69
_revision_v5_utc = """\
70
<revision committer="Martin Pool <mbp@sourcefrog.net>"
71
inventory_sha1="e79c31c1deb64c163cf660fdedd476dd579ffd41"
72
revision_id="mbp@sourcefrog.net-20050905080035-e0439293f8b6b9f9"
73
timestamp="1125907235.212"
75
<message>- start splitting code for xml (de)serialization away from objects
76
preparatory to supporting multiple formats by a single library
79
<revision_ref revision_id="mbp@sourcefrog.net-20050905063503-43948f59fa127d92"/>
84
60
_committed_inv_v5 = """<inventory>
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"/>
61
<file file_id="bar-20050901064931-73b4b1138abc9cd2"
62
name="bar" parent_id="TREE_ROOT"
63
revision="mbp@foo-123123"/>
89
64
<directory name="subdir"
90
65
file_id="foo-20050801201819-4139aa4a272f4250"
92
67
revision="mbp@foo-00"/>
93
<file executable="yes" file_id="bar-20050824000535-6bc48cfad47ed134"
94
name="bar" parent_id="foo-20050801201819-4139aa4a272f4250"
96
text_sha1="B" text_size="0"/>
68
<file file_id="bar-20050824000535-6bc48cfad47ed134"
69
name="bar" parent_id="foo-20050801201819-4139aa4a272f4250"
70
revision="mbp@foo-00"/>
100
74
_basis_inv_v5 = """<inventory revision_id="mbp@sourcefrog.net-20050905063503-43948f59fa127d92">
101
<file file_id="bar-20050901064931-73b4b1138abc9cd2"
102
name="bar" parent_id="TREE_ROOT"
75
<file file_id="bar-20050901064931-73b4b1138abc9cd2"
76
name="bar" parent_id="TREE_ROOT"
103
77
revision="mbp@foo-123123"/>
104
78
<directory name="subdir"
105
79
file_id="foo-20050801201819-4139aa4a272f4250"
106
parent_id="TREE_ROOT"
107
81
revision="mbp@foo-00"/>
108
<file file_id="bar-20050824000535-6bc48cfad47ed134"
109
name="bar" parent_id="foo-20050801201819-4139aa4a272f4250"
82
<file file_id="bar-20050824000535-6bc48cfad47ed134"
83
name="bar" parent_id="foo-20050801201819-4139aa4a272f4250"
110
84
revision="mbp@foo-00"/>
115
# DO NOT REFLOW THIS. Its the exact revision we want.
116
_expected_rev_v5 = """<revision committer="Martin Pool <mbp@sourcefrog.net>" format="5" inventory_sha1="e79c31c1deb64c163cf660fdedd476dd579ffd41" revision_id="mbp@sourcefrog.net-20050905080035-e0439293f8b6b9f9" timestamp="1125907235.212" timezone="36000">
117
<message>- start splitting code for xml (de)serialization away from objects
118
preparatory to supporting multiple formats by a single library
121
<revision_ref revision_id="mbp@sourcefrog.net-20050905063503-43948f59fa127d92" />
127
# DO NOT REFLOW THIS. Its the exact inventory we want.
128
_expected_inv_v5 = """<inventory format="5">
129
<file file_id="bar-20050901064931-73b4b1138abc9cd2" name="bar" revision="mbp@foo-123123" text_sha1="A" text_size="1" />
130
<directory file_id="foo-20050801201819-4139aa4a272f4250" name="subdir" 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" />
136
_expected_inv_v5_root = """<inventory file_id="f<" format="5" revision_id="mother!">
137
<file file_id="bar-20050901064931-73b4b1138abc9cd2" name="bar" parent_id="f<" revision="mbp@foo-123123" text_sha1="A" text_size="1" />
138
<directory file_id="foo-20050801201819-4139aa4a272f4250" name="subdir" parent_id="f<" revision="mbp@foo-00" />
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" />
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" />
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" />
161
_expected_rev_v8 = """<revision committer="Martin Pool <mbp@sourcefrog.net>" 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
166
<revision_ref revision_id="mbp@sourcefrog.net-20050905063503-43948f59fa127d92" />
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" />
179
_revision_utf8_v5 = """<revision committer="Erik Bågfors <erik@foo.net>"
180
inventory_sha1="e79c31c1deb64c163cf660fdedd476dd579ffd41"
181
revision_id="erik@bågfors-02"
182
timestamp="1125907235.212"
184
<message>Include µnicode characters
187
<revision_ref revision_id="erik@bågfors-01"/>
192
_inventory_utf8_v5 = """<inventory file_id="TREé_ROOT" format="5"
193
revision_id="erik@bågfors-02">
194
<file file_id="bår-01"
195
name="bår" parent_id="TREé_ROOT"
196
revision="erik@bågfors-01"/>
197
<directory name="sµbdir"
198
file_id="sµbdir-01"
199
parent_id="TREé_ROOT"
200
revision="erik@bågfors-01"/>
201
<file executable="yes" file_id="bår-02"
202
name="bår" parent_id="sµbdir-01"
203
revision="erik@bågfors-02"/>
207
# Before revision_id was always stored as an attribute
208
_inventory_v5a = """<inventory format="5">
212
# Before revision_id was always stored as an attribute
213
_inventory_v5b = """<inventory format="5" revision_id="a-rev-id">
218
88
class TestSerializer(TestCase):
219
89
"""Test XML serialization"""
221
90
def test_canned_inventory(self):
222
91
"""Test unpacked a canned inventory v4 file."""
223
92
inp = StringIO(_working_inventory_v4)
286
144
eq(ie.name, 'bar')
287
145
eq(inv[ie.parent_id].kind, 'directory')
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)
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
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)
308
self.assertIsNot(entry, entry_cache[key])
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
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)
324
self.assertIs(entry, entry_cache[key])
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)
331
147
def test_repack_inventory_5(self):
332
148
inp = StringIO(_committed_inv_v5)
333
149
inv = bzrlib.xml5.serializer_v5.read_inventory(inp)
334
150
outp = StringIO()
335
151
bzrlib.xml5.serializer_v5.write_inventory(inv, outp)
336
self.assertEqualDiff(_expected_inv_v5, outp.getvalue())
337
inv2 = bzrlib.xml5.serializer_v5.read_inventory(StringIO(outp.getvalue()))
338
self.assertEqual(inv, inv2)
340
def assertRoundTrips(self, xml_string):
341
inp = StringIO(xml_string)
342
inv = bzrlib.xml5.serializer_v5.read_inventory(inp)
344
bzrlib.xml5.serializer_v5.write_inventory(inv, outp)
345
self.assertEqualDiff(xml_string, outp.getvalue())
346
lines = bzrlib.xml5.serializer_v5.write_inventory_to_lines(inv)
348
self.assertEqual(outp.readlines(), lines)
349
inv2 = bzrlib.xml5.serializer_v5.read_inventory(StringIO(outp.getvalue()))
350
self.assertEqual(inv, inv2)
352
def tests_serialize_inventory_v5_with_root(self):
353
self.assertRoundTrips(_expected_inv_v5_root)
355
def check_repack_revision(self, txt):
356
"""Check that repacking a revision yields the same information"""
152
inv2 = bzrlib.xml5.serializer_v5.read_inventory(StringIO(outp.getvalue()))
153
self.assertEqual(inv, inv2)
155
def test_repack_revision_5(self):
156
"""Round-trip revision to XML v5"""
157
inp = StringIO(_revision_v5)
358
158
rev = bzrlib.xml5.serializer_v5.read_revision(inp)
359
159
outp = StringIO()
360
160
bzrlib.xml5.serializer_v5.write_revision(rev, outp)
378
171
outfile_contents = outp.getvalue()
379
172
self.assertEqual(outfile_contents[-1], '\n')
380
173
self.assertEqualDiff(outfile_contents, bzrlib.xml5.serializer_v5.write_revision_to_string(rev))
381
self.assertEqualDiff(outfile_contents, _expected_rev_v5)
383
def test_empty_property_value(self):
384
"""Create an empty property value check that it serializes correctly"""
385
s_v5 = bzrlib.xml5.serializer_v5
386
rev = s_v5.read_revision_from_string(_revision_v5)
388
props = {'empty':'', 'one':'one'}
389
rev.properties = props
390
txt = s_v5.write_revision_to_string(rev)
391
new_rev = s_v5.read_revision_from_string(txt)
392
self.assertEqual(props, new_rev.properties)
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',
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'
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])
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])
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)
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"'))
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])
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)
483
def test_revision_text_v6(self):
484
"""Pack revision to XML v6"""
485
rev = bzrlib.xml6.serializer_v6.read_revision_from_string(
487
serialized = bzrlib.xml6.serializer_v6.write_revision_to_string(rev)
488
self.assertEqualDiff(serialized, _expected_rev_v5)
490
def test_revision_text_v7(self):
491
"""Pack revision to XML v7"""
492
rev = bzrlib.xml7.serializer_v7.read_revision_from_string(
494
serialized = bzrlib.xml7.serializer_v7.write_revision_to_string(rev)
495
self.assertEqualDiff(serialized, _expected_rev_v5)
497
def test_revision_text_v8(self):
498
"""Pack revision to XML v8"""
499
rev = bzrlib.xml8.serializer_v8.read_revision_from_string(
501
serialized = bzrlib.xml8.serializer_v8.write_revision_to_string(rev)
502
self.assertEqualDiff(serialized, _expected_rev_v8)
504
def test_revision_ids_are_utf8(self):
505
"""Parsed revision_ids should all be utf-8 strings, not unicode."""
506
s_v5 = bzrlib.xml5.serializer_v5
507
rev = s_v5.read_revision_from_string(_revision_utf8_v5)
508
self.assertEqual('erik@b\xc3\xa5gfors-02', rev.revision_id)
509
self.assertIsInstance(rev.revision_id, str)
510
self.assertEqual(['erik@b\xc3\xa5gfors-01'], rev.parent_ids)
511
for parent_id in rev.parent_ids:
512
self.assertIsInstance(parent_id, str)
513
self.assertEqual(u'Include \xb5nicode characters\n', rev.message)
514
self.assertIsInstance(rev.message, unicode)
516
# ie.revision should either be None or a utf-8 revision id
517
inv = s_v5.read_inventory_from_string(_inventory_utf8_v5)
518
rev_id_1 = u'erik@b\xe5gfors-01'.encode('utf8')
519
rev_id_2 = u'erik@b\xe5gfors-02'.encode('utf8')
520
fid_root = u'TRE\xe9_ROOT'.encode('utf8')
521
fid_bar1 = u'b\xe5r-01'.encode('utf8')
522
fid_sub = u's\xb5bdir-01'.encode('utf8')
523
fid_bar2 = u'b\xe5r-02'.encode('utf8')
524
expected = [(u'', fid_root, None, rev_id_2),
525
(u'b\xe5r', fid_bar1, fid_root, rev_id_1),
526
(u's\xb5bdir', fid_sub, fid_root, rev_id_1),
527
(u's\xb5bdir/b\xe5r', fid_bar2, fid_sub, rev_id_2),
529
self.assertEqual(rev_id_2, inv.revision_id)
530
self.assertIsInstance(inv.revision_id, str)
532
actual = list(inv.iter_entries_by_dir())
533
for ((exp_path, exp_file_id, exp_parent_id, exp_rev_id),
534
(act_path, act_ie)) in zip(expected, actual):
535
self.assertEqual(exp_path, act_path)
536
self.assertIsInstance(act_path, unicode)
537
self.assertEqual(exp_file_id, act_ie.file_id)
538
self.assertIsInstance(act_ie.file_id, str)
539
self.assertEqual(exp_parent_id, act_ie.parent_id)
540
if exp_parent_id is not None:
541
self.assertIsInstance(act_ie.parent_id, str)
542
self.assertEqual(exp_rev_id, act_ie.revision)
543
if exp_rev_id is not None:
544
self.assertIsInstance(act_ie.revision, str)
546
self.assertEqual(len(expected), len(actual))
549
class TestEncodeAndEscape(TestCase):
550
"""Whitebox testing of the _encode_and_escape function."""
554
# Keep the cache clear before and after the test
555
bzrlib.xml8._ensure_utf8_re()
556
bzrlib.xml8._clear_cache()
557
self.addCleanup(bzrlib.xml8._clear_cache)
559
def test_simple_ascii(self):
560
# _encode_and_escape always appends a final ", because these parameters
561
# are being used in xml attributes, and by returning it now, we have to
562
# do fewer string operations later.
563
val = bzrlib.xml8._encode_and_escape('foo bar')
564
self.assertEqual('foo bar"', val)
565
# The second time should be cached
566
val2 = bzrlib.xml8._encode_and_escape('foo bar')
567
self.assertIs(val2, val)
569
def test_ascii_with_xml(self):
570
self.assertEqual('&'"<>"',
571
bzrlib.xml8._encode_and_escape('&\'"<>'))
573
def test_utf8_with_xml(self):
575
utf8_str = '\xc2\xb5\xc3\xa5&\xd8\xac'
576
self.assertEqual('µå&ج"',
577
bzrlib.xml8._encode_and_escape(utf8_str))
579
def test_unicode(self):
580
uni_str = u'\xb5\xe5&\u062c'
581
self.assertEqual('µå&ج"',
582
bzrlib.xml8._encode_and_escape(uni_str))
added
removed
bzrlib/tests/test_xml.py
