~bzr-pqm/bzr/bzr.dev

« back to all changes in this revision

Viewing changes to bzrlib/tests/test_xml.py

  • Committer: Danny van Heumen
  • Date: 2010-03-09 21:42:11 UTC
  • mto: (4634.139.5 2.0)
  • mto: This revision was merged to the branch mainline in revision 5160.
  • Revision ID: danny@dannyvanheumen.nl-20100309214211-iqh42x6qcikgd9p3
Reverted now-useless TODO list.

Show diffs side-by-side

added added

removed removed

Lines of Context:
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
 
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
 
15
# Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
16
16
 
17
17
from cStringIO import StringIO
18
18
 
 
19
from bzrlib import (
 
20
    errors,
 
21
    inventory,
 
22
    xml6,
 
23
    xml7,
 
24
    xml8,
 
25
    serializer,
 
26
    )
19
27
from bzrlib.tests import TestCase
20
28
from bzrlib.inventory import Inventory, InventoryEntry
21
29
from bzrlib.xml4 import serializer_v4
73
81
"""
74
82
 
75
83
_committed_inv_v5 = """<inventory>
76
 
<file file_id="bar-20050901064931-73b4b1138abc9cd2" 
77
 
      name="bar" parent_id="TREE_ROOT" 
78
 
      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"/>
79
88
<directory name="subdir"
80
89
           file_id="foo-20050801201819-4139aa4a272f4250"
81
 
           parent_id="TREE_ROOT" 
 
90
           parent_id="TREE_ROOT"
82
91
           revision="mbp@foo-00"/>
83
 
<file executable="yes" file_id="bar-20050824000535-6bc48cfad47ed134" 
84
 
      name="bar" parent_id="foo-20050801201819-4139aa4a272f4250" 
85
 
      revision="mbp@foo-00"/>
 
92
<file executable="yes" file_id="bar-20050824000535-6bc48cfad47ed134"
 
93
      name="bar" parent_id="foo-20050801201819-4139aa4a272f4250"
 
94
      revision="mbp@foo-00"
 
95
      text_sha1="B" text_size="0"/>
86
96
</inventory>
87
97
"""
88
98
 
89
99
_basis_inv_v5 = """<inventory revision_id="mbp@sourcefrog.net-20050905063503-43948f59fa127d92">
90
 
<file file_id="bar-20050901064931-73b4b1138abc9cd2" 
91
 
      name="bar" parent_id="TREE_ROOT" 
 
100
<file file_id="bar-20050901064931-73b4b1138abc9cd2"
 
101
      name="bar" parent_id="TREE_ROOT"
92
102
      revision="mbp@foo-123123"/>
93
103
<directory name="subdir"
94
104
           file_id="foo-20050801201819-4139aa4a272f4250"
95
 
           parent_id="TREE_ROOT" 
 
105
           parent_id="TREE_ROOT"
96
106
           revision="mbp@foo-00"/>
97
 
<file file_id="bar-20050824000535-6bc48cfad47ed134" 
98
 
      name="bar" parent_id="foo-20050801201819-4139aa4a272f4250" 
 
107
<file file_id="bar-20050824000535-6bc48cfad47ed134"
 
108
      name="bar" parent_id="foo-20050801201819-4139aa4a272f4250"
99
109
      revision="mbp@foo-00"/>
100
110
</inventory>
101
111
"""
115
125
 
116
126
# DO NOT REFLOW THIS. Its the exact inventory we want.
117
127
_expected_inv_v5 = """<inventory format="5">
118
 
<file file_id="bar-20050901064931-73b4b1138abc9cd2" name="bar" revision="mbp@foo-123123" />
 
128
<file file_id="bar-20050901064931-73b4b1138abc9cd2" name="bar" revision="mbp@foo-123123" text_sha1="A" text_size="1" />
119
129
<directory file_id="foo-20050801201819-4139aa4a272f4250" name="subdir" revision="mbp@foo-00" />
120
 
<file executable="yes" file_id="bar-20050824000535-6bc48cfad47ed134" name="bar" parent_id="foo-20050801201819-4139aa4a272f4250" 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" />
121
131
</inventory>
122
132
"""
123
133
 
124
134
 
125
135
_expected_inv_v5_root = """<inventory file_id="f&lt;" format="5" revision_id="mother!">
126
 
<file file_id="bar-20050901064931-73b4b1138abc9cd2" name="bar" parent_id="f&lt;" revision="mbp@foo-123123" />
 
136
<file file_id="bar-20050901064931-73b4b1138abc9cd2" name="bar" parent_id="f&lt;" revision="mbp@foo-123123" text_sha1="A" text_size="1" />
127
137
<directory file_id="foo-20050801201819-4139aa4a272f4250" name="subdir" parent_id="f&lt;" revision="mbp@foo-00" />
128
 
<file executable="yes" file_id="bar-20050824000535-6bc48cfad47ed134" name="bar" parent_id="foo-20050801201819-4139aa4a272f4250" revision="mbp@foo-00" />
129
 
</inventory>
130
 
"""
131
 
 
 
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" />
 
140
</inventory>
 
141
"""
 
142
 
 
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" />
 
148
</inventory>
 
149
"""
 
150
 
 
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" />
 
157
</inventory>
 
158
"""
 
159
 
 
160
_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">
 
161
<message>- start splitting code for xml (de)serialization away from objects
 
162
  preparatory to supporting multiple formats by a single library
 
163
</message>
 
164
<parents>
 
165
<revision_ref revision_id="mbp@sourcefrog.net-20050905063503-43948f59fa127d92" />
 
166
</parents>
 
167
</revision>
 
168
"""
 
169
 
 
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" />
 
175
</inventory>
 
176
"""
132
177
 
133
178
_revision_utf8_v5 = """<revision committer="Erik B&#229;gfors &lt;erik@foo.net&gt;"
134
179
    inventory_sha1="e79c31c1deb64c163cf660fdedd476dd579ffd41"
158
203
</inventory>
159
204
"""
160
205
 
 
206
# Before revision_id was always stored as an attribute
 
207
_inventory_v5a = """<inventory format="5">
 
208
</inventory>
 
209
"""
 
210
 
 
211
# Before revision_id was always stored as an attribute
 
212
_inventory_v5b = """<inventory format="5" revision_id="a-rev-id">
 
213
</inventory>
 
214
"""
 
215
 
161
216
 
162
217
class TestSerializer(TestCase):
163
218
    """Test XML serialization"""
 
219
 
164
220
    def test_canned_inventory(self):
165
221
        """Test unpacked a canned inventory v4 file."""
166
222
        inp = StringIO(_working_inventory_v4)
229
285
        eq(ie.name, 'bar')
230
286
        eq(inv[ie.parent_id].kind, 'directory')
231
287
 
 
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)
 
292
 
 
293
    def test_unpack_inventory_5b(self):
 
294
        inv = bzrlib.xml5.serializer_v5.read_inventory_from_string(
 
295
                _inventory_v5b, revision_id='test-rev-id')
 
296
        self.assertEqual('a-rev-id', inv.root.revision)
 
297
 
232
298
    def test_repack_inventory_5(self):
233
299
        inp = StringIO(_committed_inv_v5)
234
300
        inv = bzrlib.xml5.serializer_v5.read_inventory(inp)
237
303
        self.assertEqualDiff(_expected_inv_v5, outp.getvalue())
238
304
        inv2 = bzrlib.xml5.serializer_v5.read_inventory(StringIO(outp.getvalue()))
239
305
        self.assertEqual(inv, inv2)
240
 
    
 
306
 
241
307
    def assertRoundTrips(self, xml_string):
242
308
        inp = StringIO(xml_string)
243
309
        inv = bzrlib.xml5.serializer_v5.read_inventory(inp)
244
310
        outp = StringIO()
245
311
        bzrlib.xml5.serializer_v5.write_inventory(inv, outp)
246
312
        self.assertEqualDiff(xml_string, outp.getvalue())
 
313
        lines = bzrlib.xml5.serializer_v5.write_inventory_to_lines(inv)
 
314
        outp.seek(0)
 
315
        self.assertEqual(outp.readlines(), lines)
247
316
        inv2 = bzrlib.xml5.serializer_v5.read_inventory(StringIO(outp.getvalue()))
248
317
        self.assertEqual(inv, inv2)
249
318
 
289
358
        new_rev = s_v5.read_revision_from_string(txt)
290
359
        self.assertEqual(props, new_rev.properties)
291
360
 
 
361
    def get_sample_inventory(self):
 
362
        inv = Inventory('tree-root-321', revision_id='rev_outer')
 
363
        inv.add(inventory.InventoryFile('file-id', 'file', 'tree-root-321'))
 
364
        inv.add(inventory.InventoryDirectory('dir-id', 'dir',
 
365
                                             'tree-root-321'))
 
366
        inv.add(inventory.InventoryLink('link-id', 'link', 'tree-root-321'))
 
367
        inv['tree-root-321'].revision = 'rev_outer'
 
368
        inv['dir-id'].revision = 'rev_outer'
 
369
        inv['file-id'].revision = 'rev_outer'
 
370
        inv['file-id'].text_sha1 = 'A'
 
371
        inv['file-id'].text_size = 1
 
372
        inv['link-id'].revision = 'rev_outer'
 
373
        inv['link-id'].symlink_target = 'a'
 
374
        return inv
 
375
 
 
376
    def test_roundtrip_inventory_v7(self):
 
377
        inv = self.get_sample_inventory()
 
378
        inv.add(inventory.TreeReference('nested-id', 'nested', 'tree-root-321',
 
379
                                        'rev_outer', 'rev_inner'))
 
380
        txt = xml7.serializer_v7.write_inventory_to_string(inv)
 
381
        lines = xml7.serializer_v7.write_inventory_to_lines(inv)
 
382
        self.assertEqual(bzrlib.osutils.split_lines(txt), lines)
 
383
        self.assertEqualDiff(_expected_inv_v7, txt)
 
384
        inv2 = xml7.serializer_v7.read_inventory_from_string(txt)
 
385
        self.assertEqual(5, len(inv2))
 
386
        for path, ie in inv.iter_entries():
 
387
            self.assertEqual(ie, inv2[ie.file_id])
 
388
 
 
389
    def test_roundtrip_inventory_v6(self):
 
390
        inv = self.get_sample_inventory()
 
391
        txt = xml6.serializer_v6.write_inventory_to_string(inv)
 
392
        lines = xml6.serializer_v6.write_inventory_to_lines(inv)
 
393
        self.assertEqual(bzrlib.osutils.split_lines(txt), lines)
 
394
        self.assertEqualDiff(_expected_inv_v6, txt)
 
395
        inv2 = xml6.serializer_v6.read_inventory_from_string(txt)
 
396
        self.assertEqual(4, len(inv2))
 
397
        for path, ie in inv.iter_entries():
 
398
            self.assertEqual(ie, inv2[ie.file_id])
 
399
 
 
400
    def test_wrong_format_v7(self):
 
401
        """Can't accidentally open a file with wrong serializer"""
 
402
        s_v6 = bzrlib.xml6.serializer_v6
 
403
        s_v7 = xml7.serializer_v7
 
404
        self.assertRaises(errors.UnexpectedInventoryFormat,
 
405
                          s_v7.read_inventory_from_string, _expected_inv_v5)
 
406
        self.assertRaises(errors.UnexpectedInventoryFormat,
 
407
                          s_v6.read_inventory_from_string, _expected_inv_v7)
 
408
 
 
409
    def test_tree_reference(self):
 
410
        s_v5 = bzrlib.xml5.serializer_v5
 
411
        s_v6 = bzrlib.xml6.serializer_v6
 
412
        s_v7 = xml7.serializer_v7
 
413
        inv = Inventory('tree-root-321', revision_id='rev-outer')
 
414
        inv.root.revision = 'root-rev'
 
415
        inv.add(inventory.TreeReference('nested-id', 'nested', 'tree-root-321',
 
416
                                        'rev-outer', 'rev-inner'))
 
417
        self.assertRaises(errors.UnsupportedInventoryKind,
 
418
                          s_v5.write_inventory_to_string, inv)
 
419
        self.assertRaises(errors.UnsupportedInventoryKind,
 
420
                          s_v6.write_inventory_to_string, inv)
 
421
        txt = s_v7.write_inventory_to_string(inv)
 
422
        lines = s_v7.write_inventory_to_lines(inv)
 
423
        self.assertEqual(bzrlib.osutils.split_lines(txt), lines)
 
424
        inv2 = s_v7.read_inventory_from_string(txt)
 
425
        self.assertEqual('tree-root-321', inv2['nested-id'].parent_id)
 
426
        self.assertEqual('rev-outer', inv2['nested-id'].revision)
 
427
        self.assertEqual('rev-inner', inv2['nested-id'].reference_revision)
 
428
        self.assertRaises(errors.UnsupportedInventoryKind,
 
429
                          s_v6.read_inventory_from_string,
 
430
                          txt.replace('format="7"', 'format="6"'))
 
431
        self.assertRaises(errors.UnsupportedInventoryKind,
 
432
                          s_v5.read_inventory_from_string,
 
433
                          txt.replace('format="7"', 'format="5"'))
 
434
 
 
435
    def test_roundtrip_inventory_v8(self):
 
436
        inv = self.get_sample_inventory()
 
437
        txt = xml8.serializer_v8.write_inventory_to_string(inv)
 
438
        inv2 = xml8.serializer_v8.read_inventory_from_string(txt)
 
439
        self.assertEqual(4, len(inv2))
 
440
        for path, ie in inv.iter_entries():
 
441
            self.assertEqual(ie, inv2[ie.file_id])
 
442
 
 
443
    def test_inventory_text_v8(self):
 
444
        inv = self.get_sample_inventory()
 
445
        txt = xml8.serializer_v8.write_inventory_to_string(inv)
 
446
        lines = xml8.serializer_v8.write_inventory_to_lines(inv)
 
447
        self.assertEqual(bzrlib.osutils.split_lines(txt), lines)
 
448
        self.assertEqualDiff(_expected_inv_v8, txt)
 
449
 
 
450
    def test_revision_text_v6(self):
 
451
        """Pack revision to XML v6"""
 
452
        rev = bzrlib.xml6.serializer_v6.read_revision_from_string(
 
453
            _expected_rev_v5)
 
454
        serialized = bzrlib.xml6.serializer_v6.write_revision_to_string(rev)
 
455
        self.assertEqualDiff(serialized, _expected_rev_v5)
 
456
 
 
457
    def test_revision_text_v7(self):
 
458
        """Pack revision to XML v7"""
 
459
        rev = bzrlib.xml7.serializer_v7.read_revision_from_string(
 
460
            _expected_rev_v5)
 
461
        serialized = bzrlib.xml7.serializer_v7.write_revision_to_string(rev)
 
462
        self.assertEqualDiff(serialized, _expected_rev_v5)
 
463
 
 
464
    def test_revision_text_v8(self):
 
465
        """Pack revision to XML v8"""
 
466
        rev = bzrlib.xml8.serializer_v8.read_revision_from_string(
 
467
            _expected_rev_v8)
 
468
        serialized = bzrlib.xml8.serializer_v8.write_revision_to_string(rev)
 
469
        self.assertEqualDiff(serialized, _expected_rev_v8)
 
470
 
292
471
    def test_revision_ids_are_utf8(self):
293
472
        """Parsed revision_ids should all be utf-8 strings, not unicode."""
294
473
        s_v5 = bzrlib.xml5.serializer_v5
309
488
        fid_bar1 = u'b\xe5r-01'.encode('utf8')
310
489
        fid_sub = u's\xb5bdir-01'.encode('utf8')
311
490
        fid_bar2 = u'b\xe5r-02'.encode('utf8')
312
 
        expected = [(u'', fid_root, None, None),
 
491
        expected = [(u'', fid_root, None, rev_id_2),
313
492
                    (u'b\xe5r', fid_bar1, fid_root, rev_id_1),
314
493
                    (u's\xb5bdir', fid_sub, fid_root, rev_id_1),
315
494
                    (u's\xb5bdir/b\xe5r', fid_bar2, fid_sub, rev_id_2),
338
517
    """Whitebox testing of the _encode_and_escape function."""
339
518
 
340
519
    def setUp(self):
 
520
        TestCase.setUp(self)
341
521
        # Keep the cache clear before and after the test
342
 
        bzrlib.xml5._ensure_utf8_re()
343
 
        bzrlib.xml5._clear_cache()
344
 
        self.addCleanup(bzrlib.xml5._clear_cache)
 
522
        bzrlib.xml8._ensure_utf8_re()
 
523
        bzrlib.xml8._clear_cache()
 
524
        self.addCleanup(bzrlib.xml8._clear_cache)
345
525
 
346
526
    def test_simple_ascii(self):
347
527
        # _encode_and_escape always appends a final ", because these parameters
348
528
        # are being used in xml attributes, and by returning it now, we have to
349
529
        # do fewer string operations later.
350
 
        val = bzrlib.xml5._encode_and_escape('foo bar')
 
530
        val = bzrlib.xml8._encode_and_escape('foo bar')
351
531
        self.assertEqual('foo bar"', val)
352
532
        # The second time should be cached
353
 
        val2 = bzrlib.xml5._encode_and_escape('foo bar')
 
533
        val2 = bzrlib.xml8._encode_and_escape('foo bar')
354
534
        self.assertIs(val2, val)
355
535
 
356
536
    def test_ascii_with_xml(self):
357
537
        self.assertEqual('&amp;&apos;&quot;&lt;&gt;"',
358
 
                         bzrlib.xml5._encode_and_escape('&\'"<>'))
 
538
                         bzrlib.xml8._encode_and_escape('&\'"<>'))
359
539
 
360
540
    def test_utf8_with_xml(self):
361
541
        # u'\xb5\xe5&\u062c'
362
542
        utf8_str = '\xc2\xb5\xc3\xa5&\xd8\xac'
363
543
        self.assertEqual('&#181;&#229;&amp;&#1580;"',
364
 
                         bzrlib.xml5._encode_and_escape(utf8_str))
 
544
                         bzrlib.xml8._encode_and_escape(utf8_str))
365
545
 
366
546
    def test_unicode(self):
367
547
        uni_str = u'\xb5\xe5&\u062c'
368
548
        self.assertEqual('&#181;&#229;&amp;&#1580;"',
369
 
                         bzrlib.xml5._encode_and_escape(uni_str))
 
549
                         bzrlib.xml8._encode_and_escape(uni_str))