← Back to branch summary

~bzr-pqm/bzr/bzr.dev

« back to all changes in this revision

Viewing changes to bzrlib/diff.py

  • Committer: John Arbash Meinel
  • Date: 2007-11-13 20:37:09 UTC
  • mto: This revision was merged to the branch mainline in revision 3001.
  • Revision ID: john@arbash-meinel.com-20071113203709-kysdte0emqv84pnj
https://launchpad.net/bugs/162486
Fix bug #162486, by having RemoteBranch properly initialize self._revision_id_to_revno_map.

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
19
 
from bzrlib.delta import compare_trees
 
17
import difflib
 
18
import os
 
19
import re
 
20
import sys
 
21
 
 
22
from bzrlib.lazy_import import lazy_import
 
23
lazy_import(globals(), """
 
24
import errno
 
25
import subprocess
 
26
import tempfile
 
27
import time
 
28
 
 
29
from bzrlib import (
 
30
    errors,
 
31
    osutils,
 
32
    patiencediff,
 
33
    textfile,
 
34
    timestamp,
 
35
    )
 
36
""")
 
37
 
 
38
from bzrlib.symbol_versioning import (
 
39
        deprecated_function,
 
40
        )
 
41
from bzrlib.trace import mutter, warning
 
42
 
20
43
 
21
44
# TODO: Rather than building a changeset object, we should probably
22
45
# invoke callbacks on an object.  That object can either accumulate a
23
46
# list, write them out directly, etc etc.
24
47
 
25
 
def internal_diff(old_filename, oldlines, new_filename, newlines, to_file):
26
 
    import difflib
27
 
    
 
48
 
 
49
class _PrematchedMatcher(difflib.SequenceMatcher):
 
50
    """Allow SequenceMatcher operations to use predetermined blocks"""
 
51
 
 
52
    def __init__(self, matching_blocks):
 
53
        difflib.SequenceMatcher(self, None, None)
 
54
        self.matching_blocks = matching_blocks
 
55
        self.opcodes = None
 
56
 
 
57
 
 
58
def internal_diff(old_filename, oldlines, new_filename, newlines, to_file,
 
59
                  allow_binary=False, sequence_matcher=None,
 
60
                  path_encoding='utf8'):
28
61
    # FIXME: difflib is wrong if there is no trailing newline.
29
62
    # The syntax used by patch seems to be "\ No newline at
30
63
    # end of file" following the last diff line from that
40
73
    # both sequences are empty.
41
74
    if not oldlines and not newlines:
42
75
        return
 
76
    
 
77
    if allow_binary is False:
 
78
        textfile.check_text_lines(oldlines)
 
79
        textfile.check_text_lines(newlines)
43
80
 
44
 
    ud = difflib.unified_diff(oldlines, newlines,
45
 
                              fromfile=old_filename+'\t', 
46
 
                              tofile=new_filename+'\t')
 
81
    if sequence_matcher is None:
 
82
        sequence_matcher = patiencediff.PatienceSequenceMatcher
 
83
    ud = patiencediff.unified_diff(oldlines, newlines,
 
84
                      fromfile=old_filename.encode(path_encoding),
 
85
                      tofile=new_filename.encode(path_encoding),
 
86
                      sequencematcher=sequence_matcher)
47
87
 
48
88
    ud = list(ud)
49
89
    # work-around for difflib being too smart for its own good
60
100
        to_file.write(line)
61
101
        if not line.endswith('\n'):
62
102
            to_file.write("\n\\ No newline at end of file\n")
63
 
    print >>to_file
 
103
    to_file.write('\n')
 
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
        and LC_ALL=C. This lets us read and understand the output of diff,
 
112
        and respond to any errors.
 
113
    :return: A Popen object.
 
114
    """
 
115
    if capture_errors:
 
116
        # construct minimal environment
 
117
        env = {}
 
118
        path = os.environ.get('PATH')
 
119
        if path is not None:
 
120
            env['PATH'] = path
 
121
        env['LANGUAGE'] = 'C'   # on win32 only LANGUAGE has effect
 
122
        env['LANG'] = 'C'
 
123
        env['LC_ALL'] = 'C'
 
124
        stderr = subprocess.PIPE
 
125
    else:
 
126
        env = None
 
127
        stderr = None
 
128
 
 
129
    try:
 
130
        pipe = subprocess.Popen(diffcmd,
 
131
                                stdin=subprocess.PIPE,
 
132
                                stdout=subprocess.PIPE,
 
133
                                stderr=stderr,
 
134
                                env=env)
 
135
    except OSError, e:
 
136
        if e.errno == errno.ENOENT:
 
137
            raise errors.NoDiff(str(e))
 
138
        raise
 
139
 
 
140
    return pipe
64
141
 
65
142
 
66
143
def external_diff(old_filename, oldlines, new_filename, newlines, to_file,
67
144
                  diff_opts):
68
145
    """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
146
    # make sure our own output is properly ordered before the diff
76
147
    to_file.flush()
77
148
 
78
 
    from tempfile import NamedTemporaryFile
79
 
    import os
80
 
 
81
 
    oldtmpf = NamedTemporaryFile()
82
 
    newtmpf = NamedTemporaryFile()
 
149
    oldtmp_fd, old_abspath = tempfile.mkstemp(prefix='bzr-diff-old-')
 
150
    newtmp_fd, new_abspath = tempfile.mkstemp(prefix='bzr-diff-new-')
 
151
    oldtmpf = os.fdopen(oldtmp_fd, 'wb')
 
152
    newtmpf = os.fdopen(newtmp_fd, 'wb')
83
153
 
84
154
    try:
85
155
        # TODO: perhaps a special case for comparing to or from the empty
92
162
        oldtmpf.writelines(oldlines)
93
163
        newtmpf.writelines(newlines)
94
164
 
95
 
        oldtmpf.flush()
96
 
        newtmpf.flush()
 
165
        oldtmpf.close()
 
166
        newtmpf.close()
97
167
 
98
168
        if not diff_opts:
99
169
            diff_opts = []
100
170
        diffcmd = ['diff',
101
 
                   '--label', old_filename+'\t',
102
 
                   oldtmpf.name,
103
 
                   '--label', new_filename+'\t',
104
 
                   newtmpf.name]
 
171
                   '--label', old_filename,
 
172
                   old_abspath,
 
173
                   '--label', new_filename,
 
174
                   new_abspath,
 
175
                   '--binary',
 
176
                  ]
105
177
 
106
178
        # diff only allows one style to be specified; they don't override.
107
179
        # note that some of these take optargs, and the optargs can be
127
199
        if diff_opts:
128
200
            diffcmd.extend(diff_opts)
129
201
 
130
 
        rc = os.spawnvp(os.P_WAIT, 'diff', diffcmd)
 
202
        pipe = _spawn_external_diff(diffcmd, capture_errors=True)
 
203
        out,err = pipe.communicate()
 
204
        rc = pipe.returncode
131
205
        
132
 
        if rc != 0 and rc != 1:
 
206
        # internal_diff() adds a trailing newline, add one here for consistency
 
207
        out += '\n'
 
208
        if rc == 2:
 
209
            # 'diff' gives retcode == 2 for all sorts of errors
 
210
            # one of those is 'Binary files differ'.
 
211
            # Bad options could also be the problem.
 
212
            # 'Binary files' is not a real error, so we suppress that error.
 
213
            lang_c_out = out
 
214
 
 
215
            # Since we got here, we want to make sure to give an i18n error
 
216
            pipe = _spawn_external_diff(diffcmd, capture_errors=False)
 
217
            out, err = pipe.communicate()
 
218
 
 
219
            # Write out the new i18n diff response
 
220
            to_file.write(out+'\n')
 
221
            if pipe.returncode != 2:
 
222
                raise errors.BzrError(
 
223
                               'external diff failed with exit code 2'
 
224
                               ' when run with LANG=C and LC_ALL=C,'
 
225
                               ' but not when run natively: %r' % (diffcmd,))
 
226
 
 
227
            first_line = lang_c_out.split('\n', 1)[0]
 
228
            # Starting with diffutils 2.8.4 the word "binary" was dropped.
 
229
            m = re.match('^(binary )?files.*differ$', first_line, re.I)
 
230
            if m is None:
 
231
                raise errors.BzrError('external diff failed with exit code 2;'
 
232
                                      ' command: %r' % (diffcmd,))
 
233
            else:
 
234
                # Binary files differ, just return
 
235
                return
 
236
 
 
237
        # If we got to here, we haven't written out the output of diff
 
238
        # do so now
 
239
        to_file.write(out)
 
240
        if rc not in (0, 1):
133
241
            # returns 1 if files differ; that's OK
134
242
            if rc < 0:
135
243
                msg = 'signal %d' % (-rc)
136
244
            else:
137
245
                msg = 'exit code %d' % rc
138
246
                
139
 
            raise BzrError('external diff failed with %s; command: %r' % (rc, diffcmd))
 
247
            raise errors.BzrError('external diff failed with %s; command: %r' 
 
248
                                  % (rc, diffcmd))
 
249
 
 
250
 
140
251
    finally:
141
252
        oldtmpf.close()                 # and delete
142
253
        newtmpf.close()
143
 
 
144
 
def show_diff(b, from_spec, specific_files, external_diff_options=None,
145
 
              revision2=None, output=None, b2=None):
146
 
    """Shortcut for showing the diff to the working tree.
147
 
 
148
 
    b
149
 
        Branch.
150
 
 
151
 
    revision
152
 
        None for 'basis tree', or otherwise the old revision to compare against.
 
254
        # Clean up. Warn in case the files couldn't be deleted
 
255
        # (in case windows still holds the file open, but not
 
256
        # if the files have already been deleted)
 
257
        try:
 
258
            os.remove(old_abspath)
 
259
        except OSError, e:
 
260
            if e.errno not in (errno.ENOENT,):
 
261
                warning('Failed to delete temporary file: %s %s',
 
262
                        old_abspath, e)
 
263
        try:
 
264
            os.remove(new_abspath)
 
265
        except OSError:
 
266
            if e.errno not in (errno.ENOENT,):
 
267
                warning('Failed to delete temporary file: %s %s',
 
268
                        new_abspath, e)
 
269
 
 
270
 
 
271
def diff_cmd_helper(tree, specific_files, external_diff_options, 
 
272
                    old_revision_spec=None, new_revision_spec=None,
 
273
                    revision_specs=None,
 
274
                    old_label='a/', new_label='b/'):
 
275
    """Helper for cmd_diff.
 
276
 
 
277
    :param tree:
 
278
        A WorkingTree
 
279
 
 
280
    :param specific_files:
 
281
        The specific files to compare, or None
 
282
 
 
283
    :param external_diff_options:
 
284
        If non-None, run an external diff, and pass it these options
 
285
 
 
286
    :param old_revision_spec:
 
287
        If None, use basis tree as old revision, otherwise use the tree for
 
288
        the specified revision. 
 
289
 
 
290
    :param new_revision_spec:
 
291
        If None, use working tree as new revision, otherwise use the tree for
 
292
        the specified revision.
153
293
    
 
294
    :param revision_specs: 
 
295
        Zero, one or two RevisionSpecs from the command line, saying what revisions 
 
296
        to compare.  This can be passed as an alternative to the old_revision_spec 
 
297
        and new_revision_spec parameters.
 
298
 
154
299
    The more general form is show_diff_trees(), where the caller
155
300
    supplies any two trees.
156
301
    """
157
 
    if output is None:
158
 
        import sys
159
 
        output = sys.stdout
160
 
 
161
 
    if from_spec is None:
162
 
        if b2 is None:
163
 
            old_tree = b.basis_tree()
164
 
        else:
165
 
            old_tree = b.working_tree()
166
 
    else:
167
 
        old_tree = b.revision_tree(from_spec.in_history(b).rev_id)
168
 
 
169
 
    if revision2 is None:
170
 
        if b2 is None:
171
 
            new_tree = b.working_tree()
172
 
        else:
173
 
            new_tree = b2.working_tree()
174
 
    else:
175
 
        new_tree = b.revision_tree(revision2.in_history(b).rev_id)
176
 
 
177
 
    return show_diff_trees(old_tree, new_tree, output, specific_files,
178
 
                           external_diff_options)
179
 
 
 
302
 
 
303
    # TODO: perhaps remove the old parameters old_revision_spec and
 
304
    # new_revision_spec, since this is only really for use from cmd_diff and
 
305
    # it now always passes through a sequence of revision_specs -- mbp
 
306
    # 20061221
 
307
 
 
308
    def spec_tree(spec):
 
309
        if tree:
 
310
            revision = spec.in_store(tree.branch)
 
311
        else:
 
312
            revision = spec.in_store(None)
 
313
        revision_id = revision.rev_id
 
314
        branch = revision.branch
 
315
        return branch.repository.revision_tree(revision_id)
 
316
 
 
317
    if revision_specs is not None:
 
318
        assert (old_revision_spec is None
 
319
                and new_revision_spec is None)
 
320
        if len(revision_specs) > 0:
 
321
            old_revision_spec = revision_specs[0]
 
322
        if len(revision_specs) > 1:
 
323
            new_revision_spec = revision_specs[1]
 
324
 
 
325
    if old_revision_spec is None:
 
326
        old_tree = tree.basis_tree()
 
327
    else:
 
328
        old_tree = spec_tree(old_revision_spec)
 
329
 
 
330
    if (new_revision_spec is None
 
331
        or new_revision_spec.spec is None):
 
332
        new_tree = tree
 
333
    else:
 
334
        new_tree = spec_tree(new_revision_spec)
 
335
 
 
336
    if new_tree is not tree:
 
337
        extra_trees = (tree,)
 
338
    else:
 
339
        extra_trees = None
 
340
 
 
341
    return show_diff_trees(old_tree, new_tree, sys.stdout, specific_files,
 
342
                           external_diff_options,
 
343
                           old_label=old_label, new_label=new_label,
 
344
                           extra_trees=extra_trees)
180
345
 
181
346
 
182
347
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None,
183
 
                    external_diff_options=None):
 
348
                    external_diff_options=None,
 
349
                    old_label='a/', new_label='b/',
 
350
                    extra_trees=None,
 
351
                    path_encoding='utf8'):
184
352
    """Show in text form the changes from one tree to another.
185
353
 
186
354
    to_files
188
356
 
189
357
    external_diff_options
190
358
        If set, use an external GNU diff and pass these options.
 
359
 
 
360
    extra_trees
 
361
        If set, more Trees to use for looking up file ids
 
362
 
 
363
    path_encoding
 
364
        If set, the path will be encoded as specified, otherwise is supposed
 
365
        to be utf8
191
366
    """
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.
 
367
    old_tree.lock_read()
 
368
    try:
 
369
        if extra_trees is not None:
 
370
            for tree in extra_trees:
 
371
                tree.lock_read()
 
372
        new_tree.lock_read()
 
373
        try:
 
374
            return _show_diff_trees(old_tree, new_tree, to_file,
 
375
                                    specific_files, external_diff_options,
 
376
                                    old_label=old_label, new_label=new_label,
 
377
                                    extra_trees=extra_trees,
 
378
                                    path_encoding=path_encoding)
 
379
        finally:
 
380
            new_tree.unlock()
 
381
            if extra_trees is not None:
 
382
                for tree in extra_trees:
 
383
                    tree.unlock()
 
384
    finally:
 
385
        old_tree.unlock()
 
386
 
 
387
 
 
388
def _show_diff_trees(old_tree, new_tree, to_file,
 
389
                     specific_files, external_diff_options, path_encoding,
 
390
                     old_label='a/', new_label='b/', extra_trees=None):
 
391
 
 
392
    # GNU Patch uses the epoch date to detect files that are being added
 
393
    # or removed in a diff.
 
394
    EPOCH_DATE = '1970-01-01 00:00:00 +0000'
201
395
 
202
396
    # TODO: Generation of pseudo-diffs for added/deleted files could
203
397
    # be usefully made into a much faster special case.
210
404
    else:
211
405
        diff_file = internal_diff
212
406
    
213
 
 
214
 
    delta = compare_trees(old_tree, new_tree, want_unchanged=False,
215
 
                          specific_files=specific_files)
 
407
    delta = new_tree.changes_from(old_tree,
 
408
        specific_files=specific_files,
 
409
        extra_trees=extra_trees, require_versioned=True)
216
410
 
217
411
    has_changes = 0
218
412
    for path, file_id, kind in delta.removed:
219
413
        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)
 
414
        path_encoded = path.encode(path_encoding, "replace")
 
415
        to_file.write("=== removed %s '%s'\n" % (kind, path_encoded))
 
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)
223
421
    for path, file_id, kind in delta.added:
224
422
        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, 
 
423
        path_encoded = path.encode(path_encoding, "replace")
 
424
        to_file.write("=== added %s '%s'\n" % (kind, path_encoded))
 
425
        old_name = '%s%s\t%s' % (old_label, path, EPOCH_DATE)
 
426
        new_name = '%s%s\t%s' % (new_label, path,
 
427
                                 _patch_header_date(new_tree, file_id, path))
 
428
        new_tree.inventory[file_id].diff(diff_file, new_name, new_tree,
 
429
                                         old_name, None, None, to_file, 
228
430
                                         reverse=True)
229
431
    for (old_path, new_path, file_id, kind,
230
432
         text_modified, meta_modified) in delta.renamed:
231
433
        has_changes = 1
232
434
        prop_str = get_prop_change(meta_modified)
233
 
        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,
 
435
        oldpath_encoded = old_path.encode(path_encoding, "replace")
 
436
        newpath_encoded = new_path.encode(path_encoding, "replace")
 
437
        to_file.write("=== renamed %s '%s' => '%s'%s\n" % (kind,
 
438
                            oldpath_encoded, newpath_encoded, prop_str))
 
439
        old_name = '%s%s\t%s' % (old_label, old_path,
 
440
                                 _patch_header_date(old_tree, file_id,
 
441
                                                    old_path))
 
442
        new_name = '%s%s\t%s' % (new_label, new_path,
 
443
                                 _patch_header_date(new_tree, file_id,
 
444
                                                    new_path))
 
445
        _maybe_diff_file_or_symlink(old_name, old_tree, file_id,
 
446
                                    new_name, new_tree,
237
447
                                    text_modified, kind, to_file, diff_file)
238
448
    for path, file_id, kind, text_modified, meta_modified in delta.modified:
239
449
        has_changes = 1
240
450
        prop_str = get_prop_change(meta_modified)
241
 
        print >>to_file, '=== modified %s %r%s' % (kind, path, prop_str)
 
451
        path_encoded = path.encode(path_encoding, "replace")
 
452
        to_file.write("=== modified %s '%s'%s\n" % (kind,
 
453
                            path_encoded, prop_str))
 
454
        # The file may be in a different location in the old tree (because
 
455
        # the containing dir was renamed, but the file itself was not)
 
456
        old_path = old_tree.id2path(file_id)
 
457
        old_name = '%s%s\t%s' % (old_label, old_path,
 
458
                                 _patch_header_date(old_tree, file_id, old_path))
 
459
        new_name = '%s%s\t%s' % (new_label, path,
 
460
                                 _patch_header_date(new_tree, file_id, path))
242
461
        if text_modified:
243
 
            _maybe_diff_file_or_symlink(old_label, path, old_tree, file_id,
244
 
                                        new_label, path, new_tree,
 
462
            _maybe_diff_file_or_symlink(old_name, old_tree, file_id,
 
463
                                        new_name, new_tree,
245
464
                                        True, kind, to_file, diff_file)
 
465
 
246
466
    return has_changes
 
467
 
 
468
 
 
469
def _patch_header_date(tree, file_id, path):
 
470
    """Returns a timestamp suitable for use in a patch header."""
 
471
    mtime = tree.get_file_mtime(file_id, path)
 
472
    assert mtime is not None, \
 
473
        "got an mtime of None for file-id %s, path %s in tree %s" % (
 
474
                file_id, path, tree)
 
475
    return timestamp.format_patch_date(mtime)
 
476
 
 
477
 
 
478
def _raise_if_nonexistent(paths, old_tree, new_tree):
 
479
    """Complain if paths are not in either inventory or tree.
 
480
 
 
481
    It's OK with the files exist in either tree's inventory, or 
 
482
    if they exist in the tree but are not versioned.
247
483
    
 
484
    This can be used by operations such as bzr status that can accept
 
485
    unknown or ignored files.
 
486
    """
 
487
    mutter("check paths: %r", paths)
 
488
    if not paths:
 
489
        return
 
490
    s = old_tree.filter_unversioned_files(paths)
 
491
    s = new_tree.filter_unversioned_files(s)
 
492
    s = [path for path in s if not new_tree.has_filename(path)]
 
493
    if s:
 
494
        raise errors.PathsDoNotExist(sorted(s))
 
495
 
248
496
 
249
497
def get_prop_change(meta_modified):
250
498
    if meta_modified:
253
501
        return  ""
254
502
 
255
503
 
256
 
def _maybe_diff_file_or_symlink(old_label, old_path, old_tree, file_id,
257
 
                                new_label, new_path, new_tree, text_modified,
 
504
def _maybe_diff_file_or_symlink(old_path, old_tree, file_id,
 
505
                                new_path, new_tree, text_modified,
258
506
                                kind, to_file, diff_file):
259
507
    if text_modified:
260
508
        new_entry = new_tree.inventory[file_id]
261
509
        old_tree.inventory[file_id].diff(diff_file,
262
 
                                         old_label + old_path, old_tree,
263
 
                                         new_label + new_path, new_entry, 
 
510
                                         old_path, old_tree,
 
511
                                         new_path, new_entry, 
264
512
                                         new_tree, to_file)