← 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: 2006-11-10 01:55:55 UTC
  • mto: This revision was merged to the branch mainline in revision 2127.
  • Revision ID: aaron.bentley@utoronto.ca-20061110015555-f48202744b630209
Ignore html docs (both kinds)

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
18
 
from bzrlib.errors import BzrError
 
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
19
38
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
 
20
45
 
21
46
# TODO: Rather than building a changeset object, we should probably
22
47
# invoke callbacks on an object.  That object can either accumulate a
23
48
# list, write them out directly, etc etc.
24
49
 
25
 
def internal_diff(old_label, oldlines, new_label, newlines, to_file):
26
 
    import difflib
27
 
    
 
50
def internal_diff(old_filename, oldlines, new_filename, newlines, to_file,
 
51
                  allow_binary=False, sequence_matcher=None,
 
52
                  path_encoding='utf8'):
28
53
    # FIXME: difflib is wrong if there is no trailing newline.
29
54
    # The syntax used by patch seems to be "\ No newline at
30
55
    # end of file" following the last diff line from that
40
65
    # both sequences are empty.
41
66
    if not oldlines and not newlines:
42
67
        return
 
68
    
 
69
    if allow_binary is False:
 
70
        textfile.check_text_lines(oldlines)
 
71
        textfile.check_text_lines(newlines)
43
72
 
44
 
    ud = difflib.unified_diff(oldlines, newlines,
45
 
                              fromfile=old_label, tofile=new_label)
 
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)
46
79
 
47
80
    ud = list(ud)
48
81
    # work-around for difflib being too smart for its own good
62
95
    print >>to_file
63
96
 
64
97
 
65
 
def external_diff(old_label, oldlines, new_label, newlines, to_file,
 
98
def _set_lang_C():
 
99
    """Set the env var LANG=C"""
 
100
    osutils.set_or_unset_env('LANG', 'C')
 
101
    osutils.set_or_unset_env('LC_ALL', None)
 
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
        This lets us read and understand the output of diff, and respond 
 
112
        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 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
def external_diff(old_filename, oldlines, new_filename, newlines, to_file,
66
142
                  diff_opts):
67
143
    """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
144
    # make sure our own output is properly ordered before the diff
75
145
    to_file.flush()
76
146
 
77
 
    from tempfile import NamedTemporaryFile
78
 
    import os
79
 
 
80
 
    oldtmpf = NamedTemporaryFile()
81
 
    newtmpf = NamedTemporaryFile()
 
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')
82
151
 
83
152
    try:
84
153
        # TODO: perhaps a special case for comparing to or from the empty
91
160
        oldtmpf.writelines(oldlines)
92
161
        newtmpf.writelines(newlines)
93
162
 
94
 
        oldtmpf.flush()
95
 
        newtmpf.flush()
 
163
        oldtmpf.close()
 
164
        newtmpf.close()
96
165
 
97
166
        if not diff_opts:
98
167
            diff_opts = []
99
168
        diffcmd = ['diff',
100
 
                   '--label', old_label,
101
 
                   oldtmpf.name,
102
 
                   '--label', new_label,
103
 
                   newtmpf.name]
 
169
                   '--label', old_filename,
 
170
                   old_abspath,
 
171
                   '--label', new_filename,
 
172
                   new_abspath,
 
173
                   '--binary',
 
174
                  ]
104
175
 
105
176
        # diff only allows one style to be specified; they don't override.
106
177
        # note that some of these take optargs, and the optargs can be
126
197
        if diff_opts:
127
198
            diffcmd.extend(diff_opts)
128
199
 
129
 
        rc = os.spawnvp(os.P_WAIT, 'diff', diffcmd)
 
200
        pipe = _spawn_external_diff(diffcmd, capture_errors=True)
 
201
        out,err = pipe.communicate()
 
202
        rc = pipe.returncode
130
203
        
131
 
        if rc != 0 and rc != 1:
 
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, but not when run'
 
223
                               ' 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):
132
239
            # returns 1 if files differ; that's OK
133
240
            if rc < 0:
134
241
                msg = 'signal %d' % (-rc)
135
242
            else:
136
243
                msg = 'exit code %d' % rc
137
244
                
138
 
            raise BzrError('external diff failed with %s; command: %r' % (rc, diffcmd))
 
245
            raise errors.BzrError('external diff failed with %s; command: %r' 
 
246
                                  % (rc, diffcmd))
 
247
 
 
248
 
139
249
    finally:
140
250
        oldtmpf.close()                 # and delete
141
251
        newtmpf.close()
142
 
 
 
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
 
 
268
 
 
269
@deprecated_function(zero_eight)
143
270
def show_diff(b, from_spec, specific_files, external_diff_options=None,
144
 
              revision2=None, output=None):
 
271
              revision2=None, output=None, b2=None):
145
272
    """Shortcut for showing the diff to the working tree.
146
273
 
 
274
    Please use show_diff_trees instead.
 
275
 
147
276
    b
148
277
        Branch.
149
278
 
154
283
    supplies any two trees.
155
284
    """
156
285
    if output is None:
157
 
        import sys
158
286
        output = sys.stdout
159
287
 
160
288
    if from_spec is None:
161
 
        old_tree = b.basis_tree()
 
289
        old_tree = b.bzrdir.open_workingtree()
 
290
        if b2 is None:
 
291
            old_tree = old_tree = old_tree.basis_tree()
162
292
    else:
163
 
        old_tree = b.revision_tree(from_spec.in_history(b).rev_id)
 
293
        old_tree = b.repository.revision_tree(from_spec.in_history(b).rev_id)
164
294
 
165
295
    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
 
 
 
296
        if b2 is None:
 
297
            new_tree = b.bzrdir.open_workingtree()
 
298
        else:
 
299
            new_tree = b2.bzrdir.open_workingtree()
 
300
    else:
 
301
        new_tree = b.repository.revision_tree(revision2.in_history(b).rev_id)
 
302
 
 
303
    return show_diff_trees(old_tree, new_tree, output, specific_files,
 
304
                           external_diff_options)
 
305
 
 
306
 
 
307
def diff_cmd_helper(tree, specific_files, external_diff_options, 
 
308
                    old_revision_spec=None, new_revision_spec=None,
 
309
                    old_label='a/', new_label='b/'):
 
310
    """Helper for cmd_diff.
 
311
 
 
312
   tree 
 
313
        A WorkingTree
 
314
 
 
315
    specific_files
 
316
        The specific files to compare, or None
 
317
 
 
318
    external_diff_options
 
319
        If non-None, run an external diff, and pass it these options
 
320
 
 
321
    old_revision_spec
 
322
        If None, use basis tree as old revision, otherwise use the tree for
 
323
        the specified revision. 
 
324
 
 
325
    new_revision_spec
 
326
        If None, use working tree as new revision, otherwise use the tree for
 
327
        the specified revision.
 
328
    
 
329
    The more general form is show_diff_trees(), where the caller
 
330
    supplies any two trees.
 
331
    """
 
332
    def spec_tree(spec):
 
333
        if tree:
 
334
            revision = spec.in_store(tree.branch)
 
335
        else:
 
336
            revision = spec.in_store(None)
 
337
        revision_id = revision.rev_id
 
338
        branch = revision.branch
 
339
        return branch.repository.revision_tree(revision_id)
 
340
    if old_revision_spec is None:
 
341
        old_tree = tree.basis_tree()
 
342
    else:
 
343
        old_tree = spec_tree(old_revision_spec)
 
344
 
 
345
    if new_revision_spec is None:
 
346
        new_tree = tree
 
347
    else:
 
348
        new_tree = spec_tree(new_revision_spec)
 
349
    if new_tree is not tree:
 
350
        extra_trees = (tree,)
 
351
    else:
 
352
        extra_trees = None
 
353
 
 
354
    return show_diff_trees(old_tree, new_tree, sys.stdout, specific_files,
 
355
                           external_diff_options,
 
356
                           old_label=old_label, new_label=new_label,
 
357
                           extra_trees=extra_trees)
173
358
 
174
359
 
175
360
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None,
176
 
                    external_diff_options=None):
 
361
                    external_diff_options=None,
 
362
                    old_label='a/', new_label='b/',
 
363
                    extra_trees=None):
177
364
    """Show in text form the changes from one tree to another.
178
365
 
179
366
    to_files
181
368
 
182
369
    external_diff_options
183
370
        If set, use an external GNU diff and pass these options.
 
371
 
 
372
    extra_trees
 
373
        If set, more Trees to use for looking up file ids
184
374
    """
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.
 
375
    old_tree.lock_read()
 
376
    try:
 
377
        new_tree.lock_read()
 
378
        try:
 
379
            return _show_diff_trees(old_tree, new_tree, to_file,
 
380
                                    specific_files, external_diff_options,
 
381
                                    old_label=old_label, new_label=new_label,
 
382
                                    extra_trees=extra_trees)
 
383
        finally:
 
384
            new_tree.unlock()
 
385
    finally:
 
386
        old_tree.unlock()
 
387
 
 
388
 
 
389
def _show_diff_trees(old_tree, new_tree, to_file,
 
390
                     specific_files, external_diff_options, 
 
391
                     old_label='a/', new_label='b/', extra_trees=None):
 
392
 
 
393
    # GNU Patch uses the epoch date to detect files that are being added
 
394
    # or removed in a diff.
 
395
    EPOCH_DATE = '1970-01-01 00:00:00 +0000'
194
396
 
195
397
    # TODO: Generation of pseudo-diffs for added/deleted files could
196
398
    # be usefully made into a much faster special case.
203
405
    else:
204
406
        diff_file = internal_diff
205
407
    
206
 
 
207
 
    delta = compare_trees(old_tree, new_tree, want_unchanged=False,
208
 
                          specific_files=specific_files)
209
 
 
 
408
    delta = new_tree.changes_from(old_tree,
 
409
        specific_files=specific_files,
 
410
        extra_trees=extra_trees, require_versioned=True)
 
411
 
 
412
    has_changes = 0
210
413
    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)
 
414
        has_changes = 1
 
415
        print >>to_file, '=== removed %s %r' % (kind, path.encode('utf8'))
 
416
        old_name = '%s%s\t%s' % (old_label, path,
 
417
                                 _patch_header_date(old_tree, file_id, path))
 
418
        new_name = '%s%s\t%s' % (new_label, path, EPOCH_DATE)
 
419
        old_tree.inventory[file_id].diff(diff_file, old_name, old_tree,
 
420
                                         new_name, None, None, to_file)
214
421
    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, 
 
422
        has_changes = 1
 
423
        print >>to_file, '=== added %s %r' % (kind, path.encode('utf8'))
 
424
        old_name = '%s%s\t%s' % (old_label, path, EPOCH_DATE)
 
425
        new_name = '%s%s\t%s' % (new_label, path,
 
426
                                 _patch_header_date(new_tree, file_id, path))
 
427
        new_tree.inventory[file_id].diff(diff_file, new_name, new_tree,
 
428
                                         old_name, None, None, to_file, 
218
429
                                         reverse=True)
219
430
    for (old_path, new_path, file_id, kind,
220
431
         text_modified, meta_modified) in delta.renamed:
 
432
        has_changes = 1
221
433
        prop_str = get_prop_change(meta_modified)
222
434
        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,
 
435
                    kind, old_path.encode('utf8'),
 
436
                    new_path.encode('utf8'), prop_str)
 
437
        old_name = '%s%s\t%s' % (old_label, old_path,
 
438
                                 _patch_header_date(old_tree, file_id,
 
439
                                                    old_path))
 
440
        new_name = '%s%s\t%s' % (new_label, new_path,
 
441
                                 _patch_header_date(new_tree, file_id,
 
442
                                                    new_path))
 
443
        _maybe_diff_file_or_symlink(old_name, old_tree, file_id,
 
444
                                    new_name, new_tree,
226
445
                                    text_modified, kind, to_file, diff_file)
227
446
    for path, file_id, kind, text_modified, meta_modified in delta.modified:
 
447
        has_changes = 1
228
448
        prop_str = get_prop_change(meta_modified)
229
 
        print >>to_file, '=== modified %s %r%s' % (kind, path, prop_str)
 
449
        print >>to_file, '=== modified %s %r%s' % (kind, path.encode('utf8'), prop_str)
 
450
        old_name = '%s%s\t%s' % (old_label, path,
 
451
                                 _patch_header_date(old_tree, file_id, path))
 
452
        new_name = '%s%s\t%s' % (new_label, path,
 
453
                                 _patch_header_date(new_tree, file_id, path))
230
454
        if text_modified:
231
 
            _maybe_diff_file_or_symlink(old_label, path, old_tree, file_id,
232
 
                                        new_label, path, new_tree,
 
455
            _maybe_diff_file_or_symlink(old_name, old_tree, file_id,
 
456
                                        new_name, new_tree,
233
457
                                        True, kind, to_file, diff_file)
 
458
 
 
459
    return has_changes
 
460
 
 
461
 
 
462
def _patch_header_date(tree, file_id, path):
 
463
    """Returns a timestamp suitable for use in a patch header."""
 
464
    tm = time.gmtime(tree.get_file_mtime(file_id, path))
 
465
    return time.strftime('%Y-%m-%d %H:%M:%S +0000', tm)
 
466
 
 
467
 
 
468
def _raise_if_nonexistent(paths, old_tree, new_tree):
 
469
    """Complain if paths are not in either inventory or tree.
 
470
 
 
471
    It's OK with the files exist in either tree's inventory, or 
 
472
    if they exist in the tree but are not versioned.
234
473
    
 
474
    This can be used by operations such as bzr status that can accept
 
475
    unknown or ignored files.
 
476
    """
 
477
    mutter("check paths: %r", paths)
 
478
    if not paths:
 
479
        return
 
480
    s = old_tree.filter_unversioned_files(paths)
 
481
    s = new_tree.filter_unversioned_files(s)
 
482
    s = [path for path in s if not new_tree.has_filename(path)]
 
483
    if s:
 
484
        raise errors.PathsDoNotExist(sorted(s))
 
485
 
235
486
 
236
487
def get_prop_change(meta_modified):
237
488
    if meta_modified:
240
491
        return  ""
241
492
 
242
493
 
243
 
def _maybe_diff_file_or_symlink(old_label, old_path, old_tree, file_id,
244
 
                                new_label, new_path, new_tree, text_modified,
 
494
def _maybe_diff_file_or_symlink(old_path, old_tree, file_id,
 
495
                                new_path, new_tree, text_modified,
245
496
                                kind, to_file, diff_file):
246
497
    if text_modified:
247
498
        new_entry = new_tree.inventory[file_id]
248
499
        old_tree.inventory[file_id].diff(diff_file,
249
 
                                         old_label + old_path, old_tree,
250
 
                                         new_label + new_path, new_entry, 
 
500
                                         old_path, old_tree,
 
501
                                         new_path, new_entry, 
251
502
                                         new_tree, to_file)