~bzr-pqm/bzr/bzr.dev

« back to all changes in this revision

Viewing changes to bzrlib/tests/test_diff.py

  • Committer: Canonical.com Patch Queue Manager
  • Date: 2008-10-31 03:15:38 UTC
  • mfrom: (3806.1.1 contrib)
  • Revision ID: pqm@pqm.ubuntu.com-20081031031538-0ih0zi2owevf6fwu
Add contrib/bzr_ssh_path_limiter. (Andrew Bennetts)

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.revision as _mod_revision
 
39
import bzrlib.transform as transform
 
40
import bzrlib.patiencediff
 
41
import bzrlib._patiencediff_py
 
42
from bzrlib.tests import (Feature, TestCase, TestCaseWithTransport,
 
43
                          TestCaseInTempDir, TestSkipped)
 
44
 
 
45
 
 
46
class _AttribFeature(Feature):
 
47
 
 
48
    def _probe(self):
 
49
        if (sys.platform not in ('cygwin', 'win32')):
 
50
            return False
 
51
        try:
 
52
            proc = subprocess.Popen(['attrib', '.'], stdout=subprocess.PIPE)
 
53
        except OSError, e:
 
54
            return False
 
55
        return (0 == proc.wait())
 
56
 
 
57
    def feature_name(self):
 
58
        return 'attrib Windows command-line tool'
 
59
 
 
60
AttribFeature = _AttribFeature()
 
61
 
 
62
 
 
63
class _CompiledPatienceDiffFeature(Feature):
 
64
 
 
65
    def _probe(self):
 
66
        try:
 
67
            import bzrlib._patiencediff_c
 
68
        except ImportError:
 
69
            return False
 
70
        return True
 
71
 
 
72
    def feature_name(self):
 
73
        return 'bzrlib._patiencediff_c'
 
74
 
 
75
CompiledPatienceDiffFeature = _CompiledPatienceDiffFeature()
 
76
 
 
77
 
 
78
def udiff_lines(old, new, allow_binary=False):
 
79
    output = StringIO()
 
80
    internal_diff('old', old, 'new', new, output, allow_binary)
 
81
    output.seek(0, 0)
 
82
    return output.readlines()
 
83
 
 
84
 
 
85
def external_udiff_lines(old, new, use_stringio=False):
 
86
    if use_stringio:
 
87
        # StringIO has no fileno, so it tests a different codepath
 
88
        output = StringIO()
 
89
    else:
 
90
        output = TemporaryFile()
 
91
    try:
 
92
        external_diff('old', old, 'new', new, output, diff_opts=['-u'])
 
93
    except NoDiff:
 
94
        raise TestSkipped('external "diff" not present to test')
 
95
    output.seek(0, 0)
 
96
    lines = output.readlines()
 
97
    output.close()
 
98
    return lines
 
99
 
 
100
 
 
101
class TestDiff(TestCase):
 
102
 
 
103
    def test_add_nl(self):
 
104
        """diff generates a valid diff for patches that add a newline"""
 
105
        lines = udiff_lines(['boo'], ['boo\n'])
 
106
        self.check_patch(lines)
 
107
        self.assertEquals(lines[4], '\\ No newline at end of file\n')
 
108
            ## "expected no-nl, got %r" % lines[4]
 
109
 
 
110
    def test_add_nl_2(self):
 
111
        """diff generates a valid diff for patches that change last line and
 
112
        add a newline.
 
113
        """
 
114
        lines = udiff_lines(['boo'], ['goo\n'])
 
115
        self.check_patch(lines)
 
116
        self.assertEquals(lines[4], '\\ No newline at end of file\n')
 
117
            ## "expected no-nl, got %r" % lines[4]
 
118
 
 
119
    def test_remove_nl(self):
 
120
        """diff generates a valid diff for patches that change last line and
 
121
        add a newline.
 
122
        """
 
123
        lines = udiff_lines(['boo\n'], ['boo'])
 
124
        self.check_patch(lines)
 
125
        self.assertEquals(lines[5], '\\ No newline at end of file\n')
 
126
            ## "expected no-nl, got %r" % lines[5]
 
127
 
 
128
    def check_patch(self, lines):
 
129
        self.assert_(len(lines) > 1)
 
130
            ## "Not enough lines for a file header for patch:\n%s" % "".join(lines)
 
131
        self.assert_(lines[0].startswith ('---'))
 
132
            ## 'No orig line for patch:\n%s' % "".join(lines)
 
133
        self.assert_(lines[1].startswith ('+++'))
 
134
            ## 'No mod line for patch:\n%s' % "".join(lines)
 
135
        self.assert_(len(lines) > 2)
 
136
            ## "No hunks for patch:\n%s" % "".join(lines)
 
137
        self.assert_(lines[2].startswith('@@'))
 
138
            ## "No hunk header for patch:\n%s" % "".join(lines)
 
139
        self.assert_('@@' in lines[2][2:])
 
140
            ## "Unterminated hunk header for patch:\n%s" % "".join(lines)
 
141
 
 
142
    def test_binary_lines(self):
 
143
        self.assertRaises(BinaryFile, udiff_lines, [1023 * 'a' + '\x00'], [])
 
144
        self.assertRaises(BinaryFile, udiff_lines, [], [1023 * 'a' + '\x00'])
 
145
        udiff_lines([1023 * 'a' + '\x00'], [], allow_binary=True)
 
146
        udiff_lines([], [1023 * 'a' + '\x00'], allow_binary=True)
 
147
 
 
148
    def test_external_diff(self):
 
149
        lines = external_udiff_lines(['boo\n'], ['goo\n'])
 
150
        self.check_patch(lines)
 
151
        self.assertEqual('\n', lines[-1])
 
152
 
 
153
    def test_external_diff_no_fileno(self):
 
154
        # Make sure that we can handle not having a fileno, even
 
155
        # if the diff is large
 
156
        lines = external_udiff_lines(['boo\n']*10000,
 
157
                                     ['goo\n']*10000,
 
158
                                     use_stringio=True)
 
159
        self.check_patch(lines)
 
160
 
 
161
    def test_external_diff_binary_lang_c(self):
 
162
        old_env = {}
 
163
        for lang in ('LANG', 'LC_ALL', 'LANGUAGE'):
 
164
            old_env[lang] = osutils.set_or_unset_env(lang, 'C')
 
165
        try:
 
166
            lines = external_udiff_lines(['\x00foobar\n'], ['foo\x00bar\n'])
 
167
            # Older versions of diffutils say "Binary files", newer
 
168
            # versions just say "Files".
 
169
            self.assertContainsRe(lines[0],
 
170
                                  '(Binary f|F)iles old and new differ\n')
 
171
            self.assertEquals(lines[1:], ['\n'])
 
172
        finally:
 
173
            for lang, old_val in old_env.iteritems():
 
174
                osutils.set_or_unset_env(lang, old_val)
 
175
 
 
176
    def test_no_external_diff(self):
 
177
        """Check that NoDiff is raised when diff is not available"""
 
178
        # Use os.environ['PATH'] to make sure no 'diff' command is available
 
179
        orig_path = os.environ['PATH']
 
180
        try:
 
181
            os.environ['PATH'] = ''
 
182
            self.assertRaises(NoDiff, external_diff,
 
183
                              'old', ['boo\n'], 'new', ['goo\n'],
 
184
                              StringIO(), diff_opts=['-u'])
 
185
        finally:
 
186
            os.environ['PATH'] = orig_path
 
187
        
 
188
    def test_internal_diff_default(self):
 
189
        # Default internal diff encoding is utf8
 
190
        output = StringIO()
 
191
        internal_diff(u'old_\xb5', ['old_text\n'],
 
192
                    u'new_\xe5', ['new_text\n'], output)
 
193
        lines = output.getvalue().splitlines(True)
 
194
        self.check_patch(lines)
 
195
        self.assertEquals(['--- old_\xc2\xb5\n',
 
196
                           '+++ new_\xc3\xa5\n',
 
197
                           '@@ -1,1 +1,1 @@\n',
 
198
                           '-old_text\n',
 
199
                           '+new_text\n',
 
200
                           '\n',
 
201
                          ]
 
202
                          , lines)
 
203
 
 
204
    def test_internal_diff_utf8(self):
 
205
        output = StringIO()
 
206
        internal_diff(u'old_\xb5', ['old_text\n'],
 
207
                    u'new_\xe5', ['new_text\n'], output,
 
208
                    path_encoding='utf8')
 
209
        lines = output.getvalue().splitlines(True)
 
210
        self.check_patch(lines)
 
211
        self.assertEquals(['--- old_\xc2\xb5\n',
 
212
                           '+++ new_\xc3\xa5\n',
 
213
                           '@@ -1,1 +1,1 @@\n',
 
214
                           '-old_text\n',
 
215
                           '+new_text\n',
 
216
                           '\n',
 
217
                          ]
 
218
                          , lines)
 
219
 
 
220
    def test_internal_diff_iso_8859_1(self):
 
221
        output = StringIO()
 
222
        internal_diff(u'old_\xb5', ['old_text\n'],
 
223
                    u'new_\xe5', ['new_text\n'], output,
 
224
                    path_encoding='iso-8859-1')
 
225
        lines = output.getvalue().splitlines(True)
 
226
        self.check_patch(lines)
 
227
        self.assertEquals(['--- old_\xb5\n',
 
228
                           '+++ new_\xe5\n',
 
229
                           '@@ -1,1 +1,1 @@\n',
 
230
                           '-old_text\n',
 
231
                           '+new_text\n',
 
232
                           '\n',
 
233
                          ]
 
234
                          , lines)
 
235
 
 
236
    def test_internal_diff_no_content(self):
 
237
        output = StringIO()
 
238
        internal_diff(u'old', [], u'new', [], output)
 
239
        self.assertEqual('', output.getvalue())
 
240
 
 
241
    def test_internal_diff_no_changes(self):
 
242
        output = StringIO()
 
243
        internal_diff(u'old', ['text\n', 'contents\n'],
 
244
                      u'new', ['text\n', 'contents\n'],
 
245
                      output)
 
246
        self.assertEqual('', output.getvalue())
 
247
 
 
248
    def test_internal_diff_returns_bytes(self):
 
249
        import StringIO
 
250
        output = StringIO.StringIO()
 
251
        internal_diff(u'old_\xb5', ['old_text\n'],
 
252
                    u'new_\xe5', ['new_text\n'], output)
 
253
        self.failUnless(isinstance(output.getvalue(), str),
 
254
            'internal_diff should return bytestrings')
 
255
 
 
256
 
 
257
class TestDiffFiles(TestCaseInTempDir):
 
258
 
 
259
    def test_external_diff_binary(self):
 
260
        """The output when using external diff should use diff's i18n error"""
 
261
        # Make sure external_diff doesn't fail in the current LANG
 
262
        lines = external_udiff_lines(['\x00foobar\n'], ['foo\x00bar\n'])
 
263
 
 
264
        cmd = ['diff', '-u', '--binary', 'old', 'new']
 
265
        open('old', 'wb').write('\x00foobar\n')
 
266
        open('new', 'wb').write('foo\x00bar\n')
 
267
        pipe = subprocess.Popen(cmd, stdout=subprocess.PIPE,
 
268
                                     stdin=subprocess.PIPE)
 
269
        out, err = pipe.communicate()
 
270
        # Diff returns '2' on Binary files.
 
271
        self.assertEqual(2, pipe.returncode)
 
272
        # We should output whatever diff tells us, plus a trailing newline
 
273
        self.assertEqual(out.splitlines(True) + ['\n'], lines)
 
274
 
 
275
 
 
276
class TestShowDiffTreesHelper(TestCaseWithTransport):
 
277
    """Has a helper for running show_diff_trees"""
 
278
 
 
279
    def get_diff(self, tree1, tree2, specific_files=None, working_tree=None):
 
280
        output = StringIO()
 
281
        if working_tree is not None:
 
282
            extra_trees = (working_tree,)
 
283
        else:
 
284
            extra_trees = ()
 
285
        show_diff_trees(tree1, tree2, output, specific_files=specific_files,
 
286
                        extra_trees=extra_trees, old_label='old/',
 
287
                        new_label='new/')
 
288
        return output.getvalue()
 
289
 
 
290
 
 
291
class TestDiffDates(TestShowDiffTreesHelper):
 
292
 
 
293
    def setUp(self):
 
294
        super(TestDiffDates, self).setUp()
 
295
        self.wt = self.make_branch_and_tree('.')
 
296
        self.b = self.wt.branch
 
297
        self.build_tree_contents([
 
298
            ('file1', 'file1 contents at rev 1\n'),
 
299
            ('file2', 'file2 contents at rev 1\n')
 
300
            ])
 
301
        self.wt.add(['file1', 'file2'])
 
302
        self.wt.commit(
 
303
            message='Revision 1',
 
304
            timestamp=1143849600, # 2006-04-01 00:00:00 UTC
 
305
            timezone=0,
 
306
            rev_id='rev-1')
 
307
        self.build_tree_contents([('file1', 'file1 contents at rev 2\n')])
 
308
        self.wt.commit(
 
309
            message='Revision 2',
 
310
            timestamp=1143936000, # 2006-04-02 00:00:00 UTC
 
311
            timezone=28800,
 
312
            rev_id='rev-2')
 
313
        self.build_tree_contents([('file2', 'file2 contents at rev 3\n')])
 
314
        self.wt.commit(
 
315
            message='Revision 3',
 
316
            timestamp=1144022400, # 2006-04-03 00:00:00 UTC
 
317
            timezone=-3600,
 
318
            rev_id='rev-3')
 
319
        self.wt.remove(['file2'])
 
320
        self.wt.commit(
 
321
            message='Revision 4',
 
322
            timestamp=1144108800, # 2006-04-04 00:00:00 UTC
 
323
            timezone=0,
 
324
            rev_id='rev-4')
 
325
        self.build_tree_contents([
 
326
            ('file1', 'file1 contents in working tree\n')
 
327
            ])
 
328
        # set the date stamps for files in the working tree to known values
 
329
        os.utime('file1', (1144195200, 1144195200)) # 2006-04-05 00:00:00 UTC
 
330
 
 
331
    def test_diff_rev_tree_working_tree(self):
 
332
        output = self.get_diff(self.wt.basis_tree(), self.wt)
 
333
        # note that the date for old/file1 is from rev 2 rather than from
 
334
        # the basis revision (rev 4)
 
335
        self.assertEqualDiff(output, '''\
 
336
=== modified file 'file1'
 
337
--- old/file1\t2006-04-02 00:00:00 +0000
 
338
+++ new/file1\t2006-04-05 00:00:00 +0000
 
339
@@ -1,1 +1,1 @@
 
340
-file1 contents at rev 2
 
341
+file1 contents in working tree
 
342
 
 
343
''')
 
344
 
 
345
    def test_diff_rev_tree_rev_tree(self):
 
346
        tree1 = self.b.repository.revision_tree('rev-2')
 
347
        tree2 = self.b.repository.revision_tree('rev-3')
 
348
        output = self.get_diff(tree1, tree2)
 
349
        self.assertEqualDiff(output, '''\
 
350
=== modified file 'file2'
 
351
--- old/file2\t2006-04-01 00:00:00 +0000
 
352
+++ new/file2\t2006-04-03 00:00:00 +0000
 
353
@@ -1,1 +1,1 @@
 
354
-file2 contents at rev 1
 
355
+file2 contents at rev 3
 
356
 
 
357
''')
 
358
        
 
359
    def test_diff_add_files(self):
 
360
        tree1 = self.b.repository.revision_tree(_mod_revision.NULL_REVISION)
 
361
        tree2 = self.b.repository.revision_tree('rev-1')
 
362
        output = self.get_diff(tree1, tree2)
 
363
        # the files have the epoch time stamp for the tree in which
 
364
        # they don't exist.
 
365
        self.assertEqualDiff(output, '''\
 
366
=== added file 'file1'
 
367
--- old/file1\t1970-01-01 00:00:00 +0000
 
368
+++ new/file1\t2006-04-01 00:00:00 +0000
 
369
@@ -0,0 +1,1 @@
 
370
+file1 contents at rev 1
 
371
 
 
372
=== added file 'file2'
 
373
--- old/file2\t1970-01-01 00:00:00 +0000
 
374
+++ new/file2\t2006-04-01 00:00:00 +0000
 
375
@@ -0,0 +1,1 @@
 
376
+file2 contents at rev 1
 
377
 
 
378
''')
 
379
 
 
380
    def test_diff_remove_files(self):
 
381
        tree1 = self.b.repository.revision_tree('rev-3')
 
382
        tree2 = self.b.repository.revision_tree('rev-4')
 
383
        output = self.get_diff(tree1, tree2)
 
384
        # the file has the epoch time stamp for the tree in which
 
385
        # it doesn't exist.
 
386
        self.assertEqualDiff(output, '''\
 
387
=== removed file 'file2'
 
388
--- old/file2\t2006-04-03 00:00:00 +0000
 
389
+++ new/file2\t1970-01-01 00:00:00 +0000
 
390
@@ -1,1 +0,0 @@
 
391
-file2 contents at rev 3
 
392
 
 
393
''')
 
394
 
 
395
    def test_show_diff_specified(self):
 
396
        """A working tree filename can be used to identify a file"""
 
397
        self.wt.rename_one('file1', 'file1b')
 
398
        old_tree = self.b.repository.revision_tree('rev-1')
 
399
        new_tree = self.b.repository.revision_tree('rev-4')
 
400
        out = self.get_diff(old_tree, new_tree, specific_files=['file1b'], 
 
401
                            working_tree=self.wt)
 
402
        self.assertContainsRe(out, 'file1\t')
 
403
 
 
404
    def test_recursive_diff(self):
 
405
        """Children of directories are matched"""
 
406
        os.mkdir('dir1')
 
407
        os.mkdir('dir2')
 
408
        self.wt.add(['dir1', 'dir2'])
 
409
        self.wt.rename_one('file1', 'dir1/file1')
 
410
        old_tree = self.b.repository.revision_tree('rev-1')
 
411
        new_tree = self.b.repository.revision_tree('rev-4')
 
412
        out = self.get_diff(old_tree, new_tree, specific_files=['dir1'], 
 
413
                            working_tree=self.wt)
 
414
        self.assertContainsRe(out, 'file1\t')
 
415
        out = self.get_diff(old_tree, new_tree, specific_files=['dir2'], 
 
416
                            working_tree=self.wt)
 
417
        self.assertNotContainsRe(out, 'file1\t')
 
418
 
 
419
 
 
420
 
 
421
class TestShowDiffTrees(TestShowDiffTreesHelper):
 
422
    """Direct tests for show_diff_trees"""
 
423
 
 
424
    def test_modified_file(self):
 
425
        """Test when a file is modified."""
 
426
        tree = self.make_branch_and_tree('tree')
 
427
        self.build_tree_contents([('tree/file', 'contents\n')])
 
428
        tree.add(['file'], ['file-id'])
 
429
        tree.commit('one', rev_id='rev-1')
 
430
 
 
431
        self.build_tree_contents([('tree/file', 'new contents\n')])
 
432
        diff = self.get_diff(tree.basis_tree(), tree)
 
433
        self.assertContainsRe(diff, "=== modified file 'file'\n")
 
434
        self.assertContainsRe(diff, '--- old/file\t')
 
435
        self.assertContainsRe(diff, '\\+\\+\\+ new/file\t')
 
436
        self.assertContainsRe(diff, '-contents\n'
 
437
                                    '\\+new contents\n')
 
438
 
 
439
    def test_modified_file_in_renamed_dir(self):
 
440
        """Test when a file is modified in a renamed directory."""
 
441
        tree = self.make_branch_and_tree('tree')
 
442
        self.build_tree(['tree/dir/'])
 
443
        self.build_tree_contents([('tree/dir/file', 'contents\n')])
 
444
        tree.add(['dir', 'dir/file'], ['dir-id', 'file-id'])
 
445
        tree.commit('one', rev_id='rev-1')
 
446
 
 
447
        tree.rename_one('dir', 'other')
 
448
        self.build_tree_contents([('tree/other/file', 'new contents\n')])
 
449
        diff = self.get_diff(tree.basis_tree(), tree)
 
450
        self.assertContainsRe(diff, "=== renamed directory 'dir' => 'other'\n")
 
451
        self.assertContainsRe(diff, "=== modified file 'other/file'\n")
 
452
        # XXX: This is technically incorrect, because it used to be at another
 
453
        # location. What to do?
 
454
        self.assertContainsRe(diff, '--- old/dir/file\t')
 
455
        self.assertContainsRe(diff, '\\+\\+\\+ new/other/file\t')
 
456
        self.assertContainsRe(diff, '-contents\n'
 
457
                                    '\\+new contents\n')
 
458
 
 
459
    def test_renamed_directory(self):
 
460
        """Test when only a directory is only renamed."""
 
461
        tree = self.make_branch_and_tree('tree')
 
462
        self.build_tree(['tree/dir/'])
 
463
        self.build_tree_contents([('tree/dir/file', 'contents\n')])
 
464
        tree.add(['dir', 'dir/file'], ['dir-id', 'file-id'])
 
465
        tree.commit('one', rev_id='rev-1')
 
466
 
 
467
        tree.rename_one('dir', 'newdir')
 
468
        diff = self.get_diff(tree.basis_tree(), tree)
 
469
        # Renaming a directory should be a single "you renamed this dir" even
 
470
        # when there are files inside.
 
471
        self.assertEqual("=== renamed directory 'dir' => 'newdir'\n", diff)
 
472
 
 
473
    def test_renamed_file(self):
 
474
        """Test when a file is only renamed."""
 
475
        tree = self.make_branch_and_tree('tree')
 
476
        self.build_tree_contents([('tree/file', 'contents\n')])
 
477
        tree.add(['file'], ['file-id'])
 
478
        tree.commit('one', rev_id='rev-1')
 
479
 
 
480
        tree.rename_one('file', 'newname')
 
481
        diff = self.get_diff(tree.basis_tree(), tree)
 
482
        self.assertContainsRe(diff, "=== renamed file 'file' => 'newname'\n")
 
483
        # We shouldn't have a --- or +++ line, because there is no content
 
484
        # change
 
485
        self.assertNotContainsRe(diff, '---')
 
486
 
 
487
    def test_renamed_and_modified_file(self):
 
488
        """Test when a file is only renamed."""
 
489
        tree = self.make_branch_and_tree('tree')
 
490
        self.build_tree_contents([('tree/file', 'contents\n')])
 
491
        tree.add(['file'], ['file-id'])
 
492
        tree.commit('one', rev_id='rev-1')
 
493
 
 
494
        tree.rename_one('file', 'newname')
 
495
        self.build_tree_contents([('tree/newname', 'new contents\n')])
 
496
        diff = self.get_diff(tree.basis_tree(), tree)
 
497
        self.assertContainsRe(diff, "=== renamed file 'file' => 'newname'\n")
 
498
        self.assertContainsRe(diff, '--- old/file\t')
 
499
        self.assertContainsRe(diff, '\\+\\+\\+ new/newname\t')
 
500
        self.assertContainsRe(diff, '-contents\n'
 
501
                                    '\\+new contents\n')
 
502
 
 
503
 
 
504
    def test_internal_diff_exec_property(self):
 
505
        tree = self.make_branch_and_tree('tree')
 
506
 
 
507
        tt = transform.TreeTransform(tree)
 
508
        tt.new_file('a', tt.root, 'contents\n', 'a-id', True)
 
509
        tt.new_file('b', tt.root, 'contents\n', 'b-id', False)
 
510
        tt.new_file('c', tt.root, 'contents\n', 'c-id', True)
 
511
        tt.new_file('d', tt.root, 'contents\n', 'd-id', False)
 
512
        tt.new_file('e', tt.root, 'contents\n', 'control-e-id', True)
 
513
        tt.new_file('f', tt.root, 'contents\n', 'control-f-id', False)
 
514
        tt.apply()
 
515
        tree.commit('one', rev_id='rev-1')
 
516
 
 
517
        tt = transform.TreeTransform(tree)
 
518
        tt.set_executability(False, tt.trans_id_file_id('a-id'))
 
519
        tt.set_executability(True, tt.trans_id_file_id('b-id'))
 
520
        tt.set_executability(False, tt.trans_id_file_id('c-id'))
 
521
        tt.set_executability(True, tt.trans_id_file_id('d-id'))
 
522
        tt.apply()
 
523
        tree.rename_one('c', 'new-c')
 
524
        tree.rename_one('d', 'new-d')
 
525
 
 
526
        diff = self.get_diff(tree.basis_tree(), tree)
 
527
 
 
528
        self.assertContainsRe(diff, r"file 'a'.*\(properties changed:.*\+x to -x.*\)")
 
529
        self.assertContainsRe(diff, r"file 'b'.*\(properties changed:.*-x to \+x.*\)")
 
530
        self.assertContainsRe(diff, r"file 'c'.*\(properties changed:.*\+x to -x.*\)")
 
531
        self.assertContainsRe(diff, r"file 'd'.*\(properties changed:.*-x to \+x.*\)")
 
532
        self.assertNotContainsRe(diff, r"file 'e'")
 
533
        self.assertNotContainsRe(diff, r"file 'f'")
 
534
 
 
535
 
 
536
    def test_binary_unicode_filenames(self):
 
537
        """Test that contents of files are *not* encoded in UTF-8 when there
 
538
        is a binary file in the diff.
 
539
        """
 
540
        # See https://bugs.launchpad.net/bugs/110092.
 
541
        self.requireFeature(tests.UnicodeFilenameFeature)
 
542
 
 
543
        # This bug isn't triggered with cStringIO.
 
544
        from StringIO import StringIO
 
545
        tree = self.make_branch_and_tree('tree')
 
546
        alpha, omega = u'\u03b1', u'\u03c9'
 
547
        alpha_utf8, omega_utf8 = alpha.encode('utf8'), omega.encode('utf8')
 
548
        self.build_tree_contents(
 
549
            [('tree/' + alpha, chr(0)),
 
550
             ('tree/' + omega,
 
551
              ('The %s and the %s\n' % (alpha_utf8, omega_utf8)))])
 
552
        tree.add([alpha], ['file-id'])
 
553
        tree.add([omega], ['file-id-2'])
 
554
        diff_content = StringIO()
 
555
        show_diff_trees(tree.basis_tree(), tree, diff_content)
 
556
        diff = diff_content.getvalue()
 
557
        self.assertContainsRe(diff, r"=== added file '%s'" % alpha_utf8)
 
558
        self.assertContainsRe(
 
559
            diff, "Binary files a/%s.*and b/%s.* differ\n" % (alpha_utf8, alpha_utf8))
 
560
        self.assertContainsRe(diff, r"=== added file '%s'" % omega_utf8)
 
561
        self.assertContainsRe(diff, r"--- a/%s" % (omega_utf8,))
 
562
        self.assertContainsRe(diff, r"\+\+\+ b/%s" % (omega_utf8,))
 
563
 
 
564
    def test_unicode_filename(self):
 
565
        """Test when the filename are unicode."""
 
566
        self.requireFeature(tests.UnicodeFilenameFeature)
 
567
 
 
568
        alpha, omega = u'\u03b1', u'\u03c9'
 
569
        autf8, outf8 = alpha.encode('utf8'), omega.encode('utf8')
 
570
 
 
571
        tree = self.make_branch_and_tree('tree')
 
572
        self.build_tree_contents([('tree/ren_'+alpha, 'contents\n')])
 
573
        tree.add(['ren_'+alpha], ['file-id-2'])
 
574
        self.build_tree_contents([('tree/del_'+alpha, 'contents\n')])
 
575
        tree.add(['del_'+alpha], ['file-id-3'])
 
576
        self.build_tree_contents([('tree/mod_'+alpha, 'contents\n')])
 
577
        tree.add(['mod_'+alpha], ['file-id-4'])
 
578
 
 
579
        tree.commit('one', rev_id='rev-1')
 
580
 
 
581
        tree.rename_one('ren_'+alpha, 'ren_'+omega)
 
582
        tree.remove('del_'+alpha)
 
583
        self.build_tree_contents([('tree/add_'+alpha, 'contents\n')])
 
584
        tree.add(['add_'+alpha], ['file-id'])
 
585
        self.build_tree_contents([('tree/mod_'+alpha, 'contents_mod\n')])
 
586
 
 
587
        diff = self.get_diff(tree.basis_tree(), tree)
 
588
        self.assertContainsRe(diff,
 
589
                "=== renamed file 'ren_%s' => 'ren_%s'\n"%(autf8, outf8))
 
590
        self.assertContainsRe(diff, "=== added file 'add_%s'"%autf8)
 
591
        self.assertContainsRe(diff, "=== modified file 'mod_%s'"%autf8)
 
592
        self.assertContainsRe(diff, "=== removed file 'del_%s'"%autf8)
 
593
 
 
594
 
 
595
class DiffWasIs(DiffPath):
 
596
 
 
597
    def diff(self, file_id, old_path, new_path, old_kind, new_kind):
 
598
        self.to_file.write('was: ')
 
599
        self.to_file.write(self.old_tree.get_file(file_id).read())
 
600
        self.to_file.write('is: ')
 
601
        self.to_file.write(self.new_tree.get_file(file_id).read())
 
602
        pass
 
603
 
 
604
 
 
605
class TestDiffTree(TestCaseWithTransport):
 
606
 
 
607
    def setUp(self):
 
608
        TestCaseWithTransport.setUp(self)
 
609
        self.old_tree = self.make_branch_and_tree('old-tree')
 
610
        self.old_tree.lock_write()
 
611
        self.addCleanup(self.old_tree.unlock)
 
612
        self.new_tree = self.make_branch_and_tree('new-tree')
 
613
        self.new_tree.lock_write()
 
614
        self.addCleanup(self.new_tree.unlock)
 
615
        self.differ = DiffTree(self.old_tree, self.new_tree, StringIO())
 
616
 
 
617
    def test_diff_text(self):
 
618
        self.build_tree_contents([('old-tree/olddir/',),
 
619
                                  ('old-tree/olddir/oldfile', 'old\n')])
 
620
        self.old_tree.add('olddir')
 
621
        self.old_tree.add('olddir/oldfile', 'file-id')
 
622
        self.build_tree_contents([('new-tree/newdir/',),
 
623
                                  ('new-tree/newdir/newfile', 'new\n')])
 
624
        self.new_tree.add('newdir')
 
625
        self.new_tree.add('newdir/newfile', 'file-id')
 
626
        differ = DiffText(self.old_tree, self.new_tree, StringIO())
 
627
        differ.diff_text('file-id', None, 'old label', 'new label')
 
628
        self.assertEqual(
 
629
            '--- old label\n+++ new label\n@@ -1,1 +0,0 @@\n-old\n\n',
 
630
            differ.to_file.getvalue())
 
631
        differ.to_file.seek(0)
 
632
        differ.diff_text(None, 'file-id', 'old label', 'new label')
 
633
        self.assertEqual(
 
634
            '--- old label\n+++ new label\n@@ -0,0 +1,1 @@\n+new\n\n',
 
635
            differ.to_file.getvalue())
 
636
        differ.to_file.seek(0)
 
637
        differ.diff_text('file-id', 'file-id', 'old label', 'new label')
 
638
        self.assertEqual(
 
639
            '--- old label\n+++ new label\n@@ -1,1 +1,1 @@\n-old\n+new\n\n',
 
640
            differ.to_file.getvalue())
 
641
 
 
642
    def test_diff_deletion(self):
 
643
        self.build_tree_contents([('old-tree/file', 'contents'),
 
644
                                  ('new-tree/file', 'contents')])
 
645
        self.old_tree.add('file', 'file-id')
 
646
        self.new_tree.add('file', 'file-id')
 
647
        os.unlink('new-tree/file')
 
648
        self.differ.show_diff(None)
 
649
        self.assertContainsRe(self.differ.to_file.getvalue(), '-contents')
 
650
 
 
651
    def test_diff_creation(self):
 
652
        self.build_tree_contents([('old-tree/file', 'contents'),
 
653
                                  ('new-tree/file', 'contents')])
 
654
        self.old_tree.add('file', 'file-id')
 
655
        self.new_tree.add('file', 'file-id')
 
656
        os.unlink('old-tree/file')
 
657
        self.differ.show_diff(None)
 
658
        self.assertContainsRe(self.differ.to_file.getvalue(), '\+contents')
 
659
 
 
660
    def test_diff_symlink(self):
 
661
        differ = DiffSymlink(self.old_tree, self.new_tree, StringIO())
 
662
        differ.diff_symlink('old target', None)
 
663
        self.assertEqual("=== target was 'old target'\n",
 
664
                         differ.to_file.getvalue())
 
665
 
 
666
        differ = DiffSymlink(self.old_tree, self.new_tree, StringIO())
 
667
        differ.diff_symlink(None, 'new target')
 
668
        self.assertEqual("=== target is 'new target'\n",
 
669
                         differ.to_file.getvalue())
 
670
 
 
671
        differ = DiffSymlink(self.old_tree, self.new_tree, StringIO())
 
672
        differ.diff_symlink('old target', 'new target')
 
673
        self.assertEqual("=== target changed 'old target' => 'new target'\n",
 
674
                         differ.to_file.getvalue())
 
675
 
 
676
    def test_diff(self):
 
677
        self.build_tree_contents([('old-tree/olddir/',),
 
678
                                  ('old-tree/olddir/oldfile', 'old\n')])
 
679
        self.old_tree.add('olddir')
 
680
        self.old_tree.add('olddir/oldfile', 'file-id')
 
681
        self.build_tree_contents([('new-tree/newdir/',),
 
682
                                  ('new-tree/newdir/newfile', 'new\n')])
 
683
        self.new_tree.add('newdir')
 
684
        self.new_tree.add('newdir/newfile', 'file-id')
 
685
        self.differ.diff('file-id', 'olddir/oldfile', 'newdir/newfile')
 
686
        self.assertContainsRe(
 
687
            self.differ.to_file.getvalue(),
 
688
            r'--- olddir/oldfile.*\n\+\+\+ newdir/newfile.*\n\@\@ -1,1 \+1,1'
 
689
             ' \@\@\n-old\n\+new\n\n')
 
690
 
 
691
    def test_diff_kind_change(self):
 
692
        self.requireFeature(tests.SymlinkFeature)
 
693
        self.build_tree_contents([('old-tree/olddir/',),
 
694
                                  ('old-tree/olddir/oldfile', 'old\n')])
 
695
        self.old_tree.add('olddir')
 
696
        self.old_tree.add('olddir/oldfile', 'file-id')
 
697
        self.build_tree(['new-tree/newdir/'])
 
698
        os.symlink('new', 'new-tree/newdir/newfile')
 
699
        self.new_tree.add('newdir')
 
700
        self.new_tree.add('newdir/newfile', 'file-id')
 
701
        self.differ.diff('file-id', 'olddir/oldfile', 'newdir/newfile')
 
702
        self.assertContainsRe(
 
703
            self.differ.to_file.getvalue(),
 
704
            r'--- olddir/oldfile.*\n\+\+\+ newdir/newfile.*\n\@\@ -1,1 \+0,0'
 
705
             ' \@\@\n-old\n\n')
 
706
        self.assertContainsRe(self.differ.to_file.getvalue(),
 
707
                              "=== target is 'new'\n")
 
708
 
 
709
    def test_diff_directory(self):
 
710
        self.build_tree(['new-tree/new-dir/'])
 
711
        self.new_tree.add('new-dir', 'new-dir-id')
 
712
        self.differ.diff('new-dir-id', None, 'new-dir')
 
713
        self.assertEqual(self.differ.to_file.getvalue(), '')
 
714
 
 
715
    def create_old_new(self):
 
716
        self.build_tree_contents([('old-tree/olddir/',),
 
717
                                  ('old-tree/olddir/oldfile', 'old\n')])
 
718
        self.old_tree.add('olddir')
 
719
        self.old_tree.add('olddir/oldfile', 'file-id')
 
720
        self.build_tree_contents([('new-tree/newdir/',),
 
721
                                  ('new-tree/newdir/newfile', 'new\n')])
 
722
        self.new_tree.add('newdir')
 
723
        self.new_tree.add('newdir/newfile', 'file-id')
 
724
 
 
725
    def test_register_diff(self):
 
726
        self.create_old_new()
 
727
        old_diff_factories = DiffTree.diff_factories
 
728
        DiffTree.diff_factories=old_diff_factories[:]
 
729
        DiffTree.diff_factories.insert(0, DiffWasIs.from_diff_tree)
 
730
        try:
 
731
            differ = DiffTree(self.old_tree, self.new_tree, StringIO())
 
732
        finally:
 
733
            DiffTree.diff_factories = old_diff_factories
 
734
        differ.diff('file-id', 'olddir/oldfile', 'newdir/newfile')
 
735
        self.assertNotContainsRe(
 
736
            differ.to_file.getvalue(),
 
737
            r'--- olddir/oldfile.*\n\+\+\+ newdir/newfile.*\n\@\@ -1,1 \+1,1'
 
738
             ' \@\@\n-old\n\+new\n\n')
 
739
        self.assertContainsRe(differ.to_file.getvalue(),
 
740
                              'was: old\nis: new\n')
 
741
 
 
742
    def test_extra_factories(self):
 
743
        self.create_old_new()
 
744
        differ = DiffTree(self.old_tree, self.new_tree, StringIO(),
 
745
                            extra_factories=[DiffWasIs.from_diff_tree])
 
746
        differ.diff('file-id', 'olddir/oldfile', 'newdir/newfile')
 
747
        self.assertNotContainsRe(
 
748
            differ.to_file.getvalue(),
 
749
            r'--- olddir/oldfile.*\n\+\+\+ newdir/newfile.*\n\@\@ -1,1 \+1,1'
 
750
             ' \@\@\n-old\n\+new\n\n')
 
751
        self.assertContainsRe(differ.to_file.getvalue(),
 
752
                              'was: old\nis: new\n')
 
753
 
 
754
    def test_alphabetical_order(self):
 
755
        self.build_tree(['new-tree/a-file'])
 
756
        self.new_tree.add('a-file')
 
757
        self.build_tree(['old-tree/b-file'])
 
758
        self.old_tree.add('b-file')
 
759
        self.differ.show_diff(None)
 
760
        self.assertContainsRe(self.differ.to_file.getvalue(),
 
761
            '.*a-file(.|\n)*b-file')
 
762
 
 
763
 
 
764
class TestPatienceDiffLib(TestCase):
 
765
 
 
766
    def setUp(self):
 
767
        super(TestPatienceDiffLib, self).setUp()
 
768
        self._unique_lcs = bzrlib._patiencediff_py.unique_lcs_py
 
769
        self._recurse_matches = bzrlib._patiencediff_py.recurse_matches_py
 
770
        self._PatienceSequenceMatcher = \
 
771
            bzrlib._patiencediff_py.PatienceSequenceMatcher_py
 
772
 
 
773
    def test_diff_unicode_string(self):
 
774
        a = ''.join([unichr(i) for i in range(4000, 4500, 3)])
 
775
        b = ''.join([unichr(i) for i in range(4300, 4800, 2)])
 
776
        sm = self._PatienceSequenceMatcher(None, a, b)
 
777
        mb = sm.get_matching_blocks()
 
778
        self.assertEquals(35, len(mb))
 
779
 
 
780
    def test_unique_lcs(self):
 
781
        unique_lcs = self._unique_lcs
 
782
        self.assertEquals(unique_lcs('', ''), [])
 
783
        self.assertEquals(unique_lcs('', 'a'), [])
 
784
        self.assertEquals(unique_lcs('a', ''), [])
 
785
        self.assertEquals(unique_lcs('a', 'a'), [(0,0)])
 
786
        self.assertEquals(unique_lcs('a', 'b'), [])
 
787
        self.assertEquals(unique_lcs('ab', 'ab'), [(0,0), (1,1)])
 
788
        self.assertEquals(unique_lcs('abcde', 'cdeab'), [(2,0), (3,1), (4,2)])
 
789
        self.assertEquals(unique_lcs('cdeab', 'abcde'), [(0,2), (1,3), (2,4)])
 
790
        self.assertEquals(unique_lcs('abXde', 'abYde'), [(0,0), (1,1), 
 
791
                                                         (3,3), (4,4)])
 
792
        self.assertEquals(unique_lcs('acbac', 'abc'), [(2,1)])
 
793
 
 
794
    def test_recurse_matches(self):
 
795
        def test_one(a, b, matches):
 
796
            test_matches = []
 
797
            self._recurse_matches(
 
798
                a, b, 0, 0, len(a), len(b), test_matches, 10)
 
799
            self.assertEquals(test_matches, matches)
 
800
 
 
801
        test_one(['a', '', 'b', '', 'c'], ['a', 'a', 'b', 'c', 'c'],
 
802
                 [(0, 0), (2, 2), (4, 4)])
 
803
        test_one(['a', 'c', 'b', 'a', 'c'], ['a', 'b', 'c'],
 
804
                 [(0, 0), (2, 1), (4, 2)])
 
805
        # Even though 'bc' is not unique globally, and is surrounded by
 
806
        # non-matching lines, we should still match, because they are locally
 
807
        # unique
 
808
        test_one('abcdbce', 'afbcgdbce', [(0,0), (1, 2), (2, 3), (3, 5),
 
809
                                          (4, 6), (5, 7), (6, 8)])
 
810
 
 
811
        # recurse_matches doesn't match non-unique 
 
812
        # lines surrounded by bogus text.
 
813
        # The update has been done in patiencediff.SequenceMatcher instead
 
814
 
 
815
        # This is what it could be
 
816
        #test_one('aBccDe', 'abccde', [(0,0), (2,2), (3,3), (5,5)])
 
817
 
 
818
        # This is what it currently gives:
 
819
        test_one('aBccDe', 'abccde', [(0,0), (5,5)])
 
820
 
 
821
    def assertDiffBlocks(self, a, b, expected_blocks):
 
822
        """Check that the sequence matcher returns the correct blocks.
 
823
 
 
824
        :param a: A sequence to match
 
825
        :param b: Another sequence to match
 
826
        :param expected_blocks: The expected output, not including the final
 
827
            matching block (len(a), len(b), 0)
 
828
        """
 
829
        matcher = self._PatienceSequenceMatcher(None, a, b)
 
830
        blocks = matcher.get_matching_blocks()
 
831
        last = blocks.pop()
 
832
        self.assertEqual((len(a), len(b), 0), last)
 
833
        self.assertEqual(expected_blocks, blocks)
 
834
 
 
835
    def test_matching_blocks(self):
 
836
        # Some basic matching tests
 
837
        self.assertDiffBlocks('', '', [])
 
838
        self.assertDiffBlocks([], [], [])
 
839
        self.assertDiffBlocks('abc', '', [])
 
840
        self.assertDiffBlocks('', 'abc', [])
 
841
        self.assertDiffBlocks('abcd', 'abcd', [(0, 0, 4)])
 
842
        self.assertDiffBlocks('abcd', 'abce', [(0, 0, 3)])
 
843
        self.assertDiffBlocks('eabc', 'abce', [(1, 0, 3)])
 
844
        self.assertDiffBlocks('eabce', 'abce', [(1, 0, 4)])
 
845
        self.assertDiffBlocks('abcde', 'abXde', [(0, 0, 2), (3, 3, 2)])
 
846
        self.assertDiffBlocks('abcde', 'abXYZde', [(0, 0, 2), (3, 5, 2)])
 
847
        self.assertDiffBlocks('abde', 'abXYZde', [(0, 0, 2), (2, 5, 2)])
 
848
        # This may check too much, but it checks to see that
 
849
        # a copied block stays attached to the previous section,
 
850
        # not the later one.
 
851
        # difflib would tend to grab the trailing longest match
 
852
        # which would make the diff not look right
 
853
        self.assertDiffBlocks('abcdefghijklmnop', 'abcdefxydefghijklmnop',
 
854
                              [(0, 0, 6), (6, 11, 10)])
 
855
 
 
856
        # make sure it supports passing in lists
 
857
        self.assertDiffBlocks(
 
858
                   ['hello there\n',
 
859
                    'world\n',
 
860
                    'how are you today?\n'],
 
861
                   ['hello there\n',
 
862
                    'how are you today?\n'],
 
863
                [(0, 0, 1), (2, 1, 1)])
 
864
 
 
865
        # non unique lines surrounded by non-matching lines
 
866
        # won't be found
 
867
        self.assertDiffBlocks('aBccDe', 'abccde', [(0,0,1), (5,5,1)])
 
868
 
 
869
        # But they only need to be locally unique
 
870
        self.assertDiffBlocks('aBcDec', 'abcdec', [(0,0,1), (2,2,1), (4,4,2)])
 
871
 
 
872
        # non unique blocks won't be matched
 
873
        self.assertDiffBlocks('aBcdEcdFg', 'abcdecdfg', [(0,0,1), (8,8,1)])
 
874
 
 
875
        # but locally unique ones will
 
876
        self.assertDiffBlocks('aBcdEeXcdFg', 'abcdecdfg', [(0,0,1), (2,2,2),
 
877
                                              (5,4,1), (7,5,2), (10,8,1)])
 
878
 
 
879
        self.assertDiffBlocks('abbabbXd', 'cabbabxd', [(7,7,1)])
 
880
        self.assertDiffBlocks('abbabbbb', 'cabbabbc', [])
 
881
        self.assertDiffBlocks('bbbbbbbb', 'cbbbbbbc', [])
 
882
 
 
883
    def test_matching_blocks_tuples(self):
 
884
        # Some basic matching tests
 
885
        self.assertDiffBlocks([], [], [])
 
886
        self.assertDiffBlocks([('a',), ('b',), ('c,')], [], [])
 
887
        self.assertDiffBlocks([], [('a',), ('b',), ('c,')], [])
 
888
        self.assertDiffBlocks([('a',), ('b',), ('c,')],
 
889
                              [('a',), ('b',), ('c,')],
 
890
                              [(0, 0, 3)])
 
891
        self.assertDiffBlocks([('a',), ('b',), ('c,')],
 
892
                              [('a',), ('b',), ('d,')],
 
893
                              [(0, 0, 2)])
 
894
        self.assertDiffBlocks([('d',), ('b',), ('c,')],
 
895
                              [('a',), ('b',), ('c,')],
 
896
                              [(1, 1, 2)])
 
897
        self.assertDiffBlocks([('d',), ('a',), ('b',), ('c,')],
 
898
                              [('a',), ('b',), ('c,')],
 
899
                              [(1, 0, 3)])
 
900
        self.assertDiffBlocks([('a', 'b'), ('c', 'd'), ('e', 'f')],
 
901
                              [('a', 'b'), ('c', 'X'), ('e', 'f')],
 
902
                              [(0, 0, 1), (2, 2, 1)])
 
903
        self.assertDiffBlocks([('a', 'b'), ('c', 'd'), ('e', 'f')],
 
904
                              [('a', 'b'), ('c', 'dX'), ('e', 'f')],
 
905
                              [(0, 0, 1), (2, 2, 1)])
 
906
 
 
907
    def test_opcodes(self):
 
908
        def chk_ops(a, b, expected_codes):
 
909
            s = self._PatienceSequenceMatcher(None, a, b)
 
910
            self.assertEquals(expected_codes, s.get_opcodes())
 
911
 
 
912
        chk_ops('', '', [])
 
913
        chk_ops([], [], [])
 
914
        chk_ops('abc', '', [('delete', 0,3, 0,0)])
 
915
        chk_ops('', 'abc', [('insert', 0,0, 0,3)])
 
916
        chk_ops('abcd', 'abcd', [('equal',    0,4, 0,4)])
 
917
        chk_ops('abcd', 'abce', [('equal',   0,3, 0,3),
 
918
                                 ('replace', 3,4, 3,4)
 
919
                                ])
 
920
        chk_ops('eabc', 'abce', [('delete', 0,1, 0,0),
 
921
                                 ('equal',  1,4, 0,3),
 
922
                                 ('insert', 4,4, 3,4)
 
923
                                ])
 
924
        chk_ops('eabce', 'abce', [('delete', 0,1, 0,0),
 
925
                                  ('equal',  1,5, 0,4)
 
926
                                 ])
 
927
        chk_ops('abcde', 'abXde', [('equal',   0,2, 0,2),
 
928
                                   ('replace', 2,3, 2,3),
 
929
                                   ('equal',   3,5, 3,5)
 
930
                                  ])
 
931
        chk_ops('abcde', 'abXYZde', [('equal',   0,2, 0,2),
 
932
                                     ('replace', 2,3, 2,5),
 
933
                                     ('equal',   3,5, 5,7)
 
934
                                    ])
 
935
        chk_ops('abde', 'abXYZde', [('equal',  0,2, 0,2),
 
936
                                    ('insert', 2,2, 2,5),
 
937
                                    ('equal',  2,4, 5,7)
 
938
                                   ])
 
939
        chk_ops('abcdefghijklmnop', 'abcdefxydefghijklmnop',
 
940
                [('equal',  0,6,  0,6),
 
941
                 ('insert', 6,6,  6,11),
 
942
                 ('equal',  6,16, 11,21)
 
943
                ])
 
944
        chk_ops(
 
945
                [ 'hello there\n'
 
946
                , 'world\n'
 
947
                , 'how are you today?\n'],
 
948
                [ 'hello there\n'
 
949
                , 'how are you today?\n'],
 
950
                [('equal',  0,1, 0,1),
 
951
                 ('delete', 1,2, 1,1),
 
952
                 ('equal',  2,3, 1,2),
 
953
                ])
 
954
        chk_ops('aBccDe', 'abccde', 
 
955
                [('equal',   0,1, 0,1),
 
956
                 ('replace', 1,5, 1,5),
 
957
                 ('equal',   5,6, 5,6),
 
958
                ])
 
959
        chk_ops('aBcDec', 'abcdec', 
 
960
                [('equal',   0,1, 0,1),
 
961
                 ('replace', 1,2, 1,2),
 
962
                 ('equal',   2,3, 2,3),
 
963
                 ('replace', 3,4, 3,4),
 
964
                 ('equal',   4,6, 4,6),
 
965
                ])
 
966
        chk_ops('aBcdEcdFg', 'abcdecdfg', 
 
967
                [('equal',   0,1, 0,1),
 
968
                 ('replace', 1,8, 1,8),
 
969
                 ('equal',   8,9, 8,9)
 
970
                ])
 
971
        chk_ops('aBcdEeXcdFg', 'abcdecdfg', 
 
972
                [('equal',   0,1, 0,1),
 
973
                 ('replace', 1,2, 1,2),
 
974
                 ('equal',   2,4, 2,4),
 
975
                 ('delete', 4,5, 4,4),
 
976
                 ('equal',   5,6, 4,5),
 
977
                 ('delete', 6,7, 5,5),
 
978
                 ('equal',   7,9, 5,7),
 
979
                 ('replace', 9,10, 7,8),
 
980
                 ('equal',   10,11, 8,9)
 
981
                ])
 
982
 
 
983
    def test_grouped_opcodes(self):
 
984
        def chk_ops(a, b, expected_codes, n=3):
 
985
            s = self._PatienceSequenceMatcher(None, a, b)
 
986
            self.assertEquals(expected_codes, list(s.get_grouped_opcodes(n)))
 
987
 
 
988
        chk_ops('', '', [])
 
989
        chk_ops([], [], [])
 
990
        chk_ops('abc', '', [[('delete', 0,3, 0,0)]])
 
991
        chk_ops('', 'abc', [[('insert', 0,0, 0,3)]])
 
992
        chk_ops('abcd', 'abcd', [])
 
993
        chk_ops('abcd', 'abce', [[('equal',   0,3, 0,3),
 
994
                                  ('replace', 3,4, 3,4)
 
995
                                 ]])
 
996
        chk_ops('eabc', 'abce', [[('delete', 0,1, 0,0),
 
997
                                 ('equal',  1,4, 0,3),
 
998
                                 ('insert', 4,4, 3,4)
 
999
                                ]])
 
1000
        chk_ops('abcdefghijklmnop', 'abcdefxydefghijklmnop',
 
1001
                [[('equal',  3,6, 3,6),
 
1002
                  ('insert', 6,6, 6,11),
 
1003
                  ('equal',  6,9, 11,14)
 
1004
                  ]])
 
1005
        chk_ops('abcdefghijklmnop', 'abcdefxydefghijklmnop',
 
1006
                [[('equal',  2,6, 2,6),
 
1007
                  ('insert', 6,6, 6,11),
 
1008
                  ('equal',  6,10, 11,15)
 
1009
                  ]], 4)
 
1010
        chk_ops('Xabcdef', 'abcdef',
 
1011
                [[('delete', 0,1, 0,0),
 
1012
                  ('equal',  1,4, 0,3)
 
1013
                  ]])
 
1014
        chk_ops('abcdef', 'abcdefX',
 
1015
                [[('equal',  3,6, 3,6),
 
1016
                  ('insert', 6,6, 6,7)
 
1017
                  ]])
 
1018
 
 
1019
 
 
1020
    def test_multiple_ranges(self):
 
1021
        # There was an earlier bug where we used a bad set of ranges,
 
1022
        # this triggers that specific bug, to make sure it doesn't regress
 
1023
        self.assertDiffBlocks('abcdefghijklmnop',
 
1024
                              'abcXghiYZQRSTUVWXYZijklmnop',
 
1025
                              [(0, 0, 3), (6, 4, 3), (9, 20, 7)])
 
1026
 
 
1027
        self.assertDiffBlocks('ABCd efghIjk  L',
 
1028
                              'AxyzBCn mo pqrstuvwI1 2  L',
 
1029
                              [(0,0,1), (1, 4, 2), (9, 19, 1), (12, 23, 3)])
 
1030
 
 
1031
        # These are rot13 code snippets.
 
1032
        self.assertDiffBlocks('''\
 
1033
    trg nqqrq jura lbh nqq n svyr va gur qverpgbel.
 
1034
    """
 
1035
    gnxrf_netf = ['svyr*']
 
1036
    gnxrf_bcgvbaf = ['ab-erphefr']
 
1037
  
 
1038
    qrs eha(frys, svyr_yvfg, ab_erphefr=Snyfr):
 
1039
        sebz omeyvo.nqq vzcbeg fzneg_nqq, nqq_ercbegre_cevag, nqq_ercbegre_ahyy
 
1040
        vs vf_dhvrg():
 
1041
            ercbegre = nqq_ercbegre_ahyy
 
1042
        ryfr:
 
1043
            ercbegre = nqq_ercbegre_cevag
 
1044
        fzneg_nqq(svyr_yvfg, abg ab_erphefr, ercbegre)
 
1045
 
 
1046
 
 
1047
pynff pzq_zxqve(Pbzznaq):
 
1048
'''.splitlines(True), '''\
 
1049
    trg nqqrq jura lbh nqq n svyr va gur qverpgbel.
 
1050
 
 
1051
    --qel-eha jvyy fubj juvpu svyrf jbhyq or nqqrq, ohg abg npghnyyl 
 
1052
    nqq gurz.
 
1053
    """
 
1054
    gnxrf_netf = ['svyr*']
 
1055
    gnxrf_bcgvbaf = ['ab-erphefr', 'qel-eha']
 
1056
 
 
1057
    qrs eha(frys, svyr_yvfg, ab_erphefr=Snyfr, qel_eha=Snyfr):
 
1058
        vzcbeg omeyvo.nqq
 
1059
 
 
1060
        vs qel_eha:
 
1061
            vs vf_dhvrg():
 
1062
                # Guvf vf cbvagyrff, ohg V'q engure abg envfr na reebe
 
1063
                npgvba = omeyvo.nqq.nqq_npgvba_ahyy
 
1064
            ryfr:
 
1065
  npgvba = omeyvo.nqq.nqq_npgvba_cevag
 
1066
        ryvs vf_dhvrg():
 
1067
            npgvba = omeyvo.nqq.nqq_npgvba_nqq
 
1068
        ryfr:
 
1069
       npgvba = omeyvo.nqq.nqq_npgvba_nqq_naq_cevag
 
1070
 
 
1071
        omeyvo.nqq.fzneg_nqq(svyr_yvfg, abg ab_erphefr, npgvba)
 
1072
 
 
1073
 
 
1074
pynff pzq_zxqve(Pbzznaq):
 
1075
'''.splitlines(True)
 
1076
, [(0,0,1), (1, 4, 2), (9, 19, 1), (12, 23, 3)])
 
1077
 
 
1078
    def test_patience_unified_diff(self):
 
1079
        txt_a = ['hello there\n',
 
1080
                 'world\n',
 
1081
                 'how are you today?\n']
 
1082
        txt_b = ['hello there\n',
 
1083
                 'how are you today?\n']
 
1084
        unified_diff = bzrlib.patiencediff.unified_diff
 
1085
        psm = self._PatienceSequenceMatcher
 
1086
        self.assertEquals([ '---  \n',
 
1087
                           '+++  \n',
 
1088
                           '@@ -1,3 +1,2 @@\n',
 
1089
                           ' hello there\n',
 
1090
                           '-world\n',
 
1091
                           ' how are you today?\n'
 
1092
                          ]
 
1093
                          , list(unified_diff(txt_a, txt_b,
 
1094
                                 sequencematcher=psm)))
 
1095
        txt_a = map(lambda x: x+'\n', 'abcdefghijklmnop')
 
1096
        txt_b = map(lambda x: x+'\n', 'abcdefxydefghijklmnop')
 
1097
        # This is the result with LongestCommonSubstring matching
 
1098
        self.assertEquals(['---  \n',
 
1099
                           '+++  \n',
 
1100
                           '@@ -1,6 +1,11 @@\n',
 
1101
                           ' a\n',
 
1102
                           ' b\n',
 
1103
                           ' c\n',
 
1104
                           '+d\n',
 
1105
                           '+e\n',
 
1106
                           '+f\n',
 
1107
                           '+x\n',
 
1108
                           '+y\n',
 
1109
                           ' d\n',
 
1110
                           ' e\n',
 
1111
                           ' f\n']
 
1112
                          , list(unified_diff(txt_a, txt_b)))
 
1113
        # And the patience diff
 
1114
        self.assertEquals(['---  \n',
 
1115
                           '+++  \n',
 
1116
                           '@@ -4,6 +4,11 @@\n',
 
1117
                           ' d\n',
 
1118
                           ' e\n',
 
1119
                           ' f\n',
 
1120
                           '+x\n',
 
1121
                           '+y\n',
 
1122
                           '+d\n',
 
1123
                           '+e\n',
 
1124
                           '+f\n',
 
1125
                           ' g\n',
 
1126
                           ' h\n',
 
1127
                           ' i\n',
 
1128
                          ]
 
1129
                          , list(unified_diff(txt_a, txt_b,
 
1130
                                 sequencematcher=psm)))
 
1131
 
 
1132
 
 
1133
class TestPatienceDiffLib_c(TestPatienceDiffLib):
 
1134
 
 
1135
    _test_needs_features = [CompiledPatienceDiffFeature]
 
1136
 
 
1137
    def setUp(self):
 
1138
        super(TestPatienceDiffLib_c, self).setUp()
 
1139
        import bzrlib._patiencediff_c
 
1140
        self._unique_lcs = bzrlib._patiencediff_c.unique_lcs_c
 
1141
        self._recurse_matches = bzrlib._patiencediff_c.recurse_matches_c
 
1142
        self._PatienceSequenceMatcher = \
 
1143
            bzrlib._patiencediff_c.PatienceSequenceMatcher_c
 
1144
 
 
1145
    def test_unhashable(self):
 
1146
        """We should get a proper exception here."""
 
1147
        # We need to be able to hash items in the sequence, lists are
 
1148
        # unhashable, and thus cannot be diffed
 
1149
        e = self.assertRaises(TypeError, self._PatienceSequenceMatcher,
 
1150
                                         None, [[]], [])
 
1151
        e = self.assertRaises(TypeError, self._PatienceSequenceMatcher,
 
1152
                                         None, ['valid', []], [])
 
1153
        e = self.assertRaises(TypeError, self._PatienceSequenceMatcher,
 
1154
                                         None, ['valid'], [[]])
 
1155
        e = self.assertRaises(TypeError, self._PatienceSequenceMatcher,
 
1156
                                         None, ['valid'], ['valid', []])
 
1157
 
 
1158
 
 
1159
class TestPatienceDiffLibFiles(TestCaseInTempDir):
 
1160
 
 
1161
    def setUp(self):
 
1162
        super(TestPatienceDiffLibFiles, self).setUp()
 
1163
        self._PatienceSequenceMatcher = \
 
1164
            bzrlib._patiencediff_py.PatienceSequenceMatcher_py
 
1165
 
 
1166
    def test_patience_unified_diff_files(self):
 
1167
        txt_a = ['hello there\n',
 
1168
                 'world\n',
 
1169
                 'how are you today?\n']
 
1170
        txt_b = ['hello there\n',
 
1171
                 'how are you today?\n']
 
1172
        open('a1', 'wb').writelines(txt_a)
 
1173
        open('b1', 'wb').writelines(txt_b)
 
1174
 
 
1175
        unified_diff_files = bzrlib.patiencediff.unified_diff_files
 
1176
        psm = self._PatienceSequenceMatcher
 
1177
        self.assertEquals(['--- a1 \n',
 
1178
                           '+++ b1 \n',
 
1179
                           '@@ -1,3 +1,2 @@\n',
 
1180
                           ' hello there\n',
 
1181
                           '-world\n',
 
1182
                           ' how are you today?\n',
 
1183
                          ]
 
1184
                          , list(unified_diff_files('a1', 'b1',
 
1185
                                 sequencematcher=psm)))
 
1186
 
 
1187
        txt_a = map(lambda x: x+'\n', 'abcdefghijklmnop')
 
1188
        txt_b = map(lambda x: x+'\n', 'abcdefxydefghijklmnop')
 
1189
        open('a2', 'wb').writelines(txt_a)
 
1190
        open('b2', 'wb').writelines(txt_b)
 
1191
 
 
1192
        # This is the result with LongestCommonSubstring matching
 
1193
        self.assertEquals(['--- a2 \n',
 
1194
                           '+++ b2 \n',
 
1195
                           '@@ -1,6 +1,11 @@\n',
 
1196
                           ' a\n',
 
1197
                           ' b\n',
 
1198
                           ' c\n',
 
1199
                           '+d\n',
 
1200
                           '+e\n',
 
1201
                           '+f\n',
 
1202
                           '+x\n',
 
1203
                           '+y\n',
 
1204
                           ' d\n',
 
1205
                           ' e\n',
 
1206
                           ' f\n']
 
1207
                          , list(unified_diff_files('a2', 'b2')))
 
1208
 
 
1209
        # And the patience diff
 
1210
        self.assertEquals(['--- a2 \n',
 
1211
                           '+++ b2 \n',
 
1212
                           '@@ -4,6 +4,11 @@\n',
 
1213
                           ' d\n',
 
1214
                           ' e\n',
 
1215
                           ' f\n',
 
1216
                           '+x\n',
 
1217
                           '+y\n',
 
1218
                           '+d\n',
 
1219
                           '+e\n',
 
1220
                           '+f\n',
 
1221
                           ' g\n',
 
1222
                           ' h\n',
 
1223
                           ' i\n',
 
1224
                          ]
 
1225
                          , list(unified_diff_files('a2', 'b2',
 
1226
                                 sequencematcher=psm)))
 
1227
 
 
1228
 
 
1229
class TestPatienceDiffLibFiles_c(TestPatienceDiffLibFiles):
 
1230
 
 
1231
    _test_needs_features = [CompiledPatienceDiffFeature]
 
1232
 
 
1233
    def setUp(self):
 
1234
        super(TestPatienceDiffLibFiles_c, self).setUp()
 
1235
        import bzrlib._patiencediff_c
 
1236
        self._PatienceSequenceMatcher = \
 
1237
            bzrlib._patiencediff_c.PatienceSequenceMatcher_c
 
1238
 
 
1239
 
 
1240
class TestUsingCompiledIfAvailable(TestCase):
 
1241
 
 
1242
    def test_PatienceSequenceMatcher(self):
 
1243
        if CompiledPatienceDiffFeature.available():
 
1244
            from bzrlib._patiencediff_c import PatienceSequenceMatcher_c
 
1245
            self.assertIs(PatienceSequenceMatcher_c,
 
1246
                          bzrlib.patiencediff.PatienceSequenceMatcher)
 
1247
        else:
 
1248
            from bzrlib._patiencediff_py import PatienceSequenceMatcher_py
 
1249
            self.assertIs(PatienceSequenceMatcher_py,
 
1250
                          bzrlib.patiencediff.PatienceSequenceMatcher)
 
1251
 
 
1252
    def test_unique_lcs(self):
 
1253
        if CompiledPatienceDiffFeature.available():
 
1254
            from bzrlib._patiencediff_c import unique_lcs_c
 
1255
            self.assertIs(unique_lcs_c,
 
1256
                          bzrlib.patiencediff.unique_lcs)
 
1257
        else:
 
1258
            from bzrlib._patiencediff_py import unique_lcs_py
 
1259
            self.assertIs(unique_lcs_py,
 
1260
                          bzrlib.patiencediff.unique_lcs)
 
1261
 
 
1262
    def test_recurse_matches(self):
 
1263
        if CompiledPatienceDiffFeature.available():
 
1264
            from bzrlib._patiencediff_c import recurse_matches_c
 
1265
            self.assertIs(recurse_matches_c,
 
1266
                          bzrlib.patiencediff.recurse_matches)
 
1267
        else:
 
1268
            from bzrlib._patiencediff_py import recurse_matches_py
 
1269
            self.assertIs(recurse_matches_py,
 
1270
                          bzrlib.patiencediff.recurse_matches)
 
1271
 
 
1272
 
 
1273
class TestDiffFromTool(TestCaseWithTransport):
 
1274
 
 
1275
    def test_from_string(self):
 
1276
        diff_obj = DiffFromTool.from_string('diff', None, None, None)
 
1277
        self.addCleanup(diff_obj.finish)
 
1278
        self.assertEqual(['diff', '%(old_path)s', '%(new_path)s'],
 
1279
            diff_obj.command_template)
 
1280
 
 
1281
    def test_from_string_u5(self):
 
1282
        diff_obj = DiffFromTool.from_string('diff -u\\ 5', None, None, None)
 
1283
        self.addCleanup(diff_obj.finish)
 
1284
        self.assertEqual(['diff', '-u 5', '%(old_path)s', '%(new_path)s'],
 
1285
                         diff_obj.command_template)
 
1286
        self.assertEqual(['diff', '-u 5', 'old-path', 'new-path'],
 
1287
                         diff_obj._get_command('old-path', 'new-path'))
 
1288
 
 
1289
    def test_execute(self):
 
1290
        output = StringIO()
 
1291
        diff_obj = DiffFromTool(['python', '-c',
 
1292
                                 'print "%(old_path)s %(new_path)s"'],
 
1293
                                None, None, output)
 
1294
        self.addCleanup(diff_obj.finish)
 
1295
        diff_obj._execute('old', 'new')
 
1296
        self.assertEqual(output.getvalue().rstrip(), 'old new')
 
1297
 
 
1298
    def test_excute_missing(self):
 
1299
        diff_obj = DiffFromTool(['a-tool-which-is-unlikely-to-exist'],
 
1300
                                None, None, None)
 
1301
        self.addCleanup(diff_obj.finish)
 
1302
        e = self.assertRaises(ExecutableMissing, diff_obj._execute, 'old',
 
1303
                              'new')
 
1304
        self.assertEqual('a-tool-which-is-unlikely-to-exist could not be found'
 
1305
                         ' on this machine', str(e))
 
1306
 
 
1307
    def test_prepare_files_creates_paths_readable_by_windows_tool(self):
 
1308
        self.requireFeature(AttribFeature)
 
1309
        output = StringIO()
 
1310
        tree = self.make_branch_and_tree('tree')
 
1311
        self.build_tree_contents([('tree/file', 'content')])
 
1312
        tree.add('file', 'file-id')
 
1313
        tree.commit('old tree')
 
1314
        tree.lock_read()
 
1315
        self.addCleanup(tree.unlock)
 
1316
        diff_obj = DiffFromTool(['python', '-c',
 
1317
                                 'print "%(old_path)s %(new_path)s"'],
 
1318
                                tree, tree, output)
 
1319
        diff_obj._prepare_files('file-id', 'file', 'file')
 
1320
        self.assertReadableByAttrib(diff_obj._root, 'old\\file', r'old\\file')
 
1321
        self.assertReadableByAttrib(diff_obj._root, 'new\\file', r'new\\file')
 
1322
 
 
1323
    def assertReadableByAttrib(self, cwd, relpath, regex):
 
1324
        proc = subprocess.Popen(['attrib', relpath],
 
1325
                                stdout=subprocess.PIPE,
 
1326
                                cwd=cwd)
 
1327
        proc.wait()
 
1328
        result = proc.stdout.read()
 
1329
        self.assertContainsRe(result, regex)
 
1330
 
 
1331
    def test_prepare_files(self):
 
1332
        output = StringIO()
 
1333
        tree = self.make_branch_and_tree('tree')
 
1334
        self.build_tree_contents([('tree/oldname', 'oldcontent')])
 
1335
        self.build_tree_contents([('tree/oldname2', 'oldcontent2')])
 
1336
        tree.add('oldname', 'file-id')
 
1337
        tree.add('oldname2', 'file2-id')
 
1338
        tree.commit('old tree', timestamp=0)
 
1339
        tree.rename_one('oldname', 'newname')
 
1340
        tree.rename_one('oldname2', 'newname2')
 
1341
        self.build_tree_contents([('tree/newname', 'newcontent')])
 
1342
        self.build_tree_contents([('tree/newname2', 'newcontent2')])
 
1343
        old_tree = tree.basis_tree()
 
1344
        old_tree.lock_read()
 
1345
        self.addCleanup(old_tree.unlock)
 
1346
        tree.lock_read()
 
1347
        self.addCleanup(tree.unlock)
 
1348
        diff_obj = DiffFromTool(['python', '-c',
 
1349
                                 'print "%(old_path)s %(new_path)s"'],
 
1350
                                old_tree, tree, output)
 
1351
        self.addCleanup(diff_obj.finish)
 
1352
        self.assertContainsRe(diff_obj._root, 'bzr-diff-[^/]*')
 
1353
        old_path, new_path = diff_obj._prepare_files('file-id', 'oldname',
 
1354
                                                     'newname')
 
1355
        self.assertContainsRe(old_path, 'old/oldname$')
 
1356
        self.assertEqual(0, os.stat(old_path).st_mtime)
 
1357
        self.assertContainsRe(new_path, 'new/newname$')
 
1358
        self.assertFileEqual('oldcontent', old_path)
 
1359
        self.assertFileEqual('newcontent', new_path)
 
1360
        if osutils.host_os_dereferences_symlinks():
 
1361
            self.assertTrue(os.path.samefile('tree/newname', new_path))
 
1362
        # make sure we can create files with the same parent directories
 
1363
        diff_obj._prepare_files('file2-id', 'oldname2', 'newname2')