2
# -*- coding: UTF-8 -*-
1
# Copyright (C) 2004, 2005, 2006 Canonical Ltd.
4
3
# This program is free software; you can redistribute it and/or modify
5
4
# it under the terms of the GNU General Public License as published by
6
5
# the Free Software Foundation; either version 2 of the License, or
7
6
# (at your option) any later version.
9
8
# This program is distributed in the hope that it will be useful,
10
9
# but WITHOUT ANY WARRANTY; without even the implied warranty of
11
10
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12
11
# GNU General Public License for more details.
14
13
# You should have received a copy of the GNU General Public License
15
14
# along with this program; if not, write to the Free Software
16
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
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.
21
from bzrlib.lazy_import import lazy_import
22
lazy_import(globals(), """
37
# compatability - plugins import compare_trees from diff!!!
38
# deprecated as of 0.10
39
from bzrlib.delta import compare_trees
40
from bzrlib.symbol_versioning import (
44
from bzrlib.trace import mutter, warning
47
# TODO: Rather than building a changeset object, we should probably
48
# invoke callbacks on an object. That object can either accumulate a
49
# list, write them out directly, etc etc.
51
def internal_diff(old_filename, oldlines, new_filename, newlines, to_file,
52
allow_binary=False, sequence_matcher=None,
53
path_encoding='utf8'):
54
# FIXME: difflib is wrong if there is no trailing newline.
55
# The syntax used by patch seems to be "\ No newline at
56
# end of file" following the last diff line from that
57
# file. This is not trivial to insert into the
58
# unified_diff output and it might be better to just fix
59
# or replace that function.
61
# In the meantime we at least make sure the patch isn't
65
# Special workaround for Python2.3, where difflib fails if
66
# both sequences are empty.
67
if not oldlines and not newlines:
70
if allow_binary is False:
71
textfile.check_text_lines(oldlines)
72
textfile.check_text_lines(newlines)
74
if sequence_matcher is None:
75
sequence_matcher = patiencediff.PatienceSequenceMatcher
76
ud = patiencediff.unified_diff(oldlines, newlines,
77
fromfile=old_filename.encode(path_encoding),
78
tofile=new_filename.encode(path_encoding),
79
sequencematcher=sequence_matcher)
82
# work-around for difflib being too smart for its own good
83
# if /dev/null is "1,0", patch won't recognize it as /dev/null
85
ud[2] = ud[2].replace('-1,0', '-0,0')
87
ud[2] = ud[2].replace('+1,0', '+0,0')
88
# work around for difflib emitting random spaces after the label
89
ud[0] = ud[0][:-2] + '\n'
90
ud[1] = ud[1][:-2] + '\n'
94
if not line.endswith('\n'):
95
to_file.write("\n\\ No newline at end of file\n")
99
def _spawn_external_diff(diffcmd, capture_errors=True):
100
"""Spawn the externall diff process, and return the child handle.
102
:param diffcmd: The command list to spawn
103
:param capture_errors: Capture stderr as well as setting LANG=C
104
and LC_ALL=C. This lets us read and understand the output of diff,
105
and respond to any errors.
106
:return: A Popen object.
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
147
old_tree = b.basis_tree()
109
# construct minimal environment
111
path = os.environ.get('PATH')
114
env['LANGUAGE'] = 'C' # on win32 only LANGUAGE has effect
117
stderr = subprocess.PIPE
149
old_tree = b.revision_tree(b.lookup_revision(revision))
123
pipe = subprocess.Popen(diffcmd,
124
stdin=subprocess.PIPE,
125
stdout=subprocess.PIPE,
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()
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.
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 errors.BzrError(
216
'external diff failed with exit code 2'
217
' when run with LANG=C and LC_ALL=C,'
218
' but not when run natively: %r' % (diffcmd,))
220
first_line = lang_c_out.split('\n', 1)[0]
221
# Starting with diffutils 2.8.4 the word "binary" was dropped.
222
m = re.match('^(binary )?files.*differ$', first_line, re.I)
224
raise errors.BzrError('external diff failed with exit code 2;'
225
' command: %r' % (diffcmd,))
227
# Binary files differ, just return
230
# If we got to here, we haven't written out the output of diff
234
# returns 1 if files differ; that's OK
236
msg = 'signal %d' % (-rc)
238
msg = 'exit code %d' % rc
240
raise errors.BzrError('external diff failed with %s; command: %r'
245
oldtmpf.close() # and delete
247
# Clean up. Warn in case the files couldn't be deleted
248
# (in case windows still holds the file open, but not
249
# if the files have already been deleted)
251
os.remove(old_abspath)
253
if e.errno not in (errno.ENOENT,):
254
warning('Failed to delete temporary file: %s %s',
257
os.remove(new_abspath)
259
if e.errno not in (errno.ENOENT,):
260
warning('Failed to delete temporary file: %s %s',
264
@deprecated_function(zero_eight)
265
def show_diff(b, from_spec, specific_files, external_diff_options=None,
266
revision2=None, output=None, b2=None):
267
"""Shortcut for showing the diff to the working tree.
269
Please use show_diff_trees instead.
275
None for 'basis tree', or otherwise the old revision to compare against.
277
The more general form is show_diff_trees(), where the caller
278
supplies any two trees.
283
if from_spec is None:
284
old_tree = b.bzrdir.open_workingtree()
286
old_tree = old_tree = old_tree.basis_tree()
288
old_tree = b.repository.revision_tree(from_spec.in_history(b).rev_id)
290
if revision2 is None:
292
new_tree = b.bzrdir.open_workingtree()
294
new_tree = b2.bzrdir.open_workingtree()
296
new_tree = b.repository.revision_tree(revision2.in_history(b).rev_id)
298
return show_diff_trees(old_tree, new_tree, output, specific_files,
299
external_diff_options)
302
def diff_cmd_helper(tree, specific_files, external_diff_options,
303
old_revision_spec=None, new_revision_spec=None,
305
old_label='a/', new_label='b/'):
306
"""Helper for cmd_diff.
311
:param specific_files:
312
The specific files to compare, or None
314
:param external_diff_options:
315
If non-None, run an external diff, and pass it these options
317
:param old_revision_spec:
318
If None, use basis tree as old revision, otherwise use the tree for
319
the specified revision.
321
:param new_revision_spec:
322
If None, use working tree as new revision, otherwise use the tree for
323
the specified revision.
325
:param revision_specs:
326
Zero, one or two RevisionSpecs from the command line, saying what revisions
327
to compare. This can be passed as an alternative to the old_revision_spec
328
and new_revision_spec parameters.
330
The more general form is show_diff_trees(), where the caller
331
supplies any two trees.
334
# TODO: perhaps remove the old parameters old_revision_spec and
335
# new_revision_spec, since this is only really for use from cmd_diff and
336
# it now always passes through a sequence of revision_specs -- mbp
341
revision = spec.in_store(tree.branch)
343
revision = spec.in_store(None)
344
revision_id = revision.rev_id
345
branch = revision.branch
346
return branch.repository.revision_tree(revision_id)
348
if revision_specs is not None:
349
assert (old_revision_spec is None
350
and new_revision_spec is None)
351
if len(revision_specs) > 0:
352
old_revision_spec = revision_specs[0]
353
if len(revision_specs) > 1:
354
new_revision_spec = revision_specs[1]
356
if old_revision_spec is None:
357
old_tree = tree.basis_tree()
359
old_tree = spec_tree(old_revision_spec)
361
if (new_revision_spec is None
362
or new_revision_spec.spec is None):
365
new_tree = spec_tree(new_revision_spec)
367
if new_tree is not tree:
368
extra_trees = (tree,)
372
return show_diff_trees(old_tree, new_tree, sys.stdout, specific_files,
373
external_diff_options,
374
old_label=old_label, new_label=new_label,
375
extra_trees=extra_trees)
378
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None,
379
external_diff_options=None,
380
old_label='a/', new_label='b/',
382
"""Show in text form the changes from one tree to another.
385
If set, include only changes to these files.
387
external_diff_options
388
If set, use an external GNU diff and pass these options.
391
If set, more Trees to use for looking up file ids
395
if extra_trees is not None:
396
for tree in extra_trees:
400
return _show_diff_trees(old_tree, new_tree, to_file,
401
specific_files, external_diff_options,
402
old_label=old_label, new_label=new_label,
403
extra_trees=extra_trees)
406
if extra_trees is not None:
407
for tree in extra_trees:
413
def _show_diff_trees(old_tree, new_tree, to_file,
414
specific_files, external_diff_options,
415
old_label='a/', new_label='b/', extra_trees=None):
417
# GNU Patch uses the epoch date to detect files that are being added
418
# or removed in a diff.
419
EPOCH_DATE = '1970-01-01 00:00:00 +0000'
162
421
# TODO: Generation of pseudo-diffs for added/deleted files could
163
422
# be usefully made into a much faster special case.
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()
424
if external_diff_options:
425
assert isinstance(external_diff_options, basestring)
426
opts = external_diff_options.split()
427
def diff_file(olab, olines, nlab, nlines, to_file):
428
external_diff(olab, olines, nlab, nlines, to_file, opts)
430
diff_file = internal_diff
432
delta = new_tree.changes_from(old_tree,
433
specific_files=specific_files,
434
extra_trees=extra_trees, require_versioned=True)
437
for path, file_id, kind in delta.removed:
439
print >>to_file, '=== removed %s %r' % (kind, path.encode('utf8'))
440
old_name = '%s%s\t%s' % (old_label, path,
441
_patch_header_date(old_tree, file_id, path))
442
new_name = '%s%s\t%s' % (new_label, path, EPOCH_DATE)
443
old_tree.inventory[file_id].diff(diff_file, old_name, old_tree,
444
new_name, None, None, to_file)
445
for path, file_id, kind in delta.added:
447
print >>to_file, '=== added %s %r' % (kind, path.encode('utf8'))
448
old_name = '%s%s\t%s' % (old_label, path, EPOCH_DATE)
449
new_name = '%s%s\t%s' % (new_label, path,
450
_patch_header_date(new_tree, file_id, path))
451
new_tree.inventory[file_id].diff(diff_file, new_name, new_tree,
452
old_name, None, None, to_file,
454
for (old_path, new_path, file_id, kind,
455
text_modified, meta_modified) in delta.renamed:
457
prop_str = get_prop_change(meta_modified)
458
print >>to_file, '=== renamed %s %r => %r%s' % (
459
kind, old_path.encode('utf8'),
460
new_path.encode('utf8'), prop_str)
461
old_name = '%s%s\t%s' % (old_label, old_path,
462
_patch_header_date(old_tree, file_id,
464
new_name = '%s%s\t%s' % (new_label, new_path,
465
_patch_header_date(new_tree, file_id,
467
_maybe_diff_file_or_symlink(old_name, old_tree, file_id,
469
text_modified, kind, to_file, diff_file)
470
for path, file_id, kind, text_modified, meta_modified in delta.modified:
472
prop_str = get_prop_change(meta_modified)
473
print >>to_file, '=== modified %s %r%s' % (kind, path.encode('utf8'), prop_str)
474
old_name = '%s%s\t%s' % (old_label, path,
475
_patch_header_date(old_tree, file_id, path))
476
new_name = '%s%s\t%s' % (new_label, path,
477
_patch_header_date(new_tree, file_id, path))
479
_maybe_diff_file_or_symlink(old_name, old_tree, file_id,
481
True, kind, to_file, diff_file)
486
def _patch_header_date(tree, file_id, path):
487
"""Returns a timestamp suitable for use in a patch header."""
488
return timestamp.format_patch_date(tree.get_file_mtime(file_id, path))
491
def _raise_if_nonexistent(paths, old_tree, new_tree):
492
"""Complain if paths are not in either inventory or tree.
494
It's OK with the files exist in either tree's inventory, or
495
if they exist in the tree but are not versioned.
497
This can be used by operations such as bzr status that can accept
498
unknown or ignored files.
500
mutter("check paths: %r", paths)
503
s = old_tree.filter_unversioned_files(paths)
504
s = new_tree.filter_unversioned_files(s)
505
s = [path for path in s if not new_tree.has_filename(path)]
507
raise errors.PathsDoNotExist(sorted(s))
510
def get_prop_change(meta_modified):
512
return " (properties changed)"
517
def _maybe_diff_file_or_symlink(old_path, old_tree, file_id,
518
new_path, new_tree, text_modified,
519
kind, to_file, diff_file):
521
new_entry = new_tree.inventory[file_id]
522
old_tree.inventory[file_id].diff(diff_file,