~bzr-pqm/bzr/bzr.dev

« back to all changes in this revision

Viewing changes to bzrlib/tests/test_xml.py

  • Committer: Andrew Bennetts
  • Date: 2010-10-08 04:25:10 UTC
  • mto: This revision was merged to the branch mainline in revision 5472.
  • Revision ID: andrew.bennetts@canonical.com-20101008042510-sg9vdhmnggilzxsk
Fix stray TAB in source.

Show diffs side-by-side

added added

removed removed

Lines of Context:
1
 
# Copyright (C) 2005 Canonical Ltd
 
1
# Copyright (C) 2005-2010 Canonical Ltd
2
2
#
3
3
# This program is free software; you can redistribute it and/or modify
4
4
# it under the terms of the GNU General Public License as published by
12
12
#
13
13
# You should have received a copy of the GNU General Public License
14
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
 
15
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
16
16
 
17
17
from cStringIO import StringIO
18
18
 
 
19
from bzrlib import (
 
20
    errors,
 
21
    fifo_cache,
 
22
    inventory,
 
23
    xml6,
 
24
    xml7,
 
25
    xml8,
 
26
    serializer,
 
27
    )
19
28
from bzrlib.tests import TestCase
20
29
from bzrlib.inventory import Inventory, InventoryEntry
21
30
from bzrlib.xml4 import serializer_v4
73
82
"""
74
83
 
75
84
_committed_inv_v5 = """<inventory>
76
 
<file file_id="bar-20050901064931-73b4b1138abc9cd2" 
77
 
      name="bar" parent_id="TREE_ROOT" 
78
 
      revision="mbp@foo-123123"/>
 
85
<file file_id="bar-20050901064931-73b4b1138abc9cd2"
 
86
      name="bar" parent_id="TREE_ROOT"
 
87
      revision="mbp@foo-123123"
 
88
      text_sha1="A" text_size="1"/>
79
89
<directory name="subdir"
80
90
           file_id="foo-20050801201819-4139aa4a272f4250"
81
 
           parent_id="TREE_ROOT" 
 
91
           parent_id="TREE_ROOT"
82
92
           revision="mbp@foo-00"/>
83
 
<file executable="yes" file_id="bar-20050824000535-6bc48cfad47ed134" 
84
 
      name="bar" parent_id="foo-20050801201819-4139aa4a272f4250" 
85
 
      revision="mbp@foo-00"/>
 
93
<file executable="yes" file_id="bar-20050824000535-6bc48cfad47ed134"
 
94
      name="bar" parent_id="foo-20050801201819-4139aa4a272f4250"
 
95
      revision="mbp@foo-00"
 
96
      text_sha1="B" text_size="0"/>
86
97
</inventory>
87
98
"""
88
99
 
89
100
_basis_inv_v5 = """<inventory revision_id="mbp@sourcefrog.net-20050905063503-43948f59fa127d92">
90
 
<file file_id="bar-20050901064931-73b4b1138abc9cd2" 
91
 
      name="bar" parent_id="TREE_ROOT" 
 
101
<file file_id="bar-20050901064931-73b4b1138abc9cd2"
 
102
      name="bar" parent_id="TREE_ROOT"
92
103
      revision="mbp@foo-123123"/>
93
104
<directory name="subdir"
94
105
           file_id="foo-20050801201819-4139aa4a272f4250"
95
 
           parent_id="TREE_ROOT" 
 
106
           parent_id="TREE_ROOT"
96
107
           revision="mbp@foo-00"/>
97
 
<file file_id="bar-20050824000535-6bc48cfad47ed134" 
98
 
      name="bar" parent_id="foo-20050801201819-4139aa4a272f4250" 
 
108
<file file_id="bar-20050824000535-6bc48cfad47ed134"
 
109
      name="bar" parent_id="foo-20050801201819-4139aa4a272f4250"
99
110
      revision="mbp@foo-00"/>
100
111
</inventory>
101
112
"""
115
126
 
116
127
# DO NOT REFLOW THIS. Its the exact inventory we want.
117
128
_expected_inv_v5 = """<inventory format="5">
118
 
<file file_id="bar-20050901064931-73b4b1138abc9cd2" name="bar" revision="mbp@foo-123123" />
 
129
<file file_id="bar-20050901064931-73b4b1138abc9cd2" name="bar" revision="mbp@foo-123123" text_sha1="A" text_size="1" />
119
130
<directory file_id="foo-20050801201819-4139aa4a272f4250" name="subdir" revision="mbp@foo-00" />
120
 
<file executable="yes" file_id="bar-20050824000535-6bc48cfad47ed134" name="bar" parent_id="foo-20050801201819-4139aa4a272f4250" revision="mbp@foo-00" />
 
131
<file executable="yes" file_id="bar-20050824000535-6bc48cfad47ed134" name="bar" parent_id="foo-20050801201819-4139aa4a272f4250" revision="mbp@foo-00" text_sha1="B" text_size="0" />
121
132
</inventory>
122
133
"""
123
134
 
124
135
 
125
136
_expected_inv_v5_root = """<inventory file_id="f&lt;" format="5" revision_id="mother!">
126
 
<file file_id="bar-20050901064931-73b4b1138abc9cd2" name="bar" parent_id="f&lt;" revision="mbp@foo-123123" />
 
137
<file file_id="bar-20050901064931-73b4b1138abc9cd2" name="bar" parent_id="f&lt;" revision="mbp@foo-123123" text_sha1="A" text_size="1" />
127
138
<directory file_id="foo-20050801201819-4139aa4a272f4250" name="subdir" parent_id="f&lt;" revision="mbp@foo-00" />
128
 
<file executable="yes" file_id="bar-20050824000535-6bc48cfad47ed134" name="bar" parent_id="foo-20050801201819-4139aa4a272f4250" revision="mbp@foo-00" />
129
 
</inventory>
130
 
"""
131
 
 
 
139
<file executable="yes" file_id="bar-20050824000535-6bc48cfad47ed134" name="bar" parent_id="foo-20050801201819-4139aa4a272f4250" revision="mbp@foo-00" text_sha1="B" text_size="0" />
 
140
<symlink file_id="link-1" name="link" parent_id="foo-20050801201819-4139aa4a272f4250" revision="mbp@foo-00" symlink_target="a" />
 
141
</inventory>
 
142
"""
 
143
 
 
144
_expected_inv_v6 = """<inventory format="6" revision_id="rev_outer">
 
145
<directory file_id="tree-root-321" name="" revision="rev_outer" />
 
146
<directory file_id="dir-id" name="dir" parent_id="tree-root-321" revision="rev_outer" />
 
147
<file file_id="file-id" name="file" parent_id="tree-root-321" revision="rev_outer" text_sha1="A" text_size="1" />
 
148
<symlink file_id="link-id" name="link" parent_id="tree-root-321" revision="rev_outer" symlink_target="a" />
 
149
</inventory>
 
150
"""
 
151
 
 
152
_expected_inv_v7 = """<inventory format="7" revision_id="rev_outer">
 
153
<directory file_id="tree-root-321" name="" revision="rev_outer" />
 
154
<directory file_id="dir-id" name="dir" parent_id="tree-root-321" revision="rev_outer" />
 
155
<file file_id="file-id" name="file" parent_id="tree-root-321" revision="rev_outer" text_sha1="A" text_size="1" />
 
156
<symlink file_id="link-id" name="link" parent_id="tree-root-321" revision="rev_outer" symlink_target="a" />
 
157
<tree-reference file_id="nested-id" name="nested" parent_id="tree-root-321" revision="rev_outer" reference_revision="rev_inner" />
 
158
</inventory>
 
159
"""
 
160
 
 
161
_expected_rev_v8 = """<revision committer="Martin Pool &lt;mbp@sourcefrog.net&gt;" format="8" inventory_sha1="e79c31c1deb64c163cf660fdedd476dd579ffd41" revision_id="mbp@sourcefrog.net-20050905080035-e0439293f8b6b9f9" timestamp="1125907235.212" timezone="36000">
 
162
<message>- start splitting code for xml (de)serialization away from objects
 
163
  preparatory to supporting multiple formats by a single library
 
164
</message>
 
165
<parents>
 
166
<revision_ref revision_id="mbp@sourcefrog.net-20050905063503-43948f59fa127d92" />
 
167
</parents>
 
168
</revision>
 
169
"""
 
170
 
 
171
_expected_inv_v8 = """<inventory format="8" revision_id="rev_outer">
 
172
<directory file_id="tree-root-321" name="" revision="rev_outer" />
 
173
<directory file_id="dir-id" name="dir" parent_id="tree-root-321" revision="rev_outer" />
 
174
<file file_id="file-id" name="file" parent_id="tree-root-321" revision="rev_outer" text_sha1="A" text_size="1" />
 
175
<symlink file_id="link-id" name="link" parent_id="tree-root-321" revision="rev_outer" symlink_target="a" />
 
176
</inventory>
 
177
"""
132
178
 
133
179
_revision_utf8_v5 = """<revision committer="Erik B&#229;gfors &lt;erik@foo.net&gt;"
134
180
    inventory_sha1="e79c31c1deb64c163cf660fdedd476dd579ffd41"
158
204
</inventory>
159
205
"""
160
206
 
 
207
# Before revision_id was always stored as an attribute
 
208
_inventory_v5a = """<inventory format="5">
 
209
</inventory>
 
210
"""
 
211
 
 
212
# Before revision_id was always stored as an attribute
 
213
_inventory_v5b = """<inventory format="5" revision_id="a-rev-id">
 
214
</inventory>
 
215
"""
 
216
 
161
217
 
162
218
class TestSerializer(TestCase):
163
219
    """Test XML serialization"""
 
220
 
164
221
    def test_canned_inventory(self):
165
222
        """Test unpacked a canned inventory v4 file."""
166
223
        inp = StringIO(_working_inventory_v4)
229
286
        eq(ie.name, 'bar')
230
287
        eq(inv[ie.parent_id].kind, 'directory')
231
288
 
 
289
    def test_unpack_inventory_5a(self):
 
290
        inv = bzrlib.xml5.serializer_v5.read_inventory_from_string(
 
291
                _inventory_v5a, revision_id='test-rev-id')
 
292
        self.assertEqual('test-rev-id', inv.root.revision)
 
293
 
 
294
    def test_unpack_inventory_5a_cache_and_copy(self):
 
295
        # Passing an entry_cache should get populated with the objects
 
296
        # But the returned objects should be copies if return_from_cache is
 
297
        # False
 
298
        entry_cache = fifo_cache.FIFOCache()
 
299
        inv = bzrlib.xml5.serializer_v5.read_inventory_from_string(
 
300
            _inventory_v5a, revision_id='test-rev-id',
 
301
            entry_cache=entry_cache, return_from_cache=False)
 
302
        for entry in inv.iter_just_entries():
 
303
            key = (entry.file_id, entry.revision)
 
304
            if entry.file_id is inv.root.file_id:
 
305
                # The root id is inferred for xml v5
 
306
                self.assertFalse(key in entry_cache)
 
307
            else:
 
308
                self.assertIsNot(entry, entry_cache[key])
 
309
 
 
310
    def test_unpack_inventory_5a_cache_no_copy(self):
 
311
        # Passing an entry_cache should get populated with the objects
 
312
        # The returned objects should be exact if return_from_cache is
 
313
        # True
 
314
        entry_cache = fifo_cache.FIFOCache()
 
315
        inv = bzrlib.xml5.serializer_v5.read_inventory_from_string(
 
316
            _inventory_v5a, revision_id='test-rev-id',
 
317
            entry_cache=entry_cache, return_from_cache=True)
 
318
        for entry in inv.iter_just_entries():
 
319
            key = (entry.file_id, entry.revision)
 
320
            if entry.file_id is inv.root.file_id:
 
321
                # The root id is inferred for xml v5
 
322
                self.assertFalse(key in entry_cache)
 
323
            else:
 
324
                self.assertIs(entry, entry_cache[key])
 
325
 
 
326
    def test_unpack_inventory_5b(self):
 
327
        inv = bzrlib.xml5.serializer_v5.read_inventory_from_string(
 
328
                _inventory_v5b, revision_id='test-rev-id')
 
329
        self.assertEqual('a-rev-id', inv.root.revision)
 
330
 
232
331
    def test_repack_inventory_5(self):
233
332
        inp = StringIO(_committed_inv_v5)
234
333
        inv = bzrlib.xml5.serializer_v5.read_inventory(inp)
237
336
        self.assertEqualDiff(_expected_inv_v5, outp.getvalue())
238
337
        inv2 = bzrlib.xml5.serializer_v5.read_inventory(StringIO(outp.getvalue()))
239
338
        self.assertEqual(inv, inv2)
240
 
    
 
339
 
241
340
    def assertRoundTrips(self, xml_string):
242
341
        inp = StringIO(xml_string)
243
342
        inv = bzrlib.xml5.serializer_v5.read_inventory(inp)
244
343
        outp = StringIO()
245
344
        bzrlib.xml5.serializer_v5.write_inventory(inv, outp)
246
345
        self.assertEqualDiff(xml_string, outp.getvalue())
 
346
        lines = bzrlib.xml5.serializer_v5.write_inventory_to_lines(inv)
 
347
        outp.seek(0)
 
348
        self.assertEqual(outp.readlines(), lines)
247
349
        inv2 = bzrlib.xml5.serializer_v5.read_inventory(StringIO(outp.getvalue()))
248
350
        self.assertEqual(inv, inv2)
249
351
 
289
391
        new_rev = s_v5.read_revision_from_string(txt)
290
392
        self.assertEqual(props, new_rev.properties)
291
393
 
 
394
    def get_sample_inventory(self):
 
395
        inv = Inventory('tree-root-321', revision_id='rev_outer')
 
396
        inv.add(inventory.InventoryFile('file-id', 'file', 'tree-root-321'))
 
397
        inv.add(inventory.InventoryDirectory('dir-id', 'dir',
 
398
                                             'tree-root-321'))
 
399
        inv.add(inventory.InventoryLink('link-id', 'link', 'tree-root-321'))
 
400
        inv['tree-root-321'].revision = 'rev_outer'
 
401
        inv['dir-id'].revision = 'rev_outer'
 
402
        inv['file-id'].revision = 'rev_outer'
 
403
        inv['file-id'].text_sha1 = 'A'
 
404
        inv['file-id'].text_size = 1
 
405
        inv['link-id'].revision = 'rev_outer'
 
406
        inv['link-id'].symlink_target = 'a'
 
407
        return inv
 
408
 
 
409
    def test_roundtrip_inventory_v7(self):
 
410
        inv = self.get_sample_inventory()
 
411
        inv.add(inventory.TreeReference('nested-id', 'nested', 'tree-root-321',
 
412
                                        'rev_outer', 'rev_inner'))
 
413
        txt = xml7.serializer_v7.write_inventory_to_string(inv)
 
414
        lines = xml7.serializer_v7.write_inventory_to_lines(inv)
 
415
        self.assertEqual(bzrlib.osutils.split_lines(txt), lines)
 
416
        self.assertEqualDiff(_expected_inv_v7, txt)
 
417
        inv2 = xml7.serializer_v7.read_inventory_from_string(txt)
 
418
        self.assertEqual(5, len(inv2))
 
419
        for path, ie in inv.iter_entries():
 
420
            self.assertEqual(ie, inv2[ie.file_id])
 
421
 
 
422
    def test_roundtrip_inventory_v6(self):
 
423
        inv = self.get_sample_inventory()
 
424
        txt = xml6.serializer_v6.write_inventory_to_string(inv)
 
425
        lines = xml6.serializer_v6.write_inventory_to_lines(inv)
 
426
        self.assertEqual(bzrlib.osutils.split_lines(txt), lines)
 
427
        self.assertEqualDiff(_expected_inv_v6, txt)
 
428
        inv2 = xml6.serializer_v6.read_inventory_from_string(txt)
 
429
        self.assertEqual(4, len(inv2))
 
430
        for path, ie in inv.iter_entries():
 
431
            self.assertEqual(ie, inv2[ie.file_id])
 
432
 
 
433
    def test_wrong_format_v7(self):
 
434
        """Can't accidentally open a file with wrong serializer"""
 
435
        s_v6 = bzrlib.xml6.serializer_v6
 
436
        s_v7 = xml7.serializer_v7
 
437
        self.assertRaises(errors.UnexpectedInventoryFormat,
 
438
                          s_v7.read_inventory_from_string, _expected_inv_v5)
 
439
        self.assertRaises(errors.UnexpectedInventoryFormat,
 
440
                          s_v6.read_inventory_from_string, _expected_inv_v7)
 
441
 
 
442
    def test_tree_reference(self):
 
443
        s_v5 = bzrlib.xml5.serializer_v5
 
444
        s_v6 = bzrlib.xml6.serializer_v6
 
445
        s_v7 = xml7.serializer_v7
 
446
        inv = Inventory('tree-root-321', revision_id='rev-outer')
 
447
        inv.root.revision = 'root-rev'
 
448
        inv.add(inventory.TreeReference('nested-id', 'nested', 'tree-root-321',
 
449
                                        'rev-outer', 'rev-inner'))
 
450
        self.assertRaises(errors.UnsupportedInventoryKind,
 
451
                          s_v5.write_inventory_to_string, inv)
 
452
        self.assertRaises(errors.UnsupportedInventoryKind,
 
453
                          s_v6.write_inventory_to_string, inv)
 
454
        txt = s_v7.write_inventory_to_string(inv)
 
455
        lines = s_v7.write_inventory_to_lines(inv)
 
456
        self.assertEqual(bzrlib.osutils.split_lines(txt), lines)
 
457
        inv2 = s_v7.read_inventory_from_string(txt)
 
458
        self.assertEqual('tree-root-321', inv2['nested-id'].parent_id)
 
459
        self.assertEqual('rev-outer', inv2['nested-id'].revision)
 
460
        self.assertEqual('rev-inner', inv2['nested-id'].reference_revision)
 
461
        self.assertRaises(errors.UnsupportedInventoryKind,
 
462
                          s_v6.read_inventory_from_string,
 
463
                          txt.replace('format="7"', 'format="6"'))
 
464
        self.assertRaises(errors.UnsupportedInventoryKind,
 
465
                          s_v5.read_inventory_from_string,
 
466
                          txt.replace('format="7"', 'format="5"'))
 
467
 
 
468
    def test_roundtrip_inventory_v8(self):
 
469
        inv = self.get_sample_inventory()
 
470
        txt = xml8.serializer_v8.write_inventory_to_string(inv)
 
471
        inv2 = xml8.serializer_v8.read_inventory_from_string(txt)
 
472
        self.assertEqual(4, len(inv2))
 
473
        for path, ie in inv.iter_entries():
 
474
            self.assertEqual(ie, inv2[ie.file_id])
 
475
 
 
476
    def test_inventory_text_v8(self):
 
477
        inv = self.get_sample_inventory()
 
478
        txt = xml8.serializer_v8.write_inventory_to_string(inv)
 
479
        lines = xml8.serializer_v8.write_inventory_to_lines(inv)
 
480
        self.assertEqual(bzrlib.osutils.split_lines(txt), lines)
 
481
        self.assertEqualDiff(_expected_inv_v8, txt)
 
482
 
 
483
    def test_revision_text_v6(self):
 
484
        """Pack revision to XML v6"""
 
485
        rev = bzrlib.xml6.serializer_v6.read_revision_from_string(
 
486
            _expected_rev_v5)
 
487
        serialized = bzrlib.xml6.serializer_v6.write_revision_to_string(rev)
 
488
        self.assertEqualDiff(serialized, _expected_rev_v5)
 
489
 
 
490
    def test_revision_text_v7(self):
 
491
        """Pack revision to XML v7"""
 
492
        rev = bzrlib.xml7.serializer_v7.read_revision_from_string(
 
493
            _expected_rev_v5)
 
494
        serialized = bzrlib.xml7.serializer_v7.write_revision_to_string(rev)
 
495
        self.assertEqualDiff(serialized, _expected_rev_v5)
 
496
 
 
497
    def test_revision_text_v8(self):
 
498
        """Pack revision to XML v8"""
 
499
        rev = bzrlib.xml8.serializer_v8.read_revision_from_string(
 
500
            _expected_rev_v8)
 
501
        serialized = bzrlib.xml8.serializer_v8.write_revision_to_string(rev)
 
502
        self.assertEqualDiff(serialized, _expected_rev_v8)
 
503
 
292
504
    def test_revision_ids_are_utf8(self):
293
505
        """Parsed revision_ids should all be utf-8 strings, not unicode."""
294
506
        s_v5 = bzrlib.xml5.serializer_v5
309
521
        fid_bar1 = u'b\xe5r-01'.encode('utf8')
310
522
        fid_sub = u's\xb5bdir-01'.encode('utf8')
311
523
        fid_bar2 = u'b\xe5r-02'.encode('utf8')
312
 
        expected = [(u'', fid_root, None, None),
 
524
        expected = [(u'', fid_root, None, rev_id_2),
313
525
                    (u'b\xe5r', fid_bar1, fid_root, rev_id_1),
314
526
                    (u's\xb5bdir', fid_sub, fid_root, rev_id_1),
315
527
                    (u's\xb5bdir/b\xe5r', fid_bar2, fid_sub, rev_id_2),
338
550
    """Whitebox testing of the _encode_and_escape function."""
339
551
 
340
552
    def setUp(self):
 
553
        TestCase.setUp(self)
341
554
        # Keep the cache clear before and after the test
342
 
        bzrlib.xml5._ensure_utf8_re()
343
 
        bzrlib.xml5._clear_cache()
344
 
        self.addCleanup(bzrlib.xml5._clear_cache)
 
555
        bzrlib.xml8._ensure_utf8_re()
 
556
        bzrlib.xml8._clear_cache()
 
557
        self.addCleanup(bzrlib.xml8._clear_cache)
345
558
 
346
559
    def test_simple_ascii(self):
347
560
        # _encode_and_escape always appends a final ", because these parameters
348
561
        # are being used in xml attributes, and by returning it now, we have to
349
562
        # do fewer string operations later.
350
 
        val = bzrlib.xml5._encode_and_escape('foo bar')
 
563
        val = bzrlib.xml8._encode_and_escape('foo bar')
351
564
        self.assertEqual('foo bar"', val)
352
565
        # The second time should be cached
353
 
        val2 = bzrlib.xml5._encode_and_escape('foo bar')
 
566
        val2 = bzrlib.xml8._encode_and_escape('foo bar')
354
567
        self.assertIs(val2, val)
355
568
 
356
569
    def test_ascii_with_xml(self):
357
570
        self.assertEqual('&amp;&apos;&quot;&lt;&gt;"',
358
 
                         bzrlib.xml5._encode_and_escape('&\'"<>'))
 
571
                         bzrlib.xml8._encode_and_escape('&\'"<>'))
359
572
 
360
573
    def test_utf8_with_xml(self):
361
574
        # u'\xb5\xe5&\u062c'
362
575
        utf8_str = '\xc2\xb5\xc3\xa5&\xd8\xac'
363
576
        self.assertEqual('&#181;&#229;&amp;&#1580;"',
364
 
                         bzrlib.xml5._encode_and_escape(utf8_str))
 
577
                         bzrlib.xml8._encode_and_escape(utf8_str))
365
578
 
366
579
    def test_unicode(self):
367
580
        uni_str = u'\xb5\xe5&\u062c'
368
581
        self.assertEqual('&#181;&#229;&amp;&#1580;"',
369
 
                         bzrlib.xml5._encode_and_escape(uni_str))
 
582
                         bzrlib.xml8._encode_and_escape(uni_str))