← Back to branch summary

~bzr-pqm/bzr/bzr.dev

« back to all changes in this revision

Viewing changes to bzrlib/diff.py

  • Committer: Aaron Bentley
  • Date: 2006-08-25 21:49:47 UTC
  • mfrom: (1910.8.1 format-bumps)
  • mto: This revision was merged to the branch mainline in revision 1997.
  • Revision ID: aaron.bentley@utoronto.ca-20060825214947-62b9eea88621dc01
Merge from other branch

Show diffs side-by-side

added added

removed removed

Lines of Context:
bzrlib/diff.py
1
 
# -*- coding: UTF-8 -*-
2
 
 
 
1
# Copyright (C) 2004, 2005, 2006 Canonical Ltd.
 
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
 
from bzrlib.trace import mutter
 
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
 
27
from bzrlib.delta import compare_trees
18
28
from bzrlib.errors import BzrError
19
 
from bzrlib.delta import compare_trees
 
29
import bzrlib.errors as errors
 
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
 
20
38
 
21
39
# TODO: Rather than building a changeset object, we should probably
22
40
# invoke callbacks on an object.  That object can either accumulate a
23
41
# list, write them out directly, etc etc.
24
42
 
25
 
def internal_diff(old_label, oldlines, new_label, newlines, to_file):
26
 
    import difflib
27
 
    
 
43
def internal_diff(old_filename, oldlines, new_filename, newlines, to_file,
 
44
                  allow_binary=False, sequence_matcher=None,
 
45
                  path_encoding='utf8'):
28
46
    # FIXME: difflib is wrong if there is no trailing newline.
29
47
    # The syntax used by patch seems to be "\ No newline at
30
48
    # end of file" following the last diff line from that
40
58
    # both sequences are empty.
41
59
    if not oldlines and not newlines:
42
60
        return
 
61
    
 
62
    if allow_binary is False:
 
63
        check_text_lines(oldlines)
 
64
        check_text_lines(newlines)
43
65
 
44
 
    ud = difflib.unified_diff(oldlines, newlines,
45
 
                              fromfile=old_label, tofile=new_label)
 
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)
46
72
 
47
73
    ud = list(ud)
48
74
    # work-around for difflib being too smart for its own good
62
88
    print >>to_file
63
89
 
64
90
 
65
 
def external_diff(old_label, oldlines, new_label, newlines, to_file,
 
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
 
 
126
def external_diff(old_filename, oldlines, new_filename, newlines, to_file,
66
127
                  diff_opts):
67
128
    """Display a diff by calling out to the external diff program."""
68
 
    import sys
69
 
    
70
 
    if to_file != sys.stdout:
71
 
        raise NotImplementedError("sorry, can't send external diff other than to stdout yet",
72
 
                                  to_file)
73
 
 
74
129
    # make sure our own output is properly ordered before the diff
75
130
    to_file.flush()
76
131
 
77
 
    from tempfile import NamedTemporaryFile
78
 
    import os
79
 
 
80
 
    oldtmpf = NamedTemporaryFile()
81
 
    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')
82
136
 
83
137
    try:
84
138
        # TODO: perhaps a special case for comparing to or from the empty
91
145
        oldtmpf.writelines(oldlines)
92
146
        newtmpf.writelines(newlines)
93
147
 
94
 
        oldtmpf.flush()
95
 
        newtmpf.flush()
 
148
        oldtmpf.close()
 
149
        newtmpf.close()
96
150
 
97
151
        if not diff_opts:
98
152
            diff_opts = []
99
153
        diffcmd = ['diff',
100
 
                   '--label', old_label,
101
 
                   oldtmpf.name,
102
 
                   '--label', new_label,
103
 
                   newtmpf.name]
 
154
                   '--label', old_filename,
 
155
                   old_abspath,
 
156
                   '--label', new_filename,
 
157
                   new_abspath,
 
158
                   '--binary',
 
159
                  ]
104
160
 
105
161
        # diff only allows one style to be specified; they don't override.
106
162
        # note that some of these take optargs, and the optargs can be
126
182
        if diff_opts:
127
183
            diffcmd.extend(diff_opts)
128
184
 
129
 
        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
130
188
        
131
 
        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
            m = re.match('^binary files.*differ$', first_line, re.I)
 
211
            if m is None:
 
212
                raise BzrError('external diff failed with exit code 2;'
 
213
                               ' command: %r' % (diffcmd,))
 
214
            else:
 
215
                # Binary files differ, just return
 
216
                return
 
217
 
 
218
        # If we got to here, we haven't written out the output of diff
 
219
        # do so now
 
220
        to_file.write(out)
 
221
        if rc not in (0, 1):
132
222
            # returns 1 if files differ; that's OK
133
223
            if rc < 0:
134
224
                msg = 'signal %d' % (-rc)
135
225
            else:
136
226
                msg = 'exit code %d' % rc
137
227
                
138
 
            raise BzrError('external diff failed with %s; command: %r' % (rc, diffcmd))
 
228
            raise BzrError('external diff failed with %s; command: %r' 
 
229
                           % (rc, diffcmd))
 
230
 
 
231
 
139
232
    finally:
140
233
        oldtmpf.close()                 # and delete
141
234
        newtmpf.close()
142
 
 
 
235
        # Clean up. Warn in case the files couldn't be deleted
 
236
        # (in case windows still holds the file open, but not
 
237
        # if the files have already been deleted)
 
238
        try:
 
239
            os.remove(old_abspath)
 
240
        except OSError, e:
 
241
            if e.errno not in (errno.ENOENT,):
 
242
                warning('Failed to delete temporary file: %s %s',
 
243
                        old_abspath, e)
 
244
        try:
 
245
            os.remove(new_abspath)
 
246
        except OSError:
 
247
            if e.errno not in (errno.ENOENT,):
 
248
                warning('Failed to delete temporary file: %s %s',
 
249
                        new_abspath, e)
 
250
 
 
251
 
 
252
@deprecated_function(zero_eight)
143
253
def show_diff(b, from_spec, specific_files, external_diff_options=None,
144
 
              revision2=None, output=None):
 
254
              revision2=None, output=None, b2=None):
145
255
    """Shortcut for showing the diff to the working tree.
146
256
 
 
257
    Please use show_diff_trees instead.
 
258
 
147
259
    b
148
260
        Branch.
149
261
 
154
266
    supplies any two trees.
155
267
    """
156
268
    if output is None:
157
 
        import sys
158
269
        output = sys.stdout
159
270
 
160
271
    if from_spec is None:
161
 
        old_tree = b.basis_tree()
 
272
        old_tree = b.bzrdir.open_workingtree()
 
273
        if b2 is None:
 
274
            old_tree = old_tree = old_tree.basis_tree()
162
275
    else:
163
 
        old_tree = b.revision_tree(from_spec.in_history(b).rev_id)
 
276
        old_tree = b.repository.revision_tree(from_spec.in_history(b).rev_id)
164
277
 
165
278
    if revision2 is None:
166
 
        new_tree = b.working_tree()
167
 
    else:
168
 
        new_tree = b.revision_tree(revision2.in_history(b).rev_id)
169
 
 
170
 
    show_diff_trees(old_tree, new_tree, output, specific_files,
171
 
                    external_diff_options)
172
 
 
 
279
        if b2 is None:
 
280
            new_tree = b.bzrdir.open_workingtree()
 
281
        else:
 
282
            new_tree = b2.bzrdir.open_workingtree()
 
283
    else:
 
284
        new_tree = b.repository.revision_tree(revision2.in_history(b).rev_id)
 
285
 
 
286
    return show_diff_trees(old_tree, new_tree, output, specific_files,
 
287
                           external_diff_options)
 
288
 
 
289
 
 
290
def diff_cmd_helper(tree, specific_files, external_diff_options, 
 
291
                    old_revision_spec=None, new_revision_spec=None,
 
292
                    old_label='a/', new_label='b/'):
 
293
    """Helper for cmd_diff.
 
294
 
 
295
   tree 
 
296
        A WorkingTree
 
297
 
 
298
    specific_files
 
299
        The specific files to compare, or None
 
300
 
 
301
    external_diff_options
 
302
        If non-None, run an external diff, and pass it these options
 
303
 
 
304
    old_revision_spec
 
305
        If None, use basis tree as old revision, otherwise use the tree for
 
306
        the specified revision. 
 
307
 
 
308
    new_revision_spec
 
309
        If None, use working tree as new revision, otherwise use the tree for
 
310
        the specified revision.
 
311
    
 
312
    The more general form is show_diff_trees(), where the caller
 
313
    supplies any two trees.
 
314
    """
 
315
    def spec_tree(spec):
 
316
        if tree:
 
317
            revision = spec.in_store(tree.branch)
 
318
        else:
 
319
            revision = spec.in_store(None)
 
320
        revision_id = revision.rev_id
 
321
        branch = revision.branch
 
322
        return branch.repository.revision_tree(revision_id)
 
323
    if old_revision_spec is None:
 
324
        old_tree = tree.basis_tree()
 
325
    else:
 
326
        old_tree = spec_tree(old_revision_spec)
 
327
 
 
328
    if new_revision_spec is None:
 
329
        new_tree = tree
 
330
    else:
 
331
        new_tree = spec_tree(new_revision_spec)
 
332
    if new_tree is not tree:
 
333
        extra_trees = (tree,)
 
334
    else:
 
335
        extra_trees = None
 
336
 
 
337
    return show_diff_trees(old_tree, new_tree, sys.stdout, specific_files,
 
338
                           external_diff_options,
 
339
                           old_label=old_label, new_label=new_label,
 
340
                           extra_trees=extra_trees)
173
341
 
174
342
 
175
343
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None,
176
 
                    external_diff_options=None):
 
344
                    external_diff_options=None,
 
345
                    old_label='a/', new_label='b/',
 
346
                    extra_trees=None):
177
347
    """Show in text form the changes from one tree to another.
178
348
 
179
349
    to_files
181
351
 
182
352
    external_diff_options
183
353
        If set, use an external GNU diff and pass these options.
 
354
 
 
355
    extra_trees
 
356
        If set, more Trees to use for looking up file ids
184
357
    """
185
 
 
186
 
    # TODO: Options to control putting on a prefix or suffix, perhaps as a format string
187
 
    old_label = ''
188
 
    new_label = ''
189
 
 
190
 
    DEVNULL = '/dev/null'
191
 
    # Windows users, don't panic about this filename -- it is a
192
 
    # special signal to GNU patch that the file should be created or
193
 
    # deleted respectively.
 
358
    old_tree.lock_read()
 
359
    try:
 
360
        new_tree.lock_read()
 
361
        try:
 
362
            return _show_diff_trees(old_tree, new_tree, to_file,
 
363
                                    specific_files, external_diff_options,
 
364
                                    old_label=old_label, new_label=new_label,
 
365
                                    extra_trees=extra_trees)
 
366
        finally:
 
367
            new_tree.unlock()
 
368
    finally:
 
369
        old_tree.unlock()
 
370
 
 
371
 
 
372
def _show_diff_trees(old_tree, new_tree, to_file,
 
373
                     specific_files, external_diff_options, 
 
374
                     old_label='a/', new_label='b/', extra_trees=None):
 
375
 
 
376
    # GNU Patch uses the epoch date to detect files that are being added
 
377
    # or removed in a diff.
 
378
    EPOCH_DATE = '1970-01-01 00:00:00 +0000'
194
379
 
195
380
    # TODO: Generation of pseudo-diffs for added/deleted files could
196
381
    # be usefully made into a much faster special case.
203
388
    else:
204
389
        diff_file = internal_diff
205
390
    
206
 
 
207
 
    delta = compare_trees(old_tree, new_tree, want_unchanged=False,
208
 
                          specific_files=specific_files)
209
 
 
 
391
    delta = new_tree.changes_from(old_tree,
 
392
        specific_files=specific_files,
 
393
        extra_trees=extra_trees, require_versioned=True)
 
394
 
 
395
    has_changes = 0
210
396
    for path, file_id, kind in delta.removed:
211
 
        print >>to_file, '=== removed %s %r' % (kind, path)
212
 
        old_tree.inventory[file_id].diff(diff_file, old_label + path, old_tree,
213
 
                                         DEVNULL, None, None, to_file)
 
397
        has_changes = 1
 
398
        print >>to_file, '=== removed %s %r' % (kind, path.encode('utf8'))
 
399
        old_name = '%s%s\t%s' % (old_label, path,
 
400
                                 _patch_header_date(old_tree, file_id, path))
 
401
        new_name = '%s%s\t%s' % (new_label, path, EPOCH_DATE)
 
402
        old_tree.inventory[file_id].diff(diff_file, old_name, old_tree,
 
403
                                         new_name, None, None, to_file)
214
404
    for path, file_id, kind in delta.added:
215
 
        print >>to_file, '=== added %s %r' % (kind, path)
216
 
        new_tree.inventory[file_id].diff(diff_file, new_label + path, new_tree,
217
 
                                         DEVNULL, None, None, to_file, 
 
405
        has_changes = 1
 
406
        print >>to_file, '=== added %s %r' % (kind, path.encode('utf8'))
 
407
        old_name = '%s%s\t%s' % (old_label, path, EPOCH_DATE)
 
408
        new_name = '%s%s\t%s' % (new_label, path,
 
409
                                 _patch_header_date(new_tree, file_id, path))
 
410
        new_tree.inventory[file_id].diff(diff_file, new_name, new_tree,
 
411
                                         old_name, None, None, to_file, 
218
412
                                         reverse=True)
219
413
    for (old_path, new_path, file_id, kind,
220
414
         text_modified, meta_modified) in delta.renamed:
 
415
        has_changes = 1
221
416
        prop_str = get_prop_change(meta_modified)
222
417
        print >>to_file, '=== renamed %s %r => %r%s' % (
223
 
                          kind, old_path, new_path, prop_str)
224
 
        _maybe_diff_file_or_symlink(old_label, old_path, old_tree, file_id,
225
 
                                    new_label, new_path, new_tree,
 
418
                    kind, old_path.encode('utf8'),
 
419
                    new_path.encode('utf8'), prop_str)
 
420
        old_name = '%s%s\t%s' % (old_label, old_path,
 
421
                                 _patch_header_date(old_tree, file_id,
 
422
                                                    old_path))
 
423
        new_name = '%s%s\t%s' % (new_label, new_path,
 
424
                                 _patch_header_date(new_tree, file_id,
 
425
                                                    new_path))
 
426
        _maybe_diff_file_or_symlink(old_name, old_tree, file_id,
 
427
                                    new_name, new_tree,
226
428
                                    text_modified, kind, to_file, diff_file)
227
429
    for path, file_id, kind, text_modified, meta_modified in delta.modified:
 
430
        has_changes = 1
228
431
        prop_str = get_prop_change(meta_modified)
229
 
        print >>to_file, '=== modified %s %r%s' % (kind, path, prop_str)
 
432
        print >>to_file, '=== modified %s %r%s' % (kind, path.encode('utf8'), prop_str)
 
433
        old_name = '%s%s\t%s' % (old_label, path,
 
434
                                 _patch_header_date(old_tree, file_id, path))
 
435
        new_name = '%s%s\t%s' % (new_label, path,
 
436
                                 _patch_header_date(new_tree, file_id, path))
230
437
        if text_modified:
231
 
            _maybe_diff_file_or_symlink(old_label, path, old_tree, file_id,
232
 
                                        new_label, path, new_tree,
 
438
            _maybe_diff_file_or_symlink(old_name, old_tree, file_id,
 
439
                                        new_name, new_tree,
233
440
                                        True, kind, to_file, diff_file)
 
441
 
 
442
    return has_changes
 
443
 
 
444
 
 
445
def _patch_header_date(tree, file_id, path):
 
446
    """Returns a timestamp suitable for use in a patch header."""
 
447
    tm = time.gmtime(tree.get_file_mtime(file_id, path))
 
448
    return time.strftime('%Y-%m-%d %H:%M:%S +0000', tm)
 
449
 
 
450
 
 
451
def _raise_if_nonexistent(paths, old_tree, new_tree):
 
452
    """Complain if paths are not in either inventory or tree.
 
453
 
 
454
    It's OK with the files exist in either tree's inventory, or 
 
455
    if they exist in the tree but are not versioned.
234
456
    
 
457
    This can be used by operations such as bzr status that can accept
 
458
    unknown or ignored files.
 
459
    """
 
460
    mutter("check paths: %r", paths)
 
461
    if not paths:
 
462
        return
 
463
    s = old_tree.filter_unversioned_files(paths)
 
464
    s = new_tree.filter_unversioned_files(s)
 
465
    s = [path for path in s if not new_tree.has_filename(path)]
 
466
    if s:
 
467
        raise errors.PathsDoNotExist(sorted(s))
 
468
 
235
469
 
236
470
def get_prop_change(meta_modified):
237
471
    if meta_modified:
240
474
        return  ""
241
475
 
242
476
 
243
 
def _maybe_diff_file_or_symlink(old_label, old_path, old_tree, file_id,
244
 
                                new_label, new_path, new_tree, text_modified,
 
477
def _maybe_diff_file_or_symlink(old_path, old_tree, file_id,
 
478
                                new_path, new_tree, text_modified,
245
479
                                kind, to_file, diff_file):
246
480
    if text_modified:
247
481
        new_entry = new_tree.inventory[file_id]
248
482
        old_tree.inventory[file_id].diff(diff_file,
249
 
                                         old_label + old_path, old_tree,
250
 
                                         new_label + new_path, new_entry, 
 
483
                                         old_path, old_tree,
 
484
                                         new_path, new_entry, 
251
485
                                         new_tree, to_file)