bfile.py 25.7 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
# Copyright (c) 2020 VisualDL Authors. All Rights Reserve.
#
# 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.
# =======================================================================

import os
import tempfile
走神的阿圆's avatar
走神的阿圆 已提交
18 19
import hashlib
import base64
走神的阿圆's avatar
走神的阿圆 已提交
20
import time
走神的阿圆's avatar
走神的阿圆 已提交
21 22 23 24 25 26 27 28 29 30

try:
    import hdfs
    from hdfs.util import HdfsError
    HDFS_ENABLED = True
except ImportError:
    HDFS_ENABLED = False
try:
    from baidubce.services.bos.bos_client import BosClient
    from baidubce import exception
走神的阿圆's avatar
走神的阿圆 已提交
31 32
    from baidubce.bce_client_configuration import BceClientConfiguration
    from baidubce.auth.bce_credentials import BceCredentials
走神的阿圆's avatar
走神的阿圆 已提交
33 34 35
    BOS_ENABLED = True
except ImportError:
    BOS_ENABLED = False
36 37 38 39 40 41 42

# Note: Some codes here refer to TensorBoardX.
# A good default block size depends on the system in question.
# A somewhat conservative default chosen here.
_DEFAULT_BLOCK_SIZE = 16 * 1024 * 1024


走神的阿圆's avatar
走神的阿圆 已提交
43 44 45 46 47 48
def content_md5(buffer):
    md5 = hashlib.md5()
    md5.update(buffer)
    return base64.standard_b64encode(md5.digest())


49 50 51 52 53 54 55 56
class FileFactory(object):
    def __init__(self):
        self._register_factories = {}

    def register_filesystem(self, path, filesystem):
        self._register_factories.update({path: filesystem})

    def get_filesystem(self, path):
走神的阿圆's avatar
走神的阿圆 已提交
57 58
        if path.startswith(
                'hdfs://') and "hdfs" not in self._register_factories:
走神的阿圆's avatar
走神的阿圆 已提交
59 60
            if not HDFS_ENABLED:
                raise RuntimeError('Please install module named "hdfs".')
走神的阿圆's avatar
走神的阿圆 已提交
61
            try:
62 63
                default_file_factory.register_filesystem(
                    "hdfs", HDFileSystem())
走神的阿圆's avatar
走神的阿圆 已提交
64 65 66
            except hdfs.util.HdfsError:
                raise RuntimeError(
                    "Please initialize `~/.hdfscli.cfg` for HDFS.")
走神的阿圆's avatar
走神的阿圆 已提交
67 68 69 70 71 72 73
        elif path.startswith(
                'bos://') and "bos" not in self._register_factories:
            if not BOS_ENABLED:
                raise RuntimeError(
                    'Please install module named "bce-python-sdk".')
            default_file_factory.register_filesystem("bos", BosFileSystem())

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
        prefix = ""
        index = path.find("://")
        if index >= 0:
            prefix = path[:index]
        fs = self._register_factories.get(prefix, None)
        if fs is None:
            raise ValueError("No recognized filesystem for prefix %s" % prefix)
        return fs


default_file_factory = FileFactory()


class LocalFileSystem(object):
    def __init__(self):
        pass

    @staticmethod
    def exists(path):
        return os.path.exists(path)

    @staticmethod
    def makedirs(path):
        os.makedirs(path)

    @staticmethod
    def join(path, *paths):
        return os.path.join(path, *paths)

走神的阿圆's avatar
走神的阿圆 已提交
103 104 105 106 107 108 109 110 111
    def isfile(self, filename):
        return os.path.isfile(filename)

    def read_file(self, filename, binary_mode=True):
        mode = "rb" if binary_mode else "r"
        with open(filename, mode) as reader:
            data = reader.read()
        return data

112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130
    def read(self, filename, binary_mode=False, size=None, continue_from=None):
        mode = "rb" if binary_mode else "r"
        encoding = None if binary_mode else "utf-8"
        offset = None
        if continue_from is not None:
            offset = continue_from.get("last_offset", None)
        with open(filename, mode=mode, encoding=encoding) as fp:
            if offset is not None:
                fp.seek(offset)
            data = fp.read(size)
            continue_from_token = {"last_offset": fp.tell()}
            return data, continue_from_token

    def _write(self, filename, file_content, mode):
        encoding = None if "b" in mode else "utf-8"
        with open(filename, mode, encoding=encoding) as fp:
            fp.write(file_content)

    def append(self, filename, file_content, binary_mode=False):
131 132 133 134
        try:
            self._write(filename, file_content, "ab" if binary_mode else "a")
        except FileNotFoundError:
            self.makedirs(os.path.dirname(filename))
135 136

    def write(self, filename, file_content, binary_mode=False):
137 138 139 140
        try:
            self._write(filename, file_content, "ab" if binary_mode else "a")
        except FileNotFoundError:
            self.makedirs(os.path.dirname(filename))
141
        # self._write(filename, file_content, "wb" if binary_mode else "w")
142 143

    def walk(self, dir):
144 145
        if 'posix' == os.name:
            return os.walk(dir, followlinks=True)
146 147 148 149 150 151
        return os.walk(dir)


default_file_factory.register_filesystem("", LocalFileSystem())


152 153 154 155 156 157 158 159 160 161
class HDFileSystem(object):
    def __init__(self):
        self.cli = hdfs.config.Config().get_client('dev')

    def exists(self, path):
        if self.cli.status(hdfs_path=path[7:], strict=False) is None:
            return False
        else:
            return True

走神的阿圆's avatar
走神的阿圆 已提交
162 163 164 165 166 167 168 169
    def isfile(self, filename):
        return exists(filename)

    def read_file(self, filename, binary_mode=True):
        with self.cli.read(hdfs_path=filename[7:]) as reader:
            data = reader.read()
        return data

170 171 172 173 174
    def makedirs(self, path):
        self.cli.makedirs(hdfs_path=path[7:])

    @staticmethod
    def join(path, *paths):
走神的阿圆's avatar
走神的阿圆 已提交
175 176 177
        result = os.path.join(path, *paths)
        result.replace('\\', '/')
        return result
178 179 180 181 182 183 184

    def read(self, filename, binary_mode=False, size=0, continue_from=None):
        offset = 0
        if continue_from is not None:
            offset = continue_from.get("last_offset", 0)

        encoding = None if binary_mode else "utf-8"
走神的阿圆's avatar
走神的阿圆 已提交
185
        try:
186 187 188
            with self.cli.read(
                    hdfs_path=filename[7:], offset=offset,
                    encoding=encoding) as reader:
走神的阿圆's avatar
走神的阿圆 已提交
189 190 191 192 193
                data = reader.read()
                continue_from_token = {"last_offset": offset + len(data)}
                return data, continue_from_token
        except HdfsError:
            raise EOFError('No more events to read on HDFS.')
194 195 196 197 198

    def append(self, filename, file_content, binary_mode=False):
        self.cli.write(hdfs_path=filename[7:], data=file_content, append=True)

    def write(self, filename, file_content, binary_mode=False):
199 200
        self.cli.write(hdfs_path=filename[7:], data=file_content, append=True)
        # self.cli.write(hdfs_path=filename[7:], data=file_content)
201 202 203

    def walk(self, dir):
        walks = self.cli.walk(hdfs_path=dir[7:])
走神的阿圆's avatar
走神的阿圆 已提交
204
        return (['hdfs://' + root, dirs, files] for root, dirs, files in walks)
205 206


走神的阿圆's avatar
走神的阿圆 已提交
207 208 209 210 211 212 213 214 215 216 217 218
def get_object_info(path):
    path = path[6:]
    index = path.index('/')
    bucket_name = path[0:index]
    object_key = path[index + 1:]
    return bucket_name, object_key


class BosConfigClient(object):
    def __init__(self, bos_ak, bos_sk, bos_sts, bos_host="bj.bcebos.com"):
        self.config = BceClientConfiguration(
            credentials=BceCredentials(bos_ak, bos_sk),
219 220
            endpoint=bos_host,
            security_token=bos_sts)
走神的阿圆's avatar
走神的阿圆 已提交
221
        self.bos_client = BosClient(self.config)
走神的阿圆's avatar
走神的阿圆 已提交
222

走神的阿圆's avatar
走神的阿圆 已提交
223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239
    def exists(self, path):
        bucket_name, object_key = get_object_info(path)
        try:
            self.bos_client.get_object_meta_data(bucket_name, object_key)
            return True
        except exception.BceError:
            return False

    def makedirs(self, path):
        if not path.endswith('/'):
            path += '/'
        if self.exists(path):
            return
        bucket_name, object_key = get_object_info(path)
        if not object_key.endswith('/'):
            object_key += '/'
        init_data = b''
240 241 242 243 244 245
        self.bos_client.append_object(
            bucket_name=bucket_name,
            key=object_key,
            data=init_data,
            content_md5=content_md5(init_data),
            content_length=len(init_data))
走神的阿圆's avatar
走神的阿圆 已提交
246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261

    @staticmethod
    def join(path, *paths):
        result = os.path.join(path, *paths)
        result.replace('\\', '/')
        return result

    def upload_object_from_file(self, path, filename):
        if not self.exists(path):
            self.makedirs(path)
        bucket_name, object_key = get_object_info(path)

        object_key = self.join(object_key, filename)
        # if not object_key.endswith('/'):
        #     object_key += '/'
        print('Uploading file `%s`' % filename)
262 263
        self.bos_client.put_object_from_file(
            bucket=bucket_name, key=object_key, file_name=filename)
走神的阿圆's avatar
走神的阿圆 已提交
264 265 266 267 268 269 270 271 272 273 274 275 276 277


class BosFileSystem(object):
    def __init__(self, write_flag=True):
        if write_flag:
            self.max_contents_count = 1
            self.max_contents_time = 1
            self.get_bos_config()
            self.bos_client = BosClient(self.config)
            self.file_length_map = {}

            self._file_contents_to_add = b''
            self._file_contents_count = 0
            self._start_append_time = time.time()
走神的阿圆's avatar
走神的阿圆 已提交
278 279 280 281 282 283 284 285 286 287 288 289 290

    def get_bos_config(self):
        bos_host = os.getenv("BOS_HOST")
        if not bos_host:
            raise KeyError('${BOS_HOST} is not found.')
        access_key_id = os.getenv("BOS_AK")
        if not access_key_id:
            raise KeyError('${BOS_AK} is not found.')
        secret_access_key = os.getenv("BOS_SK")
        if not secret_access_key:
            raise KeyError('${BOS_SK} is not found.')
        self.max_contents_count = int(os.getenv('BOS_CACHE_COUNT', 1))
        self.max_contents_time = int(os.getenv('BOS_CACHE_TIME', 1))
走神的阿圆's avatar
走神的阿圆 已提交
291
        bos_sts = os.getenv("BOS_STS")
走神的阿圆's avatar
走神的阿圆 已提交
292 293
        self.config = BceClientConfiguration(
            credentials=BceCredentials(access_key_id, secret_access_key),
294 295
            endpoint=bos_host,
            security_token=bos_sts)
走神的阿圆's avatar
走神的阿圆 已提交
296

297 298
    def set_bos_config(self, bos_ak, bos_sk, bos_sts,
                       bos_host="bj.bcebos.com"):
走神的阿圆's avatar
走神的阿圆 已提交
299 300
        self.config = BceClientConfiguration(
            credentials=BceCredentials(bos_ak, bos_sk),
301 302
            endpoint=bos_host,
            security_token=bos_sts)
走神的阿圆's avatar
走神的阿圆 已提交
303 304
        self.bos_client = BosClient(self.config)

305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323
    def renew_bos_client_from_server(self):
        import requests
        import json
        from visualdl.utils.dir import CONFIG_PATH
        with open(CONFIG_PATH, 'r') as fp:
            server_url = json.load(fp)['server_url']
        url = server_url + '/sts/'
        res = requests.post(url=url).json()
        err_code = res.get('code')
        msg = res.get('msg')
        if '000000' == err_code:
            sts_ak = msg.get('sts_ak')
            sts_sk = msg.get('sts_sk')
            sts_token = msg.get('token')
            self.set_bos_config(sts_ak, sts_sk, sts_token)
        else:
            print('Renew bos client error. Error msg: {}'.format(msg))
            return

走神的阿圆's avatar
走神的阿圆 已提交
324 325 326 327
    def isfile(self, filename):
        return exists(filename)

    def read_file(self, filename, binary=True):
走神的阿圆's avatar
走神的阿圆 已提交
328
        bucket_name, object_key = get_object_info(filename)
329 330
        result = self.bos_client.get_object_as_string(bucket_name, object_key)
        return result
走神的阿圆's avatar
走神的阿圆 已提交
331

走神的阿圆's avatar
走神的阿圆 已提交
332
    def exists(self, path):
走神的阿圆's avatar
走神的阿圆 已提交
333
        bucket_name, object_key = get_object_info(path)
走神的阿圆's avatar
走神的阿圆 已提交
334 335 336 337 338 339 340 341 342 343 344 345 346 347
        try:
            self.bos_client.get_object_meta_data(bucket_name, object_key)
            return True
        except exception.BceError:
            return False

    def get_meta(self, bucket_name, object_key):
        return self.bos_client.get_object_meta_data(bucket_name, object_key)

    def makedirs(self, path):
        if not path.endswith('/'):
            path += '/'
        if self.exists(path):
            return
走神的阿圆's avatar
走神的阿圆 已提交
348
        bucket_name, object_key = get_object_info(path)
走神的阿圆's avatar
走神的阿圆 已提交
349 350 351
        if not object_key.endswith('/'):
            object_key += '/'
        init_data = b''
352 353 354 355 356 357
        self.bos_client.append_object(
            bucket_name=bucket_name,
            key=object_key,
            data=init_data,
            content_md5=content_md5(init_data),
            content_length=len(init_data))
走神的阿圆's avatar
走神的阿圆 已提交
358 359 360 361 362 363 364 365

    @staticmethod
    def join(path, *paths):
        result = os.path.join(path, *paths)
        result.replace('\\', '/')
        return result

    def read(self, filename, binary_mode=False, size=0, continue_from=None):
走神的阿圆's avatar
走神的阿圆 已提交
366
        bucket_name, object_key = get_object_info(filename)
走神的阿圆's avatar
走神的阿圆 已提交
367 368 369
        offset = 0
        if continue_from is not None:
            offset = continue_from.get("last_offset", 0)
走神的阿圆's avatar
走神的阿圆 已提交
370 371 372
        length = int(
            self.get_meta(bucket_name, object_key).metadata.content_length)
        if offset < length:
373 374 375 376
            data = self.bos_client.get_object_as_string(
                bucket_name=bucket_name,
                key=object_key,
                range=[offset, length - 1])
走神的阿圆's avatar
走神的阿圆 已提交
377 378 379 380
        else:
            data = b''

        continue_from_token = {"last_offset": length}
走神的阿圆's avatar
走神的阿圆 已提交
381 382
        return data, continue_from_token

走神的阿圆's avatar
走神的阿圆 已提交
383 384 385 386 387 388 389 390 391 392 393 394 395 396
    def ready_to_append(self):
        if self._file_contents_count >= self.max_contents_count or \
                time.time() - self._start_append_time > self.max_contents_time:
            return True
        else:
            return False

    def append(self, filename, file_content, binary_mode=False, force=False):
        self._file_contents_to_add += file_content
        self._file_contents_count += 1

        if not force and not self.ready_to_append():
            return
        file_content = self._file_contents_to_add
走神的阿圆's avatar
走神的阿圆 已提交
397
        bucket_name, object_key = get_object_info(filename)
走神的阿圆's avatar
走神的阿圆 已提交
398 399
        if not self.exists(filename):
            init_data = b''
400 401 402
            # Two cases will come here
            # 1. the file not exist, then we need to create the file
            # 2. sts token invalid in self.exists, we should renew the token, and continue to write
403 404 405 406 407 408 409
            try:
                self.bos_client.append_object(
                    bucket_name=bucket_name,
                    key=object_key,
                    data=init_data,
                    content_md5=content_md5(init_data),
                    content_length=len(init_data))
410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426
            except (exception.BceServerError, exception.BceHttpClientError) as e:
                if bucket_name == 'visualdl-server':  # only sts token from visualdl-server, we can renew automatically
                    self.renew_bos_client_from_server()
                    # we should add a judgement for case 2
                    try:
                        self.bos_client.get_object_meta_data(bucket_name, object_key)
                    except exception.BceError:
                        # the file not exists, then create the file
                        self.bos_client.append_object(
                            bucket_name=bucket_name,
                            key=object_key,
                            data=init_data,
                            content_md5=content_md5(init_data),
                            content_length=len(init_data))
                        return
                else:
                    raise e  # user defined bos token, we have no idea to renew the token, so throw the exception
走神的阿圆's avatar
走神的阿圆 已提交
427 428
        content_length = len(file_content)

429 430 431
        try:
            offset = self.get_meta(bucket_name,
                                   object_key).metadata.content_length
432 433 434 435 436 437 438
            self.bos_client.append_object(
                bucket_name=bucket_name,
                key=object_key,
                data=file_content,
                content_md5=content_md5(file_content),
                content_length=content_length,
                offset=offset)
439 440 441 442 443 444 445 446 447 448 449 450 451 452
        except (exception.BceServerError, exception.BceHttpClientError) as e:
            if bucket_name == 'visualdl-server':  # only sts token from visualdl-server, we can renew automatically
                self.renew_bos_client_from_server()
                offset = self.get_meta(bucket_name,
                                       object_key).metadata.content_length
                self.bos_client.append_object(
                    bucket_name=bucket_name,
                    key=object_key,
                    data=file_content,
                    content_md5=content_md5(file_content),
                    content_length=content_length,
                    offset=offset)
            else:
                raise e  # user defined bos token, we have no idea to renew the token, so throw the exception
走神的阿圆's avatar
走神的阿圆 已提交
453 454 455
        self._file_contents_to_add = b''
        self._file_contents_count = 0
        self._start_append_time = time.time()
走神的阿圆's avatar
走神的阿圆 已提交
456 457

    def write(self, filename, file_content, binary_mode=False):
458 459 460 461 462 463 464 465 466
        self.append(filename, file_content, binary_mode=False)

        # bucket_name, object_key = BosFileSystem._get_object_info(filename)
        #
        # self.bos_client.append_object(bucket_name=bucket_name,
        #                               key=object_key,
        #                               data=file_content,
        #                               content_md5=content_md5(file_content),
        #                               content_length=len(file_content))
走神的阿圆's avatar
走神的阿圆 已提交
467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492

    def walk(self, dir):
        class WalkGenerator():
            def __init__(self, bucket_name, contents):
                self.contents = None
                self.length = 0
                self.bucket = bucket_name
                self.handle_contents(contents)
                self.count = 0

            def handle_contents(self, contents):
                contents_map = {}
                for item in contents:
                    try:
                        rindex = item.rindex('/')
                        key = item[0:rindex]
                        value = item[rindex + 1:]
                    except ValueError:
                        key = '.'
                        value = item
                    if key in contents_map.keys():
                        contents_map[key].append(value)
                    else:
                        contents_map[key] = [value]
                temp_walk = []
                for key, value in contents_map.items():
493 494 495 496
                    temp_walk.append([
                        BosFileSystem.join('bos://' + self.bucket, key), [],
                        value
                    ])
走神的阿圆's avatar
走神的阿圆 已提交
497 498 499 500 501 502 503 504 505 506 507 508 509
                self.length = len(temp_walk)
                self.contents = temp_walk

            def __iter__(self):
                return self

            def __next__(self):
                if self.count < self.length:
                    self.count += 1
                    return self.contents[self.count - 1]
                else:
                    raise StopIteration

走神的阿圆's avatar
走神的阿圆 已提交
510
        bucket_name, object_key = get_object_info(dir)
走神的阿圆's avatar
走神的阿圆 已提交
511 512 513 514 515 516

        if object_key in ['.', './']:
            prefix = None
        else:
            prefix = object_key if object_key.endswith(
                '/') else object_key + '/'
517
        response = self.bos_client.list_objects(bucket_name, prefix=prefix)
走神的阿圆's avatar
走神的阿圆 已提交
518 519 520 521
        contents = [content.key for content in response.contents]
        return WalkGenerator(bucket_name, contents)


522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543
class BFile(object):
    def __init__(self, filename, mode):
        if mode not in ('r', 'rb', 'br', 'w', 'wb', 'bw'):
            raise NotImplementedError("mode {} not supported by "
                                      "BFile.".format(mode))
        self._filename = filename
        self.fs = default_file_factory.get_filesystem(filename)
        self.fs_supports_append = hasattr(self.fs, 'append')
        self.buff = None
        self.buff_chunk_size = _DEFAULT_BLOCK_SIZE
        self.buff_offset = 0
        self.continuation_token = None
        self.write_temp = None
        self.write_started = False
        self.binary_mode = 'b' in mode
        self.write_mode = 'w' in mode
        self.closed = False

    def __enter__(self):
        return self

    def __exit__(self, exc_type, exc_val, exc_tb):
544 545 546 547
        self.close()
        self.buff = None
        self.buff_offset = 0
        self.continuation_token = None
548 549 550 551

    def __iter__(self):
        return self

走神的阿圆's avatar
走神的阿圆 已提交
552 553 554
    def isfile(self, filename):
        return self.fs.isfile(filename)

555 556 557 558 559 560 561 562 563 564 565 566 567 568 569
    def _read_buffer_to_offset(self, new_buff_offset):
        """Read buffer from index self.buffer_offset to index new_buff_offset.

        self.buff_offset marks the last position of the last read,
        new_buff_offset indicates the last position of this read.
        self.buff_offset will be updated by new_buff_offset after this read.

        Returns:
            self.buff[i1: i2]: Content of self.buff.
        """
        old_buff_offset = self.buff_offset
        read_size = min(len(self.buff), new_buff_offset) - old_buff_offset
        self.buff_offset += read_size
        return self.buff[old_buff_offset:old_buff_offset + read_size]

走神的阿圆's avatar
走神的阿圆 已提交
570 571 572
    def read_file(self, filename, binnary=True):
        return self.fs.read_file(filename, binnary)

573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688
    def read(self, n=None):
        """Read `n` or all contents of self.buff or file.

        Returns:
            result: Data from self.buff or file.
        """
        result = None
        # If self.buff is not none and length of self.buff more than
        # self.buff_offset, means there are some content in self.buff have
        # not been read.
        if self.buff and len(self.buff) > self.buff_offset:
            if n is not None:
                chunk = self._read_buffer_to_offset(self.buff_offset + n)
                # If length of data in self.buff is more than `n`, then read `n`
                # data from local buffer.
                if len(chunk) == n:
                    return chunk
                result = chunk
                # The length of all data in self.buff may less than `n`,
                # so we should read other `n-length(self.buff)` data.
                n -= len(chunk)
            # If n is none, read all data in self.buff.
            else:
                # add all local buffer and update offsets
                result = self._read_buffer_to_offset(len(self.buff))

        # self.buff is empty if program is here.
        # Read from filesystem.
        # If n is not none, read max(n, self.buff_chunk_size) data from file,
        # otherwise read all data from file.
        # TODO(shenhuhan) N is limited to max_buff, but all-data is unlimited?
        read_size = max(self.buff_chunk_size, n) if n is not None else None
        self.buff, self.continuation_token = self.fs.read(
            self._filename, self.binary_mode, read_size,
            self.continuation_token)
        self.buff_offset = 0

        if n is not None:
            chunk = self._read_buffer_to_offset(n)
        else:
            # add all local buffer and update offsets
            chunk = self._read_buffer_to_offset(len(self.buff))
        result = result + chunk if result else chunk

        return result

    def write(self, file_content):
        """Write contents to file.

        Args:
            file_content: Contents waiting to be written to file.
        """
        if not self.write_mode:
            raise RuntimeError("File not opened in write mode")
        if self.closed:
            raise RuntimeError("File already closed")

        if self.fs_supports_append:
            if not self.write_started:
                self.fs.write(self._filename, file_content, self.binary_mode)
                self.write_started = True
            else:
                self.fs.append(self._filename, file_content, self.binary_mode)
        else:
            # add to temp file, but wait for flush to write to final filesystem
            if self.write_temp is None:
                mode = "w+b" if self.binary_mode else "w+"
                self.write_temp = tempfile.TemporaryFile(mode)
            self.write_temp.write(file_content)

    def __next__(self):
        line = None
        while True:
            if not self.buff:
                # read one unit into the buffer
                line = self.read(1)
                if line and (line[-1] == '\n' or not self.buff):
                    return line
                if not self.buff:
                    raise StopIteration()
            else:
                index = self.buff.find('\n', self.buff_offset)
                if index != -1:
                    # include line until now plus newline
                    chunk = self.read(index + 1 - self.buff_offset)
                    line = line + chunk if line else chunk
                    return line

                # read one unit past end of buffer
                chunk = self.read(len(self.buff) + 1 - self.buff_offset)
                line = line + chunk if line else chunk
                if line and (line[-1] == '\n' or not self.buff):
                    return line
                if not self.buff:
                    raise StopIteration()

    def next(self):
        return self.__next__()

    def flush(self):
        """Flush data to disk.
        """
        if self.closed:
            raise RuntimeError("File already closed")
        if not self.fs_supports_append:
            if self.write_temp is not None:
                # read temp file from the beginning
                self.write_temp.flush()
                self.write_temp.seek(0)
                chunk = self.write_temp.read()
                if chunk is not None:
                    # write full contents and keep in temp file
                    self.fs.write(self._filename, chunk, self.binary_mode)
                    self.write_temp.seek(len(chunk))

    def close(self):
走神的阿圆's avatar
走神的阿圆 已提交
689
        if isinstance(self.fs, BosFileSystem):
690
            try:
691 692
                self.fs.append(
                    self._filename, b'', self.binary_mode, force=True)
P
Peter Pan 已提交
693
            except Exception:
694
                pass
695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716
        self.flush()
        if self.write_temp is not None:
            self.write_temp.close()
            self.write_temp = None
            self.write_started = False
        self.closed = True


def exists(path):
    return default_file_factory.get_filesystem(path).exists(path)


def makedirs(path):
    return default_file_factory.get_filesystem(path).makedirs(path)


def join(path, *paths):
    return default_file_factory.get_filesystem(path).join(path, *paths)


def walk(dir):
    return default_file_factory.get_filesystem(dir).walk(dir)