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 (
32
from bzrlib.errors import BinaryFile, NoDiff
33
import bzrlib.osutils as osutils
34
import bzrlib.patiencediff
35
import bzrlib._patiencediff_py
36
from bzrlib.tests import (Feature, TestCase, TestCaseWithTransport,
37
TestCaseInTempDir, TestSkipped)
40
class _CompiledPatienceDiffFeature(Feature):
44
import bzrlib._patiencediff_c
49
def feature_name(self):
50
return 'bzrlib._patiencediff_c'
52
CompiledPatienceDiffFeature = _CompiledPatienceDiffFeature()
55
class _UnicodeFilename(Feature):
56
"""Does the filesystem support Unicode filenames?"""
61
except UnicodeEncodeError:
63
except (IOError, OSError):
64
# The filesystem allows the Unicode filename but the file doesn't
68
# The filesystem allows the Unicode filename and the file exists,
72
UnicodeFilename = _UnicodeFilename()
75
class TestUnicodeFilename(TestCase):
77
def test_probe_passes(self):
78
"""UnicodeFilename._probe passes."""
79
# We can't test much more than that because the behaviour depends
81
UnicodeFilename._probe()
84
def udiff_lines(old, new, allow_binary=False):
86
internal_diff('old', old, 'new', new, output, allow_binary)
88
return output.readlines()
91
def external_udiff_lines(old, new, use_stringio=False):
93
# StringIO has no fileno, so it tests a different codepath
96
output = TemporaryFile()
98
external_diff('old', old, 'new', new, output, diff_opts=['-u'])
100
raise TestSkipped('external "diff" not present to test')
102
lines = output.readlines()
107
class TestDiff(TestCase):
109
def test_add_nl(self):
110
"""diff generates a valid diff for patches that add a newline"""
111
lines = udiff_lines(['boo'], ['boo\n'])
112
self.check_patch(lines)
113
self.assertEquals(lines[4], '\\ No newline at end of file\n')
114
## "expected no-nl, got %r" % lines[4]
116
def test_add_nl_2(self):
117
"""diff generates a valid diff for patches that change last line and
120
lines = udiff_lines(['boo'], ['goo\n'])
121
self.check_patch(lines)
122
self.assertEquals(lines[4], '\\ No newline at end of file\n')
123
## "expected no-nl, got %r" % lines[4]
125
def test_remove_nl(self):
126
"""diff generates a valid diff for patches that change last line and
129
lines = udiff_lines(['boo\n'], ['boo'])
130
self.check_patch(lines)
131
self.assertEquals(lines[5], '\\ No newline at end of file\n')
132
## "expected no-nl, got %r" % lines[5]
134
def check_patch(self, lines):
135
self.assert_(len(lines) > 1)
136
## "Not enough lines for a file header for patch:\n%s" % "".join(lines)
137
self.assert_(lines[0].startswith ('---'))
138
## 'No orig line for patch:\n%s' % "".join(lines)
139
self.assert_(lines[1].startswith ('+++'))
140
## 'No mod line for patch:\n%s' % "".join(lines)
141
self.assert_(len(lines) > 2)
142
## "No hunks for patch:\n%s" % "".join(lines)
143
self.assert_(lines[2].startswith('@@'))
144
## "No hunk header for patch:\n%s" % "".join(lines)
145
self.assert_('@@' in lines[2][2:])
146
## "Unterminated hunk header for patch:\n%s" % "".join(lines)
148
def test_binary_lines(self):
149
self.assertRaises(BinaryFile, udiff_lines, [1023 * 'a' + '\x00'], [])
150
self.assertRaises(BinaryFile, udiff_lines, [], [1023 * 'a' + '\x00'])
151
udiff_lines([1023 * 'a' + '\x00'], [], allow_binary=True)
152
udiff_lines([], [1023 * 'a' + '\x00'], allow_binary=True)
154
def test_external_diff(self):
155
lines = external_udiff_lines(['boo\n'], ['goo\n'])
156
self.check_patch(lines)
157
self.assertEqual('\n', lines[-1])
159
def test_external_diff_no_fileno(self):
160
# Make sure that we can handle not having a fileno, even
161
# if the diff is large
162
lines = external_udiff_lines(['boo\n']*10000,
165
self.check_patch(lines)
167
def test_external_diff_binary_lang_c(self):
169
for lang in ('LANG', 'LC_ALL', 'LANGUAGE'):
170
old_env[lang] = osutils.set_or_unset_env(lang, 'C')
172
lines = external_udiff_lines(['\x00foobar\n'], ['foo\x00bar\n'])
173
# Older versions of diffutils say "Binary files", newer
174
# versions just say "Files".
175
self.assertContainsRe(lines[0],
176
'(Binary f|F)iles old and new differ\n')
177
self.assertEquals(lines[1:], ['\n'])
179
for lang, old_val in old_env.iteritems():
180
osutils.set_or_unset_env(lang, old_val)
182
def test_no_external_diff(self):
183
"""Check that NoDiff is raised when diff is not available"""
184
# Use os.environ['PATH'] to make sure no 'diff' command is available
185
orig_path = os.environ['PATH']
187
os.environ['PATH'] = ''
188
self.assertRaises(NoDiff, external_diff,
189
'old', ['boo\n'], 'new', ['goo\n'],
190
StringIO(), diff_opts=['-u'])
192
os.environ['PATH'] = orig_path
194
def test_internal_diff_default(self):
195
# Default internal diff encoding is utf8
197
internal_diff(u'old_\xb5', ['old_text\n'],
198
u'new_\xe5', ['new_text\n'], output)
199
lines = output.getvalue().splitlines(True)
200
self.check_patch(lines)
201
self.assertEquals(['--- old_\xc2\xb5\n',
202
'+++ new_\xc3\xa5\n',
210
def test_internal_diff_utf8(self):
212
internal_diff(u'old_\xb5', ['old_text\n'],
213
u'new_\xe5', ['new_text\n'], output,
214
path_encoding='utf8')
215
lines = output.getvalue().splitlines(True)
216
self.check_patch(lines)
217
self.assertEquals(['--- old_\xc2\xb5\n',
218
'+++ new_\xc3\xa5\n',
226
def test_internal_diff_iso_8859_1(self):
228
internal_diff(u'old_\xb5', ['old_text\n'],
229
u'new_\xe5', ['new_text\n'], output,
230
path_encoding='iso-8859-1')
231
lines = output.getvalue().splitlines(True)
232
self.check_patch(lines)
233
self.assertEquals(['--- old_\xb5\n',
242
def test_internal_diff_no_content(self):
244
internal_diff(u'old', [], u'new', [], output)
245
self.assertEqual('', output.getvalue())
247
def test_internal_diff_no_changes(self):
249
internal_diff(u'old', ['text\n', 'contents\n'],
250
u'new', ['text\n', 'contents\n'],
252
self.assertEqual('', output.getvalue())
254
def test_internal_diff_returns_bytes(self):
256
output = StringIO.StringIO()
257
internal_diff(u'old_\xb5', ['old_text\n'],
258
u'new_\xe5', ['new_text\n'], output)
259
self.failUnless(isinstance(output.getvalue(), str),
260
'internal_diff should return bytestrings')
263
class TestDiffFiles(TestCaseInTempDir):
265
def test_external_diff_binary(self):
266
"""The output when using external diff should use diff's i18n error"""
267
# Make sure external_diff doesn't fail in the current LANG
268
lines = external_udiff_lines(['\x00foobar\n'], ['foo\x00bar\n'])
270
cmd = ['diff', '-u', '--binary', 'old', 'new']
271
open('old', 'wb').write('\x00foobar\n')
272
open('new', 'wb').write('foo\x00bar\n')
273
pipe = subprocess.Popen(cmd, stdout=subprocess.PIPE,
274
stdin=subprocess.PIPE)
275
out, err = pipe.communicate()
276
# Diff returns '2' on Binary files.
277
self.assertEqual(2, pipe.returncode)
278
# We should output whatever diff tells us, plus a trailing newline
279
self.assertEqual(out.splitlines(True) + ['\n'], lines)
282
class TestShowDiffTreesHelper(TestCaseWithTransport):
283
"""Has a helper for running show_diff_trees"""
285
def get_diff(self, tree1, tree2, specific_files=None, working_tree=None):
287
if working_tree is not None:
288
extra_trees = (working_tree,)
291
show_diff_trees(tree1, tree2, output, specific_files=specific_files,
292
extra_trees=extra_trees, old_label='old/',
294
return output.getvalue()
297
class TestDiffDates(TestShowDiffTreesHelper):
300
super(TestDiffDates, self).setUp()
301
self.wt = self.make_branch_and_tree('.')
302
self.b = self.wt.branch
303
self.build_tree_contents([
304
('file1', 'file1 contents at rev 1\n'),
305
('file2', 'file2 contents at rev 1\n')
307
self.wt.add(['file1', 'file2'])
309
message='Revision 1',
310
timestamp=1143849600, # 2006-04-01 00:00:00 UTC
313
self.build_tree_contents([('file1', 'file1 contents at rev 2\n')])
315
message='Revision 2',
316
timestamp=1143936000, # 2006-04-02 00:00:00 UTC
319
self.build_tree_contents([('file2', 'file2 contents at rev 3\n')])
321
message='Revision 3',
322
timestamp=1144022400, # 2006-04-03 00:00:00 UTC
325
self.wt.remove(['file2'])
327
message='Revision 4',
328
timestamp=1144108800, # 2006-04-04 00:00:00 UTC
331
self.build_tree_contents([
332
('file1', 'file1 contents in working tree\n')
334
# set the date stamps for files in the working tree to known values
335
os.utime('file1', (1144195200, 1144195200)) # 2006-04-05 00:00:00 UTC
337
def test_diff_rev_tree_working_tree(self):
338
output = self.get_diff(self.wt.basis_tree(), self.wt)
339
# note that the date for old/file1 is from rev 2 rather than from
340
# the basis revision (rev 4)
341
self.assertEqualDiff(output, '''\
342
=== modified file 'file1'
343
--- old/file1\t2006-04-02 00:00:00 +0000
344
+++ new/file1\t2006-04-05 00:00:00 +0000
346
-file1 contents at rev 2
347
+file1 contents in working tree
351
def test_diff_rev_tree_rev_tree(self):
352
tree1 = self.b.repository.revision_tree('rev-2')
353
tree2 = self.b.repository.revision_tree('rev-3')
354
output = self.get_diff(tree1, tree2)
355
self.assertEqualDiff(output, '''\
356
=== modified file 'file2'
357
--- old/file2\t2006-04-01 00:00:00 +0000
358
+++ new/file2\t2006-04-03 00:00:00 +0000
360
-file2 contents at rev 1
361
+file2 contents at rev 3
365
def test_diff_add_files(self):
366
tree1 = self.b.repository.revision_tree(None)
367
tree2 = self.b.repository.revision_tree('rev-1')
368
output = self.get_diff(tree1, tree2)
369
# the files have the epoch time stamp for the tree in which
371
self.assertEqualDiff(output, '''\
372
=== added file 'file1'
373
--- old/file1\t1970-01-01 00:00:00 +0000
374
+++ new/file1\t2006-04-01 00:00:00 +0000
376
+file1 contents at rev 1
378
=== added file 'file2'
379
--- old/file2\t1970-01-01 00:00:00 +0000
380
+++ new/file2\t2006-04-01 00:00:00 +0000
382
+file2 contents at rev 1
386
def test_diff_remove_files(self):
387
tree1 = self.b.repository.revision_tree('rev-3')
388
tree2 = self.b.repository.revision_tree('rev-4')
389
output = self.get_diff(tree1, tree2)
390
# the file has the epoch time stamp for the tree in which
392
self.assertEqualDiff(output, '''\
393
=== removed file 'file2'
394
--- old/file2\t2006-04-03 00:00:00 +0000
395
+++ new/file2\t1970-01-01 00:00:00 +0000
397
-file2 contents at rev 3
401
def test_show_diff_specified(self):
402
"""A working tree filename can be used to identify a file"""
403
self.wt.rename_one('file1', 'file1b')
404
old_tree = self.b.repository.revision_tree('rev-1')
405
new_tree = self.b.repository.revision_tree('rev-4')
406
out = self.get_diff(old_tree, new_tree, specific_files=['file1b'],
407
working_tree=self.wt)
408
self.assertContainsRe(out, 'file1\t')
410
def test_recursive_diff(self):
411
"""Children of directories are matched"""
414
self.wt.add(['dir1', 'dir2'])
415
self.wt.rename_one('file1', 'dir1/file1')
416
old_tree = self.b.repository.revision_tree('rev-1')
417
new_tree = self.b.repository.revision_tree('rev-4')
418
out = self.get_diff(old_tree, new_tree, specific_files=['dir1'],
419
working_tree=self.wt)
420
self.assertContainsRe(out, 'file1\t')
421
out = self.get_diff(old_tree, new_tree, specific_files=['dir2'],
422
working_tree=self.wt)
423
self.assertNotContainsRe(out, 'file1\t')
427
class TestShowDiffTrees(TestShowDiffTreesHelper):
428
"""Direct tests for show_diff_trees"""
430
def test_modified_file(self):
431
"""Test when a file is modified."""
432
tree = self.make_branch_and_tree('tree')
433
self.build_tree_contents([('tree/file', 'contents\n')])
434
tree.add(['file'], ['file-id'])
435
tree.commit('one', rev_id='rev-1')
437
self.build_tree_contents([('tree/file', 'new contents\n')])
438
diff = self.get_diff(tree.basis_tree(), tree)
439
self.assertContainsRe(diff, "=== modified file 'file'\n")
440
self.assertContainsRe(diff, '--- old/file\t')
441
self.assertContainsRe(diff, '\\+\\+\\+ new/file\t')
442
self.assertContainsRe(diff, '-contents\n'
445
def test_modified_file_in_renamed_dir(self):
446
"""Test when a file is modified in a renamed directory."""
447
tree = self.make_branch_and_tree('tree')
448
self.build_tree(['tree/dir/'])
449
self.build_tree_contents([('tree/dir/file', 'contents\n')])
450
tree.add(['dir', 'dir/file'], ['dir-id', 'file-id'])
451
tree.commit('one', rev_id='rev-1')
453
tree.rename_one('dir', 'other')
454
self.build_tree_contents([('tree/other/file', 'new contents\n')])
455
diff = self.get_diff(tree.basis_tree(), tree)
456
self.assertContainsRe(diff, "=== renamed directory 'dir' => 'other'\n")
457
self.assertContainsRe(diff, "=== modified file 'other/file'\n")
458
# XXX: This is technically incorrect, because it used to be at another
459
# location. What to do?
460
self.assertContainsRe(diff, '--- old/dir/file\t')
461
self.assertContainsRe(diff, '\\+\\+\\+ new/other/file\t')
462
self.assertContainsRe(diff, '-contents\n'
465
def test_renamed_directory(self):
466
"""Test when only a directory is only renamed."""
467
tree = self.make_branch_and_tree('tree')
468
self.build_tree(['tree/dir/'])
469
self.build_tree_contents([('tree/dir/file', 'contents\n')])
470
tree.add(['dir', 'dir/file'], ['dir-id', 'file-id'])
471
tree.commit('one', rev_id='rev-1')
473
tree.rename_one('dir', 'newdir')
474
diff = self.get_diff(tree.basis_tree(), tree)
475
# Renaming a directory should be a single "you renamed this dir" even
476
# when there are files inside.
477
self.assertEqual("=== renamed directory 'dir' => 'newdir'\n", diff)
479
def test_renamed_file(self):
480
"""Test when a file is only renamed."""
481
tree = self.make_branch_and_tree('tree')
482
self.build_tree_contents([('tree/file', 'contents\n')])
483
tree.add(['file'], ['file-id'])
484
tree.commit('one', rev_id='rev-1')
486
tree.rename_one('file', 'newname')
487
diff = self.get_diff(tree.basis_tree(), tree)
488
self.assertContainsRe(diff, "=== renamed file 'file' => 'newname'\n")
489
# We shouldn't have a --- or +++ line, because there is no content
491
self.assertNotContainsRe(diff, '---')
493
def test_renamed_and_modified_file(self):
494
"""Test when a file is only renamed."""
495
tree = self.make_branch_and_tree('tree')
496
self.build_tree_contents([('tree/file', 'contents\n')])
497
tree.add(['file'], ['file-id'])
498
tree.commit('one', rev_id='rev-1')
500
tree.rename_one('file', 'newname')
501
self.build_tree_contents([('tree/newname', 'new contents\n')])
502
diff = self.get_diff(tree.basis_tree(), tree)
503
self.assertContainsRe(diff, "=== renamed file 'file' => 'newname'\n")
504
self.assertContainsRe(diff, '--- old/file\t')
505
self.assertContainsRe(diff, '\\+\\+\\+ new/newname\t')
506
self.assertContainsRe(diff, '-contents\n'
509
def test_binary_unicode_filenames(self):
510
"""Test that contents of files are *not* encoded in UTF-8 when there
511
is a binary file in the diff.
513
# See https://bugs.launchpad.net/bugs/110092.
514
self.requireFeature(UnicodeFilename)
516
# This bug isn't triggered with cStringIO.
517
from StringIO import StringIO
518
tree = self.make_branch_and_tree('tree')
519
alpha, omega = u'\u03b1', u'\u03c9'
520
alpha_utf8, omega_utf8 = alpha.encode('utf8'), omega.encode('utf8')
521
self.build_tree_contents(
522
[('tree/' + alpha, chr(0)),
524
('The %s and the %s\n' % (alpha_utf8, omega_utf8)))])
525
tree.add([alpha], ['file-id'])
526
tree.add([omega], ['file-id-2'])
527
diff_content = StringIO()
528
show_diff_trees(tree.basis_tree(), tree, diff_content)
529
diff = diff_content.getvalue()
530
self.assertContainsRe(diff, r"=== added file '%s'" % alpha_utf8)
531
self.assertContainsRe(
532
diff, "Binary files a/%s.*and b/%s.* differ\n" % (alpha_utf8, alpha_utf8))
533
self.assertContainsRe(diff, r"=== added file '%s'" % omega_utf8)
534
self.assertContainsRe(diff, r"--- a/%s" % (omega_utf8,))
535
self.assertContainsRe(diff, r"\+\+\+ b/%s" % (omega_utf8,))
537
def test_unicode_filename(self):
538
"""Test when the filename are unicode."""
539
self.requireFeature(UnicodeFilename)
541
alpha, omega = u'\u03b1', u'\u03c9'
542
autf8, outf8 = alpha.encode('utf8'), omega.encode('utf8')
544
tree = self.make_branch_and_tree('tree')
545
self.build_tree_contents([('tree/ren_'+alpha, 'contents\n')])
546
tree.add(['ren_'+alpha], ['file-id-2'])
547
self.build_tree_contents([('tree/del_'+alpha, 'contents\n')])
548
tree.add(['del_'+alpha], ['file-id-3'])
549
self.build_tree_contents([('tree/mod_'+alpha, 'contents\n')])
550
tree.add(['mod_'+alpha], ['file-id-4'])
552
tree.commit('one', rev_id='rev-1')
554
tree.rename_one('ren_'+alpha, 'ren_'+omega)
555
tree.remove('del_'+alpha)
556
self.build_tree_contents([('tree/add_'+alpha, 'contents\n')])
557
tree.add(['add_'+alpha], ['file-id'])
558
self.build_tree_contents([('tree/mod_'+alpha, 'contents_mod\n')])
560
diff = self.get_diff(tree.basis_tree(), tree)
561
self.assertContainsRe(diff,
562
"=== renamed file 'ren_%s' => 'ren_%s'\n"%(autf8, outf8))
563
self.assertContainsRe(diff, "=== added file 'add_%s'"%autf8)
564
self.assertContainsRe(diff, "=== modified file 'mod_%s'"%autf8)
565
self.assertContainsRe(diff, "=== removed file 'del_%s'"%autf8)
568
class DiffWasIs(DiffPath):
570
def diff(self, file_id, old_path, new_path, old_kind, new_kind):
571
self.to_file.write('was: ')
572
self.to_file.write(self.old_tree.get_file(file_id).read())
573
self.to_file.write('is: ')
574
self.to_file.write(self.new_tree.get_file(file_id).read())
578
class TestDiffTree(TestCaseWithTransport):
581
TestCaseWithTransport.setUp(self)
582
self.old_tree = self.make_branch_and_tree('old-tree')
583
self.old_tree.lock_write()
584
self.addCleanup(self.old_tree.unlock)
585
self.new_tree = self.make_branch_and_tree('new-tree')
586
self.new_tree.lock_write()
587
self.addCleanup(self.new_tree.unlock)
588
self.differ = DiffTree(self.old_tree, self.new_tree, StringIO())
590
def test_diff_text(self):
591
self.build_tree_contents([('old-tree/olddir/',),
592
('old-tree/olddir/oldfile', 'old\n')])
593
self.old_tree.add('olddir')
594
self.old_tree.add('olddir/oldfile', 'file-id')
595
self.build_tree_contents([('new-tree/newdir/',),
596
('new-tree/newdir/newfile', 'new\n')])
597
self.new_tree.add('newdir')
598
self.new_tree.add('newdir/newfile', 'file-id')
599
differ = DiffText(self.old_tree, self.new_tree, StringIO())
600
differ.diff_text('file-id', None, 'old label', 'new label')
602
'--- old label\n+++ new label\n@@ -1,1 +0,0 @@\n-old\n\n',
603
differ.to_file.getvalue())
604
differ.to_file.seek(0)
605
differ.diff_text(None, 'file-id', 'old label', 'new label')
607
'--- old label\n+++ new label\n@@ -0,0 +1,1 @@\n+new\n\n',
608
differ.to_file.getvalue())
609
differ.to_file.seek(0)
610
differ.diff_text('file-id', 'file-id', 'old label', 'new label')
612
'--- old label\n+++ new label\n@@ -1,1 +1,1 @@\n-old\n+new\n\n',
613
differ.to_file.getvalue())
615
def test_diff_deletion(self):
616
self.build_tree_contents([('old-tree/file', 'contents'),
617
('new-tree/file', 'contents')])
618
self.old_tree.add('file', 'file-id')
619
self.new_tree.add('file', 'file-id')
620
os.unlink('new-tree/file')
621
self.differ.show_diff(None)
622
self.assertContainsRe(self.differ.to_file.getvalue(), '-contents')
624
def test_diff_creation(self):
625
self.build_tree_contents([('old-tree/file', 'contents'),
626
('new-tree/file', 'contents')])
627
self.old_tree.add('file', 'file-id')
628
self.new_tree.add('file', 'file-id')
629
os.unlink('old-tree/file')
630
self.differ.show_diff(None)
631
self.assertContainsRe(self.differ.to_file.getvalue(), '\+contents')
633
def test_diff_symlink(self):
634
differ = DiffSymlink(self.old_tree, self.new_tree, StringIO())
635
differ.diff_symlink('old target', None)
636
self.assertEqual("=== target was 'old target'\n",
637
differ.to_file.getvalue())
639
differ = DiffSymlink(self.old_tree, self.new_tree, StringIO())
640
differ.diff_symlink(None, 'new target')
641
self.assertEqual("=== target is 'new target'\n",
642
differ.to_file.getvalue())
644
differ = DiffSymlink(self.old_tree, self.new_tree, StringIO())
645
differ.diff_symlink('old target', 'new target')
646
self.assertEqual("=== target changed 'old target' => 'new target'\n",
647
differ.to_file.getvalue())
650
self.build_tree_contents([('old-tree/olddir/',),
651
('old-tree/olddir/oldfile', 'old\n')])
652
self.old_tree.add('olddir')
653
self.old_tree.add('olddir/oldfile', 'file-id')
654
self.build_tree_contents([('new-tree/newdir/',),
655
('new-tree/newdir/newfile', 'new\n')])
656
self.new_tree.add('newdir')
657
self.new_tree.add('newdir/newfile', 'file-id')
658
self.differ.diff('file-id', 'olddir/oldfile', 'newdir/newfile')
659
self.assertContainsRe(
660
self.differ.to_file.getvalue(),
661
r'--- olddir/oldfile.*\n\+\+\+ newdir/newfile.*\n\@\@ -1,1 \+1,1'
662
' \@\@\n-old\n\+new\n\n')
664
def test_diff_kind_change(self):
665
self.build_tree_contents([('old-tree/olddir/',),
666
('old-tree/olddir/oldfile', 'old\n')])
667
self.old_tree.add('olddir')
668
self.old_tree.add('olddir/oldfile', 'file-id')
669
self.build_tree(['new-tree/newdir/'])
670
os.symlink('new', 'new-tree/newdir/newfile')
671
self.new_tree.add('newdir')
672
self.new_tree.add('newdir/newfile', 'file-id')
673
self.differ.diff('file-id', 'olddir/oldfile', 'newdir/newfile')
674
self.assertContainsRe(
675
self.differ.to_file.getvalue(),
676
r'--- olddir/oldfile.*\n\+\+\+ newdir/newfile.*\n\@\@ -1,1 \+0,0'
678
self.assertContainsRe(self.differ.to_file.getvalue(),
679
"=== target is 'new'\n")
681
def test_diff_directory(self):
682
self.build_tree(['new-tree/new-dir/'])
683
self.new_tree.add('new-dir', 'new-dir-id')
684
self.differ.diff('new-dir-id', None, 'new-dir')
685
self.assertEqual(self.differ.to_file.getvalue(), '')
687
def create_old_new(self):
688
self.build_tree_contents([('old-tree/olddir/',),
689
('old-tree/olddir/oldfile', 'old\n')])
690
self.old_tree.add('olddir')
691
self.old_tree.add('olddir/oldfile', 'file-id')
692
self.build_tree_contents([('new-tree/newdir/',),
693
('new-tree/newdir/newfile', 'new\n')])
694
self.new_tree.add('newdir')
695
self.new_tree.add('newdir/newfile', 'file-id')
697
def test_register_diff(self):
698
self.create_old_new()
699
old_diff_factories = DiffTree.diff_factories
700
DiffTree.diff_factories=old_diff_factories[:]
701
DiffTree.diff_factories.insert(0, DiffWasIs.from_diff_tree)
703
differ = DiffTree(self.old_tree, self.new_tree, StringIO())
705
DiffTree.diff_factories = old_diff_factories
706
differ.diff('file-id', 'olddir/oldfile', 'newdir/newfile')
707
self.assertNotContainsRe(
708
differ.to_file.getvalue(),
709
r'--- olddir/oldfile.*\n\+\+\+ newdir/newfile.*\n\@\@ -1,1 \+1,1'
710
' \@\@\n-old\n\+new\n\n')
711
self.assertContainsRe(differ.to_file.getvalue(),
712
'was: old\nis: new\n')
714
def test_extra_factories(self):
715
self.create_old_new()
716
differ = DiffTree(self.old_tree, self.new_tree, StringIO(),
717
extra_factories=[DiffWasIs.from_diff_tree])
718
differ.diff('file-id', 'olddir/oldfile', 'newdir/newfile')
719
self.assertNotContainsRe(
720
differ.to_file.getvalue(),
721
r'--- olddir/oldfile.*\n\+\+\+ newdir/newfile.*\n\@\@ -1,1 \+1,1'
722
' \@\@\n-old\n\+new\n\n')
723
self.assertContainsRe(differ.to_file.getvalue(),
724
'was: old\nis: new\n')
727
class TestPatienceDiffLib(TestCase):
730
super(TestPatienceDiffLib, self).setUp()
731
self._unique_lcs = bzrlib._patiencediff_py.unique_lcs_py
732
self._recurse_matches = bzrlib._patiencediff_py.recurse_matches_py
733
self._PatienceSequenceMatcher = \
734
bzrlib._patiencediff_py.PatienceSequenceMatcher_py
736
def test_unique_lcs(self):
737
unique_lcs = self._unique_lcs
738
self.assertEquals(unique_lcs('', ''), [])
739
self.assertEquals(unique_lcs('', 'a'), [])
740
self.assertEquals(unique_lcs('a', ''), [])
741
self.assertEquals(unique_lcs('a', 'a'), [(0,0)])
742
self.assertEquals(unique_lcs('a', 'b'), [])
743
self.assertEquals(unique_lcs('ab', 'ab'), [(0,0), (1,1)])
744
self.assertEquals(unique_lcs('abcde', 'cdeab'), [(2,0), (3,1), (4,2)])
745
self.assertEquals(unique_lcs('cdeab', 'abcde'), [(0,2), (1,3), (2,4)])
746
self.assertEquals(unique_lcs('abXde', 'abYde'), [(0,0), (1,1),
748
self.assertEquals(unique_lcs('acbac', 'abc'), [(2,1)])
750
def test_recurse_matches(self):
751
def test_one(a, b, matches):
753
self._recurse_matches(
754
a, b, 0, 0, len(a), len(b), test_matches, 10)
755
self.assertEquals(test_matches, matches)
757
test_one(['a', '', 'b', '', 'c'], ['a', 'a', 'b', 'c', 'c'],
758
[(0, 0), (2, 2), (4, 4)])
759
test_one(['a', 'c', 'b', 'a', 'c'], ['a', 'b', 'c'],
760
[(0, 0), (2, 1), (4, 2)])
761
# Even though 'bc' is not unique globally, and is surrounded by
762
# non-matching lines, we should still match, because they are locally
764
test_one('abcdbce', 'afbcgdbce', [(0,0), (1, 2), (2, 3), (3, 5),
765
(4, 6), (5, 7), (6, 8)])
767
# recurse_matches doesn't match non-unique
768
# lines surrounded by bogus text.
769
# The update has been done in patiencediff.SequenceMatcher instead
771
# This is what it could be
772
#test_one('aBccDe', 'abccde', [(0,0), (2,2), (3,3), (5,5)])
774
# This is what it currently gives:
775
test_one('aBccDe', 'abccde', [(0,0), (5,5)])
777
def test_matching_blocks(self):
778
def chk_blocks(a, b, expected_blocks):
779
# difflib always adds a signature of the total
780
# length, with no matching entries at the end
781
s = self._PatienceSequenceMatcher(None, a, b)
782
blocks = s.get_matching_blocks()
783
self.assertEquals((len(a), len(b), 0), blocks[-1])
784
self.assertEquals(expected_blocks, blocks[:-1])
786
# Some basic matching tests
787
chk_blocks('', '', [])
788
chk_blocks([], [], [])
789
chk_blocks('abc', '', [])
790
chk_blocks('', 'abc', [])
791
chk_blocks('abcd', 'abcd', [(0, 0, 4)])
792
chk_blocks('abcd', 'abce', [(0, 0, 3)])
793
chk_blocks('eabc', 'abce', [(1, 0, 3)])
794
chk_blocks('eabce', 'abce', [(1, 0, 4)])
795
chk_blocks('abcde', 'abXde', [(0, 0, 2), (3, 3, 2)])
796
chk_blocks('abcde', 'abXYZde', [(0, 0, 2), (3, 5, 2)])
797
chk_blocks('abde', 'abXYZde', [(0, 0, 2), (2, 5, 2)])
798
# This may check too much, but it checks to see that
799
# a copied block stays attached to the previous section,
801
# difflib would tend to grab the trailing longest match
802
# which would make the diff not look right
803
chk_blocks('abcdefghijklmnop', 'abcdefxydefghijklmnop',
804
[(0, 0, 6), (6, 11, 10)])
806
# make sure it supports passing in lists
810
'how are you today?\n'],
812
'how are you today?\n'],
813
[(0, 0, 1), (2, 1, 1)])
815
# non unique lines surrounded by non-matching lines
817
chk_blocks('aBccDe', 'abccde', [(0,0,1), (5,5,1)])
819
# But they only need to be locally unique
820
chk_blocks('aBcDec', 'abcdec', [(0,0,1), (2,2,1), (4,4,2)])
822
# non unique blocks won't be matched
823
chk_blocks('aBcdEcdFg', 'abcdecdfg', [(0,0,1), (8,8,1)])
825
# but locally unique ones will
826
chk_blocks('aBcdEeXcdFg', 'abcdecdfg', [(0,0,1), (2,2,2),
827
(5,4,1), (7,5,2), (10,8,1)])
829
chk_blocks('abbabbXd', 'cabbabxd', [(7,7,1)])
830
chk_blocks('abbabbbb', 'cabbabbc', [])
831
chk_blocks('bbbbbbbb', 'cbbbbbbc', [])
833
def test_opcodes(self):
834
def chk_ops(a, b, expected_codes):
835
s = self._PatienceSequenceMatcher(None, a, b)
836
self.assertEquals(expected_codes, s.get_opcodes())
840
chk_ops('abc', '', [('delete', 0,3, 0,0)])
841
chk_ops('', 'abc', [('insert', 0,0, 0,3)])
842
chk_ops('abcd', 'abcd', [('equal', 0,4, 0,4)])
843
chk_ops('abcd', 'abce', [('equal', 0,3, 0,3),
844
('replace', 3,4, 3,4)
846
chk_ops('eabc', 'abce', [('delete', 0,1, 0,0),
850
chk_ops('eabce', 'abce', [('delete', 0,1, 0,0),
853
chk_ops('abcde', 'abXde', [('equal', 0,2, 0,2),
854
('replace', 2,3, 2,3),
857
chk_ops('abcde', 'abXYZde', [('equal', 0,2, 0,2),
858
('replace', 2,3, 2,5),
861
chk_ops('abde', 'abXYZde', [('equal', 0,2, 0,2),
862
('insert', 2,2, 2,5),
865
chk_ops('abcdefghijklmnop', 'abcdefxydefghijklmnop',
866
[('equal', 0,6, 0,6),
867
('insert', 6,6, 6,11),
868
('equal', 6,16, 11,21)
873
, 'how are you today?\n'],
875
, 'how are you today?\n'],
876
[('equal', 0,1, 0,1),
877
('delete', 1,2, 1,1),
880
chk_ops('aBccDe', 'abccde',
881
[('equal', 0,1, 0,1),
882
('replace', 1,5, 1,5),
885
chk_ops('aBcDec', 'abcdec',
886
[('equal', 0,1, 0,1),
887
('replace', 1,2, 1,2),
889
('replace', 3,4, 3,4),
892
chk_ops('aBcdEcdFg', 'abcdecdfg',
893
[('equal', 0,1, 0,1),
894
('replace', 1,8, 1,8),
897
chk_ops('aBcdEeXcdFg', 'abcdecdfg',
898
[('equal', 0,1, 0,1),
899
('replace', 1,2, 1,2),
901
('delete', 4,5, 4,4),
903
('delete', 6,7, 5,5),
905
('replace', 9,10, 7,8),
906
('equal', 10,11, 8,9)
909
def test_grouped_opcodes(self):
910
def chk_ops(a, b, expected_codes, n=3):
911
s = self._PatienceSequenceMatcher(None, a, b)
912
self.assertEquals(expected_codes, list(s.get_grouped_opcodes(n)))
916
chk_ops('abc', '', [[('delete', 0,3, 0,0)]])
917
chk_ops('', 'abc', [[('insert', 0,0, 0,3)]])
918
chk_ops('abcd', 'abcd', [])
919
chk_ops('abcd', 'abce', [[('equal', 0,3, 0,3),
920
('replace', 3,4, 3,4)
922
chk_ops('eabc', 'abce', [[('delete', 0,1, 0,0),
926
chk_ops('abcdefghijklmnop', 'abcdefxydefghijklmnop',
927
[[('equal', 3,6, 3,6),
928
('insert', 6,6, 6,11),
929
('equal', 6,9, 11,14)
931
chk_ops('abcdefghijklmnop', 'abcdefxydefghijklmnop',
932
[[('equal', 2,6, 2,6),
933
('insert', 6,6, 6,11),
934
('equal', 6,10, 11,15)
936
chk_ops('Xabcdef', 'abcdef',
937
[[('delete', 0,1, 0,0),
940
chk_ops('abcdef', 'abcdefX',
941
[[('equal', 3,6, 3,6),
946
def test_multiple_ranges(self):
947
# There was an earlier bug where we used a bad set of ranges,
948
# this triggers that specific bug, to make sure it doesn't regress
949
def chk_blocks(a, b, expected_blocks):
950
# difflib always adds a signature of the total
951
# length, with no matching entries at the end
952
s = self._PatienceSequenceMatcher(None, a, b)
953
blocks = s.get_matching_blocks()
955
self.assertEquals(x, (len(a), len(b), 0))
956
self.assertEquals(expected_blocks, blocks)
958
chk_blocks('abcdefghijklmnop'
959
, 'abcXghiYZQRSTUVWXYZijklmnop'
960
, [(0, 0, 3), (6, 4, 3), (9, 20, 7)])
962
chk_blocks('ABCd efghIjk L'
963
, 'AxyzBCn mo pqrstuvwI1 2 L'
964
, [(0,0,1), (1, 4, 2), (9, 19, 1), (12, 23, 3)])
966
# These are rot13 code snippets.
968
trg nqqrq jura lbh nqq n svyr va gur qverpgbel.
970
gnxrf_netf = ['svyr*']
971
gnxrf_bcgvbaf = ['ab-erphefr']
973
qrs eha(frys, svyr_yvfg, ab_erphefr=Snyfr):
974
sebz omeyvo.nqq vzcbeg fzneg_nqq, nqq_ercbegre_cevag, nqq_ercbegre_ahyy
976
ercbegre = nqq_ercbegre_ahyy
978
ercbegre = nqq_ercbegre_cevag
979
fzneg_nqq(svyr_yvfg, abg ab_erphefr, ercbegre)
982
pynff pzq_zxqve(Pbzznaq):
983
'''.splitlines(True), '''\
984
trg nqqrq jura lbh nqq n svyr va gur qverpgbel.
986
--qel-eha jvyy fubj juvpu svyrf jbhyq or nqqrq, ohg abg npghnyyl
989
gnxrf_netf = ['svyr*']
990
gnxrf_bcgvbaf = ['ab-erphefr', 'qel-eha']
992
qrs eha(frys, svyr_yvfg, ab_erphefr=Snyfr, qel_eha=Snyfr):
997
# Guvf vf cbvagyrff, ohg V'q engure abg envfr na reebe
998
npgvba = omeyvo.nqq.nqq_npgvba_ahyy
1000
npgvba = omeyvo.nqq.nqq_npgvba_cevag
1002
npgvba = omeyvo.nqq.nqq_npgvba_nqq
1004
npgvba = omeyvo.nqq.nqq_npgvba_nqq_naq_cevag
1006
omeyvo.nqq.fzneg_nqq(svyr_yvfg, abg ab_erphefr, npgvba)
1009
pynff pzq_zxqve(Pbzznaq):
1010
'''.splitlines(True)
1011
, [(0,0,1), (1, 4, 2), (9, 19, 1), (12, 23, 3)])
1013
def test_patience_unified_diff(self):
1014
txt_a = ['hello there\n',
1016
'how are you today?\n']
1017
txt_b = ['hello there\n',
1018
'how are you today?\n']
1019
unified_diff = bzrlib.patiencediff.unified_diff
1020
psm = self._PatienceSequenceMatcher
1021
self.assertEquals([ '--- \n',
1023
'@@ -1,3 +1,2 @@\n',
1026
' how are you today?\n'
1028
, list(unified_diff(txt_a, txt_b,
1029
sequencematcher=psm)))
1030
txt_a = map(lambda x: x+'\n', 'abcdefghijklmnop')
1031
txt_b = map(lambda x: x+'\n', 'abcdefxydefghijklmnop')
1032
# This is the result with LongestCommonSubstring matching
1033
self.assertEquals(['--- \n',
1035
'@@ -1,6 +1,11 @@\n',
1047
, list(unified_diff(txt_a, txt_b)))
1048
# And the patience diff
1049
self.assertEquals(['--- \n',
1051
'@@ -4,6 +4,11 @@\n',
1064
, list(unified_diff(txt_a, txt_b,
1065
sequencematcher=psm)))
1068
class TestPatienceDiffLib_c(TestPatienceDiffLib):
1070
_test_needs_features = [CompiledPatienceDiffFeature]
1073
super(TestPatienceDiffLib_c, self).setUp()
1074
import bzrlib._patiencediff_c
1075
self._unique_lcs = bzrlib._patiencediff_c.unique_lcs_c
1076
self._recurse_matches = bzrlib._patiencediff_c.recurse_matches_c
1077
self._PatienceSequenceMatcher = \
1078
bzrlib._patiencediff_c.PatienceSequenceMatcher_c
1081
class TestPatienceDiffLibFiles(TestCaseInTempDir):
1084
super(TestPatienceDiffLibFiles, self).setUp()
1085
self._PatienceSequenceMatcher = \
1086
bzrlib._patiencediff_py.PatienceSequenceMatcher_py
1088
def test_patience_unified_diff_files(self):
1089
txt_a = ['hello there\n',
1091
'how are you today?\n']
1092
txt_b = ['hello there\n',
1093
'how are you today?\n']
1094
open('a1', 'wb').writelines(txt_a)
1095
open('b1', 'wb').writelines(txt_b)
1097
unified_diff_files = bzrlib.patiencediff.unified_diff_files
1098
psm = self._PatienceSequenceMatcher
1099
self.assertEquals(['--- a1 \n',
1101
'@@ -1,3 +1,2 @@\n',
1104
' how are you today?\n',
1106
, list(unified_diff_files('a1', 'b1',
1107
sequencematcher=psm)))
1109
txt_a = map(lambda x: x+'\n', 'abcdefghijklmnop')
1110
txt_b = map(lambda x: x+'\n', 'abcdefxydefghijklmnop')
1111
open('a2', 'wb').writelines(txt_a)
1112
open('b2', 'wb').writelines(txt_b)
1114
# This is the result with LongestCommonSubstring matching
1115
self.assertEquals(['--- a2 \n',
1117
'@@ -1,6 +1,11 @@\n',
1129
, list(unified_diff_files('a2', 'b2')))
1131
# And the patience diff
1132
self.assertEquals(['--- a2 \n',
1134
'@@ -4,6 +4,11 @@\n',
1147
, list(unified_diff_files('a2', 'b2',
1148
sequencematcher=psm)))
1151
class TestPatienceDiffLibFiles_c(TestPatienceDiffLibFiles):
1153
_test_needs_features = [CompiledPatienceDiffFeature]
1156
super(TestPatienceDiffLibFiles_c, self).setUp()
1157
import bzrlib._patiencediff_c
1158
self._PatienceSequenceMatcher = \
1159
bzrlib._patiencediff_c.PatienceSequenceMatcher_c
1162
class TestUsingCompiledIfAvailable(TestCase):
1164
def test_PatienceSequenceMatcher(self):
1165
if CompiledPatienceDiffFeature.available():
1166
from bzrlib._patiencediff_c import PatienceSequenceMatcher_c
1167
self.assertIs(PatienceSequenceMatcher_c,
1168
bzrlib.patiencediff.PatienceSequenceMatcher)
1170
from bzrlib._patiencediff_py import PatienceSequenceMatcher_py
1171
self.assertIs(PatienceSequenceMatcher_py,
1172
bzrlib.patiencediff.PatienceSequenceMatcher)
1174
def test_unique_lcs(self):
1175
if CompiledPatienceDiffFeature.available():
1176
from bzrlib._patiencediff_c import unique_lcs_c
1177
self.assertIs(unique_lcs_c,
1178
bzrlib.patiencediff.unique_lcs)
1180
from bzrlib._patiencediff_py import unique_lcs_py
1181
self.assertIs(unique_lcs_py,
1182
bzrlib.patiencediff.unique_lcs)
1184
def test_recurse_matches(self):
1185
if CompiledPatienceDiffFeature.available():
1186
from bzrlib._patiencediff_c import recurse_matches_c
1187
self.assertIs(recurse_matches_c,
1188
bzrlib.patiencediff.recurse_matches)
1190
from bzrlib._patiencediff_py import recurse_matches_py
1191
self.assertIs(recurse_matches_py,
1192
bzrlib.patiencediff.recurse_matches)
added
removed
bzrlib/tests/test_diff.py
