annotate mercurial/mdiff.py @ 4102:544838cc1158

Don't lie that "binary file has changed" Without -a option to "hg diff", mdiff.unidiff reported that "Binary file foo has changed" without even trying to compare things. Now it computes MD5 of old and new files, compares them and makes the conclusion.
author tailgunner@smtp.ru
date Sat, 17 Feb 2007 09:54:56 -0200
parents 096f1c73cdc3
children 226df1808f16
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
239
75840796e8e2 mdiff.py: kill #! line, add copyright notice
mpm@selenic.com
parents: 184
diff changeset
1 # mdiff.py - diff and patch routines for mercurial
75840796e8e2 mdiff.py: kill #! line, add copyright notice
mpm@selenic.com
parents: 184
diff changeset
2 #
2858
345bac2bc4ec update copyrights.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2580
diff changeset
3 # Copyright 2005, 2006 Matt Mackall <mpm@selenic.com>
239
75840796e8e2 mdiff.py: kill #! line, add copyright notice
mpm@selenic.com
parents: 184
diff changeset
4 #
75840796e8e2 mdiff.py: kill #! line, add copyright notice
mpm@selenic.com
parents: 184
diff changeset
5 # This software may be used and distributed according to the terms
75840796e8e2 mdiff.py: kill #! line, add copyright notice
mpm@selenic.com
parents: 184
diff changeset
6 # of the GNU General Public License, incorporated herein by reference.
75840796e8e2 mdiff.py: kill #! line, add copyright notice
mpm@selenic.com
parents: 184
diff changeset
7
1637
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
8 from demandload import demandload
2470
fe1689273f84 use demandload more.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2251
diff changeset
9 import bdiff, mpatch
4102
544838cc1158 Don't lie that "binary file has changed"
tailgunner@smtp.ru
parents: 3192
diff changeset
10 demandload(globals(), "re struct util md5")
0
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
11
2251
35fb62a3a673 fix speed regression in mdiff caused by line split bugfix.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2248
diff changeset
12 def splitnewlines(text):
2248
b914f0557832 fix diffs containing embedded "\r".
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2078
diff changeset
13 '''like str.splitlines, but only split on newlines.'''
2251
35fb62a3a673 fix speed regression in mdiff caused by line split bugfix.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2248
diff changeset
14 lines = [l + '\n' for l in text.split('\n')]
35fb62a3a673 fix speed regression in mdiff caused by line split bugfix.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2248
diff changeset
15 if lines:
35fb62a3a673 fix speed regression in mdiff caused by line split bugfix.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2248
diff changeset
16 if lines[-1] == '\n':
35fb62a3a673 fix speed regression in mdiff caused by line split bugfix.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2248
diff changeset
17 lines.pop()
35fb62a3a673 fix speed regression in mdiff caused by line split bugfix.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2248
diff changeset
18 else:
35fb62a3a673 fix speed regression in mdiff caused by line split bugfix.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2248
diff changeset
19 lines[-1] = lines[-1][:-1]
35fb62a3a673 fix speed regression in mdiff caused by line split bugfix.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2248
diff changeset
20 return lines
2248
b914f0557832 fix diffs containing embedded "\r".
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2078
diff changeset
21
2873
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
22 class diffopts(object):
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
23 '''context is the number of context lines
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
24 text treats all files as text
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
25 showfunc enables diff -p output
2892
8b02af865990 Add diff --git option
Brendan Cully <brendan@kublai.com>
parents: 2873
diff changeset
26 git enables the git extended patch format
3192
096f1c73cdc3 Add -D/--nodates options to hg diff/export that removes dates from diff headers
Stephen Darnell <stephen@darnell.plus.com>
parents: 3023
diff changeset
27 nodates removes dates from diff headers
2873
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
28 ignorews ignores all whitespace changes in the diff
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
29 ignorewsamount ignores changes in the amount of whitespace
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
30 ignoreblanklines ignores changes whose lines are all blank'''
396
8f8bb77d560e Show revisions in diffs like CVS, based on a patch from Goffredo Baroncelli.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 361
diff changeset
31
2873
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
32 defaults = {
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
33 'context': 3,
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
34 'text': False,
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
35 'showfunc': True,
2892
8b02af865990 Add diff --git option
Brendan Cully <brendan@kublai.com>
parents: 2873
diff changeset
36 'git': False,
3192
096f1c73cdc3 Add -D/--nodates options to hg diff/export that removes dates from diff headers
Stephen Darnell <stephen@darnell.plus.com>
parents: 3023
diff changeset
37 'nodates': False,
2873
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
38 'ignorews': False,
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
39 'ignorewsamount': False,
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
40 'ignoreblanklines': False,
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
41 }
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
42
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
43 __slots__ = defaults.keys()
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
44
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
45 def __init__(self, **opts):
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
46 for k in self.__slots__:
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
47 v = opts.get(k)
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
48 if v is None:
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
49 v = self.defaults[k]
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
50 setattr(self, k, v)
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
51
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
52 defaultopts = diffopts()
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
53
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
54 def unidiff(a, ad, b, bd, fn, r=None, opts=defaultopts):
3023
d838bfac668d Remove dates from git export file lines - they confuse git-apply
Brendan Cully <brendan@kublai.com>
parents: 2892
diff changeset
55 def datetag(date):
3192
096f1c73cdc3 Add -D/--nodates options to hg diff/export that removes dates from diff headers
Stephen Darnell <stephen@darnell.plus.com>
parents: 3023
diff changeset
56 return (opts.git or opts.nodates) and '\n' or '\t%s\n' % date
3023
d838bfac668d Remove dates from git export file lines - they confuse git-apply
Brendan Cully <brendan@kublai.com>
parents: 2892
diff changeset
57
35
9197c26a414b unidiff: punt on comparing empty files
mpm@selenic.com
parents: 0
diff changeset
58 if not a and not b: return ""
1379
8ee7ce877be2 Clean up mdiff imports
Matt Mackall <mpm@selenic.com>
parents: 1378
diff changeset
59 epoch = util.datestr((0, 0))
264
4c1d7072d5cd Attempt to make diff deal with null sources properly
mpm@selenic.com
parents: 249
diff changeset
60
2873
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
61 if not opts.text and (util.binary(a) or util.binary(b)):
4102
544838cc1158 Don't lie that "binary file has changed"
tailgunner@smtp.ru
parents: 3192
diff changeset
62 def h(v):
544838cc1158 Don't lie that "binary file has changed"
tailgunner@smtp.ru
parents: 3192
diff changeset
63 # md5 is used instead of sha1 because md5 is supposedly faster
544838cc1158 Don't lie that "binary file has changed"
tailgunner@smtp.ru
parents: 3192
diff changeset
64 return md5.new(v).digest()
544838cc1158 Don't lie that "binary file has changed"
tailgunner@smtp.ru
parents: 3192
diff changeset
65 if a and b and len(a) == len(b) and h(a) == h(b):
544838cc1158 Don't lie that "binary file has changed"
tailgunner@smtp.ru
parents: 3192
diff changeset
66 return ""
1015
22571b8d35d3 Add automatic binary file detection to diff and export
mpm@selenic.com
parents: 582
diff changeset
67 l = ['Binary file %s has changed\n' % fn]
1723
fde8fb2cbede Fix diff against an empty file (issue124) and add a test for this.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 1637
diff changeset
68 elif not a:
2251
35fb62a3a673 fix speed regression in mdiff caused by line split bugfix.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2248
diff changeset
69 b = splitnewlines(b)
1723
fde8fb2cbede Fix diff against an empty file (issue124) and add a test for this.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 1637
diff changeset
70 if a is None:
3023
d838bfac668d Remove dates from git export file lines - they confuse git-apply
Brendan Cully <brendan@kublai.com>
parents: 2892
diff changeset
71 l1 = '--- /dev/null%s' % datetag(epoch)
1723
fde8fb2cbede Fix diff against an empty file (issue124) and add a test for this.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 1637
diff changeset
72 else:
3023
d838bfac668d Remove dates from git export file lines - they confuse git-apply
Brendan Cully <brendan@kublai.com>
parents: 2892
diff changeset
73 l1 = "--- %s%s" % ("a/" + fn, datetag(ad))
d838bfac668d Remove dates from git export file lines - they confuse git-apply
Brendan Cully <brendan@kublai.com>
parents: 2892
diff changeset
74 l2 = "+++ %s%s" % ("b/" + fn, datetag(bd))
264
4c1d7072d5cd Attempt to make diff deal with null sources properly
mpm@selenic.com
parents: 249
diff changeset
75 l3 = "@@ -0,0 +1,%d @@\n" % len(b)
4c1d7072d5cd Attempt to make diff deal with null sources properly
mpm@selenic.com
parents: 249
diff changeset
76 l = [l1, l2, l3] + ["+" + e for e in b]
1723
fde8fb2cbede Fix diff against an empty file (issue124) and add a test for this.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 1637
diff changeset
77 elif not b:
2251
35fb62a3a673 fix speed regression in mdiff caused by line split bugfix.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2248
diff changeset
78 a = splitnewlines(a)
3023
d838bfac668d Remove dates from git export file lines - they confuse git-apply
Brendan Cully <brendan@kublai.com>
parents: 2892
diff changeset
79 l1 = "--- %s%s" % ("a/" + fn, datetag(ad))
1723
fde8fb2cbede Fix diff against an empty file (issue124) and add a test for this.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 1637
diff changeset
80 if b is None:
3023
d838bfac668d Remove dates from git export file lines - they confuse git-apply
Brendan Cully <brendan@kublai.com>
parents: 2892
diff changeset
81 l2 = '+++ /dev/null%s' % datetag(epoch)
1723
fde8fb2cbede Fix diff against an empty file (issue124) and add a test for this.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 1637
diff changeset
82 else:
3023
d838bfac668d Remove dates from git export file lines - they confuse git-apply
Brendan Cully <brendan@kublai.com>
parents: 2892
diff changeset
83 l2 = "+++ %s%s" % ("b/" + fn, datetag(bd))
264
4c1d7072d5cd Attempt to make diff deal with null sources properly
mpm@selenic.com
parents: 249
diff changeset
84 l3 = "@@ -1,%d +0,0 @@\n" % len(a)
4c1d7072d5cd Attempt to make diff deal with null sources properly
mpm@selenic.com
parents: 249
diff changeset
85 l = [l1, l2, l3] + ["-" + e for e in a]
4c1d7072d5cd Attempt to make diff deal with null sources properly
mpm@selenic.com
parents: 249
diff changeset
86 else:
2251
35fb62a3a673 fix speed regression in mdiff caused by line split bugfix.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2248
diff changeset
87 al = splitnewlines(a)
35fb62a3a673 fix speed regression in mdiff caused by line split bugfix.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2248
diff changeset
88 bl = splitnewlines(b)
2873
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
89 l = list(bunidiff(a, b, al, bl, "a/" + fn, "b/" + fn, opts=opts))
278
777e388c06d6 unidiff: handle empty diffs more gracefully
mpm@selenic.com
parents: 272
diff changeset
90 if not l: return ""
272
467cea2bf2d8 diff: use tab to separate date from filename
mpm@selenic.com
parents: 264
diff changeset
91 # difflib uses a space, rather than a tab
3023
d838bfac668d Remove dates from git export file lines - they confuse git-apply
Brendan Cully <brendan@kublai.com>
parents: 2892
diff changeset
92 l[0] = "%s%s" % (l[0][:-2], datetag(ad))
d838bfac668d Remove dates from git export file lines - they confuse git-apply
Brendan Cully <brendan@kublai.com>
parents: 2892
diff changeset
93 l[1] = "%s%s" % (l[1][:-2], datetag(bd))
170
e6c621a825f2 hg diff: fix missing final newline bug
mpm@selenic.com
parents: 127
diff changeset
94
e6c621a825f2 hg diff: fix missing final newline bug
mpm@selenic.com
parents: 127
diff changeset
95 for ln in xrange(len(l)):
e6c621a825f2 hg diff: fix missing final newline bug
mpm@selenic.com
parents: 127
diff changeset
96 if l[ln][-1] != '\n':
e6c621a825f2 hg diff: fix missing final newline bug
mpm@selenic.com
parents: 127
diff changeset
97 l[ln] += "\n\ No newline at end of file\n"
e6c621a825f2 hg diff: fix missing final newline bug
mpm@selenic.com
parents: 127
diff changeset
98
396
8f8bb77d560e Show revisions in diffs like CVS, based on a patch from Goffredo Baroncelli.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 361
diff changeset
99 if r:
8f8bb77d560e Show revisions in diffs like CVS, based on a patch from Goffredo Baroncelli.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 361
diff changeset
100 l.insert(0, "diff %s %s\n" %
8f8bb77d560e Show revisions in diffs like CVS, based on a patch from Goffredo Baroncelli.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 361
diff changeset
101 (' '.join(["-r %s" % rev for rev in r]), fn))
8f8bb77d560e Show revisions in diffs like CVS, based on a patch from Goffredo Baroncelli.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 361
diff changeset
102
0
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
103 return "".join(l)
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
104
1637
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
105 # somewhat self contained replacement for difflib.unified_diff
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
106 # t1 and t2 are the text to be diffed
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
107 # l1 and l2 are the text broken up into lines
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
108 # header1 and header2 are the filenames for the diff output
2873
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
109 def bunidiff(t1, t2, l1, l2, header1, header2, opts=defaultopts):
1637
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
110 def contextend(l, len):
2873
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
111 ret = l + opts.context
1637
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
112 if ret > len:
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
113 ret = len
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
114 return ret
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
115
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
116 def contextstart(l):
2873
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
117 ret = l - opts.context
1637
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
118 if ret < 0:
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
119 return 0
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
120 return ret
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
121
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
122 def yieldhunk(hunk, header):
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
123 if header:
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
124 for x in header:
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
125 yield x
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
126 (astart, a2, bstart, b2, delta) = hunk
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
127 aend = contextend(a2, len(l1))
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
128 alen = aend - astart
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
129 blen = b2 - bstart + aend - a2
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
130
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
131 func = ""
2873
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
132 if opts.showfunc:
1637
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
133 # walk backwards from the start of the context
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
134 # to find a line starting with an alphanumeric char.
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
135 for x in xrange(astart, -1, -1):
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
136 t = l1[x].rstrip()
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
137 if funcre.match(t):
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
138 func = ' ' + t[:40]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
139 break
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
140
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
141 yield "@@ -%d,%d +%d,%d @@%s\n" % (astart + 1, alen,
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
142 bstart + 1, blen, func)
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
143 for x in delta:
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
144 yield x
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
145 for x in xrange(a2, aend):
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
146 yield ' ' + l1[x]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
147
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
148 header = [ "--- %s\t\n" % header1, "+++ %s\t\n" % header2 ]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
149
2873
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
150 if opts.showfunc:
1637
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
151 funcre = re.compile('\w')
2873
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
152 if opts.ignorewsamount:
2580
a20a1bb0c396 diff: add -b/-B options
Haakon Riiser <haakon.riiser@fys.uio.no>
parents: 2470
diff changeset
153 wsamountre = re.compile('[ \t]+')
a20a1bb0c396 diff: add -b/-B options
Haakon Riiser <haakon.riiser@fys.uio.no>
parents: 2470
diff changeset
154 wsappendedre = re.compile(' \n')
2873
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
155 if opts.ignoreblanklines:
2580
a20a1bb0c396 diff: add -b/-B options
Haakon Riiser <haakon.riiser@fys.uio.no>
parents: 2470
diff changeset
156 wsblanklinesre = re.compile('\n')
2873
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
157 if opts.ignorews:
1637
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
158 wsre = re.compile('[ \t]')
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
159
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
160 # bdiff.blocks gives us the matching sequences in the files. The loop
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
161 # below finds the spaces between those matching sequences and translates
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
162 # them into diff output.
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
163 #
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
164 diff = bdiff.blocks(t1, t2)
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
165 hunk = None
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
166 for i in xrange(len(diff)):
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
167 # The first match is special.
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
168 # we've either found a match starting at line 0 or a match later
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
169 # in the file. If it starts later, old and new below will both be
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
170 # empty and we'll continue to the next match.
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
171 if i > 0:
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
172 s = diff[i-1]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
173 else:
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
174 s = [0, 0, 0, 0]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
175 delta = []
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
176 s1 = diff[i]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
177 a1 = s[1]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
178 a2 = s1[0]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
179 b1 = s[3]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
180 b2 = s1[2]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
181
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
182 old = l1[a1:a2]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
183 new = l2[b1:b2]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
184
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
185 # bdiff sometimes gives huge matches past eof, this check eats them,
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
186 # and deals with the special first match case described above
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
187 if not old and not new:
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
188 continue
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
189
2873
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
190 if opts.ignoreblanklines:
2580
a20a1bb0c396 diff: add -b/-B options
Haakon Riiser <haakon.riiser@fys.uio.no>
parents: 2470
diff changeset
191 wsold = wsblanklinesre.sub('', "".join(old))
a20a1bb0c396 diff: add -b/-B options
Haakon Riiser <haakon.riiser@fys.uio.no>
parents: 2470
diff changeset
192 wsnew = wsblanklinesre.sub('', "".join(new))
a20a1bb0c396 diff: add -b/-B options
Haakon Riiser <haakon.riiser@fys.uio.no>
parents: 2470
diff changeset
193 if wsold == wsnew:
a20a1bb0c396 diff: add -b/-B options
Haakon Riiser <haakon.riiser@fys.uio.no>
parents: 2470
diff changeset
194 continue
a20a1bb0c396 diff: add -b/-B options
Haakon Riiser <haakon.riiser@fys.uio.no>
parents: 2470
diff changeset
195
2873
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
196 if opts.ignorewsamount:
2580
a20a1bb0c396 diff: add -b/-B options
Haakon Riiser <haakon.riiser@fys.uio.no>
parents: 2470
diff changeset
197 wsold = wsamountre.sub(' ', "".join(old))
a20a1bb0c396 diff: add -b/-B options
Haakon Riiser <haakon.riiser@fys.uio.no>
parents: 2470
diff changeset
198 wsold = wsappendedre.sub('\n', wsold)
a20a1bb0c396 diff: add -b/-B options
Haakon Riiser <haakon.riiser@fys.uio.no>
parents: 2470
diff changeset
199 wsnew = wsamountre.sub(' ', "".join(new))
a20a1bb0c396 diff: add -b/-B options
Haakon Riiser <haakon.riiser@fys.uio.no>
parents: 2470
diff changeset
200 wsnew = wsappendedre.sub('\n', wsnew)
a20a1bb0c396 diff: add -b/-B options
Haakon Riiser <haakon.riiser@fys.uio.no>
parents: 2470
diff changeset
201 if wsold == wsnew:
a20a1bb0c396 diff: add -b/-B options
Haakon Riiser <haakon.riiser@fys.uio.no>
parents: 2470
diff changeset
202 continue
a20a1bb0c396 diff: add -b/-B options
Haakon Riiser <haakon.riiser@fys.uio.no>
parents: 2470
diff changeset
203
2873
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
204 if opts.ignorews:
1637
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
205 wsold = wsre.sub('', "".join(old))
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
206 wsnew = wsre.sub('', "".join(new))
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
207 if wsold == wsnew:
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
208 continue
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
209
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
210 astart = contextstart(a1)
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
211 bstart = contextstart(b1)
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
212 prev = None
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
213 if hunk:
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
214 # join with the previous hunk if it falls inside the context
2873
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
215 if astart < hunk[1] + opts.context + 1:
1637
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
216 prev = hunk
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
217 astart = hunk[1]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
218 bstart = hunk[3]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
219 else:
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
220 for x in yieldhunk(hunk, header):
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
221 yield x
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
222 # we only want to yield the header if the files differ, and
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
223 # we only want to yield it once.
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
224 header = None
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
225 if prev:
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
226 # we've joined the previous hunk, record the new ending points.
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
227 hunk[1] = a2
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
228 hunk[3] = b2
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
229 delta = hunk[4]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
230 else:
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
231 # create a new hunk
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
232 hunk = [ astart, a2, bstart, b2, delta ]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
233
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
234 delta[len(delta):] = [ ' ' + x for x in l1[astart:a1] ]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
235 delta[len(delta):] = [ '-' + x for x in old ]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
236 delta[len(delta):] = [ '+' + x for x in new ]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
237
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
238 if hunk:
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
239 for x in yieldhunk(hunk, header):
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
240 yield x
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
241
120
bae6f0328f63 Add a function to return the new text from a binary diff
mpm@selenic.com
parents: 75
diff changeset
242 def patchtext(bin):
bae6f0328f63 Add a function to return the new text from a binary diff
mpm@selenic.com
parents: 75
diff changeset
243 pos = 0
bae6f0328f63 Add a function to return the new text from a binary diff
mpm@selenic.com
parents: 75
diff changeset
244 t = []
bae6f0328f63 Add a function to return the new text from a binary diff
mpm@selenic.com
parents: 75
diff changeset
245 while pos < len(bin):
bae6f0328f63 Add a function to return the new text from a binary diff
mpm@selenic.com
parents: 75
diff changeset
246 p1, p2, l = struct.unpack(">lll", bin[pos:pos + 12])
bae6f0328f63 Add a function to return the new text from a binary diff
mpm@selenic.com
parents: 75
diff changeset
247 pos += 12
bae6f0328f63 Add a function to return the new text from a binary diff
mpm@selenic.com
parents: 75
diff changeset
248 t.append(bin[pos:pos + l])
bae6f0328f63 Add a function to return the new text from a binary diff
mpm@selenic.com
parents: 75
diff changeset
249 pos += l
bae6f0328f63 Add a function to return the new text from a binary diff
mpm@selenic.com
parents: 75
diff changeset
250 return "".join(t)
bae6f0328f63 Add a function to return the new text from a binary diff
mpm@selenic.com
parents: 75
diff changeset
251
0
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
252 def patch(a, bin):
1379
8ee7ce877be2 Clean up mdiff imports
Matt Mackall <mpm@selenic.com>
parents: 1378
diff changeset
253 return mpatch.patches(a, [bin])
432
3b9e3d3d2810 Start using bdiff for generating deltas
mpm@selenic.com
parents: 396
diff changeset
254
1379
8ee7ce877be2 Clean up mdiff imports
Matt Mackall <mpm@selenic.com>
parents: 1378
diff changeset
255 patches = mpatch.patches
2078
441ea218414e Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents: 1723
diff changeset
256 patchedsize = mpatch.patchedsize
432
3b9e3d3d2810 Start using bdiff for generating deltas
mpm@selenic.com
parents: 396
diff changeset
257 textdiff = bdiff.bdiff