66
_revision_v5_utc = """\
67
<revision committer="Martin Pool <mbp@sourcefrog.net>"
68
inventory_sha1="e79c31c1deb64c163cf660fdedd476dd579ffd41"
69
revision_id="mbp@sourcefrog.net-20050905080035-e0439293f8b6b9f9"
70
timestamp="1125907235.212"
72
<message>- start splitting code for xml (de)serialization away from objects
73
preparatory to supporting multiple formats by a single library
76
<revision_ref revision_id="mbp@sourcefrog.net-20050905063503-43948f59fa127d92"/>
60
81
_committed_inv_v5 = """<inventory>
61
82
<file file_id="bar-20050901064931-73b4b1138abc9cd2"
62
83
name="bar" parent_id="TREE_ROOT"
63
text_version="mbp@foo-123123" entry_version="mbp@foo-123123"
65
<directory name="subdir"
66
file_id="foo-20050801201819-4139aa4a272f4250"
68
entry_version="mbp@foo-00"/>
84
revision="mbp@foo-123123"
85
text_sha1="A" text_size="1"/>
86
<directory name="subdir"
87
file_id="foo-20050801201819-4139aa4a272f4250"
89
revision="mbp@foo-00"/>
90
<file executable="yes" file_id="bar-20050824000535-6bc48cfad47ed134"
91
name="bar" parent_id="foo-20050801201819-4139aa4a272f4250"
93
text_sha1="B" text_size="0"/>
97
_basis_inv_v5 = """<inventory revision_id="mbp@sourcefrog.net-20050905063503-43948f59fa127d92">
98
<file file_id="bar-20050901064931-73b4b1138abc9cd2"
99
name="bar" parent_id="TREE_ROOT"
100
revision="mbp@foo-123123"/>
101
<directory name="subdir"
102
file_id="foo-20050801201819-4139aa4a272f4250"
103
parent_id="TREE_ROOT"
104
revision="mbp@foo-00"/>
69
105
<file file_id="bar-20050824000535-6bc48cfad47ed134"
70
106
name="bar" parent_id="foo-20050801201819-4139aa4a272f4250"
71
entry_version="mbp@foo-00"
72
text_version="mbp@foo-123123"/>
107
revision="mbp@foo-00"/>
112
# DO NOT REFLOW THIS. Its the exact revision we want.
113
_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">
114
<message>- start splitting code for xml (de)serialization away from objects
115
preparatory to supporting multiple formats by a single library
118
<revision_ref revision_id="mbp@sourcefrog.net-20050905063503-43948f59fa127d92" />
124
# DO NOT REFLOW THIS. Its the exact inventory we want.
125
_expected_inv_v5 = """<inventory format="5">
126
<file file_id="bar-20050901064931-73b4b1138abc9cd2" name="bar" revision="mbp@foo-123123" text_sha1="A" text_size="1" />
127
<directory file_id="foo-20050801201819-4139aa4a272f4250" name="subdir" revision="mbp@foo-00" />
128
<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" />
133
_expected_inv_v5_root = """<inventory file_id="f<" format="5" revision_id="mother!">
134
<file file_id="bar-20050901064931-73b4b1138abc9cd2" name="bar" parent_id="f<" revision="mbp@foo-123123" text_sha1="A" text_size="1" />
135
<directory file_id="foo-20050801201819-4139aa4a272f4250" name="subdir" parent_id="f<" revision="mbp@foo-00" />
136
<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" />
137
<symlink file_id="link-1" name="link" parent_id="foo-20050801201819-4139aa4a272f4250" revision="mbp@foo-00" symlink_target="a" />
141
_expected_inv_v7 = """<inventory format="7" revision_id="rev_outer">
142
<directory file_id="tree-root-321" name="" revision="rev_outer" />
143
<directory file_id="dir-id" name="dir" parent_id="tree-root-321" revision="rev_outer" />
144
<file file_id="file-id" name="file" parent_id="tree-root-321" revision="rev_outer" text_sha1="A" text_size="1" />
145
<symlink file_id="link-id" name="link" parent_id="tree-root-321" revision="rev_outer" symlink_target="a" />
146
<tree-reference file_id="nested-id" name="nested" parent_id="tree-root-321" revision="rev_outer" reference_revision="rev_inner" />
150
_revision_utf8_v5 = """<revision committer="Erik Bågfors <erik@foo.net>"
151
inventory_sha1="e79c31c1deb64c163cf660fdedd476dd579ffd41"
152
revision_id="erik@bågfors-02"
153
timestamp="1125907235.212"
155
<message>Include µnicode characters
158
<revision_ref revision_id="erik@bågfors-01"/>
163
_inventory_utf8_v5 = """<inventory file_id="TREé_ROOT" format="5"
164
revision_id="erik@bågfors-02">
165
<file file_id="bår-01"
166
name="bår" parent_id="TREé_ROOT"
167
revision="erik@bågfors-01"/>
168
<directory name="sµbdir"
169
file_id="sµbdir-01"
170
parent_id="TREé_ROOT"
171
revision="erik@bågfors-01"/>
172
<file executable="yes" file_id="bår-02"
173
name="bår" parent_id="sµbdir-01"
174
revision="erik@bågfors-02"/>
76
179
class TestSerializer(TestCase):
77
180
"""Test XML serialization"""
78
182
def test_canned_inventory(self):
79
183
"""Test unpacked a canned inventory v4 file."""
80
184
inp = StringIO(_working_inventory_v4)
91
195
"Martin Pool <mbp@sourcefrog.net>")
92
196
eq(rev.inventory_id,
93
197
"mbp@sourcefrog.net-20050905080035-e0439293f8b6b9f9")
94
eq(len(rev.parents), 1)
95
eq(rev.parents[0].revision_id,
198
eq(len(rev.parent_ids), 1)
199
eq(rev.parent_ids[0],
96
200
"mbp@sourcefrog.net-20050905063503-43948f59fa127d92")
98
202
def test_unpack_revision_5(self):
99
203
"""Test unpacking a canned revision v5"""
100
204
inp = StringIO(_revision_v5)
101
rev = serializer_v5.read_revision(inp)
205
rev = bzrlib.xml5.serializer_v5.read_revision(inp)
102
206
eq = self.assertEqual
103
207
eq(rev.committer,
104
208
"Martin Pool <mbp@sourcefrog.net>")
106
"mbp@sourcefrog.net-20050905080035-e0439293f8b6b9f9")
107
eq(len(rev.parents), 1)
209
eq(len(rev.parent_ids), 1)
108
210
eq(rev.timezone, 36000)
109
eq(rev.parents[0].revision_id,
211
eq(rev.parent_ids[0],
212
"mbp@sourcefrog.net-20050905063503-43948f59fa127d92")
214
def test_unpack_revision_5_utc(self):
215
inp = StringIO(_revision_v5_utc)
216
rev = bzrlib.xml5.serializer_v5.read_revision(inp)
217
eq = self.assertEqual
219
"Martin Pool <mbp@sourcefrog.net>")
220
eq(len(rev.parent_ids), 1)
222
eq(rev.parent_ids[0],
110
223
"mbp@sourcefrog.net-20050905063503-43948f59fa127d92")
112
225
def test_unpack_inventory_5(self):
113
226
"""Unpack canned new-style inventory"""
114
227
inp = StringIO(_committed_inv_v5)
115
inv = serializer_v5.read_inventory(inp)
116
eq = self.assertEqual
118
ie = inv['bar-20050824000535-6bc48cfad47ed134']
120
eq(ie.text_version, 'mbp@foo-123123')
121
eq(ie.entry_version, 'mbp@foo-00')
228
inv = bzrlib.xml5.serializer_v5.read_inventory(inp)
229
eq = self.assertEqual
231
ie = inv['bar-20050824000535-6bc48cfad47ed134']
233
eq(ie.revision, 'mbp@foo-00')
235
eq(inv[ie.parent_id].kind, 'directory')
237
def test_unpack_basis_inventory_5(self):
238
"""Unpack canned new-style inventory"""
239
inp = StringIO(_basis_inv_v5)
240
inv = bzrlib.xml5.serializer_v5.read_inventory(inp)
241
eq = self.assertEqual
243
eq(inv.revision_id, 'mbp@sourcefrog.net-20050905063503-43948f59fa127d92')
244
ie = inv['bar-20050824000535-6bc48cfad47ed134']
246
eq(ie.revision, 'mbp@foo-00')
122
247
eq(ie.name, 'bar')
123
248
eq(inv[ie.parent_id].kind, 'directory')
125
250
def test_repack_inventory_5(self):
126
251
inp = StringIO(_committed_inv_v5)
127
inv = serializer_v5.read_inventory(inp)
129
serializer_v5.write_inventory(inv, outp)
130
inv2 = serializer_v5.read_inventory(StringIO(outp.getvalue()))
131
self.assertEqual(inv, inv2)
252
inv = bzrlib.xml5.serializer_v5.read_inventory(inp)
254
bzrlib.xml5.serializer_v5.write_inventory(inv, outp)
255
self.assertEqualDiff(_expected_inv_v5, outp.getvalue())
256
inv2 = bzrlib.xml5.serializer_v5.read_inventory(StringIO(outp.getvalue()))
257
self.assertEqual(inv, inv2)
259
def assertRoundTrips(self, xml_string):
260
inp = StringIO(xml_string)
261
inv = bzrlib.xml5.serializer_v5.read_inventory(inp)
263
bzrlib.xml5.serializer_v5.write_inventory(inv, outp)
264
self.assertEqualDiff(xml_string, outp.getvalue())
265
lines = bzrlib.xml5.serializer_v5.write_inventory_to_lines(inv)
267
self.assertEqual(outp.readlines(), lines)
268
inv2 = bzrlib.xml5.serializer_v5.read_inventory(StringIO(outp.getvalue()))
269
self.assertEqual(inv, inv2)
271
def tests_serialize_inventory_v5_with_root(self):
272
self.assertRoundTrips(_expected_inv_v5_root)
274
def check_repack_revision(self, txt):
275
"""Check that repacking a revision yields the same information"""
277
rev = bzrlib.xml5.serializer_v5.read_revision(inp)
279
bzrlib.xml5.serializer_v5.write_revision(rev, outp)
280
outfile_contents = outp.getvalue()
281
rev2 = bzrlib.xml5.serializer_v5.read_revision(StringIO(outfile_contents))
282
self.assertEqual(rev, rev2)
133
284
def test_repack_revision_5(self):
134
inp = StringIO(_revision_v5)
135
rev = serializer_v5.read_revision(inp)
137
serializer_v5.write_revision(rev, outp)
138
rev2 = serializer_v5.read_revision(StringIO(outp.getvalue()))
139
self.assertEqual(rev, rev2)
285
"""Round-trip revision to XML v5"""
286
self.check_repack_revision(_revision_v5)
288
def test_repack_revision_5_utc(self):
289
self.check_repack_revision(_revision_v5_utc)
291
def test_pack_revision_5(self):
292
"""Pack revision to XML v5"""
293
# fixed 20051025, revisions should have final newline
294
rev = bzrlib.xml5.serializer_v5.read_revision_from_string(_revision_v5)
296
bzrlib.xml5.serializer_v5.write_revision(rev, outp)
297
outfile_contents = outp.getvalue()
298
self.assertEqual(outfile_contents[-1], '\n')
299
self.assertEqualDiff(outfile_contents, bzrlib.xml5.serializer_v5.write_revision_to_string(rev))
300
self.assertEqualDiff(outfile_contents, _expected_rev_v5)
302
def test_empty_property_value(self):
303
"""Create an empty property value check that it serializes correctly"""
304
s_v5 = bzrlib.xml5.serializer_v5
305
rev = s_v5.read_revision_from_string(_revision_v5)
307
props = {'empty':'', 'one':'one'}
308
rev.properties = props
309
txt = s_v5.write_revision_to_string(rev)
310
new_rev = s_v5.read_revision_from_string(txt)
311
self.assertEqual(props, new_rev.properties)
313
def test_roundtrip_inventory_v7(self):
314
inv = Inventory('tree-root-321', revision_id='rev_outer')
315
inv.add(inventory.TreeReference('nested-id', 'nested', 'tree-root-321',
316
'rev_outer', 'rev_inner'))
317
inv.add(inventory.InventoryFile('file-id', 'file', 'tree-root-321'))
318
inv.add(inventory.InventoryDirectory('dir-id', 'dir',
320
inv.add(inventory.InventoryLink('link-id', 'link', 'tree-root-321'))
321
inv['tree-root-321'].revision = 'rev_outer'
322
inv['dir-id'].revision = 'rev_outer'
323
inv['file-id'].revision = 'rev_outer'
324
inv['file-id'].text_sha1 = 'A'
325
inv['file-id'].text_size = 1
326
inv['link-id'].revision = 'rev_outer'
327
inv['link-id'].symlink_target = 'a'
328
txt = xml7.serializer_v7.write_inventory_to_string(inv)
329
lines = xml7.serializer_v7.write_inventory_to_lines(inv)
330
self.assertEqual(bzrlib.osutils.split_lines(txt), lines)
331
self.assertEqualDiff(_expected_inv_v7, txt)
332
inv2 = xml7.serializer_v7.read_inventory_from_string(txt)
333
self.assertEqual(5, len(inv2))
334
for path, ie in inv.iter_entries():
335
self.assertEqual(ie, inv2[ie.file_id])
337
def test_wrong_format_v7(self):
338
"""Can't accidentally open a file with wrong serializer"""
339
s_v6 = bzrlib.xml6.serializer_v6
340
s_v7 = xml7.serializer_v7
341
self.assertRaises(errors.UnexpectedInventoryFormat,
342
s_v7.read_inventory_from_string, _expected_inv_v5)
343
self.assertRaises(errors.UnexpectedInventoryFormat,
344
s_v6.read_inventory_from_string, _expected_inv_v7)
346
def test_tree_reference(self):
347
s_v5 = bzrlib.xml5.serializer_v5
348
s_v6 = bzrlib.xml6.serializer_v6
349
s_v7 = xml7.serializer_v7
350
inv = Inventory('tree-root-321', revision_id='rev-outer')
351
inv.root.revision = 'root-rev'
352
inv.add(inventory.TreeReference('nested-id', 'nested', 'tree-root-321',
353
'rev-outer', 'rev-inner'))
354
self.assertRaises(errors.UnsupportedInventoryKind,
355
s_v5.write_inventory_to_string, inv)
356
self.assertRaises(errors.UnsupportedInventoryKind,
357
s_v6.write_inventory_to_string, inv)
358
txt = s_v7.write_inventory_to_string(inv)
359
lines = s_v7.write_inventory_to_lines(inv)
360
self.assertEqual(bzrlib.osutils.split_lines(txt), lines)
361
inv2 = s_v7.read_inventory_from_string(txt)
362
self.assertEqual('tree-root-321', inv2['nested-id'].parent_id)
363
self.assertEqual('rev-outer', inv2['nested-id'].revision)
364
self.assertEqual('rev-inner', inv2['nested-id'].reference_revision)
365
self.assertRaises(errors.UnsupportedInventoryKind,
366
s_v6.read_inventory_from_string,
367
txt.replace('format="7"', 'format="6"'))
368
self.assertRaises(errors.UnsupportedInventoryKind,
369
s_v5.read_inventory_from_string,
370
txt.replace('format="7"', 'format="5"'))
372
def test_revision_ids_are_utf8(self):
373
"""Parsed revision_ids should all be utf-8 strings, not unicode."""
374
s_v5 = bzrlib.xml5.serializer_v5
375
rev = s_v5.read_revision_from_string(_revision_utf8_v5)
376
self.assertEqual('erik@b\xc3\xa5gfors-02', rev.revision_id)
377
self.assertIsInstance(rev.revision_id, str)
378
self.assertEqual(['erik@b\xc3\xa5gfors-01'], rev.parent_ids)
379
for parent_id in rev.parent_ids:
380
self.assertIsInstance(parent_id, str)
381
self.assertEqual(u'Include \xb5nicode characters\n', rev.message)
382
self.assertIsInstance(rev.message, unicode)
384
# ie.revision should either be None or a utf-8 revision id
385
inv = s_v5.read_inventory_from_string(_inventory_utf8_v5)
386
rev_id_1 = u'erik@b\xe5gfors-01'.encode('utf8')
387
rev_id_2 = u'erik@b\xe5gfors-02'.encode('utf8')
388
fid_root = u'TRE\xe9_ROOT'.encode('utf8')
389
fid_bar1 = u'b\xe5r-01'.encode('utf8')
390
fid_sub = u's\xb5bdir-01'.encode('utf8')
391
fid_bar2 = u'b\xe5r-02'.encode('utf8')
392
expected = [(u'', fid_root, None, rev_id_2),
393
(u'b\xe5r', fid_bar1, fid_root, rev_id_1),
394
(u's\xb5bdir', fid_sub, fid_root, rev_id_1),
395
(u's\xb5bdir/b\xe5r', fid_bar2, fid_sub, rev_id_2),
397
self.assertEqual(rev_id_2, inv.revision_id)
398
self.assertIsInstance(inv.revision_id, str)
400
actual = list(inv.iter_entries_by_dir())
401
for ((exp_path, exp_file_id, exp_parent_id, exp_rev_id),
402
(act_path, act_ie)) in zip(expected, actual):
403
self.assertEqual(exp_path, act_path)
404
self.assertIsInstance(act_path, unicode)
405
self.assertEqual(exp_file_id, act_ie.file_id)
406
self.assertIsInstance(act_ie.file_id, str)
407
self.assertEqual(exp_parent_id, act_ie.parent_id)
408
if exp_parent_id is not None:
409
self.assertIsInstance(act_ie.parent_id, str)
410
self.assertEqual(exp_rev_id, act_ie.revision)
411
if exp_rev_id is not None:
412
self.assertIsInstance(act_ie.revision, str)
414
self.assertEqual(len(expected), len(actual))
416
def test_registry(self):
417
self.assertIs(serializer_v4,
418
xml_serializer.format_registry.get('4'))
419
self.assertIs(bzrlib.xml5.serializer_v5,
420
xml_serializer.format_registry.get('5'))
421
self.assertIs(bzrlib.xml6.serializer_v6,
422
xml_serializer.format_registry.get('6'))
423
self.assertIs(bzrlib.xml7.serializer_v7,
424
xml_serializer.format_registry.get('7'))
427
class TestEncodeAndEscape(TestCase):
428
"""Whitebox testing of the _encode_and_escape function."""
431
# Keep the cache clear before and after the test
432
bzrlib.xml5._ensure_utf8_re()
433
bzrlib.xml5._clear_cache()
434
self.addCleanup(bzrlib.xml5._clear_cache)
436
def test_simple_ascii(self):
437
# _encode_and_escape always appends a final ", because these parameters
438
# are being used in xml attributes, and by returning it now, we have to
439
# do fewer string operations later.
440
val = bzrlib.xml5._encode_and_escape('foo bar')
441
self.assertEqual('foo bar"', val)
442
# The second time should be cached
443
val2 = bzrlib.xml5._encode_and_escape('foo bar')
444
self.assertIs(val2, val)
446
def test_ascii_with_xml(self):
447
self.assertEqual('&'"<>"',
448
bzrlib.xml5._encode_and_escape('&\'"<>'))
450
def test_utf8_with_xml(self):
452
utf8_str = '\xc2\xb5\xc3\xa5&\xd8\xac'
453
self.assertEqual('µå&ج"',
454
bzrlib.xml5._encode_and_escape(utf8_str))
456
def test_unicode(self):
457
uni_str = u'\xb5\xe5&\u062c'
458
self.assertEqual('µå&ج"',
459
bzrlib.xml5._encode_and_escape(uni_str))