1 /* $Id: mdoc_macro.c,v 1.235 2022/04/14 16:43:44 schwarze Exp $ */ 2 /* 3 * Copyright (c) 2010, 2012-2021 Ingo Schwarze <schwarze@openbsd.org> 4 * Copyright (c) 2008-2012 Kristaps Dzonsons <kristaps@bsd.lv> 5 * 6 * Permission to use, copy, modify, and distribute this software for any 7 * purpose with or without fee is hereby granted, provided that the above 8 * copyright notice and this permission notice appear in all copies. 9 * 10 * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES 11 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF 12 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR 13 * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES 14 * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN 15 * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF 16 * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. 17 */ 18 #include "config.h" 19 20 #include <sys/types.h> 21 22 #include <assert.h> 23 #include <ctype.h> 24 #include <stdlib.h> 25 #include <stdio.h> 26 #include <string.h> 27 #include <time.h> 28 29 #if DEBUG_MEMORY 30 #include "mandoc_dbg.h" 31 #endif 32 #include "mandoc.h" 33 #include "roff.h" 34 #include "mdoc.h" 35 #include "libmandoc.h" 36 #include "roff_int.h" 37 #include "libmdoc.h" 38 39 static void blk_full(MACRO_PROT_ARGS); 40 static void blk_exp_close(MACRO_PROT_ARGS); 41 static void blk_part_exp(MACRO_PROT_ARGS); 42 static void blk_part_imp(MACRO_PROT_ARGS); 43 static void ctx_synopsis(MACRO_PROT_ARGS); 44 static void in_line_eoln(MACRO_PROT_ARGS); 45 static void in_line_argn(MACRO_PROT_ARGS); 46 static void in_line(MACRO_PROT_ARGS); 47 static void phrase_ta(MACRO_PROT_ARGS); 48 49 static void append_delims(struct roff_man *, int, int *, char *); 50 static void dword(struct roff_man *, int, int, const char *, 51 enum mdelim, int); 52 static int find_pending(struct roff_man *, enum roff_tok, 53 int, int, struct roff_node *); 54 static int lookup(struct roff_man *, int, int, int, const char *); 55 static int macro_or_word(MACRO_PROT_ARGS, char *, int); 56 static void break_intermediate(struct roff_node *, 57 struct roff_node *); 58 static int parse_rest(struct roff_man *, enum roff_tok, 59 int, int *, char *); 60 static enum roff_tok rew_alt(enum roff_tok); 61 static void rew_elem(struct roff_man *, enum roff_tok); 62 static void rew_last(struct roff_man *, const struct roff_node *); 63 static void rew_pending(struct roff_man *, 64 const struct roff_node *); 65 66 static const struct mdoc_macro mdoc_macros[MDOC_MAX - MDOC_Dd] = { 67 { in_line_eoln, MDOC_PROLOGUE | MDOC_JOIN }, /* Dd */ 68 { in_line_eoln, MDOC_PROLOGUE }, /* Dt */ 69 { in_line_eoln, MDOC_PROLOGUE }, /* Os */ 70 { blk_full, MDOC_PARSED | MDOC_JOIN }, /* Sh */ 71 { blk_full, MDOC_PARSED | MDOC_JOIN }, /* Ss */ 72 { in_line_eoln, 0 }, /* Pp */ 73 { blk_part_imp, MDOC_PARSED | MDOC_JOIN }, /* D1 */ 74 { blk_part_imp, MDOC_PARSED | MDOC_JOIN }, /* Dl */ 75 { blk_full, MDOC_EXPLICIT }, /* Bd */ 76 { blk_exp_close, MDOC_EXPLICIT | MDOC_JOIN }, /* Ed */ 77 { blk_full, MDOC_EXPLICIT }, /* Bl */ 78 { blk_exp_close, MDOC_EXPLICIT | MDOC_JOIN }, /* El */ 79 { blk_full, MDOC_PARSED | MDOC_JOIN }, /* It */ 80 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ad */ 81 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* An */ 82 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED | 83 MDOC_IGNDELIM | MDOC_JOIN }, /* Ap */ 84 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ar */ 85 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Cd */ 86 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Cm */ 87 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Dv */ 88 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Er */ 89 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ev */ 90 { in_line_eoln, 0 }, /* Ex */ 91 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fa */ 92 { in_line_eoln, 0 }, /* Fd */ 93 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fl */ 94 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fn */ 95 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ft */ 96 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Ic */ 97 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* In */ 98 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Li */ 99 { blk_full, MDOC_JOIN }, /* Nd */ 100 { ctx_synopsis, MDOC_CALLABLE | MDOC_PARSED }, /* Nm */ 101 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Op */ 102 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ot */ 103 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Pa */ 104 { in_line_eoln, 0 }, /* Rv */ 105 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* St */ 106 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Va */ 107 { ctx_synopsis, MDOC_CALLABLE | MDOC_PARSED }, /* Vt */ 108 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Xr */ 109 { in_line_eoln, MDOC_JOIN }, /* %A */ 110 { in_line_eoln, MDOC_JOIN }, /* %B */ 111 { in_line_eoln, MDOC_JOIN }, /* %D */ 112 { in_line_eoln, MDOC_JOIN }, /* %I */ 113 { in_line_eoln, MDOC_JOIN }, /* %J */ 114 { in_line_eoln, 0 }, /* %N */ 115 { in_line_eoln, MDOC_JOIN }, /* %O */ 116 { in_line_eoln, 0 }, /* %P */ 117 { in_line_eoln, MDOC_JOIN }, /* %R */ 118 { in_line_eoln, MDOC_JOIN }, /* %T */ 119 { in_line_eoln, 0 }, /* %V */ 120 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED | 121 MDOC_EXPLICIT | MDOC_JOIN }, /* Ac */ 122 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | 123 MDOC_EXPLICIT | MDOC_JOIN }, /* Ao */ 124 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Aq */ 125 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* At */ 126 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED | 127 MDOC_EXPLICIT | MDOC_JOIN }, /* Bc */ 128 { blk_full, MDOC_EXPLICIT }, /* Bf */ 129 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | 130 MDOC_EXPLICIT | MDOC_JOIN }, /* Bo */ 131 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Bq */ 132 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Bsx */ 133 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Bx */ 134 { in_line_eoln, 0 }, /* Db */ 135 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED | 136 MDOC_EXPLICIT | MDOC_JOIN }, /* Dc */ 137 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | 138 MDOC_EXPLICIT | MDOC_JOIN }, /* Do */ 139 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Dq */ 140 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Ec */ 141 { blk_exp_close, MDOC_EXPLICIT | MDOC_JOIN }, /* Ef */ 142 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Em */ 143 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Eo */ 144 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Fx */ 145 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ms */ 146 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* No */ 147 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED | 148 MDOC_IGNDELIM | MDOC_JOIN }, /* Ns */ 149 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Nx */ 150 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Ox */ 151 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED | 152 MDOC_EXPLICIT | MDOC_JOIN }, /* Pc */ 153 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED | MDOC_IGNDELIM }, /* Pf */ 154 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | 155 MDOC_EXPLICIT | MDOC_JOIN }, /* Po */ 156 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Pq */ 157 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED | 158 MDOC_EXPLICIT | MDOC_JOIN }, /* Qc */ 159 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Ql */ 160 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | 161 MDOC_EXPLICIT | MDOC_JOIN }, /* Qo */ 162 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Qq */ 163 { blk_exp_close, MDOC_EXPLICIT | MDOC_JOIN }, /* Re */ 164 { blk_full, MDOC_EXPLICIT }, /* Rs */ 165 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED | 166 MDOC_EXPLICIT | MDOC_JOIN }, /* Sc */ 167 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | 168 MDOC_EXPLICIT | MDOC_JOIN }, /* So */ 169 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Sq */ 170 { in_line_argn, 0 }, /* Sm */ 171 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Sx */ 172 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Sy */ 173 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Tn */ 174 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Ux */ 175 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Xc */ 176 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Xo */ 177 { blk_full, MDOC_EXPLICIT | MDOC_CALLABLE }, /* Fo */ 178 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED | 179 MDOC_EXPLICIT | MDOC_JOIN }, /* Fc */ 180 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | 181 MDOC_EXPLICIT | MDOC_JOIN }, /* Oo */ 182 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED | 183 MDOC_EXPLICIT | MDOC_JOIN }, /* Oc */ 184 { blk_full, MDOC_EXPLICIT }, /* Bk */ 185 { blk_exp_close, MDOC_EXPLICIT | MDOC_JOIN }, /* Ek */ 186 { in_line_eoln, 0 }, /* Bt */ 187 { in_line_eoln, 0 }, /* Hf */ 188 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fr */ 189 { in_line_eoln, 0 }, /* Ud */ 190 { in_line, 0 }, /* Lb */ 191 { in_line_eoln, 0 }, /* Lp */ 192 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Lk */ 193 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Mt */ 194 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Brq */ 195 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | 196 MDOC_EXPLICIT | MDOC_JOIN }, /* Bro */ 197 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED | 198 MDOC_EXPLICIT | MDOC_JOIN }, /* Brc */ 199 { in_line_eoln, MDOC_JOIN }, /* %C */ 200 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Es */ 201 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* En */ 202 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Dx */ 203 { in_line_eoln, MDOC_JOIN }, /* %Q */ 204 { in_line_eoln, 0 }, /* %U */ 205 { phrase_ta, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Ta */ 206 { in_line_eoln, 0 }, /* Tg */ 207 }; 208 209 210 const struct mdoc_macro * 211 mdoc_macro(enum roff_tok tok) 212 { 213 assert(tok >= MDOC_Dd && tok < MDOC_MAX); 214 return mdoc_macros + (tok - MDOC_Dd); 215 } 216 217 /* 218 * This is called at the end of parsing. It must traverse up the tree, 219 * closing out open [implicit] scopes. Obviously, open explicit scopes 220 * are errors. 221 */ 222 void 223 mdoc_endparse(struct roff_man *mdoc) 224 { 225 struct roff_node *n; 226 227 /* Scan for open explicit scopes. */ 228 229 n = mdoc->last->flags & NODE_VALID ? 230 mdoc->last->parent : mdoc->last; 231 232 for ( ; n; n = n->parent) 233 if (n->type == ROFFT_BLOCK && 234 mdoc_macro(n->tok)->flags & MDOC_EXPLICIT) 235 mandoc_msg(MANDOCERR_BLK_NOEND, 236 n->line, n->pos, "%s", roff_name[n->tok]); 237 238 /* Rewind to the first. */ 239 240 rew_last(mdoc, mdoc->meta.first); 241 } 242 243 /* 244 * Look up the macro at *p called by "from", 245 * or as a line macro if from == TOKEN_NONE. 246 */ 247 static int 248 lookup(struct roff_man *mdoc, int from, int line, int ppos, const char *p) 249 { 250 enum roff_tok res; 251 252 if (mdoc->flags & MDOC_PHRASEQF) { 253 mdoc->flags &= ~MDOC_PHRASEQF; 254 return TOKEN_NONE; 255 } 256 if (from == TOKEN_NONE || mdoc_macro(from)->flags & MDOC_PARSED) { 257 res = roffhash_find(mdoc->mdocmac, p, 0); 258 if (res != TOKEN_NONE) { 259 if (mdoc_macro(res)->flags & MDOC_CALLABLE) 260 return res; 261 mandoc_msg(MANDOCERR_MACRO_CALL, line, ppos, "%s", p); 262 } 263 } 264 return TOKEN_NONE; 265 } 266 267 /* 268 * Rewind up to and including a specific node. 269 */ 270 static void 271 rew_last(struct roff_man *mdoc, const struct roff_node *to) 272 { 273 274 if (to->flags & NODE_VALID) 275 return; 276 277 while (mdoc->last != to) { 278 mdoc_state(mdoc, mdoc->last); 279 mdoc->last->flags |= NODE_VALID | NODE_ENDED; 280 mdoc->last = mdoc->last->parent; 281 } 282 mdoc_state(mdoc, mdoc->last); 283 mdoc->last->flags |= NODE_VALID | NODE_ENDED; 284 mdoc->next = ROFF_NEXT_SIBLING; 285 } 286 287 /* 288 * Rewind up to a specific block, including all blocks that broke it. 289 */ 290 static void 291 rew_pending(struct roff_man *mdoc, const struct roff_node *n) 292 { 293 294 for (;;) { 295 rew_last(mdoc, n); 296 297 if (mdoc->last == n) { 298 switch (n->type) { 299 case ROFFT_HEAD: 300 roff_body_alloc(mdoc, n->line, n->pos, 301 n->tok); 302 if (n->tok == MDOC_Ss) 303 mdoc->flags &= ~ROFF_NONOFILL; 304 break; 305 case ROFFT_BLOCK: 306 break; 307 default: 308 return; 309 } 310 if ( ! (n->flags & NODE_BROKEN)) 311 return; 312 } else 313 n = mdoc->last; 314 315 for (;;) { 316 if ((n = n->parent) == NULL) 317 return; 318 319 if (n->type == ROFFT_BLOCK || 320 n->type == ROFFT_HEAD) { 321 if (n->flags & NODE_ENDED) 322 break; 323 else 324 return; 325 } 326 } 327 } 328 } 329 330 /* 331 * For a block closing macro, return the corresponding opening one. 332 * Otherwise, return the macro itself. 333 */ 334 static enum roff_tok 335 rew_alt(enum roff_tok tok) 336 { 337 switch (tok) { 338 case MDOC_Ac: 339 return MDOC_Ao; 340 case MDOC_Bc: 341 return MDOC_Bo; 342 case MDOC_Brc: 343 return MDOC_Bro; 344 case MDOC_Dc: 345 return MDOC_Do; 346 case MDOC_Ec: 347 return MDOC_Eo; 348 case MDOC_Ed: 349 return MDOC_Bd; 350 case MDOC_Ef: 351 return MDOC_Bf; 352 case MDOC_Ek: 353 return MDOC_Bk; 354 case MDOC_El: 355 return MDOC_Bl; 356 case MDOC_Fc: 357 return MDOC_Fo; 358 case MDOC_Oc: 359 return MDOC_Oo; 360 case MDOC_Pc: 361 return MDOC_Po; 362 case MDOC_Qc: 363 return MDOC_Qo; 364 case MDOC_Re: 365 return MDOC_Rs; 366 case MDOC_Sc: 367 return MDOC_So; 368 case MDOC_Xc: 369 return MDOC_Xo; 370 default: 371 return tok; 372 } 373 } 374 375 static void 376 rew_elem(struct roff_man *mdoc, enum roff_tok tok) 377 { 378 struct roff_node *n; 379 380 n = mdoc->last; 381 if (n->type != ROFFT_ELEM) 382 n = n->parent; 383 assert(n->type == ROFFT_ELEM); 384 assert(tok == n->tok); 385 rew_last(mdoc, n); 386 } 387 388 static void 389 break_intermediate(struct roff_node *n, struct roff_node *breaker) 390 { 391 if (n != breaker && 392 n->type != ROFFT_BLOCK && n->type != ROFFT_HEAD && 393 (n->type != ROFFT_BODY || n->end != ENDBODY_NOT)) 394 n = n->parent; 395 while (n != breaker) { 396 if ( ! (n->flags & NODE_VALID)) 397 n->flags |= NODE_BROKEN; 398 n = n->parent; 399 } 400 } 401 402 /* 403 * If there is an open sub-block of the target requiring 404 * explicit close-out, postpone closing out the target until 405 * the rew_pending() call closing out the sub-block. 406 */ 407 static int 408 find_pending(struct roff_man *mdoc, enum roff_tok tok, int line, int ppos, 409 struct roff_node *target) 410 { 411 struct roff_node *n; 412 int irc; 413 414 if (target->flags & NODE_VALID) 415 return 0; 416 417 irc = 0; 418 for (n = mdoc->last; n != NULL && n != target; n = n->parent) { 419 if (n->flags & NODE_ENDED) 420 continue; 421 if (n->type == ROFFT_BLOCK && 422 mdoc_macro(n->tok)->flags & MDOC_EXPLICIT) { 423 irc = 1; 424 break_intermediate(mdoc->last, target); 425 if (target->type == ROFFT_HEAD) 426 target->flags |= NODE_ENDED; 427 else if ( ! (target->flags & NODE_ENDED)) { 428 mandoc_msg(MANDOCERR_BLK_NEST, 429 line, ppos, "%s breaks %s", 430 roff_name[tok], roff_name[n->tok]); 431 mdoc_endbody_alloc(mdoc, line, ppos, 432 tok, target); 433 } 434 } 435 } 436 return irc; 437 } 438 439 /* 440 * Allocate a word and check whether it's punctuation or not. 441 * Punctuation consists of those tokens found in mdoc_isdelim(). 442 */ 443 static void 444 dword(struct roff_man *mdoc, int line, int col, const char *p, 445 enum mdelim d, int may_append) 446 { 447 448 if (d == DELIM_MAX) 449 d = mdoc_isdelim(p); 450 451 if (may_append && 452 ! (mdoc->flags & (MDOC_SYNOPSIS | MDOC_KEEP | MDOC_SMOFF)) && 453 d == DELIM_NONE && mdoc->last->type == ROFFT_TEXT && 454 mdoc_isdelim(mdoc->last->string) == DELIM_NONE) { 455 roff_word_append(mdoc, p); 456 return; 457 } 458 459 roff_word_alloc(mdoc, line, col, p); 460 461 /* 462 * If the word consists of a bare delimiter, 463 * flag the new node accordingly, 464 * unless doing so was vetoed by the invoking macro. 465 * Always clear the veto, it is only valid for one word. 466 */ 467 468 if (d == DELIM_OPEN) 469 mdoc->last->flags |= NODE_DELIMO; 470 else if (d == DELIM_CLOSE && 471 ! (mdoc->flags & MDOC_NODELIMC) && 472 mdoc->last->parent->tok != MDOC_Fd) 473 mdoc->last->flags |= NODE_DELIMC; 474 mdoc->flags &= ~MDOC_NODELIMC; 475 } 476 477 static void 478 append_delims(struct roff_man *mdoc, int line, int *pos, char *buf) 479 { 480 char *p; 481 int la; 482 enum margserr ac; 483 484 if (buf[*pos] == '\0') 485 return; 486 487 for (;;) { 488 la = *pos; 489 ac = mdoc_args(mdoc, line, pos, buf, TOKEN_NONE, &p); 490 if (ac == ARGS_EOLN) 491 break; 492 dword(mdoc, line, la, p, DELIM_MAX, 1); 493 494 /* 495 * If we encounter end-of-sentence symbols, then trigger 496 * the double-space. 497 * 498 * XXX: it's easy to allow this to propagate outward to 499 * the last symbol, such that `. )' will cause the 500 * correct double-spacing. However, (1) groff isn't 501 * smart enough to do this and (2) it would require 502 * knowing which symbols break this behaviour, for 503 * example, `. ;' shouldn't propagate the double-space. 504 */ 505 506 if (mandoc_eos(p, strlen(p))) 507 mdoc->last->flags |= NODE_EOS; 508 if (ac == ARGS_ALLOC) 509 free(p); 510 } 511 } 512 513 /* 514 * Parse one word. 515 * If it is a macro, call it and return 1. 516 * Otherwise, allocate it and return 0. 517 */ 518 static int 519 macro_or_word(MACRO_PROT_ARGS, char *p, int parsed) 520 { 521 int ntok; 522 523 ntok = buf[ppos] == '"' || parsed == 0 || 524 mdoc->flags & MDOC_PHRASELIT ? TOKEN_NONE : 525 lookup(mdoc, tok, line, ppos, p); 526 527 if (ntok == TOKEN_NONE) { 528 dword(mdoc, line, ppos, p, DELIM_MAX, tok == TOKEN_NONE || 529 mdoc_macro(tok)->flags & MDOC_JOIN); 530 return 0; 531 } else { 532 if (tok != TOKEN_NONE && 533 mdoc_macro(tok)->fp == in_line_eoln) 534 rew_elem(mdoc, tok); 535 (*mdoc_macro(ntok)->fp)(mdoc, ntok, line, ppos, pos, buf); 536 if (tok == TOKEN_NONE) 537 append_delims(mdoc, line, pos, buf); 538 return 1; 539 } 540 } 541 542 /* 543 * Close out block partial/full explicit. 544 */ 545 static void 546 blk_exp_close(MACRO_PROT_ARGS) 547 { 548 struct roff_node *body; /* Our own body. */ 549 struct roff_node *endbody; /* Our own end marker. */ 550 struct roff_node *itblk; /* An It block starting later. */ 551 struct roff_node *later; /* A sub-block starting later. */ 552 struct roff_node *n; /* Search back to our block. */ 553 struct roff_node *target; /* For find_pending(). */ 554 555 int j, lastarg, maxargs, nl, pending; 556 enum margserr ac; 557 enum roff_tok atok, ntok; 558 char *p; 559 560 nl = MDOC_NEWLINE & mdoc->flags; 561 562 switch (tok) { 563 case MDOC_Ec: 564 maxargs = 1; 565 break; 566 case MDOC_Ek: 567 mdoc->flags &= ~MDOC_KEEP; 568 /* FALLTHROUGH */ 569 default: 570 maxargs = 0; 571 break; 572 } 573 574 /* Search backwards for the beginning of our own body. */ 575 576 atok = rew_alt(tok); 577 body = NULL; 578 for (n = mdoc->last; n; n = n->parent) { 579 if (n->flags & NODE_ENDED || n->tok != atok || 580 n->type != ROFFT_BODY || n->end != ENDBODY_NOT) 581 continue; 582 body = n; 583 break; 584 } 585 586 /* 587 * Search backwards for beginnings of blocks, 588 * both of our own and of pending sub-blocks. 589 */ 590 591 endbody = itblk = later = NULL; 592 for (n = mdoc->last; n; n = n->parent) { 593 if (n->flags & NODE_ENDED) 594 continue; 595 596 /* 597 * Mismatching end macros can never break anything 598 * and we only care about the breaking of BLOCKs. 599 */ 600 601 if (body == NULL || n->type != ROFFT_BLOCK) 602 continue; 603 604 /* 605 * SYNOPSIS name blocks can not be broken themselves, 606 * but they do get broken together with a broken child. 607 */ 608 609 if (n->tok == MDOC_Nm) { 610 if (later != NULL) 611 n->flags |= NODE_BROKEN | NODE_ENDED; 612 continue; 613 } 614 615 if (n->tok == MDOC_It) { 616 itblk = n; 617 continue; 618 } 619 620 if (atok == n->tok) { 621 622 /* 623 * Found the start of our own block. 624 * When there is no pending sub block, 625 * just proceed to closing out. 626 */ 627 628 if (later == NULL || 629 (tok == MDOC_El && itblk == NULL)) 630 break; 631 632 /* 633 * When there is a pending sub block, postpone 634 * closing out the current block until the 635 * rew_pending() closing out the sub-block. 636 * Mark the place where the formatting - but not 637 * the scope - of the current block ends. 638 */ 639 640 mandoc_msg(MANDOCERR_BLK_NEST, 641 line, ppos, "%s breaks %s", 642 roff_name[atok], roff_name[later->tok]); 643 644 endbody = mdoc_endbody_alloc(mdoc, line, ppos, 645 atok, body); 646 647 if (tok == MDOC_El) 648 itblk->flags |= NODE_ENDED | NODE_BROKEN; 649 650 /* 651 * If a block closing macro taking arguments 652 * breaks another block, put the arguments 653 * into the end marker. 654 */ 655 656 if (maxargs) 657 mdoc->next = ROFF_NEXT_CHILD; 658 break; 659 } 660 661 /* 662 * Explicit blocks close out description lines, but 663 * even those can get broken together with a child. 664 */ 665 666 if (n->tok == MDOC_Nd) { 667 if (later != NULL) 668 n->flags |= NODE_BROKEN | NODE_ENDED; 669 else 670 rew_last(mdoc, n); 671 continue; 672 } 673 674 /* Breaking an open sub block. */ 675 676 break_intermediate(mdoc->last, body); 677 n->flags |= NODE_BROKEN; 678 if (later == NULL) 679 later = n; 680 } 681 682 if (body == NULL) { 683 mandoc_msg(MANDOCERR_BLK_NOTOPEN, line, ppos, 684 "%s", roff_name[tok]); 685 if (maxargs && endbody == NULL) { 686 /* 687 * Stray .Ec without previous .Eo: 688 * Break the output line, keep the arguments. 689 */ 690 roff_elem_alloc(mdoc, line, ppos, ROFF_br); 691 rew_elem(mdoc, ROFF_br); 692 } 693 } else if (endbody == NULL) { 694 rew_last(mdoc, body); 695 if (maxargs) 696 mdoc_tail_alloc(mdoc, line, ppos, atok); 697 } 698 699 if ((mdoc_macro(tok)->flags & MDOC_PARSED) == 0) { 700 if (buf[*pos] != '\0') 701 mandoc_msg(MANDOCERR_ARG_SKIP, line, ppos, 702 "%s %s", roff_name[tok], buf + *pos); 703 if (endbody == NULL && n != NULL) 704 rew_pending(mdoc, n); 705 706 /* 707 * Restore the fill mode that was set before the display. 708 * This needs to be done here rather than during validation 709 * such that subsequent nodes get the right flags. 710 */ 711 712 if (tok == MDOC_Ed && body != NULL) { 713 if (body->flags & NODE_NOFILL) 714 mdoc->flags |= ROFF_NOFILL; 715 else 716 mdoc->flags &= ~ROFF_NOFILL; 717 } 718 return; 719 } 720 721 if (endbody != NULL) 722 n = endbody; 723 724 ntok = TOKEN_NONE; 725 for (j = 0; ; j++) { 726 lastarg = *pos; 727 728 if (j == maxargs && n != NULL) 729 rew_last(mdoc, n); 730 731 ac = mdoc_args(mdoc, line, pos, buf, tok, &p); 732 if (ac == ARGS_PUNCT || ac == ARGS_EOLN) 733 break; 734 735 ntok = lookup(mdoc, tok, line, lastarg, p); 736 737 if (ntok == TOKEN_NONE) { 738 dword(mdoc, line, lastarg, p, DELIM_MAX, 739 mdoc_macro(tok)->flags & MDOC_JOIN); 740 if (ac == ARGS_ALLOC) 741 free(p); 742 continue; 743 } 744 if (ac == ARGS_ALLOC) 745 free(p); 746 747 if (n != NULL) 748 rew_last(mdoc, n); 749 mdoc->flags &= ~MDOC_NEWLINE; 750 (*mdoc_macro(ntok)->fp)(mdoc, ntok, line, lastarg, pos, buf); 751 break; 752 } 753 754 if (n != NULL) { 755 pending = 0; 756 if (ntok != TOKEN_NONE && n->flags & NODE_BROKEN) { 757 target = n; 758 do 759 target = target->parent; 760 while ( ! (target->flags & NODE_ENDED)); 761 pending = find_pending(mdoc, ntok, line, ppos, target); 762 } 763 if ( ! pending) 764 rew_pending(mdoc, n); 765 } 766 if (nl) 767 append_delims(mdoc, line, pos, buf); 768 } 769 770 static void 771 in_line(MACRO_PROT_ARGS) 772 { 773 int la, scope, cnt, firstarg, mayopen, nc, nl; 774 enum roff_tok ntok; 775 enum margserr ac; 776 enum mdelim d; 777 struct mdoc_arg *arg; 778 char *p; 779 780 nl = MDOC_NEWLINE & mdoc->flags; 781 782 /* 783 * Whether we allow ignored elements (those without content, 784 * usually because of reserved words) to squeak by. 785 */ 786 787 switch (tok) { 788 case MDOC_An: 789 case MDOC_Ar: 790 case MDOC_Fl: 791 case MDOC_Mt: 792 case MDOC_Nm: 793 case MDOC_Pa: 794 nc = 1; 795 break; 796 default: 797 nc = 0; 798 break; 799 } 800 801 mdoc_argv(mdoc, line, tok, &arg, pos, buf); 802 803 d = DELIM_NONE; 804 firstarg = 1; 805 mayopen = 1; 806 for (cnt = scope = 0;; ) { 807 la = *pos; 808 ac = mdoc_args(mdoc, line, pos, buf, tok, &p); 809 810 /* 811 * At the end of a macro line, 812 * opening delimiters do not suppress spacing. 813 */ 814 815 if (ac == ARGS_EOLN) { 816 if (d == DELIM_OPEN) 817 mdoc->last->flags &= ~NODE_DELIMO; 818 break; 819 } 820 821 /* 822 * The rest of the macro line is only punctuation, 823 * to be handled by append_delims(). 824 * If there were no other arguments, 825 * do not allow the first one to suppress spacing, 826 * even if it turns out to be a closing one. 827 */ 828 829 if (ac == ARGS_PUNCT) { 830 if (cnt == 0 && (nc == 0 || tok == MDOC_An)) 831 mdoc->flags |= MDOC_NODELIMC; 832 break; 833 } 834 835 ntok = (tok == MDOC_Fn && !cnt) ? 836 TOKEN_NONE : lookup(mdoc, tok, line, la, p); 837 838 /* 839 * In this case, we've located a submacro and must 840 * execute it. Close out scope, if open. If no 841 * elements have been generated, either create one (nc) 842 * or raise a warning. 843 */ 844 845 if (ntok != TOKEN_NONE) { 846 if (scope) 847 rew_elem(mdoc, tok); 848 if (nc && ! cnt) { 849 mdoc_elem_alloc(mdoc, line, ppos, tok, arg); 850 rew_last(mdoc, mdoc->last); 851 } else if ( ! nc && ! cnt) { 852 mdoc_argv_free(arg); 853 mandoc_msg(MANDOCERR_MACRO_EMPTY, 854 line, ppos, "%s", roff_name[tok]); 855 } 856 (*mdoc_macro(ntok)->fp)(mdoc, ntok, 857 line, la, pos, buf); 858 if (nl) 859 append_delims(mdoc, line, pos, buf); 860 if (ac == ARGS_ALLOC) 861 free(p); 862 return; 863 } 864 865 /* 866 * Handle punctuation. Set up our scope, if a word; 867 * rewind the scope, if a delimiter; then append the word. 868 */ 869 870 if ((d = mdoc_isdelim(p)) != DELIM_NONE) { 871 /* 872 * If we encounter closing punctuation, no word 873 * has been emitted, no scope is open, and we're 874 * allowed to have an empty element, then start 875 * a new scope. 876 */ 877 if ((d == DELIM_CLOSE || 878 (d == DELIM_MIDDLE && tok == MDOC_Fl)) && 879 !cnt && !scope && nc && mayopen) { 880 mdoc_elem_alloc(mdoc, line, ppos, tok, arg); 881 scope = 1; 882 cnt++; 883 if (tok == MDOC_Nm) 884 mayopen = 0; 885 } 886 /* 887 * Close out our scope, if one is open, before 888 * any punctuation. 889 */ 890 if (scope && tok != MDOC_Lk) { 891 rew_elem(mdoc, tok); 892 scope = 0; 893 if (tok == MDOC_Fn) 894 mayopen = 0; 895 } 896 } else if (mayopen && !scope) { 897 mdoc_elem_alloc(mdoc, line, ppos, tok, arg); 898 scope = 1; 899 cnt++; 900 } 901 902 dword(mdoc, line, la, p, d, 903 mdoc_macro(tok)->flags & MDOC_JOIN); 904 905 if (ac == ARGS_ALLOC) 906 free(p); 907 908 /* 909 * If the first argument is a closing delimiter, 910 * do not suppress spacing before it. 911 */ 912 913 if (firstarg && d == DELIM_CLOSE && !nc) 914 mdoc->last->flags &= ~NODE_DELIMC; 915 firstarg = 0; 916 917 /* 918 * `Fl' macros have their scope re-opened with each new 919 * word so that the `-' can be added to each one without 920 * having to parse out spaces. 921 */ 922 if (scope && tok == MDOC_Fl) { 923 rew_elem(mdoc, tok); 924 scope = 0; 925 } 926 } 927 928 if (scope && tok != MDOC_Lk) { 929 rew_elem(mdoc, tok); 930 scope = 0; 931 } 932 933 /* 934 * If no elements have been collected and we're allowed to have 935 * empties (nc), open a scope and close it out. Otherwise, 936 * raise a warning. 937 */ 938 939 if ( ! cnt) { 940 if (nc) { 941 mdoc_elem_alloc(mdoc, line, ppos, tok, arg); 942 rew_last(mdoc, mdoc->last); 943 } else { 944 mdoc_argv_free(arg); 945 mandoc_msg(MANDOCERR_MACRO_EMPTY, 946 line, ppos, "%s", roff_name[tok]); 947 } 948 } 949 if (nl) 950 append_delims(mdoc, line, pos, buf); 951 if (scope) 952 rew_elem(mdoc, tok); 953 } 954 955 static void 956 blk_full(MACRO_PROT_ARGS) 957 { 958 struct mdoc_arg *arg; 959 struct roff_node *blk; /* Our own or a broken block. */ 960 struct roff_node *head; /* Our own head. */ 961 struct roff_node *body; /* Our own body. */ 962 struct roff_node *n; 963 char *p; 964 size_t iarg; 965 int done, la, nl, parsed; 966 enum margserr ac, lac; 967 968 nl = MDOC_NEWLINE & mdoc->flags; 969 970 if (buf[*pos] == '\0' && (tok == MDOC_Sh || tok == MDOC_Ss)) { 971 mandoc_msg(MANDOCERR_MACRO_EMPTY, 972 line, ppos, "%s", roff_name[tok]); 973 return; 974 } 975 976 if ((mdoc_macro(tok)->flags & MDOC_EXPLICIT) == 0) { 977 978 /* Here, tok is one of Sh Ss Nm Nd It. */ 979 980 blk = NULL; 981 for (n = mdoc->last; n != NULL; n = n->parent) { 982 if (n->flags & NODE_ENDED) { 983 if ( ! (n->flags & NODE_VALID)) 984 n->flags |= NODE_BROKEN; 985 continue; 986 } 987 if (n->type != ROFFT_BLOCK) 988 continue; 989 990 if (tok == MDOC_It && n->tok == MDOC_Bl) { 991 if (blk != NULL) { 992 mandoc_msg(MANDOCERR_BLK_BROKEN, 993 line, ppos, "It breaks %s", 994 roff_name[blk->tok]); 995 rew_pending(mdoc, blk); 996 } 997 break; 998 } 999 1000 if (mdoc_macro(n->tok)->flags & MDOC_EXPLICIT) { 1001 switch (tok) { 1002 case MDOC_Sh: 1003 case MDOC_Ss: 1004 mandoc_msg(MANDOCERR_BLK_BROKEN, 1005 line, ppos, 1006 "%s breaks %s", roff_name[tok], 1007 roff_name[n->tok]); 1008 rew_pending(mdoc, n); 1009 n = mdoc->last; 1010 continue; 1011 case MDOC_It: 1012 /* Delay in case it's astray. */ 1013 blk = n; 1014 continue; 1015 default: 1016 break; 1017 } 1018 break; 1019 } 1020 1021 /* Here, n is one of Sh Ss Nm Nd It. */ 1022 1023 if (tok != MDOC_Sh && (n->tok == MDOC_Sh || 1024 (tok != MDOC_Ss && (n->tok == MDOC_Ss || 1025 (tok != MDOC_It && n->tok == MDOC_It))))) 1026 break; 1027 1028 /* Item breaking an explicit block. */ 1029 1030 if (blk != NULL) { 1031 mandoc_msg(MANDOCERR_BLK_BROKEN, line, ppos, 1032 "It breaks %s", roff_name[blk->tok]); 1033 rew_pending(mdoc, blk); 1034 blk = NULL; 1035 } 1036 1037 /* Close out prior implicit scopes. */ 1038 1039 rew_pending(mdoc, n); 1040 } 1041 1042 /* Skip items outside lists. */ 1043 1044 if (tok == MDOC_It && (n == NULL || n->tok != MDOC_Bl)) { 1045 mandoc_msg(MANDOCERR_IT_STRAY, 1046 line, ppos, "It %s", buf + *pos); 1047 roff_elem_alloc(mdoc, line, ppos, ROFF_br); 1048 rew_elem(mdoc, ROFF_br); 1049 return; 1050 } 1051 } 1052 1053 /* 1054 * This routine accommodates implicitly- and explicitly-scoped 1055 * macro openings. Implicit ones first close out prior scope 1056 * (seen above). Delay opening the head until necessary to 1057 * allow leading punctuation to print. Special consideration 1058 * for `It -column', which has phrase-part syntax instead of 1059 * regular child nodes. 1060 */ 1061 1062 switch (tok) { 1063 case MDOC_Sh: 1064 mdoc->flags &= ~ROFF_NOFILL; 1065 break; 1066 case MDOC_Ss: 1067 mdoc->flags |= ROFF_NONOFILL; 1068 break; 1069 default: 1070 break; 1071 } 1072 mdoc_argv(mdoc, line, tok, &arg, pos, buf); 1073 blk = mdoc_block_alloc(mdoc, line, ppos, tok, arg); 1074 head = body = NULL; 1075 1076 /* 1077 * Exception: Heads of `It' macros in `-diag' lists are not 1078 * parsed, even though `It' macros in general are parsed. 1079 */ 1080 1081 parsed = tok != MDOC_It || 1082 mdoc->last->parent->tok != MDOC_Bl || 1083 mdoc->last->parent->norm->Bl.type != LIST_diag; 1084 1085 /* 1086 * The `Nd' macro has all arguments in its body: it's a hybrid 1087 * of block partial-explicit and full-implicit. Stupid. 1088 */ 1089 1090 if (tok == MDOC_Nd) { 1091 head = roff_head_alloc(mdoc, line, ppos, tok); 1092 rew_last(mdoc, head); 1093 body = roff_body_alloc(mdoc, line, ppos, tok); 1094 } 1095 1096 if (tok == MDOC_Bk) 1097 mdoc->flags |= MDOC_KEEP; 1098 1099 ac = ARGS_EOLN; 1100 for (;;) { 1101 1102 /* 1103 * If we are right after a tab character, 1104 * do not parse the first word for macros. 1105 */ 1106 1107 if (mdoc->flags & MDOC_PHRASEQN) { 1108 mdoc->flags &= ~MDOC_PHRASEQN; 1109 mdoc->flags |= MDOC_PHRASEQF; 1110 } 1111 1112 la = *pos; 1113 lac = ac; 1114 ac = mdoc_args(mdoc, line, pos, buf, tok, &p); 1115 if (ac == ARGS_EOLN) { 1116 if (lac != ARGS_PHRASE || 1117 ! (mdoc->flags & MDOC_PHRASEQF)) 1118 break; 1119 1120 /* 1121 * This line ends in a tab; start the next 1122 * column now, with a leading blank. 1123 */ 1124 1125 if (body != NULL) 1126 rew_last(mdoc, body); 1127 body = roff_body_alloc(mdoc, line, ppos, tok); 1128 roff_word_alloc(mdoc, line, ppos, "\\&"); 1129 break; 1130 } 1131 1132 if (tok == MDOC_Bd || tok == MDOC_Bk) { 1133 mandoc_msg(MANDOCERR_ARG_EXCESS, line, la, 1134 "%s ... %s", roff_name[tok], buf + la); 1135 if (ac == ARGS_ALLOC) 1136 free(p); 1137 break; 1138 } 1139 if (tok == MDOC_Rs) { 1140 mandoc_msg(MANDOCERR_ARG_SKIP, 1141 line, la, "Rs %s", buf + la); 1142 if (ac == ARGS_ALLOC) 1143 free(p); 1144 break; 1145 } 1146 if (ac == ARGS_PUNCT) 1147 break; 1148 1149 /* 1150 * Emit leading punctuation (i.e., punctuation before 1151 * the ROFFT_HEAD) for non-phrase types. 1152 */ 1153 1154 if (head == NULL && 1155 ac != ARGS_PHRASE && 1156 mdoc_isdelim(p) == DELIM_OPEN) { 1157 dword(mdoc, line, la, p, DELIM_OPEN, 0); 1158 if (ac == ARGS_ALLOC) 1159 free(p); 1160 continue; 1161 } 1162 1163 /* Open a head if one hasn't been opened. */ 1164 1165 if (head == NULL) 1166 head = roff_head_alloc(mdoc, line, ppos, tok); 1167 1168 if (ac == ARGS_PHRASE) { 1169 1170 /* 1171 * If we haven't opened a body yet, rewind the 1172 * head; if we have, rewind that instead. 1173 */ 1174 1175 rew_last(mdoc, body == NULL ? head : body); 1176 body = roff_body_alloc(mdoc, line, ppos, tok); 1177 1178 /* Process to the tab or to the end of the line. */ 1179 1180 mdoc->flags |= MDOC_PHRASE; 1181 parse_rest(mdoc, TOKEN_NONE, line, &la, buf); 1182 mdoc->flags &= ~MDOC_PHRASE; 1183 1184 /* There may have been `Ta' macros. */ 1185 1186 while (body->next != NULL) 1187 body = body->next; 1188 continue; 1189 } 1190 1191 done = macro_or_word(mdoc, tok, line, la, pos, buf, p, parsed); 1192 if (ac == ARGS_ALLOC) 1193 free(p); 1194 if (done) 1195 break; 1196 } 1197 1198 if (blk->flags & NODE_VALID) 1199 return; 1200 if (head == NULL) 1201 head = roff_head_alloc(mdoc, line, ppos, tok); 1202 if (nl && tok != MDOC_Bd && tok != MDOC_Bl && tok != MDOC_Rs) 1203 append_delims(mdoc, line, pos, buf); 1204 if (body != NULL) 1205 goto out; 1206 if (find_pending(mdoc, tok, line, ppos, head)) 1207 return; 1208 1209 /* Close out scopes to remain in a consistent state. */ 1210 1211 rew_last(mdoc, head); 1212 body = roff_body_alloc(mdoc, line, ppos, tok); 1213 if (tok == MDOC_Ss) 1214 mdoc->flags &= ~ROFF_NONOFILL; 1215 1216 /* 1217 * Set up fill mode for display blocks. 1218 * This needs to be done here up front rather than during 1219 * validation such that child nodes get the right flags. 1220 */ 1221 1222 if (tok == MDOC_Bd && arg != NULL) { 1223 for (iarg = 0; iarg < arg->argc; iarg++) { 1224 switch (arg->argv[iarg].arg) { 1225 case MDOC_Unfilled: 1226 case MDOC_Literal: 1227 mdoc->flags |= ROFF_NOFILL; 1228 break; 1229 case MDOC_Filled: 1230 case MDOC_Ragged: 1231 case MDOC_Centred: 1232 mdoc->flags &= ~ROFF_NOFILL; 1233 break; 1234 default: 1235 continue; 1236 } 1237 break; 1238 } 1239 } 1240 out: 1241 if (mdoc->flags & MDOC_FREECOL) { 1242 rew_last(mdoc, body); 1243 rew_last(mdoc, blk); 1244 mdoc->flags &= ~MDOC_FREECOL; 1245 } 1246 } 1247 1248 static void 1249 blk_part_imp(MACRO_PROT_ARGS) 1250 { 1251 int done, la, nl; 1252 enum margserr ac; 1253 char *p; 1254 struct roff_node *blk; /* saved block context */ 1255 struct roff_node *body; /* saved body context */ 1256 struct roff_node *n; 1257 1258 nl = MDOC_NEWLINE & mdoc->flags; 1259 1260 /* 1261 * A macro that spans to the end of the line. This is generally 1262 * (but not necessarily) called as the first macro. The block 1263 * has a head as the immediate child, which is always empty, 1264 * followed by zero or more opening punctuation nodes, then the 1265 * body (which may be empty, depending on the macro), then zero 1266 * or more closing punctuation nodes. 1267 */ 1268 1269 blk = mdoc_block_alloc(mdoc, line, ppos, tok, NULL); 1270 rew_last(mdoc, roff_head_alloc(mdoc, line, ppos, tok)); 1271 1272 /* 1273 * Open the body scope "on-demand", that is, after we've 1274 * processed all our the leading delimiters (open parenthesis, 1275 * etc.). 1276 */ 1277 1278 for (body = NULL; ; ) { 1279 la = *pos; 1280 ac = mdoc_args(mdoc, line, pos, buf, tok, &p); 1281 if (ac == ARGS_EOLN || ac == ARGS_PUNCT) 1282 break; 1283 1284 if (body == NULL && mdoc_isdelim(p) == DELIM_OPEN) { 1285 dword(mdoc, line, la, p, DELIM_OPEN, 0); 1286 if (ac == ARGS_ALLOC) 1287 free(p); 1288 continue; 1289 } 1290 1291 if (body == NULL) 1292 body = roff_body_alloc(mdoc, line, ppos, tok); 1293 1294 done = macro_or_word(mdoc, tok, line, la, pos, buf, p, 1); 1295 if (ac == ARGS_ALLOC) 1296 free(p); 1297 if (done) 1298 break; 1299 } 1300 if (body == NULL) 1301 body = roff_body_alloc(mdoc, line, ppos, tok); 1302 1303 if (find_pending(mdoc, tok, line, ppos, body)) 1304 return; 1305 1306 rew_last(mdoc, body); 1307 if (nl) 1308 append_delims(mdoc, line, pos, buf); 1309 rew_pending(mdoc, blk); 1310 1311 /* Move trailing .Ns out of scope. */ 1312 1313 for (n = body->child; n && n->next; n = n->next) 1314 /* Do nothing. */ ; 1315 if (n && n->tok == MDOC_Ns) 1316 roff_node_relink(mdoc, n); 1317 } 1318 1319 static void 1320 blk_part_exp(MACRO_PROT_ARGS) 1321 { 1322 int done, la, nl; 1323 enum margserr ac; 1324 struct roff_node *head; /* keep track of head */ 1325 char *p; 1326 1327 nl = MDOC_NEWLINE & mdoc->flags; 1328 1329 /* 1330 * The opening of an explicit macro having zero or more leading 1331 * punctuation nodes; a head with optional single element (the 1332 * case of `Eo'); and a body that may be empty. 1333 */ 1334 1335 roff_block_alloc(mdoc, line, ppos, tok); 1336 head = NULL; 1337 for (;;) { 1338 la = *pos; 1339 ac = mdoc_args(mdoc, line, pos, buf, tok, &p); 1340 if (ac == ARGS_PUNCT || ac == ARGS_EOLN) 1341 break; 1342 1343 /* Flush out leading punctuation. */ 1344 1345 if (head == NULL && mdoc_isdelim(p) == DELIM_OPEN) { 1346 dword(mdoc, line, la, p, DELIM_OPEN, 0); 1347 if (ac == ARGS_ALLOC) 1348 free(p); 1349 continue; 1350 } 1351 1352 if (head == NULL) { 1353 head = roff_head_alloc(mdoc, line, ppos, tok); 1354 if (tok == MDOC_Eo) /* Not parsed. */ 1355 dword(mdoc, line, la, p, DELIM_MAX, 0); 1356 rew_last(mdoc, head); 1357 roff_body_alloc(mdoc, line, ppos, tok); 1358 if (tok == MDOC_Eo) { 1359 if (ac == ARGS_ALLOC) 1360 free(p); 1361 continue; 1362 } 1363 } 1364 1365 done = macro_or_word(mdoc, tok, line, la, pos, buf, p, 1); 1366 if (ac == ARGS_ALLOC) 1367 free(p); 1368 if (done) 1369 break; 1370 } 1371 1372 /* Clean-up to leave in a consistent state. */ 1373 1374 if (head == NULL) { 1375 rew_last(mdoc, roff_head_alloc(mdoc, line, ppos, tok)); 1376 roff_body_alloc(mdoc, line, ppos, tok); 1377 } 1378 if (nl) 1379 append_delims(mdoc, line, pos, buf); 1380 } 1381 1382 static void 1383 in_line_argn(MACRO_PROT_ARGS) 1384 { 1385 struct mdoc_arg *arg; 1386 char *p; 1387 enum margserr ac; 1388 enum roff_tok ntok; 1389 int state; /* arg#; -1: not yet open; -2: closed */ 1390 int la, maxargs, nl; 1391 1392 nl = mdoc->flags & MDOC_NEWLINE; 1393 1394 /* 1395 * A line macro that has a fixed number of arguments (maxargs). 1396 * Only open the scope once the first non-leading-punctuation is 1397 * found (unless MDOC_IGNDELIM is noted, like in `Pf'), then 1398 * keep it open until the maximum number of arguments are 1399 * exhausted. 1400 */ 1401 1402 switch (tok) { 1403 case MDOC_Ap: 1404 case MDOC_Ns: 1405 case MDOC_Ux: 1406 maxargs = 0; 1407 break; 1408 case MDOC_Bx: 1409 case MDOC_Es: 1410 case MDOC_Xr: 1411 maxargs = 2; 1412 break; 1413 default: 1414 maxargs = 1; 1415 break; 1416 } 1417 1418 mdoc_argv(mdoc, line, tok, &arg, pos, buf); 1419 1420 state = -1; 1421 p = NULL; 1422 for (;;) { 1423 la = *pos; 1424 ac = mdoc_args(mdoc, line, pos, buf, tok, &p); 1425 1426 if ((ac == ARGS_WORD || ac == ARGS_ALLOC) && state == -1 && 1427 (mdoc_macro(tok)->flags & MDOC_IGNDELIM) == 0 && 1428 mdoc_isdelim(p) == DELIM_OPEN) { 1429 dword(mdoc, line, la, p, DELIM_OPEN, 0); 1430 if (ac == ARGS_ALLOC) 1431 free(p); 1432 continue; 1433 } 1434 1435 if (state == -1 && tok != MDOC_In && 1436 tok != MDOC_St && tok != MDOC_Xr) { 1437 mdoc_elem_alloc(mdoc, line, ppos, tok, arg); 1438 state = 0; 1439 } 1440 1441 if (ac == ARGS_PUNCT || ac == ARGS_EOLN) { 1442 if (abs(state) < 2 && tok == MDOC_Pf) 1443 mandoc_msg(MANDOCERR_PF_SKIP, 1444 line, ppos, "Pf %s", 1445 p == NULL ? "at eol" : p); 1446 break; 1447 } 1448 1449 if (state == maxargs) { 1450 rew_elem(mdoc, tok); 1451 state = -2; 1452 } 1453 1454 ntok = (tok == MDOC_Pf && state == 0) ? 1455 TOKEN_NONE : lookup(mdoc, tok, line, la, p); 1456 1457 if (ntok != TOKEN_NONE) { 1458 if (state >= 0) { 1459 rew_elem(mdoc, tok); 1460 state = -2; 1461 } 1462 (*mdoc_macro(ntok)->fp)(mdoc, ntok, 1463 line, la, pos, buf); 1464 if (ac == ARGS_ALLOC) 1465 free(p); 1466 break; 1467 } 1468 1469 if (mdoc_macro(tok)->flags & MDOC_IGNDELIM || 1470 mdoc_isdelim(p) == DELIM_NONE) { 1471 if (state == -1) { 1472 mdoc_elem_alloc(mdoc, line, ppos, tok, arg); 1473 state = 1; 1474 } else if (state >= 0) 1475 state++; 1476 } else if (state >= 0) { 1477 rew_elem(mdoc, tok); 1478 state = -2; 1479 } 1480 1481 dword(mdoc, line, la, p, DELIM_MAX, 1482 mdoc_macro(tok)->flags & MDOC_JOIN); 1483 if (ac == ARGS_ALLOC) 1484 free(p); 1485 p = mdoc->last->string; 1486 } 1487 1488 if (state == -1) { 1489 mandoc_msg(MANDOCERR_MACRO_EMPTY, 1490 line, ppos, "%s", roff_name[tok]); 1491 return; 1492 } 1493 1494 if (state == 0 && tok == MDOC_Pf) 1495 append_delims(mdoc, line, pos, buf); 1496 if (state >= 0) 1497 rew_elem(mdoc, tok); 1498 if (nl) 1499 append_delims(mdoc, line, pos, buf); 1500 } 1501 1502 static void 1503 in_line_eoln(MACRO_PROT_ARGS) 1504 { 1505 struct roff_node *n; 1506 struct mdoc_arg *arg; 1507 1508 if ((tok == MDOC_Pp || tok == MDOC_Lp) && 1509 ! (mdoc->flags & MDOC_SYNOPSIS)) { 1510 n = mdoc->last; 1511 if (mdoc->next == ROFF_NEXT_SIBLING) 1512 n = n->parent; 1513 if (n->tok == MDOC_Nm) 1514 rew_last(mdoc, n->parent); 1515 } 1516 1517 #if DEBUG_MEMORY 1518 if (tok == MDOC_Dt) 1519 mandoc_dbg_name(buf); 1520 #endif 1521 1522 if (buf[*pos] == '\0' && 1523 (tok == MDOC_Fd || *roff_name[tok] == '%')) { 1524 mandoc_msg(MANDOCERR_MACRO_EMPTY, 1525 line, ppos, "%s", roff_name[tok]); 1526 return; 1527 } 1528 1529 mdoc_argv(mdoc, line, tok, &arg, pos, buf); 1530 mdoc_elem_alloc(mdoc, line, ppos, tok, arg); 1531 if (parse_rest(mdoc, tok, line, pos, buf)) 1532 return; 1533 rew_elem(mdoc, tok); 1534 } 1535 1536 /* 1537 * The simplest argument parser available: Parse the remaining 1538 * words until the end of the phrase or line and return 0 1539 * or until the next macro, call that macro, and return 1. 1540 */ 1541 static int 1542 parse_rest(struct roff_man *mdoc, enum roff_tok tok, 1543 int line, int *pos, char *buf) 1544 { 1545 char *p; 1546 int done, la; 1547 enum margserr ac; 1548 1549 for (;;) { 1550 la = *pos; 1551 ac = mdoc_args(mdoc, line, pos, buf, tok, &p); 1552 if (ac == ARGS_EOLN) 1553 return 0; 1554 done = macro_or_word(mdoc, tok, line, la, pos, buf, p, 1); 1555 if (ac == ARGS_ALLOC) 1556 free(p); 1557 if (done) 1558 return 1; 1559 } 1560 } 1561 1562 static void 1563 ctx_synopsis(MACRO_PROT_ARGS) 1564 { 1565 1566 if (~mdoc->flags & (MDOC_SYNOPSIS | MDOC_NEWLINE)) 1567 in_line(mdoc, tok, line, ppos, pos, buf); 1568 else if (tok == MDOC_Nm) 1569 blk_full(mdoc, tok, line, ppos, pos, buf); 1570 else { 1571 assert(tok == MDOC_Vt); 1572 blk_part_imp(mdoc, tok, line, ppos, pos, buf); 1573 } 1574 } 1575 1576 /* 1577 * Phrases occur within `Bl -column' entries, separated by `Ta' or tabs. 1578 * They're unusual because they're basically free-form text until a 1579 * macro is encountered. 1580 */ 1581 static void 1582 phrase_ta(MACRO_PROT_ARGS) 1583 { 1584 struct roff_node *body, *n; 1585 1586 /* Make sure we are in a column list or ignore this macro. */ 1587 1588 body = NULL; 1589 for (n = mdoc->last; n != NULL; n = n->parent) { 1590 if (n->flags & NODE_ENDED) 1591 continue; 1592 if (n->tok == MDOC_It && n->type == ROFFT_BODY) 1593 body = n; 1594 if (n->tok == MDOC_Bl && n->end == ENDBODY_NOT) 1595 break; 1596 } 1597 1598 if (n == NULL || n->norm->Bl.type != LIST_column) { 1599 mandoc_msg(MANDOCERR_TA_STRAY, line, ppos, "Ta"); 1600 return; 1601 } 1602 1603 /* Advance to the next column. */ 1604 1605 rew_last(mdoc, body); 1606 roff_body_alloc(mdoc, line, ppos, MDOC_It); 1607 parse_rest(mdoc, TOKEN_NONE, line, pos, buf); 1608 } 1609