/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: 2006-09-08 00:19:48 UTC
  • mfrom: (1992 +trunk)
  • mto: This revision was merged to the branch mainline in revision 1993.
  • Revision ID: robertc@robertcollins.net-20060908001948-87b1b268cec30ede
Merge bzr.dev and undeprecated WorkingTree.last_revision as per review feedback.

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
65
88
    print >>to_file
66
89
 
67
90
 
 
91
def _set_lang_C():
 
92
    """Set the env var LANG=C"""
 
93
    os.environ['LANG'] = 'C'
 
94
 
 
95
 
 
96
def _spawn_external_diff(diffcmd, capture_errors=True):
 
97
    """Spawn the externall diff process, and return the child handle.
 
98
 
 
99
    :param diffcmd: The command list to spawn
 
100
    :param capture_errors: Capture stderr as well as setting LANG=C.
 
101
        This lets us read and understand the output of diff, and respond 
 
102
        to any errors.
 
103
    :return: A Popen object.
 
104
    """
 
105
    if capture_errors:
 
106
        if sys.platform == 'win32':
 
107
            # Win32 doesn't support preexec_fn, but that is
 
108
            # okay, because it doesn't support LANG either.
 
109
            preexec_fn = None
 
110
        else:
 
111
            preexec_fn = _set_lang_C
 
112
        stderr = subprocess.PIPE
 
113
    else:
 
114
        preexec_fn = None
 
115
        stderr = None
 
116
 
 
117
    try:
 
118
        pipe = subprocess.Popen(diffcmd,
 
119
                                stdin=subprocess.PIPE,
 
120
                                stdout=subprocess.PIPE,
 
121
                                stderr=stderr,
 
122
                                preexec_fn=preexec_fn)
 
123
    except OSError, e:
 
124
        if e.errno == errno.ENOENT:
 
125
            raise errors.NoDiff(str(e))
 
126
        raise
 
127
 
 
128
    return pipe
 
129
 
 
130
 
68
131
def external_diff(old_filename, oldlines, new_filename, newlines, to_file,
69
132
                  diff_opts):
70
133
    """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
134
    # make sure our own output is properly ordered before the diff
78
135
    to_file.flush()
79
136
 
80
 
    from tempfile import NamedTemporaryFile
81
 
    import os
82
 
 
83
 
    oldtmpf = NamedTemporaryFile()
84
 
    newtmpf = NamedTemporaryFile()
 
137
    oldtmp_fd, old_abspath = tempfile.mkstemp(prefix='bzr-diff-old-')
 
138
    newtmp_fd, new_abspath = tempfile.mkstemp(prefix='bzr-diff-new-')
 
139
    oldtmpf = os.fdopen(oldtmp_fd, 'wb')
 
140
    newtmpf = os.fdopen(newtmp_fd, 'wb')
85
141
 
86
142
    try:
87
143
        # TODO: perhaps a special case for comparing to or from the empty
94
150
        oldtmpf.writelines(oldlines)
95
151
        newtmpf.writelines(newlines)
96
152
 
97
 
        oldtmpf.flush()
98
 
        newtmpf.flush()
 
153
        oldtmpf.close()
 
154
        newtmpf.close()
99
155
 
100
156
        if not diff_opts:
101
157
            diff_opts = []
102
158
        diffcmd = ['diff',
103
 
                   '--label', old_filename+'\t',
104
 
                   oldtmpf.name,
105
 
                   '--label', new_filename+'\t',
106
 
                   newtmpf.name]
 
159
                   '--label', old_filename,
 
160
                   old_abspath,
 
161
                   '--label', new_filename,
 
162
                   new_abspath,
 
163
                   '--binary',
 
164
                  ]
107
165
 
108
166
        # diff only allows one style to be specified; they don't override.
109
167
        # note that some of these take optargs, and the optargs can be
129
187
        if diff_opts:
130
188
            diffcmd.extend(diff_opts)
131
189
 
132
 
        rc = os.spawnvp(os.P_WAIT, 'diff', diffcmd)
 
190
        pipe = _spawn_external_diff(diffcmd, capture_errors=True)
 
191
        out,err = pipe.communicate()
 
192
        rc = pipe.returncode
133
193
        
134
 
        if rc != 0 and rc != 1:
 
194
        # internal_diff() adds a trailing newline, add one here for consistency
 
195
        out += '\n'
 
196
        if rc == 2:
 
197
            # 'diff' gives retcode == 2 for all sorts of errors
 
198
            # one of those is 'Binary files differ'.
 
199
            # Bad options could also be the problem.
 
200
            # 'Binary files' is not a real error, so we suppress that error.
 
201
            lang_c_out = out
 
202
 
 
203
            # Since we got here, we want to make sure to give an i18n error
 
204
            pipe = _spawn_external_diff(diffcmd, capture_errors=False)
 
205
            out, err = pipe.communicate()
 
206
 
 
207
            # Write out the new i18n diff response
 
208
            to_file.write(out+'\n')
 
209
            if pipe.returncode != 2:
 
210
                raise BzrError('external diff failed with exit code 2'
 
211
                               ' when run with LANG=C, but not when run'
 
212
                               ' natively: %r' % (diffcmd,))
 
213
 
 
214
            first_line = lang_c_out.split('\n', 1)[0]
 
215
            # Starting with diffutils 2.8.4 the word "binary" was dropped.
 
216
            m = re.match('^(binary )?files.*differ$', first_line, re.I)
 
217
            if m is None:
 
218
                raise BzrError('external diff failed with exit code 2;'
 
219
                               ' command: %r' % (diffcmd,))
 
220
            else:
 
221
                # Binary files differ, just return
 
222
                return
 
223
 
 
224
        # If we got to here, we haven't written out the output of diff
 
225
        # do so now
 
226
        to_file.write(out)
 
227
        if rc not in (0, 1):
135
228
            # returns 1 if files differ; that's OK
136
229
            if rc < 0:
137
230
                msg = 'signal %d' % (-rc)
138
231
            else:
139
232
                msg = 'exit code %d' % rc
140
233
                
141
 
            raise BzrError('external diff failed with %s; command: %r' % (rc, diffcmd))
 
234
            raise BzrError('external diff failed with %s; command: %r' 
 
235
                           % (rc, diffcmd))
 
236
 
 
237
 
142
238
    finally:
143
239
        oldtmpf.close()                 # and delete
144
240
        newtmpf.close()
 
241
        # Clean up. Warn in case the files couldn't be deleted
 
242
        # (in case windows still holds the file open, but not
 
243
        # if the files have already been deleted)
 
244
        try:
 
245
            os.remove(old_abspath)
 
246
        except OSError, e:
 
247
            if e.errno not in (errno.ENOENT,):
 
248
                warning('Failed to delete temporary file: %s %s',
 
249
                        old_abspath, e)
 
250
        try:
 
251
            os.remove(new_abspath)
 
252
        except OSError:
 
253
            if e.errno not in (errno.ENOENT,):
 
254
                warning('Failed to delete temporary file: %s %s',
 
255
                        new_abspath, e)
145
256
 
146
257
 
147
258
@deprecated_function(zero_eight)
161
272
    supplies any two trees.
162
273
    """
163
274
    if output is None:
164
 
        import sys
165
275
        output = sys.stdout
166
276
 
167
277
    if from_spec is None:
184
294
 
185
295
 
186
296
def diff_cmd_helper(tree, specific_files, external_diff_options, 
187
 
                    old_revision_spec=None, new_revision_spec=None):
 
297
                    old_revision_spec=None, new_revision_spec=None,
 
298
                    old_label='a/', new_label='b/'):
188
299
    """Helper for cmd_diff.
189
300
 
190
301
   tree 
207
318
    The more general form is show_diff_trees(), where the caller
208
319
    supplies any two trees.
209
320
    """
210
 
    import sys
211
 
    output = sys.stdout
212
321
    def spec_tree(spec):
213
 
        revision_id = spec.in_store(tree.branch).rev_id
214
 
        return tree.branch.repository.revision_tree(revision_id)
 
322
        if tree:
 
323
            revision = spec.in_store(tree.branch)
 
324
        else:
 
325
            revision = spec.in_store(None)
 
326
        revision_id = revision.rev_id
 
327
        branch = revision.branch
 
328
        return branch.repository.revision_tree(revision_id)
215
329
    if old_revision_spec is None:
216
330
        old_tree = tree.basis_tree()
217
331
    else:
221
335
        new_tree = tree
222
336
    else:
223
337
        new_tree = spec_tree(new_revision_spec)
 
338
    if new_tree is not tree:
 
339
        extra_trees = (tree,)
 
340
    else:
 
341
        extra_trees = None
224
342
 
225
343
    return show_diff_trees(old_tree, new_tree, sys.stdout, specific_files,
226
 
                           external_diff_options)
 
344
                           external_diff_options,
 
345
                           old_label=old_label, new_label=new_label,
 
346
                           extra_trees=extra_trees)
227
347
 
228
348
 
229
349
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None,
230
 
                    external_diff_options=None):
 
350
                    external_diff_options=None,
 
351
                    old_label='a/', new_label='b/',
 
352
                    extra_trees=None):
231
353
    """Show in text form the changes from one tree to another.
232
354
 
233
355
    to_files
235
357
 
236
358
    external_diff_options
237
359
        If set, use an external GNU diff and pass these options.
 
360
 
 
361
    extra_trees
 
362
        If set, more Trees to use for looking up file ids
238
363
    """
239
364
    old_tree.lock_read()
240
365
    try:
241
366
        new_tree.lock_read()
242
367
        try:
243
368
            return _show_diff_trees(old_tree, new_tree, to_file,
244
 
                                    specific_files, external_diff_options)
 
369
                                    specific_files, external_diff_options,
 
370
                                    old_label=old_label, new_label=new_label,
 
371
                                    extra_trees=extra_trees)
245
372
        finally:
246
373
            new_tree.unlock()
247
374
    finally:
249
376
 
250
377
 
251
378
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.
 
379
                     specific_files, external_diff_options, 
 
380
                     old_label='a/', new_label='b/', extra_trees=None):
 
381
 
 
382
    # GNU Patch uses the epoch date to detect files that are being added
 
383
    # or removed in a diff.
 
384
    EPOCH_DATE = '1970-01-01 00:00:00 +0000'
263
385
 
264
386
    # TODO: Generation of pseudo-diffs for added/deleted files could
265
387
    # be usefully made into a much faster special case.
266
388
 
267
 
    _raise_if_doubly_unversioned(specific_files, old_tree, new_tree)
268
 
 
269
389
    if external_diff_options:
270
390
        assert isinstance(external_diff_options, basestring)
271
391
        opts = external_diff_options.split()
274
394
    else:
275
395
        diff_file = internal_diff
276
396
    
277
 
    delta = compare_trees(old_tree, new_tree, want_unchanged=False,
278
 
                          specific_files=specific_files)
 
397
    delta = new_tree.changes_from(old_tree,
 
398
        specific_files=specific_files,
 
399
        extra_trees=extra_trees, require_versioned=True)
279
400
 
280
401
    has_changes = 0
281
402
    for path, file_id, kind in delta.removed:
282
403
        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)
 
404
        print >>to_file, '=== removed %s %r' % (kind, path.encode('utf8'))
 
405
        old_name = '%s%s\t%s' % (old_label, path,
 
406
                                 _patch_header_date(old_tree, file_id, path))
 
407
        new_name = '%s%s\t%s' % (new_label, path, EPOCH_DATE)
 
408
        old_tree.inventory[file_id].diff(diff_file, old_name, old_tree,
 
409
                                         new_name, None, None, to_file)
286
410
    for path, file_id, kind in delta.added:
287
411
        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, 
 
412
        print >>to_file, '=== added %s %r' % (kind, path.encode('utf8'))
 
413
        old_name = '%s%s\t%s' % (old_label, path, EPOCH_DATE)
 
414
        new_name = '%s%s\t%s' % (new_label, path,
 
415
                                 _patch_header_date(new_tree, file_id, path))
 
416
        new_tree.inventory[file_id].diff(diff_file, new_name, new_tree,
 
417
                                         old_name, None, None, to_file, 
291
418
                                         reverse=True)
292
419
    for (old_path, new_path, file_id, kind,
293
420
         text_modified, meta_modified) in delta.renamed:
294
421
        has_changes = 1
295
422
        prop_str = get_prop_change(meta_modified)
296
423
        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,
 
424
                    kind, old_path.encode('utf8'),
 
425
                    new_path.encode('utf8'), prop_str)
 
426
        old_name = '%s%s\t%s' % (old_label, old_path,
 
427
                                 _patch_header_date(old_tree, file_id,
 
428
                                                    old_path))
 
429
        new_name = '%s%s\t%s' % (new_label, new_path,
 
430
                                 _patch_header_date(new_tree, file_id,
 
431
                                                    new_path))
 
432
        _maybe_diff_file_or_symlink(old_name, old_tree, file_id,
 
433
                                    new_name, new_tree,
300
434
                                    text_modified, kind, to_file, diff_file)
301
435
    for path, file_id, kind, text_modified, meta_modified in delta.modified:
302
436
        has_changes = 1
303
437
        prop_str = get_prop_change(meta_modified)
304
 
        print >>to_file, '=== modified %s %r%s' % (kind, old_label + path,
305
 
                    prop_str)
 
438
        print >>to_file, '=== modified %s %r%s' % (kind, path.encode('utf8'), prop_str)
 
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,
 
442
                                 _patch_header_date(new_tree, file_id, path))
306
443
        if text_modified:
307
 
            _maybe_diff_file_or_symlink(old_label, path, old_tree, file_id,
308
 
                                        new_label, path, new_tree,
 
444
            _maybe_diff_file_or_symlink(old_name, old_tree, file_id,
 
445
                                        new_name, new_tree,
309
446
                                        True, kind, to_file, diff_file)
310
447
 
311
448
    return has_changes
312
449
 
313
450
 
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:
 
451
def _patch_header_date(tree, file_id, path):
 
452
    """Returns a timestamp suitable for use in a patch header."""
 
453
    tm = time.gmtime(tree.get_file_mtime(file_id, path))
 
454
    return time.strftime('%Y-%m-%d %H:%M:%S +0000', tm)
 
455
 
 
456
 
 
457
def _raise_if_nonexistent(paths, old_tree, new_tree):
 
458
    """Complain if paths are not in either inventory or tree.
 
459
 
 
460
    It's OK with the files exist in either tree's inventory, or 
 
461
    if they exist in the tree but are not versioned.
 
462
    
 
463
    This can be used by operations such as bzr status that can accept
 
464
    unknown or ignored files.
 
465
    """
 
466
    mutter("check paths: %r", paths)
 
467
    if not paths:
317
468
        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
 
    
 
469
    s = old_tree.filter_unversioned_files(paths)
 
470
    s = new_tree.filter_unversioned_files(s)
 
471
    s = [path for path in s if not new_tree.has_filename(path)]
 
472
    if s:
 
473
        raise errors.PathsDoNotExist(sorted(s))
 
474
 
324
475
 
325
476
def get_prop_change(meta_modified):
326
477
    if meta_modified:
329
480
        return  ""
330
481
 
331
482
 
332
 
def _maybe_diff_file_or_symlink(old_label, old_path, old_tree, file_id,
333
 
                                new_label, new_path, new_tree, text_modified,
 
483
def _maybe_diff_file_or_symlink(old_path, old_tree, file_id,
 
484
                                new_path, new_tree, text_modified,
334
485
                                kind, to_file, diff_file):
335
486
    if text_modified:
336
487
        new_entry = new_tree.inventory[file_id]
337
488
        old_tree.inventory[file_id].diff(diff_file,
338
 
                                         old_label + old_path, old_tree,
339
 
                                         new_label + new_path, new_entry, 
 
489
                                         old_path, old_tree,
 
490
                                         new_path, new_entry, 
340
491
                                         new_tree, to_file)