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
24
# compatability - plugins import compare_trees from diff!!!
25
# deprecated as of 0.10
26
from bzrlib.delta import compare_trees
27
from bzrlib.errors import BzrError
28
import bzrlib.errors as errors
30
from bzrlib.patiencediff import unified_diff
31
import bzrlib.patiencediff
32
from bzrlib.symbol_versioning import (deprecated_function,
34
from bzrlib.textfile import check_text_lines
35
from bzrlib.trace import mutter, warning
38
# TODO: Rather than building a changeset object, we should probably
39
# invoke callbacks on an object. That object can either accumulate a
40
# list, write them out directly, etc etc.
42
def internal_diff(old_filename, oldlines, new_filename, newlines, to_file,
43
allow_binary=False, sequence_matcher=None,
44
path_encoding='utf8'):
45
# FIXME: difflib is wrong if there is no trailing newline.
46
# The syntax used by patch seems to be "\ No newline at
47
# end of file" following the last diff line from that
48
# file. This is not trivial to insert into the
49
# unified_diff output and it might be better to just fix
50
# or replace that function.
52
# In the meantime we at least make sure the patch isn't
56
# Special workaround for Python2.3, where difflib fails if
57
# both sequences are empty.
58
if not oldlines and not newlines:
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.
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):
61
if allow_binary is False:
62
check_text_lines(oldlines)
63
check_text_lines(newlines)
65
if sequence_matcher is None:
66
sequence_matcher = bzrlib.patiencediff.PatienceSequenceMatcher
67
ud = unified_diff(oldlines, newlines,
68
fromfile=old_filename.encode(path_encoding),
69
tofile=new_filename.encode(path_encoding),
70
sequencematcher=sequence_matcher)
73
# work-around for difflib being too smart for its own good
74
# if /dev/null is "1,0", patch won't recognize it as /dev/null
76
ud[2] = ud[2].replace('-1,0', '-0,0')
78
ud[2] = ud[2].replace('+1,0', '+0,0')
79
# work around for difflib emitting random spaces after the label
80
ud[0] = ud[0][:-2] + '\n'
81
ud[1] = ud[1][:-2] + '\n'
85
if not line.endswith('\n'):
86
to_file.write("\n\\ No newline at end of file\n")
90
def external_diff(old_filename, oldlines, new_filename, newlines, to_file,
92
"""Display a diff by calling out to the external diff program."""
93
if hasattr(to_file, 'fileno'):
147
old_tree = b.basis_tree()
97
out_file = subprocess.PIPE
100
# make sure our own output is properly ordered before the diff
103
oldtmp_fd, old_abspath = tempfile.mkstemp(prefix='bzr-diff-old-')
104
newtmp_fd, new_abspath = tempfile.mkstemp(prefix='bzr-diff-new-')
105
oldtmpf = os.fdopen(oldtmp_fd, 'wb')
106
newtmpf = os.fdopen(newtmp_fd, 'wb')
109
# TODO: perhaps a special case for comparing to or from the empty
110
# sequence; can just use /dev/null on Unix
112
# TODO: if either of the files being compared already exists as a
113
# regular named file (e.g. in the working directory) then we can
114
# compare directly to that, rather than copying it.
116
oldtmpf.writelines(oldlines)
117
newtmpf.writelines(newlines)
125
'--label', old_filename,
127
'--label', new_filename,
132
# diff only allows one style to be specified; they don't override.
133
# note that some of these take optargs, and the optargs can be
134
# directly appended to the options.
135
# this is only an approximate parser; it doesn't properly understand
137
for s in ['-c', '-u', '-C', '-U',
142
'-y', '--side-by-side',
154
diffcmd.extend(diff_opts)
157
pipe = subprocess.Popen(diffcmd,
158
stdin=subprocess.PIPE,
161
if e.errno == errno.ENOENT:
162
raise errors.NoDiff(str(e))
167
bzrlib.osutils.pumpfile(pipe.stdout, to_file)
149
old_tree = b.revision_tree(b.lookup_revision(revision))
170
if rc != 0 and rc != 1:
171
# returns 1 if files differ; that's OK
173
msg = 'signal %d' % (-rc)
175
msg = 'exit code %d' % rc
177
raise BzrError('external diff failed with %s; command: %r' % (rc, diffcmd))
179
oldtmpf.close() # and delete
181
# Clean up. Warn in case the files couldn't be deleted
182
# (in case windows still holds the file open, but not
183
# if the files have already been deleted)
185
os.remove(old_abspath)
187
if e.errno not in (errno.ENOENT,):
188
warning('Failed to delete temporary file: %s %s',
191
os.remove(new_abspath)
193
if e.errno not in (errno.ENOENT,):
194
warning('Failed to delete temporary file: %s %s',
198
@deprecated_function(zero_eight)
199
def show_diff(b, from_spec, specific_files, external_diff_options=None,
200
revision2=None, output=None, b2=None):
201
"""Shortcut for showing the diff to the working tree.
203
Please use show_diff_trees instead.
209
None for 'basis tree', or otherwise the old revision to compare against.
211
The more general form is show_diff_trees(), where the caller
212
supplies any two trees.
217
if from_spec is None:
218
old_tree = b.bzrdir.open_workingtree()
220
old_tree = old_tree = old_tree.basis_tree()
222
old_tree = b.repository.revision_tree(from_spec.in_history(b).rev_id)
224
if revision2 is None:
226
new_tree = b.bzrdir.open_workingtree()
228
new_tree = b2.bzrdir.open_workingtree()
230
new_tree = b.repository.revision_tree(revision2.in_history(b).rev_id)
232
return show_diff_trees(old_tree, new_tree, output, specific_files,
233
external_diff_options)
236
def diff_cmd_helper(tree, specific_files, external_diff_options,
237
old_revision_spec=None, new_revision_spec=None,
238
old_label='a/', new_label='b/'):
239
"""Helper for cmd_diff.
245
The specific files to compare, or None
247
external_diff_options
248
If non-None, run an external diff, and pass it these options
251
If None, use basis tree as old revision, otherwise use the tree for
252
the specified revision.
255
If None, use working tree as new revision, otherwise use the tree for
256
the specified revision.
258
The more general form is show_diff_trees(), where the caller
259
supplies any two trees.
263
revision = spec.in_store(tree.branch)
265
revision = spec.in_store(None)
266
revision_id = revision.rev_id
267
branch = revision.branch
268
return branch.repository.revision_tree(revision_id)
269
if old_revision_spec is None:
270
old_tree = tree.basis_tree()
272
old_tree = spec_tree(old_revision_spec)
274
if new_revision_spec is None:
277
new_tree = spec_tree(new_revision_spec)
278
if new_tree is not tree:
279
extra_trees = (tree,)
283
return show_diff_trees(old_tree, new_tree, sys.stdout, specific_files,
284
external_diff_options,
285
old_label=old_label, new_label=new_label,
286
extra_trees=extra_trees)
289
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None,
290
external_diff_options=None,
291
old_label='a/', new_label='b/',
293
"""Show in text form the changes from one tree to another.
296
If set, include only changes to these files.
298
external_diff_options
299
If set, use an external GNU diff and pass these options.
302
If set, more Trees to use for looking up file ids
308
return _show_diff_trees(old_tree, new_tree, to_file,
309
specific_files, external_diff_options,
310
old_label=old_label, new_label=new_label,
311
extra_trees=extra_trees)
318
def _show_diff_trees(old_tree, new_tree, to_file,
319
specific_files, external_diff_options,
320
old_label='a/', new_label='b/', extra_trees=None):
322
# GNU Patch uses the epoch date to detect files that are being added
323
# or removed in a diff.
324
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.
326
162
# TODO: Generation of pseudo-diffs for added/deleted files could
327
163
# be usefully made into a much faster special case.
329
if external_diff_options:
330
assert isinstance(external_diff_options, basestring)
331
opts = external_diff_options.split()
332
def diff_file(olab, olines, nlab, nlines, to_file):
333
external_diff(olab, olines, nlab, nlines, to_file, opts)
335
diff_file = internal_diff
337
delta = new_tree.changes_from(old_tree,
338
specific_files=specific_files,
339
extra_trees=extra_trees, require_versioned=True)
342
for path, file_id, kind in delta.removed:
344
print >>to_file, '=== removed %s %r' % (kind, path.encode('utf8'))
345
old_name = '%s%s\t%s' % (old_label, path,
346
_patch_header_date(old_tree, file_id, path))
347
new_name = '%s%s\t%s' % (new_label, path, EPOCH_DATE)
348
old_tree.inventory[file_id].diff(diff_file, old_name, old_tree,
349
new_name, None, None, to_file)
350
for path, file_id, kind in delta.added:
352
print >>to_file, '=== added %s %r' % (kind, path.encode('utf8'))
353
old_name = '%s%s\t%s' % (old_label, path, EPOCH_DATE)
354
new_name = '%s%s\t%s' % (new_label, path,
355
_patch_header_date(new_tree, file_id, path))
356
new_tree.inventory[file_id].diff(diff_file, new_name, new_tree,
357
old_name, None, None, to_file,
359
for (old_path, new_path, file_id, kind,
360
text_modified, meta_modified) in delta.renamed:
362
prop_str = get_prop_change(meta_modified)
363
print >>to_file, '=== renamed %s %r => %r%s' % (
364
kind, old_path.encode('utf8'),
365
new_path.encode('utf8'), prop_str)
366
old_name = '%s%s\t%s' % (old_label, old_path,
367
_patch_header_date(old_tree, file_id,
369
new_name = '%s%s\t%s' % (new_label, new_path,
370
_patch_header_date(new_tree, file_id,
372
_maybe_diff_file_or_symlink(old_name, old_tree, file_id,
374
text_modified, kind, to_file, diff_file)
375
for path, file_id, kind, text_modified, meta_modified in delta.modified:
377
prop_str = get_prop_change(meta_modified)
378
print >>to_file, '=== modified %s %r%s' % (kind, path.encode('utf8'), prop_str)
379
old_name = '%s%s\t%s' % (old_label, path,
380
_patch_header_date(old_tree, file_id, path))
381
new_name = '%s%s\t%s' % (new_label, path,
382
_patch_header_date(new_tree, file_id, path))
384
_maybe_diff_file_or_symlink(old_name, old_tree, file_id,
386
True, kind, to_file, diff_file)
391
def _patch_header_date(tree, file_id, path):
392
"""Returns a timestamp suitable for use in a patch header."""
393
tm = time.gmtime(tree.get_file_mtime(file_id, path))
394
return time.strftime('%Y-%m-%d %H:%M:%S +0000', tm)
397
def _raise_if_nonexistent(paths, old_tree, new_tree):
398
"""Complain if paths are not in either inventory or tree.
400
It's OK with the files exist in either tree's inventory, or
401
if they exist in the tree but are not versioned.
403
This can be used by operations such as bzr status that can accept
404
unknown or ignored files.
406
mutter("check paths: %r", paths)
409
s = old_tree.filter_unversioned_files(paths)
410
s = new_tree.filter_unversioned_files(s)
411
s = [path for path in s if not new_tree.has_filename(path)]
413
raise errors.PathsDoNotExist(sorted(s))
416
def get_prop_change(meta_modified):
418
return " (properties changed)"
423
def _maybe_diff_file_or_symlink(old_path, old_tree, file_id,
424
new_path, new_tree, text_modified,
425
kind, to_file, diff_file):
427
new_entry = new_tree.inventory[file_id]
428
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
bailout("can't represent state %s {%s}" % (file_state, fid))
added
removed
bzrlib/diff.py
