« back to all changes in this revision
Viewing changes to bzrlib/diff.py
- browse files at revision 316
- compare with another revision
- download diff
- download tarball
- view history from revision 316
- Committer: Martin Pool
- Date: 2005-05-02 04:24:33 UTC
- Revision ID: mbp@sourcefrog.net-20050502042433-c825a7f7235f6b15
doc: notes on merge
- files added:
- doc/faq.txt
- files removed:
- bzrlib/atomicfile.py
- contrib
- contrib/bash
- contrib/zsh
- files modified:
- .bzrignore
added
removed
bzrlib/diff.py
18
18
from sets import Set
19
19
20
20
from trace import mutter
21
from errors import BzrError
21
22
23
24
22
25
23
26
24
27
def diff_trees(old_tree, new_tree):
27
30
They may be in different branches and may be working or historical
28
31
trees.
29
32
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
33
Yields a sequence of (state, id, old_name, new_name, kind).
35
34
Each filename and each id is listed only once.
36
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
37
44
## TODO: Allow specifying a list of files to compare, rather than
38
45
## doing the whole tree? (Not urgent.)
39
46
42
49
## stores to look for the files if diffing two branches. That
43
50
## might imply this shouldn't be primarily a Branch method.
44
51
45
sha_match_cnt = modified_cnt = 0
52
## XXX: This doesn't report on unknown files; that can be done
53
## from a separate method.
46
54
47
55
old_it = old_tree.list_files()
48
56
new_it = new_tree.list_files()
72
80
else:
73
81
new_name = None
74
82
83
mutter(" diff pairwise %r" % (old_item,))
84
mutter(" %r" % (new_item,))
85
75
86
if old_item:
76
87
# can't handle the old tree being a WorkingTree
77
88
assert old_class == 'V'
80
91
yield new_class, None, None, new_name, new_kind
81
92
new_item = next(new_it)
82
93
elif (not new_item) or (old_item and (old_name < new_name)):
94
mutter(" extra entry in old-tree sequence")
83
95
if new_tree.has_id(old_id):
84
96
# will be mentioned as renamed under new name
85
97
pass
87
99
yield 'D', old_id, old_name, None, old_kind
88
100
old_item = next(old_it)
89
101
elif (not old_item) or (new_item and (new_name < old_name)):
102
mutter(" extra entry in new-tree sequence")
90
103
if old_tree.has_id(new_id):
91
104
yield 'R', new_id, old_tree.id2path(new_id), new_name, new_kind
92
105
else:
114
127
115
128
if old_kind == 'directory':
116
129
yield '.', new_id, old_name, new_name, new_kind
130
elif old_tree.get_file_size(old_id) != new_tree.get_file_size(old_id):
131
mutter(" file size has changed, must be different")
132
yield 'M', new_id, old_name, new_name, new_kind
117
133
elif old_tree.get_file_sha1(old_id) == new_tree.get_file_sha1(old_id):
118
sha_match_cnt += 1
134
mutter(" SHA1 indicates they're identical")
135
## assert compare_files(old_tree.get_file(i), new_tree.get_file(i))
119
136
yield '.', new_id, old_name, new_name, new_kind
120
137
else:
121
modified_cnt += 1
138
mutter(" quick compare shows different")
122
139
yield 'M', new_id, old_name, new_name, new_kind
123
140
124
141
new_item = next(new_it)
125
142
old_item = next(old_it)
126
143
127
144
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
136
if revision == None:
137
old_tree = b.basis_tree()
138
else:
139
old_tree = b.revision_tree(b.lookup_revision(revision))
140
141
new_tree = b.working_tree()
142
143
# TODO: Options to control putting on a prefix or suffix, perhaps as a format string
144
old_label = ''
145
new_label = ''
146
147
DEVNULL = '/dev/null'
148
# Windows users, don't panic about this filename -- it is a
149
# special signal to GNU patch that the file should be created or
150
# deleted respectively.
151
152
# TODO: Generation of pseudo-diffs for added/deleted files could
153
# be usefully made into a much faster special case.
154
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.
162
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:
250
"""Describes changes from one tree to another.
251
252
Contains four lists:
253
254
added
255
(path, id)
256
removed
257
(path, id)
258
renamed
259
(oldpath, newpath, id, text_modified)
260
modified
261
(path, id)
262
unchanged
263
(path, id)
264
265
Each id is listed only once.
266
267
Files that are both modified and renamed are listed only in
268
renamed, with the text_modified flag true.
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 show(self, to_file, show_ids=False, show_unchanged=False):
280
def show_list(files):
281
for path, fid in files:
282
if show_ids:
283
print >>to_file, ' %-30s %s' % (path, fid)
284
else:
285
print >>to_file, ' ', path
286
287
if self.removed:
288
print >>to_file, 'removed files:'
289
show_list(self.removed)
290
291
if self.added:
292
print >>to_file, 'added files:'
293
show_list(self.added)
294
295
if self.renamed:
296
print >>to_file, 'renamed files:'
297
for oldpath, newpath, fid, text_modified in self.renamed:
298
if show_ids:
299
print >>to_file, ' %s => %s %s' % (oldpath, newpath, fid)
300
else:
301
print >>to_file, ' %s => %s' % (oldpath, newpath)
302
303
if self.modified:
304
print >>to_file, 'modified files:'
305
show_list(self.modified)
306
307
if show_unchanged and self.unchanged:
308
print >>to_file, 'unchanged files:'
309
show_list(self.unchanged)
310
311
312
313
def compare_trees(old_tree, new_tree, want_unchanged):
314
old_inv = old_tree.inventory
315
new_inv = new_tree.inventory
316
delta = TreeDelta()
317
for file_id in old_tree:
318
if file_id in new_tree:
319
old_path = old_inv.id2path(file_id)
320
new_path = new_inv.id2path(file_id)
321
322
kind = old_inv.get_file_kind(file_id)
323
assert kind in ('file', 'directory', 'symlink', 'root_directory'), \
324
'invalid file kind %r' % kind
325
if kind == 'file':
326
old_sha1 = old_tree.get_file_sha1(file_id)
327
new_sha1 = new_tree.get_file_sha1(file_id)
328
text_modified = (old_sha1 != new_sha1)
329
else:
330
## mutter("no text to check for %r %r" % (file_id, kind))
331
text_modified = False
332
333
# TODO: Can possibly avoid calculating path strings if the
334
# two files are unchanged and their names and parents are
335
# the same and the parents are unchanged all the way up.
336
# May not be worthwhile.
337
338
if old_path != new_path:
339
delta.renamed.append((old_path, new_path, file_id, text_modified))
340
elif text_modified:
341
delta.modified.append((new_path, file_id))
342
elif want_unchanged:
343
delta.unchanged.append((new_path, file_id))
344
else:
345
delta.removed.append((old_inv.id2path(file_id), file_id))
346
for file_id in new_inv:
347
if file_id in old_inv:
348
continue
349
delta.added.append((new_inv.id2path(file_id), file_id))
350
351
delta.removed.sort()
352
delta.added.sort()
353
delta.renamed.sort()
354
delta.modified.sort()
355
356
return delta
Loggerhead is a web-based interface for Breezy
Version: 2.0.1