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.
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()
24
from bzrlib.delta import compare_trees
25
from bzrlib.errors import BzrError
26
import bzrlib.errors as errors
28
from bzrlib.patiencediff import unified_diff
29
import bzrlib.patiencediff
30
from bzrlib.symbol_versioning import (deprecated_function,
32
from bzrlib.textfile import check_text_lines
33
from bzrlib.trace import mutter, warning
36
# TODO: Rather than building a changeset object, we should probably
37
# invoke callbacks on an object. That object can either accumulate a
38
# list, write them out directly, etc etc.
40
def internal_diff(old_filename, oldlines, new_filename, newlines, to_file,
41
allow_binary=False, sequence_matcher=None,
42
path_encoding='utf8'):
43
# FIXME: difflib is wrong if there is no trailing newline.
44
# The syntax used by patch seems to be "\ No newline at
45
# end of file" following the last diff line from that
46
# file. This is not trivial to insert into the
47
# unified_diff output and it might be better to just fix
48
# or replace that function.
50
# In the meantime we at least make sure the patch isn't
54
# Special workaround for Python2.3, where difflib fails if
55
# both sequences are empty.
56
if not oldlines and not newlines:
59
if allow_binary is False:
60
check_text_lines(oldlines)
61
check_text_lines(newlines)
63
if sequence_matcher is None:
64
sequence_matcher = bzrlib.patiencediff.PatienceSequenceMatcher
65
ud = unified_diff(oldlines, newlines,
66
fromfile=old_filename.encode(path_encoding),
67
tofile=new_filename.encode(path_encoding),
68
sequencematcher=sequence_matcher)
71
# work-around for difflib being too smart for its own good
72
# if /dev/null is "1,0", patch won't recognize it as /dev/null
74
ud[2] = ud[2].replace('-1,0', '-0,0')
76
ud[2] = ud[2].replace('+1,0', '+0,0')
77
# work around for difflib emitting random spaces after the label
78
ud[0] = ud[0][:-2] + '\n'
79
ud[1] = ud[1][:-2] + '\n'
83
if not line.endswith('\n'):
84
to_file.write("\n\\ No newline at end of file\n")
88
def external_diff(old_filename, oldlines, new_filename, newlines, to_file,
90
"""Display a diff by calling out to the external diff program."""
91
if hasattr(to_file, 'fileno'):
95
out_file = subprocess.PIPE
98
# make sure our own output is properly ordered before the diff
101
oldtmp_fd, old_abspath = tempfile.mkstemp(prefix='bzr-diff-old-')
102
newtmp_fd, new_abspath = tempfile.mkstemp(prefix='bzr-diff-new-')
103
oldtmpf = os.fdopen(oldtmp_fd, 'wb')
104
newtmpf = os.fdopen(newtmp_fd, 'wb')
107
# TODO: perhaps a special case for comparing to or from the empty
108
# sequence; can just use /dev/null on Unix
110
# TODO: if either of the files being compared already exists as a
111
# regular named file (e.g. in the working directory) then we can
112
# compare directly to that, rather than copying it.
114
oldtmpf.writelines(oldlines)
115
newtmpf.writelines(newlines)
123
'--label', old_filename,
125
'--label', new_filename,
130
# diff only allows one style to be specified; they don't override.
131
# note that some of these take optargs, and the optargs can be
132
# directly appended to the options.
133
# this is only an approximate parser; it doesn't properly understand
135
for s in ['-c', '-u', '-C', '-U',
140
'-y', '--side-by-side',
152
diffcmd.extend(diff_opts)
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()
149
old_tree = b.revision_tree(b.lookup_revision(revision))
155
pipe = subprocess.Popen(diffcmd,
156
stdin=subprocess.PIPE,
159
if e.errno == errno.ENOENT:
160
raise errors.NoDiff(str(e))
165
bzrlib.osutils.pumpfile(pipe.stdout, to_file)
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.
168
if rc != 0 and rc != 1:
169
# returns 1 if files differ; that's OK
171
msg = 'signal %d' % (-rc)
173
msg = 'exit code %d' % rc
175
raise BzrError('external diff failed with %s; command: %r' % (rc, diffcmd))
177
oldtmpf.close() # and delete
179
# Clean up. Warn in case the files couldn't be deleted
180
# (in case windows still holds the file open, but not
181
# if the files have already been deleted)
183
os.remove(old_abspath)
185
if e.errno not in (errno.ENOENT,):
186
warning('Failed to delete temporary file: %s %s',
189
os.remove(new_abspath)
191
if e.errno not in (errno.ENOENT,):
192
warning('Failed to delete temporary file: %s %s',
196
@deprecated_function(zero_eight)
197
def show_diff(b, from_spec, specific_files, external_diff_options=None,
198
revision2=None, output=None, b2=None):
199
"""Shortcut for showing the diff to the working tree.
201
Please use show_diff_trees instead.
207
None for 'basis tree', or otherwise the old revision to compare against.
209
The more general form is show_diff_trees(), where the caller
210
supplies any two trees.
215
if from_spec is None:
216
old_tree = b.bzrdir.open_workingtree()
218
old_tree = old_tree = old_tree.basis_tree()
220
old_tree = b.repository.revision_tree(from_spec.in_history(b).rev_id)
222
if revision2 is None:
224
new_tree = b.bzrdir.open_workingtree()
226
new_tree = b2.bzrdir.open_workingtree()
228
new_tree = b.repository.revision_tree(revision2.in_history(b).rev_id)
230
return show_diff_trees(old_tree, new_tree, output, specific_files,
231
external_diff_options)
234
def diff_cmd_helper(tree, specific_files, external_diff_options,
235
old_revision_spec=None, new_revision_spec=None,
236
old_label='a/', new_label='b/'):
237
"""Helper for cmd_diff.
243
The specific files to compare, or None
245
external_diff_options
246
If non-None, run an external diff, and pass it these options
249
If None, use basis tree as old revision, otherwise use the tree for
250
the specified revision.
253
If None, use working tree as new revision, otherwise use the tree for
254
the specified revision.
256
The more general form is show_diff_trees(), where the caller
257
supplies any two trees.
261
revision_id = spec.in_store(tree.branch).rev_id
262
return tree.branch.repository.revision_tree(revision_id)
263
if old_revision_spec is None:
264
old_tree = tree.basis_tree()
266
old_tree = spec_tree(old_revision_spec)
268
if new_revision_spec is None:
271
new_tree = spec_tree(new_revision_spec)
273
return show_diff_trees(old_tree, new_tree, sys.stdout, specific_files,
274
external_diff_options,
275
old_label=old_label, new_label=new_label)
278
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None,
279
external_diff_options=None,
280
old_label='a/', new_label='b/'):
281
"""Show in text form the changes from one tree to another.
284
If set, include only changes to these files.
286
external_diff_options
287
If set, use an external GNU diff and pass these options.
293
return _show_diff_trees(old_tree, new_tree, to_file,
294
specific_files, external_diff_options,
295
old_label=old_label, new_label=new_label)
302
def _show_diff_trees(old_tree, new_tree, to_file,
303
specific_files, external_diff_options,
304
old_label='a/', new_label='b/' ):
306
# GNU Patch uses the epoch date to detect files that are being added
307
# or removed in a diff.
308
EPOCH_DATE = '1970-01-01 00:00:00 +0000'
162
310
# TODO: Generation of pseudo-diffs for added/deleted files could
163
311
# 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()
313
_raise_if_doubly_unversioned(specific_files, old_tree, new_tree)
315
if external_diff_options:
316
assert isinstance(external_diff_options, basestring)
317
opts = external_diff_options.split()
318
def diff_file(olab, olines, nlab, nlines, to_file):
319
external_diff(olab, olines, nlab, nlines, to_file, opts)
321
diff_file = internal_diff
323
delta = compare_trees(old_tree, new_tree, want_unchanged=False,
324
specific_files=specific_files)
327
for path, file_id, kind in delta.removed:
329
print >>to_file, '=== removed %s %r' % (kind, path.encode('utf8'))
330
old_name = '%s%s\t%s' % (old_label, path,
331
_patch_header_date(old_tree, file_id, path))
332
new_name = '%s%s\t%s' % (new_label, path, EPOCH_DATE)
333
old_tree.inventory[file_id].diff(diff_file, old_name, old_tree,
334
new_name, None, None, to_file)
335
for path, file_id, kind in delta.added:
337
print >>to_file, '=== added %s %r' % (kind, path.encode('utf8'))
338
old_name = '%s%s\t%s' % (old_label, path, EPOCH_DATE)
339
new_name = '%s%s\t%s' % (new_label, path,
340
_patch_header_date(new_tree, file_id, path))
341
new_tree.inventory[file_id].diff(diff_file, new_name, new_tree,
342
old_name, None, None, to_file,
344
for (old_path, new_path, file_id, kind,
345
text_modified, meta_modified) in delta.renamed:
347
prop_str = get_prop_change(meta_modified)
348
print >>to_file, '=== renamed %s %r => %r%s' % (
349
kind, old_path.encode('utf8'),
350
new_path.encode('utf8'), prop_str)
351
old_name = '%s%s\t%s' % (old_label, old_path,
352
_patch_header_date(old_tree, file_id,
354
new_name = '%s%s\t%s' % (new_label, new_path,
355
_patch_header_date(new_tree, file_id,
357
_maybe_diff_file_or_symlink(old_name, old_tree, file_id,
359
text_modified, kind, to_file, diff_file)
360
for path, file_id, kind, text_modified, meta_modified in delta.modified:
362
prop_str = get_prop_change(meta_modified)
363
print >>to_file, '=== modified %s %r%s' % (kind, path.encode('utf8'), prop_str)
364
old_name = '%s%s\t%s' % (old_label, path,
365
_patch_header_date(old_tree, file_id, path))
366
new_name = '%s%s\t%s' % (new_label, path,
367
_patch_header_date(new_tree, file_id, path))
369
_maybe_diff_file_or_symlink(old_name, old_tree, file_id,
371
True, kind, to_file, diff_file)
376
def _patch_header_date(tree, file_id, path):
377
"""Returns a timestamp suitable for use in a patch header."""
378
tm = time.gmtime(tree.get_file_mtime(file_id, path))
379
return time.strftime('%Y-%m-%d %H:%M:%S +0000', tm)
382
def _raise_if_doubly_unversioned(specific_files, old_tree, new_tree):
383
"""Complain if paths are not versioned in either tree."""
384
if not specific_files:
386
old_unversioned = old_tree.filter_unversioned_files(specific_files)
387
new_unversioned = new_tree.filter_unversioned_files(specific_files)
388
unversioned = old_unversioned.intersection(new_unversioned)
390
raise errors.PathsNotVersionedError(sorted(unversioned))
393
def _raise_if_nonexistent(paths, old_tree, new_tree):
394
"""Complain if paths are not in either inventory or tree.
396
It's OK with the files exist in either tree's inventory, or
397
if they exist in the tree but are not versioned.
399
This can be used by operations such as bzr status that can accept
400
unknown or ignored files.
402
mutter("check paths: %r", paths)
405
s = old_tree.filter_unversioned_files(paths)
406
s = new_tree.filter_unversioned_files(s)
407
s = [path for path in s if not new_tree.has_filename(path)]
409
raise errors.PathsDoNotExist(sorted(s))
412
def get_prop_change(meta_modified):
414
return " (properties changed)"
419
def _maybe_diff_file_or_symlink(old_path, old_tree, file_id,
420
new_path, new_tree, text_modified,
421
kind, to_file, diff_file):
423
new_entry = new_tree.inventory[file_id]
424
old_tree.inventory[file_id].diff(diff_file,
added
removed
bzrlib/diff.py
