« back to all changes in this revision
Viewing changes to bzrlib/diff.py
- browse files at revision 974.1.17
- compare with another revision
- download diff
- download tarball
- view history from revision 974.1.17
- Committer: Aaron Bentley
- Date: 2005-08-10 15:05:26 UTC
- mto: (1092.1.41) (1185.3.4) (974.1.47)
- mto: This revision was merged to the branch mainline in revision 1110.
- Revision ID: abentley@panoramicfeedback.com-20050810150526-468a7e2e3dc299e4
Switched to using a set of interesting file_ids instead of SourceFile attribute
- files added:
- bzr-man.py
- bzrlib/selftest
- effbot
- effbot/org
- patches
- plugins
- plugins/changeset
- tools
- 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
22
23
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
Yields a sequence of (state, id, old_name, new_name, kind).
31
Each filename and each id is listed only once.
19
from errors import BzrError
20
21
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
ud = difflib.unified_diff(oldlines, newlines,
46
fromfile=old_label, tofile=new_label)
47
48
# work-around for difflib being too smart for its own good
49
# if /dev/null is "1,0", patch won't recognize it as /dev/null
50
if not oldlines:
51
ud = list(ud)
52
ud[2] = ud[2].replace('-1,0', '-0,0')
53
elif not newlines:
54
ud = list(ud)
55
ud[2] = ud[2].replace('+1,0', '+0,0')
56
57
for line in ud:
58
to_file.write(line)
59
if not line.endswith('\n'):
60
to_file.write("\n\\ No newline at end of file\n")
61
print >>to_file
62
63
64
65
66
def external_diff(old_label, oldlines, new_label, newlines, to_file,
67
diff_opts):
68
"""Display a diff by calling out to the external diff program."""
69
import sys
70
71
if to_file != sys.stdout:
72
raise NotImplementedError("sorry, can't send external diff other than to stdout yet",
73
to_file)
74
75
# make sure our own output is properly ordered before the diff
76
to_file.flush()
77
78
from tempfile import NamedTemporaryFile
79
import os
80
81
oldtmpf = NamedTemporaryFile()
82
newtmpf = NamedTemporaryFile()
83
84
try:
85
# TODO: perhaps a special case for comparing to or from the empty
86
# sequence; can just use /dev/null on Unix
87
88
# TODO: if either of the files being compared already exists as a
89
# regular named file (e.g. in the working directory) then we can
90
# compare directly to that, rather than copying it.
91
92
oldtmpf.writelines(oldlines)
93
newtmpf.writelines(newlines)
94
95
oldtmpf.flush()
96
newtmpf.flush()
97
98
if not diff_opts:
99
diff_opts = []
100
diffcmd = ['diff',
101
'--label', old_label,
102
oldtmpf.name,
103
'--label', new_label,
104
newtmpf.name]
105
106
# diff only allows one style to be specified; they don't override.
107
# note that some of these take optargs, and the optargs can be
108
# directly appended to the options.
109
# this is only an approximate parser; it doesn't properly understand
110
# the grammar.
111
for s in ['-c', '-u', '-C', '-U',
112
'-e', '--ed',
113
'-q', '--brief',
114
'--normal',
115
'-n', '--rcs',
116
'-y', '--side-by-side',
117
'-D', '--ifdef']:
118
for j in diff_opts:
119
if j.startswith(s):
120
break
121
else:
122
continue
123
break
124
else:
125
diffcmd.append('-u')
126
127
if diff_opts:
128
diffcmd.extend(diff_opts)
129
130
rc = os.spawnvp(os.P_WAIT, 'diff', diffcmd)
131
132
if rc != 0 and rc != 1:
133
# returns 1 if files differ; that's OK
134
if rc < 0:
135
msg = 'signal %d' % (-rc)
136
else:
137
msg = 'exit code %d' % rc
138
139
raise BzrError('external diff failed with %s; command: %r' % (rc, diffcmd))
140
finally:
141
oldtmpf.close() # and delete
142
newtmpf.close()
143
144
145
146
def show_diff(b, revision, specific_files, external_diff_options=None):
147
"""Shortcut for showing the diff to the working tree.
148
149
b
150
Branch.
151
152
revision
153
None for each, or otherwise the old revision to compare against.
154
155
The more general form is show_diff_trees(), where the caller
156
supplies any two trees.
32
157
"""
33
34
## TODO: Compare files before diffing; only mention those that have changed
35
36
## TODO: Set nice names in the headers, maybe include diffstat
37
38
## TODO: Perhaps make this a generator rather than using
39
## a callback object?
40
41
## TODO: Allow specifying a list of files to compare, rather than
42
## doing the whole tree? (Not urgent.)
43
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.
48
49
## XXX: This doesn't report on unknown files; that can be done
50
## from a separate method.
51
52
old_it = old_tree.list_files()
53
new_it = new_tree.list_files()
54
55
def next(it):
56
try:
57
return it.next()
58
except StopIteration:
59
return None
60
61
old_item = next(old_it)
62
new_item = next(new_it)
63
64
# We step through the two sorted iterators in parallel, trying to
65
# keep them lined up.
66
67
while (old_item != None) or (new_item != None):
68
# OK, we still have some remaining on both, but they may be
69
# out of step.
70
if old_item != None:
71
old_name, old_class, old_kind, old_id = old_item
72
else:
73
old_name = None
74
75
if new_item != None:
76
new_name, new_class, new_kind, new_id = new_item
77
else:
78
new_name = None
79
80
mutter(" diff pairwise %r" % (old_item,))
81
mutter(" %r" % (new_item,))
82
83
if old_item:
84
# can't handle the old tree being a WorkingTree
85
assert old_class == 'V'
86
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
94
pass
95
else:
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
102
else:
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
114
else:
115
yield 'A', new_id, None, new_name, new_kind
116
117
new_item = next(new_it)
118
old_item = next(old_it)
119
else:
120
assert old_id == new_id
121
assert old_id != None
122
assert old_name == new_name
123
assert old_kind == new_kind
124
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
134
else:
135
mutter(" quick compare shows different")
136
yield 'M', new_id, old_name, new_name, new_kind
137
138
new_item = next(new_it)
139
old_item = next(old_it)
140
141
142
143
def show_diff(b, revision, file_list):
144
import difflib, sys
145
158
import sys
159
146
160
if revision == None:
147
161
old_tree = b.basis_tree()
148
162
else:
150
164
151
165
new_tree = b.working_tree()
152
166
167
show_diff_trees(old_tree, new_tree, sys.stdout, specific_files,
168
external_diff_options)
169
170
171
172
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None,
173
external_diff_options=None):
174
"""Show in text form the changes from one tree to another.
175
176
to_files
177
If set, include only changes to these files.
178
179
external_diff_options
180
If set, use an external GNU diff and pass these options.
181
"""
182
153
183
# TODO: Options to control putting on a prefix or suffix, perhaps as a format string
154
184
old_label = ''
155
185
new_label = ''
162
192
# TODO: Generation of pseudo-diffs for added/deleted files could
163
193
# be usefully made into a much faster special case.
164
194
165
# TODO: Better to return them in sorted order I think.
166
167
if file_list:
168
file_list = [b.relpath(f) for f in file_list]
169
170
# FIXME: If given a file list, compare only those files rather
171
# than comparing everything and then throwing stuff away.
172
173
for file_state, fid, old_name, new_name, kind in diff_trees(old_tree, new_tree):
174
175
if file_list and (new_name not in file_list):
176
continue
177
178
# Don't show this by default; maybe do it if an option is passed
179
# idlabel = ' {%s}' % fid
180
idlabel = ''
181
182
# FIXME: Something about the diff format makes patch unhappy
183
# with newly-added files.
184
185
def diffit(oldlines, newlines, **kw):
186
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.
193
194
# In the meantime we at least make sure the patch isn't
195
# mangled.
196
197
198
# Special workaround for Python2.3, where difflib fails if
199
# both sequences are empty.
200
if not oldlines and not newlines:
201
return
202
203
nonl = False
204
205
if oldlines and (oldlines[-1][-1] != '\n'):
206
oldlines[-1] += '\n'
207
nonl = True
208
if newlines and (newlines[-1][-1] != '\n'):
209
newlines[-1] += '\n'
210
nonl = True
211
212
ud = difflib.unified_diff(oldlines, newlines, **kw)
213
sys.stdout.writelines(ud)
214
if nonl:
215
print "\\ No newline at end of file"
216
sys.stdout.write('\n')
217
218
if file_state in ['.', '?', 'I']:
219
continue
220
elif file_state == 'A':
221
print '*** added %s %r' % (kind, new_name)
222
if kind == 'file':
223
diffit([],
224
new_tree.get_file(fid).readlines(),
225
fromfile=DEVNULL,
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)
230
if kind == 'file':
231
diffit(old_tree.get_file(fid).readlines(), [],
232
fromfile=old_label + old_name + idlabel,
233
tofile=DEVNULL)
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)
241
242
if kind == 'file':
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)
195
if external_diff_options:
196
assert isinstance(external_diff_options, basestring)
197
opts = external_diff_options.split()
198
def diff_file(olab, olines, nlab, nlines, to_file):
199
external_diff(olab, olines, nlab, nlines, to_file, opts)
200
else:
201
diff_file = internal_diff
202
203
204
delta = compare_trees(old_tree, new_tree, want_unchanged=False,
205
specific_files=specific_files)
206
207
for path, file_id, kind in delta.removed:
208
print >>to_file, '*** removed %s %r' % (kind, path)
209
if kind == 'file':
210
diff_file(old_label + path,
211
old_tree.get_file(file_id).readlines(),
212
DEVNULL,
213
[],
214
to_file)
215
216
for path, file_id, kind in delta.added:
217
print >>to_file, '*** added %s %r' % (kind, path)
218
if kind == 'file':
219
diff_file(DEVNULL,
220
[],
221
new_label + path,
222
new_tree.get_file(file_id).readlines(),
223
to_file)
224
225
for old_path, new_path, file_id, kind, text_modified in delta.renamed:
226
print >>to_file, '*** renamed %s %r => %r' % (kind, old_path, new_path)
227
if text_modified:
228
diff_file(old_label + old_path,
229
old_tree.get_file(file_id).readlines(),
230
new_label + new_path,
231
new_tree.get_file(file_id).readlines(),
232
to_file)
233
234
for path, file_id, kind in delta.modified:
235
print >>to_file, '*** modified %s %r' % (kind, path)
236
if kind == 'file':
237
diff_file(old_label + path,
238
old_tree.get_file(file_id).readlines(),
239
new_label + path,
240
new_tree.get_file(file_id).readlines(),
241
to_file)
242
243
244
245
class TreeDelta(object):
246
"""Describes changes from one tree to another.
247
248
Contains four lists:
249
250
added
251
(path, id, kind)
252
removed
253
(path, id, kind)
254
renamed
255
(oldpath, newpath, id, kind, text_modified)
256
modified
257
(path, id, kind)
258
unchanged
259
(path, id, kind)
260
261
Each id is listed only once.
262
263
Files that are both modified and renamed are listed only in
264
renamed, with the text_modified flag true.
265
266
Files are only considered renamed if their name has changed or
267
their parent directory has changed. Renaming a directory
268
does not count as renaming all its contents.
269
270
The lists are normally sorted when the delta is created.
271
"""
272
def __init__(self):
273
self.added = []
274
self.removed = []
275
self.renamed = []
276
self.modified = []
277
self.unchanged = []
278
279
def __eq__(self, other):
280
if not isinstance(other, TreeDelta):
281
return False
282
return self.added == other.added \
283
and self.removed == other.removed \
284
and self.renamed == other.renamed \
285
and self.modified == other.modified \
286
and self.unchanged == other.unchanged
287
288
def __ne__(self, other):
289
return not (self == other)
290
291
def __repr__(self):
292
return "TreeDelta(added=%r, removed=%r, renamed=%r, modified=%r," \
293
" unchanged=%r)" % (self.added, self.removed, self.renamed,
294
self.modified, self.unchanged)
295
296
def has_changed(self):
297
changes = len(self.added) + len(self.removed) + len(self.renamed)
298
changes += len(self.modified)
299
return (changes != 0)
300
301
def touches_file_id(self, file_id):
302
"""Return True if file_id is modified by this delta."""
303
for l in self.added, self.removed, self.modified:
304
for v in l:
305
if v[1] == file_id:
306
return True
307
for v in self.renamed:
308
if v[2] == file_id:
309
return True
310
return False
311
312
313
def show(self, to_file, show_ids=False, show_unchanged=False):
314
def show_list(files):
315
for path, fid, kind in files:
316
if kind == 'directory':
317
path += '/'
318
elif kind == 'symlink':
319
path += '@'
320
321
if show_ids:
322
print >>to_file, ' %-30s %s' % (path, fid)
323
else:
324
print >>to_file, ' ', path
325
326
if self.removed:
327
print >>to_file, 'removed:'
328
show_list(self.removed)
329
330
if self.added:
331
print >>to_file, 'added:'
332
show_list(self.added)
333
334
if self.renamed:
335
print >>to_file, 'renamed:'
336
for oldpath, newpath, fid, kind, text_modified in self.renamed:
337
if show_ids:
338
print >>to_file, ' %s => %s %s' % (oldpath, newpath, fid)
339
else:
340
print >>to_file, ' %s => %s' % (oldpath, newpath)
341
342
if self.modified:
343
print >>to_file, 'modified:'
344
show_list(self.modified)
345
346
if show_unchanged and self.unchanged:
347
print >>to_file, 'unchanged:'
348
show_list(self.unchanged)
349
350
351
352
def compare_trees(old_tree, new_tree, want_unchanged=False, specific_files=None):
353
"""Describe changes from one tree to another.
354
355
Returns a TreeDelta with details of added, modified, renamed, and
356
deleted entries.
357
358
The root entry is specifically exempt.
359
360
This only considers versioned files.
361
362
want_unchanged
363
If true, also list files unchanged from one version to
364
the next.
365
366
specific_files
367
If true, only check for changes to specified names or
368
files within them.
369
"""
370
371
from osutils import is_inside_any
372
373
old_inv = old_tree.inventory
374
new_inv = new_tree.inventory
375
delta = TreeDelta()
376
mutter('start compare_trees')
377
378
# TODO: match for specific files can be rather smarter by finding
379
# the IDs of those files up front and then considering only that.
380
381
for file_id in old_tree:
382
if file_id in new_tree:
383
old_ie = old_inv[file_id]
384
new_ie = new_inv[file_id]
385
386
kind = old_ie.kind
387
assert kind == new_ie.kind
388
389
assert kind in ('file', 'directory', 'symlink', 'root_directory'), \
390
'invalid file kind %r' % kind
391
392
if kind == 'root_directory':
393
continue
394
395
if specific_files:
396
if (not is_inside_any(specific_files, old_inv.id2path(file_id))
397
and not is_inside_any(specific_files, new_inv.id2path(file_id))):
398
continue
399
400
if kind == 'file':
401
old_sha1 = old_tree.get_file_sha1(file_id)
402
new_sha1 = new_tree.get_file_sha1(file_id)
403
text_modified = (old_sha1 != new_sha1)
404
else:
405
## mutter("no text to check for %r %r" % (file_id, kind))
406
text_modified = False
407
408
# TODO: Can possibly avoid calculating path strings if the
409
# two files are unchanged and their names and parents are
410
# the same and the parents are unchanged all the way up.
411
# May not be worthwhile.
412
413
if (old_ie.name != new_ie.name
414
or old_ie.parent_id != new_ie.parent_id):
415
delta.renamed.append((old_inv.id2path(file_id),
416
new_inv.id2path(file_id),
417
file_id, kind,
418
text_modified))
419
elif text_modified:
420
delta.modified.append((new_inv.id2path(file_id), file_id, kind))
421
elif want_unchanged:
422
delta.unchanged.append((new_inv.id2path(file_id), file_id, kind))
247
423
else:
248
bailout("can't represent state %s {%s}" % (file_state, fid))
249
250
424
kind = old_inv.get_file_kind(file_id)
425
old_path = old_inv.id2path(file_id)
426
if specific_files:
427
if not is_inside_any(specific_files, old_path):
428
continue
429
delta.removed.append((old_path, file_id, kind))
430
431
mutter('start looking for new files')
432
for file_id in new_inv:
433
if file_id in old_inv:
434
continue
435
new_path = new_inv.id2path(file_id)
436
if specific_files:
437
if not is_inside_any(specific_files, new_path):
438
continue
439
kind = new_inv.get_file_kind(file_id)
440
delta.added.append((new_path, file_id, kind))
441
442
delta.removed.sort()
443
delta.added.sort()
444
delta.renamed.sort()
445
delta.modified.sort()
446
delta.unchanged.sort()
447
448
return delta
Loggerhead is a web-based interface for Breezy
Version: 2.0.1