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
34
from ...sixish import (
39
class ResponseFile(object):
40
"""A wrapper around the http socket containing the result of a GET request.
42
Only read() and seek() (forward) are supported.
45
def __init__(self, path, infile):
48
:param path: File url, for error reports.
50
:param infile: File-like socket set at body start.
59
Dummy implementation for consistency with the 'file' API.
62
def read(self, size=-1):
63
"""Read size bytes from the current position in the file.
65
:param size: The number of bytes to read. Leave unspecified or pass
68
data = self._file.read(size)
69
self._pos += len(data)
73
data = self._file.readline()
74
self._pos += len(data)
79
line = self.readline()
87
def seek(self, offset, whence=os.SEEK_SET):
88
if whence == os.SEEK_SET:
89
if offset < self._pos:
91
"Can't seek backwards, pos: %s, offset: %s"
92
% (self._pos, offset))
93
to_discard = offset - self._pos
94
elif whence == os.SEEK_CUR:
97
raise AssertionError("Can't seek backwards")
99
# Just discard the unwanted bytes
100
self.read(to_discard)
102
# A RangeFile expects the following grammar (simplified to outline the
103
# assumptions we rely upon).
108
# single_range: content_range_header data
110
# multiple_range: boundary_header boundary (content_range_header data boundary)+
112
class RangeFile(ResponseFile):
113
"""File-like object that allow access to partial available data.
115
All accesses should happen sequentially since the acquisition occurs during
116
an http response reception (as sockets can't be seeked, we simulate the
117
seek by just reading and discarding the data).
119
The access pattern is defined by a set of ranges discovered as reading
120
progress. Only one range is available at a given time, so all accesses
121
should happen with monotonically increasing offsets.
124
# in _checked_read() below, we may have to discard several MB in the worst
125
# case. To avoid buffering that much, we read and discard by chunks
126
# instead. The underlying file is either a socket or a BytesIO, so reading
127
# 8k chunks should be fine.
128
_discarded_buf_size = 8192
130
# maximum size of read requests -- used to avoid MemoryError issues in recv
131
_max_read_size = 512 * 1024
133
def __init__(self, path, infile):
136
:param path: File url, for error reports.
138
:param infile: File-like socket set at body start.
140
super(RangeFile, self).__init__(path, infile)
141
self._boundary = None
142
# When using multi parts response, this will be set with the headers
143
# associated with the range currently read.
145
# Default to the whole file of unspecified size
146
self.set_range(0, -1)
148
def set_range(self, start, size):
149
"""Change the range mapping"""
152
# Set the new _pos since that's what we want to expose
153
self._pos = self._start
155
def set_boundary(self, boundary):
156
"""Define the boundary used in a multi parts message.
158
The file should be at the beginning of the body, the first range
159
definition is read and taken into account.
161
self._boundary = boundary
162
# Decode the headers and setup the first range
164
self.read_range_definition()
166
def read_boundary(self):
167
"""Read the boundary headers defining a new range"""
168
boundary_line = '\r\n'
169
while boundary_line == '\r\n':
170
# RFC2616 19.2 Additional CRLFs may precede the first boundary
172
# To be on the safe side we allow it before any boundary line
173
boundary_line = self._file.readline()
175
if boundary_line == '':
176
# A timeout in the proxy server caused the response to end early.
177
# See launchpad bug 198646.
178
raise errors.HttpBoundaryMissing(
182
if boundary_line != '--' + self._boundary + '\r\n':
183
# rfc822.unquote() incorrectly unquotes strings enclosed in <>
184
# IIS 6 and 7 incorrectly wrap boundary strings in <>
185
# together they make a beautiful bug, which we will be gracious
187
if (self._unquote_boundary(boundary_line) !=
188
'--' + self._boundary + '\r\n'):
189
raise errors.InvalidHttpResponse(
191
"Expected a boundary (%s) line, got '%s'"
192
% (self._boundary, boundary_line))
194
def _unquote_boundary(self, b):
195
return b[:2] + rfc822.unquote(b[2:-2]) + b[-2:]
197
def read_range_definition(self):
198
"""Read a new range definition in a multi parts message.
200
Parse the headers including the empty line following them so that we
201
are ready to read the data itself.
203
self._headers = httplib.HTTPMessage(self._file, seekable=0)
204
# Extract the range definition
205
content_range = self._headers.getheader('content-range', None)
206
if content_range is None:
207
raise errors.InvalidHttpResponse(
209
'Content-Range header missing in a multi-part response')
210
self.set_range_from_header(content_range)
212
def set_range_from_header(self, content_range):
213
"""Helper to set the new range from its description in the headers"""
215
rtype, values = content_range.split()
217
raise errors.InvalidHttpRange(self._path, content_range,
220
raise errors.InvalidHttpRange(self._path, content_range,
221
"Unsupported range type '%s'" % rtype)
223
# We don't need total, but note that it may be either the file size
224
# or '*' if the server can't or doesn't want to return the file
226
start_end, total = values.split('/')
227
start, end = start_end.split('-')
231
raise errors.InvalidHttpRange(self._path, content_range,
232
'Invalid range values')
233
size = end - start + 1
235
raise errors.InvalidHttpRange(self._path, content_range,
236
'Invalid range, size <= 0')
237
self.set_range(start, size)
239
def _checked_read(self, size):
240
"""Read the file checking for short reads.
242
The data read is discarded along the way.
247
data = self._file.read(min(remaining, self._discarded_buf_size))
248
remaining -= len(data)
250
raise errors.ShortReadvError(self._path, pos, size,
254
def _seek_to_next_range(self):
255
# We will cross range boundaries
256
if self._boundary is None:
257
# If we don't have a boundary, we can't find another range
258
raise errors.InvalidRange(self._path, self._pos,
259
"Range (%s, %s) exhausted"
260
% (self._start, self._size))
262
self.read_range_definition()
264
def read(self, size=-1):
265
"""Read size bytes from the current position in the file.
267
Reading across ranges is not supported. We rely on the underlying http
268
client to clean the socket if we leave bytes unread. This may occur for
269
the final boundary line of a multipart response or for any range
270
request not entirely consumed by the client (due to offset coalescing)
272
:param size: The number of bytes to read. Leave unspecified or pass
276
and self._pos == self._start + self._size):
280
self._seek_to_next_range()
281
elif self._pos < self._start:
282
raise errors.InvalidRange(
283
self._path, self._pos,
284
"Can't read %s bytes before range (%s, %s)"
285
% (size, self._start, self._size))
287
if size > 0 and self._pos + size > self._start + self._size:
288
raise errors.InvalidRange(
289
self._path, self._pos,
290
"Can't read %s bytes across range (%s, %s)"
291
% (size, self._start, self._size))
293
# read data from file
297
# Don't read past the range definition
298
limited = self._start + self._size - self._pos
300
limited = min(limited, size)
301
osutils.pumpfile(self._file, buf, limited, self._max_read_size)
302
data = buf.getvalue()
304
# Update _pos respecting the data effectively read
305
self._pos += len(data)
308
def seek(self, offset, whence=0):
309
start_pos = self._pos
313
final_pos = start_pos + offset
316
final_pos = self._start + self._size + offset # offset < 0
318
raise errors.InvalidRange(
319
self._path, self._pos,
320
"RangeFile: can't seek from end while size is unknown")
322
raise ValueError("Invalid value %s for whence." % whence)
324
if final_pos < self._pos:
325
# Can't seek backwards
326
raise errors.InvalidRange(
327
self._path, self._pos,
328
'RangeFile: trying to seek backwards to %s' % final_pos)
331
cur_limit = self._start + self._size
332
while final_pos > cur_limit:
333
# We will cross range boundaries
334
remain = cur_limit - self._pos
336
# Finish reading the current range
337
self._checked_read(remain)
338
self._seek_to_next_range()
339
cur_limit = self._start + self._size
341
size = final_pos - self._pos
342
if size > 0: # size can be < 0 if we crossed a range boundary
343
# We don't need the data, just read it and throw it away
344
self._checked_read(size)
350
def handle_response(url, code, msg, data):
351
"""Interpret the code & headers and wrap the provided data in a RangeFile.
353
This is a factory method which returns an appropriate RangeFile based on
354
the code & headers it's given.
356
:param url: The url being processed. Mostly for error reporting
357
:param code: The integer HTTP response code
358
:param msg: An HTTPMessage containing the headers for the response
359
:param data: A file-like object that can be read() to get the
361
:return: A file-like object that can seek()+read() the
362
ranges indicated by the headers.
366
rfile = ResponseFile(url, data)
368
rfile = RangeFile(url, data)
369
content_type = msg.getheader('content-type', None)
370
if content_type is None:
371
# When there is no content-type header we treat the response as
372
# being of type 'application/octet-stream' as per RFC2616 section
374
# Therefore it is obviously not multipart
375
content_type = 'application/octet-stream'
378
is_multipart = (msg.getmaintype() == 'multipart'
379
and msg.getsubtype() == 'byteranges')
382
# Full fledged multipart response
383
rfile.set_boundary(msg.getparam('boundary'))
385
# A response to a range request, but not multipart
386
content_range = msg.getheader('content-range', None)
387
if content_range is None:
388
raise errors.InvalidHttpResponse(url,
389
'Missing the Content-Range header in a 206 range response')
390
rfile.set_range_from_header(content_range)
392
raise errors.InvalidHttpResponse(url,
393
'Unknown response code %s' % code)