1
# Copyright (C) 2006-2011 Canonical Ltd
3
# This program is free software; you can redistribute it and/or modify
4
# it under the terms of the GNU General Public License as published by
5
# the Free Software Foundation; either version 2 of the License, or
6
# (at your option) any later version.
8
# This program is distributed in the hope that it will be useful,
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
# GNU General Public License for more details.
13
# You should have received a copy of the GNU General Public License
14
# along with this program; if not, write to the Free Software
15
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
17
"""Handlers for HTTP Responses.
19
The purpose of these classes is to provide a uniform interface for clients
20
to standard HTTP responses, single range responses and multipart range
24
from __future__ import absolute_import
28
import http.client as http_client
29
except ImportError: # python < 3
30
import httplib as http_client
32
import email.utils as email_utils
33
except ImportError: # python < 3
34
import rfc822 as email_utils
40
from ...sixish import (
46
class ResponseFile(object):
47
"""A wrapper around the http socket containing the result of a GET request.
49
Only read() and seek() (forward) are supported.
53
def __init__(self, path, infile):
56
:param path: File url, for error reports.
58
:param infile: File-like socket set at body start.
67
Dummy implementation for consistency with the 'file' API.
73
def __exit__(self, exc_type, exc_val, exc_tb):
74
return False # propogate exceptions.
76
def read(self, size=None):
77
"""Read size bytes from the current position in the file.
79
:param size: The number of bytes to read. Leave unspecified or pass
82
if size is None and not PY3:
84
data = self._file.read(size)
85
self._pos += len(data)
89
data = self._file.readline()
90
self._pos += len(data)
95
line = self.readline()
103
def seek(self, offset, whence=os.SEEK_SET):
104
if whence == os.SEEK_SET:
105
if offset < self._pos:
106
raise AssertionError(
107
"Can't seek backwards, pos: %s, offset: %s"
108
% (self._pos, offset))
109
to_discard = offset - self._pos
110
elif whence == os.SEEK_CUR:
113
raise AssertionError("Can't seek backwards")
115
# Just discard the unwanted bytes
116
self.read(to_discard)
118
# A RangeFile expects the following grammar (simplified to outline the
119
# assumptions we rely upon).
124
# single_range: content_range_header data
126
# multiple_range: boundary_header boundary (content_range_header data boundary)+
129
class RangeFile(ResponseFile):
130
"""File-like object that allow access to partial available data.
132
All accesses should happen sequentially since the acquisition occurs during
133
an http response reception (as sockets can't be seeked, we simulate the
134
seek by just reading and discarding the data).
136
The access pattern is defined by a set of ranges discovered as reading
137
progress. Only one range is available at a given time, so all accesses
138
should happen with monotonically increasing offsets.
141
# in _checked_read() below, we may have to discard several MB in the worst
142
# case. To avoid buffering that much, we read and discard by chunks
143
# instead. The underlying file is either a socket or a BytesIO, so reading
144
# 8k chunks should be fine.
145
_discarded_buf_size = 8192
147
# maximum size of read requests -- used to avoid MemoryError issues in recv
148
_max_read_size = 512 * 1024
150
def __init__(self, path, infile):
153
:param path: File url, for error reports.
155
:param infile: File-like socket set at body start.
157
super(RangeFile, self).__init__(path, infile)
158
self._boundary = None
159
# When using multi parts response, this will be set with the headers
160
# associated with the range currently read.
162
# Default to the whole file of unspecified size
163
self.set_range(0, -1)
165
def set_range(self, start, size):
166
"""Change the range mapping"""
169
# Set the new _pos since that's what we want to expose
170
self._pos = self._start
172
def set_boundary(self, boundary):
173
"""Define the boundary used in a multi parts message.
175
The file should be at the beginning of the body, the first range
176
definition is read and taken into account.
178
if not isinstance(boundary, bytes):
179
raise TypeError(boundary)
180
self._boundary = boundary
181
# Decode the headers and setup the first range
183
self.read_range_definition()
185
def read_boundary(self):
186
"""Read the boundary headers defining a new range"""
187
boundary_line = b'\r\n'
188
while boundary_line == b'\r\n':
189
# RFC2616 19.2 Additional CRLFs may precede the first boundary
191
# To be on the safe side we allow it before any boundary line
192
boundary_line = self._file.readline()
194
if boundary_line == b'':
195
# A timeout in the proxy server caused the response to end early.
196
# See launchpad bug 198646.
197
raise errors.HttpBoundaryMissing(
201
if boundary_line != b'--' + self._boundary + b'\r\n':
202
# email_utils.unquote() incorrectly unquotes strings enclosed in <>
203
# IIS 6 and 7 incorrectly wrap boundary strings in <>
204
# together they make a beautiful bug, which we will be gracious
206
if (self._unquote_boundary(boundary_line) !=
207
b'--' + self._boundary + b'\r\n'):
208
raise errors.InvalidHttpResponse(
210
"Expected a boundary (%s) line, got '%s'"
211
% (self._boundary, boundary_line))
213
def _unquote_boundary(self, b):
214
return b[:2] + email_utils.unquote(b[2:-2].decode('ascii')).encode('ascii') + b[-2:]
216
def read_range_definition(self):
217
"""Read a new range definition in a multi parts message.
219
Parse the headers including the empty line following them so that we
220
are ready to read the data itself.
223
self._headers = http_client.parse_headers(self._file)
225
self._headers = http_client.HTTPMessage(self._file, seekable=0)
226
# Extract the range definition
227
content_range = self._headers.get('content-range', None)
228
if content_range is None:
229
raise errors.InvalidHttpResponse(
231
'Content-Range header missing in a multi-part response')
232
self.set_range_from_header(content_range)
234
def set_range_from_header(self, content_range):
235
"""Helper to set the new range from its description in the headers"""
237
rtype, values = content_range.split()
239
raise errors.InvalidHttpRange(self._path, content_range,
242
raise errors.InvalidHttpRange(self._path, content_range,
243
"Unsupported range type '%s'" % rtype)
245
# We don't need total, but note that it may be either the file size
246
# or '*' if the server can't or doesn't want to return the file
248
start_end, total = values.split('/')
249
start, end = start_end.split('-')
253
raise errors.InvalidHttpRange(self._path, content_range,
254
'Invalid range values')
255
size = end - start + 1
257
raise errors.InvalidHttpRange(self._path, content_range,
258
'Invalid range, size <= 0')
259
self.set_range(start, size)
261
def _checked_read(self, size):
262
"""Read the file checking for short reads.
264
The data read is discarded along the way.
269
data = self._file.read(min(remaining, self._discarded_buf_size))
270
remaining -= len(data)
272
raise errors.ShortReadvError(self._path, pos, size,
276
def _seek_to_next_range(self):
277
# We will cross range boundaries
278
if self._boundary is None:
279
# If we don't have a boundary, we can't find another range
280
raise errors.InvalidRange(self._path, self._pos,
281
"Range (%s, %s) exhausted"
282
% (self._start, self._size))
284
self.read_range_definition()
286
def read(self, size=-1):
287
"""Read size bytes from the current position in the file.
289
Reading across ranges is not supported. We rely on the underlying http
290
client to clean the socket if we leave bytes unread. This may occur for
291
the final boundary line of a multipart response or for any range
292
request not entirely consumed by the client (due to offset coalescing)
294
:param size: The number of bytes to read. Leave unspecified or pass
298
and self._pos == self._start + self._size):
302
self._seek_to_next_range()
303
elif self._pos < self._start:
304
raise errors.InvalidRange(
305
self._path, self._pos,
306
"Can't read %s bytes before range (%s, %s)"
307
% (size, self._start, self._size))
309
if size > 0 and self._pos + size > self._start + self._size:
310
raise errors.InvalidRange(
311
self._path, self._pos,
312
"Can't read %s bytes across range (%s, %s)"
313
% (size, self._start, self._size))
315
# read data from file
319
# Don't read past the range definition
320
limited = self._start + self._size - self._pos
322
limited = min(limited, size)
323
osutils.pumpfile(self._file, buf, limited, self._max_read_size)
324
data = buf.getvalue()
326
# Update _pos respecting the data effectively read
327
self._pos += len(data)
330
def seek(self, offset, whence=0):
331
start_pos = self._pos
335
final_pos = start_pos + offset
338
final_pos = self._start + self._size + offset # offset < 0
340
raise errors.InvalidRange(
341
self._path, self._pos,
342
"RangeFile: can't seek from end while size is unknown")
344
raise ValueError("Invalid value %s for whence." % whence)
346
if final_pos < self._pos:
347
# Can't seek backwards
348
raise errors.InvalidRange(
349
self._path, self._pos,
350
'RangeFile: trying to seek backwards to %s' % final_pos)
353
cur_limit = self._start + self._size
354
while final_pos > cur_limit:
355
# We will cross range boundaries
356
remain = cur_limit - self._pos
358
# Finish reading the current range
359
self._checked_read(remain)
360
self._seek_to_next_range()
361
cur_limit = self._start + self._size
363
size = final_pos - self._pos
364
if size > 0: # size can be < 0 if we crossed a range boundary
365
# We don't need the data, just read it and throw it away
366
self._checked_read(size)
372
def handle_response(url, code, msg, data):
373
"""Interpret the code & headers and wrap the provided data in a RangeFile.
375
This is a factory method which returns an appropriate RangeFile based on
376
the code & headers it's given.
378
:param url: The url being processed. Mostly for error reporting
379
:param code: The integer HTTP response code
380
:param msg: An HTTPMessage containing the headers for the response
381
:param data: A file-like object that can be read() to get the
383
:return: A file-like object that can seek()+read() the
384
ranges indicated by the headers.
388
rfile = ResponseFile(url, data)
390
rfile = RangeFile(url, data)
391
content_type = msg.get('content-type', None)
392
if content_type is None:
393
# When there is no content-type header we treat the response as
394
# being of type 'application/octet-stream' as per RFC2616 section
396
# Therefore it is obviously not multipart
397
content_type = 'application/octet-stream'
401
is_multipart = (msg.get_content_maintype() == 'multipart'
402
and msg.get_content_subtype() == 'byteranges')
404
is_multipart = (msg.getmaintype() == 'multipart'
405
and msg.getsubtype() == 'byteranges')
408
# Full fledged multipart response
410
boundary = msg.get_param('boundary')
412
boundary = msg.getparam('boundary')
413
rfile.set_boundary(boundary.encode('ascii'))
415
# A response to a range request, but not multipart
416
content_range = msg.get('content-range', None)
417
if content_range is None:
418
raise errors.InvalidHttpResponse(url,
419
'Missing the Content-Range header in a 206 range response')
420
rfile.set_range_from_header(content_range)
422
raise errors.InvalidHttpResponse(url,
423
'Unknown response code %s' % code)