| | 1 | #ifdef RCSID |
| | 2 | static char RCSid[] = |
| | 3 | "$Header: d:/cvsroot/tads/TADS2/TOK.C,v 1.3 1999/07/11 00:46:30 MJRoberts Exp $"; |
| | 4 | #endif |
| | 5 | |
| | 6 | /* |
| | 7 | * Copyright (c) 1987, 1988 by Michael J. Roberts. All Rights Reserved. |
| | 8 | * |
| | 9 | * Please see the accompanying license file, LICENSE.TXT, for information |
| | 10 | * on using and copying this software. |
| | 11 | */ |
| | 12 | /* |
| | 13 | Name |
| | 14 | tok - tokenizer |
| | 15 | Function |
| | 16 | Tokenizes an input stream |
| | 17 | Notes |
| | 18 | None |
| | 19 | Modified |
| | 20 | 04/11/99 CNebel - Enforce identifier length limits. |
| | 21 | 08/14/91 MJRoberts - created |
| | 22 | */ |
| | 23 | |
| | 24 | #include "os.h" |
| | 25 | #include "std.h" |
| | 26 | #include "mch.h" |
| | 27 | #include "mcm.h" |
| | 28 | #include "tok.h" |
| | 29 | #include "lin.h" |
| | 30 | #include "linf.h" |
| | 31 | #include "dbg.h" |
| | 32 | #include <ctype.h> |
| | 33 | #include <string.h> |
| | 34 | #include <stdlib.h> |
| | 35 | |
| | 36 | |
| | 37 | /* special temporary buffers for <<expr>> macro expansion */ |
| | 38 | static char tokmac1[] = ",say(("; |
| | 39 | static char tokmac1s[] = "(say(("; |
| | 40 | static char tokmac2[] = "),nil),\""; |
| | 41 | static char tokmac3[] = "),nil))"; |
| | 42 | static char tokmac4[] = ")"; |
| | 43 | |
| | 44 | /* forward definition of static functions */ |
| | 45 | static int tokdfhsh(char *sym, int len); |
| | 46 | |
| | 47 | |
| | 48 | /* find a #define symbol */ |
| | 49 | static tokdfdef *tok_find_define(tokcxdef *ctx, char *sym, int len) |
| | 50 | { |
| | 51 | int hsh; |
| | 52 | tokdfdef *df; |
| | 53 | |
| | 54 | /* find the appropriate chain the hash table */ |
| | 55 | hsh = tokdfhsh(sym, len); |
| | 56 | |
| | 57 | /* search the chain for this symbol */ |
| | 58 | for (df = ctx->tokcxdf[hsh] ; df ; df = df->nxt) |
| | 59 | { |
| | 60 | /* if this one matches, return it */ |
| | 61 | if (df->len == len && !memcmp(df->nm, sym, (size_t)len)) |
| | 62 | { |
| | 63 | /* fix it up if it's the special __FILE__ or __LINE__ symbol */ |
| | 64 | if (len == 8) |
| | 65 | { |
| | 66 | if (!memcmp(sym, "__FILE__", (size_t)8)) |
| | 67 | { |
| | 68 | size_t len; |
| | 69 | |
| | 70 | /* |
| | 71 | * put in the opening single quote, since we want |
| | 72 | * the expanded result to be a string |
| | 73 | */ |
| | 74 | df->expan[0] = '\''; |
| | 75 | |
| | 76 | /* get the name */ |
| | 77 | linnam(ctx->tokcxlin, df->expan+1); |
| | 78 | |
| | 79 | /* get the length, and add the closing quote */ |
| | 80 | len = strlen(df->expan); |
| | 81 | df->expan[len] = '\''; |
| | 82 | |
| | 83 | /* |
| | 84 | * set the length of the expansion, including the |
| | 85 | * quotes (the first quote was measured in the |
| | 86 | * length originally, but the second quote hasn't |
| | 87 | * been counted yet, so add one to our original |
| | 88 | * length) |
| | 89 | */ |
| | 90 | df->explen = (int)len + 1; |
| | 91 | |
| | 92 | /* if the expansion is too long, it's an error */ |
| | 93 | if (df->explen >= OSFNMAX) |
| | 94 | errsig(ctx->tokcxerr, ERR_LONG_FILE_MACRO); |
| | 95 | } |
| | 96 | else if (!memcmp(sym, "__LINE__", (size_t)8)) |
| | 97 | { |
| | 98 | ulong l; |
| | 99 | |
| | 100 | /* get the line number */ |
| | 101 | l = linlnum(ctx->tokcxlin); |
| | 102 | |
| | 103 | /* convert it to a textual format for the expansion */ |
| | 104 | sprintf(df->expan, "%lu", l); |
| | 105 | |
| | 106 | /* set the expanded value's length */ |
| | 107 | df->explen = strlen(df->expan); |
| | 108 | |
| | 109 | /* make sure the expansion isn't too long */ |
| | 110 | if (df->explen >= 40) |
| | 111 | errsig(ctx->tokcxerr, ERR_LONG_LINE_MACRO); |
| | 112 | } |
| | 113 | } |
| | 114 | |
| | 115 | /* return it */ |
| | 116 | return df; |
| | 117 | } |
| | 118 | } |
| | 119 | |
| | 120 | /* didn't find anything */ |
| | 121 | return 0; |
| | 122 | } |
| | 123 | |
| | 124 | /* |
| | 125 | * Write preprocessor state to a file |
| | 126 | */ |
| | 127 | void tok_write_defines(tokcxdef *ctx, osfildef *fp, errcxdef *ec) |
| | 128 | { |
| | 129 | int i; |
| | 130 | tokdfdef **dfp; |
| | 131 | tokdfdef *df; |
| | 132 | char buf[4]; |
| | 133 | |
| | 134 | /* write each element of the hash chains */ |
| | 135 | for (i = TOKDFHSHSIZ, dfp = ctx->tokcxdf ; i ; ++dfp, --i) |
| | 136 | { |
| | 137 | /* write each entry in this hash chain */ |
| | 138 | for (df = *dfp ; df ; df = df->nxt) |
| | 139 | { |
| | 140 | /* write this entry */ |
| | 141 | oswp2(buf, df->len); |
| | 142 | oswp2(buf + 2, df->explen); |
| | 143 | if (osfwb(fp, buf, 4) |
| | 144 | || osfwb(fp, df->nm, df->len) |
| | 145 | || (df->explen != 0 && osfwb(fp, df->expan, df->explen))) |
| | 146 | errsig(ec, ERR_WRTGAM); |
| | 147 | } |
| | 148 | |
| | 149 | /* write a zero-length entry to indicate the end of this chain */ |
| | 150 | oswp2(buf, 0); |
| | 151 | if (osfwb(fp, buf, 4)) errsig(ec, ERR_WRTGAM); |
| | 152 | } |
| | 153 | } |
| | 154 | |
| | 155 | /* |
| | 156 | * Read preprocessor state from a file |
| | 157 | */ |
| | 158 | void tok_read_defines(tokcxdef *ctx, osfildef *fp, errcxdef *ec) |
| | 159 | { |
| | 160 | int i; |
| | 161 | tokdfdef **dfp; |
| | 162 | tokdfdef *df; |
| | 163 | char buf[4]; |
| | 164 | |
| | 165 | /* write each element of the hash chains */ |
| | 166 | for (i = TOKDFHSHSIZ, dfp = ctx->tokcxdf ; i ; ++dfp, --i) |
| | 167 | { |
| | 168 | /* read this hash chain */ |
| | 169 | for (;;) |
| | 170 | { |
| | 171 | /* read the next entry's header, and stop if this is the end */ |
| | 172 | if (osfrb(fp, buf, 4)) errsig(ec, ERR_RDGAM); |
| | 173 | if (osrp2(buf) == 0) break; |
| | 174 | |
| | 175 | /* set up a new symbol of the appropriate size */ |
| | 176 | df = (tokdfdef *)mchalo(ec, |
| | 177 | (ushort)(sizeof(tokdfdef) + osrp2(buf) |
| | 178 | + osrp2(buf+2) - 1), |
| | 179 | "tok_read_defines"); |
| | 180 | df->explen = osrp2(buf+2); |
| | 181 | df->nm = df->expan + df->explen; |
| | 182 | df->len = osrp2(buf); |
| | 183 | |
| | 184 | /* read the rest of the symbol */ |
| | 185 | if (osfrb(fp, df->nm, df->len) |
| | 186 | || (df->explen != 0 && osfrb(fp, df->expan, df->explen))) |
| | 187 | errsig(ec, ERR_RDGAM); |
| | 188 | |
| | 189 | /* |
| | 190 | * If a symbol with this name already exists in the table, |
| | 191 | * discard the new one -- the symbols defined by -D and the |
| | 192 | * current set of built-in symbols takes precedence over the |
| | 193 | * set loaded from the file. |
| | 194 | */ |
| | 195 | if (tok_find_define(ctx, df->nm, df->len)) |
| | 196 | { |
| | 197 | /* simply discard this symbol */ |
| | 198 | mchfre(df); |
| | 199 | } |
| | 200 | else |
| | 201 | { |
| | 202 | /* link it into this hash chain */ |
| | 203 | df->nxt = *dfp; |
| | 204 | *dfp = df; |
| | 205 | } |
| | 206 | } |
| | 207 | } |
| | 208 | } |
| | 209 | |
| | 210 | |
| | 211 | |
| | 212 | /* compute a #define symbol's hash value */ |
| | 213 | static int tokdfhsh(char *sym, int len) |
| | 214 | { |
| | 215 | uint hsh; |
| | 216 | |
| | 217 | for (hsh = 0 ; len ; ++sym, --len) |
| | 218 | hsh = (hsh + *sym) & TOKDFHSHMASK; |
| | 219 | return hsh; |
| | 220 | } |
| | 221 | |
| | 222 | /* convert a #define symbol to lower case if folding case */ |
| | 223 | static char *tok_casefold_defsym(tokcxdef *ctx, char *outbuf, |
| | 224 | char *src, int len) |
| | 225 | { |
| | 226 | if (ctx->tokcxflg & TOKCXCASEFOLD) |
| | 227 | { |
| | 228 | char *dst; |
| | 229 | int rem; |
| | 230 | |
| | 231 | /* make a lower-case copy of the symbol */ |
| | 232 | rem = (len > TOKNAMMAX ? TOKNAMMAX : len); |
| | 233 | for (dst = outbuf ; rem > 0 ; ++dst, ++src, --rem) |
| | 234 | *dst = (isupper((uchar)*src) ? tolower((uchar)*src) : *src); |
| | 235 | |
| | 236 | /* use the lower-case copy instead of the original */ |
| | 237 | return outbuf; |
| | 238 | } |
| | 239 | else |
| | 240 | { |
| | 241 | /* return the original unchanged */ |
| | 242 | return src; |
| | 243 | } |
| | 244 | } |
| | 245 | |
| | 246 | /* |
| | 247 | * convert a token to lower-case if we're folding case |
| | 248 | */ |
| | 249 | void tok_case_fold(tokcxdef *ctx, tokdef *tok) |
| | 250 | { |
| | 251 | /* if we're in case-insensitive mode, convert the token to lower-case */ |
| | 252 | if (ctx->tokcxflg & TOKCXCASEFOLD) |
| | 253 | { |
| | 254 | char *p; |
| | 255 | int len; |
| | 256 | |
| | 257 | /* convert each character in the token to lower-case */ |
| | 258 | for (p = tok->toknam, len = tok->toklen ; len != 0 ; ++p, --len) |
| | 259 | { |
| | 260 | /* if this character is upper-case, convert it to lower-case */ |
| | 261 | if (isupper((uchar)*p)) |
| | 262 | *p = tolower((uchar)*p); |
| | 263 | } |
| | 264 | } |
| | 265 | } |
| | 266 | |
| | 267 | /* add a symbol to the #define symbol table, folding case if necessary */ |
| | 268 | void tok_add_define_cvtcase(tokcxdef *ctx, char *sym, int len, |
| | 269 | char *expan, int explen) |
| | 270 | { |
| | 271 | char mysym[TOKNAMMAX]; |
| | 272 | |
| | 273 | /* convert to lower-case if necessary */ |
| | 274 | sym = tok_casefold_defsym(ctx, mysym, sym, len); |
| | 275 | |
| | 276 | /* add the symbol */ |
| | 277 | tok_add_define(ctx, sym, len, expan, explen); |
| | 278 | } |
| | 279 | |
| | 280 | /* add a symbol to the #define symbol table */ |
| | 281 | void tok_add_define(tokcxdef *ctx, char *sym, int len, |
| | 282 | char *expan, int explen) |
| | 283 | { |
| | 284 | int hsh; |
| | 285 | tokdfdef *df; |
| | 286 | |
| | 287 | /* if it's already defined, ignore it */ |
| | 288 | if (tok_find_define(ctx, sym, len)) |
| | 289 | return; |
| | 290 | |
| | 291 | /* find the appropriate entry in the hash table */ |
| | 292 | hsh = tokdfhsh(sym, len); |
| | 293 | |
| | 294 | /* allocate space for the symbol */ |
| | 295 | df = (tokdfdef *)mchalo(ctx->tokcxerr, |
| | 296 | (ushort)(sizeof(tokdfdef) + len + explen - 1), |
| | 297 | "tok_add_define"); |
| | 298 | |
| | 299 | /* set up the new symbol */ |
| | 300 | df->nm = df->expan + explen; |
| | 301 | df->len = len; |
| | 302 | df->explen = explen; |
| | 303 | memcpy(df->expan, expan, explen); |
| | 304 | memcpy(df->nm, sym, len); |
| | 305 | |
| | 306 | /* link it into the hash chain */ |
| | 307 | df->nxt = ctx->tokcxdf[hsh]; |
| | 308 | ctx->tokcxdf[hsh] = df; |
| | 309 | } |
| | 310 | |
| | 311 | /* add a #define symbol with a numeric value */ |
| | 312 | void tok_add_define_num_cvtcase(tokcxdef *ctx, char *sym, int len, int num) |
| | 313 | { |
| | 314 | char buf[20]; |
| | 315 | |
| | 316 | /* convert the value to a string */ |
| | 317 | sprintf(buf, "%d", num); |
| | 318 | |
| | 319 | /* add the text value */ |
| | 320 | tok_add_define_cvtcase(ctx, sym, len, buf, strlen(buf)); |
| | 321 | } |
| | 322 | |
| | 323 | /* undefine a #define symbol */ |
| | 324 | void tok_del_define(tokcxdef *ctx, char *sym, int len) |
| | 325 | { |
| | 326 | int hsh; |
| | 327 | tokdfdef *df; |
| | 328 | tokdfdef *prv; |
| | 329 | |
| | 330 | /* find the appropriate chain the hash table */ |
| | 331 | hsh = tokdfhsh(sym, len); |
| | 332 | |
| | 333 | /* search the chain for this symbol */ |
| | 334 | for (prv = 0, df = ctx->tokcxdf[hsh] ; df ; prv = df, df = df->nxt) |
| | 335 | { |
| | 336 | /* if this one matches, delete it */ |
| | 337 | if (df->len == len && !memcmp(df->nm, sym, (size_t)len)) |
| | 338 | { |
| | 339 | /* unlink it from the chain */ |
| | 340 | if (prv) |
| | 341 | prv->nxt = df->nxt; |
| | 342 | else |
| | 343 | ctx->tokcxdf[hsh] = df->nxt; |
| | 344 | |
| | 345 | /* delete this symbol, and we're done */ |
| | 346 | mchfre(df); |
| | 347 | break; |
| | 348 | } |
| | 349 | } |
| | 350 | } |
| | 351 | |
| | 352 | /* scan a #define symbol to see how long it is */ |
| | 353 | static int tok_scan_defsym(tokcxdef *ctx, char *p, int len) |
| | 354 | { |
| | 355 | int symlen; |
| | 356 | |
| | 357 | /* make sure it's a valid symbol */ |
| | 358 | if (!(isalpha((uchar)*p) || *p == '_' || *p == '$')) |
| | 359 | { |
| | 360 | errlog(ctx->tokcxerr, ERR_REQSYM); |
| | 361 | return 0; |
| | 362 | } |
| | 363 | |
| | 364 | /* count characters as long as we have valid symbol characters */ |
| | 365 | for (symlen = 0 ; len && TOKISSYM(*p) ; ++p, --len, ++symlen) ; |
| | 366 | return symlen; |
| | 367 | } |
| | 368 | |
| | 369 | /* process a #define */ |
| | 370 | static void tokdefine(tokcxdef *ctx, char *p, int len) |
| | 371 | { |
| | 372 | char *sym; |
| | 373 | int symlen; |
| | 374 | char *expan; |
| | 375 | char mysym[TOKNAMMAX]; |
| | 376 | |
| | 377 | /* get the symbol */ |
| | 378 | sym = p; |
| | 379 | if (!(symlen = tok_scan_defsym(ctx, p, len))) |
| | 380 | return; |
| | 381 | |
| | 382 | /* if it's already in the table, log an error */ |
| | 383 | if (tok_find_define(ctx, sym, symlen)) |
| | 384 | { |
| | 385 | errlog(ctx->tokcxerr, ERR_DEFREDEF); |
| | 386 | return; |
| | 387 | } |
| | 388 | |
| | 389 | /* skip whitespace following the symbol */ |
| | 390 | expan = sym + symlen; |
| | 391 | len -= symlen; |
| | 392 | while (len && t_isspace(*expan)) --len, ++expan; |
| | 393 | |
| | 394 | /* if we're folding case, convert the symbol to lower case */ |
| | 395 | sym = tok_casefold_defsym(ctx, mysym, sym, symlen); |
| | 396 | |
| | 397 | /* define the symbol */ |
| | 398 | tok_add_define(ctx, sym, symlen, expan, len); |
| | 399 | } |
| | 400 | |
| | 401 | /* |
| | 402 | * Update the #if status for the current nesting. Any enclosing |
| | 403 | * negative #if will override everything inside, so we need to look |
| | 404 | * through the nesting from the outside in until we either determine |
| | 405 | * that everything is affirmative or we find a negative anywhere in the |
| | 406 | * nesting. |
| | 407 | */ |
| | 408 | static void tok_update_if_stat(tokcxdef *ctx) |
| | 409 | { |
| | 410 | int i; |
| | 411 | |
| | 412 | /* look through nesting from the outermost level */ |
| | 413 | for (i = 0 ; i < ctx->tokcxifcnt ; ++i) |
| | 414 | { |
| | 415 | /* assume this level will apply to everything inside */ |
| | 416 | ctx->tokcxifcur = ctx->tokcxif[i]; |
| | 417 | |
| | 418 | /* if this level is off, everything inside is off */ |
| | 419 | switch (ctx->tokcxif[i]) |
| | 420 | { |
| | 421 | case TOKIF_IF_NO: |
| | 422 | case TOKIF_ELSE_NO: |
| | 423 | /* |
| | 424 | * this level is off, hence everything inside is off -- stop |
| | 425 | * here with the current (negative) determination |
| | 426 | */ |
| | 427 | return; |
| | 428 | |
| | 429 | default: |
| | 430 | /* so far we're in the "on" section, so keep looking */ |
| | 431 | break; |
| | 432 | } |
| | 433 | } |
| | 434 | } |
| | 435 | |
| | 436 | /* process an #ifdef or a #ifndef */ |
| | 437 | static void tok_ifdef_ifndef(tokcxdef *ctx, char *p, int len, int is_ifdef) |
| | 438 | { |
| | 439 | int symlen; |
| | 440 | char *sym; |
| | 441 | int stat; |
| | 442 | int found; |
| | 443 | char mysym[TOKNAMMAX]; |
| | 444 | |
| | 445 | /* get the symbol */ |
| | 446 | sym = p; |
| | 447 | if (!(symlen = tok_scan_defsym(ctx, p, len))) |
| | 448 | return; |
| | 449 | |
| | 450 | /* if we're folding case, convert the symbol to lower case */ |
| | 451 | sym = tok_casefold_defsym(ctx, mysym, sym, symlen); |
| | 452 | |
| | 453 | /* see if we can find it in the table, and set the status accordingly */ |
| | 454 | found = (tok_find_define(ctx, sym, symlen) != 0); |
| | 455 | |
| | 456 | /* invert the test if this is an ifndef */ |
| | 457 | if (!is_ifdef) found = !found; |
| | 458 | |
| | 459 | /* set the #if status accordingly */ |
| | 460 | if (found) |
| | 461 | stat = TOKIF_IF_YES; |
| | 462 | else |
| | 463 | stat = TOKIF_IF_NO; |
| | 464 | ctx->tokcxif[ctx->tokcxifcnt] = stat; |
| | 465 | |
| | 466 | /* allocate a new #if level (making sure we have room) */ |
| | 467 | if (ctx->tokcxifcnt >= TOKIFNEST) |
| | 468 | { |
| | 469 | errlog(ctx->tokcxerr, ERR_MANYPIF); |
| | 470 | return; |
| | 471 | } |
| | 472 | ctx->tokcxifcnt++; |
| | 473 | |
| | 474 | /* update the current status */ |
| | 475 | tok_update_if_stat(ctx); |
| | 476 | } |
| | 477 | |
| | 478 | /* process a #error */ |
| | 479 | static void tok_p_error(tokcxdef *ctx, char *p, int len) |
| | 480 | { |
| | 481 | errlog1(ctx->tokcxerr, ERR_P_ERROR, |
| | 482 | ERRTSTR, errstr(ctx->tokcxerr, p, len)); |
| | 483 | } |
| | 484 | |
| | 485 | /* process a #ifdef */ |
| | 486 | static void tokifdef(tokcxdef *ctx, char *p, int len) |
| | 487 | { |
| | 488 | tok_ifdef_ifndef(ctx, p, len, TRUE); |
| | 489 | } |
| | 490 | |
| | 491 | /* process a #ifndef */ |
| | 492 | static void tokifndef(tokcxdef *ctx, char *p, int len) |
| | 493 | { |
| | 494 | tok_ifdef_ifndef(ctx, p, len, FALSE); |
| | 495 | } |
| | 496 | |
| | 497 | /* process a #if */ |
| | 498 | static void tokif(tokcxdef *ctx, char *p, int len) |
| | 499 | { |
| | 500 | errsig(ctx->tokcxerr, ERR_PIF_NA); |
| | 501 | } |
| | 502 | |
| | 503 | /* process a #elif */ |
| | 504 | static void tokelif(tokcxdef *ctx, char *p, int len) |
| | 505 | { |
| | 506 | errsig(ctx->tokcxerr, ERR_PELIF_NA); |
| | 507 | } |
| | 508 | |
| | 509 | /* process a #else */ |
| | 510 | static void tokelse(tokcxdef *ctx, char *p, int len) |
| | 511 | { |
| | 512 | int cnt; |
| | 513 | |
| | 514 | /* if we're not expecting #else, it's an error */ |
| | 515 | cnt = ctx->tokcxifcnt; |
| | 516 | if (cnt == 0 || ctx->tokcxif[cnt-1] == TOKIF_ELSE_YES |
| | 517 | || ctx->tokcxif[cnt-1] == TOKIF_ELSE_NO) |
| | 518 | { |
| | 519 | errlog(ctx->tokcxerr, ERR_BADPELSE); |
| | 520 | return; |
| | 521 | } |
| | 522 | |
| | 523 | /* switch to the appropriate #else state (opposite the #if state) */ |
| | 524 | if (ctx->tokcxif[cnt-1] == TOKIF_IF_YES) |
| | 525 | ctx->tokcxif[cnt-1] = TOKIF_ELSE_NO; |
| | 526 | else |
| | 527 | ctx->tokcxif[cnt-1] = TOKIF_ELSE_YES; |
| | 528 | |
| | 529 | /* update the current status */ |
| | 530 | tok_update_if_stat(ctx); |
| | 531 | } |
| | 532 | |
| | 533 | /* process a #endif */ |
| | 534 | static void tokendif(tokcxdef *ctx, char *p, int len) |
| | 535 | { |
| | 536 | /* if we're not expecting #endif, it's an error */ |
| | 537 | if (ctx->tokcxifcnt == 0) |
| | 538 | { |
| | 539 | errlog(ctx->tokcxerr, ERR_BADENDIF); |
| | 540 | return; |
| | 541 | } |
| | 542 | |
| | 543 | /* remove the #if level */ |
| | 544 | ctx->tokcxifcnt--; |
| | 545 | |
| | 546 | /* update the current status */ |
| | 547 | tok_update_if_stat(ctx); |
| | 548 | } |
| | 549 | |
| | 550 | /* process a #undef */ |
| | 551 | static void tokundef(tokcxdef *ctx, char *p, int len) |
| | 552 | { |
| | 553 | char *sym; |
| | 554 | int symlen; |
| | 555 | char mysym[TOKNAMMAX]; |
| | 556 | |
| | 557 | /* get the symbol */ |
| | 558 | sym = p; |
| | 559 | if (!(symlen = tok_scan_defsym(ctx, p, len))) |
| | 560 | return; |
| | 561 | |
| | 562 | /* if we're folding case, convert the symbol to lower case */ |
| | 563 | sym = tok_casefold_defsym(ctx, mysym, sym, symlen); |
| | 564 | |
| | 565 | /* if it's not defined, log a warning */ |
| | 566 | if (!tok_find_define(ctx, sym, symlen)) |
| | 567 | { |
| | 568 | errlog(ctx->tokcxerr, ERR_PUNDEF); |
| | 569 | return; |
| | 570 | } |
| | 571 | |
| | 572 | /* undefine the symbol */ |
| | 573 | tok_del_define(ctx, sym, symlen); |
| | 574 | } |
| | 575 | |
| | 576 | /* process a #pragma directive */ |
| | 577 | static void tokpragma(tokcxdef *ctx, char *p, int len) |
| | 578 | { |
| | 579 | /* ignore empty pragmas */ |
| | 580 | if (len == 0) |
| | 581 | { |
| | 582 | errlog(ctx->tokcxerr, ERR_PRAGMA); |
| | 583 | return; |
| | 584 | } |
| | 585 | |
| | 586 | /* see what we have */ |
| | 587 | if (len > 1 && (*p == 'c' || *p == 'C') |
| | 588 | && (*(p+1) == '+' || *(p+1) == '-') || t_isspace(*(p+1))) |
| | 589 | { |
| | 590 | /* skip spaces after the 'C', if any */ |
| | 591 | for (++p, --len ; len && t_isspace(*p) ; ++p, --len) ; |
| | 592 | |
| | 593 | /* look for the + or - flag */ |
| | 594 | if (len && *p == '+') |
| | 595 | ctx->tokcxflg |= TOKCXFCMODE; |
| | 596 | else if (len && *p == '-') |
| | 597 | ctx->tokcxflg &= ~TOKCXFCMODE; |
| | 598 | else |
| | 599 | { |
| | 600 | errlog(ctx->tokcxerr, ERR_PRAGMA); |
| | 601 | return; |
| | 602 | } |
| | 603 | } |
| | 604 | else |
| | 605 | { |
| | 606 | errlog(ctx->tokcxerr, ERR_PRAGMA); |
| | 607 | } |
| | 608 | } |
| | 609 | |
| | 610 | /* process a #include directive */ |
| | 611 | static void tokinclude(tokcxdef *ctx, char *p, int len) |
| | 612 | { |
| | 613 | linfdef *child; |
| | 614 | tokpdef *path; |
| | 615 | char *fname; |
| | 616 | int match; |
| | 617 | int flen; |
| | 618 | linfdef *lin; |
| | 619 | char *q; |
| | 620 | size_t flen2; |
| | 621 | |
| | 622 | /* find the filename portion */ |
| | 623 | fname = p + 1; /* remember start of filename */ |
| | 624 | path = ctx->tokcxinc; /* start with first path entry */ |
| | 625 | |
| | 626 | if (!len) |
| | 627 | { |
| | 628 | errlog(ctx->tokcxerr, ERR_INCNOFN); |
| | 629 | return; |
| | 630 | } |
| | 631 | |
| | 632 | switch(*p) |
| | 633 | { |
| | 634 | case '<': |
| | 635 | match = '>'; |
| | 636 | if (path && path->tokpnxt) path = path->tokpnxt; /* skip 1st path */ |
| | 637 | goto find_matching_delim; |
| | 638 | |
| | 639 | case '"': |
| | 640 | match = '"'; |
| | 641 | |
| | 642 | find_matching_delim: |
| | 643 | for (++p, --len ; len && *p != match ; --len, ++p) ; |
| | 644 | if (len == 0 || *p != match) errlog(ctx->tokcxerr, ERR_INCMTCH); |
| | 645 | break; |
| | 646 | |
| | 647 | default: |
| | 648 | errlog(ctx->tokcxerr, ERR_INCSYN); |
| | 649 | return; |
| | 650 | } |
| | 651 | |
| | 652 | flen = p - fname; /* compute length of filename */ |
| | 653 | for (q = p, flen2 = 0 ; |
| | 654 | q > fname && *(q-1) != OSPATHCHAR && !strchr(OSPATHALT, *(q-1)) ; |
| | 655 | --q, ++flen2) ; |
| | 656 | |
| | 657 | /* check to see if this file has already been included */ |
| | 658 | for (lin = ctx->tokcxhdr ; lin ; lin = (linfdef *)lin->linflin.linnxt) |
| | 659 | { |
| | 660 | char *p = lin->linfnam; |
| | 661 | |
| | 662 | p += strlen(p); |
| | 663 | |
| | 664 | while (p > lin->linfnam && *(p-1) != OSPATHCHAR |
| | 665 | && !strchr(OSPATHALT, *(p-1))) |
| | 666 | --p; |
| | 667 | if (strlen(p) == flen2 |
| | 668 | && !memicmp(p, q, flen2)) |
| | 669 | { |
| | 670 | errlog1(ctx->tokcxerr, ERR_INCRPT, ERRTSTR, |
| | 671 | errstr(ctx->tokcxerr, fname, flen)); |
| | 672 | return; |
| | 673 | } |
| | 674 | } |
| | 675 | |
| | 676 | /* initialize the line source */ |
| | 677 | child = linfini(ctx->tokcxmem, ctx->tokcxerr, fname, flen, path, TRUE, |
| | 678 | (ctx->tokcxflg & TOKCXFLIN2) != 0); |
| | 679 | |
| | 680 | /* if not found, signal an error */ |
| | 681 | if (!child) errsig1(ctx->tokcxerr, ERR_INCSEAR, |
| | 682 | ERRTSTR, errstr(ctx->tokcxerr, fname, flen)); |
| | 683 | |
| | 684 | /* link into tokenizer list of line records */ |
| | 685 | child->linflin.linnxt = (lindef *)ctx->tokcxhdr; |
| | 686 | ctx->tokcxhdr = child; |
| | 687 | |
| | 688 | /* if we're tracking sources for debugging, add into the chain */ |
| | 689 | if (ctx->tokcxdbg) |
| | 690 | { |
| | 691 | ctx->tokcxdbg->dbgcxlin = &child->linflin; |
| | 692 | child->linflin.linid = ctx->tokcxdbg->dbgcxfid++; |
| | 693 | } |
| | 694 | |
| | 695 | /* remember my C-mode setting */ |
| | 696 | if (ctx->tokcxflg & TOKCXFCMODE) |
| | 697 | ctx->tokcxlin->linflg |= LINFCMODE; |
| | 698 | else |
| | 699 | ctx->tokcxlin->linflg &= ~LINFCMODE; |
| | 700 | |
| | 701 | child->linflin.linpar = ctx->tokcxlin; /* remember parent line source */ |
| | 702 | ctx->tokcxlin = &child->linflin; /* make the child the current source */ |
| | 703 | } |
| | 704 | |
| | 705 | /* get a new line from line source, processing '#' directives */ |
| | 706 | static int tokgetlin(tokcxdef *ctx, int dopound) |
| | 707 | { |
| | 708 | for (;;) |
| | 709 | { |
| | 710 | if (linget(ctx->tokcxlin)) |
| | 711 | { |
| | 712 | /* at eof in current source; resume parent if there is one */ |
| | 713 | if (ctx->tokcxlin->linpar) |
| | 714 | { |
| | 715 | lindef *parent; |
| | 716 | |
| | 717 | parent = ctx->tokcxlin->linpar; /* remember parent */ |
| | 718 | lincls(ctx->tokcxlin); /* close included file */ |
| | 719 | if (!ctx->tokcxdbg) /* if no debug context... */ |
| | 720 | mchfre(ctx->tokcxlin); /* free line source */ |
| | 721 | ctx->tokcxlin = parent; /* reset to parent line source */ |
| | 722 | if (parent->linflg & LINFCMODE) |
| | 723 | ctx->tokcxflg |= TOKCXFCMODE; |
| | 724 | else |
| | 725 | ctx->tokcxflg &= ~TOKCXFCMODE; |
| | 726 | continue; /* back for another attempt */ |
| | 727 | } |
| | 728 | else |
| | 729 | { |
| | 730 | /* check for outstanding #if/#ifdef */ |
| | 731 | if (ctx->tokcxifcnt) |
| | 732 | errlog(ctx->tokcxerr, ERR_NOENDIF); |
| | 733 | |
| | 734 | /* return end-of-file indication */ |
| | 735 | return TRUE; |
| | 736 | } |
| | 737 | } |
| | 738 | |
| | 739 | /* if this is a multi-segment line, copy it into our own buffer */ |
| | 740 | if (ctx->tokcxlin->linflg & LINFMORE) |
| | 741 | { |
| | 742 | char *p; |
| | 743 | uint rem; |
| | 744 | int done; |
| | 745 | |
| | 746 | if (!ctx->tokcxbuf) |
| | 747 | { |
| | 748 | /* allocate 1k as a default buffer */ |
| | 749 | ctx->tokcxbuf = (char *)mchalo(ctx->tokcxerr, (ushort)1024, |
| | 750 | "tok"); |
| | 751 | ctx->tokcxbsz = 1024; |
| | 752 | } |
| | 753 | ctx->tokcxlen = 0; |
| | 754 | |
| | 755 | for (done = FALSE, p = ctx->tokcxbuf, rem = ctx->tokcxbsz ; |
| | 756 | !done ; ) |
| | 757 | { |
| | 758 | size_t len = ctx->tokcxlin->linlen; |
| | 759 | |
| | 760 | /* add the current segment's length into line length */ |
| | 761 | ctx->tokcxlen += len; |
| | 762 | |
| | 763 | /* we're done after this piece if the last fetch was all */ |
| | 764 | done = !(ctx->tokcxlin->linflg & LINFMORE); |
| | 765 | if (len + 1 > rem) |
| | 766 | { |
| | 767 | char *newp; |
| | 768 | |
| | 769 | /* increase the size of the buffer */ |
| | 770 | if (ctx->tokcxbsz > (unsigned)0x8000) |
| | 771 | errsig(ctx->tokcxerr, ERR_LONGLIN); |
| | 772 | rem += 4096; |
| | 773 | ctx->tokcxbsz += 4096; |
| | 774 | |
| | 775 | /* allocate a new buffer and copy line into it */ |
| | 776 | newp = (char *)mchalo(ctx->tokcxerr, ctx->tokcxbsz, "tok"); |
| | 777 | memcpy(newp, ctx->tokcxbuf, (size_t)(p - ctx->tokcxbuf)); |
| | 778 | |
| | 779 | /* free the original buffer, and use the new one */ |
| | 780 | p = (p - ctx->tokcxbuf) + newp; |
| | 781 | mchfre(ctx->tokcxbuf); |
| | 782 | ctx->tokcxbuf = newp; |
| | 783 | } |
| | 784 | |
| | 785 | /* add the line to the buffer */ |
| | 786 | memcpy(p, ctx->tokcxlin->linbuf, len); |
| | 787 | p += len; |
| | 788 | rem -= len; |
| | 789 | |
| | 790 | /* get the next piece of the line if there is one */ |
| | 791 | if (!done) |
| | 792 | { |
| | 793 | if (linget(ctx->tokcxlin)) break; |
| | 794 | } |
| | 795 | } |
| | 796 | |
| | 797 | /* null-terminate the buffer, and use it for input */ |
| | 798 | *p = '\0'; |
| | 799 | ctx->tokcxptr = ctx->tokcxbuf; |
| | 800 | } |
| | 801 | else |
| | 802 | { |
| | 803 | ctx->tokcxptr = ctx->tokcxlin->linbuf; |
| | 804 | ctx->tokcxlen = ctx->tokcxlin->linlen; |
| | 805 | } |
| | 806 | |
| | 807 | /* check for preprocessor directives */ |
| | 808 | if (dopound && ctx->tokcxlen != 0 && ctx->tokcxptr[0] == '#' |
| | 809 | && !(ctx->tokcxlin->linflg & LINFNOINC)) |
| | 810 | { |
| | 811 | char *p; |
| | 812 | int len; |
| | 813 | static struct |
| | 814 | { |
| | 815 | char *nm; |
| | 816 | int len; |
| | 817 | int ok_in_if; |
| | 818 | void (*fn)(tokcxdef *, char *, int); |
| | 819 | } |
| | 820 | *dirp, dir[] = |
| | 821 | { |
| | 822 | { "include", 7, FALSE, tokinclude }, |
| | 823 | { "pragma", 6, FALSE, tokpragma }, |
| | 824 | { "define", 6, FALSE, tokdefine }, |
| | 825 | { "ifdef", 5, TRUE, tokifdef }, |
| | 826 | { "ifndef", 6, TRUE, tokifndef }, |
| | 827 | { "if", 2, TRUE, tokif }, |
| | 828 | { "else", 4, TRUE, tokelse }, |
| | 829 | { "elif", 4, TRUE, tokelif }, |
| | 830 | { "endif", 5, TRUE, tokendif }, |
| | 831 | { "undef", 5, FALSE, tokundef }, |
| | 832 | { "error", 5, FALSE, tok_p_error } |
| | 833 | }; |
| | 834 | int i; |
| | 835 | |
| | 836 | /* scan off spaces between '#' and directive */ |
| | 837 | for (len = ctx->tokcxlen - 1, p = &ctx->tokcxptr[1] ; |
| | 838 | len && t_isspace(*p) ; --len, ++p) ; |
| | 839 | |
| | 840 | /* find and process the directive */ |
| | 841 | for (dirp = dir, i = sizeof(dir)/sizeof(dir[0]) ; i ; --i, ++dirp) |
| | 842 | { |
| | 843 | /* compare this directive; if it wins, call its function */ |
| | 844 | if (len >= dirp->len && !memcmp(p, dirp->nm, (size_t)dirp->len) |
| | 845 | && (len == dirp->len || t_isspace(*(p + dirp->len)))) |
| | 846 | { |
| | 847 | int cnt; |
| | 848 | int stat; |
| | 849 | |
| | 850 | /* |
| | 851 | * if we're not in a #if's false part, or if the |
| | 852 | * directive is processed even in #if false parts, |
| | 853 | * process the line, otherwise skip it |
| | 854 | */ |
| | 855 | cnt = ctx->tokcxifcnt; |
| | 856 | if (dirp->ok_in_if || cnt == 0 |
| | 857 | || ((stat = ctx->tokcxifcur) == TOKIF_IF_YES |
| | 858 | || stat == TOKIF_ELSE_YES)) |
| | 859 | { |
| | 860 | /* skip whitespace following the directive */ |
| | 861 | for (p += dirp->len, len -= dirp->len ; |
| | 862 | len && t_isspace(*p) ; |
| | 863 | --len, ++p) ; |
| | 864 | |
| | 865 | /* invoke the function to process this directive */ |
| | 866 | (*dirp->fn)(ctx, p, len); |
| | 867 | } |
| | 868 | |
| | 869 | /* there's no need to look at more directives */ |
| | 870 | break; |
| | 871 | } |
| | 872 | } |
| | 873 | |
| | 874 | /* if we didn't find anything, flag the error */ |
| | 875 | if (i == 0) |
| | 876 | errlog(ctx->tokcxerr, ERR_PRPDIR); |
| | 877 | |
| | 878 | /* ignore this line */ |
| | 879 | continue; |
| | 880 | } |
| | 881 | else |
| | 882 | { |
| | 883 | /* |
| | 884 | * Check the #if level. If we're in an #if, and we're to |
| | 885 | * ignore lines (because of a false condition or an #else |
| | 886 | * part for a true condition), skip this line. |
| | 887 | */ |
| | 888 | if (ctx->tokcxifcnt != 0) |
| | 889 | { |
| | 890 | switch(ctx->tokcxifcur) |
| | 891 | { |
| | 892 | case TOKIF_IF_NO: |
| | 893 | case TOKIF_ELSE_NO: |
| | 894 | /* ignore this line */ |
| | 895 | continue; |
| | 896 | |
| | 897 | default: |
| | 898 | /* we're in a true part - keep the line */ |
| | 899 | break; |
| | 900 | } |
| | 901 | } |
| | 902 | |
| | 903 | ctx->tokcxlin->linflg &= ~LINFDBG; /* no debug record yet */ |
| | 904 | return(FALSE); /* return the line we found */ |
| | 905 | } |
| | 906 | } |
| | 907 | } |
| | 908 | |
| | 909 | /* get the next token, removing it from the input stream */ |
| | 910 | int toknext(tokcxdef *ctx) |
| | 911 | { |
| | 912 | char *p; |
| | 913 | tokdef *tok = &ctx->tokcxcur; |
| | 914 | int len; |
| | 915 | |
| | 916 | /* |
| | 917 | * Check for the special case that we pushed an open paren prior to |
| | 918 | * a string containing an embedded expression. If this is the case, |
| | 919 | * immediately return the string we previously parsed. |
| | 920 | */ |
| | 921 | if ((ctx->tokcxflg & TOKCXF_EMBED_PAREN_PRE) != 0) |
| | 922 | { |
| | 923 | /* |
| | 924 | * convert the token to a string - note that the offset |
| | 925 | * information for the string is already in the current token |
| | 926 | * structure, since we set everything up for it on the previous |
| | 927 | * call where we actually parsed the beginning of the string |
| | 928 | */ |
| | 929 | tok->toktyp = TOKTDSTRING; |
| | 930 | |
| | 931 | /* clear the special flag - we've now consumed the pushed string */ |
| | 932 | ctx->tokcxflg &= ~TOKCXF_EMBED_PAREN_PRE; |
| | 933 | |
| | 934 | /* immediately return the string */ |
| | 935 | return tok->toktyp; |
| | 936 | } |
| | 937 | |
| | 938 | /* set up at the current scanning position */ |
| | 939 | p = ctx->tokcxptr; |
| | 940 | len = ctx->tokcxlen; |
| | 941 | |
| | 942 | /* scan off whitespace and comments until we find something */ |
| | 943 | do |
| | 944 | { |
| | 945 | skipblanks: |
| | 946 | /* if there's nothing on this line, get the next one */ |
| | 947 | if (len == 0) |
| | 948 | { |
| | 949 | /* if we're in a macro expansion, continue after it */ |
| | 950 | if (ctx->tokcxmlvl) |
| | 951 | { |
| | 952 | ctx->tokcxmlvl--; |
| | 953 | p = ctx->tokcxmsav[ctx->tokcxmlvl]; |
| | 954 | len = ctx->tokcxmsvl[ctx->tokcxmlvl]; |
| | 955 | } |
| | 956 | else |
| | 957 | { |
| | 958 | if (tokgetlin(ctx, TRUE)) |
| | 959 | { |
| | 960 | tok->toktyp = TOKTEOF; |
| | 961 | goto done; |
| | 962 | } |
| | 963 | p = ctx->tokcxptr; |
| | 964 | len = ctx->tokcxlen; |
| | 965 | } |
| | 966 | } |
| | 967 | while (len && t_isspace(*p)) ++p, --len; /* scan off whitespace */ |
| | 968 | |
| | 969 | /* check for comments, and remove if present */ |
| | 970 | if (len >= 2 && *p == '/' && *(p+1) == '/') |
| | 971 | len = 0; |
| | 972 | else if (len >= 2 && *p == '/' && *(p+1) == '*') |
| | 973 | { |
| | 974 | while (len < 2 || *p != '*' || *(p+1) != '/') |
| | 975 | { |
| | 976 | if (len != 0) |
| | 977 | ++p, --len; |
| | 978 | |
| | 979 | if (len == 0) |
| | 980 | { |
| | 981 | if (ctx->tokcxmlvl != 0) |
| | 982 | { |
| | 983 | ctx->tokcxmlvl--; |
| | 984 | p = ctx->tokcxmsav[ctx->tokcxmlvl]; |
| | 985 | len = ctx->tokcxmsvl[ctx->tokcxmlvl]; |
| | 986 | } |
| | 987 | else |
| | 988 | { |
| | 989 | if (tokgetlin(ctx, FALSE)) |
| | 990 | { |
| | 991 | ctx->tokcxptr = p; |
| | 992 | tok->toktyp = TOKTEOF; |
| | 993 | goto done; |
| | 994 | } |
| | 995 | p = ctx->tokcxptr; |
| | 996 | len = ctx->tokcxlen; |
| | 997 | } |
| | 998 | } |
| | 999 | } |
| | 1000 | p += 2; |
| | 1001 | len -= 2; |
| | 1002 | goto skipblanks; |
| | 1003 | } |
| | 1004 | } while (len == 0); |
| | 1005 | |
| | 1006 | nexttoken: |
| | 1007 | if (isalpha((uchar)*p) || *p == '_' || *p == '$') |
| | 1008 | { |
| | 1009 | int l; |
| | 1010 | int hash; |
| | 1011 | char *q; |
| | 1012 | toktdef *tab; |
| | 1013 | int found = FALSE; |
| | 1014 | uchar thischar; |
| | 1015 | tokdfdef *df; |
| | 1016 | |
| | 1017 | for (hash = 0, l = 0, q = tok->toknam ; |
| | 1018 | len != 0 && TOKISSYM(*p) && l < TOKNAMMAX ; |
| | 1019 | (thischar = ((isupper((uchar)*p) |
| | 1020 | && (ctx->tokcxflg & TOKCXCASEFOLD)) |
| | 1021 | ? tolower((uchar)*p) : *p)), |
| | 1022 | (hash = ((hash + thischar) & (TOKHASHSIZE - 1))), |
| | 1023 | (*q++ = thischar), ++p, --len, ++l) ; |
| | 1024 | *q = '\0'; |
| | 1025 | if (len != 0 && TOKISSYM(*p)) |
| | 1026 | { |
| | 1027 | while (len != 0 && TOKISSYM(*p)) ++p, --len; |
| | 1028 | errlog1(ctx->tokcxerr, ERR_TRUNC, ERRTSTR, |
| | 1029 | errstr(ctx->tokcxerr, tok->toknam, tok->toklen)); |
| | 1030 | } |
| | 1031 | tok->toklen = l; |
| | 1032 | tok->tokhash = hash; |
| | 1033 | |
| | 1034 | /* |
| | 1035 | * check for the special defined() preprocessor operator |
| | 1036 | */ |
| | 1037 | if (l == 9 && !memcmp(tok->toknam, |
| | 1038 | ((ctx->tokcxflg & TOKCXCASEFOLD) |
| | 1039 | ? "__defined" : "__DEFINED"), |
| | 1040 | (size_t)9) |
| | 1041 | && len > 2 && *p == '(' && TOKISSYM(*(p+1)) |
| | 1042 | && !isdigit((uchar)*(p+1))) |
| | 1043 | { |
| | 1044 | int symlen; |
| | 1045 | char mysym[TOKNAMMAX]; |
| | 1046 | |
| | 1047 | /* find the matching ')', allowing only symbolic characters */ |
| | 1048 | ++p, --len; |
| | 1049 | for (symlen = 0, q = p ; len && *p != ')' && TOKISSYM(*p) ; |
| | 1050 | ++p, --len, ++symlen) ; |
| | 1051 | |
| | 1052 | /* make sure we found the closing paren */ |
| | 1053 | if (!len || *p != ')') |
| | 1054 | errsig(ctx->tokcxerr, ERR_BADISDEF); |
| | 1055 | ++p, --len; |
| | 1056 | |
| | 1057 | /* if we're folding case, convert the symbol to lower case */ |
| | 1058 | q = tok_casefold_defsym(ctx, mysym, q, symlen); |
| | 1059 | |
| | 1060 | /* check to see if it's defined */ |
| | 1061 | tok->toktyp = TOKTNUMBER; |
| | 1062 | tok->tokval = (tok_find_define(ctx, q, symlen) != 0); |
| | 1063 | goto done; |
| | 1064 | } |
| | 1065 | |
| | 1066 | /* substitute the preprocessor #define, if any */ |
| | 1067 | if ((df = tok_find_define(ctx, tok->toknam, l)) != 0) |
| | 1068 | { |
| | 1069 | /* save the current parsing position */ |
| | 1070 | if (ctx->tokcxmlvl >= TOKMACNEST) |
| | 1071 | errsig(ctx->tokcxerr, ERR_MACNEST); |
| | 1072 | ctx->tokcxmsav[ctx->tokcxmlvl] = p; |
| | 1073 | ctx->tokcxmsvl[ctx->tokcxmlvl] = len; |
| | 1074 | ctx->tokcxmlvl++; |
| | 1075 | |
| | 1076 | /* point to the token's expansion and keep going */ |
| | 1077 | p = df->expan; |
| | 1078 | len = df->explen; |
| | 1079 | goto nexttoken; |
| | 1080 | } |
| | 1081 | |
| | 1082 | /* look up in symbol table(s), if any */ |
| | 1083 | for (tab = ctx->tokcxstab ; tab ; tab = tab->toktnxt) |
| | 1084 | { |
| | 1085 | if ((found = (*tab->toktfsea)(tab, tok->toknam, l, hash, |
| | 1086 | &tok->toksym)) != 0) |
| | 1087 | break; |
| | 1088 | } |
| | 1089 | |
| | 1090 | if (found && tok->toksym.tokstyp == TOKSTKW) |
| | 1091 | tok->toktyp = tok->toksym.toksval; |
| | 1092 | else |
| | 1093 | { |
| | 1094 | tok->toktyp = TOKTSYMBOL; |
| | 1095 | if (!found) tok->toksym.tokstyp = TOKSTUNK; |
| | 1096 | } |
| | 1097 | goto done; |
| | 1098 | } |
| | 1099 | else if (isdigit((uchar)*p)) |
| | 1100 | { |
| | 1101 | long acc = 0; |
| | 1102 | |
| | 1103 | /* check for octal/hex */ |
| | 1104 | if (*p == '0') |
| | 1105 | { |
| | 1106 | ++p, --len; |
| | 1107 | if (len && (*p == 'x' || *p == 'X')) |
| | 1108 | { |
| | 1109 | /* hex */ |
| | 1110 | ++p, --len; |
| | 1111 | while (len && TOKISHEX(*p)) |
| | 1112 | { |
| | 1113 | acc = (acc << 4) + TOKHEX2INT(*p); |
| | 1114 | ++p, --len; |
| | 1115 | } |
| | 1116 | } |
| | 1117 | else |
| | 1118 | { |
| | 1119 | /* octal */ |
| | 1120 | while (len && TOKISOCT(*p)) |
| | 1121 | { |
| | 1122 | acc = (acc << 3) + TOKOCT2INT(*p); |
| | 1123 | ++p, --len; |
| | 1124 | } |
| | 1125 | } |
| | 1126 | } |
| | 1127 | else |
| | 1128 | { |
| | 1129 | /* decimal */ |
| | 1130 | while (len && isdigit((uchar)*p)) |
| | 1131 | { |
| | 1132 | acc = (acc << 1) + (acc << 3) + TOKDEC2INT(*p); |
| | 1133 | ++p, --len; |
| | 1134 | } |
| | 1135 | } |
| | 1136 | tok->tokval = acc; |
| | 1137 | tok->toktyp = TOKTNUMBER; |
| | 1138 | goto done; |
| | 1139 | } |
| | 1140 | else if (*p == '"' || *p == '\'') |
| | 1141 | { |
| | 1142 | char delim; /* closing delimiter we're looking for */ |
| | 1143 | char *strstart; /* pointer to start of string */ |
| | 1144 | int warned; |
| | 1145 | |
| | 1146 | delim = *p; |
| | 1147 | --len; |
| | 1148 | strstart = ++p; |
| | 1149 | |
| | 1150 | if (delim == '"' && len >= 2 && *p == '<' && *(p+1) == '<') |
| | 1151 | { |
| | 1152 | /* save the current parsing position */ |
| | 1153 | if (ctx->tokcxmlvl >= TOKMACNEST) |
| | 1154 | errsig(ctx->tokcxerr, ERR_MACNEST); |
| | 1155 | ctx->tokcxmsav[ctx->tokcxmlvl] = p + 2; |
| | 1156 | ctx->tokcxmsvl[ctx->tokcxmlvl] = len - 2; |
| | 1157 | ctx->tokcxmlvl++; |
| | 1158 | |
| | 1159 | /* |
| | 1160 | * read from the special "<<" expansion string - use the |
| | 1161 | * version for a "<<" at the very beginning of the string |
| | 1162 | */ |
| | 1163 | p = tokmac1s; |
| | 1164 | len = strlen(p); |
| | 1165 | ctx->tokcxflg |= TOKCXFINMAC; |
| | 1166 | goto nexttoken; |
| | 1167 | } |
| | 1168 | tok->toktyp = (delim == '"' ? TOKTDSTRING : TOKTSSTRING); |
| | 1169 | |
| | 1170 | tok->tokofs = (*ctx->tokcxsst)(ctx->tokcxscx); /* start the string */ |
| | 1171 | for (warned = FALSE ;; ) |
| | 1172 | { |
| | 1173 | if (len >= 2 && *p == '\\') |
| | 1174 | { |
| | 1175 | if (*(p+1) == '"' || *(p+1) == '\'') |
| | 1176 | { |
| | 1177 | (*ctx->tokcxsad)(ctx->tokcxscx, strstart, |
| | 1178 | (ushort)(p - strstart)); |
| | 1179 | strstart = p + 1; |
| | 1180 | } |
| | 1181 | p += 2; |
| | 1182 | len -= 2; |
| | 1183 | } |
| | 1184 | else if (len == 0 || *p == delim || |
| | 1185 | (delim == '"' && len >= 2 && *p == '<' && *(p+1) == '<' |
| | 1186 | && !(ctx->tokcxflg & TOKCXFINMAC))) |
| | 1187 | { |
| | 1188 | (*ctx->tokcxsad)(ctx->tokcxscx, strstart, |
| | 1189 | (ushort)(p - strstart)); |
| | 1190 | if (len == 0) |
| | 1191 | { |
| | 1192 | if (ctx->tokcxmlvl) |
| | 1193 | { |
| | 1194 | ctx->tokcxmlvl--; |
| | 1195 | p = ctx->tokcxmsav[ctx->tokcxmlvl]; |
| | 1196 | len = ctx->tokcxmsvl[ctx->tokcxmlvl]; |
| | 1197 | } |
| | 1198 | else |
| | 1199 | (*ctx->tokcxsad)(ctx->tokcxscx, " ", (ushort)1); |
| | 1200 | |
| | 1201 | while (len == 0) |
| | 1202 | { |
| | 1203 | if (tokgetlin(ctx, FALSE)) |
| | 1204 | errsig(ctx->tokcxerr, ERR_STREOF); |
| | 1205 | p = ctx->tokcxptr; |
| | 1206 | len = ctx->tokcxlen; |
| | 1207 | |
| | 1208 | /* warn if it looks like the end of an object */ |
| | 1209 | if (!warned && len && (*p == ';' || *p == '}')) |
| | 1210 | { |
| | 1211 | errlog(ctx->tokcxerr, ERR_STREND); |
| | 1212 | warned = TRUE; /* warn only once per string */ |
| | 1213 | } |
| | 1214 | |
| | 1215 | /* scan past whitespace at start of line */ |
| | 1216 | while (len && t_isspace(*p)) ++p, --len; |
| | 1217 | } |
| | 1218 | strstart = p; |
| | 1219 | } |
| | 1220 | else break; |
| | 1221 | } |
| | 1222 | else |
| | 1223 | ++p, --len; |
| | 1224 | } |
| | 1225 | |
| | 1226 | /* end the string */ |
| | 1227 | (*ctx->tokcxsend)(ctx->tokcxscx); |
| | 1228 | |
| | 1229 | /* check to see how it ended */ |
| | 1230 | if (len != 0 && *p == delim) |
| | 1231 | { |
| | 1232 | /* |
| | 1233 | * We ended with the matching delimiter. Move past the |
| | 1234 | * closing delimiter. |
| | 1235 | */ |
| | 1236 | ++p; |
| | 1237 | --len; |
| | 1238 | |
| | 1239 | /* |
| | 1240 | * If we have a pending close paren we need to put in |
| | 1241 | * because of an embedded expression that occurred earlier |
| | 1242 | * in the string, parse the macro to provide the paren. |
| | 1243 | */ |
| | 1244 | if ((ctx->tokcxflg & TOKCXF_EMBED_PAREN_AFT) != 0 |
| | 1245 | && !(ctx->tokcxflg & TOKCXFINMAC)) |
| | 1246 | { |
| | 1247 | /* clear the flag */ |
| | 1248 | ctx->tokcxflg &= ~TOKCXF_EMBED_PAREN_AFT; |
| | 1249 | |
| | 1250 | /* push the current parsing position */ |
| | 1251 | if (ctx->tokcxmlvl >= TOKMACNEST) |
| | 1252 | errsig(ctx->tokcxerr, ERR_MACNEST); |
| | 1253 | ctx->tokcxmsav[ctx->tokcxmlvl] = p; |
| | 1254 | ctx->tokcxmsvl[ctx->tokcxmlvl] = len; |
| | 1255 | ctx->tokcxmlvl++; |
| | 1256 | |
| | 1257 | /* parse the macro */ |
| | 1258 | p = tokmac4; |
| | 1259 | len = strlen(p); |
| | 1260 | } |
| | 1261 | } |
| | 1262 | else if (len != 0 && *p == '<') |
| | 1263 | { |
| | 1264 | /* save the current parsing position */ |
| | 1265 | if (ctx->tokcxmlvl >= TOKMACNEST) |
| | 1266 | errsig(ctx->tokcxerr, ERR_MACNEST); |
| | 1267 | ctx->tokcxmsav[ctx->tokcxmlvl] = p + 2; |
| | 1268 | ctx->tokcxmsvl[ctx->tokcxmlvl] = len - 2; |
| | 1269 | ctx->tokcxmlvl++; |
| | 1270 | |
| | 1271 | /* read from the "<<" expansion */ |
| | 1272 | p = tokmac1; |
| | 1273 | len = strlen(p); |
| | 1274 | ctx->tokcxflg |= TOKCXFINMAC; |
| | 1275 | |
| | 1276 | /* |
| | 1277 | * Set the special push-a-paren flag: we'll return an open |
| | 1278 | * paren now, so that we have an open paren before the |
| | 1279 | * string, and then on the next call to toknext() we'll |
| | 1280 | * immediately return the string we've already parsed here. |
| | 1281 | * This will ensure that everything in the string is |
| | 1282 | * properly grouped together as a single indivisible |
| | 1283 | * expression. |
| | 1284 | * |
| | 1285 | * Note that we only need to do this for the first embedded |
| | 1286 | * expression in a string. Once we have a close paren |
| | 1287 | * pending, we don't need more open parens. |
| | 1288 | */ |
| | 1289 | if (!(ctx->tokcxflg & TOKCXF_EMBED_PAREN_AFT)) |
| | 1290 | { |
| | 1291 | ctx->tokcxflg |= TOKCXF_EMBED_PAREN_PRE; |
| | 1292 | tok->toktyp = TOKTLPAR; |
| | 1293 | } |
| | 1294 | } |
| | 1295 | goto done; |
| | 1296 | } |
| | 1297 | else if (len >= 2 && *p == '>' && *(p+1) == '>' |
| | 1298 | && (ctx->tokcxflg & TOKCXFINMAC) != 0) |
| | 1299 | { |
| | 1300 | /* skip the ">>" */ |
| | 1301 | ctx->tokcxflg &= ~TOKCXFINMAC; |
| | 1302 | p += 2; |
| | 1303 | len -= 2; |
| | 1304 | |
| | 1305 | /* save the current parsing position */ |
| | 1306 | if (ctx->tokcxmlvl >= TOKMACNEST) |
| | 1307 | errsig(ctx->tokcxerr, ERR_MACNEST); |
| | 1308 | ctx->tokcxmsav[ctx->tokcxmlvl] = p; |
| | 1309 | ctx->tokcxmsvl[ctx->tokcxmlvl] = len; |
| | 1310 | ctx->tokcxmlvl++; |
| | 1311 | |
| | 1312 | if (*p == '"') |
| | 1313 | { |
| | 1314 | ++(ctx->tokcxmsav[ctx->tokcxmlvl - 1]); |
| | 1315 | --(ctx->tokcxmsvl[ctx->tokcxmlvl - 1]); |
| | 1316 | p = tokmac3; |
| | 1317 | |
| | 1318 | /* |
| | 1319 | * we won't need an extra closing paren now, since tokmac3 |
| | 1320 | * provides it |
| | 1321 | */ |
| | 1322 | ctx->tokcxflg &= ~TOKCXF_EMBED_PAREN_AFT; |
| | 1323 | } |
| | 1324 | else |
| | 1325 | { |
| | 1326 | /* |
| | 1327 | * The string is continuing. Set a flag to note that we |
| | 1328 | * need to provide a close paren after the end of the |
| | 1329 | * string, and parse the glue (tokmac2) that goes between |
| | 1330 | * the expression and the resumption of the string. |
| | 1331 | */ |
| | 1332 | ctx->tokcxflg |= TOKCXF_EMBED_PAREN_AFT; |
| | 1333 | p = tokmac2; |
| | 1334 | } |
| | 1335 | |
| | 1336 | len = strlen(p); |
| | 1337 | goto nexttoken; |
| | 1338 | } |
| | 1339 | else |
| | 1340 | { |
| | 1341 | tokscdef *sc; |
| | 1342 | |
| | 1343 | for (sc = ctx->tokcxsc[ctx->tokcxinx[(uchar)*p]] ; sc ; |
| | 1344 | sc = sc->tokscnxt) |
| | 1345 | { |
| | 1346 | if (toksceq(sc->tokscstr, p, sc->toksclen, len)) |
| | 1347 | { |
| | 1348 | tok->toktyp = sc->toksctyp; |
| | 1349 | p += sc->toksclen; |
| | 1350 | len -= sc->toksclen; |
| | 1351 | goto done; |
| | 1352 | } |
| | 1353 | } |
| | 1354 | errsig(ctx->tokcxerr, ERR_INVTOK); |
| | 1355 | } |
| | 1356 | |
| | 1357 | done: |
| | 1358 | ctx->tokcxptr = p; |
| | 1359 | ctx->tokcxlen = len; |
| | 1360 | return(tok->toktyp); |
| | 1361 | } |
| | 1362 | |
| | 1363 | /* initialize a linear symbol table */ |
| | 1364 | void toktlini(errcxdef *errctx, toktldef *toktab, uchar *mem, uint siz) |
| | 1365 | { |
| | 1366 | CLRSTRUCT(*toktab); |
| | 1367 | |
| | 1368 | /* initialize superclass data */ |
| | 1369 | toktab->toktlsc.toktfadd = toktladd; /* set add-symbol method */ |
| | 1370 | toktab->toktlsc.toktfsea = toktlsea; /* set search-table method */ |
| | 1371 | toktab->toktlsc.toktfeach = toktleach; /* set 'each' method */ |
| | 1372 | toktab->toktlsc.toktfset = toktlset; /* set 'update' method */ |
| | 1373 | toktab->toktlsc.tokterr = errctx; /* set error handling context */ |
| | 1374 | |
| | 1375 | /* initialize class data */ |
| | 1376 | toktab->toktlptr = mem; |
| | 1377 | toktab->toktlnxt = mem; |
| | 1378 | toktab->toktlsiz = siz; |
| | 1379 | } |
| | 1380 | |
| | 1381 | /* add a symbol to a linear symbol table */ |
| | 1382 | void toktladd(toktdef *toktab1, char *name, int namel, |
| | 1383 | int typ, int val, int hash) |
| | 1384 | { |
| | 1385 | uint siz = sizeof(toks1def) + namel; |
| | 1386 | toksdef *sym; |
| | 1387 | toktldef *toktab = (toktldef *)toktab1; |
| | 1388 | |
| | 1389 | VARUSED(hash); |
| | 1390 | |
| | 1391 | if (toktab->toktlsiz < siz) |
| | 1392 | errsig(toktab->toktlsc.tokterr, ERR_NOLCLSY); |
| | 1393 | |
| | 1394 | sym = (toksdef *)toktab->toktlnxt; |
| | 1395 | siz = osrndsz(siz); |
| | 1396 | toktab->toktlnxt += siz; |
| | 1397 | if (siz > toktab->toktlsiz) toktab->toktlsiz = 0; |
| | 1398 | else toktab->toktlsiz -= siz; |
| | 1399 | |
| | 1400 | /* set up symbol */ |
| | 1401 | sym->toksval = val; |
| | 1402 | sym->tokslen = namel; |
| | 1403 | sym->tokstyp = typ; |
| | 1404 | sym->toksfr = 0; |
| | 1405 | memcpy(sym->toksnam, name, (size_t)(namel + 1)); |
| | 1406 | |
| | 1407 | /* indicate there's one more symbol in the table */ |
| | 1408 | ++(toktab->toktlcnt); |
| | 1409 | } |
| | 1410 | |
| | 1411 | /* delete all symbols from a linear symbol table */ |
| | 1412 | void toktldel(toktldef *tab) |
| | 1413 | { |
| | 1414 | tab->toktlcnt = 0; |
| | 1415 | tab->toktlsiz += tab->toktlnxt - tab->toktlptr; |
| | 1416 | tab->toktlnxt = tab->toktlptr; |
| | 1417 | } |
| | 1418 | |
| | 1419 | /* call a function for every symbol in a linear symbol table */ |
| | 1420 | void toktleach(toktdef *tab1, |
| | 1421 | void (*cb)(void *ctx, toksdef *sym), void *ctx) |
| | 1422 | { |
| | 1423 | toksdef *p; |
| | 1424 | uint cnt; |
| | 1425 | toktldef *tab = (toktldef *)tab1; |
| | 1426 | |
| | 1427 | for (p = (toksdef *)tab->toktlptr, cnt = tab->toktlcnt ; cnt ; --cnt ) |
| | 1428 | { |
| | 1429 | (*cb)(ctx, p); |
| | 1430 | p = (toksdef *)(((uchar *)p) |
| | 1431 | + osrndsz(p->tokslen + sizeof(toks1def))); |
| | 1432 | } |
| | 1433 | } |
| | 1434 | |
| | 1435 | /* search a linear symbol table */ |
| | 1436 | int toktlsea(toktdef *tab1, char *name, int namel, int hash, toksdef *ret) |
| | 1437 | { |
| | 1438 | toksdef *p; |
| | 1439 | uint cnt; |
| | 1440 | toktldef *tab = (toktldef *)tab1; |
| | 1441 | |
| | 1442 | VARUSED(hash); |
| | 1443 | |
| | 1444 | for (p = (toksdef *)tab->toktlptr, cnt = tab->toktlcnt ; cnt ; --cnt ) |
| | 1445 | { |
| | 1446 | if (p->tokslen == namel && !memcmp(p->toksnam, name, (size_t)namel)) |
| | 1447 | { |
| | 1448 | memcpy(ret, p, (size_t)(sizeof(toks1def) + namel)); |
| | 1449 | return(TRUE); |
| | 1450 | } |
| | 1451 | |
| | 1452 | p = (toksdef *)(((uchar *)p) |
| | 1453 | + osrndsz(p->tokslen + sizeof(toks1def))); |
| | 1454 | } |
| | 1455 | |
| | 1456 | /* nothing found - indicate by returning FALSE */ |
| | 1457 | return(FALSE); |
| | 1458 | } |
| | 1459 | |
| | 1460 | /* update a symbol in a linear symbol table */ |
| | 1461 | void toktlset(toktdef *tab1, toksdef *newsym) |
| | 1462 | { |
| | 1463 | toksdef *p; |
| | 1464 | uint cnt; |
| | 1465 | toktldef *tab = (toktldef *)tab1; |
| | 1466 | |
| | 1467 | for (p = (toksdef *)tab->toktlptr, cnt = tab->toktlcnt ; cnt ; --cnt ) |
| | 1468 | { |
| | 1469 | if (p->tokslen == newsym->tokslen |
| | 1470 | && !memcmp(p->toksnam, newsym->toksnam, (size_t)newsym->tokslen)) |
| | 1471 | { |
| | 1472 | p->toksval = newsym->toksval; |
| | 1473 | p->tokstyp = newsym->tokstyp; |
| | 1474 | return; |
| | 1475 | } |
| | 1476 | |
| | 1477 | p = (toksdef *)(((uchar *)p) |
| | 1478 | + osrndsz(p->tokslen + sizeof(toks1def))); |
| | 1479 | } |
| | 1480 | } |
| | 1481 | |
| | 1482 | tokcxdef *tokcxini(errcxdef *errctx, mcmcxdef *mcmctx, tokldef *sctab) |
| | 1483 | { |
| | 1484 | int i; |
| | 1485 | int cnt; |
| | 1486 | tokldef *p; |
| | 1487 | uchar c; |
| | 1488 | uchar index[256]; |
| | 1489 | tokcxdef *ret; |
| | 1490 | tokscdef *sc; |
| | 1491 | ushort siz; |
| | 1492 | |
| | 1493 | /* set up index table: finds tokcxsc entry from character value */ |
| | 1494 | memset(index, 0, (size_t)sizeof(index)); |
| | 1495 | for (i = cnt = 0, p = sctab ; (c = p->toklstr[0]) != 0 ; ++cnt, ++p) |
| | 1496 | if (!index[c]) index[c] = ++i; |
| | 1497 | |
| | 1498 | /* allocate memory for table plus the tokscdef's */ |
| | 1499 | siz = sizeof(tokcxdef) + (i * sizeof(tokscdef *)) |
| | 1500 | + ((cnt + 1) * sizeof(tokscdef)); |
| | 1501 | ret = (tokcxdef *)mchalo(errctx, siz, "tokcxini"); |
| | 1502 | memset(ret, 0, (size_t)siz); |
| | 1503 | |
| | 1504 | /* copy the index, set up fixed part */ |
| | 1505 | memcpy(ret->tokcxinx, index, sizeof(ret->tokcxinx)); |
| | 1506 | ret->tokcxerr = errctx; |
| | 1507 | ret->tokcxmem = mcmctx; |
| | 1508 | |
| | 1509 | /* start out without an #if */ |
| | 1510 | ret->tokcxifcur = TOKIF_IF_YES; |
| | 1511 | |
| | 1512 | /* force the first toknext() to read a line */ |
| | 1513 | ret->tokcxptr = "\000"; |
| | 1514 | |
| | 1515 | /* figure where the tokscdef's go (right after sc pointer array) */ |
| | 1516 | sc = (tokscdef *)&ret->tokcxsc[i+1]; |
| | 1517 | |
| | 1518 | /* set up the individual tokscdef entries, and link into lists */ |
| | 1519 | for (p = sctab ; (c = p->toklstr[0]) != 0 ; ++p, ++sc) |
| | 1520 | { |
| | 1521 | size_t len; |
| | 1522 | |
| | 1523 | sc->toksctyp = p->tokltyp; |
| | 1524 | len = sc->toksclen = strlen(p->toklstr); |
| | 1525 | memcpy(sc->tokscstr, p->toklstr, len); |
| | 1526 | sc->tokscnxt = ret->tokcxsc[index[c]]; |
| | 1527 | ret->tokcxsc[index[c]] = sc; |
| | 1528 | } |
| | 1529 | |
| | 1530 | return(ret); |
| | 1531 | } |
| | 1532 | |
| | 1533 | /* add an include path to a tokdef */ |
| | 1534 | void tokaddinc(tokcxdef *ctx, char *path, int pathlen) |
| | 1535 | { |
| | 1536 | tokpdef *newpath; |
| | 1537 | tokpdef *last; |
| | 1538 | |
| | 1539 | /* find the tail of the include path list, if any */ |
| | 1540 | for (last = ctx->tokcxinc ; last && last->tokpnxt ; |
| | 1541 | last = last->tokpnxt) ; |
| | 1542 | |
| | 1543 | /* allocate storage for and set up a new path structure */ |
| | 1544 | newpath = (tokpdef *)mchalo(ctx->tokcxerr, |
| | 1545 | (ushort)(sizeof(tokpdef) + pathlen - 1), |
| | 1546 | "tokaddinc"); |
| | 1547 | newpath->tokplen = pathlen; |
| | 1548 | newpath->tokpnxt = (tokpdef *)0; |
| | 1549 | memcpy(newpath->tokpdir, path, (size_t)pathlen); |
| | 1550 | |
| | 1551 | /* link in at end of list (if no list yet, newpath becomes first entry) */ |
| | 1552 | if (last) |
| | 1553 | last->tokpnxt = newpath; |
| | 1554 | else |
| | 1555 | ctx->tokcxinc = newpath; |
| | 1556 | } |
| | 1557 | |