← Back to branch summary

~bzr-pqm/bzr/bzr.dev

« back to all changes in this revision

Viewing changes to bzrlib/tests/test_xml.py

  • Committer: Robert Collins
  • Date: 2006-03-01 03:26:23 UTC
  • mto: (1594.2.4 integration)
  • mto: This revision was merged to the branch mainline in revision 1596.
  • Revision ID: robertc@robertcollins.net-20060301032623-9d3c073e102f2239
Move WeaveStore down into bzrlib.store.versioned.weave.

Show diffs side-by-side

added added

removed removed

Lines of Context:
bzrlib/tests/test_xml.py
1
 
# Copyright (C) 2005 Canonical Ltd
2
 
#
 
1
# Copyright (C) 2005 by 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
19
19
from bzrlib.tests import TestCase
20
20
from bzrlib.inventory import Inventory, InventoryEntry
21
21
from bzrlib.xml4 import serializer_v4
22
 
import bzrlib.xml5
 
22
from bzrlib.xml5 import serializer_v5
23
23
 
24
24
_working_inventory_v4 = """<inventory file_id="TREE_ROOT">
25
25
<entry file_id="bar-20050901064931-73b4b1138abc9cd2" kind="file" name="bar" parent_id="TREE_ROOT" />
32
32
    inventory_id="mbp@sourcefrog.net-20050905080035-e0439293f8b6b9f9"
33
33
    inventory_sha1="e79c31c1deb64c163cf660fdedd476dd579ffd41"
34
34
    revision_id="mbp@sourcefrog.net-20050905080035-e0439293f8b6b9f9"
35
 
    timestamp="1125907235.212"
 
35
    timestamp="1125907235.211783886"
36
36
    timezone="36000">
37
37
<message>- start splitting code for xml (de)serialization away from objects
38
38
  preparatory to supporting multiple formats by a single library
46
46
_revision_v5 = """<revision committer="Martin Pool &lt;mbp@sourcefrog.net&gt;"
47
47
    inventory_sha1="e79c31c1deb64c163cf660fdedd476dd579ffd41"
48
48
    revision_id="mbp@sourcefrog.net-20050905080035-e0439293f8b6b9f9"
49
 
    timestamp="1125907235.212"
 
49
    timestamp="1125907235.211783886"
50
50
    timezone="36000">
51
51
<message>- start splitting code for xml (de)serialization away from objects
52
52
  preparatory to supporting multiple formats by a single library
57
57
</revision>
58
58
"""
59
59
 
60
 
_revision_v5_utc = """\
61
 
<revision committer="Martin Pool &lt;mbp@sourcefrog.net&gt;"
62
 
    inventory_sha1="e79c31c1deb64c163cf660fdedd476dd579ffd41"
63
 
    revision_id="mbp@sourcefrog.net-20050905080035-e0439293f8b6b9f9"
64
 
    timestamp="1125907235.212"
65
 
    timezone="0">
66
 
<message>- start splitting code for xml (de)serialization away from objects
67
 
  preparatory to supporting multiple formats by a single library
68
 
</message>
69
 
<parents>
70
 
<revision_ref revision_id="mbp@sourcefrog.net-20050905063503-43948f59fa127d92"/>
71
 
</parents>
72
 
</revision>
73
 
"""
74
 
 
75
60
_committed_inv_v5 = """<inventory>
76
61
<file file_id="bar-20050901064931-73b4b1138abc9cd2" 
77
62
      name="bar" parent_id="TREE_ROOT" 
80
65
           file_id="foo-20050801201819-4139aa4a272f4250"
81
66
           parent_id="TREE_ROOT" 
82
67
           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"/>
86
 
</inventory>
87
 
"""
88
 
 
89
 
_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" 
92
 
      revision="mbp@foo-123123"/>
93
 
<directory name="subdir"
94
 
           file_id="foo-20050801201819-4139aa4a272f4250"
95
 
           parent_id="TREE_ROOT" 
96
 
           revision="mbp@foo-00"/>
97
68
<file file_id="bar-20050824000535-6bc48cfad47ed134" 
98
69
      name="bar" parent_id="foo-20050801201819-4139aa4a272f4250" 
99
70
      revision="mbp@foo-00"/>
100
71
</inventory>
101
72
"""
102
73
 
103
 
 
104
 
# DO NOT REFLOW THIS. Its the exact revision we want.
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="1125907235.212" timezone="36000">
106
 
<message>- start splitting code for xml (de)serialization away from objects
107
 
  preparatory to supporting multiple formats by a single library
108
 
</message>
109
 
<parents>
110
 
<revision_ref revision_id="mbp@sourcefrog.net-20050905063503-43948f59fa127d92" />
111
 
</parents>
112
 
</revision>
113
 
"""
114
 
 
115
 
 
116
 
# DO NOT REFLOW THIS. Its the exact inventory we want.
117
 
_expected_inv_v5 = """<inventory format="5">
118
 
<file file_id="bar-20050901064931-73b4b1138abc9cd2" name="bar" revision="mbp@foo-123123" />
119
 
<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" />
121
 
</inventory>
122
 
"""
123
 
 
124
 
 
125
 
_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" />
127
 
<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
 
 
132
 
 
133
 
_revision_utf8_v5 = """<revision committer="Erik B&#229;gfors &lt;erik@foo.net&gt;"
134
 
    inventory_sha1="e79c31c1deb64c163cf660fdedd476dd579ffd41"
135
 
    revision_id="erik@b&#229;gfors-02"
136
 
    timestamp="1125907235.212"
137
 
    timezone="36000">
138
 
<message>Include &#181;nicode characters
139
 
</message>
140
 
<parents>
141
 
<revision_ref revision_id="erik@b&#229;gfors-01"/>
142
 
</parents>
143
 
</revision>
144
 
"""
145
 
 
146
 
_inventory_utf8_v5 = """<inventory file_id="TRE&#233;_ROOT" format="5"
147
 
                                   revision_id="erik@b&#229;gfors-02">
148
 
<file file_id="b&#229;r-01"
149
 
      name="b&#229;r" parent_id="TRE&#233;_ROOT"
150
 
      revision="erik@b&#229;gfors-01"/>
151
 
<directory name="s&#181;bdir"
152
 
           file_id="s&#181;bdir-01"
153
 
           parent_id="TRE&#233;_ROOT"
154
 
           revision="erik@b&#229;gfors-01"/>
155
 
<file executable="yes" file_id="b&#229;r-02"
156
 
      name="b&#229;r" parent_id="s&#181;bdir-01"
157
 
      revision="erik@b&#229;gfors-02"/>
158
 
</inventory>
159
 
"""
160
 
 
161
 
 
162
74
class TestSerializer(TestCase):
163
75
    """Test XML serialization"""
164
76
    def test_canned_inventory(self):
184
96
    def test_unpack_revision_5(self):
185
97
        """Test unpacking a canned revision v5"""
186
98
        inp = StringIO(_revision_v5)
187
 
        rev = bzrlib.xml5.serializer_v5.read_revision(inp)
 
99
        rev = serializer_v5.read_revision(inp)
188
100
        eq = self.assertEqual
189
101
        eq(rev.committer,
190
102
           "Martin Pool <mbp@sourcefrog.net>")
193
105
        eq(rev.parent_ids[0],
194
106
           "mbp@sourcefrog.net-20050905063503-43948f59fa127d92")
195
107
 
196
 
    def test_unpack_revision_5_utc(self):
197
 
        inp = StringIO(_revision_v5_utc)
198
 
        rev = bzrlib.xml5.serializer_v5.read_revision(inp)
199
 
        eq = self.assertEqual
200
 
        eq(rev.committer,
201
 
           "Martin Pool <mbp@sourcefrog.net>")
202
 
        eq(len(rev.parent_ids), 1)
203
 
        eq(rev.timezone, 0)
204
 
        eq(rev.parent_ids[0],
205
 
           "mbp@sourcefrog.net-20050905063503-43948f59fa127d92")
206
 
 
207
108
    def test_unpack_inventory_5(self):
208
109
        """Unpack canned new-style inventory"""
209
110
        inp = StringIO(_committed_inv_v5)
210
 
        inv = bzrlib.xml5.serializer_v5.read_inventory(inp)
211
 
        eq = self.assertEqual
212
 
        eq(len(inv), 4)
213
 
        ie = inv['bar-20050824000535-6bc48cfad47ed134']
214
 
        eq(ie.kind, 'file')
215
 
        eq(ie.revision, 'mbp@foo-00')
216
 
        eq(ie.name, 'bar')
217
 
        eq(inv[ie.parent_id].kind, 'directory')
218
 
 
219
 
    def test_unpack_basis_inventory_5(self):
220
 
        """Unpack canned new-style inventory"""
221
 
        inp = StringIO(_basis_inv_v5)
222
 
        inv = bzrlib.xml5.serializer_v5.read_inventory(inp)
223
 
        eq = self.assertEqual
224
 
        eq(len(inv), 4)
225
 
        eq(inv.revision_id, 'mbp@sourcefrog.net-20050905063503-43948f59fa127d92')
 
111
        inv = serializer_v5.read_inventory(inp)
 
112
        eq = self.assertEqual
 
113
        eq(len(inv), 4)
226
114
        ie = inv['bar-20050824000535-6bc48cfad47ed134']
227
115
        eq(ie.kind, 'file')
228
116
        eq(ie.revision, 'mbp@foo-00')
231
119
 
232
120
    def test_repack_inventory_5(self):
233
121
        inp = StringIO(_committed_inv_v5)
234
 
        inv = bzrlib.xml5.serializer_v5.read_inventory(inp)
235
 
        outp = StringIO()
236
 
        bzrlib.xml5.serializer_v5.write_inventory(inv, outp)
237
 
        self.assertEqualDiff(_expected_inv_v5, outp.getvalue())
238
 
        inv2 = bzrlib.xml5.serializer_v5.read_inventory(StringIO(outp.getvalue()))
239
 
        self.assertEqual(inv, inv2)
240
 
    
241
 
    def assertRoundTrips(self, xml_string):
242
 
        inp = StringIO(xml_string)
243
 
        inv = bzrlib.xml5.serializer_v5.read_inventory(inp)
244
 
        outp = StringIO()
245
 
        bzrlib.xml5.serializer_v5.write_inventory(inv, outp)
246
 
        self.assertEqualDiff(xml_string, outp.getvalue())
247
 
        inv2 = bzrlib.xml5.serializer_v5.read_inventory(StringIO(outp.getvalue()))
248
 
        self.assertEqual(inv, inv2)
249
 
 
250
 
    def tests_serialize_inventory_v5_with_root(self):
251
 
        self.assertRoundTrips(_expected_inv_v5_root)
252
 
 
253
 
    def check_repack_revision(self, txt):
254
 
        """Check that repacking a revision yields the same information"""
255
 
        inp = StringIO(txt)
256
 
        rev = bzrlib.xml5.serializer_v5.read_revision(inp)
257
 
        outp = StringIO()
258
 
        bzrlib.xml5.serializer_v5.write_revision(rev, outp)
259
 
        outfile_contents = outp.getvalue()
260
 
        rev2 = bzrlib.xml5.serializer_v5.read_revision(StringIO(outfile_contents))
261
 
        self.assertEqual(rev, rev2)
 
122
        inv = serializer_v5.read_inventory(inp)
 
123
        outp = StringIO()
 
124
        serializer_v5.write_inventory(inv, outp)
 
125
        inv2 = serializer_v5.read_inventory(StringIO(outp.getvalue()))
 
126
        self.assertEqual(inv, inv2)
262
127
 
263
128
    def test_repack_revision_5(self):
264
129
        """Round-trip revision to XML v5"""
265
 
        self.check_repack_revision(_revision_v5)
266
 
 
267
 
    def test_repack_revision_5_utc(self):
268
 
        self.check_repack_revision(_revision_v5_utc)
 
130
        inp = StringIO(_revision_v5)
 
131
        rev = serializer_v5.read_revision(inp)
 
132
        outp = StringIO()
 
133
        serializer_v5.write_revision(rev, outp)
 
134
        outfile_contents = outp.getvalue()
 
135
        rev2 = serializer_v5.read_revision(StringIO(outfile_contents))
 
136
        self.assertEqual(rev, rev2)
269
137
 
270
138
    def test_pack_revision_5(self):
271
139
        """Pack revision to XML v5"""
272
140
        # fixed 20051025, revisions should have final newline
273
 
        rev = bzrlib.xml5.serializer_v5.read_revision_from_string(_revision_v5)
 
141
        rev = serializer_v5.read_revision_from_string(_revision_v5)
274
142
        outp = StringIO()
275
 
        bzrlib.xml5.serializer_v5.write_revision(rev, outp)
 
143
        serializer_v5.write_revision(rev, outp)
276
144
        outfile_contents = outp.getvalue()
277
145
        self.assertEqual(outfile_contents[-1], '\n')
278
 
        self.assertEqualDiff(outfile_contents, bzrlib.xml5.serializer_v5.write_revision_to_string(rev))
279
 
        self.assertEqualDiff(outfile_contents, _expected_rev_v5)
280
 
 
281
 
    def test_empty_property_value(self):
282
 
        """Create an empty property value check that it serializes correctly"""
283
 
        s_v5 = bzrlib.xml5.serializer_v5
284
 
        rev = s_v5.read_revision_from_string(_revision_v5)
285
 
        outp = StringIO()
286
 
        props = {'empty':'', 'one':'one'}
287
 
        rev.properties = props
288
 
        txt = s_v5.write_revision_to_string(rev)
289
 
        new_rev = s_v5.read_revision_from_string(txt)
290
 
        self.assertEqual(props, new_rev.properties)
291
 
 
292
 
    def test_revision_ids_are_utf8(self):
293
 
        """Parsed revision_ids should all be utf-8 strings, not unicode."""
294
 
        s_v5 = bzrlib.xml5.serializer_v5
295
 
        rev = s_v5.read_revision_from_string(_revision_utf8_v5)
296
 
        self.assertEqual('erik@b\xc3\xa5gfors-02', rev.revision_id)
297
 
        self.assertIsInstance(rev.revision_id, str)
298
 
        self.assertEqual(['erik@b\xc3\xa5gfors-01'], rev.parent_ids)
299
 
        for parent_id in rev.parent_ids:
300
 
            self.assertIsInstance(parent_id, str)
301
 
        self.assertEqual(u'Include \xb5nicode characters\n', rev.message)
302
 
        self.assertIsInstance(rev.message, unicode)
303
 
 
304
 
        # ie.revision should either be None or a utf-8 revision id
305
 
        inv = s_v5.read_inventory_from_string(_inventory_utf8_v5)
306
 
        rev_id_1 = u'erik@b\xe5gfors-01'.encode('utf8')
307
 
        rev_id_2 = u'erik@b\xe5gfors-02'.encode('utf8')
308
 
        fid_root = u'TRE\xe9_ROOT'.encode('utf8')
309
 
        fid_bar1 = u'b\xe5r-01'.encode('utf8')
310
 
        fid_sub = u's\xb5bdir-01'.encode('utf8')
311
 
        fid_bar2 = u'b\xe5r-02'.encode('utf8')
312
 
        expected = [(u'', fid_root, None, None),
313
 
                    (u'b\xe5r', fid_bar1, fid_root, rev_id_1),
314
 
                    (u's\xb5bdir', fid_sub, fid_root, rev_id_1),
315
 
                    (u's\xb5bdir/b\xe5r', fid_bar2, fid_sub, rev_id_2),
316
 
                   ]
317
 
        self.assertEqual(rev_id_2, inv.revision_id)
318
 
        self.assertIsInstance(inv.revision_id, str)
319
 
 
320
 
        actual = list(inv.iter_entries_by_dir())
321
 
        for ((exp_path, exp_file_id, exp_parent_id, exp_rev_id),
322
 
             (act_path, act_ie)) in zip(expected, actual):
323
 
            self.assertEqual(exp_path, act_path)
324
 
            self.assertIsInstance(act_path, unicode)
325
 
            self.assertEqual(exp_file_id, act_ie.file_id)
326
 
            self.assertIsInstance(act_ie.file_id, str)
327
 
            self.assertEqual(exp_parent_id, act_ie.parent_id)
328
 
            if exp_parent_id is not None:
329
 
                self.assertIsInstance(act_ie.parent_id, str)
330
 
            self.assertEqual(exp_rev_id, act_ie.revision)
331
 
            if exp_rev_id is not None:
332
 
                self.assertIsInstance(act_ie.revision, str)
333
 
 
334
 
        self.assertEqual(len(expected), len(actual))
335
 
 
336
 
 
337
 
class TestEncodeAndEscape(TestCase):
338
 
    """Whitebox testing of the _encode_and_escape function."""
339
 
 
340
 
    def setUp(self):
341
 
        # 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)
345
 
 
346
 
    def test_simple_ascii(self):
347
 
        # _encode_and_escape always appends a final ", because these parameters
348
 
        # are being used in xml attributes, and by returning it now, we have to
349
 
        # do fewer string operations later.
350
 
        val = bzrlib.xml5._encode_and_escape('foo bar')
351
 
        self.assertEqual('foo bar"', val)
352
 
        # The second time should be cached
353
 
        val2 = bzrlib.xml5._encode_and_escape('foo bar')
354
 
        self.assertIs(val2, val)
355
 
 
356
 
    def test_ascii_with_xml(self):
357
 
        self.assertEqual('&amp;&apos;&quot;&lt;&gt;"',
358
 
                         bzrlib.xml5._encode_and_escape('&\'"<>'))
359
 
 
360
 
    def test_utf8_with_xml(self):
361
 
        # u'\xb5\xe5&\u062c'
362
 
        utf8_str = '\xc2\xb5\xc3\xa5&\xd8\xac'
363
 
        self.assertEqual('&#181;&#229;&amp;&#1580;"',
364
 
                         bzrlib.xml5._encode_and_escape(utf8_str))
365
 
 
366
 
    def test_unicode(self):
367
 
        uni_str = u'\xb5\xe5&\u062c'
368
 
        self.assertEqual('&#181;&#229;&amp;&#1580;"',
369
 
                         bzrlib.xml5._encode_and_escape(uni_str))
 
146
        self.assertEqualDiff(outfile_contents, serializer_v5.write_revision_to_string(rev))