2588 2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602
        # Note: This test can result in false positives if #ifdef constructs
        # get in the way of brace matching. See the testBuildClass test in
        # cpplint_unittest.py for an example of this.
        for obj in self.stack:
            if isinstance(obj, _ClassInfo):
                error(filename, obj.starting_linenum, 'build/class', 5,
                      'Failed to find complete declaration of class %s' %
                      obj.name)
            elif isinstance(obj, _NamespaceInfo):
                error(filename, obj.starting_linenum, 'build/namespaces', 5,
                      'Failed to find complete declaration of namespace %s' %
                      obj.name)


def CheckForNonStandardConstructs(filename, clean_lines, linenum, nesting_state,
                                  error):
    r"""Logs an error if we see certain non-ANSI constructs ignored by gcc-2.
Z
zhangjinchao01 已提交
2603 2604 2605 2606 2607 2608 2609 2610 2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 2627 2628

  Complain about several constructs which gcc-2 accepts, but which are
  not standard C++.  Warning about these in lint is one way to ease the
  transition to new compilers.
  - put storage class first (e.g. "static const" instead of "const static").
  - "%lld" instead of %qd" in printf-type functions.
  - "%1$d" is non-standard in printf-type functions.
  - "\%" is an undefined character escape sequence.
  - text after #endif is not allowed.
  - invalid inner-style forward declaration.
  - >? and <? operators, and their >?= and <?= cousins.

  Additionally, check for constructor/destructor style violations and reference
  members, as it is very convenient to do so while checking for
  gcc-2 compliance.

  Args:
    filename: The name of the current file.
    clean_lines: A CleansedLines instance containing the file.
    linenum: The number of the line to check.
    nesting_state: A NestingState instance which maintains information about
                   the current stack of nested blocks being parsed.
    error: A callable to which errors are reported, which takes 4 arguments:
           filename, line number, error level, and message
  """

Y
Yu Yang 已提交
2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 2641 2642 2643 2644 2645 2646 2647 2648 2649 2650 2651 2652 2653 2654 2655 2656 2657 2658 2659 2660 2661 2662 2663 2664 2665 2666 2667 2668 2669 2670 2671 2672 2673 2674 2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748 2749 2750 2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762
    # Remove comments from the line, but leave in strings for now.
    line = clean_lines.lines[linenum]

    if Search(r'printf\s*\(.*".*%[-+ ]?\d*q', line):
        error(filename, linenum, 'runtime/printf_format', 3,
              '%q in format strings is deprecated.  Use %ll instead.')

    if Search(r'printf\s*\(.*".*%\d+\$', line):
        error(filename, linenum, 'runtime/printf_format', 2,
              '%N$ formats are unconventional.  Try rewriting to avoid them.')

    # Remove escaped backslashes before looking for undefined escapes.
    line = line.replace('\\\\', '')

    if Search(r'("|\').*\\(%|\[|\(|{)', line):
        error(filename, linenum, 'build/printf_format', 3,
              '%, [, (, and { are undefined character escapes.  Unescape them.')

    # For the rest, work with both comments and strings removed.
    line = clean_lines.elided[linenum]

    if Search(r'\b(const|volatile|void|char|short|int|long'
              r'|float|double|signed|unsigned'
              r'|schar|u?int8|u?int16|u?int32|u?int64)'
              r'\s+(register|static|extern|typedef)\b', line):
        error(filename, linenum, 'build/storage_class', 5,
              'Storage class (static, extern, typedef, etc) should be first.')

    if Match(r'\s*#\s*endif\s*[^/\s]+', line):
        error(filename, linenum, 'build/endif_comment', 5,
              'Uncommented text after #endif is non-standard.  Use a comment.')

    if Match(r'\s*class\s+(\w+\s*::\s*)+\w+\s*;', line):
        error(
            filename, linenum, 'build/forward_decl', 5,
            'Inner-style forward declarations are invalid.  Remove this line.')

    if Search(r'(\w+|[+-]?\d+(\.\d*)?)\s*(<|>)\?=?\s*(\w+|[+-]?\d+)(\.\d*)?',
              line):
        error(
            filename, linenum, 'build/deprecated', 3,
            '>? and <? (max and min) operators are non-standard and deprecated.')

    if Search(r'^\s*const\s*string\s*&\s*\w+\s*;', line):
        # TODO(unknown): Could it be expanded safely to arbitrary references,
        # without triggering too many false positives? The first
        # attempt triggered 5 warnings for mostly benign code in the regtest, hence
        # the restriction.
        # Here's the original regexp, for the reference:
        # type_name = r'\w+((\s*::\s*\w+)|(\s*<\s*\w+?\s*>))?'
        # r'\s*const\s*' + type_name + '\s*&\s*\w+\s*;'
        error(filename, linenum, 'runtime/member_string_references', 2,
              'const string& members are dangerous. It is much better to use '
              'alternatives, such as pointers or simple constants.')

    # Everything else in this function operates on class declarations.
    # Return early if the top of the nesting stack is not a class, or if
    # the class head is not completed yet.
    classinfo = nesting_state.InnermostClass()
    if not classinfo or not classinfo.seen_open_brace:
        return

    # The class may have been declared with namespace or classname qualifiers.
    # The constructor and destructor will not have those qualifiers.
    base_classname = classinfo.name.split('::')[-1]

    # Look for single-argument constructors that aren't marked explicit.
    # Technically a valid construct, but against style. Also look for
    # non-single-argument constructors which are also technically valid, but
    # strongly suggest something is wrong.
    explicit_constructor_match = Match(
        r'\s+(?:inline\s+)?(explicit\s+)?(?:inline\s+)?%s\s*'
        r'\(((?:[^()]|\([^()]*\))*)\)' % re.escape(base_classname), line)

    if explicit_constructor_match:
        is_marked_explicit = explicit_constructor_match.group(1)

        if not explicit_constructor_match.group(2):
            constructor_args = []
        else:
            constructor_args = explicit_constructor_match.group(2).split(',')

        # collapse arguments so that commas in template parameter lists and function
        # argument parameter lists don't split arguments in two
        i = 0
        while i < len(constructor_args):
            constructor_arg = constructor_args[i]
            while (constructor_arg.count('<') > constructor_arg.count('>') or
                   constructor_arg.count('(') > constructor_arg.count(')')):
                constructor_arg += ',' + constructor_args[i + 1]
                del constructor_args[i + 1]
            constructor_args[i] = constructor_arg
            i += 1

        defaulted_args = [arg for arg in constructor_args if '=' in arg]
        noarg_constructor = (
            not constructor_args or  # empty arg list
            # 'void' arg specifier
            (len(constructor_args) == 1 and
             constructor_args[0].strip() == 'void'))
        onearg_constructor = (
            (
                len(constructor_args) == 1 and  # exactly one arg
                not noarg_constructor) or
            # all but at most one arg defaulted
            (len(constructor_args) >= 1 and not noarg_constructor and
             len(defaulted_args) >= len(constructor_args) - 1))
        initializer_list_constructor = bool(
            onearg_constructor and
            Search(r'\bstd\s*::\s*initializer_list\b', constructor_args[0]))
        copy_constructor = bool(
            onearg_constructor and
            Match(r'(const\s+)?%s(\s*<[^>]*>)?(\s+const)?\s*(?:<\w+>\s*)?&' %
                  re.escape(base_classname), constructor_args[0].strip()))

        if (not is_marked_explicit and onearg_constructor and
                not initializer_list_constructor and not copy_constructor):
            if defaulted_args:
                error(filename, linenum, 'runtime/explicit', 5,
                      'Constructors callable with one argument '
                      'should be marked explicit.')
            else:
                error(
                    filename, linenum, 'runtime/explicit', 5,
                    'Single-parameter constructors should be marked explicit.')
        elif is_marked_explicit and not onearg_constructor:
            if noarg_constructor:
                error(
                    filename, linenum, 'runtime/explicit', 5,
                    'Zero-parameter constructors should not be marked explicit.')
            else:
                error(filename, linenum, 'runtime/explicit', 0,
                      'Constructors that require multiple arguments '
                      'should not be marked explicit.')
Z
zhangjinchao01 已提交
2763 2764 2765


def CheckSpacingForFunctionCall(filename, clean_lines, linenum, error):
Y
Yu Yang 已提交
2766
    """Checks for the correctness of various spacing around function calls.
Z
zhangjinchao01 已提交
2767 2768 2769 2770 2771 2772 2773

  Args:
    filename: The name of the current file.
    clean_lines: A CleansedLines instance containing the file.
    linenum: The number of the line to check.
    error: The function to call with any errors found.
  """
Y
Yu Yang 已提交
2774 2775 2776 2777 2778 2779 2780 2781 2782 2783 2784 2785 2786 2787 2788 2789 2790 2791 2792 2793 2794 2795 2796 2797 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810 2811 2812 2813 2814 2815 2816 2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837
    line = clean_lines.elided[linenum]

    # Since function calls often occur inside if/for/while/switch
    # expressions - which have their own, more liberal conventions - we
    # first see if we should be looking inside such an expression for a
    # function call, to which we can apply more strict standards.
    fncall = line  # if there's no control flow construct, look at whole line
    for pattern in (r'\bif\s*\((.*)\)\s*{', r'\bfor\s*\((.*)\)\s*{',
                    r'\bwhile\s*\((.*)\)\s*[{;]', r'\bswitch\s*\((.*)\)\s*{'):
        match = Search(pattern, line)
        if match:
            fncall = match.group(1)  # look inside the parens for function calls
            break

    # Except in if/for/while/switch, there should never be space
    # immediately inside parens (eg "f( 3, 4 )").  We make an exception
    # for nested parens ( (a+b) + c ).  Likewise, there should never be
    # a space before a ( when it's a function argument.  I assume it's a
    # function argument when the char before the whitespace is legal in
    # a function name (alnum + _) and we're not starting a macro. Also ignore
    # pointers and references to arrays and functions coz they're too tricky:
    # we use a very simple way to recognize these:
    # " (something)(maybe-something)" or
    # " (something)(maybe-something," or
    # " (something)[something]"
    # Note that we assume the contents of [] to be short enough that
    # they'll never need to wrap.
    if (  # Ignore control structures.
            not Search(
                r'\b(if|for|while|switch|return|new|delete|catch|sizeof)\b',
                fncall) and
            # Ignore pointers/references to functions.
            not Search(r' \([^)]+\)\([^)]*(\)|,$)', fncall) and
            # Ignore pointers/references to arrays.
            not Search(r' \([^)]+\)\[[^\]]+\]', fncall)):
        if Search(r'\w\s*\(\s(?!\s*\\$)', fncall):  # a ( used for a fn call
            error(filename, linenum, 'whitespace/parens', 4,
                  'Extra space after ( in function call')
        elif Search(r'\(\s+(?!(\s*\\)|\()', fncall):
            error(filename, linenum, 'whitespace/parens', 2,
                  'Extra space after (')
        if (Search(r'\w\s+\(', fncall) and
                not Search(r'#\s*define|typedef|using\s+\w+\s*=', fncall) and
                not Search(r'\w\s+\((\w+::)*\*\w+\)\(', fncall) and
                not Search(r'\bcase\s+\(', fncall)):
            # TODO(unknown): Space after an operator function seem to be a common
            # error, silence those for now by restricting them to highest verbosity.
            if Search(r'\boperator_*\b', line):
                error(filename, linenum, 'whitespace/parens', 0,
                      'Extra space before ( in function call')
            else:
                error(filename, linenum, 'whitespace/parens', 4,
                      'Extra space before ( in function call')
        # If the ) is followed only by a newline or a { + newline, assume it's
        # part of a control statement (if/while/etc), and don't complain
        if Search(r'[^)]\s+\)\s*[^{\s]', fncall):
            # If the closing parenthesis is preceded by only whitespaces,
            # try to give a more descriptive error message.
            if Search(r'^\s+\)', fncall):
                error(filename, linenum, 'whitespace/parens', 2,
                      'Closing ) should be moved to the previous line')
            else:
                error(filename, linenum, 'whitespace/parens', 2,
                      'Extra space before )')
Z
zhangjinchao01 已提交
2838 2839 2840


def IsBlankLine(line):
Y
Yu Yang 已提交
2841
    """Returns true if the given line is blank.
Z
zhangjinchao01 已提交
2842 2843 2844 2845 2846 2847 2848 2849 2850 2851

  We consider a line to be blank if the line is empty or consists of
  only white spaces.

  Args:
    line: A line of a string.

  Returns:
    True, if the given line is blank.
  """
Y
Yu Yang 已提交
2852
    return not line or line.isspace()
Z
zhangjinchao01 已提交
2853 2854 2855 2856


def CheckForNamespaceIndentation(filename, nesting_state, clean_lines, line,
                                 error):
Y
Yu Yang 已提交
2857 2858 2859 2860 2861
    is_namespace_indent_item = (
        len(nesting_state.stack) > 1 and
        nesting_state.stack[-1].check_namespace_indentation and
        isinstance(nesting_state.previous_stack_top, _NamespaceInfo) and
        nesting_state.previous_stack_top == nesting_state.stack[-2])
Z
zhangjinchao01 已提交
2862

Y
Yu Yang 已提交
2863 2864 2865 2866
    if ShouldCheckNamespaceIndentation(nesting_state, is_namespace_indent_item,
                                       clean_lines.elided, line):
        CheckItemIndentationInNamespace(filename, clean_lines.elided, line,
                                        error)
Z
zhangjinchao01 已提交
2867 2868


Y
Yu Yang 已提交
2869 2870 2871
def CheckForFunctionLengths(filename, clean_lines, linenum, function_state,
                            error):
    """Reports for long function bodies.
Z
zhangjinchao01 已提交
2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 2886 2887 2888 2889 2890 2891

  For an overview why this is done, see:
  http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml#Write_Short_Functions

  Uses a simplistic algorithm assuming other style guidelines
  (especially spacing) are followed.
  Only checks unindented functions, so class members are unchecked.
  Trivial bodies are unchecked, so constructors with huge initializer lists
  may be missed.
  Blank/comment lines are not counted so as to avoid encouraging the removal
  of vertical space and comments just to get through a lint check.
  NOLINT *on the last line of a function* disables this check.

  Args:
    filename: The name of the current file.
    clean_lines: A CleansedLines instance containing the file.
    linenum: The number of the line to check.
    function_state: Current function name and lines in body so far.
    error: The function to call with any errors found.
  """
Y
Yu Yang 已提交
2892 2893 2894 2895 2896 2897 2898 2899 2900 2901 2902 2903 2904 2905 2906 2907 2908 2909 2910 2911 2912 2913 2914 2915 2916 2917 2918 2919 2920 2921 2922 2923 2924 2925 2926 2927 2928 2929 2930 2931 2932 2933 2934 2935
    lines = clean_lines.lines
    line = lines[linenum]
    joined_line = ''

    starting_func = False
    regexp = r'(\w(\w|::|\*|\&|\s)*)\('  # decls * & space::name( ...
    match_result = Match(regexp, line)
    if match_result:
        # If the name is all caps and underscores, figure it's a macro and
        # ignore it, unless it's TEST or TEST_F.
        function_name = match_result.group(1).split()[-1]
        if function_name == 'TEST' or function_name == 'TEST_F' or (
                not Match(r'[A-Z_]+$', function_name)):
            starting_func = True

    if starting_func:
        body_found = False
        for start_linenum in xrange(linenum, clean_lines.NumLines()):
            start_line = lines[start_linenum]
            joined_line += ' ' + start_line.lstrip()
            if Search(r'(;|})',
                      start_line):  # Declarations and trivial functions
                body_found = True
                break  # ... ignore
            elif Search(r'{', start_line):
                body_found = True
                function = Search(r'((\w|:)*)\(', line).group(1)
                if Match(r'TEST', function):  # Handle TEST... macros
                    parameter_regexp = Search(r'(\(.*\))', joined_line)
                    if parameter_regexp:  # Ignore bad syntax
                        function += parameter_regexp.group(1)
                else:
                    function += '()'
                function_state.Begin(function)
                break
        if not body_found:
            # No body for the function (or evidence of a non-function) was found.
            error(filename, linenum, 'readability/fn_size', 5,
                  'Lint failed to find start of function body.')
    elif Match(r'^\}\s*$', line):  # function end
        function_state.Check(error, filename, linenum)
        function_state.End()
    elif not Match(r'^\s*$', line):
        function_state.Count()  # Count non-blank/non-comment lines.
Z
zhangjinchao01 已提交
2936 2937 2938 2939 2940 2941


_RE_PATTERN_TODO = re.compile(r'^//(\s*)TODO(\(.+?\))?:?(\s|$)?')


def CheckComment(line, filename, linenum, next_line_start, error):
Y
Yu Yang 已提交
2942
    """Checks for common mistakes in comments.
Z
zhangjinchao01 已提交
2943 2944 2945 2946 2947 2948 2949 2950

  Args:
    line: The line in question.
    filename: The name of the current file.
    linenum: The number of the line to check.
    next_line_start: The first non-whitespace column of the next line.
    error: The function to call with any errors found.
  """
Y
Yu Yang 已提交
2951 2952 2953 2954 2955 2956 2957 2958 2959 2960 2961 2962 2963 2964 2965 2966 2967 2968 2969 2970 2971 2972 2973 2974 2975 2976 2977 2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 2991 2992 2993 2994
    commentpos = line.find('//')
    if commentpos != -1:
        # Check if the // may be in quotes.  If so, ignore it
        # Comparisons made explicit for clarity -- pylint: disable=g-explicit-bool-comparison
        if (line.count('"', 0, commentpos) - line.count('\\"', 0, commentpos)
            ) % 2 == 0:  # not in quotes
            # Allow one space for new scopes, two spaces otherwise:
            if (not (Match(r'^.*{ *//', line) and next_line_start == commentpos)
                    and ((commentpos >= 1 and
                          line[commentpos - 1] not in string.whitespace) or
                         (commentpos >= 2 and
                          line[commentpos - 2] not in string.whitespace))):
                error(filename, linenum, 'whitespace/comments', 2,
                      'At least two spaces is best between code and comments')

            # Checks for common mistakes in TODO comments.
            comment = line[commentpos:]
            match = _RE_PATTERN_TODO.match(comment)
            if match:
                # One whitespace is correct; zero whitespace is handled elsewhere.
                leading_whitespace = match.group(1)
                if len(leading_whitespace) > 1:
                    error(filename, linenum, 'whitespace/todo', 2,
                          'Too many spaces before TODO')

                username = match.group(2)
                if not username:
                    error(filename, linenum, 'readability/todo', 2,
                          'Missing username in TODO; it should look like '
                          '"// TODO(my_username): Stuff."')

                middle_whitespace = match.group(3)
                # Comparisons made explicit for correctness -- pylint: disable=g-explicit-bool-comparison
                if middle_whitespace != ' ' and middle_whitespace != '':
                    error(filename, linenum, 'whitespace/todo', 2,
                          'TODO(my_username) should be followed by a space')

            # If the comment contains an alphanumeric character, there
            # should be a space somewhere between it and the // unless
            # it's a /// or //! Doxygen comment.
            if (Match(r'//[^ ]*\w', comment) and
                    not Match(r'(///|//\!)(\s+|$)', comment)):
                error(filename, linenum, 'whitespace/comments', 4,
                      'Should have a space between // and comment')
Z
zhangjinchao01 已提交
2995 2996 2997


def CheckAccess(filename, clean_lines, linenum, nesting_state, error):
Y
Yu Yang 已提交
2998
    """Checks for improper use of DISALLOW* macros.
Z
zhangjinchao01 已提交
2999 3000 3001 3002 3003 3004 3005 3006 3007

  Args:
    filename: The name of the current file.
    clean_lines: A CleansedLines instance containing the file.
    linenum: The number of the line to check.
    nesting_state: A NestingState instance which maintains information about
                   the current stack of nested blocks being parsed.
    error: The function to call with any errors found.
  """
Y
Yu Yang 已提交
3008 3009 3010 3011 3012 3013 3014 3015 3016 3017 3018 3019 3020 3021 3022 3023 3024
    line = clean_lines.elided[linenum]  # get rid of comments and strings

    matched = Match((r'\s*(DISALLOW_COPY_AND_ASSIGN|'
                     r'DISALLOW_IMPLICIT_CONSTRUCTORS)'), line)
    if not matched:
        return
    if nesting_state.stack and isinstance(nesting_state.stack[-1], _ClassInfo):
        if nesting_state.stack[-1].access != 'private':
            error(filename, linenum, 'readability/constructors', 3,
                  '%s must be in the private: section' % matched.group(1))

    else:
        # Found DISALLOW* macro outside a class declaration, or perhaps it
        # was used inside a function when it should have been part of the
        # class declaration.  We could issue a warning here, but it
        # probably resulted in a compiler error already.
        pass
Z
zhangjinchao01 已提交
3025 3026 3027


def CheckSpacing(filename, clean_lines, linenum, nesting_state, error):
Y
Yu Yang 已提交
3028
    """Checks for the correctness of various spacing issues in the code.
Z
zhangjinchao01 已提交
3029 3030 3031 3032 3033 3034 3035 3036 3037 3038 3039 3040 3041 3042 3043 3044

  Things we check for: spaces around operators, spaces after
  if/for/while/switch, no spaces around parens in function calls, two
  spaces between code and comment, don't start a block with a blank
  line, don't end a function with a blank line, don't add a blank line
  after public/protected/private, don't have too many blank lines in a row.

  Args:
    filename: The name of the current file.
    clean_lines: A CleansedLines instance containing the file.
    linenum: The number of the line to check.
    nesting_state: A NestingState instance which maintains information about
                   the current stack of nested blocks being parsed.
    error: The function to call with any errors found.
  """

Y
Yu Yang 已提交
3045 3046 3047 3048 3049 3050 3051 3052 3053 3054 3055 3056 3057 3058
    # Don't use "elided" lines here, otherwise we can't check commented lines.
    # Don't want to use "raw" either, because we don't want to check inside C++11
    # raw strings,
    raw = clean_lines.lines_without_raw_strings
    line = raw[linenum]

    # Before nixing comments, check if the line is blank for no good
    # reason.  This includes the first line after a block is opened, and
    # blank lines at the end of a function (ie, right before a line like '}'
    #
    # Skip all the blank line checks if we are immediately inside a
    # namespace body.  In other words, don't issue blank line warnings
    # for this block:
    #   namespace {
Z
zhangjinchao01 已提交
3059 3060
    #
    #   }
Y
Yu Yang 已提交
3061 3062 3063 3064 3065 3066 3067 3068 3069 3070 3071 3072 3073 3074 3075 3076 3077 3078 3079 3080 3081 3082 3083 3084 3085 3086 3087 3088 3089 3090 3091 3092 3093 3094 3095 3096 3097 3098 3099 3100 3101 3102 3103 3104 3105 3106 3107 3108 3109 3110 3111 3112 3113 3114 3115 3116 3117 3118 3119 3120 3121 3122 3123 3124 3125 3126 3127 3128 3129
    #
    # A warning about missing end of namespace comments will be issued instead.
    #
    # Also skip blank line checks for 'extern "C"' blocks, which are formatted
    # like namespaces.
    if (IsBlankLine(line) and not nesting_state.InNamespaceBody() and
            not nesting_state.InExternC()):
        elided = clean_lines.elided
        prev_line = elided[linenum - 1]
        prevbrace = prev_line.rfind('{')
        # TODO(unknown): Don't complain if line before blank line, and line after,
        #                both start with alnums and are indented the same amount.
        #                This ignores whitespace at the start of a namespace block
        #                because those are not usually indented.
        if prevbrace != -1 and prev_line[prevbrace:].find('}') == -1:
            # OK, we have a blank line at the start of a code block.  Before we
            # complain, we check if it is an exception to the rule: The previous
            # non-empty line has the parameters of a function header that are indented
            # 4 spaces (because they did not fit in a 80 column line when placed on
            # the same line as the function name).  We also check for the case where
            # the previous line is indented 6 spaces, which may happen when the
            # initializers of a constructor do not fit into a 80 column line.
            exception = False
            if Match(r' {6}\w', prev_line):  # Initializer list?
                # We are looking for the opening column of initializer list, which
                # should be indented 4 spaces to cause 6 space indentation afterwards.
                search_position = linenum - 2
                while (search_position >= 0 and
                       Match(r' {6}\w', elided[search_position])):
                    search_position -= 1
                exception = (search_position >= 0 and
                             elided[search_position][:5] == '    :')
            else:
                # Search for the function arguments or an initializer list.  We use a
                # simple heuristic here: If the line is indented 4 spaces; and we have a
                # closing paren, without the opening paren, followed by an opening brace
                # or colon (for initializer lists) we assume that it is the last line of
                # a function header.  If we have a colon indented 4 spaces, it is an
                # initializer list.
                exception = (Match(r' {4}\w[^\(]*\)\s*(const\s*)?(\{\s*$|:)',
                                   prev_line) or Match(r' {4}:', prev_line))

            if not exception:
                error(filename, linenum, 'whitespace/blank_line', 2,
                      'Redundant blank line at the start of a code block '
                      'should be deleted.')
        # Ignore blank lines at the end of a block in a long if-else
        # chain, like this:
        #   if (condition1) {
        #     // Something followed by a blank line
        #
        #   } else if (condition2) {
        #     // Something else
        #   }
        if linenum + 1 < clean_lines.NumLines():
            next_line = raw[linenum + 1]
            if (next_line and Match(r'\s*}', next_line) and
                    next_line.find('} else ') == -1):
                error(filename, linenum, 'whitespace/blank_line', 3,
                      'Redundant blank line at the end of a code block '
                      'should be deleted.')

        matched = Match(r'\s*(public|protected|private):', prev_line)
        if matched:
            error(filename, linenum, 'whitespace/blank_line', 3,
                  'Do not leave a blank line after "%s:"' % matched.group(1))

    # Next, check comments
    next_line_start = 0
Z
zhangjinchao01 已提交
3130
    if linenum + 1 < clean_lines.NumLines():
Y
Yu Yang 已提交
3131 3132 3133
        next_line = raw[linenum + 1]
        next_line_start = len(next_line) - len(next_line.lstrip())
    CheckComment(line, filename, linenum, next_line_start, error)
Z
zhangjinchao01 已提交
3134

Y
Yu Yang 已提交
3135 3136
    # get rid of comments and strings
    line = clean_lines.elided[linenum]
Z
zhangjinchao01 已提交
3137

Y
Yu Yang 已提交
3138 3139 3140 3141
    # You shouldn't have spaces before your brackets, except maybe after
    # 'delete []' or 'return []() {};'
    if Search(r'\w\s+\[', line) and not Search(r'(?:delete|return)\s+\[', line):
        error(filename, linenum, 'whitespace/braces', 5, 'Extra space before [')
Z
zhangjinchao01 已提交
3142

Y
Yu Yang 已提交
3143 3144 3145 3146 3147 3148
    # In range-based for, we wanted spaces before and after the colon, but
    # not around "::" tokens that might appear.
    if (Search(r'for *\(.*[^:]:[^: ]', line) or
            Search(r'for *\(.*[^: ]:[^:]', line)):
        error(filename, linenum, 'whitespace/forcolon', 2,
              'Missing space around colon in range-based for loop')
Z
zhangjinchao01 已提交
3149 3150 3151


def CheckOperatorSpacing(filename, clean_lines, linenum, error):
Y
Yu Yang 已提交
3152
    """Checks for horizontal spacing around operators.
Z
zhangjinchao01 已提交
3153 3154 3155 3156 3157 3158 3159

  Args:
    filename: The name of the current file.
    clean_lines: A CleansedLines instance containing the file.
    linenum: The number of the line to check.
    error: The function to call with any errors found.
  """
Y
Yu Yang 已提交
3160 3161 3162 3163 3164 3165 3166 3167 3168 3169 3170 3171 3172 3173 3174 3175 3176 3177 3178 3179 3180 3181 3182 3183 3184 3185 3186 3187 3188 3189 3190 3191 3192 3193 3194 3195 3196 3197 3198 3199 3200 3201 3202 3203
    line = clean_lines.elided[linenum]

    # Don't try to do spacing checks for operator methods.  Do this by
    # replacing the troublesome characters with something else,
    # preserving column position for all other characters.
    #
    # The replacement is done repeatedly to avoid false positives from
    # operators that call operators.
    while True:
        match = Match(r'^(.*\boperator\b)(\S+)(\s*\(.*)$', line)
        if match:
            line = match.group(1) + ('_' * len(match.group(2))) + match.group(3)
        else:
            break

    # We allow no-spaces around = within an if: "if ( (a=Foo()) == 0 )".
    # Otherwise not.  Note we only check for non-spaces on *both* sides;
    # sometimes people put non-spaces on one side when aligning ='s among
    # many lines (not that this is behavior that I approve of...)
    if ((Search(r'[\w.]=', line) or
         Search(r'=[\w.]', line)) and not Search(r'\b(if|while|for) ', line)
            # Operators taken from [lex.operators] in C++11 standard.
            and
            not Search(r'(>=|<=|==|!=|&=|\^=|\|=|\+=|\*=|\/=|\%=)', line) and
            not Search(r'operator=', line)):
        error(filename, linenum, 'whitespace/operators', 4,
              'Missing spaces around =')

    # It's ok not to have spaces around binary operators like + - * /, but if
    # there's too little whitespace, we get concerned.  It's hard to tell,
    # though, so we punt on this one for now.  TODO.

    # You should always have whitespace around binary operators.
    #
    # Check <= and >= first to avoid false positives with < and >, then
    # check non-include lines for spacing around < and >.
    #
    # If the operator is followed by a comma, assume it's be used in a
    # macro context and don't do any checks.  This avoids false
    # positives.
    #
    # Note that && is not included here.  Those are checked separately
    # in CheckRValueReference
    match = Search(r'[^<>=!\s](==|!=|<=|>=|\|\|)[^<>=!\s,;\)]', line)
Z
zhangjinchao01 已提交
3204 3205
    if match:
        error(filename, linenum, 'whitespace/operators', 3,
Y
Yu Yang 已提交
3206 3207 3208 3209 3210 3211 3212 3213 3214 3215 3216 3217 3218 3219 3220 3221 3222 3223 3224 3225 3226 3227 3228 3229 3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241 3242
              'Missing spaces around %s' % match.group(1))
    elif not Match(r'#.*include', line):
        # Look for < that is not surrounded by spaces.  This is only
        # triggered if both sides are missing spaces, even though
        # technically should should flag if at least one side is missing a
        # space.  This is done to avoid some false positives with shifts.
        match = Match(r'^(.*[^\s<])<[^\s=<,]', line)
        if match:
            (_, _, end_pos) = CloseExpression(clean_lines, linenum,
                                              len(match.group(1)))
            if end_pos <= -1:
                error(filename, linenum, 'whitespace/operators', 3,
                      'Missing spaces around <')

        # Look for > that is not surrounded by spaces.  Similar to the
        # above, we only trigger if both sides are missing spaces to avoid
        # false positives with shifts.
        match = Match(r'^(.*[^-\s>])>[^\s=>,]', line)
        if match:
            (_, _, start_pos) = ReverseCloseExpression(clean_lines, linenum,
                                                       len(match.group(1)))
            if start_pos <= -1:
                error(filename, linenum, 'whitespace/operators', 3,
                      'Missing spaces around >')

    # We allow no-spaces around << when used like this: 10<<20, but
    # not otherwise (particularly, not when used as streams)
    #
    # We also allow operators following an opening parenthesis, since
    # those tend to be macros that deal with operators.
    match = Search(r'(operator|[^\s(<])(?:L|UL|ULL|l|ul|ull)?<<([^\s,=<])',
                   line)
    if (match and
            not (match.group(1).isdigit() and match.group(2).isdigit()) and
            not (match.group(1) == 'operator' and match.group(2) == ';')):
        error(filename, linenum, 'whitespace/operators', 3,
              'Missing spaces around <<')
Z
zhangjinchao01 已提交
3243

Y
Yu Yang 已提交
3244 3245 3246 3247 3248 3249 3250 3251 3252 3253 3254 3255 3256
    # We allow no-spaces around >> for almost anything.  This is because
    # C++11 allows ">>" to close nested templates, which accounts for
    # most cases when ">>" is not followed by a space.
    #
    # We still warn on ">>" followed by alpha character, because that is
    # likely due to ">>" being used for right shifts, e.g.:
    #   value >> alpha
    #
    # When ">>" is used to close templates, the alphanumeric letter that
    # follows would be part of an identifier, and there should still be
    # a space separating the template type and the identifier.
    #   type<type<type>> alpha
    match = Search(r'>>[a-zA-Z_]', line)
Z
zhangjinchao01 已提交
3257 3258
    if match:
        error(filename, linenum, 'whitespace/operators', 3,
Y
Yu Yang 已提交
3259 3260 3261 3262 3263 3264 3265
              'Missing spaces around >>')

    # There shouldn't be space around unary operators
    match = Search(r'(!\s|~\s|[\s]--[\s;]|[\s]\+\+[\s;])', line)
    if match:
        error(filename, linenum, 'whitespace/operators', 4,
              'Extra space for operator %s' % match.group(1))
Z
zhangjinchao01 已提交
3266 3267 3268


def CheckParenthesisSpacing(filename, clean_lines, linenum, error):
Y
Yu Yang 已提交
3269
    """Checks for horizontal spacing around parentheses.
Z
zhangjinchao01 已提交
3270 3271 3272 3273 3274 3275 3276

  Args:
    filename: The name of the current file.
    clean_lines: A CleansedLines instance containing the file.
    linenum: The number of the line to check.
    error: The function to call with any errors found.
  """
Y
Yu Yang 已提交
3277 3278 3279 3280 3281
    line = clean_lines.elided[linenum]

    # No spaces after an if, while, switch, or for
    match = Search(r' (if\(|for\(|while\(|switch\()', line)
    if match:
Z
zhangjinchao01 已提交
3282
        error(filename, linenum, 'whitespace/parens', 5,
Y
Yu Yang 已提交
3283 3284 3285 3286 3287 3288 3289 3290 3291 3292 3293 3294 3295 3296 3297 3298 3299 3300 3301 3302
              'Missing space before ( in %s' % match.group(1))

    # For if/for/while/switch, the left and right parens should be
    # consistent about how many spaces are inside the parens, and
    # there should either be zero or one spaces inside the parens.
    # We don't want: "if ( foo)" or "if ( foo   )".
    # Exception: "for ( ; foo; bar)" and "for (foo; bar; )" are allowed.
    match = Search(r'\b(if|for|while|switch)\s*'
                   r'\(([ ]*)(.).*[^ ]+([ ]*)\)\s*{\s*$', line)
    if match:
        if len(match.group(2)) != len(match.group(4)):
            if not (match.group(3) == ';' and
                    len(match.group(2)) == 1 + len(match.group(4)) or
                    not match.group(2) and Search(r'\bfor\s*\(.*; \)', line)):
                error(filename, linenum, 'whitespace/parens', 5,
                      'Mismatching spaces inside () in %s' % match.group(1))
        if len(match.group(2)) not in [0, 1]:
            error(filename, linenum, 'whitespace/parens', 5,
                  'Should have zero or one spaces inside ( and ) in %s' %
                  match.group(1))
Z
zhangjinchao01 已提交
3303 3304 3305


def CheckCommaSpacing(filename, clean_lines, linenum, error):
Y
Yu Yang 已提交
3306
    """Checks for horizontal spacing near commas and semicolons.
Z
zhangjinchao01 已提交
3307 3308 3309 3310 3311 3312 3313

  Args:
    filename: The name of the current file.
    clean_lines: A CleansedLines instance containing the file.
    linenum: The number of the line to check.
    error: The function to call with any errors found.
  """
Y
Yu Yang 已提交
3314 3315 3316 3317 3318 3319 3320 3321 3322 3323 3324 3325 3326 3327 3328 3329 3330 3331 3332 3333 3334 3335 3336 3337
    raw = clean_lines.lines_without_raw_strings
    line = clean_lines.elided[linenum]

    # You should always have a space after a comma (either as fn arg or operator)
    #
    # This does not apply when the non-space character following the
    # comma is another comma, since the only time when that happens is
    # for empty macro arguments.
    #
    # We run this check in two passes: first pass on elided lines to
    # verify that lines contain missing whitespaces, second pass on raw
    # lines to confirm that those missing whitespaces are not due to
    # elided comments.
    if (Search(r',[^,\s]', ReplaceAll(r'\boperator\s*,\s*\(', 'F(', line)) and
            Search(r',[^,\s]', raw[linenum])):
        error(filename, linenum, 'whitespace/comma', 3, 'Missing space after ,')

    # You should always have a space after a semicolon
    # except for few corner cases
    # TODO(unknown): clarify if 'if (1) { return 1;}' is requires one more
    # space after ;
    if Search(r';[^\s};\\)/]', line):
        error(filename, linenum, 'whitespace/semicolon', 3,
              'Missing space after ;')
Z
zhangjinchao01 已提交
3338 3339 3340


def CheckBracesSpacing(filename, clean_lines, linenum, error):
Y
Yu Yang 已提交
3341
    """Checks for horizontal spacing near commas.
Z
zhangjinchao01 已提交
3342 3343 3344 3345 3346 3347 3348

  Args:
    filename: The name of the current file.
    clean_lines: A CleansedLines instance containing the file.
    linenum: The number of the line to check.
    error: The function to call with any errors found.
  """
Y
Yu Yang 已提交
3349 3350 3351 3352 3353 3354 3355 3356 3357 3358 3359 3360 3361 3362 3363 3364 3365 3366 3367 3368 3369 3370 3371 3372 3373 3374 3375 3376 3377 3378 3379 3380 3381 3382 3383 3384 3385 3386 3387 3388 3389 3390 3391 3392 3393 3394 3395 3396 3397 3398 3399 3400 3401 3402 3403 3404 3405 3406 3407 3408 3409 3410 3411 3412 3413 3414 3415 3416
    line = clean_lines.elided[linenum]

    # Except after an opening paren, or after another opening brace (in case of
    # an initializer list, for instance), you should have spaces before your
    # braces. And since you should never have braces at the beginning of a line,
    # this is an easy test.
    match = Match(r'^(.*[^ ({>]){', line)
    if match:
        # Try a bit harder to check for brace initialization.  This
        # happens in one of the following forms:
        #   Constructor() : initializer_list_{} { ... }
        #   Constructor{}.MemberFunction()
        #   Type variable{};
        #   FunctionCall(type{}, ...);
        #   LastArgument(..., type{});
        #   LOG(INFO) << type{} << " ...";
        #   map_of_type[{...}] = ...;
        #   ternary = expr ? new type{} : nullptr;
        #   OuterTemplate<InnerTemplateConstructor<Type>{}>
        #
        # We check for the character following the closing brace, and
        # silence the warning if it's one of those listed above, i.e.
        # "{.;,)<>]:".
        #
        # To account for nested initializer list, we allow any number of
        # closing braces up to "{;,)<".  We can't simply silence the
        # warning on first sight of closing brace, because that would
        # cause false negatives for things that are not initializer lists.
        #   Silence this:         But not this:
        #     Outer{                if (...) {
        #       Inner{...}            if (...){  // Missing space before {
        #     };                    }
        #
        # There is a false negative with this approach if people inserted
        # spurious semicolons, e.g. "if (cond){};", but we will catch the
        # spurious semicolon with a separate check.
        (endline, endlinenum, endpos) = CloseExpression(clean_lines, linenum,
                                                        len(match.group(1)))
        trailing_text = ''
        if endpos > -1:
            trailing_text = endline[endpos:]
        for offset in xrange(endlinenum + 1,
                             min(endlinenum + 3, clean_lines.NumLines() - 1)):
            trailing_text += clean_lines.elided[offset]
        if not Match(r'^[\s}]*[{.;,)<>\]:]', trailing_text):
            error(filename, linenum, 'whitespace/braces', 5,
                  'Missing space before {')

    # Make sure '} else {' has spaces.
    if Search(r'}else', line):
        error(filename, linenum, 'whitespace/braces', 5,
              'Missing space before else')

    # You shouldn't have a space before a semicolon at the end of the line.
    # There's a special case for "for" since the style guide allows space before
    # the semicolon there.
    if Search(r':\s*;\s*$', line):
        error(filename, linenum, 'whitespace/semicolon', 5,
              'Semicolon defining empty statement. Use {} instead.')
    elif Search(r'^\s*;\s*$', line):
        error(
            filename, linenum, 'whitespace/semicolon', 5,
            'Line contains only semicolon. If this should be an empty statement, '
            'use {} instead.')
    elif (Search(r'\s+;\s*$', line) and not Search(r'\bfor\b', line)):
        error(filename, linenum, 'whitespace/semicolon', 5,
              'Extra space before last semicolon. If this should be an empty '
              'statement, use {} instead.')
Z
zhangjinchao01 已提交
3417 3418 3419


def IsDecltype(clean_lines, linenum, column):
Y
Yu Yang 已提交
3420
    """Check if the token ending on (linenum, column) is decltype().
Z
zhangjinchao01 已提交
3421 3422 3423 3424 3425 3426 3427 3428

  Args:
    clean_lines: A CleansedLines instance containing the file.
    linenum: the number of the line to check.
    column: end column of the token to check.
  Returns:
    True if this token is decltype() expression, False otherwise.
  """
Y
Yu Yang 已提交
3429 3430 3431 3432 3433
    (text, _, start_col) = ReverseCloseExpression(clean_lines, linenum, column)
    if start_col < 0:
        return False
    if Search(r'\bdecltype\s*$', text[0:start_col]):
        return True
Z
zhangjinchao01 已提交
3434 3435 3436 3437
    return False


def IsTemplateParameterList(clean_lines, linenum, column):
Y
Yu Yang 已提交
3438
    """Check if the token ending on (linenum, column) is the end of template<>.
Z
zhangjinchao01 已提交
3439 3440 3441 3442 3443 3444 3445 3446

  Args:
    clean_lines: A CleansedLines instance containing the file.
    linenum: the number of the line to check.
    column: end column of the token to check.
  Returns:
    True if this token is end of a template parameter list, False otherwise.
  """
Y
Yu Yang 已提交
3447 3448 3449 3450 3451 3452
    (_, startline, startpos) = ReverseCloseExpression(clean_lines, linenum,
                                                      column)
    if (startpos > -1 and Search(r'\btemplate\s*$',
                                 clean_lines.elided[startline][0:startpos])):
        return True
    return False
Z
zhangjinchao01 已提交
3453 3454 3455


def IsRValueType(typenames, clean_lines, nesting_state, linenum, column):
Y
Yu Yang 已提交
3456
    """Check if the token ending on (linenum, column) is a type.
Z
zhangjinchao01 已提交
3457 3458 3459 3460 3461 3462 3463 3464 3465 3466 3467 3468 3469 3470

  Assumes that text to the right of the column is "&&" or a function
  name.

  Args:
    typenames: set of type names from template-argument-list.
    clean_lines: A CleansedLines instance containing the file.
    nesting_state: A NestingState instance which maintains information about
                   the current stack of nested blocks being parsed.
    linenum: the number of the line to check.
    column: end column of the token to check.
  Returns:
    True if this token is a type, False if we are not sure.
  """
Y
Yu Yang 已提交
3471
    prefix = clean_lines.elided[linenum][0:column]
Z
zhangjinchao01 已提交
3472

Y
Yu Yang 已提交
3473 3474 3475 3476 3477 3478
    # Get one word to the left.  If we failed to do so, this is most
    # likely not a type, since it's unlikely that the type name and "&&"
    # would be split across multiple lines.
    match = Match(r'^(.*)(\b\w+|[>*)&])\s*$', prefix)
    if not match:
        return False
Z
zhangjinchao01 已提交
3479

Y
Yu Yang 已提交
3480 3481 3482 3483 3484
    # Check text following the token.  If it's "&&>" or "&&," or "&&...", it's
    # most likely a rvalue reference used inside a template.
    suffix = clean_lines.elided[linenum][column:]
    if Match(r'&&\s*(?:[>,]|\.\.\.)', suffix):
        return True
Z
zhangjinchao01 已提交
3485

Y
Yu Yang 已提交
3486 3487 3488 3489 3490 3491 3492 3493 3494 3495 3496 3497 3498 3499
    # Check for known types and end of templates:
    #   int&& variable
    #   vector<int>&& variable
    #
    # Because this function is called recursively, we also need to
    # recognize pointer and reference types:
    #   int* Function()
    #   int& Function()
    if (match.group(2) in typenames or match.group(2) in [
            'char', 'char16_t', 'char32_t', 'wchar_t', 'bool', 'short', 'int',
            'long', 'signed', 'unsigned', 'float', 'double', 'void', 'auto',
            '>', '*', '&'
    ]):
        return True
Z
zhangjinchao01 已提交
3500

Y
Yu Yang 已提交
3501 3502 3503 3504 3505 3506 3507 3508 3509 3510 3511 3512 3513 3514 3515 3516
    # If we see a close parenthesis, look for decltype on the other side.
    # decltype would unambiguously identify a type, anything else is
    # probably a parenthesized expression and not a type.
    if match.group(2) == ')':
        return IsDecltype(clean_lines, linenum,
                          len(match.group(1)) + len(match.group(2)) - 1)

    # Check for casts and cv-qualifiers.
    #   match.group(1)  remainder
    #   --------------  ---------
    #   const_cast<     type&&
    #   const           type&&
    #   type            const&&
    if Search(r'\b(?:const_cast\s*<|static_cast\s*<|dynamic_cast\s*<|'
              r'reinterpret_cast\s*<|\w+\s)\s*$', match.group(1)):
        return True
Z
zhangjinchao01 已提交
3517

Y
Yu Yang 已提交
3518 3519 3520 3521 3522 3523 3524 3525 3526 3527 3528 3529 3530 3531 3532 3533 3534 3535 3536 3537 3538 3539 3540 3541 3542 3543
    # Look for a preceding symbol that might help differentiate the context.
    # These are the cases that would be ambiguous:
    #   match.group(1)  remainder
    #   --------------  ---------
    #   Call         (   expression &&
    #   Declaration  (   type&&
    #   sizeof       (   type&&
    #   if           (   expression &&
    #   while        (   expression &&
    #   for          (   type&&
    #   for(         ;   expression &&
    #   statement    ;   type&&
    #   block        {   type&&
    #   constructor  {   expression &&
    start = linenum
    line = match.group(1)
    match_symbol = None
    while start >= 0:
        # We want to skip over identifiers and commas to get to a symbol.
        # Commas are skipped so that we can find the opening parenthesis
        # for function parameter lists.
        match_symbol = Match(r'^(.*)([^\w\s,])[\w\s,]*$', line)
        if match_symbol:
            break
        start -= 1
        line = clean_lines.elided[start]
Z
zhangjinchao01 已提交
3544

Y
Yu Yang 已提交
3545 3546 3547
    if not match_symbol:
        # Probably the first statement in the file is an rvalue reference
        return True
Z
zhangjinchao01 已提交
3548

Y
Yu Yang 已提交
3549 3550 3551 3552
    if match_symbol.group(2) == '}':
        # Found closing brace, probably an indicate of this:
        #   block{} type&&
        return True
Z
zhangjinchao01 已提交
3553

Y
Yu Yang 已提交
3554 3555 3556 3557 3558 3559 3560 3561 3562 3563 3564
    if match_symbol.group(2) == ';':
        # Found semicolon, probably one of these:
        #   for(; expression &&
        #   statement; type&&

        # Look for the previous 'for(' in the previous lines.
        before_text = match_symbol.group(1)
        for i in xrange(start - 1, max(start - 6, 0), -1):
            before_text = clean_lines.elided[i] + before_text
        if Search(r'for\s*\([^{};]*$', before_text):
            # This is the condition inside a for-loop
Z
zhangjinchao01 已提交
3565 3566
            return False

Y
Yu Yang 已提交
3567 3568
        # Did not find a for-init-statement before this semicolon, so this
        # is probably a new statement and not a condition.
Z
zhangjinchao01 已提交
3569 3570
        return True

Y
Yu Yang 已提交
3571 3572 3573 3574 3575 3576 3577 3578 3579 3580 3581 3582 3583 3584 3585 3586 3587 3588 3589 3590 3591 3592 3593 3594 3595 3596
    if match_symbol.group(2) == '{':
        # Found opening brace, probably one of these:
        #   block{ type&& = ... ; }
        #   constructor{ expression && expression }

        # Look for a closing brace or a semicolon.  If we see a semicolon
        # first, this is probably a rvalue reference.
        line = clean_lines.elided[start][0:len(match_symbol.group(1)) + 1]
        end = start
        depth = 1
        while True:
            for ch in line:
                if ch == ';':
                    return True
                elif ch == '{':
                    depth += 1
                elif ch == '}':
                    depth -= 1
                    if depth == 0:
                        return False
            end += 1
            if end >= clean_lines.NumLines():
                break
            line = clean_lines.elided[end]
        # Incomplete program?
        return False
Z
zhangjinchao01 已提交
3597

Y
Yu Yang 已提交
3598 3599 3600 3601 3602 3603 3604 3605 3606 3607 3608 3609 3610 3611 3612 3613 3614 3615 3616 3617 3618 3619 3620 3621 3622 3623 3624 3625 3626 3627
    if match_symbol.group(2) == '(':
        # Opening parenthesis.  Need to check what's to the left of the
        # parenthesis.  Look back one extra line for additional context.
        before_text = match_symbol.group(1)
        if linenum > 1:
            before_text = clean_lines.elided[linenum - 1] + before_text
        before_text = match_symbol.group(1)

        # Patterns that are likely to be types:
        #   [](type&&
        #   for (type&&
        #   sizeof(type&&
        #   operator=(type&&
        #
        if Search(r'(?:\]|\bfor|\bsizeof|\boperator\s*\S+\s*)\s*$',
                  before_text):
            return True

        # Patterns that are likely to be expressions:
        #   if (expression &&
        #   while (expression &&
        #   : initializer(expression &&
        #   , initializer(expression &&
        #   ( FunctionCall(expression &&
        #   + FunctionCall(expression &&
        #   + (expression &&
        #
        # The last '+' represents operators such as '+' and '-'.
        if Search(r'(?:\bif|\bwhile|[-+=%^(<!?:,&*]\s*)$', before_text):
            return False
Z
zhangjinchao01 已提交
3628

Y
Yu Yang 已提交
3629 3630 3631 3632 3633 3634 3635 3636 3637 3638 3639 3640 3641 3642 3643 3644 3645 3646 3647 3648 3649 3650 3651 3652 3653 3654 3655 3656 3657 3658
        # Something else.  Check that tokens to the left look like
        #   return_type function_name
        match_func = Match(r'^(.*\S.*)\s+\w(?:\w|::)*(?:<[^<>]*>)?\s*$',
                           match_symbol.group(1))
        if match_func:
            # Check for constructors, which don't have return types.
            if Search(r'\b(?:explicit|inline)$', match_func.group(1)):
                return True
            implicit_constructor = Match(r'\s*(\w+)\((?:const\s+)?(\w+)',
                                         prefix)
            if (implicit_constructor and implicit_constructor.group(1) ==
                    implicit_constructor.group(2)):
                return True
            return IsRValueType(typenames, clean_lines, nesting_state, linenum,
                                len(match_func.group(1)))

        # Nothing before the function name.  If this is inside a block scope,
        # this is probably a function call.
        return not (nesting_state.previous_stack_top and
                    nesting_state.previous_stack_top.IsBlockInfo())

    if match_symbol.group(2) == '>':
        # Possibly a closing bracket, check that what's on the other side
        # looks like the start of a template.
        return IsTemplateParameterList(clean_lines, start,
                                       len(match_symbol.group(1)))

    # Some other symbol, usually something like "a=b&&c".  This is most
    # likely not a type.
    return False
Z
zhangjinchao01 已提交
3659 3660 3661


def IsDeletedOrDefault(clean_lines, linenum):
Y
Yu Yang 已提交
3662
    """Check if current constructor or operator is deleted or default.
Z
zhangjinchao01 已提交
3663 3664 3665 3666 3667 3668 3669

  Args:
    clean_lines: A CleansedLines instance containing the file.
    linenum: The number of the line to check.
  Returns:
    True if this is a deleted or default constructor.
  """
Y
Yu Yang 已提交
3670 3671 3672 3673 3674 3675 3676 3677
    open_paren = clean_lines.elided[linenum].find('(')
    if open_paren < 0:
        return False
    (close_line, _, close_paren) = CloseExpression(clean_lines, linenum,
                                                   open_paren)
    if close_paren < 0:
        return False
    return Match(r'\s*=\s*(?:delete|default)\b', close_line[close_paren:])
Z
zhangjinchao01 已提交
3678 3679 3680


def IsRValueAllowed(clean_lines, linenum, typenames):
Y
Yu Yang 已提交
3681
    """Check if RValue reference is allowed on a particular line.
Z
zhangjinchao01 已提交
3682 3683 3684 3685 3686 3687 3688 3689

  Args:
    clean_lines: A CleansedLines instance containing the file.
    linenum: The number of the line to check.
    typenames: set of type names from template-argument-list.
  Returns:
    True if line is within the region where RValue references are allowed.
  """
Y
Yu Yang 已提交
3690 3691 3692
    # Allow region marked by PUSH/POP macros
    for i in xrange(linenum, 0, -1):
        line = clean_lines.elided[i]
Z
zhangjinchao01 已提交
3693
        if Match(r'GOOGLE_ALLOW_RVALUE_REFERENCES_(?:PUSH|POP)', line):
Y
Yu Yang 已提交
3694 3695 3696 3697 3698 3699
            if not line.endswith('PUSH'):
                return False
            for j in xrange(linenum, clean_lines.NumLines(), 1):
                line = clean_lines.elided[j]
                if Match(r'GOOGLE_ALLOW_RVALUE_REFERENCES_(?:PUSH|POP)', line):
                    return line.endswith('POP')
Z
zhangjinchao01 已提交
3700

Y
Yu Yang 已提交
3701 3702 3703 3704 3705 3706 3707 3708 3709 3710 3711 3712 3713 3714 3715 3716 3717 3718 3719 3720 3721 3722 3723 3724 3725 3726 3727 3728
    # Allow operator=
    line = clean_lines.elided[linenum]
    if Search(r'\boperator\s*=\s*\(', line):
        return IsDeletedOrDefault(clean_lines, linenum)

    # Allow constructors
    match = Match(r'\s*(?:[\w<>]+::)*([\w<>]+)\s*::\s*([\w<>]+)\s*\(', line)
    if match and match.group(1) == match.group(2):
        return IsDeletedOrDefault(clean_lines, linenum)
    if Search(r'\b(?:explicit|inline)\s+[\w<>]+\s*\(', line):
        return IsDeletedOrDefault(clean_lines, linenum)

    if Match(r'\s*[\w<>]+\s*\(', line):
        previous_line = 'ReturnType'
        if linenum > 0:
            previous_line = clean_lines.elided[linenum - 1]
        if Match(r'^\s*$', previous_line) or Search(r'[{}:;]\s*$',
                                                    previous_line):
            return IsDeletedOrDefault(clean_lines, linenum)

    # Reject types not mentioned in template-argument-list
    while line:
        match = Match(r'^.*?(\w+)\s*&&(.*)$', line)
        if not match:
            break
        if match.group(1) not in typenames:
            return False
        line = match.group(2)
Z
zhangjinchao01 已提交
3729

Y
Yu Yang 已提交
3730 3731 3732 3733 3734 3735 3736
    # All RValue types that were in template-argument-list should have
    # been removed by now.  Those were allowed, assuming that they will
    # be forwarded.
    #
    # If there are no remaining RValue types left (i.e. types that were
    # not found in template-argument-list), flag those as not allowed.
    return line.find('&&') < 0
Z
zhangjinchao01 已提交
3737 3738 3739


def GetTemplateArgs(clean_lines, linenum):
Y
Yu Yang 已提交
3740
    """Find list of template arguments associated with this function declaration.
Z
zhangjinchao01 已提交
3741 3742 3743 3744 3745 3746 3747 3748 3749

  Args:
    clean_lines: A CleansedLines instance containing the file.
    linenum: Line number containing the start of the function declaration,
             usually one line after the end of the template-argument-list.
  Returns:
    Set of type names, or empty set if this does not appear to have
    any template parameters.
  """
Y
Yu Yang 已提交
3750 3751 3752 3753 3754 3755 3756 3757 3758 3759 3760 3761 3762 3763 3764
    # Find start of function
    func_line = linenum
    while func_line > 0:
        line = clean_lines.elided[func_line]
        if Match(r'^\s*$', line):
            return set()
        if line.find('(') >= 0:
            break
        func_line -= 1
    if func_line == 0:
        return set()

    # Collapse template-argument-list into a single string
    argument_list = ''
    match = Match(r'^(\s*template\s*)<', clean_lines.elided[func_line])
Z
zhangjinchao01 已提交
3765
    if match:
Y
Yu Yang 已提交
3766 3767 3768 3769 3770 3771 3772 3773 3774 3775 3776 3777 3778 3779 3780 3781 3782 3783 3784 3785 3786 3787 3788 3789 3790 3791 3792 3793 3794 3795 3796 3797 3798 3799 3800 3801 3802
        # template-argument-list on the same line as function name
        start_col = len(match.group(1))
        _, end_line, end_col = CloseExpression(clean_lines, func_line,
                                               start_col)
        if end_col > -1 and end_line == func_line:
            start_col += 1  # Skip the opening bracket
            argument_list = clean_lines.elided[func_line][start_col:end_col]

    elif func_line > 1:
        # template-argument-list one line before function name
        match = Match(r'^(.*)>\s*$', clean_lines.elided[func_line - 1])
        if match:
            end_col = len(match.group(1))
            _, start_line, start_col = ReverseCloseExpression(
                clean_lines, func_line - 1, end_col)
            if start_col > -1:
                start_col += 1  # Skip the opening bracket
                while start_line < func_line - 1:
                    argument_list += clean_lines.elided[start_line][start_col:]
                    start_col = 0
                    start_line += 1
                argument_list += clean_lines.elided[func_line - 1][start_col:
                                                                   end_col]

    if not argument_list:
        return set()

    # Extract type names
    typenames = set()
    while True:
        match = Match(r'^[,\s]*(?:typename|class)(?:\.\.\.)?\s+(\w+)(.*)$',
                      argument_list)
        if not match:
            break
        typenames.add(match.group(1))
        argument_list = match.group(2)
    return typenames
Z
zhangjinchao01 已提交
3803 3804 3805


def CheckRValueReference(filename, clean_lines, linenum, nesting_state, error):
Y
Yu Yang 已提交
3806
    """Check for rvalue references.
Z
zhangjinchao01 已提交
3807 3808 3809 3810 3811 3812 3813 3814 3815

  Args:
    filename: The name of the current file.
    clean_lines: A CleansedLines instance containing the file.
    linenum: The number of the line to check.
    nesting_state: A NestingState instance which maintains information about
                   the current stack of nested blocks being parsed.
    error: The function to call with any errors found.
  """
Y
Yu Yang 已提交
3816 3817 3818 3819 3820 3821 3822 3823 3824 3825 3826 3827 3828 3829 3830 3831 3832 3833 3834 3835 3836 3837 3838 3839
    # Find lines missing spaces around &&.
    # TODO(unknown): currently we don't check for rvalue references
    # with spaces surrounding the && to avoid false positives with
    # boolean expressions.
    line = clean_lines.elided[linenum]
    match = Match(r'^(.*\S)&&', line)
    if not match:
        match = Match(r'(.*)&&\S', line)
    if (not match) or '(&&)' in line or Search(r'\boperator\s*$',
                                               match.group(1)):
        return

    # Either poorly formed && or an rvalue reference, check the context
    # to get a more accurate error message.  Mostly we want to determine
    # if what's to the left of "&&" is a type or not.
    typenames = GetTemplateArgs(clean_lines, linenum)
    and_pos = len(match.group(1))
    if IsRValueType(typenames, clean_lines, nesting_state, linenum, and_pos):
        if not IsRValueAllowed(clean_lines, linenum, typenames):
            error(filename, linenum, 'build/c++11', 3,
                  'RValue references are an unapproved C++ feature.')
    else:
        error(filename, linenum, 'whitespace/operators', 3,
              'Missing spaces around &&')
Z
zhangjinchao01 已提交
3840 3841 3842


def CheckSectionSpacing(filename, clean_lines, class_info, linenum, error):
Y
Yu Yang 已提交
3843
    """Checks for additional blank line issues related to sections.
Z
zhangjinchao01 已提交
3844 3845 3846 3847 3848 3849 3850 3851 3852 3853

  Currently the only thing checked here is blank line before protected/private.

  Args:
    filename: The name of the current file.
    clean_lines: A CleansedLines instance containing the file.
    class_info: A _ClassInfo objects.
    linenum: The number of the line to check.
    error: The function to call with any errors found.
  """
Y
Yu Yang 已提交
3854 3855 3856 3857 3858 3859 3860 3861 3862 3863 3864 3865 3866 3867 3868 3869 3870 3871 3872 3873 3874 3875 3876 3877 3878 3879 3880 3881 3882 3883 3884 3885 3886 3887 3888 3889 3890 3891 3892 3893 3894 3895 3896
    # Skip checks if the class is small, where small means 25 lines or less.
    # 25 lines seems like a good cutoff since that's the usual height of
    # terminals, and any class that can't fit in one screen can't really
    # be considered "small".
    #
    # Also skip checks if we are on the first line.  This accounts for
    # classes that look like
    #   class Foo { public: ... };
    #
    # If we didn't find the end of the class, last_line would be zero,
    # and the check will be skipped by the first condition.
    if (class_info.last_line - class_info.starting_linenum <= 24 or
            linenum <= class_info.starting_linenum):
        return

    matched = Match(r'\s*(public|protected|private):',
                    clean_lines.lines[linenum])
    if matched:
        # Issue warning if the line before public/protected/private was
        # not a blank line, but don't do this if the previous line contains
        # "class" or "struct".  This can happen two ways:
        #  - We are at the beginning of the class.
        #  - We are forward-declaring an inner class that is semantically
        #    private, but needed to be public for implementation reasons.
        # Also ignores cases where the previous line ends with a backslash as can be
        # common when defining classes in C macros.
        prev_line = clean_lines.lines[linenum - 1]
        if (not IsBlankLine(prev_line) and
                not Search(r'\b(class|struct)\b', prev_line) and
                not Search(r'\\$', prev_line)):
            # Try a bit harder to find the beginning of the class.  This is to
            # account for multi-line base-specifier lists, e.g.:
            #   class Derived
            #       : public Base {
            end_class_head = class_info.starting_linenum
            for i in range(class_info.starting_linenum, linenum):
                if Search(r'\{\s*$', clean_lines.lines[i]):
                    end_class_head = i
                    break
            if end_class_head < linenum - 1:
                error(filename, linenum, 'whitespace/blank_line', 3,
                      '"%s:" should be preceded by a blank line' %
                      matched.group(1))
Z
zhangjinchao01 已提交
3897 3898 3899


def GetPreviousNonBlankLine(clean_lines, linenum):
Y
Yu Yang 已提交
3900
    """Return the most recent non-blank line and its line number.
Z
zhangjinchao01 已提交
3901 3902 3903 3904 3905 3906 3907 3908 3909 3910 3911 3912

  Args:
    clean_lines: A CleansedLines instance containing the file contents.
    linenum: The number of the line to check.

  Returns:
    A tuple with two elements.  The first element is the contents of the last
    non-blank line before the current line, or the empty string if this is the
    first non-blank line.  The second is the line number of that line, or -1
    if this is the first non-blank line.
  """

Y
Yu Yang 已提交
3913 3914 3915 3916 3917 3918 3919
    prevlinenum = linenum - 1
    while prevlinenum >= 0:
        prevline = clean_lines.elided[prevlinenum]
        if not IsBlankLine(prevline):  # if not a blank line...
            return (prevline, prevlinenum)
        prevlinenum -= 1
    return ('', -1)
Z
zhangjinchao01 已提交
3920 3921 3922


def CheckBraces(filename, clean_lines, linenum, error):
Y
Yu Yang 已提交
3923
    """Looks for misplaced braces (e.g. at the end of line).
Z
zhangjinchao01 已提交
3924 3925 3926 3927 3928 3929 3930 3931

  Args:
    filename: The name of the current file.
    clean_lines: A CleansedLines instance containing the file.
    linenum: The number of the line to check.
    error: The function to call with any errors found.
  """

Y
Yu Yang 已提交
3932 3933 3934 3935 3936 3937 3938 3939 3940 3941 3942 3943 3944 3945 3946 3947 3948 3949 3950 3951 3952 3953 3954 3955 3956 3957 3958 3959 3960 3961 3962 3963 3964 3965 3966 3967 3968 3969 3970
    line = clean_lines.elided[linenum]  # get rid of comments and strings

    if Match(r'\s*{\s*$', line):
        # We allow an open brace to start a line in the case where someone is using
        # braces in a block to explicitly create a new scope, which is commonly used
        # to control the lifetime of stack-allocated variables.  Braces are also
        # used for brace initializers inside function calls.  We don't detect this
        # perfectly: we just don't complain if the last non-whitespace character on
        # the previous non-blank line is ',', ';', ':', '(', '{', or '}', or if the
        # previous line starts a preprocessor block.
        prevline = GetPreviousNonBlankLine(clean_lines, linenum)[0]
        if (not Search(r'[,;:}{(]\s*$', prevline) and
                not Match(r'\s*#', prevline)):
            error(filename, linenum, 'whitespace/braces', 4,
                  '{ should almost always be at the end of the previous line')

    # An else clause should be on the same line as the preceding closing brace.
    if Match(r'\s*else\b\s*(?:if\b|\{|$)', line):
        prevline = GetPreviousNonBlankLine(clean_lines, linenum)[0]
        if Match(r'\s*}\s*$', prevline):
            error(filename, linenum, 'whitespace/newline', 4,
                  'An else should appear on the same line as the preceding }')

    # If braces come on one side of an else, they should be on both.
    # However, we have to worry about "else if" that spans multiple lines!
    if Search(r'else if\s*\(', line):  # could be multi-line if
        brace_on_left = bool(Search(r'}\s*else if\s*\(', line))
        # find the ( after the if
        pos = line.find('else if')
        pos = line.find('(', pos)
        if pos > 0:
            (endline, _, endpos) = CloseExpression(clean_lines, linenum, pos)
            brace_on_right = endline[endpos:].find('{') != -1
            if brace_on_left != brace_on_right:  # must be brace after if
                error(
                    filename, linenum, 'readability/braces', 5,
                    'If an else has a brace on one side, it should have it on both'
                )
    elif Search(r'}\s*else[^{]*$', line) or Match(r'[^}]*else\s*{', line):
Z
zhangjinchao01 已提交
3971 3972
        error(filename, linenum, 'readability/braces', 5,
              'If an else has a brace on one side, it should have it on both')
Y
Yu Yang 已提交
3973 3974 3975 3976 3977 3978 3979 3980 3981 3982 3983 3984 3985 3986 3987 3988 3989 3990 3991 3992 3993 3994 3995 3996 3997 3998 3999 4000 4001 4002 4003 4004 4005 4006 4007 4008 4009 4010 4011 4012 4013 4014 4015 4016 4017 4018 4019 4020 4021 4022 4023 4024 4025 4026 4027 4028 4029 4030 4031 4032 4033 4034 4035 4036 4037 4038 4039 4040 4041 4042 4043 4044

    # Likewise, an else should never have the else clause on the same line
    if Search(r'\belse [^\s{]', line) and not Search(r'\belse if\b', line):
        error(filename, linenum, 'whitespace/newline', 4,
              'Else clause should never be on same line as else (use 2 lines)')

    # In the same way, a do/while should never be on one line
    if Match(r'\s*do [^\s{]', line):
        error(filename, linenum, 'whitespace/newline', 4,
              'do/while clauses should not be on a single line')

    # Check single-line if/else bodies. The style guide says 'curly braces are not
    # required for single-line statements'. We additionally allow multi-line,
    # single statements, but we reject anything with more than one semicolon in
    # it. This means that the first semicolon after the if should be at the end of
    # its line, and the line after that should have an indent level equal to or
    # lower than the if. We also check for ambiguous if/else nesting without
    # braces.
    if_else_match = Search(r'\b(if\s*\(|else\b)', line)
    if if_else_match and not Match(r'\s*#', line):
        if_indent = GetIndentLevel(line)
        endline, endlinenum, endpos = line, linenum, if_else_match.end()
        if_match = Search(r'\bif\s*\(', line)
        if if_match:
            # This could be a multiline if condition, so find the end first.
            pos = if_match.end() - 1
            (endline, endlinenum, endpos) = CloseExpression(clean_lines,
                                                            linenum, pos)
        # Check for an opening brace, either directly after the if or on the next
        # line. If found, this isn't a single-statement conditional.
        if (not Match(r'\s*{', endline[endpos:]) and
                not (Match(r'\s*$', endline[endpos:]) and endlinenum <
                     (len(clean_lines.elided) - 1) and
                     Match(r'\s*{', clean_lines.elided[endlinenum + 1]))):
            while (endlinenum < len(clean_lines.elided) and
                   ';' not in clean_lines.elided[endlinenum][endpos:]):
                endlinenum += 1
                endpos = 0
            if endlinenum < len(clean_lines.elided):
                endline = clean_lines.elided[endlinenum]
                # We allow a mix of whitespace and closing braces (e.g. for one-liner
                # methods) and a single \ after the semicolon (for macros)
                endpos = endline.find(';')
                if not Match(r';[\s}]*(\\?)$', endline[endpos:]):
                    # Semicolon isn't the last character, there's something trailing.
                    # Output a warning if the semicolon is not contained inside
                    # a lambda expression.
                    if not Match(
                            r'^[^{};]*\[[^\[\]]*\][^{}]*\{[^{}]*\}\s*\)*[;,]\s*$',
                            endline):
                        error(
                            filename, linenum, 'readability/braces', 4,
                            'If/else bodies with multiple statements require braces'
                        )
                elif endlinenum < len(clean_lines.elided) - 1:
                    # Make sure the next line is dedented
                    next_line = clean_lines.elided[endlinenum + 1]
                    next_indent = GetIndentLevel(next_line)
                    # With ambiguous nested if statements, this will error out on the
                    # if that *doesn't* match the else, regardless of whether it's the
                    # inner one or outer one.
                    if (if_match and Match(r'\s*else\b', next_line) and
                            next_indent != if_indent):
                        error(
                            filename, linenum, 'readability/braces', 4,
                            'Else clause should be indented at the same level as if. '
                            'Ambiguous nested if/else chains require braces.')
                    elif next_indent > if_indent:
                        error(
                            filename, linenum, 'readability/braces', 4,
                            'If/else bodies with multiple statements require braces'
                        )
Z
zhangjinchao01 已提交
4045 4046 4047


def CheckTrailingSemicolon(filename, clean_lines, linenum, error):
Y
Yu Yang 已提交
4048
    """Looks for redundant trailing semicolon.
Z
zhangjinchao01 已提交
4049 4050 4051 4052 4053 4054 4055 4056

  Args:
    filename: The name of the current file.
    clean_lines: A CleansedLines instance containing the file.
    linenum: The number of the line to check.
    error: The function to call with any errors found.
  """

Y
Yu Yang 已提交
4057 4058 4059 4060 4061 4062 4063 4064 4065 4066 4067 4068 4069 4070
    line = clean_lines.elided[linenum]

    # Block bodies should not be followed by a semicolon.  Due to C++11
    # brace initialization, there are more places where semicolons are
    # required than not, so we use a whitelist approach to check these
    # rather than a blacklist.  These are the places where "};" should
    # be replaced by just "}":
    # 1. Some flavor of block following closing parenthesis:
    #    for (;;) {};
    #    while (...) {};
    #    switch (...) {};
    #    Function(...) {};
    #    if (...) {};
    #    if (...) else if (...) {};
Z
zhangjinchao01 已提交
4071
    #
Y
Yu Yang 已提交
4072 4073
    # 2. else block:
    #    if (...) else {};
Z
zhangjinchao01 已提交
4074
    #
Y
Yu Yang 已提交
4075 4076
    # 3. const member function:
    #    Function(...) const {};
Z
zhangjinchao01 已提交
4077
    #
Y
Yu Yang 已提交
4078 4079 4080 4081 4082 4083 4084 4085 4086 4087 4088 4089 4090 4091 4092 4093 4094 4095 4096 4097 4098 4099 4100 4101 4102 4103 4104 4105 4106 4107 4108 4109 4110 4111 4112 4113 4114 4115 4116 4117 4118 4119 4120 4121 4122 4123 4124 4125 4126 4127 4128 4129 4130 4131 4132 4133 4134 4135 4136 4137 4138 4139 4140 4141 4142 4143 4144 4145 4146 4147 4148 4149 4150 4151 4152 4153 4154 4155 4156 4157 4158 4159 4160 4161 4162 4163 4164 4165 4166 4167 4168 4169 4170 4171 4172 4173 4174 4175 4176 4177 4178 4179
    # 4. Block following some statement:
    #    x = 42;
    #    {};
    #
    # 5. Block at the beginning of a function:
    #    Function(...) {
    #      {};
    #    }
    #
    #    Note that naively checking for the preceding "{" will also match
    #    braces inside multi-dimensional arrays, but this is fine since
    #    that expression will not contain semicolons.
    #
    # 6. Block following another block:
    #    while (true) {}
    #    {};
    #
    # 7. End of namespaces:
    #    namespace {};
    #
    #    These semicolons seems far more common than other kinds of
    #    redundant semicolons, possibly due to people converting classes
    #    to namespaces.  For now we do not warn for this case.
    #
    # Try matching case 1 first.
    match = Match(r'^(.*\)\s*)\{', line)
    if match:
        # Matched closing parenthesis (case 1).  Check the token before the
        # matching opening parenthesis, and don't warn if it looks like a
        # macro.  This avoids these false positives:
        #  - macro that defines a base class
        #  - multi-line macro that defines a base class
        #  - macro that defines the whole class-head
        #
        # But we still issue warnings for macros that we know are safe to
        # warn, specifically:
        #  - TEST, TEST_F, TEST_P, MATCHER, MATCHER_P
        #  - TYPED_TEST
        #  - INTERFACE_DEF
        #  - EXCLUSIVE_LOCKS_REQUIRED, SHARED_LOCKS_REQUIRED, LOCKS_EXCLUDED:
        #
        # We implement a whitelist of safe macros instead of a blacklist of
        # unsafe macros, even though the latter appears less frequently in
        # google code and would have been easier to implement.  This is because
        # the downside for getting the whitelist wrong means some extra
        # semicolons, while the downside for getting the blacklist wrong
        # would result in compile errors.
        #
        # In addition to macros, we also don't want to warn on
        #  - Compound literals
        #  - Lambdas
        #  - alignas specifier with anonymous structs:
        closing_brace_pos = match.group(1).rfind(')')
        opening_parenthesis = ReverseCloseExpression(clean_lines, linenum,
                                                     closing_brace_pos)
        if opening_parenthesis[2] > -1:
            line_prefix = opening_parenthesis[0][0:opening_parenthesis[2]]
            macro = Search(r'\b([A-Z_]+)\s*$', line_prefix)
            func = Match(r'^(.*\])\s*$', line_prefix)
            if ((macro and macro.group(1) not in
                 ('TEST', 'TEST_F', 'MATCHER', 'MATCHER_P', 'TYPED_TEST',
                  'EXCLUSIVE_LOCKS_REQUIRED', 'SHARED_LOCKS_REQUIRED',
                  'LOCKS_EXCLUDED', 'INTERFACE_DEF')) or
                (func and not Search(r'\boperator\s*\[\s*\]', func.group(1))) or
                    Search(r'\b(?:struct|union)\s+alignas\s*$', line_prefix) or
                    Search(r'\s+=\s*$', line_prefix)):
                match = None
        if (match and opening_parenthesis[1] > 1 and Search(
                r'\]\s*$', clean_lines.elided[opening_parenthesis[1] - 1])):
            # Multi-line lambda-expression
            match = None

    else:
        # Try matching cases 2-3.
        match = Match(r'^(.*(?:else|\)\s*const)\s*)\{', line)
        if not match:
            # Try matching cases 4-6.  These are always matched on separate lines.
            #
            # Note that we can't simply concatenate the previous line to the
            # current line and do a single match, otherwise we may output
            # duplicate warnings for the blank line case:
            #   if (cond) {
            #     // blank line
            #   }
            prevline = GetPreviousNonBlankLine(clean_lines, linenum)[0]
            if prevline and Search(r'[;{}]\s*$', prevline):
                match = Match(r'^(\s*)\{', line)

    # Check matching closing brace
    if match:
        (endline, endlinenum, endpos) = CloseExpression(clean_lines, linenum,
                                                        len(match.group(1)))
        if endpos > -1 and Match(r'^\s*;', endline[endpos:]):
            # Current {} pair is eligible for semicolon check, and we have found
            # the redundant semicolon, output warning here.
            #
            # Note: because we are scanning forward for opening braces, and
            # outputting warnings for the matching closing brace, if there are
            # nested blocks with trailing semicolons, we will get the error
            # messages in reversed order.
            error(filename, endlinenum, 'readability/braces', 4,
                  "You don't need a ; after a }")
Z
zhangjinchao01 已提交
4180 4181 4182


def CheckEmptyBlockBody(filename, clean_lines, linenum, error):
Y
Yu Yang 已提交
4183
    """Look for empty loop/conditional body with only a single semicolon.
Z
zhangjinchao01 已提交
4184 4185 4186 4187 4188 4189 4190 4191

  Args:
    filename: The name of the current file.
    clean_lines: A CleansedLines instance containing the file.
    linenum: The number of the line to check.
    error: The function to call with any errors found.
  """

Y
Yu Yang 已提交
4192 4193 4194 4195 4196 4197 4198 4199 4200 4201 4202 4203 4204 4205 4206 4207 4208 4209 4210 4211 4212 4213 4214 4215
    # Search for loop keywords at the beginning of the line.  Because only
    # whitespaces are allowed before the keywords, this will also ignore most
    # do-while-loops, since those lines should start with closing brace.
    #
    # We also check "if" blocks here, since an empty conditional block
    # is likely an error.
    line = clean_lines.elided[linenum]
    matched = Match(r'\s*(for|while|if)\s*\(', line)
    if matched:
        # Find the end of the conditional expression
        (end_line, end_linenum, end_pos) = CloseExpression(clean_lines, linenum,
                                                           line.find('('))

        # Output warning if what follows the condition expression is a semicolon.
        # No warning for all other cases, including whitespace or newline, since we
        # have a separate check for semicolons preceded by whitespace.
        if end_pos >= 0 and Match(r';', end_line[end_pos:]):
            if matched.group(1) == 'if':
                error(filename, end_linenum,
                      'whitespace/empty_conditional_body', 5,
                      'Empty conditional bodies should use {}')
            else:
                error(filename, end_linenum, 'whitespace/empty_loop_body', 5,
                      'Empty loop bodies should use {} or continue')
Z
zhangjinchao01 已提交
4216 4217 4218


def FindCheckMacro(line):
Y
Yu Yang 已提交
4219
    """Find a replaceable CHECK-like macro.
Z
zhangjinchao01 已提交
4220 4221 4222 4223 4224 4225 4226

  Args:
    line: line to search on.
  Returns:
    (macro name, start position), or (None, -1) if no replaceable
    macro is found.
  """
Y
Yu Yang 已提交
4227 4228 4229 4230 4231 4232 4233 4234 4235 4236 4237 4238
    for macro in _CHECK_MACROS:
        i = line.find(macro)
        if i >= 0:
            # Find opening parenthesis.  Do a regular expression match here
            # to make sure that we are matching the expected CHECK macro, as
            # opposed to some other macro that happens to contain the CHECK
            # substring.
            matched = Match(r'^(.*\b' + macro + r'\s*)\(', line)
            if not matched:
                continue
            return (macro, len(matched.group(1)))
    return (None, -1)
Z
zhangjinchao01 已提交
4239 4240 4241


def CheckCheck(filename, clean_lines, linenum, error):
Y
Yu Yang 已提交
4242
    """Checks the use of CHECK and EXPECT macros.
Z
zhangjinchao01 已提交
4243 4244 4245 4246 4247 4248 4249 4250

  Args:
    filename: The name of the current file.
    clean_lines: A CleansedLines instance containing the file.
    linenum: The number of the line to check.
    error: The function to call with any errors found.
  """

Y
Yu Yang 已提交
4251 4252 4253 4254
    # Decide the set of replacement macros that should be suggested
    lines = clean_lines.elided
    (check_macro, start_pos) = FindCheckMacro(lines[linenum])
    if not check_macro:
Z
zhangjinchao01 已提交
4255
        return
Y
Yu Yang 已提交
4256 4257 4258 4259 4260 4261 4262 4263 4264 4265 4266 4267 4268 4269 4270

    # Find end of the boolean expression by matching parentheses
    (last_line, end_line, end_pos) = CloseExpression(clean_lines, linenum,
                                                     start_pos)
    if end_pos < 0:
        return

    # If the check macro is followed by something other than a
    # semicolon, assume users will log their own custom error messages
    # and don't suggest any replacements.
    if not Match(r'\s*;', last_line[end_pos:]):
        return

    if linenum == end_line:
        expression = lines[linenum][start_pos + 1:end_pos - 1]
Z
zhangjinchao01 已提交
4271
    else:
Y
Yu Yang 已提交
4272 4273 4274 4275 4276 4277 4278 4279 4280 4281 4282 4283 4284 4285 4286 4287 4288 4289 4290 4291 4292 4293 4294 4295 4296 4297 4298 4299 4300 4301 4302 4303 4304 4305 4306 4307 4308 4309 4310 4311 4312 4313 4314 4315 4316 4317 4318 4319 4320 4321 4322 4323 4324 4325 4326 4327 4328 4329 4330 4331 4332 4333 4334 4335 4336 4337
        expression = lines[linenum][start_pos + 1:]
        for i in xrange(linenum + 1, end_line):
            expression += lines[i]
        expression += last_line[0:end_pos - 1]

    # Parse expression so that we can take parentheses into account.
    # This avoids false positives for inputs like "CHECK((a < 4) == b)",
    # which is not replaceable by CHECK_LE.
    lhs = ''
    rhs = ''
    operator = None
    while expression:
        matched = Match(r'^\s*(<<|<<=|>>|>>=|->\*|->|&&|\|\||'
                        r'==|!=|>=|>|<=|<|\()(.*)$', expression)
        if matched:
            token = matched.group(1)
            if token == '(':
                # Parenthesized operand
                expression = matched.group(2)
                (end, _) = FindEndOfExpressionInLine(expression, 0, ['('])
                if end < 0:
                    return  # Unmatched parenthesis
                lhs += '(' + expression[0:end]
                expression = expression[end:]
            elif token in ('&&', '||'):
                # Logical and/or operators.  This means the expression
                # contains more than one term, for example:
                #   CHECK(42 < a && a < b);
                #
                # These are not replaceable with CHECK_LE, so bail out early.
                return
            elif token in ('<<', '<<=', '>>', '>>=', '->*', '->'):
                # Non-relational operator
                lhs += token
                expression = matched.group(2)
            else:
                # Relational operator
                operator = token
                rhs = matched.group(2)
                break
        else:
            # Unparenthesized operand.  Instead of appending to lhs one character
            # at a time, we do another regular expression match to consume several
            # characters at once if possible.  Trivial benchmark shows that this
            # is more efficient when the operands are longer than a single
            # character, which is generally the case.
            matched = Match(r'^([^-=!<>()&|]+)(.*)$', expression)
            if not matched:
                matched = Match(r'^(\s*\S)(.*)$', expression)
                if not matched:
                    break
            lhs += matched.group(1)
            expression = matched.group(2)

    # Only apply checks if we got all parts of the boolean expression
    if not (lhs and operator and rhs):
        return

    # Check that rhs do not contain logical operators.  We already know
    # that lhs is fine since the loop above parses out && and ||.
    if rhs.find('&&') > -1 or rhs.find('||') > -1:
        return

    # At least one of the operands must be a constant literal.  This is
    # to avoid suggesting replacements for unprintable things like
    # CHECK(variable != iterator)
Z
zhangjinchao01 已提交
4338
    #
Y
Yu Yang 已提交
4339 4340 4341 4342 4343 4344 4345 4346 4347 4348 4349 4350 4351 4352 4353 4354 4355 4356
    # The following pattern matches decimal, hex integers, strings, and
    # characters (in that order).
    lhs = lhs.strip()
    rhs = rhs.strip()
    match_constant = r'^([-+]?(\d+|0[xX][0-9a-fA-F]+)[lLuU]{0,3}|".*"|\'.*\')$'
    if Match(match_constant, lhs) or Match(match_constant, rhs):
        # Note: since we know both lhs and rhs, we can provide a more
        # descriptive error message like:
        #   Consider using CHECK_EQ(x, 42) instead of CHECK(x == 42)
        # Instead of:
        #   Consider using CHECK_EQ instead of CHECK(a == b)
        #
        # We are still keeping the less descriptive message because if lhs
        # or rhs gets long, the error message might become unreadable.
        error(filename, linenum, 'readability/check', 2,
              'Consider using %s instead of %s(a %s b)' %
              (_CHECK_REPLACEMENT[check_macro][operator], check_macro,
               operator))
Z
zhangjinchao01 已提交
4357 4358 4359


def CheckAltTokens(filename, clean_lines, linenum, error):
Y
Yu Yang 已提交
4360
    """Check alternative keywords being used in boolean expressions.
Z
zhangjinchao01 已提交
4361 4362 4363 4364 4365 4366 4367

  Args:
    filename: The name of the current file.
    clean_lines: A CleansedLines instance containing the file.
    linenum: The number of the line to check.
    error: The function to call with any errors found.
  """
Y
Yu Yang 已提交
4368
    line = clean_lines.elided[linenum]
Z
zhangjinchao01 已提交
4369

Y
Yu Yang 已提交
4370 4371 4372
    # Avoid preprocessor lines
    if Match(r'^\s*#', line):
        return
Z
zhangjinchao01 已提交
4373

Y
Yu Yang 已提交
4374 4375 4376 4377 4378 4379 4380 4381 4382 4383
    # Last ditch effort to avoid multi-line comments.  This will not help
    # if the comment started before the current line or ended after the
    # current line, but it catches most of the false positives.  At least,
    # it provides a way to workaround this warning for people who use
    # multi-line comments in preprocessor macros.
    #
    # TODO(unknown): remove this once cpplint has better support for
    # multi-line comments.
    if line.find('/*') >= 0 or line.find('*/') >= 0:
        return
Z
zhangjinchao01 已提交
4384

Y
Yu Yang 已提交
4385 4386 4387 4388
    for match in _ALT_TOKEN_REPLACEMENT_PATTERN.finditer(line):
        error(filename, linenum, 'readability/alt_tokens', 2,
              'Use operator %s instead of %s' % (
                  _ALT_TOKEN_REPLACEMENT[match.group(1)], match.group(1)))
Z
zhangjinchao01 已提交
4389 4390 4391


def GetLineWidth(line):
Y
Yu Yang 已提交
4392
    """Determines the width of the line in column positions.
Z
zhangjinchao01 已提交
4393 4394 4395 4396 4397 4398 4399 4400

  Args:
    line: A string, which may be a Unicode string.

  Returns:
    The width of the line in column positions, accounting for Unicode
    combining characters and wide characters.
  """
Y
Yu Yang 已提交
4401 4402 4403 4404 4405 4406 4407 4408 4409 4410
    if isinstance(line, unicode):
        width = 0
        for uc in unicodedata.normalize('NFC', line):
            if unicodedata.east_asian_width(uc) in ('W', 'F'):
                width += 2
            elif not unicodedata.combining(uc):
                width += 1
        return width
    else:
        return len(line)
Z
zhangjinchao01 已提交
4411 4412 4413 4414


def CheckStyle(filename, clean_lines, linenum, file_extension, nesting_state,
               error):
Y
Yu Yang 已提交
4415
    """Checks rules from the 'C++ style rules' section of cppguide.html.
Z
zhangjinchao01 已提交
4416 4417 4418 4419 4420 4421 4422 4423 4424 4425 4426 4427 4428 4429 4430

  Most of these rules are hard to test (naming, comment style), but we
  do what we can.  In particular we check for 2-space indents, line lengths,
  tab usage, spaces inside code, etc.

  Args:
    filename: The name of the current file.
    clean_lines: A CleansedLines instance containing the file.
    linenum: The number of the line to check.
    file_extension: The extension (without the dot) of the filename.
    nesting_state: A NestingState instance which maintains information about
                   the current stack of nested blocks being parsed.
    error: The function to call with any errors found.
  """

Y
Yu Yang 已提交
4431 4432 4433 4434 4435 4436 4437 4438 4439 4440 4441 4442 4443 4444 4445 4446 4447 4448 4449 4450 4451 4452 4453 4454 4455 4456 4457 4458 4459 4460 4461 4462 4463 4464 4465 4466 4467 4468 4469 4470 4471 4472 4473 4474 4475 4476 4477 4478 4479 4480 4481 4482 4483 4484 4485 4486 4487 4488 4489 4490 4491 4492 4493 4494 4495 4496 4497 4498 4499 4500 4501 4502 4503 4504 4505 4506 4507 4508 4509 4510 4511 4512 4513 4514 4515 4516 4517 4518 4519 4520 4521 4522 4523 4524 4525 4526 4527 4528 4529
    # Don't use "elided" lines here, otherwise we can't check commented lines.
    # Don't want to use "raw" either, because we don't want to check inside C++11
    # raw strings,
    raw_lines = clean_lines.lines_without_raw_strings
    line = raw_lines[linenum]

    if line.find('\t') != -1:
        error(filename, linenum, 'whitespace/tab', 1,
              'Tab found; better to use spaces')

    # One or three blank spaces at the beginning of the line is weird; it's
    # hard to reconcile that with 2-space indents.
    # NOTE: here are the conditions rob pike used for his tests.  Mine aren't
    # as sophisticated, but it may be worth becoming so:  RLENGTH==initial_spaces
    # if(RLENGTH > 20) complain = 0;
    # if(match($0, " +(error|private|public|protected):")) complain = 0;
    # if(match(prev, "&& *$")) complain = 0;
    # if(match(prev, "\\|\\| *$")) complain = 0;
    # if(match(prev, "[\",=><] *$")) complain = 0;
    # if(match($0, " <<")) complain = 0;
    # if(match(prev, " +for \\(")) complain = 0;
    # if(prevodd && match(prevprev, " +for \\(")) complain = 0;
    scope_or_label_pattern = r'\s*\w+\s*:\s*\\?$'
    classinfo = nesting_state.InnermostClass()
    initial_spaces = 0
    cleansed_line = clean_lines.elided[linenum]
    while initial_spaces < len(line) and line[initial_spaces] == ' ':
        initial_spaces += 1
    if line and line[-1].isspace():
        error(filename, linenum, 'whitespace/end_of_line', 4,
              'Line ends in whitespace.  Consider deleting these extra spaces.')
    # There are certain situations we allow one space, notably for
    # section labels, and also lines containing multi-line raw strings.
    elif ((initial_spaces == 1 or initial_spaces == 3) and
          not Match(scope_or_label_pattern, cleansed_line) and
          not (clean_lines.raw_lines[linenum] != line and
               Match(r'^\s*""', line))):
        error(filename, linenum, 'whitespace/indent', 3,
              'Weird number of spaces at line-start.  '
              'Are you using a 2-space indent?')

    # Check if the line is a header guard.
    is_header_guard = False
    if file_extension == 'h':
        cppvar = GetHeaderGuardCPPVariable(filename)
        if (line.startswith('#ifndef %s' % cppvar) or
                line.startswith('#define %s' % cppvar) or
                line.startswith('#endif  // %s' % cppvar)):
            is_header_guard = True
    # #include lines and header guards can be long, since there's no clean way to
    # split them.
    #
    # URLs can be long too.  It's possible to split these, but it makes them
    # harder to cut&paste.
    #
    # The "$Id:...$" comment may also get very long without it being the
    # developers fault.
    if (not line.startswith('#include') and not is_header_guard and
            not Match(r'^\s*//.*http(s?)://\S*$', line) and
            not Match(r'^// \$Id:.*#[0-9]+ \$$', line)):
        line_width = GetLineWidth(line)
        extended_length = int((_line_length * 1.25))
        if line_width > extended_length:
            error(filename, linenum, 'whitespace/line_length', 4,
                  'Lines should very rarely be longer than %i characters' %
                  extended_length)
        elif line_width > _line_length:
            error(filename, linenum, 'whitespace/line_length', 2,
                  'Lines should be <= %i characters long' % _line_length)

    if (cleansed_line.count(';') > 1 and
            # for loops are allowed two ;'s (and may run over two lines).
            cleansed_line.find('for') == -1 and
        (GetPreviousNonBlankLine(clean_lines, linenum)[0].find('for') == -1 or
         GetPreviousNonBlankLine(clean_lines, linenum)[0].find(';') != -1) and
            # It's ok to have many commands in a switch case that fits in 1 line
            not ((cleansed_line.find('case ') != -1 or
                  cleansed_line.find('default:') != -1) and
                 cleansed_line.find('break;') != -1)):
        error(filename, linenum, 'whitespace/newline', 0,
              'More than one command on the same line')

    # Some more style checks
    CheckBraces(filename, clean_lines, linenum, error)
    CheckTrailingSemicolon(filename, clean_lines, linenum, error)
    CheckEmptyBlockBody(filename, clean_lines, linenum, error)
    CheckAccess(filename, clean_lines, linenum, nesting_state, error)
    CheckSpacing(filename, clean_lines, linenum, nesting_state, error)
    CheckOperatorSpacing(filename, clean_lines, linenum, error)
    CheckParenthesisSpacing(filename, clean_lines, linenum, error)
    CheckCommaSpacing(filename, clean_lines, linenum, error)
    CheckBracesSpacing(filename, clean_lines, linenum, error)
    CheckSpacingForFunctionCall(filename, clean_lines, linenum, error)
    CheckRValueReference(filename, clean_lines, linenum, nesting_state, error)
    CheckCheck(filename, clean_lines, linenum, error)
    CheckAltTokens(filename, clean_lines, linenum, error)
    classinfo = nesting_state.InnermostClass()
    if classinfo:
        CheckSectionSpacing(filename, clean_lines, classinfo, linenum, error)
Z
zhangjinchao01 已提交
4530 4531 4532 4533 4534 4535 4536 4537 4538 4539 4540 4541


_RE_PATTERN_INCLUDE = re.compile(r'^\s*#\s*include\s*([<"])([^>"]*)[>"].*$')
# Matches the first component of a filename delimited by -s and _s. That is:
#  _RE_FIRST_COMPONENT.match('foo').group(0) == 'foo'
#  _RE_FIRST_COMPONENT.match('foo.cc').group(0) == 'foo'
#  _RE_FIRST_COMPONENT.match('foo-bar_baz.cc').group(0) == 'foo'
#  _RE_FIRST_COMPONENT.match('foo_bar-baz.cc').group(0) == 'foo'
_RE_FIRST_COMPONENT = re.compile(r'^[^-_.]+')


def _DropCommonSuffixes(filename):
Y
Yu Yang 已提交
4542
    """Drops common suffixes like _test.cc or -inl.h from filename.
Z
zhangjinchao01 已提交
4543 4544 4545 4546 4547 4548 4549 4550 4551 4552 4553 4554 4555 4556 4557 4558 4559

  For example:
    >>> _DropCommonSuffixes('foo/foo-inl.h')
    'foo/foo'
    >>> _DropCommonSuffixes('foo/bar/foo.cc')
    'foo/bar/foo'
    >>> _DropCommonSuffixes('foo/foo_internal.h')
    'foo/foo'
    >>> _DropCommonSuffixes('foo/foo_unusualinternal.h')
    'foo/foo_unusualinternal'

  Args:
    filename: The input filename.

  Returns:
    The filename with the common suffix removed.
  """
Y
Yu Yang 已提交
4560 4561 4562 4563 4564 4565
    for suffix in ('test.cc', 'regtest.cc', 'unittest.cc', 'inl.h', 'impl.h',
                   'internal.h'):
        if (filename.endswith(suffix) and len(filename) > len(suffix) and
                filename[-len(suffix) - 1] in ('-', '_')):
            return filename[:-len(suffix) - 1]
    return os.path.splitext(filename)[0]
Z
zhangjinchao01 已提交
4566 4567 4568


def _IsTestFilename(filename):
Y
Yu Yang 已提交
4569
    """Determines if the given filename has a suffix that identifies it as a test.
Z
zhangjinchao01 已提交
4570 4571 4572 4573 4574 4575 4576

  Args:
    filename: The input filename.

  Returns:
    True if 'filename' looks like a test, False otherwise.
  """
Y
Yu Yang 已提交
4577 4578 4579 4580 4581
    if (filename.endswith('_test.cc') or filename.endswith('_unittest.cc') or
            filename.endswith('_regtest.cc')):
        return True
    else:
        return False
Z
zhangjinchao01 已提交
4582 4583 4584


def _ClassifyInclude(fileinfo, include, is_system):
Y
Yu Yang 已提交
4585
    """Figures out what kind of header 'include' is.
Z
zhangjinchao01 已提交
4586 4587 4588 4589 4590 4591 4592 4593 4594 4595 4596 4597 4598 4599 4600 4601 4602 4603 4604 4605 4606 4607

  Args:
    fileinfo: The current file cpplint is running over. A FileInfo instance.
    include: The path to a #included file.
    is_system: True if the #include used <> rather than "".

  Returns:
    One of the _XXX_HEADER constants.

  For example:
    >>> _ClassifyInclude(FileInfo('foo/foo.cc'), 'stdio.h', True)
    _C_SYS_HEADER
    >>> _ClassifyInclude(FileInfo('foo/foo.cc'), 'string', True)
    _CPP_SYS_HEADER
    >>> _ClassifyInclude(FileInfo('foo/foo.cc'), 'foo/foo.h', False)
    _LIKELY_MY_HEADER
    >>> _ClassifyInclude(FileInfo('foo/foo_unknown_extension.cc'),
    ...                  'bar/foo_other_ext.h', False)
    _POSSIBLE_MY_HEADER
    >>> _ClassifyInclude(FileInfo('foo/foo.cc'), 'foo/bar.h', False)
    _OTHER_HEADER
  """
Y
Yu Yang 已提交
4608 4609 4610
    # This is a list of all standard c++ header files, except
    # those already checked for above.
    is_cpp_h = include in _CPP_HEADERS
Z
zhangjinchao01 已提交
4611

Y
Yu Yang 已提交
4612 4613 4614 4615 4616 4617 4618 4619 4620 4621 4622 4623 4624 4625 4626 4627 4628 4629 4630 4631 4632 4633 4634 4635 4636 4637 4638 4639 4640
    if is_system:
        if is_cpp_h:
            return _CPP_SYS_HEADER
        else:
            return _C_SYS_HEADER

    # If the target file and the include we're checking share a
    # basename when we drop common extensions, and the include
    # lives in . , then it's likely to be owned by the target file.
    target_dir, target_base = (
        os.path.split(_DropCommonSuffixes(fileinfo.RepositoryName())))
    include_dir, include_base = os.path.split(_DropCommonSuffixes(include))
    if target_base == include_base and (
            include_dir == target_dir or
            include_dir == os.path.normpath(target_dir + '/../public')):
        return _LIKELY_MY_HEADER

    # If the target and include share some initial basename
    # component, it's possible the target is implementing the
    # include, so it's allowed to be first, but we'll never
    # complain if it's not there.
    target_first_component = _RE_FIRST_COMPONENT.match(target_base)
    include_first_component = _RE_FIRST_COMPONENT.match(include_base)
    if (target_first_component and include_first_component and
            target_first_component.group(0) ==
            include_first_component.group(0)):
        return _POSSIBLE_MY_HEADER

    return _OTHER_HEADER
Z
zhangjinchao01 已提交
4641 4642 4643


def CheckIncludeLine(filename, clean_lines, linenum, include_state, error):
Y
Yu Yang 已提交
4644
    """Check rules that are applicable to #include lines.
Z
zhangjinchao01 已提交
4645 4646 4647 4648 4649 4650 4651 4652 4653 4654 4655 4656

  Strings on #include lines are NOT removed from elided line, to make
  certain tasks easier. However, to prevent false positives, checks
  applicable to #include lines in CheckLanguage must be put here.

  Args:
    filename: The name of the current file.
    clean_lines: A CleansedLines instance containing the file.
    linenum: The number of the line to check.
    include_state: An _IncludeState instance in which the headers are inserted.
    error: The function to call with any errors found.
  """
Y
Yu Yang 已提交
4657 4658
    fileinfo = FileInfo(filename)
    line = clean_lines.lines[linenum]
Z
zhangjinchao01 已提交
4659

Y
Yu Yang 已提交
4660 4661 4662 4663 4664 4665 4666 4667 4668 4669 4670 4671 4672 4673 4674 4675 4676 4677 4678 4679 4680 4681 4682 4683 4684 4685 4686 4687 4688 4689 4690 4691 4692 4693 4694 4695 4696 4697 4698 4699 4700 4701 4702 4703 4704 4705 4706 4707 4708 4709 4710 4711 4712 4713 4714 4715
    # "include" should use the new style "foo/bar.h" instead of just "bar.h"
    # Only do this check if the included header follows google naming
    # conventions.  If not, assume that it's a 3rd party API that
    # requires special include conventions.
    #
    # We also make an exception for Lua headers, which follow google
    # naming convention but not the include convention.
    match = Match(r'#include\s*"([^/]+\.h)"', line)
    if match and not _THIRD_PARTY_HEADERS_PATTERN.match(match.group(1)):
        error(filename, linenum, 'build/include', 4,
              'Include the directory when naming .h files')

    # we shouldn't include a file more than once. actually, there are a
    # handful of instances where doing so is okay, but in general it's
    # not.
    match = _RE_PATTERN_INCLUDE.search(line)
    if match:
        include = match.group(2)
        is_system = (match.group(1) == '<')
        duplicate_line = include_state.FindHeader(include)
        if duplicate_line >= 0:
            error(filename, linenum, 'build/include', 4,
                  '"%s" already included at %s:%s' %
                  (include, filename, duplicate_line))
        elif (include.endswith('.cc') and
              os.path.dirname(fileinfo.RepositoryName()) !=
              os.path.dirname(include)):
            error(filename, linenum, 'build/include', 4,
                  'Do not include .cc files from other packages')
        elif not _THIRD_PARTY_HEADERS_PATTERN.match(include):
            include_state.include_list[-1].append((include, linenum))

            # We want to ensure that headers appear in the right order:
            # 1) for foo.cc, foo.h  (preferred location)
            # 2) c system files
            # 3) cpp system files
            # 4) for foo.cc, foo.h  (deprecated location)
            # 5) other google headers
            #
            # We classify each include statement as one of those 5 types
            # using a number of techniques. The include_state object keeps
            # track of the highest type seen, and complains if we see a
            # lower type after that.
            error_message = include_state.CheckNextIncludeOrder(
                _ClassifyInclude(fileinfo, include, is_system))
            if error_message:
                error(filename, linenum, 'build/include_order', 4,
                      '%s. Should be: %s.h, c system, c++ system, other.' %
                      (error_message, fileinfo.BaseName()))
            canonical_include = include_state.CanonicalizeAlphabeticalOrder(
                include)
            if not include_state.IsInAlphabeticalOrder(clean_lines, linenum,
                                                       canonical_include):
                error(filename, linenum, 'build/include_alpha', 4,
                      'Include "%s" not in alphabetical order' % include)
            include_state.SetLastHeader(canonical_include)
Z
zhangjinchao01 已提交
4716 4717 4718


def _GetTextInside(text, start_pattern):
Y
Yu Yang 已提交
4719
    r"""Retrieves all the text between matching open and close parentheses.
Z
zhangjinchao01 已提交
4720 4721 4722 4723 4724 4725 4726 4727 4728 4729 4730 4731 4732 4733 4734 4735 4736 4737

  Given a string of lines and a regular expression string, retrieve all the text
  following the expression and between opening punctuation symbols like
  (, [, or {, and the matching close-punctuation symbol. This properly nested
  occurrences of the punctuations, so for the text like
    printf(a(), b(c()));
  a call to _GetTextInside(text, r'printf\(') will return 'a(), b(c())'.
  start_pattern must match string having an open punctuation symbol at the end.

  Args:
    text: The lines to extract text. Its comments and strings must be elided.
           It can be single line and can span multiple lines.
    start_pattern: The regexp string indicating where to start extracting
                   the text.
  Returns:
    The extracted text.
    None if either the opening string or ending punctuation could not be found.
  """
Y
Yu Yang 已提交
4738 4739 4740 4741 4742 4743 4744 4745 4746 4747 4748 4749 4750 4751 4752 4753 4754 4755 4756 4757 4758 4759 4760 4761 4762 4763 4764 4765 4766 4767 4768 4769 4770 4771
    # TODO(unknown): Audit cpplint.py to see what places could be profitably
    # rewritten to use _GetTextInside (and use inferior regexp matching today).

    # Give opening punctuations to get the matching close-punctuations.
    matching_punctuation = {'(': ')', '{': '}', '[': ']'}
    closing_punctuation = set(matching_punctuation.itervalues())

    # Find the position to start extracting text.
    match = re.search(start_pattern, text, re.M)
    if not match:  # start_pattern not found in text.
        return None
    start_position = match.end(0)

    assert start_position > 0, (
        'start_pattern must ends with an opening punctuation.')
    assert text[start_position - 1] in matching_punctuation, (
        'start_pattern must ends with an opening punctuation.')
    # Stack of closing punctuations we expect to have in text after position.
    punctuation_stack = [matching_punctuation[text[start_position - 1]]]
    position = start_position
    while punctuation_stack and position < len(text):
        if text[position] == punctuation_stack[-1]:
            punctuation_stack.pop()
        elif text[position] in closing_punctuation:
            # A closing punctuation without matching opening punctuations.
            return None
        elif text[position] in matching_punctuation:
            punctuation_stack.append(matching_punctuation[text[position]])
        position += 1
    if punctuation_stack:
        # Opening punctuations left without matching close-punctuations.
        return None
    # punctuations match.
    return text[start_position:position - 1]
Z
zhangjinchao01 已提交
4772 4773 4774 4775 4776 4777 4778 4779 4780 4781 4782 4783 4784 4785 4786 4787 4788 4789 4790 4791 4792 4793 4794 4795


# Patterns for matching call-by-reference parameters.
#
# Supports nested templates up to 2 levels deep using this messy pattern:
#   < (?: < (?: < [^<>]*
#               >
#           |   [^<>] )*
#         >
#     |   [^<>] )*
#   >
_RE_PATTERN_IDENT = r'[_a-zA-Z]\w*'  # =~ [[:alpha:]][[:alnum:]]*
_RE_PATTERN_TYPE = (
    r'(?:const\s+)?(?:typename\s+|class\s+|struct\s+|union\s+|enum\s+)?'
    r'(?:\w|'
    r'\s*<(?:<(?:<[^<>]*>|[^<>])*>|[^<>])*>|'
    r'::)+')
# A call-by-reference parameter ends with '& identifier'.
_RE_PATTERN_REF_PARAM = re.compile(
    r'(' + _RE_PATTERN_TYPE + r'(?:\s*(?:\bconst\b|[*]))*\s*'
    r'&\s*' + _RE_PATTERN_IDENT + r')\s*(?:=[^,()]+)?[,)]')
# A call-by-const-reference parameter either ends with 'const& identifier'
# or looks like 'const type& identifier' when 'type' is atomic.
_RE_PATTERN_CONST_REF_PARAM = (
Y
Yu Yang 已提交
4796 4797
    r'(?:.*\s*\bconst\s*&\s*' + _RE_PATTERN_IDENT + r'|const\s+' +
    _RE_PATTERN_TYPE + r'\s*&\s*' + _RE_PATTERN_IDENT + r')')
Z
zhangjinchao01 已提交
4798 4799


Y
Yu Yang 已提交
4800 4801 4802
def CheckLanguage(filename, clean_lines, linenum, file_extension, include_state,
                  nesting_state, error):
    """Checks rules from the 'C++ language rules' section of cppguide.html.
Z
zhangjinchao01 已提交
4803 4804 4805 4806 4807 4808 4809 4810 4811 4812 4813 4814 4815 4816

  Some of these rules are hard to test (function overloading, using
  uint32 inappropriately), but we do the best we can.

  Args:
    filename: The name of the current file.
    clean_lines: A CleansedLines instance containing the file.
    linenum: The number of the line to check.
    file_extension: The extension (without the dot) of the filename.
    include_state: An _IncludeState instance in which the headers are inserted.
    nesting_state: A NestingState instance which maintains information about
                   the current stack of nested blocks being parsed.
    error: The function to call with any errors found.
  """
Y
Yu Yang 已提交
4817 4818 4819 4820 4821
    # If the line is empty or consists of entirely a comment, no need to
    # check it.
    line = clean_lines.elided[linenum]
    if not line:
        return
Z
zhangjinchao01 已提交
4822

Y
Yu Yang 已提交
4823
    match = _RE_PATTERN_INCLUDE.search(line)
Z
zhangjinchao01 已提交
4824
    if match:
Y
Yu Yang 已提交
4825 4826 4827 4828 4829 4830 4831 4832 4833 4834 4835 4836 4837 4838 4839 4840 4841 4842 4843 4844 4845 4846 4847 4848 4849 4850 4851 4852 4853 4854 4855 4856 4857 4858 4859 4860 4861 4862 4863 4864 4865 4866 4867 4868 4869 4870 4871 4872 4873 4874 4875 4876 4877 4878 4879 4880 4881 4882 4883 4884 4885 4886 4887 4888 4889 4890 4891 4892 4893 4894 4895 4896 4897 4898 4899 4900 4901 4902 4903 4904 4905 4906 4907 4908 4909 4910 4911 4912 4913 4914 4915 4916
        CheckIncludeLine(filename, clean_lines, linenum, include_state, error)
        return

    # Reset include state across preprocessor directives.  This is meant
    # to silence warnings for conditional includes.
    match = Match(r'^\s*#\s*(if|ifdef|ifndef|elif|else|endif)\b', line)
    if match:
        include_state.ResetSection(match.group(1))

    # Make Windows paths like Unix.
    fullname = os.path.abspath(filename).replace('\\', '/')

    # Perform other checks now that we are sure that this is not an include line
    CheckCasts(filename, clean_lines, linenum, error)
    CheckGlobalStatic(filename, clean_lines, linenum, error)
    CheckPrintf(filename, clean_lines, linenum, error)

    if file_extension == 'h':
        # TODO(unknown): check that 1-arg constructors are explicit.
        #                How to tell it's a constructor?
        #                (handled in CheckForNonStandardConstructs for now)
        # TODO(unknown): check that classes declare or disable copy/assign
        #                (level 1 error)
        pass

    # Check if people are using the verboten C basic types.  The only exception
    # we regularly allow is "unsigned short port" for port.
    if Search(r'\bshort port\b', line):
        if not Search(r'\bunsigned short port\b', line):
            error(filename, linenum, 'runtime/int', 4,
                  'Use "unsigned short" for ports, not "short"')
    else:
        match = Search(r'\b(short|long(?! +double)|long long)\b', line)
        if match:
            error(filename, linenum, 'runtime/int', 4,
                  'Use int16/int64/etc, rather than the C type %s' %
                  match.group(1))

    # Check if some verboten operator overloading is going on
    # TODO(unknown): catch out-of-line unary operator&:
    #   class X {};
    #   int operator&(const X& x) { return 42; }  // unary operator&
    # The trick is it's hard to tell apart from binary operator&:
    #   class Y { int operator&(const Y& x) { return 23; } }; // binary operator&
    if Search(r'\boperator\s*&\s*\(\s*\)', line):
        error(filename, linenum, 'runtime/operator', 4,
              'Unary operator& is dangerous.  Do not use it.')

    # Check for suspicious usage of "if" like
    # } if (a == b) {
    if Search(r'\}\s*if\s*\(', line):
        error(filename, linenum, 'readability/braces', 4,
              'Did you mean "else if"? If not, start a new line for "if".')

    # Check for potential format string bugs like printf(foo).
    # We constrain the pattern not to pick things like DocidForPrintf(foo).
    # Not perfect but it can catch printf(foo.c_str()) and printf(foo->c_str())
    # TODO(unknown): Catch the following case. Need to change the calling
    # convention of the whole function to process multiple line to handle it.
    #   printf(
    #       boy_this_is_a_really_long_variable_that_cannot_fit_on_the_prev_line);
    printf_args = _GetTextInside(line, r'(?i)\b(string)?printf\s*\(')
    if printf_args:
        match = Match(r'([\w.\->()]+)$', printf_args)
        if match and match.group(1) != '__VA_ARGS__':
            function_name = re.search(r'\b((?:string)?printf)\s*\(', line,
                                      re.I).group(1)
            error(filename, linenum, 'runtime/printf', 4,
                  'Potential format string bug. Do %s("%%s", %s) instead.' %
                  (function_name, match.group(1)))

    # Check for potential memset bugs like memset(buf, sizeof(buf), 0).
    match = Search(r'memset\s*\(([^,]*),\s*([^,]*),\s*0\s*\)', line)
    if match and not Match(r"^''|-?[0-9]+|0x[0-9A-Fa-f]$", match.group(2)):
        error(filename, linenum, 'runtime/memset', 4,
              'Did you mean "memset(%s, 0, %s)"?' %
              (match.group(1), match.group(2)))

    if Search(r'\busing namespace\b', line):
        error(filename, linenum, 'build/namespaces', 5,
              'Do not use namespace using-directives.  '
              'Use using-declarations instead.')

    # Detect variable-length arrays.
    match = Match(r'\s*(.+::)?(\w+) [a-z]\w*\[(.+)];', line)
    if (match and match.group(2) != 'return' and match.group(2) != 'delete' and
            match.group(3).find(']') == -1):
        # Split the size using space and arithmetic operators as delimiters.
        # If any of the resulting tokens are not compile time constants then
        # report the error.
        tokens = re.split(r'\s|\+|\-|\*|\/|<<|>>]', match.group(3))
        is_const = True
Z
zhangjinchao01 已提交
4917
        skip_next = False
Y
Yu Yang 已提交
4918 4919 4920 4921 4922 4923 4924 4925 4926 4927 4928 4929 4930 4931 4932 4933 4934 4935 4936 4937 4938 4939 4940 4941 4942 4943 4944 4945 4946 4947 4948 4949 4950 4951 4952 4953 4954 4955 4956 4957 4958
        for tok in tokens:
            if skip_next:
                skip_next = False
                continue

            if Search(r'sizeof\(.+\)', tok): continue
            if Search(r'arraysize\(\w+\)', tok): continue

            tok = tok.lstrip('(')
            tok = tok.rstrip(')')
            if not tok: continue
            if Match(r'\d+', tok): continue
            if Match(r'0[xX][0-9a-fA-F]+', tok): continue
            if Match(r'k[A-Z0-9]\w*', tok): continue
            if Match(r'(.+::)?k[A-Z0-9]\w*', tok): continue
            if Match(r'(.+::)?[A-Z][A-Z0-9_]*', tok): continue
            # A catch all for tricky sizeof cases, including 'sizeof expression',
            # 'sizeof(*type)', 'sizeof(const type)', 'sizeof(struct StructName)'
            # requires skipping the next token because we split on ' ' and '*'.
            if tok.startswith('sizeof'):
                skip_next = True
                continue
            is_const = False
            break
        if not is_const:
            error(
                filename, linenum, 'runtime/arrays', 1,
                'Do not use variable-length arrays.  Use an appropriately named '
                "('k' followed by CamelCase) compile-time constant for the size."
            )

    # Check for use of unnamed namespaces in header files.  Registration
    # macros are typically OK, so we allow use of "namespace {" on lines
    # that end with backslashes.
    if (file_extension == 'h' and Search(r'\bnamespace\s*{', line) and
            line[-1] != '\\'):
        error(
            filename, linenum, 'build/namespaces', 4,
            'Do not use unnamed namespaces in header files.  See '
            'http://google-styleguide.googlecode.com/svn/trunk/cppguide.xml#Namespaces'
            ' for more information.')
Z
zhangjinchao01 已提交
4959 4960 4961


def CheckGlobalStatic(filename, clean_lines, linenum, error):
Y
Yu Yang 已提交
4962
    """Check for unsafe global or static objects.
Z
zhangjinchao01 已提交
4963 4964 4965 4966 4967 4968 4969

  Args:
    filename: The name of the current file.
    clean_lines: A CleansedLines instance containing the file.
    linenum: The number of the line to check.
    error: The function to call with any errors found.
  """
Y
Yu Yang 已提交
4970 4971 4972 4973 4974 4975 4976 4977 4978 4979 4980 4981 4982 4983 4984 4985 4986 4987 4988 4989 4990 4991 4992 4993 4994 4995 4996 4997 4998 4999 5000 5001 5002 5003 5004 5005 5006 5007 5008 5009
    line = clean_lines.elided[linenum]

    # Match two lines at a time to support multiline declarations
    if linenum + 1 < clean_lines.NumLines() and not Search(r'[;({]', line):
        line += clean_lines.elided[linenum + 1].strip()

    # Check for people declaring static/global STL strings at the top level.
    # This is dangerous because the C++ language does not guarantee that
    # globals with constructors are initialized before the first access.
    match = Match(r'((?:|static +)(?:|const +))string +([a-zA-Z0-9_:]+)\b(.*)',
                  line)

    # Remove false positives:
    # - String pointers (as opposed to values).
    #    string *pointer
    #    const string *pointer
    #    string const *pointer
    #    string *const pointer
    #
    # - Functions and template specializations.
    #    string Function<Type>(...
    #    string Class<Type>::Method(...
    #
    # - Operators.  These are matched separately because operator names
    #   cross non-word boundaries, and trying to match both operators
    #   and functions at the same time would decrease accuracy of
    #   matching identifiers.
    #    string Class::operator*()
    if (match and
            not Search(r'\bstring\b(\s+const)?\s*\*\s*(const\s+)?\w', line) and
            not Search(r'\boperator\W', line) and not Match(
                r'\s*(<.*>)?(::[a-zA-Z0-9_]+)*\s*\(([^"]|$)', match.group(3))):
        error(
            filename, linenum, 'runtime/string', 4,
            'For a static/global string constant, use a C style string instead: '
            '"%schar %s[]".' % (match.group(1), match.group(2)))

    if Search(r'\b([A-Za-z0-9_]*_)\(\1\)', line):
        error(filename, linenum, 'runtime/init', 4,
              'You seem to be initializing a member variable with itself.')
Z
zhangjinchao01 已提交
5010 5011 5012


def CheckPrintf(filename, clean_lines, linenum, error):
Y
Yu Yang 已提交
5013
    """Check for printf related issues.
Z
zhangjinchao01 已提交
5014 5015 5016 5017 5018 5019 5020

  Args:
    filename: The name of the current file.
    clean_lines: A CleansedLines instance containing the file.
    linenum: The number of the line to check.
    error: The function to call with any errors found.
  """
Y
Yu Yang 已提交
5021 5022 5023 5024 5025 5026 5027 5028 5029 5030 5031 5032 5033 5034 5035 5036 5037 5038
    line = clean_lines.elided[linenum]

    # When snprintf is used, the second argument shouldn't be a literal.
    match = Search(r'snprintf\s*\(([^,]*),\s*([0-9]*)\s*,', line)
    if match and match.group(2) != '0':
        # If 2nd arg is zero, snprintf is used to calculate size.
        error(filename, linenum, 'runtime/printf', 3,
              'If you can, use sizeof(%s) instead of %s as the 2nd arg '
              'to snprintf.' % (match.group(1), match.group(2)))

    # Check if some verboten C functions are being used.
    if Search(r'\bsprintf\s*\(', line):
        error(filename, linenum, 'runtime/printf', 5,
              'Never use sprintf. Use snprintf instead.')
    match = Search(r'\b(strcpy|strcat)\s*\(', line)
    if match:
        error(filename, linenum, 'runtime/printf', 4,
              'Almost always, snprintf is better than %s' % match.group(1))
Z
zhangjinchao01 已提交
5039 5040 5041


def IsDerivedFunction(clean_lines, linenum):
Y
Yu Yang 已提交
5042
    """Check if current line contains an inherited function.
Z
zhangjinchao01 已提交
5043 5044 5045 5046 5047 5048 5049 5050

  Args:
    clean_lines: A CleansedLines instance containing the file.
    linenum: The number of the line to check.
  Returns:
    True if current line contains a function with "override"
    virt-specifier.
  """
Y
Yu Yang 已提交
5051 5052 5053 5054 5055 5056 5057 5058 5059 5060
    # Scan back a few lines for start of current function
    for i in xrange(linenum, max(-1, linenum - 10), -1):
        match = Match(r'^([^()]*\w+)\(', clean_lines.elided[i])
        if match:
            # Look for "override" after the matching closing parenthesis
            line, _, closing_paren = CloseExpression(clean_lines, i,
                                                     len(match.group(1)))
            return (closing_paren >= 0 and
                    Search(r'\boverride\b', line[closing_paren:]))
    return False
Z
zhangjinchao01 已提交
5061 5062 5063


def IsOutOfLineMethodDefinition(clean_lines, linenum):
Y
Yu Yang 已提交
5064
    """Check if current line contains an out-of-line method definition.
Z
zhangjinchao01 已提交
5065 5066 5067 5068 5069 5070 5071

  Args:
    clean_lines: A CleansedLines instance containing the file.
    linenum: The number of the line to check.
  Returns:
    True if current line contains an out-of-line method definition.
  """
Y
Yu Yang 已提交
5072 5073 5074 5075 5076 5077
    # Scan back a few lines for start of current function
    for i in xrange(linenum, max(-1, linenum - 10), -1):
        if Match(r'^([^()]*\w+)\(', clean_lines.elided[i]):
            return Match(r'^[^()]*\w+::\w+\(',
                         clean_lines.elided[i]) is not None
    return False
Z
zhangjinchao01 已提交
5078 5079 5080


def IsInitializerList(clean_lines, linenum):
Y
Yu Yang 已提交
5081
    """Check if current line is inside constructor initializer list.
Z
zhangjinchao01 已提交
5082 5083 5084 5085 5086 5087 5088 5089

  Args:
    clean_lines: A CleansedLines instance containing the file.
    linenum: The number of the line to check.
  Returns:
    True if current line appears to be inside constructor initializer
    list, False otherwise.
  """
Y
Yu Yang 已提交
5090 5091 5092 5093 5094 5095 5096 5097 5098 5099 5100 5101 5102 5103 5104 5105 5106 5107 5108 5109 5110 5111 5112 5113 5114 5115 5116 5117 5118 5119 5120 5121 5122 5123 5124
    for i in xrange(linenum, 1, -1):
        line = clean_lines.elided[i]
        if i == linenum:
            remove_function_body = Match(r'^(.*)\{\s*$', line)
            if remove_function_body:
                line = remove_function_body.group(1)

        if Search(r'\s:\s*\w+[({]', line):
            # A lone colon tend to indicate the start of a constructor
            # initializer list.  It could also be a ternary operator, which
            # also tend to appear in constructor initializer lists as
            # opposed to parameter lists.
            return True
        if Search(r'\}\s*,\s*$', line):
            # A closing brace followed by a comma is probably the end of a
            # brace-initialized member in constructor initializer list.
            return True
        if Search(r'[{};]\s*$', line):
            # Found one of the following:
            # - A closing brace or semicolon, probably the end of the previous
            #   function.
            # - An opening brace, probably the start of current class or namespace.
            #
            # Current line is probably not inside an initializer list since
            # we saw one of those things without seeing the starting colon.
            return False

    # Got to the beginning of the file without seeing the start of
    # constructor initializer list.
    return False


def CheckForNonConstReference(filename, clean_lines, linenum, nesting_state,
                              error):
    """Check for non-const references.
Z
zhangjinchao01 已提交
5125 5126 5127 5128 5129 5130 5131 5132 5133 5134 5135 5136

  Separate from CheckLanguage since it scans backwards from current
  line, instead of scanning forward.

  Args:
    filename: The name of the current file.
    clean_lines: A CleansedLines instance containing the file.
    linenum: The number of the line to check.
    nesting_state: A NestingState instance which maintains information about
                   the current stack of nested blocks being parsed.
    error: The function to call with any errors found.
  """
Y
Yu Yang 已提交
5137 5138 5139 5140 5141 5142 5143 5144 5145 5146 5147 5148 5149 5150 5151 5152 5153 5154 5155 5156 5157 5158 5159 5160 5161 5162 5163 5164 5165 5166 5167 5168 5169 5170 5171 5172 5173 5174 5175 5176 5177 5178 5179 5180 5181 5182 5183 5184 5185 5186 5187 5188 5189 5190 5191 5192 5193 5194 5195 5196 5197 5198 5199 5200 5201 5202 5203 5204 5205 5206 5207 5208 5209 5210 5211 5212 5213 5214 5215 5216 5217 5218 5219 5220 5221 5222 5223
    # Do nothing if there is no '&' on current line.
    line = clean_lines.elided[linenum]
    if '&' not in line:
        return

    # If a function is inherited, current function doesn't have much of
    # a choice, so any non-const references should not be blamed on
    # derived function.
    if IsDerivedFunction(clean_lines, linenum):
        return

    # Don't warn on out-of-line method definitions, as we would warn on the
    # in-line declaration, if it isn't marked with 'override'.
    if IsOutOfLineMethodDefinition(clean_lines, linenum):
        return

    # Long type names may be broken across multiple lines, usually in one
    # of these forms:
    #   LongType
    #       ::LongTypeContinued &identifier
    #   LongType::
    #       LongTypeContinued &identifier
    #   LongType<
    #       ...>::LongTypeContinued &identifier
    #
    # If we detected a type split across two lines, join the previous
    # line to current line so that we can match const references
    # accordingly.
    #
    # Note that this only scans back one line, since scanning back
    # arbitrary number of lines would be expensive.  If you have a type
    # that spans more than 2 lines, please use a typedef.
    if linenum > 1:
        previous = None
        if Match(r'\s*::(?:[\w<>]|::)+\s*&\s*\S', line):
            # previous_line\n + ::current_line
            previous = Search(r'\b((?:const\s*)?(?:[\w<>]|::)+[\w<>])\s*$',
                              clean_lines.elided[linenum - 1])
        elif Match(r'\s*[a-zA-Z_]([\w<>]|::)+\s*&\s*\S', line):
            # previous_line::\n + current_line
            previous = Search(r'\b((?:const\s*)?(?:[\w<>]|::)+::)\s*$',
                              clean_lines.elided[linenum - 1])
        if previous:
            line = previous.group(1) + line.lstrip()
        else:
            # Check for templated parameter that is split across multiple lines
            endpos = line.rfind('>')
            if endpos > -1:
                (_, startline, startpos) = ReverseCloseExpression(
                    clean_lines, linenum, endpos)
                if startpos > -1 and startline < linenum:
                    # Found the matching < on an earlier line, collect all
                    # pieces up to current line.
                    line = ''
                    for i in xrange(startline, linenum + 1):
                        line += clean_lines.elided[i].strip()

    # Check for non-const references in function parameters.  A single '&' may
    # found in the following places:
    #   inside expression: binary & for bitwise AND
    #   inside expression: unary & for taking the address of something
    #   inside declarators: reference parameter
    # We will exclude the first two cases by checking that we are not inside a
    # function body, including one that was just introduced by a trailing '{'.
    # TODO(unknown): Doesn't account for 'catch(Exception& e)' [rare].
    if (nesting_state.previous_stack_top and
            not (isinstance(nesting_state.previous_stack_top, _ClassInfo) or
                 isinstance(nesting_state.previous_stack_top, _NamespaceInfo))):
        # Not at toplevel, not within a class, and not within a namespace
        return

    # Avoid initializer lists.  We only need to scan back from the
    # current line for something that starts with ':'.
    #
    # We don't need to check the current line, since the '&' would
    # appear inside the second set of parentheses on the current line as
    # opposed to the first set.
    if linenum > 0:
        for i in xrange(linenum - 1, max(0, linenum - 10), -1):
            previous_line = clean_lines.elided[i]
            if not Search(r'[),]\s*$', previous_line):
                break
            if Match(r'^\s*:\s+\S', previous_line):
                return

    # Avoid preprocessors
    if Search(r'\\\s*$', line):
Z
zhangjinchao01 已提交
5224 5225
        return

Y
Yu Yang 已提交
5226 5227
    # Avoid constructor initializer lists
    if IsInitializerList(clean_lines, linenum):
Z
zhangjinchao01 已提交
5228 5229
        return

Y
Yu Yang 已提交
5230 5231 5232 5233 5234 5235 5236 5237 5238 5239 5240 5241 5242 5243 5244 5245 5246 5247 5248 5249 5250 5251 5252 5253 5254 5255 5256 5257
    # We allow non-const references in a few standard places, like functions
    # called "swap()" or iostream operators like "<<" or ">>".  Do not check
    # those function parameters.
    #
    # We also accept & in static_assert, which looks like a function but
    # it's actually a declaration expression.
    whitelisted_functions = (r'(?:[sS]wap(?:<\w:+>)?|'
                             r'operator\s*[<>][<>]|'
                             r'static_assert|COMPILE_ASSERT'
                             r')\s*\(')
    if Search(whitelisted_functions, line):
        return
    elif not Search(r'\S+\([^)]*$', line):
        # Don't see a whitelisted function on this line.  Actually we
        # didn't see any function name on this line, so this is likely a
        # multi-line parameter list.  Try a bit harder to catch this case.
        for i in xrange(2):
            if (linenum > i and Search(whitelisted_functions,
                                       clean_lines.elided[linenum - i - 1])):
                return

    decls = ReplaceAll(r'{[^}]*}', ' ', line)  # exclude function body
    for parameter in re.findall(_RE_PATTERN_REF_PARAM, decls):
        if not Match(_RE_PATTERN_CONST_REF_PARAM, parameter):
            error(filename, linenum, 'runtime/references', 2,
                  'Is this a non-const reference? '
                  'If so, make const or use a pointer: ' + ReplaceAll(
                      ' *<', '<', parameter))
Z
zhangjinchao01 已提交
5258 5259 5260


def CheckCasts(filename, clean_lines, linenum, error):
Y
Yu Yang 已提交
5261
    """Various cast related checks.
Z
zhangjinchao01 已提交
5262 5263 5264 5265 5266 5267 5268

  Args:
    filename: The name of the current file.
    clean_lines: A CleansedLines instance containing the file.
    linenum: The number of the line to check.
    error: The function to call with any errors found.
  """
Y
Yu Yang 已提交
5269 5270 5271 5272 5273 5274 5275 5276 5277 5278 5279 5280 5281 5282 5283 5284 5285 5286 5287 5288 5289 5290 5291 5292 5293 5294 5295 5296 5297 5298 5299 5300 5301 5302 5303 5304 5305 5306 5307 5308 5309 5310 5311 5312 5313 5314 5315 5316 5317 5318 5319 5320 5321
    line = clean_lines.elided[linenum]

    # Check to see if they're using an conversion function cast.
    # I just try to capture the most common basic types, though there are more.
    # Parameterless conversion functions, such as bool(), are allowed as they are
    # probably a member operator declaration or default constructor.
    match = Search(r'(\bnew\s+|\S<\s*(?:const\s+)?)?\b'
                   r'(int|float|double|bool|char|int32|uint32|int64|uint64)'
                   r'(\([^)].*)', line)
    expecting_function = ExpectingFunctionArgs(clean_lines, linenum)
    if match and not expecting_function:
        matched_type = match.group(2)

        # matched_new_or_template is used to silence two false positives:
        # - New operators
        # - Template arguments with function types
        #
        # For template arguments, we match on types immediately following
        # an opening bracket without any spaces.  This is a fast way to
        # silence the common case where the function type is the first
        # template argument.  False negative with less-than comparison is
        # avoided because those operators are usually followed by a space.
        #
        #   function<double(double)>   // bracket + no space = false positive
        #   value < double(42)         // bracket + space = true positive
        matched_new_or_template = match.group(1)

        # Avoid arrays by looking for brackets that come after the closing
        # parenthesis.
        if Match(r'\([^()]+\)\s*\[', match.group(3)):
            return

        # Other things to ignore:
        # - Function pointers
        # - Casts to pointer types
        # - Placement new
        # - Alias declarations
        matched_funcptr = match.group(3)
        if (matched_new_or_template is None and not (matched_funcptr and (Match(
                r'\((?:[^() ]+::\s*\*\s*)?[^() ]+\)\s*\(',
                matched_funcptr) or matched_funcptr.startswith('(*)'))) and
                not Match(r'\s*using\s+\S+\s*=\s*' + matched_type, line) and
                not Search(r'new\(\S+\)\s*' + matched_type, line)):
            error(filename, linenum, 'readability/casting', 4,
                  'Using deprecated casting style.  '
                  'Use static_cast<%s>(...) instead' % matched_type)

    if not expecting_function:
        CheckCStyleCast(filename, clean_lines, linenum, 'static_cast',
                        r'\((int|float|double|bool|char|u?int(16|32|64))\)',
                        error)

    # This doesn't catch all cases. Consider (const char * const)"hello".
Z
zhangjinchao01 已提交
5322
    #
Y
Yu Yang 已提交
5323 5324 5325 5326 5327 5328 5329 5330 5331 5332 5333 5334 5335
    # (char *) "foo" should always be a const_cast (reinterpret_cast won't
    # compile).
    if CheckCStyleCast(filename, clean_lines, linenum, 'const_cast',
                       r'\((char\s?\*+\s?)\)\s*"', error):
        pass
    else:
        # Check pointer casts for other than string constants
        CheckCStyleCast(filename, clean_lines, linenum, 'reinterpret_cast',
                        r'\((\w+\s?\*+\s?)\)', error)

    # In addition, we look for people taking the address of a cast.  This
    # is dangerous -- casts can assign to temporaries, so the pointer doesn't
    # point where you think.
Z
zhangjinchao01 已提交
5336
    #
Y
Yu Yang 已提交
5337 5338 5339 5340 5341 5342 5343 5344 5345
    # Some non-identifier character is required before the '&' for the
    # expression to be recognized as a cast.  These are casts:
    #   expression = &static_cast<int*>(temporary());
    #   function(&(int*)(temporary()));
    #
    # This is not a cast:
    #   reference_type&(int* function_param);
    match = Search(r'(?:[^\w]&\(([^)*][^)]*)\)[\w(])|'
                   r'(?:[^\w]&(static|dynamic|down|reinterpret)_cast\b)', line)
Z
zhangjinchao01 已提交
5346
    if match:
Y
Yu Yang 已提交
5347 5348 5349 5350 5351 5352 5353 5354 5355 5356 5357 5358 5359 5360 5361 5362 5363 5364 5365 5366 5367 5368 5369 5370 5371 5372 5373 5374
        # Try a better error message when the & is bound to something
        # dereferenced by the casted pointer, as opposed to the casted
        # pointer itself.
        parenthesis_error = False
        match = Match(r'^(.*&(?:static|dynamic|down|reinterpret)_cast\b)<',
                      line)
        if match:
            _, y1, x1 = CloseExpression(clean_lines, linenum,
                                        len(match.group(1)))
            if x1 >= 0 and clean_lines.elided[y1][x1] == '(':
                _, y2, x2 = CloseExpression(clean_lines, y1, x1)
                if x2 >= 0:
                    extended_line = clean_lines.elided[y2][x2:]
                    if y2 < clean_lines.NumLines() - 1:
                        extended_line += clean_lines.elided[y2 + 1]
                    if Match(r'\s*(?:->|\[)', extended_line):
                        parenthesis_error = True

        if parenthesis_error:
            error(filename, linenum, 'readability/casting', 4,
                  ('Are you taking an address of something dereferenced '
                   'from a cast?  Wrapping the dereferenced expression in '
                   'parentheses will make the binding more obvious'))
        else:
            error(filename, linenum, 'runtime/casting', 4,
                  ('Are you taking an address of a cast?  '
                   'This is dangerous: could be a temp var.  '
                   'Take the address before doing the cast, rather than after'))
Z
zhangjinchao01 已提交
5375 5376 5377


def CheckCStyleCast(filename, clean_lines, linenum, cast_type, pattern, error):
Y
Yu Yang 已提交
5378
    """Checks for a C-style cast by looking for the pattern.
Z
zhangjinchao01 已提交
5379 5380 5381 5382 5383 5384 5385 5386 5387 5388 5389 5390 5391 5392

  Args:
    filename: The name of the current file.
    clean_lines: A CleansedLines instance containing the file.
    linenum: The number of the line to check.
    cast_type: The string for the C++ cast to recommend.  This is either
      reinterpret_cast, static_cast, or const_cast, depending.
    pattern: The regular expression used to find C-style casts.
    error: The function to call with any errors found.

  Returns:
    True if an error was emitted.
    False otherwise.
  """
Y
Yu Yang 已提交
5393 5394 5395 5396
    line = clean_lines.elided[linenum]
    match = Search(pattern, line)
    if not match:
        return False
Z
zhangjinchao01 已提交
5397

Y
Yu Yang 已提交
5398 5399 5400 5401
    # Exclude lines with keywords that tend to look like casts
    context = line[0:match.start(1) - 1]
    if Match(r'.*\b(?:sizeof|alignof|alignas|[_A-Z][_A-Z0-9]*)\s*$', context):
        return False
Z
zhangjinchao01 已提交
5402

Y
Yu Yang 已提交
5403 5404 5405 5406 5407 5408 5409
    # Try expanding current context to see if we one level of
    # parentheses inside a macro.
    if linenum > 0:
        for i in xrange(linenum - 1, max(0, linenum - 5), -1):
            context = clean_lines.elided[i] + context
    if Match(r'.*\b[_A-Z][_A-Z0-9]*\s*\((?:\([^()]*\)|[^()])*$', context):
        return False
Z
zhangjinchao01 已提交
5410

Y
Yu Yang 已提交
5411 5412 5413
    # operator++(int) and operator--(int)
    if context.endswith(' operator++') or context.endswith(' operator--'):
        return False
Z
zhangjinchao01 已提交
5414

Y
Yu Yang 已提交
5415 5416 5417 5418 5419 5420 5421 5422 5423 5424 5425 5426 5427 5428 5429 5430 5431 5432 5433 5434 5435 5436 5437 5438 5439 5440 5441 5442 5443 5444 5445 5446 5447
    # A single unnamed argument for a function tends to look like old
    # style cast.  If we see those, don't issue warnings for deprecated
    # casts, instead issue warnings for unnamed arguments where
    # appropriate.
    #
    # These are things that we want warnings for, since the style guide
    # explicitly require all parameters to be named:
    #   Function(int);
    #   Function(int) {
    #   ConstMember(int) const;
    #   ConstMember(int) const {
    #   ExceptionMember(int) throw (...);
    #   ExceptionMember(int) throw (...) {
    #   PureVirtual(int) = 0;
    #   [](int) -> bool {
    #
    # These are functions of some sort, where the compiler would be fine
    # if they had named parameters, but people often omit those
    # identifiers to reduce clutter:
    #   (FunctionPointer)(int);
    #   (FunctionPointer)(int) = value;
    #   Function((function_pointer_arg)(int))
    #   Function((function_pointer_arg)(int), int param)
    #   <TemplateArgument(int)>;
    #   <(FunctionPointerTemplateArgument)(int)>;
    remainder = line[match.end(0):]
    if Match(r'^\s*(?:;|const\b|throw\b|final\b|override\b|[=>{),]|->)',
             remainder):
        # Looks like an unnamed parameter.

        # Don't warn on any kind of template arguments.
        if Match(r'^\s*>', remainder):
            return False
Z
zhangjinchao01 已提交
5448

Y
Yu Yang 已提交
5449 5450 5451 5452 5453 5454 5455
        # Don't warn on assignments to function pointers, but keep warnings for
        # unnamed parameters to pure virtual functions.  Note that this pattern
        # will also pass on assignments of "0" to function pointers, but the
        # preferred values for those would be "nullptr" or "NULL".
        matched_zero = Match(r'^\s=\s*(\S+)\s*;', remainder)
        if matched_zero and matched_zero.group(1) != '0':
            return False
Z
zhangjinchao01 已提交
5456

Y
Yu Yang 已提交
5457 5458 5459 5460 5461 5462 5463 5464 5465 5466 5467 5468 5469 5470 5471 5472 5473 5474 5475 5476 5477 5478
        # Don't warn on function pointer declarations.  For this we need
        # to check what came before the "(type)" string.
        if Match(r'.*\)\s*$', line[0:match.start(0)]):
            return False

        # Don't warn if the parameter is named with block comments, e.g.:
        #  Function(int /*unused_param*/);
        raw_line = clean_lines.raw_lines[linenum]
        if '/*' in raw_line:
            return False

        # Passed all filters, issue warning here.
        error(filename, linenum, 'readability/function', 3,
              'All parameters should be named in a function')
        return True

    # At this point, all that should be left is actual casts.
    error(filename, linenum, 'readability/casting', 4,
          'Using C-style cast.  Use %s<%s>(...) instead' %
          (cast_type, match.group(1)))

    return True
Z
zhangjinchao01 已提交
5479 5480 5481


def ExpectingFunctionArgs(clean_lines, linenum):
Y
Yu Yang 已提交
5482
    """Checks whether where function type arguments are expected.
Z
zhangjinchao01 已提交
5483 5484 5485 5486 5487 5488 5489 5490 5491

  Args:
    clean_lines: A CleansedLines instance containing the file.
    linenum: The number of the line to check.

  Returns:
    True if the line at 'linenum' is inside something that expects arguments
    of function types.
  """
Y
Yu Yang 已提交
5492 5493 5494 5495 5496 5497 5498 5499 5500
    line = clean_lines.elided[linenum]
    return (Match(r'^\s*MOCK_(CONST_)?METHOD\d+(_T)?\(', line) or
            (linenum >= 2 and
             (Match(r'^\s*MOCK_(?:CONST_)?METHOD\d+(?:_T)?\((?:\S+,)?\s*$',
                    clean_lines.elided[linenum - 1]) or
              Match(r'^\s*MOCK_(?:CONST_)?METHOD\d+(?:_T)?\(\s*$',
                    clean_lines.elided[linenum - 2]) or
              Search(r'\bstd::m?function\s*\<\s*$',
                     clean_lines.elided[linenum - 1]))))
Z
zhangjinchao01 已提交
5501 5502 5503


_HEADERS_CONTAINING_TEMPLATES = (
Y
Yu Yang 已提交
5504 5505 5506 5507 5508 5509 5510 5511 5512 5513 5514 5515 5516 5517 5518 5519 5520 5521 5522 5523 5524 5525 5526 5527 5528 5529 5530 5531 5532 5533 5534 5535 5536 5537 5538 5539 5540 5541 5542 5543 5544 5545 5546 5547 5548 5549 5550 5551 5552 5553 5554 5555 5556 5557 5558 5559 5560
    ('<deque>', ('deque', )),
    ('<functional>', (
        'unary_function',
        'binary_function',
        'plus',
        'minus',
        'multiplies',
        'divides',
        'modulus',
        'negate',
        'equal_to',
        'not_equal_to',
        'greater',
        'less',
        'greater_equal',
        'less_equal',
        'logical_and',
        'logical_or',
        'logical_not',
        'unary_negate',
        'not1',
        'binary_negate',
        'not2',
        'bind1st',
        'bind2nd',
        'pointer_to_unary_function',
        'pointer_to_binary_function',
        'ptr_fun',
        'mem_fun_t',
        'mem_fun',
        'mem_fun1_t',
        'mem_fun1_ref_t',
        'mem_fun_ref_t',
        'const_mem_fun_t',
        'const_mem_fun1_t',
        'const_mem_fun_ref_t',
        'const_mem_fun1_ref_t',
        'mem_fun_ref', )),
    ('<limits>', ('numeric_limits', )),
    ('<list>', ('list', )),
    ('<map>', (
        'map',
        'multimap', )),
    ('<memory>', ('allocator', )),
    ('<queue>', (
        'queue',
        'priority_queue', )),
    ('<set>', (
        'set',
        'multiset', )),
    ('<stack>', ('stack', )),
    ('<string>', (
        'char_traits',
        'basic_string', )),
    ('<tuple>', ('tuple', )),
    ('<utility>', ('pair', )),
    ('<vector>', ('vector', )),
Z
zhangjinchao01 已提交
5561 5562 5563

    # gcc extensions.
    # Note: std::hash is their hash, ::hash is our hash
Y
Yu Yang 已提交
5564 5565 5566 5567 5568 5569 5570
    ('<hash_map>', (
        'hash_map',
        'hash_multimap', )),
    ('<hash_set>', (
        'hash_set',
        'hash_multiset', )),
    ('<slist>', ('slist', )), )
Z
zhangjinchao01 已提交
5571 5572 5573 5574 5575 5576

_RE_PATTERN_STRING = re.compile(r'\bstring\b')

_re_pattern_algorithm_header = []
for _template in ('copy', 'max', 'min', 'min_element', 'sort', 'swap',
                  'transform'):
Y
Yu Yang 已提交
5577 5578 5579 5580 5581
    # Match max<type>(..., ...), max(..., ...), but not foo->max, foo.max or
    # type::max().
    _re_pattern_algorithm_header.append(
        (re.compile(r'[^>.]\b' + _template + r'(<.*?>)?\([^\)]'), _template,
         '<algorithm>'))
Z
zhangjinchao01 已提交
5582 5583 5584

_re_pattern_templates = []
for _header, _templates in _HEADERS_CONTAINING_TEMPLATES:
Y
Yu Yang 已提交
5585 5586 5587 5588
    for _template in _templates:
        _re_pattern_templates.append(
            (re.compile(r'(\<|\b)' + _template + r'\s*\<'), _template + '<>',
             _header))
Z
zhangjinchao01 已提交
5589 5590 5591


def FilesBelongToSameModule(filename_cc, filename_h):
Y
Yu Yang 已提交
5592
    """Check if these two filenames belong to the same module.
Z
zhangjinchao01 已提交
5593 5594 5595 5596 5597 5598 5599 5600 5601 5602 5603 5604 5605 5606 5607 5608 5609 5610 5611 5612 5613 5614 5615 5616 5617 5618 5619 5620

  The concept of a 'module' here is a as follows:
  foo.h, foo-inl.h, foo.cc, foo_test.cc and foo_unittest.cc belong to the
  same 'module' if they are in the same directory.
  some/path/public/xyzzy and some/path/internal/xyzzy are also considered
  to belong to the same module here.

  If the filename_cc contains a longer path than the filename_h, for example,
  '/absolute/path/to/base/sysinfo.cc', and this file would include
  'base/sysinfo.h', this function also produces the prefix needed to open the
  header. This is used by the caller of this function to more robustly open the
  header file. We don't have access to the real include paths in this context,
  so we need this guesswork here.

  Known bugs: tools/base/bar.cc and base/bar.h belong to the same module
  according to this implementation. Because of this, this function gives
  some false positives. This should be sufficiently rare in practice.

  Args:
    filename_cc: is the path for the .cc file
    filename_h: is the path for the header path

  Returns:
    Tuple with a bool and a string:
    bool: True if filename_cc and filename_h belong to the same module.
    string: the additional prefix needed to open the header file.
  """

Y
Yu Yang 已提交
5621 5622 5623 5624 5625 5626 5627 5628 5629 5630 5631 5632 5633 5634 5635 5636 5637 5638 5639 5640 5641 5642 5643
    if not filename_cc.endswith('.cc'):
        return (False, '')
    filename_cc = filename_cc[:-len('.cc')]
    if filename_cc.endswith('_unittest'):
        filename_cc = filename_cc[:-len('_unittest')]
    elif filename_cc.endswith('_test'):
        filename_cc = filename_cc[:-len('_test')]
    filename_cc = filename_cc.replace('/public/', '/')
    filename_cc = filename_cc.replace('/internal/', '/')

    if not filename_h.endswith('.h'):
        return (False, '')
    filename_h = filename_h[:-len('.h')]
    if filename_h.endswith('-inl'):
        filename_h = filename_h[:-len('-inl')]
    filename_h = filename_h.replace('/public/', '/')
    filename_h = filename_h.replace('/internal/', '/')

    files_belong_to_same_module = filename_cc.endswith(filename_h)
    common_path = ''
    if files_belong_to_same_module:
        common_path = filename_cc[:-len(filename_h)]
    return files_belong_to_same_module, common_path
Z
zhangjinchao01 已提交
5644 5645 5646


def UpdateIncludeState(filename, include_dict, io=codecs):
Y
Yu Yang 已提交
5647
    """Fill up the include_dict with new includes found from the file.
Z
zhangjinchao01 已提交
5648 5649 5650 5651 5652 5653 5654 5655 5656

  Args:
    filename: the name of the header to read.
    include_dict: a dictionary in which the headers are inserted.
    io: The io factory to use to read the file. Provided for testability.

  Returns:
    True if a header was successfully added. False otherwise.
  """
Y
Yu Yang 已提交
5657 5658 5659 5660 5661 5662 5663 5664 5665 5666 5667 5668 5669 5670
    headerfile = None
    try:
        headerfile = io.open(filename, 'r', 'utf8', 'replace')
    except IOError:
        return False
    linenum = 0
    for line in headerfile:
        linenum += 1
        clean_line = CleanseComments(line)
        match = _RE_PATTERN_INCLUDE.search(clean_line)
        if match:
            include = match.group(2)
            include_dict.setdefault(include, linenum)
    return True
Z
zhangjinchao01 已提交
5671 5672


Y
Yu Yang 已提交
5673 5674 5675 5676
def CheckForIncludeWhatYouUse(filename,
                              clean_lines,
                              include_state,
                              error,
Z
zhangjinchao01 已提交
5677
                              io=codecs):
Y
Yu Yang 已提交
5678
    """Reports for missing stl includes.
Z
zhangjinchao01 已提交
5679 5680 5681 5682 5683 5684 5685 5686 5687 5688 5689 5690 5691 5692 5693

  This function will output warnings to make sure you are including the headers
  necessary for the stl containers and functions that you use. We only give one
  reason to include a header. For example, if you use both equal_to<> and
  less<> in a .h file, only one (the latter in the file) of these will be
  reported as a reason to include the <functional>.

  Args:
    filename: The name of the current file.
    clean_lines: A CleansedLines instance containing the file.
    include_state: An _IncludeState instance.
    error: The function to call with any errors found.
    io: The IO factory to use to read the header file. Provided for unittest
        injection.
  """
Y
Yu Yang 已提交
5694 5695
    required = {}  # A map of header name to linenumber and the template entity.
    # Example of required: { '<functional>': (1219, 'less<>') }
Z
zhangjinchao01 已提交
5696

Y
Yu Yang 已提交
5697 5698 5699 5700
    for linenum in xrange(clean_lines.NumLines()):
        line = clean_lines.elided[linenum]
        if not line or line[0] == '#':
            continue
Z
zhangjinchao01 已提交
5701

Y
Yu Yang 已提交
5702 5703 5704 5705 5706 5707 5708 5709 5710 5711 5712 5713 5714 5715 5716 5717 5718 5719 5720 5721 5722 5723 5724 5725 5726 5727 5728 5729 5730 5731 5732 5733 5734 5735 5736 5737 5738 5739 5740 5741 5742 5743 5744 5745 5746 5747 5748 5749 5750 5751 5752 5753 5754 5755 5756 5757 5758 5759 5760 5761 5762 5763 5764 5765 5766 5767 5768
        # String is special -- it is a non-templatized type in STL.
        matched = _RE_PATTERN_STRING.search(line)
        if matched:
            # Don't warn about strings in non-STL namespaces:
            # (We check only the first match per line; good enough.)
            prefix = line[:matched.start()]
            if prefix.endswith('std::') or not prefix.endswith('::'):
                required['<string>'] = (linenum, 'string')

        for pattern, template, header in _re_pattern_algorithm_header:
            if pattern.search(line):
                required[header] = (linenum, template)

        # The following function is just a speed up, no semantics are changed.
        if not '<' in line:  # Reduces the cpu time usage by skipping lines.
            continue

        for pattern, template, header in _re_pattern_templates:
            if pattern.search(line):
                required[header] = (linenum, template)

    # The policy is that if you #include something in foo.h you don't need to
    # include it again in foo.cc. Here, we will look at possible includes.
    # Let's flatten the include_state include_list and copy it into a dictionary.
    include_dict = dict(
        [item for sublist in include_state.include_list for item in sublist])

    # Did we find the header for this file (if any) and successfully load it?
    header_found = False

    # Use the absolute path so that matching works properly.
    abs_filename = FileInfo(filename).FullName()

    # For Emacs's flymake.
    # If cpplint is invoked from Emacs's flymake, a temporary file is generated
    # by flymake and that file name might end with '_flymake.cc'. In that case,
    # restore original file name here so that the corresponding header file can be
    # found.
    # e.g. If the file name is 'foo_flymake.cc', we should search for 'foo.h'
    # instead of 'foo_flymake.h'
    abs_filename = re.sub(r'_flymake\.cc$', '.cc', abs_filename)

    # include_dict is modified during iteration, so we iterate over a copy of
    # the keys.
    header_keys = include_dict.keys()
    for header in header_keys:
        (same_module, common_path) = FilesBelongToSameModule(abs_filename,
                                                             header)
        fullpath = common_path + header
        if same_module and UpdateIncludeState(fullpath, include_dict, io):
            header_found = True

    # If we can't find the header file for a .cc, assume it's because we don't
    # know where to look. In that case we'll give up as we're not sure they
    # didn't include it in the .h file.
    # TODO(unknown): Do a better job of finding .h files so we are confident that
    # not having the .h file means there isn't one.
    if filename.endswith('.cc') and not header_found:
        return

    # All the lines have been processed, report the errors found.
    for required_header_unstripped in required:
        template = required[required_header_unstripped][1]
        if required_header_unstripped.strip('<>"') not in include_dict:
            error(filename, required[required_header_unstripped][0],
                  'build/include_what_you_use', 4, 'Add #include ' +
                  required_header_unstripped + ' for ' + template)
Z
zhangjinchao01 已提交
5769 5770 5771 5772 5773 5774


_RE_PATTERN_EXPLICIT_MAKEPAIR = re.compile(r'\bmake_pair\s*<')


def CheckMakePairUsesDeduction(filename, clean_lines, linenum, error):
Y
Yu Yang 已提交
5775
    """Check that make_pair's template arguments are deduced.
Z
zhangjinchao01 已提交
5776 5777 5778 5779 5780 5781 5782 5783 5784 5785

  G++ 4.6 in C++11 mode fails badly if make_pair's template arguments are
  specified explicitly, and such use isn't intended in any case.

  Args:
    filename: The name of the current file.
    clean_lines: A CleansedLines instance containing the file.
    linenum: The number of the line to check.
    error: The function to call with any errors found.
  """
Y
Yu Yang 已提交
5786 5787 5788 5789 5790 5791 5792 5793 5794 5795
    line = clean_lines.elided[linenum]
    match = _RE_PATTERN_EXPLICIT_MAKEPAIR.search(line)
    if match:
        error(
            filename,
            linenum,
            'build/explicit_make_pair',
            4,  # 4 = high confidence
            'For C++11-compatibility, omit template arguments from make_pair'
            ' OR use pair directly OR if appropriate, construct a pair directly')
Z
zhangjinchao01 已提交
5796 5797 5798


def CheckDefaultLambdaCaptures(filename, clean_lines, linenum, error):
Y
Yu Yang 已提交
5799
    """Check that default lambda captures are not used.
Z
zhangjinchao01 已提交
5800 5801 5802 5803 5804 5805 5806

  Args:
    filename: The name of the current file.
    clean_lines: A CleansedLines instance containing the file.
    linenum: The number of the line to check.
    error: The function to call with any errors found.
  """
Y
Yu Yang 已提交
5807 5808 5809 5810 5811 5812 5813 5814 5815 5816 5817 5818 5819 5820 5821 5822 5823 5824
    line = clean_lines.elided[linenum]

    # A lambda introducer specifies a default capture if it starts with "[="
    # or if it starts with "[&" _not_ followed by an identifier.
    match = Match(r'^(.*)\[\s*(?:=|&[^\w])', line)
    if match:
        # Found a potential error, check what comes after the lambda-introducer.
        # If it's not open parenthesis (for lambda-declarator) or open brace
        # (for compound-statement), it's not a lambda.
        line, _, pos = CloseExpression(clean_lines, linenum,
                                       len(match.group(1)))
        if pos >= 0 and Match(r'^\s*[{(]', line[pos:]):
            error(
                filename,
                linenum,
                'build/c++11',
                4,  # 4 = high confidence
                'Default lambda captures are an unapproved C++ feature.')
Z
zhangjinchao01 已提交
5825 5826 5827


def CheckRedundantVirtual(filename, clean_lines, linenum, error):
Y
Yu Yang 已提交
5828
    """Check if line contains a redundant "virtual" function-specifier.
Z
zhangjinchao01 已提交
5829 5830 5831 5832 5833 5834 5835

  Args:
    filename: The name of the current file.
    clean_lines: A CleansedLines instance containing the file.
    linenum: The number of the line to check.
    error: The function to call with any errors found.
  """
Y
Yu Yang 已提交
5836 5837 5838 5839 5840 5841 5842 5843 5844 5845 5846
    # Look for "virtual" on current line.
    line = clean_lines.elided[linenum]
    virtual = Match(r'^(.*)(\bvirtual\b)(.*)$', line)
    if not virtual: return

    # Ignore "virtual" keywords that are near access-specifiers.  These
    # are only used in class base-specifier and do not apply to member
    # functions.
    if (Search(r'\b(public|protected|private)\s+$', virtual.group(1)) or
            Match(r'^\s+(public|protected|private)\b', virtual.group(3))):
        return
Z
zhangjinchao01 已提交
5847

Y
Yu Yang 已提交
5848 5849 5850 5851 5852 5853 5854 5855 5856 5857 5858 5859 5860 5861 5862 5863 5864 5865 5866 5867 5868 5869 5870 5871 5872 5873 5874 5875 5876 5877 5878 5879 5880 5881 5882 5883 5884 5885 5886 5887 5888 5889
    # Ignore the "virtual" keyword from virtual base classes.  Usually
    # there is a column on the same line in these cases (virtual base
    # classes are rare in google3 because multiple inheritance is rare).
    if Match(r'^.*[^:]:[^:].*$', line): return

    # Look for the next opening parenthesis.  This is the start of the
    # parameter list (possibly on the next line shortly after virtual).
    # TODO(unknown): doesn't work if there are virtual functions with
    # decltype() or other things that use parentheses, but csearch suggests
    # that this is rare.
    end_col = -1
    end_line = -1
    start_col = len(virtual.group(2))
    for start_line in xrange(linenum, min(linenum + 3, clean_lines.NumLines())):
        line = clean_lines.elided[start_line][start_col:]
        parameter_list = Match(r'^([^(]*)\(', line)
        if parameter_list:
            # Match parentheses to find the end of the parameter list
            (_, end_line, end_col) = CloseExpression(
                clean_lines, start_line,
                start_col + len(parameter_list.group(1)))
            break
        start_col = 0

    if end_col < 0:
        return  # Couldn't find end of parameter list, give up

    # Look for "override" or "final" after the parameter list
    # (possibly on the next few lines).
    for i in xrange(end_line, min(end_line + 3, clean_lines.NumLines())):
        line = clean_lines.elided[i][end_col:]
        match = Search(r'\b(override|final)\b', line)
        if match:
            error(filename, linenum, 'readability/inheritance', 4,
                  ('"virtual" is redundant since function is '
                   'already declared as "%s"' % match.group(1)))

        # Set end_col to check whole lines after we are done with the
        # first line.
        end_col = 0
        if Search(r'[^\w]\s*$', line):
            break
Z
zhangjinchao01 已提交
5890 5891 5892


def CheckRedundantOverrideOrFinal(filename, clean_lines, linenum, error):
Y
Yu Yang 已提交
5893
    """Check if line contains a redundant "override" or "final" virt-specifier.
Z
zhangjinchao01 已提交
5894 5895 5896 5897 5898 5899 5900

  Args:
    filename: The name of the current file.
    clean_lines: A CleansedLines instance containing the file.
    linenum: The number of the line to check.
    error: The function to call with any errors found.
  """
Y
Yu Yang 已提交
5901 5902 5903 5904 5905 5906 5907
    # Look for closing parenthesis nearby.  We need one to confirm where
    # the declarator ends and where the virt-specifier starts to avoid
    # false positives.
    line = clean_lines.elided[linenum]
    declarator_end = line.rfind(')')
    if declarator_end >= 0:
        fragment = line[declarator_end:]
Z
zhangjinchao01 已提交
5908
    else:
Y
Yu Yang 已提交
5909 5910 5911 5912
        if linenum > 1 and clean_lines.elided[linenum - 1].rfind(')') >= 0:
            fragment = line
        else:
            return
Z
zhangjinchao01 已提交
5913

Y
Yu Yang 已提交
5914 5915 5916 5917 5918
    # Check that at most one of "override" or "final" is present, not both
    if Search(r'\boverride\b', fragment) and Search(r'\bfinal\b', fragment):
        error(filename, linenum, 'readability/inheritance', 4,
              ('"override" is redundant since function is '
               'already declared as "final"'))
Z
zhangjinchao01 已提交
5919 5920 5921 5922 5923


# Returns true if we are at a new block, and it is directly
# inside of a namespace.
def IsBlockInNameSpace(nesting_state, is_forward_declaration):
Y
Yu Yang 已提交
5924
    """Checks that the new block is directly in a namespace.
Z
zhangjinchao01 已提交
5925 5926 5927 5928 5929 5930 5931

  Args:
    nesting_state: The _NestingState object that contains info about our state.
    is_forward_declaration: If the class is a forward declared class.
  Returns:
    Whether or not the new block is directly in a namespace.
  """
Y
Yu Yang 已提交
5932 5933 5934 5935 5936 5937
    if is_forward_declaration:
        if len(nesting_state.stack) >= 1 and (
                isinstance(nesting_state.stack[-1], _NamespaceInfo)):
            return True
        else:
            return False
Z
zhangjinchao01 已提交
5938

Y
Yu Yang 已提交
5939 5940 5941
    return (len(nesting_state.stack) > 1 and
            nesting_state.stack[-1].check_namespace_indentation and
            isinstance(nesting_state.stack[-2], _NamespaceInfo))
Z
zhangjinchao01 已提交
5942 5943 5944 5945


def ShouldCheckNamespaceIndentation(nesting_state, is_namespace_indent_item,
                                    raw_lines_no_comments, linenum):
Y
Yu Yang 已提交
5946
    """This method determines if we should apply our namespace indentation check.
Z
zhangjinchao01 已提交
5947 5948 5949 5950 5951 5952 5953 5954 5955 5956 5957 5958 5959 5960

  Args:
    nesting_state: The current nesting state.
    is_namespace_indent_item: If we just put a new class on the stack, True.
      If the top of the stack is not a class, or we did not recently
      add the class, False.
    raw_lines_no_comments: The lines without the comments.
    linenum: The current line number we are processing.

  Returns:
    True if we should apply our namespace indentation check. Currently, it
    only works for classes and namespaces inside of a namespace.
  """

Y
Yu Yang 已提交
5961 5962
    is_forward_declaration = IsForwardClassDeclaration(raw_lines_no_comments,
                                                       linenum)
Z
zhangjinchao01 已提交
5963

Y
Yu Yang 已提交
5964 5965
    if not (is_namespace_indent_item or is_forward_declaration):
        return False
Z
zhangjinchao01 已提交
5966

Y
Yu Yang 已提交
5967 5968 5969
    # If we are in a macro, we do not want to check the namespace indentation.
    if IsMacroDefinition(raw_lines_no_comments, linenum):
        return False
Z
zhangjinchao01 已提交
5970

Y
Yu Yang 已提交
5971
    return IsBlockInNameSpace(nesting_state, is_forward_declaration)
Z
zhangjinchao01 已提交
5972 5973 5974 5975 5976 5977 5978


# Call this method if the line is directly inside of a namespace.
# If the line above is blank (excluding comments) or the start of
# an inner namespace, it cannot be indented.
def CheckItemIndentationInNamespace(filename, raw_lines_no_comments, linenum,
                                    error):
Y
Yu Yang 已提交
5979 5980 5981 5982 5983 5984 5985 5986 5987 5988 5989 5990 5991 5992
    line = raw_lines_no_comments[linenum]
    if Match(r'^\s+', line):
        error(filename, linenum, 'runtime/indentation_namespace', 4,
              'Do not indent within a namespace')


def ProcessLine(filename,
                file_extension,
                clean_lines,
                line,
                include_state,
                function_state,
                nesting_state,
                error,
Z
zhangjinchao01 已提交
5993
                extra_check_functions=[]):
Y
Yu Yang 已提交
5994
    """Processes a single line in the file.
Z
zhangjinchao01 已提交
5995 5996 5997 5998 5999 6000 6001 6002 6003 6004 6005 6006 6007 6008 6009 6010 6011

  Args:
    filename: Filename of the file that is being processed.
    file_extension: The extension (dot not included) of the file.
    clean_lines: An array of strings, each representing a line of the file,
                 with comments stripped.
    line: Number of line being processed.
    include_state: An _IncludeState instance in which the headers are inserted.
    function_state: A _FunctionState instance which counts function lines, etc.
    nesting_state: A NestingState instance which maintains information about
                   the current stack of nested blocks being parsed.
    error: A callable to which errors are reported, which takes 4 arguments:
           filename, line number, error level, and message
    extra_check_functions: An array of additional check functions that will be
                           run on each source line. Each function takes 4
                           arguments: filename, clean_lines, line, error
  """
Y
Yu Yang 已提交
6012 6013 6014 6015 6016 6017 6018 6019 6020 6021 6022 6023 6024 6025 6026 6027 6028 6029 6030 6031 6032 6033 6034 6035 6036
    raw_lines = clean_lines.raw_lines
    ParseNolintSuppressions(filename, raw_lines[line], line, error)
    nesting_state.Update(filename, clean_lines, line, error)
    CheckForNamespaceIndentation(filename, nesting_state, clean_lines, line,
                                 error)
    if nesting_state.InAsmBlock(): return
    CheckForFunctionLengths(filename, clean_lines, line, function_state, error)
    CheckForMultilineCommentsAndStrings(filename, clean_lines, line, error)
    CheckStyle(filename, clean_lines, line, file_extension, nesting_state,
               error)
    CheckLanguage(filename, clean_lines, line, file_extension, include_state,
                  nesting_state, error)
    CheckForNonConstReference(filename, clean_lines, line, nesting_state, error)
    CheckForNonStandardConstructs(filename, clean_lines, line, nesting_state,
                                  error)
    CheckVlogArguments(filename, clean_lines, line, error)
    CheckPosixThreading(filename, clean_lines, line, error)
    CheckInvalidIncrement(filename, clean_lines, line, error)
    CheckMakePairUsesDeduction(filename, clean_lines, line, error)
    CheckDefaultLambdaCaptures(filename, clean_lines, line, error)
    CheckRedundantVirtual(filename, clean_lines, line, error)
    CheckRedundantOverrideOrFinal(filename, clean_lines, line, error)
    for check_fn in extra_check_functions:
        check_fn(filename, clean_lines, line, error)

Z
zhangjinchao01 已提交
6037 6038

def FlagCxx11Features(filename, clean_lines, linenum, error):
Y
Yu Yang 已提交
6039
    """Flag those c++11 features that we only allow in certain places.
Z
zhangjinchao01 已提交
6040 6041 6042 6043 6044 6045 6046

  Args:
    filename: The name of the current file.
    clean_lines: A CleansedLines instance containing the file.
    linenum: The number of the line to check.
    error: The function to call with any errors found.
  """
Y
Yu Yang 已提交
6047 6048 6049 6050 6051 6052 6053 6054 6055 6056 6057 6058 6059 6060 6061 6062 6063 6064 6065 6066 6067 6068 6069 6070 6071 6072 6073 6074 6075 6076 6077 6078 6079 6080 6081 6082 6083 6084 6085 6086
    line = clean_lines.elided[linenum]

    # Flag unapproved C++11 headers.
    include = Match(r'\s*#\s*include\s+[<"]([^<"]+)[">]', line)
    if include and include.group(1) in (
            'cfenv',
            'condition_variable',
            'fenv.h',
            'future',
            'mutex',
            'thread',
            'chrono',
            'ratio',
            'regex',
            'system_error', ):
        error(filename, linenum, 'build/c++11', 5,
              ('<%s> is an unapproved C++11 header.') % include.group(1))

    # The only place where we need to worry about C++11 keywords and library
    # features in preprocessor directives is in macro definitions.
    if Match(r'\s*#', line) and not Match(r'\s*#\s*define\b', line): return

    # These are classes and free functions.  The classes are always
    # mentioned as std::*, but we only catch the free functions if
    # they're not found by ADL.  They're alphabetical by header.
    for top_name in (
            # type_traits
            'alignment_of',
            'aligned_union', ):
        if Search(r'\bstd::%s\b' % top_name, line):
            error(filename, linenum, 'build/c++11', 5, (
                'std::%s is an unapproved C++11 class or function.  Send c-style '
                'an example of where it would make your code more readable, and '
                'they may let you use it.') % top_name)


def ProcessFileData(filename,
                    file_extension,
                    lines,
                    error,
Z
zhangjinchao01 已提交
6087
                    extra_check_functions=[]):
Y
Yu Yang 已提交
6088
    """Performs lint checks and reports any errors to the given error function.
Z
zhangjinchao01 已提交
6089 6090 6091 6092 6093 6094 6095 6096 6097 6098 6099 6100

  Args:
    filename: Filename of the file that is being processed.
    file_extension: The extension (dot not included) of the file.
    lines: An array of strings, each representing a line of the file, with the
           last element being empty if the file is terminated with a newline.
    error: A callable to which errors are reported, which takes 4 arguments:
           filename, line number, error level, and message
    extra_check_functions: An array of additional check functions that will be
                           run on each source line. Each function takes 4
                           arguments: filename, clean_lines, line, error
  """
Y
Yu Yang 已提交
6101 6102
    lines = (['// marker so line numbers and indices both start at 1'] + lines +
             ['// marker so line numbers end in a known way'])
Z
zhangjinchao01 已提交
6103

Y
Yu Yang 已提交
6104 6105 6106
    include_state = _IncludeState()
    function_state = _FunctionState()
    nesting_state = NestingState()
Z
zhangjinchao01 已提交
6107

Y
Yu Yang 已提交
6108
    ResetNolintSuppressions()
Z
zhangjinchao01 已提交
6109

Y
Yu Yang 已提交
6110
    CheckForCopyright(filename, lines, error)
Z
zhangjinchao01 已提交
6111

Y
Yu Yang 已提交
6112 6113
    RemoveMultiLineComments(filename, lines, error)
    clean_lines = CleansedLines(lines)
Z
zhangjinchao01 已提交
6114

Y
Yu Yang 已提交
6115 6116
    if file_extension == 'h':
        CheckForHeaderGuard(filename, clean_lines, error)
Z
zhangjinchao01 已提交
6117

Y
Yu Yang 已提交
6118 6119 6120 6121 6122
    for line in xrange(clean_lines.NumLines()):
        ProcessLine(filename, file_extension, clean_lines, line, include_state,
                    function_state, nesting_state, error, extra_check_functions)
        FlagCxx11Features(filename, clean_lines, line, error)
    nesting_state.CheckCompletedBlocks(filename, error)
Z
zhangjinchao01 已提交
6123

Y
Yu Yang 已提交
6124
    CheckForIncludeWhatYouUse(filename, clean_lines, include_state, error)
Z
zhangjinchao01 已提交
6125

Y
Yu Yang 已提交
6126 6127 6128 6129 6130 6131 6132 6133 6134
    # Check that the .cc file has included its header if it exists.
    if file_extension == 'cc':
        CheckHeaderFileIncluded(filename, include_state, error)

    # We check here rather than inside ProcessLine so that we see raw
    # lines rather than "cleaned" lines.
    CheckForBadCharacters(filename, lines, error)

    CheckForNewlineAtEOF(filename, lines, error)
Z
zhangjinchao01 已提交
6135 6136 6137


def ProcessConfigOverrides(filename):
Y
Yu Yang 已提交
6138
    """ Loads the configuration files and processes the config overrides.
Z
zhangjinchao01 已提交
6139 6140 6141 6142 6143 6144 6145 6146

  Args:
    filename: The name of the file being processed by the linter.

  Returns:
    False if the current |filename| should not be processed further.
  """

Y
Yu Yang 已提交
6147 6148 6149 6150 6151 6152 6153 6154 6155 6156 6157
    abs_filename = os.path.abspath(filename)
    cfg_filters = []
    keep_looking = True
    while keep_looking:
        abs_path, base_name = os.path.split(abs_filename)
        if not base_name:
            break  # Reached the root directory.

        cfg_file = os.path.join(abs_path, "CPPLINT.cfg")
        abs_filename = abs_path
        if not os.path.isfile(cfg_file):
Z
zhangjinchao01 已提交
6158 6159
            continue

Y
Yu Yang 已提交
6160 6161 6162 6163 6164 6165 6166 6167 6168 6169 6170 6171 6172 6173 6174 6175 6176 6177 6178 6179 6180 6181 6182 6183 6184 6185 6186 6187 6188 6189 6190 6191 6192 6193 6194 6195 6196 6197 6198 6199 6200 6201
        try:
            with open(cfg_file) as file_handle:
                for line in file_handle:
                    line, _, _ = line.partition('#')  # Remove comments.
                    if not line.strip():
                        continue

                    name, _, val = line.partition('=')
                    name = name.strip()
                    val = val.strip()
                    if name == 'set noparent':
                        keep_looking = False
                    elif name == 'filter':
                        cfg_filters.append(val)
                    elif name == 'exclude_files':
                        # When matching exclude_files pattern, use the base_name of
                        # the current file name or the directory name we are processing.
                        # For example, if we are checking for lint errors in /foo/bar/baz.cc
                        # and we found the .cfg file at /foo/CPPLINT.cfg, then the config
                        # file's "exclude_files" filter is meant to be checked against "bar"
                        # and not "baz" nor "bar/baz.cc".
                        if base_name:
                            pattern = re.compile(val)
                            if pattern.match(base_name):
                                sys.stderr.write(
                                    'Ignoring "%s": file excluded by "%s". '
                                    'File path component "%s" matches '
                                    'pattern "%s"\n' %
                                    (filename, cfg_file, base_name, val))
                                return False
                    elif name == 'linelength':
                        global _line_length
                        try:
                            _line_length = int(val)
                        except ValueError:
                            sys.stderr.write('Line length must be numeric.')
                    else:
                        sys.stderr.write(
                            'Invalid configuration option (%s) in file %s\n' %
                            (name, cfg_file))

        except IOError:
Z
zhangjinchao01 已提交
6202
            sys.stderr.write(
Y
Yu Yang 已提交
6203 6204 6205
                "Skipping config file '%s': Can't open for reading\n" %
                cfg_file)
            keep_looking = False
Z
zhangjinchao01 已提交
6206

Y
Yu Yang 已提交
6207 6208 6209 6210
    # Apply all the accumulated filters in reverse order (top-level directory
    # config options having the least priority).
    for filter in reversed(cfg_filters):
        _AddFilters(filter)
Z
zhangjinchao01 已提交
6211

Y
Yu Yang 已提交
6212
    return True
Z
zhangjinchao01 已提交
6213 6214 6215


def ProcessFile(filename, vlevel, extra_check_functions=[]):
Y
Yu Yang 已提交
6216
    """Does google-lint on a single file.
Z
zhangjinchao01 已提交
6217 6218 6219 6220 6221 6222 6223 6224 6225 6226 6227 6228

  Args:
    filename: The name of the file to parse.

    vlevel: The level of errors to report.  Every error of confidence
    >= verbose_level will be reported.  0 is a good default.

    extra_check_functions: An array of additional check functions that will be
                           run on each source line. Each function takes 4
                           arguments: filename, clean_lines, line, error
  """

Y
Yu Yang 已提交
6229 6230
    _SetVerboseLevel(vlevel)
    _BackupFilters()
Z
zhangjinchao01 已提交
6231

Y
Yu Yang 已提交
6232 6233 6234 6235 6236 6237 6238 6239 6240 6241 6242 6243 6244 6245 6246 6247 6248 6249 6250 6251 6252 6253 6254 6255 6256 6257 6258 6259 6260 6261 6262 6263 6264 6265 6266 6267 6268 6269 6270 6271 6272 6273 6274 6275 6276 6277
    if not ProcessConfigOverrides(filename):
        _RestoreFilters()
        return

    lf_lines = []
    crlf_lines = []
    try:
        # Support the UNIX convention of using "-" for stdin.  Note that
        # we are not opening the file with universal newline support
        # (which codecs doesn't support anyway), so the resulting lines do
        # contain trailing '\r' characters if we are reading a file that
        # has CRLF endings.
        # If after the split a trailing '\r' is present, it is removed
        # below.
        if filename == '-':
            lines = codecs.StreamReaderWriter(sys.stdin,
                                              codecs.getreader('utf8'),
                                              codecs.getwriter('utf8'),
                                              'replace').read().split('\n')
        else:
            lines = codecs.open(filename, 'r', 'utf8',
                                'replace').read().split('\n')

        # Remove trailing '\r'.
        # The -1 accounts for the extra trailing blank line we get from split()
        for linenum in range(len(lines) - 1):
            if lines[linenum].endswith('\r'):
                lines[linenum] = lines[linenum].rstrip('\r')
                crlf_lines.append(linenum + 1)
            else:
                lf_lines.append(linenum + 1)

    except IOError:
        sys.stderr.write("Skipping input '%s': Can't open for reading\n" %
                         filename)
        _RestoreFilters()
        return

    # Note, if no dot is found, this will give the entire filename as the ext.
    file_extension = filename[filename.rfind('.') + 1:]

    # When reading from stdin, the extension is unknown, so no cpplint tests
    # should rely on the extension.
    if filename != '-' and file_extension not in _valid_extensions:
        sys.stderr.write('Ignoring %s; not a valid file name '
                         '(%s)\n' % (filename, ', '.join(_valid_extensions)))
Z
zhangjinchao01 已提交
6278
    else:
Y
Yu Yang 已提交
6279 6280
        ProcessFileData(filename, file_extension, lines, Error,
                        extra_check_functions)
Z
zhangjinchao01 已提交
6281

Y
Yu Yang 已提交
6282 6283 6284 6285 6286 6287 6288 6289 6290 6291 6292 6293 6294 6295 6296 6297 6298 6299 6300 6301
        # If end-of-line sequences are a mix of LF and CR-LF, issue
        # warnings on the lines with CR.
        #
        # Don't issue any warnings if all lines are uniformly LF or CR-LF,
        # since critique can handle these just fine, and the style guide
        # doesn't dictate a particular end of line sequence.
        #
        # We can't depend on os.linesep to determine what the desired
        # end-of-line sequence should be, since that will return the
        # server-side end-of-line sequence.
        if lf_lines and crlf_lines:
            # Warn on every line with CR.  An alternative approach might be to
            # check whether the file is mostly CRLF or just LF, and warn on the
            # minority, we bias toward LF here since most tools prefer LF.
            for linenum in crlf_lines:
                Error(filename, linenum, 'whitespace/newline', 1,
                      'Unexpected \\r (^M) found; better to use only \\n')

    sys.stdout.write('Done processing %s\n' % filename)
    _RestoreFilters()
Z
zhangjinchao01 已提交
6302 6303 6304


def PrintUsage(message):
Y
Yu Yang 已提交
6305
    """Prints a brief usage string and exits, optionally with an error message.
Z
zhangjinchao01 已提交
6306 6307 6308 6309

  Args:
    message: The optional error message.
  """
Y
Yu Yang 已提交
6310 6311 6312 6313 6314
    sys.stderr.write(_USAGE)
    if message:
        sys.exit('\nFATAL ERROR: ' + message)
    else:
        sys.exit(1)
Z
zhangjinchao01 已提交
6315 6316 6317


def PrintCategories():
Y
Yu Yang 已提交
6318
    """Prints a list of all the error-categories used by error messages.
Z
zhangjinchao01 已提交
6319 6320 6321

  These are the categories used to filter messages via --filter.
  """
Y
Yu Yang 已提交
6322 6323
    sys.stderr.write(''.join('  %s\n' % cat for cat in _ERROR_CATEGORIES))
    sys.exit(0)
Z
zhangjinchao01 已提交
6324 6325 6326


def ParseArguments(args):
Y
Yu Yang 已提交
6327
    """Parses the command line arguments.
Z
zhangjinchao01 已提交
6328 6329 6330 6331 6332 6333 6334 6335 6336

  This may set the output format and verbosity level as side-effects.

  Args:
    args: The command line arguments:

  Returns:
    The list of filenames to lint.
  """
Y
Yu Yang 已提交
6337 6338 6339 6340 6341 6342 6343 6344 6345 6346 6347 6348 6349 6350 6351 6352 6353 6354 6355 6356 6357 6358 6359 6360 6361 6362 6363 6364 6365 6366 6367 6368 6369 6370 6371 6372 6373 6374 6375 6376 6377 6378 6379 6380 6381 6382 6383 6384 6385 6386 6387 6388 6389 6390 6391 6392 6393 6394
    try:
        (opts, filenames) = getopt.getopt(args, '', [
            'help', 'output=', 'verbose=', 'counting=', 'filter=', 'root=',
            'linelength=', 'extensions='
        ])
    except getopt.GetoptError:
        PrintUsage('Invalid arguments.')

    verbosity = _VerboseLevel()
    output_format = _OutputFormat()
    filters = ''
    counting_style = ''

    for (opt, val) in opts:
        if opt == '--help':
            PrintUsage(None)
        elif opt == '--output':
            if val not in ('emacs', 'vs7', 'eclipse'):
                PrintUsage(
                    'The only allowed output formats are emacs, vs7 and eclipse.'
                )
            output_format = val
        elif opt == '--verbose':
            verbosity = int(val)
        elif opt == '--filter':
            filters = val
            if not filters:
                PrintCategories()
        elif opt == '--counting':
            if val not in ('total', 'toplevel', 'detailed'):
                PrintUsage(
                    'Valid counting options are total, toplevel, and detailed')
            counting_style = val
        elif opt == '--root':
            global _root
            _root = val
        elif opt == '--linelength':
            global _line_length
            try:
                _line_length = int(val)
            except ValueError:
                PrintUsage('Line length must be digits.')
        elif opt == '--extensions':
            global _valid_extensions
            try:
                _valid_extensions = set(val.split(','))
            except ValueError:
                PrintUsage('Extensions must be comma seperated list.')

    if not filenames:
        PrintUsage('No files were specified.')

    _SetOutputFormat(output_format)
    _SetVerboseLevel(verbosity)
    _SetFilters(filters)
    _SetCountingStyle(counting_style)

    return filenames
Z
zhangjinchao01 已提交
6395 6396 6397


def main():
Y
Yu Yang 已提交
6398
    filenames = ParseArguments(sys.argv[1:])
Z
zhangjinchao01 已提交
6399

Y
Yu Yang 已提交
6400 6401 6402 6403 6404
    # Change stderr to write with replacement characters so we don't die
    # if we try to print something containing non-ASCII characters.
    sys.stderr = codecs.StreamReaderWriter(sys.stderr,
                                           codecs.getreader('utf8'),
                                           codecs.getwriter('utf8'), 'replace')
Z
zhangjinchao01 已提交
6405

Y
Yu Yang 已提交
6406 6407 6408 6409
    _cpplint_state.ResetErrorCounts()
    for filename in filenames:
        ProcessFile(filename, _cpplint_state.verbose_level)
    _cpplint_state.PrintErrorCounts()
Z
zhangjinchao01 已提交
6410

Y
Yu Yang 已提交
6411
    sys.exit(_cpplint_state.error_count > 0)
Z
zhangjinchao01 已提交
6412 6413 6414


if __name__ == '__main__':
Y
Yu Yang 已提交
6415
    main()