# -------------------------------------------------------------------------
# Copyright (c) Microsoft Corporation. All rights reserved.
# Licensed under the MIT License. See License.txt in the project root for
# license information.
# --------------------------------------------------------------------------
import sys
import threading
import time
import warnings
from io import BytesIO
from typing import Iterator
import requests
from azure.core.exceptions import HttpResponseError, ServiceResponseError
from azure.core.tracing.common import with_current_context
from ._shared.encryption import decrypt_blob
from ._shared.request_handlers import validate_and_format_range_headers
from ._shared.response_handlers import process_storage_error, parse_length_from_content_range
from ._deserialize import get_page_ranges_result
def process_range_and_offset(start_range, end_range, length, encryption):
start_offset, end_offset = 0, 0
if encryption.get("key") is not None or encryption.get("resolver") is not None:
if start_range is not None:
# Align the start of the range along a 16 byte block
start_offset = start_range % 16
start_range -= start_offset
# Include an extra 16 bytes for the IV if necessary
# Because of the previous offsetting, start_range will always
# be a multiple of 16.
if start_range > 0:
start_offset += 16
start_range -= 16
if length is not None:
# Align the end of the range along a 16 byte block
end_offset = 15 - (end_range % 16)
end_range += end_offset
return (start_range, end_range), (start_offset, end_offset)
def process_content(data, start_offset, end_offset, encryption):
if data is None:
raise ValueError("Response cannot be None.")
content = b"".join(list(data))
if content and encryption.get("key") is not None or encryption.get("resolver") is not None:
try:
return decrypt_blob(
encryption.get("required"),
encryption.get("key"),
encryption.get("resolver"),
content,
start_offset,
end_offset,
data.response.headers,
)
except Exception as error:
raise HttpResponseError(message="Decryption failed.", response=data.response, error=error)
return content
class _ChunkDownloader(object): # pylint: disable=too-many-instance-attributes
def __init__(
self,
client=None,
non_empty_ranges=None,
total_size=None,
chunk_size=None,
current_progress=None,
start_range=None,
end_range=None,
stream=None,
parallel=None,
validate_content=None,
encryption_options=None,
**kwargs
):
self.client = client
self.non_empty_ranges = non_empty_ranges
# Information on the download range/chunk size
self.chunk_size = chunk_size
self.total_size = total_size
self.start_index = start_range
self.end_index = end_range
# The destination that we will write to
self.stream = stream
self.stream_lock = threading.Lock() if parallel else None
self.progress_lock = threading.Lock() if parallel else None
# For a parallel download, the stream is always seekable, so we note down the current position
# in order to seek to the right place when out-of-order chunks come in
self.stream_start = stream.tell() if parallel else None
# Download progress so far
self.progress_total = current_progress
# Encryption
self.encryption_options = encryption_options
# Parameters for each get operation
self.validate_content = validate_content
self.request_options = kwargs
def _calculate_range(self, chunk_start):
if chunk_start + self.chunk_size > self.end_index:
chunk_end = self.end_index
else:
chunk_end = chunk_start + self.chunk_size
return chunk_start, chunk_end
def get_chunk_offsets(self):
index = self.start_index
while index < self.end_index:
yield index
index += self.chunk_size
def process_chunk(self, chunk_start):
chunk_start, chunk_end = self._calculate_range(chunk_start)
chunk_data = self._download_chunk(chunk_start, chunk_end - 1)
length = chunk_end - chunk_start
if length > 0:
self._write_to_stream(chunk_data, chunk_start)
self._update_progress(length)
def yield_chunk(self, chunk_start):
chunk_start, chunk_end = self._calculate_range(chunk_start)
return self._download_chunk(chunk_start, chunk_end - 1)
def _update_progress(self, length):
if self.progress_lock:
with self.progress_lock: # pylint: disable=not-context-manager
self.progress_total += length
else:
self.progress_total += length
def _write_to_stream(self, chunk_data, chunk_start):
if self.stream_lock:
with self.stream_lock: # pylint: disable=not-context-manager
self.stream.seek(self.stream_start + (chunk_start - self.start_index))
self.stream.write(chunk_data)
else:
self.stream.write(chunk_data)
def _do_optimize(self, given_range_start, given_range_end):
# If we have no page range list stored, then assume there's data everywhere for that page blob
# or it's a block blob or append blob
if self.non_empty_ranges is None:
return False
for source_range in self.non_empty_ranges:
# Case 1: As the range list is sorted, if we've reached such a source_range
# we've checked all the appropriate source_range already and haven't found any overlapping.
# so the given range doesn't have any data and download optimization could be applied.
# given range: | |
# source range: | |
if given_range_end < source_range['start']: # pylint:disable=no-else-return
return True
# Case 2: the given range comes after source_range, continue checking.
# given range: | |
# source range: | |
elif source_range['end'] < given_range_start:
pass
# Case 3: source_range and given range overlap somehow, no need to optimize.
else:
return False
# Went through all src_ranges, but nothing overlapped. Optimization will be applied.
return True
def _download_chunk(self, chunk_start, chunk_end):
download_range, offset = process_range_and_offset(
chunk_start, chunk_end, chunk_end, self.encryption_options
)
# No need to download the empty chunk from server if there's no data in the chunk to be downloaded.
# Do optimize and create empty chunk locally if condition is met.
if self._do_optimize(download_range[0], download_range[1]):
chunk_data = b"\x00" * self.chunk_size
else:
range_header, range_validation = validate_and_format_range_headers(
download_range[0],
download_range[1],
check_content_md5=self.validate_content
)
retry_active = True
retry_total = 3
while retry_active:
try:
_, response = self.client.download(
range=range_header,
range_get_content_md5=range_validation,
validate_content=self.validate_content,
data_stream_total=self.total_size,
download_stream_current=self.progress_total,
**self.request_options
)
except HttpResponseError as error:
process_storage_error(error)
try:
chunk_data = process_content(response, offset[0], offset[1], self.encryption_options)
retry_active = False
except (requests.exceptions.ChunkedEncodingError, requests.exceptions.ConnectionError) as error:
retry_total -= 1
if retry_total <= 0:
raise ServiceResponseError(error, error=error)
time.sleep(1)
# This makes sure that if_match is set so that we can validate
# that subsequent downloads are to an unmodified blob
if self.request_options.get("modified_access_conditions"):
self.request_options["modified_access_conditions"].if_match = response.properties.etag
return chunk_data
class _ChunkIterator(object):
"""Async iterator for chunks in blob download stream."""
def __init__(self, size, content, downloader, chunk_size):
self.size = size
self._chunk_size = chunk_size
self._current_content = content
self._iter_downloader = downloader
self._iter_chunks = None
self._complete = (size == 0)
def __len__(self):
return self.size
def __iter__(self):
return self
def __next__(self):
"""Iterate through responses."""
if self._complete:
raise StopIteration("Download complete")
if not self._iter_downloader:
# cut the data obtained from initial GET into chunks
if len(self._current_content) > self._chunk_size:
return self._get_chunk_data()
self._complete = True
return self._current_content
if not self._iter_chunks:
self._iter_chunks = self._iter_downloader.get_chunk_offsets()
# initial GET result still has more than _chunk_size bytes of data
if len(self._current_content) >= self._chunk_size:
return self._get_chunk_data()
try:
chunk = next(self._iter_chunks)
self._current_content += self._iter_downloader.yield_chunk(chunk)
except StopIteration as e:
self._complete = True
if self._current_content:
return self._current_content
raise e
# the current content from the first get is still there but smaller than chunk size
# therefore we want to make sure its also included
return self._get_chunk_data()
next = __next__ # Python 2 compatibility.
def _get_chunk_data(self):
chunk_data = self._current_content[: self._chunk_size]
self._current_content = self._current_content[self._chunk_size:]
return chunk_data
[docs]class StorageStreamDownloader(object): # pylint: disable=too-many-instance-attributes
"""A streaming object to download from Azure Storage.
:ivar str name:
The name of the blob being downloaded.
:ivar str container:
The name of the container where the blob is.
:ivar ~azure.storage.blob.BlobProperties properties:
The properties of the blob being downloaded. If only a range of the data is being
downloaded, this will be reflected in the properties.
:ivar int size:
The size of the total data in the stream. This will be the byte range if specified,
otherwise the total size of the blob.
"""
def __init__(
self,
clients=None,
config=None,
start_range=None,
end_range=None,
validate_content=None,
encryption_options=None,
max_concurrency=1,
name=None,
container=None,
encoding=None,
**kwargs
):
self.name = name
self.container = container
self.properties = None
self.size = None
self._clients = clients
self._config = config
self._start_range = start_range
self._end_range = end_range
self._max_concurrency = max_concurrency
self._encoding = encoding
self._validate_content = validate_content
self._encryption_options = encryption_options or {}
self._request_options = kwargs
self._location_mode = None
self._download_complete = False
self._current_content = None
self._file_size = None
self._non_empty_ranges = None
self._response = None
# The service only provides transactional MD5s for chunks under 4MB.
# If validate_content is on, get only self.MAX_CHUNK_GET_SIZE for the first
# chunk so a transactional MD5 can be retrieved.
self._first_get_size = (
self._config.max_single_get_size if not self._validate_content else self._config.max_chunk_get_size
)
initial_request_start = self._start_range if self._start_range is not None else 0
if self._end_range is not None and self._end_range - self._start_range < self._first_get_size:
initial_request_end = self._end_range
else:
initial_request_end = initial_request_start + self._first_get_size - 1
self._initial_range, self._initial_offset = process_range_and_offset(
initial_request_start, initial_request_end, self._end_range, self._encryption_options
)
self._response = self._initial_request()
self.properties = self._response.properties
self.properties.name = self.name
self.properties.container = self.container
# Set the content length to the download size instead of the size of
# the last range
self.properties.size = self.size
# Overwrite the content range to the user requested range
self.properties.content_range = "bytes {0}-{1}/{2}".format(
self._start_range,
self._end_range,
self._file_size
)
# Overwrite the content MD5 as it is the MD5 for the last range instead
# of the stored MD5
# TODO: Set to the stored MD5 when the service returns this
self.properties.content_md5 = None
def __len__(self):
return self.size
def _initial_request(self):
range_header, range_validation = validate_and_format_range_headers(
self._initial_range[0],
self._initial_range[1],
start_range_required=False,
end_range_required=False,
check_content_md5=self._validate_content
)
retry_active = True
retry_total = 3
while retry_active:
try:
location_mode, response = self._clients.blob.download(
range=range_header,
range_get_content_md5=range_validation,
validate_content=self._validate_content,
data_stream_total=None,
download_stream_current=0,
**self._request_options
)
# Check the location we read from to ensure we use the same one
# for subsequent requests.
self._location_mode = location_mode
# Parse the total file size and adjust the download size if ranges
# were specified
self._file_size = parse_length_from_content_range(response.properties.content_range)
if self._end_range is not None:
# Use the end range index unless it is over the end of the file
self.size = min(self._file_size, self._end_range - self._start_range + 1)
elif self._start_range is not None:
self.size = self._file_size - self._start_range
else:
self.size = self._file_size
except HttpResponseError as error:
if self._start_range is None and error.response.status_code == 416:
# Get range will fail on an empty file. If the user did not
# request a range, do a regular get request in order to get
# any properties.
try:
_, response = self._clients.blob.download(
validate_content=self._validate_content,
data_stream_total=0,
download_stream_current=0,
**self._request_options
)
except HttpResponseError as error:
process_storage_error(error)
# Set the download size to empty
self.size = 0
self._file_size = 0
else:
process_storage_error(error)
try:
if self.size == 0:
self._current_content = b""
else:
self._current_content = process_content(
response,
self._initial_offset[0],
self._initial_offset[1],
self._encryption_options
)
retry_active = False
except (requests.exceptions.ChunkedEncodingError, requests.exceptions.ConnectionError) as error:
retry_total -= 1
if retry_total <= 0:
raise ServiceResponseError(error, error=error)
time.sleep(1)
# get page ranges to optimize downloading sparse page blob
if response.properties.blob_type == 'PageBlob':
try:
page_ranges = self._clients.page_blob.get_page_ranges()
self._non_empty_ranges = get_page_ranges_result(page_ranges)[0]
# according to the REST API documentation:
# in a highly fragmented page blob with a large number of writes,
# a Get Page Ranges request can fail due to an internal server timeout.
# thus, if the page blob is not sparse, it's ok for it to fail
except HttpResponseError:
pass
# If the file is small, the download is complete at this point.
# If file size is large, download the rest of the file in chunks.
if response.properties.size != self.size:
if self._request_options.get("modified_access_conditions"):
self._request_options["modified_access_conditions"].if_match = response.properties.etag
else:
self._download_complete = True
return response
[docs] def chunks(self):
# type: () -> Iterator[bytes]
"""Iterate over chunks in the download stream.
:rtype: Iterator[bytes]
.. admonition:: Example:
.. literalinclude:: ../samples/blob_samples_hello_world.py
:start-after: [START download_a_blob_in_chunk]
:end-before: [END download_a_blob_in_chunk]
:language: python
:dedent: 12
:caption: Download a blob using chunks().
"""
if self.size == 0 or self._download_complete:
iter_downloader = None
else:
data_end = self._file_size
if self._end_range is not None:
# Use the end range index unless it is over the end of the file
data_end = min(self._file_size, self._end_range + 1)
iter_downloader = _ChunkDownloader(
client=self._clients.blob,
non_empty_ranges=self._non_empty_ranges,
total_size=self.size,
chunk_size=self._config.max_chunk_get_size,
current_progress=self._first_get_size,
start_range=self._initial_range[1] + 1, # start where the first download ended
end_range=data_end,
stream=None,
parallel=False,
validate_content=self._validate_content,
encryption_options=self._encryption_options,
use_location=self._location_mode,
**self._request_options
)
return _ChunkIterator(
size=self.size,
content=self._current_content,
downloader=iter_downloader,
chunk_size=self._config.max_chunk_get_size)
[docs] def readall(self):
"""Download the contents of this blob.
This operation is blocking until all data is downloaded.
:rtype: bytes or str
"""
stream = BytesIO()
self.readinto(stream)
data = stream.getvalue()
if self._encoding:
return data.decode(self._encoding)
return data
[docs] def content_as_bytes(self, max_concurrency=1):
"""Download the contents of this file.
This operation is blocking until all data is downloaded.
:keyword int max_concurrency:
The number of parallel connections with which to download.
:rtype: bytes
"""
warnings.warn(
"content_as_bytes is deprecated, use readall instead",
DeprecationWarning
)
self._max_concurrency = max_concurrency
return self.readall()
[docs] def content_as_text(self, max_concurrency=1, encoding="UTF-8"):
"""Download the contents of this blob, and decode as text.
This operation is blocking until all data is downloaded.
:keyword int max_concurrency:
The number of parallel connections with which to download.
:param str encoding:
Test encoding to decode the downloaded bytes. Default is UTF-8.
:rtype: str
"""
warnings.warn(
"content_as_text is deprecated, use readall instead",
DeprecationWarning
)
self._max_concurrency = max_concurrency
self._encoding = encoding
return self.readall()
[docs] def readinto(self, stream):
"""Download the contents of this file to a stream.
:param stream:
The stream to download to. This can be an open file-handle,
or any writable stream. The stream must be seekable if the download
uses more than one parallel connection.
:returns: The number of bytes read.
:rtype: int
"""
# The stream must be seekable if parallel download is required
parallel = self._max_concurrency > 1
if parallel:
error_message = "Target stream handle must be seekable."
if sys.version_info >= (3,) and not stream.seekable():
raise ValueError(error_message)
try:
stream.seek(stream.tell())
except (NotImplementedError, AttributeError):
raise ValueError(error_message)
# Write the content to the user stream
stream.write(self._current_content)
if self._download_complete:
return self.size
data_end = self._file_size
if self._end_range is not None:
# Use the length unless it is over the end of the file
data_end = min(self._file_size, self._end_range + 1)
downloader = _ChunkDownloader(
client=self._clients.blob,
non_empty_ranges=self._non_empty_ranges,
total_size=self.size,
chunk_size=self._config.max_chunk_get_size,
current_progress=self._first_get_size,
start_range=self._initial_range[1] + 1, # Start where the first download ended
end_range=data_end,
stream=stream,
parallel=parallel,
validate_content=self._validate_content,
encryption_options=self._encryption_options,
use_location=self._location_mode,
**self._request_options
)
if parallel:
import concurrent.futures
with concurrent.futures.ThreadPoolExecutor(self._max_concurrency) as executor:
list(executor.map(
with_current_context(downloader.process_chunk),
downloader.get_chunk_offsets()
))
else:
for chunk in downloader.get_chunk_offsets():
downloader.process_chunk(chunk)
return self.size
[docs] def download_to_stream(self, stream, max_concurrency=1):
"""Download the contents of this blob to a stream.
:param stream:
The stream to download to. This can be an open file-handle,
or any writable stream. The stream must be seekable if the download
uses more than one parallel connection.
:returns: The properties of the downloaded blob.
:rtype: Any
"""
warnings.warn(
"download_to_stream is deprecated, use readinto instead",
DeprecationWarning
)
self._max_concurrency = max_concurrency
self.readinto(stream)
return self.properties