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
from cStringIO import StringIO
37
class ResponseFile(object):
38
"""A wrapper around the http socket containing the result of a GET request.
40
Only read() and seek() (forward) are supported.
42
def __init__(self, path, infile):
45
:param path: File url, for error reports.
47
:param infile: File-like socket set at body start.
56
Dummy implementation for consistency with the 'file' API.
59
def read(self, size=-1):
60
"""Read size bytes from the current position in the file.
62
:param size: The number of bytes to read. Leave unspecified or pass
65
data = self._file.read(size)
66
self._pos += len(data)
70
data = self._file.readline()
71
self._pos += len(data)
77
def seek(self, offset, whence=os.SEEK_SET):
78
if whence == os.SEEK_SET:
79
if offset < self._pos:
81
"Can't seek backwards, pos: %s, offset: %s"
82
% (self._pos, offset))
83
to_discard = offset - self._pos
84
elif whence == os.SEEK_CUR:
87
raise AssertionError("Can't seek backwards")
89
# Just discard the unwanted bytes
92
# A RangeFile expects the following grammar (simplified to outline the
93
# assumptions we rely upon).
98
# single_range: content_range_header data
100
# multiple_range: boundary_header boundary (content_range_header data boundary)+
102
class RangeFile(ResponseFile):
103
"""File-like object that allow access to partial available data.
105
All accesses should happen sequentially since the acquisition occurs during
106
an http response reception (as sockets can't be seeked, we simulate the
107
seek by just reading and discarding the data).
109
The access pattern is defined by a set of ranges discovered as reading
110
progress. Only one range is available at a given time, so all accesses
111
should happen with monotonically increasing offsets.
114
# in _checked_read() below, we may have to discard several MB in the worst
115
# case. To avoid buffering that much, we read and discard by chunks
116
# instead. The underlying file is either a socket or a StringIO, so reading
117
# 8k chunks should be fine.
118
_discarded_buf_size = 8192
120
# maximum size of read requests -- used to avoid MemoryError issues in recv
121
_max_read_size = 512 * 1024
123
def __init__(self, path, infile):
126
:param path: File url, for error reports.
128
:param infile: File-like socket set at body start.
130
super(RangeFile, self).__init__(path, infile)
131
self._boundary = None
132
# When using multi parts response, this will be set with the headers
133
# associated with the range currently read.
135
# Default to the whole file of unspecified size
136
self.set_range(0, -1)
138
def set_range(self, start, size):
139
"""Change the range mapping"""
142
# Set the new _pos since that's what we want to expose
143
self._pos = self._start
145
def set_boundary(self, boundary):
146
"""Define the boundary used in a multi parts message.
148
The file should be at the beginning of the body, the first range
149
definition is read and taken into account.
151
self._boundary = boundary
152
# Decode the headers and setup the first range
154
self.read_range_definition()
156
def read_boundary(self):
157
"""Read the boundary headers defining a new range"""
158
boundary_line = '\r\n'
159
while boundary_line == '\r\n':
160
# RFC2616 19.2 Additional CRLFs may precede the first boundary
162
# To be on the safe side we allow it before any boundary line
163
boundary_line = self._file.readline()
165
if boundary_line == '':
166
# A timeout in the proxy server caused the response to end early.
167
# See launchpad bug 198646.
168
raise errors.HttpBoundaryMissing(
172
if boundary_line != '--' + self._boundary + '\r\n':
173
# rfc822.unquote() incorrectly unquotes strings enclosed in <>
174
# IIS 6 and 7 incorrectly wrap boundary strings in <>
175
# together they make a beautiful bug, which we will be gracious
177
if (self._unquote_boundary(boundary_line) !=
178
'--' + self._boundary + '\r\n'):
179
raise errors.InvalidHttpResponse(
181
"Expected a boundary (%s) line, got '%s'"
182
% (self._boundary, boundary_line))
184
def _unquote_boundary(self, b):
185
return b[:2] + rfc822.unquote(b[2:-2]) + b[-2:]
187
def read_range_definition(self):
188
"""Read a new range definition in a multi parts message.
190
Parse the headers including the empty line following them so that we
191
are ready to read the data itself.
193
self._headers = httplib.HTTPMessage(self._file, seekable=0)
194
# Extract the range definition
195
content_range = self._headers.getheader('content-range', None)
196
if content_range is None:
197
raise errors.InvalidHttpResponse(
199
'Content-Range header missing in a multi-part response')
200
self.set_range_from_header(content_range)
202
def set_range_from_header(self, content_range):
203
"""Helper to set the new range from its description in the headers"""
205
rtype, values = content_range.split()
207
raise errors.InvalidHttpRange(self._path, content_range,
210
raise errors.InvalidHttpRange(self._path, content_range,
211
"Unsupported range type '%s'" % rtype)
213
# We don't need total, but note that it may be either the file size
214
# or '*' if the server can't or doesn't want to return the file
216
start_end, total = values.split('/')
217
start, end = start_end.split('-')
221
raise errors.InvalidHttpRange(self._path, content_range,
222
'Invalid range values')
223
size = end - start + 1
225
raise errors.InvalidHttpRange(self._path, content_range,
226
'Invalid range, size <= 0')
227
self.set_range(start, size)
229
def _checked_read(self, size):
230
"""Read the file checking for short reads.
232
The data read is discarded along the way.
237
data = self._file.read(min(remaining, self._discarded_buf_size))
238
remaining -= len(data)
240
raise errors.ShortReadvError(self._path, pos, size,
244
def _seek_to_next_range(self):
245
# We will cross range boundaries
246
if self._boundary is None:
247
# If we don't have a boundary, we can't find another range
248
raise errors.InvalidRange(self._path, self._pos,
249
"Range (%s, %s) exhausted"
250
% (self._start, self._size))
252
self.read_range_definition()
254
def read(self, size=-1):
255
"""Read size bytes from the current position in the file.
257
Reading across ranges is not supported. We rely on the underlying http
258
client to clean the socket if we leave bytes unread. This may occur for
259
the final boundary line of a multipart response or for any range
260
request not entirely consumed by the client (due to offset coalescing)
262
:param size: The number of bytes to read. Leave unspecified or pass
266
and self._pos == self._start + self._size):
270
self._seek_to_next_range()
271
elif self._pos < self._start:
272
raise errors.InvalidRange(
273
self._path, self._pos,
274
"Can't read %s bytes before range (%s, %s)"
275
% (size, self._start, self._size))
277
if size > 0 and self._pos + size > self._start + self._size:
278
raise errors.InvalidRange(
279
self._path, self._pos,
280
"Can't read %s bytes across range (%s, %s)"
281
% (size, self._start, self._size))
283
# read data from file
287
# Don't read past the range definition
288
limited = self._start + self._size - self._pos
290
limited = min(limited, size)
291
osutils.pumpfile(self._file, buffer, limited, self._max_read_size)
292
data = buffer.getvalue()
294
# Update _pos respecting the data effectively read
295
self._pos += len(data)
298
def seek(self, offset, whence=0):
299
start_pos = self._pos
303
final_pos = start_pos + offset
306
final_pos = self._start + self._size + offset # offset < 0
308
raise errors.InvalidRange(
309
self._path, self._pos,
310
"RangeFile: can't seek from end while size is unknown")
312
raise ValueError("Invalid value %s for whence." % whence)
314
if final_pos < self._pos:
315
# Can't seek backwards
316
raise errors.InvalidRange(
317
self._path, self._pos,
318
'RangeFile: trying to seek backwards to %s' % final_pos)
321
cur_limit = self._start + self._size
322
while final_pos > cur_limit:
323
# We will cross range boundaries
324
remain = cur_limit - self._pos
326
# Finish reading the current range
327
self._checked_read(remain)
328
self._seek_to_next_range()
329
cur_limit = self._start + self._size
331
size = final_pos - self._pos
332
if size > 0: # size can be < 0 if we crossed a range boundary
333
# We don't need the data, just read it and throw it away
334
self._checked_read(size)
340
def handle_response(url, code, msg, data):
341
"""Interpret the code & headers and wrap the provided data in a RangeFile.
343
This is a factory method which returns an appropriate RangeFile based on
344
the code & headers it's given.
346
:param url: The url being processed. Mostly for error reporting
347
:param code: The integer HTTP response code
348
:param msg: An HTTPMessage containing the headers for the response
349
:param data: A file-like object that can be read() to get the
351
:return: A file-like object that can seek()+read() the
352
ranges indicated by the headers.
356
rfile = ResponseFile(url, data)
358
rfile = RangeFile(url, data)
359
content_type = msg.getheader('content-type', None)
360
if content_type is None:
361
# When there is no content-type header we treat the response as
362
# being of type 'application/octet-stream' as per RFC2616 section
364
# Therefore it is obviously not multipart
365
content_type = 'application/octet-stream'
368
is_multipart = (msg.getmaintype() == 'multipart'
369
and msg.getsubtype() == 'byteranges')
372
# Full fledged multipart response
373
rfile.set_boundary(msg.getparam('boundary'))
375
# A response to a range request, but not multipart
376
content_range = msg.getheader('content-range', None)
377
if content_range is None:
378
raise errors.InvalidHttpResponse(url,
379
'Missing the Content-Range header in a 206 range response')
380
rfile.set_range_from_header(content_range)
382
raise errors.InvalidHttpResponse(url,
383
'Unknown response code %s' % code)
added
removed
bzrlib/transport/http/response.py
