/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: Robey Pointer
  • Date: 2006-09-08 18:46:29 UTC
  • mto: This revision was merged to the branch mainline in revision 1996.
  • Revision ID: robey@lag.net-20060908184629-e3fc4c61ca21508c
pychecker is on crack; go back to using 'is None'.

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
        preexec_fn = _set_lang_C
 
107
        stderr = subprocess.PIPE
 
108
    else:
 
109
        preexec_fn = None
 
110
        stderr = None
 
111
 
 
112
    try:
 
113
        pipe = subprocess.Popen(diffcmd,
 
114
                                stdin=subprocess.PIPE,
 
115
                                stdout=subprocess.PIPE,
 
116
                                stderr=stderr,
 
117
                                preexec_fn=preexec_fn)
 
118
    except OSError, e:
 
119
        if e.errno == errno.ENOENT:
 
120
            raise errors.NoDiff(str(e))
 
121
        raise
 
122
 
 
123
    return pipe
 
124
 
 
125
 
68
126
def external_diff(old_filename, oldlines, new_filename, newlines, to_file,
69
127
                  diff_opts):
70
128
    """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
129
    # make sure our own output is properly ordered before the diff
78
130
    to_file.flush()
79
131
 
80
 
    from tempfile import NamedTemporaryFile
81
 
    import os
82
 
 
83
 
    oldtmpf = NamedTemporaryFile()
84
 
    newtmpf = NamedTemporaryFile()
 
132
    oldtmp_fd, old_abspath = tempfile.mkstemp(prefix='bzr-diff-old-')
 
133
    newtmp_fd, new_abspath = tempfile.mkstemp(prefix='bzr-diff-new-')
 
134
    oldtmpf = os.fdopen(oldtmp_fd, 'wb')
 
135
    newtmpf = os.fdopen(newtmp_fd, 'wb')
85
136
 
86
137
    try:
87
138
        # TODO: perhaps a special case for comparing to or from the empty
94
145
        oldtmpf.writelines(oldlines)
95
146
        newtmpf.writelines(newlines)
96
147
 
97
 
        oldtmpf.flush()
98
 
        newtmpf.flush()
 
148
        oldtmpf.close()
 
149
        newtmpf.close()
99
150
 
100
151
        if not diff_opts:
101
152
            diff_opts = []
102
153
        diffcmd = ['diff',
103
 
                   '--label', old_filename+'\t',
104
 
                   oldtmpf.name,
105
 
                   '--label', new_filename+'\t',
106
 
                   newtmpf.name]
 
154
                   '--label', old_filename,
 
155
                   old_abspath,
 
156
                   '--label', new_filename,
 
157
                   new_abspath,
 
158
                   '--binary',
 
159
                  ]
107
160
 
108
161
        # diff only allows one style to be specified; they don't override.
109
162
        # note that some of these take optargs, and the optargs can be
129
182
        if diff_opts:
130
183
            diffcmd.extend(diff_opts)
131
184
 
132
 
        rc = os.spawnvp(os.P_WAIT, 'diff', diffcmd)
 
185
        pipe = _spawn_external_diff(diffcmd, capture_errors=True)
 
186
        out,err = pipe.communicate()
 
187
        rc = pipe.returncode
133
188
        
134
 
        if rc != 0 and rc != 1:
 
189
        # internal_diff() adds a trailing newline, add one here for consistency
 
190
        out += '\n'
 
191
        if rc == 2:
 
192
            # 'diff' gives retcode == 2 for all sorts of errors
 
193
            # one of those is 'Binary files differ'.
 
194
            # Bad options could also be the problem.
 
195
            # 'Binary files' is not a real error, so we suppress that error.
 
196
            lang_c_out = out
 
197
 
 
198
            # Since we got here, we want to make sure to give an i18n error
 
199
            pipe = _spawn_external_diff(diffcmd, capture_errors=False)
 
200
            out, err = pipe.communicate()
 
201
 
 
202
            # Write out the new i18n diff response
 
203
            to_file.write(out+'\n')
 
204
            if pipe.returncode != 2:
 
205
                raise BzrError('external diff failed with exit code 2'
 
206
                               ' when run with LANG=C, but not when run'
 
207
                               ' natively: %r' % (diffcmd,))
 
208
 
 
209
            first_line = lang_c_out.split('\n', 1)[0]
 
210
            # Starting with diffutils 2.8.4 the word "binary" was dropped.
 
211
            m = re.match('^(binary )?files.*differ$', first_line, re.I)
 
212
            if m is None:
 
213
                raise BzrError('external diff failed with exit code 2;'
 
214
                               ' command: %r' % (diffcmd,))
 
215
            else:
 
216
                # Binary files differ, just return
 
217
                return
 
218
 
 
219
        # If we got to here, we haven't written out the output of diff
 
220
        # do so now
 
221
        to_file.write(out)
 
222
        if rc not in (0, 1):
135
223
            # returns 1 if files differ; that's OK
136
224
            if rc < 0:
137
225
                msg = 'signal %d' % (-rc)
138
226
            else:
139
227
                msg = 'exit code %d' % rc
140
228
                
141
 
            raise BzrError('external diff failed with %s; command: %r' % (rc, diffcmd))
 
229
            raise BzrError('external diff failed with %s; command: %r' 
 
230
                           % (rc, diffcmd))
 
231
 
 
232
 
142
233
    finally:
143
234
        oldtmpf.close()                 # and delete
144
235
        newtmpf.close()
 
236
        # Clean up. Warn in case the files couldn't be deleted
 
237
        # (in case windows still holds the file open, but not
 
238
        # if the files have already been deleted)
 
239
        try:
 
240
            os.remove(old_abspath)
 
241
        except OSError, e:
 
242
            if e.errno not in (errno.ENOENT,):
 
243
                warning('Failed to delete temporary file: %s %s',
 
244
                        old_abspath, e)
 
245
        try:
 
246
            os.remove(new_abspath)
 
247
        except OSError:
 
248
            if e.errno not in (errno.ENOENT,):
 
249
                warning('Failed to delete temporary file: %s %s',
 
250
                        new_abspath, e)
145
251
 
146
252
 
147
253
@deprecated_function(zero_eight)
161
267
    supplies any two trees.
162
268
    """
163
269
    if output is None:
164
 
        import sys
165
270
        output = sys.stdout
166
271
 
167
272
    if from_spec is None:
184
289
 
185
290
 
186
291
def diff_cmd_helper(tree, specific_files, external_diff_options, 
187
 
                    old_revision_spec=None, new_revision_spec=None):
 
292
                    old_revision_spec=None, new_revision_spec=None,
 
293
                    old_label='a/', new_label='b/'):
188
294
    """Helper for cmd_diff.
189
295
 
190
296
   tree 
207
313
    The more general form is show_diff_trees(), where the caller
208
314
    supplies any two trees.
209
315
    """
210
 
    import sys
211
 
    output = sys.stdout
212
316
    def spec_tree(spec):
213
 
        revision_id = spec.in_store(tree.branch).rev_id
214
 
        return tree.branch.repository.revision_tree(revision_id)
 
317
        if tree:
 
318
            revision = spec.in_store(tree.branch)
 
319
        else:
 
320
            revision = spec.in_store(None)
 
321
        revision_id = revision.rev_id
 
322
        branch = revision.branch
 
323
        return branch.repository.revision_tree(revision_id)
215
324
    if old_revision_spec is None:
216
325
        old_tree = tree.basis_tree()
217
326
    else:
221
330
        new_tree = tree
222
331
    else:
223
332
        new_tree = spec_tree(new_revision_spec)
 
333
    if new_tree is not tree:
 
334
        extra_trees = (tree,)
 
335
    else:
 
336
        extra_trees = None
224
337
 
225
338
    return show_diff_trees(old_tree, new_tree, sys.stdout, specific_files,
226
 
                           external_diff_options)
 
339
                           external_diff_options,
 
340
                           old_label=old_label, new_label=new_label,
 
341
                           extra_trees=extra_trees)
227
342
 
228
343
 
229
344
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None,
230
 
                    external_diff_options=None):
 
345
                    external_diff_options=None,
 
346
                    old_label='a/', new_label='b/',
 
347
                    extra_trees=None):
231
348
    """Show in text form the changes from one tree to another.
232
349
 
233
350
    to_files
235
352
 
236
353
    external_diff_options
237
354
        If set, use an external GNU diff and pass these options.
 
355
 
 
356
    extra_trees
 
357
        If set, more Trees to use for looking up file ids
238
358
    """
239
359
    old_tree.lock_read()
240
360
    try:
241
361
        new_tree.lock_read()
242
362
        try:
243
363
            return _show_diff_trees(old_tree, new_tree, to_file,
244
 
                                    specific_files, external_diff_options)
 
364
                                    specific_files, external_diff_options,
 
365
                                    old_label=old_label, new_label=new_label,
 
366
                                    extra_trees=extra_trees)
245
367
        finally:
246
368
            new_tree.unlock()
247
369
    finally:
249
371
 
250
372
 
251
373
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.
 
374
                     specific_files, external_diff_options, 
 
375
                     old_label='a/', new_label='b/', extra_trees=None):
 
376
 
 
377
    # GNU Patch uses the epoch date to detect files that are being added
 
378
    # or removed in a diff.
 
379
    EPOCH_DATE = '1970-01-01 00:00:00 +0000'
263
380
 
264
381
    # TODO: Generation of pseudo-diffs for added/deleted files could
265
382
    # be usefully made into a much faster special case.
266
383
 
267
 
    _raise_if_doubly_unversioned(specific_files, old_tree, new_tree)
268
 
 
269
384
    if external_diff_options:
270
385
        assert isinstance(external_diff_options, basestring)
271
386
        opts = external_diff_options.split()
274
389
    else:
275
390
        diff_file = internal_diff
276
391
    
277
 
    delta = compare_trees(old_tree, new_tree, want_unchanged=False,
278
 
                          specific_files=specific_files)
 
392
    delta = new_tree.changes_from(old_tree,
 
393
        specific_files=specific_files,
 
394
        extra_trees=extra_trees, require_versioned=True)
279
395
 
280
396
    has_changes = 0
281
397
    for path, file_id, kind in delta.removed:
282
398
        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)
 
399
        print >>to_file, '=== removed %s %r' % (kind, path.encode('utf8'))
 
400
        old_name = '%s%s\t%s' % (old_label, path,
 
401
                                 _patch_header_date(old_tree, file_id, path))
 
402
        new_name = '%s%s\t%s' % (new_label, path, EPOCH_DATE)
 
403
        old_tree.inventory[file_id].diff(diff_file, old_name, old_tree,
 
404
                                         new_name, None, None, to_file)
286
405
    for path, file_id, kind in delta.added:
287
406
        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, 
 
407
        print >>to_file, '=== added %s %r' % (kind, path.encode('utf8'))
 
408
        old_name = '%s%s\t%s' % (old_label, path, EPOCH_DATE)
 
409
        new_name = '%s%s\t%s' % (new_label, path,
 
410
                                 _patch_header_date(new_tree, file_id, path))
 
411
        new_tree.inventory[file_id].diff(diff_file, new_name, new_tree,
 
412
                                         old_name, None, None, to_file, 
291
413
                                         reverse=True)
292
414
    for (old_path, new_path, file_id, kind,
293
415
         text_modified, meta_modified) in delta.renamed:
294
416
        has_changes = 1
295
417
        prop_str = get_prop_change(meta_modified)
296
418
        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,
 
419
                    kind, old_path.encode('utf8'),
 
420
                    new_path.encode('utf8'), prop_str)
 
421
        old_name = '%s%s\t%s' % (old_label, old_path,
 
422
                                 _patch_header_date(old_tree, file_id,
 
423
                                                    old_path))
 
424
        new_name = '%s%s\t%s' % (new_label, new_path,
 
425
                                 _patch_header_date(new_tree, file_id,
 
426
                                                    new_path))
 
427
        _maybe_diff_file_or_symlink(old_name, old_tree, file_id,
 
428
                                    new_name, new_tree,
300
429
                                    text_modified, kind, to_file, diff_file)
301
430
    for path, file_id, kind, text_modified, meta_modified in delta.modified:
302
431
        has_changes = 1
303
432
        prop_str = get_prop_change(meta_modified)
304
 
        print >>to_file, '=== modified %s %r%s' % (kind, old_label + path,
305
 
                    prop_str)
 
433
        print >>to_file, '=== modified %s %r%s' % (kind, path.encode('utf8'), prop_str)
 
434
        old_name = '%s%s\t%s' % (old_label, path,
 
435
                                 _patch_header_date(old_tree, file_id, path))
 
436
        new_name = '%s%s\t%s' % (new_label, path,
 
437
                                 _patch_header_date(new_tree, file_id, path))
306
438
        if text_modified:
307
 
            _maybe_diff_file_or_symlink(old_label, path, old_tree, file_id,
308
 
                                        new_label, path, new_tree,
 
439
            _maybe_diff_file_or_symlink(old_name, old_tree, file_id,
 
440
                                        new_name, new_tree,
309
441
                                        True, kind, to_file, diff_file)
310
442
 
311
443
    return has_changes
312
444
 
313
445
 
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:
 
446
def _patch_header_date(tree, file_id, path):
 
447
    """Returns a timestamp suitable for use in a patch header."""
 
448
    tm = time.gmtime(tree.get_file_mtime(file_id, path))
 
449
    return time.strftime('%Y-%m-%d %H:%M:%S +0000', tm)
 
450
 
 
451
 
 
452
def _raise_if_nonexistent(paths, old_tree, new_tree):
 
453
    """Complain if paths are not in either inventory or tree.
 
454
 
 
455
    It's OK with the files exist in either tree's inventory, or 
 
456
    if they exist in the tree but are not versioned.
 
457
    
 
458
    This can be used by operations such as bzr status that can accept
 
459
    unknown or ignored files.
 
460
    """
 
461
    mutter("check paths: %r", paths)
 
462
    if not paths:
317
463
        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
 
    
 
464
    s = old_tree.filter_unversioned_files(paths)
 
465
    s = new_tree.filter_unversioned_files(s)
 
466
    s = [path for path in s if not new_tree.has_filename(path)]
 
467
    if s:
 
468
        raise errors.PathsDoNotExist(sorted(s))
 
469
 
324
470
 
325
471
def get_prop_change(meta_modified):
326
472
    if meta_modified:
329
475
        return  ""
330
476
 
331
477
 
332
 
def _maybe_diff_file_or_symlink(old_label, old_path, old_tree, file_id,
333
 
                                new_label, new_path, new_tree, text_modified,
 
478
def _maybe_diff_file_or_symlink(old_path, old_tree, file_id,
 
479
                                new_path, new_tree, text_modified,
334
480
                                kind, to_file, diff_file):
335
481
    if text_modified:
336
482
        new_entry = new_tree.inventory[file_id]
337
483
        old_tree.inventory[file_id].diff(diff_file,
338
 
                                         old_label + old_path, old_tree,
339
 
                                         new_label + new_path, new_entry, 
 
484
                                         old_path, old_tree,
 
485
                                         new_path, new_entry, 
340
486
                                         new_tree, to_file)