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
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(), """
36
# compatability - plugins import compare_trees from diff!!!
37
# deprecated as of 0.10
38
from bzrlib.delta import compare_trees
39
from bzrlib.symbol_versioning import (
43
from bzrlib.trace import mutter, warning
46
# TODO: Rather than building a changeset object, we should probably
47
# invoke callbacks on an object. That object can either accumulate a
48
# list, write them out directly, etc etc.
50
def internal_diff(old_filename, oldlines, new_filename, newlines, to_file,
51
allow_binary=False, sequence_matcher=None,
52
path_encoding='utf8'):
53
# FIXME: difflib is wrong if there is no trailing newline.
54
# The syntax used by patch seems to be "\ No newline at
55
# end of file" following the last diff line from that
56
# file. This is not trivial to insert into the
57
# unified_diff output and it might be better to just fix
58
# or replace that function.
60
# In the meantime we at least make sure the patch isn't
64
# Special workaround for Python2.3, where difflib fails if
65
# both sequences are empty.
66
if not oldlines and not newlines:
69
if allow_binary is False:
70
textfile.check_text_lines(oldlines)
71
textfile.check_text_lines(newlines)
73
if sequence_matcher is None:
74
sequence_matcher = patiencediff.PatienceSequenceMatcher
75
ud = patiencediff.unified_diff(oldlines, newlines,
76
fromfile=old_filename.encode(path_encoding),
77
tofile=new_filename.encode(path_encoding),
78
sequencematcher=sequence_matcher)
81
# work-around for difflib being too smart for its own good
82
# if /dev/null is "1,0", patch won't recognize it as /dev/null
84
ud[2] = ud[2].replace('-1,0', '-0,0')
86
ud[2] = ud[2].replace('+1,0', '+0,0')
87
# work around for difflib emitting random spaces after the label
88
ud[0] = ud[0][:-2] + '\n'
89
ud[1] = ud[1][:-2] + '\n'
93
if not line.endswith('\n'):
94
to_file.write("\n\\ No newline at end of file\n")
99
"""Set the env vars LANG=C and LC_ALL=C."""
100
osutils.set_or_unset_env('LANG', 'C')
101
osutils.set_or_unset_env('LC_ALL', 'C')
102
osutils.set_or_unset_env('LC_CTYPE', None)
103
osutils.set_or_unset_env('LANGUAGE', None)
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.
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):
147
old_tree = b.basis_tree()
116
if sys.platform == 'win32':
117
# Win32 doesn't support preexec_fn, but that is
118
# okay, because it doesn't support LANG and LC_ALL either.
121
preexec_fn = _set_lang_C
122
stderr = subprocess.PIPE
149
old_tree = b.revision_tree(b.lookup_revision(revision))
128
pipe = subprocess.Popen(diffcmd,
129
stdin=subprocess.PIPE,
130
stdout=subprocess.PIPE,
132
preexec_fn=preexec_fn)
134
if e.errno == errno.ENOENT:
135
raise errors.NoDiff(str(e))
141
def external_diff(old_filename, oldlines, new_filename, newlines, to_file,
143
"""Display a diff by calling out to the external diff program."""
144
# make sure our own output is properly ordered before the diff
147
oldtmp_fd, old_abspath = tempfile.mkstemp(prefix='bzr-diff-old-')
148
newtmp_fd, new_abspath = tempfile.mkstemp(prefix='bzr-diff-new-')
149
oldtmpf = os.fdopen(oldtmp_fd, 'wb')
150
newtmpf = os.fdopen(newtmp_fd, 'wb')
153
# TODO: perhaps a special case for comparing to or from the empty
154
# sequence; can just use /dev/null on Unix
156
# TODO: if either of the files being compared already exists as a
157
# regular named file (e.g. in the working directory) then we can
158
# compare directly to that, rather than copying it.
160
oldtmpf.writelines(oldlines)
161
newtmpf.writelines(newlines)
169
'--label', old_filename,
171
'--label', new_filename,
176
# diff only allows one style to be specified; they don't override.
177
# note that some of these take optargs, and the optargs can be
178
# directly appended to the options.
179
# this is only an approximate parser; it doesn't properly understand
181
for s in ['-c', '-u', '-C', '-U',
186
'-y', '--side-by-side',
198
diffcmd.extend(diff_opts)
200
pipe = _spawn_external_diff(diffcmd, capture_errors=True)
201
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.
204
# internal_diff() adds a trailing newline, add one here for consistency
207
# 'diff' gives retcode == 2 for all sorts of errors
208
# one of those is 'Binary files differ'.
209
# Bad options could also be the problem.
210
# 'Binary files' is not a real error, so we suppress that error.
213
# Since we got here, we want to make sure to give an i18n error
214
pipe = _spawn_external_diff(diffcmd, capture_errors=False)
215
out, err = pipe.communicate()
217
# Write out the new i18n diff response
218
to_file.write(out+'\n')
219
if pipe.returncode != 2:
220
raise errors.BzrError(
221
'external diff failed with exit code 2'
222
' when run with LANG=C and LC_ALL=C,'
223
' but not when run natively: %r' % (diffcmd,))
225
first_line = lang_c_out.split('\n', 1)[0]
226
# Starting with diffutils 2.8.4 the word "binary" was dropped.
227
m = re.match('^(binary )?files.*differ$', first_line, re.I)
229
raise errors.BzrError('external diff failed with exit code 2;'
230
' command: %r' % (diffcmd,))
232
# Binary files differ, just return
235
# If we got to here, we haven't written out the output of diff
239
# returns 1 if files differ; that's OK
241
msg = 'signal %d' % (-rc)
243
msg = 'exit code %d' % rc
245
raise errors.BzrError('external diff failed with %s; command: %r'
250
oldtmpf.close() # and delete
252
# Clean up. Warn in case the files couldn't be deleted
253
# (in case windows still holds the file open, but not
254
# if the files have already been deleted)
256
os.remove(old_abspath)
258
if e.errno not in (errno.ENOENT,):
259
warning('Failed to delete temporary file: %s %s',
262
os.remove(new_abspath)
264
if e.errno not in (errno.ENOENT,):
265
warning('Failed to delete temporary file: %s %s',
269
@deprecated_function(zero_eight)
270
def show_diff(b, from_spec, specific_files, external_diff_options=None,
271
revision2=None, output=None, b2=None):
272
"""Shortcut for showing the diff to the working tree.
274
Please use show_diff_trees instead.
280
None for 'basis tree', or otherwise the old revision to compare against.
282
The more general form is show_diff_trees(), where the caller
283
supplies any two trees.
288
if from_spec is None:
289
old_tree = b.bzrdir.open_workingtree()
291
old_tree = old_tree = old_tree.basis_tree()
293
old_tree = b.repository.revision_tree(from_spec.in_history(b).rev_id)
295
if revision2 is None:
297
new_tree = b.bzrdir.open_workingtree()
299
new_tree = b2.bzrdir.open_workingtree()
301
new_tree = b.repository.revision_tree(revision2.in_history(b).rev_id)
303
return show_diff_trees(old_tree, new_tree, output, specific_files,
304
external_diff_options)
307
def diff_cmd_helper(tree, specific_files, external_diff_options,
308
old_revision_spec=None, new_revision_spec=None,
309
old_label='a/', new_label='b/'):
310
"""Helper for cmd_diff.
316
The specific files to compare, or None
318
external_diff_options
319
If non-None, run an external diff, and pass it these options
322
If None, use basis tree as old revision, otherwise use the tree for
323
the specified revision.
326
If None, use working tree as new revision, otherwise use the tree for
327
the specified revision.
329
The more general form is show_diff_trees(), where the caller
330
supplies any two trees.
334
revision = spec.in_store(tree.branch)
336
revision = spec.in_store(None)
337
revision_id = revision.rev_id
338
branch = revision.branch
339
return branch.repository.revision_tree(revision_id)
340
if old_revision_spec is None:
341
old_tree = tree.basis_tree()
343
old_tree = spec_tree(old_revision_spec)
345
if new_revision_spec is None:
348
new_tree = spec_tree(new_revision_spec)
349
if new_tree is not tree:
350
extra_trees = (tree,)
354
return show_diff_trees(old_tree, new_tree, sys.stdout, specific_files,
355
external_diff_options,
356
old_label=old_label, new_label=new_label,
357
extra_trees=extra_trees)
360
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None,
361
external_diff_options=None,
362
old_label='a/', new_label='b/',
364
"""Show in text form the changes from one tree to another.
367
If set, include only changes to these files.
369
external_diff_options
370
If set, use an external GNU diff and pass these options.
373
If set, more Trees to use for looking up file ids
379
return _show_diff_trees(old_tree, new_tree, to_file,
380
specific_files, external_diff_options,
381
old_label=old_label, new_label=new_label,
382
extra_trees=extra_trees)
389
def _show_diff_trees(old_tree, new_tree, to_file,
390
specific_files, external_diff_options,
391
old_label='a/', new_label='b/', extra_trees=None):
393
# GNU Patch uses the epoch date to detect files that are being added
394
# or removed in a diff.
395
EPOCH_DATE = '1970-01-01 00:00:00 +0000'
162
397
# TODO: Generation of pseudo-diffs for added/deleted files could
163
398
# 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
bailout("can't represent state %s {%s}" % (file_state, fid))
400
if external_diff_options:
401
assert isinstance(external_diff_options, basestring)
402
opts = external_diff_options.split()
403
def diff_file(olab, olines, nlab, nlines, to_file):
404
external_diff(olab, olines, nlab, nlines, to_file, opts)
406
diff_file = internal_diff
408
delta = new_tree.changes_from(old_tree,
409
specific_files=specific_files,
410
extra_trees=extra_trees, require_versioned=True)
413
for path, file_id, kind in delta.removed:
415
print >>to_file, '=== removed %s %r' % (kind, path.encode('utf8'))
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
print >>to_file, '=== added %s %r' % (kind, path.encode('utf8'))
424
old_name = '%s%s\t%s' % (old_label, path, EPOCH_DATE)
425
new_name = '%s%s\t%s' % (new_label, path,
426
_patch_header_date(new_tree, file_id, path))
427
new_tree.inventory[file_id].diff(diff_file, new_name, new_tree,
428
old_name, None, None, to_file,
430
for (old_path, new_path, file_id, kind,
431
text_modified, meta_modified) in delta.renamed:
433
prop_str = get_prop_change(meta_modified)
434
print >>to_file, '=== renamed %s %r => %r%s' % (
435
kind, old_path.encode('utf8'),
436
new_path.encode('utf8'), prop_str)
437
old_name = '%s%s\t%s' % (old_label, old_path,
438
_patch_header_date(old_tree, file_id,
440
new_name = '%s%s\t%s' % (new_label, new_path,
441
_patch_header_date(new_tree, file_id,
443
_maybe_diff_file_or_symlink(old_name, old_tree, file_id,
445
text_modified, kind, to_file, diff_file)
446
for path, file_id, kind, text_modified, meta_modified in delta.modified:
448
prop_str = get_prop_change(meta_modified)
449
print >>to_file, '=== modified %s %r%s' % (kind, path.encode('utf8'), prop_str)
450
old_name = '%s%s\t%s' % (old_label, path,
451
_patch_header_date(old_tree, file_id, path))
452
new_name = '%s%s\t%s' % (new_label, path,
453
_patch_header_date(new_tree, file_id, path))
455
_maybe_diff_file_or_symlink(old_name, old_tree, file_id,
457
True, kind, to_file, diff_file)
462
def _patch_header_date(tree, file_id, path):
463
"""Returns a timestamp suitable for use in a patch header."""
464
tm = time.gmtime(tree.get_file_mtime(file_id, path))
465
return time.strftime('%Y-%m-%d %H:%M:%S +0000', tm)
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,
added
removed
bzrlib/diff.py
