/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: Robert Collins
  • Date: 2007-06-28 02:43:50 UTC
  • mfrom: (2553 +trunk)
  • mto: This revision was merged to the branch mainline in revision 2558.
  • Revision ID: robertc@robertcollins.net-20070628024350-z8bdm0y6yz2uyf4o
Merge bzr.dev.

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
    timestamp,
 
34
    )
 
35
""")
 
36
 
 
37
# compatability - plugins import compare_trees from diff!!!
 
38
# deprecated as of 0.10
17
39
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
 
40
from bzrlib.symbol_versioning import (
 
41
        deprecated_function,
 
42
        zero_eight,
 
43
        )
 
44
from bzrlib.trace import mutter, warning
 
45
 
22
46
 
23
47
# TODO: Rather than building a changeset object, we should probably
24
48
# invoke callbacks on an object.  That object can either accumulate a
25
49
# list, write them out directly, etc etc.
26
50
 
27
 
def internal_diff(old_filename, oldlines, new_filename, newlines, to_file):
28
 
    import difflib
29
 
    
 
51
def internal_diff(old_filename, oldlines, new_filename, newlines, to_file,
 
52
                  allow_binary=False, sequence_matcher=None,
 
53
                  path_encoding='utf8'):
30
54
    # FIXME: difflib is wrong if there is no trailing newline.
31
55
    # The syntax used by patch seems to be "\ No newline at
32
56
    # end of file" following the last diff line from that
42
66
    # both sequences are empty.
43
67
    if not oldlines and not newlines:
44
68
        return
 
69
    
 
70
    if allow_binary is False:
 
71
        textfile.check_text_lines(oldlines)
 
72
        textfile.check_text_lines(newlines)
45
73
 
46
 
    ud = difflib.unified_diff(oldlines, newlines,
47
 
                              fromfile=old_filename+'\t', 
48
 
                              tofile=new_filename+'\t')
 
74
    if sequence_matcher is None:
 
75
        sequence_matcher = patiencediff.PatienceSequenceMatcher
 
76
    ud = patiencediff.unified_diff(oldlines, newlines,
 
77
                      fromfile=old_filename.encode(path_encoding),
 
78
                      tofile=new_filename.encode(path_encoding),
 
79
                      sequencematcher=sequence_matcher)
49
80
 
50
81
    ud = list(ud)
51
82
    # work-around for difflib being too smart for its own good
65
96
    print >>to_file
66
97
 
67
98
 
 
99
def _spawn_external_diff(diffcmd, capture_errors=True):
 
100
    """Spawn the externall diff process, and return the child handle.
 
101
 
 
102
    :param diffcmd: The command list to spawn
 
103
    :param capture_errors: Capture stderr as well as setting LANG=C
 
104
        and LC_ALL=C. This lets us read and understand the output of diff,
 
105
        and respond to any errors.
 
106
    :return: A Popen object.
 
107
    """
 
108
    if capture_errors:
 
109
        # construct minimal environment
 
110
        env = {}
 
111
        path = os.environ.get('PATH')
 
112
        if path is not None:
 
113
            env['PATH'] = path
 
114
        env['LANGUAGE'] = 'C'   # on win32 only LANGUAGE has effect
 
115
        env['LANG'] = 'C'
 
116
        env['LC_ALL'] = 'C'
 
117
        stderr = subprocess.PIPE
 
118
    else:
 
119
        env = None
 
120
        stderr = None
 
121
 
 
122
    try:
 
123
        pipe = subprocess.Popen(diffcmd,
 
124
                                stdin=subprocess.PIPE,
 
125
                                stdout=subprocess.PIPE,
 
126
                                stderr=stderr,
 
127
                                env=env)
 
128
    except OSError, e:
 
129
        if e.errno == errno.ENOENT:
 
130
            raise errors.NoDiff(str(e))
 
131
        raise
 
132
 
 
133
    return pipe
 
134
 
 
135
 
68
136
def external_diff(old_filename, oldlines, new_filename, newlines, to_file,
69
137
                  diff_opts):
70
138
    """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
139
    # make sure our own output is properly ordered before the diff
78
140
    to_file.flush()
79
141
 
80
 
    from tempfile import NamedTemporaryFile
81
 
    import os
82
 
 
83
 
    oldtmpf = NamedTemporaryFile()
84
 
    newtmpf = NamedTemporaryFile()
 
142
    oldtmp_fd, old_abspath = tempfile.mkstemp(prefix='bzr-diff-old-')
 
143
    newtmp_fd, new_abspath = tempfile.mkstemp(prefix='bzr-diff-new-')
 
144
    oldtmpf = os.fdopen(oldtmp_fd, 'wb')
 
145
    newtmpf = os.fdopen(newtmp_fd, 'wb')
85
146
 
86
147
    try:
87
148
        # TODO: perhaps a special case for comparing to or from the empty
94
155
        oldtmpf.writelines(oldlines)
95
156
        newtmpf.writelines(newlines)
96
157
 
97
 
        oldtmpf.flush()
98
 
        newtmpf.flush()
 
158
        oldtmpf.close()
 
159
        newtmpf.close()
99
160
 
100
161
        if not diff_opts:
101
162
            diff_opts = []
102
163
        diffcmd = ['diff',
103
 
                   '--label', old_filename+'\t',
104
 
                   oldtmpf.name,
105
 
                   '--label', new_filename+'\t',
106
 
                   newtmpf.name]
 
164
                   '--label', old_filename,
 
165
                   old_abspath,
 
166
                   '--label', new_filename,
 
167
                   new_abspath,
 
168
                   '--binary',
 
169
                  ]
107
170
 
108
171
        # diff only allows one style to be specified; they don't override.
109
172
        # note that some of these take optargs, and the optargs can be
129
192
        if diff_opts:
130
193
            diffcmd.extend(diff_opts)
131
194
 
132
 
        rc = os.spawnvp(os.P_WAIT, 'diff', diffcmd)
 
195
        pipe = _spawn_external_diff(diffcmd, capture_errors=True)
 
196
        out,err = pipe.communicate()
 
197
        rc = pipe.returncode
133
198
        
134
 
        if rc != 0 and rc != 1:
 
199
        # internal_diff() adds a trailing newline, add one here for consistency
 
200
        out += '\n'
 
201
        if rc == 2:
 
202
            # 'diff' gives retcode == 2 for all sorts of errors
 
203
            # one of those is 'Binary files differ'.
 
204
            # Bad options could also be the problem.
 
205
            # 'Binary files' is not a real error, so we suppress that error.
 
206
            lang_c_out = out
 
207
 
 
208
            # Since we got here, we want to make sure to give an i18n error
 
209
            pipe = _spawn_external_diff(diffcmd, capture_errors=False)
 
210
            out, err = pipe.communicate()
 
211
 
 
212
            # Write out the new i18n diff response
 
213
            to_file.write(out+'\n')
 
214
            if pipe.returncode != 2:
 
215
                raise errors.BzrError(
 
216
                               'external diff failed with exit code 2'
 
217
                               ' when run with LANG=C and LC_ALL=C,'
 
218
                               ' but not when run natively: %r' % (diffcmd,))
 
219
 
 
220
            first_line = lang_c_out.split('\n', 1)[0]
 
221
            # Starting with diffutils 2.8.4 the word "binary" was dropped.
 
222
            m = re.match('^(binary )?files.*differ$', first_line, re.I)
 
223
            if m is None:
 
224
                raise errors.BzrError('external diff failed with exit code 2;'
 
225
                                      ' command: %r' % (diffcmd,))
 
226
            else:
 
227
                # Binary files differ, just return
 
228
                return
 
229
 
 
230
        # If we got to here, we haven't written out the output of diff
 
231
        # do so now
 
232
        to_file.write(out)
 
233
        if rc not in (0, 1):
135
234
            # returns 1 if files differ; that's OK
136
235
            if rc < 0:
137
236
                msg = 'signal %d' % (-rc)
138
237
            else:
139
238
                msg = 'exit code %d' % rc
140
239
                
141
 
            raise BzrError('external diff failed with %s; command: %r' % (rc, diffcmd))
 
240
            raise errors.BzrError('external diff failed with %s; command: %r' 
 
241
                                  % (rc, diffcmd))
 
242
 
 
243
 
142
244
    finally:
143
245
        oldtmpf.close()                 # and delete
144
246
        newtmpf.close()
 
247
        # Clean up. Warn in case the files couldn't be deleted
 
248
        # (in case windows still holds the file open, but not
 
249
        # if the files have already been deleted)
 
250
        try:
 
251
            os.remove(old_abspath)
 
252
        except OSError, e:
 
253
            if e.errno not in (errno.ENOENT,):
 
254
                warning('Failed to delete temporary file: %s %s',
 
255
                        old_abspath, e)
 
256
        try:
 
257
            os.remove(new_abspath)
 
258
        except OSError:
 
259
            if e.errno not in (errno.ENOENT,):
 
260
                warning('Failed to delete temporary file: %s %s',
 
261
                        new_abspath, e)
145
262
 
146
263
 
147
264
@deprecated_function(zero_eight)
161
278
    supplies any two trees.
162
279
    """
163
280
    if output is None:
164
 
        import sys
165
281
        output = sys.stdout
166
282
 
167
283
    if from_spec is None:
184
300
 
185
301
 
186
302
def diff_cmd_helper(tree, specific_files, external_diff_options, 
187
 
                    old_revision_spec=None, new_revision_spec=None):
 
303
                    old_revision_spec=None, new_revision_spec=None,
 
304
                    revision_specs=None,
 
305
                    old_label='a/', new_label='b/'):
188
306
    """Helper for cmd_diff.
189
307
 
190
 
   tree 
 
308
    :param tree:
191
309
        A WorkingTree
192
310
 
193
 
    specific_files
 
311
    :param specific_files:
194
312
        The specific files to compare, or None
195
313
 
196
 
    external_diff_options
 
314
    :param external_diff_options:
197
315
        If non-None, run an external diff, and pass it these options
198
316
 
199
 
    old_revision_spec
 
317
    :param old_revision_spec:
200
318
        If None, use basis tree as old revision, otherwise use the tree for
201
319
        the specified revision. 
202
320
 
203
 
    new_revision_spec
 
321
    :param new_revision_spec:
204
322
        If None, use working tree as new revision, otherwise use the tree for
205
323
        the specified revision.
206
324
    
 
325
    :param revision_specs: 
 
326
        Zero, one or two RevisionSpecs from the command line, saying what revisions 
 
327
        to compare.  This can be passed as an alternative to the old_revision_spec 
 
328
        and new_revision_spec parameters.
 
329
 
207
330
    The more general form is show_diff_trees(), where the caller
208
331
    supplies any two trees.
209
332
    """
210
 
    import sys
211
 
    output = sys.stdout
 
333
 
 
334
    # TODO: perhaps remove the old parameters old_revision_spec and
 
335
    # new_revision_spec, since this is only really for use from cmd_diff and
 
336
    # it now always passes through a sequence of revision_specs -- mbp
 
337
    # 20061221
 
338
 
212
339
    def spec_tree(spec):
213
 
        revision_id = spec.in_store(tree.branch).rev_id
214
 
        return tree.branch.repository.revision_tree(revision_id)
 
340
        if tree:
 
341
            revision = spec.in_store(tree.branch)
 
342
        else:
 
343
            revision = spec.in_store(None)
 
344
        revision_id = revision.rev_id
 
345
        branch = revision.branch
 
346
        return branch.repository.revision_tree(revision_id)
 
347
 
 
348
    if revision_specs is not None:
 
349
        assert (old_revision_spec is None
 
350
                and new_revision_spec is None)
 
351
        if len(revision_specs) > 0:
 
352
            old_revision_spec = revision_specs[0]
 
353
        if len(revision_specs) > 1:
 
354
            new_revision_spec = revision_specs[1]
 
355
 
215
356
    if old_revision_spec is None:
216
357
        old_tree = tree.basis_tree()
217
358
    else:
218
359
        old_tree = spec_tree(old_revision_spec)
219
360
 
220
 
    if new_revision_spec is None:
 
361
    if (new_revision_spec is None
 
362
        or new_revision_spec.spec is None):
221
363
        new_tree = tree
222
364
    else:
223
365
        new_tree = spec_tree(new_revision_spec)
224
366
 
 
367
    if new_tree is not tree:
 
368
        extra_trees = (tree,)
 
369
    else:
 
370
        extra_trees = None
 
371
 
225
372
    return show_diff_trees(old_tree, new_tree, sys.stdout, specific_files,
226
 
                           external_diff_options)
 
373
                           external_diff_options,
 
374
                           old_label=old_label, new_label=new_label,
 
375
                           extra_trees=extra_trees)
227
376
 
228
377
 
229
378
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None,
230
 
                    external_diff_options=None):
 
379
                    external_diff_options=None,
 
380
                    old_label='a/', new_label='b/',
 
381
                    extra_trees=None):
231
382
    """Show in text form the changes from one tree to another.
232
383
 
233
384
    to_files
235
386
 
236
387
    external_diff_options
237
388
        If set, use an external GNU diff and pass these options.
 
389
 
 
390
    extra_trees
 
391
        If set, more Trees to use for looking up file ids
238
392
    """
239
393
    old_tree.lock_read()
240
394
    try:
 
395
        if extra_trees is not None:
 
396
            for tree in extra_trees:
 
397
                tree.lock_read()
241
398
        new_tree.lock_read()
242
399
        try:
243
400
            return _show_diff_trees(old_tree, new_tree, to_file,
244
 
                                    specific_files, external_diff_options)
 
401
                                    specific_files, external_diff_options,
 
402
                                    old_label=old_label, new_label=new_label,
 
403
                                    extra_trees=extra_trees)
245
404
        finally:
246
405
            new_tree.unlock()
 
406
            if extra_trees is not None:
 
407
                for tree in extra_trees:
 
408
                    tree.unlock()
247
409
    finally:
248
410
        old_tree.unlock()
249
411
 
250
412
 
251
413
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.
 
414
                     specific_files, external_diff_options, 
 
415
                     old_label='a/', new_label='b/', extra_trees=None):
 
416
 
 
417
    # GNU Patch uses the epoch date to detect files that are being added
 
418
    # or removed in a diff.
 
419
    EPOCH_DATE = '1970-01-01 00:00:00 +0000'
263
420
 
264
421
    # TODO: Generation of pseudo-diffs for added/deleted files could
265
422
    # be usefully made into a much faster special case.
266
423
 
267
 
    _raise_if_doubly_unversioned(specific_files, old_tree, new_tree)
268
 
 
269
424
    if external_diff_options:
270
425
        assert isinstance(external_diff_options, basestring)
271
426
        opts = external_diff_options.split()
274
429
    else:
275
430
        diff_file = internal_diff
276
431
    
277
 
    delta = compare_trees(old_tree, new_tree, want_unchanged=False,
278
 
                          specific_files=specific_files)
 
432
    delta = new_tree.changes_from(old_tree,
 
433
        specific_files=specific_files,
 
434
        extra_trees=extra_trees, require_versioned=True)
279
435
 
280
436
    has_changes = 0
281
437
    for path, file_id, kind in delta.removed:
282
438
        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)
 
439
        print >>to_file, '=== removed %s %r' % (kind, path.encode('utf8'))
 
440
        old_name = '%s%s\t%s' % (old_label, path,
 
441
                                 _patch_header_date(old_tree, file_id, path))
 
442
        new_name = '%s%s\t%s' % (new_label, path, EPOCH_DATE)
 
443
        old_tree.inventory[file_id].diff(diff_file, old_name, old_tree,
 
444
                                         new_name, None, None, to_file)
286
445
    for path, file_id, kind in delta.added:
287
446
        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, 
 
447
        print >>to_file, '=== added %s %r' % (kind, path.encode('utf8'))
 
448
        old_name = '%s%s\t%s' % (old_label, path, EPOCH_DATE)
 
449
        new_name = '%s%s\t%s' % (new_label, path,
 
450
                                 _patch_header_date(new_tree, file_id, path))
 
451
        new_tree.inventory[file_id].diff(diff_file, new_name, new_tree,
 
452
                                         old_name, None, None, to_file, 
291
453
                                         reverse=True)
292
454
    for (old_path, new_path, file_id, kind,
293
455
         text_modified, meta_modified) in delta.renamed:
294
456
        has_changes = 1
295
457
        prop_str = get_prop_change(meta_modified)
296
458
        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,
 
459
                    kind, old_path.encode('utf8'),
 
460
                    new_path.encode('utf8'), prop_str)
 
461
        old_name = '%s%s\t%s' % (old_label, old_path,
 
462
                                 _patch_header_date(old_tree, file_id,
 
463
                                                    old_path))
 
464
        new_name = '%s%s\t%s' % (new_label, new_path,
 
465
                                 _patch_header_date(new_tree, file_id,
 
466
                                                    new_path))
 
467
        _maybe_diff_file_or_symlink(old_name, old_tree, file_id,
 
468
                                    new_name, new_tree,
300
469
                                    text_modified, kind, to_file, diff_file)
301
470
    for path, file_id, kind, text_modified, meta_modified in delta.modified:
302
471
        has_changes = 1
303
472
        prop_str = get_prop_change(meta_modified)
304
 
        print >>to_file, '=== modified %s %r%s' % (kind, old_label + path,
305
 
                    prop_str)
 
473
        print >>to_file, '=== modified %s %r%s' % (kind, path.encode('utf8'), prop_str)
 
474
        # The file may be in a different location in the old tree (because
 
475
        # the containing dir was renamed, but the file itself was not)
 
476
        old_path = old_tree.id2path(file_id)
 
477
        old_name = '%s%s\t%s' % (old_label, old_path,
 
478
                                 _patch_header_date(old_tree, file_id, old_path))
 
479
        new_name = '%s%s\t%s' % (new_label, path,
 
480
                                 _patch_header_date(new_tree, file_id, path))
306
481
        if text_modified:
307
 
            _maybe_diff_file_or_symlink(old_label, path, old_tree, file_id,
308
 
                                        new_label, path, new_tree,
 
482
            _maybe_diff_file_or_symlink(old_name, old_tree, file_id,
 
483
                                        new_name, new_tree,
309
484
                                        True, kind, to_file, diff_file)
310
485
 
311
486
    return has_changes
312
487
 
313
488
 
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:
 
489
def _patch_header_date(tree, file_id, path):
 
490
    """Returns a timestamp suitable for use in a patch header."""
 
491
    mtime = tree.get_file_mtime(file_id, path)
 
492
    assert mtime is not None, \
 
493
        "got an mtime of None for file-id %s, path %s in tree %s" % (
 
494
                file_id, path, tree)
 
495
    return timestamp.format_patch_date(mtime)
 
496
 
 
497
 
 
498
def _raise_if_nonexistent(paths, old_tree, new_tree):
 
499
    """Complain if paths are not in either inventory or tree.
 
500
 
 
501
    It's OK with the files exist in either tree's inventory, or 
 
502
    if they exist in the tree but are not versioned.
 
503
    
 
504
    This can be used by operations such as bzr status that can accept
 
505
    unknown or ignored files.
 
506
    """
 
507
    mutter("check paths: %r", paths)
 
508
    if not paths:
317
509
        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
 
    
 
510
    s = old_tree.filter_unversioned_files(paths)
 
511
    s = new_tree.filter_unversioned_files(s)
 
512
    s = [path for path in s if not new_tree.has_filename(path)]
 
513
    if s:
 
514
        raise errors.PathsDoNotExist(sorted(s))
 
515
 
324
516
 
325
517
def get_prop_change(meta_modified):
326
518
    if meta_modified:
329
521
        return  ""
330
522
 
331
523
 
332
 
def _maybe_diff_file_or_symlink(old_label, old_path, old_tree, file_id,
333
 
                                new_label, new_path, new_tree, text_modified,
 
524
def _maybe_diff_file_or_symlink(old_path, old_tree, file_id,
 
525
                                new_path, new_tree, text_modified,
334
526
                                kind, to_file, diff_file):
335
527
    if text_modified:
336
528
        new_entry = new_tree.inventory[file_id]
337
529
        old_tree.inventory[file_id].diff(diff_file,
338
 
                                         old_label + old_path, old_tree,
339
 
                                         new_label + new_path, new_entry, 
 
530
                                         old_path, old_tree,
 
531
                                         new_path, new_entry, 
340
532
                                         new_tree, to_file)