Mercurial > public > mercurial-scm > hg-stable
annotate mercurial/mdiff.py @ 2470:fe1689273f84
use demandload more.
author | Vadim Gelfer <vadim.gelfer@gmail.com> |
---|---|
date | Tue, 20 Jun 2006 23:58:21 -0700 |
parents | 35fb62a3a673 |
children | a20a1bb0c396 |
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 # |
75840796e8e2
mdiff.py: kill #! line, add copyright notice
mpm@selenic.com
parents:
184
diff
changeset
|
3 # Copyright 2005 Matt Mackall <mpm@selenic.com> |
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 | 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 |
1637 | 22 def unidiff(a, ad, b, bd, fn, r=None, text=False, |
23 showfunc=False, ignorews=False): | |
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
|
24 |
35 | 25 if not a and not b: return "" |
1379 | 26 epoch = util.datestr((0, 0)) |
264
4c1d7072d5cd
Attempt to make diff deal with null sources properly
mpm@selenic.com
parents:
249
diff
changeset
|
27 |
1379 | 28 if not 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
|
29 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
|
30 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
|
31 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
|
32 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
|
33 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
|
34 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
|
35 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
|
36 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
|
37 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
|
38 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
|
39 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
|
40 a = splitnewlines(a) |
264
4c1d7072d5cd
Attempt to make diff deal with null sources properly
mpm@selenic.com
parents:
249
diff
changeset
|
41 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
|
42 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
|
43 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
|
44 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
|
45 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
|
46 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
|
47 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
|
48 else: |
2251
35fb62a3a673
fix speed regression in mdiff caused by line split bugfix.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents:
2248
diff
changeset
|
49 al = splitnewlines(a) |
35fb62a3a673
fix speed regression in mdiff caused by line split bugfix.
Vadim Gelfer <vadim.gelfer@gmail.com>
parents:
2248
diff
changeset
|
50 bl = splitnewlines(b) |
1637 | 51 l = list(bunidiff(a, b, al, bl, "a/" + fn, "b/" + fn, |
52 showfunc=showfunc, ignorews=ignorews)) | |
278
777e388c06d6
unidiff: handle empty diffs more gracefully
mpm@selenic.com
parents:
272
diff
changeset
|
53 if not l: return "" |
272
467cea2bf2d8
diff: use tab to separate date from filename
mpm@selenic.com
parents:
264
diff
changeset
|
54 # difflib uses a space, rather than a tab |
1540
8ca9f5b17257
minor optimization: save some string trash
twaldmann@thinkmo.de
parents:
1452
diff
changeset
|
55 l[0] = "%s\t%s\n" % (l[0][:-2], ad) |
8ca9f5b17257
minor optimization: save some string trash
twaldmann@thinkmo.de
parents:
1452
diff
changeset
|
56 l[1] = "%s\t%s\n" % (l[1][:-2], bd) |
170 | 57 |
58 for ln in xrange(len(l)): | |
59 if l[ln][-1] != '\n': | |
60 l[ln] += "\n\ No newline at end of file\n" | |
61 | |
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
|
62 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
|
63 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
|
64 (' '.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
|
65 |
0
9117c6561b0b
Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff
changeset
|
66 return "".join(l) |
9117c6561b0b
Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff
changeset
|
67 |
1637 | 68 # somewhat self contained replacement for difflib.unified_diff |
69 # t1 and t2 are the text to be diffed | |
70 # l1 and l2 are the text broken up into lines | |
71 # header1 and header2 are the filenames for the diff output | |
72 # context is the number of context lines | |
73 # showfunc enables diff -p output | |
74 # ignorews ignores all whitespace changes in the diff | |
75 def bunidiff(t1, t2, l1, l2, header1, header2, context=3, showfunc=False, | |
76 ignorews=False): | |
77 def contextend(l, len): | |
78 ret = l + context | |
79 if ret > len: | |
80 ret = len | |
81 return ret | |
82 | |
83 def contextstart(l): | |
84 ret = l - context | |
85 if ret < 0: | |
86 return 0 | |
87 return ret | |
88 | |
89 def yieldhunk(hunk, header): | |
90 if header: | |
91 for x in header: | |
92 yield x | |
93 (astart, a2, bstart, b2, delta) = hunk | |
94 aend = contextend(a2, len(l1)) | |
95 alen = aend - astart | |
96 blen = b2 - bstart + aend - a2 | |
97 | |
98 func = "" | |
99 if showfunc: | |
100 # walk backwards from the start of the context | |
101 # to find a line starting with an alphanumeric char. | |
102 for x in xrange(astart, -1, -1): | |
103 t = l1[x].rstrip() | |
104 if funcre.match(t): | |
105 func = ' ' + t[:40] | |
106 break | |
107 | |
108 yield "@@ -%d,%d +%d,%d @@%s\n" % (astart + 1, alen, | |
109 bstart + 1, blen, func) | |
110 for x in delta: | |
111 yield x | |
112 for x in xrange(a2, aend): | |
113 yield ' ' + l1[x] | |
114 | |
115 header = [ "--- %s\t\n" % header1, "+++ %s\t\n" % header2 ] | |
116 | |
117 if showfunc: | |
118 funcre = re.compile('\w') | |
119 if ignorews: | |
120 wsre = re.compile('[ \t]') | |
121 | |
122 # bdiff.blocks gives us the matching sequences in the files. The loop | |
123 # below finds the spaces between those matching sequences and translates | |
124 # them into diff output. | |
125 # | |
126 diff = bdiff.blocks(t1, t2) | |
127 hunk = None | |
128 for i in xrange(len(diff)): | |
129 # The first match is special. | |
130 # we've either found a match starting at line 0 or a match later | |
131 # in the file. If it starts later, old and new below will both be | |
132 # empty and we'll continue to the next match. | |
133 if i > 0: | |
134 s = diff[i-1] | |
135 else: | |
136 s = [0, 0, 0, 0] | |
137 delta = [] | |
138 s1 = diff[i] | |
139 a1 = s[1] | |
140 a2 = s1[0] | |
141 b1 = s[3] | |
142 b2 = s1[2] | |
143 | |
144 old = l1[a1:a2] | |
145 new = l2[b1:b2] | |
146 | |
147 # bdiff sometimes gives huge matches past eof, this check eats them, | |
148 # and deals with the special first match case described above | |
149 if not old and not new: | |
150 continue | |
151 | |
152 if ignorews: | |
153 wsold = wsre.sub('', "".join(old)) | |
154 wsnew = wsre.sub('', "".join(new)) | |
155 if wsold == wsnew: | |
156 continue | |
157 | |
158 astart = contextstart(a1) | |
159 bstart = contextstart(b1) | |
160 prev = None | |
161 if hunk: | |
162 # join with the previous hunk if it falls inside the context | |
163 if astart < hunk[1] + context + 1: | |
164 prev = hunk | |
165 astart = hunk[1] | |
166 bstart = hunk[3] | |
167 else: | |
168 for x in yieldhunk(hunk, header): | |
169 yield x | |
170 # we only want to yield the header if the files differ, and | |
171 # we only want to yield it once. | |
172 header = None | |
173 if prev: | |
174 # we've joined the previous hunk, record the new ending points. | |
175 hunk[1] = a2 | |
176 hunk[3] = b2 | |
177 delta = hunk[4] | |
178 else: | |
179 # create a new hunk | |
180 hunk = [ astart, a2, bstart, b2, delta ] | |
181 | |
182 delta[len(delta):] = [ ' ' + x for x in l1[astart:a1] ] | |
183 delta[len(delta):] = [ '-' + x for x in old ] | |
184 delta[len(delta):] = [ '+' + x for x in new ] | |
185 | |
186 if hunk: | |
187 for x in yieldhunk(hunk, header): | |
188 yield x | |
189 | |
120
bae6f0328f63
Add a function to return the new text from a binary diff
mpm@selenic.com
parents:
75
diff
changeset
|
190 def patchtext(bin): |
bae6f0328f63
Add a function to return the new text from a binary diff
mpm@selenic.com
parents:
75
diff
changeset
|
191 pos = 0 |
bae6f0328f63
Add a function to return the new text from a binary diff
mpm@selenic.com
parents:
75
diff
changeset
|
192 t = [] |
bae6f0328f63
Add a function to return the new text from a binary diff
mpm@selenic.com
parents:
75
diff
changeset
|
193 while pos < len(bin): |
bae6f0328f63
Add a function to return the new text from a binary diff
mpm@selenic.com
parents:
75
diff
changeset
|
194 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
|
195 pos += 12 |
bae6f0328f63
Add a function to return the new text from a binary diff
mpm@selenic.com
parents:
75
diff
changeset
|
196 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
|
197 pos += l |
bae6f0328f63
Add a function to return the new text from a binary diff
mpm@selenic.com
parents:
75
diff
changeset
|
198 return "".join(t) |
bae6f0328f63
Add a function to return the new text from a binary diff
mpm@selenic.com
parents:
75
diff
changeset
|
199 |
0
9117c6561b0b
Add back links from file revisions to changeset revisions
mpm@selenic.com
parents:
diff
changeset
|
200 def patch(a, bin): |
1379 | 201 return mpatch.patches(a, [bin]) |
432 | 202 |
1379 | 203 patches = mpatch.patches |
2078
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1723
diff
changeset
|
204 patchedsize = mpatch.patchedsize |
432 | 205 textdiff = bdiff.bdiff |