~bzr-pqm/bzr/bzr.dev

3059.2.2 by Vincent Ladeuil
Read http responses on demand without buffering the whole body
1
# Copyright (C) 2005, 2006, 2007 Canonical Ltd
1786.1.8 by John Arbash Meinel
[merge] Johan Rydberg test updates
2
#
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.
7
#
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.
12
#
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., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
16
3059.2.12 by Vincent Ladeuil
Spiv review feedback.
17
"""Tests from HTTP response parsing.
18
3059.2.18 by Vincent Ladeuil
Take spiv review comments into account.
19
The handle_response method read the response body of a GET request an returns
20
the corresponding RangeFile.
3059.2.12 by Vincent Ladeuil
Spiv review feedback.
21
22
There are four different kinds of RangeFile:
23
- a whole file whose size is unknown, seen as a simple byte stream,
24
- a whole file whose size is known, we can't read past its end,
25
- a single range file, a part of a file with a start and a size,
26
- a multiple range file, several consecutive parts with known start offset
27
  and size.
28
29
Some properties are common to all kinds:
30
- seek can only be forward (its really a socket underneath),
31
- read can't cross ranges,
32
- successive ranges are taken into account transparently,
3059.2.18 by Vincent Ladeuil
Take spiv review comments into account.
33
3059.2.12 by Vincent Ladeuil
Spiv review feedback.
34
- the expected pattern of use is either seek(offset)+read(size) or a single
3059.2.18 by Vincent Ladeuil
Take spiv review comments into account.
35
  read with no size specified. For multiple range files, multiple read() will
36
  return the corresponding ranges, trying to read further will raise
37
  InvalidHttpResponse.
3059.2.12 by Vincent Ladeuil
Spiv review feedback.
38
"""
1786.1.8 by John Arbash Meinel
[merge] Johan Rydberg test updates
39
40
from cStringIO import StringIO
3059.2.2 by Vincent Ladeuil
Read http responses on demand without buffering the whole body
41
import httplib
1786.1.8 by John Arbash Meinel
[merge] Johan Rydberg test updates
42
3059.2.12 by Vincent Ladeuil
Spiv review feedback.
43
from bzrlib import (
44
    errors,
45
    tests,
46
    )
3104.3.4 by Vincent Ladeuil
Add test.
47
from bzrlib.transport.http import (
48
    response,
49
    _urllib2_wrappers,
50
    )
51
52
53
class ReadSocket(object):
54
    """A socket-like object that can be given a predefined content."""
55
56
    def __init__(self, data):
57
        self.readfile = StringIO(data)
58
59
    def makefile(self, mode='r', bufsize=None):
60
        return self.readfile
61
62
class FakeHTTPConnection(_urllib2_wrappers.HTTPConnection):
63
64
    def __init__(self, sock):
65
        _urllib2_wrappers.HTTPConnection.__init__(self, 'localhost')
66
        # Set the socket to bypass the connection
67
        self.sock = sock
68
69
    def send(self, str):
70
        """Ignores the writes on the socket."""
71
        pass
72
73
74
class TestHTTPConnection(tests.TestCase):
75
76
    def test_cleanup_pipe(self):
77
        sock = ReadSocket("""HTTP/1.1 200 OK\r
78
Content-Type: text/plain; charset=UTF-8\r
79
Content-Length: 18
80
\r
81
0123456789
82
garbage""")
83
        conn = FakeHTTPConnection(sock)
84
        # Simulate the request sending so that the connection will be able to
85
        # read the response.
86
        conn.putrequest('GET', 'http://localhost/fictious')
87
        conn.endheaders()
88
        # Now, get the response
89
        resp = conn.getresponse()
90
        # Read part of the response
91
        self.assertEquals('0123456789\n', resp.read(11))
92
        # Override the thresold to force the warning emission
93
        conn._range_warning_thresold = 6 # There are 7 bytes pending
94
        conn.cleanup_pipe()
95
        self.assertContainsRe(self._get_log(keep_log_file=True),
96
                              'Got a 200 response when asking')
3059.2.12 by Vincent Ladeuil
Spiv review feedback.
97
98
99
class TestRangeFileMixin(object):
100
    """Tests for accessing the first range in a RangeFile."""
101
102
    # A simple string used to represent a file part (also called a range), in
103
    # which offsets are easy to calculate for test writers. It's used as a
3059.2.18 by Vincent Ladeuil
Take spiv review comments into account.
104
    # building block with slight variations but basically 'a' is the first char
3059.2.12 by Vincent Ladeuil
Spiv review feedback.
105
    # of the range and 'z' is the last.
106
    alpha = 'abcdefghijklmnopqrstuvwxyz'
107
108
    def test_can_read_at_first_access(self):
109
        """Test that the just created file can be read."""
110
        self.assertEquals(self.alpha, self._file.read())
111
112
    def test_seek_read(self):
113
        """Test seek/read inside the range."""
114
        f = self._file
115
        start = self.first_range_start
116
        # Before any use, tell() should be at the range start
3059.2.2 by Vincent Ladeuil
Read http responses on demand without buffering the whole body
117
        self.assertEquals(start, f.tell())
3059.2.12 by Vincent Ladeuil
Spiv review feedback.
118
        cur = start # For an overall offset assertion
119
        f.seek(start + 3)
120
        cur += 3
3059.2.2 by Vincent Ladeuil
Read http responses on demand without buffering the whole body
121
        self.assertEquals('def', f.read(3))
3059.2.12 by Vincent Ladeuil
Spiv review feedback.
122
        cur += len('def')
123
        f.seek(4, 1)
124
        cur += 4
125
        self.assertEquals('klmn', f.read(4))
126
        cur += len('klmn')
3059.2.18 by Vincent Ladeuil
Take spiv review comments into account.
127
        # read(0) in the middle of a range
128
        self.assertEquals('', f.read(0))
129
        # seek in place
130
        here = f.tell()
131
        f.seek(0, 1)
132
        self.assertEquals(here, f.tell())
3059.2.12 by Vincent Ladeuil
Spiv review feedback.
133
        self.assertEquals(cur, f.tell())
134
3059.2.18 by Vincent Ladeuil
Take spiv review comments into account.
135
    def test_read_zero(self):
136
        f = self._file
137
        start = self.first_range_start
138
        self.assertEquals('', f.read(0))
139
        f.seek(10, 1)
140
        self.assertEquals('', f.read(0))
141
142
    def test_seek_at_range_end(self):
143
        f = self._file
144
        f.seek(26, 1)
145
146
    def test_read_at_range_end(self):
147
        """Test read behaviour at range end."""
148
        f = self._file
149
        self.assertEquals(self.alpha, f.read())
150
        self.assertEquals('', f.read(0))
151
        self.assertRaises(errors.InvalidRange, f.read, 1)
152
3059.2.12 by Vincent Ladeuil
Spiv review feedback.
153
    def test_unbounded_read_after_seek(self):
154
        f = self._file
155
        f.seek(24, 1)
156
        # Should not cross ranges
157
        self.assertEquals('yz', f.read())
158
159
    def test_seek_backwards(self):
160
        f = self._file
161
        start = self.first_range_start
162
        f.seek(start)
163
        f.read(12)
164
        self.assertRaises(errors.InvalidRange, f.seek, start + 5)
165
166
    def test_seek_outside_single_range(self):
167
        f = self._file
168
        if f._size == -1 or f._boundary is not None:
169
            raise tests.TestNotApplicable('Needs a fully defined range')
170
        # Will seek past the range and then errors out
171
        self.assertRaises(errors.InvalidRange,
172
                          f.seek, self.first_range_start + 27)
173
174
    def test_read_past_end_of_range(self):
175
        f = self._file
176
        if f._size == -1:
177
            raise tests.TestNotApplicable("Can't check an unknown size")
178
        start = self.first_range_start
179
        f.seek(start + 20)
180
        self.assertRaises(errors.InvalidRange, f.read, 10)
181
3059.2.14 by Vincent Ladeuil
Complete coverage by adding tests for more invalid inputs. Fix a
182
    def test_seek_from_end(self):
183
       """Test seeking from the end of the file.
184
185
       The semantic is unclear in case of multiple ranges. Seeking from end
186
       exists only for the http transports, cannot be used if the file size is
187
       unknown and is not used in bzrlib itself. This test must be (and is)
188
       overridden by daughter classes.
189
190
       Reading from end makes sense only when a range has been requested from
191
       the end of the file (see HttpTransportBase._get() when using the
192
       'tail_amount' parameter). The HTTP response can only be a whole file or
193
       a single range.
194
       """
195
       f = self._file
196
       f.seek(-2, 2)
197
       self.assertEquals('yz', f.read())
198
3059.2.12 by Vincent Ladeuil
Spiv review feedback.
199
200
class TestRangeFileSizeUnknown(tests.TestCase, TestRangeFileMixin):
201
    """Test a RangeFile for a whole file whose size is not known."""
202
203
    def setUp(self):
204
        super(TestRangeFileSizeUnknown, self).setUp()
205
        self._file = response.RangeFile('Whole_file_size_known',
206
                                        StringIO(self.alpha))
207
        # We define no range, relying on RangeFile to provide default values
208
        self.first_range_start = 0 # It's the whole file
209
210
    def test_seek_from_end(self):
3059.2.14 by Vincent Ladeuil
Complete coverage by adding tests for more invalid inputs. Fix a
211
        """See TestRangeFileMixin.test_seek_from_end.
212
213
        The end of the file can't be determined since the size is unknown.
214
        """
3059.2.12 by Vincent Ladeuil
Spiv review feedback.
215
        self.assertRaises(errors.InvalidRange, self._file.seek, -1, 2)
216
3059.2.18 by Vincent Ladeuil
Take spiv review comments into account.
217
    def test_read_at_range_end(self):
218
        """Test read behaviour at range end."""
219
        f = self._file
220
        self.assertEquals(self.alpha, f.read())
221
        self.assertEquals('', f.read(0))
222
        self.assertEquals('', f.read(1))
3059.2.12 by Vincent Ladeuil
Spiv review feedback.
223
224
class TestRangeFileSizeKnown(tests.TestCase, TestRangeFileMixin):
225
    """Test a RangeFile for a whole file whose size is known."""
226
227
    def setUp(self):
228
        super(TestRangeFileSizeKnown, self).setUp()
229
        self._file = response.RangeFile('Whole_file_size_known',
230
                                        StringIO(self.alpha))
231
        self._file.set_range(0, len(self.alpha))
232
        self.first_range_start = 0 # It's the whole file
233
234
235
class TestRangeFileSingleRange(tests.TestCase, TestRangeFileMixin):
236
    """Test a RangeFile for a single range."""
237
238
    def setUp(self):
239
        super(TestRangeFileSingleRange, self).setUp()
240
        self._file = response.RangeFile('Single_range_file',
241
                                        StringIO(self.alpha))
242
        self.first_range_start = 15
243
        self._file.set_range(self.first_range_start, len(self.alpha))
244
245
3059.2.14 by Vincent Ladeuil
Complete coverage by adding tests for more invalid inputs. Fix a
246
    def test_read_before_range(self):
247
        # This can't occur under normal circumstances, we have to force it
248
        f = self._file
249
        f._pos = 0 # Force an invalid pos
250
        self.assertRaises(errors.InvalidRange, f.read, 2)
251
3146.3.2 by Vincent Ladeuil
Fix #179368 by keeping the current range hint on ShortReadvErrors.
252
class TestRangeFileMultipleRanges(tests.TestCase, TestRangeFileMixin):
3059.2.18 by Vincent Ladeuil
Take spiv review comments into account.
253
    """Test a RangeFile for multiple ranges.
254
255
    The RangeFile used for the tests contains three ranges:
256
257
    - at offset 25: alpha
258
    - at offset 100: alpha
259
    - at offset 126: alpha.upper()
260
261
    The two last ranges are contiguous. This only rarely occurs (should not in
262
    fact) in real uses but may lead to hard to track bugs.
263
    """
3059.2.12 by Vincent Ladeuil
Spiv review feedback.
264
265
    def setUp(self):
3146.3.2 by Vincent Ladeuil
Fix #179368 by keeping the current range hint on ShortReadvErrors.
266
        super(TestRangeFileMultipleRanges, self).setUp()
3059.2.12 by Vincent Ladeuil
Spiv review feedback.
267
268
        boundary = 'separation'
269
270
        content = ''
271
        self.first_range_start = 25
272
        file_size = 200 # big enough to encompass all ranges
273
        for (start, part) in [(self.first_range_start, self.alpha),
274
                              # Two contiguous ranges
275
                              (100, self.alpha),
276
                              (126, self.alpha.upper())]:
277
            content += self._multipart_byterange(part, start, boundary,
278
                                                 file_size)
279
        # Final boundary
280
        content += self._boundary_line(boundary)
281
282
        self._file = response.RangeFile('Multiple_ranges_file',
283
                                        StringIO(content))
284
        # Ranges are set by decoding the range headers, the RangeFile user is
285
        # supposed to call the following before using seek or read since it
286
        # requires knowing the *response* headers (in that case the boundary
287
        # which is part of the Content-Type header).
288
        self._file.set_boundary(boundary)
289
290
    def _boundary_line(self, boundary):
291
        """Helper to build the formatted boundary line."""
292
        return '--' + boundary + '\r\n'
293
294
    def _multipart_byterange(self, data, offset, boundary, file_size='*'):
295
        """Encode a part of a file as a multipart/byterange MIME type.
296
297
        When a range request is issued, the HTTP response body can be
298
        decomposed in parts, each one representing a range (start, size) in a
299
        file.
300
301
        :param data: The payload.
302
        :param offset: where data starts in the file
303
        :param boundary: used to separate the parts
304
        :param file_size: the size of the file containing the range (default to
305
            '*' meaning unknown)
306
307
        :return: a string containing the data encoded as it will appear in the
308
            HTTP response body.
309
        """
310
        bline = self._boundary_line(boundary)
311
        # Each range begins with a boundary line
312
        range = bline
313
        # A range is described by a set of headers, but only 'Content-Range' is
314
        # required for our implementation (TestHandleResponse below will
315
        # exercise ranges with multiple or missing headers')
316
        range += 'Content-Range: bytes %d-%d/%d\r\n' % (offset,
317
                                                        offset+len(data)-1,
318
                                                        file_size)
319
        range += '\r\n'
320
        # Finally the raw bytes
321
        range += data
322
        return range
323
324
    def test_read_all_ranges(self):
325
        f = self._file
326
        self.assertEquals(self.alpha, f.read()) # Read first range
327
        f.seek(100) # Trigger the second range recognition
328
        self.assertEquals(self.alpha, f.read()) # Read second range
329
        self.assertEquals(126, f.tell())
330
        f.seek(126) # Start of third range which is also the current pos !
331
        self.assertEquals('A', f.read(1))
332
        f.seek(10, 1)
333
        self.assertEquals('LMN', f.read(3))
334
3059.2.14 by Vincent Ladeuil
Complete coverage by adding tests for more invalid inputs. Fix a
335
    def test_seek_from_end(self):
336
        """See TestRangeFileMixin.test_seek_from_end."""
337
        # The actual implementation will seek from end for the first range only
3059.2.18 by Vincent Ladeuil
Take spiv review comments into account.
338
        # and then fail. Since seeking from end is intended to be used for a
3059.2.14 by Vincent Ladeuil
Complete coverage by adding tests for more invalid inputs. Fix a
339
        # single range only anyway, this test just document the actual
340
        # behaviour.
341
        f = self._file
342
        f.seek(-2, 2)
343
        self.assertEquals('yz', f.read())
344
        self.assertRaises(errors.InvalidRange, f.seek, -2, 2)
345
3059.2.12 by Vincent Ladeuil
Spiv review feedback.
346
    def test_seek_into_void(self):
347
        f = self._file
348
        start = self.first_range_start
349
        f.seek(start)
3059.2.2 by Vincent Ladeuil
Read http responses on demand without buffering the whole body
350
        # Seeking to a point between two ranges is possible (only once) but
351
        # reading there is forbidden
3059.2.12 by Vincent Ladeuil
Spiv review feedback.
352
        f.seek(start + 40)
3059.2.2 by Vincent Ladeuil
Read http responses on demand without buffering the whole body
353
        # We crossed a range boundary, so now the file is positioned at the
354
        # start of the new range (i.e. trying to seek below 100 will error out)
355
        f.seek(100)
3059.2.7 by Vincent Ladeuil
Allow pycurl users to watch the blinkenlights and fix a bug when ranges are contiguous.
356
        f.seek(125)
357
3059.2.18 by Vincent Ladeuil
Take spiv review comments into account.
358
    def test_seek_across_ranges(self):
3059.2.12 by Vincent Ladeuil
Spiv review feedback.
359
        f = self._file
360
        start = self.first_range_start
361
        f.seek(126) # skip the two first ranges
362
        self.assertEquals('AB', f.read(2))
363
3146.3.2 by Vincent Ladeuil
Fix #179368 by keeping the current range hint on ShortReadvErrors.
364
    def test_checked_read_dont_overflow_buffers(self):
365
        f = self._file
366
        start = self.first_range_start
367
        # We force a very low value to exercise all code paths in _checked_read
368
        f._discarded_buf_size = 8
369
        f.seek(126) # skip the two first ranges
370
        self.assertEquals('AB', f.read(2))
371
3059.2.12 by Vincent Ladeuil
Spiv review feedback.
372
    def test_seek_twice_between_ranges(self):
373
        f = self._file
374
        start = self.first_range_start
375
        f.seek(start + 40) # Past the first range but before the second
3059.2.2 by Vincent Ladeuil
Read http responses on demand without buffering the whole body
376
        # Now the file is positioned at the second range start (100)
3059.2.12 by Vincent Ladeuil
Spiv review feedback.
377
        self.assertRaises(errors.InvalidRange, f.seek, start + 41)
378
3059.2.18 by Vincent Ladeuil
Take spiv review comments into account.
379
    def test_seek_at_range_end(self):
380
        """Test seek behavior at range end."""
381
        f = self._file
382
        f.seek(25 + 25)
383
        f.seek(100 + 25)
384
        f.seek(126 + 25)
385
386
    def test_read_at_range_end(self):
387
        f = self._file
388
        self.assertEquals(self.alpha, f.read())
389
        self.assertEquals(self.alpha, f.read())
390
        self.assertEquals(self.alpha.upper(), f.read())
391
        self.assertRaises(errors.InvalidHttpResponse, f.read, 1)
392
3059.2.12 by Vincent Ladeuil
Spiv review feedback.
393
3059.2.14 by Vincent Ladeuil
Complete coverage by adding tests for more invalid inputs. Fix a
394
class TestRangeFileVarious(tests.TestCase):
395
    """Tests RangeFile aspects not covered elsewhere."""
396
397
    def test_seek_whence(self):
398
        """Test the seek whence parameter values."""
399
        f = response.RangeFile('foo', StringIO('abc'))
400
        f.set_range(0, 3)
401
        f.seek(0)
402
        f.seek(1, 1)
403
        f.seek(-1, 2)
404
        self.assertRaises(ValueError, f.seek, 0, 14)
3059.2.2 by Vincent Ladeuil
Read http responses on demand without buffering the whole body
405
406
    def test_range_syntax(self):
3059.2.14 by Vincent Ladeuil
Complete coverage by adding tests for more invalid inputs. Fix a
407
        """Test the Content-Range scanning."""
3059.2.2 by Vincent Ladeuil
Read http responses on demand without buffering the whole body
408
3059.2.14 by Vincent Ladeuil
Complete coverage by adding tests for more invalid inputs. Fix a
409
        f = response.RangeFile('foo', StringIO())
3059.2.2 by Vincent Ladeuil
Read http responses on demand without buffering the whole body
410
411
        def ok(expected, header_value):
3059.2.14 by Vincent Ladeuil
Complete coverage by adding tests for more invalid inputs. Fix a
412
            f.set_range_from_header(header_value)
3059.2.2 by Vincent Ladeuil
Read http responses on demand without buffering the whole body
413
            # Slightly peek under the covers to get the size
3059.2.14 by Vincent Ladeuil
Complete coverage by adding tests for more invalid inputs. Fix a
414
            self.assertEquals(expected, (f.tell(), f._size))
3059.2.2 by Vincent Ladeuil
Read http responses on demand without buffering the whole body
415
416
        ok((1, 10), 'bytes 1-10/11')
417
        ok((1, 10), 'bytes 1-10/*')
418
        ok((12, 2), '\tbytes 12-13/*')
419
        ok((28, 1), '  bytes 28-28/*')
420
        ok((2123, 2120), 'bytes  2123-4242/12310')
3059.2.14 by Vincent Ladeuil
Complete coverage by adding tests for more invalid inputs. Fix a
421
        ok((1, 10), 'bytes 1-10/ttt') # We don't check total (ttt)
3059.2.2 by Vincent Ladeuil
Read http responses on demand without buffering the whole body
422
423
        def nok(header_value):
424
            self.assertRaises(errors.InvalidHttpRange,
3059.2.14 by Vincent Ladeuil
Complete coverage by adding tests for more invalid inputs. Fix a
425
                              f.set_range_from_header, header_value)
3059.2.2 by Vincent Ladeuil
Read http responses on demand without buffering the whole body
426
3059.2.14 by Vincent Ladeuil
Complete coverage by adding tests for more invalid inputs. Fix a
427
        nok('bytes 10-2/3')
3059.2.2 by Vincent Ladeuil
Read http responses on demand without buffering the whole body
428
        nok('chars 1-2/3')
429
        nok('bytes xx-yyy/zzz')
430
        nok('bytes xx-12/zzz')
431
        nok('bytes 11-yy/zzz')
3059.2.14 by Vincent Ladeuil
Complete coverage by adding tests for more invalid inputs. Fix a
432
        nok('bytes10-2/3')
1786.1.21 by John Arbash Meinel
(broken) Work on factoring out handle_response so we can test with fake headers.
433
3059.2.12 by Vincent Ladeuil
Spiv review feedback.
434
1786.1.21 by John Arbash Meinel
(broken) Work on factoring out handle_response so we can test with fake headers.
435
# Taken from real request responses
1786.1.26 by John Arbash Meinel
Update and test handle_response.
436
_full_text_response = (200, """HTTP/1.1 200 OK\r
1786.1.21 by John Arbash Meinel
(broken) Work on factoring out handle_response so we can test with fake headers.
437
Date: Tue, 11 Jul 2006 04:32:56 GMT\r
438
Server: Apache/2.0.54 (Fedora)\r
439
Last-Modified: Sun, 23 Apr 2006 19:35:20 GMT\r
440
ETag: "56691-23-38e9ae00"\r
441
Accept-Ranges: bytes\r
442
Content-Length: 35\r
443
Connection: close\r
444
Content-Type: text/plain; charset=UTF-8\r
445
\r
1786.1.25 by John Arbash Meinel
Test that we can extract headers properly.
446
""", """Bazaar-NG meta directory, format 1
447
""")
448
449
1786.1.26 by John Arbash Meinel
Update and test handle_response.
450
_single_range_response = (206, """HTTP/1.1 206 Partial Content\r
1786.1.21 by John Arbash Meinel
(broken) Work on factoring out handle_response so we can test with fake headers.
451
Date: Tue, 11 Jul 2006 04:45:22 GMT\r
452
Server: Apache/2.0.54 (Fedora)\r
453
Last-Modified: Thu, 06 Jul 2006 20:22:05 GMT\r
454
ETag: "238a3c-16ec2-805c5540"\r
455
Accept-Ranges: bytes\r
456
Content-Length: 100\r
1786.1.26 by John Arbash Meinel
Update and test handle_response.
457
Content-Range: bytes 100-199/93890\r
1786.1.21 by John Arbash Meinel
(broken) Work on factoring out handle_response so we can test with fake headers.
458
Connection: close\r
459
Content-Type: text/plain; charset=UTF-8\r
460
\r
1786.1.25 by John Arbash Meinel
Test that we can extract headers properly.
461
""", """mbp@sourcefrog.net-20050309040815-13242001617e4a06
1786.1.26 by John Arbash Meinel
Update and test handle_response.
462
mbp@sourcefrog.net-20050309040929-eee0eb3e6d1e762""")
463
464
2070.1.1 by John Arbash Meinel
Fix bug #62473 by not requiring content-type in range responses
465
_single_range_no_content_type = (206, """HTTP/1.1 206 Partial Content\r
466
Date: Tue, 11 Jul 2006 04:45:22 GMT\r
467
Server: Apache/2.0.54 (Fedora)\r
468
Last-Modified: Thu, 06 Jul 2006 20:22:05 GMT\r
469
ETag: "238a3c-16ec2-805c5540"\r
470
Accept-Ranges: bytes\r
471
Content-Length: 100\r
472
Content-Range: bytes 100-199/93890\r
473
Connection: close\r
474
\r
475
""", """mbp@sourcefrog.net-20050309040815-13242001617e4a06
476
mbp@sourcefrog.net-20050309040929-eee0eb3e6d1e762""")
477
478
1786.1.26 by John Arbash Meinel
Update and test handle_response.
479
_multipart_range_response = (206, """HTTP/1.1 206 Partial Content\r
1786.1.21 by John Arbash Meinel
(broken) Work on factoring out handle_response so we can test with fake headers.
480
Date: Tue, 11 Jul 2006 04:49:48 GMT\r
481
Server: Apache/2.0.54 (Fedora)\r
482
Last-Modified: Thu, 06 Jul 2006 20:22:05 GMT\r
483
ETag: "238a3c-16ec2-805c5540"\r
484
Accept-Ranges: bytes\r
485
Content-Length: 1534\r
486
Connection: close\r
487
Content-Type: multipart/byteranges; boundary=418470f848b63279b\r
488
\r
1786.1.25 by John Arbash Meinel
Test that we can extract headers properly.
489
\r""", """--418470f848b63279b\r
1786.1.21 by John Arbash Meinel
(broken) Work on factoring out handle_response so we can test with fake headers.
490
Content-type: text/plain; charset=UTF-8\r
491
Content-range: bytes 0-254/93890\r
492
\r
493
mbp@sourcefrog.net-20050309040815-13242001617e4a06
494
mbp@sourcefrog.net-20050309040929-eee0eb3e6d1e7627
495
mbp@sourcefrog.net-20050309040957-6cad07f466bb0bb8
496
mbp@sourcefrog.net-20050309041501-c840e09071de3b67
497
mbp@sourcefrog.net-20050309044615-c24a3250be83220a
498
\r
499
--418470f848b63279b\r
500
Content-type: text/plain; charset=UTF-8\r
501
Content-range: bytes 1000-2049/93890\r
502
\r
503
40-fd4ec249b6b139ab
504
mbp@sourcefrog.net-20050311063625-07858525021f270b
505
mbp@sourcefrog.net-20050311231934-aa3776aff5200bb9
506
mbp@sourcefrog.net-20050311231953-73aeb3a131c3699a
507
mbp@sourcefrog.net-20050311232353-f5e33da490872c6a
508
mbp@sourcefrog.net-20050312071639-0a8f59a34a024ff0
509
mbp@sourcefrog.net-20050312073432-b2c16a55e0d6e9fb
510
mbp@sourcefrog.net-20050312073831-a47c3335ece1920f
511
mbp@sourcefrog.net-20050312085412-13373aa129ccbad3
512
mbp@sourcefrog.net-20050313052251-2bf004cb96b39933
513
mbp@sourcefrog.net-20050313052856-3edd84094687cb11
514
mbp@sourcefrog.net-20050313053233-e30a4f28aef48f9d
515
mbp@sourcefrog.net-20050313053853-7c64085594ff3072
516
mbp@sourcefrog.net-20050313054757-a86c3f5871069e22
517
mbp@sourcefrog.net-20050313061422-418f1f73b94879b9
518
mbp@sourcefrog.net-20050313120651-497bd231b19df600
519
mbp@sourcefrog.net-20050314024931-eae0170ef25a5d1a
520
mbp@sourcefrog.net-20050314025438-d52099f915fe65fc
521
mbp@sourcefrog.net-20050314025539-637a636692c055cf
522
mbp@sourcefrog.net-20050314025737-55eb441f430ab4ba
523
mbp@sourcefrog.net-20050314025901-d74aa93bb7ee8f62
524
mbp@source\r
1979.1.1 by John Arbash Meinel
Fix bug #57723, parse boundary="" correctly, since Squid uses it
525
--418470f848b63279b--\r
526
""")
527
3059.2.14 by Vincent Ladeuil
Complete coverage by adding tests for more invalid inputs. Fix a
528
1979.1.1 by John Arbash Meinel
Fix bug #57723, parse boundary="" correctly, since Squid uses it
529
_multipart_squid_range_response = (206, """HTTP/1.0 206 Partial Content\r
530
Date: Thu, 31 Aug 2006 21:16:22 GMT\r
531
Server: Apache/2.2.2 (Unix) DAV/2\r
532
Last-Modified: Thu, 31 Aug 2006 17:57:06 GMT\r
533
Accept-Ranges: bytes\r
534
Content-Type: multipart/byteranges; boundary="squid/2.5.STABLE12:C99323425AD4FE26F726261FA6C24196"\r
535
Content-Length: 598\r
536
X-Cache: MISS from localhost.localdomain\r
537
X-Cache-Lookup: HIT from localhost.localdomain:3128\r
538
Proxy-Connection: keep-alive\r
539
\r
540
""",
541
"""\r
542
--squid/2.5.STABLE12:C99323425AD4FE26F726261FA6C24196\r
543
Content-Type: text/plain\r
544
Content-Range: bytes 0-99/18672\r
545
\r
546
# bzr knit index 8
547
548
scott@netsplit.com-20050708230047-47c7868f276b939f fulltext 0 863  :
549
scott@netsp\r
550
--squid/2.5.STABLE12:C99323425AD4FE26F726261FA6C24196\r
551
Content-Type: text/plain\r
552
Content-Range: bytes 300-499/18672\r
553
\r
554
com-20050708231537-2b124b835395399a :
555
scott@netsplit.com-20050820234126-551311dbb7435b51 line-delta 1803 479 .scott@netsplit.com-20050820232911-dc4322a084eadf7e :
556
scott@netsplit.com-20050821213706-c86\r
557
--squid/2.5.STABLE12:C99323425AD4FE26F726261FA6C24196--\r
1786.1.25 by John Arbash Meinel
Test that we can extract headers properly.
558
""")
559
560
1786.1.26 by John Arbash Meinel
Update and test handle_response.
561
# This is made up
3059.2.2 by Vincent Ladeuil
Read http responses on demand without buffering the whole body
562
_full_text_response_no_content_type = (200, """HTTP/1.1 200 OK\r
563
Date: Tue, 11 Jul 2006 04:32:56 GMT\r
564
Server: Apache/2.0.54 (Fedora)\r
565
Last-Modified: Sun, 23 Apr 2006 19:35:20 GMT\r
566
ETag: "56691-23-38e9ae00"\r
567
Accept-Ranges: bytes\r
568
Content-Length: 35\r
569
Connection: close\r
570
\r
571
""", """Bazaar-NG meta directory, format 1
572
""")
573
574
3059.2.14 by Vincent Ladeuil
Complete coverage by adding tests for more invalid inputs. Fix a
575
_full_text_response_no_content_length = (200, """HTTP/1.1 200 OK\r
576
Date: Tue, 11 Jul 2006 04:32:56 GMT\r
577
Server: Apache/2.0.54 (Fedora)\r
578
Last-Modified: Sun, 23 Apr 2006 19:35:20 GMT\r
579
ETag: "56691-23-38e9ae00"\r
580
Accept-Ranges: bytes\r
581
Connection: close\r
582
Content-Type: text/plain; charset=UTF-8\r
583
\r
584
""", """Bazaar-NG meta directory, format 1
585
""")
586
587
3059.2.2 by Vincent Ladeuil
Read http responses on demand without buffering the whole body
588
_single_range_no_content_range = (206, """HTTP/1.1 206 Partial Content\r
589
Date: Tue, 11 Jul 2006 04:45:22 GMT\r
590
Server: Apache/2.0.54 (Fedora)\r
591
Last-Modified: Thu, 06 Jul 2006 20:22:05 GMT\r
592
ETag: "238a3c-16ec2-805c5540"\r
593
Accept-Ranges: bytes\r
594
Content-Length: 100\r
595
Connection: close\r
596
\r
597
""", """mbp@sourcefrog.net-20050309040815-13242001617e4a06
598
mbp@sourcefrog.net-20050309040929-eee0eb3e6d1e762""")
599
600
3059.2.14 by Vincent Ladeuil
Complete coverage by adding tests for more invalid inputs. Fix a
601
_single_range_response_truncated = (206, """HTTP/1.1 206 Partial Content\r
602
Date: Tue, 11 Jul 2006 04:45:22 GMT\r
603
Server: Apache/2.0.54 (Fedora)\r
604
Last-Modified: Thu, 06 Jul 2006 20:22:05 GMT\r
605
ETag: "238a3c-16ec2-805c5540"\r
606
Accept-Ranges: bytes\r
607
Content-Length: 100\r
608
Content-Range: bytes 100-199/93890\r
609
Connection: close\r
610
Content-Type: text/plain; charset=UTF-8\r
611
\r
612
""", """mbp@sourcefrog.net-20050309040815-13242001617e4a06""")
613
614
1786.1.26 by John Arbash Meinel
Update and test handle_response.
615
_invalid_response = (444, """HTTP/1.1 444 Bad Response\r
616
Date: Tue, 11 Jul 2006 04:32:56 GMT\r
617
Connection: close\r
618
Content-Type: text/html; charset=iso-8859-1\r
619
\r
620
""", """<!DOCTYPE HTML PUBLIC "-//IETF//DTD HTML 2.0//EN">
621
<html><head>
622
<title>404 Not Found</title>
623
</head><body>
624
<h1>Not Found</h1>
625
<p>I don't know what I'm doing</p>
626
<hr>
627
</body></html>
628
""")
629
630
3059.2.14 by Vincent Ladeuil
Complete coverage by adding tests for more invalid inputs. Fix a
631
_multipart_no_content_range = (206, """HTTP/1.0 206 Partial Content\r
632
Content-Type: multipart/byteranges; boundary=THIS_SEPARATES\r
633
Content-Length: 598\r
634
\r
635
""",
636
"""\r
637
--THIS_SEPARATES\r
638
Content-Type: text/plain\r
639
\r
640
# bzr knit index 8
641
--THIS_SEPARATES\r
642
""")
643
644
645
_multipart_no_boundary = (206, """HTTP/1.0 206 Partial Content\r
646
Content-Type: multipart/byteranges; boundary=THIS_SEPARATES\r
647
Content-Length: 598\r
648
\r
649
""",
650
"""\r
651
--THIS_SEPARATES\r
652
Content-Type: text/plain\r
653
Content-Range: bytes 0-18/18672\r
654
\r
655
# bzr knit index 8
656
657
The range ended at the line above, this text is garbage instead of a boundary
658
line
659
""")
660
661
3059.2.12 by Vincent Ladeuil
Spiv review feedback.
662
class TestHandleResponse(tests.TestCase):
3059.2.2 by Vincent Ladeuil
Read http responses on demand without buffering the whole body
663
664
    def _build_HTTPMessage(self, raw_headers):
665
        status_and_headers = StringIO(raw_headers)
3059.2.11 by Vincent Ladeuil
Fix typos mentioned by spiv.
666
        # Get rid of the status line
3059.2.2 by Vincent Ladeuil
Read http responses on demand without buffering the whole body
667
        status_and_headers.readline()
668
        msg = httplib.HTTPMessage(status_and_headers)
669
        return msg
670
1786.1.26 by John Arbash Meinel
Update and test handle_response.
671
    def get_response(self, a_response):
672
        """Process a supplied response, and return the result."""
3059.2.2 by Vincent Ladeuil
Read http responses on demand without buffering the whole body
673
        code, raw_headers, body = a_response
674
        msg = self._build_HTTPMessage(raw_headers)
675
        return response.handle_response('http://foo', code, msg,
1786.1.26 by John Arbash Meinel
Update and test handle_response.
676
                                        StringIO(a_response[2]))
677
678
    def test_full_text(self):
679
        out = self.get_response(_full_text_response)
680
        # It is a StringIO from the original data
681
        self.assertEqual(_full_text_response[2], out.read())
682
683
    def test_single_range(self):
684
        out = self.get_response(_single_range_response)
685
686
        out.seek(100)
687
        self.assertEqual(_single_range_response[2], out.read(100))
688
2070.1.1 by John Arbash Meinel
Fix bug #62473 by not requiring content-type in range responses
689
    def test_single_range_no_content(self):
690
        out = self.get_response(_single_range_no_content_type)
691
692
        out.seek(100)
693
        self.assertEqual(_single_range_no_content_type[2], out.read(100))
694
3059.2.14 by Vincent Ladeuil
Complete coverage by adding tests for more invalid inputs. Fix a
695
    def test_single_range_truncated(self):
696
        out = self.get_response(_single_range_response_truncated)
697
        # Content-Range declares 100 but only 51 present
698
        self.assertRaises(errors.ShortReadvError, out.seek, out.tell() + 51)
699
1786.1.26 by John Arbash Meinel
Update and test handle_response.
700
    def test_multi_range(self):
701
        out = self.get_response(_multipart_range_response)
702
703
        # Just make sure we can read the right contents
704
        out.seek(0)
705
        out.read(255)
706
707
        out.seek(1000)
708
        out.read(1050)
709
1979.1.1 by John Arbash Meinel
Fix bug #57723, parse boundary="" correctly, since Squid uses it
710
    def test_multi_squid_range(self):
711
        out = self.get_response(_multipart_squid_range_response)
712
713
        # Just make sure we can read the right contents
714
        out.seek(0)
715
        out.read(100)
716
717
        out.seek(300)
718
        out.read(200)
719
1786.1.26 by John Arbash Meinel
Update and test handle_response.
720
    def test_invalid_response(self):
721
        self.assertRaises(errors.InvalidHttpResponse,
3059.2.2 by Vincent Ladeuil
Read http responses on demand without buffering the whole body
722
                          self.get_response, _invalid_response)
1786.1.26 by John Arbash Meinel
Update and test handle_response.
723
724
    def test_full_text_no_content_type(self):
725
        # We should not require Content-Type for a full response
3059.2.2 by Vincent Ladeuil
Read http responses on demand without buffering the whole body
726
        code, raw_headers, body = _full_text_response_no_content_type
727
        msg = self._build_HTTPMessage(raw_headers)
728
        out = response.handle_response('http://foo', code, msg, StringIO(body))
729
        self.assertEqual(body, out.read())
1786.1.26 by John Arbash Meinel
Update and test handle_response.
730
3059.2.14 by Vincent Ladeuil
Complete coverage by adding tests for more invalid inputs. Fix a
731
    def test_full_text_no_content_length(self):
732
        code, raw_headers, body = _full_text_response_no_content_length
733
        msg = self._build_HTTPMessage(raw_headers)
734
        out = response.handle_response('http://foo', code, msg, StringIO(body))
735
        self.assertEqual(body, out.read())
736
1786.1.26 by John Arbash Meinel
Update and test handle_response.
737
    def test_missing_content_range(self):
3059.2.2 by Vincent Ladeuil
Read http responses on demand without buffering the whole body
738
        code, raw_headers, body = _single_range_no_content_range
739
        msg = self._build_HTTPMessage(raw_headers)
1786.1.26 by John Arbash Meinel
Update and test handle_response.
740
        self.assertRaises(errors.InvalidHttpResponse,
3059.2.2 by Vincent Ladeuil
Read http responses on demand without buffering the whole body
741
                          response.handle_response,
3059.2.14 by Vincent Ladeuil
Complete coverage by adding tests for more invalid inputs. Fix a
742
                          'http://bogus', code, msg, StringIO(body))
743
744
    def test_multipart_no_content_range(self):
745
        code, raw_headers, body = _multipart_no_content_range
746
        msg = self._build_HTTPMessage(raw_headers)
747
        self.assertRaises(errors.InvalidHttpResponse,
748
                          response.handle_response,
749
                          'http://bogus', code, msg, StringIO(body))
750
751
    def test_multipart_no_boundary(self):
752
        out = self.get_response(_multipart_no_boundary)
753
        out.read()  # Read the whole range
754
        # Fail to find the boundary line
755
        self.assertRaises(errors.InvalidHttpResponse, out.seek, 1, 1)