~bzr-pqm/bzr/bzr.dev

« back to all changes in this revision

Viewing changes to bzrlib/tests/test_xml.py

  • Committer: John Arbash Meinel
  • Date: 2006-10-11 00:23:23 UTC
  • mfrom: (2070 +trunk)
  • mto: This revision was merged to the branch mainline in revision 2071.
  • Revision ID: john@arbash-meinel.com-20061011002323-82ba88c293d7caff
[merge] bzr.dev 2070

Show diffs side-by-side

added added

removed removed

Lines of Context:
1
 
# Copyright (C) 2005 Canonical Ltd
 
1
# Copyright (C) 2005 by 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
 
    )
24
19
from bzrlib.tests import TestCase
25
20
from bzrlib.inventory import Inventory, InventoryEntry
26
21
from bzrlib.xml4 import serializer_v4
37
32
    inventory_id="mbp@sourcefrog.net-20050905080035-e0439293f8b6b9f9"
38
33
    inventory_sha1="e79c31c1deb64c163cf660fdedd476dd579ffd41"
39
34
    revision_id="mbp@sourcefrog.net-20050905080035-e0439293f8b6b9f9"
40
 
    timestamp="1125907235.212"
 
35
    timestamp="1125907235.211783886"
41
36
    timezone="36000">
42
37
<message>- start splitting code for xml (de)serialization away from objects
43
38
  preparatory to supporting multiple formats by a single library
51
46
_revision_v5 = """<revision committer="Martin Pool &lt;mbp@sourcefrog.net&gt;"
52
47
    inventory_sha1="e79c31c1deb64c163cf660fdedd476dd579ffd41"
53
48
    revision_id="mbp@sourcefrog.net-20050905080035-e0439293f8b6b9f9"
54
 
    timestamp="1125907235.212"
 
49
    timestamp="1125907235.211783886"
55
50
    timezone="36000">
56
51
<message>- start splitting code for xml (de)serialization away from objects
57
52
  preparatory to supporting multiple formats by a single library
66
61
<revision committer="Martin Pool &lt;mbp@sourcefrog.net&gt;"
67
62
    inventory_sha1="e79c31c1deb64c163cf660fdedd476dd579ffd41"
68
63
    revision_id="mbp@sourcefrog.net-20050905080035-e0439293f8b6b9f9"
69
 
    timestamp="1125907235.212"
 
64
    timestamp="1125907235.211783886"
70
65
    timezone="0">
71
66
<message>- start splitting code for xml (de)serialization away from objects
72
67
  preparatory to supporting multiple formats by a single library
107
102
 
108
103
 
109
104
# DO NOT REFLOW THIS. Its the exact revision we want.
110
 
_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">
 
105
_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="1125907235.211783886" timezone="36000">
111
106
<message>- start splitting code for xml (de)serialization away from objects
112
107
  preparatory to supporting multiple formats by a single library
113
108
</message>
134
129
</inventory>
135
130
"""
136
131
 
137
 
_expected_inv_v7 = """<inventory format="7" revision_id="rev_outer">
138
 
<directory file_id="tree-root-321" name="" revision="rev_outer" />
139
 
<directory file_id="dir-id" name="dir" parent_id="tree-root-321" revision="rev_outer" />
140
 
<file file_id="file-id" name="file" parent_id="tree-root-321" revision="rev_outer" />
141
 
<symlink file_id="link-id" name="link" parent_id="tree-root-321" revision="rev_outer" />
142
 
<tree-reference file_id="nested-id" name="nested" parent_id="tree-root-321" revision="rev_outer" reference_revision="rev_inner" />
143
 
</inventory>
144
 
"""
145
 
 
146
 
_revision_utf8_v5 = """<revision committer="Erik B&#229;gfors &lt;erik@foo.net&gt;"
147
 
    inventory_sha1="e79c31c1deb64c163cf660fdedd476dd579ffd41"
148
 
    revision_id="erik@b&#229;gfors-02"
149
 
    timestamp="1125907235.212"
150
 
    timezone="36000">
151
 
<message>Include &#181;nicode characters
152
 
</message>
153
 
<parents>
154
 
<revision_ref revision_id="erik@b&#229;gfors-01"/>
155
 
</parents>
156
 
</revision>
157
 
"""
158
 
 
159
 
_inventory_utf8_v5 = """<inventory file_id="TRE&#233;_ROOT" format="5"
160
 
                                   revision_id="erik@b&#229;gfors-02">
161
 
<file file_id="b&#229;r-01"
162
 
      name="b&#229;r" parent_id="TRE&#233;_ROOT"
163
 
      revision="erik@b&#229;gfors-01"/>
164
 
<directory name="s&#181;bdir"
165
 
           file_id="s&#181;bdir-01"
166
 
           parent_id="TRE&#233;_ROOT"
167
 
           revision="erik@b&#229;gfors-01"/>
168
 
<file executable="yes" file_id="b&#229;r-02"
169
 
      name="b&#229;r" parent_id="s&#181;bdir-01"
170
 
      revision="erik@b&#229;gfors-02"/>
171
 
</inventory>
172
 
"""
173
 
 
174
132
 
175
133
class TestSerializer(TestCase):
176
134
    """Test XML serialization"""
301
259
        txt = s_v5.write_revision_to_string(rev)
302
260
        new_rev = s_v5.read_revision_from_string(txt)
303
261
        self.assertEqual(props, new_rev.properties)
304
 
 
305
 
    def test_roundtrip_inventory_v7(self):
306
 
        inv = Inventory('tree-root-321', revision_id='rev_outer')
307
 
        inv.add(inventory.TreeReference('nested-id', 'nested', 'tree-root-321',
308
 
                                        'rev_outer', 'rev_inner'))
309
 
        inv.add(inventory.InventoryFile('file-id', 'file', 'tree-root-321'))
310
 
        inv.add(inventory.InventoryDirectory('dir-id', 'dir', 
311
 
                                             'tree-root-321'))
312
 
        inv.add(inventory.InventoryLink('link-id', 'link', 'tree-root-321'))
313
 
        inv['tree-root-321'].revision = 'rev_outer'
314
 
        inv['dir-id'].revision = 'rev_outer'
315
 
        inv['file-id'].revision = 'rev_outer'
316
 
        inv['link-id'].revision = 'rev_outer'
317
 
        txt = xml7.serializer_v7.write_inventory_to_string(inv)
318
 
        self.assertEqualDiff(_expected_inv_v7, txt)
319
 
        inv2 = xml7.serializer_v7.read_inventory_from_string(txt)
320
 
        self.assertEqual(5, len(inv2))
321
 
        for path, ie in inv.iter_entries():
322
 
            self.assertEqual(ie, inv2[ie.file_id])
323
 
 
324
 
    def test_wrong_format_v7(self):
325
 
        """Can't accidentally open a file with wrong serializer"""
326
 
        s_v6 = bzrlib.xml6.serializer_v6
327
 
        s_v7 = xml7.serializer_v7
328
 
        self.assertRaises(errors.UnexpectedInventoryFormat, 
329
 
                          s_v7.read_inventory_from_string, _expected_inv_v5)
330
 
        self.assertRaises(errors.UnexpectedInventoryFormat, 
331
 
                          s_v6.read_inventory_from_string, _expected_inv_v7)
332
 
 
333
 
    def test_tree_reference(self):
334
 
        s_v5 = bzrlib.xml5.serializer_v5
335
 
        s_v6 = bzrlib.xml6.serializer_v6
336
 
        s_v7 = xml7.serializer_v7
337
 
        inv = Inventory('tree-root-321')
338
 
        inv.add(inventory.TreeReference('nested-id', 'nested', 'tree-root-321',
339
 
                                        'rev-outer', 'rev-inner'))
340
 
        self.assertRaises(errors.UnsupportedInventoryKind, 
341
 
                          s_v5.write_inventory_to_string, inv)
342
 
        self.assertRaises(errors.UnsupportedInventoryKind, 
343
 
                          s_v6.write_inventory_to_string, inv)
344
 
        txt = s_v7.write_inventory_to_string(inv)
345
 
        inv2 = s_v7.read_inventory_from_string(txt)
346
 
        self.assertEqual('tree-root-321', inv2['nested-id'].parent_id)
347
 
        self.assertEqual('rev-outer', inv2['nested-id'].revision)
348
 
        self.assertEqual('rev-inner', inv2['nested-id'].reference_revision)
349
 
        self.assertRaises(errors.UnsupportedInventoryKind, 
350
 
                          s_v6.read_inventory_from_string,
351
 
                          txt.replace('format="7"', 'format="6"'))
352
 
        self.assertRaises(errors.UnsupportedInventoryKind, 
353
 
                          s_v5.read_inventory_from_string,
354
 
                          txt.replace('format="7"', 'format="5"'))
355
 
 
356
 
    def test_revision_ids_are_utf8(self):
357
 
        """Parsed revision_ids should all be utf-8 strings, not unicode."""
358
 
        s_v5 = bzrlib.xml5.serializer_v5
359
 
        rev = s_v5.read_revision_from_string(_revision_utf8_v5)
360
 
        self.assertEqual('erik@b\xc3\xa5gfors-02', rev.revision_id)
361
 
        self.assertIsInstance(rev.revision_id, str)
362
 
        self.assertEqual(['erik@b\xc3\xa5gfors-01'], rev.parent_ids)
363
 
        for parent_id in rev.parent_ids:
364
 
            self.assertIsInstance(parent_id, str)
365
 
        self.assertEqual(u'Include \xb5nicode characters\n', rev.message)
366
 
        self.assertIsInstance(rev.message, unicode)
367
 
 
368
 
        # ie.revision should either be None or a utf-8 revision id
369
 
        inv = s_v5.read_inventory_from_string(_inventory_utf8_v5)
370
 
        rev_id_1 = u'erik@b\xe5gfors-01'.encode('utf8')
371
 
        rev_id_2 = u'erik@b\xe5gfors-02'.encode('utf8')
372
 
        fid_root = u'TRE\xe9_ROOT'.encode('utf8')
373
 
        fid_bar1 = u'b\xe5r-01'.encode('utf8')
374
 
        fid_sub = u's\xb5bdir-01'.encode('utf8')
375
 
        fid_bar2 = u'b\xe5r-02'.encode('utf8')
376
 
        expected = [(u'', fid_root, None, None),
377
 
                    (u'b\xe5r', fid_bar1, fid_root, rev_id_1),
378
 
                    (u's\xb5bdir', fid_sub, fid_root, rev_id_1),
379
 
                    (u's\xb5bdir/b\xe5r', fid_bar2, fid_sub, rev_id_2),
380
 
                   ]
381
 
        self.assertEqual(rev_id_2, inv.revision_id)
382
 
        self.assertIsInstance(inv.revision_id, str)
383
 
 
384
 
        actual = list(inv.iter_entries_by_dir())
385
 
        for ((exp_path, exp_file_id, exp_parent_id, exp_rev_id),
386
 
             (act_path, act_ie)) in zip(expected, actual):
387
 
            self.assertEqual(exp_path, act_path)
388
 
            self.assertIsInstance(act_path, unicode)
389
 
            self.assertEqual(exp_file_id, act_ie.file_id)
390
 
            self.assertIsInstance(act_ie.file_id, str)
391
 
            self.assertEqual(exp_parent_id, act_ie.parent_id)
392
 
            if exp_parent_id is not None:
393
 
                self.assertIsInstance(act_ie.parent_id, str)
394
 
            self.assertEqual(exp_rev_id, act_ie.revision)
395
 
            if exp_rev_id is not None:
396
 
                self.assertIsInstance(act_ie.revision, str)
397
 
 
398
 
        self.assertEqual(len(expected), len(actual))
399
 
 
400
 
 
401
 
class TestEncodeAndEscape(TestCase):
402
 
    """Whitebox testing of the _encode_and_escape function."""
403
 
 
404
 
    def setUp(self):
405
 
        # Keep the cache clear before and after the test
406
 
        bzrlib.xml5._ensure_utf8_re()
407
 
        bzrlib.xml5._clear_cache()
408
 
        self.addCleanup(bzrlib.xml5._clear_cache)
409
 
 
410
 
    def test_simple_ascii(self):
411
 
        # _encode_and_escape always appends a final ", because these parameters
412
 
        # are being used in xml attributes, and by returning it now, we have to
413
 
        # do fewer string operations later.
414
 
        val = bzrlib.xml5._encode_and_escape('foo bar')
415
 
        self.assertEqual('foo bar"', val)
416
 
        # The second time should be cached
417
 
        val2 = bzrlib.xml5._encode_and_escape('foo bar')
418
 
        self.assertIs(val2, val)
419
 
 
420
 
    def test_ascii_with_xml(self):
421
 
        self.assertEqual('&amp;&apos;&quot;&lt;&gt;"',
422
 
                         bzrlib.xml5._encode_and_escape('&\'"<>'))
423
 
 
424
 
    def test_utf8_with_xml(self):
425
 
        # u'\xb5\xe5&\u062c'
426
 
        utf8_str = '\xc2\xb5\xc3\xa5&\xd8\xac'
427
 
        self.assertEqual('&#181;&#229;&amp;&#1580;"',
428
 
                         bzrlib.xml5._encode_and_escape(utf8_str))
429
 
 
430
 
    def test_unicode(self):
431
 
        uni_str = u'\xb5\xe5&\u062c'
432
 
        self.assertEqual('&#181;&#229;&amp;&#1580;"',
433
 
                         bzrlib.xml5._encode_and_escape(uni_str))