~bzr-pqm/bzr/bzr.dev

« back to all changes in this revision

Viewing changes to bzrlib/diff.py

  • Committer: Aaron Bentley
  • Date: 2006-09-21 20:21:36 UTC
  • mto: (2027.1.2 revert-subpath-56549)
  • mto: This revision was merged to the branch mainline in revision 2031.
  • Revision ID: abentley@panoramicfeedback.com-20060921202136-e5a8deaadfa00021
Added test for preserving file mode

Show diffs side-by-side

added added

removed removed

Lines of Context:
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
from bzrlib import (
 
26
    errors,
 
27
    osutils,
 
28
    )
 
29
# compatability - plugins import compare_trees from diff!!!
 
30
# deprecated as of 0.10
 
31
from bzrlib.delta import compare_trees
18
32
from bzrlib.errors import BzrError
19
 
from bzrlib.delta import compare_trees
 
33
from bzrlib.patiencediff import unified_diff
 
34
import bzrlib.patiencediff
 
35
from bzrlib.symbol_versioning import (deprecated_function,
 
36
        zero_eight)
 
37
from bzrlib.textfile import check_text_lines
 
38
from bzrlib.trace import mutter, warning
 
39
 
20
40
 
21
41
# TODO: Rather than building a changeset object, we should probably
22
42
# invoke callbacks on an object.  That object can either accumulate a
23
43
# list, write them out directly, etc etc.
24
44
 
25
 
def internal_diff(old_filename, oldlines, new_filename, newlines, to_file):
26
 
    import difflib
27
 
    
 
45
def internal_diff(old_filename, oldlines, new_filename, newlines, to_file,
 
46
                  allow_binary=False, sequence_matcher=None,
 
47
                  path_encoding='utf8'):
28
48
    # FIXME: difflib is wrong if there is no trailing newline.
29
49
    # The syntax used by patch seems to be "\ No newline at
30
50
    # end of file" following the last diff line from that
40
60
    # both sequences are empty.
41
61
    if not oldlines and not newlines:
42
62
        return
 
63
    
 
64
    if allow_binary is False:
 
65
        check_text_lines(oldlines)
 
66
        check_text_lines(newlines)
43
67
 
44
 
    ud = difflib.unified_diff(oldlines, newlines,
45
 
                              fromfile=old_filename+'\t', 
46
 
                              tofile=new_filename+'\t')
 
68
    if sequence_matcher is None:
 
69
        sequence_matcher = bzrlib.patiencediff.PatienceSequenceMatcher
 
70
    ud = unified_diff(oldlines, newlines,
 
71
                      fromfile=old_filename.encode(path_encoding),
 
72
                      tofile=new_filename.encode(path_encoding),
 
73
                      sequencematcher=sequence_matcher)
47
74
 
48
75
    ud = list(ud)
49
76
    # work-around for difflib being too smart for its own good
63
90
    print >>to_file
64
91
 
65
92
 
 
93
def _set_lang_C():
 
94
    """Set the env var LANG=C"""
 
95
    osutils.set_or_unset_env('LANG', 'C')
 
96
    osutils.set_or_unset_env('LC_ALL', None)
 
97
    osutils.set_or_unset_env('LC_CTYPE', None)
 
98
    osutils.set_or_unset_env('LANGUAGE', None)
 
99
 
 
100
 
 
101
def _spawn_external_diff(diffcmd, capture_errors=True):
 
102
    """Spawn the externall diff process, and return the child handle.
 
103
 
 
104
    :param diffcmd: The command list to spawn
 
105
    :param capture_errors: Capture stderr as well as setting LANG=C.
 
106
        This lets us read and understand the output of diff, and respond 
 
107
        to any errors.
 
108
    :return: A Popen object.
 
109
    """
 
110
    if capture_errors:
 
111
        if sys.platform == 'win32':
 
112
            # Win32 doesn't support preexec_fn, but that is
 
113
            # okay, because it doesn't support LANG either.
 
114
            preexec_fn = None
 
115
        else:
 
116
            preexec_fn = _set_lang_C
 
117
        stderr = subprocess.PIPE
 
118
    else:
 
119
        preexec_fn = None
 
120
        stderr = None
 
121
 
 
122
    try:
 
123
        pipe = subprocess.Popen(diffcmd,
 
124
                                stdin=subprocess.PIPE,
 
125
                                stdout=subprocess.PIPE,
 
126
                                stderr=stderr,
 
127
                                preexec_fn=preexec_fn)
 
128
    except OSError, e:
 
129
        if e.errno == errno.ENOENT:
 
130
            raise errors.NoDiff(str(e))
 
131
        raise
 
132
 
 
133
    return pipe
 
134
 
 
135
 
66
136
def external_diff(old_filename, oldlines, new_filename, newlines, to_file,
67
137
                  diff_opts):
68
138
    """Display a diff by calling out to the external diff program."""
69
 
    import sys
70
 
    
71
 
    if to_file != sys.stdout:
72
 
        raise NotImplementedError("sorry, can't send external diff other than to stdout yet",
73
 
                                  to_file)
74
 
 
75
139
    # make sure our own output is properly ordered before the diff
76
140
    to_file.flush()
77
141
 
78
 
    from tempfile import NamedTemporaryFile
79
 
    import os
80
 
 
81
 
    oldtmpf = NamedTemporaryFile()
82
 
    newtmpf = NamedTemporaryFile()
 
142
    oldtmp_fd, old_abspath = tempfile.mkstemp(prefix='bzr-diff-old-')
 
143
    newtmp_fd, new_abspath = tempfile.mkstemp(prefix='bzr-diff-new-')
 
144
    oldtmpf = os.fdopen(oldtmp_fd, 'wb')
 
145
    newtmpf = os.fdopen(newtmp_fd, 'wb')
83
146
 
84
147
    try:
85
148
        # TODO: perhaps a special case for comparing to or from the empty
92
155
        oldtmpf.writelines(oldlines)
93
156
        newtmpf.writelines(newlines)
94
157
 
95
 
        oldtmpf.flush()
96
 
        newtmpf.flush()
 
158
        oldtmpf.close()
 
159
        newtmpf.close()
97
160
 
98
161
        if not diff_opts:
99
162
            diff_opts = []
100
163
        diffcmd = ['diff',
101
 
                   '--label', old_filename+'\t',
102
 
                   oldtmpf.name,
103
 
                   '--label', new_filename+'\t',
104
 
                   newtmpf.name]
 
164
                   '--label', old_filename,
 
165
                   old_abspath,
 
166
                   '--label', new_filename,
 
167
                   new_abspath,
 
168
                   '--binary',
 
169
                  ]
105
170
 
106
171
        # diff only allows one style to be specified; they don't override.
107
172
        # note that some of these take optargs, and the optargs can be
127
192
        if diff_opts:
128
193
            diffcmd.extend(diff_opts)
129
194
 
130
 
        rc = os.spawnvp(os.P_WAIT, 'diff', diffcmd)
 
195
        pipe = _spawn_external_diff(diffcmd, capture_errors=True)
 
196
        out,err = pipe.communicate()
 
197
        rc = pipe.returncode
131
198
        
132
 
        if rc != 0 and rc != 1:
 
199
        # internal_diff() adds a trailing newline, add one here for consistency
 
200
        out += '\n'
 
201
        if rc == 2:
 
202
            # 'diff' gives retcode == 2 for all sorts of errors
 
203
            # one of those is 'Binary files differ'.
 
204
            # Bad options could also be the problem.
 
205
            # 'Binary files' is not a real error, so we suppress that error.
 
206
            lang_c_out = out
 
207
 
 
208
            # Since we got here, we want to make sure to give an i18n error
 
209
            pipe = _spawn_external_diff(diffcmd, capture_errors=False)
 
210
            out, err = pipe.communicate()
 
211
 
 
212
            # Write out the new i18n diff response
 
213
            to_file.write(out+'\n')
 
214
            if pipe.returncode != 2:
 
215
                raise BzrError('external diff failed with exit code 2'
 
216
                               ' when run with LANG=C, but not when run'
 
217
                               ' natively: %r' % (diffcmd,))
 
218
 
 
219
            first_line = lang_c_out.split('\n', 1)[0]
 
220
            # Starting with diffutils 2.8.4 the word "binary" was dropped.
 
221
            m = re.match('^(binary )?files.*differ$', first_line, re.I)
 
222
            if m is None:
 
223
                raise BzrError('external diff failed with exit code 2;'
 
224
                               ' command: %r' % (diffcmd,))
 
225
            else:
 
226
                # Binary files differ, just return
 
227
                return
 
228
 
 
229
        # If we got to here, we haven't written out the output of diff
 
230
        # do so now
 
231
        to_file.write(out)
 
232
        if rc not in (0, 1):
133
233
            # returns 1 if files differ; that's OK
134
234
            if rc < 0:
135
235
                msg = 'signal %d' % (-rc)
136
236
            else:
137
237
                msg = 'exit code %d' % rc
138
238
                
139
 
            raise BzrError('external diff failed with %s; command: %r' % (rc, diffcmd))
 
239
            raise BzrError('external diff failed with %s; command: %r' 
 
240
                           % (rc, diffcmd))
 
241
 
 
242
 
140
243
    finally:
141
244
        oldtmpf.close()                 # and delete
142
245
        newtmpf.close()
143
 
 
 
246
        # Clean up. Warn in case the files couldn't be deleted
 
247
        # (in case windows still holds the file open, but not
 
248
        # if the files have already been deleted)
 
249
        try:
 
250
            os.remove(old_abspath)
 
251
        except OSError, e:
 
252
            if e.errno not in (errno.ENOENT,):
 
253
                warning('Failed to delete temporary file: %s %s',
 
254
                        old_abspath, e)
 
255
        try:
 
256
            os.remove(new_abspath)
 
257
        except OSError:
 
258
            if e.errno not in (errno.ENOENT,):
 
259
                warning('Failed to delete temporary file: %s %s',
 
260
                        new_abspath, e)
 
261
 
 
262
 
 
263
@deprecated_function(zero_eight)
144
264
def show_diff(b, from_spec, specific_files, external_diff_options=None,
145
265
              revision2=None, output=None, b2=None):
146
266
    """Shortcut for showing the diff to the working tree.
147
267
 
 
268
    Please use show_diff_trees instead.
 
269
 
148
270
    b
149
271
        Branch.
150
272
 
155
277
    supplies any two trees.
156
278
    """
157
279
    if output is None:
158
 
        import sys
159
280
        output = sys.stdout
160
281
 
161
282
    if from_spec is None:
 
283
        old_tree = b.bzrdir.open_workingtree()
162
284
        if b2 is None:
163
 
            old_tree = b.basis_tree()
164
 
        else:
165
 
            old_tree = b.working_tree()
 
285
            old_tree = old_tree = old_tree.basis_tree()
166
286
    else:
167
 
        old_tree = b.revision_tree(from_spec.in_history(b).rev_id)
 
287
        old_tree = b.repository.revision_tree(from_spec.in_history(b).rev_id)
168
288
 
169
289
    if revision2 is None:
170
290
        if b2 is None:
171
 
            new_tree = b.working_tree()
 
291
            new_tree = b.bzrdir.open_workingtree()
172
292
        else:
173
 
            new_tree = b2.working_tree()
 
293
            new_tree = b2.bzrdir.open_workingtree()
174
294
    else:
175
 
        new_tree = b.revision_tree(revision2.in_history(b).rev_id)
 
295
        new_tree = b.repository.revision_tree(revision2.in_history(b).rev_id)
176
296
 
177
297
    return show_diff_trees(old_tree, new_tree, output, specific_files,
178
298
                           external_diff_options)
179
299
 
180
300
 
 
301
def diff_cmd_helper(tree, specific_files, external_diff_options, 
 
302
                    old_revision_spec=None, new_revision_spec=None,
 
303
                    old_label='a/', new_label='b/'):
 
304
    """Helper for cmd_diff.
 
305
 
 
306
   tree 
 
307
        A WorkingTree
 
308
 
 
309
    specific_files
 
310
        The specific files to compare, or None
 
311
 
 
312
    external_diff_options
 
313
        If non-None, run an external diff, and pass it these options
 
314
 
 
315
    old_revision_spec
 
316
        If None, use basis tree as old revision, otherwise use the tree for
 
317
        the specified revision. 
 
318
 
 
319
    new_revision_spec
 
320
        If None, use working tree as new revision, otherwise use the tree for
 
321
        the specified revision.
 
322
    
 
323
    The more general form is show_diff_trees(), where the caller
 
324
    supplies any two trees.
 
325
    """
 
326
    def spec_tree(spec):
 
327
        if tree:
 
328
            revision = spec.in_store(tree.branch)
 
329
        else:
 
330
            revision = spec.in_store(None)
 
331
        revision_id = revision.rev_id
 
332
        branch = revision.branch
 
333
        return branch.repository.revision_tree(revision_id)
 
334
    if old_revision_spec is None:
 
335
        old_tree = tree.basis_tree()
 
336
    else:
 
337
        old_tree = spec_tree(old_revision_spec)
 
338
 
 
339
    if new_revision_spec is None:
 
340
        new_tree = tree
 
341
    else:
 
342
        new_tree = spec_tree(new_revision_spec)
 
343
    if new_tree is not tree:
 
344
        extra_trees = (tree,)
 
345
    else:
 
346
        extra_trees = None
 
347
 
 
348
    return show_diff_trees(old_tree, new_tree, sys.stdout, specific_files,
 
349
                           external_diff_options,
 
350
                           old_label=old_label, new_label=new_label,
 
351
                           extra_trees=extra_trees)
 
352
 
181
353
 
182
354
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None,
183
 
                    external_diff_options=None):
 
355
                    external_diff_options=None,
 
356
                    old_label='a/', new_label='b/',
 
357
                    extra_trees=None):
184
358
    """Show in text form the changes from one tree to another.
185
359
 
186
360
    to_files
188
362
 
189
363
    external_diff_options
190
364
        If set, use an external GNU diff and pass these options.
 
365
 
 
366
    extra_trees
 
367
        If set, more Trees to use for looking up file ids
191
368
    """
192
 
 
193
 
    # TODO: Options to control putting on a prefix or suffix, perhaps as a format string
194
 
    old_label = ''
195
 
    new_label = ''
196
 
 
197
 
    DEVNULL = '/dev/null'
198
 
    # Windows users, don't panic about this filename -- it is a
199
 
    # special signal to GNU patch that the file should be created or
200
 
    # deleted respectively.
 
369
    old_tree.lock_read()
 
370
    try:
 
371
        new_tree.lock_read()
 
372
        try:
 
373
            return _show_diff_trees(old_tree, new_tree, to_file,
 
374
                                    specific_files, external_diff_options,
 
375
                                    old_label=old_label, new_label=new_label,
 
376
                                    extra_trees=extra_trees)
 
377
        finally:
 
378
            new_tree.unlock()
 
379
    finally:
 
380
        old_tree.unlock()
 
381
 
 
382
 
 
383
def _show_diff_trees(old_tree, new_tree, to_file,
 
384
                     specific_files, external_diff_options, 
 
385
                     old_label='a/', new_label='b/', extra_trees=None):
 
386
 
 
387
    # GNU Patch uses the epoch date to detect files that are being added
 
388
    # or removed in a diff.
 
389
    EPOCH_DATE = '1970-01-01 00:00:00 +0000'
201
390
 
202
391
    # TODO: Generation of pseudo-diffs for added/deleted files could
203
392
    # be usefully made into a much faster special case.
210
399
    else:
211
400
        diff_file = internal_diff
212
401
    
213
 
 
214
 
    delta = compare_trees(old_tree, new_tree, want_unchanged=False,
215
 
                          specific_files=specific_files)
 
402
    delta = new_tree.changes_from(old_tree,
 
403
        specific_files=specific_files,
 
404
        extra_trees=extra_trees, require_versioned=True)
216
405
 
217
406
    has_changes = 0
218
407
    for path, file_id, kind in delta.removed:
219
408
        has_changes = 1
220
 
        print >>to_file, '=== removed %s %r' % (kind, path)
221
 
        old_tree.inventory[file_id].diff(diff_file, old_label + path, old_tree,
222
 
                                         DEVNULL, None, None, to_file)
 
409
        print >>to_file, '=== removed %s %r' % (kind, path.encode('utf8'))
 
410
        old_name = '%s%s\t%s' % (old_label, path,
 
411
                                 _patch_header_date(old_tree, file_id, path))
 
412
        new_name = '%s%s\t%s' % (new_label, path, EPOCH_DATE)
 
413
        old_tree.inventory[file_id].diff(diff_file, old_name, old_tree,
 
414
                                         new_name, None, None, to_file)
223
415
    for path, file_id, kind in delta.added:
224
416
        has_changes = 1
225
 
        print >>to_file, '=== added %s %r' % (kind, path)
226
 
        new_tree.inventory[file_id].diff(diff_file, new_label + path, new_tree,
227
 
                                         DEVNULL, None, None, to_file, 
 
417
        print >>to_file, '=== added %s %r' % (kind, path.encode('utf8'))
 
418
        old_name = '%s%s\t%s' % (old_label, path, EPOCH_DATE)
 
419
        new_name = '%s%s\t%s' % (new_label, path,
 
420
                                 _patch_header_date(new_tree, file_id, path))
 
421
        new_tree.inventory[file_id].diff(diff_file, new_name, new_tree,
 
422
                                         old_name, None, None, to_file, 
228
423
                                         reverse=True)
229
424
    for (old_path, new_path, file_id, kind,
230
425
         text_modified, meta_modified) in delta.renamed:
231
426
        has_changes = 1
232
427
        prop_str = get_prop_change(meta_modified)
233
428
        print >>to_file, '=== renamed %s %r => %r%s' % (
234
 
                          kind, old_path, new_path, prop_str)
235
 
        _maybe_diff_file_or_symlink(old_label, old_path, old_tree, file_id,
236
 
                                    new_label, new_path, new_tree,
 
429
                    kind, old_path.encode('utf8'),
 
430
                    new_path.encode('utf8'), prop_str)
 
431
        old_name = '%s%s\t%s' % (old_label, old_path,
 
432
                                 _patch_header_date(old_tree, file_id,
 
433
                                                    old_path))
 
434
        new_name = '%s%s\t%s' % (new_label, new_path,
 
435
                                 _patch_header_date(new_tree, file_id,
 
436
                                                    new_path))
 
437
        _maybe_diff_file_or_symlink(old_name, old_tree, file_id,
 
438
                                    new_name, new_tree,
237
439
                                    text_modified, kind, to_file, diff_file)
238
440
    for path, file_id, kind, text_modified, meta_modified in delta.modified:
239
441
        has_changes = 1
240
442
        prop_str = get_prop_change(meta_modified)
241
 
        print >>to_file, '=== modified %s %r%s' % (kind, path, prop_str)
 
443
        print >>to_file, '=== modified %s %r%s' % (kind, path.encode('utf8'), prop_str)
 
444
        old_name = '%s%s\t%s' % (old_label, path,
 
445
                                 _patch_header_date(old_tree, file_id, path))
 
446
        new_name = '%s%s\t%s' % (new_label, path,
 
447
                                 _patch_header_date(new_tree, file_id, path))
242
448
        if text_modified:
243
 
            _maybe_diff_file_or_symlink(old_label, path, old_tree, file_id,
244
 
                                        new_label, path, new_tree,
 
449
            _maybe_diff_file_or_symlink(old_name, old_tree, file_id,
 
450
                                        new_name, new_tree,
245
451
                                        True, kind, to_file, diff_file)
 
452
 
246
453
    return has_changes
 
454
 
 
455
 
 
456
def _patch_header_date(tree, file_id, path):
 
457
    """Returns a timestamp suitable for use in a patch header."""
 
458
    tm = time.gmtime(tree.get_file_mtime(file_id, path))
 
459
    return time.strftime('%Y-%m-%d %H:%M:%S +0000', tm)
 
460
 
 
461
 
 
462
def _raise_if_nonexistent(paths, old_tree, new_tree):
 
463
    """Complain if paths are not in either inventory or tree.
 
464
 
 
465
    It's OK with the files exist in either tree's inventory, or 
 
466
    if they exist in the tree but are not versioned.
247
467
    
 
468
    This can be used by operations such as bzr status that can accept
 
469
    unknown or ignored files.
 
470
    """
 
471
    mutter("check paths: %r", paths)
 
472
    if not paths:
 
473
        return
 
474
    s = old_tree.filter_unversioned_files(paths)
 
475
    s = new_tree.filter_unversioned_files(s)
 
476
    s = [path for path in s if not new_tree.has_filename(path)]
 
477
    if s:
 
478
        raise errors.PathsDoNotExist(sorted(s))
 
479
 
248
480
 
249
481
def get_prop_change(meta_modified):
250
482
    if meta_modified:
253
485
        return  ""
254
486
 
255
487
 
256
 
def _maybe_diff_file_or_symlink(old_label, old_path, old_tree, file_id,
257
 
                                new_label, new_path, new_tree, text_modified,
 
488
def _maybe_diff_file_or_symlink(old_path, old_tree, file_id,
 
489
                                new_path, new_tree, text_modified,
258
490
                                kind, to_file, diff_file):
259
491
    if text_modified:
260
492
        new_entry = new_tree.inventory[file_id]
261
493
        old_tree.inventory[file_id].diff(diff_file,
262
 
                                         old_label + old_path, old_tree,
263
 
                                         new_label + new_path, new_entry, 
 
494
                                         old_path, old_tree,
 
495
                                         new_path, new_entry, 
264
496
                                         new_tree, to_file)