68
_revision_v5_utc = """\
69
<revision committer="Martin Pool <mbp@sourcefrog.net>"
70
inventory_sha1="e79c31c1deb64c163cf660fdedd476dd579ffd41"
71
revision_id="mbp@sourcefrog.net-20050905080035-e0439293f8b6b9f9"
72
timestamp="1125907235.212"
74
<message>- start splitting code for xml (de)serialization away from objects
75
preparatory to supporting multiple formats by a single library
78
<revision_ref revision_id="mbp@sourcefrog.net-20050905063503-43948f59fa127d92"/>
60
83
_committed_inv_v5 = """<inventory>
61
<file file_id="bar-20050901064931-73b4b1138abc9cd2"
62
name="bar" parent_id="TREE_ROOT"
63
revision="mbp@foo-123123"/>
84
<file file_id="bar-20050901064931-73b4b1138abc9cd2"
85
name="bar" parent_id="TREE_ROOT"
86
revision="mbp@foo-123123"
87
text_sha1="A" text_size="1"/>
64
88
<directory name="subdir"
65
89
file_id="foo-20050801201819-4139aa4a272f4250"
67
91
revision="mbp@foo-00"/>
68
<file file_id="bar-20050824000535-6bc48cfad47ed134"
69
name="bar" parent_id="foo-20050801201819-4139aa4a272f4250"
70
revision="mbp@foo-00"/>
92
<file executable="yes" file_id="bar-20050824000535-6bc48cfad47ed134"
93
name="bar" parent_id="foo-20050801201819-4139aa4a272f4250"
95
text_sha1="B" text_size="0"/>
74
99
_basis_inv_v5 = """<inventory revision_id="mbp@sourcefrog.net-20050905063503-43948f59fa127d92">
75
<file file_id="bar-20050901064931-73b4b1138abc9cd2"
76
name="bar" parent_id="TREE_ROOT"
100
<file file_id="bar-20050901064931-73b4b1138abc9cd2"
101
name="bar" parent_id="TREE_ROOT"
77
102
revision="mbp@foo-123123"/>
78
103
<directory name="subdir"
79
104
file_id="foo-20050801201819-4139aa4a272f4250"
105
parent_id="TREE_ROOT"
81
106
revision="mbp@foo-00"/>
82
<file file_id="bar-20050824000535-6bc48cfad47ed134"
83
name="bar" parent_id="foo-20050801201819-4139aa4a272f4250"
107
<file file_id="bar-20050824000535-6bc48cfad47ed134"
108
name="bar" parent_id="foo-20050801201819-4139aa4a272f4250"
84
109
revision="mbp@foo-00"/>
114
# DO NOT REFLOW THIS. Its the exact revision we want.
115
_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">
116
<message>- start splitting code for xml (de)serialization away from objects
117
preparatory to supporting multiple formats by a single library
120
<revision_ref revision_id="mbp@sourcefrog.net-20050905063503-43948f59fa127d92" />
126
# DO NOT REFLOW THIS. Its the exact inventory we want.
127
_expected_inv_v5 = """<inventory format="5">
128
<file file_id="bar-20050901064931-73b4b1138abc9cd2" name="bar" revision="mbp@foo-123123" text_sha1="A" text_size="1" />
129
<directory file_id="foo-20050801201819-4139aa4a272f4250" name="subdir" revision="mbp@foo-00" />
130
<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" />
135
_expected_inv_v5_root = """<inventory file_id="f<" format="5" revision_id="mother!">
136
<file file_id="bar-20050901064931-73b4b1138abc9cd2" name="bar" parent_id="f<" revision="mbp@foo-123123" text_sha1="A" text_size="1" />
137
<directory file_id="foo-20050801201819-4139aa4a272f4250" name="subdir" parent_id="f<" revision="mbp@foo-00" />
138
<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" />
139
<symlink file_id="link-1" name="link" parent_id="foo-20050801201819-4139aa4a272f4250" revision="mbp@foo-00" symlink_target="a" />
143
_expected_inv_v6 = """<inventory format="6" revision_id="rev_outer">
144
<directory file_id="tree-root-321" name="" revision="rev_outer" />
145
<directory file_id="dir-id" name="dir" parent_id="tree-root-321" revision="rev_outer" />
146
<file file_id="file-id" name="file" parent_id="tree-root-321" revision="rev_outer" text_sha1="A" text_size="1" />
147
<symlink file_id="link-id" name="link" parent_id="tree-root-321" revision="rev_outer" symlink_target="a" />
151
_expected_inv_v7 = """<inventory format="7" revision_id="rev_outer">
152
<directory file_id="tree-root-321" name="" revision="rev_outer" />
153
<directory file_id="dir-id" name="dir" parent_id="tree-root-321" revision="rev_outer" />
154
<file file_id="file-id" name="file" parent_id="tree-root-321" revision="rev_outer" text_sha1="A" text_size="1" />
155
<symlink file_id="link-id" name="link" parent_id="tree-root-321" revision="rev_outer" symlink_target="a" />
156
<tree-reference file_id="nested-id" name="nested" parent_id="tree-root-321" revision="rev_outer" reference_revision="rev_inner" />
160
_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">
161
<message>- start splitting code for xml (de)serialization away from objects
162
preparatory to supporting multiple formats by a single library
165
<revision_ref revision_id="mbp@sourcefrog.net-20050905063503-43948f59fa127d92" />
170
_expected_inv_v8 = """<inventory format="8" revision_id="rev_outer">
171
<directory file_id="tree-root-321" name="" revision="rev_outer" />
172
<directory file_id="dir-id" name="dir" parent_id="tree-root-321" revision="rev_outer" />
173
<file file_id="file-id" name="file" parent_id="tree-root-321" revision="rev_outer" text_sha1="A" text_size="1" />
174
<symlink file_id="link-id" name="link" parent_id="tree-root-321" revision="rev_outer" symlink_target="a" />
178
_revision_utf8_v5 = """<revision committer="Erik Bågfors <erik@foo.net>"
179
inventory_sha1="e79c31c1deb64c163cf660fdedd476dd579ffd41"
180
revision_id="erik@bågfors-02"
181
timestamp="1125907235.212"
183
<message>Include µnicode characters
186
<revision_ref revision_id="erik@bågfors-01"/>
191
_inventory_utf8_v5 = """<inventory file_id="TREé_ROOT" format="5"
192
revision_id="erik@bågfors-02">
193
<file file_id="bår-01"
194
name="bår" parent_id="TREé_ROOT"
195
revision="erik@bågfors-01"/>
196
<directory name="sµbdir"
197
file_id="sµbdir-01"
198
parent_id="TREé_ROOT"
199
revision="erik@bågfors-01"/>
200
<file executable="yes" file_id="bår-02"
201
name="bår" parent_id="sµbdir-01"
202
revision="erik@bågfors-02"/>
206
# Before revision_id was always stored as an attribute
207
_inventory_v5a = """<inventory format="5">
211
# Before revision_id was always stored as an attribute
212
_inventory_v5b = """<inventory format="5" revision_id="a-rev-id">
88
217
class TestSerializer(TestCase):
89
218
"""Test XML serialization"""
90
220
def test_canned_inventory(self):
91
221
"""Test unpacked a canned inventory v4 file."""
92
222
inp = StringIO(_working_inventory_v4)
144
285
eq(ie.name, 'bar')
145
286
eq(inv[ie.parent_id].kind, 'directory')
288
def test_unpack_inventory_5a(self):
289
inv = bzrlib.xml5.serializer_v5.read_inventory_from_string(
290
_inventory_v5a, revision_id='test-rev-id')
291
self.assertEqual('test-rev-id', inv.root.revision)
293
def test_unpack_inventory_5a_cache_and_copy(self):
294
# Passing an entry_cache should get populated with the objects
295
# But the returned objects should be copies if return_from_cache is
297
entry_cache = fifo_cache.FIFOCache()
298
inv = bzrlib.xml5.serializer_v5.read_inventory_from_string(
299
_inventory_v5a, revision_id='test-rev-id',
300
entry_cache=entry_cache, return_from_cache=False)
301
for entry in inv.iter_just_entries():
302
key = (entry.file_id, entry.revision)
303
if entry.file_id is inv.root.file_id:
304
# The root id is inferred for xml v5
305
self.assertFalse(key in entry_cache)
307
self.assertIsNot(entry, entry_cache[key])
309
def test_unpack_inventory_5a_cache_no_copy(self):
310
# Passing an entry_cache should get populated with the objects
311
# The returned objects should be exact if return_from_cache is
313
entry_cache = fifo_cache.FIFOCache()
314
inv = bzrlib.xml5.serializer_v5.read_inventory_from_string(
315
_inventory_v5a, revision_id='test-rev-id',
316
entry_cache=entry_cache, return_from_cache=True)
317
for entry in inv.iter_just_entries():
318
key = (entry.file_id, entry.revision)
319
if entry.file_id is inv.root.file_id:
320
# The root id is inferred for xml v5
321
self.assertFalse(key in entry_cache)
323
self.assertIs(entry, entry_cache[key])
325
def test_unpack_inventory_5b(self):
326
inv = bzrlib.xml5.serializer_v5.read_inventory_from_string(
327
_inventory_v5b, revision_id='test-rev-id')
328
self.assertEqual('a-rev-id', inv.root.revision)
147
330
def test_repack_inventory_5(self):
148
331
inp = StringIO(_committed_inv_v5)
149
inv = serializer_v5.read_inventory(inp)
151
serializer_v5.write_inventory(inv, outp)
152
inv2 = serializer_v5.read_inventory(StringIO(outp.getvalue()))
153
self.assertEqual(inv, inv2)
332
inv = bzrlib.xml5.serializer_v5.read_inventory(inp)
334
bzrlib.xml5.serializer_v5.write_inventory(inv, outp)
335
self.assertEqualDiff(_expected_inv_v5, outp.getvalue())
336
inv2 = bzrlib.xml5.serializer_v5.read_inventory(StringIO(outp.getvalue()))
337
self.assertEqual(inv, inv2)
339
def assertRoundTrips(self, xml_string):
340
inp = StringIO(xml_string)
341
inv = bzrlib.xml5.serializer_v5.read_inventory(inp)
343
bzrlib.xml5.serializer_v5.write_inventory(inv, outp)
344
self.assertEqualDiff(xml_string, outp.getvalue())
345
lines = bzrlib.xml5.serializer_v5.write_inventory_to_lines(inv)
347
self.assertEqual(outp.readlines(), lines)
348
inv2 = bzrlib.xml5.serializer_v5.read_inventory(StringIO(outp.getvalue()))
349
self.assertEqual(inv, inv2)
351
def tests_serialize_inventory_v5_with_root(self):
352
self.assertRoundTrips(_expected_inv_v5_root)
354
def check_repack_revision(self, txt):
355
"""Check that repacking a revision yields the same information"""
357
rev = bzrlib.xml5.serializer_v5.read_revision(inp)
359
bzrlib.xml5.serializer_v5.write_revision(rev, outp)
360
outfile_contents = outp.getvalue()
361
rev2 = bzrlib.xml5.serializer_v5.read_revision(StringIO(outfile_contents))
362
self.assertEqual(rev, rev2)
155
364
def test_repack_revision_5(self):
156
365
"""Round-trip revision to XML v5"""
157
inp = StringIO(_revision_v5)
158
rev = serializer_v5.read_revision(inp)
160
serializer_v5.write_revision(rev, outp)
161
outfile_contents = outp.getvalue()
162
rev2 = serializer_v5.read_revision(StringIO(outfile_contents))
163
self.assertEqual(rev, rev2)
366
self.check_repack_revision(_revision_v5)
368
def test_repack_revision_5_utc(self):
369
self.check_repack_revision(_revision_v5_utc)
165
371
def test_pack_revision_5(self):
166
372
"""Pack revision to XML v5"""
167
373
# fixed 20051025, revisions should have final newline
168
rev = serializer_v5.read_revision_from_string(_revision_v5)
374
rev = bzrlib.xml5.serializer_v5.read_revision_from_string(_revision_v5)
169
375
outp = StringIO()
170
serializer_v5.write_revision(rev, outp)
376
bzrlib.xml5.serializer_v5.write_revision(rev, outp)
171
377
outfile_contents = outp.getvalue()
172
378
self.assertEqual(outfile_contents[-1], '\n')
173
self.assertEqualDiff(outfile_contents, serializer_v5.write_revision_to_string(rev))
379
self.assertEqualDiff(outfile_contents, bzrlib.xml5.serializer_v5.write_revision_to_string(rev))
380
self.assertEqualDiff(outfile_contents, _expected_rev_v5)
382
def test_empty_property_value(self):
383
"""Create an empty property value check that it serializes correctly"""
384
s_v5 = bzrlib.xml5.serializer_v5
385
rev = s_v5.read_revision_from_string(_revision_v5)
387
props = {'empty':'', 'one':'one'}
388
rev.properties = props
389
txt = s_v5.write_revision_to_string(rev)
390
new_rev = s_v5.read_revision_from_string(txt)
391
self.assertEqual(props, new_rev.properties)
393
def get_sample_inventory(self):
394
inv = Inventory('tree-root-321', revision_id='rev_outer')
395
inv.add(inventory.InventoryFile('file-id', 'file', 'tree-root-321'))
396
inv.add(inventory.InventoryDirectory('dir-id', 'dir',
398
inv.add(inventory.InventoryLink('link-id', 'link', 'tree-root-321'))
399
inv['tree-root-321'].revision = 'rev_outer'
400
inv['dir-id'].revision = 'rev_outer'
401
inv['file-id'].revision = 'rev_outer'
402
inv['file-id'].text_sha1 = 'A'
403
inv['file-id'].text_size = 1
404
inv['link-id'].revision = 'rev_outer'
405
inv['link-id'].symlink_target = 'a'
408
def test_roundtrip_inventory_v7(self):
409
inv = self.get_sample_inventory()
410
inv.add(inventory.TreeReference('nested-id', 'nested', 'tree-root-321',
411
'rev_outer', 'rev_inner'))
412
txt = xml7.serializer_v7.write_inventory_to_string(inv)
413
lines = xml7.serializer_v7.write_inventory_to_lines(inv)
414
self.assertEqual(bzrlib.osutils.split_lines(txt), lines)
415
self.assertEqualDiff(_expected_inv_v7, txt)
416
inv2 = xml7.serializer_v7.read_inventory_from_string(txt)
417
self.assertEqual(5, len(inv2))
418
for path, ie in inv.iter_entries():
419
self.assertEqual(ie, inv2[ie.file_id])
421
def test_roundtrip_inventory_v6(self):
422
inv = self.get_sample_inventory()
423
txt = xml6.serializer_v6.write_inventory_to_string(inv)
424
lines = xml6.serializer_v6.write_inventory_to_lines(inv)
425
self.assertEqual(bzrlib.osutils.split_lines(txt), lines)
426
self.assertEqualDiff(_expected_inv_v6, txt)
427
inv2 = xml6.serializer_v6.read_inventory_from_string(txt)
428
self.assertEqual(4, len(inv2))
429
for path, ie in inv.iter_entries():
430
self.assertEqual(ie, inv2[ie.file_id])
432
def test_wrong_format_v7(self):
433
"""Can't accidentally open a file with wrong serializer"""
434
s_v6 = bzrlib.xml6.serializer_v6
435
s_v7 = xml7.serializer_v7
436
self.assertRaises(errors.UnexpectedInventoryFormat,
437
s_v7.read_inventory_from_string, _expected_inv_v5)
438
self.assertRaises(errors.UnexpectedInventoryFormat,
439
s_v6.read_inventory_from_string, _expected_inv_v7)
441
def test_tree_reference(self):
442
s_v5 = bzrlib.xml5.serializer_v5
443
s_v6 = bzrlib.xml6.serializer_v6
444
s_v7 = xml7.serializer_v7
445
inv = Inventory('tree-root-321', revision_id='rev-outer')
446
inv.root.revision = 'root-rev'
447
inv.add(inventory.TreeReference('nested-id', 'nested', 'tree-root-321',
448
'rev-outer', 'rev-inner'))
449
self.assertRaises(errors.UnsupportedInventoryKind,
450
s_v5.write_inventory_to_string, inv)
451
self.assertRaises(errors.UnsupportedInventoryKind,
452
s_v6.write_inventory_to_string, inv)
453
txt = s_v7.write_inventory_to_string(inv)
454
lines = s_v7.write_inventory_to_lines(inv)
455
self.assertEqual(bzrlib.osutils.split_lines(txt), lines)
456
inv2 = s_v7.read_inventory_from_string(txt)
457
self.assertEqual('tree-root-321', inv2['nested-id'].parent_id)
458
self.assertEqual('rev-outer', inv2['nested-id'].revision)
459
self.assertEqual('rev-inner', inv2['nested-id'].reference_revision)
460
self.assertRaises(errors.UnsupportedInventoryKind,
461
s_v6.read_inventory_from_string,
462
txt.replace('format="7"', 'format="6"'))
463
self.assertRaises(errors.UnsupportedInventoryKind,
464
s_v5.read_inventory_from_string,
465
txt.replace('format="7"', 'format="5"'))
467
def test_roundtrip_inventory_v8(self):
468
inv = self.get_sample_inventory()
469
txt = xml8.serializer_v8.write_inventory_to_string(inv)
470
inv2 = xml8.serializer_v8.read_inventory_from_string(txt)
471
self.assertEqual(4, len(inv2))
472
for path, ie in inv.iter_entries():
473
self.assertEqual(ie, inv2[ie.file_id])
475
def test_inventory_text_v8(self):
476
inv = self.get_sample_inventory()
477
txt = xml8.serializer_v8.write_inventory_to_string(inv)
478
lines = xml8.serializer_v8.write_inventory_to_lines(inv)
479
self.assertEqual(bzrlib.osutils.split_lines(txt), lines)
480
self.assertEqualDiff(_expected_inv_v8, txt)
482
def test_revision_text_v6(self):
483
"""Pack revision to XML v6"""
484
rev = bzrlib.xml6.serializer_v6.read_revision_from_string(
486
serialized = bzrlib.xml6.serializer_v6.write_revision_to_string(rev)
487
self.assertEqualDiff(serialized, _expected_rev_v5)
489
def test_revision_text_v7(self):
490
"""Pack revision to XML v7"""
491
rev = bzrlib.xml7.serializer_v7.read_revision_from_string(
493
serialized = bzrlib.xml7.serializer_v7.write_revision_to_string(rev)
494
self.assertEqualDiff(serialized, _expected_rev_v5)
496
def test_revision_text_v8(self):
497
"""Pack revision to XML v8"""
498
rev = bzrlib.xml8.serializer_v8.read_revision_from_string(
500
serialized = bzrlib.xml8.serializer_v8.write_revision_to_string(rev)
501
self.assertEqualDiff(serialized, _expected_rev_v8)
503
def test_revision_ids_are_utf8(self):
504
"""Parsed revision_ids should all be utf-8 strings, not unicode."""
505
s_v5 = bzrlib.xml5.serializer_v5
506
rev = s_v5.read_revision_from_string(_revision_utf8_v5)
507
self.assertEqual('erik@b\xc3\xa5gfors-02', rev.revision_id)
508
self.assertIsInstance(rev.revision_id, str)
509
self.assertEqual(['erik@b\xc3\xa5gfors-01'], rev.parent_ids)
510
for parent_id in rev.parent_ids:
511
self.assertIsInstance(parent_id, str)
512
self.assertEqual(u'Include \xb5nicode characters\n', rev.message)
513
self.assertIsInstance(rev.message, unicode)
515
# ie.revision should either be None or a utf-8 revision id
516
inv = s_v5.read_inventory_from_string(_inventory_utf8_v5)
517
rev_id_1 = u'erik@b\xe5gfors-01'.encode('utf8')
518
rev_id_2 = u'erik@b\xe5gfors-02'.encode('utf8')
519
fid_root = u'TRE\xe9_ROOT'.encode('utf8')
520
fid_bar1 = u'b\xe5r-01'.encode('utf8')
521
fid_sub = u's\xb5bdir-01'.encode('utf8')
522
fid_bar2 = u'b\xe5r-02'.encode('utf8')
523
expected = [(u'', fid_root, None, rev_id_2),
524
(u'b\xe5r', fid_bar1, fid_root, rev_id_1),
525
(u's\xb5bdir', fid_sub, fid_root, rev_id_1),
526
(u's\xb5bdir/b\xe5r', fid_bar2, fid_sub, rev_id_2),
528
self.assertEqual(rev_id_2, inv.revision_id)
529
self.assertIsInstance(inv.revision_id, str)
531
actual = list(inv.iter_entries_by_dir())
532
for ((exp_path, exp_file_id, exp_parent_id, exp_rev_id),
533
(act_path, act_ie)) in zip(expected, actual):
534
self.assertEqual(exp_path, act_path)
535
self.assertIsInstance(act_path, unicode)
536
self.assertEqual(exp_file_id, act_ie.file_id)
537
self.assertIsInstance(act_ie.file_id, str)
538
self.assertEqual(exp_parent_id, act_ie.parent_id)
539
if exp_parent_id is not None:
540
self.assertIsInstance(act_ie.parent_id, str)
541
self.assertEqual(exp_rev_id, act_ie.revision)
542
if exp_rev_id is not None:
543
self.assertIsInstance(act_ie.revision, str)
545
self.assertEqual(len(expected), len(actual))
548
class TestEncodeAndEscape(TestCase):
549
"""Whitebox testing of the _encode_and_escape function."""
553
# Keep the cache clear before and after the test
554
bzrlib.xml8._ensure_utf8_re()
555
bzrlib.xml8._clear_cache()
556
self.addCleanup(bzrlib.xml8._clear_cache)
558
def test_simple_ascii(self):
559
# _encode_and_escape always appends a final ", because these parameters
560
# are being used in xml attributes, and by returning it now, we have to
561
# do fewer string operations later.
562
val = bzrlib.xml8._encode_and_escape('foo bar')
563
self.assertEqual('foo bar"', val)
564
# The second time should be cached
565
val2 = bzrlib.xml8._encode_and_escape('foo bar')
566
self.assertIs(val2, val)
568
def test_ascii_with_xml(self):
569
self.assertEqual('&'"<>"',
570
bzrlib.xml8._encode_and_escape('&\'"<>'))
572
def test_utf8_with_xml(self):
574
utf8_str = '\xc2\xb5\xc3\xa5&\xd8\xac'
575
self.assertEqual('µå&ج"',
576
bzrlib.xml8._encode_and_escape(utf8_str))
578
def test_unicode(self):
579
uni_str = u'\xb5\xe5&\u062c'
580
self.assertEqual('µå&ج"',
581
bzrlib.xml8._encode_and_escape(uni_str))
added
removed
bzrlib/tests/test_xml.py
