« back to all changes in this revision
Viewing changes to bzrlib/diff.py
- browse files at revision 542
- compare with another revision
- download diff
- download tarball
- view history from revision 542
- Committer: Martin Pool
- Date: 2005-05-19 08:31:06 UTC
- Revision ID: mbp@sourcefrog.net-20050519083106-ebe71562d3bda4a7
- fix typo
- files added:
- bzrlib/atomicfile.py
- patches
- 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
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.
32
"""
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
19
from errors import BzrError
20
21
22
23
def _diff_one(oldlines, newlines, to_file, **kw):
24
import difflib
145
25
26
# FIXME: difflib is wrong if there is no trailing newline.
27
# The syntax used by patch seems to be "\ No newline at
28
# end of file" following the last diff line from that
29
# file. This is not trivial to insert into the
30
# unified_diff output and it might be better to just fix
31
# or replace that function.
32
33
# In the meantime we at least make sure the patch isn't
34
# mangled.
35
36
37
# Special workaround for Python2.3, where difflib fails if
38
# both sequences are empty.
39
if not oldlines and not newlines:
40
return
41
42
nonl = False
43
44
if oldlines and (oldlines[-1][-1] != '\n'):
45
oldlines[-1] += '\n'
46
nonl = True
47
if newlines and (newlines[-1][-1] != '\n'):
48
newlines[-1] += '\n'
49
nonl = True
50
51
ud = difflib.unified_diff(oldlines, newlines, **kw)
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
def show_diff(b, revision, specific_files):
69
import sys
70
146
71
if revision == None:
147
72
old_tree = b.basis_tree()
148
73
else:
162
87
# TODO: Generation of pseudo-diffs for added/deleted files could
163
88
# be usefully made into a much faster special case.
164
89
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.
90
delta = compare_trees(old_tree, new_tree, want_unchanged=False,
91
specific_files=specific_files)
92
93
for path, file_id, kind in delta.removed:
94
print '*** removed %s %r' % (kind, path)
95
if kind == 'file':
96
_diff_one(old_tree.get_file(file_id).readlines(),
97
[],
98
sys.stdout,
99
fromfile=old_label + path,
100
tofile=DEVNULL)
101
102
for path, file_id, kind in delta.added:
103
print '*** added %s %r' % (kind, path)
104
if kind == 'file':
105
_diff_one([],
106
new_tree.get_file(file_id).readlines(),
107
sys.stdout,
108
fromfile=DEVNULL,
109
tofile=new_label + path)
110
111
for old_path, new_path, file_id, kind, text_modified in delta.renamed:
112
print '*** renamed %s %r => %r' % (kind, old_path, new_path)
113
if text_modified:
114
_diff_one(old_tree.get_file(file_id).readlines(),
115
new_tree.get_file(file_id).readlines(),
116
sys.stdout,
117
fromfile=old_label + old_path,
118
tofile=new_label + new_path)
119
120
for path, file_id, kind in delta.modified:
121
print '*** modified %s %r' % (kind, path)
122
if kind == 'file':
123
_diff_one(old_tree.get_file(file_id).readlines(),
124
new_tree.get_file(file_id).readlines(),
125
sys.stdout,
126
fromfile=old_label + path,
127
tofile=new_label + path)
128
129
130
131
class TreeDelta:
132
"""Describes changes from one tree to another.
133
134
Contains four lists:
135
136
added
137
(path, id, kind)
138
removed
139
(path, id, kind)
140
renamed
141
(oldpath, newpath, id, kind, text_modified)
142
modified
143
(path, id, kind)
144
unchanged
145
(path, id, kind)
146
147
Each id is listed only once.
148
149
Files that are both modified and renamed are listed only in
150
renamed, with the text_modified flag true.
151
152
The lists are normally sorted when the delta is created.
153
"""
154
def __init__(self):
155
self.added = []
156
self.removed = []
157
self.renamed = []
158
self.modified = []
159
self.unchanged = []
160
161
162
def touches_file_id(self, file_id):
163
"""Return True if file_id is modified by this delta."""
164
for l in self.added, self.removed, self.modified:
165
for v in l:
166
if v[1] == file_id:
167
return True
168
for v in self.renamed:
169
if v[2] == file_id:
170
return True
171
return False
172
173
174
def show(self, to_file, show_ids=False, show_unchanged=False):
175
def show_list(files):
176
for path, fid, kind in files:
177
if kind == 'directory':
178
path += '/'
179
elif kind == 'symlink':
180
path += '@'
181
182
if show_ids:
183
print >>to_file, ' %-30s %s' % (path, fid)
184
else:
185
print >>to_file, ' ', path
186
187
if self.removed:
188
print >>to_file, 'removed:'
189
show_list(self.removed)
190
191
if self.added:
192
print >>to_file, 'added:'
193
show_list(self.added)
194
195
if self.renamed:
196
print >>to_file, 'renamed:'
197
for oldpath, newpath, fid, kind, text_modified in self.renamed:
198
if show_ids:
199
print >>to_file, ' %s => %s %s' % (oldpath, newpath, fid)
200
else:
201
print >>to_file, ' %s => %s' % (oldpath, newpath)
202
203
if self.modified:
204
print >>to_file, 'modified:'
205
show_list(self.modified)
206
207
if show_unchanged and self.unchanged:
208
print >>to_file, 'unchanged:'
209
show_list(self.unchanged)
210
211
212
213
def compare_trees(old_tree, new_tree, want_unchanged, specific_files=None):
214
"""Describe changes from one tree to another.
215
216
Returns a TreeDelta with details of added, modified, renamed, and
217
deleted entries.
218
219
The root entry is specifically exempt.
220
221
This only considers versioned files.
222
223
want_unchanged
224
If true, also list files unchanged from one version to
225
the next.
226
227
specific_files
228
If true, only check for changes to specified names or
229
files within them.
230
"""
231
232
from osutils import is_inside_any
172
233
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)
234
old_inv = old_tree.inventory
235
new_inv = new_tree.inventory
236
delta = TreeDelta()
237
mutter('start compare_trees')
238
239
# TODO: match for specific files can be rather smarter by finding
240
# the IDs of those files up front and then considering only that.
241
242
for file_id in old_tree:
243
if file_id in new_tree:
244
kind = old_inv.get_file_kind(file_id)
245
assert kind == new_inv.get_file_kind(file_id)
246
247
assert kind in ('file', 'directory', 'symlink', 'root_directory'), \
248
'invalid file kind %r' % kind
249
250
if kind == 'root_directory':
251
continue
252
253
old_path = old_inv.id2path(file_id)
254
new_path = new_inv.id2path(file_id)
255
256
if specific_files:
257
if (not is_inside_any(specific_files, old_path)
258
and not is_inside_any(specific_files, new_path)):
259
continue
260
261
if kind == 'file':
262
old_sha1 = old_tree.get_file_sha1(file_id)
263
new_sha1 = new_tree.get_file_sha1(file_id)
264
text_modified = (old_sha1 != new_sha1)
265
else:
266
## mutter("no text to check for %r %r" % (file_id, kind))
267
text_modified = False
268
269
# TODO: Can possibly avoid calculating path strings if the
270
# two files are unchanged and their names and parents are
271
# the same and the parents are unchanged all the way up.
272
# May not be worthwhile.
273
274
if old_path != new_path:
275
delta.renamed.append((old_path, new_path, file_id, kind,
276
text_modified))
277
elif text_modified:
278
delta.modified.append((new_path, file_id, kind))
279
elif want_unchanged:
280
delta.unchanged.append((new_path, file_id, kind))
247
281
else:
248
bailout("can't represent state %s {%s}" % (file_state, fid))
249
250
282
old_path = old_inv.id2path(file_id)
283
if specific_files:
284
if not is_inside_any(specific_files, old_path):
285
continue
286
delta.removed.append((old_path, file_id, kind))
287
288
mutter('start looking for new files')
289
for file_id in new_inv:
290
if file_id in old_inv:
291
continue
292
new_path = new_inv.id2path(file_id)
293
if specific_files:
294
if not is_inside_any(specific_files, new_path):
295
continue
296
kind = new_inv.get_file_kind(file_id)
297
delta.added.append((new_path, file_id, kind))
298
299
delta.removed.sort()
300
delta.added.sort()
301
delta.renamed.sort()
302
delta.modified.sort()
303
delta.unchanged.sort()
304
305
return delta
Loggerhead is a web-based interface for Breezy
Version: 2.0.1