« back to all changes in this revision
Viewing changes to bzrlib/diff.py
- browse files at revision 572
- compare with another revision
- download diff
- download tarball
- view history from revision 572
- Committer: Martin Pool
- Date: 2005-05-27 03:44:32 UTC
- Revision ID: mbp@sourcefrog.net-20050527034432-647e59832d01909f
- trim imports
- files added:
- .rsyncexclude
- contrib
- contrib/bash
- contrib/zsh
- patches
- urlgrabber
- files removed:
- bzrlib/tests.py
- 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
25
26
27
def diff_trees(old_tree, new_tree):
28
"""Compute diff between two trees.
29
30
They may be in different branches and may be working or historical
31
trees.
32
33
Yields a sequence of (state, id, old_name, new_name, kind).
34
Each filename and each id is listed only once.
35
"""
36
37
## TODO: Compare files before diffing; only mention those that have changed
38
39
## TODO: Set nice names in the headers, maybe include diffstat
40
41
## TODO: Perhaps make this a generator rather than using
42
## a callback object?
43
44
## TODO: Allow specifying a list of files to compare, rather than
45
## doing the whole tree? (Not urgent.)
46
47
## TODO: Allow diffing any two inventories, not just the
48
## current one against one. We mgiht need to specify two
49
## stores to look for the files if diffing two branches. That
50
## might imply this shouldn't be primarily a Branch method.
51
52
## XXX: This doesn't report on unknown files; that can be done
53
## from a separate method.
54
55
old_it = old_tree.list_files()
56
new_it = new_tree.list_files()
57
58
def next(it):
59
try:
60
return it.next()
61
except StopIteration:
62
return None
63
64
old_item = next(old_it)
65
new_item = next(new_it)
66
67
# We step through the two sorted iterators in parallel, trying to
68
# keep them lined up.
69
70
while (old_item != None) or (new_item != None):
71
# OK, we still have some remaining on both, but they may be
72
# out of step.
73
if old_item != None:
74
old_name, old_class, old_kind, old_id = old_item
75
else:
76
old_name = None
77
78
if new_item != None:
79
new_name, new_class, new_kind, new_id = new_item
80
else:
81
new_name = None
82
83
mutter(" diff pairwise %r" % (old_item,))
84
mutter(" %r" % (new_item,))
85
86
if old_item:
87
# can't handle the old tree being a WorkingTree
88
assert old_class == 'V'
89
90
if new_item and (new_class != 'V'):
91
yield new_class, None, None, new_name, new_kind
92
new_item = next(new_it)
93
elif (not new_item) or (old_item and (old_name < new_name)):
94
mutter(" extra entry in old-tree sequence")
95
if new_tree.has_id(old_id):
96
# will be mentioned as renamed under new name
97
pass
98
else:
99
yield 'D', old_id, old_name, None, old_kind
100
old_item = next(old_it)
101
elif (not old_item) or (new_item and (new_name < old_name)):
102
mutter(" extra entry in new-tree sequence")
103
if old_tree.has_id(new_id):
104
yield 'R', new_id, old_tree.id2path(new_id), new_name, new_kind
105
else:
106
yield 'A', new_id, None, new_name, new_kind
107
new_item = next(new_it)
108
elif old_id != new_id:
109
assert old_name == new_name
110
# both trees have a file of this name, but it is not the
111
# same file. in other words, the old filename has been
112
# overwritten by either a newly-added or a renamed file.
113
# (should we return something about the overwritten file?)
114
if old_tree.has_id(new_id):
115
# renaming, overlying a deleted file
116
yield 'R', new_id, old_tree.id2path(new_id), new_name, new_kind
117
else:
118
yield 'A', new_id, None, new_name, new_kind
119
120
new_item = next(new_it)
121
old_item = next(old_it)
122
else:
123
assert old_id == new_id
124
assert old_name == new_name
125
assert old_kind == new_kind
126
127
if old_kind == 'directory':
128
yield '.', new_id, old_name, new_name, new_kind
129
elif old_tree.get_file_size(old_id) != new_tree.get_file_size(old_id):
130
mutter(" file size has changed, must be different")
131
yield 'M', new_id, old_name, new_name, new_kind
132
elif old_tree.get_file_sha1(old_id) == new_tree.get_file_sha1(old_id):
133
mutter(" SHA1 indicates they're identical")
134
## assert compare_files(old_tree.get_file(i), new_tree.get_file(i))
135
yield '.', new_id, old_name, new_name, new_kind
136
else:
137
mutter(" quick compare shows different")
138
yield 'M', new_id, old_name, new_name, new_kind
139
140
new_item = next(new_it)
141
old_item = next(old_it)
142
143
19
from errors import BzrError
20
21
22
def internal_diff(old_label, oldlines, new_label, newlines, to_file):
23
import difflib
24
25
# FIXME: difflib is wrong if there is no trailing newline.
26
# The syntax used by patch seems to be "\ No newline at
27
# end of file" following the last diff line from that
28
# file. This is not trivial to insert into the
29
# unified_diff output and it might be better to just fix
30
# or replace that function.
31
32
# In the meantime we at least make sure the patch isn't
33
# mangled.
34
35
36
# Special workaround for Python2.3, where difflib fails if
37
# both sequences are empty.
38
if not oldlines and not newlines:
39
return
40
41
nonl = False
42
43
if oldlines and (oldlines[-1][-1] != '\n'):
44
oldlines[-1] += '\n'
45
nonl = True
46
if newlines and (newlines[-1][-1] != '\n'):
47
newlines[-1] += '\n'
48
nonl = True
49
50
ud = difflib.unified_diff(oldlines, newlines,
51
fromfile=old_label, tofile=new_label)
52
53
# work-around for difflib being too smart for its own good
54
# if /dev/null is "1,0", patch won't recognize it as /dev/null
55
if not oldlines:
56
ud = list(ud)
57
ud[2] = ud[2].replace('-1,0', '-0,0')
58
elif not newlines:
59
ud = list(ud)
60
ud[2] = ud[2].replace('+1,0', '+0,0')
61
62
to_file.writelines(ud)
63
if nonl:
64
print >>to_file, "\\ No newline at end of file"
65
print >>to_file
66
67
68
69
70
def external_diff(old_label, oldlines, new_label, newlines, to_file,
71
diff_opts):
72
"""Display a diff by calling out to the external diff program."""
73
import sys
74
75
if to_file != sys.stdout:
76
raise NotImplementedError("sorry, can't send external diff other than to stdout yet",
77
to_file)
78
79
from tempfile import NamedTemporaryFile
80
import os
81
82
oldtmpf = NamedTemporaryFile()
83
newtmpf = NamedTemporaryFile()
84
85
try:
86
# TODO: perhaps a special case for comparing to or from the empty
87
# sequence; can just use /dev/null on Unix
88
89
# TODO: if either of the files being compared already exists as a
90
# regular named file (e.g. in the working directory) then we can
91
# compare directly to that, rather than copying it.
92
93
oldtmpf.writelines(oldlines)
94
newtmpf.writelines(newlines)
95
96
oldtmpf.flush()
97
newtmpf.flush()
98
99
if not diff_opts:
100
diff_opts = []
101
diffcmd = ['diff',
102
'--label', old_label,
103
oldtmpf.name,
104
'--label', new_label,
105
newtmpf.name]
106
107
# diff only allows one style to be specified; they don't override.
108
# note that some of these take optargs, and the optargs can be
109
# directly appended to the options.
110
# this is only an approximate parser; it doesn't properly understand
111
# the grammar.
112
for s in ['-c', '-u', '-C', '-U',
113
'-e', '--ed',
114
'-q', '--brief',
115
'--normal',
116
'-n', '--rcs',
117
'-y', '--side-by-side',
118
'-D', '--ifdef']:
119
for j in diff_opts:
120
if j.startswith(s):
121
break
122
else:
123
continue
124
break
125
else:
126
diffcmd.append('-u')
127
128
if diff_opts:
129
diffcmd.extend(diff_opts)
130
131
rc = os.spawnvp(os.P_WAIT, 'diff', diffcmd)
132
133
if rc != 0 and rc != 1:
134
# returns 1 if files differ; that's OK
135
if rc < 0:
136
msg = 'signal %d' % (-rc)
137
else:
138
msg = 'exit code %d' % rc
139
140
raise BzrError('external diff failed with %s; command: %r' % (rc, diffcmd))
141
finally:
142
oldtmpf.close() # and delete
143
newtmpf.close()
144
145
146
147
def show_diff(b, revision, specific_files, external_diff_options=None):
148
import sys
149
150
if revision == None:
151
old_tree = b.basis_tree()
152
else:
153
old_tree = b.revision_tree(b.lookup_revision(revision))
154
155
new_tree = b.working_tree()
156
157
show_diff_trees(old_tree, new_tree, sys.stdout, specific_files,
158
external_diff_options)
159
160
161
162
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None,
163
external_diff_options=None):
164
"""Show in text form the changes from one tree to another.
165
166
to_files
167
If set, include only changes to these files.
168
169
external_diff_options
170
If set, use an external GNU diff and pass these options.
171
"""
172
173
# TODO: Options to control putting on a prefix or suffix, perhaps as a format string
174
old_label = ''
175
new_label = ''
176
177
DEVNULL = '/dev/null'
178
# Windows users, don't panic about this filename -- it is a
179
# special signal to GNU patch that the file should be created or
180
# deleted respectively.
181
182
# TODO: Generation of pseudo-diffs for added/deleted files could
183
# be usefully made into a much faster special case.
184
185
if external_diff_options:
186
assert isinstance(external_diff_options, basestring)
187
opts = external_diff_options.split()
188
def diff_file(olab, olines, nlab, nlines, to_file):
189
external_diff(olab, olines, nlab, nlines, to_file, opts)
190
else:
191
diff_file = internal_diff
192
193
194
delta = compare_trees(old_tree, new_tree, want_unchanged=False,
195
specific_files=specific_files)
196
197
for path, file_id, kind in delta.removed:
198
print '*** removed %s %r' % (kind, path)
199
if kind == 'file':
200
diff_file(old_label + path,
201
old_tree.get_file(file_id).readlines(),
202
DEVNULL,
203
[],
204
to_file)
205
206
for path, file_id, kind in delta.added:
207
print '*** added %s %r' % (kind, path)
208
if kind == 'file':
209
diff_file(DEVNULL,
210
[],
211
new_label + path,
212
new_tree.get_file(file_id).readlines(),
213
to_file)
214
215
for old_path, new_path, file_id, kind, text_modified in delta.renamed:
216
print '*** renamed %s %r => %r' % (kind, old_path, new_path)
217
if text_modified:
218
diff_file(old_label + old_path,
219
old_tree.get_file(file_id).readlines(),
220
new_label + new_path,
221
new_tree.get_file(file_id).readlines(),
222
to_file)
223
224
for path, file_id, kind in delta.modified:
225
print '*** modified %s %r' % (kind, path)
226
if kind == 'file':
227
diff_file(old_label + path,
228
old_tree.get_file(file_id).readlines(),
229
new_label + path,
230
new_tree.get_file(file_id).readlines(),
231
to_file)
232
233
234
235
class TreeDelta(object):
236
"""Describes changes from one tree to another.
237
238
Contains four lists:
239
240
added
241
(path, id, kind)
242
removed
243
(path, id, kind)
244
renamed
245
(oldpath, newpath, id, kind, text_modified)
246
modified
247
(path, id, kind)
248
unchanged
249
(path, id, kind)
250
251
Each id is listed only once.
252
253
Files that are both modified and renamed are listed only in
254
renamed, with the text_modified flag true.
255
256
The lists are normally sorted when the delta is created.
257
"""
258
def __init__(self):
259
self.added = []
260
self.removed = []
261
self.renamed = []
262
self.modified = []
263
self.unchanged = []
264
265
266
def touches_file_id(self, file_id):
267
"""Return True if file_id is modified by this delta."""
268
for l in self.added, self.removed, self.modified:
269
for v in l:
270
if v[1] == file_id:
271
return True
272
for v in self.renamed:
273
if v[2] == file_id:
274
return True
275
return False
276
277
278
def show(self, to_file, show_ids=False, show_unchanged=False):
279
def show_list(files):
280
for path, fid, kind in files:
281
if kind == 'directory':
282
path += '/'
283
elif kind == 'symlink':
284
path += '@'
285
286
if show_ids:
287
print >>to_file, ' %-30s %s' % (path, fid)
288
else:
289
print >>to_file, ' ', path
290
291
if self.removed:
292
print >>to_file, 'removed:'
293
show_list(self.removed)
294
295
if self.added:
296
print >>to_file, 'added:'
297
show_list(self.added)
298
299
if self.renamed:
300
print >>to_file, 'renamed:'
301
for oldpath, newpath, fid, kind, text_modified in self.renamed:
302
if show_ids:
303
print >>to_file, ' %s => %s %s' % (oldpath, newpath, fid)
304
else:
305
print >>to_file, ' %s => %s' % (oldpath, newpath)
306
307
if self.modified:
308
print >>to_file, 'modified:'
309
show_list(self.modified)
310
311
if show_unchanged and self.unchanged:
312
print >>to_file, 'unchanged:'
313
show_list(self.unchanged)
314
315
316
317
def compare_trees(old_tree, new_tree, want_unchanged, specific_files=None):
318
"""Describe changes from one tree to another.
319
320
Returns a TreeDelta with details of added, modified, renamed, and
321
deleted entries.
322
323
The root entry is specifically exempt.
324
325
This only considers versioned files.
326
327
want_unchanged
328
If true, also list files unchanged from one version to
329
the next.
330
331
specific_files
332
If true, only check for changes to specified names or
333
files within them.
334
"""
335
336
from osutils import is_inside_any
337
338
old_inv = old_tree.inventory
339
new_inv = new_tree.inventory
340
delta = TreeDelta()
341
mutter('start compare_trees')
342
343
# TODO: match for specific files can be rather smarter by finding
344
# the IDs of those files up front and then considering only that.
345
346
for file_id in old_tree:
347
if file_id in new_tree:
348
kind = old_inv.get_file_kind(file_id)
349
assert kind == new_inv.get_file_kind(file_id)
350
351
assert kind in ('file', 'directory', 'symlink', 'root_directory'), \
352
'invalid file kind %r' % kind
353
354
if kind == 'root_directory':
355
continue
356
357
old_path = old_inv.id2path(file_id)
358
new_path = new_inv.id2path(file_id)
359
360
if specific_files:
361
if (not is_inside_any(specific_files, old_path)
362
and not is_inside_any(specific_files, new_path)):
363
continue
364
365
if kind == 'file':
366
old_sha1 = old_tree.get_file_sha1(file_id)
367
new_sha1 = new_tree.get_file_sha1(file_id)
368
text_modified = (old_sha1 != new_sha1)
369
else:
370
## mutter("no text to check for %r %r" % (file_id, kind))
371
text_modified = False
372
373
# TODO: Can possibly avoid calculating path strings if the
374
# two files are unchanged and their names and parents are
375
# the same and the parents are unchanged all the way up.
376
# May not be worthwhile.
377
378
if old_path != new_path:
379
delta.renamed.append((old_path, new_path, file_id, kind,
380
text_modified))
381
elif text_modified:
382
delta.modified.append((new_path, file_id, kind))
383
elif want_unchanged:
384
delta.unchanged.append((new_path, file_id, kind))
385
else:
386
kind = old_inv.get_file_kind(file_id)
387
old_path = old_inv.id2path(file_id)
388
if specific_files:
389
if not is_inside_any(specific_files, old_path):
390
continue
391
delta.removed.append((old_path, file_id, kind))
392
393
mutter('start looking for new files')
394
for file_id in new_inv:
395
if file_id in old_inv:
396
continue
397
new_path = new_inv.id2path(file_id)
398
if specific_files:
399
if not is_inside_any(specific_files, new_path):
400
continue
401
kind = new_inv.get_file_kind(file_id)
402
delta.added.append((new_path, file_id, kind))
403
404
delta.removed.sort()
405
delta.added.sort()
406
delta.renamed.sort()
407
delta.modified.sort()
408
delta.unchanged.sort()
409
410
return delta
Loggerhead is a web-based interface for Breezy
Version: 2.0.1