1 /* $OpenBSD: mdoc_macro.c,v 1.156 2015/05/01 16:56:36 schwarze Exp $ */ 2 /* 3 * Copyright (c) 2008-2012 Kristaps Dzonsons <kristaps@bsd.lv> 4 * Copyright (c) 2010, 2012-2015 Ingo Schwarze <schwarze@openbsd.org> 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 <sys/types.h> 19 20 #include <assert.h> 21 #include <ctype.h> 22 #include <stdlib.h> 23 #include <stdio.h> 24 #include <string.h> 25 #include <time.h> 26 27 #include "mandoc.h" 28 #include "roff.h" 29 #include "mdoc.h" 30 #include "libmandoc.h" 31 #include "roff_int.h" 32 #include "libmdoc.h" 33 34 static void blk_full(MACRO_PROT_ARGS); 35 static void blk_exp_close(MACRO_PROT_ARGS); 36 static void blk_part_exp(MACRO_PROT_ARGS); 37 static void blk_part_imp(MACRO_PROT_ARGS); 38 static void ctx_synopsis(MACRO_PROT_ARGS); 39 static void in_line_eoln(MACRO_PROT_ARGS); 40 static void in_line_argn(MACRO_PROT_ARGS); 41 static void in_line(MACRO_PROT_ARGS); 42 static void phrase_ta(MACRO_PROT_ARGS); 43 44 static void append_delims(struct roff_man *, int, int *, char *); 45 static void dword(struct roff_man *, int, int, const char *, 46 enum mdelim, int); 47 static int find_pending(struct roff_man *, int, int, int, 48 struct roff_node *); 49 static int lookup(struct roff_man *, int, int, int, const char *); 50 static int macro_or_word(MACRO_PROT_ARGS, int); 51 static int parse_rest(struct roff_man *, int, int, int *, char *); 52 static int rew_alt(int); 53 static void rew_elem(struct roff_man *, int); 54 static void rew_last(struct roff_man *, const struct roff_node *); 55 static void rew_pending(struct roff_man *, 56 const struct roff_node *); 57 58 const struct mdoc_macro __mdoc_macros[MDOC_MAX] = { 59 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Ap */ 60 { in_line_eoln, MDOC_PROLOGUE }, /* Dd */ 61 { in_line_eoln, MDOC_PROLOGUE }, /* Dt */ 62 { in_line_eoln, MDOC_PROLOGUE }, /* Os */ 63 { blk_full, MDOC_PARSED | MDOC_JOIN }, /* Sh */ 64 { blk_full, MDOC_PARSED | MDOC_JOIN }, /* Ss */ 65 { in_line_eoln, 0 }, /* Pp */ 66 { blk_part_imp, MDOC_PARSED | MDOC_JOIN }, /* D1 */ 67 { blk_part_imp, MDOC_PARSED | MDOC_JOIN }, /* Dl */ 68 { blk_full, MDOC_EXPLICIT }, /* Bd */ 69 { blk_exp_close, MDOC_EXPLICIT | MDOC_JOIN }, /* Ed */ 70 { blk_full, MDOC_EXPLICIT }, /* Bl */ 71 { blk_exp_close, MDOC_EXPLICIT | MDOC_JOIN }, /* El */ 72 { blk_full, MDOC_PARSED | MDOC_JOIN }, /* It */ 73 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ad */ 74 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* An */ 75 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ar */ 76 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Cd */ 77 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Cm */ 78 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Dv */ 79 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Er */ 80 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ev */ 81 { in_line_eoln, 0 }, /* Ex */ 82 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fa */ 83 { in_line_eoln, 0 }, /* Fd */ 84 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fl */ 85 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fn */ 86 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ft */ 87 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Ic */ 88 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* In */ 89 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Li */ 90 { blk_full, MDOC_JOIN }, /* Nd */ 91 { ctx_synopsis, MDOC_CALLABLE | MDOC_PARSED }, /* Nm */ 92 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Op */ 93 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ot */ 94 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Pa */ 95 { in_line_eoln, 0 }, /* Rv */ 96 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* St */ 97 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Va */ 98 { ctx_synopsis, MDOC_CALLABLE | MDOC_PARSED }, /* Vt */ 99 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Xr */ 100 { in_line_eoln, MDOC_JOIN }, /* %A */ 101 { in_line_eoln, MDOC_JOIN }, /* %B */ 102 { in_line_eoln, MDOC_JOIN }, /* %D */ 103 { in_line_eoln, MDOC_JOIN }, /* %I */ 104 { in_line_eoln, MDOC_JOIN }, /* %J */ 105 { in_line_eoln, 0 }, /* %N */ 106 { in_line_eoln, MDOC_JOIN }, /* %O */ 107 { in_line_eoln, 0 }, /* %P */ 108 { in_line_eoln, MDOC_JOIN }, /* %R */ 109 { in_line_eoln, MDOC_JOIN }, /* %T */ 110 { in_line_eoln, 0 }, /* %V */ 111 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED | 112 MDOC_EXPLICIT | MDOC_JOIN }, /* Ac */ 113 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | 114 MDOC_EXPLICIT | MDOC_JOIN }, /* Ao */ 115 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Aq */ 116 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* At */ 117 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED | 118 MDOC_EXPLICIT | MDOC_JOIN }, /* Bc */ 119 { blk_full, MDOC_EXPLICIT }, /* Bf */ 120 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | 121 MDOC_EXPLICIT | MDOC_JOIN }, /* Bo */ 122 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Bq */ 123 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Bsx */ 124 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Bx */ 125 { in_line_eoln, 0 }, /* Db */ 126 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED | 127 MDOC_EXPLICIT | MDOC_JOIN }, /* Dc */ 128 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | 129 MDOC_EXPLICIT | MDOC_JOIN }, /* Do */ 130 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Dq */ 131 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Ec */ 132 { blk_exp_close, MDOC_EXPLICIT | MDOC_JOIN }, /* Ef */ 133 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Em */ 134 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Eo */ 135 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Fx */ 136 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ms */ 137 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* No */ 138 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED | 139 MDOC_IGNDELIM | MDOC_JOIN }, /* Ns */ 140 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Nx */ 141 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Ox */ 142 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED | 143 MDOC_EXPLICIT | MDOC_JOIN }, /* Pc */ 144 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED | MDOC_IGNDELIM }, /* Pf */ 145 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | 146 MDOC_EXPLICIT | MDOC_JOIN }, /* Po */ 147 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Pq */ 148 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED | 149 MDOC_EXPLICIT | MDOC_JOIN }, /* Qc */ 150 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Ql */ 151 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | 152 MDOC_EXPLICIT | MDOC_JOIN }, /* Qo */ 153 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Qq */ 154 { blk_exp_close, MDOC_EXPLICIT | MDOC_JOIN }, /* Re */ 155 { blk_full, MDOC_EXPLICIT }, /* Rs */ 156 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED | 157 MDOC_EXPLICIT | MDOC_JOIN }, /* Sc */ 158 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | 159 MDOC_EXPLICIT | MDOC_JOIN }, /* So */ 160 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Sq */ 161 { in_line_argn, 0 }, /* Sm */ 162 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Sx */ 163 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Sy */ 164 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Tn */ 165 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Ux */ 166 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Xc */ 167 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Xo */ 168 { blk_full, MDOC_EXPLICIT | MDOC_CALLABLE }, /* Fo */ 169 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED | 170 MDOC_EXPLICIT | MDOC_JOIN }, /* Fc */ 171 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | 172 MDOC_EXPLICIT | MDOC_JOIN }, /* Oo */ 173 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED | 174 MDOC_EXPLICIT | MDOC_JOIN }, /* Oc */ 175 { blk_full, MDOC_EXPLICIT }, /* Bk */ 176 { blk_exp_close, MDOC_EXPLICIT | MDOC_JOIN }, /* Ek */ 177 { in_line_eoln, 0 }, /* Bt */ 178 { in_line_eoln, 0 }, /* Hf */ 179 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fr */ 180 { in_line_eoln, 0 }, /* Ud */ 181 { in_line, 0 }, /* Lb */ 182 { in_line_eoln, 0 }, /* Lp */ 183 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Lk */ 184 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Mt */ 185 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Brq */ 186 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | 187 MDOC_EXPLICIT | MDOC_JOIN }, /* Bro */ 188 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED | 189 MDOC_EXPLICIT | MDOC_JOIN }, /* Brc */ 190 { in_line_eoln, MDOC_JOIN }, /* %C */ 191 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Es */ 192 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* En */ 193 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Dx */ 194 { in_line_eoln, MDOC_JOIN }, /* %Q */ 195 { in_line_eoln, 0 }, /* br */ 196 { in_line_eoln, 0 }, /* sp */ 197 { in_line_eoln, 0 }, /* %U */ 198 { phrase_ta, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Ta */ 199 { in_line_eoln, MDOC_PROLOGUE }, /* ll */ 200 }; 201 202 const struct mdoc_macro * const mdoc_macros = __mdoc_macros; 203 204 205 /* 206 * This is called at the end of parsing. It must traverse up the tree, 207 * closing out open [implicit] scopes. Obviously, open explicit scopes 208 * are errors. 209 */ 210 void 211 mdoc_endparse(struct roff_man *mdoc) 212 { 213 struct roff_node *n; 214 215 /* Scan for open explicit scopes. */ 216 217 n = mdoc->last->flags & MDOC_VALID ? 218 mdoc->last->parent : mdoc->last; 219 220 for ( ; n; n = n->parent) 221 if (n->type == ROFFT_BLOCK && 222 mdoc_macros[n->tok].flags & MDOC_EXPLICIT) 223 mandoc_msg(MANDOCERR_BLK_NOEND, mdoc->parse, 224 n->line, n->pos, mdoc_macronames[n->tok]); 225 226 /* Rewind to the first. */ 227 228 rew_last(mdoc, mdoc->first); 229 } 230 231 /* 232 * Look up the macro at *p called by "from", 233 * or as a line macro if from == TOKEN_NONE. 234 */ 235 static int 236 lookup(struct roff_man *mdoc, int from, int line, int ppos, const char *p) 237 { 238 int res; 239 240 if (from == TOKEN_NONE || mdoc_macros[from].flags & MDOC_PARSED) { 241 res = mdoc_hash_find(p); 242 if (res != TOKEN_NONE) { 243 if (mdoc_macros[res].flags & MDOC_CALLABLE) 244 return(res); 245 if (res != MDOC_br && res != MDOC_sp && res != MDOC_ll) 246 mandoc_msg(MANDOCERR_MACRO_CALL, 247 mdoc->parse, line, ppos, p); 248 } 249 } 250 return(TOKEN_NONE); 251 } 252 253 /* 254 * Rewind up to and including a specific node. 255 */ 256 static void 257 rew_last(struct roff_man *mdoc, const struct roff_node *to) 258 { 259 struct roff_node *np; 260 261 if (to->flags & MDOC_VALID) 262 return; 263 264 while (mdoc->last != to) { 265 /* 266 * Save the parent here, because we may delete the 267 * mdoc->last node in the post-validation phase and reset 268 * it to mdoc->last->parent, causing a step in the closing 269 * out to be lost. 270 */ 271 np = mdoc->last->parent; 272 mdoc_valid_post(mdoc); 273 mdoc->last = np; 274 assert(mdoc->last); 275 } 276 mdoc->next = ROFF_NEXT_SIBLING; 277 mdoc_valid_post(mdoc); 278 } 279 280 /* 281 * Rewind up to a specific block, including all blocks that broke it. 282 */ 283 static void 284 rew_pending(struct roff_man *mdoc, const struct roff_node *n) 285 { 286 287 for (;;) { 288 rew_last(mdoc, n); 289 290 if (mdoc->last == n) { 291 switch (n->type) { 292 case ROFFT_HEAD: 293 roff_body_alloc(mdoc, n->line, n->pos, 294 n->tok); 295 return; 296 case ROFFT_BLOCK: 297 break; 298 default: 299 return; 300 } 301 if ( ! (n->flags & MDOC_BROKEN)) 302 return; 303 } else 304 n = mdoc->last; 305 306 for (;;) { 307 if ((n = n->parent) == NULL) 308 return; 309 310 if (n->type == ROFFT_BLOCK || 311 n->type == ROFFT_HEAD) { 312 if (n->flags & MDOC_ENDED) 313 break; 314 else 315 return; 316 } 317 } 318 } 319 } 320 321 /* 322 * For a block closing macro, return the corresponding opening one. 323 * Otherwise, return the macro itself. 324 */ 325 static int 326 rew_alt(int tok) 327 { 328 switch (tok) { 329 case MDOC_Ac: 330 return(MDOC_Ao); 331 case MDOC_Bc: 332 return(MDOC_Bo); 333 case MDOC_Brc: 334 return(MDOC_Bro); 335 case MDOC_Dc: 336 return(MDOC_Do); 337 case MDOC_Ec: 338 return(MDOC_Eo); 339 case MDOC_Ed: 340 return(MDOC_Bd); 341 case MDOC_Ef: 342 return(MDOC_Bf); 343 case MDOC_Ek: 344 return(MDOC_Bk); 345 case MDOC_El: 346 return(MDOC_Bl); 347 case MDOC_Fc: 348 return(MDOC_Fo); 349 case MDOC_Oc: 350 return(MDOC_Oo); 351 case MDOC_Pc: 352 return(MDOC_Po); 353 case MDOC_Qc: 354 return(MDOC_Qo); 355 case MDOC_Re: 356 return(MDOC_Rs); 357 case MDOC_Sc: 358 return(MDOC_So); 359 case MDOC_Xc: 360 return(MDOC_Xo); 361 default: 362 return(tok); 363 } 364 /* NOTREACHED */ 365 } 366 367 static void 368 rew_elem(struct roff_man *mdoc, int tok) 369 { 370 struct roff_node *n; 371 372 n = mdoc->last; 373 if (n->type != ROFFT_ELEM) 374 n = n->parent; 375 assert(n->type == ROFFT_ELEM); 376 assert(tok == n->tok); 377 rew_last(mdoc, n); 378 } 379 380 /* 381 * If there is an open sub-block of the target requiring 382 * explicit close-out, postpone closing out the target until 383 * the rew_pending() call closing out the sub-block. 384 */ 385 static int 386 find_pending(struct roff_man *mdoc, int tok, int line, int ppos, 387 struct roff_node *target) 388 { 389 struct roff_node *n; 390 int irc; 391 392 irc = 0; 393 for (n = mdoc->last; n != NULL && n != target; n = n->parent) { 394 if (n->flags & MDOC_ENDED) { 395 if ( ! (n->flags & MDOC_VALID)) 396 n->flags |= MDOC_BROKEN; 397 continue; 398 } 399 if (n->type == ROFFT_BLOCK && 400 mdoc_macros[n->tok].flags & MDOC_EXPLICIT) { 401 irc = 1; 402 n->flags = MDOC_BROKEN; 403 if (target->type == ROFFT_HEAD) 404 target->flags = MDOC_ENDED; 405 else if ( ! (target->flags & MDOC_ENDED)) { 406 mandoc_vmsg(MANDOCERR_BLK_NEST, 407 mdoc->parse, line, ppos, 408 "%s breaks %s", mdoc_macronames[tok], 409 mdoc_macronames[n->tok]); 410 mdoc_endbody_alloc(mdoc, line, ppos, 411 tok, target, ENDBODY_NOSPACE); 412 } 413 } 414 } 415 return(irc); 416 } 417 418 /* 419 * Allocate a word and check whether it's punctuation or not. 420 * Punctuation consists of those tokens found in mdoc_isdelim(). 421 */ 422 static void 423 dword(struct roff_man *mdoc, int line, int col, const char *p, 424 enum mdelim d, int may_append) 425 { 426 427 if (d == DELIM_MAX) 428 d = mdoc_isdelim(p); 429 430 if (may_append && 431 ! (mdoc->flags & (MDOC_SYNOPSIS | MDOC_KEEP | MDOC_SMOFF)) && 432 d == DELIM_NONE && mdoc->last->type == ROFFT_TEXT && 433 mdoc_isdelim(mdoc->last->string) == DELIM_NONE) { 434 roff_word_append(mdoc, p); 435 return; 436 } 437 438 roff_word_alloc(mdoc, line, col, p); 439 440 /* 441 * If the word consists of a bare delimiter, 442 * flag the new node accordingly, 443 * unless doing so was vetoed by the invoking macro. 444 * Always clear the veto, it is only valid for one word. 445 */ 446 447 if (d == DELIM_OPEN) 448 mdoc->last->flags |= MDOC_DELIMO; 449 else if (d == DELIM_CLOSE && 450 ! (mdoc->flags & MDOC_NODELIMC) && 451 mdoc->last->parent->tok != MDOC_Fd) 452 mdoc->last->flags |= MDOC_DELIMC; 453 mdoc->flags &= ~MDOC_NODELIMC; 454 } 455 456 static void 457 append_delims(struct roff_man *mdoc, int line, int *pos, char *buf) 458 { 459 char *p; 460 int la; 461 462 if (buf[*pos] == '\0') 463 return; 464 465 for (;;) { 466 la = *pos; 467 if (mdoc_args(mdoc, line, pos, buf, TOKEN_NONE, &p) == 468 ARGS_EOLN) 469 break; 470 dword(mdoc, line, la, p, DELIM_MAX, 1); 471 472 /* 473 * If we encounter end-of-sentence symbols, then trigger 474 * the double-space. 475 * 476 * XXX: it's easy to allow this to propagate outward to 477 * the last symbol, such that `. )' will cause the 478 * correct double-spacing. However, (1) groff isn't 479 * smart enough to do this and (2) it would require 480 * knowing which symbols break this behaviour, for 481 * example, `. ;' shouldn't propagate the double-space. 482 */ 483 484 if (mandoc_eos(p, strlen(p))) 485 mdoc->last->flags |= MDOC_EOS; 486 } 487 } 488 489 /* 490 * Parse one word. 491 * If it is a macro, call it and return 1. 492 * Otherwise, allocate it and return 0. 493 */ 494 static int 495 macro_or_word(MACRO_PROT_ARGS, int parsed) 496 { 497 char *p; 498 int ntok; 499 500 p = buf + ppos; 501 ntok = TOKEN_NONE; 502 if (*p == '"') 503 p++; 504 else if (parsed && ! (mdoc->flags & MDOC_PHRASELIT)) 505 ntok = lookup(mdoc, tok, line, ppos, p); 506 507 if (ntok == TOKEN_NONE) { 508 dword(mdoc, line, ppos, p, DELIM_MAX, tok == TOKEN_NONE || 509 mdoc_macros[tok].flags & MDOC_JOIN); 510 return(0); 511 } else { 512 if (mdoc_macros[tok].fp == in_line_eoln) 513 rew_elem(mdoc, tok); 514 mdoc_macro(mdoc, ntok, line, ppos, pos, buf); 515 if (tok == TOKEN_NONE) 516 append_delims(mdoc, line, pos, buf); 517 return(1); 518 } 519 } 520 521 /* 522 * Close out block partial/full explicit. 523 */ 524 static void 525 blk_exp_close(MACRO_PROT_ARGS) 526 { 527 struct roff_node *body; /* Our own body. */ 528 struct roff_node *endbody; /* Our own end marker. */ 529 struct roff_node *itblk; /* An It block starting later. */ 530 struct roff_node *later; /* A sub-block starting later. */ 531 struct roff_node *n; /* Search back to our block. */ 532 struct roff_node *target; /* For find_pending(). */ 533 534 int j, lastarg, maxargs, nl, pending; 535 enum margserr ac; 536 int atok, ntok; 537 char *p; 538 539 nl = MDOC_NEWLINE & mdoc->flags; 540 541 switch (tok) { 542 case MDOC_Ec: 543 maxargs = 1; 544 break; 545 case MDOC_Ek: 546 mdoc->flags &= ~MDOC_KEEP; 547 /* FALLTHROUGH */ 548 default: 549 maxargs = 0; 550 break; 551 } 552 553 /* 554 * Search backwards for beginnings of blocks, 555 * both of our own and of pending sub-blocks. 556 */ 557 558 atok = rew_alt(tok); 559 body = endbody = itblk = later = NULL; 560 for (n = mdoc->last; n; n = n->parent) { 561 if (n->flags & MDOC_ENDED) { 562 if ( ! (n->flags & MDOC_VALID)) 563 n->flags |= MDOC_BROKEN; 564 continue; 565 } 566 567 /* Remember the start of our own body. */ 568 569 if (n->type == ROFFT_BODY && atok == n->tok) { 570 if (n->end == ENDBODY_NOT) 571 body = n; 572 continue; 573 } 574 575 if (n->type != ROFFT_BLOCK || n->tok == MDOC_Nm) 576 continue; 577 578 if (n->tok == MDOC_It) { 579 itblk = n; 580 continue; 581 } 582 583 if (atok == n->tok) { 584 assert(body); 585 586 /* 587 * Found the start of our own block. 588 * When there is no pending sub block, 589 * just proceed to closing out. 590 */ 591 592 if (later == NULL || 593 (tok == MDOC_El && itblk == NULL)) 594 break; 595 596 /* 597 * When there is a pending sub block, postpone 598 * closing out the current block until the 599 * rew_pending() closing out the sub-block. 600 * Mark the place where the formatting - but not 601 * the scope - of the current block ends. 602 */ 603 604 mandoc_vmsg(MANDOCERR_BLK_NEST, mdoc->parse, 605 line, ppos, "%s breaks %s", 606 mdoc_macronames[atok], 607 mdoc_macronames[later->tok]); 608 609 endbody = mdoc_endbody_alloc(mdoc, line, ppos, 610 atok, body, ENDBODY_SPACE); 611 612 if (tok == MDOC_El) 613 itblk->flags |= MDOC_ENDED | MDOC_BROKEN; 614 615 /* 616 * If a block closing macro taking arguments 617 * breaks another block, put the arguments 618 * into the end marker. 619 */ 620 621 if (maxargs) 622 mdoc->next = ROFF_NEXT_CHILD; 623 break; 624 } 625 626 /* Explicit blocks close out description lines. */ 627 628 if (n->tok == MDOC_Nd) { 629 rew_last(mdoc, n); 630 continue; 631 } 632 633 /* Breaking an open sub block. */ 634 635 n->flags |= MDOC_BROKEN; 636 if (later == NULL) 637 later = n; 638 } 639 640 if (body == NULL) { 641 mandoc_msg(MANDOCERR_BLK_NOTOPEN, mdoc->parse, 642 line, ppos, mdoc_macronames[tok]); 643 if (later != NULL) 644 later->flags &= ~MDOC_BROKEN; 645 if (maxargs && endbody == NULL) { 646 /* 647 * Stray .Ec without previous .Eo: 648 * Break the output line, keep the arguments. 649 */ 650 roff_elem_alloc(mdoc, line, ppos, MDOC_br); 651 rew_elem(mdoc, MDOC_br); 652 } 653 } else if (endbody == NULL) { 654 rew_last(mdoc, body); 655 if (maxargs) 656 mdoc_tail_alloc(mdoc, line, ppos, atok); 657 } 658 659 if ( ! (mdoc_macros[tok].flags & MDOC_PARSED)) { 660 if (buf[*pos] != '\0') 661 mandoc_vmsg(MANDOCERR_ARG_SKIP, 662 mdoc->parse, line, ppos, 663 "%s %s", mdoc_macronames[tok], 664 buf + *pos); 665 if (endbody == NULL && n != NULL) 666 rew_pending(mdoc, n); 667 return; 668 } 669 670 if (endbody != NULL) 671 n = endbody; 672 673 ntok = TOKEN_NONE; 674 for (j = 0; ; j++) { 675 lastarg = *pos; 676 677 if (j == maxargs && n != NULL) 678 rew_last(mdoc, n); 679 680 ac = mdoc_args(mdoc, line, pos, buf, tok, &p); 681 if (ac == ARGS_PUNCT || ac == ARGS_EOLN) 682 break; 683 684 ntok = ac == ARGS_QWORD ? TOKEN_NONE : 685 lookup(mdoc, tok, line, lastarg, p); 686 687 if (ntok == TOKEN_NONE) { 688 dword(mdoc, line, lastarg, p, DELIM_MAX, 689 MDOC_JOIN & mdoc_macros[tok].flags); 690 continue; 691 } 692 693 if (n != NULL) 694 rew_last(mdoc, n); 695 mdoc->flags &= ~MDOC_NEWLINE; 696 mdoc_macro(mdoc, ntok, line, lastarg, pos, buf); 697 break; 698 } 699 700 if (n != NULL) { 701 if (ntok != TOKEN_NONE && n->flags & MDOC_BROKEN) { 702 target = n; 703 do 704 target = target->parent; 705 while ( ! (target->flags & MDOC_ENDED)); 706 pending = find_pending(mdoc, ntok, line, ppos, 707 target); 708 } else 709 pending = 0; 710 if ( ! pending) 711 rew_pending(mdoc, n); 712 } 713 if (nl) 714 append_delims(mdoc, line, pos, buf); 715 } 716 717 static void 718 in_line(MACRO_PROT_ARGS) 719 { 720 int la, scope, cnt, firstarg, mayopen, nc, nl; 721 int ntok; 722 enum margserr ac; 723 enum mdelim d; 724 struct mdoc_arg *arg; 725 char *p; 726 727 nl = MDOC_NEWLINE & mdoc->flags; 728 729 /* 730 * Whether we allow ignored elements (those without content, 731 * usually because of reserved words) to squeak by. 732 */ 733 734 switch (tok) { 735 case MDOC_An: 736 /* FALLTHROUGH */ 737 case MDOC_Ar: 738 /* FALLTHROUGH */ 739 case MDOC_Fl: 740 /* FALLTHROUGH */ 741 case MDOC_Mt: 742 /* FALLTHROUGH */ 743 case MDOC_Nm: 744 /* FALLTHROUGH */ 745 case MDOC_Pa: 746 nc = 1; 747 break; 748 default: 749 nc = 0; 750 break; 751 } 752 753 mdoc_argv(mdoc, line, tok, &arg, pos, buf); 754 755 d = DELIM_NONE; 756 firstarg = 1; 757 mayopen = 1; 758 for (cnt = scope = 0;; ) { 759 la = *pos; 760 ac = mdoc_args(mdoc, line, pos, buf, tok, &p); 761 762 /* 763 * At the end of a macro line, 764 * opening delimiters do not suppress spacing. 765 */ 766 767 if (ac == ARGS_EOLN) { 768 if (d == DELIM_OPEN) 769 mdoc->last->flags &= ~MDOC_DELIMO; 770 break; 771 } 772 773 /* 774 * The rest of the macro line is only punctuation, 775 * to be handled by append_delims(). 776 * If there were no other arguments, 777 * do not allow the first one to suppress spacing, 778 * even if it turns out to be a closing one. 779 */ 780 781 if (ac == ARGS_PUNCT) { 782 if (cnt == 0 && (nc == 0 || tok == MDOC_An)) 783 mdoc->flags |= MDOC_NODELIMC; 784 break; 785 } 786 787 ntok = (ac == ARGS_QWORD || (tok == MDOC_Fn && !cnt)) ? 788 TOKEN_NONE : lookup(mdoc, tok, line, la, p); 789 790 /* 791 * In this case, we've located a submacro and must 792 * execute it. Close out scope, if open. If no 793 * elements have been generated, either create one (nc) 794 * or raise a warning. 795 */ 796 797 if (ntok != TOKEN_NONE) { 798 if (scope) 799 rew_elem(mdoc, tok); 800 if (nc && ! cnt) { 801 mdoc_elem_alloc(mdoc, line, ppos, tok, arg); 802 rew_last(mdoc, mdoc->last); 803 } else if ( ! nc && ! cnt) { 804 mdoc_argv_free(arg); 805 mandoc_msg(MANDOCERR_MACRO_EMPTY, 806 mdoc->parse, line, ppos, 807 mdoc_macronames[tok]); 808 } 809 mdoc_macro(mdoc, ntok, line, la, pos, buf); 810 if (nl) 811 append_delims(mdoc, line, pos, buf); 812 return; 813 } 814 815 /* 816 * Non-quote-enclosed punctuation. Set up our scope, if 817 * a word; rewind the scope, if a delimiter; then append 818 * the word. 819 */ 820 821 d = ac == ARGS_QWORD ? DELIM_NONE : mdoc_isdelim(p); 822 823 if (DELIM_NONE != d) { 824 /* 825 * If we encounter closing punctuation, no word 826 * has been emitted, no scope is open, and we're 827 * allowed to have an empty element, then start 828 * a new scope. 829 */ 830 if ((d == DELIM_CLOSE || 831 (d == DELIM_MIDDLE && tok == MDOC_Fl)) && 832 !cnt && !scope && nc && mayopen) { 833 mdoc_elem_alloc(mdoc, line, ppos, tok, arg); 834 scope = 1; 835 cnt++; 836 if (tok == MDOC_Nm) 837 mayopen = 0; 838 } 839 /* 840 * Close out our scope, if one is open, before 841 * any punctuation. 842 */ 843 if (scope) 844 rew_elem(mdoc, tok); 845 scope = 0; 846 if (tok == MDOC_Fn) 847 mayopen = 0; 848 } else if (mayopen && !scope) { 849 mdoc_elem_alloc(mdoc, line, ppos, tok, arg); 850 scope = 1; 851 cnt++; 852 } 853 854 dword(mdoc, line, la, p, d, 855 MDOC_JOIN & mdoc_macros[tok].flags); 856 857 /* 858 * If the first argument is a closing delimiter, 859 * do not suppress spacing before it. 860 */ 861 862 if (firstarg && d == DELIM_CLOSE && !nc) 863 mdoc->last->flags &= ~MDOC_DELIMC; 864 firstarg = 0; 865 866 /* 867 * `Fl' macros have their scope re-opened with each new 868 * word so that the `-' can be added to each one without 869 * having to parse out spaces. 870 */ 871 if (scope && tok == MDOC_Fl) { 872 rew_elem(mdoc, tok); 873 scope = 0; 874 } 875 } 876 877 if (scope) 878 rew_elem(mdoc, tok); 879 880 /* 881 * If no elements have been collected and we're allowed to have 882 * empties (nc), open a scope and close it out. Otherwise, 883 * raise a warning. 884 */ 885 886 if ( ! cnt) { 887 if (nc) { 888 mdoc_elem_alloc(mdoc, line, ppos, tok, arg); 889 rew_last(mdoc, mdoc->last); 890 } else { 891 mdoc_argv_free(arg); 892 mandoc_msg(MANDOCERR_MACRO_EMPTY, mdoc->parse, 893 line, ppos, mdoc_macronames[tok]); 894 } 895 } 896 if (nl) 897 append_delims(mdoc, line, pos, buf); 898 } 899 900 static void 901 blk_full(MACRO_PROT_ARGS) 902 { 903 int la, nl, parsed; 904 struct mdoc_arg *arg; 905 struct roff_node *blk; /* Our own or a broken block. */ 906 struct roff_node *head; /* Our own head. */ 907 struct roff_node *body; /* Our own body. */ 908 struct roff_node *n; 909 enum margserr ac, lac; 910 char *p; 911 912 nl = MDOC_NEWLINE & mdoc->flags; 913 914 if (buf[*pos] == '\0' && (tok == MDOC_Sh || tok == MDOC_Ss)) { 915 mandoc_msg(MANDOCERR_MACRO_EMPTY, mdoc->parse, 916 line, ppos, mdoc_macronames[tok]); 917 return; 918 } 919 920 if ( ! (mdoc_macros[tok].flags & MDOC_EXPLICIT)) { 921 922 /* Here, tok is one of Sh Ss Nm Nd It. */ 923 924 blk = NULL; 925 for (n = mdoc->last; n != NULL; n = n->parent) { 926 if (n->flags & MDOC_ENDED) { 927 if ( ! (n->flags & MDOC_VALID)) 928 n->flags |= MDOC_BROKEN; 929 continue; 930 } 931 if (n->type != ROFFT_BLOCK) 932 continue; 933 934 if (tok == MDOC_It && n->tok == MDOC_Bl) { 935 if (blk != NULL) { 936 mandoc_vmsg(MANDOCERR_BLK_BROKEN, 937 mdoc->parse, line, ppos, 938 "It breaks %s", 939 mdoc_macronames[blk->tok]); 940 rew_pending(mdoc, blk); 941 } 942 break; 943 } 944 945 if (mdoc_macros[n->tok].flags & MDOC_EXPLICIT) { 946 switch (tok) { 947 case MDOC_Sh: 948 /* FALLTHROUGH */ 949 case MDOC_Ss: 950 mandoc_vmsg(MANDOCERR_BLK_BROKEN, 951 mdoc->parse, line, ppos, 952 "%s breaks %s", 953 mdoc_macronames[tok], 954 mdoc_macronames[n->tok]); 955 rew_pending(mdoc, n); 956 n = mdoc->last; 957 continue; 958 case MDOC_It: 959 /* Delay in case it's astray. */ 960 blk = n; 961 continue; 962 default: 963 break; 964 } 965 break; 966 } 967 968 /* Here, n is one of Sh Ss Nm Nd It. */ 969 970 if (tok != MDOC_Sh && (n->tok == MDOC_Sh || 971 (tok != MDOC_Ss && (n->tok == MDOC_Ss || 972 (tok != MDOC_It && n->tok == MDOC_It))))) 973 break; 974 975 /* Item breaking an explicit block. */ 976 977 if (blk != NULL) { 978 mandoc_vmsg(MANDOCERR_BLK_BROKEN, 979 mdoc->parse, line, ppos, 980 "It breaks %s", 981 mdoc_macronames[blk->tok]); 982 rew_pending(mdoc, blk); 983 blk = NULL; 984 } 985 986 /* Close out prior implicit scopes. */ 987 988 rew_last(mdoc, n); 989 } 990 991 /* Skip items outside lists. */ 992 993 if (tok == MDOC_It && (n == NULL || n->tok != MDOC_Bl)) { 994 mandoc_vmsg(MANDOCERR_IT_STRAY, mdoc->parse, 995 line, ppos, "It %s", buf + *pos); 996 roff_elem_alloc(mdoc, line, ppos, MDOC_br); 997 rew_elem(mdoc, MDOC_br); 998 return; 999 } 1000 } 1001 1002 /* 1003 * This routine accommodates implicitly- and explicitly-scoped 1004 * macro openings. Implicit ones first close out prior scope 1005 * (seen above). Delay opening the head until necessary to 1006 * allow leading punctuation to print. Special consideration 1007 * for `It -column', which has phrase-part syntax instead of 1008 * regular child nodes. 1009 */ 1010 1011 mdoc_argv(mdoc, line, tok, &arg, pos, buf); 1012 blk = mdoc_block_alloc(mdoc, line, ppos, tok, arg); 1013 head = body = NULL; 1014 1015 /* 1016 * Exception: Heads of `It' macros in `-diag' lists are not 1017 * parsed, even though `It' macros in general are parsed. 1018 */ 1019 1020 parsed = tok != MDOC_It || 1021 mdoc->last->parent->tok != MDOC_Bl || 1022 mdoc->last->parent->norm->Bl.type != LIST_diag; 1023 1024 /* 1025 * The `Nd' macro has all arguments in its body: it's a hybrid 1026 * of block partial-explicit and full-implicit. Stupid. 1027 */ 1028 1029 if (tok == MDOC_Nd) { 1030 head = roff_head_alloc(mdoc, line, ppos, tok); 1031 rew_last(mdoc, head); 1032 body = roff_body_alloc(mdoc, line, ppos, tok); 1033 } 1034 1035 if (tok == MDOC_Bk) 1036 mdoc->flags |= MDOC_KEEP; 1037 1038 ac = ARGS_PEND; 1039 for (;;) { 1040 la = *pos; 1041 lac = ac; 1042 ac = mdoc_args(mdoc, line, pos, buf, tok, &p); 1043 if (ac == ARGS_EOLN) { 1044 if (lac != ARGS_PPHRASE && lac != ARGS_PHRASE) 1045 break; 1046 /* 1047 * This is necessary: if the last token on a 1048 * line is a `Ta' or tab, then we'll get 1049 * ARGS_EOLN, so we must be smart enough to 1050 * reopen our scope if the last parse was a 1051 * phrase or partial phrase. 1052 */ 1053 if (body != NULL) 1054 rew_last(mdoc, body); 1055 body = roff_body_alloc(mdoc, line, ppos, tok); 1056 break; 1057 } 1058 if (tok == MDOC_Bd || tok == MDOC_Bk) { 1059 mandoc_vmsg(MANDOCERR_ARG_EXCESS, 1060 mdoc->parse, line, la, "%s ... %s", 1061 mdoc_macronames[tok], buf + la); 1062 break; 1063 } 1064 if (tok == MDOC_Rs) { 1065 mandoc_vmsg(MANDOCERR_ARG_SKIP, mdoc->parse, 1066 line, la, "Rs %s", buf + la); 1067 break; 1068 } 1069 if (ac == ARGS_PUNCT) 1070 break; 1071 1072 /* 1073 * Emit leading punctuation (i.e., punctuation before 1074 * the ROFFT_HEAD) for non-phrase types. 1075 */ 1076 1077 if (head == NULL && 1078 ac != ARGS_PEND && 1079 ac != ARGS_PHRASE && 1080 ac != ARGS_PPHRASE && 1081 ac != ARGS_QWORD && 1082 mdoc_isdelim(p) == DELIM_OPEN) { 1083 dword(mdoc, line, la, p, DELIM_OPEN, 0); 1084 continue; 1085 } 1086 1087 /* Open a head if one hasn't been opened. */ 1088 1089 if (head == NULL) 1090 head = roff_head_alloc(mdoc, line, ppos, tok); 1091 1092 if (ac == ARGS_PHRASE || 1093 ac == ARGS_PEND || 1094 ac == ARGS_PPHRASE) { 1095 1096 /* 1097 * If we haven't opened a body yet, rewind the 1098 * head; if we have, rewind that instead. 1099 */ 1100 1101 rew_last(mdoc, body == NULL ? head : body); 1102 body = roff_body_alloc(mdoc, line, ppos, tok); 1103 1104 /* 1105 * Process phrases: set whether we're in a 1106 * partial-phrase (this effects line handling) 1107 * then call down into the phrase parser. 1108 */ 1109 1110 if (ac == ARGS_PPHRASE) 1111 mdoc->flags |= MDOC_PPHRASE; 1112 if (ac == ARGS_PEND && lac == ARGS_PPHRASE) 1113 mdoc->flags |= MDOC_PPHRASE; 1114 parse_rest(mdoc, TOKEN_NONE, line, &la, buf); 1115 mdoc->flags &= ~MDOC_PPHRASE; 1116 continue; 1117 } 1118 1119 if (macro_or_word(mdoc, tok, line, la, pos, buf, parsed)) 1120 break; 1121 } 1122 1123 if (blk->flags & MDOC_VALID) 1124 return; 1125 if (head == NULL) 1126 head = roff_head_alloc(mdoc, line, ppos, tok); 1127 if (nl && tok != MDOC_Bd && tok != MDOC_Bl && tok != MDOC_Rs) 1128 append_delims(mdoc, line, pos, buf); 1129 if (body != NULL) 1130 goto out; 1131 if (find_pending(mdoc, tok, line, ppos, head)) 1132 return; 1133 1134 /* Close out scopes to remain in a consistent state. */ 1135 1136 rew_last(mdoc, head); 1137 body = roff_body_alloc(mdoc, line, ppos, tok); 1138 out: 1139 if (mdoc->flags & MDOC_FREECOL) { 1140 rew_last(mdoc, body); 1141 rew_last(mdoc, blk); 1142 mdoc->flags &= ~MDOC_FREECOL; 1143 } 1144 } 1145 1146 static void 1147 blk_part_imp(MACRO_PROT_ARGS) 1148 { 1149 int la, nl; 1150 enum margserr ac; 1151 char *p; 1152 struct roff_node *blk; /* saved block context */ 1153 struct roff_node *body; /* saved body context */ 1154 struct roff_node *n; 1155 1156 nl = MDOC_NEWLINE & mdoc->flags; 1157 1158 /* 1159 * A macro that spans to the end of the line. This is generally 1160 * (but not necessarily) called as the first macro. The block 1161 * has a head as the immediate child, which is always empty, 1162 * followed by zero or more opening punctuation nodes, then the 1163 * body (which may be empty, depending on the macro), then zero 1164 * or more closing punctuation nodes. 1165 */ 1166 1167 blk = mdoc_block_alloc(mdoc, line, ppos, tok, NULL); 1168 rew_last(mdoc, roff_head_alloc(mdoc, line, ppos, tok)); 1169 1170 /* 1171 * Open the body scope "on-demand", that is, after we've 1172 * processed all our the leading delimiters (open parenthesis, 1173 * etc.). 1174 */ 1175 1176 for (body = NULL; ; ) { 1177 la = *pos; 1178 ac = mdoc_args(mdoc, line, pos, buf, tok, &p); 1179 if (ac == ARGS_EOLN || ac == ARGS_PUNCT) 1180 break; 1181 1182 if (body == NULL && ac != ARGS_QWORD && 1183 mdoc_isdelim(p) == DELIM_OPEN) { 1184 dword(mdoc, line, la, p, DELIM_OPEN, 0); 1185 continue; 1186 } 1187 1188 if (body == NULL) 1189 body = roff_body_alloc(mdoc, line, ppos, tok); 1190 1191 if (macro_or_word(mdoc, tok, line, la, pos, buf, 1)) 1192 break; 1193 } 1194 if (body == NULL) 1195 body = roff_body_alloc(mdoc, line, ppos, tok); 1196 1197 if (find_pending(mdoc, tok, line, ppos, body)) 1198 return; 1199 1200 rew_last(mdoc, body); 1201 if (nl) 1202 append_delims(mdoc, line, pos, buf); 1203 rew_pending(mdoc, blk); 1204 1205 /* Move trailing .Ns out of scope. */ 1206 1207 for (n = body->child; n && n->next; n = n->next) 1208 /* Do nothing. */ ; 1209 if (n && n->tok == MDOC_Ns) 1210 mdoc_node_relink(mdoc, n); 1211 } 1212 1213 static void 1214 blk_part_exp(MACRO_PROT_ARGS) 1215 { 1216 int la, nl; 1217 enum margserr ac; 1218 struct roff_node *head; /* keep track of head */ 1219 char *p; 1220 1221 nl = MDOC_NEWLINE & mdoc->flags; 1222 1223 /* 1224 * The opening of an explicit macro having zero or more leading 1225 * punctuation nodes; a head with optional single element (the 1226 * case of `Eo'); and a body that may be empty. 1227 */ 1228 1229 roff_block_alloc(mdoc, line, ppos, tok); 1230 head = NULL; 1231 for (;;) { 1232 la = *pos; 1233 ac = mdoc_args(mdoc, line, pos, buf, tok, &p); 1234 if (ac == ARGS_PUNCT || ac == ARGS_EOLN) 1235 break; 1236 1237 /* Flush out leading punctuation. */ 1238 1239 if (head == NULL && ac != ARGS_QWORD && 1240 mdoc_isdelim(p) == DELIM_OPEN) { 1241 dword(mdoc, line, la, p, DELIM_OPEN, 0); 1242 continue; 1243 } 1244 1245 if (head == NULL) { 1246 head = roff_head_alloc(mdoc, line, ppos, tok); 1247 if (tok == MDOC_Eo) /* Not parsed. */ 1248 dword(mdoc, line, la, p, DELIM_MAX, 0); 1249 rew_last(mdoc, head); 1250 roff_body_alloc(mdoc, line, ppos, tok); 1251 if (tok == MDOC_Eo) 1252 continue; 1253 } 1254 1255 if (macro_or_word(mdoc, tok, line, la, pos, buf, 1)) 1256 break; 1257 } 1258 1259 /* Clean-up to leave in a consistent state. */ 1260 1261 if (head == NULL) { 1262 rew_last(mdoc, roff_head_alloc(mdoc, line, ppos, tok)); 1263 roff_body_alloc(mdoc, line, ppos, tok); 1264 } 1265 if (nl) 1266 append_delims(mdoc, line, pos, buf); 1267 } 1268 1269 static void 1270 in_line_argn(MACRO_PROT_ARGS) 1271 { 1272 struct mdoc_arg *arg; 1273 char *p; 1274 enum margserr ac; 1275 int ntok; 1276 int state; /* arg#; -1: not yet open; -2: closed */ 1277 int la, maxargs, nl; 1278 1279 nl = mdoc->flags & MDOC_NEWLINE; 1280 1281 /* 1282 * A line macro that has a fixed number of arguments (maxargs). 1283 * Only open the scope once the first non-leading-punctuation is 1284 * found (unless MDOC_IGNDELIM is noted, like in `Pf'), then 1285 * keep it open until the maximum number of arguments are 1286 * exhausted. 1287 */ 1288 1289 switch (tok) { 1290 case MDOC_Ap: 1291 /* FALLTHROUGH */ 1292 case MDOC_Ns: 1293 /* FALLTHROUGH */ 1294 case MDOC_Ux: 1295 maxargs = 0; 1296 break; 1297 case MDOC_Bx: 1298 /* FALLTHROUGH */ 1299 case MDOC_Es: 1300 /* FALLTHROUGH */ 1301 case MDOC_Xr: 1302 maxargs = 2; 1303 break; 1304 default: 1305 maxargs = 1; 1306 break; 1307 } 1308 1309 mdoc_argv(mdoc, line, tok, &arg, pos, buf); 1310 1311 state = -1; 1312 p = NULL; 1313 for (;;) { 1314 la = *pos; 1315 ac = mdoc_args(mdoc, line, pos, buf, tok, &p); 1316 1317 if (ac == ARGS_WORD && state == -1 && 1318 ! (mdoc_macros[tok].flags & MDOC_IGNDELIM) && 1319 mdoc_isdelim(p) == DELIM_OPEN) { 1320 dword(mdoc, line, la, p, DELIM_OPEN, 0); 1321 continue; 1322 } 1323 1324 if (state == -1 && tok != MDOC_In && 1325 tok != MDOC_St && tok != MDOC_Xr) { 1326 mdoc_elem_alloc(mdoc, line, ppos, tok, arg); 1327 state = 0; 1328 } 1329 1330 if (ac == ARGS_PUNCT || ac == ARGS_EOLN) { 1331 if (abs(state) < 2 && tok == MDOC_Pf) 1332 mandoc_vmsg(MANDOCERR_PF_SKIP, 1333 mdoc->parse, line, ppos, "Pf %s", 1334 p == NULL ? "at eol" : p); 1335 break; 1336 } 1337 1338 if (state == maxargs) { 1339 rew_elem(mdoc, tok); 1340 state = -2; 1341 } 1342 1343 ntok = (ac == ARGS_QWORD || (tok == MDOC_Pf && state == 0)) ? 1344 TOKEN_NONE : lookup(mdoc, tok, line, la, p); 1345 1346 if (ntok != TOKEN_NONE) { 1347 if (state >= 0) { 1348 rew_elem(mdoc, tok); 1349 state = -2; 1350 } 1351 mdoc_macro(mdoc, ntok, line, la, pos, buf); 1352 break; 1353 } 1354 1355 if (ac == ARGS_QWORD || 1356 mdoc_macros[tok].flags & MDOC_IGNDELIM || 1357 mdoc_isdelim(p) == DELIM_NONE) { 1358 if (state == -1) { 1359 mdoc_elem_alloc(mdoc, line, ppos, tok, arg); 1360 state = 1; 1361 } else if (state >= 0) 1362 state++; 1363 } else if (state >= 0) { 1364 rew_elem(mdoc, tok); 1365 state = -2; 1366 } 1367 1368 dword(mdoc, line, la, p, DELIM_MAX, 1369 MDOC_JOIN & mdoc_macros[tok].flags); 1370 } 1371 1372 if (state == -1) { 1373 mandoc_msg(MANDOCERR_MACRO_EMPTY, mdoc->parse, 1374 line, ppos, mdoc_macronames[tok]); 1375 return; 1376 } 1377 1378 if (state == 0 && tok == MDOC_Pf) 1379 append_delims(mdoc, line, pos, buf); 1380 if (state >= 0) 1381 rew_elem(mdoc, tok); 1382 if (nl) 1383 append_delims(mdoc, line, pos, buf); 1384 } 1385 1386 static void 1387 in_line_eoln(MACRO_PROT_ARGS) 1388 { 1389 struct roff_node *n; 1390 struct mdoc_arg *arg; 1391 1392 if ((tok == MDOC_Pp || tok == MDOC_Lp) && 1393 ! (mdoc->flags & MDOC_SYNOPSIS)) { 1394 n = mdoc->last; 1395 if (mdoc->next == ROFF_NEXT_SIBLING) 1396 n = n->parent; 1397 if (n->tok == MDOC_Nm) 1398 rew_last(mdoc, n->parent); 1399 } 1400 1401 if (buf[*pos] == '\0' && 1402 (tok == MDOC_Fd || mdoc_macronames[tok][0] == '%')) { 1403 mandoc_msg(MANDOCERR_MACRO_EMPTY, mdoc->parse, 1404 line, ppos, mdoc_macronames[tok]); 1405 return; 1406 } 1407 1408 mdoc_argv(mdoc, line, tok, &arg, pos, buf); 1409 mdoc_elem_alloc(mdoc, line, ppos, tok, arg); 1410 if (parse_rest(mdoc, tok, line, pos, buf)) 1411 return; 1412 rew_elem(mdoc, tok); 1413 } 1414 1415 /* 1416 * The simplest argument parser available: Parse the remaining 1417 * words until the end of the phrase or line and return 0 1418 * or until the next macro, call that macro, and return 1. 1419 */ 1420 static int 1421 parse_rest(struct roff_man *mdoc, int tok, int line, int *pos, char *buf) 1422 { 1423 int la; 1424 1425 for (;;) { 1426 la = *pos; 1427 if (mdoc_args(mdoc, line, pos, buf, tok, NULL) == ARGS_EOLN) 1428 return(0); 1429 if (macro_or_word(mdoc, tok, line, la, pos, buf, 1)) 1430 return(1); 1431 } 1432 } 1433 1434 static void 1435 ctx_synopsis(MACRO_PROT_ARGS) 1436 { 1437 1438 if (~mdoc->flags & (MDOC_SYNOPSIS | MDOC_NEWLINE)) 1439 in_line(mdoc, tok, line, ppos, pos, buf); 1440 else if (tok == MDOC_Nm) 1441 blk_full(mdoc, tok, line, ppos, pos, buf); 1442 else { 1443 assert(tok == MDOC_Vt); 1444 blk_part_imp(mdoc, tok, line, ppos, pos, buf); 1445 } 1446 } 1447 1448 /* 1449 * Phrases occur within `Bl -column' entries, separated by `Ta' or tabs. 1450 * They're unusual because they're basically free-form text until a 1451 * macro is encountered. 1452 */ 1453 static void 1454 phrase_ta(MACRO_PROT_ARGS) 1455 { 1456 struct roff_node *body, *n; 1457 1458 /* Make sure we are in a column list or ignore this macro. */ 1459 1460 body = NULL; 1461 for (n = mdoc->last; n != NULL; n = n->parent) { 1462 if (n->flags & MDOC_ENDED) 1463 continue; 1464 if (n->tok == MDOC_It && n->type == ROFFT_BODY) 1465 body = n; 1466 if (n->tok == MDOC_Bl) 1467 break; 1468 } 1469 1470 if (n == NULL || n->norm->Bl.type != LIST_column) { 1471 mandoc_msg(MANDOCERR_TA_STRAY, mdoc->parse, 1472 line, ppos, "Ta"); 1473 return; 1474 } 1475 1476 /* Advance to the next column. */ 1477 1478 rew_last(mdoc, body); 1479 roff_body_alloc(mdoc, line, ppos, MDOC_It); 1480 parse_rest(mdoc, TOKEN_NONE, line, pos, buf); 1481 } 1482