1
# Copyright (C) 2004, 2005, 2006 Canonical Ltd.
2
# -*- coding: UTF-8 -*-
3
4
# This program is free software; you can redistribute it and/or modify
4
5
# it under the terms of the GNU General Public License as published by
5
6
# the Free Software Foundation; either version 2 of the License, or
6
7
# (at your option) any later version.
8
9
# This program is distributed in the hope that it will be useful,
9
10
# but WITHOUT ANY WARRANTY; without even the implied warranty of
10
11
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
11
12
# GNU General Public License for more details.
13
14
# You should have received a copy of the GNU General Public License
14
15
# along with this program; if not, write to the Free Software
15
16
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
22
from bzrlib.lazy_import import lazy_import
23
lazy_import(globals(), """
38
from bzrlib.symbol_versioning import (
41
from bzrlib.trace import mutter, warning
44
# TODO: Rather than building a changeset object, we should probably
45
# invoke callbacks on an object. That object can either accumulate a
46
# list, write them out directly, etc etc.
49
class _PrematchedMatcher(difflib.SequenceMatcher):
50
"""Allow SequenceMatcher operations to use predetermined blocks"""
52
def __init__(self, matching_blocks):
53
difflib.SequenceMatcher(self, None, None)
54
self.matching_blocks = matching_blocks
58
def internal_diff(old_filename, oldlines, new_filename, newlines, to_file,
59
allow_binary=False, sequence_matcher=None,
60
path_encoding='utf8'):
61
# FIXME: difflib is wrong if there is no trailing newline.
62
# The syntax used by patch seems to be "\ No newline at
63
# end of file" following the last diff line from that
64
# file. This is not trivial to insert into the
65
# unified_diff output and it might be better to just fix
66
# or replace that function.
68
# In the meantime we at least make sure the patch isn't
72
# Special workaround for Python2.3, where difflib fails if
73
# both sequences are empty.
74
if not oldlines and not newlines:
77
if allow_binary is False:
78
textfile.check_text_lines(oldlines)
79
textfile.check_text_lines(newlines)
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)
89
# work-around for difflib being too smart for its own good
90
# if /dev/null is "1,0", patch won't recognize it as /dev/null
92
ud[2] = ud[2].replace('-1,0', '-0,0')
94
ud[2] = ud[2].replace('+1,0', '+0,0')
95
# work around for difflib emitting random spaces after the label
96
ud[0] = ud[0][:-2] + '\n'
97
ud[1] = ud[1][:-2] + '\n'
101
if not line.endswith('\n'):
102
to_file.write("\n\\ No newline at end of file\n")
106
def _spawn_external_diff(diffcmd, capture_errors=True):
107
"""Spawn the externall diff process, and return the child handle.
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.
116
# construct minimal environment
118
path = os.environ.get('PATH')
121
env['LANGUAGE'] = 'C' # on win32 only LANGUAGE has effect
124
stderr = subprocess.PIPE
130
pipe = subprocess.Popen(diffcmd,
131
stdin=subprocess.PIPE,
132
stdout=subprocess.PIPE,
136
if e.errno == errno.ENOENT:
137
raise errors.NoDiff(str(e))
143
def external_diff(old_filename, oldlines, new_filename, newlines, to_file,
145
"""Display a diff by calling out to the external diff program."""
146
# make sure our own output is properly ordered before the diff
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')
155
# TODO: perhaps a special case for comparing to or from the empty
156
# sequence; can just use /dev/null on Unix
158
# TODO: if either of the files being compared already exists as a
159
# regular named file (e.g. in the working directory) then we can
160
# compare directly to that, rather than copying it.
162
oldtmpf.writelines(oldlines)
163
newtmpf.writelines(newlines)
171
'--label', old_filename,
173
'--label', new_filename,
178
# diff only allows one style to be specified; they don't override.
179
# note that some of these take optargs, and the optargs can be
180
# directly appended to the options.
181
# this is only an approximate parser; it doesn't properly understand
183
for s in ['-c', '-u', '-C', '-U',
188
'-y', '--side-by-side',
200
diffcmd.extend(diff_opts)
202
pipe = _spawn_external_diff(diffcmd, capture_errors=True)
203
out,err = pipe.communicate()
206
# internal_diff() adds a trailing newline, add one here for consistency
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.
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()
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,))
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)
231
raise errors.BzrError('external diff failed with exit code 2;'
232
' command: %r' % (diffcmd,))
234
# Binary files differ, just return
237
# If we got to here, we haven't written out the output of diff
241
# returns 1 if files differ; that's OK
243
msg = 'signal %d' % (-rc)
245
msg = 'exit code %d' % rc
247
raise errors.BzrError('external diff failed with %s; command: %r'
252
oldtmpf.close() # and delete
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)
258
os.remove(old_abspath)
260
if e.errno not in (errno.ENOENT,):
261
warning('Failed to delete temporary file: %s %s',
264
os.remove(new_abspath)
266
if e.errno not in (errno.ENOENT,):
267
warning('Failed to delete temporary file: %s %s',
271
def diff_cmd_helper(tree, specific_files, external_diff_options,
272
old_revision_spec=None, new_revision_spec=None,
274
old_label='a/', new_label='b/'):
275
"""Helper for cmd_diff.
280
:param specific_files:
281
The specific files to compare, or None
283
:param external_diff_options:
284
If non-None, run an external diff, and pass it these options
286
:param old_revision_spec:
287
If None, use basis tree as old revision, otherwise use the tree for
288
the specified revision.
290
:param new_revision_spec:
291
If None, use working tree as new revision, otherwise use the tree for
292
the specified revision.
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.
299
The more general form is show_diff_trees(), where the caller
300
supplies any two trees.
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
310
revision = spec.in_store(tree.branch)
312
revision = spec.in_store(None)
313
revision_id = revision.rev_id
314
branch = revision.branch
315
return branch.repository.revision_tree(revision_id)
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]
325
if old_revision_spec is None:
326
old_tree = tree.basis_tree()
328
old_tree = spec_tree(old_revision_spec)
330
if (new_revision_spec is None
331
or new_revision_spec.spec is None):
334
new_tree = spec_tree(new_revision_spec)
336
if new_tree is not tree:
337
extra_trees = (tree,)
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)
347
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None,
348
external_diff_options=None,
349
old_label='a/', new_label='b/',
351
"""Show in text form the changes from one tree to another.
354
If set, include only changes to these files.
356
external_diff_options
357
If set, use an external GNU diff and pass these options.
360
If set, more Trees to use for looking up file ids
364
if extra_trees is not None:
365
for tree in extra_trees:
369
return _show_diff_trees(old_tree, new_tree, to_file,
370
specific_files, external_diff_options,
371
old_label=old_label, new_label=new_label,
372
extra_trees=extra_trees)
375
if extra_trees is not None:
376
for tree in extra_trees:
382
def _show_diff_trees(old_tree, new_tree, to_file,
383
specific_files, external_diff_options,
384
old_label='a/', new_label='b/', extra_trees=None):
386
# GNU Patch uses the epoch date to detect files that are being added
387
# or removed in a diff.
388
EPOCH_DATE = '1970-01-01 00:00:00 +0000'
390
# TODO: Generation of pseudo-diffs for added/deleted files could
391
# be usefully made into a much faster special case.
393
if external_diff_options:
394
assert isinstance(external_diff_options, basestring)
395
opts = external_diff_options.split()
396
def diff_file(olab, olines, nlab, nlines, to_file):
397
external_diff(olab, olines, nlab, nlines, to_file, opts)
399
diff_file = internal_diff
401
delta = new_tree.changes_from(old_tree,
402
specific_files=specific_files,
403
extra_trees=extra_trees, require_versioned=True)
406
for path, file_id, kind in delta.removed:
408
print >>to_file, "=== removed %s '%s'" % (kind, path.encode('utf8'))
409
old_name = '%s%s\t%s' % (old_label, path,
410
_patch_header_date(old_tree, file_id, path))
411
new_name = '%s%s\t%s' % (new_label, path, EPOCH_DATE)
412
old_tree.inventory[file_id].diff(diff_file, old_name, old_tree,
413
new_name, None, None, to_file)
414
for path, file_id, kind in delta.added:
416
print >>to_file, "=== added %s '%s'" % (kind, path.encode('utf8'))
417
old_name = '%s%s\t%s' % (old_label, path, EPOCH_DATE)
418
new_name = '%s%s\t%s' % (new_label, path,
419
_patch_header_date(new_tree, file_id, path))
420
new_tree.inventory[file_id].diff(diff_file, new_name, new_tree,
421
old_name, None, None, to_file,
423
for (old_path, new_path, file_id, kind,
424
text_modified, meta_modified) in delta.renamed:
426
prop_str = get_prop_change(meta_modified)
427
print >>to_file, "=== renamed %s '%s' => %r%s" % (
428
kind, old_path.encode('utf8'),
429
new_path.encode('utf8'), prop_str)
430
old_name = '%s%s\t%s' % (old_label, old_path,
431
_patch_header_date(old_tree, file_id,
433
new_name = '%s%s\t%s' % (new_label, new_path,
434
_patch_header_date(new_tree, file_id,
436
_maybe_diff_file_or_symlink(old_name, old_tree, file_id,
438
text_modified, kind, to_file, diff_file)
439
for path, file_id, kind, text_modified, meta_modified in delta.modified:
441
prop_str = get_prop_change(meta_modified)
442
print >>to_file, "=== modified %s '%s'%s" % (kind, path.encode('utf8'),
444
# The file may be in a different location in the old tree (because
445
# the containing dir was renamed, but the file itself was not)
446
old_path = old_tree.id2path(file_id)
447
old_name = '%s%s\t%s' % (old_label, old_path,
448
_patch_header_date(old_tree, file_id, old_path))
449
new_name = '%s%s\t%s' % (new_label, path,
450
_patch_header_date(new_tree, file_id, path))
452
_maybe_diff_file_or_symlink(old_name, old_tree, file_id,
454
True, kind, to_file, diff_file)
459
def _patch_header_date(tree, file_id, path):
460
"""Returns a timestamp suitable for use in a patch header."""
461
mtime = tree.get_file_mtime(file_id, path)
462
assert mtime is not None, \
463
"got an mtime of None for file-id %s, path %s in tree %s" % (
465
return timestamp.format_patch_date(mtime)
468
def _raise_if_nonexistent(paths, old_tree, new_tree):
469
"""Complain if paths are not in either inventory or tree.
471
It's OK with the files exist in either tree's inventory, or
472
if they exist in the tree but are not versioned.
474
This can be used by operations such as bzr status that can accept
475
unknown or ignored files.
477
mutter("check paths: %r", paths)
480
s = old_tree.filter_unversioned_files(paths)
481
s = new_tree.filter_unversioned_files(s)
482
s = [path for path in s if not new_tree.has_filename(path)]
484
raise errors.PathsDoNotExist(sorted(s))
487
def get_prop_change(meta_modified):
489
return " (properties changed)"
494
def _maybe_diff_file_or_symlink(old_path, old_tree, file_id,
495
new_path, new_tree, text_modified,
496
kind, to_file, diff_file):
498
new_entry = new_tree.inventory[file_id]
499
old_tree.inventory[file_id].diff(diff_file,
20
from trace import mutter
27
def diff_trees(old_tree, new_tree):
28
"""Compute diff between two trees.
30
They may be in different branches and may be working or historical
33
Yields a sequence of (state, id, old_name, new_name, kind).
34
Each filename and each id is listed only once.
37
## TODO: Compare files before diffing; only mention those that have changed
39
## TODO: Set nice names in the headers, maybe include diffstat
41
## TODO: Perhaps make this a generator rather than using
44
## TODO: Allow specifying a list of files to compare, rather than
45
## doing the whole tree? (Not urgent.)
47
## TODO: Allow diffing any two inventories, not just the
48
## current one against one. We mgiht need to specify two
49
## stores to look for the files if diffing two branches. That
50
## might imply this shouldn't be primarily a Branch method.
52
## XXX: This doesn't report on unknown files; that can be done
53
## from a separate method.
55
old_it = old_tree.list_files()
56
new_it = new_tree.list_files()
64
old_item = next(old_it)
65
new_item = next(new_it)
67
# We step through the two sorted iterators in parallel, trying to
70
while (old_item != None) or (new_item != None):
71
# OK, we still have some remaining on both, but they may be
74
old_name, old_class, old_kind, old_id = old_item
79
new_name, new_class, new_kind, new_id = new_item
83
mutter(" diff pairwise %r" % (old_item,))
84
mutter(" %r" % (new_item,))
87
# can't handle the old tree being a WorkingTree
88
assert old_class == 'V'
90
if new_item and (new_class != 'V'):
91
yield new_class, None, None, new_name, new_kind
92
new_item = next(new_it)
93
elif (not new_item) or (old_item and (old_name < new_name)):
94
mutter(" extra entry in old-tree sequence")
95
if new_tree.has_id(old_id):
96
# will be mentioned as renamed under new name
99
yield 'D', old_id, old_name, None, old_kind
100
old_item = next(old_it)
101
elif (not old_item) or (new_item and (new_name < old_name)):
102
mutter(" extra entry in new-tree sequence")
103
if old_tree.has_id(new_id):
104
yield 'R', new_id, old_tree.id2path(new_id), new_name, new_kind
106
yield 'A', new_id, None, new_name, new_kind
107
new_item = next(new_it)
108
elif old_id != new_id:
109
assert old_name == new_name
110
# both trees have a file of this name, but it is not the
111
# same file. in other words, the old filename has been
112
# overwritten by either a newly-added or a renamed file.
113
# (should we return something about the overwritten file?)
114
if old_tree.has_id(new_id):
115
# renaming, overlying a deleted file
116
yield 'R', new_id, old_tree.id2path(new_id), new_name, new_kind
118
yield 'A', new_id, None, new_name, new_kind
120
new_item = next(new_it)
121
old_item = next(old_it)
123
assert old_id == new_id
124
assert old_name == new_name
125
assert old_kind == new_kind
127
if old_kind == 'directory':
128
yield '.', new_id, old_name, new_name, new_kind
129
elif old_tree.get_file_size(old_id) != new_tree.get_file_size(old_id):
130
mutter(" file size has changed, must be different")
131
yield 'M', new_id, old_name, new_name, new_kind
132
elif old_tree.get_file_sha1(old_id) == new_tree.get_file_sha1(old_id):
133
mutter(" SHA1 indicates they're identical")
134
## assert compare_files(old_tree.get_file(i), new_tree.get_file(i))
135
yield '.', new_id, old_name, new_name, new_kind
137
mutter(" quick compare shows different")
138
yield 'M', new_id, old_name, new_name, new_kind
140
new_item = next(new_it)
141
old_item = next(old_it)
added
removed
bzrlib/diff.py
