~bzr-pqm/bzr/bzr.dev

« back to all changes in this revision

Viewing changes to bzrlib/tests/test_xml.py

  • Committer: Canonical.com Patch Queue Manager
  • Date: 2007-10-24 12:49:17 UTC
  • mfrom: (2935.1.1 ianc-integration)
  • Revision ID: pqm@pqm.ubuntu.com-20071024124917-xb75eckyxx6vkrlg
Makefile fixes - hooks.html generation & allow python to be overridden (Ian Clatworthy)

Show diffs side-by-side

added added

removed removed

Lines of Context:
1
 
# Copyright (C) 2005 by Canonical Ltd
 
1
# Copyright (C) 2005 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
16
16
 
17
17
from cStringIO import StringIO
18
18
 
 
19
from bzrlib import (
 
20
    errors, 
 
21
    inventory, 
 
22
    xml7,
 
23
    xml_serializer,
 
24
    )
19
25
from bzrlib.tests import TestCase
20
26
from bzrlib.inventory import Inventory, InventoryEntry
21
27
from bzrlib.xml4 import serializer_v4
32
38
    inventory_id="mbp@sourcefrog.net-20050905080035-e0439293f8b6b9f9"
33
39
    inventory_sha1="e79c31c1deb64c163cf660fdedd476dd579ffd41"
34
40
    revision_id="mbp@sourcefrog.net-20050905080035-e0439293f8b6b9f9"
35
 
    timestamp="1125907235.211783886"
 
41
    timestamp="1125907235.212"
36
42
    timezone="36000">
37
43
<message>- start splitting code for xml (de)serialization away from objects
38
44
  preparatory to supporting multiple formats by a single library
46
52
_revision_v5 = """<revision committer="Martin Pool &lt;mbp@sourcefrog.net&gt;"
47
53
    inventory_sha1="e79c31c1deb64c163cf660fdedd476dd579ffd41"
48
54
    revision_id="mbp@sourcefrog.net-20050905080035-e0439293f8b6b9f9"
49
 
    timestamp="1125907235.211783886"
 
55
    timestamp="1125907235.212"
50
56
    timezone="36000">
51
57
<message>- start splitting code for xml (de)serialization away from objects
52
58
  preparatory to supporting multiple formats by a single library
61
67
<revision committer="Martin Pool &lt;mbp@sourcefrog.net&gt;"
62
68
    inventory_sha1="e79c31c1deb64c163cf660fdedd476dd579ffd41"
63
69
    revision_id="mbp@sourcefrog.net-20050905080035-e0439293f8b6b9f9"
64
 
    timestamp="1125907235.211783886"
 
70
    timestamp="1125907235.212"
65
71
    timezone="0">
66
72
<message>- start splitting code for xml (de)serialization away from objects
67
73
  preparatory to supporting multiple formats by a single library
75
81
_committed_inv_v5 = """<inventory>
76
82
<file file_id="bar-20050901064931-73b4b1138abc9cd2" 
77
83
      name="bar" parent_id="TREE_ROOT" 
78
 
      revision="mbp@foo-123123"/>
 
84
      revision="mbp@foo-123123"
 
85
      text_sha1="A" text_size="1"/>
79
86
<directory name="subdir"
80
87
           file_id="foo-20050801201819-4139aa4a272f4250"
81
88
           parent_id="TREE_ROOT" 
82
89
           revision="mbp@foo-00"/>
83
 
<file file_id="bar-20050824000535-6bc48cfad47ed134" 
 
90
<file executable="yes" file_id="bar-20050824000535-6bc48cfad47ed134" 
84
91
      name="bar" parent_id="foo-20050801201819-4139aa4a272f4250" 
85
 
      revision="mbp@foo-00"/>
 
92
      revision="mbp@foo-00"
 
93
      text_sha1="B" text_size="0"/>
86
94
</inventory>
87
95
"""
88
96
 
101
109
"""
102
110
 
103
111
 
 
112
# DO NOT REFLOW THIS. Its the exact revision we want.
 
113
_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">
 
114
<message>- start splitting code for xml (de)serialization away from objects
 
115
  preparatory to supporting multiple formats by a single library
 
116
</message>
 
117
<parents>
 
118
<revision_ref revision_id="mbp@sourcefrog.net-20050905063503-43948f59fa127d92" />
 
119
</parents>
 
120
</revision>
 
121
"""
 
122
 
 
123
 
 
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" />
 
129
</inventory>
 
130
"""
 
131
 
 
132
 
 
133
_expected_inv_v5_root = """<inventory file_id="f&lt;" format="5" revision_id="mother!">
 
134
<file file_id="bar-20050901064931-73b4b1138abc9cd2" name="bar" parent_id="f&lt;" revision="mbp@foo-123123" text_sha1="A" text_size="1" />
 
135
<directory file_id="foo-20050801201819-4139aa4a272f4250" name="subdir" parent_id="f&lt;" 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" />
 
138
</inventory>
 
139
"""
 
140
 
 
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" />
 
147
</inventory>
 
148
"""
 
149
 
 
150
_revision_utf8_v5 = """<revision committer="Erik B&#229;gfors &lt;erik@foo.net&gt;"
 
151
    inventory_sha1="e79c31c1deb64c163cf660fdedd476dd579ffd41"
 
152
    revision_id="erik@b&#229;gfors-02"
 
153
    timestamp="1125907235.212"
 
154
    timezone="36000">
 
155
<message>Include &#181;nicode characters
 
156
</message>
 
157
<parents>
 
158
<revision_ref revision_id="erik@b&#229;gfors-01"/>
 
159
</parents>
 
160
</revision>
 
161
"""
 
162
 
 
163
_inventory_utf8_v5 = """<inventory file_id="TRE&#233;_ROOT" format="5"
 
164
                                   revision_id="erik@b&#229;gfors-02">
 
165
<file file_id="b&#229;r-01"
 
166
      name="b&#229;r" parent_id="TRE&#233;_ROOT"
 
167
      revision="erik@b&#229;gfors-01"/>
 
168
<directory name="s&#181;bdir"
 
169
           file_id="s&#181;bdir-01"
 
170
           parent_id="TRE&#233;_ROOT"
 
171
           revision="erik@b&#229;gfors-01"/>
 
172
<file executable="yes" file_id="b&#229;r-02"
 
173
      name="b&#229;r" parent_id="s&#181;bdir-01"
 
174
      revision="erik@b&#229;gfors-02"/>
 
175
</inventory>
 
176
"""
 
177
 
 
178
# Before revision_id was always stored as an attribute
 
179
_inventory_v5a = """<inventory format="5">
 
180
</inventory>
 
181
"""
 
182
 
 
183
# Before revision_id was always stored as an attribute
 
184
_inventory_v5b = """<inventory format="5" revision_id="a-rev-id">
 
185
</inventory>
 
186
"""
 
187
 
 
188
 
104
189
class TestSerializer(TestCase):
105
190
    """Test XML serialization"""
 
191
 
106
192
    def test_canned_inventory(self):
107
193
        """Test unpacked a canned inventory v4 file."""
108
194
        inp = StringIO(_working_inventory_v4)
171
257
        eq(ie.name, 'bar')
172
258
        eq(inv[ie.parent_id].kind, 'directory')
173
259
 
 
260
    def test_unpack_inventory_5a(self):
 
261
        inv = bzrlib.xml5.serializer_v5.read_inventory_from_string(
 
262
                _inventory_v5a, revision_id='test-rev-id')
 
263
        self.assertEqual('test-rev-id', inv.root.revision)
 
264
 
 
265
    def test_unpack_inventory_5b(self):
 
266
        inv = bzrlib.xml5.serializer_v5.read_inventory_from_string(
 
267
                _inventory_v5b, revision_id='test-rev-id')
 
268
        self.assertEqual('a-rev-id', inv.root.revision)
 
269
 
174
270
    def test_repack_inventory_5(self):
175
271
        inp = StringIO(_committed_inv_v5)
176
272
        inv = bzrlib.xml5.serializer_v5.read_inventory(inp)
177
273
        outp = StringIO()
178
274
        bzrlib.xml5.serializer_v5.write_inventory(inv, outp)
179
 
        inv2 = bzrlib.xml5.serializer_v5.read_inventory(StringIO(outp.getvalue()))
180
 
        self.assertEqual(inv, inv2)
 
275
        self.assertEqualDiff(_expected_inv_v5, outp.getvalue())
 
276
        inv2 = bzrlib.xml5.serializer_v5.read_inventory(StringIO(outp.getvalue()))
 
277
        self.assertEqual(inv, inv2)
 
278
    
 
279
    def assertRoundTrips(self, xml_string):
 
280
        inp = StringIO(xml_string)
 
281
        inv = bzrlib.xml5.serializer_v5.read_inventory(inp)
 
282
        outp = StringIO()
 
283
        bzrlib.xml5.serializer_v5.write_inventory(inv, outp)
 
284
        self.assertEqualDiff(xml_string, outp.getvalue())
 
285
        lines = bzrlib.xml5.serializer_v5.write_inventory_to_lines(inv)
 
286
        outp.seek(0)
 
287
        self.assertEqual(outp.readlines(), lines)
 
288
        inv2 = bzrlib.xml5.serializer_v5.read_inventory(StringIO(outp.getvalue()))
 
289
        self.assertEqual(inv, inv2)
 
290
 
 
291
    def tests_serialize_inventory_v5_with_root(self):
 
292
        self.assertRoundTrips(_expected_inv_v5_root)
181
293
 
182
294
    def check_repack_revision(self, txt):
183
295
        """Check that repacking a revision yields the same information"""
205
317
        outfile_contents = outp.getvalue()
206
318
        self.assertEqual(outfile_contents[-1], '\n')
207
319
        self.assertEqualDiff(outfile_contents, bzrlib.xml5.serializer_v5.write_revision_to_string(rev))
 
320
        self.assertEqualDiff(outfile_contents, _expected_rev_v5)
208
321
 
209
322
    def test_empty_property_value(self):
210
323
        """Create an empty property value check that it serializes correctly"""
216
329
        txt = s_v5.write_revision_to_string(rev)
217
330
        new_rev = s_v5.read_revision_from_string(txt)
218
331
        self.assertEqual(props, new_rev.properties)
 
332
 
 
333
    def test_roundtrip_inventory_v7(self):
 
334
        inv = Inventory('tree-root-321', revision_id='rev_outer')
 
335
        inv.add(inventory.TreeReference('nested-id', 'nested', 'tree-root-321',
 
336
                                        'rev_outer', 'rev_inner'))
 
337
        inv.add(inventory.InventoryFile('file-id', 'file', 'tree-root-321'))
 
338
        inv.add(inventory.InventoryDirectory('dir-id', 'dir', 
 
339
                                             'tree-root-321'))
 
340
        inv.add(inventory.InventoryLink('link-id', 'link', 'tree-root-321'))
 
341
        inv['tree-root-321'].revision = 'rev_outer'
 
342
        inv['dir-id'].revision = 'rev_outer'
 
343
        inv['file-id'].revision = 'rev_outer'
 
344
        inv['file-id'].text_sha1 = 'A'
 
345
        inv['file-id'].text_size = 1
 
346
        inv['link-id'].revision = 'rev_outer'
 
347
        inv['link-id'].symlink_target = 'a'
 
348
        txt = xml7.serializer_v7.write_inventory_to_string(inv)
 
349
        lines = xml7.serializer_v7.write_inventory_to_lines(inv)
 
350
        self.assertEqual(bzrlib.osutils.split_lines(txt), lines)
 
351
        self.assertEqualDiff(_expected_inv_v7, txt)
 
352
        inv2 = xml7.serializer_v7.read_inventory_from_string(txt)
 
353
        self.assertEqual(5, len(inv2))
 
354
        for path, ie in inv.iter_entries():
 
355
            self.assertEqual(ie, inv2[ie.file_id])
 
356
 
 
357
    def test_wrong_format_v7(self):
 
358
        """Can't accidentally open a file with wrong serializer"""
 
359
        s_v6 = bzrlib.xml6.serializer_v6
 
360
        s_v7 = xml7.serializer_v7
 
361
        self.assertRaises(errors.UnexpectedInventoryFormat, 
 
362
                          s_v7.read_inventory_from_string, _expected_inv_v5)
 
363
        self.assertRaises(errors.UnexpectedInventoryFormat, 
 
364
                          s_v6.read_inventory_from_string, _expected_inv_v7)
 
365
 
 
366
    def test_tree_reference(self):
 
367
        s_v5 = bzrlib.xml5.serializer_v5
 
368
        s_v6 = bzrlib.xml6.serializer_v6
 
369
        s_v7 = xml7.serializer_v7
 
370
        inv = Inventory('tree-root-321', revision_id='rev-outer')
 
371
        inv.root.revision = 'root-rev'
 
372
        inv.add(inventory.TreeReference('nested-id', 'nested', 'tree-root-321',
 
373
                                        'rev-outer', 'rev-inner'))
 
374
        self.assertRaises(errors.UnsupportedInventoryKind,
 
375
                          s_v5.write_inventory_to_string, inv)
 
376
        self.assertRaises(errors.UnsupportedInventoryKind,
 
377
                          s_v6.write_inventory_to_string, inv)
 
378
        txt = s_v7.write_inventory_to_string(inv)
 
379
        lines = s_v7.write_inventory_to_lines(inv)
 
380
        self.assertEqual(bzrlib.osutils.split_lines(txt), lines)
 
381
        inv2 = s_v7.read_inventory_from_string(txt)
 
382
        self.assertEqual('tree-root-321', inv2['nested-id'].parent_id)
 
383
        self.assertEqual('rev-outer', inv2['nested-id'].revision)
 
384
        self.assertEqual('rev-inner', inv2['nested-id'].reference_revision)
 
385
        self.assertRaises(errors.UnsupportedInventoryKind, 
 
386
                          s_v6.read_inventory_from_string,
 
387
                          txt.replace('format="7"', 'format="6"'))
 
388
        self.assertRaises(errors.UnsupportedInventoryKind, 
 
389
                          s_v5.read_inventory_from_string,
 
390
                          txt.replace('format="7"', 'format="5"'))
 
391
 
 
392
    def test_revision_ids_are_utf8(self):
 
393
        """Parsed revision_ids should all be utf-8 strings, not unicode."""
 
394
        s_v5 = bzrlib.xml5.serializer_v5
 
395
        rev = s_v5.read_revision_from_string(_revision_utf8_v5)
 
396
        self.assertEqual('erik@b\xc3\xa5gfors-02', rev.revision_id)
 
397
        self.assertIsInstance(rev.revision_id, str)
 
398
        self.assertEqual(['erik@b\xc3\xa5gfors-01'], rev.parent_ids)
 
399
        for parent_id in rev.parent_ids:
 
400
            self.assertIsInstance(parent_id, str)
 
401
        self.assertEqual(u'Include \xb5nicode characters\n', rev.message)
 
402
        self.assertIsInstance(rev.message, unicode)
 
403
 
 
404
        # ie.revision should either be None or a utf-8 revision id
 
405
        inv = s_v5.read_inventory_from_string(_inventory_utf8_v5)
 
406
        rev_id_1 = u'erik@b\xe5gfors-01'.encode('utf8')
 
407
        rev_id_2 = u'erik@b\xe5gfors-02'.encode('utf8')
 
408
        fid_root = u'TRE\xe9_ROOT'.encode('utf8')
 
409
        fid_bar1 = u'b\xe5r-01'.encode('utf8')
 
410
        fid_sub = u's\xb5bdir-01'.encode('utf8')
 
411
        fid_bar2 = u'b\xe5r-02'.encode('utf8')
 
412
        expected = [(u'', fid_root, None, rev_id_2),
 
413
                    (u'b\xe5r', fid_bar1, fid_root, rev_id_1),
 
414
                    (u's\xb5bdir', fid_sub, fid_root, rev_id_1),
 
415
                    (u's\xb5bdir/b\xe5r', fid_bar2, fid_sub, rev_id_2),
 
416
                   ]
 
417
        self.assertEqual(rev_id_2, inv.revision_id)
 
418
        self.assertIsInstance(inv.revision_id, str)
 
419
 
 
420
        actual = list(inv.iter_entries_by_dir())
 
421
        for ((exp_path, exp_file_id, exp_parent_id, exp_rev_id),
 
422
             (act_path, act_ie)) in zip(expected, actual):
 
423
            self.assertEqual(exp_path, act_path)
 
424
            self.assertIsInstance(act_path, unicode)
 
425
            self.assertEqual(exp_file_id, act_ie.file_id)
 
426
            self.assertIsInstance(act_ie.file_id, str)
 
427
            self.assertEqual(exp_parent_id, act_ie.parent_id)
 
428
            if exp_parent_id is not None:
 
429
                self.assertIsInstance(act_ie.parent_id, str)
 
430
            self.assertEqual(exp_rev_id, act_ie.revision)
 
431
            if exp_rev_id is not None:
 
432
                self.assertIsInstance(act_ie.revision, str)
 
433
 
 
434
        self.assertEqual(len(expected), len(actual))
 
435
 
 
436
    def test_registry(self):
 
437
        self.assertIs(serializer_v4,
 
438
                      xml_serializer.format_registry.get('4'))
 
439
        self.assertIs(bzrlib.xml5.serializer_v5,
 
440
                      xml_serializer.format_registry.get('5'))
 
441
        self.assertIs(bzrlib.xml6.serializer_v6,
 
442
                      xml_serializer.format_registry.get('6'))
 
443
        self.assertIs(bzrlib.xml7.serializer_v7,
 
444
                      xml_serializer.format_registry.get('7'))
 
445
 
 
446
 
 
447
class TestEncodeAndEscape(TestCase):
 
448
    """Whitebox testing of the _encode_and_escape function."""
 
449
 
 
450
    def setUp(self):
 
451
        # Keep the cache clear before and after the test
 
452
        bzrlib.xml5._ensure_utf8_re()
 
453
        bzrlib.xml5._clear_cache()
 
454
        self.addCleanup(bzrlib.xml5._clear_cache)
 
455
 
 
456
    def test_simple_ascii(self):
 
457
        # _encode_and_escape always appends a final ", because these parameters
 
458
        # are being used in xml attributes, and by returning it now, we have to
 
459
        # do fewer string operations later.
 
460
        val = bzrlib.xml5._encode_and_escape('foo bar')
 
461
        self.assertEqual('foo bar"', val)
 
462
        # The second time should be cached
 
463
        val2 = bzrlib.xml5._encode_and_escape('foo bar')
 
464
        self.assertIs(val2, val)
 
465
 
 
466
    def test_ascii_with_xml(self):
 
467
        self.assertEqual('&amp;&apos;&quot;&lt;&gt;"',
 
468
                         bzrlib.xml5._encode_and_escape('&\'"<>'))
 
469
 
 
470
    def test_utf8_with_xml(self):
 
471
        # u'\xb5\xe5&\u062c'
 
472
        utf8_str = '\xc2\xb5\xc3\xa5&\xd8\xac'
 
473
        self.assertEqual('&#181;&#229;&amp;&#1580;"',
 
474
                         bzrlib.xml5._encode_and_escape(utf8_str))
 
475
 
 
476
    def test_unicode(self):
 
477
        uni_str = u'\xb5\xe5&\u062c'
 
478
        self.assertEqual('&#181;&#229;&amp;&#1580;"',
 
479
                         bzrlib.xml5._encode_and_escape(uni_str))