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
29
# compatability - plugins import compare_trees from diff!!!
30
# deprecated as of 0.10
31
from bzrlib.delta import compare_trees
32
from bzrlib.errors import BzrError
33
from bzrlib.patiencediff import unified_diff
34
import bzrlib.patiencediff
35
from bzrlib.symbol_versioning import (deprecated_function,
37
from bzrlib.textfile import check_text_lines
38
from bzrlib.trace import mutter, warning
41
# TODO: Rather than building a changeset object, we should probably
42
# invoke callbacks on an object. That object can either accumulate a
43
# list, write them out directly, etc etc.
45
def internal_diff(old_filename, oldlines, new_filename, newlines, to_file,
46
allow_binary=False, sequence_matcher=None,
47
path_encoding='utf8'):
48
# FIXME: difflib is wrong if there is no trailing newline.
49
# The syntax used by patch seems to be "\ No newline at
50
# end of file" following the last diff line from that
51
# file. This is not trivial to insert into the
52
# unified_diff output and it might be better to just fix
53
# or replace that function.
55
# In the meantime we at least make sure the patch isn't
59
# Special workaround for Python2.3, where difflib fails if
60
# both sequences are empty.
61
if not oldlines and not newlines:
20
from trace import mutter
21
from errors import BzrError
24
def diff_trees(old_tree, new_tree):
25
"""Compute diff between two trees.
27
They may be in different branches and may be working or historical
30
Yields a sequence of (state, id, old_name, new_name, kind).
31
Each filename and each id is listed only once.
34
## TODO: Compare files before diffing; only mention those that have changed
36
## TODO: Set nice names in the headers, maybe include diffstat
38
## TODO: Perhaps make this a generator rather than using
41
## TODO: Allow specifying a list of files to compare, rather than
42
## doing the whole tree? (Not urgent.)
44
## TODO: Allow diffing any two inventories, not just the
45
## current one against one. We mgiht need to specify two
46
## stores to look for the files if diffing two branches. That
47
## might imply this shouldn't be primarily a Branch method.
49
## XXX: This doesn't report on unknown files; that can be done
50
## from a separate method.
52
old_it = old_tree.list_files()
53
new_it = new_tree.list_files()
61
old_item = next(old_it)
62
new_item = next(new_it)
64
# We step through the two sorted iterators in parallel, trying to
67
while (old_item != None) or (new_item != None):
68
# OK, we still have some remaining on both, but they may be
71
old_name, old_class, old_kind, old_id = old_item
76
new_name, new_class, new_kind, new_id = new_item
80
mutter(" diff pairwise %r" % (old_item,))
81
mutter(" %r" % (new_item,))
84
# can't handle the old tree being a WorkingTree
85
assert old_class == 'V'
87
if new_item and (new_class != 'V'):
88
yield new_class, None, None, new_name, new_kind
89
new_item = next(new_it)
90
elif (not new_item) or (old_item and (old_name < new_name)):
91
mutter(" extra entry in old-tree sequence")
92
if new_tree.has_id(old_id):
93
# will be mentioned as renamed under new name
96
yield 'D', old_id, old_name, None, old_kind
97
old_item = next(old_it)
98
elif (not old_item) or (new_item and (new_name < old_name)):
99
mutter(" extra entry in new-tree sequence")
100
if old_tree.has_id(new_id):
101
yield 'R', new_id, old_tree.id2path(new_id), new_name, new_kind
103
yield 'A', new_id, None, new_name, new_kind
104
new_item = next(new_it)
105
elif old_id != new_id:
106
assert old_name == new_name
107
# both trees have a file of this name, but it is not the
108
# same file. in other words, the old filename has been
109
# overwritten by either a newly-added or a renamed file.
110
# (should we return something about the overwritten file?)
111
if old_tree.has_id(new_id):
112
# renaming, overlying a deleted file
113
yield 'R', new_id, old_tree.id2path(new_id), new_name, new_kind
115
yield 'A', new_id, None, new_name, new_kind
117
new_item = next(new_it)
118
old_item = next(old_it)
120
assert old_id == new_id
121
assert old_id != None
122
assert old_name == new_name
123
assert old_kind == new_kind
125
if old_kind == 'directory':
126
yield '.', new_id, old_name, new_name, new_kind
127
elif old_tree.get_file_size(old_id) != new_tree.get_file_size(old_id):
128
mutter(" file size has changed, must be different")
129
yield 'M', new_id, old_name, new_name, new_kind
130
elif old_tree.get_file_sha1(old_id) == new_tree.get_file_sha1(old_id):
131
mutter(" SHA1 indicates they're identical")
132
## assert compare_files(old_tree.get_file(i), new_tree.get_file(i))
133
yield '.', new_id, old_name, new_name, new_kind
135
mutter(" quick compare shows different")
136
yield 'M', new_id, old_name, new_name, new_kind
138
new_item = next(new_it)
139
old_item = next(old_it)
143
def show_diff(b, revision, file_list):
144
import difflib, sys, types
64
if allow_binary is False:
65
check_text_lines(oldlines)
66
check_text_lines(newlines)
68
if sequence_matcher is None:
69
sequence_matcher = bzrlib.patiencediff.PatienceSequenceMatcher
70
ud = unified_diff(oldlines, newlines,
71
fromfile=old_filename.encode(path_encoding),
72
tofile=new_filename.encode(path_encoding),
73
sequencematcher=sequence_matcher)
76
# work-around for difflib being too smart for its own good
77
# if /dev/null is "1,0", patch won't recognize it as /dev/null
79
ud[2] = ud[2].replace('-1,0', '-0,0')
81
ud[2] = ud[2].replace('+1,0', '+0,0')
82
# work around for difflib emitting random spaces after the label
83
ud[0] = ud[0][:-2] + '\n'
84
ud[1] = ud[1][:-2] + '\n'
88
if not line.endswith('\n'):
89
to_file.write("\n\\ No newline at end of file\n")
94
"""Set the env var LANG=C"""
95
osutils.set_or_unset_env('LANG', 'C')
96
osutils.set_or_unset_env('LC_ALL', None)
97
osutils.set_or_unset_env('LC_CTYPE', None)
98
osutils.set_or_unset_env('LANGUAGE', None)
101
def _spawn_external_diff(diffcmd, capture_errors=True):
102
"""Spawn the externall diff process, and return the child handle.
104
:param diffcmd: The command list to spawn
105
:param capture_errors: Capture stderr as well as setting LANG=C.
106
This lets us read and understand the output of diff, and respond
108
:return: A Popen object.
111
if sys.platform == 'win32':
112
# Win32 doesn't support preexec_fn, but that is
113
# okay, because it doesn't support LANG either.
116
preexec_fn = _set_lang_C
117
stderr = subprocess.PIPE
147
old_tree = b.basis_tree()
123
pipe = subprocess.Popen(diffcmd,
124
stdin=subprocess.PIPE,
125
stdout=subprocess.PIPE,
127
preexec_fn=preexec_fn)
129
if e.errno == errno.ENOENT:
130
raise errors.NoDiff(str(e))
136
def external_diff(old_filename, oldlines, new_filename, newlines, to_file,
138
"""Display a diff by calling out to the external diff program."""
139
# make sure our own output is properly ordered before the diff
142
oldtmp_fd, old_abspath = tempfile.mkstemp(prefix='bzr-diff-old-')
143
newtmp_fd, new_abspath = tempfile.mkstemp(prefix='bzr-diff-new-')
144
oldtmpf = os.fdopen(oldtmp_fd, 'wb')
145
newtmpf = os.fdopen(newtmp_fd, 'wb')
148
# TODO: perhaps a special case for comparing to or from the empty
149
# sequence; can just use /dev/null on Unix
151
# TODO: if either of the files being compared already exists as a
152
# regular named file (e.g. in the working directory) then we can
153
# compare directly to that, rather than copying it.
155
oldtmpf.writelines(oldlines)
156
newtmpf.writelines(newlines)
164
'--label', old_filename,
166
'--label', new_filename,
171
# diff only allows one style to be specified; they don't override.
172
# note that some of these take optargs, and the optargs can be
173
# directly appended to the options.
174
# this is only an approximate parser; it doesn't properly understand
176
for s in ['-c', '-u', '-C', '-U',
181
'-y', '--side-by-side',
193
diffcmd.extend(diff_opts)
195
pipe = _spawn_external_diff(diffcmd, capture_errors=True)
196
out,err = pipe.communicate()
149
old_tree = b.revision_tree(b.lookup_revision(revision))
199
# internal_diff() adds a trailing newline, add one here for consistency
202
# 'diff' gives retcode == 2 for all sorts of errors
203
# one of those is 'Binary files differ'.
204
# Bad options could also be the problem.
205
# 'Binary files' is not a real error, so we suppress that error.
208
# Since we got here, we want to make sure to give an i18n error
209
pipe = _spawn_external_diff(diffcmd, capture_errors=False)
210
out, err = pipe.communicate()
212
# Write out the new i18n diff response
213
to_file.write(out+'\n')
214
if pipe.returncode != 2:
215
raise BzrError('external diff failed with exit code 2'
216
' when run with LANG=C, but not when run'
217
' natively: %r' % (diffcmd,))
219
first_line = lang_c_out.split('\n', 1)[0]
220
# Starting with diffutils 2.8.4 the word "binary" was dropped.
221
m = re.match('^(binary )?files.*differ$', first_line, re.I)
223
raise BzrError('external diff failed with exit code 2;'
224
' command: %r' % (diffcmd,))
226
# Binary files differ, just return
229
# If we got to here, we haven't written out the output of diff
233
# returns 1 if files differ; that's OK
235
msg = 'signal %d' % (-rc)
237
msg = 'exit code %d' % rc
239
raise BzrError('external diff failed with %s; command: %r'
244
oldtmpf.close() # and delete
246
# Clean up. Warn in case the files couldn't be deleted
247
# (in case windows still holds the file open, but not
248
# if the files have already been deleted)
250
os.remove(old_abspath)
252
if e.errno not in (errno.ENOENT,):
253
warning('Failed to delete temporary file: %s %s',
256
os.remove(new_abspath)
258
if e.errno not in (errno.ENOENT,):
259
warning('Failed to delete temporary file: %s %s',
263
@deprecated_function(zero_eight)
264
def show_diff(b, from_spec, specific_files, external_diff_options=None,
265
revision2=None, output=None, b2=None):
266
"""Shortcut for showing the diff to the working tree.
268
Please use show_diff_trees instead.
274
None for 'basis tree', or otherwise the old revision to compare against.
276
The more general form is show_diff_trees(), where the caller
277
supplies any two trees.
282
if from_spec is None:
283
old_tree = b.bzrdir.open_workingtree()
285
old_tree = old_tree = old_tree.basis_tree()
287
old_tree = b.repository.revision_tree(from_spec.in_history(b).rev_id)
289
if revision2 is None:
291
new_tree = b.bzrdir.open_workingtree()
293
new_tree = b2.bzrdir.open_workingtree()
295
new_tree = b.repository.revision_tree(revision2.in_history(b).rev_id)
297
return show_diff_trees(old_tree, new_tree, output, specific_files,
298
external_diff_options)
301
def diff_cmd_helper(tree, specific_files, external_diff_options,
302
old_revision_spec=None, new_revision_spec=None,
303
old_label='a/', new_label='b/'):
304
"""Helper for cmd_diff.
310
The specific files to compare, or None
312
external_diff_options
313
If non-None, run an external diff, and pass it these options
316
If None, use basis tree as old revision, otherwise use the tree for
317
the specified revision.
320
If None, use working tree as new revision, otherwise use the tree for
321
the specified revision.
323
The more general form is show_diff_trees(), where the caller
324
supplies any two trees.
328
revision = spec.in_store(tree.branch)
330
revision = spec.in_store(None)
331
revision_id = revision.rev_id
332
branch = revision.branch
333
return branch.repository.revision_tree(revision_id)
334
if old_revision_spec is None:
335
old_tree = tree.basis_tree()
337
old_tree = spec_tree(old_revision_spec)
339
if new_revision_spec is None:
342
new_tree = spec_tree(new_revision_spec)
343
if new_tree is not tree:
344
extra_trees = (tree,)
348
return show_diff_trees(old_tree, new_tree, sys.stdout, specific_files,
349
external_diff_options,
350
old_label=old_label, new_label=new_label,
351
extra_trees=extra_trees)
354
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None,
355
external_diff_options=None,
356
old_label='a/', new_label='b/',
358
"""Show in text form the changes from one tree to another.
361
If set, include only changes to these files.
363
external_diff_options
364
If set, use an external GNU diff and pass these options.
367
If set, more Trees to use for looking up file ids
373
return _show_diff_trees(old_tree, new_tree, to_file,
374
specific_files, external_diff_options,
375
old_label=old_label, new_label=new_label,
376
extra_trees=extra_trees)
383
def _show_diff_trees(old_tree, new_tree, to_file,
384
specific_files, external_diff_options,
385
old_label='a/', new_label='b/', extra_trees=None):
387
# GNU Patch uses the epoch date to detect files that are being added
388
# or removed in a diff.
389
EPOCH_DATE = '1970-01-01 00:00:00 +0000'
151
new_tree = b.working_tree()
153
# TODO: Options to control putting on a prefix or suffix, perhaps as a format string
157
DEVNULL = '/dev/null'
158
# Windows users, don't panic about this filename -- it is a
159
# special signal to GNU patch that the file should be created or
160
# deleted respectively.
391
162
# TODO: Generation of pseudo-diffs for added/deleted files could
392
163
# be usefully made into a much faster special case.
394
if external_diff_options:
395
assert isinstance(external_diff_options, basestring)
396
opts = external_diff_options.split()
397
def diff_file(olab, olines, nlab, nlines, to_file):
398
external_diff(olab, olines, nlab, nlines, to_file, opts)
400
diff_file = internal_diff
402
delta = new_tree.changes_from(old_tree,
403
specific_files=specific_files,
404
extra_trees=extra_trees, require_versioned=True)
407
for path, file_id, kind in delta.removed:
409
print >>to_file, '=== removed %s %r' % (kind, path.encode('utf8'))
410
old_name = '%s%s\t%s' % (old_label, path,
411
_patch_header_date(old_tree, file_id, path))
412
new_name = '%s%s\t%s' % (new_label, path, EPOCH_DATE)
413
old_tree.inventory[file_id].diff(diff_file, old_name, old_tree,
414
new_name, None, None, to_file)
415
for path, file_id, kind in delta.added:
417
print >>to_file, '=== added %s %r' % (kind, path.encode('utf8'))
418
old_name = '%s%s\t%s' % (old_label, path, EPOCH_DATE)
419
new_name = '%s%s\t%s' % (new_label, path,
420
_patch_header_date(new_tree, file_id, path))
421
new_tree.inventory[file_id].diff(diff_file, new_name, new_tree,
422
old_name, None, None, to_file,
424
for (old_path, new_path, file_id, kind,
425
text_modified, meta_modified) in delta.renamed:
427
prop_str = get_prop_change(meta_modified)
428
print >>to_file, '=== renamed %s %r => %r%s' % (
429
kind, old_path.encode('utf8'),
430
new_path.encode('utf8'), prop_str)
431
old_name = '%s%s\t%s' % (old_label, old_path,
432
_patch_header_date(old_tree, file_id,
434
new_name = '%s%s\t%s' % (new_label, new_path,
435
_patch_header_date(new_tree, file_id,
437
_maybe_diff_file_or_symlink(old_name, old_tree, file_id,
439
text_modified, kind, to_file, diff_file)
440
for path, file_id, kind, text_modified, meta_modified in delta.modified:
442
prop_str = get_prop_change(meta_modified)
443
print >>to_file, '=== modified %s %r%s' % (kind, path.encode('utf8'), prop_str)
444
old_name = '%s%s\t%s' % (old_label, path,
445
_patch_header_date(old_tree, file_id, path))
446
new_name = '%s%s\t%s' % (new_label, path,
447
_patch_header_date(new_tree, file_id, path))
449
_maybe_diff_file_or_symlink(old_name, old_tree, file_id,
451
True, kind, to_file, diff_file)
456
def _patch_header_date(tree, file_id, path):
457
"""Returns a timestamp suitable for use in a patch header."""
458
tm = time.gmtime(tree.get_file_mtime(file_id, path))
459
return time.strftime('%Y-%m-%d %H:%M:%S +0000', tm)
462
def _raise_if_nonexistent(paths, old_tree, new_tree):
463
"""Complain if paths are not in either inventory or tree.
465
It's OK with the files exist in either tree's inventory, or
466
if they exist in the tree but are not versioned.
468
This can be used by operations such as bzr status that can accept
469
unknown or ignored files.
471
mutter("check paths: %r", paths)
474
s = old_tree.filter_unversioned_files(paths)
475
s = new_tree.filter_unversioned_files(s)
476
s = [path for path in s if not new_tree.has_filename(path)]
478
raise errors.PathsDoNotExist(sorted(s))
481
def get_prop_change(meta_modified):
483
return " (properties changed)"
488
def _maybe_diff_file_or_symlink(old_path, old_tree, file_id,
489
new_path, new_tree, text_modified,
490
kind, to_file, diff_file):
492
new_entry = new_tree.inventory[file_id]
493
old_tree.inventory[file_id].diff(diff_file,
165
# TODO: Better to return them in sorted order I think.
168
file_list = [b.relpath(f) for f in file_list]
170
# FIXME: If given a file list, compare only those files rather
171
# than comparing everything and then throwing stuff away.
173
for file_state, fid, old_name, new_name, kind in diff_trees(old_tree, new_tree):
175
if file_list and (new_name not in file_list):
178
# Don't show this by default; maybe do it if an option is passed
179
# idlabel = ' {%s}' % fid
182
# FIXME: Something about the diff format makes patch unhappy
183
# with newly-added files.
185
def diffit(oldlines, newlines, **kw):
187
# FIXME: difflib is wrong if there is no trailing newline.
188
# The syntax used by patch seems to be "\ No newline at
189
# end of file" following the last diff line from that
190
# file. This is not trivial to insert into the
191
# unified_diff output and it might be better to just fix
192
# or replace that function.
194
# In the meantime we at least make sure the patch isn't
198
# Special workaround for Python2.3, where difflib fails if
199
# both sequences are empty.
200
if not oldlines and not newlines:
205
if oldlines and (oldlines[-1][-1] != '\n'):
208
if newlines and (newlines[-1][-1] != '\n'):
212
ud = difflib.unified_diff(oldlines, newlines, **kw)
213
sys.stdout.writelines(ud)
215
print "\\ No newline at end of file"
216
sys.stdout.write('\n')
218
if file_state in ['.', '?', 'I']:
220
elif file_state == 'A':
221
print '*** added %s %r' % (kind, new_name)
224
new_tree.get_file(fid).readlines(),
226
tofile=new_label + new_name + idlabel)
227
elif file_state == 'D':
228
assert isinstance(old_name, types.StringTypes)
229
print '*** deleted %s %r' % (kind, old_name)
231
diffit(old_tree.get_file(fid).readlines(), [],
232
fromfile=old_label + old_name + idlabel,
234
elif file_state in ['M', 'R']:
235
if file_state == 'M':
236
assert kind == 'file'
237
assert old_name == new_name
238
print '*** modified %s %r' % (kind, new_name)
239
elif file_state == 'R':
240
print '*** renamed %s %r => %r' % (kind, old_name, new_name)
243
diffit(old_tree.get_file(fid).readlines(),
244
new_tree.get_file(fid).readlines(),
245
fromfile=old_label + old_name + idlabel,
246
tofile=new_label + new_name)
248
raise BzrError("can't represent state %s {%s}" % (file_state, fid))
253
"""Describes changes from one tree to another.
262
(oldpath, newpath, id)
266
A path may occur in more than one list if it was e.g. deleted
267
under an old id and renamed into place in a new id.
269
Files are listed in either modified or renamed, not both. In
270
other words, renamed files may also be modified.
279
def compare_inventories(old_inv, new_inv):
280
"""Return a TreeDelta object describing changes between inventories.
282
This only describes changes in the shape of the tree, not the
285
This is an alternative to diff_trees() and should probably
286
eventually replace it.
288
old_ids = old_inv.id_set()
289
new_ids = new_inv.id_set()
292
delta.removed = [(old_inv.id2path(fid), fid) for fid in (old_ids - new_ids)]
295
delta.added = [(new_inv.id2path(fid), fid) for fid in (new_ids - old_ids)]
298
for fid in old_ids & new_ids:
299
old_ie = old_inv[fid]
300
new_ie = new_inv[fid]
301
old_path = old_inv.id2path(fid)
302
new_path = new_inv.id2path(fid)
304
if old_path != new_path:
305
delta.renamed.append((old_path, new_path, fid))
306
elif old_ie.text_sha1 != new_ie.text_sha1:
307
delta.modified.append((new_path, fid))
309
delta.modified.sort()
added
removed
bzrlib/diff.py
