← Back to branch summary

~bzr-pqm/bzr/bzr.dev

« back to all changes in this revision

Viewing changes to bzrlib/diff.py

  • Committer: Michael Ellerman
  • Date: 2006-02-28 14:45:51 UTC
  • mto: (1558.1.18 Aaron's integration)
  • mto: This revision was merged to the branch mainline in revision 1586.
  • Revision ID: michael@ellerman.id.au-20060228144551-3d9941ecde4a0b0a
Update contrib/pwk for -p1 diffs from bzr

Show diffs side-by-side

added added

removed removed

Lines of Context:
bzrlib/diff.py
1
 
# Copyright (C) 2004, 2005, 2006 Canonical Ltd.
2
 
#
 
1
# -*- coding: UTF-8 -*-
 
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 subprocess
20
 
import sys
21
 
import tempfile
22
 
import time
23
 
 
24
17
from bzrlib.delta import compare_trees
25
18
from bzrlib.errors import BzrError
26
 
import bzrlib.errors as errors
27
 
import bzrlib.osutils
28
 
from bzrlib.patiencediff import unified_diff
29
 
import bzrlib.patiencediff
30
 
from bzrlib.symbol_versioning import (deprecated_function,
31
 
        zero_eight)
32
 
from bzrlib.textfile import check_text_lines
33
 
from bzrlib.trace import mutter, warning
34
 
 
 
19
from bzrlib.symbol_versioning import *
 
20
from bzrlib.trace import mutter
35
21
 
36
22
# TODO: Rather than building a changeset object, we should probably
37
23
# invoke callbacks on an object.  That object can either accumulate a
38
24
# list, write them out directly, etc etc.
39
25
 
40
 
def internal_diff(old_filename, oldlines, new_filename, newlines, to_file,
41
 
                  allow_binary=False, sequence_matcher=None,
42
 
                  path_encoding='utf8'):
 
26
def internal_diff(old_filename, oldlines, new_filename, newlines, to_file):
 
27
    import difflib
 
28
    
43
29
    # FIXME: difflib is wrong if there is no trailing newline.
44
30
    # The syntax used by patch seems to be "\ No newline at
45
31
    # end of file" following the last diff line from that
55
41
    # both sequences are empty.
56
42
    if not oldlines and not newlines:
57
43
        return
58
 
    
59
 
    if allow_binary is False:
60
 
        check_text_lines(oldlines)
61
 
        check_text_lines(newlines)
62
44
 
63
 
    if sequence_matcher is None:
64
 
        sequence_matcher = bzrlib.patiencediff.PatienceSequenceMatcher
65
 
    ud = unified_diff(oldlines, newlines,
66
 
                      fromfile=old_filename.encode(path_encoding),
67
 
                      tofile=new_filename.encode(path_encoding),
68
 
                      sequencematcher=sequence_matcher)
 
45
    ud = difflib.unified_diff(oldlines, newlines,
 
46
                              fromfile=old_filename+'\t', 
 
47
                              tofile=new_filename+'\t')
69
48
 
70
49
    ud = list(ud)
71
50
    # work-around for difflib being too smart for its own good
88
67
def external_diff(old_filename, oldlines, new_filename, newlines, to_file,
89
68
                  diff_opts):
90
69
    """Display a diff by calling out to the external diff program."""
91
 
    if hasattr(to_file, 'fileno'):
92
 
        out_file = to_file
93
 
        have_fileno = True
94
 
    else:
95
 
        out_file = subprocess.PIPE
96
 
        have_fileno = False
 
70
    import sys
97
71
    
 
72
    if to_file != sys.stdout:
 
73
        raise NotImplementedError("sorry, can't send external diff other than to stdout yet",
 
74
                                  to_file)
 
75
 
98
76
    # make sure our own output is properly ordered before the diff
99
77
    to_file.flush()
100
78
 
101
 
    oldtmp_fd, old_abspath = tempfile.mkstemp(prefix='bzr-diff-old-')
102
 
    newtmp_fd, new_abspath = tempfile.mkstemp(prefix='bzr-diff-new-')
103
 
    oldtmpf = os.fdopen(oldtmp_fd, 'wb')
104
 
    newtmpf = os.fdopen(newtmp_fd, 'wb')
 
79
    from tempfile import NamedTemporaryFile
 
80
    import os
 
81
 
 
82
    oldtmpf = NamedTemporaryFile()
 
83
    newtmpf = NamedTemporaryFile()
105
84
 
106
85
    try:
107
86
        # TODO: perhaps a special case for comparing to or from the empty
114
93
        oldtmpf.writelines(oldlines)
115
94
        newtmpf.writelines(newlines)
116
95
 
117
 
        oldtmpf.close()
118
 
        newtmpf.close()
 
96
        oldtmpf.flush()
 
97
        newtmpf.flush()
119
98
 
120
99
        if not diff_opts:
121
100
            diff_opts = []
122
101
        diffcmd = ['diff',
123
 
                   '--label', old_filename,
124
 
                   old_abspath,
125
 
                   '--label', new_filename,
126
 
                   new_abspath,
127
 
                   '--binary',
128
 
                  ]
 
102
                   '--label', old_filename+'\t',
 
103
                   oldtmpf.name,
 
104
                   '--label', new_filename+'\t',
 
105
                   newtmpf.name]
129
106
 
130
107
        # diff only allows one style to be specified; they don't override.
131
108
        # note that some of these take optargs, and the optargs can be
151
128
        if diff_opts:
152
129
            diffcmd.extend(diff_opts)
153
130
 
154
 
        try:
155
 
            pipe = subprocess.Popen(diffcmd,
156
 
                                    stdin=subprocess.PIPE,
157
 
                                    stdout=out_file)
158
 
        except OSError, e:
159
 
            if e.errno == errno.ENOENT:
160
 
                raise errors.NoDiff(str(e))
161
 
            raise
162
 
        pipe.stdin.close()
163
 
 
164
 
        if not have_fileno:
165
 
            bzrlib.osutils.pumpfile(pipe.stdout, to_file)
166
 
        rc = pipe.wait()
 
131
        rc = os.spawnvp(os.P_WAIT, 'diff', diffcmd)
167
132
        
168
133
        if rc != 0 and rc != 1:
169
134
            # returns 1 if files differ; that's OK
176
141
    finally:
177
142
        oldtmpf.close()                 # and delete
178
143
        newtmpf.close()
179
 
        # Clean up. Warn in case the files couldn't be deleted
180
 
        # (in case windows still holds the file open, but not
181
 
        # if the files have already been deleted)
182
 
        try:
183
 
            os.remove(old_abspath)
184
 
        except OSError, e:
185
 
            if e.errno not in (errno.ENOENT,):
186
 
                warning('Failed to delete temporary file: %s %s',
187
 
                        old_abspath, e)
188
 
        try:
189
 
            os.remove(new_abspath)
190
 
        except OSError:
191
 
            if e.errno not in (errno.ENOENT,):
192
 
                warning('Failed to delete temporary file: %s %s',
193
 
                        new_abspath, e)
194
144
 
195
145
 
196
146
@deprecated_function(zero_eight)
210
160
    supplies any two trees.
211
161
    """
212
162
    if output is None:
 
163
        import sys
213
164
        output = sys.stdout
214
165
 
215
166
    if from_spec is None:
232
183
 
233
184
 
234
185
def diff_cmd_helper(tree, specific_files, external_diff_options, 
235
 
                    old_revision_spec=None, new_revision_spec=None,
236
 
                    old_label='a/', new_label='b/'):
 
186
                    old_revision_spec=None, new_revision_spec=None):
237
187
    """Helper for cmd_diff.
238
188
 
239
189
   tree 
256
206
    The more general form is show_diff_trees(), where the caller
257
207
    supplies any two trees.
258
208
    """
 
209
    import sys
 
210
    output = sys.stdout
259
211
    def spec_tree(spec):
260
 
        if tree:
261
 
            revision = spec.in_store(tree.branch)
262
 
        else:
263
 
            revision = spec.in_store(None)
264
 
        revision_id = revision.rev_id
265
 
        branch = revision.branch
266
 
        return branch.repository.revision_tree(revision_id)
 
212
        revision_id = spec.in_store(tree.branch).rev_id
 
213
        return tree.branch.repository.revision_tree(revision_id)
267
214
    if old_revision_spec is None:
268
215
        old_tree = tree.basis_tree()
269
216
    else:
273
220
        new_tree = tree
274
221
    else:
275
222
        new_tree = spec_tree(new_revision_spec)
276
 
    if new_tree is not tree:
277
 
        extra_trees = (tree,)
278
 
    else:
279
 
        extra_trees = None
280
223
 
281
224
    return show_diff_trees(old_tree, new_tree, sys.stdout, specific_files,
282
 
                           external_diff_options,
283
 
                           old_label=old_label, new_label=new_label,
284
 
                           extra_trees=extra_trees)
 
225
                           external_diff_options)
285
226
 
286
227
 
287
228
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None,
288
 
                    external_diff_options=None,
289
 
                    old_label='a/', new_label='b/',
290
 
                    extra_trees=None):
 
229
                    external_diff_options=None):
291
230
    """Show in text form the changes from one tree to another.
292
231
 
293
232
    to_files
295
234
 
296
235
    external_diff_options
297
236
        If set, use an external GNU diff and pass these options.
 
237
    """
298
238
 
299
 
    extra_trees
300
 
        If set, more Trees to use for looking up file ids
301
 
    """
302
239
    old_tree.lock_read()
303
240
    try:
304
241
        new_tree.lock_read()
305
242
        try:
306
243
            return _show_diff_trees(old_tree, new_tree, to_file,
307
 
                                    specific_files, external_diff_options,
308
 
                                    old_label=old_label, new_label=new_label,
309
 
                                    extra_trees=extra_trees)
 
244
                                    specific_files, external_diff_options)
310
245
        finally:
311
246
            new_tree.unlock()
312
247
    finally:
314
249
 
315
250
 
316
251
def _show_diff_trees(old_tree, new_tree, to_file,
317
 
                     specific_files, external_diff_options, 
318
 
                     old_label='a/', new_label='b/', extra_trees=None):
319
 
 
320
 
    # GNU Patch uses the epoch date to detect files that are being added
321
 
    # or removed in a diff.
322
 
    EPOCH_DATE = '1970-01-01 00:00:00 +0000'
 
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.
323
263
 
324
264
    # TODO: Generation of pseudo-diffs for added/deleted files could
325
265
    # be usefully made into a much faster special case.
332
272
    else:
333
273
        diff_file = internal_diff
334
274
    
 
275
 
335
276
    delta = compare_trees(old_tree, new_tree, want_unchanged=False,
336
 
                          specific_files=specific_files, 
337
 
                          extra_trees=extra_trees, require_versioned=True)
 
277
                          specific_files=specific_files)
338
278
 
339
279
    has_changes = 0
340
280
    for path, file_id, kind in delta.removed:
341
281
        has_changes = 1
342
 
        print >>to_file, '=== removed %s %r' % (kind, path.encode('utf8'))
343
 
        old_name = '%s%s\t%s' % (old_label, path,
344
 
                                 _patch_header_date(old_tree, file_id, path))
345
 
        new_name = '%s%s\t%s' % (new_label, path, EPOCH_DATE)
346
 
        old_tree.inventory[file_id].diff(diff_file, old_name, old_tree,
347
 
                                         new_name, None, None, to_file)
 
282
        print >>to_file, '=== removed %s %r' % (kind, old_label + path)
 
283
        old_tree.inventory[file_id].diff(diff_file, old_label + path, old_tree,
 
284
                                         DEVNULL, None, None, to_file)
348
285
    for path, file_id, kind in delta.added:
349
286
        has_changes = 1
350
 
        print >>to_file, '=== added %s %r' % (kind, path.encode('utf8'))
351
 
        old_name = '%s%s\t%s' % (old_label, path, EPOCH_DATE)
352
 
        new_name = '%s%s\t%s' % (new_label, path,
353
 
                                 _patch_header_date(new_tree, file_id, path))
354
 
        new_tree.inventory[file_id].diff(diff_file, new_name, new_tree,
355
 
                                         old_name, None, None, to_file, 
 
287
        print >>to_file, '=== added %s %r' % (kind, new_label + path)
 
288
        new_tree.inventory[file_id].diff(diff_file, new_label + path, new_tree,
 
289
                                         DEVNULL, None, None, to_file, 
356
290
                                         reverse=True)
357
291
    for (old_path, new_path, file_id, kind,
358
292
         text_modified, meta_modified) in delta.renamed:
359
293
        has_changes = 1
360
294
        prop_str = get_prop_change(meta_modified)
361
295
        print >>to_file, '=== renamed %s %r => %r%s' % (
362
 
                    kind, old_path.encode('utf8'),
363
 
                    new_path.encode('utf8'), prop_str)
364
 
        old_name = '%s%s\t%s' % (old_label, old_path,
365
 
                                 _patch_header_date(old_tree, file_id,
366
 
                                                    old_path))
367
 
        new_name = '%s%s\t%s' % (new_label, new_path,
368
 
                                 _patch_header_date(new_tree, file_id,
369
 
                                                    new_path))
370
 
        _maybe_diff_file_or_symlink(old_name, old_tree, file_id,
371
 
                                    new_name, new_tree,
 
296
                    kind, old_label + old_path, new_label + new_path, prop_str)
 
297
        _maybe_diff_file_or_symlink(old_label, old_path, old_tree, file_id,
 
298
                                    new_label, new_path, new_tree,
372
299
                                    text_modified, kind, to_file, diff_file)
373
300
    for path, file_id, kind, text_modified, meta_modified in delta.modified:
374
301
        has_changes = 1
375
302
        prop_str = get_prop_change(meta_modified)
376
 
        print >>to_file, '=== modified %s %r%s' % (kind, path.encode('utf8'), prop_str)
377
 
        old_name = '%s%s\t%s' % (old_label, path,
378
 
                                 _patch_header_date(old_tree, file_id, path))
379
 
        new_name = '%s%s\t%s' % (new_label, path,
380
 
                                 _patch_header_date(new_tree, file_id, path))
 
303
        print >>to_file, '=== modified %s %r%s' % (kind, old_label + path,
 
304
                    prop_str)
381
305
        if text_modified:
382
 
            _maybe_diff_file_or_symlink(old_name, old_tree, file_id,
383
 
                                        new_name, new_tree,
 
306
            _maybe_diff_file_or_symlink(old_label, path, old_tree, file_id,
 
307
                                        new_label, path, new_tree,
384
308
                                        True, kind, to_file, diff_file)
385
 
 
386
309
    return has_changes
387
 
 
388
 
 
389
 
def _patch_header_date(tree, file_id, path):
390
 
    """Returns a timestamp suitable for use in a patch header."""
391
 
    tm = time.gmtime(tree.get_file_mtime(file_id, path))
392
 
    return time.strftime('%Y-%m-%d %H:%M:%S +0000', tm)
393
 
 
394
 
 
395
 
def _raise_if_nonexistent(paths, old_tree, new_tree):
396
 
    """Complain if paths are not in either inventory or tree.
397
 
 
398
 
    It's OK with the files exist in either tree's inventory, or 
399
 
    if they exist in the tree but are not versioned.
400
310
    
401
 
    This can be used by operations such as bzr status that can accept
402
 
    unknown or ignored files.
403
 
    """
404
 
    mutter("check paths: %r", paths)
405
 
    if not paths:
406
 
        return
407
 
    s = old_tree.filter_unversioned_files(paths)
408
 
    s = new_tree.filter_unversioned_files(s)
409
 
    s = [path for path in s if not new_tree.has_filename(path)]
410
 
    if s:
411
 
        raise errors.PathsDoNotExist(sorted(s))
412
 
 
413
311
 
414
312
def get_prop_change(meta_modified):
415
313
    if meta_modified:
418
316
        return  ""
419
317
 
420
318
 
421
 
def _maybe_diff_file_or_symlink(old_path, old_tree, file_id,
422
 
                                new_path, new_tree, text_modified,
 
319
def _maybe_diff_file_or_symlink(old_label, old_path, old_tree, file_id,
 
320
                                new_label, new_path, new_tree, text_modified,
423
321
                                kind, to_file, diff_file):
424
322
    if text_modified:
425
323
        new_entry = new_tree.inventory[file_id]
426
324
        old_tree.inventory[file_id].diff(diff_file,
427
 
                                         old_path, old_tree,
428
 
                                         new_path, new_entry, 
 
325
                                         old_label + old_path, old_tree,
 
326
                                         new_label + new_path, new_entry, 
429
327
                                         new_tree, to_file)