Mercurial > hg > mercurial-crew-with-dirclash
annotate mercurial/mpatch.c @ 464:50da4bb9cab6
Merge from http://moffetthome.net:8012/
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Merge from http://moffetthome.net:8012/
manifest hash: 3a67864af6b0276eabe640274633ca2625dbd4a5
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.0 (GNU/Linux)
iD8DBQFCvQKkywK+sNU5EO8RArUHAKCSGtpEfJwYGoAIpj4mBDvcPted9wCgnzML
xr8WQ8DdPkJ9DVsLjvsbhJU=
=O7Fd
-----END PGP SIGNATURE-----
| author | mpm@selenic.com |
|---|---|
| date | Fri, 24 Jun 2005 23:07:16 -0800 |
| parents | 688d03d6997a ea93402b81b9 |
| children | aa3d592df9b9 |
| 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 | |
| 17 Copyright 2005 Matt Mackall <mpm@selenic.com> | |
| 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> | |
|
410
7c678976df3e
Make mpatch.c compilable under the other `OS'
mpm@selenic.com
parents:
384
diff
changeset
|
26 #ifdef _WIN32 |
|
7c678976df3e
Make mpatch.c compilable under the other `OS'
mpm@selenic.com
parents:
384
diff
changeset
|
27 |
|
7c678976df3e
Make mpatch.c compilable under the other `OS'
mpm@selenic.com
parents:
384
diff
changeset
|
28 typedef unsigned long uint32_t; |
|
7c678976df3e
Make mpatch.c compilable under the other `OS'
mpm@selenic.com
parents:
384
diff
changeset
|
29 |
|
411
9e9f7ab43ce2
Add 'other OS' bits to bdiff.c / style cleanups
mpm@selenic.com
parents:
410
diff
changeset
|
30 static uint32_t ntohl(uint32_t x) |
|
9e9f7ab43ce2
Add 'other OS' bits to bdiff.c / style cleanups
mpm@selenic.com
parents:
410
diff
changeset
|
31 { |
|
9e9f7ab43ce2
Add 'other OS' bits to bdiff.c / style cleanups
mpm@selenic.com
parents:
410
diff
changeset
|
32 return ((x & 0x000000ffUL) << 24) | |
|
9e9f7ab43ce2
Add 'other OS' bits to bdiff.c / style cleanups
mpm@selenic.com
parents:
410
diff
changeset
|
33 ((x & 0x0000ff00UL) << 8) | |
|
9e9f7ab43ce2
Add 'other OS' bits to bdiff.c / style cleanups
mpm@selenic.com
parents:
410
diff
changeset
|
34 ((x & 0x00ff0000UL) >> 8) | |
|
9e9f7ab43ce2
Add 'other OS' bits to bdiff.c / style cleanups
mpm@selenic.com
parents:
410
diff
changeset
|
35 ((x & 0xff000000UL) >> 24); |
|
410
7c678976df3e
Make mpatch.c compilable under the other `OS'
mpm@selenic.com
parents:
384
diff
changeset
|
36 } |
|
7c678976df3e
Make mpatch.c compilable under the other `OS'
mpm@selenic.com
parents:
384
diff
changeset
|
37 |
|
7c678976df3e
Make mpatch.c compilable under the other `OS'
mpm@selenic.com
parents:
384
diff
changeset
|
38 #else |
|
7c678976df3e
Make mpatch.c compilable under the other `OS'
mpm@selenic.com
parents:
384
diff
changeset
|
39 #include <netinet/in.h> |
|
7c678976df3e
Make mpatch.c compilable under the other `OS'
mpm@selenic.com
parents:
384
diff
changeset
|
40 #include <sys/types.h> |
|
463
ea93402b81b9
Added stdint.h include to fix build on Mac OS X Tiger [v10.4]
kyle@zeus.moffetthome.net
parents:
429
diff
changeset
|
41 #include <stdint.h> |
|
410
7c678976df3e
Make mpatch.c compilable under the other `OS'
mpm@selenic.com
parents:
384
diff
changeset
|
42 #endif |
| 72 | 43 |
| 44 static char mpatch_doc[] = "Efficient binary patching."; | |
| 45 | |
| 46 struct frag { | |
| 47 int start, end, len; | |
| 48 char *data; | |
| 49 }; | |
| 50 | |
| 51 struct flist { | |
| 52 struct frag *base, *head, *tail; | |
| 53 }; | |
| 54 | |
| 55 static struct flist *lalloc(int size) | |
| 56 { | |
| 128 | 57 struct flist *a = NULL; |
| 72 | 58 |
| 59 a = malloc(sizeof(struct flist)); | |
| 128 | 60 if (a) { |
| 61 a->base = malloc(sizeof(struct frag) * size); | |
| 282 | 62 if (!a->base) { |
| 128 | 63 free(a); |
| 282 | 64 a = NULL; |
| 65 } else | |
| 128 | 66 a->head = a->tail = a->base; |
| 67 } | |
| 72 | 68 return a; |
| 69 } | |
| 70 | |
| 71 static void lfree(struct flist *a) | |
| 72 { | |
| 128 | 73 if (a) { |
| 74 free(a->base); | |
| 75 free(a); | |
| 76 } | |
| 72 | 77 } |
| 78 | |
| 79 static int lsize(struct flist *a) | |
| 80 { | |
| 81 return a->tail - a->head; | |
| 82 } | |
| 83 | |
| 84 /* move hunks in source that are less cut to dest, compensating | |
| 85 for changes in offset. the last hunk may be split if necessary. | |
| 86 */ | |
| 87 static int gather(struct flist *dest, struct flist *src, int cut, int offset) | |
| 88 { | |
| 89 struct frag *d = dest->tail, *s = src->head; | |
| 90 int postend, c, l; | |
| 91 | |
| 92 while (s != src->tail) { | |
| 93 if (s->start + offset >= cut) | |
| 82 | 94 break; /* we've gone far enough */ |
| 72 | 95 |
| 96 postend = offset + s->start + s->len; | |
| 97 if (postend <= cut) { | |
| 98 /* save this hunk */ | |
| 99 offset += s->start + s->len - s->end; | |
| 100 *d++ = *s++; | |
| 101 } | |
| 102 else { | |
| 103 /* break up this hunk */ | |
| 104 c = cut - offset; | |
| 105 if (s->end < c) | |
| 106 c = s->end; | |
| 107 l = cut - offset - s->start; | |
| 108 if (s->len < l) | |
| 109 l = s->len; | |
| 110 | |
| 111 offset += s->start + l - c; | |
| 112 | |
| 113 d->start = s->start; | |
| 114 d->end = c; | |
| 115 d->len = l; | |
| 116 d->data = s->data; | |
| 117 d++; | |
| 118 s->start = c; | |
| 119 s->len = s->len - l; | |
| 120 s->data = s->data + l; | |
| 121 | |
| 82 | 122 break; |
| 72 | 123 } |
| 124 } | |
| 125 | |
| 126 dest->tail = d; | |
| 127 src->head = s; | |
| 128 return offset; | |
| 129 } | |
| 130 | |
| 131 /* like gather, but with no output list */ | |
| 132 static int discard(struct flist *src, int cut, int offset) | |
| 133 { | |
| 134 struct frag *s = src->head; | |
| 135 int postend, c, l; | |
| 136 | |
| 137 while (s != src->tail) { | |
| 138 if (s->start + offset >= cut) | |
| 82 | 139 break; |
| 72 | 140 |
| 141 postend = offset + s->start + s->len; | |
| 142 if (postend <= cut) { | |
| 143 offset += s->start + s->len - s->end; | |
| 144 s++; | |
| 145 } | |
| 146 else { | |
| 147 c = cut - offset; | |
| 148 if (s->end < c) | |
| 149 c = s->end; | |
| 150 l = cut - offset - s->start; | |
| 151 if (s->len < l) | |
| 152 l = s->len; | |
| 153 | |
| 154 offset += s->start + l - c; | |
| 155 s->start = c; | |
| 156 s->len = s->len - l; | |
| 157 s->data = s->data + l; | |
| 158 | |
| 82 | 159 break; |
| 72 | 160 } |
| 161 } | |
| 162 | |
| 163 src->head = s; | |
| 164 return offset; | |
| 165 } | |
| 166 | |
| 167 /* combine hunk lists a and b, while adjusting b for offset changes in a/ | |
| 168 this deletes a and b and returns the resultant list. */ | |
| 169 static struct flist *combine(struct flist *a, struct flist *b) | |
| 170 { | |
| 128 | 171 struct flist *c = NULL; |
| 172 struct frag *bh, *ct; | |
| 72 | 173 int offset = 0, post; |
| 174 | |
| 128 | 175 if (a && b) |
| 176 c = lalloc((lsize(a) + lsize(b)) * 2); | |
| 177 | |
| 178 if (c) { | |
| 72 | 179 |
| 128 | 180 for (bh = b->head; bh != b->tail; bh++) { |
| 181 /* save old hunks */ | |
| 182 offset = gather(c, a, bh->start, offset); | |
| 72 | 183 |
| 128 | 184 /* discard replaced hunks */ |
| 185 post = discard(a, bh->end, offset); | |
| 72 | 186 |
| 128 | 187 /* insert new hunk */ |
| 188 ct = c->tail; | |
| 189 ct->start = bh->start - offset; | |
| 190 ct->end = bh->end - post; | |
| 191 ct->len = bh->len; | |
| 192 ct->data = bh->data; | |
| 193 c->tail++; | |
| 194 offset = post; | |
| 195 } | |
| 196 | |
| 197 /* hold on to tail from a */ | |
| 198 memcpy(c->tail, a->head, sizeof(struct frag) * lsize(a)); | |
| 199 c->tail += lsize(a); | |
| 72 | 200 } |
| 201 | |
| 202 lfree(a); | |
| 203 lfree(b); | |
| 204 return c; | |
| 205 } | |
| 206 | |
| 207 /* decode a binary patch into a hunk list */ | |
| 208 static struct flist *decode(char *bin, int len) | |
| 209 { | |
| 210 struct flist *l; | |
| 211 struct frag *lt; | |
| 212 char *end = bin + len; | |
|
384
a29decbf7475
mpatch: attempt to handle unpack alignment issues on Solaris
mpm@selenic.com
parents:
282
diff
changeset
|
213 char decode[12]; /* for dealing with alignment issues */ |
| 72 | 214 |
| 215 /* assume worst case size, we won't have many of these lists */ | |
| 216 l = lalloc(len / 12); | |
| 217 lt = l->tail; | |
| 218 | |
| 219 while (bin < end) { | |
|
384
a29decbf7475
mpatch: attempt to handle unpack alignment issues on Solaris
mpm@selenic.com
parents:
282
diff
changeset
|
220 memcpy(decode, bin, 12); |
|
a29decbf7475
mpatch: attempt to handle unpack alignment issues on Solaris
mpm@selenic.com
parents:
282
diff
changeset
|
221 lt->start = ntohl(*(uint32_t *)decode); |
|
a29decbf7475
mpatch: attempt to handle unpack alignment issues on Solaris
mpm@selenic.com
parents:
282
diff
changeset
|
222 lt->end = ntohl(*(uint32_t *)(decode + 4)); |
|
a29decbf7475
mpatch: attempt to handle unpack alignment issues on Solaris
mpm@selenic.com
parents:
282
diff
changeset
|
223 lt->len = ntohl(*(uint32_t *)(decode + 8)); |
| 72 | 224 lt->data = bin + 12; |
| 225 bin += 12 + lt->len; | |
| 226 lt++; | |
| 227 } | |
| 228 | |
| 229 l->tail = lt; | |
| 230 return l; | |
| 231 } | |
| 232 | |
| 233 /* calculate the size of resultant text */ | |
| 234 static int calcsize(int len, struct flist *l) | |
| 235 { | |
| 236 int outlen = 0, last = 0; | |
| 237 struct frag *f = l->head; | |
| 238 | |
| 239 while (f != l->tail) { | |
| 240 outlen += f->start - last; | |
| 241 last = f->end; | |
| 242 outlen += f->len; | |
| 243 f++; | |
| 244 } | |
| 245 | |
| 246 outlen += len - last; | |
| 247 return outlen; | |
| 248 } | |
| 249 | |
| 250 static void apply(char *buf, char *orig, int len, struct flist *l) | |
| 251 { | |
| 252 struct frag *f = l->head; | |
| 253 int last = 0; | |
| 254 char *p = buf; | |
| 255 | |
| 256 while (f != l->tail) { | |
| 257 memcpy(p, orig + last, f->start - last); | |
| 258 p += f->start - last; | |
| 259 memcpy(p, f->data, f->len); | |
| 260 last = f->end; | |
| 261 p += f->len; | |
| 262 f++; | |
| 263 } | |
| 264 memcpy(p, orig + last, len - last); | |
| 265 } | |
| 266 | |
| 267 /* recursively generate a patch of all bins between start and end */ | |
| 268 static struct flist *fold(PyObject *bins, int start, int end) | |
| 269 { | |
| 270 int len; | |
| 271 | |
| 272 if (start + 1 == end) { | |
| 273 /* trivial case, output a decoded list */ | |
| 274 PyObject *tmp = PyList_GetItem(bins, start); | |
| 128 | 275 if (!tmp) |
| 276 return NULL; | |
| 72 | 277 return decode(PyString_AsString(tmp), PyString_Size(tmp)); |
| 278 } | |
| 279 | |
| 280 /* divide and conquer, memory management is elsewhere */ | |
| 281 len = (end - start) / 2; | |
| 282 return combine(fold(bins, start, start + len), | |
| 283 fold(bins, start + len, end)); | |
| 284 } | |
| 285 | |
| 286 static PyObject * | |
| 287 patches(PyObject *self, PyObject *args) | |
| 288 { | |
| 289 PyObject *text, *bins, *result; | |
| 290 struct flist *patch; | |
| 291 char *in, *out; | |
| 292 int len, outlen; | |
| 293 | |
| 128 | 294 if (!PyArg_ParseTuple(args, "SO:mpatch", &text, &bins)) |
| 72 | 295 return NULL; |
| 296 | |
| 297 len = PyList_Size(bins); | |
| 298 if (!len) { | |
| 299 /* nothing to do */ | |
| 300 Py_INCREF(text); | |
| 301 return text; | |
| 302 } | |
| 303 | |
| 304 patch = fold(bins, 0, len); | |
| 128 | 305 if (!patch) |
| 306 return PyErr_NoMemory(); | |
| 307 | |
| 72 | 308 outlen = calcsize(PyString_Size(text), patch); |
| 309 result = PyString_FromStringAndSize(NULL, outlen); | |
| 128 | 310 if (result) { |
| 311 in = PyString_AsString(text); | |
| 312 out = PyString_AsString(result); | |
| 313 apply(out, in, PyString_Size(text), patch); | |
| 314 } | |
| 315 | |
| 72 | 316 lfree(patch); |
| 317 return result; | |
| 318 } | |
| 319 | |
| 320 static PyMethodDef methods[] = { | |
| 321 {"patches", patches, METH_VARARGS, "apply a series of patches\n"}, | |
| 322 {NULL, NULL} | |
| 323 }; | |
| 324 | |
| 325 PyMODINIT_FUNC | |
| 326 initmpatch(void) | |
| 327 { | |
| 328 Py_InitModule3("mpatch", methods, mpatch_doc); | |
| 329 } | |
| 330 |