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
26
from cStringIO import StringIO
36
# A RangeFile expects the following grammar (simplified to outline the
37
# assumptions we rely upon).
43
# whole_file: [content_length_header] data
45
# single_range: content_range_header data
47
# multiple_range: boundary_header boundary (content_range_header data boundary)+
49
class RangeFile(object):
50
"""File-like object that allow access to partial available data.
52
All accesses should happen sequentially since the acquisition occurs during
53
an http response reception (as sockets can't be seeked, we simulate the
54
seek by just reading and discarding the data).
56
The access pattern is defined by a set of ranges discovered as reading
57
progress. Only one range is available at a given time, so all accesses
58
should happen with monotonically increasing offsets.
61
# in _checked_read() below, we may have to discard several MB in the worst
62
# case. To avoid buffering that much, we read and discard by chunks
63
# instead. The underlying file is either a socket or a StringIO, so reading
64
# 8k chunks should be fine.
65
_discarded_buf_size = 8192
67
# maximum size of read requests -- used to avoid MemoryError issues in recv
68
_max_read_size = 512 * 1024
70
def __init__(self, path, infile):
73
:param path: File url, for error reports.
74
:param infile: File-like socket set at body start.
79
# When using multi parts response, this will be set with the headers
80
# associated with the range currently read.
82
# Default to the whole file of unspecified size
85
def set_range(self, start, size):
86
"""Change the range mapping"""
89
# Set the new _pos since that's what we want to expose
90
self._pos = self._start
92
def set_boundary(self, boundary):
93
"""Define the boundary used in a multi parts message.
95
The file should be at the beginning of the body, the first range
96
definition is read and taken into account.
98
self._boundary = boundary
99
# Decode the headers and setup the first range
101
self.read_range_definition()
103
def read_boundary(self):
104
"""Read the boundary headers defining a new range"""
105
boundary_line = '\r\n'
106
while boundary_line == '\r\n':
107
# RFC2616 19.2 Additional CRLFs may precede the first boundary
109
# To be on the safe side we allow it before any boundary line
110
boundary_line = self._file.readline()
112
if boundary_line == '':
113
# A timeout in the proxy server caused the response to end early.
114
# See launchpad bug 198646.
115
raise errors.HttpBoundaryMissing(
119
if boundary_line != '--' + self._boundary + '\r\n':
120
# rfc822.unquote() incorrectly unquotes strings enclosed in <>
121
# IIS 6 and 7 incorrectly wrap boundary strings in <>
122
# together they make a beautiful bug, which we will be gracious
124
if (self._unquote_boundary(boundary_line) !=
125
'--' + self._boundary + '\r\n'):
126
raise errors.InvalidHttpResponse(
128
"Expected a boundary (%s) line, got '%s'"
129
% (self._boundary, boundary_line))
131
def _unquote_boundary(self, b):
132
return b[:2] + rfc822.unquote(b[2:-2]) + b[-2:]
134
def read_range_definition(self):
135
"""Read a new range definition in a multi parts message.
137
Parse the headers including the empty line following them so that we
138
are ready to read the data itself.
140
self._headers = httplib.HTTPMessage(self._file, seekable=0)
141
# Extract the range definition
142
content_range = self._headers.getheader('content-range', None)
143
if content_range is None:
144
raise errors.InvalidHttpResponse(
146
'Content-Range header missing in a multi-part response')
147
self.set_range_from_header(content_range)
149
def set_range_from_header(self, content_range):
150
"""Helper to set the new range from its description in the headers"""
152
rtype, values = content_range.split()
154
raise errors.InvalidHttpRange(self._path, content_range,
157
raise errors.InvalidHttpRange(self._path, content_range,
158
"Unsupported range type '%s'" % rtype)
160
# We don't need total, but note that it may be either the file size
161
# or '*' if the server can't or doesn't want to return the file
163
start_end, total = values.split('/')
164
start, end = start_end.split('-')
168
raise errors.InvalidHttpRange(self._path, content_range,
169
'Invalid range values')
170
size = end - start + 1
172
raise errors.InvalidHttpRange(self._path, content_range,
173
'Invalid range, size <= 0')
174
self.set_range(start, size)
176
def _checked_read(self, size):
177
"""Read the file checking for short reads.
179
The data read is discarded along the way.
184
data = self._file.read(min(remaining, self._discarded_buf_size))
185
remaining -= len(data)
187
raise errors.ShortReadvError(self._path, pos, size,
191
def _seek_to_next_range(self):
192
# We will cross range boundaries
193
if self._boundary is None:
194
# If we don't have a boundary, we can't find another range
195
raise errors.InvalidRange(self._path, self._pos,
196
"Range (%s, %s) exhausted"
197
% (self._start, self._size))
199
self.read_range_definition()
201
def read(self, size=-1):
202
"""Read size bytes from the current position in the file.
204
Reading across ranges is not supported. We rely on the underlying http
205
client to clean the socket if we leave bytes unread. This may occur for
206
the final boundary line of a multipart response or for any range
207
request not entirely consumed by the client (due to offset coalescing)
209
:param size: The number of bytes to read. Leave unspecified or pass
213
and self._pos == self._start + self._size):
217
self._seek_to_next_range()
218
elif self._pos < self._start:
219
raise errors.InvalidRange(
220
self._path, self._pos,
221
"Can't read %s bytes before range (%s, %s)"
222
% (size, self._start, self._size))
224
if size > 0 and self._pos + size > self._start + self._size:
225
raise errors.InvalidRange(
226
self._path, self._pos,
227
"Can't read %s bytes across range (%s, %s)"
228
% (size, self._start, self._size))
230
# read data from file
234
# Don't read past the range definition
235
limited = self._start + self._size - self._pos
237
limited = min(limited, size)
238
osutils.pumpfile(self._file, buffer, limited, self._max_read_size)
239
data = buffer.getvalue()
241
# Update _pos respecting the data effectively read
242
self._pos += len(data)
245
def seek(self, offset, whence=0):
246
start_pos = self._pos
250
final_pos = start_pos + offset
253
final_pos = self._start + self._size + offset # offset < 0
255
raise errors.InvalidRange(
256
self._path, self._pos,
257
"RangeFile: can't seek from end while size is unknown")
259
raise ValueError("Invalid value %s for whence." % whence)
261
if final_pos < self._pos:
262
# Can't seek backwards
263
raise errors.InvalidRange(
264
self._path, self._pos,
265
'RangeFile: trying to seek backwards to %s' % final_pos)
268
cur_limit = self._start + self._size
269
while final_pos > cur_limit:
270
# We will cross range boundaries
271
remain = cur_limit - self._pos
273
# Finish reading the current range
274
self._checked_read(remain)
275
self._seek_to_next_range()
276
cur_limit = self._start + self._size
278
size = final_pos - self._pos
279
if size > 0: # size can be < 0 if we crossed a range boundary
280
# We don't need the data, just read it and throw it away
281
self._checked_read(size)
287
def handle_response(url, code, msg, data):
288
"""Interpret the code & headers and wrap the provided data in a RangeFile.
290
This is a factory method which returns an appropriate RangeFile based on
291
the code & headers it's given.
293
:param url: The url being processed. Mostly for error reporting
294
:param code: The integer HTTP response code
295
:param msg: An HTTPMessage containing the headers for the response
296
:param data: A file-like object that can be read() to get the
298
:return: A file-like object that can seek()+read() the
299
ranges indicated by the headers.
301
rfile = RangeFile(url, data)
304
size = msg.getheader('content-length', None)
309
rfile.set_range(0, size)
311
content_type = msg.getheader('content-type', None)
312
if content_type is None:
313
# When there is no content-type header we treat the response as
314
# being of type 'application/octet-stream' as per RFC2616 section
316
# Therefore it is obviously not multipart
317
content_type = 'application/octet-stream'
320
is_multipart = (msg.getmaintype() == 'multipart'
321
and msg.getsubtype() == 'byteranges')
324
# Full fledged multipart response
325
rfile.set_boundary(msg.getparam('boundary'))
327
# A response to a range request, but not multipart
328
content_range = msg.getheader('content-range', None)
329
if content_range is None:
330
raise errors.InvalidHttpResponse(url,
331
'Missing the Content-Range header in a 206 range response')
332
rfile.set_range_from_header(content_range)
334
raise errors.InvalidHttpResponse(url,
335
'Unknown response code %s' % code)
added
removed
bzrlib/transport/http/response.py
