|
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>
|
|
|
26 #include <netinet/in.h>
|
|
|
27 #include <sys/types.h>
|
|
|
28
|
|
|
29 static char mpatch_doc[] = "Efficient binary patching.";
|
|
|
30
|
|
|
31 struct frag {
|
|
|
32 int start, end, len;
|
|
|
33 char *data;
|
|
|
34 };
|
|
|
35
|
|
|
36 struct flist {
|
|
|
37 struct frag *base, *head, *tail;
|
|
|
38 };
|
|
|
39
|
|
|
40 static struct flist *lalloc(int size)
|
|
|
41 {
|
|
|
42 struct flist *a;
|
|
|
43
|
|
|
44 a = malloc(sizeof(struct flist));
|
|
|
45 a->head = a->tail = a->base = malloc(sizeof(struct frag) * size);
|
|
|
46 return a;
|
|
|
47 }
|
|
|
48
|
|
|
49 static void lfree(struct flist *a)
|
|
|
50 {
|
|
|
51 free(a->base);
|
|
|
52 free(a);
|
|
|
53 }
|
|
|
54
|
|
|
55 static int lsize(struct flist *a)
|
|
|
56 {
|
|
|
57 return a->tail - a->head;
|
|
|
58 }
|
|
|
59
|
|
|
60 /* move hunks in source that are less cut to dest, compensating
|
|
|
61 for changes in offset. the last hunk may be split if necessary.
|
|
|
62 */
|
|
|
63 static int gather(struct flist *dest, struct flist *src, int cut, int offset)
|
|
|
64 {
|
|
|
65 struct frag *d = dest->tail, *s = src->head;
|
|
|
66 int postend, c, l;
|
|
|
67
|
|
|
68 while (s != src->tail) {
|
|
|
69 if (s->start + offset >= cut)
|
|
82
|
70 break; /* we've gone far enough */
|
|
72
|
71
|
|
|
72 postend = offset + s->start + s->len;
|
|
|
73 if (postend <= cut) {
|
|
|
74 /* save this hunk */
|
|
|
75 offset += s->start + s->len - s->end;
|
|
|
76 *d++ = *s++;
|
|
|
77 }
|
|
|
78 else {
|
|
|
79 /* break up this hunk */
|
|
|
80 c = cut - offset;
|
|
|
81 if (s->end < c)
|
|
|
82 c = s->end;
|
|
|
83 l = cut - offset - s->start;
|
|
|
84 if (s->len < l)
|
|
|
85 l = s->len;
|
|
|
86
|
|
|
87 offset += s->start + l - c;
|
|
|
88
|
|
|
89 d->start = s->start;
|
|
|
90 d->end = c;
|
|
|
91 d->len = l;
|
|
|
92 d->data = s->data;
|
|
|
93 d++;
|
|
|
94 s->start = c;
|
|
|
95 s->len = s->len - l;
|
|
|
96 s->data = s->data + l;
|
|
|
97
|
|
82
|
98 break;
|
|
72
|
99 }
|
|
|
100 }
|
|
|
101
|
|
|
102 dest->tail = d;
|
|
|
103 src->head = s;
|
|
|
104 return offset;
|
|
|
105 }
|
|
|
106
|
|
|
107 /* like gather, but with no output list */
|
|
|
108 static int discard(struct flist *src, int cut, int offset)
|
|
|
109 {
|
|
|
110 struct frag *s = src->head;
|
|
|
111 int postend, c, l;
|
|
|
112
|
|
|
113 while (s != src->tail) {
|
|
|
114 if (s->start + offset >= cut)
|
|
82
|
115 break;
|
|
72
|
116
|
|
|
117 postend = offset + s->start + s->len;
|
|
|
118 if (postend <= cut) {
|
|
|
119 offset += s->start + s->len - s->end;
|
|
|
120 s++;
|
|
|
121 }
|
|
|
122 else {
|
|
|
123 c = cut - offset;
|
|
|
124 if (s->end < c)
|
|
|
125 c = s->end;
|
|
|
126 l = cut - offset - s->start;
|
|
|
127 if (s->len < l)
|
|
|
128 l = s->len;
|
|
|
129
|
|
|
130 offset += s->start + l - c;
|
|
|
131 s->start = c;
|
|
|
132 s->len = s->len - l;
|
|
|
133 s->data = s->data + l;
|
|
|
134
|
|
82
|
135 break;
|
|
72
|
136 }
|
|
|
137 }
|
|
|
138
|
|
|
139 src->head = s;
|
|
|
140 return offset;
|
|
|
141 }
|
|
|
142
|
|
|
143 /* combine hunk lists a and b, while adjusting b for offset changes in a/
|
|
|
144 this deletes a and b and returns the resultant list. */
|
|
|
145 static struct flist *combine(struct flist *a, struct flist *b)
|
|
|
146 {
|
|
|
147 struct flist *c;
|
|
|
148 struct frag *bh = b->head, *ct;
|
|
|
149 int offset = 0, post;
|
|
|
150
|
|
|
151 c = lalloc((lsize(a) + lsize(b)) * 2);
|
|
|
152
|
|
|
153 while (bh != b->tail) {
|
|
|
154 /* save old hunks */
|
|
|
155 offset = gather(c, a, bh->start, offset);
|
|
|
156
|
|
|
157 /* discard replaced hunks */
|
|
|
158 post = discard(a, bh->end, offset);
|
|
|
159
|
|
|
160 /* insert new hunk */
|
|
|
161 ct = c->tail;
|
|
|
162 ct->start = bh->start - offset;
|
|
|
163 ct->end = bh->end - post;
|
|
|
164 ct->len = bh->len;
|
|
|
165 ct->data = bh->data;
|
|
|
166 c->tail++;
|
|
|
167 bh++;
|
|
|
168 offset = post;
|
|
|
169 }
|
|
|
170
|
|
|
171 /* hold on to tail from a */
|
|
|
172 memcpy(c->tail, a->head, sizeof(struct frag) * lsize(a));
|
|
|
173 c->tail += lsize(a);
|
|
|
174 lfree(a);
|
|
|
175 lfree(b);
|
|
|
176 return c;
|
|
|
177 }
|
|
|
178
|
|
|
179 /* decode a binary patch into a hunk list */
|
|
|
180 static struct flist *decode(char *bin, int len)
|
|
|
181 {
|
|
|
182 struct flist *l;
|
|
|
183 struct frag *lt;
|
|
|
184 char *end = bin + len;
|
|
|
185
|
|
|
186 /* assume worst case size, we won't have many of these lists */
|
|
|
187 l = lalloc(len / 12);
|
|
|
188 lt = l->tail;
|
|
|
189
|
|
|
190 while (bin < end) {
|
|
|
191 lt->start = ntohl(*(uint32_t *)bin);
|
|
|
192 lt->end = ntohl(*(uint32_t *)(bin + 4));
|
|
|
193 lt->len = ntohl(*(uint32_t *)(bin + 8));
|
|
|
194 lt->data = bin + 12;
|
|
|
195 bin += 12 + lt->len;
|
|
|
196 lt++;
|
|
|
197 }
|
|
|
198
|
|
|
199 l->tail = lt;
|
|
|
200 return l;
|
|
|
201 }
|
|
|
202
|
|
|
203 /* calculate the size of resultant text */
|
|
|
204 static int calcsize(int len, struct flist *l)
|
|
|
205 {
|
|
|
206 int outlen = 0, last = 0;
|
|
|
207 struct frag *f = l->head;
|
|
|
208
|
|
|
209 while (f != l->tail) {
|
|
|
210 outlen += f->start - last;
|
|
|
211 last = f->end;
|
|
|
212 outlen += f->len;
|
|
|
213 f++;
|
|
|
214 }
|
|
|
215
|
|
|
216 outlen += len - last;
|
|
|
217 return outlen;
|
|
|
218 }
|
|
|
219
|
|
|
220 static void apply(char *buf, char *orig, int len, struct flist *l)
|
|
|
221 {
|
|
|
222 struct frag *f = l->head;
|
|
|
223 int last = 0;
|
|
|
224 char *p = buf;
|
|
|
225
|
|
|
226 while (f != l->tail) {
|
|
|
227 memcpy(p, orig + last, f->start - last);
|
|
|
228 p += f->start - last;
|
|
|
229 memcpy(p, f->data, f->len);
|
|
|
230 last = f->end;
|
|
|
231 p += f->len;
|
|
|
232 f++;
|
|
|
233 }
|
|
|
234 memcpy(p, orig + last, len - last);
|
|
|
235 }
|
|
|
236
|
|
|
237 /* recursively generate a patch of all bins between start and end */
|
|
|
238 static struct flist *fold(PyObject *bins, int start, int end)
|
|
|
239 {
|
|
|
240 int len;
|
|
|
241
|
|
|
242 if (start + 1 == end) {
|
|
|
243 /* trivial case, output a decoded list */
|
|
|
244 PyObject *tmp = PyList_GetItem(bins, start);
|
|
|
245 return decode(PyString_AsString(tmp), PyString_Size(tmp));
|
|
|
246 }
|
|
|
247
|
|
|
248 /* divide and conquer, memory management is elsewhere */
|
|
|
249 len = (end - start) / 2;
|
|
|
250 return combine(fold(bins, start, start + len),
|
|
|
251 fold(bins, start + len, end));
|
|
|
252 }
|
|
|
253
|
|
|
254 static PyObject *
|
|
|
255 patches(PyObject *self, PyObject *args)
|
|
|
256 {
|
|
|
257 PyObject *text, *bins, *result;
|
|
|
258 struct flist *patch;
|
|
|
259 char *in, *out;
|
|
|
260 int len, outlen;
|
|
|
261
|
|
|
262 if (!PyArg_ParseTuple(args, "OO:mpatch", &text, &bins))
|
|
|
263 return NULL;
|
|
|
264
|
|
|
265 len = PyList_Size(bins);
|
|
|
266 if (!len) {
|
|
|
267 /* nothing to do */
|
|
|
268 Py_INCREF(text);
|
|
|
269 return text;
|
|
|
270 }
|
|
|
271
|
|
|
272 patch = fold(bins, 0, len);
|
|
|
273 outlen = calcsize(PyString_Size(text), patch);
|
|
|
274 result = PyString_FromStringAndSize(NULL, outlen);
|
|
|
275 in = PyString_AsString(text);
|
|
|
276 out = PyString_AsString(result);
|
|
|
277 apply(out, in, PyString_Size(text), patch);
|
|
|
278 lfree(patch);
|
|
|
279
|
|
|
280 return result;
|
|
|
281 }
|
|
|
282
|
|
|
283 static PyMethodDef methods[] = {
|
|
|
284 {"patches", patches, METH_VARARGS, "apply a series of patches\n"},
|
|
|
285 {NULL, NULL}
|
|
|
286 };
|
|
|
287
|
|
|
288 PyMODINIT_FUNC
|
|
|
289 initmpatch(void)
|
|
|
290 {
|
|
|
291 Py_InitModule3("mpatch", methods, mpatch_doc);
|
|
|
292 }
|
|
|
293
|
