146
_revision_utf8_v5 = """<revision committer="Erik Bågfors <erik@foo.net>"
147
inventory_sha1="e79c31c1deb64c163cf660fdedd476dd579ffd41"
148
revision_id="erik@bågfors-02"
149
timestamp="1125907235.212"
151
<message>Include µnicode characters
154
<revision_ref revision_id="erik@bågfors-01"/>
159
_inventory_utf8_v5 = """<inventory file_id="TREé_ROOT" format="5"
160
revision_id="erik@bågfors-02">
161
<file file_id="bår-01"
162
name="bår" parent_id="TREé_ROOT"
163
revision="erik@bågfors-01"/>
164
<directory name="sµbdir"
165
file_id="sµbdir-01"
166
parent_id="TREé_ROOT"
167
revision="erik@bågfors-01"/>
168
<file executable="yes" file_id="bår-02"
169
name="bår" parent_id="sµbdir-01"
170
revision="erik@bågfors-02"/>
146
175
class TestSerializer(TestCase):
147
176
"""Test XML serialization"""
148
177
def test_canned_inventory(self):
327
356
def test_revision_ids_are_utf8(self):
328
357
"""Parsed revision_ids should all be utf-8 strings, not unicode."""
329
358
s_v5 = bzrlib.xml5.serializer_v5
330
rev = s_v5.read_revision_from_string(_revision_v5)
359
rev = s_v5.read_revision_from_string(_revision_utf8_v5)
360
self.assertEqual('erik@b\xc3\xa5gfors-02', rev.revision_id)
331
361
self.assertIsInstance(rev.revision_id, str)
362
self.assertEqual(['erik@b\xc3\xa5gfors-01'], rev.parent_ids)
332
363
for parent_id in rev.parent_ids:
333
364
self.assertIsInstance(parent_id, str)
365
self.assertEqual(u'Include \xb5nicode characters\n', rev.message)
366
self.assertIsInstance(rev.message, unicode)
335
368
# ie.revision should either be None or a utf-8 revision id
336
inv = s_v5.read_inventory_from_string(_committed_inv_v5)
337
for path, ie in inv.iter_entries():
338
if ie.revision is None:
340
self.assertIsInstance(ie.revision, str)
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),
381
self.assertEqual(rev_id_2, inv.revision_id)
382
self.assertIsInstance(inv.revision_id, str)
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)
398
self.assertEqual(len(expected), len(actual))
343
401
class TestEncodeAndEscape(TestCase):