~bzr-pqm/bzr/bzr.dev

« back to all changes in this revision

Viewing changes to bzrlib/tests/test_diff.py

  • Committer: John Arbash Meinel
  • Author(s): Mark Hammond
  • Date: 2008-09-09 17:02:21 UTC
  • mto: This revision was merged to the branch mainline in revision 3697.
  • Revision ID: john@arbash-meinel.com-20080909170221-svim3jw2mrz0amp3
An updated transparent icon for bzr.

Show diffs side-by-side

added added

removed removed

Lines of Context:
 
1
# Copyright (C) 2005, 2006 Canonical Ltd
 
2
#
 
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.
 
7
#
 
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.
 
12
#
 
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
 
16
 
 
17
import os
 
18
import os.path
 
19
from cStringIO import StringIO
 
20
import errno
 
21
import subprocess
 
22
import sys
 
23
from tempfile import TemporaryFile
 
24
 
 
25
from bzrlib import tests
 
26
from bzrlib.diff import (
 
27
    DiffFromTool,
 
28
    DiffPath,
 
29
    DiffSymlink,
 
30
    DiffTree,
 
31
    DiffText,
 
32
    external_diff,
 
33
    internal_diff,
 
34
    show_diff_trees,
 
35
    )
 
36
from bzrlib.errors import BinaryFile, NoDiff, ExecutableMissing
 
37
import bzrlib.osutils as osutils
 
38
import bzrlib.transform as transform
 
39
import bzrlib.patiencediff
 
40
import bzrlib._patiencediff_py
 
41
from bzrlib.tests import (Feature, TestCase, TestCaseWithTransport,
 
42
                          TestCaseInTempDir, TestSkipped)
 
43
 
 
44
 
 
45
class _AttribFeature(Feature):
 
46
 
 
47
    def _probe(self):
 
48
        if (sys.platform not in ('cygwin', 'win32')):
 
49
            return False
 
50
        try:
 
51
            proc = subprocess.Popen(['attrib', '.'], stdout=subprocess.PIPE)
 
52
        except OSError, e:
 
53
            return False
 
54
        return (0 == proc.wait())
 
55
 
 
56
    def feature_name(self):
 
57
        return 'attrib Windows command-line tool'
 
58
 
 
59
AttribFeature = _AttribFeature()
 
60
 
 
61
 
 
62
class _CompiledPatienceDiffFeature(Feature):
 
63
 
 
64
    def _probe(self):
 
65
        try:
 
66
            import bzrlib._patiencediff_c
 
67
        except ImportError:
 
68
            return False
 
69
        return True
 
70
 
 
71
    def feature_name(self):
 
72
        return 'bzrlib._patiencediff_c'
 
73
 
 
74
CompiledPatienceDiffFeature = _CompiledPatienceDiffFeature()
 
75
 
 
76
 
 
77
def udiff_lines(old, new, allow_binary=False):
 
78
    output = StringIO()
 
79
    internal_diff('old', old, 'new', new, output, allow_binary)
 
80
    output.seek(0, 0)
 
81
    return output.readlines()
 
82
 
 
83
 
 
84
def external_udiff_lines(old, new, use_stringio=False):
 
85
    if use_stringio:
 
86
        # StringIO has no fileno, so it tests a different codepath
 
87
        output = StringIO()
 
88
    else:
 
89
        output = TemporaryFile()
 
90
    try:
 
91
        external_diff('old', old, 'new', new, output, diff_opts=['-u'])
 
92
    except NoDiff:
 
93
        raise TestSkipped('external "diff" not present to test')
 
94
    output.seek(0, 0)
 
95
    lines = output.readlines()
 
96
    output.close()
 
97
    return lines
 
98
 
 
99
 
 
100
class TestDiff(TestCase):
 
101
 
 
102
    def test_add_nl(self):
 
103
        """diff generates a valid diff for patches that add a newline"""
 
104
        lines = udiff_lines(['boo'], ['boo\n'])
 
105
        self.check_patch(lines)
 
106
        self.assertEquals(lines[4], '\\ No newline at end of file\n')
 
107
            ## "expected no-nl, got %r" % lines[4]
 
108
 
 
109
    def test_add_nl_2(self):
 
110
        """diff generates a valid diff for patches that change last line and
 
111
        add a newline.
 
112
        """
 
113
        lines = udiff_lines(['boo'], ['goo\n'])
 
114
        self.check_patch(lines)
 
115
        self.assertEquals(lines[4], '\\ No newline at end of file\n')
 
116
            ## "expected no-nl, got %r" % lines[4]
 
117
 
 
118
    def test_remove_nl(self):
 
119
        """diff generates a valid diff for patches that change last line and
 
120
        add a newline.
 
121
        """
 
122
        lines = udiff_lines(['boo\n'], ['boo'])
 
123
        self.check_patch(lines)
 
124
        self.assertEquals(lines[5], '\\ No newline at end of file\n')
 
125
            ## "expected no-nl, got %r" % lines[5]
 
126
 
 
127
    def check_patch(self, lines):
 
128
        self.assert_(len(lines) > 1)
 
129
            ## "Not enough lines for a file header for patch:\n%s" % "".join(lines)
 
130
        self.assert_(lines[0].startswith ('---'))
 
131
            ## 'No orig line for patch:\n%s' % "".join(lines)
 
132
        self.assert_(lines[1].startswith ('+++'))
 
133
            ## 'No mod line for patch:\n%s' % "".join(lines)
 
134
        self.assert_(len(lines) > 2)
 
135
            ## "No hunks for patch:\n%s" % "".join(lines)
 
136
        self.assert_(lines[2].startswith('@@'))
 
137
            ## "No hunk header for patch:\n%s" % "".join(lines)
 
138
        self.assert_('@@' in lines[2][2:])
 
139
            ## "Unterminated hunk header for patch:\n%s" % "".join(lines)
 
140
 
 
141
    def test_binary_lines(self):
 
142
        self.assertRaises(BinaryFile, udiff_lines, [1023 * 'a' + '\x00'], [])
 
143
        self.assertRaises(BinaryFile, udiff_lines, [], [1023 * 'a' + '\x00'])
 
144
        udiff_lines([1023 * 'a' + '\x00'], [], allow_binary=True)
 
145
        udiff_lines([], [1023 * 'a' + '\x00'], allow_binary=True)
 
146
 
 
147
    def test_external_diff(self):
 
148
        lines = external_udiff_lines(['boo\n'], ['goo\n'])
 
149
        self.check_patch(lines)
 
150
        self.assertEqual('\n', lines[-1])
 
151
 
 
152
    def test_external_diff_no_fileno(self):
 
153
        # Make sure that we can handle not having a fileno, even
 
154
        # if the diff is large
 
155
        lines = external_udiff_lines(['boo\n']*10000,
 
156
                                     ['goo\n']*10000,
 
157
                                     use_stringio=True)
 
158
        self.check_patch(lines)
 
159
 
 
160
    def test_external_diff_binary_lang_c(self):
 
161
        old_env = {}
 
162
        for lang in ('LANG', 'LC_ALL', 'LANGUAGE'):
 
163
            old_env[lang] = osutils.set_or_unset_env(lang, 'C')
 
164
        try:
 
165
            lines = external_udiff_lines(['\x00foobar\n'], ['foo\x00bar\n'])
 
166
            # Older versions of diffutils say "Binary files", newer
 
167
            # versions just say "Files".
 
168
            self.assertContainsRe(lines[0],
 
169
                                  '(Binary f|F)iles old and new differ\n')
 
170
            self.assertEquals(lines[1:], ['\n'])
 
171
        finally:
 
172
            for lang, old_val in old_env.iteritems():
 
173
                osutils.set_or_unset_env(lang, old_val)
 
174
 
 
175
    def test_no_external_diff(self):
 
176
        """Check that NoDiff is raised when diff is not available"""
 
177
        # Use os.environ['PATH'] to make sure no 'diff' command is available
 
178
        orig_path = os.environ['PATH']
 
179
        try:
 
180
            os.environ['PATH'] = ''
 
181
            self.assertRaises(NoDiff, external_diff,
 
182
                              'old', ['boo\n'], 'new', ['goo\n'],
 
183
                              StringIO(), diff_opts=['-u'])
 
184
        finally:
 
185
            os.environ['PATH'] = orig_path
 
186
        
 
187
    def test_internal_diff_default(self):
 
188
        # Default internal diff encoding is utf8
 
189
        output = StringIO()
 
190
        internal_diff(u'old_\xb5', ['old_text\n'],
 
191
                    u'new_\xe5', ['new_text\n'], output)
 
192
        lines = output.getvalue().splitlines(True)
 
193
        self.check_patch(lines)
 
194
        self.assertEquals(['--- old_\xc2\xb5\n',
 
195
                           '+++ new_\xc3\xa5\n',
 
196
                           '@@ -1,1 +1,1 @@\n',
 
197
                           '-old_text\n',
 
198
                           '+new_text\n',
 
199
                           '\n',
 
200
                          ]
 
201
                          , lines)
 
202
 
 
203
    def test_internal_diff_utf8(self):
 
204
        output = StringIO()
 
205
        internal_diff(u'old_\xb5', ['old_text\n'],
 
206
                    u'new_\xe5', ['new_text\n'], output,
 
207
                    path_encoding='utf8')
 
208
        lines = output.getvalue().splitlines(True)
 
209
        self.check_patch(lines)
 
210
        self.assertEquals(['--- old_\xc2\xb5\n',
 
211
                           '+++ new_\xc3\xa5\n',
 
212
                           '@@ -1,1 +1,1 @@\n',
 
213
                           '-old_text\n',
 
214
                           '+new_text\n',
 
215
                           '\n',
 
216
                          ]
 
217
                          , lines)
 
218
 
 
219
    def test_internal_diff_iso_8859_1(self):
 
220
        output = StringIO()
 
221
        internal_diff(u'old_\xb5', ['old_text\n'],
 
222
                    u'new_\xe5', ['new_text\n'], output,
 
223
                    path_encoding='iso-8859-1')
 
224
        lines = output.getvalue().splitlines(True)
 
225
        self.check_patch(lines)
 
226
        self.assertEquals(['--- old_\xb5\n',
 
227
                           '+++ new_\xe5\n',
 
228
                           '@@ -1,1 +1,1 @@\n',
 
229
                           '-old_text\n',
 
230
                           '+new_text\n',
 
231
                           '\n',
 
232
                          ]
 
233
                          , lines)
 
234
 
 
235
    def test_internal_diff_no_content(self):
 
236
        output = StringIO()
 
237
        internal_diff(u'old', [], u'new', [], output)
 
238
        self.assertEqual('', output.getvalue())
 
239
 
 
240
    def test_internal_diff_no_changes(self):
 
241
        output = StringIO()
 
242
        internal_diff(u'old', ['text\n', 'contents\n'],
 
243
                      u'new', ['text\n', 'contents\n'],
 
244
                      output)
 
245
        self.assertEqual('', output.getvalue())
 
246
 
 
247
    def test_internal_diff_returns_bytes(self):
 
248
        import StringIO
 
249
        output = StringIO.StringIO()
 
250
        internal_diff(u'old_\xb5', ['old_text\n'],
 
251
                    u'new_\xe5', ['new_text\n'], output)
 
252
        self.failUnless(isinstance(output.getvalue(), str),
 
253
            'internal_diff should return bytestrings')
 
254
 
 
255
 
 
256
class TestDiffFiles(TestCaseInTempDir):
 
257
 
 
258
    def test_external_diff_binary(self):
 
259
        """The output when using external diff should use diff's i18n error"""
 
260
        # Make sure external_diff doesn't fail in the current LANG
 
261
        lines = external_udiff_lines(['\x00foobar\n'], ['foo\x00bar\n'])
 
262
 
 
263
        cmd = ['diff', '-u', '--binary', 'old', 'new']
 
264
        open('old', 'wb').write('\x00foobar\n')
 
265
        open('new', 'wb').write('foo\x00bar\n')
 
266
        pipe = subprocess.Popen(cmd, stdout=subprocess.PIPE,
 
267
                                     stdin=subprocess.PIPE)
 
268
        out, err = pipe.communicate()
 
269
        # Diff returns '2' on Binary files.
 
270
        self.assertEqual(2, pipe.returncode)
 
271
        # We should output whatever diff tells us, plus a trailing newline
 
272
        self.assertEqual(out.splitlines(True) + ['\n'], lines)
 
273
 
 
274
 
 
275
class TestShowDiffTreesHelper(TestCaseWithTransport):
 
276
    """Has a helper for running show_diff_trees"""
 
277
 
 
278
    def get_diff(self, tree1, tree2, specific_files=None, working_tree=None):
 
279
        output = StringIO()
 
280
        if working_tree is not None:
 
281
            extra_trees = (working_tree,)
 
282
        else:
 
283
            extra_trees = ()
 
284
        show_diff_trees(tree1, tree2, output, specific_files=specific_files,
 
285
                        extra_trees=extra_trees, old_label='old/',
 
286
                        new_label='new/')
 
287
        return output.getvalue()
 
288
 
 
289
 
 
290
class TestDiffDates(TestShowDiffTreesHelper):
 
291
 
 
292
    def setUp(self):
 
293
        super(TestDiffDates, self).setUp()
 
294
        self.wt = self.make_branch_and_tree('.')
 
295
        self.b = self.wt.branch
 
296
        self.build_tree_contents([
 
297
            ('file1', 'file1 contents at rev 1\n'),
 
298
            ('file2', 'file2 contents at rev 1\n')
 
299
            ])
 
300
        self.wt.add(['file1', 'file2'])
 
301
        self.wt.commit(
 
302
            message='Revision 1',
 
303
            timestamp=1143849600, # 2006-04-01 00:00:00 UTC
 
304
            timezone=0,
 
305
            rev_id='rev-1')
 
306
        self.build_tree_contents([('file1', 'file1 contents at rev 2\n')])
 
307
        self.wt.commit(
 
308
            message='Revision 2',
 
309
            timestamp=1143936000, # 2006-04-02 00:00:00 UTC
 
310
            timezone=28800,
 
311
            rev_id='rev-2')
 
312
        self.build_tree_contents([('file2', 'file2 contents at rev 3\n')])
 
313
        self.wt.commit(
 
314
            message='Revision 3',
 
315
            timestamp=1144022400, # 2006-04-03 00:00:00 UTC
 
316
            timezone=-3600,
 
317
            rev_id='rev-3')
 
318
        self.wt.remove(['file2'])
 
319
        self.wt.commit(
 
320
            message='Revision 4',
 
321
            timestamp=1144108800, # 2006-04-04 00:00:00 UTC
 
322
            timezone=0,
 
323
            rev_id='rev-4')
 
324
        self.build_tree_contents([
 
325
            ('file1', 'file1 contents in working tree\n')
 
326
            ])
 
327
        # set the date stamps for files in the working tree to known values
 
328
        os.utime('file1', (1144195200, 1144195200)) # 2006-04-05 00:00:00 UTC
 
329
 
 
330
    def test_diff_rev_tree_working_tree(self):
 
331
        output = self.get_diff(self.wt.basis_tree(), self.wt)
 
332
        # note that the date for old/file1 is from rev 2 rather than from
 
333
        # the basis revision (rev 4)
 
334
        self.assertEqualDiff(output, '''\
 
335
=== modified file 'file1'
 
336
--- old/file1\t2006-04-02 00:00:00 +0000
 
337
+++ new/file1\t2006-04-05 00:00:00 +0000
 
338
@@ -1,1 +1,1 @@
 
339
-file1 contents at rev 2
 
340
+file1 contents in working tree
 
341
 
 
342
''')
 
343
 
 
344
    def test_diff_rev_tree_rev_tree(self):
 
345
        tree1 = self.b.repository.revision_tree('rev-2')
 
346
        tree2 = self.b.repository.revision_tree('rev-3')
 
347
        output = self.get_diff(tree1, tree2)
 
348
        self.assertEqualDiff(output, '''\
 
349
=== modified file 'file2'
 
350
--- old/file2\t2006-04-01 00:00:00 +0000
 
351
+++ new/file2\t2006-04-03 00:00:00 +0000
 
352
@@ -1,1 +1,1 @@
 
353
-file2 contents at rev 1
 
354
+file2 contents at rev 3
 
355
 
 
356
''')
 
357
        
 
358
    def test_diff_add_files(self):
 
359
        tree1 = self.b.repository.revision_tree(None)
 
360
        tree2 = self.b.repository.revision_tree('rev-1')
 
361
        output = self.get_diff(tree1, tree2)
 
362
        # the files have the epoch time stamp for the tree in which
 
363
        # they don't exist.
 
364
        self.assertEqualDiff(output, '''\
 
365
=== added file 'file1'
 
366
--- old/file1\t1970-01-01 00:00:00 +0000
 
367
+++ new/file1\t2006-04-01 00:00:00 +0000
 
368
@@ -0,0 +1,1 @@
 
369
+file1 contents at rev 1
 
370
 
 
371
=== added file 'file2'
 
372
--- old/file2\t1970-01-01 00:00:00 +0000
 
373
+++ new/file2\t2006-04-01 00:00:00 +0000
 
374
@@ -0,0 +1,1 @@
 
375
+file2 contents at rev 1
 
376
 
 
377
''')
 
378
 
 
379
    def test_diff_remove_files(self):
 
380
        tree1 = self.b.repository.revision_tree('rev-3')
 
381
        tree2 = self.b.repository.revision_tree('rev-4')
 
382
        output = self.get_diff(tree1, tree2)
 
383
        # the file has the epoch time stamp for the tree in which
 
384
        # it doesn't exist.
 
385
        self.assertEqualDiff(output, '''\
 
386
=== removed file 'file2'
 
387
--- old/file2\t2006-04-03 00:00:00 +0000
 
388
+++ new/file2\t1970-01-01 00:00:00 +0000
 
389
@@ -1,1 +0,0 @@
 
390
-file2 contents at rev 3
 
391
 
 
392
''')
 
393
 
 
394
    def test_show_diff_specified(self):
 
395
        """A working tree filename can be used to identify a file"""
 
396
        self.wt.rename_one('file1', 'file1b')
 
397
        old_tree = self.b.repository.revision_tree('rev-1')
 
398
        new_tree = self.b.repository.revision_tree('rev-4')
 
399
        out = self.get_diff(old_tree, new_tree, specific_files=['file1b'], 
 
400
                            working_tree=self.wt)
 
401
        self.assertContainsRe(out, 'file1\t')
 
402
 
 
403
    def test_recursive_diff(self):
 
404
        """Children of directories are matched"""
 
405
        os.mkdir('dir1')
 
406
        os.mkdir('dir2')
 
407
        self.wt.add(['dir1', 'dir2'])
 
408
        self.wt.rename_one('file1', 'dir1/file1')
 
409
        old_tree = self.b.repository.revision_tree('rev-1')
 
410
        new_tree = self.b.repository.revision_tree('rev-4')
 
411
        out = self.get_diff(old_tree, new_tree, specific_files=['dir1'], 
 
412
                            working_tree=self.wt)
 
413
        self.assertContainsRe(out, 'file1\t')
 
414
        out = self.get_diff(old_tree, new_tree, specific_files=['dir2'], 
 
415
                            working_tree=self.wt)
 
416
        self.assertNotContainsRe(out, 'file1\t')
 
417
 
 
418
 
 
419
 
 
420
class TestShowDiffTrees(TestShowDiffTreesHelper):
 
421
    """Direct tests for show_diff_trees"""
 
422
 
 
423
    def test_modified_file(self):
 
424
        """Test when a file is modified."""
 
425
        tree = self.make_branch_and_tree('tree')
 
426
        self.build_tree_contents([('tree/file', 'contents\n')])
 
427
        tree.add(['file'], ['file-id'])
 
428
        tree.commit('one', rev_id='rev-1')
 
429
 
 
430
        self.build_tree_contents([('tree/file', 'new contents\n')])
 
431
        diff = self.get_diff(tree.basis_tree(), tree)
 
432
        self.assertContainsRe(diff, "=== modified file 'file'\n")
 
433
        self.assertContainsRe(diff, '--- old/file\t')
 
434
        self.assertContainsRe(diff, '\\+\\+\\+ new/file\t')
 
435
        self.assertContainsRe(diff, '-contents\n'
 
436
                                    '\\+new contents\n')
 
437
 
 
438
    def test_modified_file_in_renamed_dir(self):
 
439
        """Test when a file is modified in a renamed directory."""
 
440
        tree = self.make_branch_and_tree('tree')
 
441
        self.build_tree(['tree/dir/'])
 
442
        self.build_tree_contents([('tree/dir/file', 'contents\n')])
 
443
        tree.add(['dir', 'dir/file'], ['dir-id', 'file-id'])
 
444
        tree.commit('one', rev_id='rev-1')
 
445
 
 
446
        tree.rename_one('dir', 'other')
 
447
        self.build_tree_contents([('tree/other/file', 'new contents\n')])
 
448
        diff = self.get_diff(tree.basis_tree(), tree)
 
449
        self.assertContainsRe(diff, "=== renamed directory 'dir' => 'other'\n")
 
450
        self.assertContainsRe(diff, "=== modified file 'other/file'\n")
 
451
        # XXX: This is technically incorrect, because it used to be at another
 
452
        # location. What to do?
 
453
        self.assertContainsRe(diff, '--- old/dir/file\t')
 
454
        self.assertContainsRe(diff, '\\+\\+\\+ new/other/file\t')
 
455
        self.assertContainsRe(diff, '-contents\n'
 
456
                                    '\\+new contents\n')
 
457
 
 
458
    def test_renamed_directory(self):
 
459
        """Test when only a directory is only renamed."""
 
460
        tree = self.make_branch_and_tree('tree')
 
461
        self.build_tree(['tree/dir/'])
 
462
        self.build_tree_contents([('tree/dir/file', 'contents\n')])
 
463
        tree.add(['dir', 'dir/file'], ['dir-id', 'file-id'])
 
464
        tree.commit('one', rev_id='rev-1')
 
465
 
 
466
        tree.rename_one('dir', 'newdir')
 
467
        diff = self.get_diff(tree.basis_tree(), tree)
 
468
        # Renaming a directory should be a single "you renamed this dir" even
 
469
        # when there are files inside.
 
470
        self.assertEqual("=== renamed directory 'dir' => 'newdir'\n", diff)
 
471
 
 
472
    def test_renamed_file(self):
 
473
        """Test when a file is only renamed."""
 
474
        tree = self.make_branch_and_tree('tree')
 
475
        self.build_tree_contents([('tree/file', 'contents\n')])
 
476
        tree.add(['file'], ['file-id'])
 
477
        tree.commit('one', rev_id='rev-1')
 
478
 
 
479
        tree.rename_one('file', 'newname')
 
480
        diff = self.get_diff(tree.basis_tree(), tree)
 
481
        self.assertContainsRe(diff, "=== renamed file 'file' => 'newname'\n")
 
482
        # We shouldn't have a --- or +++ line, because there is no content
 
483
        # change
 
484
        self.assertNotContainsRe(diff, '---')
 
485
 
 
486
    def test_renamed_and_modified_file(self):
 
487
        """Test when a file is only renamed."""
 
488
        tree = self.make_branch_and_tree('tree')
 
489
        self.build_tree_contents([('tree/file', 'contents\n')])
 
490
        tree.add(['file'], ['file-id'])
 
491
        tree.commit('one', rev_id='rev-1')
 
492
 
 
493
        tree.rename_one('file', 'newname')
 
494
        self.build_tree_contents([('tree/newname', 'new contents\n')])
 
495
        diff = self.get_diff(tree.basis_tree(), tree)
 
496
        self.assertContainsRe(diff, "=== renamed file 'file' => 'newname'\n")
 
497
        self.assertContainsRe(diff, '--- old/file\t')
 
498
        self.assertContainsRe(diff, '\\+\\+\\+ new/newname\t')
 
499
        self.assertContainsRe(diff, '-contents\n'
 
500
                                    '\\+new contents\n')
 
501
 
 
502
 
 
503
    def test_internal_diff_exec_property(self):
 
504
        tree = self.make_branch_and_tree('tree')
 
505
 
 
506
        tt = transform.TreeTransform(tree)
 
507
        tt.new_file('a', tt.root, 'contents\n', 'a-id', True)
 
508
        tt.new_file('b', tt.root, 'contents\n', 'b-id', False)
 
509
        tt.new_file('c', tt.root, 'contents\n', 'c-id', True)
 
510
        tt.new_file('d', tt.root, 'contents\n', 'd-id', False)
 
511
        tt.new_file('e', tt.root, 'contents\n', 'control-e-id', True)
 
512
        tt.new_file('f', tt.root, 'contents\n', 'control-f-id', False)
 
513
        tt.apply()
 
514
        tree.commit('one', rev_id='rev-1')
 
515
 
 
516
        tt = transform.TreeTransform(tree)
 
517
        tt.set_executability(False, tt.trans_id_file_id('a-id'))
 
518
        tt.set_executability(True, tt.trans_id_file_id('b-id'))
 
519
        tt.set_executability(False, tt.trans_id_file_id('c-id'))
 
520
        tt.set_executability(True, tt.trans_id_file_id('d-id'))
 
521
        tt.apply()
 
522
        tree.rename_one('c', 'new-c')
 
523
        tree.rename_one('d', 'new-d')
 
524
 
 
525
        diff = self.get_diff(tree.basis_tree(), tree)
 
526
 
 
527
        self.assertContainsRe(diff, r"file 'a'.*\(properties changed:.*\+x to -x.*\)")
 
528
        self.assertContainsRe(diff, r"file 'b'.*\(properties changed:.*-x to \+x.*\)")
 
529
        self.assertContainsRe(diff, r"file 'c'.*\(properties changed:.*\+x to -x.*\)")
 
530
        self.assertContainsRe(diff, r"file 'd'.*\(properties changed:.*-x to \+x.*\)")
 
531
        self.assertNotContainsRe(diff, r"file 'e'")
 
532
        self.assertNotContainsRe(diff, r"file 'f'")
 
533
 
 
534
 
 
535
    def test_binary_unicode_filenames(self):
 
536
        """Test that contents of files are *not* encoded in UTF-8 when there
 
537
        is a binary file in the diff.
 
538
        """
 
539
        # See https://bugs.launchpad.net/bugs/110092.
 
540
        self.requireFeature(tests.UnicodeFilenameFeature)
 
541
 
 
542
        # This bug isn't triggered with cStringIO.
 
543
        from StringIO import StringIO
 
544
        tree = self.make_branch_and_tree('tree')
 
545
        alpha, omega = u'\u03b1', u'\u03c9'
 
546
        alpha_utf8, omega_utf8 = alpha.encode('utf8'), omega.encode('utf8')
 
547
        self.build_tree_contents(
 
548
            [('tree/' + alpha, chr(0)),
 
549
             ('tree/' + omega,
 
550
              ('The %s and the %s\n' % (alpha_utf8, omega_utf8)))])
 
551
        tree.add([alpha], ['file-id'])
 
552
        tree.add([omega], ['file-id-2'])
 
553
        diff_content = StringIO()
 
554
        show_diff_trees(tree.basis_tree(), tree, diff_content)
 
555
        diff = diff_content.getvalue()
 
556
        self.assertContainsRe(diff, r"=== added file '%s'" % alpha_utf8)
 
557
        self.assertContainsRe(
 
558
            diff, "Binary files a/%s.*and b/%s.* differ\n" % (alpha_utf8, alpha_utf8))
 
559
        self.assertContainsRe(diff, r"=== added file '%s'" % omega_utf8)
 
560
        self.assertContainsRe(diff, r"--- a/%s" % (omega_utf8,))
 
561
        self.assertContainsRe(diff, r"\+\+\+ b/%s" % (omega_utf8,))
 
562
 
 
563
    def test_unicode_filename(self):
 
564
        """Test when the filename are unicode."""
 
565
        self.requireFeature(tests.UnicodeFilenameFeature)
 
566
 
 
567
        alpha, omega = u'\u03b1', u'\u03c9'
 
568
        autf8, outf8 = alpha.encode('utf8'), omega.encode('utf8')
 
569
 
 
570
        tree = self.make_branch_and_tree('tree')
 
571
        self.build_tree_contents([('tree/ren_'+alpha, 'contents\n')])
 
572
        tree.add(['ren_'+alpha], ['file-id-2'])
 
573
        self.build_tree_contents([('tree/del_'+alpha, 'contents\n')])
 
574
        tree.add(['del_'+alpha], ['file-id-3'])
 
575
        self.build_tree_contents([('tree/mod_'+alpha, 'contents\n')])
 
576
        tree.add(['mod_'+alpha], ['file-id-4'])
 
577
 
 
578
        tree.commit('one', rev_id='rev-1')
 
579
 
 
580
        tree.rename_one('ren_'+alpha, 'ren_'+omega)
 
581
        tree.remove('del_'+alpha)
 
582
        self.build_tree_contents([('tree/add_'+alpha, 'contents\n')])
 
583
        tree.add(['add_'+alpha], ['file-id'])
 
584
        self.build_tree_contents([('tree/mod_'+alpha, 'contents_mod\n')])
 
585
 
 
586
        diff = self.get_diff(tree.basis_tree(), tree)
 
587
        self.assertContainsRe(diff,
 
588
                "=== renamed file 'ren_%s' => 'ren_%s'\n"%(autf8, outf8))
 
589
        self.assertContainsRe(diff, "=== added file 'add_%s'"%autf8)
 
590
        self.assertContainsRe(diff, "=== modified file 'mod_%s'"%autf8)
 
591
        self.assertContainsRe(diff, "=== removed file 'del_%s'"%autf8)
 
592
 
 
593
 
 
594
class DiffWasIs(DiffPath):
 
595
 
 
596
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
 
597
        self.to_file.write('was: ')
 
598
        self.to_file.write(self.old_tree.get_file(file_id).read())
 
599
        self.to_file.write('is: ')
 
600
        self.to_file.write(self.new_tree.get_file(file_id).read())
 
601
        pass
 
602
 
 
603
 
 
604
class TestDiffTree(TestCaseWithTransport):
 
605
 
 
606
    def setUp(self):
 
607
        TestCaseWithTransport.setUp(self)
 
608
        self.old_tree = self.make_branch_and_tree('old-tree')
 
609
        self.old_tree.lock_write()
 
610
        self.addCleanup(self.old_tree.unlock)
 
611
        self.new_tree = self.make_branch_and_tree('new-tree')
 
612
        self.new_tree.lock_write()
 
613
        self.addCleanup(self.new_tree.unlock)
 
614
        self.differ = DiffTree(self.old_tree, self.new_tree, StringIO())
 
615
 
 
616
    def test_diff_text(self):
 
617
        self.build_tree_contents([('old-tree/olddir/',),
 
618
                                  ('old-tree/olddir/oldfile', 'old\n')])
 
619
        self.old_tree.add('olddir')
 
620
        self.old_tree.add('olddir/oldfile', 'file-id')
 
621
        self.build_tree_contents([('new-tree/newdir/',),
 
622
                                  ('new-tree/newdir/newfile', 'new\n')])
 
623
        self.new_tree.add('newdir')
 
624
        self.new_tree.add('newdir/newfile', 'file-id')
 
625
        differ = DiffText(self.old_tree, self.new_tree, StringIO())
 
626
        differ.diff_text('file-id', None, 'old label', 'new label')
 
627
        self.assertEqual(
 
628
            '--- old label\n+++ new label\n@@ -1,1 +0,0 @@\n-old\n\n',
 
629
            differ.to_file.getvalue())
 
630
        differ.to_file.seek(0)
 
631
        differ.diff_text(None, 'file-id', 'old label', 'new label')
 
632
        self.assertEqual(
 
633
            '--- old label\n+++ new label\n@@ -0,0 +1,1 @@\n+new\n\n',
 
634
            differ.to_file.getvalue())
 
635
        differ.to_file.seek(0)
 
636
        differ.diff_text('file-id', 'file-id', 'old label', 'new label')
 
637
        self.assertEqual(
 
638
            '--- old label\n+++ new label\n@@ -1,1 +1,1 @@\n-old\n+new\n\n',
 
639
            differ.to_file.getvalue())
 
640
 
 
641
    def test_diff_deletion(self):
 
642
        self.build_tree_contents([('old-tree/file', 'contents'),
 
643
                                  ('new-tree/file', 'contents')])
 
644
        self.old_tree.add('file', 'file-id')
 
645
        self.new_tree.add('file', 'file-id')
 
646
        os.unlink('new-tree/file')
 
647
        self.differ.show_diff(None)
 
648
        self.assertContainsRe(self.differ.to_file.getvalue(), '-contents')
 
649
 
 
650
    def test_diff_creation(self):
 
651
        self.build_tree_contents([('old-tree/file', 'contents'),
 
652
                                  ('new-tree/file', 'contents')])
 
653
        self.old_tree.add('file', 'file-id')
 
654
        self.new_tree.add('file', 'file-id')
 
655
        os.unlink('old-tree/file')
 
656
        self.differ.show_diff(None)
 
657
        self.assertContainsRe(self.differ.to_file.getvalue(), '\+contents')
 
658
 
 
659
    def test_diff_symlink(self):
 
660
        differ = DiffSymlink(self.old_tree, self.new_tree, StringIO())
 
661
        differ.diff_symlink('old target', None)
 
662
        self.assertEqual("=== target was 'old target'\n",
 
663
                         differ.to_file.getvalue())
 
664
 
 
665
        differ = DiffSymlink(self.old_tree, self.new_tree, StringIO())
 
666
        differ.diff_symlink(None, 'new target')
 
667
        self.assertEqual("=== target is 'new target'\n",
 
668
                         differ.to_file.getvalue())
 
669
 
 
670
        differ = DiffSymlink(self.old_tree, self.new_tree, StringIO())
 
671
        differ.diff_symlink('old target', 'new target')
 
672
        self.assertEqual("=== target changed 'old target' => 'new target'\n",
 
673
                         differ.to_file.getvalue())
 
674
 
 
675
    def test_diff(self):
 
676
        self.build_tree_contents([('old-tree/olddir/',),
 
677
                                  ('old-tree/olddir/oldfile', 'old\n')])
 
678
        self.old_tree.add('olddir')
 
679
        self.old_tree.add('olddir/oldfile', 'file-id')
 
680
        self.build_tree_contents([('new-tree/newdir/',),
 
681
                                  ('new-tree/newdir/newfile', 'new\n')])
 
682
        self.new_tree.add('newdir')
 
683
        self.new_tree.add('newdir/newfile', 'file-id')
 
684
        self.differ.diff('file-id', 'olddir/oldfile', 'newdir/newfile')
 
685
        self.assertContainsRe(
 
686
            self.differ.to_file.getvalue(),
 
687
            r'--- olddir/oldfile.*\n\+\+\+ newdir/newfile.*\n\@\@ -1,1 \+1,1'
 
688
             ' \@\@\n-old\n\+new\n\n')
 
689
 
 
690
    def test_diff_kind_change(self):
 
691
        self.requireFeature(tests.SymlinkFeature)
 
692
        self.build_tree_contents([('old-tree/olddir/',),
 
693
                                  ('old-tree/olddir/oldfile', 'old\n')])
 
694
        self.old_tree.add('olddir')
 
695
        self.old_tree.add('olddir/oldfile', 'file-id')
 
696
        self.build_tree(['new-tree/newdir/'])
 
697
        os.symlink('new', 'new-tree/newdir/newfile')
 
698
        self.new_tree.add('newdir')
 
699
        self.new_tree.add('newdir/newfile', 'file-id')
 
700
        self.differ.diff('file-id', 'olddir/oldfile', 'newdir/newfile')
 
701
        self.assertContainsRe(
 
702
            self.differ.to_file.getvalue(),
 
703
            r'--- olddir/oldfile.*\n\+\+\+ newdir/newfile.*\n\@\@ -1,1 \+0,0'
 
704
             ' \@\@\n-old\n\n')
 
705
        self.assertContainsRe(self.differ.to_file.getvalue(),
 
706
                              "=== target is 'new'\n")
 
707
 
 
708
    def test_diff_directory(self):
 
709
        self.build_tree(['new-tree/new-dir/'])
 
710
        self.new_tree.add('new-dir', 'new-dir-id')
 
711
        self.differ.diff('new-dir-id', None, 'new-dir')
 
712
        self.assertEqual(self.differ.to_file.getvalue(), '')
 
713
 
 
714
    def create_old_new(self):
 
715
        self.build_tree_contents([('old-tree/olddir/',),
 
716
                                  ('old-tree/olddir/oldfile', 'old\n')])
 
717
        self.old_tree.add('olddir')
 
718
        self.old_tree.add('olddir/oldfile', 'file-id')
 
719
        self.build_tree_contents([('new-tree/newdir/',),
 
720
                                  ('new-tree/newdir/newfile', 'new\n')])
 
721
        self.new_tree.add('newdir')
 
722
        self.new_tree.add('newdir/newfile', 'file-id')
 
723
 
 
724
    def test_register_diff(self):
 
725
        self.create_old_new()
 
726
        old_diff_factories = DiffTree.diff_factories
 
727
        DiffTree.diff_factories=old_diff_factories[:]
 
728
        DiffTree.diff_factories.insert(0, DiffWasIs.from_diff_tree)
 
729
        try:
 
730
            differ = DiffTree(self.old_tree, self.new_tree, StringIO())
 
731
        finally:
 
732
            DiffTree.diff_factories = old_diff_factories
 
733
        differ.diff('file-id', 'olddir/oldfile', 'newdir/newfile')
 
734
        self.assertNotContainsRe(
 
735
            differ.to_file.getvalue(),
 
736
            r'--- olddir/oldfile.*\n\+\+\+ newdir/newfile.*\n\@\@ -1,1 \+1,1'
 
737
             ' \@\@\n-old\n\+new\n\n')
 
738
        self.assertContainsRe(differ.to_file.getvalue(),
 
739
                              'was: old\nis: new\n')
 
740
 
 
741
    def test_extra_factories(self):
 
742
        self.create_old_new()
 
743
        differ = DiffTree(self.old_tree, self.new_tree, StringIO(),
 
744
                            extra_factories=[DiffWasIs.from_diff_tree])
 
745
        differ.diff('file-id', 'olddir/oldfile', 'newdir/newfile')
 
746
        self.assertNotContainsRe(
 
747
            differ.to_file.getvalue(),
 
748
            r'--- olddir/oldfile.*\n\+\+\+ newdir/newfile.*\n\@\@ -1,1 \+1,1'
 
749
             ' \@\@\n-old\n\+new\n\n')
 
750
        self.assertContainsRe(differ.to_file.getvalue(),
 
751
                              'was: old\nis: new\n')
 
752
 
 
753
    def test_alphabetical_order(self):
 
754
        self.build_tree(['new-tree/a-file'])
 
755
        self.new_tree.add('a-file')
 
756
        self.build_tree(['old-tree/b-file'])
 
757
        self.old_tree.add('b-file')
 
758
        self.differ.show_diff(None)
 
759
        self.assertContainsRe(self.differ.to_file.getvalue(),
 
760
            '.*a-file(.|\n)*b-file')
 
761
 
 
762
 
 
763
class TestPatienceDiffLib(TestCase):
 
764
 
 
765
    def setUp(self):
 
766
        super(TestPatienceDiffLib, self).setUp()
 
767
        self._unique_lcs = bzrlib._patiencediff_py.unique_lcs_py
 
768
        self._recurse_matches = bzrlib._patiencediff_py.recurse_matches_py
 
769
        self._PatienceSequenceMatcher = \
 
770
            bzrlib._patiencediff_py.PatienceSequenceMatcher_py
 
771
 
 
772
    def test_diff_unicode_string(self):
 
773
        a = ''.join([unichr(i) for i in range(4000, 4500, 3)])
 
774
        b = ''.join([unichr(i) for i in range(4300, 4800, 2)])
 
775
        sm = self._PatienceSequenceMatcher(None, a, b)
 
776
        mb = sm.get_matching_blocks()
 
777
        self.assertEquals(35, len(mb))
 
778
 
 
779
    def test_unique_lcs(self):
 
780
        unique_lcs = self._unique_lcs
 
781
        self.assertEquals(unique_lcs('', ''), [])
 
782
        self.assertEquals(unique_lcs('', 'a'), [])
 
783
        self.assertEquals(unique_lcs('a', ''), [])
 
784
        self.assertEquals(unique_lcs('a', 'a'), [(0,0)])
 
785
        self.assertEquals(unique_lcs('a', 'b'), [])
 
786
        self.assertEquals(unique_lcs('ab', 'ab'), [(0,0), (1,1)])
 
787
        self.assertEquals(unique_lcs('abcde', 'cdeab'), [(2,0), (3,1), (4,2)])
 
788
        self.assertEquals(unique_lcs('cdeab', 'abcde'), [(0,2), (1,3), (2,4)])
 
789
        self.assertEquals(unique_lcs('abXde', 'abYde'), [(0,0), (1,1), 
 
790
                                                         (3,3), (4,4)])
 
791
        self.assertEquals(unique_lcs('acbac', 'abc'), [(2,1)])
 
792
 
 
793
    def test_recurse_matches(self):
 
794
        def test_one(a, b, matches):
 
795
            test_matches = []
 
796
            self._recurse_matches(
 
797
                a, b, 0, 0, len(a), len(b), test_matches, 10)
 
798
            self.assertEquals(test_matches, matches)
 
799
 
 
800
        test_one(['a', '', 'b', '', 'c'], ['a', 'a', 'b', 'c', 'c'],
 
801
                 [(0, 0), (2, 2), (4, 4)])
 
802
        test_one(['a', 'c', 'b', 'a', 'c'], ['a', 'b', 'c'],
 
803
                 [(0, 0), (2, 1), (4, 2)])
 
804
        # Even though 'bc' is not unique globally, and is surrounded by
 
805
        # non-matching lines, we should still match, because they are locally
 
806
        # unique
 
807
        test_one('abcdbce', 'afbcgdbce', [(0,0), (1, 2), (2, 3), (3, 5),
 
808
                                          (4, 6), (5, 7), (6, 8)])
 
809
 
 
810
        # recurse_matches doesn't match non-unique 
 
811
        # lines surrounded by bogus text.
 
812
        # The update has been done in patiencediff.SequenceMatcher instead
 
813
 
 
814
        # This is what it could be
 
815
        #test_one('aBccDe', 'abccde', [(0,0), (2,2), (3,3), (5,5)])
 
816
 
 
817
        # This is what it currently gives:
 
818
        test_one('aBccDe', 'abccde', [(0,0), (5,5)])
 
819
 
 
820
    def assertDiffBlocks(self, a, b, expected_blocks):
 
821
        """Check that the sequence matcher returns the correct blocks.
 
822
 
 
823
        :param a: A sequence to match
 
824
        :param b: Another sequence to match
 
825
        :param expected_blocks: The expected output, not including the final
 
826
            matching block (len(a), len(b), 0)
 
827
        """
 
828
        matcher = self._PatienceSequenceMatcher(None, a, b)
 
829
        blocks = matcher.get_matching_blocks()
 
830
        last = blocks.pop()
 
831
        self.assertEqual((len(a), len(b), 0), last)
 
832
        self.assertEqual(expected_blocks, blocks)
 
833
 
 
834
    def test_matching_blocks(self):
 
835
        # Some basic matching tests
 
836
        self.assertDiffBlocks('', '', [])
 
837
        self.assertDiffBlocks([], [], [])
 
838
        self.assertDiffBlocks('abc', '', [])
 
839
        self.assertDiffBlocks('', 'abc', [])
 
840
        self.assertDiffBlocks('abcd', 'abcd', [(0, 0, 4)])
 
841
        self.assertDiffBlocks('abcd', 'abce', [(0, 0, 3)])
 
842
        self.assertDiffBlocks('eabc', 'abce', [(1, 0, 3)])
 
843
        self.assertDiffBlocks('eabce', 'abce', [(1, 0, 4)])
 
844
        self.assertDiffBlocks('abcde', 'abXde', [(0, 0, 2), (3, 3, 2)])
 
845
        self.assertDiffBlocks('abcde', 'abXYZde', [(0, 0, 2), (3, 5, 2)])
 
846
        self.assertDiffBlocks('abde', 'abXYZde', [(0, 0, 2), (2, 5, 2)])
 
847
        # This may check too much, but it checks to see that
 
848
        # a copied block stays attached to the previous section,
 
849
        # not the later one.
 
850
        # difflib would tend to grab the trailing longest match
 
851
        # which would make the diff not look right
 
852
        self.assertDiffBlocks('abcdefghijklmnop', 'abcdefxydefghijklmnop',
 
853
                              [(0, 0, 6), (6, 11, 10)])
 
854
 
 
855
        # make sure it supports passing in lists
 
856
        self.assertDiffBlocks(
 
857
                   ['hello there\n',
 
858
                    'world\n',
 
859
                    'how are you today?\n'],
 
860
                   ['hello there\n',
 
861
                    'how are you today?\n'],
 
862
                [(0, 0, 1), (2, 1, 1)])
 
863
 
 
864
        # non unique lines surrounded by non-matching lines
 
865
        # won't be found
 
866
        self.assertDiffBlocks('aBccDe', 'abccde', [(0,0,1), (5,5,1)])
 
867
 
 
868
        # But they only need to be locally unique
 
869
        self.assertDiffBlocks('aBcDec', 'abcdec', [(0,0,1), (2,2,1), (4,4,2)])
 
870
 
 
871
        # non unique blocks won't be matched
 
872
        self.assertDiffBlocks('aBcdEcdFg', 'abcdecdfg', [(0,0,1), (8,8,1)])
 
873
 
 
874
        # but locally unique ones will
 
875
        self.assertDiffBlocks('aBcdEeXcdFg', 'abcdecdfg', [(0,0,1), (2,2,2),
 
876
                                              (5,4,1), (7,5,2), (10,8,1)])
 
877
 
 
878
        self.assertDiffBlocks('abbabbXd', 'cabbabxd', [(7,7,1)])
 
879
        self.assertDiffBlocks('abbabbbb', 'cabbabbc', [])
 
880
        self.assertDiffBlocks('bbbbbbbb', 'cbbbbbbc', [])
 
881
 
 
882
    def test_matching_blocks_tuples(self):
 
883
        # Some basic matching tests
 
884
        self.assertDiffBlocks([], [], [])
 
885
        self.assertDiffBlocks([('a',), ('b',), ('c,')], [], [])
 
886
        self.assertDiffBlocks([], [('a',), ('b',), ('c,')], [])
 
887
        self.assertDiffBlocks([('a',), ('b',), ('c,')],
 
888
                              [('a',), ('b',), ('c,')],
 
889
                              [(0, 0, 3)])
 
890
        self.assertDiffBlocks([('a',), ('b',), ('c,')],
 
891
                              [('a',), ('b',), ('d,')],
 
892
                              [(0, 0, 2)])
 
893
        self.assertDiffBlocks([('d',), ('b',), ('c,')],
 
894
                              [('a',), ('b',), ('c,')],
 
895
                              [(1, 1, 2)])
 
896
        self.assertDiffBlocks([('d',), ('a',), ('b',), ('c,')],
 
897
                              [('a',), ('b',), ('c,')],
 
898
                              [(1, 0, 3)])
 
899
        self.assertDiffBlocks([('a', 'b'), ('c', 'd'), ('e', 'f')],
 
900
                              [('a', 'b'), ('c', 'X'), ('e', 'f')],
 
901
                              [(0, 0, 1), (2, 2, 1)])
 
902
        self.assertDiffBlocks([('a', 'b'), ('c', 'd'), ('e', 'f')],
 
903
                              [('a', 'b'), ('c', 'dX'), ('e', 'f')],
 
904
                              [(0, 0, 1), (2, 2, 1)])
 
905
 
 
906
    def test_opcodes(self):
 
907
        def chk_ops(a, b, expected_codes):
 
908
            s = self._PatienceSequenceMatcher(None, a, b)
 
909
            self.assertEquals(expected_codes, s.get_opcodes())
 
910
 
 
911
        chk_ops('', '', [])
 
912
        chk_ops([], [], [])
 
913
        chk_ops('abc', '', [('delete', 0,3, 0,0)])
 
914
        chk_ops('', 'abc', [('insert', 0,0, 0,3)])
 
915
        chk_ops('abcd', 'abcd', [('equal',    0,4, 0,4)])
 
916
        chk_ops('abcd', 'abce', [('equal',   0,3, 0,3),
 
917
                                 ('replace', 3,4, 3,4)
 
918
                                ])
 
919
        chk_ops('eabc', 'abce', [('delete', 0,1, 0,0),
 
920
                                 ('equal',  1,4, 0,3),
 
921
                                 ('insert', 4,4, 3,4)
 
922
                                ])
 
923
        chk_ops('eabce', 'abce', [('delete', 0,1, 0,0),
 
924
                                  ('equal',  1,5, 0,4)
 
925
                                 ])
 
926
        chk_ops('abcde', 'abXde', [('equal',   0,2, 0,2),
 
927
                                   ('replace', 2,3, 2,3),
 
928
                                   ('equal',   3,5, 3,5)
 
929
                                  ])
 
930
        chk_ops('abcde', 'abXYZde', [('equal',   0,2, 0,2),
 
931
                                     ('replace', 2,3, 2,5),
 
932
                                     ('equal',   3,5, 5,7)
 
933
                                    ])
 
934
        chk_ops('abde', 'abXYZde', [('equal',  0,2, 0,2),
 
935
                                    ('insert', 2,2, 2,5),
 
936
                                    ('equal',  2,4, 5,7)
 
937
                                   ])
 
938
        chk_ops('abcdefghijklmnop', 'abcdefxydefghijklmnop',
 
939
                [('equal',  0,6,  0,6),
 
940
                 ('insert', 6,6,  6,11),
 
941
                 ('equal',  6,16, 11,21)
 
942
                ])
 
943
        chk_ops(
 
944
                [ 'hello there\n'
 
945
                , 'world\n'
 
946
                , 'how are you today?\n'],
 
947
                [ 'hello there\n'
 
948
                , 'how are you today?\n'],
 
949
                [('equal',  0,1, 0,1),
 
950
                 ('delete', 1,2, 1,1),
 
951
                 ('equal',  2,3, 1,2),
 
952
                ])
 
953
        chk_ops('aBccDe', 'abccde', 
 
954
                [('equal',   0,1, 0,1),
 
955
                 ('replace', 1,5, 1,5),
 
956
                 ('equal',   5,6, 5,6),
 
957
                ])
 
958
        chk_ops('aBcDec', 'abcdec', 
 
959
                [('equal',   0,1, 0,1),
 
960
                 ('replace', 1,2, 1,2),
 
961
                 ('equal',   2,3, 2,3),
 
962
                 ('replace', 3,4, 3,4),
 
963
                 ('equal',   4,6, 4,6),
 
964
                ])
 
965
        chk_ops('aBcdEcdFg', 'abcdecdfg', 
 
966
                [('equal',   0,1, 0,1),
 
967
                 ('replace', 1,8, 1,8),
 
968
                 ('equal',   8,9, 8,9)
 
969
                ])
 
970
        chk_ops('aBcdEeXcdFg', 'abcdecdfg', 
 
971
                [('equal',   0,1, 0,1),
 
972
                 ('replace', 1,2, 1,2),
 
973
                 ('equal',   2,4, 2,4),
 
974
                 ('delete', 4,5, 4,4),
 
975
                 ('equal',   5,6, 4,5),
 
976
                 ('delete', 6,7, 5,5),
 
977
                 ('equal',   7,9, 5,7),
 
978
                 ('replace', 9,10, 7,8),
 
979
                 ('equal',   10,11, 8,9)
 
980
                ])
 
981
 
 
982
    def test_grouped_opcodes(self):
 
983
        def chk_ops(a, b, expected_codes, n=3):
 
984
            s = self._PatienceSequenceMatcher(None, a, b)
 
985
            self.assertEquals(expected_codes, list(s.get_grouped_opcodes(n)))
 
986
 
 
987
        chk_ops('', '', [])
 
988
        chk_ops([], [], [])
 
989
        chk_ops('abc', '', [[('delete', 0,3, 0,0)]])
 
990
        chk_ops('', 'abc', [[('insert', 0,0, 0,3)]])
 
991
        chk_ops('abcd', 'abcd', [])
 
992
        chk_ops('abcd', 'abce', [[('equal',   0,3, 0,3),
 
993
                                  ('replace', 3,4, 3,4)
 
994
                                 ]])
 
995
        chk_ops('eabc', 'abce', [[('delete', 0,1, 0,0),
 
996
                                 ('equal',  1,4, 0,3),
 
997
                                 ('insert', 4,4, 3,4)
 
998
                                ]])
 
999
        chk_ops('abcdefghijklmnop', 'abcdefxydefghijklmnop',
 
1000
                [[('equal',  3,6, 3,6),
 
1001
                  ('insert', 6,6, 6,11),
 
1002
                  ('equal',  6,9, 11,14)
 
1003
                  ]])
 
1004
        chk_ops('abcdefghijklmnop', 'abcdefxydefghijklmnop',
 
1005
                [[('equal',  2,6, 2,6),
 
1006
                  ('insert', 6,6, 6,11),
 
1007
                  ('equal',  6,10, 11,15)
 
1008
                  ]], 4)
 
1009
        chk_ops('Xabcdef', 'abcdef',
 
1010
                [[('delete', 0,1, 0,0),
 
1011
                  ('equal',  1,4, 0,3)
 
1012
                  ]])
 
1013
        chk_ops('abcdef', 'abcdefX',
 
1014
                [[('equal',  3,6, 3,6),
 
1015
                  ('insert', 6,6, 6,7)
 
1016
                  ]])
 
1017
 
 
1018
 
 
1019
    def test_multiple_ranges(self):
 
1020
        # There was an earlier bug where we used a bad set of ranges,
 
1021
        # this triggers that specific bug, to make sure it doesn't regress
 
1022
        self.assertDiffBlocks('abcdefghijklmnop',
 
1023
                              'abcXghiYZQRSTUVWXYZijklmnop',
 
1024
                              [(0, 0, 3), (6, 4, 3), (9, 20, 7)])
 
1025
 
 
1026
        self.assertDiffBlocks('ABCd efghIjk  L',
 
1027
                              'AxyzBCn mo pqrstuvwI1 2  L',
 
1028
                              [(0,0,1), (1, 4, 2), (9, 19, 1), (12, 23, 3)])
 
1029
 
 
1030
        # These are rot13 code snippets.
 
1031
        self.assertDiffBlocks('''\
 
1032
    trg nqqrq jura lbh nqq n svyr va gur qverpgbel.
 
1033
    """
 
1034
    gnxrf_netf = ['svyr*']
 
1035
    gnxrf_bcgvbaf = ['ab-erphefr']
 
1036
  
 
1037
    qrs eha(frys, svyr_yvfg, ab_erphefr=Snyfr):
 
1038
        sebz omeyvo.nqq vzcbeg fzneg_nqq, nqq_ercbegre_cevag, nqq_ercbegre_ahyy
 
1039
        vs vf_dhvrg():
 
1040
            ercbegre = nqq_ercbegre_ahyy
 
1041
        ryfr:
 
1042
            ercbegre = nqq_ercbegre_cevag
 
1043
        fzneg_nqq(svyr_yvfg, abg ab_erphefr, ercbegre)
 
1044
 
 
1045
 
 
1046
pynff pzq_zxqve(Pbzznaq):
 
1047
'''.splitlines(True), '''\
 
1048
    trg nqqrq jura lbh nqq n svyr va gur qverpgbel.
 
1049
 
 
1050
    --qel-eha jvyy fubj juvpu svyrf jbhyq or nqqrq, ohg abg npghnyyl 
 
1051
    nqq gurz.
 
1052
    """
 
1053
    gnxrf_netf = ['svyr*']
 
1054
    gnxrf_bcgvbaf = ['ab-erphefr', 'qel-eha']
 
1055
 
 
1056
    qrs eha(frys, svyr_yvfg, ab_erphefr=Snyfr, qel_eha=Snyfr):
 
1057
        vzcbeg omeyvo.nqq
 
1058
 
 
1059
        vs qel_eha:
 
1060
            vs vf_dhvrg():
 
1061
                # Guvf vf cbvagyrff, ohg V'q engure abg envfr na reebe
 
1062
                npgvba = omeyvo.nqq.nqq_npgvba_ahyy
 
1063
            ryfr:
 
1064
  npgvba = omeyvo.nqq.nqq_npgvba_cevag
 
1065
        ryvs vf_dhvrg():
 
1066
            npgvba = omeyvo.nqq.nqq_npgvba_nqq
 
1067
        ryfr:
 
1068
       npgvba = omeyvo.nqq.nqq_npgvba_nqq_naq_cevag
 
1069
 
 
1070
        omeyvo.nqq.fzneg_nqq(svyr_yvfg, abg ab_erphefr, npgvba)
 
1071
 
 
1072
 
 
1073
pynff pzq_zxqve(Pbzznaq):
 
1074
'''.splitlines(True)
 
1075
, [(0,0,1), (1, 4, 2), (9, 19, 1), (12, 23, 3)])
 
1076
 
 
1077
    def test_patience_unified_diff(self):
 
1078
        txt_a = ['hello there\n',
 
1079
                 'world\n',
 
1080
                 'how are you today?\n']
 
1081
        txt_b = ['hello there\n',
 
1082
                 'how are you today?\n']
 
1083
        unified_diff = bzrlib.patiencediff.unified_diff
 
1084
        psm = self._PatienceSequenceMatcher
 
1085
        self.assertEquals([ '---  \n',
 
1086
                           '+++  \n',
 
1087
                           '@@ -1,3 +1,2 @@\n',
 
1088
                           ' hello there\n',
 
1089
                           '-world\n',
 
1090
                           ' how are you today?\n'
 
1091
                          ]
 
1092
                          , list(unified_diff(txt_a, txt_b,
 
1093
                                 sequencematcher=psm)))
 
1094
        txt_a = map(lambda x: x+'\n', 'abcdefghijklmnop')
 
1095
        txt_b = map(lambda x: x+'\n', 'abcdefxydefghijklmnop')
 
1096
        # This is the result with LongestCommonSubstring matching
 
1097
        self.assertEquals(['---  \n',
 
1098
                           '+++  \n',
 
1099
                           '@@ -1,6 +1,11 @@\n',
 
1100
                           ' a\n',
 
1101
                           ' b\n',
 
1102
                           ' c\n',
 
1103
                           '+d\n',
 
1104
                           '+e\n',
 
1105
                           '+f\n',
 
1106
                           '+x\n',
 
1107
                           '+y\n',
 
1108
                           ' d\n',
 
1109
                           ' e\n',
 
1110
                           ' f\n']
 
1111
                          , list(unified_diff(txt_a, txt_b)))
 
1112
        # And the patience diff
 
1113
        self.assertEquals(['---  \n',
 
1114
                           '+++  \n',
 
1115
                           '@@ -4,6 +4,11 @@\n',
 
1116
                           ' d\n',
 
1117
                           ' e\n',
 
1118
                           ' f\n',
 
1119
                           '+x\n',
 
1120
                           '+y\n',
 
1121
                           '+d\n',
 
1122
                           '+e\n',
 
1123
                           '+f\n',
 
1124
                           ' g\n',
 
1125
                           ' h\n',
 
1126
                           ' i\n',
 
1127
                          ]
 
1128
                          , list(unified_diff(txt_a, txt_b,
 
1129
                                 sequencematcher=psm)))
 
1130
 
 
1131
 
 
1132
class TestPatienceDiffLib_c(TestPatienceDiffLib):
 
1133
 
 
1134
    _test_needs_features = [CompiledPatienceDiffFeature]
 
1135
 
 
1136
    def setUp(self):
 
1137
        super(TestPatienceDiffLib_c, self).setUp()
 
1138
        import bzrlib._patiencediff_c
 
1139
        self._unique_lcs = bzrlib._patiencediff_c.unique_lcs_c
 
1140
        self._recurse_matches = bzrlib._patiencediff_c.recurse_matches_c
 
1141
        self._PatienceSequenceMatcher = \
 
1142
            bzrlib._patiencediff_c.PatienceSequenceMatcher_c
 
1143
 
 
1144
    def test_unhashable(self):
 
1145
        """We should get a proper exception here."""
 
1146
        # We need to be able to hash items in the sequence, lists are
 
1147
        # unhashable, and thus cannot be diffed
 
1148
        e = self.assertRaises(TypeError, self._PatienceSequenceMatcher,
 
1149
                                         None, [[]], [])
 
1150
        e = self.assertRaises(TypeError, self._PatienceSequenceMatcher,
 
1151
                                         None, ['valid', []], [])
 
1152
        e = self.assertRaises(TypeError, self._PatienceSequenceMatcher,
 
1153
                                         None, ['valid'], [[]])
 
1154
        e = self.assertRaises(TypeError, self._PatienceSequenceMatcher,
 
1155
                                         None, ['valid'], ['valid', []])
 
1156
 
 
1157
 
 
1158
class TestPatienceDiffLibFiles(TestCaseInTempDir):
 
1159
 
 
1160
    def setUp(self):
 
1161
        super(TestPatienceDiffLibFiles, self).setUp()
 
1162
        self._PatienceSequenceMatcher = \
 
1163
            bzrlib._patiencediff_py.PatienceSequenceMatcher_py
 
1164
 
 
1165
    def test_patience_unified_diff_files(self):
 
1166
        txt_a = ['hello there\n',
 
1167
                 'world\n',
 
1168
                 'how are you today?\n']
 
1169
        txt_b = ['hello there\n',
 
1170
                 'how are you today?\n']
 
1171
        open('a1', 'wb').writelines(txt_a)
 
1172
        open('b1', 'wb').writelines(txt_b)
 
1173
 
 
1174
        unified_diff_files = bzrlib.patiencediff.unified_diff_files
 
1175
        psm = self._PatienceSequenceMatcher
 
1176
        self.assertEquals(['--- a1 \n',
 
1177
                           '+++ b1 \n',
 
1178
                           '@@ -1,3 +1,2 @@\n',
 
1179
                           ' hello there\n',
 
1180
                           '-world\n',
 
1181
                           ' how are you today?\n',
 
1182
                          ]
 
1183
                          , list(unified_diff_files('a1', 'b1',
 
1184
                                 sequencematcher=psm)))
 
1185
 
 
1186
        txt_a = map(lambda x: x+'\n', 'abcdefghijklmnop')
 
1187
        txt_b = map(lambda x: x+'\n', 'abcdefxydefghijklmnop')
 
1188
        open('a2', 'wb').writelines(txt_a)
 
1189
        open('b2', 'wb').writelines(txt_b)
 
1190
 
 
1191
        # This is the result with LongestCommonSubstring matching
 
1192
        self.assertEquals(['--- a2 \n',
 
1193
                           '+++ b2 \n',
 
1194
                           '@@ -1,6 +1,11 @@\n',
 
1195
                           ' a\n',
 
1196
                           ' b\n',
 
1197
                           ' c\n',
 
1198
                           '+d\n',
 
1199
                           '+e\n',
 
1200
                           '+f\n',
 
1201
                           '+x\n',
 
1202
                           '+y\n',
 
1203
                           ' d\n',
 
1204
                           ' e\n',
 
1205
                           ' f\n']
 
1206
                          , list(unified_diff_files('a2', 'b2')))
 
1207
 
 
1208
        # And the patience diff
 
1209
        self.assertEquals(['--- a2 \n',
 
1210
                           '+++ b2 \n',
 
1211
                           '@@ -4,6 +4,11 @@\n',
 
1212
                           ' d\n',
 
1213
                           ' e\n',
 
1214
                           ' f\n',
 
1215
                           '+x\n',
 
1216
                           '+y\n',
 
1217
                           '+d\n',
 
1218
                           '+e\n',
 
1219
                           '+f\n',
 
1220
                           ' g\n',
 
1221
                           ' h\n',
 
1222
                           ' i\n',
 
1223
                          ]
 
1224
                          , list(unified_diff_files('a2', 'b2',
 
1225
                                 sequencematcher=psm)))
 
1226
 
 
1227
 
 
1228
class TestPatienceDiffLibFiles_c(TestPatienceDiffLibFiles):
 
1229
 
 
1230
    _test_needs_features = [CompiledPatienceDiffFeature]
 
1231
 
 
1232
    def setUp(self):
 
1233
        super(TestPatienceDiffLibFiles_c, self).setUp()
 
1234
        import bzrlib._patiencediff_c
 
1235
        self._PatienceSequenceMatcher = \
 
1236
            bzrlib._patiencediff_c.PatienceSequenceMatcher_c
 
1237
 
 
1238
 
 
1239
class TestUsingCompiledIfAvailable(TestCase):
 
1240
 
 
1241
    def test_PatienceSequenceMatcher(self):
 
1242
        if CompiledPatienceDiffFeature.available():
 
1243
            from bzrlib._patiencediff_c import PatienceSequenceMatcher_c
 
1244
            self.assertIs(PatienceSequenceMatcher_c,
 
1245
                          bzrlib.patiencediff.PatienceSequenceMatcher)
 
1246
        else:
 
1247
            from bzrlib._patiencediff_py import PatienceSequenceMatcher_py
 
1248
            self.assertIs(PatienceSequenceMatcher_py,
 
1249
                          bzrlib.patiencediff.PatienceSequenceMatcher)
 
1250
 
 
1251
    def test_unique_lcs(self):
 
1252
        if CompiledPatienceDiffFeature.available():
 
1253
            from bzrlib._patiencediff_c import unique_lcs_c
 
1254
            self.assertIs(unique_lcs_c,
 
1255
                          bzrlib.patiencediff.unique_lcs)
 
1256
        else:
 
1257
            from bzrlib._patiencediff_py import unique_lcs_py
 
1258
            self.assertIs(unique_lcs_py,
 
1259
                          bzrlib.patiencediff.unique_lcs)
 
1260
 
 
1261
    def test_recurse_matches(self):
 
1262
        if CompiledPatienceDiffFeature.available():
 
1263
            from bzrlib._patiencediff_c import recurse_matches_c
 
1264
            self.assertIs(recurse_matches_c,
 
1265
                          bzrlib.patiencediff.recurse_matches)
 
1266
        else:
 
1267
            from bzrlib._patiencediff_py import recurse_matches_py
 
1268
            self.assertIs(recurse_matches_py,
 
1269
                          bzrlib.patiencediff.recurse_matches)
 
1270
 
 
1271
 
 
1272
class TestDiffFromTool(TestCaseWithTransport):
 
1273
 
 
1274
    def test_from_string(self):
 
1275
        diff_obj = DiffFromTool.from_string('diff', None, None, None)
 
1276
        self.addCleanup(diff_obj.finish)
 
1277
        self.assertEqual(['diff', '%(old_path)s', '%(new_path)s'],
 
1278
            diff_obj.command_template)
 
1279
 
 
1280
    def test_from_string_u5(self):
 
1281
        diff_obj = DiffFromTool.from_string('diff -u\\ 5', None, None, None)
 
1282
        self.addCleanup(diff_obj.finish)
 
1283
        self.assertEqual(['diff', '-u 5', '%(old_path)s', '%(new_path)s'],
 
1284
                         diff_obj.command_template)
 
1285
        self.assertEqual(['diff', '-u 5', 'old-path', 'new-path'],
 
1286
                         diff_obj._get_command('old-path', 'new-path'))
 
1287
 
 
1288
    def test_execute(self):
 
1289
        output = StringIO()
 
1290
        diff_obj = DiffFromTool(['python', '-c',
 
1291
                                 'print "%(old_path)s %(new_path)s"'],
 
1292
                                None, None, output)
 
1293
        self.addCleanup(diff_obj.finish)
 
1294
        diff_obj._execute('old', 'new')
 
1295
        self.assertEqual(output.getvalue().rstrip(), 'old new')
 
1296
 
 
1297
    def test_excute_missing(self):
 
1298
        diff_obj = DiffFromTool(['a-tool-which-is-unlikely-to-exist'],
 
1299
                                None, None, None)
 
1300
        self.addCleanup(diff_obj.finish)
 
1301
        e = self.assertRaises(ExecutableMissing, diff_obj._execute, 'old',
 
1302
                              'new')
 
1303
        self.assertEqual('a-tool-which-is-unlikely-to-exist could not be found'
 
1304
                         ' on this machine', str(e))
 
1305
 
 
1306
    def test_prepare_files_creates_paths_readable_by_windows_tool(self):
 
1307
        self.requireFeature(AttribFeature)
 
1308
        output = StringIO()
 
1309
        tree = self.make_branch_and_tree('tree')
 
1310
        self.build_tree_contents([('tree/file', 'content')])
 
1311
        tree.add('file', 'file-id')
 
1312
        tree.commit('old tree')
 
1313
        tree.lock_read()
 
1314
        self.addCleanup(tree.unlock)
 
1315
        diff_obj = DiffFromTool(['python', '-c',
 
1316
                                 'print "%(old_path)s %(new_path)s"'],
 
1317
                                tree, tree, output)
 
1318
        diff_obj._prepare_files('file-id', 'file', 'file')
 
1319
        self.assertReadableByAttrib(diff_obj._root, 'old\\file', r'old\\file')
 
1320
        self.assertReadableByAttrib(diff_obj._root, 'new\\file', r'new\\file')
 
1321
 
 
1322
    def assertReadableByAttrib(self, cwd, relpath, regex):
 
1323
        proc = subprocess.Popen(['attrib', relpath],
 
1324
                                stdout=subprocess.PIPE,
 
1325
                                cwd=cwd)
 
1326
        proc.wait()
 
1327
        result = proc.stdout.read()
 
1328
        self.assertContainsRe(result, regex)
 
1329
 
 
1330
    def test_prepare_files(self):
 
1331
        output = StringIO()
 
1332
        tree = self.make_branch_and_tree('tree')
 
1333
        self.build_tree_contents([('tree/oldname', 'oldcontent')])
 
1334
        self.build_tree_contents([('tree/oldname2', 'oldcontent2')])
 
1335
        tree.add('oldname', 'file-id')
 
1336
        tree.add('oldname2', 'file2-id')
 
1337
        tree.commit('old tree', timestamp=0)
 
1338
        tree.rename_one('oldname', 'newname')
 
1339
        tree.rename_one('oldname2', 'newname2')
 
1340
        self.build_tree_contents([('tree/newname', 'newcontent')])
 
1341
        self.build_tree_contents([('tree/newname2', 'newcontent2')])
 
1342
        old_tree = tree.basis_tree()
 
1343
        old_tree.lock_read()
 
1344
        self.addCleanup(old_tree.unlock)
 
1345
        tree.lock_read()
 
1346
        self.addCleanup(tree.unlock)
 
1347
        diff_obj = DiffFromTool(['python', '-c',
 
1348
                                 'print "%(old_path)s %(new_path)s"'],
 
1349
                                old_tree, tree, output)
 
1350
        self.addCleanup(diff_obj.finish)
 
1351
        self.assertContainsRe(diff_obj._root, 'bzr-diff-[^/]*')
 
1352
        old_path, new_path = diff_obj._prepare_files('file-id', 'oldname',
 
1353
                                                     'newname')
 
1354
        self.assertContainsRe(old_path, 'old/oldname$')
 
1355
        self.assertEqual(0, os.stat(old_path).st_mtime)
 
1356
        self.assertContainsRe(new_path, 'new/newname$')
 
1357
        self.assertFileEqual('oldcontent', old_path)
 
1358
        self.assertFileEqual('newcontent', new_path)
 
1359
        if osutils.host_os_dereferences_symlinks():
 
1360
            self.assertTrue(os.path.samefile('tree/newname', new_path))
 
1361
        # make sure we can create files with the same parent directories
 
1362
        diff_obj._prepare_files('file2-id', 'oldname2', 'newname2')