~bzr-pqm/bzr/bzr.dev

« back to all changes in this revision

Viewing changes to bzrlib/diff.py

  • Committer: Robert Collins
  • Date: 2006-03-01 03:26:23 UTC
  • mto: (1594.2.4 integration)
  • mto: This revision was merged to the branch mainline in revision 1596.
  • Revision ID: robertc@robertcollins.net-20060301032623-9d3c073e102f2239
Move WeaveStore down into bzrlib.store.versioned.weave.

Show diffs side-by-side

added added

removed removed

Lines of Context:
1
 
# Copyright (C) 2004, 2005, 2006 Canonical Ltd.
2
 
#
 
1
# -*- coding: UTF-8 -*-
 
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
38
17
from bzrlib.delta import compare_trees
39
 
from bzrlib.symbol_versioning import (
40
 
        deprecated_function,
41
 
        zero_eight,
42
 
        )
43
 
from bzrlib.trace import mutter, warning
44
 
 
 
18
from bzrlib.errors import BzrError
 
19
from bzrlib.symbol_versioning import *
 
20
from bzrlib.trace import mutter
45
21
 
46
22
# TODO: Rather than building a changeset object, we should probably
47
23
# invoke callbacks on an object.  That object can either accumulate a
48
24
# list, write them out directly, etc etc.
49
25
 
50
 
def internal_diff(old_filename, oldlines, new_filename, newlines, to_file,
51
 
                  allow_binary=False, sequence_matcher=None,
52
 
                  path_encoding='utf8'):
 
26
def internal_diff(old_filename, oldlines, new_filename, newlines, to_file):
 
27
    import difflib
 
28
    
53
29
    # FIXME: difflib is wrong if there is no trailing newline.
54
30
    # The syntax used by patch seems to be "\ No newline at
55
31
    # end of file" following the last diff line from that
65
41
    # both sequences are empty.
66
42
    if not oldlines and not newlines:
67
43
        return
68
 
    
69
 
    if allow_binary is False:
70
 
        textfile.check_text_lines(oldlines)
71
 
        textfile.check_text_lines(newlines)
72
44
 
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)
 
45
    ud = difflib.unified_diff(oldlines, newlines,
 
46
                              fromfile=old_filename+'\t', 
 
47
                              tofile=new_filename+'\t')
79
48
 
80
49
    ud = list(ud)
81
50
    # work-around for difflib being too smart for its own good
95
64
    print >>to_file
96
65
 
97
66
 
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
 
 
141
67
def external_diff(old_filename, oldlines, new_filename, newlines, to_file,
142
68
                  diff_opts):
143
69
    """Display a diff by calling out to the external diff program."""
 
70
    import sys
 
71
    
 
72
    if to_file != sys.stdout:
 
73
        raise NotImplementedError("sorry, can't send external diff other than to stdout yet",
 
74
                                  to_file)
 
75
 
144
76
    # make sure our own output is properly ordered before the diff
145
77
    to_file.flush()
146
78
 
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')
 
79
    from tempfile import NamedTemporaryFile
 
80
    import os
 
81
 
 
82
    oldtmpf = NamedTemporaryFile()
 
83
    newtmpf = NamedTemporaryFile()
151
84
 
152
85
    try:
153
86
        # TODO: perhaps a special case for comparing to or from the empty
160
93
        oldtmpf.writelines(oldlines)
161
94
        newtmpf.writelines(newlines)
162
95
 
163
 
        oldtmpf.close()
164
 
        newtmpf.close()
 
96
        oldtmpf.flush()
 
97
        newtmpf.flush()
165
98
 
166
99
        if not diff_opts:
167
100
            diff_opts = []
168
101
        diffcmd = ['diff',
169
 
                   '--label', old_filename,
170
 
                   old_abspath,
171
 
                   '--label', new_filename,
172
 
                   new_abspath,
173
 
                   '--binary',
174
 
                  ]
 
102
                   '--label', old_filename+'\t',
 
103
                   oldtmpf.name,
 
104
                   '--label', new_filename+'\t',
 
105
                   newtmpf.name]
175
106
 
176
107
        # diff only allows one style to be specified; they don't override.
177
108
        # note that some of these take optargs, and the optargs can be
197
128
        if diff_opts:
198
129
            diffcmd.extend(diff_opts)
199
130
 
200
 
        pipe = _spawn_external_diff(diffcmd, capture_errors=True)
201
 
        out,err = pipe.communicate()
202
 
        rc = pipe.returncode
 
131
        rc = os.spawnvp(os.P_WAIT, 'diff', diffcmd)
203
132
        
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):
 
133
        if rc != 0 and rc != 1:
239
134
            # returns 1 if files differ; that's OK
240
135
            if rc < 0:
241
136
                msg = 'signal %d' % (-rc)
242
137
            else:
243
138
                msg = 'exit code %d' % rc
244
139
                
245
 
            raise errors.BzrError('external diff failed with %s; command: %r' 
246
 
                                  % (rc, diffcmd))
247
 
 
248
 
 
 
140
            raise BzrError('external diff failed with %s; command: %r' % (rc, diffcmd))
249
141
    finally:
250
142
        oldtmpf.close()                 # and delete
251
143
        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)
267
144
 
268
145
 
269
146
@deprecated_function(zero_eight)
283
160
    supplies any two trees.
284
161
    """
285
162
    if output is None:
 
163
        import sys
286
164
        output = sys.stdout
287
165
 
288
166
    if from_spec is None:
305
183
 
306
184
 
307
185
def diff_cmd_helper(tree, specific_files, external_diff_options, 
308
 
                    old_revision_spec=None, new_revision_spec=None,
309
 
                    revision_specs=None,
310
 
                    old_label='a/', new_label='b/'):
 
186
                    old_revision_spec=None, new_revision_spec=None):
311
187
    """Helper for cmd_diff.
312
188
 
313
 
    :param tree:
 
189
   tree 
314
190
        A WorkingTree
315
191
 
316
 
    :param specific_files:
 
192
    specific_files
317
193
        The specific files to compare, or None
318
194
 
319
 
    :param external_diff_options:
 
195
    external_diff_options
320
196
        If non-None, run an external diff, and pass it these options
321
197
 
322
 
    :param old_revision_spec:
 
198
    old_revision_spec
323
199
        If None, use basis tree as old revision, otherwise use the tree for
324
200
        the specified revision. 
325
201
 
326
 
    :param new_revision_spec:
 
202
    new_revision_spec
327
203
        If None, use working tree as new revision, otherwise use the tree for
328
204
        the specified revision.
329
205
    
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
 
 
335
206
    The more general form is show_diff_trees(), where the caller
336
207
    supplies any two trees.
337
208
    """
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
 
 
 
209
    import sys
 
210
    output = sys.stdout
344
211
    def spec_tree(spec):
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
 
 
 
212
        revision_id = spec.in_store(tree.branch).rev_id
 
213
        return tree.branch.repository.revision_tree(revision_id)
361
214
    if old_revision_spec is None:
362
215
        old_tree = tree.basis_tree()
363
216
    else:
364
217
        old_tree = spec_tree(old_revision_spec)
365
218
 
366
 
    if (new_revision_spec is None
367
 
        or new_revision_spec.spec is None):
 
219
    if new_revision_spec is None:
368
220
        new_tree = tree
369
221
    else:
370
222
        new_tree = spec_tree(new_revision_spec)
371
223
 
372
 
    if new_tree is not tree:
373
 
        extra_trees = (tree,)
374
 
    else:
375
 
        extra_trees = None
376
 
 
377
224
    return show_diff_trees(old_tree, new_tree, sys.stdout, specific_files,
378
 
                           external_diff_options,
379
 
                           old_label=old_label, new_label=new_label,
380
 
                           extra_trees=extra_trees)
 
225
                           external_diff_options)
381
226
 
382
227
 
383
228
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None,
384
 
                    external_diff_options=None,
385
 
                    old_label='a/', new_label='b/',
386
 
                    extra_trees=None):
 
229
                    external_diff_options=None):
387
230
    """Show in text form the changes from one tree to another.
388
231
 
389
232
    to_files
391
234
 
392
235
    external_diff_options
393
236
        If set, use an external GNU diff and pass these options.
 
237
    """
394
238
 
395
 
    extra_trees
396
 
        If set, more Trees to use for looking up file ids
397
 
    """
398
239
    old_tree.lock_read()
399
240
    try:
400
 
        if extra_trees is not None:
401
 
            for tree in extra_trees:
402
 
                tree.lock_read()
403
241
        new_tree.lock_read()
404
242
        try:
405
243
            return _show_diff_trees(old_tree, new_tree, to_file,
406
 
                                    specific_files, external_diff_options,
407
 
                                    old_label=old_label, new_label=new_label,
408
 
                                    extra_trees=extra_trees)
 
244
                                    specific_files, external_diff_options)
409
245
        finally:
410
246
            new_tree.unlock()
411
 
            if extra_trees is not None:
412
 
                for tree in extra_trees:
413
 
                    tree.unlock()
414
247
    finally:
415
248
        old_tree.unlock()
416
249
 
417
250
 
418
251
def _show_diff_trees(old_tree, new_tree, to_file,
419
 
                     specific_files, external_diff_options, 
420
 
                     old_label='a/', new_label='b/', extra_trees=None):
421
 
 
422
 
    # GNU Patch uses the epoch date to detect files that are being added
423
 
    # or removed in a diff.
424
 
    EPOCH_DATE = '1970-01-01 00:00:00 +0000'
 
252
                     specific_files, external_diff_options):
 
253
 
 
254
    # TODO: Options to control putting on a prefix or suffix, perhaps as a format string
 
255
    old_label = ''
 
256
    new_label = ''
 
257
 
 
258
    DEVNULL = '/dev/null'
 
259
    # Windows users, don't panic about this filename -- it is a
 
260
    # special signal to GNU patch that the file should be created or
 
261
    # deleted respectively.
425
262
 
426
263
    # TODO: Generation of pseudo-diffs for added/deleted files could
427
264
    # be usefully made into a much faster special case.
434
271
    else:
435
272
        diff_file = internal_diff
436
273
    
437
 
    delta = new_tree.changes_from(old_tree,
438
 
        specific_files=specific_files,
439
 
        extra_trees=extra_trees, require_versioned=True)
 
274
 
 
275
    delta = compare_trees(old_tree, new_tree, want_unchanged=False,
 
276
                          specific_files=specific_files)
440
277
 
441
278
    has_changes = 0
442
279
    for path, file_id, kind in delta.removed:
443
280
        has_changes = 1
444
 
        print >>to_file, '=== removed %s %r' % (kind, path.encode('utf8'))
445
 
        old_name = '%s%s\t%s' % (old_label, path,
446
 
                                 _patch_header_date(old_tree, file_id, path))
447
 
        new_name = '%s%s\t%s' % (new_label, path, EPOCH_DATE)
448
 
        old_tree.inventory[file_id].diff(diff_file, old_name, old_tree,
449
 
                                         new_name, None, None, to_file)
 
281
        print >>to_file, '=== removed %s %r' % (kind, path)
 
282
        old_tree.inventory[file_id].diff(diff_file, old_label + path, old_tree,
 
283
                                         DEVNULL, None, None, to_file)
450
284
    for path, file_id, kind in delta.added:
451
285
        has_changes = 1
452
 
        print >>to_file, '=== added %s %r' % (kind, path.encode('utf8'))
453
 
        old_name = '%s%s\t%s' % (old_label, path, EPOCH_DATE)
454
 
        new_name = '%s%s\t%s' % (new_label, path,
455
 
                                 _patch_header_date(new_tree, file_id, path))
456
 
        new_tree.inventory[file_id].diff(diff_file, new_name, new_tree,
457
 
                                         old_name, None, None, to_file, 
 
286
        print >>to_file, '=== added %s %r' % (kind, path)
 
287
        new_tree.inventory[file_id].diff(diff_file, new_label + path, new_tree,
 
288
                                         DEVNULL, None, None, to_file, 
458
289
                                         reverse=True)
459
290
    for (old_path, new_path, file_id, kind,
460
291
         text_modified, meta_modified) in delta.renamed:
461
292
        has_changes = 1
462
293
        prop_str = get_prop_change(meta_modified)
463
294
        print >>to_file, '=== renamed %s %r => %r%s' % (
464
 
                    kind, old_path.encode('utf8'),
465
 
                    new_path.encode('utf8'), prop_str)
466
 
        old_name = '%s%s\t%s' % (old_label, old_path,
467
 
                                 _patch_header_date(old_tree, file_id,
468
 
                                                    old_path))
469
 
        new_name = '%s%s\t%s' % (new_label, new_path,
470
 
                                 _patch_header_date(new_tree, file_id,
471
 
                                                    new_path))
472
 
        _maybe_diff_file_or_symlink(old_name, old_tree, file_id,
473
 
                                    new_name, new_tree,
 
295
                          kind, old_path, new_path, prop_str)
 
296
        _maybe_diff_file_or_symlink(old_label, old_path, old_tree, file_id,
 
297
                                    new_label, new_path, new_tree,
474
298
                                    text_modified, kind, to_file, diff_file)
475
299
    for path, file_id, kind, text_modified, meta_modified in delta.modified:
476
300
        has_changes = 1
477
301
        prop_str = get_prop_change(meta_modified)
478
 
        print >>to_file, '=== modified %s %r%s' % (kind, path.encode('utf8'), prop_str)
479
 
        old_name = '%s%s\t%s' % (old_label, path,
480
 
                                 _patch_header_date(old_tree, file_id, path))
481
 
        new_name = '%s%s\t%s' % (new_label, path,
482
 
                                 _patch_header_date(new_tree, file_id, path))
 
302
        print >>to_file, '=== modified %s %r%s' % (kind, path, prop_str)
483
303
        if text_modified:
484
 
            _maybe_diff_file_or_symlink(old_name, old_tree, file_id,
485
 
                                        new_name, new_tree,
 
304
            _maybe_diff_file_or_symlink(old_label, path, old_tree, file_id,
 
305
                                        new_label, path, new_tree,
486
306
                                        True, kind, to_file, diff_file)
487
 
 
488
307
    return has_changes
489
 
 
490
 
 
491
 
def _patch_header_date(tree, file_id, path):
492
 
    """Returns a timestamp suitable for use in a patch header."""
493
 
    tm = time.gmtime(tree.get_file_mtime(file_id, path))
494
 
    return time.strftime('%Y-%m-%d %H:%M:%S +0000', tm)
495
 
 
496
 
 
497
 
def _raise_if_nonexistent(paths, old_tree, new_tree):
498
 
    """Complain if paths are not in either inventory or tree.
499
 
 
500
 
    It's OK with the files exist in either tree's inventory, or 
501
 
    if they exist in the tree but are not versioned.
502
308
    
503
 
    This can be used by operations such as bzr status that can accept
504
 
    unknown or ignored files.
505
 
    """
506
 
    mutter("check paths: %r", paths)
507
 
    if not paths:
508
 
        return
509
 
    s = old_tree.filter_unversioned_files(paths)
510
 
    s = new_tree.filter_unversioned_files(s)
511
 
    s = [path for path in s if not new_tree.has_filename(path)]
512
 
    if s:
513
 
        raise errors.PathsDoNotExist(sorted(s))
514
 
 
515
309
 
516
310
def get_prop_change(meta_modified):
517
311
    if meta_modified:
520
314
        return  ""
521
315
 
522
316
 
523
 
def _maybe_diff_file_or_symlink(old_path, old_tree, file_id,
524
 
                                new_path, new_tree, text_modified,
 
317
def _maybe_diff_file_or_symlink(old_label, old_path, old_tree, file_id,
 
318
                                new_label, new_path, new_tree, text_modified,
525
319
                                kind, to_file, diff_file):
526
320
    if text_modified:
527
321
        new_entry = new_tree.inventory[file_id]
528
322
        old_tree.inventory[file_id].diff(diff_file,
529
 
                                         old_path, old_tree,
530
 
                                         new_path, new_entry, 
 
323
                                         old_label + old_path, old_tree,
 
324
                                         new_label + new_path, new_entry, 
531
325
                                         new_tree, to_file)