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
18
from sets import Set, ImmutableSet
20
20
from trace import mutter
21
21
from errors import BzrError
24
def diff_trees(old_tree, new_tree):
25
"""Compute diff between two trees.
27
They may be in different branches and may be working or historical
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.
34
Yields a sequence of (state, id, old_name, new_name, kind).
35
Each filename and each id is listed only once.
37
## TODO: Allow specifying a list of files to compare, rather than
38
## doing the whole tree? (Not urgent.)
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.
45
sha_match_cnt = modified_cnt = 0
47
old_it = old_tree.list_files()
48
new_it = new_tree.list_files()
56
old_item = next(old_it)
57
new_item = next(new_it)
59
# We step through the two sorted iterators in parallel, trying to
62
while (old_item != None) or (new_item != None):
63
# OK, we still have some remaining on both, but they may be
66
old_name, old_class, old_kind, old_id = old_item
71
new_name, new_class, new_kind, new_id = new_item
76
# can't handle the old tree being a WorkingTree
77
assert old_class == 'V'
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
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
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
105
yield 'A', new_id, None, new_name, new_kind
107
new_item = next(new_it)
108
old_item = next(old_it)
110
assert old_id == new_id
111
assert old_id != None
112
assert old_name == new_name
113
assert old_kind == new_kind
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):
119
yield '.', new_id, old_name, new_name, new_kind
122
yield 'M', new_id, old_name, new_name, new_kind
124
new_item = next(new_it)
125
old_item = next(old_it)
128
mutter("diff finished: %d SHA matches, %d modified"
129
% (sha_match_cnt, modified_cnt))
133
def show_diff(b, revision, file_list):
134
import difflib, sys, types
25
def _diff_one(oldlines, newlines, to_file, **kw):
28
# FIXME: difflib is wrong if there is no trailing newline.
29
# The syntax used by patch seems to be "\ No newline at
30
# end of file" following the last diff line from that
31
# file. This is not trivial to insert into the
32
# unified_diff output and it might be better to just fix
33
# or replace that function.
35
# In the meantime we at least make sure the patch isn't
39
# Special workaround for Python2.3, where difflib fails if
40
# both sequences are empty.
41
if not oldlines and not newlines:
46
if oldlines and (oldlines[-1][-1] != '\n'):
49
if newlines and (newlines[-1][-1] != '\n'):
53
ud = difflib.unified_diff(oldlines, newlines, **kw)
55
# work-around for difflib being too smart for its own good
56
# if /dev/null is "1,0", patch won't recognize it as /dev/null
59
ud[2] = ud[2].replace('-1,0', '-0,0')
62
ud[2] = ud[2].replace('+1,0', '+0,0')
64
to_file.writelines(ud)
66
print >>to_file, "\\ No newline at end of file"
70
def show_diff(b, revision, specific_files):
136
73
if revision == None:
137
74
old_tree = b.basis_tree()
152
89
# TODO: Generation of pseudo-diffs for added/deleted files could
153
90
# be usefully made into a much faster special case.
155
# TODO: Better to return them in sorted order I think.
158
file_list = [b.relpath(f) for f in file_list]
160
# FIXME: If given a file list, compare only those files rather
161
# than comparing everything and then throwing stuff away.
163
for file_state, fid, old_name, new_name, kind in diff_trees(old_tree, new_tree):
165
if file_list and (new_name not in file_list):
168
# Don't show this by default; maybe do it if an option is passed
169
# idlabel = ' {%s}' % fid
172
def diffit(oldlines, newlines, **kw):
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.
181
# In the meantime we at least make sure the patch isn't
185
# Special workaround for Python2.3, where difflib fails if
186
# both sequences are empty.
187
if not oldlines and not newlines:
192
if oldlines and (oldlines[-1][-1] != '\n'):
195
if newlines and (newlines[-1][-1] != '\n'):
199
ud = difflib.unified_diff(oldlines, newlines, **kw)
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
205
ud[2] = ud[2].replace('-1,0', '-0,0')
208
ud[2] = ud[2].replace('+1,0', '+0,0')
210
sys.stdout.writelines(ud)
212
print "\\ No newline at end of file"
213
sys.stdout.write('\n')
215
if file_state in ['.', '?', 'I']:
217
elif file_state == 'A':
218
print '*** added %s %r' % (kind, new_name)
221
new_tree.get_file(fid).readlines(),
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)
228
diffit(old_tree.get_file(fid).readlines(), [],
229
fromfile=old_label + old_name + idlabel,
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)
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)
245
raise BzrError("can't represent state %s {%s}" % (file_state, fid))
92
delta = compare_trees(old_tree, new_tree, want_unchanged=False,
93
specific_files=specific_files)
95
for path, file_id, kind in delta.removed:
96
print '*** removed %s %r' % (kind, path)
98
_diff_one(old_tree.get_file(file_id).readlines(),
101
fromfile=old_label + path,
104
for path, file_id, kind in delta.added:
105
print '*** added %s %r' % (kind, path)
108
new_tree.get_file(file_id).readlines(),
111
tofile=new_label + path)
113
for old_path, new_path, file_id, kind, text_modified in delta.renamed:
114
print '*** renamed %s %r => %r' % (kind, old_path, new_path)
116
_diff_one(old_tree.get_file(file_id).readlines(),
117
new_tree.get_file(file_id).readlines(),
119
fromfile=old_label + old_path,
120
tofile=new_label + new_path)
122
for path, file_id, kind in delta.modified:
123
print '*** modified %s %r' % (kind, path)
125
_diff_one(old_tree.get_file(file_id).readlines(),
126
new_tree.get_file(file_id).readlines(),
128
fromfile=old_label + path,
129
tofile=new_label + path)
279
163
def show(self, to_file, show_ids=False, show_unchanged=False):
280
164
def show_list(files):
281
for path, fid in files:
165
for path, fid, kind in files:
166
if kind == 'directory':
168
elif kind == 'symlink':
283
172
print >>to_file, ' %-30s %s' % (path, fid)
285
174
print >>to_file, ' ', path
288
print >>to_file, 'removed files:'
177
print >>to_file, 'removed:'
289
178
show_list(self.removed)
292
print >>to_file, 'added files:'
181
print >>to_file, 'added:'
293
182
show_list(self.added)
296
print >>to_file, 'renamed files:'
297
for oldpath, newpath, fid, text_modified in self.renamed:
185
print >>to_file, 'renamed:'
186
for oldpath, newpath, fid, kind, text_modified in self.renamed:
299
188
print >>to_file, ' %s => %s %s' % (oldpath, newpath, fid)
301
190
print >>to_file, ' %s => %s' % (oldpath, newpath)
303
192
if self.modified:
304
print >>to_file, 'modified files:'
193
print >>to_file, 'modified:'
305
194
show_list(self.modified)
307
196
if show_unchanged and self.unchanged:
308
print >>to_file, 'unchanged files:'
197
print >>to_file, 'unchanged:'
309
198
show_list(self.unchanged)
313
def compare_trees(old_tree, new_tree, want_unchanged):
202
def compare_trees(old_tree, new_tree, want_unchanged, specific_files=None):
203
"""Describe changes from one tree to another.
205
Returns a TreeDelta with details of added, modified, renamed, and
208
The root entry is specifically exempt.
210
This only considers versioned files.
213
If true, also list files unchanged from one version to
217
If true, only check for changes to specified names or
221
from osutils import is_inside_any
314
223
old_inv = old_tree.inventory
315
224
new_inv = new_tree.inventory
316
225
delta = TreeDelta()
226
mutter('start compare_trees')
228
# TODO: match for specific files can be rather smarter by finding
229
# the IDs of those files up front and then considering only that.
317
231
for file_id in old_tree:
318
232
if file_id in new_tree:
319
old_path = old_inv.id2path(file_id)
320
new_path = new_inv.id2path(file_id)
322
233
kind = old_inv.get_file_kind(file_id)
234
assert kind == new_inv.get_file_kind(file_id)
323
236
assert kind in ('file', 'directory', 'symlink', 'root_directory'), \
324
237
'invalid file kind %r' % kind
239
if kind == 'root_directory':
242
old_path = old_inv.id2path(file_id)
243
new_path = new_inv.id2path(file_id)
246
if (not is_inside_any(specific_files, old_path)
247
and not is_inside_any(specific_files, new_path)):
325
250
if kind == 'file':
326
251
old_sha1 = old_tree.get_file_sha1(file_id)
327
252
new_sha1 = new_tree.get_file_sha1(file_id)
336
261
# May not be worthwhile.
338
263
if old_path != new_path:
339
delta.renamed.append((old_path, new_path, file_id, text_modified))
264
delta.renamed.append((old_path, new_path, file_id, kind,
340
266
elif text_modified:
341
delta.modified.append((new_path, file_id))
267
delta.modified.append((new_path, file_id, kind))
342
268
elif want_unchanged:
343
delta.unchanged.append((new_path, file_id))
269
delta.unchanged.append((new_path, file_id, kind))
345
delta.removed.append((old_inv.id2path(file_id), file_id))
271
old_path = old_inv.id2path(file_id)
273
if not is_inside_any(specific_files, old_path):
275
delta.removed.append((old_path, file_id, kind))
277
mutter('start looking for new files')
346
278
for file_id in new_inv:
347
279
if file_id in old_inv:
349
delta.added.append((new_inv.id2path(file_id), file_id))
281
new_path = new_inv.id2path(file_id)
283
if not is_inside_any(specific_files, new_path):
285
kind = new_inv.get_file_kind(file_id)
286
delta.added.append((new_path, file_id, kind))
351
288
delta.removed.sort()
352
289
delta.added.sort()
353
290
delta.renamed.sort()
354
291
delta.modified.sort()
292
delta.unchanged.sort()