/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: Aaron Bentley
  • Date: 2007-01-17 15:39:21 UTC
  • mfrom: (1551.9.35 Aaron's mergeable stuff)
  • mto: This revision was merged to the branch mainline in revision 2239.
  • Revision ID: abentley@panoramicfeedback.com-20070117153921-6pp9ssa2r8n5izoo
Merge bzr.ab, to avoid conflicts submitting

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 os
 
18
import re
 
19
import sys
 
20
 
 
21
from bzrlib.lazy_import import lazy_import
 
22
lazy_import(globals(), """
 
23
import errno
 
24
import subprocess
 
25
import tempfile
 
26
import time
 
27
 
 
28
from bzrlib import (
 
29
    errors,
 
30
    osutils,
 
31
    patiencediff,
 
32
    textfile,
 
33
    )
 
34
""")
 
35
 
 
36
# compatability - plugins import compare_trees from diff!!!
 
37
# deprecated as of 0.10
17
38
from bzrlib.delta import compare_trees
18
 
from bzrlib.errors import BzrError
19
 
import bzrlib.errors as errors
20
 
from bzrlib.symbol_versioning import *
21
 
from bzrlib.trace import mutter
 
39
from bzrlib.symbol_versioning import (
 
40
        deprecated_function,
 
41
        zero_eight,
 
42
        )
 
43
from bzrlib.trace import mutter, warning
 
44
 
22
45
 
23
46
# TODO: Rather than building a changeset object, we should probably
24
47
# invoke callbacks on an object.  That object can either accumulate a
25
48
# list, write them out directly, etc etc.
26
49
 
27
 
def internal_diff(old_filename, oldlines, new_filename, newlines, to_file):
28
 
    import difflib
29
 
    
 
50
def internal_diff(old_filename, oldlines, new_filename, newlines, to_file,
 
51
                  allow_binary=False, sequence_matcher=None,
 
52
                  path_encoding='utf8'):
30
53
    # FIXME: difflib is wrong if there is no trailing newline.
31
54
    # The syntax used by patch seems to be "\ No newline at
32
55
    # end of file" following the last diff line from that
42
65
    # both sequences are empty.
43
66
    if not oldlines and not newlines:
44
67
        return
 
68
    
 
69
    if allow_binary is False:
 
70
        textfile.check_text_lines(oldlines)
 
71
        textfile.check_text_lines(newlines)
45
72
 
46
 
    ud = difflib.unified_diff(oldlines, newlines,
47
 
                              fromfile=old_filename+'\t', 
48
 
                              tofile=new_filename+'\t')
 
73
    if sequence_matcher is None:
 
74
        sequence_matcher = patiencediff.PatienceSequenceMatcher
 
75
    ud = patiencediff.unified_diff(oldlines, newlines,
 
76
                      fromfile=old_filename.encode(path_encoding),
 
77
                      tofile=new_filename.encode(path_encoding),
 
78
                      sequencematcher=sequence_matcher)
49
79
 
50
80
    ud = list(ud)
51
81
    # work-around for difflib being too smart for its own good
65
95
    print >>to_file
66
96
 
67
97
 
 
98
def _set_lang_C():
 
99
    """Set the env vars LANG=C and LC_ALL=C."""
 
100
    osutils.set_or_unset_env('LANG', 'C')
 
101
    osutils.set_or_unset_env('LC_ALL', 'C')
 
102
    osutils.set_or_unset_env('LC_CTYPE', None)
 
103
    osutils.set_or_unset_env('LANGUAGE', None)
 
104
 
 
105
 
 
106
def _spawn_external_diff(diffcmd, capture_errors=True):
 
107
    """Spawn the externall diff process, and return the child handle.
 
108
 
 
109
    :param diffcmd: The command list to spawn
 
110
    :param capture_errors: Capture stderr as well as setting LANG=C
 
111
        and LC_ALL=C. This lets us read and understand the output of diff,
 
112
        and respond to any errors.
 
113
    :return: A Popen object.
 
114
    """
 
115
    if capture_errors:
 
116
        if sys.platform == 'win32':
 
117
            # Win32 doesn't support preexec_fn, but that is
 
118
            # okay, because it doesn't support LANG and LC_ALL either.
 
119
            preexec_fn = None
 
120
        else:
 
121
            preexec_fn = _set_lang_C
 
122
        stderr = subprocess.PIPE
 
123
    else:
 
124
        preexec_fn = None
 
125
        stderr = None
 
126
 
 
127
    try:
 
128
        pipe = subprocess.Popen(diffcmd,
 
129
                                stdin=subprocess.PIPE,
 
130
                                stdout=subprocess.PIPE,
 
131
                                stderr=stderr,
 
132
                                preexec_fn=preexec_fn)
 
133
    except OSError, e:
 
134
        if e.errno == errno.ENOENT:
 
135
            raise errors.NoDiff(str(e))
 
136
        raise
 
137
 
 
138
    return pipe
 
139
 
 
140
 
68
141
def external_diff(old_filename, oldlines, new_filename, newlines, to_file,
69
142
                  diff_opts):
70
143
    """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
144
    # make sure our own output is properly ordered before the diff
78
145
    to_file.flush()
79
146
 
80
 
    from tempfile import NamedTemporaryFile
81
 
    import os
82
 
 
83
 
    oldtmpf = NamedTemporaryFile()
84
 
    newtmpf = NamedTemporaryFile()
 
147
    oldtmp_fd, old_abspath = tempfile.mkstemp(prefix='bzr-diff-old-')
 
148
    newtmp_fd, new_abspath = tempfile.mkstemp(prefix='bzr-diff-new-')
 
149
    oldtmpf = os.fdopen(oldtmp_fd, 'wb')
 
150
    newtmpf = os.fdopen(newtmp_fd, 'wb')
85
151
 
86
152
    try:
87
153
        # TODO: perhaps a special case for comparing to or from the empty
94
160
        oldtmpf.writelines(oldlines)
95
161
        newtmpf.writelines(newlines)
96
162
 
97
 
        oldtmpf.flush()
98
 
        newtmpf.flush()
 
163
        oldtmpf.close()
 
164
        newtmpf.close()
99
165
 
100
166
        if not diff_opts:
101
167
            diff_opts = []
102
168
        diffcmd = ['diff',
103
 
                   '--label', old_filename+'\t',
104
 
                   oldtmpf.name,
105
 
                   '--label', new_filename+'\t',
106
 
                   newtmpf.name]
 
169
                   '--label', old_filename,
 
170
                   old_abspath,
 
171
                   '--label', new_filename,
 
172
                   new_abspath,
 
173
                   '--binary',
 
174
                  ]
107
175
 
108
176
        # diff only allows one style to be specified; they don't override.
109
177
        # note that some of these take optargs, and the optargs can be
129
197
        if diff_opts:
130
198
            diffcmd.extend(diff_opts)
131
199
 
132
 
        rc = os.spawnvp(os.P_WAIT, 'diff', diffcmd)
 
200
        pipe = _spawn_external_diff(diffcmd, capture_errors=True)
 
201
        out,err = pipe.communicate()
 
202
        rc = pipe.returncode
133
203
        
134
 
        if rc != 0 and rc != 1:
 
204
        # internal_diff() adds a trailing newline, add one here for consistency
 
205
        out += '\n'
 
206
        if rc == 2:
 
207
            # 'diff' gives retcode == 2 for all sorts of errors
 
208
            # one of those is 'Binary files differ'.
 
209
            # Bad options could also be the problem.
 
210
            # 'Binary files' is not a real error, so we suppress that error.
 
211
            lang_c_out = out
 
212
 
 
213
            # Since we got here, we want to make sure to give an i18n error
 
214
            pipe = _spawn_external_diff(diffcmd, capture_errors=False)
 
215
            out, err = pipe.communicate()
 
216
 
 
217
            # Write out the new i18n diff response
 
218
            to_file.write(out+'\n')
 
219
            if pipe.returncode != 2:
 
220
                raise errors.BzrError(
 
221
                               'external diff failed with exit code 2'
 
222
                               ' when run with LANG=C and LC_ALL=C,'
 
223
                               ' but not when run natively: %r' % (diffcmd,))
 
224
 
 
225
            first_line = lang_c_out.split('\n', 1)[0]
 
226
            # Starting with diffutils 2.8.4 the word "binary" was dropped.
 
227
            m = re.match('^(binary )?files.*differ$', first_line, re.I)
 
228
            if m is None:
 
229
                raise errors.BzrError('external diff failed with exit code 2;'
 
230
                                      ' command: %r' % (diffcmd,))
 
231
            else:
 
232
                # Binary files differ, just return
 
233
                return
 
234
 
 
235
        # If we got to here, we haven't written out the output of diff
 
236
        # do so now
 
237
        to_file.write(out)
 
238
        if rc not in (0, 1):
135
239
            # returns 1 if files differ; that's OK
136
240
            if rc < 0:
137
241
                msg = 'signal %d' % (-rc)
138
242
            else:
139
243
                msg = 'exit code %d' % rc
140
244
                
141
 
            raise BzrError('external diff failed with %s; command: %r' % (rc, diffcmd))
 
245
            raise errors.BzrError('external diff failed with %s; command: %r' 
 
246
                                  % (rc, diffcmd))
 
247
 
 
248
 
142
249
    finally:
143
250
        oldtmpf.close()                 # and delete
144
251
        newtmpf.close()
 
252
        # Clean up. Warn in case the files couldn't be deleted
 
253
        # (in case windows still holds the file open, but not
 
254
        # if the files have already been deleted)
 
255
        try:
 
256
            os.remove(old_abspath)
 
257
        except OSError, e:
 
258
            if e.errno not in (errno.ENOENT,):
 
259
                warning('Failed to delete temporary file: %s %s',
 
260
                        old_abspath, e)
 
261
        try:
 
262
            os.remove(new_abspath)
 
263
        except OSError:
 
264
            if e.errno not in (errno.ENOENT,):
 
265
                warning('Failed to delete temporary file: %s %s',
 
266
                        new_abspath, e)
145
267
 
146
268
 
147
269
@deprecated_function(zero_eight)
161
283
    supplies any two trees.
162
284
    """
163
285
    if output is None:
164
 
        import sys
165
286
        output = sys.stdout
166
287
 
167
288
    if from_spec is None:
184
305
 
185
306
 
186
307
def diff_cmd_helper(tree, specific_files, external_diff_options, 
187
 
                    old_revision_spec=None, new_revision_spec=None):
 
308
                    old_revision_spec=None, new_revision_spec=None,
 
309
                    revision_specs=None,
 
310
                    old_label='a/', new_label='b/'):
188
311
    """Helper for cmd_diff.
189
312
 
190
 
   tree 
 
313
    :param tree:
191
314
        A WorkingTree
192
315
 
193
 
    specific_files
 
316
    :param specific_files:
194
317
        The specific files to compare, or None
195
318
 
196
 
    external_diff_options
 
319
    :param external_diff_options:
197
320
        If non-None, run an external diff, and pass it these options
198
321
 
199
 
    old_revision_spec
 
322
    :param old_revision_spec:
200
323
        If None, use basis tree as old revision, otherwise use the tree for
201
324
        the specified revision. 
202
325
 
203
 
    new_revision_spec
 
326
    :param new_revision_spec:
204
327
        If None, use working tree as new revision, otherwise use the tree for
205
328
        the specified revision.
206
329
    
 
330
    :param revision_specs: 
 
331
        Zero, one or two RevisionSpecs from the command line, saying what revisions 
 
332
        to compare.  This can be passed as an alternative to the old_revision_spec 
 
333
        and new_revision_spec parameters.
 
334
 
207
335
    The more general form is show_diff_trees(), where the caller
208
336
    supplies any two trees.
209
337
    """
210
 
    import sys
211
 
    output = sys.stdout
 
338
 
 
339
    # TODO: perhaps remove the old parameters old_revision_spec and
 
340
    # new_revision_spec, since this is only really for use from cmd_diff and
 
341
    # it now always passes through a sequence of revision_specs -- mbp
 
342
    # 20061221
 
343
 
212
344
    def spec_tree(spec):
213
 
        revision_id = spec.in_store(tree.branch).rev_id
214
 
        return tree.branch.repository.revision_tree(revision_id)
 
345
        if tree:
 
346
            revision = spec.in_store(tree.branch)
 
347
        else:
 
348
            revision = spec.in_store(None)
 
349
        revision_id = revision.rev_id
 
350
        branch = revision.branch
 
351
        return branch.repository.revision_tree(revision_id)
 
352
 
 
353
    if revision_specs is not None:
 
354
        assert (old_revision_spec is None
 
355
                and new_revision_spec is None)
 
356
        if len(revision_specs) > 0:
 
357
            old_revision_spec = revision_specs[0]
 
358
        if len(revision_specs) > 1:
 
359
            new_revision_spec = revision_specs[1]
 
360
 
215
361
    if old_revision_spec is None:
216
362
        old_tree = tree.basis_tree()
217
363
    else:
218
364
        old_tree = spec_tree(old_revision_spec)
219
365
 
220
 
    if new_revision_spec is None:
 
366
    if (new_revision_spec is None
 
367
        or new_revision_spec.spec is None):
221
368
        new_tree = tree
222
369
    else:
223
370
        new_tree = spec_tree(new_revision_spec)
224
371
 
 
372
    if new_tree is not tree:
 
373
        extra_trees = (tree,)
 
374
    else:
 
375
        extra_trees = None
 
376
 
225
377
    return show_diff_trees(old_tree, new_tree, sys.stdout, specific_files,
226
 
                           external_diff_options)
 
378
                           external_diff_options,
 
379
                           old_label=old_label, new_label=new_label,
 
380
                           extra_trees=extra_trees)
227
381
 
228
382
 
229
383
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None,
230
 
                    external_diff_options=None):
 
384
                    external_diff_options=None,
 
385
                    old_label='a/', new_label='b/',
 
386
                    extra_trees=None):
231
387
    """Show in text form the changes from one tree to another.
232
388
 
233
389
    to_files
235
391
 
236
392
    external_diff_options
237
393
        If set, use an external GNU diff and pass these options.
 
394
 
 
395
    extra_trees
 
396
        If set, more Trees to use for looking up file ids
238
397
    """
239
398
    old_tree.lock_read()
240
399
    try:
241
400
        new_tree.lock_read()
242
401
        try:
243
402
            return _show_diff_trees(old_tree, new_tree, to_file,
244
 
                                    specific_files, external_diff_options)
 
403
                                    specific_files, external_diff_options,
 
404
                                    old_label=old_label, new_label=new_label,
 
405
                                    extra_trees=extra_trees)
245
406
        finally:
246
407
            new_tree.unlock()
247
408
    finally:
249
410
 
250
411
 
251
412
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.
 
413
                     specific_files, external_diff_options, 
 
414
                     old_label='a/', new_label='b/', extra_trees=None):
 
415
 
 
416
    # GNU Patch uses the epoch date to detect files that are being added
 
417
    # or removed in a diff.
 
418
    EPOCH_DATE = '1970-01-01 00:00:00 +0000'
263
419
 
264
420
    # TODO: Generation of pseudo-diffs for added/deleted files could
265
421
    # be usefully made into a much faster special case.
266
422
 
267
 
    _raise_if_doubly_unversioned(specific_files, old_tree, new_tree)
268
 
 
269
423
    if external_diff_options:
270
424
        assert isinstance(external_diff_options, basestring)
271
425
        opts = external_diff_options.split()
274
428
    else:
275
429
        diff_file = internal_diff
276
430
    
277
 
    delta = compare_trees(old_tree, new_tree, want_unchanged=False,
278
 
                          specific_files=specific_files)
 
431
    delta = new_tree.changes_from(old_tree,
 
432
        specific_files=specific_files,
 
433
        extra_trees=extra_trees, require_versioned=True)
279
434
 
280
435
    has_changes = 0
281
436
    for path, file_id, kind in delta.removed:
282
437
        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)
 
438
        print >>to_file, '=== removed %s %r' % (kind, path.encode('utf8'))
 
439
        old_name = '%s%s\t%s' % (old_label, path,
 
440
                                 _patch_header_date(old_tree, file_id, path))
 
441
        new_name = '%s%s\t%s' % (new_label, path, EPOCH_DATE)
 
442
        old_tree.inventory[file_id].diff(diff_file, old_name, old_tree,
 
443
                                         new_name, None, None, to_file)
286
444
    for path, file_id, kind in delta.added:
287
445
        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, 
 
446
        print >>to_file, '=== added %s %r' % (kind, path.encode('utf8'))
 
447
        old_name = '%s%s\t%s' % (old_label, path, EPOCH_DATE)
 
448
        new_name = '%s%s\t%s' % (new_label, path,
 
449
                                 _patch_header_date(new_tree, file_id, path))
 
450
        new_tree.inventory[file_id].diff(diff_file, new_name, new_tree,
 
451
                                         old_name, None, None, to_file, 
291
452
                                         reverse=True)
292
453
    for (old_path, new_path, file_id, kind,
293
454
         text_modified, meta_modified) in delta.renamed:
294
455
        has_changes = 1
295
456
        prop_str = get_prop_change(meta_modified)
296
457
        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,
 
458
                    kind, old_path.encode('utf8'),
 
459
                    new_path.encode('utf8'), prop_str)
 
460
        old_name = '%s%s\t%s' % (old_label, old_path,
 
461
                                 _patch_header_date(old_tree, file_id,
 
462
                                                    old_path))
 
463
        new_name = '%s%s\t%s' % (new_label, new_path,
 
464
                                 _patch_header_date(new_tree, file_id,
 
465
                                                    new_path))
 
466
        _maybe_diff_file_or_symlink(old_name, old_tree, file_id,
 
467
                                    new_name, new_tree,
300
468
                                    text_modified, kind, to_file, diff_file)
301
469
    for path, file_id, kind, text_modified, meta_modified in delta.modified:
302
470
        has_changes = 1
303
471
        prop_str = get_prop_change(meta_modified)
304
 
        print >>to_file, '=== modified %s %r%s' % (kind, old_label + path,
305
 
                    prop_str)
 
472
        print >>to_file, '=== modified %s %r%s' % (kind, path.encode('utf8'), prop_str)
 
473
        old_name = '%s%s\t%s' % (old_label, path,
 
474
                                 _patch_header_date(old_tree, file_id, path))
 
475
        new_name = '%s%s\t%s' % (new_label, path,
 
476
                                 _patch_header_date(new_tree, file_id, path))
306
477
        if text_modified:
307
 
            _maybe_diff_file_or_symlink(old_label, path, old_tree, file_id,
308
 
                                        new_label, path, new_tree,
 
478
            _maybe_diff_file_or_symlink(old_name, old_tree, file_id,
 
479
                                        new_name, new_tree,
309
480
                                        True, kind, to_file, diff_file)
310
481
 
311
482
    return has_changes
312
483
 
313
484
 
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:
 
485
def _patch_header_date(tree, file_id, path):
 
486
    """Returns a timestamp suitable for use in a patch header."""
 
487
    tm = time.gmtime(tree.get_file_mtime(file_id, path))
 
488
    return time.strftime('%Y-%m-%d %H:%M:%S +0000', tm)
 
489
 
 
490
 
 
491
def _raise_if_nonexistent(paths, old_tree, new_tree):
 
492
    """Complain if paths are not in either inventory or tree.
 
493
 
 
494
    It's OK with the files exist in either tree's inventory, or 
 
495
    if they exist in the tree but are not versioned.
 
496
    
 
497
    This can be used by operations such as bzr status that can accept
 
498
    unknown or ignored files.
 
499
    """
 
500
    mutter("check paths: %r", paths)
 
501
    if not paths:
317
502
        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
 
    
 
503
    s = old_tree.filter_unversioned_files(paths)
 
504
    s = new_tree.filter_unversioned_files(s)
 
505
    s = [path for path in s if not new_tree.has_filename(path)]
 
506
    if s:
 
507
        raise errors.PathsDoNotExist(sorted(s))
 
508
 
324
509
 
325
510
def get_prop_change(meta_modified):
326
511
    if meta_modified:
329
514
        return  ""
330
515
 
331
516
 
332
 
def _maybe_diff_file_or_symlink(old_label, old_path, old_tree, file_id,
333
 
                                new_label, new_path, new_tree, text_modified,
 
517
def _maybe_diff_file_or_symlink(old_path, old_tree, file_id,
 
518
                                new_path, new_tree, text_modified,
334
519
                                kind, to_file, diff_file):
335
520
    if text_modified:
336
521
        new_entry = new_tree.inventory[file_id]
337
522
        old_tree.inventory[file_id].diff(diff_file,
338
 
                                         old_label + old_path, old_tree,
339
 
                                         new_label + new_path, new_entry, 
 
523
                                         old_path, old_tree,
 
524
                                         new_path, new_entry, 
340
525
                                         new_tree, to_file)