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:
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
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
147
old_tree = b.basis_tree()
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()
149
old_tree = b.revision_tree(b.lookup_revision(revision))
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
path_encoding='utf8'):
352
"""Show in text form the changes from one tree to another.
355
If set, include only changes to these files.
357
external_diff_options
358
If set, use an external GNU diff and pass these options.
361
If set, more Trees to use for looking up file ids
364
If set, the path will be encoded as specified, otherwise is supposed
369
if extra_trees is not None:
370
for tree in extra_trees:
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)
381
if extra_trees is not None:
382
for tree in extra_trees:
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):
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'
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.
396
162
# TODO: Generation of pseudo-diffs for added/deleted files could
397
163
# be usefully made into a much faster special case.
399
if external_diff_options:
400
assert isinstance(external_diff_options, basestring)
401
opts = external_diff_options.split()
402
def diff_file(olab, olines, nlab, nlines, to_file):
403
external_diff(olab, olines, nlab, nlines, to_file, opts)
405
diff_file = internal_diff
407
delta = new_tree.changes_from(old_tree,
408
specific_files=specific_files,
409
extra_trees=extra_trees, require_versioned=True)
412
for path, file_id, kind in delta.removed:
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)
421
for path, file_id, kind in delta.added:
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,
431
for (old_path, new_path, file_id, kind,
432
text_modified, meta_modified) in delta.renamed:
434
prop_str = get_prop_change(meta_modified)
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,
442
new_name = '%s%s\t%s' % (new_label, new_path,
443
_patch_header_date(new_tree, file_id,
445
_maybe_diff_file_or_symlink(old_name, old_tree, file_id,
447
text_modified, kind, to_file, diff_file)
448
for path, file_id, kind, text_modified, meta_modified in delta.modified:
450
prop_str = get_prop_change(meta_modified)
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))
462
_maybe_diff_file_or_symlink(old_name, old_tree, file_id,
464
True, kind, to_file, diff_file)
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" % (
475
return timestamp.format_patch_date(mtime)
478
def _raise_if_nonexistent(paths, old_tree, new_tree):
479
"""Complain if paths are not in either inventory or tree.
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.
484
This can be used by operations such as bzr status that can accept
485
unknown or ignored files.
487
mutter("check paths: %r", paths)
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)]
494
raise errors.PathsDoNotExist(sorted(s))
497
def get_prop_change(meta_modified):
499
return " (properties changed)"
504
def _maybe_diff_file_or_symlink(old_path, old_tree, file_id,
505
new_path, new_tree, text_modified,
506
kind, to_file, diff_file):
508
new_entry = new_tree.inventory[file_id]
509
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
