/brz/remove-bazaar

To get this branch, use:
bzr branch http://gegoxaren.bato24.eu/bzr/brz/remove-bazaar

« back to all changes in this revision

Viewing changes to bzrlib/diff.py

  • Committer: Carl Friedrich Bolz
  • Date: 2006-08-11 16:02:36 UTC
  • mto: (1908.3.5 bench_usecases)
  • mto: This revision was merged to the branch mainline in revision 2068.
  • Revision ID: cfbolz@gmx.de-20060811160236-0ce4b4864fbdb46a
(cfbolz, hpk): Add caching mechanism and add benchmark for bundle-reading.

Show diffs side-by-side

added added

removed removed

Lines of Context:
bzrlib/diff.py
1
1
# Copyright (C) 2004, 2005, 2006 Canonical Ltd.
2
 
 
 
2
#
3
3
# This program is free software; you can redistribute it and/or modify
4
4
# it under the terms of the GNU General Public License as published by
5
5
# the Free Software Foundation; either version 2 of the License, or
6
6
# (at your option) any later version.
7
 
 
 
7
#
8
8
# This program is distributed in the hope that it will be useful,
9
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
11
11
# GNU General Public License for more details.
12
 
 
 
12
#
13
13
# You should have received a copy of the GNU General Public License
14
14
# along with this program; if not, write to the Free Software
15
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
16
16
 
 
17
import errno
 
18
import os
 
19
import re
 
20
import subprocess
 
21
import sys
 
22
import tempfile
 
23
import time
 
24
 
 
25
# compatability - plugins import compare_trees from diff!!!
 
26
# deprecated as of 0.10
17
27
from bzrlib.delta import compare_trees
18
28
from bzrlib.errors import BzrError
19
29
import bzrlib.errors as errors
20
 
from bzrlib.symbol_versioning import *
21
 
from bzrlib.trace import mutter
 
30
import bzrlib.osutils
 
31
from bzrlib.patiencediff import unified_diff
 
32
import bzrlib.patiencediff
 
33
from bzrlib.symbol_versioning import (deprecated_function,
 
34
        zero_eight)
 
35
from bzrlib.textfile import check_text_lines
 
36
from bzrlib.trace import mutter, warning
 
37
 
22
38
 
23
39
# TODO: Rather than building a changeset object, we should probably
24
40
# invoke callbacks on an object.  That object can either accumulate a
25
41
# list, write them out directly, etc etc.
26
42
 
27
 
def internal_diff(old_filename, oldlines, new_filename, newlines, to_file):
28
 
    import difflib
29
 
    
 
43
def internal_diff(old_filename, oldlines, new_filename, newlines, to_file,
 
44
                  allow_binary=False, sequence_matcher=None,
 
45
                  path_encoding='utf8'):
30
46
    # FIXME: difflib is wrong if there is no trailing newline.
31
47
    # The syntax used by patch seems to be "\ No newline at
32
48
    # end of file" following the last diff line from that
42
58
    # both sequences are empty.
43
59
    if not oldlines and not newlines:
44
60
        return
 
61
    
 
62
    if allow_binary is False:
 
63
        check_text_lines(oldlines)
 
64
        check_text_lines(newlines)
45
65
 
46
 
    ud = difflib.unified_diff(oldlines, newlines,
47
 
                              fromfile=old_filename+'\t', 
48
 
                              tofile=new_filename+'\t')
 
66
    if sequence_matcher is None:
 
67
        sequence_matcher = bzrlib.patiencediff.PatienceSequenceMatcher
 
68
    ud = unified_diff(oldlines, newlines,
 
69
                      fromfile=old_filename.encode(path_encoding),
 
70
                      tofile=new_filename.encode(path_encoding),
 
71
                      sequencematcher=sequence_matcher)
49
72
 
50
73
    ud = list(ud)
51
74
    # work-around for difflib being too smart for its own good
68
91
def external_diff(old_filename, oldlines, new_filename, newlines, to_file,
69
92
                  diff_opts):
70
93
    """Display a diff by calling out to the external diff program."""
71
 
    import sys
72
 
    
73
 
    if to_file != sys.stdout:
74
 
        raise NotImplementedError("sorry, can't send external diff other than to stdout yet",
75
 
                                  to_file)
76
 
 
77
94
    # make sure our own output is properly ordered before the diff
78
95
    to_file.flush()
79
96
 
80
 
    from tempfile import NamedTemporaryFile
81
 
    import os
82
 
 
83
 
    oldtmpf = NamedTemporaryFile()
84
 
    newtmpf = NamedTemporaryFile()
 
97
    oldtmp_fd, old_abspath = tempfile.mkstemp(prefix='bzr-diff-old-')
 
98
    newtmp_fd, new_abspath = tempfile.mkstemp(prefix='bzr-diff-new-')
 
99
    oldtmpf = os.fdopen(oldtmp_fd, 'wb')
 
100
    newtmpf = os.fdopen(newtmp_fd, 'wb')
85
101
 
86
102
    try:
87
103
        # TODO: perhaps a special case for comparing to or from the empty
94
110
        oldtmpf.writelines(oldlines)
95
111
        newtmpf.writelines(newlines)
96
112
 
97
 
        oldtmpf.flush()
98
 
        newtmpf.flush()
 
113
        oldtmpf.close()
 
114
        newtmpf.close()
99
115
 
100
116
        if not diff_opts:
101
117
            diff_opts = []
102
118
        diffcmd = ['diff',
103
 
                   '--label', old_filename+'\t',
104
 
                   oldtmpf.name,
105
 
                   '--label', new_filename+'\t',
106
 
                   newtmpf.name]
 
119
                   '--label', old_filename,
 
120
                   old_abspath,
 
121
                   '--label', new_filename,
 
122
                   new_abspath,
 
123
                   '--binary',
 
124
                  ]
107
125
 
108
126
        # diff only allows one style to be specified; they don't override.
109
127
        # note that some of these take optargs, and the optargs can be
129
147
        if diff_opts:
130
148
            diffcmd.extend(diff_opts)
131
149
 
132
 
        rc = os.spawnvp(os.P_WAIT, 'diff', diffcmd)
 
150
        try:
 
151
            pipe = subprocess.Popen(diffcmd,
 
152
                                    stdin=subprocess.PIPE,
 
153
                                    stdout=subprocess.PIPE)
 
154
        except OSError, e:
 
155
            if e.errno == errno.ENOENT:
 
156
                raise errors.NoDiff(str(e))
 
157
            raise
 
158
        pipe.stdin.close()
 
159
 
 
160
        first_line = pipe.stdout.readline()
 
161
        to_file.write(first_line)
 
162
        bzrlib.osutils.pumpfile(pipe.stdout, to_file)
 
163
        rc = pipe.wait()
133
164
        
134
 
        if rc != 0 and rc != 1:
 
165
        if rc == 2:
 
166
            # 'diff' gives retcode == 2 for all sorts of errors
 
167
            # one of those is 'Binary files differ'.
 
168
            # Bad options could also be the problem.
 
169
            # 'Binary files' is not a real error, so we suppress that error
 
170
            m = re.match('^binary files.*differ$', first_line, re.I)
 
171
            if not m:
 
172
                raise BzrError('external diff failed with exit code 2;'
 
173
                               ' command: %r' % (diffcmd,))
 
174
        elif rc not in (0, 1):
135
175
            # returns 1 if files differ; that's OK
136
176
            if rc < 0:
137
177
                msg = 'signal %d' % (-rc)
138
178
            else:
139
179
                msg = 'exit code %d' % rc
140
180
                
141
 
            raise BzrError('external diff failed with %s; command: %r' % (rc, diffcmd))
 
181
            raise BzrError('external diff failed with %s; command: %r' 
 
182
                           % (rc, diffcmd))
 
183
 
 
184
        # internal_diff() adds a trailing newline, add one here for consistency
 
185
        to_file.write('\n')
 
186
 
142
187
    finally:
143
188
        oldtmpf.close()                 # and delete
144
189
        newtmpf.close()
 
190
        # Clean up. Warn in case the files couldn't be deleted
 
191
        # (in case windows still holds the file open, but not
 
192
        # if the files have already been deleted)
 
193
        try:
 
194
            os.remove(old_abspath)
 
195
        except OSError, e:
 
196
            if e.errno not in (errno.ENOENT,):
 
197
                warning('Failed to delete temporary file: %s %s',
 
198
                        old_abspath, e)
 
199
        try:
 
200
            os.remove(new_abspath)
 
201
        except OSError:
 
202
            if e.errno not in (errno.ENOENT,):
 
203
                warning('Failed to delete temporary file: %s %s',
 
204
                        new_abspath, e)
145
205
 
146
206
 
147
207
@deprecated_function(zero_eight)
161
221
    supplies any two trees.
162
222
    """
163
223
    if output is None:
164
 
        import sys
165
224
        output = sys.stdout
166
225
 
167
226
    if from_spec is None:
184
243
 
185
244
 
186
245
def diff_cmd_helper(tree, specific_files, external_diff_options, 
187
 
                    old_revision_spec=None, new_revision_spec=None):
 
246
                    old_revision_spec=None, new_revision_spec=None,
 
247
                    old_label='a/', new_label='b/'):
188
248
    """Helper for cmd_diff.
189
249
 
190
250
   tree 
207
267
    The more general form is show_diff_trees(), where the caller
208
268
    supplies any two trees.
209
269
    """
210
 
    import sys
211
 
    output = sys.stdout
212
270
    def spec_tree(spec):
213
 
        revision_id = spec.in_store(tree.branch).rev_id
214
 
        return tree.branch.repository.revision_tree(revision_id)
 
271
        if tree:
 
272
            revision = spec.in_store(tree.branch)
 
273
        else:
 
274
            revision = spec.in_store(None)
 
275
        revision_id = revision.rev_id
 
276
        branch = revision.branch
 
277
        return branch.repository.revision_tree(revision_id)
215
278
    if old_revision_spec is None:
216
279
        old_tree = tree.basis_tree()
217
280
    else:
221
284
        new_tree = tree
222
285
    else:
223
286
        new_tree = spec_tree(new_revision_spec)
 
287
    if new_tree is not tree:
 
288
        extra_trees = (tree,)
 
289
    else:
 
290
        extra_trees = None
224
291
 
225
292
    return show_diff_trees(old_tree, new_tree, sys.stdout, specific_files,
226
 
                           external_diff_options)
 
293
                           external_diff_options,
 
294
                           old_label=old_label, new_label=new_label,
 
295
                           extra_trees=extra_trees)
227
296
 
228
297
 
229
298
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None,
230
 
                    external_diff_options=None):
 
299
                    external_diff_options=None,
 
300
                    old_label='a/', new_label='b/',
 
301
                    extra_trees=None):
231
302
    """Show in text form the changes from one tree to another.
232
303
 
233
304
    to_files
235
306
 
236
307
    external_diff_options
237
308
        If set, use an external GNU diff and pass these options.
 
309
 
 
310
    extra_trees
 
311
        If set, more Trees to use for looking up file ids
238
312
    """
239
313
    old_tree.lock_read()
240
314
    try:
241
315
        new_tree.lock_read()
242
316
        try:
243
317
            return _show_diff_trees(old_tree, new_tree, to_file,
244
 
                                    specific_files, external_diff_options)
 
318
                                    specific_files, external_diff_options,
 
319
                                    old_label=old_label, new_label=new_label,
 
320
                                    extra_trees=extra_trees)
245
321
        finally:
246
322
            new_tree.unlock()
247
323
    finally:
249
325
 
250
326
 
251
327
def _show_diff_trees(old_tree, new_tree, to_file,
252
 
                     specific_files, external_diff_options):
253
 
 
254
 
    # TODO: Options to control putting on a prefix or suffix, perhaps
255
 
    # as a format string?
256
 
    old_label = 'a/'
257
 
    new_label = 'b/'
258
 
 
259
 
    DEVNULL = '/dev/null'
260
 
    # Windows users, don't panic about this filename -- it is a
261
 
    # special signal to GNU patch that the file should be created or
262
 
    # deleted respectively.
 
328
                     specific_files, external_diff_options, 
 
329
                     old_label='a/', new_label='b/', extra_trees=None):
 
330
 
 
331
    # GNU Patch uses the epoch date to detect files that are being added
 
332
    # or removed in a diff.
 
333
    EPOCH_DATE = '1970-01-01 00:00:00 +0000'
263
334
 
264
335
    # TODO: Generation of pseudo-diffs for added/deleted files could
265
336
    # be usefully made into a much faster special case.
266
337
 
267
 
    _raise_if_doubly_unversioned(specific_files, old_tree, new_tree)
268
 
 
269
338
    if external_diff_options:
270
339
        assert isinstance(external_diff_options, basestring)
271
340
        opts = external_diff_options.split()
274
343
    else:
275
344
        diff_file = internal_diff
276
345
    
277
 
    delta = compare_trees(old_tree, new_tree, want_unchanged=False,
278
 
                          specific_files=specific_files)
 
346
    delta = new_tree.changes_from(old_tree,
 
347
        specific_files=specific_files,
 
348
        extra_trees=extra_trees, require_versioned=True)
279
349
 
280
350
    has_changes = 0
281
351
    for path, file_id, kind in delta.removed:
282
352
        has_changes = 1
283
 
        print >>to_file, '=== removed %s %r' % (kind, old_label + path)
284
 
        old_tree.inventory[file_id].diff(diff_file, old_label + path, old_tree,
285
 
                                         DEVNULL, None, None, to_file)
 
353
        print >>to_file, '=== removed %s %r' % (kind, path.encode('utf8'))
 
354
        old_name = '%s%s\t%s' % (old_label, path,
 
355
                                 _patch_header_date(old_tree, file_id, path))
 
356
        new_name = '%s%s\t%s' % (new_label, path, EPOCH_DATE)
 
357
        old_tree.inventory[file_id].diff(diff_file, old_name, old_tree,
 
358
                                         new_name, None, None, to_file)
286
359
    for path, file_id, kind in delta.added:
287
360
        has_changes = 1
288
 
        print >>to_file, '=== added %s %r' % (kind, new_label + path)
289
 
        new_tree.inventory[file_id].diff(diff_file, new_label + path, new_tree,
290
 
                                         DEVNULL, None, None, to_file, 
 
361
        print >>to_file, '=== added %s %r' % (kind, path.encode('utf8'))
 
362
        old_name = '%s%s\t%s' % (old_label, path, EPOCH_DATE)
 
363
        new_name = '%s%s\t%s' % (new_label, path,
 
364
                                 _patch_header_date(new_tree, file_id, path))
 
365
        new_tree.inventory[file_id].diff(diff_file, new_name, new_tree,
 
366
                                         old_name, None, None, to_file, 
291
367
                                         reverse=True)
292
368
    for (old_path, new_path, file_id, kind,
293
369
         text_modified, meta_modified) in delta.renamed:
294
370
        has_changes = 1
295
371
        prop_str = get_prop_change(meta_modified)
296
372
        print >>to_file, '=== renamed %s %r => %r%s' % (
297
 
                    kind, old_label + old_path, new_label + new_path, prop_str)
298
 
        _maybe_diff_file_or_symlink(old_label, old_path, old_tree, file_id,
299
 
                                    new_label, new_path, new_tree,
 
373
                    kind, old_path.encode('utf8'),
 
374
                    new_path.encode('utf8'), prop_str)
 
375
        old_name = '%s%s\t%s' % (old_label, old_path,
 
376
                                 _patch_header_date(old_tree, file_id,
 
377
                                                    old_path))
 
378
        new_name = '%s%s\t%s' % (new_label, new_path,
 
379
                                 _patch_header_date(new_tree, file_id,
 
380
                                                    new_path))
 
381
        _maybe_diff_file_or_symlink(old_name, old_tree, file_id,
 
382
                                    new_name, new_tree,
300
383
                                    text_modified, kind, to_file, diff_file)
301
384
    for path, file_id, kind, text_modified, meta_modified in delta.modified:
302
385
        has_changes = 1
303
386
        prop_str = get_prop_change(meta_modified)
304
 
        print >>to_file, '=== modified %s %r%s' % (kind, old_label + path,
305
 
                    prop_str)
 
387
        print >>to_file, '=== modified %s %r%s' % (kind, path.encode('utf8'), prop_str)
 
388
        old_name = '%s%s\t%s' % (old_label, path,
 
389
                                 _patch_header_date(old_tree, file_id, path))
 
390
        new_name = '%s%s\t%s' % (new_label, path,
 
391
                                 _patch_header_date(new_tree, file_id, path))
306
392
        if text_modified:
307
 
            _maybe_diff_file_or_symlink(old_label, path, old_tree, file_id,
308
 
                                        new_label, path, new_tree,
 
393
            _maybe_diff_file_or_symlink(old_name, old_tree, file_id,
 
394
                                        new_name, new_tree,
309
395
                                        True, kind, to_file, diff_file)
310
396
 
311
397
    return has_changes
312
398
 
313
399
 
314
 
def _raise_if_doubly_unversioned(specific_files, old_tree, new_tree):
315
 
    """Complain if paths are not versioned in either tree."""
316
 
    if not specific_files:
 
400
def _patch_header_date(tree, file_id, path):
 
401
    """Returns a timestamp suitable for use in a patch header."""
 
402
    tm = time.gmtime(tree.get_file_mtime(file_id, path))
 
403
    return time.strftime('%Y-%m-%d %H:%M:%S +0000', tm)
 
404
 
 
405
 
 
406
def _raise_if_nonexistent(paths, old_tree, new_tree):
 
407
    """Complain if paths are not in either inventory or tree.
 
408
 
 
409
    It's OK with the files exist in either tree's inventory, or 
 
410
    if they exist in the tree but are not versioned.
 
411
    
 
412
    This can be used by operations such as bzr status that can accept
 
413
    unknown or ignored files.
 
414
    """
 
415
    mutter("check paths: %r", paths)
 
416
    if not paths:
317
417
        return
318
 
    old_unversioned = old_tree.filter_unversioned_files(specific_files)
319
 
    new_unversioned = new_tree.filter_unversioned_files(specific_files)
320
 
    unversioned = old_unversioned.intersection(new_unversioned)
321
 
    if unversioned:
322
 
        raise errors.PathsNotVersionedError(sorted(unversioned))
323
 
    
 
418
    s = old_tree.filter_unversioned_files(paths)
 
419
    s = new_tree.filter_unversioned_files(s)
 
420
    s = [path for path in s if not new_tree.has_filename(path)]
 
421
    if s:
 
422
        raise errors.PathsDoNotExist(sorted(s))
 
423
 
324
424
 
325
425
def get_prop_change(meta_modified):
326
426
    if meta_modified:
329
429
        return  ""
330
430
 
331
431
 
332
 
def _maybe_diff_file_or_symlink(old_label, old_path, old_tree, file_id,
333
 
                                new_label, new_path, new_tree, text_modified,
 
432
def _maybe_diff_file_or_symlink(old_path, old_tree, file_id,
 
433
                                new_path, new_tree, text_modified,
334
434
                                kind, to_file, diff_file):
335
435
    if text_modified:
336
436
        new_entry = new_tree.inventory[file_id]
337
437
        old_tree.inventory[file_id].diff(diff_file,
338
 
                                         old_label + old_path, old_tree,
339
 
                                         new_label + new_path, new_entry, 
 
438
                                         old_path, old_tree,
 
439
                                         new_path, new_entry, 
340
440
                                         new_tree, to_file)