annotate mercurial/mdiff.py @ 2873:4ec58b157265

refactor text diff/patch code. rename commands.dodiff to patch.diff. rename commands.doexport to patch.export. move some functions from commands to new mercurial.cmdutil module. turn list of diff options into mdiff.diffopts class. patch.diff and patch.export now has clean api for call from 3rd party python code.
author Vadim Gelfer <vadim.gelfer@gmail.com>
date Sat, 12 Aug 2006 16:13:27 -0700
parents 345bac2bc4ec
children 8b02af865990
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
fe1689273f84 use demandload more.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2251
diff changeset
10 demandload(globals(), "re struct util")
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
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
26 ignorews ignores all whitespace changes in the diff
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
27 ignorewsamount ignores changes in the amount of whitespace
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
28 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
29
2873
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
30 defaults = {
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
31 'context': 3,
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
32 'text': False,
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
33 'showfunc': True,
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
34 'ignorews': False,
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
35 'ignorewsamount': False,
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
36 'ignoreblanklines': False,
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
37 }
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
38
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
39 __slots__ = defaults.keys()
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
40
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
41 def __init__(self, **opts):
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
42 for k in self.__slots__:
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
43 v = opts.get(k)
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
44 if v is None:
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
45 v = self.defaults[k]
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
46 setattr(self, k, v)
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
47
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
48 defaultopts = diffopts()
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
49
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
50 def unidiff(a, ad, b, bd, fn, r=None, opts=defaultopts):
35
9197c26a414b unidiff: punt on comparing empty files
mpm@selenic.com
parents: 0
diff changeset
51 if not a and not b: return ""
1379
8ee7ce877be2 Clean up mdiff imports
Matt Mackall <mpm@selenic.com>
parents: 1378
diff changeset
52 epoch = util.datestr((0, 0))
264
4c1d7072d5cd Attempt to make diff deal with null sources properly
mpm@selenic.com
parents: 249
diff changeset
53
2873
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
54 if not opts.text and (util.binary(a) or util.binary(b)):
1015
22571b8d35d3 Add automatic binary file detection to diff and export
mpm@selenic.com
parents: 582
diff changeset
55 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
56 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
57 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
58 if a is None:
fde8fb2cbede Fix diff against an empty file (issue124) and add a test for this.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 1637
diff changeset
59 l1 = "--- %s\t%s\n" % ("/dev/null", epoch)
fde8fb2cbede Fix diff against an empty file (issue124) and add a test for this.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 1637
diff changeset
60 else:
fde8fb2cbede Fix diff against an empty file (issue124) and add a test for this.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 1637
diff changeset
61 l1 = "--- %s\t%s\n" % ("a/" + fn, ad)
264
4c1d7072d5cd Attempt to make diff deal with null sources properly
mpm@selenic.com
parents: 249
diff changeset
62 l2 = "+++ %s\t%s\n" % ("b/" + fn, bd)
4c1d7072d5cd Attempt to make diff deal with null sources properly
mpm@selenic.com
parents: 249
diff changeset
63 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
64 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
65 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
66 a = splitnewlines(a)
264
4c1d7072d5cd Attempt to make diff deal with null sources properly
mpm@selenic.com
parents: 249
diff changeset
67 l1 = "--- %s\t%s\n" % ("a/" + fn, 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
68 if b is None:
fde8fb2cbede Fix diff against an empty file (issue124) and add a test for this.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 1637
diff changeset
69 l2 = "+++ %s\t%s\n" % ("/dev/null", epoch)
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 else:
fde8fb2cbede Fix diff against an empty file (issue124) and add a test for this.
Thomas Arendsen Hein <thomas@intevation.de>
parents: 1637
diff changeset
71 l2 = "+++ %s\t%s\n" % ("b/" + fn, bd)
264
4c1d7072d5cd Attempt to make diff deal with null sources properly
mpm@selenic.com
parents: 249
diff changeset
72 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
73 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
74 else:
2251
35fb62a3a673 fix speed regression in mdiff caused by line split bugfix.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2248
diff changeset
75 al = splitnewlines(a)
35fb62a3a673 fix speed regression in mdiff caused by line split bugfix.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2248
diff changeset
76 bl = splitnewlines(b)
2873
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
77 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
78 if not l: return ""
272
467cea2bf2d8 diff: use tab to separate date from filename
mpm@selenic.com
parents: 264
diff changeset
79 # difflib uses a space, rather than a tab
1540
8ca9f5b17257 minor optimization: save some string trash
twaldmann@thinkmo.de
parents: 1452
diff changeset
80 l[0] = "%s\t%s\n" % (l[0][:-2], ad)
8ca9f5b17257 minor optimization: save some string trash
twaldmann@thinkmo.de
parents: 1452
diff changeset
81 l[1] = "%s\t%s\n" % (l[1][:-2], bd)
170
e6c621a825f2 hg diff: fix missing final newline bug
mpm@selenic.com
parents: 127
diff changeset
82
e6c621a825f2 hg diff: fix missing final newline bug
mpm@selenic.com
parents: 127
diff changeset
83 for ln in xrange(len(l)):
e6c621a825f2 hg diff: fix missing final newline bug
mpm@selenic.com
parents: 127
diff changeset
84 if l[ln][-1] != '\n':
e6c621a825f2 hg diff: fix missing final newline bug
mpm@selenic.com
parents: 127
diff changeset
85 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
86
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
87 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
88 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
89 (' '.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
90
0
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
91 return "".join(l)
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
92
1637
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
93 # somewhat self contained replacement for difflib.unified_diff
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
94 # t1 and t2 are the text to be diffed
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
95 # l1 and l2 are the text broken up into lines
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
96 # 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
97 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
98 def contextend(l, len):
2873
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
99 ret = l + opts.context
1637
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
100 if ret > len:
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
101 ret = len
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
102 return ret
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
103
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
104 def contextstart(l):
2873
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
105 ret = l - opts.context
1637
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
106 if ret < 0:
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
107 return 0
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
108 return ret
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
109
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
110 def yieldhunk(hunk, header):
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
111 if header:
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
112 for x in header:
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
113 yield x
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
114 (astart, a2, bstart, b2, delta) = hunk
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
115 aend = contextend(a2, len(l1))
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
116 alen = aend - astart
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
117 blen = b2 - bstart + aend - a2
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
118
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
119 func = ""
2873
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
120 if opts.showfunc:
1637
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
121 # walk backwards from the start of the context
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
122 # to find a line starting with an alphanumeric char.
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
123 for x in xrange(astart, -1, -1):
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
124 t = l1[x].rstrip()
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
125 if funcre.match(t):
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
126 func = ' ' + t[:40]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
127 break
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
128
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
129 yield "@@ -%d,%d +%d,%d @@%s\n" % (astart + 1, alen,
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
130 bstart + 1, blen, func)
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
131 for x in delta:
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
132 yield x
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
133 for x in xrange(a2, aend):
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
134 yield ' ' + l1[x]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
135
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
136 header = [ "--- %s\t\n" % header1, "+++ %s\t\n" % header2 ]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
137
2873
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
138 if opts.showfunc:
1637
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
139 funcre = re.compile('\w')
2873
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
140 if opts.ignorewsamount:
2580
a20a1bb0c396 diff: add -b/-B options
Haakon Riiser <haakon.riiser@fys.uio.no>
parents: 2470
diff changeset
141 wsamountre = re.compile('[ \t]+')
a20a1bb0c396 diff: add -b/-B options
Haakon Riiser <haakon.riiser@fys.uio.no>
parents: 2470
diff changeset
142 wsappendedre = re.compile(' \n')
2873
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
143 if opts.ignoreblanklines:
2580
a20a1bb0c396 diff: add -b/-B options
Haakon Riiser <haakon.riiser@fys.uio.no>
parents: 2470
diff changeset
144 wsblanklinesre = re.compile('\n')
2873
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
145 if opts.ignorews:
1637
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
146 wsre = re.compile('[ \t]')
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 # 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
149 # below finds the spaces between those matching sequences and translates
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
150 # them into diff output.
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
151 #
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
152 diff = bdiff.blocks(t1, t2)
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
153 hunk = None
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
154 for i in xrange(len(diff)):
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
155 # The first match is special.
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
156 # 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
157 # 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
158 # empty and we'll continue to the next match.
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
159 if i > 0:
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
160 s = diff[i-1]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
161 else:
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
162 s = [0, 0, 0, 0]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
163 delta = []
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
164 s1 = diff[i]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
165 a1 = s[1]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
166 a2 = s1[0]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
167 b1 = s[3]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
168 b2 = s1[2]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
169
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
170 old = l1[a1:a2]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
171 new = l2[b1:b2]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
172
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
173 # 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
174 # and deals with the special first match case described above
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
175 if not old and not new:
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
176 continue
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
177
2873
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
178 if opts.ignoreblanklines:
2580
a20a1bb0c396 diff: add -b/-B options
Haakon Riiser <haakon.riiser@fys.uio.no>
parents: 2470
diff changeset
179 wsold = wsblanklinesre.sub('', "".join(old))
a20a1bb0c396 diff: add -b/-B options
Haakon Riiser <haakon.riiser@fys.uio.no>
parents: 2470
diff changeset
180 wsnew = wsblanklinesre.sub('', "".join(new))
a20a1bb0c396 diff: add -b/-B options
Haakon Riiser <haakon.riiser@fys.uio.no>
parents: 2470
diff changeset
181 if wsold == wsnew:
a20a1bb0c396 diff: add -b/-B options
Haakon Riiser <haakon.riiser@fys.uio.no>
parents: 2470
diff changeset
182 continue
a20a1bb0c396 diff: add -b/-B options
Haakon Riiser <haakon.riiser@fys.uio.no>
parents: 2470
diff changeset
183
2873
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
184 if opts.ignorewsamount:
2580
a20a1bb0c396 diff: add -b/-B options
Haakon Riiser <haakon.riiser@fys.uio.no>
parents: 2470
diff changeset
185 wsold = wsamountre.sub(' ', "".join(old))
a20a1bb0c396 diff: add -b/-B options
Haakon Riiser <haakon.riiser@fys.uio.no>
parents: 2470
diff changeset
186 wsold = wsappendedre.sub('\n', wsold)
a20a1bb0c396 diff: add -b/-B options
Haakon Riiser <haakon.riiser@fys.uio.no>
parents: 2470
diff changeset
187 wsnew = wsamountre.sub(' ', "".join(new))
a20a1bb0c396 diff: add -b/-B options
Haakon Riiser <haakon.riiser@fys.uio.no>
parents: 2470
diff changeset
188 wsnew = wsappendedre.sub('\n', wsnew)
a20a1bb0c396 diff: add -b/-B options
Haakon Riiser <haakon.riiser@fys.uio.no>
parents: 2470
diff changeset
189 if wsold == wsnew:
a20a1bb0c396 diff: add -b/-B options
Haakon Riiser <haakon.riiser@fys.uio.no>
parents: 2470
diff changeset
190 continue
a20a1bb0c396 diff: add -b/-B options
Haakon Riiser <haakon.riiser@fys.uio.no>
parents: 2470
diff changeset
191
2873
4ec58b157265 refactor text diff/patch code.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents: 2858
diff changeset
192 if opts.ignorews:
1637
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
193 wsold = wsre.sub('', "".join(old))
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
194 wsnew = wsre.sub('', "".join(new))
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
195 if wsold == wsnew:
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
196 continue
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
197
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
198 astart = contextstart(a1)
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
199 bstart = contextstart(b1)
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
200 prev = None
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
201 if hunk:
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
202 # 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
203 if astart < hunk[1] + opts.context + 1:
1637
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
204 prev = hunk
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
205 astart = hunk[1]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
206 bstart = hunk[3]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
207 else:
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
208 for x in yieldhunk(hunk, header):
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
209 yield x
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
210 # 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
211 # we only want to yield it once.
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
212 header = None
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
213 if prev:
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
214 # 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
215 hunk[1] = a2
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
216 hunk[3] = b2
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
217 delta = hunk[4]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
218 else:
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
219 # create a new hunk
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
220 hunk = [ astart, a2, bstart, b2, delta ]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
221
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
222 delta[len(delta):] = [ ' ' + x for x in l1[astart:a1] ]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
223 delta[len(delta):] = [ '-' + x for x in old ]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
224 delta[len(delta):] = [ '+' + x for x in new ]
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
225
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
226 if hunk:
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
227 for x in yieldhunk(hunk, header):
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
228 yield x
3b1b44b917f4 Add new bdiff based unidiff generation.
mason@suse.com
parents: 1540
diff changeset
229
120
bae6f0328f63 Add a function to return the new text from a binary diff
mpm@selenic.com
parents: 75
diff changeset
230 def patchtext(bin):
bae6f0328f63 Add a function to return the new text from a binary diff
mpm@selenic.com
parents: 75
diff changeset
231 pos = 0
bae6f0328f63 Add a function to return the new text from a binary diff
mpm@selenic.com
parents: 75
diff changeset
232 t = []
bae6f0328f63 Add a function to return the new text from a binary diff
mpm@selenic.com
parents: 75
diff changeset
233 while pos < len(bin):
bae6f0328f63 Add a function to return the new text from a binary diff
mpm@selenic.com
parents: 75
diff changeset
234 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
235 pos += 12
bae6f0328f63 Add a function to return the new text from a binary diff
mpm@selenic.com
parents: 75
diff changeset
236 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
237 pos += l
bae6f0328f63 Add a function to return the new text from a binary diff
mpm@selenic.com
parents: 75
diff changeset
238 return "".join(t)
bae6f0328f63 Add a function to return the new text from a binary diff
mpm@selenic.com
parents: 75
diff changeset
239
0
9117c6561b0b Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff changeset
240 def patch(a, bin):
1379
8ee7ce877be2 Clean up mdiff imports
Matt Mackall <mpm@selenic.com>
parents: 1378
diff changeset
241 return mpatch.patches(a, [bin])
432
3b9e3d3d2810 Start using bdiff for generating deltas
mpm@selenic.com
parents: 396
diff changeset
242
1379
8ee7ce877be2 Clean up mdiff imports
Matt Mackall <mpm@selenic.com>
parents: 1378
diff changeset
243 patches = mpatch.patches
2078
441ea218414e Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents: 1723
diff changeset
244 patchedsize = mpatch.patchedsize
432
3b9e3d3d2810 Start using bdiff for generating deltas
mpm@selenic.com
parents: 396
diff changeset
245 textdiff = bdiff.bdiff