Show More
@@ -1,760 +1,760 b'' | |||
|
1 | 1 | /* |
|
2 | 2 | pathencode.c - efficient path name encoding |
|
3 | 3 | |
|
4 | 4 | Copyright 2012 Facebook |
|
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 | |
|
10 | 10 | /* |
|
11 | 11 | * An implementation of the name encoding scheme used by the fncache |
|
12 | 12 | * store. The common case is of a path < 120 bytes long, which is |
|
13 | 13 | * handled either in a single pass with no allocations or two passes |
|
14 | 14 | * with a single allocation. For longer paths, multiple passes are |
|
15 | 15 | * required. |
|
16 | 16 | */ |
|
17 | 17 | |
|
18 | 18 | #define PY_SSIZE_T_CLEAN |
|
19 | 19 | #include <Python.h> |
|
20 | 20 | #include <assert.h> |
|
21 | 21 | #include <ctype.h> |
|
22 | 22 | #include <stdlib.h> |
|
23 | 23 | #include <string.h> |
|
24 | 24 | |
|
25 | 25 | #include "util.h" |
|
26 | 26 | |
|
27 | 27 | /* state machine for the fast path */ |
|
28 | 28 | enum path_state { |
|
29 | 29 | START, /* first byte of a path component */ |
|
30 | 30 | A, /* "AUX" */ |
|
31 | 31 | AU, |
|
32 | 32 | THIRD, /* third of a 3-byte sequence, e.g. "AUX", "NUL" */ |
|
33 | 33 | C, /* "CON" or "COMn" */ |
|
34 | 34 | CO, |
|
35 | 35 | COMLPT, /* "COM" or "LPT" */ |
|
36 | 36 | COMLPTn, |
|
37 | 37 | L, |
|
38 | 38 | LP, |
|
39 | 39 | N, |
|
40 | 40 | NU, |
|
41 | 41 | P, /* "PRN" */ |
|
42 | 42 | PR, |
|
43 | 43 | LDOT, /* leading '.' */ |
|
44 | 44 | DOT, /* '.' in a non-leading position */ |
|
45 | 45 | H, /* ".h" */ |
|
46 | 46 | HGDI, /* ".hg", ".d", or ".i" */ |
|
47 | 47 | SPACE, |
|
48 | 48 | DEFAULT /* byte of a path component after the first */ |
|
49 | 49 | }; |
|
50 | 50 | |
|
51 | 51 | /* state machine for dir-encoding */ |
|
52 | 52 | enum dir_state { |
|
53 | 53 | DDOT, |
|
54 | 54 | DH, |
|
55 | 55 | DHGDI, |
|
56 | 56 | DDEFAULT |
|
57 | 57 | }; |
|
58 | 58 | |
|
59 | 59 | static inline int inset(const uint32_t bitset[], char c) |
|
60 | 60 | { |
|
61 | 61 | return bitset[((uint8_t)c) >> 5] & (1 << (((uint8_t)c) & 31)); |
|
62 | 62 | } |
|
63 | 63 | |
|
64 | 64 | static inline void charcopy(char *dest, Py_ssize_t *destlen, size_t destsize, |
|
65 | 65 | char c) |
|
66 | 66 | { |
|
67 | 67 | if (dest) { |
|
68 | 68 | assert(*destlen < destsize); |
|
69 | 69 | dest[*destlen] = c; |
|
70 | 70 | } |
|
71 | 71 | (*destlen)++; |
|
72 | 72 | } |
|
73 | 73 | |
|
74 | 74 | static inline void memcopy(char *dest, Py_ssize_t *destlen, size_t destsize, |
|
75 | 75 | const void *src, Py_ssize_t len) |
|
76 | 76 | { |
|
77 | 77 | if (dest) { |
|
78 | 78 | assert(*destlen + len < destsize); |
|
79 | 79 | memcpy((void *)&dest[*destlen], src, len); |
|
80 | 80 | } |
|
81 | 81 | *destlen += len; |
|
82 | 82 | } |
|
83 | 83 | |
|
84 | 84 | static inline void hexencode(char *dest, Py_ssize_t *destlen, size_t destsize, |
|
85 | 85 | uint8_t c) |
|
86 | 86 | { |
|
87 | 87 | static const char hexdigit[] = "0123456789abcdef"; |
|
88 | 88 | |
|
89 | 89 | charcopy(dest, destlen, destsize, hexdigit[c >> 4]); |
|
90 | 90 | charcopy(dest, destlen, destsize, hexdigit[c & 15]); |
|
91 | 91 | } |
|
92 | 92 | |
|
93 | 93 | /* 3-byte escape: tilde followed by two hex digits */ |
|
94 | 94 | static inline void escape3(char *dest, Py_ssize_t *destlen, size_t destsize, |
|
95 | 95 | char c) |
|
96 | 96 | { |
|
97 | 97 | charcopy(dest, destlen, destsize, '~'); |
|
98 | 98 | hexencode(dest, destlen, destsize, c); |
|
99 | 99 | } |
|
100 | 100 | |
|
101 | 101 | static Py_ssize_t _encodedir(char *dest, size_t destsize, |
|
102 | 102 | const char *src, Py_ssize_t len) |
|
103 | 103 | { |
|
104 | 104 | enum dir_state state = DDEFAULT; |
|
105 | 105 | Py_ssize_t i = 0, destlen = 0; |
|
106 | 106 | |
|
107 | 107 | while (i < len) { |
|
108 | 108 | switch (state) { |
|
109 | 109 | case DDOT: |
|
110 | 110 | switch (src[i]) { |
|
111 | 111 | case 'd': |
|
112 | 112 | case 'i': |
|
113 | 113 | state = DHGDI; |
|
114 | 114 | charcopy(dest, &destlen, destsize, src[i++]); |
|
115 | 115 | break; |
|
116 | 116 | case 'h': |
|
117 | 117 | state = DH; |
|
118 | 118 | charcopy(dest, &destlen, destsize, src[i++]); |
|
119 | 119 | break; |
|
120 | 120 | default: |
|
121 | 121 | state = DDEFAULT; |
|
122 | 122 | break; |
|
123 | 123 | } |
|
124 | 124 | break; |
|
125 | 125 | case DH: |
|
126 | 126 | if (src[i] == 'g') { |
|
127 | 127 | state = DHGDI; |
|
128 | 128 | charcopy(dest, &destlen, destsize, src[i++]); |
|
129 | 129 | } |
|
130 | 130 | else state = DDEFAULT; |
|
131 | 131 | break; |
|
132 | 132 | case DHGDI: |
|
133 | 133 | if (src[i] == '/') { |
|
134 | 134 | memcopy(dest, &destlen, destsize, ".hg", 3); |
|
135 | 135 | charcopy(dest, &destlen, destsize, src[i++]); |
|
136 | 136 | } |
|
137 | 137 | state = DDEFAULT; |
|
138 | 138 | break; |
|
139 | 139 | case DDEFAULT: |
|
140 | 140 | if (src[i] == '.') |
|
141 | 141 | state = DDOT; |
|
142 | 142 | charcopy(dest, &destlen, destsize, src[i++]); |
|
143 | 143 | break; |
|
144 | 144 | } |
|
145 | 145 | } |
|
146 | 146 | |
|
147 | 147 | return destlen; |
|
148 | 148 | } |
|
149 | 149 | |
|
150 | 150 | PyObject *encodedir(PyObject *self, PyObject *args) |
|
151 | 151 | { |
|
152 | 152 | Py_ssize_t len, newlen; |
|
153 | 153 | PyObject *pathobj, *newobj; |
|
154 | 154 | char *path; |
|
155 | 155 | |
|
156 | 156 | if (!PyArg_ParseTuple(args, "O:encodedir", &pathobj)) |
|
157 | 157 | return NULL; |
|
158 | 158 | |
|
159 | 159 | if (PyString_AsStringAndSize(pathobj, &path, &len) == -1) { |
|
160 | 160 | PyErr_SetString(PyExc_TypeError, "expected a string"); |
|
161 | 161 | return NULL; |
|
162 | 162 | } |
|
163 | 163 | |
|
164 | 164 | newlen = len ? _encodedir(NULL, 0, path, len + 1) : 1; |
|
165 | 165 | |
|
166 | 166 | if (newlen == len + 1) { |
|
167 | 167 | Py_INCREF(pathobj); |
|
168 | 168 | return pathobj; |
|
169 | 169 | } |
|
170 | 170 | |
|
171 | 171 | newobj = PyString_FromStringAndSize(NULL, newlen); |
|
172 | 172 | |
|
173 | 173 | if (newobj) { |
|
174 | 174 | PyString_GET_SIZE(newobj)--; |
|
175 | 175 | _encodedir(PyString_AS_STRING(newobj), newlen, path, |
|
176 | 176 | len + 1); |
|
177 | 177 | } |
|
178 | 178 | |
|
179 | 179 | return newobj; |
|
180 | 180 | } |
|
181 | 181 | |
|
182 | 182 | static Py_ssize_t _encode(const uint32_t twobytes[8], const uint32_t onebyte[8], |
|
183 | 183 | char *dest, Py_ssize_t destlen, size_t destsize, |
|
184 | 184 | const char *src, Py_ssize_t len, |
|
185 | 185 | int encodedir) |
|
186 | 186 | { |
|
187 | 187 | enum path_state state = START; |
|
188 | 188 | Py_ssize_t i = 0; |
|
189 | 189 | |
|
190 | 190 | /* |
|
191 | 191 | * Python strings end with a zero byte, which we use as a |
|
192 | 192 | * terminal token as they are not valid inside path names. |
|
193 | 193 | */ |
|
194 | 194 | |
|
195 | 195 | while (i < len) { |
|
196 | 196 | switch (state) { |
|
197 | 197 | case START: |
|
198 | 198 | switch (src[i]) { |
|
199 | 199 | case '/': |
|
200 | 200 | charcopy(dest, &destlen, destsize, src[i++]); |
|
201 | 201 | break; |
|
202 | 202 | case '.': |
|
203 | 203 | state = LDOT; |
|
204 | 204 | escape3(dest, &destlen, destsize, src[i++]); |
|
205 | 205 | break; |
|
206 | 206 | case ' ': |
|
207 | 207 | state = DEFAULT; |
|
208 | 208 | escape3(dest, &destlen, destsize, src[i++]); |
|
209 | 209 | break; |
|
210 | 210 | case 'a': |
|
211 | 211 | state = A; |
|
212 | 212 | charcopy(dest, &destlen, destsize, src[i++]); |
|
213 | 213 | break; |
|
214 | 214 | case 'c': |
|
215 | 215 | state = C; |
|
216 | 216 | charcopy(dest, &destlen, destsize, src[i++]); |
|
217 | 217 | break; |
|
218 | 218 | case 'l': |
|
219 | 219 | state = L; |
|
220 | 220 | charcopy(dest, &destlen, destsize, src[i++]); |
|
221 | 221 | break; |
|
222 | 222 | case 'n': |
|
223 | 223 | state = N; |
|
224 | 224 | charcopy(dest, &destlen, destsize, src[i++]); |
|
225 | 225 | break; |
|
226 | 226 | case 'p': |
|
227 | 227 | state = P; |
|
228 | 228 | charcopy(dest, &destlen, destsize, src[i++]); |
|
229 | 229 | break; |
|
230 | 230 | default: |
|
231 | 231 | state = DEFAULT; |
|
232 | 232 | break; |
|
233 | 233 | } |
|
234 | 234 | break; |
|
235 | 235 | case A: |
|
236 | 236 | if (src[i] == 'u') { |
|
237 | 237 | state = AU; |
|
238 | 238 | charcopy(dest, &destlen, destsize, src[i++]); |
|
239 | 239 | } |
|
240 | 240 | else state = DEFAULT; |
|
241 | 241 | break; |
|
242 | 242 | case AU: |
|
243 | 243 | if (src[i] == 'x') { |
|
244 | 244 | state = THIRD; |
|
245 | 245 | i++; |
|
246 | 246 | } |
|
247 | 247 | else state = DEFAULT; |
|
248 | 248 | break; |
|
249 | 249 | case THIRD: |
|
250 | 250 | state = DEFAULT; |
|
251 | 251 | switch (src[i]) { |
|
252 | 252 | case '.': |
|
253 | 253 | case '/': |
|
254 | 254 | case '\0': |
|
255 | 255 | escape3(dest, &destlen, destsize, src[i - 1]); |
|
256 | 256 | break; |
|
257 | 257 | default: |
|
258 | 258 | i--; |
|
259 | 259 | break; |
|
260 | 260 | } |
|
261 | 261 | break; |
|
262 | 262 | case C: |
|
263 | 263 | if (src[i] == 'o') { |
|
264 | 264 | state = CO; |
|
265 | 265 | charcopy(dest, &destlen, destsize, src[i++]); |
|
266 | 266 | } |
|
267 | 267 | else state = DEFAULT; |
|
268 | 268 | break; |
|
269 | 269 | case CO: |
|
270 | 270 | if (src[i] == 'm') { |
|
271 | 271 | state = COMLPT; |
|
272 | 272 | i++; |
|
273 | 273 | } |
|
274 | 274 | else if (src[i] == 'n') { |
|
275 | 275 | state = THIRD; |
|
276 | 276 | i++; |
|
277 | 277 | } |
|
278 | 278 | else state = DEFAULT; |
|
279 | 279 | break; |
|
280 | 280 | case COMLPT: |
|
281 | 281 | switch (src[i]) { |
|
282 | 282 | case '1': case '2': case '3': case '4': case '5': |
|
283 | 283 | case '6': case '7': case '8': case '9': |
|
284 | 284 | state = COMLPTn; |
|
285 | 285 | i++; |
|
286 | 286 | break; |
|
287 | 287 | default: |
|
288 | 288 | state = DEFAULT; |
|
289 | 289 | charcopy(dest, &destlen, destsize, src[i - 1]); |
|
290 | 290 | break; |
|
291 | 291 | } |
|
292 | 292 | break; |
|
293 | 293 | case COMLPTn: |
|
294 | 294 | state = DEFAULT; |
|
295 | 295 | switch (src[i]) { |
|
296 | 296 | case '.': |
|
297 | 297 | case '/': |
|
298 | 298 | case '\0': |
|
299 | 299 | escape3(dest, &destlen, destsize, src[i - 2]); |
|
300 | 300 | charcopy(dest, &destlen, destsize, src[i - 1]); |
|
301 | 301 | break; |
|
302 | 302 | default: |
|
303 | 303 | memcopy(dest, &destlen, destsize, |
|
304 | 304 | &src[i - 2], 2); |
|
305 | 305 | break; |
|
306 | 306 | } |
|
307 | 307 | break; |
|
308 | 308 | case L: |
|
309 | 309 | if (src[i] == 'p') { |
|
310 | 310 | state = LP; |
|
311 | 311 | charcopy(dest, &destlen, destsize, src[i++]); |
|
312 | 312 | } |
|
313 | 313 | else state = DEFAULT; |
|
314 | 314 | break; |
|
315 | 315 | case LP: |
|
316 | 316 | if (src[i] == 't') { |
|
317 | 317 | state = COMLPT; |
|
318 | 318 | i++; |
|
319 | 319 | } |
|
320 | 320 | else state = DEFAULT; |
|
321 | 321 | break; |
|
322 | 322 | case N: |
|
323 | 323 | if (src[i] == 'u') { |
|
324 | 324 | state = NU; |
|
325 | 325 | charcopy(dest, &destlen, destsize, src[i++]); |
|
326 | 326 | } |
|
327 | 327 | else state = DEFAULT; |
|
328 | 328 | break; |
|
329 | 329 | case NU: |
|
330 | 330 | if (src[i] == 'l') { |
|
331 | 331 | state = THIRD; |
|
332 | 332 | i++; |
|
333 | 333 | } |
|
334 | 334 | else state = DEFAULT; |
|
335 | 335 | break; |
|
336 | 336 | case P: |
|
337 | 337 | if (src[i] == 'r') { |
|
338 | 338 | state = PR; |
|
339 | 339 | charcopy(dest, &destlen, destsize, src[i++]); |
|
340 | 340 | } |
|
341 | 341 | else state = DEFAULT; |
|
342 | 342 | break; |
|
343 | 343 | case PR: |
|
344 | 344 | if (src[i] == 'n') { |
|
345 | 345 | state = THIRD; |
|
346 | 346 | i++; |
|
347 | 347 | } |
|
348 | 348 | else state = DEFAULT; |
|
349 | 349 | break; |
|
350 | 350 | case LDOT: |
|
351 | 351 | switch (src[i]) { |
|
352 | 352 | case 'd': |
|
353 | 353 | case 'i': |
|
354 | 354 | state = HGDI; |
|
355 | 355 | charcopy(dest, &destlen, destsize, src[i++]); |
|
356 | 356 | break; |
|
357 | 357 | case 'h': |
|
358 | 358 | state = H; |
|
359 | 359 | charcopy(dest, &destlen, destsize, src[i++]); |
|
360 | 360 | break; |
|
361 | 361 | default: |
|
362 | 362 | state = DEFAULT; |
|
363 | 363 | break; |
|
364 | 364 | } |
|
365 | 365 | break; |
|
366 | 366 | case DOT: |
|
367 | 367 | switch (src[i]) { |
|
368 | 368 | case '/': |
|
369 | 369 | case '\0': |
|
370 | 370 | state = START; |
|
371 | 371 | memcopy(dest, &destlen, destsize, "~2e", 3); |
|
372 | 372 | charcopy(dest, &destlen, destsize, src[i++]); |
|
373 | 373 | break; |
|
374 | 374 | case 'd': |
|
375 | 375 | case 'i': |
|
376 | 376 | state = HGDI; |
|
377 | 377 | charcopy(dest, &destlen, destsize, '.'); |
|
378 | 378 | charcopy(dest, &destlen, destsize, src[i++]); |
|
379 | 379 | break; |
|
380 | 380 | case 'h': |
|
381 | 381 | state = H; |
|
382 | 382 | memcopy(dest, &destlen, destsize, ".h", 2); |
|
383 | 383 | i++; |
|
384 | 384 | break; |
|
385 | 385 | default: |
|
386 | 386 | state = DEFAULT; |
|
387 | 387 | charcopy(dest, &destlen, destsize, '.'); |
|
388 | 388 | break; |
|
389 | 389 | } |
|
390 | 390 | break; |
|
391 | 391 | case H: |
|
392 | 392 | if (src[i] == 'g') { |
|
393 | 393 | state = HGDI; |
|
394 | 394 | charcopy(dest, &destlen, destsize, src[i++]); |
|
395 | 395 | } |
|
396 | 396 | else state = DEFAULT; |
|
397 | 397 | break; |
|
398 | 398 | case HGDI: |
|
399 | 399 | if (src[i] == '/') { |
|
400 | 400 | state = START; |
|
401 | 401 | if (encodedir) |
|
402 | 402 | memcopy(dest, &destlen, destsize, ".hg", |
|
403 | 403 | 3); |
|
404 | 404 | charcopy(dest, &destlen, destsize, src[i++]); |
|
405 | 405 | } |
|
406 | 406 | else state = DEFAULT; |
|
407 | 407 | break; |
|
408 | 408 | case SPACE: |
|
409 | 409 | switch (src[i]) { |
|
410 | 410 | case '/': |
|
411 | 411 | case '\0': |
|
412 | 412 | state = START; |
|
413 | 413 | memcopy(dest, &destlen, destsize, "~20", 3); |
|
414 | 414 | charcopy(dest, &destlen, destsize, src[i++]); |
|
415 | 415 | break; |
|
416 | 416 | default: |
|
417 | 417 | state = DEFAULT; |
|
418 | 418 | charcopy(dest, &destlen, destsize, ' '); |
|
419 | 419 | break; |
|
420 | 420 | } |
|
421 | 421 | break; |
|
422 | 422 | case DEFAULT: |
|
423 | 423 | while (inset(onebyte, src[i])) { |
|
424 | 424 | charcopy(dest, &destlen, destsize, src[i++]); |
|
425 | 425 | if (i == len) |
|
426 | 426 | goto done; |
|
427 | 427 | } |
|
428 | 428 | switch (src[i]) { |
|
429 | 429 | case '.': |
|
430 | 430 | state = DOT; |
|
431 | 431 | i++; |
|
432 | 432 | break; |
|
433 | 433 | case ' ': |
|
434 | 434 | state = SPACE; |
|
435 | 435 | i++; |
|
436 | 436 | break; |
|
437 | 437 | case '/': |
|
438 | 438 | state = START; |
|
439 | 439 | charcopy(dest, &destlen, destsize, '/'); |
|
440 | 440 | i++; |
|
441 | 441 | break; |
|
442 | 442 | default: |
|
443 | 443 | if (inset(onebyte, src[i])) { |
|
444 | 444 | do { |
|
445 | 445 | charcopy(dest, &destlen, |
|
446 | 446 | destsize, src[i++]); |
|
447 | 447 | } while (i < len && |
|
448 | 448 | inset(onebyte, src[i])); |
|
449 | 449 | } |
|
450 | 450 | else if (inset(twobytes, src[i])) { |
|
451 | 451 | char c = src[i++]; |
|
452 | 452 | charcopy(dest, &destlen, destsize, '_'); |
|
453 | 453 | charcopy(dest, &destlen, destsize, |
|
454 | 454 | c == '_' ? '_' : c + 32); |
|
455 | 455 | } |
|
456 | 456 | else |
|
457 | 457 | escape3(dest, &destlen, destsize, |
|
458 | 458 | src[i++]); |
|
459 | 459 | break; |
|
460 | 460 | } |
|
461 | 461 | break; |
|
462 | 462 | } |
|
463 | 463 | } |
|
464 | 464 | done: |
|
465 | 465 | return destlen; |
|
466 | 466 | } |
|
467 | 467 | |
|
468 | 468 | static Py_ssize_t basicencode(char *dest, size_t destsize, |
|
469 | 469 | const char *src, Py_ssize_t len) |
|
470 | 470 | { |
|
471 | 471 | static const uint32_t twobytes[8] = { 0, 0, 0x87fffffe }; |
|
472 | 472 | |
|
473 | 473 | static const uint32_t onebyte[8] = { |
|
474 | 474 | 1, 0x2bff3bfa, 0x68000001, 0x2fffffff, |
|
475 | 475 | }; |
|
476 | 476 | |
|
477 | 477 | Py_ssize_t destlen = 0; |
|
478 | 478 | |
|
479 | 479 | return _encode(twobytes, onebyte, dest, destlen, destsize, |
|
480 | 480 | src, len, 1); |
|
481 | 481 | } |
|
482 | 482 | |
|
483 | 483 | static const Py_ssize_t maxstorepathlen = 120; |
|
484 | 484 | |
|
485 | 485 | static Py_ssize_t _lowerencode(char *dest, size_t destsize, |
|
486 | 486 | const char *src, Py_ssize_t len) |
|
487 | 487 | { |
|
488 | 488 | static const uint32_t onebyte[8] = { |
|
489 | 489 | 1, 0x2bfffbfb, 0xe8000001, 0x2fffffff |
|
490 | 490 | }; |
|
491 | 491 | |
|
492 | 492 | static const uint32_t lower[8] = { 0, 0, 0x7fffffe }; |
|
493 | 493 | |
|
494 | 494 | Py_ssize_t i, destlen = 0; |
|
495 | 495 | |
|
496 | 496 | for (i = 0; i < len; i++) { |
|
497 | 497 | if (inset(onebyte, src[i])) |
|
498 | 498 | charcopy(dest, &destlen, destsize, src[i]); |
|
499 | 499 | else if (inset(lower, src[i])) |
|
500 | 500 | charcopy(dest, &destlen, destsize, src[i] + 32); |
|
501 | 501 | else |
|
502 | 502 | escape3(dest, &destlen, destsize, src[i]); |
|
503 | 503 | } |
|
504 | 504 | |
|
505 | 505 | return destlen; |
|
506 | 506 | } |
|
507 | 507 | |
|
508 | 508 | PyObject *lowerencode(PyObject *self, PyObject *args) |
|
509 | 509 | { |
|
510 | 510 | char *path; |
|
511 | 511 | Py_ssize_t len, newlen; |
|
512 | 512 | PyObject *ret; |
|
513 | 513 | |
|
514 | 514 | if (!PyArg_ParseTuple(args, "s#:lowerencode", &path, &len)) |
|
515 | 515 | return NULL; |
|
516 | 516 | |
|
517 | 517 | newlen = _lowerencode(NULL, 0, path, len); |
|
518 | 518 | ret = PyString_FromStringAndSize(NULL, newlen); |
|
519 | 519 | if (ret) |
|
520 | 520 | newlen = _lowerencode(PyString_AS_STRING(ret), newlen, |
|
521 | 521 | path, len); |
|
522 | 522 | |
|
523 | 523 | return ret; |
|
524 | 524 | } |
|
525 | 525 | |
|
526 | 526 | /* See store.py:_auxencode for a description. */ |
|
527 | 527 | static Py_ssize_t auxencode(char *dest, size_t destsize, |
|
528 | 528 | const char *src, Py_ssize_t len) |
|
529 | 529 | { |
|
530 | 530 | static const uint32_t twobytes[8]; |
|
531 | 531 | |
|
532 | 532 | static const uint32_t onebyte[8] = { |
|
533 | 533 | ~0, 0xffff3ffe, ~0, ~0, ~0, ~0, ~0, ~0, |
|
534 | 534 | }; |
|
535 | 535 | |
|
536 | 536 | return _encode(twobytes, onebyte, dest, 0, destsize, src, len, 0); |
|
537 | 537 | } |
|
538 | 538 | |
|
539 | 539 | static PyObject *hashmangle(const char *src, Py_ssize_t len, const char sha[20]) |
|
540 | 540 | { |
|
541 | 541 | static const Py_ssize_t dirprefixlen = 8; |
|
542 | 542 | static const Py_ssize_t maxshortdirslen = 68; |
|
543 | 543 | char *dest; |
|
544 | 544 | PyObject *ret; |
|
545 | 545 | |
|
546 | 546 | Py_ssize_t i, d, p, lastslash = len - 1, lastdot = -1; |
|
547 | 547 | Py_ssize_t destsize, destlen = 0, slop, used; |
|
548 | 548 | |
|
549 | 549 | while (lastslash >= 0 && src[lastslash] != '/') { |
|
550 | 550 | if (src[lastslash] == '.' && lastdot == -1) |
|
551 | 551 | lastdot = lastslash; |
|
552 | 552 | lastslash--; |
|
553 | 553 | } |
|
554 | 554 | |
|
555 | 555 | #if 0 |
|
556 | 556 | /* All paths should end in a suffix of ".i" or ".d". |
|
557 | 557 | Unfortunately, the file names in test-hybridencode.py |
|
558 | 558 | violate this rule. */ |
|
559 | 559 | if (lastdot != len - 3) { |
|
560 | 560 | PyErr_SetString(PyExc_ValueError, |
|
561 | 561 | "suffix missing or wrong length"); |
|
562 | 562 | return NULL; |
|
563 | 563 | } |
|
564 | 564 | #endif |
|
565 | 565 | |
|
566 | 566 | /* If src contains a suffix, we will append it to the end of |
|
567 | 567 | the new string, so make room. */ |
|
568 | 568 | destsize = 120; |
|
569 | 569 | if (lastdot >= 0) |
|
570 | 570 | destsize += len - lastdot - 1; |
|
571 | 571 | |
|
572 | 572 | ret = PyString_FromStringAndSize(NULL, destsize); |
|
573 | 573 | if (ret == NULL) |
|
574 | 574 | return NULL; |
|
575 | 575 | |
|
576 | 576 | dest = PyString_AS_STRING(ret); |
|
577 | 577 | memcopy(dest, &destlen, destsize, "dh/", 3); |
|
578 | 578 | |
|
579 | 579 | /* Copy up to dirprefixlen bytes of each path component, up to |
|
580 | 580 | a limit of maxshortdirslen bytes. */ |
|
581 | 581 | for (i = d = p = 0; i < lastslash; i++, p++) { |
|
582 | 582 | if (src[i] == '/') { |
|
583 | 583 | char d = dest[destlen - 1]; |
|
584 | 584 | /* After truncation, a directory name may end |
|
585 | 585 | in a space or dot, which are unportable. */ |
|
586 | 586 | if (d == '.' || d == ' ') |
|
587 | 587 | dest[destlen - 1] = '_'; |
|
588 | 588 | if (destlen > maxshortdirslen) |
|
589 | 589 | break; |
|
590 | 590 | charcopy(dest, &destlen, destsize, src[i]); |
|
591 | 591 | p = -1; |
|
592 | 592 | } |
|
593 | 593 | else if (p < dirprefixlen) |
|
594 | 594 | charcopy(dest, &destlen, destsize, src[i]); |
|
595 | 595 | } |
|
596 | 596 | |
|
597 | 597 | /* Rewind to just before the last slash copied. */ |
|
598 | 598 | if (destlen > maxshortdirslen + 3) |
|
599 | 599 | do { |
|
600 | 600 | destlen--; |
|
601 | 601 | } while (destlen > 0 && dest[destlen] != '/'); |
|
602 | 602 | |
|
603 | 603 | if (destlen > 3) { |
|
604 | 604 | if (lastslash > 0) { |
|
605 | 605 | char d = dest[destlen - 1]; |
|
606 | 606 | /* The last directory component may be |
|
607 | 607 | truncated, so make it safe. */ |
|
608 | 608 | if (d == '.' || d == ' ') |
|
609 | 609 | dest[destlen - 1] = '_'; |
|
610 | 610 | } |
|
611 | 611 | |
|
612 | 612 | charcopy(dest, &destlen, destsize, '/'); |
|
613 | 613 | } |
|
614 | 614 | |
|
615 | 615 | /* Add a prefix of the original file's name. Its length |
|
616 | 616 | depends on the number of bytes left after accounting for |
|
617 | 617 | hash and suffix. */ |
|
618 | 618 | used = destlen + 40; |
|
619 | 619 | if (lastdot >= 0) |
|
620 | 620 | used += len - lastdot - 1; |
|
621 | 621 | slop = maxstorepathlen - used; |
|
622 | 622 | if (slop > 0) { |
|
623 | 623 | Py_ssize_t basenamelen = |
|
624 | 624 | lastslash >= 0 ? len - lastslash - 2 : len - 1; |
|
625 | 625 | |
|
626 | 626 | if (basenamelen > slop) |
|
627 | 627 | basenamelen = slop; |
|
628 | 628 | if (basenamelen > 0) |
|
629 | 629 | memcopy(dest, &destlen, destsize, &src[lastslash + 1], |
|
630 | 630 | basenamelen); |
|
631 | 631 | } |
|
632 | 632 | |
|
633 | 633 | /* Add hash and suffix. */ |
|
634 | 634 | for (i = 0; i < 20; i++) |
|
635 | 635 | hexencode(dest, &destlen, destsize, sha[i]); |
|
636 | 636 | |
|
637 | 637 | if (lastdot >= 0) |
|
638 | 638 | memcopy(dest, &destlen, destsize, &src[lastdot], |
|
639 | 639 | len - lastdot - 1); |
|
640 | 640 | |
|
641 | 641 | PyString_GET_SIZE(ret) = destlen; |
|
642 | 642 | |
|
643 | 643 | return ret; |
|
644 | 644 | } |
|
645 | 645 | |
|
646 | 646 | /* |
|
647 | 647 | * Avoiding a trip through Python would improve performance by 50%, |
|
648 | 648 | * but we don't encounter enough long names to be worth the code. |
|
649 | 649 | */ |
|
650 | 650 | static int sha1hash(char hash[20], const char *str, Py_ssize_t len) |
|
651 | 651 | { |
|
652 | 652 | static PyObject *shafunc; |
|
653 | 653 | PyObject *shaobj, *hashobj; |
|
654 | 654 | |
|
655 | 655 | if (shafunc == NULL) { |
|
656 | 656 | PyObject *util, *name = PyString_FromString("mercurial.util"); |
|
657 | 657 | |
|
658 | 658 | if (name == NULL) |
|
659 | 659 | return -1; |
|
660 | 660 | |
|
661 | 661 | util = PyImport_Import(name); |
|
662 | 662 | Py_DECREF(name); |
|
663 | 663 | |
|
664 | 664 | if (util == NULL) { |
|
665 | 665 | PyErr_SetString(PyExc_ImportError, "mercurial.util"); |
|
666 | 666 | return -1; |
|
667 | 667 | } |
|
668 | 668 | shafunc = PyObject_GetAttrString(util, "sha1"); |
|
669 | 669 | Py_DECREF(util); |
|
670 | 670 | |
|
671 | 671 | if (shafunc == NULL) { |
|
672 | 672 | PyErr_SetString(PyExc_AttributeError, |
|
673 | 673 | "module 'mercurial.util' has no " |
|
674 | 674 | "attribute 'sha1'"); |
|
675 | 675 | return -1; |
|
676 | 676 | } |
|
677 | 677 | } |
|
678 | 678 | |
|
679 | 679 | shaobj = PyObject_CallFunction(shafunc, "s#", str, len); |
|
680 | 680 | |
|
681 | 681 | if (shaobj == NULL) |
|
682 | 682 | return -1; |
|
683 | 683 | |
|
684 | 684 | hashobj = PyObject_CallMethod(shaobj, "digest", ""); |
|
685 | 685 | Py_DECREF(shaobj); |
|
686 | 686 | |
|
687 | 687 | if (!PyString_Check(hashobj) || PyString_GET_SIZE(hashobj) != 20) { |
|
688 | 688 | PyErr_SetString(PyExc_TypeError, |
|
689 | 689 | "result of digest is not a 20-byte hash"); |
|
690 | 690 | Py_DECREF(hashobj); |
|
691 | 691 | return -1; |
|
692 | 692 | } |
|
693 | 693 | |
|
694 | 694 | memcpy(hash, PyString_AS_STRING(hashobj), 20); |
|
695 | 695 | Py_DECREF(hashobj); |
|
696 | 696 | return 0; |
|
697 | 697 | } |
|
698 | 698 | |
|
699 |
#define MAXENCODE 4096 * |
|
|
699 | #define MAXENCODE 4096 * 4 | |
|
700 | 700 | |
|
701 | 701 | static PyObject *hashencode(const char *src, Py_ssize_t len) |
|
702 | 702 | { |
|
703 | 703 | char dired[MAXENCODE]; |
|
704 | 704 | char lowered[MAXENCODE]; |
|
705 | 705 | char auxed[MAXENCODE]; |
|
706 | 706 | Py_ssize_t dirlen, lowerlen, auxlen, baselen; |
|
707 | 707 | char sha[20]; |
|
708 | 708 | |
|
709 | 709 | baselen = (len - 5) * 3; |
|
710 | 710 | if (baselen >= MAXENCODE) { |
|
711 | 711 | PyErr_SetString(PyExc_ValueError, "string too long"); |
|
712 | 712 | return NULL; |
|
713 | 713 | } |
|
714 | 714 | |
|
715 | 715 | dirlen = _encodedir(dired, baselen, src, len); |
|
716 | 716 | if (sha1hash(sha, dired, dirlen - 1) == -1) |
|
717 | 717 | return NULL; |
|
718 | 718 | lowerlen = _lowerencode(lowered, baselen, dired + 5, dirlen - 5); |
|
719 | 719 | auxlen = auxencode(auxed, baselen, lowered, lowerlen); |
|
720 | 720 | return hashmangle(auxed, auxlen, sha); |
|
721 | 721 | } |
|
722 | 722 | |
|
723 | 723 | PyObject *pathencode(PyObject *self, PyObject *args) |
|
724 | 724 | { |
|
725 | 725 | Py_ssize_t len, newlen; |
|
726 | 726 | PyObject *pathobj, *newobj; |
|
727 | 727 | char *path; |
|
728 | 728 | |
|
729 | 729 | if (!PyArg_ParseTuple(args, "O:pathencode", &pathobj)) |
|
730 | 730 | return NULL; |
|
731 | 731 | |
|
732 | 732 | if (PyString_AsStringAndSize(pathobj, &path, &len) == -1) { |
|
733 | 733 | PyErr_SetString(PyExc_TypeError, "expected a string"); |
|
734 | 734 | return NULL; |
|
735 | 735 | } |
|
736 | 736 | |
|
737 | 737 | if (len > maxstorepathlen) |
|
738 | 738 | newlen = maxstorepathlen + 2; |
|
739 | 739 | else |
|
740 | 740 | newlen = len ? basicencode(NULL, 0, path, len + 1) : 1; |
|
741 | 741 | |
|
742 | 742 | if (newlen <= maxstorepathlen + 1) { |
|
743 | 743 | if (newlen == len + 1) { |
|
744 | 744 | Py_INCREF(pathobj); |
|
745 | 745 | return pathobj; |
|
746 | 746 | } |
|
747 | 747 | |
|
748 | 748 | newobj = PyString_FromStringAndSize(NULL, newlen); |
|
749 | 749 | |
|
750 | 750 | if (newobj) { |
|
751 | 751 | PyString_GET_SIZE(newobj)--; |
|
752 | 752 | basicencode(PyString_AS_STRING(newobj), newlen, path, |
|
753 | 753 | len + 1); |
|
754 | 754 | } |
|
755 | 755 | } |
|
756 | 756 | else |
|
757 | 757 | newobj = hashencode(path, len + 1); |
|
758 | 758 | |
|
759 | 759 | return newobj; |
|
760 | 760 | } |
General Comments 0
You need to be logged in to leave comments.
Login now