To get this branch, use:
bzr branch
http://gegoxaren.bato24.eu/bzr/brz/remove-bazaar
« back to all changes in this revision
Viewing changes to bzrlib/diff.py
- browse files at revision 468
- compare with another revision
- download diff
- download tarball
- view history from revision 468
- Committer: Martin Pool
- Date: 2005-05-11 06:05:28 UTC
- Revision ID: mbp@sourcefrog.net-20050511060528-b2b35cafa4e1785d
- Interpret arguments to bzr status
- Handle giving a non-default branch to bzr status
- bzr status docs
- Handle giving a non-default branch to bzr status
- bzr status docs
- files added:
- bzrlib/statcache.py
- files removed:
- bzr-man.py
- bzrlib/selftest
- contrib/emacs
- effbot
- effbot/org
- patches
- plugins
- plugins/changeset
- tools
- files renamed:
- contrib/bash/bzr.simple => contrib/bash/bzr
- 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 bzrlib.trace import mutter
19
from bzrlib.errors import BzrError
20
from bzrlib.delta import compare_trees
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.
18
from sets import Set
19
20
from trace import mutter
21
from errors import BzrError
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
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.
157
36
"""
158
import sys
159
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
160
136
if revision == None:
161
137
old_tree = b.basis_tree()
162
138
else:
164
140
165
141
new_tree = b.working_tree()
166
142
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
183
143
# TODO: Options to control putting on a prefix or suffix, perhaps as a format string
184
144
old_label = ''
185
145
new_label = ''
192
152
# TODO: Generation of pseudo-diffs for added/deleted files could
193
153
# be usefully made into a much faster special case.
194
154
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
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.
202
162
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
246
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):
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
if old_path != new_path:
334
delta.renamed.append((old_path, new_path, file_id, text_modified))
335
elif text_modified:
336
delta.modified.append((new_path, file_id))
337
else:
338
delta.unchanged.append((new_path, file_id))
339
else:
340
delta.removed.append((old_inv.id2path(file_id), file_id))
341
for file_id in new_inv:
342
if file_id in old_inv:
343
continue
344
delta.added.append((new_inv.id2path(file_id), file_id))
345
346
delta.removed.sort()
347
delta.added.sort()
348
delta.renamed.sort()
349
delta.modified.sort()
350
351
return delta
Loggerhead is a web-based interface for Breezy
Version: 2.0.2.dev0