15
14
# along with this program; if not, write to the Free Software
16
15
# Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
20
from trace import mutter
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
Yields a sequence of (state, id, old_name, new_name, kind).
31
Each filename and each id is listed only once.
17
from bzrlib.trace import mutter
18
from bzrlib.errors import BzrError
19
from bzrlib.delta import compare_trees
21
# TODO: Rather than building a changeset object, we should probably
22
# invoke callbacks on an object. That object can either accumulate a
23
# list, write them out directly, etc etc.
25
def internal_diff(old_label, oldlines, new_label, newlines, to_file):
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:
44
ud = difflib.unified_diff(oldlines, newlines,
45
fromfile=old_label, tofile=new_label)
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
51
ud[2] = ud[2].replace('-1,0', '-0,0')
53
ud[2] = ud[2].replace('+1,0', '+0,0')
54
# work around for difflib emitting random spaces after the label
55
ud[0] = ud[0][:-2] + '\n'
56
ud[1] = ud[1][:-2] + '\n'
60
if not line.endswith('\n'):
61
to_file.write("\n\\ No newline at end of file\n")
65
def external_diff(old_label, oldlines, new_label, newlines, to_file,
67
"""Display a diff by calling out to the external diff program."""
70
if to_file != sys.stdout:
71
raise NotImplementedError("sorry, can't send external diff other than to stdout yet",
74
# make sure our own output is properly ordered before the diff
77
from tempfile import NamedTemporaryFile
80
oldtmpf = NamedTemporaryFile()
81
newtmpf = NamedTemporaryFile()
84
# TODO: perhaps a special case for comparing to or from the empty
85
# sequence; can just use /dev/null on Unix
87
# TODO: if either of the files being compared already exists as a
88
# regular named file (e.g. in the working directory) then we can
89
# compare directly to that, rather than copying it.
91
oldtmpf.writelines(oldlines)
92
newtmpf.writelines(newlines)
100
'--label', old_label,
102
'--label', new_label,
105
# diff only allows one style to be specified; they don't override.
106
# note that some of these take optargs, and the optargs can be
107
# directly appended to the options.
108
# this is only an approximate parser; it doesn't properly understand
110
for s in ['-c', '-u', '-C', '-U',
115
'-y', '--side-by-side',
127
diffcmd.extend(diff_opts)
129
rc = os.spawnvp(os.P_WAIT, 'diff', diffcmd)
131
if rc != 0 and rc != 1:
132
# returns 1 if files differ; that's OK
134
msg = 'signal %d' % (-rc)
136
msg = 'exit code %d' % rc
138
raise BzrError('external diff failed with %s; command: %r' % (rc, diffcmd))
140
oldtmpf.close() # and delete
143
def show_diff(b, from_spec, specific_files, external_diff_options=None,
144
revision2=None, output=None):
145
"""Shortcut for showing the diff to the working tree.
151
None for 'basis tree', or otherwise the old revision to compare against.
153
The more general form is show_diff_trees(), where the caller
154
supplies any two trees.
34
## TODO: Compare files before diffing; only mention those that have changed
36
## TODO: Set nice names in the headers, maybe include diffstat
38
## TODO: Perhaps make this a generator rather than using
41
## TODO: Allow specifying a list of files to compare, rather than
42
## doing the whole tree? (Not urgent.)
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.
49
## XXX: This doesn't report on unknown files; that can be done
50
## from a separate method.
52
old_it = old_tree.list_files()
53
new_it = new_tree.list_files()
61
old_item = next(old_it)
62
new_item = next(new_it)
64
# We step through the two sorted iterators in parallel, trying to
67
while (old_item != None) or (new_item != None):
68
# OK, we still have some remaining on both, but they may be
71
old_name, old_class, old_kind, old_id = old_item
76
new_name, new_class, new_kind, new_id = new_item
80
mutter(" diff pairwise %r" % (old_item,))
81
mutter(" %r" % (new_item,))
84
# can't handle the old tree being a WorkingTree
85
assert old_class == 'V'
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
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
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
115
yield 'A', new_id, None, new_name, new_kind
117
new_item = next(new_it)
118
old_item = next(old_it)
120
assert old_id == new_id
121
assert old_id != None
122
assert old_name == new_name
123
assert old_kind == new_kind
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
135
mutter(" quick compare shows different")
136
yield 'M', new_id, old_name, new_name, new_kind
138
new_item = next(new_it)
139
old_item = next(old_it)
143
def show_diff(b, revision, file_list):
144
import difflib, sys, types
160
if from_spec is None:
147
161
old_tree = b.basis_tree()
149
old_tree = b.revision_tree(b.lookup_revision(revision))
151
new_tree = b.working_tree()
163
old_tree = b.revision_tree(from_spec.in_history(b).rev_id)
165
if revision2 is None:
166
new_tree = b.working_tree()
168
new_tree = b.revision_tree(revision2.in_history(b).rev_id)
170
show_diff_trees(old_tree, new_tree, output, specific_files,
171
external_diff_options)
175
def show_diff_trees(old_tree, new_tree, to_file, specific_files=None,
176
external_diff_options=None):
177
"""Show in text form the changes from one tree to another.
180
If set, include only changes to these files.
182
external_diff_options
183
If set, use an external GNU diff and pass these options.
153
186
# TODO: Options to control putting on a prefix or suffix, perhaps as a format string
162
195
# TODO: Generation of pseudo-diffs for added/deleted files could
163
196
# be usefully made into a much faster special case.
165
# TODO: Better to return them in sorted order I think.
168
file_list = [b.relpath(f) for f in file_list]
170
# FIXME: If given a file list, compare only those files rather
171
# than comparing everything and then throwing stuff away.
173
for file_state, fid, old_name, new_name, kind in diff_trees(old_tree, new_tree):
175
if file_list and (new_name not in file_list):
178
# Don't show this by default; maybe do it if an option is passed
179
# idlabel = ' {%s}' % fid
182
# FIXME: Something about the diff format makes patch unhappy
183
# with newly-added files.
185
def diffit(oldlines, newlines, **kw):
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.
194
# In the meantime we at least make sure the patch isn't
198
# Special workaround for Python2.3, where difflib fails if
199
# both sequences are empty.
200
if not oldlines and not newlines:
205
if oldlines and (oldlines[-1][-1] != '\n'):
208
if newlines and (newlines[-1][-1] != '\n'):
212
ud = difflib.unified_diff(oldlines, newlines, **kw)
213
sys.stdout.writelines(ud)
215
print "\\ No newline at end of file"
216
sys.stdout.write('\n')
218
if file_state in ['.', '?', 'I']:
220
elif file_state == 'A':
221
print '*** added %s %r' % (kind, new_name)
224
new_tree.get_file(fid).readlines(),
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)
231
diffit(old_tree.get_file(fid).readlines(), [],
232
fromfile=old_label + old_name + idlabel,
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)
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)
248
raise BzrError("can't represent state %s {%s}" % (file_state, fid))
253
"""Describes changes from one tree to another.
262
(oldpath, newpath, id)
266
A path may occur in more than one list if it was e.g. deleted
267
under an old id and renamed into place in a new id.
269
Files are listed in either modified or renamed, not both. In
270
other words, renamed files may also be modified.
279
def compare_inventories(old_inv, new_inv):
280
"""Return a TreeDelta object describing changes between inventories.
282
This only describes changes in the shape of the tree, not the
285
This is an alternative to diff_trees() and should probably
286
eventually replace it.
288
old_ids = old_inv.id_set()
289
new_ids = new_inv.id_set()
292
delta.removed = [(old_inv.id2path(fid), fid) for fid in (old_ids - new_ids)]
295
delta.added = [(new_inv.id2path(fid), fid) for fid in (new_ids - old_ids)]
298
for fid in old_ids & new_ids:
299
old_ie = old_inv[fid]
300
new_ie = new_inv[fid]
301
old_path = old_inv.id2path(fid)
302
new_path = new_inv.id2path(fid)
304
if old_path != new_path:
305
delta.renamed.append((old_path, new_path, fid))
306
elif old_ie.text_sha1 != new_ie.text_sha1:
307
delta.modified.append((new_path, fid))
309
delta.modified.sort()
198
if external_diff_options:
199
assert isinstance(external_diff_options, basestring)
200
opts = external_diff_options.split()
201
def diff_file(olab, olines, nlab, nlines, to_file):
202
external_diff(olab, olines, nlab, nlines, to_file, opts)
204
diff_file = internal_diff
207
delta = compare_trees(old_tree, new_tree, want_unchanged=False,
208
specific_files=specific_files)
210
for path, file_id, kind in delta.removed:
211
print >>to_file, '=== removed %s %r' % (kind, path)
212
old_tree.inventory[file_id].diff(diff_file, old_label + path, old_tree,
213
DEVNULL, None, None, to_file)
214
for path, file_id, kind in delta.added:
215
print >>to_file, '=== added %s %r' % (kind, path)
216
new_tree.inventory[file_id].diff(diff_file, new_label + path, new_tree,
217
DEVNULL, None, None, to_file,
219
for (old_path, new_path, file_id, kind,
220
text_modified, meta_modified) in delta.renamed:
221
prop_str = get_prop_change(meta_modified)
222
print >>to_file, '=== renamed %s %r => %r%s' % (
223
kind, old_path, new_path, prop_str)
224
_maybe_diff_file_or_symlink(old_label, old_path, old_tree, file_id,
225
new_label, new_path, new_tree,
226
text_modified, kind, to_file, diff_file)
227
for path, file_id, kind, text_modified, meta_modified in delta.modified:
228
prop_str = get_prop_change(meta_modified)
229
print >>to_file, '=== modified %s %r%s' % (kind, path, prop_str)
231
_maybe_diff_file_or_symlink(old_label, path, old_tree, file_id,
232
new_label, path, new_tree,
233
True, kind, to_file, diff_file)
236
def get_prop_change(meta_modified):
238
return " (properties changed)"
243
def _maybe_diff_file_or_symlink(old_label, old_path, old_tree, file_id,
244
new_label, new_path, new_tree, text_modified,
245
kind, to_file, diff_file):
247
new_entry = new_tree.inventory[file_id]
248
old_tree.inventory[file_id].diff(diff_file,
249
old_label + old_path, old_tree,
250
new_label + new_path, new_entry,
added
removed
bzrlib/diff.py
