Show More
@@ -1,352 +1,350 b'' | |||
|
1 | 1 | /* |
|
2 | 2 | bdiff.c - efficient binary diff extension for Mercurial |
|
3 | 3 | |
|
4 | 4 | Copyright 2005, 2006 Matt Mackall <mpm@selenic.com> |
|
5 | 5 | |
|
6 | 6 | This software may be used and distributed according to the terms of |
|
7 | 7 | the GNU General Public License, incorporated herein by reference. |
|
8 | 8 | |
|
9 | 9 | Based roughly on Python difflib |
|
10 | 10 | */ |
|
11 | 11 | |
|
12 | 12 | #define PY_SSIZE_T_CLEAN |
|
13 | 13 | #include <Python.h> |
|
14 | 14 | #include <limits.h> |
|
15 | 15 | #include <stdlib.h> |
|
16 | 16 | #include <string.h> |
|
17 | 17 | |
|
18 | 18 | #include "bdiff.h" |
|
19 | 19 | #include "bitmanipulation.h" |
|
20 | 20 | #include "thirdparty/xdiff/xdiff.h" |
|
21 | 21 | #include "util.h" |
|
22 | 22 | |
|
23 | 23 | static PyObject *blocks(PyObject *self, PyObject *args) |
|
24 | 24 | { |
|
25 | 25 | PyObject *sa, *sb, *rl = NULL, *m; |
|
26 | 26 | struct bdiff_line *a, *b; |
|
27 | 27 | struct bdiff_hunk l, *h; |
|
28 | 28 | int an, bn, count, pos = 0; |
|
29 | 29 | |
|
30 | 30 | l.next = NULL; |
|
31 | 31 | |
|
32 | 32 | if (!PyArg_ParseTuple(args, "SS:bdiff", &sa, &sb)) |
|
33 | 33 | return NULL; |
|
34 | 34 | |
|
35 | 35 | an = bdiff_splitlines(PyBytes_AsString(sa), PyBytes_Size(sa), &a); |
|
36 | 36 | bn = bdiff_splitlines(PyBytes_AsString(sb), PyBytes_Size(sb), &b); |
|
37 | 37 | |
|
38 | 38 | if (!a || !b) |
|
39 | 39 | goto nomem; |
|
40 | 40 | |
|
41 | 41 | count = bdiff_diff(a, an, b, bn, &l); |
|
42 | 42 | if (count < 0) |
|
43 | 43 | goto nomem; |
|
44 | 44 | |
|
45 | 45 | rl = PyList_New(count); |
|
46 | 46 | if (!rl) |
|
47 | 47 | goto nomem; |
|
48 | 48 | |
|
49 | 49 | for (h = l.next; h; h = h->next) { |
|
50 | 50 | m = Py_BuildValue("iiii", h->a1, h->a2, h->b1, h->b2); |
|
51 | 51 | PyList_SetItem(rl, pos, m); |
|
52 | 52 | pos++; |
|
53 | 53 | } |
|
54 | 54 | |
|
55 | 55 | nomem: |
|
56 | 56 | free(a); |
|
57 | 57 | free(b); |
|
58 | 58 | bdiff_freehunks(l.next); |
|
59 | 59 | return rl ? rl : PyErr_NoMemory(); |
|
60 | 60 | } |
|
61 | 61 | |
|
62 | 62 | static PyObject *bdiff(PyObject *self, PyObject *args) |
|
63 | 63 | { |
|
64 | 64 | Py_buffer ba, bb; |
|
65 | 65 | char *rb, *ia, *ib; |
|
66 | 66 | PyObject *result = NULL; |
|
67 | 67 | struct bdiff_line *al = NULL, *bl = NULL; |
|
68 | 68 | struct bdiff_hunk l, *h; |
|
69 | 69 | int an, bn, count; |
|
70 | 70 | Py_ssize_t len = 0, la, lb, li = 0, lcommon = 0, lmax; |
|
71 | 71 | PyThreadState *_save = NULL; |
|
72 | 72 | |
|
73 | 73 | l.next = NULL; |
|
74 | 74 | |
|
75 | 75 | if (!PyArg_ParseTuple(args, PY23("s*s*:bdiff", "y*y*:bdiff"), &ba, &bb)) |
|
76 | 76 | return NULL; |
|
77 | 77 | |
|
78 | 78 | if (!PyBuffer_IsContiguous(&ba, 'C') || ba.ndim > 1) { |
|
79 | 79 | PyErr_SetString(PyExc_ValueError, "bdiff input not contiguous"); |
|
80 | 80 | goto cleanup; |
|
81 | 81 | } |
|
82 | 82 | |
|
83 | 83 | if (!PyBuffer_IsContiguous(&bb, 'C') || bb.ndim > 1) { |
|
84 | 84 | PyErr_SetString(PyExc_ValueError, "bdiff input not contiguous"); |
|
85 | 85 | goto cleanup; |
|
86 | 86 | } |
|
87 | 87 | |
|
88 | 88 | la = ba.len; |
|
89 | 89 | lb = bb.len; |
|
90 | 90 | |
|
91 | 91 | if (la > UINT_MAX || lb > UINT_MAX) { |
|
92 | 92 | PyErr_SetString(PyExc_ValueError, "bdiff inputs too large"); |
|
93 | 93 | goto cleanup; |
|
94 | 94 | } |
|
95 | 95 | |
|
96 | 96 | _save = PyEval_SaveThread(); |
|
97 | 97 | |
|
98 | 98 | lmax = la > lb ? lb : la; |
|
99 | 99 | for (ia = ba.buf, ib = bb.buf; li < lmax && *ia == *ib; |
|
100 | 100 | ++li, ++ia, ++ib) { |
|
101 | 101 | if (*ia == '\n') |
|
102 | 102 | lcommon = li + 1; |
|
103 | 103 | } |
|
104 | 104 | /* we can almost add: if (li == lmax) lcommon = li; */ |
|
105 | 105 | |
|
106 | 106 | an = bdiff_splitlines((char *)ba.buf + lcommon, la - lcommon, &al); |
|
107 | 107 | bn = bdiff_splitlines((char *)bb.buf + lcommon, lb - lcommon, &bl); |
|
108 | 108 | if (!al || !bl) { |
|
109 | 109 | PyErr_NoMemory(); |
|
110 | 110 | goto cleanup; |
|
111 | 111 | } |
|
112 | 112 | |
|
113 | 113 | count = bdiff_diff(al, an, bl, bn, &l); |
|
114 | 114 | if (count < 0) { |
|
115 | 115 | PyErr_NoMemory(); |
|
116 | 116 | goto cleanup; |
|
117 | 117 | } |
|
118 | 118 | |
|
119 | 119 | /* calculate length of output */ |
|
120 | 120 | la = lb = 0; |
|
121 | 121 | for (h = l.next; h; h = h->next) { |
|
122 | 122 | if (h->a1 != la || h->b1 != lb) |
|
123 | 123 | len += 12 + bl[h->b1].l - bl[lb].l; |
|
124 | 124 | la = h->a2; |
|
125 | 125 | lb = h->b2; |
|
126 | 126 | } |
|
127 | 127 | PyEval_RestoreThread(_save); |
|
128 | 128 | _save = NULL; |
|
129 | 129 | |
|
130 | 130 | result = PyBytes_FromStringAndSize(NULL, len); |
|
131 | 131 | |
|
132 | 132 | if (!result) |
|
133 | 133 | goto cleanup; |
|
134 | 134 | |
|
135 | 135 | /* build binary patch */ |
|
136 | 136 | rb = PyBytes_AsString(result); |
|
137 | 137 | la = lb = 0; |
|
138 | 138 | |
|
139 | 139 | for (h = l.next; h; h = h->next) { |
|
140 | 140 | if (h->a1 != la || h->b1 != lb) { |
|
141 | 141 | len = bl[h->b1].l - bl[lb].l; |
|
142 | 142 | putbe32((uint32_t)(al[la].l + lcommon - al->l), rb); |
|
143 | 143 | putbe32((uint32_t)(al[h->a1].l + lcommon - al->l), |
|
144 | 144 | rb + 4); |
|
145 | 145 | putbe32((uint32_t)len, rb + 8); |
|
146 | 146 | memcpy(rb + 12, bl[lb].l, len); |
|
147 | 147 | rb += 12 + len; |
|
148 | 148 | } |
|
149 | 149 | la = h->a2; |
|
150 | 150 | lb = h->b2; |
|
151 | 151 | } |
|
152 | 152 | |
|
153 | 153 | cleanup: |
|
154 | 154 | if (_save) |
|
155 | 155 | PyEval_RestoreThread(_save); |
|
156 | 156 | PyBuffer_Release(&ba); |
|
157 | 157 | PyBuffer_Release(&bb); |
|
158 | 158 | if (al) { |
|
159 | 159 | free(al); |
|
160 | 160 | } |
|
161 | 161 | if (bl) { |
|
162 | 162 | free(bl); |
|
163 | 163 | } |
|
164 | 164 | if (l.next) { |
|
165 | 165 | bdiff_freehunks(l.next); |
|
166 | 166 | } |
|
167 | 167 | return result; |
|
168 | 168 | } |
|
169 | 169 | |
|
170 | 170 | /* |
|
171 | 171 | * If allws != 0, remove all whitespace (' ', \t and \r). Otherwise, |
|
172 | 172 | * reduce whitespace sequences to a single space and trim remaining whitespace |
|
173 | 173 | * from end of lines. |
|
174 | 174 | */ |
|
175 | 175 | static PyObject *fixws(PyObject *self, PyObject *args) |
|
176 | 176 | { |
|
177 | 177 | PyObject *s, *result = NULL; |
|
178 | 178 | char allws, c; |
|
179 | 179 | const char *r; |
|
180 | 180 | Py_ssize_t i, rlen, wlen = 0; |
|
181 | 181 | char *w; |
|
182 | 182 | |
|
183 | 183 | if (!PyArg_ParseTuple(args, "Sb:fixws", &s, &allws)) |
|
184 | 184 | return NULL; |
|
185 | 185 | r = PyBytes_AsString(s); |
|
186 | 186 | rlen = PyBytes_Size(s); |
|
187 | 187 | |
|
188 | 188 | w = (char *)PyMem_Malloc(rlen ? rlen : 1); |
|
189 | 189 | if (!w) |
|
190 | 190 | goto nomem; |
|
191 | 191 | |
|
192 | 192 | for (i = 0; i != rlen; i++) { |
|
193 | 193 | c = r[i]; |
|
194 | 194 | if (c == ' ' || c == '\t' || c == '\r') { |
|
195 | 195 | if (!allws && (wlen == 0 || w[wlen - 1] != ' ')) |
|
196 | 196 | w[wlen++] = ' '; |
|
197 | 197 | } else if (c == '\n' && !allws && wlen > 0 && |
|
198 | 198 | w[wlen - 1] == ' ') { |
|
199 | 199 | w[wlen - 1] = '\n'; |
|
200 | 200 | } else { |
|
201 | 201 | w[wlen++] = c; |
|
202 | 202 | } |
|
203 | 203 | } |
|
204 | 204 | |
|
205 | 205 | result = PyBytes_FromStringAndSize(w, wlen); |
|
206 | 206 | |
|
207 | 207 | nomem: |
|
208 | 208 | PyMem_Free(w); |
|
209 | 209 | return result ? result : PyErr_NoMemory(); |
|
210 | 210 | } |
|
211 | 211 | |
|
212 | 212 | static bool sliceintolist(PyObject *list, Py_ssize_t destidx, |
|
213 | 213 | const char *source, Py_ssize_t len) |
|
214 | 214 | { |
|
215 | 215 | PyObject *sliced = PyBytes_FromStringAndSize(source, len); |
|
216 | 216 | if (sliced == NULL) |
|
217 | 217 | return false; |
|
218 | 218 | PyList_SET_ITEM(list, destidx, sliced); |
|
219 | 219 | return true; |
|
220 | 220 | } |
|
221 | 221 | |
|
222 | 222 | static PyObject *splitnewlines(PyObject *self, PyObject *args) |
|
223 | 223 | { |
|
224 | 224 | const char *text; |
|
225 | 225 | Py_ssize_t nelts = 0, size, i, start = 0; |
|
226 | 226 | PyObject *result = NULL; |
|
227 | 227 | |
|
228 | 228 | if (!PyArg_ParseTuple(args, PY23("s#", "y#"), &text, &size)) { |
|
229 | 229 | goto abort; |
|
230 | 230 | } |
|
231 | 231 | if (!size) { |
|
232 | 232 | return PyList_New(0); |
|
233 | 233 | } |
|
234 | 234 | /* This loops to size-1 because if the last byte is a newline, |
|
235 | 235 | * we don't want to perform a split there. */ |
|
236 | 236 | for (i = 0; i < size - 1; ++i) { |
|
237 | 237 | if (text[i] == '\n') { |
|
238 | 238 | ++nelts; |
|
239 | 239 | } |
|
240 | 240 | } |
|
241 | 241 | if ((result = PyList_New(nelts + 1)) == NULL) |
|
242 | 242 | goto abort; |
|
243 | 243 | nelts = 0; |
|
244 | 244 | for (i = 0; i < size - 1; ++i) { |
|
245 | 245 | if (text[i] == '\n') { |
|
246 | 246 | if (!sliceintolist(result, nelts++, text + start, |
|
247 | 247 | i - start + 1)) |
|
248 | 248 | goto abort; |
|
249 | 249 | start = i + 1; |
|
250 | 250 | } |
|
251 | 251 | } |
|
252 | 252 | if (!sliceintolist(result, nelts++, text + start, size - start)) |
|
253 | 253 | goto abort; |
|
254 | 254 | return result; |
|
255 | 255 | abort: |
|
256 | 256 | Py_XDECREF(result); |
|
257 | 257 | return NULL; |
|
258 | 258 | } |
|
259 | 259 | |
|
260 | 260 | static int hunk_consumer(long a1, long a2, long b1, long b2, void *priv) |
|
261 | 261 | { |
|
262 | 262 | PyObject *rl = (PyObject *)priv; |
|
263 | 263 | PyObject *m = Py_BuildValue("llll", a1, a2, b1, b2); |
|
264 | 264 | if (!m) |
|
265 | 265 | return -1; |
|
266 | 266 | if (PyList_Append(rl, m) != 0) { |
|
267 | 267 | Py_DECREF(m); |
|
268 | 268 | return -1; |
|
269 | 269 | } |
|
270 | 270 | return 0; |
|
271 | 271 | } |
|
272 | 272 | |
|
273 | 273 | static PyObject *xdiffblocks(PyObject *self, PyObject *args) |
|
274 | 274 | { |
|
275 | 275 | Py_ssize_t la, lb; |
|
276 | 276 | mmfile_t a, b; |
|
277 | 277 | PyObject *rl; |
|
278 | 278 | |
|
279 | 279 | xpparam_t xpp = { |
|
280 | 280 | XDF_INDENT_HEURISTIC, /* flags */ |
|
281 | NULL, /* anchors */ | |
|
282 | 0, /* anchors_nr */ | |
|
283 | 281 | }; |
|
284 | 282 | xdemitconf_t xecfg = { |
|
285 | 283 | 0, /* ctxlen */ |
|
286 | 284 | 0, /* interhunkctxlen */ |
|
287 | 285 | XDL_EMIT_BDIFFHUNK, /* flags */ |
|
288 | 286 | NULL, /* find_func */ |
|
289 | 287 | NULL, /* find_func_priv */ |
|
290 | 288 | hunk_consumer, /* hunk_consume_func */ |
|
291 | 289 | }; |
|
292 | 290 | xdemitcb_t ecb = { |
|
293 | 291 | NULL, /* priv */ |
|
294 | 292 | NULL, /* outf */ |
|
295 | 293 | }; |
|
296 | 294 | |
|
297 | 295 | if (!PyArg_ParseTuple(args, PY23("s#s#", "y#y#"), &a.ptr, &la, &b.ptr, |
|
298 | 296 | &lb)) |
|
299 | 297 | return NULL; |
|
300 | 298 | |
|
301 | 299 | a.size = la; |
|
302 | 300 | b.size = lb; |
|
303 | 301 | |
|
304 | 302 | rl = PyList_New(0); |
|
305 | 303 | if (!rl) |
|
306 | 304 | return PyErr_NoMemory(); |
|
307 | 305 | |
|
308 | 306 | ecb.priv = rl; |
|
309 | 307 | |
|
310 | 308 | if (xdl_diff(&a, &b, &xpp, &xecfg, &ecb) != 0) { |
|
311 | 309 | Py_DECREF(rl); |
|
312 | 310 | return PyErr_NoMemory(); |
|
313 | 311 | } |
|
314 | 312 | |
|
315 | 313 | return rl; |
|
316 | 314 | } |
|
317 | 315 | |
|
318 | 316 | static char mdiff_doc[] = "Efficient binary diff."; |
|
319 | 317 | |
|
320 | 318 | static PyMethodDef methods[] = { |
|
321 | 319 | {"bdiff", bdiff, METH_VARARGS, "calculate a binary diff\n"}, |
|
322 | 320 | {"blocks", blocks, METH_VARARGS, "find a list of matching lines\n"}, |
|
323 | 321 | {"fixws", fixws, METH_VARARGS, "normalize diff whitespaces\n"}, |
|
324 | 322 | {"splitnewlines", splitnewlines, METH_VARARGS, |
|
325 | 323 | "like str.splitlines, but only split on newlines\n"}, |
|
326 | 324 | {"xdiffblocks", xdiffblocks, METH_VARARGS, |
|
327 | 325 | "find a list of matching lines using xdiff algorithm\n"}, |
|
328 | 326 | {NULL, NULL}, |
|
329 | 327 | }; |
|
330 | 328 | |
|
331 | 329 | static const int version = 3; |
|
332 | 330 | |
|
333 | 331 | #ifdef IS_PY3K |
|
334 | 332 | static struct PyModuleDef bdiff_module = { |
|
335 | 333 | PyModuleDef_HEAD_INIT, "bdiff", mdiff_doc, -1, methods, |
|
336 | 334 | }; |
|
337 | 335 | |
|
338 | 336 | PyMODINIT_FUNC PyInit_bdiff(void) |
|
339 | 337 | { |
|
340 | 338 | PyObject *m; |
|
341 | 339 | m = PyModule_Create(&bdiff_module); |
|
342 | 340 | PyModule_AddIntConstant(m, "version", version); |
|
343 | 341 | return m; |
|
344 | 342 | } |
|
345 | 343 | #else |
|
346 | 344 | PyMODINIT_FUNC initbdiff(void) |
|
347 | 345 | { |
|
348 | 346 | PyObject *m; |
|
349 | 347 | m = Py_InitModule3("bdiff", methods, mdiff_doc); |
|
350 | 348 | PyModule_AddIntConstant(m, "version", version); |
|
351 | 349 | } |
|
352 | 350 | #endif |
@@ -1,139 +1,123 b'' | |||
|
1 | 1 | /* |
|
2 | 2 | * LibXDiff by Davide Libenzi ( File Differential Library ) |
|
3 | 3 | * Copyright (C) 2003 Davide Libenzi |
|
4 | 4 | * |
|
5 | 5 | * This library is free software; you can redistribute it and/or |
|
6 | 6 | * modify it under the terms of the GNU Lesser General Public |
|
7 | 7 | * License as published by the Free Software Foundation; either |
|
8 | 8 | * version 2.1 of the License, or (at your option) any later version. |
|
9 | 9 | * |
|
10 | 10 | * This library is distributed in the hope that it will be useful, |
|
11 | 11 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
|
12 | 12 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
|
13 | 13 | * Lesser General Public License for more details. |
|
14 | 14 | * |
|
15 | 15 | * You should have received a copy of the GNU Lesser General Public |
|
16 | 16 | * License along with this library; if not, see |
|
17 | 17 | * <http://www.gnu.org/licenses/>. |
|
18 | 18 | * |
|
19 | 19 | * Davide Libenzi <davidel@xmailserver.org> |
|
20 | 20 | * |
|
21 | 21 | */ |
|
22 | 22 | |
|
23 | 23 | #if !defined(XDIFF_H) |
|
24 | 24 | #define XDIFF_H |
|
25 | 25 | |
|
26 | 26 | #ifdef __cplusplus |
|
27 | 27 | extern "C" { |
|
28 | 28 | #endif /* #ifdef __cplusplus */ |
|
29 | 29 | |
|
30 | 30 | #include <stddef.h> /* size_t */ |
|
31 | 31 | |
|
32 | 32 | /* xpparm_t.flags */ |
|
33 | 33 | #define XDF_NEED_MINIMAL (1 << 0) |
|
34 | 34 | |
|
35 | 35 | #define XDF_INDENT_HEURISTIC (1 << 23) |
|
36 | 36 | |
|
37 | 37 | /* xdemitconf_t.flags */ |
|
38 | 38 | #define XDL_EMIT_FUNCNAMES (1 << 0) |
|
39 | 39 | #define XDL_EMIT_FUNCCONTEXT (1 << 2) |
|
40 | 40 | /* emit bdiff-style "matched" (a1, a2, b1, b2) hunks instead of "different" |
|
41 | 41 | * (a1, a2 - a1, b1, b2 - b1) hunks */ |
|
42 | 42 | #define XDL_EMIT_BDIFFHUNK (1 << 4) |
|
43 | 43 | |
|
44 | #define XDL_MMB_READONLY (1 << 0) | |
|
45 | ||
|
46 | #define XDL_MMF_ATOMIC (1 << 0) | |
|
47 | ||
|
48 | #define XDL_BDOP_INS 1 | |
|
49 | #define XDL_BDOP_CPY 2 | |
|
50 | #define XDL_BDOP_INSB 3 | |
|
51 | ||
|
52 | 44 | /* merge simplification levels */ |
|
53 | 45 | #define XDL_MERGE_MINIMAL 0 |
|
54 | 46 | #define XDL_MERGE_EAGER 1 |
|
55 | 47 | #define XDL_MERGE_ZEALOUS 2 |
|
56 | 48 | #define XDL_MERGE_ZEALOUS_ALNUM 3 |
|
57 | 49 | |
|
58 | 50 | /* merge favor modes */ |
|
59 | 51 | #define XDL_MERGE_FAVOR_OURS 1 |
|
60 | 52 | #define XDL_MERGE_FAVOR_THEIRS 2 |
|
61 | 53 | #define XDL_MERGE_FAVOR_UNION 3 |
|
62 | 54 | |
|
63 | 55 | /* merge output styles */ |
|
64 | 56 | #define XDL_MERGE_DIFF3 1 |
|
65 | 57 | |
|
66 | 58 | typedef struct s_mmfile { |
|
67 | 59 | char *ptr; |
|
68 | 60 | long size; |
|
69 | 61 | } mmfile_t; |
|
70 | 62 | |
|
71 | 63 | typedef struct s_mmbuffer { |
|
72 | 64 | char *ptr; |
|
73 | 65 | long size; |
|
74 | 66 | } mmbuffer_t; |
|
75 | 67 | |
|
76 | 68 | typedef struct s_xpparam { |
|
77 | 69 | unsigned long flags; |
|
78 | ||
|
79 | /* See Documentation/diff-options.txt. */ | |
|
80 | char **anchors; | |
|
81 | size_t anchors_nr; | |
|
82 | 70 | } xpparam_t; |
|
83 | 71 | |
|
84 | 72 | typedef struct s_xdemitcb { |
|
85 | 73 | void *priv; |
|
86 | 74 | int (*outf)(void *, mmbuffer_t *, int); |
|
87 | 75 | } xdemitcb_t; |
|
88 | 76 | |
|
89 | 77 | typedef long (*find_func_t)(const char *line, long line_len, char *buffer, long buffer_size, void *priv); |
|
90 | 78 | |
|
91 | 79 | typedef int (*xdl_emit_hunk_consume_func_t)(long start_a, long count_a, |
|
92 | 80 | long start_b, long count_b, |
|
93 | 81 | void *cb_data); |
|
94 | 82 | |
|
95 | 83 | typedef struct s_xdemitconf { |
|
96 | 84 | long ctxlen; |
|
97 | 85 | long interhunkctxlen; |
|
98 | 86 | unsigned long flags; |
|
99 | 87 | find_func_t find_func; |
|
100 | 88 | void *find_func_priv; |
|
101 | 89 | xdl_emit_hunk_consume_func_t hunk_func; |
|
102 | 90 | } xdemitconf_t; |
|
103 | 91 | |
|
104 | typedef struct s_bdiffparam { | |
|
105 | long bsize; | |
|
106 | } bdiffparam_t; | |
|
107 | ||
|
108 | 92 | |
|
109 | 93 | #define xdl_malloc(x) malloc(x) |
|
110 | 94 | #define xdl_free(ptr) free(ptr) |
|
111 | 95 | #define xdl_realloc(ptr,x) realloc(ptr,x) |
|
112 | 96 | |
|
113 | 97 | void *xdl_mmfile_first(mmfile_t *mmf, long *size); |
|
114 | 98 | long xdl_mmfile_size(mmfile_t *mmf); |
|
115 | 99 | |
|
116 | 100 | int xdl_diff(mmfile_t *mf1, mmfile_t *mf2, xpparam_t const *xpp, |
|
117 | 101 | xdemitconf_t const *xecfg, xdemitcb_t *ecb); |
|
118 | 102 | |
|
119 | 103 | typedef struct s_xmparam { |
|
120 | 104 | xpparam_t xpp; |
|
121 | 105 | int marker_size; |
|
122 | 106 | int level; |
|
123 | 107 | int favor; |
|
124 | 108 | int style; |
|
125 | 109 | const char *ancestor; /* label for orig */ |
|
126 | 110 | const char *file1; /* label for mf1 */ |
|
127 | 111 | const char *file2; /* label for mf2 */ |
|
128 | 112 | } xmparam_t; |
|
129 | 113 | |
|
130 | 114 | #define DEFAULT_CONFLICT_MARKER_SIZE 7 |
|
131 | 115 | |
|
132 | 116 | int xdl_merge(mmfile_t *orig, mmfile_t *mf1, mmfile_t *mf2, |
|
133 | 117 | xmparam_t const *xmp, mmbuffer_t *result); |
|
134 | 118 | |
|
135 | 119 | #ifdef __cplusplus |
|
136 | 120 | } |
|
137 | 121 | #endif /* #ifdef __cplusplus */ |
|
138 | 122 | |
|
139 | 123 | #endif /* #if !defined(XDIFF_H) */ |
@@ -1,64 +1,60 b'' | |||
|
1 | 1 | /* |
|
2 | 2 | * LibXDiff by Davide Libenzi ( File Differential Library ) |
|
3 | 3 | * Copyright (C) 2003 Davide Libenzi |
|
4 | 4 | * |
|
5 | 5 | * This library is free software; you can redistribute it and/or |
|
6 | 6 | * modify it under the terms of the GNU Lesser General Public |
|
7 | 7 | * License as published by the Free Software Foundation; either |
|
8 | 8 | * version 2.1 of the License, or (at your option) any later version. |
|
9 | 9 | * |
|
10 | 10 | * This library is distributed in the hope that it will be useful, |
|
11 | 11 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
|
12 | 12 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
|
13 | 13 | * Lesser General Public License for more details. |
|
14 | 14 | * |
|
15 | 15 | * You should have received a copy of the GNU Lesser General Public |
|
16 | 16 | * License along with this library; if not, see |
|
17 | 17 | * <http://www.gnu.org/licenses/>. |
|
18 | 18 | * |
|
19 | 19 | * Davide Libenzi <davidel@xmailserver.org> |
|
20 | 20 | * |
|
21 | 21 | */ |
|
22 | 22 | |
|
23 | 23 | #if !defined(XDIFFI_H) |
|
24 | 24 | #define XDIFFI_H |
|
25 | 25 | |
|
26 | 26 | |
|
27 | 27 | typedef struct s_diffdata { |
|
28 | 28 | long nrec; |
|
29 | 29 | unsigned long const *ha; |
|
30 | 30 | long *rindex; |
|
31 | 31 | char *rchg; |
|
32 | 32 | } diffdata_t; |
|
33 | 33 | |
|
34 | 34 | typedef struct s_xdalgoenv { |
|
35 | 35 | long mxcost; |
|
36 | 36 | long snake_cnt; |
|
37 | 37 | long heur_min; |
|
38 | 38 | } xdalgoenv_t; |
|
39 | 39 | |
|
40 | 40 | typedef struct s_xdchange { |
|
41 | 41 | struct s_xdchange *next; |
|
42 | 42 | long i1, i2; |
|
43 | 43 | long chg1, chg2; |
|
44 | 44 | int ignore; |
|
45 | 45 | } xdchange_t; |
|
46 | 46 | |
|
47 | 47 | |
|
48 | 48 | |
|
49 | 49 | int xdl_recs_cmp(diffdata_t *dd1, long off1, long lim1, |
|
50 | 50 | diffdata_t *dd2, long off2, long lim2, |
|
51 | 51 | long *kvdf, long *kvdb, int need_min, xdalgoenv_t *xenv); |
|
52 | 52 | int xdl_do_diff(mmfile_t *mf1, mmfile_t *mf2, xpparam_t const *xpp, |
|
53 | 53 | xdfenv_t *xe); |
|
54 | 54 | int xdl_change_compact(xdfile_t *xdf, xdfile_t *xdfo, long flags); |
|
55 | 55 | int xdl_build_script(xdfenv_t *xe, xdchange_t **xscr); |
|
56 | 56 | void xdl_free_script(xdchange_t *xscr); |
|
57 | 57 | int xdl_emit_diff(xdfenv_t *xe, xdchange_t *xscr, xdemitcb_t *ecb, |
|
58 | 58 | xdemitconf_t const *xecfg); |
|
59 | int xdl_do_patience_diff(mmfile_t *mf1, mmfile_t *mf2, xpparam_t const *xpp, | |
|
60 | xdfenv_t *env); | |
|
61 | int xdl_do_histogram_diff(mmfile_t *mf1, mmfile_t *mf2, xpparam_t const *xpp, | |
|
62 | xdfenv_t *env); | |
|
63 | 59 | |
|
64 | 60 | #endif /* #if !defined(XDIFFI_H) */ |
@@ -1,274 +1,243 b'' | |||
|
1 | 1 | /* |
|
2 | 2 | * LibXDiff by Davide Libenzi ( File Differential Library ) |
|
3 | 3 | * Copyright (C) 2003 Davide Libenzi |
|
4 | 4 | * |
|
5 | 5 | * This library is free software; you can redistribute it and/or |
|
6 | 6 | * modify it under the terms of the GNU Lesser General Public |
|
7 | 7 | * License as published by the Free Software Foundation; either |
|
8 | 8 | * version 2.1 of the License, or (at your option) any later version. |
|
9 | 9 | * |
|
10 | 10 | * This library is distributed in the hope that it will be useful, |
|
11 | 11 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
|
12 | 12 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
|
13 | 13 | * Lesser General Public License for more details. |
|
14 | 14 | * |
|
15 | 15 | * You should have received a copy of the GNU Lesser General Public |
|
16 | 16 | * License along with this library; if not, see |
|
17 | 17 | * <http://www.gnu.org/licenses/>. |
|
18 | 18 | * |
|
19 | 19 | * Davide Libenzi <davidel@xmailserver.org> |
|
20 | 20 | * |
|
21 | 21 | */ |
|
22 | 22 | |
|
23 | 23 | #include <limits.h> |
|
24 | 24 | #include <assert.h> |
|
25 | 25 | #include "xinclude.h" |
|
26 | 26 | |
|
27 | 27 | |
|
28 | 28 | |
|
29 | 29 | |
|
30 | 30 | long xdl_bogosqrt(long n) { |
|
31 | 31 | long i; |
|
32 | 32 | |
|
33 | 33 | /* |
|
34 | 34 | * Classical integer square root approximation using shifts. |
|
35 | 35 | */ |
|
36 | 36 | for (i = 1; n > 0; n >>= 2) |
|
37 | 37 | i <<= 1; |
|
38 | 38 | |
|
39 | 39 | return i; |
|
40 | 40 | } |
|
41 | 41 | |
|
42 | 42 | |
|
43 | 43 | int xdl_emit_diffrec(char const *rec, long size, char const *pre, long psize, |
|
44 | 44 | xdemitcb_t *ecb) { |
|
45 | 45 | int i = 2; |
|
46 | 46 | mmbuffer_t mb[3]; |
|
47 | 47 | |
|
48 | 48 | mb[0].ptr = (char *) pre; |
|
49 | 49 | mb[0].size = psize; |
|
50 | 50 | mb[1].ptr = (char *) rec; |
|
51 | 51 | mb[1].size = size; |
|
52 | 52 | if (size > 0 && rec[size - 1] != '\n') { |
|
53 | 53 | mb[2].ptr = (char *) "\n\\ No newline at end of file\n"; |
|
54 | 54 | mb[2].size = strlen(mb[2].ptr); |
|
55 | 55 | i++; |
|
56 | 56 | } |
|
57 | 57 | if (ecb->outf(ecb->priv, mb, i) < 0) { |
|
58 | 58 | |
|
59 | 59 | return -1; |
|
60 | 60 | } |
|
61 | 61 | |
|
62 | 62 | return 0; |
|
63 | 63 | } |
|
64 | 64 | |
|
65 | 65 | void *xdl_mmfile_first(mmfile_t *mmf, long *size) |
|
66 | 66 | { |
|
67 | 67 | *size = mmf->size; |
|
68 | 68 | return mmf->ptr; |
|
69 | 69 | } |
|
70 | 70 | |
|
71 | 71 | |
|
72 | 72 | long xdl_mmfile_size(mmfile_t *mmf) |
|
73 | 73 | { |
|
74 | 74 | return mmf->size; |
|
75 | 75 | } |
|
76 | 76 | |
|
77 | 77 | |
|
78 | 78 | int xdl_cha_init(chastore_t *cha, long isize, long icount) { |
|
79 | 79 | |
|
80 | 80 | cha->head = cha->tail = NULL; |
|
81 | 81 | cha->isize = isize; |
|
82 | 82 | cha->nsize = icount * isize; |
|
83 | 83 | cha->ancur = cha->sncur = NULL; |
|
84 | 84 | cha->scurr = 0; |
|
85 | 85 | |
|
86 | 86 | return 0; |
|
87 | 87 | } |
|
88 | 88 | |
|
89 | 89 | |
|
90 | 90 | void xdl_cha_free(chastore_t *cha) { |
|
91 | 91 | chanode_t *cur, *tmp; |
|
92 | 92 | |
|
93 | 93 | for (cur = cha->head; (tmp = cur) != NULL;) { |
|
94 | 94 | cur = cur->next; |
|
95 | 95 | xdl_free(tmp); |
|
96 | 96 | } |
|
97 | 97 | } |
|
98 | 98 | |
|
99 | 99 | |
|
100 | 100 | void *xdl_cha_alloc(chastore_t *cha) { |
|
101 | 101 | chanode_t *ancur; |
|
102 | 102 | void *data; |
|
103 | 103 | |
|
104 | 104 | if (!(ancur = cha->ancur) || ancur->icurr == cha->nsize) { |
|
105 | 105 | if (!(ancur = (chanode_t *) xdl_malloc(sizeof(chanode_t) + cha->nsize))) { |
|
106 | 106 | |
|
107 | 107 | return NULL; |
|
108 | 108 | } |
|
109 | 109 | ancur->icurr = 0; |
|
110 | 110 | ancur->next = NULL; |
|
111 | 111 | if (cha->tail) |
|
112 | 112 | cha->tail->next = ancur; |
|
113 | 113 | if (!cha->head) |
|
114 | 114 | cha->head = ancur; |
|
115 | 115 | cha->tail = ancur; |
|
116 | 116 | cha->ancur = ancur; |
|
117 | 117 | } |
|
118 | 118 | |
|
119 | 119 | data = (char *) ancur + sizeof(chanode_t) + ancur->icurr; |
|
120 | 120 | ancur->icurr += cha->isize; |
|
121 | 121 | |
|
122 | 122 | return data; |
|
123 | 123 | } |
|
124 | 124 | |
|
125 | 125 | long xdl_guess_lines(mmfile_t *mf, long sample) { |
|
126 | 126 | long nl = 0, size, tsize = 0; |
|
127 | 127 | char const *data, *cur, *top; |
|
128 | 128 | |
|
129 | 129 | if ((cur = data = xdl_mmfile_first(mf, &size)) != NULL) { |
|
130 | 130 | for (top = data + size; nl < sample && cur < top; ) { |
|
131 | 131 | nl++; |
|
132 | 132 | if (!(cur = memchr(cur, '\n', top - cur))) |
|
133 | 133 | cur = top; |
|
134 | 134 | else |
|
135 | 135 | cur++; |
|
136 | 136 | } |
|
137 | 137 | tsize += (long) (cur - data); |
|
138 | 138 | } |
|
139 | 139 | |
|
140 | 140 | if (nl && tsize) |
|
141 | 141 | nl = xdl_mmfile_size(mf) / (tsize / nl); |
|
142 | 142 | |
|
143 | 143 | return nl + 1; |
|
144 | 144 | } |
|
145 | 145 | |
|
146 | 146 | int xdl_recmatch(const char *l1, long s1, const char *l2, long s2, long flags) |
|
147 | 147 | { |
|
148 | 148 | if (s1 == s2 && !memcmp(l1, l2, s1)) |
|
149 | 149 | return 1; |
|
150 | 150 | return 0; |
|
151 | 151 | } |
|
152 | 152 | |
|
153 | 153 | unsigned long xdl_hash_record(char const **data, char const *top, long flags) { |
|
154 | 154 | unsigned long ha = 5381; |
|
155 | 155 | char const *ptr = *data; |
|
156 | 156 | |
|
157 | 157 | for (; ptr < top && *ptr != '\n'; ptr++) { |
|
158 | 158 | ha += (ha << 5); |
|
159 | 159 | ha ^= (unsigned long) *ptr; |
|
160 | 160 | } |
|
161 | 161 | *data = ptr < top ? ptr + 1: ptr; |
|
162 | 162 | |
|
163 | 163 | return ha; |
|
164 | 164 | } |
|
165 | 165 | |
|
166 | 166 | unsigned int xdl_hashbits(unsigned int size) { |
|
167 | 167 | unsigned int val = 1, bits = 0; |
|
168 | 168 | |
|
169 | 169 | for (; val < size && bits < CHAR_BIT * sizeof(unsigned int); val <<= 1, bits++); |
|
170 | 170 | return bits ? bits: 1; |
|
171 | 171 | } |
|
172 | 172 | |
|
173 | 173 | |
|
174 | 174 | int xdl_num_out(char *out, long val) { |
|
175 | 175 | char *ptr, *str = out; |
|
176 | 176 | char buf[32]; |
|
177 | 177 | |
|
178 | 178 | ptr = buf + sizeof(buf) - 1; |
|
179 | 179 | *ptr = '\0'; |
|
180 | 180 | if (val < 0) { |
|
181 | 181 | *--ptr = '-'; |
|
182 | 182 | val = -val; |
|
183 | 183 | } |
|
184 | 184 | for (; val && ptr > buf; val /= 10) |
|
185 | 185 | *--ptr = "0123456789"[val % 10]; |
|
186 | 186 | if (*ptr) |
|
187 | 187 | for (; *ptr; ptr++, str++) |
|
188 | 188 | *str = *ptr; |
|
189 | 189 | else |
|
190 | 190 | *str++ = '0'; |
|
191 | 191 | *str = '\0'; |
|
192 | 192 | |
|
193 | 193 | return str - out; |
|
194 | 194 | } |
|
195 | 195 | |
|
196 | 196 | int xdl_emit_hunk_hdr(long s1, long c1, long s2, long c2, |
|
197 | 197 | const char *func, long funclen, xdemitcb_t *ecb) { |
|
198 | 198 | int nb = 0; |
|
199 | 199 | mmbuffer_t mb; |
|
200 | 200 | char buf[128]; |
|
201 | 201 | |
|
202 | 202 | memcpy(buf, "@@ -", 4); |
|
203 | 203 | nb += 4; |
|
204 | 204 | |
|
205 | 205 | nb += xdl_num_out(buf + nb, c1 ? s1: s1 - 1); |
|
206 | 206 | |
|
207 | 207 | if (c1 != 1) { |
|
208 | 208 | memcpy(buf + nb, ",", 1); |
|
209 | 209 | nb += 1; |
|
210 | 210 | |
|
211 | 211 | nb += xdl_num_out(buf + nb, c1); |
|
212 | 212 | } |
|
213 | 213 | |
|
214 | 214 | memcpy(buf + nb, " +", 2); |
|
215 | 215 | nb += 2; |
|
216 | 216 | |
|
217 | 217 | nb += xdl_num_out(buf + nb, c2 ? s2: s2 - 1); |
|
218 | 218 | |
|
219 | 219 | if (c2 != 1) { |
|
220 | 220 | memcpy(buf + nb, ",", 1); |
|
221 | 221 | nb += 1; |
|
222 | 222 | |
|
223 | 223 | nb += xdl_num_out(buf + nb, c2); |
|
224 | 224 | } |
|
225 | 225 | |
|
226 | 226 | memcpy(buf + nb, " @@", 3); |
|
227 | 227 | nb += 3; |
|
228 | 228 | if (func && funclen) { |
|
229 | 229 | buf[nb++] = ' '; |
|
230 | 230 | if (funclen > sizeof(buf) - nb - 1) |
|
231 | 231 | funclen = sizeof(buf) - nb - 1; |
|
232 | 232 | memcpy(buf + nb, func, funclen); |
|
233 | 233 | nb += funclen; |
|
234 | 234 | } |
|
235 | 235 | buf[nb++] = '\n'; |
|
236 | 236 | |
|
237 | 237 | mb.ptr = buf; |
|
238 | 238 | mb.size = nb; |
|
239 | 239 | if (ecb->outf(ecb->priv, &mb, 1) < 0) |
|
240 | 240 | return -1; |
|
241 | 241 | |
|
242 | 242 | return 0; |
|
243 | 243 | } |
|
244 | ||
|
245 | int xdl_fall_back_diff(xdfenv_t *diff_env, xpparam_t const *xpp, | |
|
246 | int line1, int count1, int line2, int count2) | |
|
247 | { | |
|
248 | /* | |
|
249 | * This probably does not work outside Git, since | |
|
250 | * we have a very simple mmfile structure. | |
|
251 | * | |
|
252 | * Note: ideally, we would reuse the prepared environment, but | |
|
253 | * the libxdiff interface does not (yet) allow for diffing only | |
|
254 | * ranges of lines instead of the whole files. | |
|
255 | */ | |
|
256 | mmfile_t subfile1, subfile2; | |
|
257 | xdfenv_t env; | |
|
258 | ||
|
259 | subfile1.ptr = (char *)diff_env->xdf1.recs[line1 - 1]->ptr; | |
|
260 | subfile1.size = diff_env->xdf1.recs[line1 + count1 - 2]->ptr + | |
|
261 | diff_env->xdf1.recs[line1 + count1 - 2]->size - subfile1.ptr; | |
|
262 | subfile2.ptr = (char *)diff_env->xdf2.recs[line2 - 1]->ptr; | |
|
263 | subfile2.size = diff_env->xdf2.recs[line2 + count2 - 2]->ptr + | |
|
264 | diff_env->xdf2.recs[line2 + count2 - 2]->size - subfile2.ptr; | |
|
265 | if (xdl_do_diff(&subfile1, &subfile2, xpp, &env) < 0) | |
|
266 | return -1; | |
|
267 | ||
|
268 | memcpy(diff_env->xdf1.rchg + line1 - 1, env.xdf1.rchg, count1); | |
|
269 | memcpy(diff_env->xdf2.rchg + line2 - 1, env.xdf2.rchg, count2); | |
|
270 | ||
|
271 | xdl_free_env(&env); | |
|
272 | ||
|
273 | return 0; | |
|
274 | } |
@@ -1,46 +1,44 b'' | |||
|
1 | 1 | /* |
|
2 | 2 | * LibXDiff by Davide Libenzi ( File Differential Library ) |
|
3 | 3 | * Copyright (C) 2003 Davide Libenzi |
|
4 | 4 | * |
|
5 | 5 | * This library is free software; you can redistribute it and/or |
|
6 | 6 | * modify it under the terms of the GNU Lesser General Public |
|
7 | 7 | * License as published by the Free Software Foundation; either |
|
8 | 8 | * version 2.1 of the License, or (at your option) any later version. |
|
9 | 9 | * |
|
10 | 10 | * This library is distributed in the hope that it will be useful, |
|
11 | 11 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
|
12 | 12 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
|
13 | 13 | * Lesser General Public License for more details. |
|
14 | 14 | * |
|
15 | 15 | * You should have received a copy of the GNU Lesser General Public |
|
16 | 16 | * License along with this library; if not, see |
|
17 | 17 | * <http://www.gnu.org/licenses/>. |
|
18 | 18 | * |
|
19 | 19 | * Davide Libenzi <davidel@xmailserver.org> |
|
20 | 20 | * |
|
21 | 21 | */ |
|
22 | 22 | |
|
23 | 23 | #if !defined(XUTILS_H) |
|
24 | 24 | #define XUTILS_H |
|
25 | 25 | |
|
26 | 26 | |
|
27 | 27 | |
|
28 | 28 | long xdl_bogosqrt(long n); |
|
29 | 29 | int xdl_emit_diffrec(char const *rec, long size, char const *pre, long psize, |
|
30 | 30 | xdemitcb_t *ecb); |
|
31 | 31 | int xdl_cha_init(chastore_t *cha, long isize, long icount); |
|
32 | 32 | void xdl_cha_free(chastore_t *cha); |
|
33 | 33 | void *xdl_cha_alloc(chastore_t *cha); |
|
34 | 34 | long xdl_guess_lines(mmfile_t *mf, long sample); |
|
35 | 35 | int xdl_recmatch(const char *l1, long s1, const char *l2, long s2, long flags); |
|
36 | 36 | unsigned long xdl_hash_record(char const **data, char const *top, long flags); |
|
37 | 37 | unsigned int xdl_hashbits(unsigned int size); |
|
38 | 38 | int xdl_num_out(char *out, long val); |
|
39 | 39 | int xdl_emit_hunk_hdr(long s1, long c1, long s2, long c2, |
|
40 | 40 | const char *func, long funclen, xdemitcb_t *ecb); |
|
41 | int xdl_fall_back_diff(xdfenv_t *diff_env, xpparam_t const *xpp, | |
|
42 | int line1, int count1, int line2, int count2); | |
|
43 | 41 | |
|
44 | 42 | |
|
45 | 43 | |
|
46 | 44 | #endif /* #if !defined(XUTILS_H) */ |
General Comments 0
You need to be logged in to leave comments.
Login now