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_returns_bytes(self):
244
output = StringIO.StringIO()
245
internal_diff(u'old_\xb5', ['old_text\n'],
246
u'new_\xe5', ['new_text\n'], output)
247
self.failUnless(isinstance(output.getvalue(), str),
248
'internal_diff should return bytestrings')
251
class TestDiffFiles(TestCaseInTempDir):
253
def test_external_diff_binary(self):
254
"""The output when using external diff should use diff's i18n error"""
255
# Make sure external_diff doesn't fail in the current LANG
256
lines = external_udiff_lines(['\x00foobar\n'], ['foo\x00bar\n'])
258
cmd = ['diff', '-u', '--binary', 'old', 'new']
259
open('old', 'wb').write('\x00foobar\n')
260
open('new', 'wb').write('foo\x00bar\n')
261
pipe = subprocess.Popen(cmd, stdout=subprocess.PIPE,
262
stdin=subprocess.PIPE)
263
out, err = pipe.communicate()
264
# Diff returns '2' on Binary files.
265
self.assertEqual(2, pipe.returncode)
266
# We should output whatever diff tells us, plus a trailing newline
267
self.assertEqual(out.splitlines(True) + ['\n'], lines)
270
class TestShowDiffTreesHelper(TestCaseWithTransport):
271
"""Has a helper for running show_diff_trees"""
273
def get_diff(self, tree1, tree2, specific_files=None, working_tree=None):
275
if working_tree is not None:
276
extra_trees = (working_tree,)
279
show_diff_trees(tree1, tree2, output, specific_files=specific_files,
280
extra_trees=extra_trees, old_label='old/',
282
return output.getvalue()
285
class TestDiffDates(TestShowDiffTreesHelper):
288
super(TestDiffDates, self).setUp()
289
self.wt = self.make_branch_and_tree('.')
290
self.b = self.wt.branch
291
self.build_tree_contents([
292
('file1', 'file1 contents at rev 1\n'),
293
('file2', 'file2 contents at rev 1\n')
295
self.wt.add(['file1', 'file2'])
297
message='Revision 1',
298
timestamp=1143849600, # 2006-04-01 00:00:00 UTC
301
self.build_tree_contents([('file1', 'file1 contents at rev 2\n')])
303
message='Revision 2',
304
timestamp=1143936000, # 2006-04-02 00:00:00 UTC
307
self.build_tree_contents([('file2', 'file2 contents at rev 3\n')])
309
message='Revision 3',
310
timestamp=1144022400, # 2006-04-03 00:00:00 UTC
313
self.wt.remove(['file2'])
315
message='Revision 4',
316
timestamp=1144108800, # 2006-04-04 00:00:00 UTC
319
self.build_tree_contents([
320
('file1', 'file1 contents in working tree\n')
322
# set the date stamps for files in the working tree to known values
323
os.utime('file1', (1144195200, 1144195200)) # 2006-04-05 00:00:00 UTC
325
def test_diff_rev_tree_working_tree(self):
326
output = self.get_diff(self.wt.basis_tree(), self.wt)
327
# note that the date for old/file1 is from rev 2 rather than from
328
# the basis revision (rev 4)
329
self.assertEqualDiff(output, '''\
330
=== modified file 'file1'
331
--- old/file1\t2006-04-02 00:00:00 +0000
332
+++ new/file1\t2006-04-05 00:00:00 +0000
334
-file1 contents at rev 2
335
+file1 contents in working tree
339
def test_diff_rev_tree_rev_tree(self):
340
tree1 = self.b.repository.revision_tree('rev-2')
341
tree2 = self.b.repository.revision_tree('rev-3')
342
output = self.get_diff(tree1, tree2)
343
self.assertEqualDiff(output, '''\
344
=== modified file 'file2'
345
--- old/file2\t2006-04-01 00:00:00 +0000
346
+++ new/file2\t2006-04-03 00:00:00 +0000
348
-file2 contents at rev 1
349
+file2 contents at rev 3
353
def test_diff_add_files(self):
354
tree1 = self.b.repository.revision_tree(None)
355
tree2 = self.b.repository.revision_tree('rev-1')
356
output = self.get_diff(tree1, tree2)
357
# the files have the epoch time stamp for the tree in which
359
self.assertEqualDiff(output, '''\
360
=== added file 'file1'
361
--- old/file1\t1970-01-01 00:00:00 +0000
362
+++ new/file1\t2006-04-01 00:00:00 +0000
364
+file1 contents at rev 1
366
=== added file 'file2'
367
--- old/file2\t1970-01-01 00:00:00 +0000
368
+++ new/file2\t2006-04-01 00:00:00 +0000
370
+file2 contents at rev 1
374
def test_diff_remove_files(self):
375
tree1 = self.b.repository.revision_tree('rev-3')
376
tree2 = self.b.repository.revision_tree('rev-4')
377
output = self.get_diff(tree1, tree2)
378
# the file has the epoch time stamp for the tree in which
380
self.assertEqualDiff(output, '''\
381
=== removed file 'file2'
382
--- old/file2\t2006-04-03 00:00:00 +0000
383
+++ new/file2\t1970-01-01 00:00:00 +0000
385
-file2 contents at rev 3
389
def test_show_diff_specified(self):
390
"""A working tree filename can be used to identify a file"""
391
self.wt.rename_one('file1', 'file1b')
392
old_tree = self.b.repository.revision_tree('rev-1')
393
new_tree = self.b.repository.revision_tree('rev-4')
394
out = self.get_diff(old_tree, new_tree, specific_files=['file1b'],
395
working_tree=self.wt)
396
self.assertContainsRe(out, 'file1\t')
398
def test_recursive_diff(self):
399
"""Children of directories are matched"""
402
self.wt.add(['dir1', 'dir2'])
403
self.wt.rename_one('file1', 'dir1/file1')
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=['dir1'],
407
working_tree=self.wt)
408
self.assertContainsRe(out, 'file1\t')
409
out = self.get_diff(old_tree, new_tree, specific_files=['dir2'],
410
working_tree=self.wt)
411
self.assertNotContainsRe(out, 'file1\t')
415
class TestShowDiffTrees(TestShowDiffTreesHelper):
416
"""Direct tests for show_diff_trees"""
418
def test_modified_file(self):
419
"""Test when a file is modified."""
420
tree = self.make_branch_and_tree('tree')
421
self.build_tree_contents([('tree/file', 'contents\n')])
422
tree.add(['file'], ['file-id'])
423
tree.commit('one', rev_id='rev-1')
425
self.build_tree_contents([('tree/file', 'new contents\n')])
426
diff = self.get_diff(tree.basis_tree(), tree)
427
self.assertContainsRe(diff, "=== modified file 'file'\n")
428
self.assertContainsRe(diff, '--- old/file\t')
429
self.assertContainsRe(diff, '\\+\\+\\+ new/file\t')
430
self.assertContainsRe(diff, '-contents\n'
433
def test_modified_file_in_renamed_dir(self):
434
"""Test when a file is modified in a renamed directory."""
435
tree = self.make_branch_and_tree('tree')
436
self.build_tree(['tree/dir/'])
437
self.build_tree_contents([('tree/dir/file', 'contents\n')])
438
tree.add(['dir', 'dir/file'], ['dir-id', 'file-id'])
439
tree.commit('one', rev_id='rev-1')
441
tree.rename_one('dir', 'other')
442
self.build_tree_contents([('tree/other/file', 'new contents\n')])
443
diff = self.get_diff(tree.basis_tree(), tree)
444
self.assertContainsRe(diff, "=== renamed directory 'dir' => 'other'\n")
445
self.assertContainsRe(diff, "=== modified file 'other/file'\n")
446
# XXX: This is technically incorrect, because it used to be at another
447
# location. What to do?
448
self.assertContainsRe(diff, '--- old/dir/file\t')
449
self.assertContainsRe(diff, '\\+\\+\\+ new/other/file\t')
450
self.assertContainsRe(diff, '-contents\n'
453
def test_renamed_directory(self):
454
"""Test when only a directory is only renamed."""
455
tree = self.make_branch_and_tree('tree')
456
self.build_tree(['tree/dir/'])
457
self.build_tree_contents([('tree/dir/file', 'contents\n')])
458
tree.add(['dir', 'dir/file'], ['dir-id', 'file-id'])
459
tree.commit('one', rev_id='rev-1')
461
tree.rename_one('dir', 'newdir')
462
diff = self.get_diff(tree.basis_tree(), tree)
463
# Renaming a directory should be a single "you renamed this dir" even
464
# when there are files inside.
465
self.assertEqual("=== renamed directory 'dir' => 'newdir'\n", diff)
467
def test_renamed_file(self):
468
"""Test when a file is only renamed."""
469
tree = self.make_branch_and_tree('tree')
470
self.build_tree_contents([('tree/file', 'contents\n')])
471
tree.add(['file'], ['file-id'])
472
tree.commit('one', rev_id='rev-1')
474
tree.rename_one('file', 'newname')
475
diff = self.get_diff(tree.basis_tree(), tree)
476
self.assertContainsRe(diff, "=== renamed file 'file' => 'newname'\n")
477
# We shouldn't have a --- or +++ line, because there is no content
479
self.assertNotContainsRe(diff, '---')
481
def test_renamed_and_modified_file(self):
482
"""Test when a file is only renamed."""
483
tree = self.make_branch_and_tree('tree')
484
self.build_tree_contents([('tree/file', 'contents\n')])
485
tree.add(['file'], ['file-id'])
486
tree.commit('one', rev_id='rev-1')
488
tree.rename_one('file', 'newname')
489
self.build_tree_contents([('tree/newname', 'new contents\n')])
490
diff = self.get_diff(tree.basis_tree(), tree)
491
self.assertContainsRe(diff, "=== renamed file 'file' => 'newname'\n")
492
self.assertContainsRe(diff, '--- old/file\t')
493
self.assertContainsRe(diff, '\\+\\+\\+ new/newname\t')
494
self.assertContainsRe(diff, '-contents\n'
497
def test_binary_unicode_filenames(self):
498
"""Test that contents of files are *not* encoded in UTF-8 when there
499
is a binary file in the diff.
501
# See https://bugs.launchpad.net/bugs/110092.
502
self.requireFeature(UnicodeFilename)
504
# This bug isn't triggered with cStringIO.
505
from StringIO import StringIO
506
tree = self.make_branch_and_tree('tree')
507
alpha, omega = u'\u03b1', u'\u03c9'
508
alpha_utf8, omega_utf8 = alpha.encode('utf8'), omega.encode('utf8')
509
self.build_tree_contents(
510
[('tree/' + alpha, chr(0)),
512
('The %s and the %s\n' % (alpha_utf8, omega_utf8)))])
513
tree.add([alpha], ['file-id'])
514
tree.add([omega], ['file-id-2'])
515
diff_content = StringIO()
516
show_diff_trees(tree.basis_tree(), tree, diff_content)
517
diff = diff_content.getvalue()
518
self.assertContainsRe(diff, r"=== added file '%s'" % alpha_utf8)
519
self.assertContainsRe(
520
diff, "Binary files a/%s.*and b/%s.* differ\n" % (alpha_utf8, alpha_utf8))
521
self.assertContainsRe(diff, r"=== added file '%s'" % omega_utf8)
522
self.assertContainsRe(diff, r"--- a/%s" % (omega_utf8,))
523
self.assertContainsRe(diff, r"\+\+\+ b/%s" % (omega_utf8,))
525
def test_unicode_filename(self):
526
"""Test when the filename are unicode."""
527
self.requireFeature(UnicodeFilename)
529
alpha, omega = u'\u03b1', u'\u03c9'
530
autf8, outf8 = alpha.encode('utf8'), omega.encode('utf8')
532
tree = self.make_branch_and_tree('tree')
533
self.build_tree_contents([('tree/ren_'+alpha, 'contents\n')])
534
tree.add(['ren_'+alpha], ['file-id-2'])
535
self.build_tree_contents([('tree/del_'+alpha, 'contents\n')])
536
tree.add(['del_'+alpha], ['file-id-3'])
537
self.build_tree_contents([('tree/mod_'+alpha, 'contents\n')])
538
tree.add(['mod_'+alpha], ['file-id-4'])
540
tree.commit('one', rev_id='rev-1')
542
tree.rename_one('ren_'+alpha, 'ren_'+omega)
543
tree.remove('del_'+alpha)
544
self.build_tree_contents([('tree/add_'+alpha, 'contents\n')])
545
tree.add(['add_'+alpha], ['file-id'])
546
self.build_tree_contents([('tree/mod_'+alpha, 'contents_mod\n')])
548
diff = self.get_diff(tree.basis_tree(), tree)
549
self.assertContainsRe(diff,
550
"=== renamed file 'ren_%s' => 'ren_%s'\n"%(autf8, outf8))
551
self.assertContainsRe(diff, "=== added file 'add_%s'"%autf8)
552
self.assertContainsRe(diff, "=== modified file 'mod_%s'"%autf8)
553
self.assertContainsRe(diff, "=== removed file 'del_%s'"%autf8)
556
class DiffWasIs(DiffPath):
558
def diff(self, file_id, old_path, new_path, old_kind, new_kind):
559
self.to_file.write('was: ')
560
self.to_file.write(self.old_tree.get_file(file_id).read())
561
self.to_file.write('is: ')
562
self.to_file.write(self.new_tree.get_file(file_id).read())
566
class TestDiffTree(TestCaseWithTransport):
569
TestCaseWithTransport.setUp(self)
570
self.old_tree = self.make_branch_and_tree('old-tree')
571
self.old_tree.lock_write()
572
self.addCleanup(self.old_tree.unlock)
573
self.new_tree = self.make_branch_and_tree('new-tree')
574
self.new_tree.lock_write()
575
self.addCleanup(self.new_tree.unlock)
576
self.differ = DiffTree(self.old_tree, self.new_tree, StringIO())
578
def test_diff_text(self):
579
self.build_tree_contents([('old-tree/olddir/',),
580
('old-tree/olddir/oldfile', 'old\n')])
581
self.old_tree.add('olddir')
582
self.old_tree.add('olddir/oldfile', 'file-id')
583
self.build_tree_contents([('new-tree/newdir/',),
584
('new-tree/newdir/newfile', 'new\n')])
585
self.new_tree.add('newdir')
586
self.new_tree.add('newdir/newfile', 'file-id')
587
differ = DiffText(self.old_tree, self.new_tree, StringIO())
588
differ.diff_text('file-id', None, 'old label', 'new label')
590
'--- old label\n+++ new label\n@@ -1,1 +0,0 @@\n-old\n\n',
591
differ.to_file.getvalue())
592
differ.to_file.seek(0)
593
differ.diff_text(None, 'file-id', 'old label', 'new label')
595
'--- old label\n+++ new label\n@@ -0,0 +1,1 @@\n+new\n\n',
596
differ.to_file.getvalue())
597
differ.to_file.seek(0)
598
differ.diff_text('file-id', 'file-id', 'old label', 'new label')
600
'--- old label\n+++ new label\n@@ -1,1 +1,1 @@\n-old\n+new\n\n',
601
differ.to_file.getvalue())
603
def test_diff_symlink(self):
604
differ = DiffSymlink(self.old_tree, self.new_tree, StringIO())
605
differ.diff_symlink('old target', None)
606
self.assertEqual("=== target was 'old target'\n",
607
differ.to_file.getvalue())
609
differ = DiffSymlink(self.old_tree, self.new_tree, StringIO())
610
differ.diff_symlink(None, 'new target')
611
self.assertEqual("=== target is 'new target'\n",
612
differ.to_file.getvalue())
614
differ = DiffSymlink(self.old_tree, self.new_tree, StringIO())
615
differ.diff_symlink('old target', 'new target')
616
self.assertEqual("=== target changed 'old target' => 'new target'\n",
617
differ.to_file.getvalue())
620
self.build_tree_contents([('old-tree/olddir/',),
621
('old-tree/olddir/oldfile', 'old\n')])
622
self.old_tree.add('olddir')
623
self.old_tree.add('olddir/oldfile', 'file-id')
624
self.build_tree_contents([('new-tree/newdir/',),
625
('new-tree/newdir/newfile', 'new\n')])
626
self.new_tree.add('newdir')
627
self.new_tree.add('newdir/newfile', 'file-id')
628
self.differ.diff('file-id', 'olddir/oldfile', 'newdir/newfile')
629
self.assertContainsRe(
630
self.differ.to_file.getvalue(),
631
r'--- olddir/oldfile.*\n\+\+\+ newdir/newfile.*\n\@\@ -1,1 \+1,1'
632
' \@\@\n-old\n\+new\n\n')
634
def test_diff_kind_change(self):
635
self.build_tree_contents([('old-tree/olddir/',),
636
('old-tree/olddir/oldfile', 'old\n')])
637
self.old_tree.add('olddir')
638
self.old_tree.add('olddir/oldfile', 'file-id')
639
self.build_tree(['new-tree/newdir/'])
640
os.symlink('new', 'new-tree/newdir/newfile')
641
self.new_tree.add('newdir')
642
self.new_tree.add('newdir/newfile', 'file-id')
643
self.differ.diff('file-id', 'olddir/oldfile', 'newdir/newfile')
644
self.assertContainsRe(
645
self.differ.to_file.getvalue(),
646
r'--- olddir/oldfile.*\n\+\+\+ newdir/newfile.*\n\@\@ -1,1 \+0,0'
648
self.assertContainsRe(self.differ.to_file.getvalue(),
649
"=== target is 'new'\n")
651
def test_diff_directory(self):
652
self.build_tree(['new-tree/new-dir/'])
653
self.new_tree.add('new-dir', 'new-dir-id')
654
self.differ.diff('new-dir-id', None, 'new-dir')
655
self.assertEqual(self.differ.to_file.getvalue(), '')
657
def create_old_new(self):
658
self.build_tree_contents([('old-tree/olddir/',),
659
('old-tree/olddir/oldfile', 'old\n')])
660
self.old_tree.add('olddir')
661
self.old_tree.add('olddir/oldfile', 'file-id')
662
self.build_tree_contents([('new-tree/newdir/',),
663
('new-tree/newdir/newfile', 'new\n')])
664
self.new_tree.add('newdir')
665
self.new_tree.add('newdir/newfile', 'file-id')
667
def test_register_diff(self):
668
self.create_old_new()
669
old_diff_factories = DiffTree.diff_factories
670
DiffTree.diff_factories=old_diff_factories[:]
671
DiffTree.diff_factories.insert(0, DiffWasIs.from_diff_tree)
673
differ = DiffTree(self.old_tree, self.new_tree, StringIO())
675
DiffTree.diff_factories = old_diff_factories
676
differ.diff('file-id', 'olddir/oldfile', 'newdir/newfile')
677
self.assertNotContainsRe(
678
differ.to_file.getvalue(),
679
r'--- olddir/oldfile.*\n\+\+\+ newdir/newfile.*\n\@\@ -1,1 \+1,1'
680
' \@\@\n-old\n\+new\n\n')
681
self.assertContainsRe(differ.to_file.getvalue(),
682
'was: old\nis: new\n')
684
def test_extra_factories(self):
685
self.create_old_new()
686
differ = DiffTree(self.old_tree, self.new_tree, StringIO(),
687
extra_factories=[DiffWasIs.from_diff_tree])
688
differ.diff('file-id', 'olddir/oldfile', 'newdir/newfile')
689
self.assertNotContainsRe(
690
differ.to_file.getvalue(),
691
r'--- olddir/oldfile.*\n\+\+\+ newdir/newfile.*\n\@\@ -1,1 \+1,1'
692
' \@\@\n-old\n\+new\n\n')
693
self.assertContainsRe(differ.to_file.getvalue(),
694
'was: old\nis: new\n')
697
class TestPatienceDiffLib(TestCase):
700
super(TestPatienceDiffLib, self).setUp()
701
self._unique_lcs = bzrlib._patiencediff_py.unique_lcs_py
702
self._recurse_matches = bzrlib._patiencediff_py.recurse_matches_py
703
self._PatienceSequenceMatcher = \
704
bzrlib._patiencediff_py.PatienceSequenceMatcher_py
706
def test_unique_lcs(self):
707
unique_lcs = self._unique_lcs
708
self.assertEquals(unique_lcs('', ''), [])
709
self.assertEquals(unique_lcs('', 'a'), [])
710
self.assertEquals(unique_lcs('a', ''), [])
711
self.assertEquals(unique_lcs('a', 'a'), [(0,0)])
712
self.assertEquals(unique_lcs('a', 'b'), [])
713
self.assertEquals(unique_lcs('ab', 'ab'), [(0,0), (1,1)])
714
self.assertEquals(unique_lcs('abcde', 'cdeab'), [(2,0), (3,1), (4,2)])
715
self.assertEquals(unique_lcs('cdeab', 'abcde'), [(0,2), (1,3), (2,4)])
716
self.assertEquals(unique_lcs('abXde', 'abYde'), [(0,0), (1,1),
718
self.assertEquals(unique_lcs('acbac', 'abc'), [(2,1)])
720
def test_recurse_matches(self):
721
def test_one(a, b, matches):
723
self._recurse_matches(
724
a, b, 0, 0, len(a), len(b), test_matches, 10)
725
self.assertEquals(test_matches, matches)
727
test_one(['a', '', 'b', '', 'c'], ['a', 'a', 'b', 'c', 'c'],
728
[(0, 0), (2, 2), (4, 4)])
729
test_one(['a', 'c', 'b', 'a', 'c'], ['a', 'b', 'c'],
730
[(0, 0), (2, 1), (4, 2)])
731
# Even though 'bc' is not unique globally, and is surrounded by
732
# non-matching lines, we should still match, because they are locally
734
test_one('abcdbce', 'afbcgdbce', [(0,0), (1, 2), (2, 3), (3, 5),
735
(4, 6), (5, 7), (6, 8)])
737
# recurse_matches doesn't match non-unique
738
# lines surrounded by bogus text.
739
# The update has been done in patiencediff.SequenceMatcher instead
741
# This is what it could be
742
#test_one('aBccDe', 'abccde', [(0,0), (2,2), (3,3), (5,5)])
744
# This is what it currently gives:
745
test_one('aBccDe', 'abccde', [(0,0), (5,5)])
747
def test_matching_blocks(self):
748
def chk_blocks(a, b, expected_blocks):
749
# difflib always adds a signature of the total
750
# length, with no matching entries at the end
751
s = self._PatienceSequenceMatcher(None, a, b)
752
blocks = s.get_matching_blocks()
753
self.assertEquals((len(a), len(b), 0), blocks[-1])
754
self.assertEquals(expected_blocks, blocks[:-1])
756
# Some basic matching tests
757
chk_blocks('', '', [])
758
chk_blocks([], [], [])
759
chk_blocks('abc', '', [])
760
chk_blocks('', 'abc', [])
761
chk_blocks('abcd', 'abcd', [(0, 0, 4)])
762
chk_blocks('abcd', 'abce', [(0, 0, 3)])
763
chk_blocks('eabc', 'abce', [(1, 0, 3)])
764
chk_blocks('eabce', 'abce', [(1, 0, 4)])
765
chk_blocks('abcde', 'abXde', [(0, 0, 2), (3, 3, 2)])
766
chk_blocks('abcde', 'abXYZde', [(0, 0, 2), (3, 5, 2)])
767
chk_blocks('abde', 'abXYZde', [(0, 0, 2), (2, 5, 2)])
768
# This may check too much, but it checks to see that
769
# a copied block stays attached to the previous section,
771
# difflib would tend to grab the trailing longest match
772
# which would make the diff not look right
773
chk_blocks('abcdefghijklmnop', 'abcdefxydefghijklmnop',
774
[(0, 0, 6), (6, 11, 10)])
776
# make sure it supports passing in lists
780
'how are you today?\n'],
782
'how are you today?\n'],
783
[(0, 0, 1), (2, 1, 1)])
785
# non unique lines surrounded by non-matching lines
787
chk_blocks('aBccDe', 'abccde', [(0,0,1), (5,5,1)])
789
# But they only need to be locally unique
790
chk_blocks('aBcDec', 'abcdec', [(0,0,1), (2,2,1), (4,4,2)])
792
# non unique blocks won't be matched
793
chk_blocks('aBcdEcdFg', 'abcdecdfg', [(0,0,1), (8,8,1)])
795
# but locally unique ones will
796
chk_blocks('aBcdEeXcdFg', 'abcdecdfg', [(0,0,1), (2,2,2),
797
(5,4,1), (7,5,2), (10,8,1)])
799
chk_blocks('abbabbXd', 'cabbabxd', [(7,7,1)])
800
chk_blocks('abbabbbb', 'cabbabbc', [])
801
chk_blocks('bbbbbbbb', 'cbbbbbbc', [])
803
def test_opcodes(self):
804
def chk_ops(a, b, expected_codes):
805
s = self._PatienceSequenceMatcher(None, a, b)
806
self.assertEquals(expected_codes, s.get_opcodes())
810
chk_ops('abc', '', [('delete', 0,3, 0,0)])
811
chk_ops('', 'abc', [('insert', 0,0, 0,3)])
812
chk_ops('abcd', 'abcd', [('equal', 0,4, 0,4)])
813
chk_ops('abcd', 'abce', [('equal', 0,3, 0,3),
814
('replace', 3,4, 3,4)
816
chk_ops('eabc', 'abce', [('delete', 0,1, 0,0),
820
chk_ops('eabce', 'abce', [('delete', 0,1, 0,0),
823
chk_ops('abcde', 'abXde', [('equal', 0,2, 0,2),
824
('replace', 2,3, 2,3),
827
chk_ops('abcde', 'abXYZde', [('equal', 0,2, 0,2),
828
('replace', 2,3, 2,5),
831
chk_ops('abde', 'abXYZde', [('equal', 0,2, 0,2),
832
('insert', 2,2, 2,5),
835
chk_ops('abcdefghijklmnop', 'abcdefxydefghijklmnop',
836
[('equal', 0,6, 0,6),
837
('insert', 6,6, 6,11),
838
('equal', 6,16, 11,21)
843
, 'how are you today?\n'],
845
, 'how are you today?\n'],
846
[('equal', 0,1, 0,1),
847
('delete', 1,2, 1,1),
850
chk_ops('aBccDe', 'abccde',
851
[('equal', 0,1, 0,1),
852
('replace', 1,5, 1,5),
855
chk_ops('aBcDec', 'abcdec',
856
[('equal', 0,1, 0,1),
857
('replace', 1,2, 1,2),
859
('replace', 3,4, 3,4),
862
chk_ops('aBcdEcdFg', 'abcdecdfg',
863
[('equal', 0,1, 0,1),
864
('replace', 1,8, 1,8),
867
chk_ops('aBcdEeXcdFg', 'abcdecdfg',
868
[('equal', 0,1, 0,1),
869
('replace', 1,2, 1,2),
871
('delete', 4,5, 4,4),
873
('delete', 6,7, 5,5),
875
('replace', 9,10, 7,8),
876
('equal', 10,11, 8,9)
879
def test_grouped_opcodes(self):
880
def chk_ops(a, b, expected_codes, n=3):
881
s = self._PatienceSequenceMatcher(None, a, b)
882
self.assertEquals(expected_codes, list(s.get_grouped_opcodes(n)))
886
chk_ops('abc', '', [[('delete', 0,3, 0,0)]])
887
chk_ops('', 'abc', [[('insert', 0,0, 0,3)]])
888
chk_ops('abcd', 'abcd', [])
889
chk_ops('abcd', 'abce', [[('equal', 0,3, 0,3),
890
('replace', 3,4, 3,4)
892
chk_ops('eabc', 'abce', [[('delete', 0,1, 0,0),
896
chk_ops('abcdefghijklmnop', 'abcdefxydefghijklmnop',
897
[[('equal', 3,6, 3,6),
898
('insert', 6,6, 6,11),
899
('equal', 6,9, 11,14)
901
chk_ops('abcdefghijklmnop', 'abcdefxydefghijklmnop',
902
[[('equal', 2,6, 2,6),
903
('insert', 6,6, 6,11),
904
('equal', 6,10, 11,15)
906
chk_ops('Xabcdef', 'abcdef',
907
[[('delete', 0,1, 0,0),
910
chk_ops('abcdef', 'abcdefX',
911
[[('equal', 3,6, 3,6),
916
def test_multiple_ranges(self):
917
# There was an earlier bug where we used a bad set of ranges,
918
# this triggers that specific bug, to make sure it doesn't regress
919
def chk_blocks(a, b, expected_blocks):
920
# difflib always adds a signature of the total
921
# length, with no matching entries at the end
922
s = self._PatienceSequenceMatcher(None, a, b)
923
blocks = s.get_matching_blocks()
925
self.assertEquals(x, (len(a), len(b), 0))
926
self.assertEquals(expected_blocks, blocks)
928
chk_blocks('abcdefghijklmnop'
929
, 'abcXghiYZQRSTUVWXYZijklmnop'
930
, [(0, 0, 3), (6, 4, 3), (9, 20, 7)])
932
chk_blocks('ABCd efghIjk L'
933
, 'AxyzBCn mo pqrstuvwI1 2 L'
934
, [(0,0,1), (1, 4, 2), (9, 19, 1), (12, 23, 3)])
936
# These are rot13 code snippets.
938
trg nqqrq jura lbh nqq n svyr va gur qverpgbel.
940
gnxrf_netf = ['svyr*']
941
gnxrf_bcgvbaf = ['ab-erphefr']
943
qrs eha(frys, svyr_yvfg, ab_erphefr=Snyfr):
944
sebz omeyvo.nqq vzcbeg fzneg_nqq, nqq_ercbegre_cevag, nqq_ercbegre_ahyy
946
ercbegre = nqq_ercbegre_ahyy
948
ercbegre = nqq_ercbegre_cevag
949
fzneg_nqq(svyr_yvfg, abg ab_erphefr, ercbegre)
952
pynff pzq_zxqve(Pbzznaq):
953
'''.splitlines(True), '''\
954
trg nqqrq jura lbh nqq n svyr va gur qverpgbel.
956
--qel-eha jvyy fubj juvpu svyrf jbhyq or nqqrq, ohg abg npghnyyl
959
gnxrf_netf = ['svyr*']
960
gnxrf_bcgvbaf = ['ab-erphefr', 'qel-eha']
962
qrs eha(frys, svyr_yvfg, ab_erphefr=Snyfr, qel_eha=Snyfr):
967
# Guvf vf cbvagyrff, ohg V'q engure abg envfr na reebe
968
npgvba = omeyvo.nqq.nqq_npgvba_ahyy
970
npgvba = omeyvo.nqq.nqq_npgvba_cevag
972
npgvba = omeyvo.nqq.nqq_npgvba_nqq
974
npgvba = omeyvo.nqq.nqq_npgvba_nqq_naq_cevag
976
omeyvo.nqq.fzneg_nqq(svyr_yvfg, abg ab_erphefr, npgvba)
979
pynff pzq_zxqve(Pbzznaq):
981
, [(0,0,1), (1, 4, 2), (9, 19, 1), (12, 23, 3)])
983
def test_patience_unified_diff(self):
984
txt_a = ['hello there\n',
986
'how are you today?\n']
987
txt_b = ['hello there\n',
988
'how are you today?\n']
989
unified_diff = bzrlib.patiencediff.unified_diff
990
psm = self._PatienceSequenceMatcher
991
self.assertEquals([ '--- \n',
996
' how are you today?\n'
998
, list(unified_diff(txt_a, txt_b,
999
sequencematcher=psm)))
1000
txt_a = map(lambda x: x+'\n', 'abcdefghijklmnop')
1001
txt_b = map(lambda x: x+'\n', 'abcdefxydefghijklmnop')
1002
# This is the result with LongestCommonSubstring matching
1003
self.assertEquals(['--- \n',
1005
'@@ -1,6 +1,11 @@\n',
1017
, list(unified_diff(txt_a, txt_b)))
1018
# And the patience diff
1019
self.assertEquals(['--- \n',
1021
'@@ -4,6 +4,11 @@\n',
1034
, list(unified_diff(txt_a, txt_b,
1035
sequencematcher=psm)))
1038
class TestPatienceDiffLib_c(TestPatienceDiffLib):
1040
_test_needs_features = [CompiledPatienceDiffFeature]
1043
super(TestPatienceDiffLib_c, self).setUp()
1044
import bzrlib._patiencediff_c
1045
self._unique_lcs = bzrlib._patiencediff_c.unique_lcs_c
1046
self._recurse_matches = bzrlib._patiencediff_c.recurse_matches_c
1047
self._PatienceSequenceMatcher = \
1048
bzrlib._patiencediff_c.PatienceSequenceMatcher_c
1051
class TestPatienceDiffLibFiles(TestCaseInTempDir):
1054
super(TestPatienceDiffLibFiles, self).setUp()
1055
self._PatienceSequenceMatcher = \
1056
bzrlib._patiencediff_py.PatienceSequenceMatcher_py
1058
def test_patience_unified_diff_files(self):
1059
txt_a = ['hello there\n',
1061
'how are you today?\n']
1062
txt_b = ['hello there\n',
1063
'how are you today?\n']
1064
open('a1', 'wb').writelines(txt_a)
1065
open('b1', 'wb').writelines(txt_b)
1067
unified_diff_files = bzrlib.patiencediff.unified_diff_files
1068
psm = self._PatienceSequenceMatcher
1069
self.assertEquals(['--- a1 \n',
1071
'@@ -1,3 +1,2 @@\n',
1074
' how are you today?\n',
1076
, list(unified_diff_files('a1', 'b1',
1077
sequencematcher=psm)))
1079
txt_a = map(lambda x: x+'\n', 'abcdefghijklmnop')
1080
txt_b = map(lambda x: x+'\n', 'abcdefxydefghijklmnop')
1081
open('a2', 'wb').writelines(txt_a)
1082
open('b2', 'wb').writelines(txt_b)
1084
# This is the result with LongestCommonSubstring matching
1085
self.assertEquals(['--- a2 \n',
1087
'@@ -1,6 +1,11 @@\n',
1099
, list(unified_diff_files('a2', 'b2')))
1101
# And the patience diff
1102
self.assertEquals(['--- a2 \n',
1104
'@@ -4,6 +4,11 @@\n',
1117
, list(unified_diff_files('a2', 'b2',
1118
sequencematcher=psm)))
1121
class TestPatienceDiffLibFiles_c(TestPatienceDiffLibFiles):
1123
_test_needs_features = [CompiledPatienceDiffFeature]
1126
super(TestPatienceDiffLibFiles_c, self).setUp()
1127
import bzrlib._patiencediff_c
1128
self._PatienceSequenceMatcher = \
1129
bzrlib._patiencediff_c.PatienceSequenceMatcher_c
1132
class TestUsingCompiledIfAvailable(TestCase):
1134
def test_PatienceSequenceMatcher(self):
1135
if CompiledPatienceDiffFeature.available():
1136
from bzrlib._patiencediff_c import PatienceSequenceMatcher_c
1137
self.assertIs(PatienceSequenceMatcher_c,
1138
bzrlib.patiencediff.PatienceSequenceMatcher)
1140
from bzrlib._patiencediff_py import PatienceSequenceMatcher_py
1141
self.assertIs(PatienceSequenceMatcher_py,
1142
bzrlib.patiencediff.PatienceSequenceMatcher)
1144
def test_unique_lcs(self):
1145
if CompiledPatienceDiffFeature.available():
1146
from bzrlib._patiencediff_c import unique_lcs_c
1147
self.assertIs(unique_lcs_c,
1148
bzrlib.patiencediff.unique_lcs)
1150
from bzrlib._patiencediff_py import unique_lcs_py
1151
self.assertIs(unique_lcs_py,
1152
bzrlib.patiencediff.unique_lcs)
1154
def test_recurse_matches(self):
1155
if CompiledPatienceDiffFeature.available():
1156
from bzrlib._patiencediff_c import recurse_matches_c
1157
self.assertIs(recurse_matches_c,
1158
bzrlib.patiencediff.recurse_matches)
1160
from bzrlib._patiencediff_py import recurse_matches_py
1161
self.assertIs(recurse_matches_py,
1162
bzrlib.patiencediff.recurse_matches)
added
removed
bzrlib/tests/test_diff.py
