~bzr-pqm/bzr/bzr.dev

« back to all changes in this revision

Viewing changes to bzrlib/tests/test_xml.py

  • Committer: Ian Clatworthy
  • Date: 2009-01-19 02:24:15 UTC
  • mto: This revision was merged to the branch mainline in revision 3944.
  • Revision ID: ian.clatworthy@canonical.com-20090119022415-mo0mcfeiexfktgwt
apply jam's log --short fix (Ian Clatworthy)

Show diffs side-by-side

added added

removed removed

Lines of Context:
1
 
# Copyright (C) 2005 by Canonical Ltd
2
 
 
 
1
# Copyright (C) 2005 Canonical Ltd
 
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
5
5
# the Free Software Foundation; either version 2 of the License, or
6
6
# (at your option) any later version.
7
 
 
 
7
#
8
8
# This program is distributed in the hope that it will be useful,
9
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
11
11
# GNU General Public License for more details.
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
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
16
16
 
17
17
from cStringIO import StringIO
18
18
 
 
19
from bzrlib import (
 
20
    errors, 
 
21
    inventory, 
 
22
    xml7,
 
23
    xml8,
 
24
    xml_serializer,
 
25
    )
19
26
from bzrlib.tests import TestCase
20
27
from bzrlib.inventory import Inventory, InventoryEntry
21
28
from bzrlib.xml4 import serializer_v4
32
39
    inventory_id="mbp@sourcefrog.net-20050905080035-e0439293f8b6b9f9"
33
40
    inventory_sha1="e79c31c1deb64c163cf660fdedd476dd579ffd41"
34
41
    revision_id="mbp@sourcefrog.net-20050905080035-e0439293f8b6b9f9"
35
 
    timestamp="1125907235.211783886"
 
42
    timestamp="1125907235.212"
36
43
    timezone="36000">
37
44
<message>- start splitting code for xml (de)serialization away from objects
38
45
  preparatory to supporting multiple formats by a single library
46
53
_revision_v5 = """<revision committer="Martin Pool &lt;mbp@sourcefrog.net&gt;"
47
54
    inventory_sha1="e79c31c1deb64c163cf660fdedd476dd579ffd41"
48
55
    revision_id="mbp@sourcefrog.net-20050905080035-e0439293f8b6b9f9"
49
 
    timestamp="1125907235.211783886"
 
56
    timestamp="1125907235.212"
50
57
    timezone="36000">
51
58
<message>- start splitting code for xml (de)serialization away from objects
52
59
  preparatory to supporting multiple formats by a single library
57
64
</revision>
58
65
"""
59
66
 
 
67
_revision_v5_utc = """\
 
68
<revision committer="Martin Pool &lt;mbp@sourcefrog.net&gt;"
 
69
    inventory_sha1="e79c31c1deb64c163cf660fdedd476dd579ffd41"
 
70
    revision_id="mbp@sourcefrog.net-20050905080035-e0439293f8b6b9f9"
 
71
    timestamp="1125907235.212"
 
72
    timezone="0">
 
73
<message>- start splitting code for xml (de)serialization away from objects
 
74
  preparatory to supporting multiple formats by a single library
 
75
</message>
 
76
<parents>
 
77
<revision_ref revision_id="mbp@sourcefrog.net-20050905063503-43948f59fa127d92"/>
 
78
</parents>
 
79
</revision>
 
80
"""
 
81
 
60
82
_committed_inv_v5 = """<inventory>
61
83
<file file_id="bar-20050901064931-73b4b1138abc9cd2" 
62
84
      name="bar" parent_id="TREE_ROOT" 
63
 
      revision="mbp@foo-123123"/>
 
85
      revision="mbp@foo-123123"
 
86
      text_sha1="A" text_size="1"/>
64
87
<directory name="subdir"
65
88
           file_id="foo-20050801201819-4139aa4a272f4250"
66
89
           parent_id="TREE_ROOT" 
67
90
           revision="mbp@foo-00"/>
68
 
<file file_id="bar-20050824000535-6bc48cfad47ed134" 
 
91
<file executable="yes" file_id="bar-20050824000535-6bc48cfad47ed134" 
69
92
      name="bar" parent_id="foo-20050801201819-4139aa4a272f4250" 
70
 
      revision="mbp@foo-00"/>
 
93
      revision="mbp@foo-00"
 
94
      text_sha1="B" text_size="0"/>
71
95
</inventory>
72
96
"""
73
97
 
85
109
</inventory>
86
110
"""
87
111
 
 
112
 
 
113
# DO NOT REFLOW THIS. Its the exact revision we want.
 
114
_expected_rev_v5 = """<revision committer="Martin Pool &lt;mbp@sourcefrog.net&gt;" format="5" inventory_sha1="e79c31c1deb64c163cf660fdedd476dd579ffd41" revision_id="mbp@sourcefrog.net-20050905080035-e0439293f8b6b9f9" timestamp="1125907235.212" timezone="36000">
 
115
<message>- start splitting code for xml (de)serialization away from objects
 
116
  preparatory to supporting multiple formats by a single library
 
117
</message>
 
118
<parents>
 
119
<revision_ref revision_id="mbp@sourcefrog.net-20050905063503-43948f59fa127d92" />
 
120
</parents>
 
121
</revision>
 
122
"""
 
123
 
 
124
 
 
125
# DO NOT REFLOW THIS. Its the exact inventory we want.
 
126
_expected_inv_v5 = """<inventory format="5">
 
127
<file file_id="bar-20050901064931-73b4b1138abc9cd2" name="bar" revision="mbp@foo-123123" text_sha1="A" text_size="1" />
 
128
<directory file_id="foo-20050801201819-4139aa4a272f4250" name="subdir" revision="mbp@foo-00" />
 
129
<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" />
 
130
</inventory>
 
131
"""
 
132
 
 
133
 
 
134
_expected_inv_v5_root = """<inventory file_id="f&lt;" format="5" revision_id="mother!">
 
135
<file file_id="bar-20050901064931-73b4b1138abc9cd2" name="bar" parent_id="f&lt;" revision="mbp@foo-123123" text_sha1="A" text_size="1" />
 
136
<directory file_id="foo-20050801201819-4139aa4a272f4250" name="subdir" parent_id="f&lt;" revision="mbp@foo-00" />
 
137
<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" />
 
138
<symlink file_id="link-1" name="link" parent_id="foo-20050801201819-4139aa4a272f4250" revision="mbp@foo-00" symlink_target="a" />
 
139
</inventory>
 
140
"""
 
141
 
 
142
_expected_inv_v7 = """<inventory format="7" revision_id="rev_outer">
 
143
<directory file_id="tree-root-321" name="" revision="rev_outer" />
 
144
<directory file_id="dir-id" name="dir" parent_id="tree-root-321" revision="rev_outer" />
 
145
<file file_id="file-id" name="file" parent_id="tree-root-321" revision="rev_outer" text_sha1="A" text_size="1" />
 
146
<symlink file_id="link-id" name="link" parent_id="tree-root-321" revision="rev_outer" symlink_target="a" />
 
147
<tree-reference file_id="nested-id" name="nested" parent_id="tree-root-321" revision="rev_outer" reference_revision="rev_inner" />
 
148
</inventory>
 
149
"""
 
150
 
 
151
_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">
 
152
<message>- start splitting code for xml (de)serialization away from objects
 
153
  preparatory to supporting multiple formats by a single library
 
154
</message>
 
155
<parents>
 
156
<revision_ref revision_id="mbp@sourcefrog.net-20050905063503-43948f59fa127d92" />
 
157
</parents>
 
158
</revision>
 
159
"""
 
160
 
 
161
_expected_inv_v8 = """<inventory format="8" revision_id="rev_outer">
 
162
<directory file_id="tree-root-321" name="" revision="rev_outer" />
 
163
<directory file_id="dir-id" name="dir" parent_id="tree-root-321" revision="rev_outer" />
 
164
<file file_id="file-id" name="file" parent_id="tree-root-321" revision="rev_outer" text_sha1="A" text_size="1" />
 
165
<symlink file_id="link-id" name="link" parent_id="tree-root-321" revision="rev_outer" symlink_target="a" />
 
166
</inventory>
 
167
"""
 
168
 
 
169
_revision_utf8_v5 = """<revision committer="Erik B&#229;gfors &lt;erik@foo.net&gt;"
 
170
    inventory_sha1="e79c31c1deb64c163cf660fdedd476dd579ffd41"
 
171
    revision_id="erik@b&#229;gfors-02"
 
172
    timestamp="1125907235.212"
 
173
    timezone="36000">
 
174
<message>Include &#181;nicode characters
 
175
</message>
 
176
<parents>
 
177
<revision_ref revision_id="erik@b&#229;gfors-01"/>
 
178
</parents>
 
179
</revision>
 
180
"""
 
181
 
 
182
_inventory_utf8_v5 = """<inventory file_id="TRE&#233;_ROOT" format="5"
 
183
                                   revision_id="erik@b&#229;gfors-02">
 
184
<file file_id="b&#229;r-01"
 
185
      name="b&#229;r" parent_id="TRE&#233;_ROOT"
 
186
      revision="erik@b&#229;gfors-01"/>
 
187
<directory name="s&#181;bdir"
 
188
           file_id="s&#181;bdir-01"
 
189
           parent_id="TRE&#233;_ROOT"
 
190
           revision="erik@b&#229;gfors-01"/>
 
191
<file executable="yes" file_id="b&#229;r-02"
 
192
      name="b&#229;r" parent_id="s&#181;bdir-01"
 
193
      revision="erik@b&#229;gfors-02"/>
 
194
</inventory>
 
195
"""
 
196
 
 
197
# Before revision_id was always stored as an attribute
 
198
_inventory_v5a = """<inventory format="5">
 
199
</inventory>
 
200
"""
 
201
 
 
202
# Before revision_id was always stored as an attribute
 
203
_inventory_v5b = """<inventory format="5" revision_id="a-rev-id">
 
204
</inventory>
 
205
"""
 
206
 
 
207
 
88
208
class TestSerializer(TestCase):
89
209
    """Test XML serialization"""
 
210
 
90
211
    def test_canned_inventory(self):
91
212
        """Test unpacked a canned inventory v4 file."""
92
213
        inp = StringIO(_working_inventory_v4)
119
240
        eq(rev.parent_ids[0],
120
241
           "mbp@sourcefrog.net-20050905063503-43948f59fa127d92")
121
242
 
 
243
    def test_unpack_revision_5_utc(self):
 
244
        inp = StringIO(_revision_v5_utc)
 
245
        rev = bzrlib.xml5.serializer_v5.read_revision(inp)
 
246
        eq = self.assertEqual
 
247
        eq(rev.committer,
 
248
           "Martin Pool <mbp@sourcefrog.net>")
 
249
        eq(len(rev.parent_ids), 1)
 
250
        eq(rev.timezone, 0)
 
251
        eq(rev.parent_ids[0],
 
252
           "mbp@sourcefrog.net-20050905063503-43948f59fa127d92")
 
253
 
122
254
    def test_unpack_inventory_5(self):
123
255
        """Unpack canned new-style inventory"""
124
256
        inp = StringIO(_committed_inv_v5)
144
276
        eq(ie.name, 'bar')
145
277
        eq(inv[ie.parent_id].kind, 'directory')
146
278
 
 
279
    def test_unpack_inventory_5a(self):
 
280
        inv = bzrlib.xml5.serializer_v5.read_inventory_from_string(
 
281
                _inventory_v5a, revision_id='test-rev-id')
 
282
        self.assertEqual('test-rev-id', inv.root.revision)
 
283
 
 
284
    def test_unpack_inventory_5b(self):
 
285
        inv = bzrlib.xml5.serializer_v5.read_inventory_from_string(
 
286
                _inventory_v5b, revision_id='test-rev-id')
 
287
        self.assertEqual('a-rev-id', inv.root.revision)
 
288
 
147
289
    def test_repack_inventory_5(self):
148
290
        inp = StringIO(_committed_inv_v5)
149
291
        inv = bzrlib.xml5.serializer_v5.read_inventory(inp)
150
292
        outp = StringIO()
151
293
        bzrlib.xml5.serializer_v5.write_inventory(inv, outp)
152
 
        inv2 = bzrlib.xml5.serializer_v5.read_inventory(StringIO(outp.getvalue()))
153
 
        self.assertEqual(inv, inv2)
 
294
        self.assertEqualDiff(_expected_inv_v5, outp.getvalue())
 
295
        inv2 = bzrlib.xml5.serializer_v5.read_inventory(StringIO(outp.getvalue()))
 
296
        self.assertEqual(inv, inv2)
 
297
    
 
298
    def assertRoundTrips(self, xml_string):
 
299
        inp = StringIO(xml_string)
 
300
        inv = bzrlib.xml5.serializer_v5.read_inventory(inp)
 
301
        outp = StringIO()
 
302
        bzrlib.xml5.serializer_v5.write_inventory(inv, outp)
 
303
        self.assertEqualDiff(xml_string, outp.getvalue())
 
304
        lines = bzrlib.xml5.serializer_v5.write_inventory_to_lines(inv)
 
305
        outp.seek(0)
 
306
        self.assertEqual(outp.readlines(), lines)
 
307
        inv2 = bzrlib.xml5.serializer_v5.read_inventory(StringIO(outp.getvalue()))
 
308
        self.assertEqual(inv, inv2)
 
309
 
 
310
    def tests_serialize_inventory_v5_with_root(self):
 
311
        self.assertRoundTrips(_expected_inv_v5_root)
 
312
 
 
313
    def check_repack_revision(self, txt):
 
314
        """Check that repacking a revision yields the same information"""
 
315
        inp = StringIO(txt)
 
316
        rev = bzrlib.xml5.serializer_v5.read_revision(inp)
 
317
        outp = StringIO()
 
318
        bzrlib.xml5.serializer_v5.write_revision(rev, outp)
 
319
        outfile_contents = outp.getvalue()
 
320
        rev2 = bzrlib.xml5.serializer_v5.read_revision(StringIO(outfile_contents))
 
321
        self.assertEqual(rev, rev2)
154
322
 
155
323
    def test_repack_revision_5(self):
156
324
        """Round-trip revision to XML v5"""
157
 
        inp = StringIO(_revision_v5)
158
 
        rev = bzrlib.xml5.serializer_v5.read_revision(inp)
159
 
        outp = StringIO()
160
 
        bzrlib.xml5.serializer_v5.write_revision(rev, outp)
161
 
        outfile_contents = outp.getvalue()
162
 
        rev2 = bzrlib.xml5.serializer_v5.read_revision(StringIO(outfile_contents))
163
 
        self.assertEqual(rev, rev2)
 
325
        self.check_repack_revision(_revision_v5)
 
326
 
 
327
    def test_repack_revision_5_utc(self):
 
328
        self.check_repack_revision(_revision_v5_utc)
164
329
 
165
330
    def test_pack_revision_5(self):
166
331
        """Pack revision to XML v5"""
171
336
        outfile_contents = outp.getvalue()
172
337
        self.assertEqual(outfile_contents[-1], '\n')
173
338
        self.assertEqualDiff(outfile_contents, bzrlib.xml5.serializer_v5.write_revision_to_string(rev))
 
339
        self.assertEqualDiff(outfile_contents, _expected_rev_v5)
 
340
 
 
341
    def test_empty_property_value(self):
 
342
        """Create an empty property value check that it serializes correctly"""
 
343
        s_v5 = bzrlib.xml5.serializer_v5
 
344
        rev = s_v5.read_revision_from_string(_revision_v5)
 
345
        outp = StringIO()
 
346
        props = {'empty':'', 'one':'one'}
 
347
        rev.properties = props
 
348
        txt = s_v5.write_revision_to_string(rev)
 
349
        new_rev = s_v5.read_revision_from_string(txt)
 
350
        self.assertEqual(props, new_rev.properties)
 
351
 
 
352
    def get_sample_inventory(self):
 
353
        inv = Inventory('tree-root-321', revision_id='rev_outer')
 
354
        inv.add(inventory.InventoryFile('file-id', 'file', 'tree-root-321'))
 
355
        inv.add(inventory.InventoryDirectory('dir-id', 'dir', 
 
356
                                             'tree-root-321'))
 
357
        inv.add(inventory.InventoryLink('link-id', 'link', 'tree-root-321'))
 
358
        inv['tree-root-321'].revision = 'rev_outer'
 
359
        inv['dir-id'].revision = 'rev_outer'
 
360
        inv['file-id'].revision = 'rev_outer'
 
361
        inv['file-id'].text_sha1 = 'A'
 
362
        inv['file-id'].text_size = 1
 
363
        inv['link-id'].revision = 'rev_outer'
 
364
        inv['link-id'].symlink_target = 'a'
 
365
        return inv
 
366
 
 
367
    def test_roundtrip_inventory_v7(self):
 
368
        inv = self.get_sample_inventory()
 
369
        inv.add(inventory.TreeReference('nested-id', 'nested', 'tree-root-321',
 
370
                                        'rev_outer', 'rev_inner'))
 
371
        txt = xml7.serializer_v7.write_inventory_to_string(inv)
 
372
        lines = xml7.serializer_v7.write_inventory_to_lines(inv)
 
373
        self.assertEqual(bzrlib.osutils.split_lines(txt), lines)
 
374
        self.assertEqualDiff(_expected_inv_v7, txt)
 
375
        inv2 = xml7.serializer_v7.read_inventory_from_string(txt)
 
376
        self.assertEqual(5, len(inv2))
 
377
        for path, ie in inv.iter_entries():
 
378
            self.assertEqual(ie, inv2[ie.file_id])
 
379
 
 
380
    def test_wrong_format_v7(self):
 
381
        """Can't accidentally open a file with wrong serializer"""
 
382
        s_v6 = bzrlib.xml6.serializer_v6
 
383
        s_v7 = xml7.serializer_v7
 
384
        self.assertRaises(errors.UnexpectedInventoryFormat, 
 
385
                          s_v7.read_inventory_from_string, _expected_inv_v5)
 
386
        self.assertRaises(errors.UnexpectedInventoryFormat, 
 
387
                          s_v6.read_inventory_from_string, _expected_inv_v7)
 
388
 
 
389
    def test_tree_reference(self):
 
390
        s_v5 = bzrlib.xml5.serializer_v5
 
391
        s_v6 = bzrlib.xml6.serializer_v6
 
392
        s_v7 = xml7.serializer_v7
 
393
        inv = Inventory('tree-root-321', revision_id='rev-outer')
 
394
        inv.root.revision = 'root-rev'
 
395
        inv.add(inventory.TreeReference('nested-id', 'nested', 'tree-root-321',
 
396
                                        'rev-outer', 'rev-inner'))
 
397
        self.assertRaises(errors.UnsupportedInventoryKind,
 
398
                          s_v5.write_inventory_to_string, inv)
 
399
        self.assertRaises(errors.UnsupportedInventoryKind,
 
400
                          s_v6.write_inventory_to_string, inv)
 
401
        txt = s_v7.write_inventory_to_string(inv)
 
402
        lines = s_v7.write_inventory_to_lines(inv)
 
403
        self.assertEqual(bzrlib.osutils.split_lines(txt), lines)
 
404
        inv2 = s_v7.read_inventory_from_string(txt)
 
405
        self.assertEqual('tree-root-321', inv2['nested-id'].parent_id)
 
406
        self.assertEqual('rev-outer', inv2['nested-id'].revision)
 
407
        self.assertEqual('rev-inner', inv2['nested-id'].reference_revision)
 
408
        self.assertRaises(errors.UnsupportedInventoryKind, 
 
409
                          s_v6.read_inventory_from_string,
 
410
                          txt.replace('format="7"', 'format="6"'))
 
411
        self.assertRaises(errors.UnsupportedInventoryKind, 
 
412
                          s_v5.read_inventory_from_string,
 
413
                          txt.replace('format="7"', 'format="5"'))
 
414
 
 
415
    def test_roundtrip_inventory_v8(self):
 
416
        inv = self.get_sample_inventory()
 
417
        txt = xml8.serializer_v8.write_inventory_to_string(inv)
 
418
        inv2 = xml8.serializer_v8.read_inventory_from_string(txt)
 
419
        self.assertEqual(4, len(inv2))
 
420
        for path, ie in inv.iter_entries():
 
421
            self.assertEqual(ie, inv2[ie.file_id])
 
422
 
 
423
    def test_inventory_text_v8(self):
 
424
        inv = self.get_sample_inventory()
 
425
        txt = xml8.serializer_v8.write_inventory_to_string(inv)
 
426
        lines = xml8.serializer_v8.write_inventory_to_lines(inv)
 
427
        self.assertEqual(bzrlib.osutils.split_lines(txt), lines)
 
428
        self.assertEqualDiff(_expected_inv_v8, txt)
 
429
 
 
430
    def test_revision_text_v6(self):
 
431
        """Pack revision to XML v6"""
 
432
        rev = bzrlib.xml6.serializer_v6.read_revision_from_string(
 
433
            _expected_rev_v5)
 
434
        serialized = bzrlib.xml6.serializer_v6.write_revision_to_string(rev)
 
435
        self.assertEqualDiff(serialized, _expected_rev_v5)
 
436
 
 
437
    def test_revision_text_v7(self):
 
438
        """Pack revision to XML v7"""
 
439
        rev = bzrlib.xml7.serializer_v7.read_revision_from_string(
 
440
            _expected_rev_v5)
 
441
        serialized = bzrlib.xml7.serializer_v7.write_revision_to_string(rev)
 
442
        self.assertEqualDiff(serialized, _expected_rev_v5)
 
443
 
 
444
    def test_revision_text_v8(self):
 
445
        """Pack revision to XML v8"""
 
446
        rev = bzrlib.xml8.serializer_v8.read_revision_from_string(
 
447
            _expected_rev_v8)
 
448
        serialized = bzrlib.xml8.serializer_v8.write_revision_to_string(rev)
 
449
        self.assertEqualDiff(serialized, _expected_rev_v8)
 
450
 
 
451
    def test_revision_ids_are_utf8(self):
 
452
        """Parsed revision_ids should all be utf-8 strings, not unicode."""
 
453
        s_v5 = bzrlib.xml5.serializer_v5
 
454
        rev = s_v5.read_revision_from_string(_revision_utf8_v5)
 
455
        self.assertEqual('erik@b\xc3\xa5gfors-02', rev.revision_id)
 
456
        self.assertIsInstance(rev.revision_id, str)
 
457
        self.assertEqual(['erik@b\xc3\xa5gfors-01'], rev.parent_ids)
 
458
        for parent_id in rev.parent_ids:
 
459
            self.assertIsInstance(parent_id, str)
 
460
        self.assertEqual(u'Include \xb5nicode characters\n', rev.message)
 
461
        self.assertIsInstance(rev.message, unicode)
 
462
 
 
463
        # ie.revision should either be None or a utf-8 revision id
 
464
        inv = s_v5.read_inventory_from_string(_inventory_utf8_v5)
 
465
        rev_id_1 = u'erik@b\xe5gfors-01'.encode('utf8')
 
466
        rev_id_2 = u'erik@b\xe5gfors-02'.encode('utf8')
 
467
        fid_root = u'TRE\xe9_ROOT'.encode('utf8')
 
468
        fid_bar1 = u'b\xe5r-01'.encode('utf8')
 
469
        fid_sub = u's\xb5bdir-01'.encode('utf8')
 
470
        fid_bar2 = u'b\xe5r-02'.encode('utf8')
 
471
        expected = [(u'', fid_root, None, rev_id_2),
 
472
                    (u'b\xe5r', fid_bar1, fid_root, rev_id_1),
 
473
                    (u's\xb5bdir', fid_sub, fid_root, rev_id_1),
 
474
                    (u's\xb5bdir/b\xe5r', fid_bar2, fid_sub, rev_id_2),
 
475
                   ]
 
476
        self.assertEqual(rev_id_2, inv.revision_id)
 
477
        self.assertIsInstance(inv.revision_id, str)
 
478
 
 
479
        actual = list(inv.iter_entries_by_dir())
 
480
        for ((exp_path, exp_file_id, exp_parent_id, exp_rev_id),
 
481
             (act_path, act_ie)) in zip(expected, actual):
 
482
            self.assertEqual(exp_path, act_path)
 
483
            self.assertIsInstance(act_path, unicode)
 
484
            self.assertEqual(exp_file_id, act_ie.file_id)
 
485
            self.assertIsInstance(act_ie.file_id, str)
 
486
            self.assertEqual(exp_parent_id, act_ie.parent_id)
 
487
            if exp_parent_id is not None:
 
488
                self.assertIsInstance(act_ie.parent_id, str)
 
489
            self.assertEqual(exp_rev_id, act_ie.revision)
 
490
            if exp_rev_id is not None:
 
491
                self.assertIsInstance(act_ie.revision, str)
 
492
 
 
493
        self.assertEqual(len(expected), len(actual))
 
494
 
 
495
    def test_registry(self):
 
496
        self.assertIs(serializer_v4,
 
497
                      xml_serializer.format_registry.get('4'))
 
498
        self.assertIs(bzrlib.xml5.serializer_v5,
 
499
                      xml_serializer.format_registry.get('5'))
 
500
        self.assertIs(bzrlib.xml6.serializer_v6,
 
501
                      xml_serializer.format_registry.get('6'))
 
502
        self.assertIs(bzrlib.xml7.serializer_v7,
 
503
                      xml_serializer.format_registry.get('7'))
 
504
        self.assertIs(bzrlib.xml8.serializer_v8,
 
505
                      xml_serializer.format_registry.get('8'))
 
506
 
 
507
 
 
508
class TestEncodeAndEscape(TestCase):
 
509
    """Whitebox testing of the _encode_and_escape function."""
 
510
 
 
511
    def setUp(self):
 
512
        # Keep the cache clear before and after the test
 
513
        bzrlib.xml8._ensure_utf8_re()
 
514
        bzrlib.xml8._clear_cache()
 
515
        self.addCleanup(bzrlib.xml8._clear_cache)
 
516
 
 
517
    def test_simple_ascii(self):
 
518
        # _encode_and_escape always appends a final ", because these parameters
 
519
        # are being used in xml attributes, and by returning it now, we have to
 
520
        # do fewer string operations later.
 
521
        val = bzrlib.xml8._encode_and_escape('foo bar')
 
522
        self.assertEqual('foo bar"', val)
 
523
        # The second time should be cached
 
524
        val2 = bzrlib.xml8._encode_and_escape('foo bar')
 
525
        self.assertIs(val2, val)
 
526
 
 
527
    def test_ascii_with_xml(self):
 
528
        self.assertEqual('&amp;&apos;&quot;&lt;&gt;"',
 
529
                         bzrlib.xml8._encode_and_escape('&\'"<>'))
 
530
 
 
531
    def test_utf8_with_xml(self):
 
532
        # u'\xb5\xe5&\u062c'
 
533
        utf8_str = '\xc2\xb5\xc3\xa5&\xd8\xac'
 
534
        self.assertEqual('&#181;&#229;&amp;&#1580;"',
 
535
                         bzrlib.xml8._encode_and_escape(utf8_str))
 
536
 
 
537
    def test_unicode(self):
 
538
        uni_str = u'\xb5\xe5&\u062c'
 
539
        self.assertEqual('&#181;&#229;&amp;&#1580;"',
 
540
                         bzrlib.xml8._encode_and_escape(uni_str))