Mercurial > hg > mercurial-crew-with-dirclash
annotate mercurial/mpatch.c @ 5459:b0e5f44fdeb3
mpatch: Define Py_ssize_t for old pythons and use it instead of ssize_t.
See also PEP 353.
NOTE: Microsoft compilers (8 or earlier) does not have ssize_t.
| author | Shun-ichi GOTO <shunichi.goto@gmail.com> |
|---|---|
| date | Fri, 12 Oct 2007 11:46:49 +0900 |
| parents | a0952e4e52eb |
| children | fe9b0bb3eb1c |
| 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 | |
| 2858 | 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 |
|
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
|
27 /* 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
|
28 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
|
29 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
|
30 */ |
|
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 #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
|
32 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
|
33 #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
|
34 #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
|
35 #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
|
36 |
|
410
7c678976df3e
Make mpatch.c compilable under the other `OS'
mpm@selenic.com
parents:
384
diff
changeset
|
37 #ifdef _WIN32 |
|
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
|
38 # ifdef _MSC_VER |
|
1ac0574f1768
mac os x: fixes for 10.2 from chris monson <monpublic@gmail.com>
Vadim Gelfer <vadim.gelfer@gmail.com>
parents:
2083
diff
changeset
|
39 /* msvc 6.0 has problems */ |
|
1ac0574f1768
mac os x: fixes for 10.2 from chris monson <monpublic@gmail.com>
Vadim Gelfer <vadim.gelfer@gmail.com>
parents:
2083
diff
changeset
|
40 # define inline __inline |
| 551 | 41 typedef unsigned long uint32_t; |
|
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
|
42 # else |
|
1ac0574f1768
mac os x: fixes for 10.2 from chris monson <monpublic@gmail.com>
Vadim Gelfer <vadim.gelfer@gmail.com>
parents:
2083
diff
changeset
|
43 # include <stdint.h> |
|
1ac0574f1768
mac os x: fixes for 10.2 from chris monson <monpublic@gmail.com>
Vadim Gelfer <vadim.gelfer@gmail.com>
parents:
2083
diff
changeset
|
44 # endif |
|
411
9e9f7ab43ce2
Add 'other OS' bits to bdiff.c / style cleanups
mpm@selenic.com
parents:
410
diff
changeset
|
45 static uint32_t ntohl(uint32_t x) |
|
9e9f7ab43ce2
Add 'other OS' bits to bdiff.c / style cleanups
mpm@selenic.com
parents:
410
diff
changeset
|
46 { |
|
9e9f7ab43ce2
Add 'other OS' bits to bdiff.c / style cleanups
mpm@selenic.com
parents:
410
diff
changeset
|
47 return ((x & 0x000000ffUL) << 24) | |
|
9e9f7ab43ce2
Add 'other OS' bits to bdiff.c / style cleanups
mpm@selenic.com
parents:
410
diff
changeset
|
48 ((x & 0x0000ff00UL) << 8) | |
|
9e9f7ab43ce2
Add 'other OS' bits to bdiff.c / style cleanups
mpm@selenic.com
parents:
410
diff
changeset
|
49 ((x & 0x00ff0000UL) >> 8) | |
|
9e9f7ab43ce2
Add 'other OS' bits to bdiff.c / style cleanups
mpm@selenic.com
parents:
410
diff
changeset
|
50 ((x & 0xff000000UL) >> 24); |
|
410
7c678976df3e
Make mpatch.c compilable under the other `OS'
mpm@selenic.com
parents:
384
diff
changeset
|
51 } |
|
7c678976df3e
Make mpatch.c compilable under the other `OS'
mpm@selenic.com
parents:
384
diff
changeset
|
52 #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
|
53 /* not windows */ |
|
1ac0574f1768
mac os x: fixes for 10.2 from chris monson <monpublic@gmail.com>
Vadim Gelfer <vadim.gelfer@gmail.com>
parents:
2083
diff
changeset
|
54 # include <sys/types.h> |
|
4073
95ffa36d1d2a
BeOS compatibility support
Andrew Bachmann <andrewbachmann@gmail.com>
parents:
3138
diff
changeset
|
55 # ifdef __BEOS__ |
|
95ffa36d1d2a
BeOS compatibility support
Andrew Bachmann <andrewbachmann@gmail.com>
parents:
3138
diff
changeset
|
56 # include <ByteOrder.h> |
|
95ffa36d1d2a
BeOS compatibility support
Andrew Bachmann <andrewbachmann@gmail.com>
parents:
3138
diff
changeset
|
57 # else |
|
95ffa36d1d2a
BeOS compatibility support
Andrew Bachmann <andrewbachmann@gmail.com>
parents:
3138
diff
changeset
|
58 # include <arpa/inet.h> |
|
95ffa36d1d2a
BeOS compatibility support
Andrew Bachmann <andrewbachmann@gmail.com>
parents:
3138
diff
changeset
|
59 # endif |
|
2543
860e9c83fc59
Include inttypes.h instead of stdint.h (fixes issue299)
Thomas Arendsen Hein <thomas@intevation.de>
parents:
2468
diff
changeset
|
60 # include <inttypes.h> |
|
410
7c678976df3e
Make mpatch.c compilable under the other `OS'
mpm@selenic.com
parents:
384
diff
changeset
|
61 #endif |
| 72 | 62 |
| 63 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
|
64 static PyObject *mpatch_Error; |
| 72 | 65 |
| 66 struct frag { | |
| 67 int start, end, len; | |
|
5444
a0952e4e52eb
mpatch: allow buffer objects for input
Matt Mackall <mpm@selenic.com>
parents:
4376
diff
changeset
|
68 const char *data; |
| 72 | 69 }; |
| 70 | |
| 71 struct flist { | |
| 72 struct frag *base, *head, *tail; | |
| 73 }; | |
| 74 | |
| 75 static struct flist *lalloc(int size) | |
| 76 { | |
| 128 | 77 struct flist *a = NULL; |
| 72 | 78 |
|
3138
cc856c4d91ca
mpatch: Fix for malloc corner case on AIX
Matt Mackall <mpm@selenic.com>
parents:
2858
diff
changeset
|
79 if (size < 1) |
|
cc856c4d91ca
mpatch: Fix for malloc corner case on AIX
Matt Mackall <mpm@selenic.com>
parents:
2858
diff
changeset
|
80 size = 1; |
|
cc856c4d91ca
mpatch: Fix for malloc corner case on AIX
Matt Mackall <mpm@selenic.com>
parents:
2858
diff
changeset
|
81 |
|
1978
10606ee61107
do proper typecasting on malloc() and calloc() calls
TK Soh <teekaysoh@yahoo.com>
parents:
1746
diff
changeset
|
82 a = (struct flist *)malloc(sizeof(struct flist)); |
| 128 | 83 if (a) { |
|
1978
10606ee61107
do proper typecasting on malloc() and calloc() calls
TK Soh <teekaysoh@yahoo.com>
parents:
1746
diff
changeset
|
84 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
|
85 if (a->base) { |
| 128 | 86 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
|
87 return a; |
|
8f9660c568b8
Set correct exception for another possible malloc error in mpatch.c
Thomas Arendsen Hein <thomas@intevation.de>
parents:
1978
diff
changeset
|
88 } |
|
8f9660c568b8
Set correct exception for another possible malloc error in mpatch.c
Thomas Arendsen Hein <thomas@intevation.de>
parents:
1978
diff
changeset
|
89 free(a); |
|
8f9660c568b8
Set correct exception for another possible malloc error in mpatch.c
Thomas Arendsen Hein <thomas@intevation.de>
parents:
1978
diff
changeset
|
90 a = NULL; |
| 128 | 91 } |
|
1722
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
92 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
|
93 PyErr_NoMemory(); |
|
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
94 return NULL; |
| 72 | 95 } |
| 96 | |
| 97 static void lfree(struct flist *a) | |
| 98 { | |
| 128 | 99 if (a) { |
| 100 free(a->base); | |
| 101 free(a); | |
| 102 } | |
| 72 | 103 } |
| 104 | |
| 105 static int lsize(struct flist *a) | |
| 106 { | |
| 107 return a->tail - a->head; | |
| 108 } | |
| 109 | |
| 110 /* move hunks in source that are less cut to dest, compensating | |
| 111 for changes in offset. the last hunk may be split if necessary. | |
| 112 */ | |
| 113 static int gather(struct flist *dest, struct flist *src, int cut, int offset) | |
| 114 { | |
| 115 struct frag *d = dest->tail, *s = src->head; | |
| 116 int postend, c, l; | |
| 117 | |
| 118 while (s != src->tail) { | |
| 119 if (s->start + offset >= cut) | |
| 82 | 120 break; /* we've gone far enough */ |
| 72 | 121 |
| 122 postend = offset + s->start + s->len; | |
| 123 if (postend <= cut) { | |
| 124 /* save this hunk */ | |
| 125 offset += s->start + s->len - s->end; | |
| 126 *d++ = *s++; | |
| 127 } | |
| 128 else { | |
| 129 /* break up this hunk */ | |
| 130 c = cut - offset; | |
| 131 if (s->end < c) | |
| 132 c = s->end; | |
| 133 l = cut - offset - s->start; | |
| 134 if (s->len < l) | |
| 135 l = s->len; | |
| 136 | |
| 137 offset += s->start + l - c; | |
| 138 | |
| 139 d->start = s->start; | |
| 140 d->end = c; | |
| 141 d->len = l; | |
| 142 d->data = s->data; | |
| 143 d++; | |
| 144 s->start = c; | |
| 145 s->len = s->len - l; | |
| 146 s->data = s->data + l; | |
| 147 | |
| 82 | 148 break; |
| 72 | 149 } |
| 150 } | |
| 151 | |
| 152 dest->tail = d; | |
| 153 src->head = s; | |
| 154 return offset; | |
| 155 } | |
| 156 | |
| 157 /* like gather, but with no output list */ | |
| 158 static int discard(struct flist *src, int cut, int offset) | |
| 159 { | |
| 160 struct frag *s = src->head; | |
| 161 int postend, c, l; | |
| 162 | |
| 163 while (s != src->tail) { | |
| 164 if (s->start + offset >= cut) | |
| 82 | 165 break; |
| 72 | 166 |
| 167 postend = offset + s->start + s->len; | |
| 168 if (postend <= cut) { | |
| 169 offset += s->start + s->len - s->end; | |
| 170 s++; | |
| 171 } | |
| 172 else { | |
| 173 c = cut - offset; | |
| 174 if (s->end < c) | |
| 175 c = s->end; | |
| 176 l = cut - offset - s->start; | |
| 177 if (s->len < l) | |
| 178 l = s->len; | |
| 179 | |
| 180 offset += s->start + l - c; | |
| 181 s->start = c; | |
| 182 s->len = s->len - l; | |
| 183 s->data = s->data + l; | |
| 184 | |
| 82 | 185 break; |
| 72 | 186 } |
| 187 } | |
| 188 | |
| 189 src->head = s; | |
| 190 return offset; | |
| 191 } | |
| 192 | |
| 193 /* combine hunk lists a and b, while adjusting b for offset changes in a/ | |
| 194 this deletes a and b and returns the resultant list. */ | |
| 195 static struct flist *combine(struct flist *a, struct flist *b) | |
| 196 { | |
| 128 | 197 struct flist *c = NULL; |
| 198 struct frag *bh, *ct; | |
| 72 | 199 int offset = 0, post; |
| 200 | |
| 128 | 201 if (a && b) |
| 202 c = lalloc((lsize(a) + lsize(b)) * 2); | |
| 203 | |
| 204 if (c) { | |
| 72 | 205 |
| 128 | 206 for (bh = b->head; bh != b->tail; bh++) { |
| 207 /* save old hunks */ | |
| 208 offset = gather(c, a, bh->start, offset); | |
| 72 | 209 |
| 128 | 210 /* discard replaced hunks */ |
| 211 post = discard(a, bh->end, offset); | |
| 72 | 212 |
| 128 | 213 /* insert new hunk */ |
| 214 ct = c->tail; | |
| 215 ct->start = bh->start - offset; | |
| 216 ct->end = bh->end - post; | |
| 217 ct->len = bh->len; | |
| 218 ct->data = bh->data; | |
| 219 c->tail++; | |
| 220 offset = post; | |
| 221 } | |
| 222 | |
| 223 /* hold on to tail from a */ | |
| 224 memcpy(c->tail, a->head, sizeof(struct frag) * lsize(a)); | |
| 225 c->tail += lsize(a); | |
| 72 | 226 } |
| 227 | |
| 228 lfree(a); | |
| 229 lfree(b); | |
| 230 return c; | |
| 231 } | |
| 232 | |
| 233 /* decode a binary patch into a hunk list */ | |
|
5444
a0952e4e52eb
mpatch: allow buffer objects for input
Matt Mackall <mpm@selenic.com>
parents:
4376
diff
changeset
|
234 static struct flist *decode(const char *bin, int len) |
| 72 | 235 { |
| 236 struct flist *l; | |
| 237 struct frag *lt; | |
|
5444
a0952e4e52eb
mpatch: allow buffer objects for input
Matt Mackall <mpm@selenic.com>
parents:
4376
diff
changeset
|
238 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
|
239 char decode[12]; /* for dealing with alignment issues */ |
| 72 | 240 |
| 241 /* assume worst case size, we won't have many of these lists */ | |
| 242 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
|
243 if (!l) |
|
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
244 return NULL; |
|
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
245 |
| 72 | 246 lt = l->tail; |
| 247 | |
|
4375
11dc22eb8e8d
Fix segfaults when parsing bdiff hunks in mpatch.decode() and .patchedsize()
Thomas Arendsen Hein <thomas@intevation.de>
parents:
3138
diff
changeset
|
248 while (data <= end) { |
|
384
a29decbf7475
mpatch: attempt to handle unpack alignment issues on Solaris
mpm@selenic.com
parents:
282
diff
changeset
|
249 memcpy(decode, bin, 12); |
|
a29decbf7475
mpatch: attempt to handle unpack alignment issues on Solaris
mpm@selenic.com
parents:
282
diff
changeset
|
250 lt->start = ntohl(*(uint32_t *)decode); |
|
a29decbf7475
mpatch: attempt to handle unpack alignment issues on Solaris
mpm@selenic.com
parents:
282
diff
changeset
|
251 lt->end = ntohl(*(uint32_t *)(decode + 4)); |
|
a29decbf7475
mpatch: attempt to handle unpack alignment issues on Solaris
mpm@selenic.com
parents:
282
diff
changeset
|
252 lt->len = ntohl(*(uint32_t *)(decode + 8)); |
|
4375
11dc22eb8e8d
Fix segfaults when parsing bdiff hunks in mpatch.decode() and .patchedsize()
Thomas Arendsen Hein <thomas@intevation.de>
parents:
3138
diff
changeset
|
253 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
|
254 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
|
255 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
|
256 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
|
257 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
|
258 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
|
259 data = bin + 12; |
| 72 | 260 lt++; |
| 261 } | |
| 262 | |
|
1722
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
263 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
|
264 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
|
265 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
|
266 lfree(l); |
|
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
267 return NULL; |
|
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
268 } |
|
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
269 |
| 72 | 270 l->tail = lt; |
| 271 return l; | |
| 272 } | |
| 273 | |
| 274 /* calculate the size of resultant text */ | |
| 275 static int calcsize(int len, struct flist *l) | |
| 276 { | |
| 277 int outlen = 0, last = 0; | |
| 278 struct frag *f = l->head; | |
| 279 | |
| 280 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
|
281 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
|
282 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
|
283 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
|
284 "invalid patch"); |
|
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
285 return -1; |
|
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
286 } |
| 72 | 287 outlen += f->start - last; |
| 288 last = f->end; | |
| 289 outlen += f->len; | |
| 290 f++; | |
| 291 } | |
| 292 | |
| 293 outlen += len - last; | |
| 294 return outlen; | |
| 295 } | |
| 296 | |
|
5444
a0952e4e52eb
mpatch: allow buffer objects for input
Matt Mackall <mpm@selenic.com>
parents:
4376
diff
changeset
|
297 static int apply(char *buf, const char *orig, int len, struct flist *l) |
| 72 | 298 { |
| 299 struct frag *f = l->head; | |
| 300 int last = 0; | |
| 301 char *p = buf; | |
| 302 | |
| 303 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
|
304 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
|
305 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
|
306 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
|
307 "invalid patch"); |
|
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
308 return 0; |
|
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
309 } |
| 72 | 310 memcpy(p, orig + last, f->start - last); |
| 311 p += f->start - last; | |
| 312 memcpy(p, f->data, f->len); | |
| 313 last = f->end; | |
| 314 p += f->len; | |
| 315 f++; | |
| 316 } | |
| 317 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
|
318 return 1; |
| 72 | 319 } |
| 320 | |
| 321 /* recursively generate a patch of all bins between start and end */ | |
| 322 static struct flist *fold(PyObject *bins, int start, int end) | |
| 323 { | |
| 324 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
|
325 Py_ssize_t blen; |
|
5444
a0952e4e52eb
mpatch: allow buffer objects for input
Matt Mackall <mpm@selenic.com>
parents:
4376
diff
changeset
|
326 const char *buffer; |
| 72 | 327 |
| 328 if (start + 1 == end) { | |
| 329 /* trivial case, output a decoded list */ | |
| 330 PyObject *tmp = PyList_GetItem(bins, start); | |
| 128 | 331 if (!tmp) |
| 332 return NULL; | |
|
5444
a0952e4e52eb
mpatch: allow buffer objects for input
Matt Mackall <mpm@selenic.com>
parents:
4376
diff
changeset
|
333 if (PyObject_AsCharBuffer(tmp, &buffer, &blen)) |
|
a0952e4e52eb
mpatch: allow buffer objects for input
Matt Mackall <mpm@selenic.com>
parents:
4376
diff
changeset
|
334 return NULL; |
|
a0952e4e52eb
mpatch: allow buffer objects for input
Matt Mackall <mpm@selenic.com>
parents:
4376
diff
changeset
|
335 return decode(buffer, blen); |
| 72 | 336 } |
| 337 | |
| 338 /* divide and conquer, memory management is elsewhere */ | |
| 339 len = (end - start) / 2; | |
| 340 return combine(fold(bins, start, start + len), | |
| 341 fold(bins, start + len, end)); | |
| 342 } | |
| 343 | |
| 344 static PyObject * | |
| 345 patches(PyObject *self, PyObject *args) | |
| 346 { | |
| 347 PyObject *text, *bins, *result; | |
| 348 struct flist *patch; | |
|
5444
a0952e4e52eb
mpatch: allow buffer objects for input
Matt Mackall <mpm@selenic.com>
parents:
4376
diff
changeset
|
349 const char *in; |
|
a0952e4e52eb
mpatch: allow buffer objects for input
Matt Mackall <mpm@selenic.com>
parents:
4376
diff
changeset
|
350 char *out; |
| 72 | 351 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
|
352 Py_ssize_t inlen; |
| 72 | 353 |
|
5444
a0952e4e52eb
mpatch: allow buffer objects for input
Matt Mackall <mpm@selenic.com>
parents:
4376
diff
changeset
|
354 if (!PyArg_ParseTuple(args, "OO:mpatch", &text, &bins)) |
| 72 | 355 return NULL; |
| 356 | |
| 357 len = PyList_Size(bins); | |
| 358 if (!len) { | |
| 359 /* nothing to do */ | |
| 360 Py_INCREF(text); | |
| 361 return text; | |
| 362 } | |
| 363 | |
|
5444
a0952e4e52eb
mpatch: allow buffer objects for input
Matt Mackall <mpm@selenic.com>
parents:
4376
diff
changeset
|
364 if (PyObject_AsCharBuffer(text, &in, &inlen)) |
|
a0952e4e52eb
mpatch: allow buffer objects for input
Matt Mackall <mpm@selenic.com>
parents:
4376
diff
changeset
|
365 return NULL; |
|
a0952e4e52eb
mpatch: allow buffer objects for input
Matt Mackall <mpm@selenic.com>
parents:
4376
diff
changeset
|
366 |
| 72 | 367 patch = fold(bins, 0, len); |
| 128 | 368 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
|
369 return NULL; |
| 128 | 370 |
|
5444
a0952e4e52eb
mpatch: allow buffer objects for input
Matt Mackall <mpm@selenic.com>
parents:
4376
diff
changeset
|
371 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
|
372 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
|
373 result = NULL; |
|
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
374 goto cleanup; |
|
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
375 } |
| 72 | 376 result = PyString_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
|
377 if (!result) { |
|
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
378 result = NULL; |
|
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
379 goto cleanup; |
| 128 | 380 } |
|
1722
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
381 out = PyString_AsString(result); |
|
5444
a0952e4e52eb
mpatch: allow buffer objects for input
Matt Mackall <mpm@selenic.com>
parents:
4376
diff
changeset
|
382 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
|
383 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
|
384 result = NULL; |
|
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
385 } |
|
681c5c211b92
catch errors and throw exception with invalid binary patch data
Benoit Boissinot <benoit.boissinot@ens-lyon.org>
parents:
597
diff
changeset
|
386 cleanup: |
| 72 | 387 lfree(patch); |
| 388 return result; | |
| 389 } | |
| 390 | |
|
2078
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
391 /* calculate size of a patched file directly */ |
|
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
392 static PyObject * |
|
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
393 patchedsize(PyObject *self, PyObject *args) |
|
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
394 { |
|
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
395 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
|
396 int patchlen; |
|
4375
11dc22eb8e8d
Fix segfaults when parsing bdiff hunks in mpatch.decode() and .patchedsize()
Thomas Arendsen Hein <thomas@intevation.de>
parents:
3138
diff
changeset
|
397 char *bin, *binend, *data; |
|
2078
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
398 char decode[12]; /* for dealing with alignment issues */ |
|
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
399 |
|
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
400 if (!PyArg_ParseTuple(args, "ls#", &orig, &bin, &patchlen)) |
|
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
401 return NULL; |
|
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
402 |
|
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
403 binend = bin + patchlen; |
|
4375
11dc22eb8e8d
Fix segfaults when parsing bdiff hunks in mpatch.decode() and .patchedsize()
Thomas Arendsen Hein <thomas@intevation.de>
parents:
3138
diff
changeset
|
404 data = bin + 12; |
|
2078
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
405 |
|
4375
11dc22eb8e8d
Fix segfaults when parsing bdiff hunks in mpatch.decode() and .patchedsize()
Thomas Arendsen Hein <thomas@intevation.de>
parents:
3138
diff
changeset
|
406 while (data <= binend) { |
|
2078
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
407 memcpy(decode, bin, 12); |
|
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
408 start = ntohl(*(uint32_t *)decode); |
|
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
409 end = ntohl(*(uint32_t *)(decode + 4)); |
|
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
410 len = ntohl(*(uint32_t *)(decode + 8)); |
|
4375
11dc22eb8e8d
Fix segfaults when parsing bdiff hunks in mpatch.decode() and .patchedsize()
Thomas Arendsen Hein <thomas@intevation.de>
parents:
3138
diff
changeset
|
411 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
|
412 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
|
413 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
|
414 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
|
415 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
|
416 data = bin + 12; |
|
2078
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
417 outlen += start - last; |
|
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
418 last = end; |
|
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
419 outlen += len; |
|
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
420 } |
|
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
421 |
|
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
422 if (bin != binend) { |
|
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
423 if (!PyErr_Occurred()) |
|
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
424 PyErr_SetString(mpatch_Error, "patch cannot be decoded"); |
|
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
425 return NULL; |
|
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
426 } |
|
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
427 |
|
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
428 outlen += orig - last; |
|
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
429 return Py_BuildValue("l", outlen); |
|
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
430 } |
|
441ea218414e
Fill in the uncompressed size during revlog.addgroup
mason@suse.com
parents:
1978
diff
changeset
|
431 |
| 72 | 432 static PyMethodDef methods[] = { |
| 433 {"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
|
434 {"patchedsize", patchedsize, METH_VARARGS, "calculed patched size\n"}, |
| 72 | 435 {NULL, NULL} |
| 436 }; | |
| 437 | |
| 438 PyMODINIT_FUNC | |
| 439 initmpatch(void) | |
| 440 { | |
| 441 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
|
442 mpatch_Error = PyErr_NewException("mpatch.mpatchError", NULL, NULL); |
| 72 | 443 } |
| 444 |