get_pr_ut.py 15.4 KB
Newer Older
C
chalsliu 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
# Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
""" For the PR that only modified the unit test, get cases in pull request. """

import os
import json
18 19
import re
import sys
20 21
import time
import subprocess
22
import requests
23 24
import urllib.request
import ssl
Y
YUNSHEN XIE 已提交
25
import platform
C
chalsliu 已提交
26 27 28
from github import Github

PADDLE_ROOT = os.getenv('PADDLE_ROOT', '/paddle/')
C
chalsliu 已提交
29 30
PADDLE_ROOT += '/'
PADDLE_ROOT = PADDLE_ROOT.replace('//', '/')
31
ssl._create_default_https_context = ssl._create_unverified_context
C
chalsliu 已提交
32 33 34 35 36 37 38 39


class PRChecker(object):
    """ PR Checker. """

    def __init__(self):
        self.github = Github(os.getenv('GITHUB_API_TOKEN'), timeout=60)
        self.repo = self.github.get_repo('PaddlePaddle/Paddle')
40
        self.py_prog_oneline = re.compile('\d+\|\s*#.*')
C
chalsliu 已提交
41 42
        self.py_prog_multiline_a = re.compile('\d+\|\s*r?""".*?"""', re.DOTALL)
        self.py_prog_multiline_b = re.compile("\d+\|\s*r?'''.*?'''", re.DOTALL)
43 44 45
        self.cc_prog_online = re.compile('\d+\|\s*//.*')
        self.cc_prog_multiline = re.compile('\d+\|\s*/\*.*?\*/', re.DOTALL)
        self.lineno_prog = re.compile('@@ \-\d+,\d+ \+(\d+),(\d+) @@')
C
chalsliu 已提交
46
        self.pr = None
47
        self.suffix = ''
C
chalsliu 已提交
48
        self.full_case = False
C
chalsliu 已提交
49 50 51 52 53

    def init(self):
        """ Get pull request. """
        pr_id = os.getenv('GIT_PR_ID')
        if not pr_id:
54
            print('PREC No PR ID')
C
chalsliu 已提交
55
            exit(0)
56 57 58
        suffix = os.getenv('PREC_SUFFIX')
        if suffix:
            self.suffix = suffix
C
chalsliu 已提交
59
        self.pr = self.repo.get_pull(int(pr_id))
C
chalsliu 已提交
60 61 62 63 64 65 66 67 68
        last_commit = None
        ix = 0
        while True:
            commits = self.pr.get_commits().get_page(ix)
            for c in commits:
                last_commit = c.commit
            else:
                break
            ix = ix + 1
69 70
        if last_commit.message.find('test=allcase') != -1:
            print('PREC test=allcase is set')
C
chalsliu 已提交
71
            self.full_case = True
C
chalsliu 已提交
72

73 74 75 76 77 78 79 80
    #todo: exception
    def __wget_with_retry(self, url):
        ix = 1
        proxy = '--no-proxy'
        while ix < 6:
            if ix // 2 == 0:
                proxy = ''
            else:
81 82 83 84
                if platform.system() == 'Windows':
                    proxy = '-Y off'
                else:
                    proxy = '--no-proxy'
85 86 87 88 89 90 91 92 93 94 95 96
            code = subprocess.call(
                'wget -q {} --no-check-certificate {}'.format(proxy, url),
                shell=True)
            if code == 0:
                return True
            print(
                'PREC download {} error, retry {} time(s) after {} secs.[proxy_option={}]'.
                format(url, ix, ix * 10, proxy))
            time.sleep(ix * 10)
            ix += 1
        return False

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
    def __urlretrieve(self, url, filename):
        ix = 1
        with_proxy = urllib.request.getproxies()
        without_proxy = {'http': '', 'http': ''}
        while ix < 6:
            if ix // 2 == 0:
                cur_proxy = urllib.request.ProxyHandler(without_proxy)
            else:
                cur_proxy = urllib.request.ProxyHandler(with_proxy)
            opener = urllib.request.build_opener(cur_proxy,
                                                 urllib.request.HTTPHandler)
            urllib.request.install_opener(opener)
            try:
                urllib.request.urlretrieve(url, filename)
            except Exception as e:
                print(e)
                print(
                    'PREC download {} error, retry {} time(s) after {} secs.[proxy_option={}]'.
                    format(url, ix, ix * 10, proxy))
                continue
            else:
                return True
            time.sleep(ix * 10)
            ix += 1

        return False

C
chalsliu 已提交
124 125 126
    def get_pr_files(self):
        """ Get files in pull request. """
        page = 0
Z
zhangchunle 已提交
127
        file_dict = {}
C
chalsliu 已提交
128 129 130 131 132
        while True:
            files = self.pr.get_files().get_page(page)
            if not files:
                break
            for f in files:
Z
zhangchunle 已提交
133
                file_dict[PADDLE_ROOT + f.filename] = f.status
C
chalsliu 已提交
134
            page += 1
Z
zhangchunle 已提交
135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154
        print("pr modify files: %s" % file_dict)
        return file_dict

    def get_is_white_file(self, filename):
        """ judge is white file in pr's files. """
        isWhiteFile = False
        white_files = (PADDLE_ROOT + 'cmake/', PADDLE_ROOT + 'patches/',
                       PADDLE_ROOT + 'tools/dockerfile/',
                       PADDLE_ROOT + 'tools/windows/',
                       PADDLE_ROOT + 'tools/test_runner.py',
                       PADDLE_ROOT + 'tools/parallel_UT_rule.py',
                       PADDLE_ROOT + 'paddle/scripts/paddle_build.sh',
                       PADDLE_ROOT + 'paddle/scripts/paddle_build.bat')
        if 'cmakelist' in filename.lower():
            isWhiteFile = False
        elif filename.startswith((white_files)):
            isWhiteFile = False
        else:
            isWhiteFile = True
        return isWhiteFile
C
chalsliu 已提交
155

156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172
    def __get_comment_by_filetype(self, content, filetype):
        result = []
        if filetype == 'py':
            result = self.__get_comment_by_prog(content, self.py_prog_oneline)
            result.extend(
                self.__get_comment_by_prog(content, self.py_prog_multiline_a))
            result.extend(
                self.__get_comment_by_prog(content, self.py_prog_multiline_b))
        if filetype == 'cc':
            result = self.__get_comment_by_prog(content, self.cc_prog_oneline)
            result.extend(
                self.__get_comment_by_prog(content, self.cc_prog_multiline))
        return result

    def __get_comment_by_prog(self, content, prog):
        result_list = prog.findall(content)
        if not result_list:
C
chalsliu 已提交
173 174
            return []
        result = []
175 176 177 178 179 180
        for u in result_list:
            result.extend(u.split('\n'))
        return result

    def get_comment_of_file(self, f):
        #content = self.repo.get_contents(f.replace(PADDLE_ROOT, ''), 'pull/').decoded_content
181
        #todo: get file from github
182 183 184 185 186 187 188 189 190 191 192 193 194 195 196
        with open(f) as fd:
            lines = fd.readlines()
        lineno = 1
        inputs = ''
        for line in lines:
            #for line in content.split('\n'):
            #input += str(lineno) + '|' + line + '\n'
            inputs += str(lineno) + '|' + line
            lineno += 1
        fietype = ''
        if f.endswith('.h') or f.endswith('.cc') or f.endswith('.cu'):
            filetype = 'cc'
        if f.endswith('.py'):
            filetype = 'py'
        else:
C
chalsliu 已提交
197
            return []
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
        return self.__get_comment_by_filetype(inputs, filetype)

    def get_pr_diff_lines(self):
        file_to_diff_lines = {}
        r = requests.get(self.pr.diff_url)
        data = r.text
        data = data.split('\n')
        ix = 0
        while ix < len(data):
            if data[ix].startswith('+++'):
                if data[ix].rstrip('\r\n') == '+++ /dev/null':
                    ix += 1
                    continue
                filename = data[ix][6:]
                ix += 1
                while ix < len(data):
                    result = self.lineno_prog.match(data[ix])
                    if not result:
                        break
                    lineno = int(result.group(1))
                    length = int(result.group(2))
                    ix += 1
                    end = ix + length
                    while ix < end:
                        if data[ix][0] == '-':
                            end += 1
                        if data[ix][0] == '+':
                            line_list = file_to_diff_lines.get(filename)
C
chalsliu 已提交
226 227
                            line = '{}{}'.format(lineno,
                                                 data[ix].replace('+', '|', 1))
228 229 230 231 232 233 234 235 236 237 238 239 240
                            if line_list:
                                line_list.append(line)
                            else:
                                file_to_diff_lines[filename] = [line, ]
                        if data[ix][0] != '-':
                            lineno += 1
                        ix += 1
            ix += 1
        return file_to_diff_lines

    def is_only_comment(self, f):
        file_to_diff_lines = self.get_pr_diff_lines()
        comment_lines = self.get_comment_of_file(f)
C
chalsliu 已提交
241 242 243
        diff_lines = file_to_diff_lines.get(f.replace(PADDLE_ROOT, '', 1))
        if not diff_lines:
            return False
244 245 246
        for l in diff_lines:
            if l not in comment_lines:
                return False
247
        print('PREC {} is only comment'.format(f))
248 249
        return True

Z
zhangchunle 已提交
250 251
    def get_all_count(self):
        os.system(
252
            "cd %sbuild && ctest -N|grep 'Total Tests:' | awk -F ': ' '{print $2}' > testCount"
Z
zhangchunle 已提交
253
            % PADDLE_ROOT)
254
        f = open("%sbuild/testCount" % PADDLE_ROOT)
Z
zhangchunle 已提交
255 256 257 258
        testCount = f.read()
        f.close()
        return int(testCount.strip())

C
chalsliu 已提交
259 260
    def get_pr_ut(self):
        """ Get unit tests in pull request. """
C
chalsliu 已提交
261 262
        if self.full_case:
            return ''
C
chalsliu 已提交
263
        check_added_ut = False
C
chalsliu 已提交
264 265
        ut_list = []
        file_ut_map = None
Z
zhangchunle 已提交
266

267
        ret = self.__urlretrieve(
Z
zhangchunle 已提交
268 269
            'https://paddle-docker-tar.bj.bcebos.com/pre_test/ut_file_map.json',
            'ut_file_map.json')
270 271 272
        if not ret:
            print('PREC download file_ut.json failed')
            exit(1)
Z
zhangchunle 已提交
273

Z
zhangchunle 已提交
274
        with open('ut_file_map.json') as jsonfile:
C
chalsliu 已提交
275
            file_ut_map = json.load(jsonfile)
Z
zhangchunle 已提交
276 277 278

        current_system = platform.system()
        notHitMapFiles = []
Z
zhangchunle 已提交
279
        hitMapFiles = {}
Z
zhangchunle 已提交
280
        onlyCommentsFilesOrXpu = []
Z
zhangchunle 已提交
281 282 283 284 285 286 287 288 289 290 291 292 293 294 295
        filterFiles = []
        file_list = []
        file_dict = self.get_pr_files()
        for filename in file_dict:
            if filename.startswith(
                (PADDLE_ROOT + 'python/', PADDLE_ROOT + 'paddle/fluid/')):
                file_list.append(filename)
            else:
                isWhiteFile = self.get_is_white_file(filename)
                if isWhiteFile == False:
                    file_list.append(filename)
                else:
                    filterFiles.append(filename)
        if len(file_list) == 0:
            ut_list.append('filterfiles_placeholder')
Z
zhangchunle 已提交
296 297 298 299 300 301 302 303 304 305 306 307
            ret = self.__urlretrieve(
                'https://paddle-docker-tar.bj.bcebos.com/pre_test/prec_delta',
                'prec_delta')
            if ret:
                with open('prec_delta') as delta:
                    for ut in delta:
                        ut_list.append(ut.rstrip('\r\n'))
            else:
                print('PREC download prec_delta failed')
                exit(1)
            PRECISION_TEST_Cases_ratio = format(
                float(len(ut_list)) / float(self.get_all_count()), '.2f')
Z
zhangchunle 已提交
308 309
            print("filterFiles: %s" % filterFiles)
            print("ipipe_log_param_PRECISION_TEST: true")
Z
zhangchunle 已提交
310 311 312 313
            print("ipipe_log_param_PRECISION_TEST_Cases_count: %s" %
                  len(ut_list))
            print("ipipe_log_param_PRECISION_TEST_Cases_ratio: %s" %
                  PRECISION_TEST_Cases_ratio)
Z
zhangchunle 已提交
314
            return '\n'.join(ut_list)
Z
zhangchunle 已提交
315 316 317 318 319 320 321 322 323 324
        else:
            for f in file_list:
                if current_system == "Darwin" or current_system == "Windows" or self.suffix == ".py3":
                    f_judge = f.replace(PADDLE_ROOT, '/paddle/', 1)
                    f_judge = f_judge.replace('//', '/')
                else:
                    f_judge = f
                if f_judge not in file_ut_map:
                    if f_judge.endswith('.md'):
                        ut_list.append('md_placeholder')
Z
zhangchunle 已提交
325
                        onlyCommentsFilesOrXpu.append(f_judge)
Z
zhangchunle 已提交
326 327 328 329 330 331 332
                    elif 'tests/unittests/xpu' in f_judge or 'tests/unittests/npu' in f_judge:
                        ut_list.append('xpu_npu_placeholder')
                        onlyCommentsFilesOrXpu.append(f_judge)
                    elif f_judge.endswith(('.h', '.cu', '.cc', 'py')):
                        if f_judge.find('test_') != -1 or f_judge.find(
                                '_test') != -1:
                            check_added_ut = True
Z
zhangchunle 已提交
333 334 335 336 337 338 339 340 341 342 343 344 345 346
                        if file_dict[f] not in ['removed']:
                            if self.is_only_comment(f):
                                ut_list.append('comment_placeholder')
                                onlyCommentsFilesOrXpu.append(f_judge)
                            else:
                                notHitMapFiles.append(f_judge)
                        else:
                            print("remove file not hit mapFiles: %s" % f_judge)
                    else:
                        notHitMapFiles.append(f_judge) if file_dict[
                            f] != 'removed' else print(
                                "remove file not hit mapFiles: %s" % f_judge)
                else:
                    if file_dict[f] not in ['removed']:
Z
zhangchunle 已提交
347 348 349 350
                        if self.is_only_comment(f):
                            ut_list.append('comment_placeholder')
                            onlyCommentsFilesOrXpu.append(f_judge)
                        else:
Z
zhangchunle 已提交
351 352
                            hitMapFiles[f_judge] = len(file_ut_map[f_judge])
                            ut_list.extend(file_ut_map.get(f_judge))
353
                    else:
Z
zhangchunle 已提交
354
                        hitMapFiles[f_judge] = len(file_ut_map[f_judge])
Z
zhangchunle 已提交
355
                        ut_list.extend(file_ut_map.get(f_judge))
Z
zhangchunle 已提交
356

Z
zhangchunle 已提交
357 358 359 360
            ut_list = list(set(ut_list))
            if len(notHitMapFiles) != 0:
                print("ipipe_log_param_PRECISION_TEST: false")
                print("notHitMapFiles: %s" % notHitMapFiles)
Z
zhangchunle 已提交
361 362
                if len(filterFiles) != 0:
                    print("filterFiles: %s" % filterFiles)
Z
zhangchunle 已提交
363
                return ''
C
chalsliu 已提交
364
            else:
Z
zhangchunle 已提交
365 366 367
                if check_added_ut:
                    with open('{}/added_ut'.format(PADDLE_ROOT)) as utfile:
                        for ut in utfile:
Z
zhangchunle 已提交
368
                            ut_list.append(ut.rstrip('\r\n'))
Z
zhangchunle 已提交
369 370 371 372 373 374 375 376 377 378 379
                if ut_list:
                    ret = self.__urlretrieve(
                        'https://paddle-docker-tar.bj.bcebos.com/pre_test/prec_delta',
                        'prec_delta')
                    if ret:
                        with open('prec_delta') as delta:
                            for ut in delta:
                                ut_list.append(ut.rstrip('\r\n'))
                    else:
                        print('PREC download prec_delta failed')
                        exit(1)
Z
zhangchunle 已提交
380
                    print("hitMapFiles: %s" % hitMapFiles)
Z
zhangchunle 已提交
381 382 383 384 385 386 387 388
                    print("ipipe_log_param_PRECISION_TEST: true")
                    print("ipipe_log_param_PRECISION_TEST_Cases_count: %s" %
                          len(ut_list))
                    PRECISION_TEST_Cases_ratio = format(
                        float(len(ut_list)) / float(self.get_all_count()),
                        '.2f')
                    print("ipipe_log_param_PRECISION_TEST_Cases_ratio: %s" %
                          PRECISION_TEST_Cases_ratio)
Z
zhangchunle 已提交
389 390
                    if len(filterFiles) != 0:
                        print("filterFiles: %s" % filterFiles)
Z
zhangchunle 已提交
391
                return '\n'.join(ut_list)
C
chalsliu 已提交
392 393 394 395 396


if __name__ == '__main__':
    pr_checker = PRChecker()
    pr_checker.init()
397 398
    with open('ut_list', 'w') as f:
        f.write(pr_checker.get_pr_ut())