Mercurial > public > mercurial-scm > hg
annotate mercurial/mpatch.c @ 495:e94cebc60d96
Pull from TAH
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Pull from TAH
manifest hash: 08b55e07198b8a4272753895fe3727d9f52ebb75
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.0 (GNU/Linux)
iD8DBQFCwO5YywK+sNU5EO8RArUtAJ9gsYYNbZH22COOz9B7ppwgCDxM8ACfYWcl
n9Dim3Z7qepGUXvRDEeFZ9c=
=JMQx
-----END PGP SIGNATURE-----
| author | mpm@selenic.com |
|---|---|
| date | Mon, 27 Jun 2005 22:29:44 -0800 |
| parents | 098d1f039c18 aa3d592df9b9 |
| children | 7f3fc8fd427e |
| 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 |
|
411
9e9f7ab43ce2
Add 'other OS' bits to bdiff.c / style cleanups
mpm@selenic.com
parents:
410
diff
changeset
|
27 static uint32_t ntohl(uint32_t x) |
|
9e9f7ab43ce2
Add 'other OS' bits to bdiff.c / style cleanups
mpm@selenic.com
parents:
410
diff
changeset
|
28 { |
|
9e9f7ab43ce2
Add 'other OS' bits to bdiff.c / style cleanups
mpm@selenic.com
parents:
410
diff
changeset
|
29 return ((x & 0x000000ffUL) << 24) | |
|
9e9f7ab43ce2
Add 'other OS' bits to bdiff.c / style cleanups
mpm@selenic.com
parents:
410
diff
changeset
|
30 ((x & 0x0000ff00UL) << 8) | |
|
9e9f7ab43ce2
Add 'other OS' bits to bdiff.c / style cleanups
mpm@selenic.com
parents:
410
diff
changeset
|
31 ((x & 0x00ff0000UL) >> 8) | |
|
9e9f7ab43ce2
Add 'other OS' bits to bdiff.c / style cleanups
mpm@selenic.com
parents:
410
diff
changeset
|
32 ((x & 0xff000000UL) >> 24); |
|
410
7c678976df3e
Make mpatch.c compilable under the other `OS'
mpm@selenic.com
parents:
384
diff
changeset
|
33 } |
|
7c678976df3e
Make mpatch.c compilable under the other `OS'
mpm@selenic.com
parents:
384
diff
changeset
|
34 #else |
| 472 | 35 #include <netinet/in.h> |
|
410
7c678976df3e
Make mpatch.c compilable under the other `OS'
mpm@selenic.com
parents:
384
diff
changeset
|
36 #endif |
| 72 | 37 |
| 38 static char mpatch_doc[] = "Efficient binary patching."; | |
| 39 | |
| 40 struct frag { | |
| 41 int start, end, len; | |
| 42 char *data; | |
| 43 }; | |
| 44 | |
| 45 struct flist { | |
| 46 struct frag *base, *head, *tail; | |
| 47 }; | |
| 48 | |
| 49 static struct flist *lalloc(int size) | |
| 50 { | |
| 128 | 51 struct flist *a = NULL; |
| 72 | 52 |
| 53 a = malloc(sizeof(struct flist)); | |
| 128 | 54 if (a) { |
| 55 a->base = malloc(sizeof(struct frag) * size); | |
| 282 | 56 if (!a->base) { |
| 128 | 57 free(a); |
| 282 | 58 a = NULL; |
| 59 } else | |
| 128 | 60 a->head = a->tail = a->base; |
| 61 } | |
| 72 | 62 return a; |
| 63 } | |
| 64 | |
| 65 static void lfree(struct flist *a) | |
| 66 { | |
| 128 | 67 if (a) { |
| 68 free(a->base); | |
| 69 free(a); | |
| 70 } | |
| 72 | 71 } |
| 72 | |
| 73 static int lsize(struct flist *a) | |
| 74 { | |
| 75 return a->tail - a->head; | |
| 76 } | |
| 77 | |
| 78 /* move hunks in source that are less cut to dest, compensating | |
| 79 for changes in offset. the last hunk may be split if necessary. | |
| 80 */ | |
| 81 static int gather(struct flist *dest, struct flist *src, int cut, int offset) | |
| 82 { | |
| 83 struct frag *d = dest->tail, *s = src->head; | |
| 84 int postend, c, l; | |
| 85 | |
| 86 while (s != src->tail) { | |
| 87 if (s->start + offset >= cut) | |
| 82 | 88 break; /* we've gone far enough */ |
| 72 | 89 |
| 90 postend = offset + s->start + s->len; | |
| 91 if (postend <= cut) { | |
| 92 /* save this hunk */ | |
| 93 offset += s->start + s->len - s->end; | |
| 94 *d++ = *s++; | |
| 95 } | |
| 96 else { | |
| 97 /* break up this hunk */ | |
| 98 c = cut - offset; | |
| 99 if (s->end < c) | |
| 100 c = s->end; | |
| 101 l = cut - offset - s->start; | |
| 102 if (s->len < l) | |
| 103 l = s->len; | |
| 104 | |
| 105 offset += s->start + l - c; | |
| 106 | |
| 107 d->start = s->start; | |
| 108 d->end = c; | |
| 109 d->len = l; | |
| 110 d->data = s->data; | |
| 111 d++; | |
| 112 s->start = c; | |
| 113 s->len = s->len - l; | |
| 114 s->data = s->data + l; | |
| 115 | |
| 82 | 116 break; |
| 72 | 117 } |
| 118 } | |
| 119 | |
| 120 dest->tail = d; | |
| 121 src->head = s; | |
| 122 return offset; | |
| 123 } | |
| 124 | |
| 125 /* like gather, but with no output list */ | |
| 126 static int discard(struct flist *src, int cut, int offset) | |
| 127 { | |
| 128 struct frag *s = src->head; | |
| 129 int postend, c, l; | |
| 130 | |
| 131 while (s != src->tail) { | |
| 132 if (s->start + offset >= cut) | |
| 82 | 133 break; |
| 72 | 134 |
| 135 postend = offset + s->start + s->len; | |
| 136 if (postend <= cut) { | |
| 137 offset += s->start + s->len - s->end; | |
| 138 s++; | |
| 139 } | |
| 140 else { | |
| 141 c = cut - offset; | |
| 142 if (s->end < c) | |
| 143 c = s->end; | |
| 144 l = cut - offset - s->start; | |
| 145 if (s->len < l) | |
| 146 l = s->len; | |
| 147 | |
| 148 offset += s->start + l - c; | |
| 149 s->start = c; | |
| 150 s->len = s->len - l; | |
| 151 s->data = s->data + l; | |
| 152 | |
| 82 | 153 break; |
| 72 | 154 } |
| 155 } | |
| 156 | |
| 157 src->head = s; | |
| 158 return offset; | |
| 159 } | |
| 160 | |
| 161 /* combine hunk lists a and b, while adjusting b for offset changes in a/ | |
| 162 this deletes a and b and returns the resultant list. */ | |
| 163 static struct flist *combine(struct flist *a, struct flist *b) | |
| 164 { | |
| 128 | 165 struct flist *c = NULL; |
| 166 struct frag *bh, *ct; | |
| 72 | 167 int offset = 0, post; |
| 168 | |
| 128 | 169 if (a && b) |
| 170 c = lalloc((lsize(a) + lsize(b)) * 2); | |
| 171 | |
| 172 if (c) { | |
| 72 | 173 |
| 128 | 174 for (bh = b->head; bh != b->tail; bh++) { |
| 175 /* save old hunks */ | |
| 176 offset = gather(c, a, bh->start, offset); | |
| 72 | 177 |
| 128 | 178 /* discard replaced hunks */ |
| 179 post = discard(a, bh->end, offset); | |
| 72 | 180 |
| 128 | 181 /* insert new hunk */ |
| 182 ct = c->tail; | |
| 183 ct->start = bh->start - offset; | |
| 184 ct->end = bh->end - post; | |
| 185 ct->len = bh->len; | |
| 186 ct->data = bh->data; | |
| 187 c->tail++; | |
| 188 offset = post; | |
| 189 } | |
| 190 | |
| 191 /* hold on to tail from a */ | |
| 192 memcpy(c->tail, a->head, sizeof(struct frag) * lsize(a)); | |
| 193 c->tail += lsize(a); | |
| 72 | 194 } |
| 195 | |
| 196 lfree(a); | |
| 197 lfree(b); | |
| 198 return c; | |
| 199 } | |
| 200 | |
| 201 /* decode a binary patch into a hunk list */ | |
| 202 static struct flist *decode(char *bin, int len) | |
| 203 { | |
| 204 struct flist *l; | |
| 205 struct frag *lt; | |
| 206 char *end = bin + len; | |
|
384
a29decbf7475
mpatch: attempt to handle unpack alignment issues on Solaris
mpm@selenic.com
parents:
282
diff
changeset
|
207 char decode[12]; /* for dealing with alignment issues */ |
| 72 | 208 |
| 209 /* assume worst case size, we won't have many of these lists */ | |
| 210 l = lalloc(len / 12); | |
| 211 lt = l->tail; | |
| 212 | |
| 213 while (bin < end) { | |
|
384
a29decbf7475
mpatch: attempt to handle unpack alignment issues on Solaris
mpm@selenic.com
parents:
282
diff
changeset
|
214 memcpy(decode, bin, 12); |
|
a29decbf7475
mpatch: attempt to handle unpack alignment issues on Solaris
mpm@selenic.com
parents:
282
diff
changeset
|
215 lt->start = ntohl(*(uint32_t *)decode); |
|
a29decbf7475
mpatch: attempt to handle unpack alignment issues on Solaris
mpm@selenic.com
parents:
282
diff
changeset
|
216 lt->end = ntohl(*(uint32_t *)(decode + 4)); |
|
a29decbf7475
mpatch: attempt to handle unpack alignment issues on Solaris
mpm@selenic.com
parents:
282
diff
changeset
|
217 lt->len = ntohl(*(uint32_t *)(decode + 8)); |
| 72 | 218 lt->data = bin + 12; |
| 219 bin += 12 + lt->len; | |
| 220 lt++; | |
| 221 } | |
| 222 | |
| 223 l->tail = lt; | |
| 224 return l; | |
| 225 } | |
| 226 | |
| 227 /* calculate the size of resultant text */ | |
| 228 static int calcsize(int len, struct flist *l) | |
| 229 { | |
| 230 int outlen = 0, last = 0; | |
| 231 struct frag *f = l->head; | |
| 232 | |
| 233 while (f != l->tail) { | |
| 234 outlen += f->start - last; | |
| 235 last = f->end; | |
| 236 outlen += f->len; | |
| 237 f++; | |
| 238 } | |
| 239 | |
| 240 outlen += len - last; | |
| 241 return outlen; | |
| 242 } | |
| 243 | |
| 244 static void apply(char *buf, char *orig, int len, struct flist *l) | |
| 245 { | |
| 246 struct frag *f = l->head; | |
| 247 int last = 0; | |
| 248 char *p = buf; | |
| 249 | |
| 250 while (f != l->tail) { | |
| 251 memcpy(p, orig + last, f->start - last); | |
| 252 p += f->start - last; | |
| 253 memcpy(p, f->data, f->len); | |
| 254 last = f->end; | |
| 255 p += f->len; | |
| 256 f++; | |
| 257 } | |
| 258 memcpy(p, orig + last, len - last); | |
| 259 } | |
| 260 | |
| 261 /* recursively generate a patch of all bins between start and end */ | |
| 262 static struct flist *fold(PyObject *bins, int start, int end) | |
| 263 { | |
| 264 int len; | |
| 265 | |
| 266 if (start + 1 == end) { | |
| 267 /* trivial case, output a decoded list */ | |
| 268 PyObject *tmp = PyList_GetItem(bins, start); | |
| 128 | 269 if (!tmp) |
| 270 return NULL; | |
| 72 | 271 return decode(PyString_AsString(tmp), PyString_Size(tmp)); |
| 272 } | |
| 273 | |
| 274 /* divide and conquer, memory management is elsewhere */ | |
| 275 len = (end - start) / 2; | |
| 276 return combine(fold(bins, start, start + len), | |
| 277 fold(bins, start + len, end)); | |
| 278 } | |
| 279 | |
| 280 static PyObject * | |
| 281 patches(PyObject *self, PyObject *args) | |
| 282 { | |
| 283 PyObject *text, *bins, *result; | |
| 284 struct flist *patch; | |
| 285 char *in, *out; | |
| 286 int len, outlen; | |
| 287 | |
| 128 | 288 if (!PyArg_ParseTuple(args, "SO:mpatch", &text, &bins)) |
| 72 | 289 return NULL; |
| 290 | |
| 291 len = PyList_Size(bins); | |
| 292 if (!len) { | |
| 293 /* nothing to do */ | |
| 294 Py_INCREF(text); | |
| 295 return text; | |
| 296 } | |
| 297 | |
| 298 patch = fold(bins, 0, len); | |
| 128 | 299 if (!patch) |
| 300 return PyErr_NoMemory(); | |
| 301 | |
| 72 | 302 outlen = calcsize(PyString_Size(text), patch); |
| 303 result = PyString_FromStringAndSize(NULL, outlen); | |
| 128 | 304 if (result) { |
| 305 in = PyString_AsString(text); | |
| 306 out = PyString_AsString(result); | |
| 307 apply(out, in, PyString_Size(text), patch); | |
| 308 } | |
| 309 | |
| 72 | 310 lfree(patch); |
| 311 return result; | |
| 312 } | |
| 313 | |
| 314 static PyMethodDef methods[] = { | |
| 315 {"patches", patches, METH_VARARGS, "apply a series of patches\n"}, | |
| 316 {NULL, NULL} | |
| 317 }; | |
| 318 | |
| 319 PyMODINIT_FUNC | |
| 320 initmpatch(void) | |
| 321 { | |
| 322 Py_InitModule3("mpatch", methods, mpatch_doc); | |
| 323 } | |
| 324 |