1
# Copyright (C) 2005, 2006 Canonical Ltd
3
# This program is free software; you can redistribute it and/or modify
4
# it under the terms of the GNU General Public License as published by
5
# the Free Software Foundation; either version 2 of the License, or
6
# (at your option) any later version.
8
# This program is distributed in the hope that it will be useful,
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
# GNU General Public License for more details.
13
# You should have received a copy of the GNU General Public License
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
18
from cStringIO import StringIO
21
from tempfile import TemporaryFile
23
from bzrlib.diff import (
33
from bzrlib.errors import BinaryFile, NoDiff
34
import bzrlib.osutils as osutils
35
import bzrlib.patiencediff
36
import bzrlib._patiencediff_py
37
from bzrlib.tests import (Feature, TestCase, TestCaseWithTransport,
38
TestCaseInTempDir, TestSkipped)
41
class _CompiledPatienceDiffFeature(Feature):
45
import bzrlib._patiencediff_c
50
def feature_name(self):
51
return 'bzrlib._patiencediff_c'
53
CompiledPatienceDiffFeature = _CompiledPatienceDiffFeature()
56
class _UnicodeFilename(Feature):
57
"""Does the filesystem support Unicode filenames?"""
62
except UnicodeEncodeError:
64
except (IOError, OSError):
65
# The filesystem allows the Unicode filename but the file doesn't
69
# The filesystem allows the Unicode filename and the file exists,
73
UnicodeFilename = _UnicodeFilename()
76
class TestUnicodeFilename(TestCase):
78
def test_probe_passes(self):
79
"""UnicodeFilename._probe passes."""
80
# We can't test much more than that because the behaviour depends
82
UnicodeFilename._probe()
85
def udiff_lines(old, new, allow_binary=False):
87
internal_diff('old', old, 'new', new, output, allow_binary)
89
return output.readlines()
92
def external_udiff_lines(old, new, use_stringio=False):
94
# StringIO has no fileno, so it tests a different codepath
97
output = TemporaryFile()
99
external_diff('old', old, 'new', new, output, diff_opts=['-u'])
101
raise TestSkipped('external "diff" not present to test')
103
lines = output.readlines()
108
class TestDiff(TestCase):
110
def test_add_nl(self):
111
"""diff generates a valid diff for patches that add a newline"""
112
lines = udiff_lines(['boo'], ['boo\n'])
113
self.check_patch(lines)
114
self.assertEquals(lines[4], '\\ No newline at end of file\n')
115
## "expected no-nl, got %r" % lines[4]
117
def test_add_nl_2(self):
118
"""diff generates a valid diff for patches that change last line and
121
lines = udiff_lines(['boo'], ['goo\n'])
122
self.check_patch(lines)
123
self.assertEquals(lines[4], '\\ No newline at end of file\n')
124
## "expected no-nl, got %r" % lines[4]
126
def test_remove_nl(self):
127
"""diff generates a valid diff for patches that change last line and
130
lines = udiff_lines(['boo\n'], ['boo'])
131
self.check_patch(lines)
132
self.assertEquals(lines[5], '\\ No newline at end of file\n')
133
## "expected no-nl, got %r" % lines[5]
135
def check_patch(self, lines):
136
self.assert_(len(lines) > 1)
137
## "Not enough lines for a file header for patch:\n%s" % "".join(lines)
138
self.assert_(lines[0].startswith ('---'))
139
## 'No orig line for patch:\n%s' % "".join(lines)
140
self.assert_(lines[1].startswith ('+++'))
141
## 'No mod line for patch:\n%s' % "".join(lines)
142
self.assert_(len(lines) > 2)
143
## "No hunks for patch:\n%s" % "".join(lines)
144
self.assert_(lines[2].startswith('@@'))
145
## "No hunk header for patch:\n%s" % "".join(lines)
146
self.assert_('@@' in lines[2][2:])
147
## "Unterminated hunk header for patch:\n%s" % "".join(lines)
149
def test_binary_lines(self):
150
self.assertRaises(BinaryFile, udiff_lines, [1023 * 'a' + '\x00'], [])
151
self.assertRaises(BinaryFile, udiff_lines, [], [1023 * 'a' + '\x00'])
152
udiff_lines([1023 * 'a' + '\x00'], [], allow_binary=True)
153
udiff_lines([], [1023 * 'a' + '\x00'], allow_binary=True)
155
def test_external_diff(self):
156
lines = external_udiff_lines(['boo\n'], ['goo\n'])
157
self.check_patch(lines)
158
self.assertEqual('\n', lines[-1])
160
def test_external_diff_no_fileno(self):
161
# Make sure that we can handle not having a fileno, even
162
# if the diff is large
163
lines = external_udiff_lines(['boo\n']*10000,
166
self.check_patch(lines)
168
def test_external_diff_binary_lang_c(self):
170
for lang in ('LANG', 'LC_ALL', 'LANGUAGE'):
171
old_env[lang] = osutils.set_or_unset_env(lang, 'C')
173
lines = external_udiff_lines(['\x00foobar\n'], ['foo\x00bar\n'])
174
# Older versions of diffutils say "Binary files", newer
175
# versions just say "Files".
176
self.assertContainsRe(lines[0],
177
'(Binary f|F)iles old and new differ\n')
178
self.assertEquals(lines[1:], ['\n'])
180
for lang, old_val in old_env.iteritems():
181
osutils.set_or_unset_env(lang, old_val)
183
def test_no_external_diff(self):
184
"""Check that NoDiff is raised when diff is not available"""
185
# Use os.environ['PATH'] to make sure no 'diff' command is available
186
orig_path = os.environ['PATH']
188
os.environ['PATH'] = ''
189
self.assertRaises(NoDiff, external_diff,
190
'old', ['boo\n'], 'new', ['goo\n'],
191
StringIO(), diff_opts=['-u'])
193
os.environ['PATH'] = orig_path
195
def test_internal_diff_default(self):
196
# Default internal diff encoding is utf8
198
internal_diff(u'old_\xb5', ['old_text\n'],
199
u'new_\xe5', ['new_text\n'], output)
200
lines = output.getvalue().splitlines(True)
201
self.check_patch(lines)
202
self.assertEquals(['--- old_\xc2\xb5\n',
203
'+++ new_\xc3\xa5\n',
211
def test_internal_diff_utf8(self):
213
internal_diff(u'old_\xb5', ['old_text\n'],
214
u'new_\xe5', ['new_text\n'], output,
215
path_encoding='utf8')
216
lines = output.getvalue().splitlines(True)
217
self.check_patch(lines)
218
self.assertEquals(['--- old_\xc2\xb5\n',
219
'+++ new_\xc3\xa5\n',
227
def test_internal_diff_iso_8859_1(self):
229
internal_diff(u'old_\xb5', ['old_text\n'],
230
u'new_\xe5', ['new_text\n'], output,
231
path_encoding='iso-8859-1')
232
lines = output.getvalue().splitlines(True)
233
self.check_patch(lines)
234
self.assertEquals(['--- old_\xb5\n',
243
def test_internal_diff_no_content(self):
245
internal_diff(u'old', [], u'new', [], output)
246
self.assertEqual('', output.getvalue())
248
def test_internal_diff_no_changes(self):
250
internal_diff(u'old', ['text\n', 'contents\n'],
251
u'new', ['text\n', 'contents\n'],
253
self.assertEqual('', output.getvalue())
255
def test_internal_diff_returns_bytes(self):
257
output = StringIO.StringIO()
258
internal_diff(u'old_\xb5', ['old_text\n'],
259
u'new_\xe5', ['new_text\n'], output)
260
self.failUnless(isinstance(output.getvalue(), str),
261
'internal_diff should return bytestrings')
264
class TestDiffFiles(TestCaseInTempDir):
266
def test_external_diff_binary(self):
267
"""The output when using external diff should use diff's i18n error"""
268
# Make sure external_diff doesn't fail in the current LANG
269
lines = external_udiff_lines(['\x00foobar\n'], ['foo\x00bar\n'])
271
cmd = ['diff', '-u', '--binary', 'old', 'new']
272
open('old', 'wb').write('\x00foobar\n')
273
open('new', 'wb').write('foo\x00bar\n')
274
pipe = subprocess.Popen(cmd, stdout=subprocess.PIPE,
275
stdin=subprocess.PIPE)
276
out, err = pipe.communicate()
277
# Diff returns '2' on Binary files.
278
self.assertEqual(2, pipe.returncode)
279
# We should output whatever diff tells us, plus a trailing newline
280
self.assertEqual(out.splitlines(True) + ['\n'], lines)
283
class TestShowDiffTreesHelper(TestCaseWithTransport):
284
"""Has a helper for running show_diff_trees"""
286
def get_diff(self, tree1, tree2, specific_files=None, working_tree=None):
288
if working_tree is not None:
289
extra_trees = (working_tree,)
292
show_diff_trees(tree1, tree2, output, specific_files=specific_files,
293
extra_trees=extra_trees, old_label='old/',
295
return output.getvalue()
298
class TestDiffDates(TestShowDiffTreesHelper):
301
super(TestDiffDates, self).setUp()
302
self.wt = self.make_branch_and_tree('.')
303
self.b = self.wt.branch
304
self.build_tree_contents([
305
('file1', 'file1 contents at rev 1\n'),
306
('file2', 'file2 contents at rev 1\n')
308
self.wt.add(['file1', 'file2'])
310
message='Revision 1',
311
timestamp=1143849600, # 2006-04-01 00:00:00 UTC
314
self.build_tree_contents([('file1', 'file1 contents at rev 2\n')])
316
message='Revision 2',
317
timestamp=1143936000, # 2006-04-02 00:00:00 UTC
320
self.build_tree_contents([('file2', 'file2 contents at rev 3\n')])
322
message='Revision 3',
323
timestamp=1144022400, # 2006-04-03 00:00:00 UTC
326
self.wt.remove(['file2'])
328
message='Revision 4',
329
timestamp=1144108800, # 2006-04-04 00:00:00 UTC
332
self.build_tree_contents([
333
('file1', 'file1 contents in working tree\n')
335
# set the date stamps for files in the working tree to known values
336
os.utime('file1', (1144195200, 1144195200)) # 2006-04-05 00:00:00 UTC
338
def test_diff_rev_tree_working_tree(self):
339
output = self.get_diff(self.wt.basis_tree(), self.wt)
340
# note that the date for old/file1 is from rev 2 rather than from
341
# the basis revision (rev 4)
342
self.assertEqualDiff(output, '''\
343
=== modified file 'file1'
344
--- old/file1\t2006-04-02 00:00:00 +0000
345
+++ new/file1\t2006-04-05 00:00:00 +0000
347
-file1 contents at rev 2
348
+file1 contents in working tree
352
def test_diff_rev_tree_rev_tree(self):
353
tree1 = self.b.repository.revision_tree('rev-2')
354
tree2 = self.b.repository.revision_tree('rev-3')
355
output = self.get_diff(tree1, tree2)
356
self.assertEqualDiff(output, '''\
357
=== modified file 'file2'
358
--- old/file2\t2006-04-01 00:00:00 +0000
359
+++ new/file2\t2006-04-03 00:00:00 +0000
361
-file2 contents at rev 1
362
+file2 contents at rev 3
366
def test_diff_add_files(self):
367
tree1 = self.b.repository.revision_tree(None)
368
tree2 = self.b.repository.revision_tree('rev-1')
369
output = self.get_diff(tree1, tree2)
370
# the files have the epoch time stamp for the tree in which
372
self.assertEqualDiff(output, '''\
373
=== added file 'file1'
374
--- old/file1\t1970-01-01 00:00:00 +0000
375
+++ new/file1\t2006-04-01 00:00:00 +0000
377
+file1 contents at rev 1
379
=== added file 'file2'
380
--- old/file2\t1970-01-01 00:00:00 +0000
381
+++ new/file2\t2006-04-01 00:00:00 +0000
383
+file2 contents at rev 1
387
def test_diff_remove_files(self):
388
tree1 = self.b.repository.revision_tree('rev-3')
389
tree2 = self.b.repository.revision_tree('rev-4')
390
output = self.get_diff(tree1, tree2)
391
# the file has the epoch time stamp for the tree in which
393
self.assertEqualDiff(output, '''\
394
=== removed file 'file2'
395
--- old/file2\t2006-04-03 00:00:00 +0000
396
+++ new/file2\t1970-01-01 00:00:00 +0000
398
-file2 contents at rev 3
402
def test_show_diff_specified(self):
403
"""A working tree filename can be used to identify a file"""
404
self.wt.rename_one('file1', 'file1b')
405
old_tree = self.b.repository.revision_tree('rev-1')
406
new_tree = self.b.repository.revision_tree('rev-4')
407
out = self.get_diff(old_tree, new_tree, specific_files=['file1b'],
408
working_tree=self.wt)
409
self.assertContainsRe(out, 'file1\t')
411
def test_recursive_diff(self):
412
"""Children of directories are matched"""
415
self.wt.add(['dir1', 'dir2'])
416
self.wt.rename_one('file1', 'dir1/file1')
417
old_tree = self.b.repository.revision_tree('rev-1')
418
new_tree = self.b.repository.revision_tree('rev-4')
419
out = self.get_diff(old_tree, new_tree, specific_files=['dir1'],
420
working_tree=self.wt)
421
self.assertContainsRe(out, 'file1\t')
422
out = self.get_diff(old_tree, new_tree, specific_files=['dir2'],
423
working_tree=self.wt)
424
self.assertNotContainsRe(out, 'file1\t')
428
class TestShowDiffTrees(TestShowDiffTreesHelper):
429
"""Direct tests for show_diff_trees"""
431
def test_modified_file(self):
432
"""Test when a file is modified."""
433
tree = self.make_branch_and_tree('tree')
434
self.build_tree_contents([('tree/file', 'contents\n')])
435
tree.add(['file'], ['file-id'])
436
tree.commit('one', rev_id='rev-1')
438
self.build_tree_contents([('tree/file', 'new contents\n')])
439
diff = self.get_diff(tree.basis_tree(), tree)
440
self.assertContainsRe(diff, "=== modified file 'file'\n")
441
self.assertContainsRe(diff, '--- old/file\t')
442
self.assertContainsRe(diff, '\\+\\+\\+ new/file\t')
443
self.assertContainsRe(diff, '-contents\n'
446
def test_modified_file_in_renamed_dir(self):
447
"""Test when a file is modified in a renamed directory."""
448
tree = self.make_branch_and_tree('tree')
449
self.build_tree(['tree/dir/'])
450
self.build_tree_contents([('tree/dir/file', 'contents\n')])
451
tree.add(['dir', 'dir/file'], ['dir-id', 'file-id'])
452
tree.commit('one', rev_id='rev-1')
454
tree.rename_one('dir', 'other')
455
self.build_tree_contents([('tree/other/file', 'new contents\n')])
456
diff = self.get_diff(tree.basis_tree(), tree)
457
self.assertContainsRe(diff, "=== renamed directory 'dir' => 'other'\n")
458
self.assertContainsRe(diff, "=== modified file 'other/file'\n")
459
# XXX: This is technically incorrect, because it used to be at another
460
# location. What to do?
461
self.assertContainsRe(diff, '--- old/dir/file\t')
462
self.assertContainsRe(diff, '\\+\\+\\+ new/other/file\t')
463
self.assertContainsRe(diff, '-contents\n'
466
def test_renamed_directory(self):
467
"""Test when only a directory is only renamed."""
468
tree = self.make_branch_and_tree('tree')
469
self.build_tree(['tree/dir/'])
470
self.build_tree_contents([('tree/dir/file', 'contents\n')])
471
tree.add(['dir', 'dir/file'], ['dir-id', 'file-id'])
472
tree.commit('one', rev_id='rev-1')
474
tree.rename_one('dir', 'newdir')
475
diff = self.get_diff(tree.basis_tree(), tree)
476
# Renaming a directory should be a single "you renamed this dir" even
477
# when there are files inside.
478
self.assertEqual("=== renamed directory 'dir' => 'newdir'\n", diff)
480
def test_renamed_file(self):
481
"""Test when a file is only renamed."""
482
tree = self.make_branch_and_tree('tree')
483
self.build_tree_contents([('tree/file', 'contents\n')])
484
tree.add(['file'], ['file-id'])
485
tree.commit('one', rev_id='rev-1')
487
tree.rename_one('file', 'newname')
488
diff = self.get_diff(tree.basis_tree(), tree)
489
self.assertContainsRe(diff, "=== renamed file 'file' => 'newname'\n")
490
# We shouldn't have a --- or +++ line, because there is no content
492
self.assertNotContainsRe(diff, '---')
494
def test_renamed_and_modified_file(self):
495
"""Test when a file is only renamed."""
496
tree = self.make_branch_and_tree('tree')
497
self.build_tree_contents([('tree/file', 'contents\n')])
498
tree.add(['file'], ['file-id'])
499
tree.commit('one', rev_id='rev-1')
501
tree.rename_one('file', 'newname')
502
self.build_tree_contents([('tree/newname', 'new contents\n')])
503
diff = self.get_diff(tree.basis_tree(), tree)
504
self.assertContainsRe(diff, "=== renamed file 'file' => 'newname'\n")
505
self.assertContainsRe(diff, '--- old/file\t')
506
self.assertContainsRe(diff, '\\+\\+\\+ new/newname\t')
507
self.assertContainsRe(diff, '-contents\n'
510
def test_binary_unicode_filenames(self):
511
"""Test that contents of files are *not* encoded in UTF-8 when there
512
is a binary file in the diff.
514
# See https://bugs.launchpad.net/bugs/110092.
515
self.requireFeature(UnicodeFilename)
517
# This bug isn't triggered with cStringIO.
518
from StringIO import StringIO
519
tree = self.make_branch_and_tree('tree')
520
alpha, omega = u'\u03b1', u'\u03c9'
521
alpha_utf8, omega_utf8 = alpha.encode('utf8'), omega.encode('utf8')
522
self.build_tree_contents(
523
[('tree/' + alpha, chr(0)),
525
('The %s and the %s\n' % (alpha_utf8, omega_utf8)))])
526
tree.add([alpha], ['file-id'])
527
tree.add([omega], ['file-id-2'])
528
diff_content = StringIO()
529
show_diff_trees(tree.basis_tree(), tree, diff_content)
530
diff = diff_content.getvalue()
531
self.assertContainsRe(diff, r"=== added file '%s'" % alpha_utf8)
532
self.assertContainsRe(
533
diff, "Binary files a/%s.*and b/%s.* differ\n" % (alpha_utf8, alpha_utf8))
534
self.assertContainsRe(diff, r"=== added file '%s'" % omega_utf8)
535
self.assertContainsRe(diff, r"--- a/%s" % (omega_utf8,))
536
self.assertContainsRe(diff, r"\+\+\+ b/%s" % (omega_utf8,))
538
def test_unicode_filename(self):
539
"""Test when the filename are unicode."""
540
self.requireFeature(UnicodeFilename)
542
alpha, omega = u'\u03b1', u'\u03c9'
543
autf8, outf8 = alpha.encode('utf8'), omega.encode('utf8')
545
tree = self.make_branch_and_tree('tree')
546
self.build_tree_contents([('tree/ren_'+alpha, 'contents\n')])
547
tree.add(['ren_'+alpha], ['file-id-2'])
548
self.build_tree_contents([('tree/del_'+alpha, 'contents\n')])
549
tree.add(['del_'+alpha], ['file-id-3'])
550
self.build_tree_contents([('tree/mod_'+alpha, 'contents\n')])
551
tree.add(['mod_'+alpha], ['file-id-4'])
553
tree.commit('one', rev_id='rev-1')
555
tree.rename_one('ren_'+alpha, 'ren_'+omega)
556
tree.remove('del_'+alpha)
557
self.build_tree_contents([('tree/add_'+alpha, 'contents\n')])
558
tree.add(['add_'+alpha], ['file-id'])
559
self.build_tree_contents([('tree/mod_'+alpha, 'contents_mod\n')])
561
diff = self.get_diff(tree.basis_tree(), tree)
562
self.assertContainsRe(diff,
563
"=== renamed file 'ren_%s' => 'ren_%s'\n"%(autf8, outf8))
564
self.assertContainsRe(diff, "=== added file 'add_%s'"%autf8)
565
self.assertContainsRe(diff, "=== modified file 'mod_%s'"%autf8)
566
self.assertContainsRe(diff, "=== removed file 'del_%s'"%autf8)
569
class DiffWasIs(DiffPath):
571
def diff(self, file_id, old_path, new_path, old_kind, new_kind):
572
self.to_file.write('was: ')
573
self.to_file.write(self.old_tree.get_file(file_id).read())
574
self.to_file.write('is: ')
575
self.to_file.write(self.new_tree.get_file(file_id).read())
579
class TestDiffTree(TestCaseWithTransport):
582
TestCaseWithTransport.setUp(self)
583
self.old_tree = self.make_branch_and_tree('old-tree')
584
self.old_tree.lock_write()
585
self.addCleanup(self.old_tree.unlock)
586
self.new_tree = self.make_branch_and_tree('new-tree')
587
self.new_tree.lock_write()
588
self.addCleanup(self.new_tree.unlock)
589
self.differ = DiffTree(self.old_tree, self.new_tree, StringIO())
591
def test_diff_text(self):
592
self.build_tree_contents([('old-tree/olddir/',),
593
('old-tree/olddir/oldfile', 'old\n')])
594
self.old_tree.add('olddir')
595
self.old_tree.add('olddir/oldfile', 'file-id')
596
self.build_tree_contents([('new-tree/newdir/',),
597
('new-tree/newdir/newfile', 'new\n')])
598
self.new_tree.add('newdir')
599
self.new_tree.add('newdir/newfile', 'file-id')
600
differ = DiffText(self.old_tree, self.new_tree, StringIO())
601
differ.diff_text('file-id', None, 'old label', 'new label')
603
'--- old label\n+++ new label\n@@ -1,1 +0,0 @@\n-old\n\n',
604
differ.to_file.getvalue())
605
differ.to_file.seek(0)
606
differ.diff_text(None, 'file-id', 'old label', 'new label')
608
'--- old label\n+++ new label\n@@ -0,0 +1,1 @@\n+new\n\n',
609
differ.to_file.getvalue())
610
differ.to_file.seek(0)
611
differ.diff_text('file-id', 'file-id', 'old label', 'new label')
613
'--- old label\n+++ new label\n@@ -1,1 +1,1 @@\n-old\n+new\n\n',
614
differ.to_file.getvalue())
616
def test_diff_deletion(self):
617
self.build_tree_contents([('old-tree/file', 'contents'),
618
('new-tree/file', 'contents')])
619
self.old_tree.add('file', 'file-id')
620
self.new_tree.add('file', 'file-id')
621
os.unlink('new-tree/file')
622
self.differ.show_diff(None)
623
self.assertContainsRe(self.differ.to_file.getvalue(), '-contents')
625
def test_diff_creation(self):
626
self.build_tree_contents([('old-tree/file', 'contents'),
627
('new-tree/file', 'contents')])
628
self.old_tree.add('file', 'file-id')
629
self.new_tree.add('file', 'file-id')
630
os.unlink('old-tree/file')
631
self.differ.show_diff(None)
632
self.assertContainsRe(self.differ.to_file.getvalue(), '\+contents')
634
def test_diff_symlink(self):
635
differ = DiffSymlink(self.old_tree, self.new_tree, StringIO())
636
differ.diff_symlink('old target', None)
637
self.assertEqual("=== target was 'old target'\n",
638
differ.to_file.getvalue())
640
differ = DiffSymlink(self.old_tree, self.new_tree, StringIO())
641
differ.diff_symlink(None, 'new target')
642
self.assertEqual("=== target is 'new target'\n",
643
differ.to_file.getvalue())
645
differ = DiffSymlink(self.old_tree, self.new_tree, StringIO())
646
differ.diff_symlink('old target', 'new target')
647
self.assertEqual("=== target changed 'old target' => 'new target'\n",
648
differ.to_file.getvalue())
651
self.build_tree_contents([('old-tree/olddir/',),
652
('old-tree/olddir/oldfile', 'old\n')])
653
self.old_tree.add('olddir')
654
self.old_tree.add('olddir/oldfile', 'file-id')
655
self.build_tree_contents([('new-tree/newdir/',),
656
('new-tree/newdir/newfile', 'new\n')])
657
self.new_tree.add('newdir')
658
self.new_tree.add('newdir/newfile', 'file-id')
659
self.differ.diff('file-id', 'olddir/oldfile', 'newdir/newfile')
660
self.assertContainsRe(
661
self.differ.to_file.getvalue(),
662
r'--- olddir/oldfile.*\n\+\+\+ newdir/newfile.*\n\@\@ -1,1 \+1,1'
663
' \@\@\n-old\n\+new\n\n')
665
def test_diff_kind_change(self):
666
self.build_tree_contents([('old-tree/olddir/',),
667
('old-tree/olddir/oldfile', 'old\n')])
668
self.old_tree.add('olddir')
669
self.old_tree.add('olddir/oldfile', 'file-id')
670
self.build_tree(['new-tree/newdir/'])
671
os.symlink('new', 'new-tree/newdir/newfile')
672
self.new_tree.add('newdir')
673
self.new_tree.add('newdir/newfile', 'file-id')
674
self.differ.diff('file-id', 'olddir/oldfile', 'newdir/newfile')
675
self.assertContainsRe(
676
self.differ.to_file.getvalue(),
677
r'--- olddir/oldfile.*\n\+\+\+ newdir/newfile.*\n\@\@ -1,1 \+0,0'
679
self.assertContainsRe(self.differ.to_file.getvalue(),
680
"=== target is 'new'\n")
682
def test_diff_directory(self):
683
self.build_tree(['new-tree/new-dir/'])
684
self.new_tree.add('new-dir', 'new-dir-id')
685
self.differ.diff('new-dir-id', None, 'new-dir')
686
self.assertEqual(self.differ.to_file.getvalue(), '')
688
def create_old_new(self):
689
self.build_tree_contents([('old-tree/olddir/',),
690
('old-tree/olddir/oldfile', 'old\n')])
691
self.old_tree.add('olddir')
692
self.old_tree.add('olddir/oldfile', 'file-id')
693
self.build_tree_contents([('new-tree/newdir/',),
694
('new-tree/newdir/newfile', 'new\n')])
695
self.new_tree.add('newdir')
696
self.new_tree.add('newdir/newfile', 'file-id')
698
def test_register_diff(self):
699
self.create_old_new()
700
old_diff_factories = DiffTree.diff_factories
701
DiffTree.diff_factories=old_diff_factories[:]
702
DiffTree.diff_factories.insert(0, DiffWasIs.from_diff_tree)
704
differ = DiffTree(self.old_tree, self.new_tree, StringIO())
706
DiffTree.diff_factories = old_diff_factories
707
differ.diff('file-id', 'olddir/oldfile', 'newdir/newfile')
708
self.assertNotContainsRe(
709
differ.to_file.getvalue(),
710
r'--- olddir/oldfile.*\n\+\+\+ newdir/newfile.*\n\@\@ -1,1 \+1,1'
711
' \@\@\n-old\n\+new\n\n')
712
self.assertContainsRe(differ.to_file.getvalue(),
713
'was: old\nis: new\n')
715
def test_extra_factories(self):
716
self.create_old_new()
717
differ = DiffTree(self.old_tree, self.new_tree, StringIO(),
718
extra_factories=[DiffWasIs.from_diff_tree])
719
differ.diff('file-id', 'olddir/oldfile', 'newdir/newfile')
720
self.assertNotContainsRe(
721
differ.to_file.getvalue(),
722
r'--- olddir/oldfile.*\n\+\+\+ newdir/newfile.*\n\@\@ -1,1 \+1,1'
723
' \@\@\n-old\n\+new\n\n')
724
self.assertContainsRe(differ.to_file.getvalue(),
725
'was: old\nis: new\n')
727
def test_alphabetical_order(self):
728
self.build_tree(['new-tree/a-file'])
729
self.new_tree.add('a-file')
730
self.build_tree(['old-tree/b-file'])
731
self.old_tree.add('b-file')
732
self.differ.show_diff(None)
733
self.assertContainsRe(self.differ.to_file.getvalue(),
734
'.*a-file(.|\n)*b-file')
737
class TestPatienceDiffLib(TestCase):
740
super(TestPatienceDiffLib, self).setUp()
741
self._unique_lcs = bzrlib._patiencediff_py.unique_lcs_py
742
self._recurse_matches = bzrlib._patiencediff_py.recurse_matches_py
743
self._PatienceSequenceMatcher = \
744
bzrlib._patiencediff_py.PatienceSequenceMatcher_py
746
def test_unique_lcs(self):
747
unique_lcs = self._unique_lcs
748
self.assertEquals(unique_lcs('', ''), [])
749
self.assertEquals(unique_lcs('', 'a'), [])
750
self.assertEquals(unique_lcs('a', ''), [])
751
self.assertEquals(unique_lcs('a', 'a'), [(0,0)])
752
self.assertEquals(unique_lcs('a', 'b'), [])
753
self.assertEquals(unique_lcs('ab', 'ab'), [(0,0), (1,1)])
754
self.assertEquals(unique_lcs('abcde', 'cdeab'), [(2,0), (3,1), (4,2)])
755
self.assertEquals(unique_lcs('cdeab', 'abcde'), [(0,2), (1,3), (2,4)])
756
self.assertEquals(unique_lcs('abXde', 'abYde'), [(0,0), (1,1),
758
self.assertEquals(unique_lcs('acbac', 'abc'), [(2,1)])
760
def test_recurse_matches(self):
761
def test_one(a, b, matches):
763
self._recurse_matches(
764
a, b, 0, 0, len(a), len(b), test_matches, 10)
765
self.assertEquals(test_matches, matches)
767
test_one(['a', '', 'b', '', 'c'], ['a', 'a', 'b', 'c', 'c'],
768
[(0, 0), (2, 2), (4, 4)])
769
test_one(['a', 'c', 'b', 'a', 'c'], ['a', 'b', 'c'],
770
[(0, 0), (2, 1), (4, 2)])
771
# Even though 'bc' is not unique globally, and is surrounded by
772
# non-matching lines, we should still match, because they are locally
774
test_one('abcdbce', 'afbcgdbce', [(0,0), (1, 2), (2, 3), (3, 5),
775
(4, 6), (5, 7), (6, 8)])
777
# recurse_matches doesn't match non-unique
778
# lines surrounded by bogus text.
779
# The update has been done in patiencediff.SequenceMatcher instead
781
# This is what it could be
782
#test_one('aBccDe', 'abccde', [(0,0), (2,2), (3,3), (5,5)])
784
# This is what it currently gives:
785
test_one('aBccDe', 'abccde', [(0,0), (5,5)])
787
def assertDiffBlocks(self, a, b, expected_blocks):
788
"""Check that the sequence matcher returns the correct blocks.
790
:param a: A sequence to match
791
:param b: Another sequence to match
792
:param expected_blocks: The expected output, not including the final
793
matching block (len(a), len(b), 0)
795
matcher = self._PatienceSequenceMatcher(None, a, b)
796
blocks = matcher.get_matching_blocks()
798
self.assertEqual((len(a), len(b), 0), last)
799
self.assertEqual(expected_blocks, blocks)
801
def test_matching_blocks(self):
802
# Some basic matching tests
803
self.assertDiffBlocks('', '', [])
804
self.assertDiffBlocks([], [], [])
805
self.assertDiffBlocks('abc', '', [])
806
self.assertDiffBlocks('', 'abc', [])
807
self.assertDiffBlocks('abcd', 'abcd', [(0, 0, 4)])
808
self.assertDiffBlocks('abcd', 'abce', [(0, 0, 3)])
809
self.assertDiffBlocks('eabc', 'abce', [(1, 0, 3)])
810
self.assertDiffBlocks('eabce', 'abce', [(1, 0, 4)])
811
self.assertDiffBlocks('abcde', 'abXde', [(0, 0, 2), (3, 3, 2)])
812
self.assertDiffBlocks('abcde', 'abXYZde', [(0, 0, 2), (3, 5, 2)])
813
self.assertDiffBlocks('abde', 'abXYZde', [(0, 0, 2), (2, 5, 2)])
814
# This may check too much, but it checks to see that
815
# a copied block stays attached to the previous section,
817
# difflib would tend to grab the trailing longest match
818
# which would make the diff not look right
819
self.assertDiffBlocks('abcdefghijklmnop', 'abcdefxydefghijklmnop',
820
[(0, 0, 6), (6, 11, 10)])
822
# make sure it supports passing in lists
823
self.assertDiffBlocks(
826
'how are you today?\n'],
828
'how are you today?\n'],
829
[(0, 0, 1), (2, 1, 1)])
831
# non unique lines surrounded by non-matching lines
833
self.assertDiffBlocks('aBccDe', 'abccde', [(0,0,1), (5,5,1)])
835
# But they only need to be locally unique
836
self.assertDiffBlocks('aBcDec', 'abcdec', [(0,0,1), (2,2,1), (4,4,2)])
838
# non unique blocks won't be matched
839
self.assertDiffBlocks('aBcdEcdFg', 'abcdecdfg', [(0,0,1), (8,8,1)])
841
# but locally unique ones will
842
self.assertDiffBlocks('aBcdEeXcdFg', 'abcdecdfg', [(0,0,1), (2,2,2),
843
(5,4,1), (7,5,2), (10,8,1)])
845
self.assertDiffBlocks('abbabbXd', 'cabbabxd', [(7,7,1)])
846
self.assertDiffBlocks('abbabbbb', 'cabbabbc', [])
847
self.assertDiffBlocks('bbbbbbbb', 'cbbbbbbc', [])
849
def test_matching_blocks_tuples(self):
850
# Some basic matching tests
851
self.assertDiffBlocks([], [], [])
852
self.assertDiffBlocks([('a',), ('b',), ('c,')], [], [])
853
self.assertDiffBlocks([], [('a',), ('b',), ('c,')], [])
854
self.assertDiffBlocks([('a',), ('b',), ('c,')],
855
[('a',), ('b',), ('c,')],
857
self.assertDiffBlocks([('a',), ('b',), ('c,')],
858
[('a',), ('b',), ('d,')],
860
self.assertDiffBlocks([('d',), ('b',), ('c,')],
861
[('a',), ('b',), ('c,')],
863
self.assertDiffBlocks([('d',), ('a',), ('b',), ('c,')],
864
[('a',), ('b',), ('c,')],
866
self.assertDiffBlocks([('a', 'b'), ('c', 'd'), ('e', 'f')],
867
[('a', 'b'), ('c', 'X'), ('e', 'f')],
868
[(0, 0, 1), (2, 2, 1)])
869
self.assertDiffBlocks([('a', 'b'), ('c', 'd'), ('e', 'f')],
870
[('a', 'b'), ('c', 'dX'), ('e', 'f')],
871
[(0, 0, 1), (2, 2, 1)])
873
def test_opcodes(self):
874
def chk_ops(a, b, expected_codes):
875
s = self._PatienceSequenceMatcher(None, a, b)
876
self.assertEquals(expected_codes, s.get_opcodes())
880
chk_ops('abc', '', [('delete', 0,3, 0,0)])
881
chk_ops('', 'abc', [('insert', 0,0, 0,3)])
882
chk_ops('abcd', 'abcd', [('equal', 0,4, 0,4)])
883
chk_ops('abcd', 'abce', [('equal', 0,3, 0,3),
884
('replace', 3,4, 3,4)
886
chk_ops('eabc', 'abce', [('delete', 0,1, 0,0),
890
chk_ops('eabce', 'abce', [('delete', 0,1, 0,0),
893
chk_ops('abcde', 'abXde', [('equal', 0,2, 0,2),
894
('replace', 2,3, 2,3),
897
chk_ops('abcde', 'abXYZde', [('equal', 0,2, 0,2),
898
('replace', 2,3, 2,5),
901
chk_ops('abde', 'abXYZde', [('equal', 0,2, 0,2),
902
('insert', 2,2, 2,5),
905
chk_ops('abcdefghijklmnop', 'abcdefxydefghijklmnop',
906
[('equal', 0,6, 0,6),
907
('insert', 6,6, 6,11),
908
('equal', 6,16, 11,21)
913
, 'how are you today?\n'],
915
, 'how are you today?\n'],
916
[('equal', 0,1, 0,1),
917
('delete', 1,2, 1,1),
920
chk_ops('aBccDe', 'abccde',
921
[('equal', 0,1, 0,1),
922
('replace', 1,5, 1,5),
925
chk_ops('aBcDec', 'abcdec',
926
[('equal', 0,1, 0,1),
927
('replace', 1,2, 1,2),
929
('replace', 3,4, 3,4),
932
chk_ops('aBcdEcdFg', 'abcdecdfg',
933
[('equal', 0,1, 0,1),
934
('replace', 1,8, 1,8),
937
chk_ops('aBcdEeXcdFg', 'abcdecdfg',
938
[('equal', 0,1, 0,1),
939
('replace', 1,2, 1,2),
941
('delete', 4,5, 4,4),
943
('delete', 6,7, 5,5),
945
('replace', 9,10, 7,8),
946
('equal', 10,11, 8,9)
949
def test_grouped_opcodes(self):
950
def chk_ops(a, b, expected_codes, n=3):
951
s = self._PatienceSequenceMatcher(None, a, b)
952
self.assertEquals(expected_codes, list(s.get_grouped_opcodes(n)))
956
chk_ops('abc', '', [[('delete', 0,3, 0,0)]])
957
chk_ops('', 'abc', [[('insert', 0,0, 0,3)]])
958
chk_ops('abcd', 'abcd', [])
959
chk_ops('abcd', 'abce', [[('equal', 0,3, 0,3),
960
('replace', 3,4, 3,4)
962
chk_ops('eabc', 'abce', [[('delete', 0,1, 0,0),
966
chk_ops('abcdefghijklmnop', 'abcdefxydefghijklmnop',
967
[[('equal', 3,6, 3,6),
968
('insert', 6,6, 6,11),
969
('equal', 6,9, 11,14)
971
chk_ops('abcdefghijklmnop', 'abcdefxydefghijklmnop',
972
[[('equal', 2,6, 2,6),
973
('insert', 6,6, 6,11),
974
('equal', 6,10, 11,15)
976
chk_ops('Xabcdef', 'abcdef',
977
[[('delete', 0,1, 0,0),
980
chk_ops('abcdef', 'abcdefX',
981
[[('equal', 3,6, 3,6),
986
def test_multiple_ranges(self):
987
# There was an earlier bug where we used a bad set of ranges,
988
# this triggers that specific bug, to make sure it doesn't regress
989
self.assertDiffBlocks('abcdefghijklmnop',
990
'abcXghiYZQRSTUVWXYZijklmnop',
991
[(0, 0, 3), (6, 4, 3), (9, 20, 7)])
993
self.assertDiffBlocks('ABCd efghIjk L',
994
'AxyzBCn mo pqrstuvwI1 2 L',
995
[(0,0,1), (1, 4, 2), (9, 19, 1), (12, 23, 3)])
997
# These are rot13 code snippets.
998
self.assertDiffBlocks('''\
999
trg nqqrq jura lbh nqq n svyr va gur qverpgbel.
1001
gnxrf_netf = ['svyr*']
1002
gnxrf_bcgvbaf = ['ab-erphefr']
1004
qrs eha(frys, svyr_yvfg, ab_erphefr=Snyfr):
1005
sebz omeyvo.nqq vzcbeg fzneg_nqq, nqq_ercbegre_cevag, nqq_ercbegre_ahyy
1007
ercbegre = nqq_ercbegre_ahyy
1009
ercbegre = nqq_ercbegre_cevag
1010
fzneg_nqq(svyr_yvfg, abg ab_erphefr, ercbegre)
1013
pynff pzq_zxqve(Pbzznaq):
1014
'''.splitlines(True), '''\
1015
trg nqqrq jura lbh nqq n svyr va gur qverpgbel.
1017
--qel-eha jvyy fubj juvpu svyrf jbhyq or nqqrq, ohg abg npghnyyl
1020
gnxrf_netf = ['svyr*']
1021
gnxrf_bcgvbaf = ['ab-erphefr', 'qel-eha']
1023
qrs eha(frys, svyr_yvfg, ab_erphefr=Snyfr, qel_eha=Snyfr):
1028
# Guvf vf cbvagyrff, ohg V'q engure abg envfr na reebe
1029
npgvba = omeyvo.nqq.nqq_npgvba_ahyy
1031
npgvba = omeyvo.nqq.nqq_npgvba_cevag
1033
npgvba = omeyvo.nqq.nqq_npgvba_nqq
1035
npgvba = omeyvo.nqq.nqq_npgvba_nqq_naq_cevag
1037
omeyvo.nqq.fzneg_nqq(svyr_yvfg, abg ab_erphefr, npgvba)
1040
pynff pzq_zxqve(Pbzznaq):
1041
'''.splitlines(True)
1042
, [(0,0,1), (1, 4, 2), (9, 19, 1), (12, 23, 3)])
1044
def test_patience_unified_diff(self):
1045
txt_a = ['hello there\n',
1047
'how are you today?\n']
1048
txt_b = ['hello there\n',
1049
'how are you today?\n']
1050
unified_diff = bzrlib.patiencediff.unified_diff
1051
psm = self._PatienceSequenceMatcher
1052
self.assertEquals([ '--- \n',
1054
'@@ -1,3 +1,2 @@\n',
1057
' how are you today?\n'
1059
, list(unified_diff(txt_a, txt_b,
1060
sequencematcher=psm)))
1061
txt_a = map(lambda x: x+'\n', 'abcdefghijklmnop')
1062
txt_b = map(lambda x: x+'\n', 'abcdefxydefghijklmnop')
1063
# This is the result with LongestCommonSubstring matching
1064
self.assertEquals(['--- \n',
1066
'@@ -1,6 +1,11 @@\n',
1078
, list(unified_diff(txt_a, txt_b)))
1079
# And the patience diff
1080
self.assertEquals(['--- \n',
1082
'@@ -4,6 +4,11 @@\n',
1095
, list(unified_diff(txt_a, txt_b,
1096
sequencematcher=psm)))
1099
class TestPatienceDiffLib_c(TestPatienceDiffLib):
1101
_test_needs_features = [CompiledPatienceDiffFeature]
1104
super(TestPatienceDiffLib_c, self).setUp()
1105
import bzrlib._patiencediff_c
1106
self._unique_lcs = bzrlib._patiencediff_c.unique_lcs_c
1107
self._recurse_matches = bzrlib._patiencediff_c.recurse_matches_c
1108
self._PatienceSequenceMatcher = \
1109
bzrlib._patiencediff_c.PatienceSequenceMatcher_c
1111
def test_unhashable(self):
1112
"""We should get a proper exception here."""
1113
# We need to be able to hash items in the sequence, lists are
1114
# unhashable, and thus cannot be diffed
1115
e = self.assertRaises(TypeError, self._PatienceSequenceMatcher,
1117
e = self.assertRaises(TypeError, self._PatienceSequenceMatcher,
1118
None, ['valid', []], [])
1119
e = self.assertRaises(TypeError, self._PatienceSequenceMatcher,
1120
None, ['valid'], [[]])
1121
e = self.assertRaises(TypeError, self._PatienceSequenceMatcher,
1122
None, ['valid'], ['valid', []])
1125
class TestPatienceDiffLibFiles(TestCaseInTempDir):
1128
super(TestPatienceDiffLibFiles, self).setUp()
1129
self._PatienceSequenceMatcher = \
1130
bzrlib._patiencediff_py.PatienceSequenceMatcher_py
1132
def test_patience_unified_diff_files(self):
1133
txt_a = ['hello there\n',
1135
'how are you today?\n']
1136
txt_b = ['hello there\n',
1137
'how are you today?\n']
1138
open('a1', 'wb').writelines(txt_a)
1139
open('b1', 'wb').writelines(txt_b)
1141
unified_diff_files = bzrlib.patiencediff.unified_diff_files
1142
psm = self._PatienceSequenceMatcher
1143
self.assertEquals(['--- a1 \n',
1145
'@@ -1,3 +1,2 @@\n',
1148
' how are you today?\n',
1150
, list(unified_diff_files('a1', 'b1',
1151
sequencematcher=psm)))
1153
txt_a = map(lambda x: x+'\n', 'abcdefghijklmnop')
1154
txt_b = map(lambda x: x+'\n', 'abcdefxydefghijklmnop')
1155
open('a2', 'wb').writelines(txt_a)
1156
open('b2', 'wb').writelines(txt_b)
1158
# This is the result with LongestCommonSubstring matching
1159
self.assertEquals(['--- a2 \n',
1161
'@@ -1,6 +1,11 @@\n',
1173
, list(unified_diff_files('a2', 'b2')))
1175
# And the patience diff
1176
self.assertEquals(['--- a2 \n',
1178
'@@ -4,6 +4,11 @@\n',
1191
, list(unified_diff_files('a2', 'b2',
1192
sequencematcher=psm)))
1195
class TestPatienceDiffLibFiles_c(TestPatienceDiffLibFiles):
1197
_test_needs_features = [CompiledPatienceDiffFeature]
1200
super(TestPatienceDiffLibFiles_c, self).setUp()
1201
import bzrlib._patiencediff_c
1202
self._PatienceSequenceMatcher = \
1203
bzrlib._patiencediff_c.PatienceSequenceMatcher_c
1206
class TestUsingCompiledIfAvailable(TestCase):
1208
def test_PatienceSequenceMatcher(self):
1209
if CompiledPatienceDiffFeature.available():
1210
from bzrlib._patiencediff_c import PatienceSequenceMatcher_c
1211
self.assertIs(PatienceSequenceMatcher_c,
1212
bzrlib.patiencediff.PatienceSequenceMatcher)
1214
from bzrlib._patiencediff_py import PatienceSequenceMatcher_py
1215
self.assertIs(PatienceSequenceMatcher_py,
1216
bzrlib.patiencediff.PatienceSequenceMatcher)
1218
def test_unique_lcs(self):
1219
if CompiledPatienceDiffFeature.available():
1220
from bzrlib._patiencediff_c import unique_lcs_c
1221
self.assertIs(unique_lcs_c,
1222
bzrlib.patiencediff.unique_lcs)
1224
from bzrlib._patiencediff_py import unique_lcs_py
1225
self.assertIs(unique_lcs_py,
1226
bzrlib.patiencediff.unique_lcs)
1228
def test_recurse_matches(self):
1229
if CompiledPatienceDiffFeature.available():
1230
from bzrlib._patiencediff_c import recurse_matches_c
1231
self.assertIs(recurse_matches_c,
1232
bzrlib.patiencediff.recurse_matches)
1234
from bzrlib._patiencediff_py import recurse_matches_py
1235
self.assertIs(recurse_matches_py,
1236
bzrlib.patiencediff.recurse_matches)
1239
class TestDiffFromTool(TestCaseWithTransport):
1241
def test_from_string(self):
1242
diff_obj = DiffFromTool.from_string('diff', None, None, None)
1243
self.addCleanup(diff_obj.finish)
1244
self.assertEqual(['diff', '%(old_path)s', '%(new_path)s'],
1245
diff_obj.command_template)
1246
diff_obj = DiffFromTool.from_string('diff -u\\ 5', None, None, None)
1247
self.assertEqual(['diff', '-u 5', '%(old_path)s', '%(new_path)s'],
1248
diff_obj.command_template)
1249
self.assertEqual(['diff', '-u 5', 'old-path', 'new-path'],
1250
diff_obj._get_command('old-path', 'new-path'))
1252
def test_execute(self):
1254
diff_obj = DiffFromTool(['python', '-c',
1255
'print "%(old_path)s %(new_path)s"'],
1257
self.addCleanup(diff_obj.finish)
1258
diff_obj._execute('old', 'new')
1259
self.assertEqual(output.getvalue(), 'old new\n')
1261
def test_prepare_files(self):
1263
tree = self.make_branch_and_tree('tree')
1264
self.build_tree_contents([('tree/file', 'oldcontent')])
1265
tree.add('file', 'file-id')
1266
tree.commit('old tree')
1267
self.build_tree_contents([('tree/file', 'newcontent')])
1268
old_tree = tree.basis_tree()
1269
old_tree.lock_read()
1270
self.addCleanup(old_tree.unlock)
1271
diff_obj = DiffFromTool(['python', '-c',
1272
'print "%(old_path)s %(new_path)s"'],
1273
old_tree, tree, output)
1274
self.addCleanup(diff_obj.finish)
1275
self.assertContainsRe(diff_obj._root, 'bzr-diff-[^/]*')
1276
old_path, new_path = diff_obj._prepare_files('file-id', 'oldname',
1278
self.assertContainsRe(old_path, 'old/oldname$')
1279
self.assertContainsRe(new_path, 'new/newname$')
1280
self.assertFileEqual('oldcontent', old_path)
1281
self.assertFileEqual('newcontent', new_path)
1282
# make sure we can create files with the same parent directories
1283
diff_obj._prepare_files('file-id', 'oldname2', 'newname2')
added
removed
bzrlib/tests/test_diff.py
