Mercurial > projects > ldc
annotate dmd/macro.c @ 1638:0de4525a9ed6
Apply workaround for #395 by klickverbot.
| author | Christian Kamm <kamm incasoftware de> |
|---|---|
| date | Mon, 08 Mar 2010 20:06:08 +0100 |
| parents | b30fe7e1dbb9 |
| children |
| rev | line source |
|---|---|
| 1 | 1 |
| 2 // Copyright (c) 1999-2006 by Digital Mars | |
| 3 // All Rights Reserved | |
| 4 // written by Walter Bright | |
| 5 // http://www.digitalmars.com | |
| 6 // License for redistribution is by either the Artistic License | |
| 7 // in artistic.txt, or the GNU General Public License in gnu.txt. | |
| 8 // See the included readme.txt for details. | |
| 9 | |
| 10 /* Simple macro text processor. | |
| 11 */ | |
| 12 | |
| 13 #include <stdio.h> | |
| 14 #include <string.h> | |
| 15 #include <time.h> | |
| 16 #include <ctype.h> | |
| 17 #include <assert.h> | |
| 18 | |
|
1103
b30fe7e1dbb9
- Updated to DMD frontend 1.041.
Tomas Lindquist Olsen <tomas.l.olsen gmail.com>
parents:
571
diff
changeset
|
19 #include "rmem.h" |
|
b30fe7e1dbb9
- Updated to DMD frontend 1.041.
Tomas Lindquist Olsen <tomas.l.olsen gmail.com>
parents:
571
diff
changeset
|
20 #include "root.h" |
| 1 | 21 |
| 22 #include "macro.h" | |
| 23 | |
| 24 #define isidstart(c) (isalpha(c) || (c) == '_') | |
| 25 #define isidchar(c) (isalnum(c) || (c) == '_') | |
| 26 | |
| 27 unsigned char *memdup(unsigned char *p, size_t len) | |
| 28 { | |
| 29 return (unsigned char *)memcpy(mem.malloc(len), p, len); | |
| 30 } | |
| 31 | |
| 32 Macro::Macro(unsigned char *name, size_t namelen, unsigned char *text, size_t textlen) | |
| 33 { | |
| 34 next = NULL; | |
| 35 | |
| 36 #if 1 | |
| 37 this->name = name; | |
| 38 this->namelen = namelen; | |
| 39 | |
| 40 this->text = text; | |
| 41 this->textlen = textlen; | |
| 42 #else | |
| 43 this->name = name; | |
| 44 this->namelen = namelen; | |
| 45 | |
| 46 this->text = text; | |
| 47 this->textlen = textlen; | |
| 48 #endif | |
| 49 inuse = 0; | |
| 50 } | |
| 51 | |
| 52 | |
| 53 Macro *Macro::search(unsigned char *name, size_t namelen) | |
| 54 { Macro *table; | |
| 55 | |
| 56 //printf("Macro::search(%.*s)\n", namelen, name); | |
| 57 for (table = this; table; table = table->next) | |
| 58 { | |
| 59 if (table->namelen == namelen && | |
| 60 memcmp(table->name, name, namelen) == 0) | |
| 61 { | |
| 62 //printf("\tfound %d\n", table->textlen); | |
| 63 break; | |
| 64 } | |
| 65 } | |
| 66 return table; | |
| 67 } | |
| 68 | |
| 69 Macro *Macro::define(Macro **ptable, unsigned char *name, size_t namelen, unsigned char *text, size_t textlen) | |
| 70 { | |
| 71 //printf("Macro::define('%.*s' = '%.*s')\n", namelen, name, textlen, text); | |
| 72 | |
| 73 Macro *table; | |
| 74 | |
| 75 //assert(ptable); | |
| 76 for (table = *ptable; table; table = table->next) | |
| 77 { | |
| 78 if (table->namelen == namelen && | |
| 79 memcmp(table->name, name, namelen) == 0) | |
| 80 { | |
| 81 table->text = text; | |
| 82 table->textlen = textlen; | |
| 83 return table; | |
| 84 } | |
| 85 } | |
| 86 table = new Macro(name, namelen, text, textlen); | |
| 87 table->next = *ptable; | |
| 88 *ptable = table; | |
| 89 return table; | |
| 90 } | |
| 91 | |
| 92 /********************************************************** | |
| 93 * Given buffer p[0..end], extract argument marg[0..marglen]. | |
| 94 * Params: | |
| 95 * n 0: get entire argument | |
| 96 * 1..9: get nth argument | |
| 97 * -1: get 2nd through end | |
| 98 */ | |
| 99 | |
| 100 unsigned extractArgN(unsigned char *p, unsigned end, unsigned char **pmarg, unsigned *pmarglen, int n) | |
| 101 { | |
| 102 /* Scan forward for matching right parenthesis. | |
| 103 * Nest parentheses. | |
| 104 * Skip over $( and $) | |
| 105 * Skip over "..." and '...' strings inside HTML tags. | |
| 106 * Skip over <!-- ... --> comments. | |
| 107 * Skip over previous macro insertions | |
| 108 * Set marglen. | |
| 109 */ | |
| 110 unsigned parens = 1; | |
| 111 unsigned char instring = 0; | |
| 112 unsigned incomment = 0; | |
| 113 unsigned intag = 0; | |
| 114 unsigned inexp = 0; | |
| 115 unsigned argn = 0; | |
| 116 | |
| 117 unsigned v = 0; | |
| 118 | |
| 119 Largstart: | |
| 120 #if 1 | |
| 121 // Skip first space, if any, to find the start of the macro argument | |
| 122 if (v < end && isspace(p[v])) | |
| 123 v++; | |
| 124 #else | |
| 125 // Skip past spaces to find the start of the macro argument | |
| 126 for (; v < end && isspace(p[v]); v++) | |
| 127 ; | |
| 128 #endif | |
| 129 *pmarg = p + v; | |
| 130 | |
| 131 for (; v < end; v++) | |
| 132 { unsigned char c = p[v]; | |
| 133 | |
| 134 switch (c) | |
| 135 { | |
| 136 case ',': | |
| 137 if (!inexp && !instring && !incomment && parens == 1) | |
| 138 { | |
| 139 argn++; | |
| 140 if (argn == 1 && n == -1) | |
| 141 { v++; | |
| 142 goto Largstart; | |
| 143 } | |
| 144 if (argn == n) | |
| 145 break; | |
| 146 if (argn + 1 == n) | |
| 147 { v++; | |
| 148 goto Largstart; | |
| 149 } | |
| 150 } | |
| 151 continue; | |
| 152 | |
| 153 case '(': | |
| 154 if (!inexp && !instring && !incomment) | |
| 155 parens++; | |
| 156 continue; | |
| 157 | |
| 158 case ')': | |
| 159 if (!inexp && !instring && !incomment && --parens == 0) | |
| 160 { | |
| 161 break; | |
| 162 } | |
| 163 continue; | |
| 164 | |
| 165 case '"': | |
| 166 case '\'': | |
| 167 if (!inexp && !incomment && intag) | |
| 168 { | |
| 169 if (c == instring) | |
| 170 instring = 0; | |
| 171 else if (!instring) | |
| 172 instring = c; | |
| 173 } | |
| 174 continue; | |
| 175 | |
| 176 case '<': | |
| 177 if (!inexp && !instring && !incomment) | |
| 178 { | |
| 179 if (v + 6 < end && | |
| 180 p[v + 1] == '!' && | |
| 181 p[v + 2] == '-' && | |
| 182 p[v + 3] == '-') | |
| 183 { | |
| 184 incomment = 1; | |
| 185 v += 3; | |
| 186 } | |
| 187 else if (v + 2 < end && | |
| 188 isalpha(p[v + 1])) | |
| 189 intag = 1; | |
| 190 } | |
| 191 continue; | |
| 192 | |
| 193 case '>': | |
| 194 if (!inexp) | |
| 195 intag = 0; | |
| 196 continue; | |
| 197 | |
| 198 case '-': | |
| 199 if (!inexp && | |
| 200 !instring && | |
| 201 incomment && | |
| 202 v + 2 < end && | |
| 203 p[v + 1] == '-' && | |
| 204 p[v + 2] == '>') | |
| 205 { | |
| 206 incomment = 0; | |
| 207 v += 2; | |
| 208 } | |
| 209 continue; | |
| 210 | |
| 211 case 0xFF: | |
| 212 if (v + 1 < end) | |
| 213 { | |
| 214 if (p[v + 1] == '{') | |
| 215 inexp++; | |
| 216 else if (p[v + 1] == '}') | |
| 217 inexp--; | |
| 218 } | |
| 219 continue; | |
| 220 | |
| 221 default: | |
| 222 continue; | |
| 223 } | |
| 224 break; | |
| 225 } | |
| 226 if (argn == 0 && n == -1) | |
| 227 *pmarg = p + v; | |
| 228 *pmarglen = p + v - *pmarg; | |
| 229 //printf("extractArg%d('%.*s') = '%.*s'\n", n, end, p, *pmarglen, *pmarg); | |
| 230 return v; | |
| 231 } | |
| 232 | |
| 233 | |
| 234 /***************************************************** | |
| 235 * Expand macro in place in buf. | |
| 236 * Only look at the text in buf from start to end. | |
| 237 */ | |
| 238 | |
| 239 void Macro::expand(OutBuffer *buf, unsigned start, unsigned *pend, | |
| 240 unsigned char *arg, unsigned arglen) | |
| 241 { | |
| 242 #if 0 | |
| 243 printf("Macro::expand(buf[%d..%d], arg = '%.*s')\n", start, *pend, arglen, arg); | |
| 244 printf("Buf is: '%.*s'\n", *pend - start, buf->data + start); | |
| 245 #endif | |
| 246 | |
| 247 static int nest; | |
| 248 if (nest > 100) // limit recursive expansion | |
| 249 return; | |
| 250 nest++; | |
| 251 | |
| 252 unsigned end = *pend; | |
| 253 assert(start <= end); | |
| 254 assert(end <= buf->offset); | |
| 255 | |
| 256 /* First pass - replace $0 | |
| 257 */ | |
| 258 arg = memdup(arg, arglen); | |
| 259 for (unsigned u = start; u + 1 < end; ) | |
| 260 { | |
| 261 unsigned char *p = buf->data; // buf->data is not loop invariant | |
| 262 | |
| 263 /* Look for $0, but not $$0, and replace it with arg. | |
| 264 */ | |
| 265 if (p[u] == '$' && (isdigit(p[u + 1]) || p[u + 1] == '+')) | |
| 266 { | |
| 267 if (u > start && p[u - 1] == '$') | |
| 268 { // Don't expand $$0, but replace it with $0 | |
| 269 buf->remove(u - 1, 1); | |
| 270 end--; | |
| 271 u += 1; // now u is one past the closing '1' | |
| 272 continue; | |
| 273 } | |
| 274 | |
| 275 unsigned char c = p[u + 1]; | |
| 276 int n = (c == '+') ? -1 : c - '0'; | |
| 277 | |
| 278 unsigned char *marg; | |
| 279 unsigned marglen; | |
| 280 extractArgN(arg, arglen, &marg, &marglen, n); | |
| 281 if (marglen == 0) | |
| 282 { // Just remove macro invocation | |
| 283 //printf("Replacing '$%c' with '%.*s'\n", p[u + 1], marglen, marg); | |
| 284 buf->remove(u, 2); | |
| 285 end -= 2; | |
| 286 } | |
| 287 else if (c == '+') | |
| 288 { | |
| 289 // Replace '$+' with 'arg' | |
| 290 //printf("Replacing '$%c' with '%.*s'\n", p[u + 1], marglen, marg); | |
| 291 buf->remove(u, 2); | |
| 292 buf->insert(u, marg, marglen); | |
| 293 end += marglen - 2; | |
| 294 | |
| 295 // Scan replaced text for further expansion | |
| 296 unsigned mend = u + marglen; | |
| 297 expand(buf, u, &mend, NULL, 0); | |
| 298 end += mend - (u + marglen); | |
| 299 u = mend; | |
| 300 } | |
| 301 else | |
| 302 { | |
| 303 // Replace '$1' with '\xFF{arg\xFF}' | |
| 304 //printf("Replacing '$%c' with '\xFF{%.*s\xFF}'\n", p[u + 1], marglen, marg); | |
| 305 buf->data[u] = 0xFF; | |
| 306 buf->data[u + 1] = '{'; | |
| 307 buf->insert(u + 2, marg, marglen); | |
| 308 buf->insert(u + 2 + marglen, "\xFF}", 2); | |
| 309 end += -2 + 2 + marglen + 2; | |
| 310 | |
| 311 // Scan replaced text for further expansion | |
| 312 unsigned mend = u + 2 + marglen; | |
| 313 expand(buf, u + 2, &mend, NULL, 0); | |
| 314 end += mend - (u + 2 + marglen); | |
| 315 u = mend; | |
| 316 } | |
| 317 //printf("u = %d, end = %d\n", u, end); | |
| 318 //printf("#%.*s#\n", end, &buf->data[0]); | |
| 319 continue; | |
| 320 } | |
| 321 | |
| 322 u++; | |
| 323 } | |
| 324 | |
| 325 /* Second pass - replace other macros | |
| 326 */ | |
| 327 for (unsigned u = start; u + 4 < end; ) | |
| 328 { | |
| 329 unsigned char *p = buf->data; // buf->data is not loop invariant | |
| 330 | |
| 331 /* A valid start of macro expansion is $(c, where c is | |
| 332 * an id start character, and not $$(c. | |
| 333 */ | |
| 334 if (p[u] == '$' && p[u + 1] == '(' && isidstart(p[u + 2])) | |
| 335 { | |
| 336 //printf("\tfound macro start '%c'\n", p[u + 2]); | |
| 337 unsigned char *name = p + u + 2; | |
| 338 unsigned namelen = 0; | |
| 339 | |
| 340 unsigned char *marg; | |
| 341 unsigned marglen; | |
| 342 | |
| 343 unsigned v; | |
| 344 /* Scan forward to find end of macro name and | |
| 345 * beginning of macro argument (marg). | |
| 346 */ | |
| 347 for (v = u + 2; v < end; v++) | |
| 348 { unsigned char c = p[v]; | |
| 349 | |
| 350 if (!isidchar(c)) | |
| 351 { // We've gone past the end of the macro name. | |
| 352 namelen = v - (u + 2); | |
| 353 break; | |
| 354 } | |
| 355 } | |
| 356 | |
| 357 v += extractArgN(p + v, end - v, &marg, &marglen, 0); | |
| 358 assert(v <= end); | |
| 359 | |
| 360 if (v < end) | |
| 361 { // v is on the closing ')' | |
| 362 if (u > start && p[u - 1] == '$') | |
| 363 { // Don't expand $$(NAME), but replace it with $(NAME) | |
| 364 buf->remove(u - 1, 1); | |
| 365 end--; | |
| 366 u = v; // now u is one past the closing ')' | |
| 367 continue; | |
| 368 } | |
| 369 | |
| 370 Macro *m = search(name, namelen); | |
| 371 if (m) | |
| 372 { | |
| 373 #if 0 | |
| 374 if (m->textlen && m->text[0] == ' ') | |
| 375 { m->text++; | |
| 376 m->textlen--; | |
| 377 } | |
| 378 #endif | |
| 379 if (m->inuse && marglen == 0) | |
| 380 { // Remove macro invocation | |
| 381 buf->remove(u, v + 1 - u); | |
| 382 end -= v + 1 - u; | |
| 383 } | |
| 384 else if (m->inuse && arglen == marglen && memcmp(arg, marg, arglen) == 0) | |
| 385 { // Recursive expansion; just leave in place | |
| 386 | |
| 387 } | |
| 388 else | |
| 389 { | |
| 390 //printf("\tmacro '%.*s'(%.*s) = '%.*s'\n", m->namelen, m->name, marglen, marg, m->textlen, m->text); | |
| 391 #if 1 | |
| 392 marg = memdup(marg, marglen); | |
| 393 // Insert replacement text | |
| 394 buf->spread(v + 1, 2 + m->textlen + 2); | |
| 395 buf->data[v + 1] = 0xFF; | |
| 396 buf->data[v + 2] = '{'; | |
| 397 memcpy(buf->data + v + 3, m->text, m->textlen); | |
| 398 buf->data[v + 3 + m->textlen] = 0xFF; | |
| 399 buf->data[v + 3 + m->textlen + 1] = '}'; | |
| 400 | |
| 401 end += 2 + m->textlen + 2; | |
| 402 | |
| 403 // Scan replaced text for further expansion | |
| 404 m->inuse++; | |
| 405 unsigned mend = v + 1 + 2+m->textlen+2; | |
| 406 expand(buf, v + 1, &mend, marg, marglen); | |
| 407 end += mend - (v + 1 + 2+m->textlen+2); | |
| 408 m->inuse--; | |
| 409 | |
| 410 buf->remove(u, v + 1 - u); | |
| 411 end -= v + 1 - u; | |
| 412 u += mend - (v + 1); | |
| 413 #else | |
| 414 // Insert replacement text | |
| 415 buf->insert(v + 1, m->text, m->textlen); | |
| 416 end += m->textlen; | |
| 417 | |
| 418 // Scan replaced text for further expansion | |
| 419 m->inuse++; | |
| 420 unsigned mend = v + 1 + m->textlen; | |
| 421 expand(buf, v + 1, &mend, marg, marglen); | |
| 422 end += mend - (v + 1 + m->textlen); | |
| 423 m->inuse--; | |
| 424 | |
| 425 buf->remove(u, v + 1 - u); | |
| 426 end -= v + 1 - u; | |
| 427 u += mend - (v + 1); | |
| 428 #endif | |
| 429 mem.free(marg); | |
| 430 //printf("u = %d, end = %d\n", u, end); | |
| 431 //printf("#%.*s#\n", end - u, &buf->data[u]); | |
| 432 continue; | |
| 433 } | |
| 434 } | |
| 435 else | |
| 436 { | |
| 437 // Replace $(NAME) with nothing | |
| 438 buf->remove(u, v + 1 - u); | |
| 439 end -= (v + 1 - u); | |
| 440 continue; | |
| 441 } | |
| 442 } | |
| 443 } | |
| 444 u++; | |
| 445 } | |
| 446 mem.free(arg); | |
| 447 *pend = end; | |
| 448 nest--; | |
| 449 } |