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
25
# compatability - plugins import compare_trees from diff!!!
26
# deprecated as of 0.10
27
from bzrlib.delta import compare_trees
28
from bzrlib.errors import BzrError
29
import bzrlib.errors as errors
31
from bzrlib.patiencediff import unified_diff
32
import bzrlib.patiencediff
33
from bzrlib.symbol_versioning import (deprecated_function,
35
from bzrlib.textfile import check_text_lines
36
from bzrlib.trace import mutter, warning
39
# TODO: Rather than building a changeset object, we should probably
40
# invoke callbacks on an object. That object can either accumulate a
41
# list, write them out directly, etc etc.
43
def internal_diff(old_filename, oldlines, new_filename, newlines, to_file,
44
allow_binary=False, sequence_matcher=None,
45
path_encoding='utf8'):
18
from trace import mutter
19
from errors import BzrError
22
def internal_diff(old_label, oldlines, new_label, newlines, to_file):
46
25
# FIXME: difflib is wrong if there is no trailing newline.
47
26
# The syntax used by patch seems to be "\ No newline at
48
27
# end of file" following the last diff line from that
58
37
# both sequences are empty.
59
38
if not oldlines and not newlines:
62
if allow_binary is False:
63
check_text_lines(oldlines)
64
check_text_lines(newlines)
66
if sequence_matcher is None:
67
sequence_matcher = bzrlib.patiencediff.PatienceSequenceMatcher
68
ud = unified_diff(oldlines, newlines,
69
fromfile=old_filename.encode(path_encoding),
70
tofile=new_filename.encode(path_encoding),
71
sequencematcher=sequence_matcher)
43
if oldlines and (oldlines[-1][-1] != '\n'):
46
if newlines and (newlines[-1][-1] != '\n'):
50
ud = difflib.unified_diff(oldlines, newlines,
51
fromfile=old_label, tofile=new_label)
74
53
# work-around for difflib being too smart for its own good
75
54
# if /dev/null is "1,0", patch won't recognize it as /dev/null
77
57
ud[2] = ud[2].replace('-1,0', '-0,0')
79
60
ud[2] = ud[2].replace('+1,0', '+0,0')
80
# work around for difflib emitting random spaces after the label
81
ud[0] = ud[0][:-2] + '\n'
82
ud[1] = ud[1][:-2] + '\n'
86
if not line.endswith('\n'):
87
to_file.write("\n\\ No newline at end of file\n")
62
to_file.writelines(ud)
64
print >>to_file, "\\ No newline at end of file"
92
"""Set the env var LANG=C"""
93
os.environ['LANG'] = 'C'
96
def _spawn_external_diff(diffcmd, capture_errors=True):
97
"""Spawn the externall diff process, and return the child handle.
99
:param diffcmd: The command list to spawn
100
:param capture_errors: Capture stderr as well as setting LANG=C.
101
This lets us read and understand the output of diff, and respond
103
:return: A Popen object.
106
preexec_fn = _set_lang_C
107
stderr = subprocess.PIPE
113
pipe = subprocess.Popen(diffcmd,
114
stdin=subprocess.PIPE,
115
stdout=subprocess.PIPE,
117
preexec_fn=preexec_fn)
119
if e.errno == errno.ENOENT:
120
raise errors.NoDiff(str(e))
126
def external_diff(old_filename, oldlines, new_filename, newlines, to_file,
70
def external_diff(old_label, oldlines, new_label, newlines, to_file,
128
72
"""Display a diff by calling out to the external diff program."""
75
if to_file != sys.stdout:
76
raise NotImplementedError("sorry, can't send external diff other than to stdout yet",
129
79
# make sure our own output is properly ordered before the diff
132
oldtmp_fd, old_abspath = tempfile.mkstemp(prefix='bzr-diff-old-')
133
newtmp_fd, new_abspath = tempfile.mkstemp(prefix='bzr-diff-new-')
134
oldtmpf = os.fdopen(oldtmp_fd, 'wb')
135
newtmpf = os.fdopen(newtmp_fd, 'wb')
82
from tempfile import NamedTemporaryFile
85
oldtmpf = NamedTemporaryFile()
86
newtmpf = NamedTemporaryFile()
138
89
# TODO: perhaps a special case for comparing to or from the empty
183
132
diffcmd.extend(diff_opts)
185
pipe = _spawn_external_diff(diffcmd, capture_errors=True)
186
out,err = pipe.communicate()
134
rc = os.spawnvp(os.P_WAIT, 'diff', diffcmd)
189
# internal_diff() adds a trailing newline, add one here for consistency
192
# 'diff' gives retcode == 2 for all sorts of errors
193
# one of those is 'Binary files differ'.
194
# Bad options could also be the problem.
195
# 'Binary files' is not a real error, so we suppress that error
198
# Since we got here, we want to make sure to give an i18n error
199
pipe = _spawn_external_diff(diffcmd, capture_errors=False)
200
out, err = pipe.communicate()
202
# Write out the new i18n diff response
203
to_file.write(out+'\n')
204
if pipe.returncode != 2:
205
raise BzrError('external diff failed with exit code 2'
206
' when run with LANG=C, but not when run'
207
' natively: %r' % (diffcmd,))
209
first_line = lang_c_out.split('\n', 1)[0]
210
m = re.match('^binary files.*differ$', first_line, re.I)
212
raise BzrError('external diff failed with exit code 2;'
213
' command: %r' % (diffcmd,))
215
# Binary files differ, just return
218
# If we got to here, we haven't written out the output of diff
136
if rc != 0 and rc != 1:
222
137
# returns 1 if files differ; that's OK
224
139
msg = 'signal %d' % (-rc)
226
141
msg = 'exit code %d' % rc
228
raise BzrError('external diff failed with %s; command: %r'
143
raise BzrError('external diff failed with %s; command: %r' % (rc, diffcmd))
233
145
oldtmpf.close() # and delete
235
# Clean up. Warn in case the files couldn't be deleted
236
# (in case windows still holds the file open, but not
237
# if the files have already been deleted)
239
os.remove(old_abspath)
241
if e.errno not in (errno.ENOENT,):
242
warning('Failed to delete temporary file: %s %s',
245
os.remove(new_abspath)
247
if e.errno not in (errno.ENOENT,):
248
warning('Failed to delete temporary file: %s %s',
252
@deprecated_function(zero_eight)
253
def show_diff(b, from_spec, specific_files, external_diff_options=None,
254
revision2=None, output=None, b2=None):
255
"""Shortcut for showing the diff to the working tree.
257
Please use show_diff_trees instead.
263
None for 'basis tree', or otherwise the old revision to compare against.
265
The more general form is show_diff_trees(), where the caller
266
supplies any two trees.
271
if from_spec is None:
272
old_tree = b.bzrdir.open_workingtree()
274
old_tree = old_tree = old_tree.basis_tree()
276
old_tree = b.repository.revision_tree(from_spec.in_history(b).rev_id)
278
if revision2 is None:
280
new_tree = b.bzrdir.open_workingtree()
282
new_tree = b2.bzrdir.open_workingtree()
284
new_tree = b.repository.revision_tree(revision2.in_history(b).rev_id)
286
return show_diff_trees(old_tree, new_tree, output, specific_files,
287
external_diff_options)
290
def diff_cmd_helper(tree, specific_files, external_diff_options,
291
old_revision_spec=None, new_revision_spec=None,
292
old_label='a/', new_label='b/'):
293
"""Helper for cmd_diff.
299
The specific files to compare, or None
301
external_diff_options
302
If non-None, run an external diff, and pass it these options
305
If None, use basis tree as old revision, otherwise use the tree for
306
the specified revision.
309
If None, use working tree as new revision, otherwise use the tree for
310
the specified revision.
312
The more general form is show_diff_trees(), where the caller
313
supplies any two trees.
317
revision = spec.in_store(tree.branch)
319
revision = spec.in_store(None)
320
revision_id = revision.rev_id
321
branch = revision.branch
322
return branch.repository.revision_tree(revision_id)
323
if old_revision_spec is None:
324
old_tree = tree.basis_tree()
326
old_tree = spec_tree(old_revision_spec)
328
if new_revision_spec is None:
331
new_tree = spec_tree(new_revision_spec)
332
if new_tree is not tree:
333
extra_trees = (tree,)
337
return show_diff_trees(old_tree, new_tree, sys.stdout, specific_files,
338
external_diff_options,
339
old_label=old_label, new_label=new_label,
340
extra_trees=extra_trees)
150
def show_diff(b, revision, specific_files, external_diff_options=None):
154
old_tree = b.basis_tree()
156
old_tree = b.revision_tree(b.lookup_revision(revision))
158
new_tree = b.working_tree()
160
show_diff_trees(old_tree, new_tree, sys.stdout, specific_files,
161
external_diff_options)
343
165
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None,
344
external_diff_options=None,
345
old_label='a/', new_label='b/',
166
external_diff_options=None):
347
167
"""Show in text form the changes from one tree to another.
389
194
diff_file = internal_diff
391
delta = new_tree.changes_from(old_tree,
392
specific_files=specific_files,
393
extra_trees=extra_trees, require_versioned=True)
197
delta = compare_trees(old_tree, new_tree, want_unchanged=False,
198
specific_files=specific_files)
396
200
for path, file_id, kind in delta.removed:
398
print >>to_file, '=== removed %s %r' % (kind, path.encode('utf8'))
399
old_name = '%s%s\t%s' % (old_label, path,
400
_patch_header_date(old_tree, file_id, path))
401
new_name = '%s%s\t%s' % (new_label, path, EPOCH_DATE)
402
old_tree.inventory[file_id].diff(diff_file, old_name, old_tree,
403
new_name, None, None, to_file)
201
print '*** removed %s %r' % (kind, path)
203
diff_file(old_label + path,
204
old_tree.get_file(file_id).readlines(),
404
209
for path, file_id, kind in delta.added:
406
print >>to_file, '=== added %s %r' % (kind, path.encode('utf8'))
407
old_name = '%s%s\t%s' % (old_label, path, EPOCH_DATE)
408
new_name = '%s%s\t%s' % (new_label, path,
409
_patch_header_date(new_tree, file_id, path))
410
new_tree.inventory[file_id].diff(diff_file, new_name, new_tree,
411
old_name, None, None, to_file,
413
for (old_path, new_path, file_id, kind,
414
text_modified, meta_modified) in delta.renamed:
416
prop_str = get_prop_change(meta_modified)
417
print >>to_file, '=== renamed %s %r => %r%s' % (
418
kind, old_path.encode('utf8'),
419
new_path.encode('utf8'), prop_str)
420
old_name = '%s%s\t%s' % (old_label, old_path,
421
_patch_header_date(old_tree, file_id,
423
new_name = '%s%s\t%s' % (new_label, new_path,
424
_patch_header_date(new_tree, file_id,
426
_maybe_diff_file_or_symlink(old_name, old_tree, file_id,
428
text_modified, kind, to_file, diff_file)
429
for path, file_id, kind, text_modified, meta_modified in delta.modified:
431
prop_str = get_prop_change(meta_modified)
432
print >>to_file, '=== modified %s %r%s' % (kind, path.encode('utf8'), prop_str)
433
old_name = '%s%s\t%s' % (old_label, path,
434
_patch_header_date(old_tree, file_id, path))
435
new_name = '%s%s\t%s' % (new_label, path,
436
_patch_header_date(new_tree, file_id, path))
210
print '*** added %s %r' % (kind, path)
215
new_tree.get_file(file_id).readlines(),
218
for old_path, new_path, file_id, kind, text_modified in delta.renamed:
219
print '*** renamed %s %r => %r' % (kind, old_path, new_path)
437
220
if text_modified:
438
_maybe_diff_file_or_symlink(old_name, old_tree, file_id,
440
True, kind, to_file, diff_file)
445
def _patch_header_date(tree, file_id, path):
446
"""Returns a timestamp suitable for use in a patch header."""
447
tm = time.gmtime(tree.get_file_mtime(file_id, path))
448
return time.strftime('%Y-%m-%d %H:%M:%S +0000', tm)
451
def _raise_if_nonexistent(paths, old_tree, new_tree):
452
"""Complain if paths are not in either inventory or tree.
454
It's OK with the files exist in either tree's inventory, or
455
if they exist in the tree but are not versioned.
221
diff_file(old_label + old_path,
222
old_tree.get_file(file_id).readlines(),
223
new_label + new_path,
224
new_tree.get_file(file_id).readlines(),
227
for path, file_id, kind in delta.modified:
228
print '*** modified %s %r' % (kind, path)
230
diff_file(old_label + path,
231
old_tree.get_file(file_id).readlines(),
233
new_tree.get_file(file_id).readlines(),
238
class TreeDelta(object):
239
"""Describes changes from one tree to another.
248
(oldpath, newpath, id, kind, text_modified)
254
Each id is listed only once.
256
Files that are both modified and renamed are listed only in
257
renamed, with the text_modified flag true.
259
The lists are normally sorted when the delta is created.
269
def touches_file_id(self, file_id):
270
"""Return True if file_id is modified by this delta."""
271
for l in self.added, self.removed, self.modified:
275
for v in self.renamed:
281
def show(self, to_file, show_ids=False, show_unchanged=False):
282
def show_list(files):
283
for path, fid, kind in files:
284
if kind == 'directory':
286
elif kind == 'symlink':
290
print >>to_file, ' %-30s %s' % (path, fid)
292
print >>to_file, ' ', path
295
print >>to_file, 'removed:'
296
show_list(self.removed)
299
print >>to_file, 'added:'
300
show_list(self.added)
303
print >>to_file, 'renamed:'
304
for oldpath, newpath, fid, kind, text_modified in self.renamed:
306
print >>to_file, ' %s => %s %s' % (oldpath, newpath, fid)
308
print >>to_file, ' %s => %s' % (oldpath, newpath)
311
print >>to_file, 'modified:'
312
show_list(self.modified)
314
if show_unchanged and self.unchanged:
315
print >>to_file, 'unchanged:'
316
show_list(self.unchanged)
320
def compare_trees(old_tree, new_tree, want_unchanged, specific_files=None):
321
"""Describe changes from one tree to another.
323
Returns a TreeDelta with details of added, modified, renamed, and
326
The root entry is specifically exempt.
328
This only considers versioned files.
331
If true, also list files unchanged from one version to
335
If true, only check for changes to specified names or
339
from osutils import is_inside_any
457
This can be used by operations such as bzr status that can accept
458
unknown or ignored files.
460
mutter("check paths: %r", paths)
463
s = old_tree.filter_unversioned_files(paths)
464
s = new_tree.filter_unversioned_files(s)
465
s = [path for path in s if not new_tree.has_filename(path)]
467
raise errors.PathsDoNotExist(sorted(s))
470
def get_prop_change(meta_modified):
472
return " (properties changed)"
477
def _maybe_diff_file_or_symlink(old_path, old_tree, file_id,
478
new_path, new_tree, text_modified,
479
kind, to_file, diff_file):
481
new_entry = new_tree.inventory[file_id]
482
old_tree.inventory[file_id].diff(diff_file,
341
old_inv = old_tree.inventory
342
new_inv = new_tree.inventory
344
mutter('start compare_trees')
346
# TODO: match for specific files can be rather smarter by finding
347
# the IDs of those files up front and then considering only that.
349
for file_id in old_tree:
350
if file_id in new_tree:
351
kind = old_inv.get_file_kind(file_id)
352
assert kind == new_inv.get_file_kind(file_id)
354
assert kind in ('file', 'directory', 'symlink', 'root_directory'), \
355
'invalid file kind %r' % kind
357
if kind == 'root_directory':
360
old_path = old_inv.id2path(file_id)
361
new_path = new_inv.id2path(file_id)
364
if (not is_inside_any(specific_files, old_path)
365
and not is_inside_any(specific_files, new_path)):
369
old_sha1 = old_tree.get_file_sha1(file_id)
370
new_sha1 = new_tree.get_file_sha1(file_id)
371
text_modified = (old_sha1 != new_sha1)
373
## mutter("no text to check for %r %r" % (file_id, kind))
374
text_modified = False
376
# TODO: Can possibly avoid calculating path strings if the
377
# two files are unchanged and their names and parents are
378
# the same and the parents are unchanged all the way up.
379
# May not be worthwhile.
381
if old_path != new_path:
382
delta.renamed.append((old_path, new_path, file_id, kind,
385
delta.modified.append((new_path, file_id, kind))
387
delta.unchanged.append((new_path, file_id, kind))
389
kind = old_inv.get_file_kind(file_id)
390
old_path = old_inv.id2path(file_id)
392
if not is_inside_any(specific_files, old_path):
394
delta.removed.append((old_path, file_id, kind))
396
mutter('start looking for new files')
397
for file_id in new_inv:
398
if file_id in old_inv:
400
new_path = new_inv.id2path(file_id)
402
if not is_inside_any(specific_files, new_path):
404
kind = new_inv.get_file_kind(file_id)
405
delta.added.append((new_path, file_id, kind))
410
delta.modified.sort()
411
delta.unchanged.sort()
added
removed
bzrlib/diff.py
