~bzr-pqm/bzr/bzr.dev

« back to all changes in this revision

Viewing changes to bzrlib/diff.py

  • Committer: John Arbash Meinel
  • Date: 2006-09-10 20:39:26 UTC
  • mto: This revision was merged to the branch mainline in revision 2004.
  • Revision ID: john@arbash-meinel.com-20060910203926-ae731f6bb165d6fa
Adding a ScopeReplacer class, which can replace itself on demand

Show diffs side-by-side

added added

removed removed

Lines of Context:
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
from bzrlib import (
 
26
    errors,
 
27
    osutils,
 
28
    )
 
29
# compatability - plugins import compare_trees from diff!!!
 
30
# deprecated as of 0.10
17
31
from bzrlib.delta import compare_trees
18
32
from bzrlib.errors import BzrError
19
 
import bzrlib.errors as errors
20
 
from bzrlib.symbol_versioning import *
 
33
from bzrlib.patiencediff import unified_diff
 
34
import bzrlib.patiencediff
 
35
from bzrlib.symbol_versioning import (deprecated_function,
 
36
        zero_eight)
21
37
from bzrlib.textfile import check_text_lines
22
 
from bzrlib.trace import mutter
 
38
from bzrlib.trace import mutter, warning
 
39
 
23
40
 
24
41
# TODO: Rather than building a changeset object, we should probably
25
42
# invoke callbacks on an object.  That object can either accumulate a
26
43
# list, write them out directly, etc etc.
27
44
 
28
45
def internal_diff(old_filename, oldlines, new_filename, newlines, to_file,
29
 
                  allow_binary=False):
30
 
    import difflib
31
 
    
 
46
                  allow_binary=False, sequence_matcher=None,
 
47
                  path_encoding='utf8'):
32
48
    # FIXME: difflib is wrong if there is no trailing newline.
33
49
    # The syntax used by patch seems to be "\ No newline at
34
50
    # end of file" following the last diff line from that
49
65
        check_text_lines(oldlines)
50
66
        check_text_lines(newlines)
51
67
 
52
 
    ud = difflib.unified_diff(oldlines, newlines,
53
 
                              fromfile=old_filename+'\t', 
54
 
                              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)
55
74
 
56
75
    ud = list(ud)
57
76
    # work-around for difflib being too smart for its own good
71
90
    print >>to_file
72
91
 
73
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
 
74
136
def external_diff(old_filename, oldlines, new_filename, newlines, to_file,
75
137
                  diff_opts):
76
138
    """Display a diff by calling out to the external diff program."""
77
 
    import sys
78
 
    
79
 
    if to_file != sys.stdout:
80
 
        raise NotImplementedError("sorry, can't send external diff other than to stdout yet",
81
 
                                  to_file)
82
 
 
83
139
    # make sure our own output is properly ordered before the diff
84
140
    to_file.flush()
85
141
 
86
 
    from tempfile import NamedTemporaryFile
87
 
    import os
88
 
 
89
 
    oldtmpf = NamedTemporaryFile()
90
 
    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')
91
146
 
92
147
    try:
93
148
        # TODO: perhaps a special case for comparing to or from the empty
100
155
        oldtmpf.writelines(oldlines)
101
156
        newtmpf.writelines(newlines)
102
157
 
103
 
        oldtmpf.flush()
104
 
        newtmpf.flush()
 
158
        oldtmpf.close()
 
159
        newtmpf.close()
105
160
 
106
161
        if not diff_opts:
107
162
            diff_opts = []
108
163
        diffcmd = ['diff',
109
 
                   '--label', old_filename+'\t',
110
 
                   oldtmpf.name,
111
 
                   '--label', new_filename+'\t',
112
 
                   newtmpf.name]
 
164
                   '--label', old_filename,
 
165
                   old_abspath,
 
166
                   '--label', new_filename,
 
167
                   new_abspath,
 
168
                   '--binary',
 
169
                  ]
113
170
 
114
171
        # diff only allows one style to be specified; they don't override.
115
172
        # note that some of these take optargs, and the optargs can be
135
192
        if diff_opts:
136
193
            diffcmd.extend(diff_opts)
137
194
 
138
 
        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
139
198
        
140
 
        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):
141
233
            # returns 1 if files differ; that's OK
142
234
            if rc < 0:
143
235
                msg = 'signal %d' % (-rc)
144
236
            else:
145
237
                msg = 'exit code %d' % rc
146
238
                
147
 
            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
 
148
243
    finally:
149
244
        oldtmpf.close()                 # and delete
150
245
        newtmpf.close()
 
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)
151
261
 
152
262
 
153
263
@deprecated_function(zero_eight)
167
277
    supplies any two trees.
168
278
    """
169
279
    if output is None:
170
 
        import sys
171
280
        output = sys.stdout
172
281
 
173
282
    if from_spec is None:
214
323
    The more general form is show_diff_trees(), where the caller
215
324
    supplies any two trees.
216
325
    """
217
 
    import sys
218
 
    output = sys.stdout
219
326
    def spec_tree(spec):
220
 
        revision_id = spec.in_store(tree.branch).rev_id
221
 
        return tree.branch.repository.revision_tree(revision_id)
 
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)
222
334
    if old_revision_spec is None:
223
335
        old_tree = tree.basis_tree()
224
336
    else:
228
340
        new_tree = tree
229
341
    else:
230
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
231
347
 
232
348
    return show_diff_trees(old_tree, new_tree, sys.stdout, specific_files,
233
349
                           external_diff_options,
234
 
                           old_label=old_label, new_label=new_label)
 
350
                           old_label=old_label, new_label=new_label,
 
351
                           extra_trees=extra_trees)
235
352
 
236
353
 
237
354
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None,
238
355
                    external_diff_options=None,
239
 
                    old_label='a/', new_label='b/'):
 
356
                    old_label='a/', new_label='b/',
 
357
                    extra_trees=None):
240
358
    """Show in text form the changes from one tree to another.
241
359
 
242
360
    to_files
244
362
 
245
363
    external_diff_options
246
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
247
368
    """
248
369
    old_tree.lock_read()
249
370
    try:
251
372
        try:
252
373
            return _show_diff_trees(old_tree, new_tree, to_file,
253
374
                                    specific_files, external_diff_options,
254
 
                                    old_label=old_label, new_label=new_label)
 
375
                                    old_label=old_label, new_label=new_label,
 
376
                                    extra_trees=extra_trees)
255
377
        finally:
256
378
            new_tree.unlock()
257
379
    finally:
260
382
 
261
383
def _show_diff_trees(old_tree, new_tree, to_file,
262
384
                     specific_files, external_diff_options, 
263
 
                     old_label='a/', new_label='b/' ):
 
385
                     old_label='a/', new_label='b/', extra_trees=None):
264
386
 
265
 
    DEVNULL = '/dev/null'
266
 
    # Windows users, don't panic about this filename -- it is a
267
 
    # special signal to GNU patch that the file should be created or
268
 
    # deleted respectively.
 
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'
269
390
 
270
391
    # TODO: Generation of pseudo-diffs for added/deleted files could
271
392
    # be usefully made into a much faster special case.
272
393
 
273
 
    _raise_if_doubly_unversioned(specific_files, old_tree, new_tree)
274
 
 
275
394
    if external_diff_options:
276
395
        assert isinstance(external_diff_options, basestring)
277
396
        opts = external_diff_options.split()
280
399
    else:
281
400
        diff_file = internal_diff
282
401
    
283
 
    delta = compare_trees(old_tree, new_tree, want_unchanged=False,
284
 
                          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)
285
405
 
286
406
    has_changes = 0
287
407
    for path, file_id, kind in delta.removed:
288
408
        has_changes = 1
289
 
        print >>to_file, '=== removed %s %r' % (kind, path)
290
 
        old_tree.inventory[file_id].diff(diff_file, old_label + path, old_tree,
291
 
                                         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)
292
415
    for path, file_id, kind in delta.added:
293
416
        has_changes = 1
294
 
        print >>to_file, '=== added %s %r' % (kind, path)
295
 
        new_tree.inventory[file_id].diff(diff_file, new_label + path, new_tree,
296
 
                                         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, 
297
423
                                         reverse=True)
298
424
    for (old_path, new_path, file_id, kind,
299
425
         text_modified, meta_modified) in delta.renamed:
300
426
        has_changes = 1
301
427
        prop_str = get_prop_change(meta_modified)
302
428
        print >>to_file, '=== renamed %s %r => %r%s' % (
303
 
                    kind, old_path, new_path, prop_str)
304
 
        _maybe_diff_file_or_symlink(old_label, old_path, old_tree, file_id,
305
 
                                    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,
306
439
                                    text_modified, kind, to_file, diff_file)
307
440
    for path, file_id, kind, text_modified, meta_modified in delta.modified:
308
441
        has_changes = 1
309
442
        prop_str = get_prop_change(meta_modified)
310
 
        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))
311
448
        if text_modified:
312
 
            _maybe_diff_file_or_symlink(old_label, path, old_tree, file_id,
313
 
                                        new_label, path, new_tree,
 
449
            _maybe_diff_file_or_symlink(old_name, old_tree, file_id,
 
450
                                        new_name, new_tree,
314
451
                                        True, kind, to_file, diff_file)
315
452
 
316
453
    return has_changes
317
454
 
318
455
 
319
 
def _raise_if_doubly_unversioned(specific_files, old_tree, new_tree):
320
 
    """Complain if paths are not versioned in either tree."""
321
 
    if not specific_files:
322
 
        return
323
 
    old_unversioned = old_tree.filter_unversioned_files(specific_files)
324
 
    new_unversioned = new_tree.filter_unversioned_files(specific_files)
325
 
    unversioned = old_unversioned.intersection(new_unversioned)
326
 
    if unversioned:
327
 
        raise errors.PathsNotVersionedError(sorted(unversioned))
328
 
    
 
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
 
329
461
 
330
462
def _raise_if_nonexistent(paths, old_tree, new_tree):
331
463
    """Complain if paths are not in either inventory or tree.
353
485
        return  ""
354
486
 
355
487
 
356
 
def _maybe_diff_file_or_symlink(old_label, old_path, old_tree, file_id,
357
 
                                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,
358
490
                                kind, to_file, diff_file):
359
491
    if text_modified:
360
492
        new_entry = new_tree.inventory[file_id]
361
493
        old_tree.inventory[file_id].diff(diff_file,
362
 
                                         old_label + old_path, old_tree,
363
 
                                         new_label + new_path, new_entry, 
 
494
                                         old_path, old_tree,
 
495
                                         new_path, new_entry, 
364
496
                                         new_tree, to_file)