~bzr-pqm/bzr/bzr.dev

« back to all changes in this revision

Viewing changes to bzrlib/diff.py

  • Committer: Robey Pointer
  • Date: 2006-06-29 23:25:48 UTC
  • mto: This revision was merged to the branch mainline in revision 1839.
  • Revision ID: robey@lag.net-20060629232548-25aaee5fe809193c
fix typo

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 subprocess
 
20
import sys
 
21
import tempfile
 
22
import time
 
23
 
 
24
from bzrlib.delta import compare_trees
18
25
from bzrlib.errors import BzrError
19
 
from bzrlib.delta import compare_trees
 
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
 
20
35
 
21
36
# TODO: Rather than building a changeset object, we should probably
22
37
# invoke callbacks on an object.  That object can either accumulate a
23
38
# list, write them out directly, etc etc.
24
39
 
25
 
def internal_diff(old_label, oldlines, new_label, newlines, to_file):
26
 
    import difflib
27
 
    
 
40
def internal_diff(old_filename, oldlines, new_filename, newlines, to_file,
 
41
                  allow_binary=False, sequence_matcher=None,
 
42
                  path_encoding='utf8'):
28
43
    # FIXME: difflib is wrong if there is no trailing newline.
29
44
    # The syntax used by patch seems to be "\ No newline at
30
45
    # end of file" following the last diff line from that
40
55
    # both sequences are empty.
41
56
    if not oldlines and not newlines:
42
57
        return
 
58
    
 
59
    if allow_binary is False:
 
60
        check_text_lines(oldlines)
 
61
        check_text_lines(newlines)
43
62
 
44
 
    ud = difflib.unified_diff(oldlines, newlines,
45
 
                              fromfile=old_label, tofile=new_label)
 
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)
46
69
 
47
70
    ud = list(ud)
48
71
    # work-around for difflib being too smart for its own good
62
85
    print >>to_file
63
86
 
64
87
 
65
 
def external_diff(old_label, oldlines, new_label, newlines, to_file,
 
88
def external_diff(old_filename, oldlines, new_filename, newlines, to_file,
66
89
                  diff_opts):
67
90
    """Display a diff by calling out to the external diff program."""
68
 
    import sys
 
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
69
97
    
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
98
    # make sure our own output is properly ordered before the diff
75
99
    to_file.flush()
76
100
 
77
 
    from tempfile import NamedTemporaryFile
78
 
    import os
79
 
 
80
 
    oldtmpf = NamedTemporaryFile()
81
 
    newtmpf = NamedTemporaryFile()
 
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')
82
105
 
83
106
    try:
84
107
        # TODO: perhaps a special case for comparing to or from the empty
91
114
        oldtmpf.writelines(oldlines)
92
115
        newtmpf.writelines(newlines)
93
116
 
94
 
        oldtmpf.flush()
95
 
        newtmpf.flush()
 
117
        oldtmpf.close()
 
118
        newtmpf.close()
96
119
 
97
120
        if not diff_opts:
98
121
            diff_opts = []
99
122
        diffcmd = ['diff',
100
 
                   '--label', old_label,
101
 
                   oldtmpf.name,
102
 
                   '--label', new_label,
103
 
                   newtmpf.name]
 
123
                   '--label', old_filename,
 
124
                   old_abspath,
 
125
                   '--label', new_filename,
 
126
                   new_abspath,
 
127
                   '--binary',
 
128
                  ]
104
129
 
105
130
        # diff only allows one style to be specified; they don't override.
106
131
        # note that some of these take optargs, and the optargs can be
126
151
        if diff_opts:
127
152
            diffcmd.extend(diff_opts)
128
153
 
129
 
        rc = os.spawnvp(os.P_WAIT, 'diff', diffcmd)
 
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()
130
167
        
131
168
        if rc != 0 and rc != 1:
132
169
            # returns 1 if files differ; that's OK
139
176
    finally:
140
177
        oldtmpf.close()                 # and delete
141
178
        newtmpf.close()
142
 
 
143
 
def show_diff(b, revision, specific_files, external_diff_options=None,
144
 
              revision2=None, output=None):
 
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
 
 
195
 
 
196
@deprecated_function(zero_eight)
 
197
def show_diff(b, from_spec, specific_files, external_diff_options=None,
 
198
              revision2=None, output=None, b2=None):
145
199
    """Shortcut for showing the diff to the working tree.
146
200
 
 
201
    Please use show_diff_trees instead.
 
202
 
147
203
    b
148
204
        Branch.
149
205
 
150
206
    revision
151
 
        None for each, or otherwise the old revision to compare against.
 
207
        None for 'basis tree', or otherwise the old revision to compare against.
152
208
    
153
209
    The more general form is show_diff_trees(), where the caller
154
210
    supplies any two trees.
155
211
    """
156
212
    if output is None:
157
 
        import sys
158
213
        output = sys.stdout
159
214
 
160
 
    if revision is None:
161
 
        old_tree = b.basis_tree()
 
215
    if from_spec is None:
 
216
        old_tree = b.bzrdir.open_workingtree()
 
217
        if b2 is None:
 
218
            old_tree = old_tree = old_tree.basis_tree()
162
219
    else:
163
 
        old_tree = b.revision_tree(revision.in_history(b).rev_id)
 
220
        old_tree = b.repository.revision_tree(from_spec.in_history(b).rev_id)
164
221
 
165
222
    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
 
 
 
223
        if b2 is None:
 
224
            new_tree = b.bzrdir.open_workingtree()
 
225
        else:
 
226
            new_tree = b2.bzrdir.open_workingtree()
 
227
    else:
 
228
        new_tree = b.repository.revision_tree(revision2.in_history(b).rev_id)
 
229
 
 
230
    return show_diff_trees(old_tree, new_tree, output, specific_files,
 
231
                           external_diff_options)
 
232
 
 
233
 
 
234
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/'):
 
237
    """Helper for cmd_diff.
 
238
 
 
239
   tree 
 
240
        A WorkingTree
 
241
 
 
242
    specific_files
 
243
        The specific files to compare, or None
 
244
 
 
245
    external_diff_options
 
246
        If non-None, run an external diff, and pass it these options
 
247
 
 
248
    old_revision_spec
 
249
        If None, use basis tree as old revision, otherwise use the tree for
 
250
        the specified revision. 
 
251
 
 
252
    new_revision_spec
 
253
        If None, use working tree as new revision, otherwise use the tree for
 
254
        the specified revision.
 
255
    
 
256
    The more general form is show_diff_trees(), where the caller
 
257
    supplies any two trees.
 
258
    """
 
259
    output = sys.stdout
 
260
    def spec_tree(spec):
 
261
        revision_id = spec.in_store(tree.branch).rev_id
 
262
        return tree.branch.repository.revision_tree(revision_id)
 
263
    if old_revision_spec is None:
 
264
        old_tree = tree.basis_tree()
 
265
    else:
 
266
        old_tree = spec_tree(old_revision_spec)
 
267
 
 
268
    if new_revision_spec is None:
 
269
        new_tree = tree
 
270
    else:
 
271
        new_tree = spec_tree(new_revision_spec)
 
272
 
 
273
    return show_diff_trees(old_tree, new_tree, sys.stdout, specific_files,
 
274
                           external_diff_options,
 
275
                           old_label=old_label, new_label=new_label)
173
276
 
174
277
 
175
278
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None,
176
 
                    external_diff_options=None):
 
279
                    external_diff_options=None,
 
280
                    old_label='a/', new_label='b/'):
177
281
    """Show in text form the changes from one tree to another.
178
282
 
179
283
    to_files
182
286
    external_diff_options
183
287
        If set, use an external GNU diff and pass these options.
184
288
    """
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.
 
289
    old_tree.lock_read()
 
290
    try:
 
291
        new_tree.lock_read()
 
292
        try:
 
293
            return _show_diff_trees(old_tree, new_tree, to_file,
 
294
                                    specific_files, external_diff_options,
 
295
                                    old_label=old_label, new_label=new_label)
 
296
        finally:
 
297
            new_tree.unlock()
 
298
    finally:
 
299
        old_tree.unlock()
 
300
 
 
301
 
 
302
def _show_diff_trees(old_tree, new_tree, to_file,
 
303
                     specific_files, external_diff_options, 
 
304
                     old_label='a/', new_label='b/' ):
 
305
 
 
306
    # GNU Patch uses the epoch date to detect files that are being added
 
307
    # or removed in a diff.
 
308
    EPOCH_DATE = '1970-01-01 00:00:00 +0000'
194
309
 
195
310
    # TODO: Generation of pseudo-diffs for added/deleted files could
196
311
    # be usefully made into a much faster special case.
197
312
 
 
313
    _raise_if_doubly_unversioned(specific_files, old_tree, new_tree)
 
314
 
198
315
    if external_diff_options:
199
316
        assert isinstance(external_diff_options, basestring)
200
317
        opts = external_diff_options.split()
203
320
    else:
204
321
        diff_file = internal_diff
205
322
    
206
 
 
207
323
    delta = compare_trees(old_tree, new_tree, want_unchanged=False,
208
324
                          specific_files=specific_files)
209
325
 
 
326
    has_changes = 0
210
327
    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)
 
328
        has_changes = 1
 
329
        print >>to_file, '=== removed %s %r' % (kind, path.encode('utf8'))
 
330
        old_name = '%s%s\t%s' % (old_label, path,
 
331
                                 _patch_header_date(old_tree, file_id, path))
 
332
        new_name = '%s%s\t%s' % (new_label, path, EPOCH_DATE)
 
333
        old_tree.inventory[file_id].diff(diff_file, old_name, old_tree,
 
334
                                         new_name, None, None, to_file)
214
335
    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, 
 
336
        has_changes = 1
 
337
        print >>to_file, '=== added %s %r' % (kind, path.encode('utf8'))
 
338
        old_name = '%s%s\t%s' % (old_label, path, EPOCH_DATE)
 
339
        new_name = '%s%s\t%s' % (new_label, path,
 
340
                                 _patch_header_date(new_tree, file_id, path))
 
341
        new_tree.inventory[file_id].diff(diff_file, new_name, new_tree,
 
342
                                         old_name, None, None, to_file, 
218
343
                                         reverse=True)
219
344
    for (old_path, new_path, file_id, kind,
220
345
         text_modified, meta_modified) in delta.renamed:
 
346
        has_changes = 1
221
347
        prop_str = get_prop_change(meta_modified)
222
348
        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,
 
349
                    kind, old_path.encode('utf8'),
 
350
                    new_path.encode('utf8'), prop_str)
 
351
        old_name = '%s%s\t%s' % (old_label, old_path,
 
352
                                 _patch_header_date(old_tree, file_id,
 
353
                                                    old_path))
 
354
        new_name = '%s%s\t%s' % (new_label, new_path,
 
355
                                 _patch_header_date(new_tree, file_id,
 
356
                                                    new_path))
 
357
        _maybe_diff_file_or_symlink(old_name, old_tree, file_id,
 
358
                                    new_name, new_tree,
226
359
                                    text_modified, kind, to_file, diff_file)
227
360
    for path, file_id, kind, text_modified, meta_modified in delta.modified:
 
361
        has_changes = 1
228
362
        prop_str = get_prop_change(meta_modified)
229
 
        print >>to_file, '=== modified %s %r%s' % (kind, path, prop_str)
 
363
        print >>to_file, '=== modified %s %r%s' % (kind, path.encode('utf8'), prop_str)
 
364
        old_name = '%s%s\t%s' % (old_label, path,
 
365
                                 _patch_header_date(old_tree, file_id, path))
 
366
        new_name = '%s%s\t%s' % (new_label, path,
 
367
                                 _patch_header_date(new_tree, file_id, path))
230
368
        if text_modified:
231
 
            _maybe_diff_file_or_symlink(old_label, path, old_tree, file_id,
232
 
                                        new_label, path, new_tree,
 
369
            _maybe_diff_file_or_symlink(old_name, old_tree, file_id,
 
370
                                        new_name, new_tree,
233
371
                                        True, kind, to_file, diff_file)
234
 
    
 
372
 
 
373
    return has_changes
 
374
 
 
375
 
 
376
def _patch_header_date(tree, file_id, path):
 
377
    """Returns a timestamp suitable for use in a patch header."""
 
378
    tm = time.gmtime(tree.get_file_mtime(file_id, path))
 
379
    return time.strftime('%Y-%m-%d %H:%M:%S +0000', tm)
 
380
 
 
381
 
 
382
def _raise_if_doubly_unversioned(specific_files, old_tree, new_tree):
 
383
    """Complain if paths are not versioned in either tree."""
 
384
    if not specific_files:
 
385
        return
 
386
    old_unversioned = old_tree.filter_unversioned_files(specific_files)
 
387
    new_unversioned = new_tree.filter_unversioned_files(specific_files)
 
388
    unversioned = old_unversioned.intersection(new_unversioned)
 
389
    if unversioned:
 
390
        raise errors.PathsNotVersionedError(sorted(unversioned))
 
391
    
 
392
 
 
393
def _raise_if_nonexistent(paths, old_tree, new_tree):
 
394
    """Complain if paths are not in either inventory or tree.
 
395
 
 
396
    It's OK with the files exist in either tree's inventory, or 
 
397
    if they exist in the tree but are not versioned.
 
398
    
 
399
    This can be used by operations such as bzr status that can accept
 
400
    unknown or ignored files.
 
401
    """
 
402
    mutter("check paths: %r", paths)
 
403
    if not paths:
 
404
        return
 
405
    s = old_tree.filter_unversioned_files(paths)
 
406
    s = new_tree.filter_unversioned_files(s)
 
407
    s = [path for path in s if not new_tree.has_filename(path)]
 
408
    if s:
 
409
        raise errors.PathsDoNotExist(sorted(s))
 
410
 
235
411
 
236
412
def get_prop_change(meta_modified):
237
413
    if meta_modified:
240
416
        return  ""
241
417
 
242
418
 
243
 
def _maybe_diff_file_or_symlink(old_label, old_path, old_tree, file_id,
244
 
                                new_label, new_path, new_tree, text_modified,
 
419
def _maybe_diff_file_or_symlink(old_path, old_tree, file_id,
 
420
                                new_path, new_tree, text_modified,
245
421
                                kind, to_file, diff_file):
246
422
    if text_modified:
247
423
        new_entry = new_tree.inventory[file_id]
248
424
        old_tree.inventory[file_id].diff(diff_file,
249
 
                                         old_label + old_path, old_tree,
250
 
                                         new_label + new_path, new_entry, 
 
425
                                         old_path, old_tree,
 
426
                                         new_path, new_entry, 
251
427
                                         new_tree, to_file)