Mercurial > hg
annotate mercurial/mpatch.c @ 12583:903828be7397
mq: explain qpush -f better
author | Mads Kiilerich <mads@kiilerich.com> |
---|---|
date | Wed, 29 Sep 2010 01:32:50 +0200 |
parents | 2ac98313b26c |
children | 613b8bd2284e |
rev | line source |
---|---|
72 | 1 /* |
2 mpatch.c - efficient binary patching for Mercurial | |
3 | |
4 This implements a patch algorithm that's O(m + nlog n) where m is the | |
5 size of the output and n is the number of patches. | |
6 | |
7 Given a list of binary patches, it unpacks each into a hunk list, | |
8 then combines the hunk lists with a treewise recursion to form a | |
9 single hunk list. This hunk list is then applied to the original | |
10 text. | |
11 | |
12 The text (or binary) fragments are copied directly from their source | |
13 Python objects into a preallocated output string to avoid the | |
14 allocation of intermediate Python objects. Working memory is about 2x | |
15 the total number of hunks. | |
16 | |
2859 | 17 Copyright 2005, 2006 Matt Mackall <mpm@selenic.com> |
72 | 18 |
19 This software may be used and distributed according to the terms | |
20 of the GNU General Public License, incorporated herein by reference. | |
21 */ | |
22 | |
23 #include <Python.h> | |
24 #include <stdlib.h> | |
25 #include <string.h> | |
2468
1ac0574f1768
mac os x: fixes for 10.2 from chris monson <monpublic@gmail.com>
Vadim Gelfer <vadim.gelfer@gmail.com>
parents:
2083
diff
changeset
|
26 |
11360
2ac98313b26c
mpatch.c: Added preliminary support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
27 #include "util.h" |
2ac98313b26c
mpatch.c: Added preliminary support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
28 |
5459
b0e5f44fdeb3
mpatch: Define Py_ssize_t for old pythons and use it instead of ssize_t.
Shun-ichi GOTO <shunichi.goto@gmail.com>
parents:
5444
diff
changeset
|
29 /* Definitions to get compatibility with python 2.4 and earlier which |
b0e5f44fdeb3
mpatch: Define Py_ssize_t for old pythons and use it instead of ssize_t.
Shun-ichi GOTO <shunichi.goto@gmail.com>
parents:
5444
diff
changeset
|
30 does not have Py_ssize_t. See also PEP 353. |
b0e5f44fdeb3
mpatch: Define Py_ssize_t for old pythons and use it instead of ssize_t.
Shun-ichi GOTO <shunichi.goto@gmail.com>
parents:
5444
diff
changeset
|
31 Note: msvc (8 or earlier) does not have ssize_t, so we use Py_ssize_t. |
b0e5f44fdeb3
mpatch: Define Py_ssize_t for old pythons and use it instead of ssize_t.
Shun-ichi GOTO <shunichi.goto@gmail.com>
parents:
5444
diff
changeset
|
32 */ |
b0e5f44fdeb3
mpatch: Define Py_ssize_t for old pythons and use it instead of ssize_t.
Shun-ichi GOTO <shunichi.goto@gmail.com>
parents:
5444
diff
changeset
|
33 #if PY_VERSION_HEX < 0x02050000 && !defined(PY_SSIZE_T_MIN) |
b0e5f44fdeb3
mpatch: Define Py_ssize_t for old pythons and use it instead of ssize_t.
Shun-ichi GOTO <shunichi.goto@gmail.com>
parents:
5444
diff
changeset
|
34 typedef int Py_ssize_t; |
b0e5f44fdeb3
mpatch: Define Py_ssize_t for old pythons and use it instead of ssize_t.
Shun-ichi GOTO <shunichi.goto@gmail.com>
parents:
5444
diff
changeset
|
35 #define PY_SSIZE_T_MAX INT_MAX |
b0e5f44fdeb3
mpatch: Define Py_ssize_t for old pythons and use it instead of ssize_t.
Shun-ichi GOTO <shunichi.goto@gmail.com>
parents:
5444
diff
changeset
|
36 #define PY_SSIZE_T_MIN INT_MIN |
b0e5f44fdeb3
mpatch: Define Py_ssize_t for old pythons and use it instead of ssize_t.
Shun-ichi GOTO <shunichi.goto@gmail.com>
parents:
5444
diff
changeset
|
37 #endif |
b0e5f44fdeb3
mpatch: Define Py_ssize_t for old pythons and use it instead of ssize_t.
Shun-ichi GOTO <shunichi.goto@gmail.com>
parents:
5444
diff
changeset
|
38 |
410
7c678976df3e
Make mpatch.c compilable under the other `OS'
mpm@selenic.com
parents:
384
diff
changeset
|
39 #ifdef _WIN32 |
10282
08a0f04b56bd
many, many trivial check-code fixups
Matt Mackall <mpm@selenic.com>
parents:
7036
diff
changeset
|
40 #ifdef _MSC_VER |
2468
1ac0574f1768
mac os x: fixes for 10.2 from chris monson <monpublic@gmail.com>
Vadim Gelfer <vadim.gelfer@gmail.com>
parents:
2083
diff
changeset
|
41 /* msvc 6.0 has problems */ |
10282
08a0f04b56bd
many, many trivial check-code fixups
Matt Mackall <mpm@selenic.com>
parents:
7036
diff
changeset
|
42 #define inline __inline |
551 | 43 typedef unsigned long uint32_t; |
10282
08a0f04b56bd
many, many trivial check-code fixups
Matt Mackall <mpm@selenic.com>
parents:
7036
diff
changeset
|
44 #else |
08a0f04b56bd
many, many trivial check-code fixups
Matt Mackall <mpm@selenic.com>
parents:
7036
diff
changeset
|
45 #include <stdint.h> |
08a0f04b56bd
many, many trivial check-code fixups
Matt Mackall <mpm@selenic.com>
parents:
7036
diff
changeset
|
46 #endif |
411
9e9f7ab43ce2
Add 'other OS' bits to bdiff.c / style cleanups
mpm@selenic.com
parents:
410
diff
changeset
|
47 static uint32_t ntohl(uint32_t x) |
9e9f7ab43ce2
Add 'other OS' bits to bdiff.c / style cleanups
mpm@selenic.com
parents:
410
diff
changeset
|
48 { |
9e9f7ab43ce2
Add 'other OS' bits to bdiff.c / style cleanups
mpm@selenic.com
parents:
410
diff
changeset
|
49 return ((x & 0x000000ffUL) << 24) | |
9e9f7ab43ce2
Add 'other OS' bits to bdiff.c / style cleanups
mpm@selenic.com
parents:
410
diff
changeset
|
50 ((x & 0x0000ff00UL) << 8) | |
9e9f7ab43ce2
Add 'other OS' bits to bdiff.c / style cleanups
mpm@selenic.com
parents:
410
diff
changeset
|
51 ((x & 0x00ff0000UL) >> 8) | |
9e9f7ab43ce2
Add 'other OS' bits to bdiff.c / style cleanups
mpm@selenic.com
parents:
410
diff
changeset
|
52 ((x & 0xff000000UL) >> 24); |
410
7c678976df3e
Make mpatch.c compilable under the other `OS'
mpm@selenic.com
parents:
384
diff
changeset
|
53 } |
7c678976df3e
Make mpatch.c compilable under the other `OS'
mpm@selenic.com
parents:
384
diff
changeset
|
54 #else |
2468
1ac0574f1768
mac os x: fixes for 10.2 from chris monson <monpublic@gmail.com>
Vadim Gelfer <vadim.gelfer@gmail.com>
parents:
2083
diff
changeset
|
55 /* not windows */ |
10282
08a0f04b56bd
many, many trivial check-code fixups
Matt Mackall <mpm@selenic.com>
parents:
7036
diff
changeset
|
56 #include <sys/types.h> |
08a0f04b56bd
many, many trivial check-code fixups
Matt Mackall <mpm@selenic.com>
parents:
7036
diff
changeset
|
57 #if defined __BEOS__ && !defined __HAIKU__ |
08a0f04b56bd
many, many trivial check-code fixups
Matt Mackall <mpm@selenic.com>
parents:
7036
diff
changeset
|
58 #include <ByteOrder.h> |
08a0f04b56bd
many, many trivial check-code fixups
Matt Mackall <mpm@selenic.com>
parents:
7036
diff
changeset
|
59 #else |
08a0f04b56bd
many, many trivial check-code fixups
Matt Mackall <mpm@selenic.com>
parents:
7036
diff
changeset
|
60 #include <arpa/inet.h> |
08a0f04b56bd
many, many trivial check-code fixups
Matt Mackall <mpm@selenic.com>
parents:
7036
diff
changeset
|
61 #endif |
08a0f04b56bd
many, many trivial check-code fixups
Matt Mackall <mpm@selenic.com>
parents:
7036
diff
changeset
|
62 #include <inttypes.h> |
410
7c678976df3e
Make mpatch.c compilable under the other `OS'
mpm@selenic.com
parents:
384
diff
changeset
|
63 #endif |
72 | 64 |
65 static char mpatch_doc[] = "Efficient binary patching."; | |
1722
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
66 static PyObject *mpatch_Error; |
72 | 67 |
68 struct frag { | |
69 int start, end, len; | |
5444
a0952e4e52eb
mpatch: allow buffer objects for input
Matt Mackall <mpm@selenic.com>
parents:
4377
diff
changeset
|
70 const char *data; |
72 | 71 }; |
72 | |
73 struct flist { | |
74 struct frag *base, *head, *tail; | |
75 }; | |
76 | |
77 static struct flist *lalloc(int size) | |
78 { | |
128 | 79 struct flist *a = NULL; |
72 | 80 |
3138
cc856c4d91ca
mpatch: Fix for malloc corner case on AIX
Matt Mackall <mpm@selenic.com>
parents:
2859
diff
changeset
|
81 if (size < 1) |
cc856c4d91ca
mpatch: Fix for malloc corner case on AIX
Matt Mackall <mpm@selenic.com>
parents:
2859
diff
changeset
|
82 size = 1; |
cc856c4d91ca
mpatch: Fix for malloc corner case on AIX
Matt Mackall <mpm@selenic.com>
parents:
2859
diff
changeset
|
83 |
1978
10606ee61107
do proper typecasting on malloc() and calloc() calls
TK Soh <teekaysoh@yahoo.com>
parents:
1746
diff
changeset
|
84 a = (struct flist *)malloc(sizeof(struct flist)); |
128 | 85 if (a) { |
1978
10606ee61107
do proper typecasting on malloc() and calloc() calls
TK Soh <teekaysoh@yahoo.com>
parents:
1746
diff
changeset
|
86 a->base = (struct frag *)malloc(sizeof(struct frag) * size); |
2048
8f9660c568b8
Set correct exception for another possible malloc error in mpatch.c
Thomas Arendsen Hein <thomas@intevation.de>
parents:
1978
diff
changeset
|
87 if (a->base) { |
128 | 88 a->head = a->tail = a->base; |
2048
8f9660c568b8
Set correct exception for another possible malloc error in mpatch.c
Thomas Arendsen Hein <thomas@intevation.de>
parents:
1978
diff
changeset
|
89 return a; |
8f9660c568b8
Set correct exception for another possible malloc error in mpatch.c
Thomas Arendsen Hein <thomas@intevation.de>
parents:
1978
diff
changeset
|
90 } |
8f9660c568b8
Set correct exception for another possible malloc error in mpatch.c
Thomas Arendsen Hein <thomas@intevation.de>
parents:
1978
diff
changeset
|
91 free(a); |
8f9660c568b8
Set correct exception for another possible malloc error in mpatch.c
Thomas Arendsen Hein <thomas@intevation.de>
parents:
1978
diff
changeset
|
92 a = NULL; |
128 | 93 } |
1722
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
94 if (!PyErr_Occurred()) |
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
95 PyErr_NoMemory(); |
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
96 return NULL; |
72 | 97 } |
98 | |
99 static void lfree(struct flist *a) | |
100 { | |
128 | 101 if (a) { |
102 free(a->base); | |
103 free(a); | |
104 } | |
72 | 105 } |
106 | |
107 static int lsize(struct flist *a) | |
108 { | |
109 return a->tail - a->head; | |
110 } | |
111 | |
112 /* move hunks in source that are less cut to dest, compensating | |
113 for changes in offset. the last hunk may be split if necessary. | |
114 */ | |
115 static int gather(struct flist *dest, struct flist *src, int cut, int offset) | |
116 { | |
117 struct frag *d = dest->tail, *s = src->head; | |
118 int postend, c, l; | |
119 | |
120 while (s != src->tail) { | |
121 if (s->start + offset >= cut) | |
82 | 122 break; /* we've gone far enough */ |
72 | 123 |
124 postend = offset + s->start + s->len; | |
125 if (postend <= cut) { | |
126 /* save this hunk */ | |
127 offset += s->start + s->len - s->end; | |
128 *d++ = *s++; | |
129 } | |
130 else { | |
131 /* break up this hunk */ | |
132 c = cut - offset; | |
133 if (s->end < c) | |
134 c = s->end; | |
135 l = cut - offset - s->start; | |
136 if (s->len < l) | |
137 l = s->len; | |
138 | |
139 offset += s->start + l - c; | |
140 | |
141 d->start = s->start; | |
142 d->end = c; | |
143 d->len = l; | |
144 d->data = s->data; | |
145 d++; | |
146 s->start = c; | |
147 s->len = s->len - l; | |
148 s->data = s->data + l; | |
149 | |
82 | 150 break; |
72 | 151 } |
152 } | |
153 | |
154 dest->tail = d; | |
155 src->head = s; | |
156 return offset; | |
157 } | |
158 | |
159 /* like gather, but with no output list */ | |
160 static int discard(struct flist *src, int cut, int offset) | |
161 { | |
162 struct frag *s = src->head; | |
163 int postend, c, l; | |
164 | |
165 while (s != src->tail) { | |
166 if (s->start + offset >= cut) | |
82 | 167 break; |
72 | 168 |
169 postend = offset + s->start + s->len; | |
170 if (postend <= cut) { | |
171 offset += s->start + s->len - s->end; | |
172 s++; | |
173 } | |
174 else { | |
175 c = cut - offset; | |
176 if (s->end < c) | |
177 c = s->end; | |
178 l = cut - offset - s->start; | |
179 if (s->len < l) | |
180 l = s->len; | |
181 | |
182 offset += s->start + l - c; | |
183 s->start = c; | |
184 s->len = s->len - l; | |
185 s->data = s->data + l; | |
186 | |
82 | 187 break; |
72 | 188 } |
189 } | |
190 | |
191 src->head = s; | |
192 return offset; | |
193 } | |
194 | |
195 /* combine hunk lists a and b, while adjusting b for offset changes in a/ | |
196 this deletes a and b and returns the resultant list. */ | |
197 static struct flist *combine(struct flist *a, struct flist *b) | |
198 { | |
128 | 199 struct flist *c = NULL; |
200 struct frag *bh, *ct; | |
72 | 201 int offset = 0, post; |
202 | |
128 | 203 if (a && b) |
204 c = lalloc((lsize(a) + lsize(b)) * 2); | |
205 | |
206 if (c) { | |
72 | 207 |
128 | 208 for (bh = b->head; bh != b->tail; bh++) { |
209 /* save old hunks */ | |
210 offset = gather(c, a, bh->start, offset); | |
72 | 211 |
128 | 212 /* discard replaced hunks */ |
213 post = discard(a, bh->end, offset); | |
72 | 214 |
128 | 215 /* insert new hunk */ |
216 ct = c->tail; | |
217 ct->start = bh->start - offset; | |
218 ct->end = bh->end - post; | |
219 ct->len = bh->len; | |
220 ct->data = bh->data; | |
221 c->tail++; | |
222 offset = post; | |
223 } | |
224 | |
225 /* hold on to tail from a */ | |
226 memcpy(c->tail, a->head, sizeof(struct frag) * lsize(a)); | |
227 c->tail += lsize(a); | |
72 | 228 } |
229 | |
230 lfree(a); | |
231 lfree(b); | |
232 return c; | |
233 } | |
234 | |
235 /* decode a binary patch into a hunk list */ | |
5444
a0952e4e52eb
mpatch: allow buffer objects for input
Matt Mackall <mpm@selenic.com>
parents:
4377
diff
changeset
|
236 static struct flist *decode(const char *bin, int len) |
72 | 237 { |
238 struct flist *l; | |
239 struct frag *lt; | |
5444
a0952e4e52eb
mpatch: allow buffer objects for input
Matt Mackall <mpm@selenic.com>
parents:
4377
diff
changeset
|
240 const char *data = bin + 12, *end = bin + len; |
384
a29decbf7475
mpatch: attempt to handle unpack alignment issues on Solaris
mpm@selenic.com
parents:
282
diff
changeset
|
241 char decode[12]; /* for dealing with alignment issues */ |
72 | 242 |
243 /* assume worst case size, we won't have many of these lists */ | |
244 l = lalloc(len / 12); | |
1722
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
245 if (!l) |
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
246 return NULL; |
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
247 |
72 | 248 lt = l->tail; |
249 | |
4358
11dc22eb8e8d
Fix segfaults when parsing bdiff hunks in mpatch.decode() and .patchedsize()
Thomas Arendsen Hein <thomas@intevation.de>
parents:
3138
diff
changeset
|
250 while (data <= end) { |
384
a29decbf7475
mpatch: attempt to handle unpack alignment issues on Solaris
mpm@selenic.com
parents:
282
diff
changeset
|
251 memcpy(decode, bin, 12); |
a29decbf7475
mpatch: attempt to handle unpack alignment issues on Solaris
mpm@selenic.com
parents:
282
diff
changeset
|
252 lt->start = ntohl(*(uint32_t *)decode); |
a29decbf7475
mpatch: attempt to handle unpack alignment issues on Solaris
mpm@selenic.com
parents:
282
diff
changeset
|
253 lt->end = ntohl(*(uint32_t *)(decode + 4)); |
a29decbf7475
mpatch: attempt to handle unpack alignment issues on Solaris
mpm@selenic.com
parents:
282
diff
changeset
|
254 lt->len = ntohl(*(uint32_t *)(decode + 8)); |
4358
11dc22eb8e8d
Fix segfaults when parsing bdiff hunks in mpatch.decode() and .patchedsize()
Thomas Arendsen Hein <thomas@intevation.de>
parents:
3138
diff
changeset
|
255 if (lt->start > lt->end) |
11dc22eb8e8d
Fix segfaults when parsing bdiff hunks in mpatch.decode() and .patchedsize()
Thomas Arendsen Hein <thomas@intevation.de>
parents:
3138
diff
changeset
|
256 break; /* sanity check */ |
11dc22eb8e8d
Fix segfaults when parsing bdiff hunks in mpatch.decode() and .patchedsize()
Thomas Arendsen Hein <thomas@intevation.de>
parents:
3138
diff
changeset
|
257 bin = data + lt->len; |
11dc22eb8e8d
Fix segfaults when parsing bdiff hunks in mpatch.decode() and .patchedsize()
Thomas Arendsen Hein <thomas@intevation.de>
parents:
3138
diff
changeset
|
258 if (bin < data) |
11dc22eb8e8d
Fix segfaults when parsing bdiff hunks in mpatch.decode() and .patchedsize()
Thomas Arendsen Hein <thomas@intevation.de>
parents:
3138
diff
changeset
|
259 break; /* big data + big (bogus) len can wrap around */ |
11dc22eb8e8d
Fix segfaults when parsing bdiff hunks in mpatch.decode() and .patchedsize()
Thomas Arendsen Hein <thomas@intevation.de>
parents:
3138
diff
changeset
|
260 lt->data = data; |
11dc22eb8e8d
Fix segfaults when parsing bdiff hunks in mpatch.decode() and .patchedsize()
Thomas Arendsen Hein <thomas@intevation.de>
parents:
3138
diff
changeset
|
261 data = bin + 12; |
72 | 262 lt++; |
263 } | |
264 | |
1722
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
265 if (bin != end) { |
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
266 if (!PyErr_Occurred()) |
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
267 PyErr_SetString(mpatch_Error, "patch cannot be decoded"); |
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
268 lfree(l); |
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
269 return NULL; |
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
270 } |
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
271 |
72 | 272 l->tail = lt; |
273 return l; | |
274 } | |
275 | |
276 /* calculate the size of resultant text */ | |
277 static int calcsize(int len, struct flist *l) | |
278 { | |
279 int outlen = 0, last = 0; | |
280 struct frag *f = l->head; | |
281 | |
282 while (f != l->tail) { | |
1722
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
283 if (f->start < last || f->end > len) { |
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
284 if (!PyErr_Occurred()) |
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
285 PyErr_SetString(mpatch_Error, |
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
286 "invalid patch"); |
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
287 return -1; |
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
288 } |
72 | 289 outlen += f->start - last; |
290 last = f->end; | |
291 outlen += f->len; | |
292 f++; | |
293 } | |
294 | |
295 outlen += len - last; | |
296 return outlen; | |
297 } | |
298 | |
5444
a0952e4e52eb
mpatch: allow buffer objects for input
Matt Mackall <mpm@selenic.com>
parents:
4377
diff
changeset
|
299 static int apply(char *buf, const char *orig, int len, struct flist *l) |
72 | 300 { |
301 struct frag *f = l->head; | |
302 int last = 0; | |
303 char *p = buf; | |
304 | |
305 while (f != l->tail) { | |
1722
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
306 if (f->start < last || f->end > len) { |
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
307 if (!PyErr_Occurred()) |
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
308 PyErr_SetString(mpatch_Error, |
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
309 "invalid patch"); |
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
310 return 0; |
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
311 } |
72 | 312 memcpy(p, orig + last, f->start - last); |
313 p += f->start - last; | |
314 memcpy(p, f->data, f->len); | |
315 last = f->end; | |
316 p += f->len; | |
317 f++; | |
318 } | |
319 memcpy(p, orig + last, len - last); | |
1722
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
320 return 1; |
72 | 321 } |
322 | |
323 /* recursively generate a patch of all bins between start and end */ | |
324 static struct flist *fold(PyObject *bins, int start, int end) | |
325 { | |
326 int len; | |
5459
b0e5f44fdeb3
mpatch: Define Py_ssize_t for old pythons and use it instead of ssize_t.
Shun-ichi GOTO <shunichi.goto@gmail.com>
parents:
5444
diff
changeset
|
327 Py_ssize_t blen; |
5444
a0952e4e52eb
mpatch: allow buffer objects for input
Matt Mackall <mpm@selenic.com>
parents:
4377
diff
changeset
|
328 const char *buffer; |
72 | 329 |
330 if (start + 1 == end) { | |
331 /* trivial case, output a decoded list */ | |
332 PyObject *tmp = PyList_GetItem(bins, start); | |
128 | 333 if (!tmp) |
334 return NULL; | |
5444
a0952e4e52eb
mpatch: allow buffer objects for input
Matt Mackall <mpm@selenic.com>
parents:
4377
diff
changeset
|
335 if (PyObject_AsCharBuffer(tmp, &buffer, &blen)) |
a0952e4e52eb
mpatch: allow buffer objects for input
Matt Mackall <mpm@selenic.com>
parents:
4377
diff
changeset
|
336 return NULL; |
a0952e4e52eb
mpatch: allow buffer objects for input
Matt Mackall <mpm@selenic.com>
parents:
4377
diff
changeset
|
337 return decode(buffer, blen); |
72 | 338 } |
339 | |
340 /* divide and conquer, memory management is elsewhere */ | |
341 len = (end - start) / 2; | |
342 return combine(fold(bins, start, start + len), | |
343 fold(bins, start + len, end)); | |
344 } | |
345 | |
346 static PyObject * | |
347 patches(PyObject *self, PyObject *args) | |
348 { | |
349 PyObject *text, *bins, *result; | |
350 struct flist *patch; | |
5444
a0952e4e52eb
mpatch: allow buffer objects for input
Matt Mackall <mpm@selenic.com>
parents:
4377
diff
changeset
|
351 const char *in; |
a0952e4e52eb
mpatch: allow buffer objects for input
Matt Mackall <mpm@selenic.com>
parents:
4377
diff
changeset
|
352 char *out; |
72 | 353 int len, outlen; |
5459
b0e5f44fdeb3
mpatch: Define Py_ssize_t for old pythons and use it instead of ssize_t.
Shun-ichi GOTO <shunichi.goto@gmail.com>
parents:
5444
diff
changeset
|
354 Py_ssize_t inlen; |
72 | 355 |
5444
a0952e4e52eb
mpatch: allow buffer objects for input
Matt Mackall <mpm@selenic.com>
parents:
4377
diff
changeset
|
356 if (!PyArg_ParseTuple(args, "OO:mpatch", &text, &bins)) |
72 | 357 return NULL; |
358 | |
359 len = PyList_Size(bins); | |
360 if (!len) { | |
361 /* nothing to do */ | |
362 Py_INCREF(text); | |
363 return text; | |
364 } | |
365 | |
5444
a0952e4e52eb
mpatch: allow buffer objects for input
Matt Mackall <mpm@selenic.com>
parents:
4377
diff
changeset
|
366 if (PyObject_AsCharBuffer(text, &in, &inlen)) |
a0952e4e52eb
mpatch: allow buffer objects for input
Matt Mackall <mpm@selenic.com>
parents:
4377
diff
changeset
|
367 return NULL; |
a0952e4e52eb
mpatch: allow buffer objects for input
Matt Mackall <mpm@selenic.com>
parents:
4377
diff
changeset
|
368 |
72 | 369 patch = fold(bins, 0, len); |
128 | 370 if (!patch) |
1722
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
371 return NULL; |
128 | 372 |
5444
a0952e4e52eb
mpatch: allow buffer objects for input
Matt Mackall <mpm@selenic.com>
parents:
4377
diff
changeset
|
373 outlen = calcsize(inlen, patch); |
1722
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
374 if (outlen < 0) { |
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
375 result = NULL; |
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
376 goto cleanup; |
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
377 } |
11360
2ac98313b26c
mpatch.c: Added preliminary support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
378 result = PyBytes_FromStringAndSize(NULL, outlen); |
1722
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
379 if (!result) { |
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
380 result = NULL; |
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
381 goto cleanup; |
128 | 382 } |
11360
2ac98313b26c
mpatch.c: Added preliminary support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
383 out = PyBytes_AsString(result); |
5444
a0952e4e52eb
mpatch: allow buffer objects for input
Matt Mackall <mpm@selenic.com>
parents:
4377
diff
changeset
|
384 if (!apply(out, in, inlen, patch)) { |
1722
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
385 Py_DECREF(result); |
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
386 result = NULL; |
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
387 } |
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
388 cleanup: |
72 | 389 lfree(patch); |
390 return result; | |
391 } | |
392 | |
2078
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
393 /* calculate size of a patched file directly */ |
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
394 static PyObject * |
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
395 patchedsize(PyObject *self, PyObject *args) |
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
396 { |
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
397 long orig, start, end, len, outlen = 0, last = 0; |
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
398 int patchlen; |
4358
11dc22eb8e8d
Fix segfaults when parsing bdiff hunks in mpatch.decode() and .patchedsize()
Thomas Arendsen Hein <thomas@intevation.de>
parents:
3138
diff
changeset
|
399 char *bin, *binend, *data; |
2078
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
400 char decode[12]; /* for dealing with alignment issues */ |
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
401 |
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
402 if (!PyArg_ParseTuple(args, "ls#", &orig, &bin, &patchlen)) |
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
403 return NULL; |
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
404 |
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
405 binend = bin + patchlen; |
4358
11dc22eb8e8d
Fix segfaults when parsing bdiff hunks in mpatch.decode() and .patchedsize()
Thomas Arendsen Hein <thomas@intevation.de>
parents:
3138
diff
changeset
|
406 data = bin + 12; |
2078
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
407 |
4358
11dc22eb8e8d
Fix segfaults when parsing bdiff hunks in mpatch.decode() and .patchedsize()
Thomas Arendsen Hein <thomas@intevation.de>
parents:
3138
diff
changeset
|
408 while (data <= binend) { |
2078
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
409 memcpy(decode, bin, 12); |
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
410 start = ntohl(*(uint32_t *)decode); |
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
411 end = ntohl(*(uint32_t *)(decode + 4)); |
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
412 len = ntohl(*(uint32_t *)(decode + 8)); |
4358
11dc22eb8e8d
Fix segfaults when parsing bdiff hunks in mpatch.decode() and .patchedsize()
Thomas Arendsen Hein <thomas@intevation.de>
parents:
3138
diff
changeset
|
413 if (start > end) |
11dc22eb8e8d
Fix segfaults when parsing bdiff hunks in mpatch.decode() and .patchedsize()
Thomas Arendsen Hein <thomas@intevation.de>
parents:
3138
diff
changeset
|
414 break; /* sanity check */ |
11dc22eb8e8d
Fix segfaults when parsing bdiff hunks in mpatch.decode() and .patchedsize()
Thomas Arendsen Hein <thomas@intevation.de>
parents:
3138
diff
changeset
|
415 bin = data + len; |
11dc22eb8e8d
Fix segfaults when parsing bdiff hunks in mpatch.decode() and .patchedsize()
Thomas Arendsen Hein <thomas@intevation.de>
parents:
3138
diff
changeset
|
416 if (bin < data) |
11dc22eb8e8d
Fix segfaults when parsing bdiff hunks in mpatch.decode() and .patchedsize()
Thomas Arendsen Hein <thomas@intevation.de>
parents:
3138
diff
changeset
|
417 break; /* big data + big (bogus) len can wrap around */ |
11dc22eb8e8d
Fix segfaults when parsing bdiff hunks in mpatch.decode() and .patchedsize()
Thomas Arendsen Hein <thomas@intevation.de>
parents:
3138
diff
changeset
|
418 data = bin + 12; |
2078
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
419 outlen += start - last; |
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
420 last = end; |
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
421 outlen += len; |
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
422 } |
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
423 |
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
424 if (bin != binend) { |
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
425 if (!PyErr_Occurred()) |
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
426 PyErr_SetString(mpatch_Error, "patch cannot be decoded"); |
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
427 return NULL; |
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
428 } |
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
429 |
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
430 outlen += orig - last; |
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
431 return Py_BuildValue("l", outlen); |
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
432 } |
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
433 |
72 | 434 static PyMethodDef methods[] = { |
435 {"patches", patches, METH_VARARGS, "apply a series of patches\n"}, | |
2078
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
436 {"patchedsize", patchedsize, METH_VARARGS, "calculed patched size\n"}, |
72 | 437 {NULL, NULL} |
438 }; | |
439 | |
11360
2ac98313b26c
mpatch.c: Added preliminary support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
440 #ifdef IS_PY3K |
2ac98313b26c
mpatch.c: Added preliminary support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
441 static struct PyModuleDef mpatch_module = { |
2ac98313b26c
mpatch.c: Added preliminary support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
442 PyModuleDef_HEAD_INIT, |
2ac98313b26c
mpatch.c: Added preliminary support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
443 "mpatch", |
2ac98313b26c
mpatch.c: Added preliminary support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
444 mpatch_doc, |
2ac98313b26c
mpatch.c: Added preliminary support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
445 -1, |
2ac98313b26c
mpatch.c: Added preliminary support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
446 methods |
2ac98313b26c
mpatch.c: Added preliminary support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
447 }; |
2ac98313b26c
mpatch.c: Added preliminary support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
448 |
2ac98313b26c
mpatch.c: Added preliminary support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
449 PyMODINIT_FUNC PyInit_mpatch(void) |
2ac98313b26c
mpatch.c: Added preliminary support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
450 { |
2ac98313b26c
mpatch.c: Added preliminary support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
451 PyObject *m; |
2ac98313b26c
mpatch.c: Added preliminary support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
452 |
2ac98313b26c
mpatch.c: Added preliminary support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
453 m = PyModule_Create(&mpatch_module); |
2ac98313b26c
mpatch.c: Added preliminary support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
454 if (m == NULL) |
2ac98313b26c
mpatch.c: Added preliminary support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
455 return NULL; |
2ac98313b26c
mpatch.c: Added preliminary support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
456 |
2ac98313b26c
mpatch.c: Added preliminary support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
457 mpatch_Error = PyErr_NewException("mpatch.mpatchError", NULL, NULL); |
2ac98313b26c
mpatch.c: Added preliminary support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
458 Py_INCREF(mpatch_Error); |
2ac98313b26c
mpatch.c: Added preliminary support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
459 PyModule_AddObject(m, "mpatchError", mpatch_Error); |
2ac98313b26c
mpatch.c: Added preliminary support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
460 |
2ac98313b26c
mpatch.c: Added preliminary support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
461 return m; |
2ac98313b26c
mpatch.c: Added preliminary support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
462 } |
2ac98313b26c
mpatch.c: Added preliminary support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
463 #else |
72 | 464 PyMODINIT_FUNC |
465 initmpatch(void) | |
466 { | |
467 Py_InitModule3("mpatch", methods, mpatch_doc); | |
1722
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
468 mpatch_Error = PyErr_NewException("mpatch.mpatchError", NULL, NULL); |
72 | 469 } |
11360
2ac98313b26c
mpatch.c: Added preliminary support for py3k.
Renato Cunha <renatoc@gmail.com>
parents:
10282
diff
changeset
|
470 #endif |