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
25
# compatability - plugins import compare_trees from diff!!!
26
# deprecated as of 0.10
27
from bzrlib.delta import compare_trees
28
from bzrlib.errors import BzrError
29
import bzrlib.errors as errors
31
from bzrlib.patiencediff import unified_diff
32
import bzrlib.patiencediff
33
from bzrlib.symbol_versioning import (deprecated_function,
35
from bzrlib.textfile import check_text_lines
36
from bzrlib.trace import mutter, warning
39
# TODO: Rather than building a changeset object, we should probably
40
# invoke callbacks on an object. That object can either accumulate a
41
# list, write them out directly, etc etc.
43
def internal_diff(old_filename, oldlines, new_filename, newlines, to_file,
44
allow_binary=False, sequence_matcher=None,
45
path_encoding='utf8'):
46
# FIXME: difflib is wrong if there is no trailing newline.
47
# The syntax used by patch seems to be "\ No newline at
48
# end of file" following the last diff line from that
49
# file. This is not trivial to insert into the
50
# unified_diff output and it might be better to just fix
51
# or replace that function.
53
# In the meantime we at least make sure the patch isn't
57
# Special workaround for Python2.3, where difflib fails if
58
# both sequences are empty.
59
if not oldlines and not newlines:
62
if allow_binary is False:
63
check_text_lines(oldlines)
64
check_text_lines(newlines)
66
if sequence_matcher is None:
67
sequence_matcher = bzrlib.patiencediff.PatienceSequenceMatcher
68
ud = unified_diff(oldlines, newlines,
69
fromfile=old_filename.encode(path_encoding),
70
tofile=new_filename.encode(path_encoding),
71
sequencematcher=sequence_matcher)
74
# work-around for difflib being too smart for its own good
75
# if /dev/null is "1,0", patch won't recognize it as /dev/null
77
ud[2] = ud[2].replace('-1,0', '-0,0')
79
ud[2] = ud[2].replace('+1,0', '+0,0')
80
# work around for difflib emitting random spaces after the label
81
ud[0] = ud[0][:-2] + '\n'
82
ud[1] = ud[1][:-2] + '\n'
86
if not line.endswith('\n'):
87
to_file.write("\n\\ No newline at end of file\n")
92
"""Set the env var LANG=C"""
93
os.environ['LANG'] = 'C'
96
def _spawn_external_diff(diffcmd, capture_errors=True):
97
"""Spawn the externall diff process, and return the child handle.
99
:param diffcmd: The command list to spawn
100
:param capture_errors: Capture stderr as well as setting LANG=C.
101
This lets us read and understand the output of diff, and respond
103
:return: A Popen object.
106
preexec_fn = _set_lang_C
107
stderr = subprocess.PIPE
113
pipe = subprocess.Popen(diffcmd,
114
stdin=subprocess.PIPE,
115
stdout=subprocess.PIPE,
117
preexec_fn=preexec_fn)
119
if e.errno == errno.ENOENT:
120
raise errors.NoDiff(str(e))
126
def external_diff(old_filename, oldlines, new_filename, newlines, to_file,
128
"""Display a diff by calling out to the external diff program."""
129
# make sure our own output is properly ordered before the diff
132
oldtmp_fd, old_abspath = tempfile.mkstemp(prefix='bzr-diff-old-')
133
newtmp_fd, new_abspath = tempfile.mkstemp(prefix='bzr-diff-new-')
134
oldtmpf = os.fdopen(oldtmp_fd, 'wb')
135
newtmpf = os.fdopen(newtmp_fd, 'wb')
138
# TODO: perhaps a special case for comparing to or from the empty
139
# sequence; can just use /dev/null on Unix
141
# TODO: if either of the files being compared already exists as a
142
# regular named file (e.g. in the working directory) then we can
143
# compare directly to that, rather than copying it.
145
oldtmpf.writelines(oldlines)
146
newtmpf.writelines(newlines)
154
'--label', old_filename,
156
'--label', new_filename,
161
# diff only allows one style to be specified; they don't override.
162
# note that some of these take optargs, and the optargs can be
163
# directly appended to the options.
164
# this is only an approximate parser; it doesn't properly understand
166
for s in ['-c', '-u', '-C', '-U',
171
'-y', '--side-by-side',
183
diffcmd.extend(diff_opts)
185
pipe = _spawn_external_diff(diffcmd, capture_errors=True)
186
out,err = pipe.communicate()
189
# internal_diff() adds a trailing newline, add one here for consistency
192
# 'diff' gives retcode == 2 for all sorts of errors
193
# one of those is 'Binary files differ'.
194
# Bad options could also be the problem.
195
# 'Binary files' is not a real error, so we suppress that error
198
# Since we got here, we want to make sure to give an i18n error
199
pipe = _spawn_external_diff(diffcmd, capture_errors=False)
200
out, err = pipe.communicate()
202
# Write out the new i18n diff response
203
to_file.write(out+'\n')
204
if pipe.returncode != 2:
205
raise BzrError('external diff failed with exit code 2'
206
' when run with LANG=C, but not when run'
207
' natively: %r' % (diffcmd,))
209
first_line = lang_c_out.split('\n', 1)[0]
210
m = re.match('^binary files.*differ$', first_line, re.I)
212
raise BzrError('external diff failed with exit code 2;'
213
' command: %r' % (diffcmd,))
215
# Binary files differ, just return
218
# If we got to here, we haven't written out the output of diff
222
# returns 1 if files differ; that's OK
224
msg = 'signal %d' % (-rc)
226
msg = 'exit code %d' % rc
228
raise BzrError('external diff failed with %s; command: %r'
233
oldtmpf.close() # and delete
235
# Clean up. Warn in case the files couldn't be deleted
236
# (in case windows still holds the file open, but not
237
# if the files have already been deleted)
239
os.remove(old_abspath)
241
if e.errno not in (errno.ENOENT,):
242
warning('Failed to delete temporary file: %s %s',
245
os.remove(new_abspath)
247
if e.errno not in (errno.ENOENT,):
248
warning('Failed to delete temporary file: %s %s',
252
@deprecated_function(zero_eight)
253
def show_diff(b, from_spec, specific_files, external_diff_options=None,
254
revision2=None, output=None, b2=None):
255
"""Shortcut for showing the diff to the working tree.
257
Please use show_diff_trees instead.
263
None for 'basis tree', or otherwise the old revision to compare against.
265
The more general form is show_diff_trees(), where the caller
266
supplies any two trees.
271
if from_spec is None:
272
old_tree = b.bzrdir.open_workingtree()
274
old_tree = old_tree = old_tree.basis_tree()
276
old_tree = b.repository.revision_tree(from_spec.in_history(b).rev_id)
278
if revision2 is None:
280
new_tree = b.bzrdir.open_workingtree()
282
new_tree = b2.bzrdir.open_workingtree()
284
new_tree = b.repository.revision_tree(revision2.in_history(b).rev_id)
286
return show_diff_trees(old_tree, new_tree, output, specific_files,
287
external_diff_options)
290
def diff_cmd_helper(tree, specific_files, external_diff_options,
291
old_revision_spec=None, new_revision_spec=None,
292
old_label='a/', new_label='b/'):
293
"""Helper for cmd_diff.
299
The specific files to compare, or None
301
external_diff_options
302
If non-None, run an external diff, and pass it these options
305
If None, use basis tree as old revision, otherwise use the tree for
306
the specified revision.
309
If None, use working tree as new revision, otherwise use the tree for
310
the specified revision.
312
The more general form is show_diff_trees(), where the caller
313
supplies any two trees.
317
revision = spec.in_store(tree.branch)
319
revision = spec.in_store(None)
320
revision_id = revision.rev_id
321
branch = revision.branch
322
return branch.repository.revision_tree(revision_id)
323
if old_revision_spec is None:
324
old_tree = tree.basis_tree()
326
old_tree = spec_tree(old_revision_spec)
328
if new_revision_spec is None:
331
new_tree = spec_tree(new_revision_spec)
332
if new_tree is not tree:
333
extra_trees = (tree,)
337
return show_diff_trees(old_tree, new_tree, sys.stdout, specific_files,
338
external_diff_options,
339
old_label=old_label, new_label=new_label,
340
extra_trees=extra_trees)
343
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None,
344
external_diff_options=None,
345
old_label='a/', new_label='b/',
347
"""Show in text form the changes from one tree to another.
350
If set, include only changes to these files.
352
external_diff_options
353
If set, use an external GNU diff and pass these options.
356
If set, more Trees to use for looking up file ids
362
return _show_diff_trees(old_tree, new_tree, to_file,
363
specific_files, external_diff_options,
364
old_label=old_label, new_label=new_label,
365
extra_trees=extra_trees)
372
def _show_diff_trees(old_tree, new_tree, to_file,
373
specific_files, external_diff_options,
374
old_label='a/', new_label='b/', extra_trees=None):
376
# GNU Patch uses the epoch date to detect files that are being added
377
# or removed in a diff.
378
EPOCH_DATE = '1970-01-01 00:00:00 +0000'
380
# TODO: Generation of pseudo-diffs for added/deleted files could
381
# be usefully made into a much faster special case.
383
if external_diff_options:
384
assert isinstance(external_diff_options, basestring)
385
opts = external_diff_options.split()
386
def diff_file(olab, olines, nlab, nlines, to_file):
387
external_diff(olab, olines, nlab, nlines, to_file, opts)
389
diff_file = internal_diff
391
delta = new_tree.changes_from(old_tree,
392
specific_files=specific_files,
393
extra_trees=extra_trees, require_versioned=True)
396
for path, file_id, kind in delta.removed:
398
print >>to_file, '=== removed %s %r' % (kind, path.encode('utf8'))
399
old_name = '%s%s\t%s' % (old_label, path,
400
_patch_header_date(old_tree, file_id, path))
401
new_name = '%s%s\t%s' % (new_label, path, EPOCH_DATE)
402
old_tree.inventory[file_id].diff(diff_file, old_name, old_tree,
403
new_name, None, None, to_file)
404
for path, file_id, kind in delta.added:
406
print >>to_file, '=== added %s %r' % (kind, path.encode('utf8'))
407
old_name = '%s%s\t%s' % (old_label, path, EPOCH_DATE)
408
new_name = '%s%s\t%s' % (new_label, path,
409
_patch_header_date(new_tree, file_id, path))
410
new_tree.inventory[file_id].diff(diff_file, new_name, new_tree,
411
old_name, None, None, to_file,
413
for (old_path, new_path, file_id, kind,
414
text_modified, meta_modified) in delta.renamed:
416
prop_str = get_prop_change(meta_modified)
417
print >>to_file, '=== renamed %s %r => %r%s' % (
418
kind, old_path.encode('utf8'),
419
new_path.encode('utf8'), prop_str)
420
old_name = '%s%s\t%s' % (old_label, old_path,
421
_patch_header_date(old_tree, file_id,
423
new_name = '%s%s\t%s' % (new_label, new_path,
424
_patch_header_date(new_tree, file_id,
426
_maybe_diff_file_or_symlink(old_name, old_tree, file_id,
428
text_modified, kind, to_file, diff_file)
429
for path, file_id, kind, text_modified, meta_modified in delta.modified:
431
prop_str = get_prop_change(meta_modified)
432
print >>to_file, '=== modified %s %r%s' % (kind, path.encode('utf8'), prop_str)
433
old_name = '%s%s\t%s' % (old_label, path,
434
_patch_header_date(old_tree, file_id, path))
435
new_name = '%s%s\t%s' % (new_label, path,
436
_patch_header_date(new_tree, file_id, path))
438
_maybe_diff_file_or_symlink(old_name, old_tree, file_id,
440
True, kind, to_file, diff_file)
445
def _patch_header_date(tree, file_id, path):
446
"""Returns a timestamp suitable for use in a patch header."""
447
tm = time.gmtime(tree.get_file_mtime(file_id, path))
448
return time.strftime('%Y-%m-%d %H:%M:%S +0000', tm)
451
def _raise_if_nonexistent(paths, old_tree, new_tree):
452
"""Complain if paths are not in either inventory or tree.
454
It's OK with the files exist in either tree's inventory, or
455
if they exist in the tree but are not versioned.
457
This can be used by operations such as bzr status that can accept
458
unknown or ignored files.
460
mutter("check paths: %r", paths)
463
s = old_tree.filter_unversioned_files(paths)
464
s = new_tree.filter_unversioned_files(s)
465
s = [path for path in s if not new_tree.has_filename(path)]
467
raise errors.PathsDoNotExist(sorted(s))
470
def get_prop_change(meta_modified):
472
return " (properties changed)"
477
def _maybe_diff_file_or_symlink(old_path, old_tree, file_id,
478
new_path, new_tree, text_modified,
479
kind, to_file, diff_file):
481
new_entry = new_tree.inventory[file_id]
482
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_id != None
125
assert old_name == new_name
126
assert old_kind == new_kind
128
if old_kind == 'directory':
129
yield '.', new_id, old_name, new_name, new_kind
130
elif old_tree.get_file_size(old_id) != new_tree.get_file_size(old_id):
131
mutter(" file size has changed, must be different")
132
yield 'M', new_id, old_name, new_name, new_kind
133
elif old_tree.get_file_sha1(old_id) == new_tree.get_file_sha1(old_id):
134
mutter(" SHA1 indicates they're identical")
135
## assert compare_files(old_tree.get_file(i), new_tree.get_file(i))
136
yield '.', new_id, old_name, new_name, new_kind
138
mutter(" quick compare shows different")
139
yield 'M', new_id, old_name, new_name, new_kind
141
new_item = next(new_it)
142
old_item = next(old_it)
added
removed
bzrlib/diff.py
