Mercurial > public > mercurial-scm > hg
annotate mercurial/mpatch.c @ 411:9e9f7ab43ce2
Add 'other OS' bits to bdiff.c / style cleanups
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Add 'other OS' bits to bdiff.c / style cleanups
manifest hash: 9a0ef60a59af0594f95a2b22de243fe8ef059762
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.0 (GNU/Linux)
iD8DBQFCuNL0ywK+sNU5EO8RAg+yAKCe8dG411HdZZvsxB25AtfBBApQBQCgnb7O
MGO0xvEBgAtt+3F+VJBkiU4=
=F7r7
-----END PGP SIGNATURE-----
| author | mpm@selenic.com |
|---|---|
| date | Tue, 21 Jun 2005 18:54:44 -0800 |
| parents | 7c678976df3e |
| children | 688d03d6997a |
| 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> |
|
7c678976df3e
Make mpatch.c compilable under the other `OS'
mpm@selenic.com
parents:
384
diff
changeset
|
41 #endif |
| 72 | 42 |
| 43 static char mpatch_doc[] = "Efficient binary patching."; | |
| 44 | |
| 45 struct frag { | |
| 46 int start, end, len; | |
| 47 char *data; | |
| 48 }; | |
| 49 | |
| 50 struct flist { | |
| 51 struct frag *base, *head, *tail; | |
| 52 }; | |
| 53 | |
| 54 static struct flist *lalloc(int size) | |
| 55 { | |
| 128 | 56 struct flist *a = NULL; |
| 72 | 57 |
| 58 a = malloc(sizeof(struct flist)); | |
| 128 | 59 if (a) { |
| 60 a->base = malloc(sizeof(struct frag) * size); | |
| 282 | 61 if (!a->base) { |
| 128 | 62 free(a); |
| 282 | 63 a = NULL; |
| 64 } else | |
| 128 | 65 a->head = a->tail = a->base; |
| 66 } | |
| 72 | 67 return a; |
| 68 } | |
| 69 | |
| 70 static void lfree(struct flist *a) | |
| 71 { | |
| 128 | 72 if (a) { |
| 73 free(a->base); | |
| 74 free(a); | |
| 75 } | |
| 72 | 76 } |
| 77 | |
| 78 static int lsize(struct flist *a) | |
| 79 { | |
| 80 return a->tail - a->head; | |
| 81 } | |
| 82 | |
| 83 /* move hunks in source that are less cut to dest, compensating | |
| 84 for changes in offset. the last hunk may be split if necessary. | |
| 85 */ | |
| 86 static int gather(struct flist *dest, struct flist *src, int cut, int offset) | |
| 87 { | |
| 88 struct frag *d = dest->tail, *s = src->head; | |
| 89 int postend, c, l; | |
| 90 | |
| 91 while (s != src->tail) { | |
| 92 if (s->start + offset >= cut) | |
| 82 | 93 break; /* we've gone far enough */ |
| 72 | 94 |
| 95 postend = offset + s->start + s->len; | |
| 96 if (postend <= cut) { | |
| 97 /* save this hunk */ | |
| 98 offset += s->start + s->len - s->end; | |
| 99 *d++ = *s++; | |
| 100 } | |
| 101 else { | |
| 102 /* break up this hunk */ | |
| 103 c = cut - offset; | |
| 104 if (s->end < c) | |
| 105 c = s->end; | |
| 106 l = cut - offset - s->start; | |
| 107 if (s->len < l) | |
| 108 l = s->len; | |
| 109 | |
| 110 offset += s->start + l - c; | |
| 111 | |
| 112 d->start = s->start; | |
| 113 d->end = c; | |
| 114 d->len = l; | |
| 115 d->data = s->data; | |
| 116 d++; | |
| 117 s->start = c; | |
| 118 s->len = s->len - l; | |
| 119 s->data = s->data + l; | |
| 120 | |
| 82 | 121 break; |
| 72 | 122 } |
| 123 } | |
| 124 | |
| 125 dest->tail = d; | |
| 126 src->head = s; | |
| 127 return offset; | |
| 128 } | |
| 129 | |
| 130 /* like gather, but with no output list */ | |
| 131 static int discard(struct flist *src, int cut, int offset) | |
| 132 { | |
| 133 struct frag *s = src->head; | |
| 134 int postend, c, l; | |
| 135 | |
| 136 while (s != src->tail) { | |
| 137 if (s->start + offset >= cut) | |
| 82 | 138 break; |
| 72 | 139 |
| 140 postend = offset + s->start + s->len; | |
| 141 if (postend <= cut) { | |
| 142 offset += s->start + s->len - s->end; | |
| 143 s++; | |
| 144 } | |
| 145 else { | |
| 146 c = cut - offset; | |
| 147 if (s->end < c) | |
| 148 c = s->end; | |
| 149 l = cut - offset - s->start; | |
| 150 if (s->len < l) | |
| 151 l = s->len; | |
| 152 | |
| 153 offset += s->start + l - c; | |
| 154 s->start = c; | |
| 155 s->len = s->len - l; | |
| 156 s->data = s->data + l; | |
| 157 | |
| 82 | 158 break; |
| 72 | 159 } |
| 160 } | |
| 161 | |
| 162 src->head = s; | |
| 163 return offset; | |
| 164 } | |
| 165 | |
| 166 /* combine hunk lists a and b, while adjusting b for offset changes in a/ | |
| 167 this deletes a and b and returns the resultant list. */ | |
| 168 static struct flist *combine(struct flist *a, struct flist *b) | |
| 169 { | |
| 128 | 170 struct flist *c = NULL; |
| 171 struct frag *bh, *ct; | |
| 72 | 172 int offset = 0, post; |
| 173 | |
| 128 | 174 if (a && b) |
| 175 c = lalloc((lsize(a) + lsize(b)) * 2); | |
| 176 | |
| 177 if (c) { | |
| 72 | 178 |
| 128 | 179 for (bh = b->head; bh != b->tail; bh++) { |
| 180 /* save old hunks */ | |
| 181 offset = gather(c, a, bh->start, offset); | |
| 72 | 182 |
| 128 | 183 /* discard replaced hunks */ |
| 184 post = discard(a, bh->end, offset); | |
| 72 | 185 |
| 128 | 186 /* insert new hunk */ |
| 187 ct = c->tail; | |
| 188 ct->start = bh->start - offset; | |
| 189 ct->end = bh->end - post; | |
| 190 ct->len = bh->len; | |
| 191 ct->data = bh->data; | |
| 192 c->tail++; | |
| 193 offset = post; | |
| 194 } | |
| 195 | |
| 196 /* hold on to tail from a */ | |
| 197 memcpy(c->tail, a->head, sizeof(struct frag) * lsize(a)); | |
| 198 c->tail += lsize(a); | |
| 72 | 199 } |
| 200 | |
| 201 lfree(a); | |
| 202 lfree(b); | |
| 203 return c; | |
| 204 } | |
| 205 | |
| 206 /* decode a binary patch into a hunk list */ | |
| 207 static struct flist *decode(char *bin, int len) | |
| 208 { | |
| 209 struct flist *l; | |
| 210 struct frag *lt; | |
| 211 char *end = bin + len; | |
|
384
a29decbf7475
mpatch: attempt to handle unpack alignment issues on Solaris
mpm@selenic.com
parents:
282
diff
changeset
|
212 char decode[12]; /* for dealing with alignment issues */ |
| 72 | 213 |
| 214 /* assume worst case size, we won't have many of these lists */ | |
| 215 l = lalloc(len / 12); | |
| 216 lt = l->tail; | |
| 217 | |
| 218 while (bin < end) { | |
|
384
a29decbf7475
mpatch: attempt to handle unpack alignment issues on Solaris
mpm@selenic.com
parents:
282
diff
changeset
|
219 memcpy(decode, bin, 12); |
|
a29decbf7475
mpatch: attempt to handle unpack alignment issues on Solaris
mpm@selenic.com
parents:
282
diff
changeset
|
220 lt->start = ntohl(*(uint32_t *)decode); |
|
a29decbf7475
mpatch: attempt to handle unpack alignment issues on Solaris
mpm@selenic.com
parents:
282
diff
changeset
|
221 lt->end = ntohl(*(uint32_t *)(decode + 4)); |
|
a29decbf7475
mpatch: attempt to handle unpack alignment issues on Solaris
mpm@selenic.com
parents:
282
diff
changeset
|
222 lt->len = ntohl(*(uint32_t *)(decode + 8)); |
| 72 | 223 lt->data = bin + 12; |
| 224 bin += 12 + lt->len; | |
| 225 lt++; | |
| 226 } | |
| 227 | |
| 228 l->tail = lt; | |
| 229 return l; | |
| 230 } | |
| 231 | |
| 232 /* calculate the size of resultant text */ | |
| 233 static int calcsize(int len, struct flist *l) | |
| 234 { | |
| 235 int outlen = 0, last = 0; | |
| 236 struct frag *f = l->head; | |
| 237 | |
| 238 while (f != l->tail) { | |
| 239 outlen += f->start - last; | |
| 240 last = f->end; | |
| 241 outlen += f->len; | |
| 242 f++; | |
| 243 } | |
| 244 | |
| 245 outlen += len - last; | |
| 246 return outlen; | |
| 247 } | |
| 248 | |
| 249 static void apply(char *buf, char *orig, int len, struct flist *l) | |
| 250 { | |
| 251 struct frag *f = l->head; | |
| 252 int last = 0; | |
| 253 char *p = buf; | |
| 254 | |
| 255 while (f != l->tail) { | |
| 256 memcpy(p, orig + last, f->start - last); | |
| 257 p += f->start - last; | |
| 258 memcpy(p, f->data, f->len); | |
| 259 last = f->end; | |
| 260 p += f->len; | |
| 261 f++; | |
| 262 } | |
| 263 memcpy(p, orig + last, len - last); | |
| 264 } | |
| 265 | |
| 266 /* recursively generate a patch of all bins between start and end */ | |
| 267 static struct flist *fold(PyObject *bins, int start, int end) | |
| 268 { | |
| 269 int len; | |
| 270 | |
| 271 if (start + 1 == end) { | |
| 272 /* trivial case, output a decoded list */ | |
| 273 PyObject *tmp = PyList_GetItem(bins, start); | |
| 128 | 274 if (!tmp) |
| 275 return NULL; | |
| 72 | 276 return decode(PyString_AsString(tmp), PyString_Size(tmp)); |
| 277 } | |
| 278 | |
| 279 /* divide and conquer, memory management is elsewhere */ | |
| 280 len = (end - start) / 2; | |
| 281 return combine(fold(bins, start, start + len), | |
| 282 fold(bins, start + len, end)); | |
| 283 } | |
| 284 | |
| 285 static PyObject * | |
| 286 patches(PyObject *self, PyObject *args) | |
| 287 { | |
| 288 PyObject *text, *bins, *result; | |
| 289 struct flist *patch; | |
| 290 char *in, *out; | |
| 291 int len, outlen; | |
| 292 | |
| 128 | 293 if (!PyArg_ParseTuple(args, "SO:mpatch", &text, &bins)) |
| 72 | 294 return NULL; |
| 295 | |
| 296 len = PyList_Size(bins); | |
| 297 if (!len) { | |
| 298 /* nothing to do */ | |
| 299 Py_INCREF(text); | |
| 300 return text; | |
| 301 } | |
| 302 | |
| 303 patch = fold(bins, 0, len); | |
| 128 | 304 if (!patch) |
| 305 return PyErr_NoMemory(); | |
| 306 | |
| 72 | 307 outlen = calcsize(PyString_Size(text), patch); |
| 308 result = PyString_FromStringAndSize(NULL, outlen); | |
| 128 | 309 if (result) { |
| 310 in = PyString_AsString(text); | |
| 311 out = PyString_AsString(result); | |
| 312 apply(out, in, PyString_Size(text), patch); | |
| 313 } | |
| 314 | |
| 72 | 315 lfree(patch); |
| 316 return result; | |
| 317 } | |
| 318 | |
| 319 static PyMethodDef methods[] = { | |
| 320 {"patches", patches, METH_VARARGS, "apply a series of patches\n"}, | |
| 321 {NULL, NULL} | |
| 322 }; | |
| 323 | |
| 324 PyMODINIT_FUNC | |
| 325 initmpatch(void) | |
| 326 { | |
| 327 Py_InitModule3("mpatch", methods, mpatch_doc); | |
| 328 } | |
| 329 |