« back to all changes in this revision
Viewing changes to bzrlib/diff.py
- browse files at revision 797
- compare with another revision
- download diff
- download tarball
- view history from revision 797
- Committer: Martin Pool
- Date: 2005-06-27 06:30:10 UTC
- Revision ID: mbp@sourcefrog.net-20050627063010-fe0ac45951f9e353
- run the right bzr for testing
- files added:
- bzr-man.py
- bzrlib/selftest
- effbot
- effbot/org
- patches
- plugins
- plugins/changeset
- plugins/rsync
- files removed:
- bzrlib/tests.py
- files renamed:
- contrib/bash/bzr => contrib/bash/bzr.simple
- files modified:
- .bzrignore
added
removed
bzrlib/diff.py
15
15
# along with this program; if not, write to the Free Software
16
16
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
17
17
18
from sets import Set
19
20
18
from trace import mutter
21
19
from errors import BzrError
22
20
23
21
24
def diff_trees(old_tree, new_tree):
25
"""Compute diff between two trees.
26
27
They may be in different branches and may be working or historical
28
trees.
29
30
This only compares the versioned files, paying no attention to
31
files which are ignored or unknown. Those can only be present in
32
working trees and can be reported on separately.
33
34
Yields a sequence of (state, id, old_name, new_name, kind).
35
Each filename and each id is listed only once.
22
# TODO: Rather than building a changeset object, we should probably
23
# invoke callbacks on an object. That object can either accumulate a
24
# list, write them out directly, etc etc.
25
26
def internal_diff(old_label, oldlines, new_label, newlines, to_file):
27
import difflib
28
29
# FIXME: difflib is wrong if there is no trailing newline.
30
# The syntax used by patch seems to be "\ No newline at
31
# end of file" following the last diff line from that
32
# file. This is not trivial to insert into the
33
# unified_diff output and it might be better to just fix
34
# or replace that function.
35
36
# In the meantime we at least make sure the patch isn't
37
# mangled.
38
39
40
# Special workaround for Python2.3, where difflib fails if
41
# both sequences are empty.
42
if not oldlines and not newlines:
43
return
44
45
nonl = False
46
47
if oldlines and (oldlines[-1][-1] != '\n'):
48
oldlines[-1] += '\n'
49
nonl = True
50
if newlines and (newlines[-1][-1] != '\n'):
51
newlines[-1] += '\n'
52
nonl = True
53
54
ud = difflib.unified_diff(oldlines, newlines,
55
fromfile=old_label, tofile=new_label)
56
57
# work-around for difflib being too smart for its own good
58
# if /dev/null is "1,0", patch won't recognize it as /dev/null
59
if not oldlines:
60
ud = list(ud)
61
ud[2] = ud[2].replace('-1,0', '-0,0')
62
elif not newlines:
63
ud = list(ud)
64
ud[2] = ud[2].replace('+1,0', '+0,0')
65
66
to_file.writelines(ud)
67
if nonl:
68
print >>to_file, "\\ No newline at end of file"
69
print >>to_file
70
71
72
73
74
def external_diff(old_label, oldlines, new_label, newlines, to_file,
75
diff_opts):
76
"""Display a diff by calling out to the external diff program."""
77
import sys
78
79
if to_file != sys.stdout:
80
raise NotImplementedError("sorry, can't send external diff other than to stdout yet",
81
to_file)
82
83
# make sure our own output is properly ordered before the diff
84
to_file.flush()
85
86
from tempfile import NamedTemporaryFile
87
import os
88
89
oldtmpf = NamedTemporaryFile()
90
newtmpf = NamedTemporaryFile()
91
92
try:
93
# TODO: perhaps a special case for comparing to or from the empty
94
# sequence; can just use /dev/null on Unix
95
96
# TODO: if either of the files being compared already exists as a
97
# regular named file (e.g. in the working directory) then we can
98
# compare directly to that, rather than copying it.
99
100
oldtmpf.writelines(oldlines)
101
newtmpf.writelines(newlines)
102
103
oldtmpf.flush()
104
newtmpf.flush()
105
106
if not diff_opts:
107
diff_opts = []
108
diffcmd = ['diff',
109
'--label', old_label,
110
oldtmpf.name,
111
'--label', new_label,
112
newtmpf.name]
113
114
# diff only allows one style to be specified; they don't override.
115
# note that some of these take optargs, and the optargs can be
116
# directly appended to the options.
117
# this is only an approximate parser; it doesn't properly understand
118
# the grammar.
119
for s in ['-c', '-u', '-C', '-U',
120
'-e', '--ed',
121
'-q', '--brief',
122
'--normal',
123
'-n', '--rcs',
124
'-y', '--side-by-side',
125
'-D', '--ifdef']:
126
for j in diff_opts:
127
if j.startswith(s):
128
break
129
else:
130
continue
131
break
132
else:
133
diffcmd.append('-u')
134
135
if diff_opts:
136
diffcmd.extend(diff_opts)
137
138
rc = os.spawnvp(os.P_WAIT, 'diff', diffcmd)
139
140
if rc != 0 and rc != 1:
141
# returns 1 if files differ; that's OK
142
if rc < 0:
143
msg = 'signal %d' % (-rc)
144
else:
145
msg = 'exit code %d' % rc
146
147
raise BzrError('external diff failed with %s; command: %r' % (rc, diffcmd))
148
finally:
149
oldtmpf.close() # and delete
150
newtmpf.close()
151
152
153
154
def show_diff(b, revision, specific_files, external_diff_options=None):
155
"""Shortcut for showing the diff to the working tree.
156
157
b
158
Branch.
159
160
revision
161
None for each, or otherwise the old revision to compare against.
162
163
The more general form is show_diff_trees(), where the caller
164
supplies any two trees.
36
165
"""
37
## TODO: Allow specifying a list of files to compare, rather than
38
## doing the whole tree? (Not urgent.)
39
40
## TODO: Allow diffing any two inventories, not just the
41
## current one against one. We mgiht need to specify two
42
## stores to look for the files if diffing two branches. That
43
## might imply this shouldn't be primarily a Branch method.
44
45
sha_match_cnt = modified_cnt = 0
46
47
old_it = old_tree.list_files()
48
new_it = new_tree.list_files()
49
50
def next(it):
51
try:
52
return it.next()
53
except StopIteration:
54
return None
55
56
old_item = next(old_it)
57
new_item = next(new_it)
58
59
# We step through the two sorted iterators in parallel, trying to
60
# keep them lined up.
61
62
while (old_item != None) or (new_item != None):
63
# OK, we still have some remaining on both, but they may be
64
# out of step.
65
if old_item != None:
66
old_name, old_class, old_kind, old_id = old_item
67
else:
68
old_name = None
69
70
if new_item != None:
71
new_name, new_class, new_kind, new_id = new_item
72
else:
73
new_name = None
74
75
if old_item:
76
# can't handle the old tree being a WorkingTree
77
assert old_class == 'V'
78
79
if new_item and (new_class != 'V'):
80
yield new_class, None, None, new_name, new_kind
81
new_item = next(new_it)
82
elif (not new_item) or (old_item and (old_name < new_name)):
83
if new_tree.has_id(old_id):
84
# will be mentioned as renamed under new name
85
pass
86
else:
87
yield 'D', old_id, old_name, None, old_kind
88
old_item = next(old_it)
89
elif (not old_item) or (new_item and (new_name < old_name)):
90
if old_tree.has_id(new_id):
91
yield 'R', new_id, old_tree.id2path(new_id), new_name, new_kind
92
else:
93
yield 'A', new_id, None, new_name, new_kind
94
new_item = next(new_it)
95
elif old_id != new_id:
96
assert old_name == new_name
97
# both trees have a file of this name, but it is not the
98
# same file. in other words, the old filename has been
99
# overwritten by either a newly-added or a renamed file.
100
# (should we return something about the overwritten file?)
101
if old_tree.has_id(new_id):
102
# renaming, overlying a deleted file
103
yield 'R', new_id, old_tree.id2path(new_id), new_name, new_kind
104
else:
105
yield 'A', new_id, None, new_name, new_kind
106
107
new_item = next(new_it)
108
old_item = next(old_it)
109
else:
110
assert old_id == new_id
111
assert old_id != None
112
assert old_name == new_name
113
assert old_kind == new_kind
114
115
if old_kind == 'directory':
116
yield '.', new_id, old_name, new_name, new_kind
117
elif old_tree.get_file_sha1(old_id) == new_tree.get_file_sha1(old_id):
118
sha_match_cnt += 1
119
yield '.', new_id, old_name, new_name, new_kind
120
else:
121
modified_cnt += 1
122
yield 'M', new_id, old_name, new_name, new_kind
123
124
new_item = next(new_it)
125
old_item = next(old_it)
126
127
128
mutter("diff finished: %d SHA matches, %d modified"
129
% (sha_match_cnt, modified_cnt))
130
131
132
133
def show_diff(b, revision, file_list):
134
import difflib, sys, types
135
166
import sys
167
136
168
if revision == None:
137
169
old_tree = b.basis_tree()
138
170
else:
140
172
141
173
new_tree = b.working_tree()
142
174
175
show_diff_trees(old_tree, new_tree, sys.stdout, specific_files,
176
external_diff_options)
177
178
179
180
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None,
181
external_diff_options=None):
182
"""Show in text form the changes from one tree to another.
183
184
to_files
185
If set, include only changes to these files.
186
187
external_diff_options
188
If set, use an external GNU diff and pass these options.
189
"""
190
143
191
# TODO: Options to control putting on a prefix or suffix, perhaps as a format string
144
192
old_label = ''
145
193
new_label = ''
152
200
# TODO: Generation of pseudo-diffs for added/deleted files could
153
201
# be usefully made into a much faster special case.
154
202
155
# TODO: Better to return them in sorted order I think.
156
157
if file_list:
158
file_list = [b.relpath(f) for f in file_list]
159
160
# FIXME: If given a file list, compare only those files rather
161
# than comparing everything and then throwing stuff away.
203
if external_diff_options:
204
assert isinstance(external_diff_options, basestring)
205
opts = external_diff_options.split()
206
def diff_file(olab, olines, nlab, nlines, to_file):
207
external_diff(olab, olines, nlab, nlines, to_file, opts)
208
else:
209
diff_file = internal_diff
162
210
163
for file_state, fid, old_name, new_name, kind in diff_trees(old_tree, new_tree):
164
165
if file_list and (new_name not in file_list):
166
continue
167
168
# Don't show this by default; maybe do it if an option is passed
169
# idlabel = ' {%s}' % fid
170
idlabel = ''
171
172
def diffit(oldlines, newlines, **kw):
173
174
# FIXME: difflib is wrong if there is no trailing newline.
175
# The syntax used by patch seems to be "\ No newline at
176
# end of file" following the last diff line from that
177
# file. This is not trivial to insert into the
178
# unified_diff output and it might be better to just fix
179
# or replace that function.
180
181
# In the meantime we at least make sure the patch isn't
182
# mangled.
183
184
185
# Special workaround for Python2.3, where difflib fails if
186
# both sequences are empty.
187
if not oldlines and not newlines:
188
return
189
190
nonl = False
191
192
if oldlines and (oldlines[-1][-1] != '\n'):
193
oldlines[-1] += '\n'
194
nonl = True
195
if newlines and (newlines[-1][-1] != '\n'):
196
newlines[-1] += '\n'
197
nonl = True
198
199
ud = difflib.unified_diff(oldlines, newlines, **kw)
200
201
# work-around for difflib being too smart for its own good
202
# if /dev/null is "1,0", patch won't recognize it as /dev/null
203
if not oldlines:
204
ud = list(ud)
205
ud[2] = ud[2].replace('-1,0', '-0,0')
206
elif not newlines:
207
ud = list(ud)
208
ud[2] = ud[2].replace('+1,0', '+0,0')
209
210
sys.stdout.writelines(ud)
211
if nonl:
212
print "\\ No newline at end of file"
213
sys.stdout.write('\n')
214
215
if file_state in ['.', '?', 'I']:
216
continue
217
elif file_state == 'A':
218
print '*** added %s %r' % (kind, new_name)
219
if kind == 'file':
220
diffit([],
221
new_tree.get_file(fid).readlines(),
222
fromfile=DEVNULL,
223
tofile=new_label + new_name + idlabel)
224
elif file_state == 'D':
225
assert isinstance(old_name, types.StringTypes)
226
print '*** deleted %s %r' % (kind, old_name)
227
if kind == 'file':
228
diffit(old_tree.get_file(fid).readlines(), [],
229
fromfile=old_label + old_name + idlabel,
230
tofile=DEVNULL)
231
elif file_state in ['M', 'R']:
232
if file_state == 'M':
233
assert kind == 'file'
234
assert old_name == new_name
235
print '*** modified %s %r' % (kind, new_name)
236
elif file_state == 'R':
237
print '*** renamed %s %r => %r' % (kind, old_name, new_name)
238
239
if kind == 'file':
240
diffit(old_tree.get_file(fid).readlines(),
241
new_tree.get_file(fid).readlines(),
242
fromfile=old_label + old_name + idlabel,
243
tofile=new_label + new_name)
244
else:
245
raise BzrError("can't represent state %s {%s}" % (file_state, fid))
246
247
248
249
class TreeDelta:
211
212
delta = compare_trees(old_tree, new_tree, want_unchanged=False,
213
specific_files=specific_files)
214
215
for path, file_id, kind in delta.removed:
216
print >>to_file, '*** removed %s %r' % (kind, path)
217
if kind == 'file':
218
diff_file(old_label + path,
219
old_tree.get_file(file_id).readlines(),
220
DEVNULL,
221
[],
222
to_file)
223
224
for path, file_id, kind in delta.added:
225
print >>to_file, '*** added %s %r' % (kind, path)
226
if kind == 'file':
227
diff_file(DEVNULL,
228
[],
229
new_label + path,
230
new_tree.get_file(file_id).readlines(),
231
to_file)
232
233
for old_path, new_path, file_id, kind, text_modified in delta.renamed:
234
print >>to_file, '*** renamed %s %r => %r' % (kind, old_path, new_path)
235
if text_modified:
236
diff_file(old_label + old_path,
237
old_tree.get_file(file_id).readlines(),
238
new_label + new_path,
239
new_tree.get_file(file_id).readlines(),
240
to_file)
241
242
for path, file_id, kind in delta.modified:
243
print >>to_file, '*** modified %s %r' % (kind, path)
244
if kind == 'file':
245
diff_file(old_label + path,
246
old_tree.get_file(file_id).readlines(),
247
new_label + path,
248
new_tree.get_file(file_id).readlines(),
249
to_file)
250
251
252
253
class TreeDelta(object):
250
254
"""Describes changes from one tree to another.
251
255
252
256
Contains four lists:
253
257
254
258
added
255
(path, id)
259
(path, id, kind)
256
260
removed
257
(path, id)
261
(path, id, kind)
258
262
renamed
259
(oldpath, newpath, id, text_modified)
263
(oldpath, newpath, id, kind, text_modified)
260
264
modified
261
(path, id)
265
(path, id, kind)
262
266
unchanged
263
(path, id)
267
(path, id, kind)
264
268
265
269
Each id is listed only once.
266
270
267
271
Files that are both modified and renamed are listed only in
268
272
renamed, with the text_modified flag true.
269
273
274
Files are only considered renamed if their name has changed or
275
their parent directory has changed. Renaming a directory
276
does not count as renaming all its contents.
277
270
278
The lists are normally sorted when the delta is created.
271
279
"""
272
280
def __init__(self):
276
284
self.modified = []
277
285
self.unchanged = []
278
286
287
def __eq__(self, other):
288
if not isinstance(other, TreeDelta):
289
return False
290
return self.added == other.added \
291
and self.removed == other.removed \
292
and self.renamed == other.renamed \
293
and self.modified == other.modified \
294
and self.unchanged == other.unchanged
295
296
def __ne__(self, other):
297
return not (self == other)
298
299
def __repr__(self):
300
return "TreeDelta(added=%r, removed=%r, renamed=%r, modified=%r," \
301
" unchanged=%r)" % (self.added, self.removed, self.renamed,
302
self.modified, self.unchanged)
303
304
def has_changed(self):
305
changes = len(self.added) + len(self.removed) + len(self.renamed)
306
changes += len(self.modified)
307
return (changes != 0)
308
309
def touches_file_id(self, file_id):
310
"""Return True if file_id is modified by this delta."""
311
for l in self.added, self.removed, self.modified:
312
for v in l:
313
if v[1] == file_id:
314
return True
315
for v in self.renamed:
316
if v[2] == file_id:
317
return True
318
return False
319
320
279
321
def show(self, to_file, show_ids=False, show_unchanged=False):
280
322
def show_list(files):
281
for path, fid in files:
323
for path, fid, kind in files:
324
if kind == 'directory':
325
path += '/'
326
elif kind == 'symlink':
327
path += '@'
328
282
329
if show_ids:
283
330
print >>to_file, ' %-30s %s' % (path, fid)
284
331
else:
285
332
print >>to_file, ' ', path
286
333
287
334
if self.removed:
288
print >>to_file, 'removed files:'
335
print >>to_file, 'removed:'
289
336
show_list(self.removed)
290
337
291
338
if self.added:
292
print >>to_file, 'added files:'
339
print >>to_file, 'added:'
293
340
show_list(self.added)
294
341
295
342
if self.renamed:
296
print >>to_file, 'renamed files:'
297
for oldpath, newpath, fid, text_modified in self.renamed:
343
print >>to_file, 'renamed:'
344
for oldpath, newpath, fid, kind, text_modified in self.renamed:
298
345
if show_ids:
299
346
print >>to_file, ' %s => %s %s' % (oldpath, newpath, fid)
300
347
else:
301
348
print >>to_file, ' %s => %s' % (oldpath, newpath)
302
349
303
350
if self.modified:
304
print >>to_file, 'modified files:'
351
print >>to_file, 'modified:'
305
352
show_list(self.modified)
306
353
307
354
if show_unchanged and self.unchanged:
308
print >>to_file, 'unchanged files:'
355
print >>to_file, 'unchanged:'
309
356
show_list(self.unchanged)
310
357
311
358
312
359
313
def compare_trees(old_tree, new_tree, want_unchanged):
360
def compare_trees(old_tree, new_tree, want_unchanged=False, specific_files=None):
361
"""Describe changes from one tree to another.
362
363
Returns a TreeDelta with details of added, modified, renamed, and
364
deleted entries.
365
366
The root entry is specifically exempt.
367
368
This only considers versioned files.
369
370
want_unchanged
371
If true, also list files unchanged from one version to
372
the next.
373
374
specific_files
375
If true, only check for changes to specified names or
376
files within them.
377
"""
378
379
from osutils import is_inside_any
380
314
381
old_inv = old_tree.inventory
315
382
new_inv = new_tree.inventory
316
383
delta = TreeDelta()
384
mutter('start compare_trees')
385
386
# TODO: match for specific files can be rather smarter by finding
387
# the IDs of those files up front and then considering only that.
388
317
389
for file_id in old_tree:
318
390
if file_id in new_tree:
319
old_path = old_inv.id2path(file_id)
320
new_path = new_inv.id2path(file_id)
321
322
391
kind = old_inv.get_file_kind(file_id)
392
assert kind == new_inv.get_file_kind(file_id)
393
323
394
assert kind in ('file', 'directory', 'symlink', 'root_directory'), \
324
395
'invalid file kind %r' % kind
396
397
if kind == 'root_directory':
398
continue
399
400
old_path = old_inv.id2path(file_id)
401
new_path = new_inv.id2path(file_id)
402
403
old_ie = old_inv[file_id]
404
new_ie = new_inv[file_id]
405
406
if specific_files:
407
if (not is_inside_any(specific_files, old_path)
408
and not is_inside_any(specific_files, new_path)):
409
continue
410
325
411
if kind == 'file':
326
412
old_sha1 = old_tree.get_file_sha1(file_id)
327
413
new_sha1 = new_tree.get_file_sha1(file_id)
335
421
# the same and the parents are unchanged all the way up.
336
422
# May not be worthwhile.
337
423
338
if old_path != new_path:
339
delta.renamed.append((old_path, new_path, file_id, text_modified))
424
if (old_ie.name != new_ie.name
425
or old_ie.parent_id != new_ie.parent_id):
426
delta.renamed.append((old_path, new_path, file_id, kind,
427
text_modified))
340
428
elif text_modified:
341
delta.modified.append((new_path, file_id))
429
delta.modified.append((new_path, file_id, kind))
342
430
elif want_unchanged:
343
delta.unchanged.append((new_path, file_id))
431
delta.unchanged.append((new_path, file_id, kind))
344
432
else:
345
delta.removed.append((old_inv.id2path(file_id), file_id))
433
kind = old_inv.get_file_kind(file_id)
434
old_path = old_inv.id2path(file_id)
435
if specific_files:
436
if not is_inside_any(specific_files, old_path):
437
continue
438
delta.removed.append((old_path, file_id, kind))
439
440
mutter('start looking for new files')
346
441
for file_id in new_inv:
347
442
if file_id in old_inv:
348
443
continue
349
delta.added.append((new_inv.id2path(file_id), file_id))
444
new_path = new_inv.id2path(file_id)
445
if specific_files:
446
if not is_inside_any(specific_files, new_path):
447
continue
448
kind = new_inv.get_file_kind(file_id)
449
delta.added.append((new_path, file_id, kind))
350
450
351
451
delta.removed.sort()
352
452
delta.added.sort()
353
453
delta.renamed.sort()
354
454
delta.modified.sort()
455
delta.unchanged.sort()
355
456
356
457
return delta
Loggerhead is a web-based interface for Breezy
Version: 2.0.1