1 /* $Id: mdoc_macro.c,v 1.224 2017/05/30 16:22:03 schwarze Exp $ */ 2 /* 3 * Copyright (c) 2008-2012 Kristaps Dzonsons <kristaps@bsd.lv> 4 * Copyright (c) 2010, 2012-2017 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 "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 #include "mandoc.h" 30 #include "roff.h" 31 #include "mdoc.h" 32 #include "libmandoc.h" 33 #include "roff_int.h" 34 #include "libmdoc.h" 35 36 static void blk_full(MACRO_PROT_ARGS); 37 static void blk_exp_close(MACRO_PROT_ARGS); 38 static void blk_part_exp(MACRO_PROT_ARGS); 39 static void blk_part_imp(MACRO_PROT_ARGS); 40 static void ctx_synopsis(MACRO_PROT_ARGS); 41 static void in_line_eoln(MACRO_PROT_ARGS); 42 static void in_line_argn(MACRO_PROT_ARGS); 43 static void in_line(MACRO_PROT_ARGS); 44 static void phrase_ta(MACRO_PROT_ARGS); 45 46 static void append_delims(struct roff_man *, int, int *, char *); 47 static void dword(struct roff_man *, int, int, const char *, 48 enum mdelim, int); 49 static int find_pending(struct roff_man *, enum roff_tok, 50 int, int, struct roff_node *); 51 static int lookup(struct roff_man *, int, int, int, const char *); 52 static int macro_or_word(MACRO_PROT_ARGS, int); 53 static void break_intermediate(struct roff_node *, 54 struct roff_node *); 55 static int parse_rest(struct roff_man *, enum roff_tok, 56 int, int *, char *); 57 static enum roff_tok rew_alt(enum roff_tok); 58 static void rew_elem(struct roff_man *, enum roff_tok); 59 static void rew_last(struct roff_man *, const struct roff_node *); 60 static void rew_pending(struct roff_man *, 61 const struct roff_node *); 62 63 const struct mdoc_macro __mdoc_macros[MDOC_MAX - MDOC_Dd] = { 64 { in_line_eoln, MDOC_PROLOGUE }, /* Dd */ 65 { in_line_eoln, MDOC_PROLOGUE }, /* Dt */ 66 { in_line_eoln, MDOC_PROLOGUE }, /* Os */ 67 { blk_full, MDOC_PARSED | MDOC_JOIN }, /* Sh */ 68 { blk_full, MDOC_PARSED | MDOC_JOIN }, /* Ss */ 69 { in_line_eoln, 0 }, /* Pp */ 70 { blk_part_imp, MDOC_PARSED | MDOC_JOIN }, /* D1 */ 71 { blk_part_imp, MDOC_PARSED | MDOC_JOIN }, /* Dl */ 72 { blk_full, MDOC_EXPLICIT }, /* Bd */ 73 { blk_exp_close, MDOC_EXPLICIT | MDOC_JOIN }, /* Ed */ 74 { blk_full, MDOC_EXPLICIT }, /* Bl */ 75 { blk_exp_close, MDOC_EXPLICIT | MDOC_JOIN }, /* El */ 76 { blk_full, MDOC_PARSED | MDOC_JOIN }, /* It */ 77 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ad */ 78 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* An */ 79 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED | 80 MDOC_IGNDELIM | MDOC_JOIN }, /* Ap */ 81 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ar */ 82 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Cd */ 83 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Cm */ 84 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Dv */ 85 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Er */ 86 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ev */ 87 { in_line_eoln, 0 }, /* Ex */ 88 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fa */ 89 { in_line_eoln, 0 }, /* Fd */ 90 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fl */ 91 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fn */ 92 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ft */ 93 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Ic */ 94 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* In */ 95 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Li */ 96 { blk_full, MDOC_JOIN }, /* Nd */ 97 { ctx_synopsis, MDOC_CALLABLE | MDOC_PARSED }, /* Nm */ 98 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Op */ 99 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ot */ 100 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Pa */ 101 { in_line_eoln, 0 }, /* Rv */ 102 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* St */ 103 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Va */ 104 { ctx_synopsis, MDOC_CALLABLE | MDOC_PARSED }, /* Vt */ 105 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Xr */ 106 { in_line_eoln, MDOC_JOIN }, /* %A */ 107 { in_line_eoln, MDOC_JOIN }, /* %B */ 108 { in_line_eoln, MDOC_JOIN }, /* %D */ 109 { in_line_eoln, MDOC_JOIN }, /* %I */ 110 { in_line_eoln, MDOC_JOIN }, /* %J */ 111 { in_line_eoln, 0 }, /* %N */ 112 { in_line_eoln, MDOC_JOIN }, /* %O */ 113 { in_line_eoln, 0 }, /* %P */ 114 { in_line_eoln, MDOC_JOIN }, /* %R */ 115 { in_line_eoln, MDOC_JOIN }, /* %T */ 116 { in_line_eoln, 0 }, /* %V */ 117 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED | 118 MDOC_EXPLICIT | MDOC_JOIN }, /* Ac */ 119 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | 120 MDOC_EXPLICIT | MDOC_JOIN }, /* Ao */ 121 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Aq */ 122 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* At */ 123 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED | 124 MDOC_EXPLICIT | MDOC_JOIN }, /* Bc */ 125 { blk_full, MDOC_EXPLICIT }, /* Bf */ 126 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | 127 MDOC_EXPLICIT | MDOC_JOIN }, /* Bo */ 128 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Bq */ 129 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Bsx */ 130 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Bx */ 131 { in_line_eoln, 0 }, /* Db */ 132 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED | 133 MDOC_EXPLICIT | MDOC_JOIN }, /* Dc */ 134 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | 135 MDOC_EXPLICIT | MDOC_JOIN }, /* Do */ 136 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Dq */ 137 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Ec */ 138 { blk_exp_close, MDOC_EXPLICIT | MDOC_JOIN }, /* Ef */ 139 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Em */ 140 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Eo */ 141 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Fx */ 142 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ms */ 143 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* No */ 144 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED | 145 MDOC_IGNDELIM | MDOC_JOIN }, /* Ns */ 146 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Nx */ 147 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Ox */ 148 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED | 149 MDOC_EXPLICIT | MDOC_JOIN }, /* Pc */ 150 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED | MDOC_IGNDELIM }, /* Pf */ 151 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | 152 MDOC_EXPLICIT | MDOC_JOIN }, /* Po */ 153 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Pq */ 154 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED | 155 MDOC_EXPLICIT | MDOC_JOIN }, /* Qc */ 156 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Ql */ 157 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | 158 MDOC_EXPLICIT | MDOC_JOIN }, /* Qo */ 159 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Qq */ 160 { blk_exp_close, MDOC_EXPLICIT | MDOC_JOIN }, /* Re */ 161 { blk_full, MDOC_EXPLICIT }, /* Rs */ 162 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED | 163 MDOC_EXPLICIT | MDOC_JOIN }, /* Sc */ 164 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | 165 MDOC_EXPLICIT | MDOC_JOIN }, /* So */ 166 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Sq */ 167 { in_line_argn, 0 }, /* Sm */ 168 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Sx */ 169 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Sy */ 170 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Tn */ 171 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Ux */ 172 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Xc */ 173 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Xo */ 174 { blk_full, MDOC_EXPLICIT | MDOC_CALLABLE }, /* Fo */ 175 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED | 176 MDOC_EXPLICIT | MDOC_JOIN }, /* Fc */ 177 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | 178 MDOC_EXPLICIT | MDOC_JOIN }, /* Oo */ 179 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED | 180 MDOC_EXPLICIT | MDOC_JOIN }, /* Oc */ 181 { blk_full, MDOC_EXPLICIT }, /* Bk */ 182 { blk_exp_close, MDOC_EXPLICIT | MDOC_JOIN }, /* Ek */ 183 { in_line_eoln, 0 }, /* Bt */ 184 { in_line_eoln, 0 }, /* Hf */ 185 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fr */ 186 { in_line_eoln, 0 }, /* Ud */ 187 { in_line, 0 }, /* Lb */ 188 { in_line_eoln, 0 }, /* Lp */ 189 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Lk */ 190 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Mt */ 191 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Brq */ 192 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | 193 MDOC_EXPLICIT | MDOC_JOIN }, /* Bro */ 194 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED | 195 MDOC_EXPLICIT | MDOC_JOIN }, /* Brc */ 196 { in_line_eoln, MDOC_JOIN }, /* %C */ 197 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Es */ 198 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* En */ 199 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Dx */ 200 { in_line_eoln, MDOC_JOIN }, /* %Q */ 201 { in_line_eoln, 0 }, /* %U */ 202 { phrase_ta, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Ta */ 203 }; 204 const struct mdoc_macro *const mdoc_macros = __mdoc_macros - MDOC_Dd; 205 206 207 /* 208 * This is called at the end of parsing. It must traverse up the tree, 209 * closing out open [implicit] scopes. Obviously, open explicit scopes 210 * are errors. 211 */ 212 void 213 mdoc_endparse(struct roff_man *mdoc) 214 { 215 struct roff_node *n; 216 217 /* Scan for open explicit scopes. */ 218 219 n = mdoc->last->flags & NODE_VALID ? 220 mdoc->last->parent : mdoc->last; 221 222 for ( ; n; n = n->parent) 223 if (n->type == ROFFT_BLOCK && 224 mdoc_macros[n->tok].flags & MDOC_EXPLICIT) 225 mandoc_msg(MANDOCERR_BLK_NOEND, mdoc->parse, 226 n->line, n->pos, roff_name[n->tok]); 227 228 /* Rewind to the first. */ 229 230 rew_last(mdoc, mdoc->first); 231 mdoc_state_reset(mdoc); 232 } 233 234 /* 235 * Look up the macro at *p called by "from", 236 * or as a line macro if from == TOKEN_NONE. 237 */ 238 static int 239 lookup(struct roff_man *mdoc, int from, int line, int ppos, const char *p) 240 { 241 enum roff_tok res; 242 243 if (mdoc->flags & MDOC_PHRASEQF) { 244 mdoc->flags &= ~MDOC_PHRASEQF; 245 return TOKEN_NONE; 246 } 247 if (from == TOKEN_NONE || mdoc_macros[from].flags & MDOC_PARSED) { 248 res = roffhash_find(mdoc->mdocmac, p, 0); 249 if (res != TOKEN_NONE) { 250 if (mdoc_macros[res].flags & MDOC_CALLABLE) 251 return res; 252 mandoc_msg(MANDOCERR_MACRO_CALL, 253 mdoc->parse, line, ppos, p); 254 } 255 } 256 return TOKEN_NONE; 257 } 258 259 /* 260 * Rewind up to and including a specific node. 261 */ 262 static void 263 rew_last(struct roff_man *mdoc, const struct roff_node *to) 264 { 265 266 if (to->flags & NODE_VALID) 267 return; 268 269 while (mdoc->last != to) { 270 mdoc_state(mdoc, mdoc->last); 271 mdoc->last->flags |= NODE_VALID | NODE_ENDED; 272 mdoc->last = mdoc->last->parent; 273 } 274 mdoc_state(mdoc, mdoc->last); 275 mdoc->last->flags |= NODE_VALID | NODE_ENDED; 276 mdoc->next = ROFF_NEXT_SIBLING; 277 } 278 279 /* 280 * Rewind up to a specific block, including all blocks that broke it. 281 */ 282 static void 283 rew_pending(struct roff_man *mdoc, const struct roff_node *n) 284 { 285 286 for (;;) { 287 rew_last(mdoc, n); 288 289 if (mdoc->last == n) { 290 switch (n->type) { 291 case ROFFT_HEAD: 292 roff_body_alloc(mdoc, n->line, n->pos, 293 n->tok); 294 break; 295 case ROFFT_BLOCK: 296 break; 297 default: 298 return; 299 } 300 if ( ! (n->flags & NODE_BROKEN)) 301 return; 302 } else 303 n = mdoc->last; 304 305 for (;;) { 306 if ((n = n->parent) == NULL) 307 return; 308 309 if (n->type == ROFFT_BLOCK || 310 n->type == ROFFT_HEAD) { 311 if (n->flags & NODE_ENDED) 312 break; 313 else 314 return; 315 } 316 } 317 } 318 } 319 320 /* 321 * For a block closing macro, return the corresponding opening one. 322 * Otherwise, return the macro itself. 323 */ 324 static enum roff_tok 325 rew_alt(enum roff_tok tok) 326 { 327 switch (tok) { 328 case MDOC_Ac: 329 return MDOC_Ao; 330 case MDOC_Bc: 331 return MDOC_Bo; 332 case MDOC_Brc: 333 return MDOC_Bro; 334 case MDOC_Dc: 335 return MDOC_Do; 336 case MDOC_Ec: 337 return MDOC_Eo; 338 case MDOC_Ed: 339 return MDOC_Bd; 340 case MDOC_Ef: 341 return MDOC_Bf; 342 case MDOC_Ek: 343 return MDOC_Bk; 344 case MDOC_El: 345 return MDOC_Bl; 346 case MDOC_Fc: 347 return MDOC_Fo; 348 case MDOC_Oc: 349 return MDOC_Oo; 350 case MDOC_Pc: 351 return MDOC_Po; 352 case MDOC_Qc: 353 return MDOC_Qo; 354 case MDOC_Re: 355 return MDOC_Rs; 356 case MDOC_Sc: 357 return MDOC_So; 358 case MDOC_Xc: 359 return MDOC_Xo; 360 default: 361 return tok; 362 } 363 } 364 365 static void 366 rew_elem(struct roff_man *mdoc, enum roff_tok tok) 367 { 368 struct roff_node *n; 369 370 n = mdoc->last; 371 if (n->type != ROFFT_ELEM) 372 n = n->parent; 373 assert(n->type == ROFFT_ELEM); 374 assert(tok == n->tok); 375 rew_last(mdoc, n); 376 } 377 378 static void 379 break_intermediate(struct roff_node *n, struct roff_node *breaker) 380 { 381 if (n != breaker && 382 n->type != ROFFT_BLOCK && n->type != ROFFT_HEAD && 383 (n->type != ROFFT_BODY || n->end != ENDBODY_NOT)) 384 n = n->parent; 385 while (n != breaker) { 386 if ( ! (n->flags & NODE_VALID)) 387 n->flags |= NODE_BROKEN; 388 n = n->parent; 389 } 390 } 391 392 /* 393 * If there is an open sub-block of the target requiring 394 * explicit close-out, postpone closing out the target until 395 * the rew_pending() call closing out the sub-block. 396 */ 397 static int 398 find_pending(struct roff_man *mdoc, enum roff_tok tok, int line, int ppos, 399 struct roff_node *target) 400 { 401 struct roff_node *n; 402 int irc; 403 404 if (target->flags & NODE_VALID) 405 return 0; 406 407 irc = 0; 408 for (n = mdoc->last; n != NULL && n != target; n = n->parent) { 409 if (n->flags & NODE_ENDED) 410 continue; 411 if (n->type == ROFFT_BLOCK && 412 mdoc_macros[n->tok].flags & MDOC_EXPLICIT) { 413 irc = 1; 414 break_intermediate(mdoc->last, target); 415 if (target->type == ROFFT_HEAD) 416 target->flags |= NODE_ENDED; 417 else if ( ! (target->flags & NODE_ENDED)) { 418 mandoc_vmsg(MANDOCERR_BLK_NEST, 419 mdoc->parse, line, ppos, 420 "%s breaks %s", roff_name[tok], 421 roff_name[n->tok]); 422 mdoc_endbody_alloc(mdoc, line, ppos, 423 tok, target); 424 } 425 } 426 } 427 return irc; 428 } 429 430 /* 431 * Allocate a word and check whether it's punctuation or not. 432 * Punctuation consists of those tokens found in mdoc_isdelim(). 433 */ 434 static void 435 dword(struct roff_man *mdoc, int line, int col, const char *p, 436 enum mdelim d, int may_append) 437 { 438 439 if (d == DELIM_MAX) 440 d = mdoc_isdelim(p); 441 442 if (may_append && 443 ! (mdoc->flags & (MDOC_SYNOPSIS | MDOC_KEEP | MDOC_SMOFF)) && 444 d == DELIM_NONE && mdoc->last->type == ROFFT_TEXT && 445 mdoc_isdelim(mdoc->last->string) == DELIM_NONE) { 446 roff_word_append(mdoc, p); 447 return; 448 } 449 450 roff_word_alloc(mdoc, line, col, p); 451 452 /* 453 * If the word consists of a bare delimiter, 454 * flag the new node accordingly, 455 * unless doing so was vetoed by the invoking macro. 456 * Always clear the veto, it is only valid for one word. 457 */ 458 459 if (d == DELIM_OPEN) 460 mdoc->last->flags |= NODE_DELIMO; 461 else if (d == DELIM_CLOSE && 462 ! (mdoc->flags & MDOC_NODELIMC) && 463 mdoc->last->parent->tok != MDOC_Fd) 464 mdoc->last->flags |= NODE_DELIMC; 465 mdoc->flags &= ~MDOC_NODELIMC; 466 } 467 468 static void 469 append_delims(struct roff_man *mdoc, int line, int *pos, char *buf) 470 { 471 char *p; 472 int la; 473 474 if (buf[*pos] == '\0') 475 return; 476 477 for (;;) { 478 la = *pos; 479 if (mdoc_args(mdoc, line, pos, buf, TOKEN_NONE, &p) == 480 ARGS_EOLN) 481 break; 482 dword(mdoc, line, la, p, DELIM_MAX, 1); 483 484 /* 485 * If we encounter end-of-sentence symbols, then trigger 486 * the double-space. 487 * 488 * XXX: it's easy to allow this to propagate outward to 489 * the last symbol, such that `. )' will cause the 490 * correct double-spacing. However, (1) groff isn't 491 * smart enough to do this and (2) it would require 492 * knowing which symbols break this behaviour, for 493 * example, `. ;' shouldn't propagate the double-space. 494 */ 495 496 if (mandoc_eos(p, strlen(p))) 497 mdoc->last->flags |= NODE_EOS; 498 } 499 } 500 501 /* 502 * Parse one word. 503 * If it is a macro, call it and return 1. 504 * Otherwise, allocate it and return 0. 505 */ 506 static int 507 macro_or_word(MACRO_PROT_ARGS, int parsed) 508 { 509 char *p; 510 int ntok; 511 512 p = buf + ppos; 513 ntok = TOKEN_NONE; 514 if (*p == '"') 515 p++; 516 else if (parsed && ! (mdoc->flags & MDOC_PHRASELIT)) 517 ntok = lookup(mdoc, tok, line, ppos, p); 518 519 if (ntok == TOKEN_NONE) { 520 dword(mdoc, line, ppos, p, DELIM_MAX, tok == TOKEN_NONE || 521 mdoc_macros[tok].flags & MDOC_JOIN); 522 return 0; 523 } else { 524 if (tok != TOKEN_NONE && 525 mdoc_macros[tok].fp == in_line_eoln) 526 rew_elem(mdoc, tok); 527 mdoc_macro(mdoc, ntok, line, ppos, pos, buf); 528 if (tok == TOKEN_NONE) 529 append_delims(mdoc, line, pos, buf); 530 return 1; 531 } 532 } 533 534 /* 535 * Close out block partial/full explicit. 536 */ 537 static void 538 blk_exp_close(MACRO_PROT_ARGS) 539 { 540 struct roff_node *body; /* Our own body. */ 541 struct roff_node *endbody; /* Our own end marker. */ 542 struct roff_node *itblk; /* An It block starting later. */ 543 struct roff_node *later; /* A sub-block starting later. */ 544 struct roff_node *n; /* Search back to our block. */ 545 struct roff_node *target; /* For find_pending(). */ 546 547 int j, lastarg, maxargs, nl, pending; 548 enum margserr ac; 549 enum roff_tok atok, ntok; 550 char *p; 551 552 nl = MDOC_NEWLINE & mdoc->flags; 553 554 switch (tok) { 555 case MDOC_Ec: 556 maxargs = 1; 557 break; 558 case MDOC_Ek: 559 mdoc->flags &= ~MDOC_KEEP; 560 /* FALLTHROUGH */ 561 default: 562 maxargs = 0; 563 break; 564 } 565 566 /* Search backwards for the beginning of our own body. */ 567 568 atok = rew_alt(tok); 569 body = NULL; 570 for (n = mdoc->last; n; n = n->parent) { 571 if (n->flags & NODE_ENDED || n->tok != atok || 572 n->type != ROFFT_BODY || n->end != ENDBODY_NOT) 573 continue; 574 body = n; 575 break; 576 } 577 578 /* 579 * Search backwards for beginnings of blocks, 580 * both of our own and of pending sub-blocks. 581 */ 582 583 endbody = itblk = later = NULL; 584 for (n = mdoc->last; n; n = n->parent) { 585 if (n->flags & NODE_ENDED) 586 continue; 587 588 /* 589 * Mismatching end macros can never break anything 590 * and we only care about the breaking of BLOCKs. 591 */ 592 593 if (body == NULL || n->type != ROFFT_BLOCK) 594 continue; 595 596 /* 597 * SYNOPSIS name blocks can not be broken themselves, 598 * but they do get broken together with a broken child. 599 */ 600 601 if (n->tok == MDOC_Nm) { 602 if (later != NULL) 603 n->flags |= NODE_BROKEN | NODE_ENDED; 604 continue; 605 } 606 607 if (n->tok == MDOC_It) { 608 itblk = n; 609 continue; 610 } 611 612 if (atok == n->tok) { 613 614 /* 615 * Found the start of our own block. 616 * When there is no pending sub block, 617 * just proceed to closing out. 618 */ 619 620 if (later == NULL || 621 (tok == MDOC_El && itblk == NULL)) 622 break; 623 624 /* 625 * When there is a pending sub block, postpone 626 * closing out the current block until the 627 * rew_pending() closing out the sub-block. 628 * Mark the place where the formatting - but not 629 * the scope - of the current block ends. 630 */ 631 632 mandoc_vmsg(MANDOCERR_BLK_NEST, mdoc->parse, 633 line, ppos, "%s breaks %s", 634 roff_name[atok], roff_name[later->tok]); 635 636 endbody = mdoc_endbody_alloc(mdoc, line, ppos, 637 atok, body); 638 639 if (tok == MDOC_El) 640 itblk->flags |= NODE_ENDED | NODE_BROKEN; 641 642 /* 643 * If a block closing macro taking arguments 644 * breaks another block, put the arguments 645 * into the end marker. 646 */ 647 648 if (maxargs) 649 mdoc->next = ROFF_NEXT_CHILD; 650 break; 651 } 652 653 /* 654 * Explicit blocks close out description lines, but 655 * even those can get broken together with a child. 656 */ 657 658 if (n->tok == MDOC_Nd) { 659 if (later != NULL) 660 n->flags |= NODE_BROKEN | NODE_ENDED; 661 else 662 rew_last(mdoc, n); 663 continue; 664 } 665 666 /* Breaking an open sub block. */ 667 668 break_intermediate(mdoc->last, body); 669 n->flags |= NODE_BROKEN; 670 if (later == NULL) 671 later = n; 672 } 673 674 if (body == NULL) { 675 mandoc_msg(MANDOCERR_BLK_NOTOPEN, mdoc->parse, 676 line, ppos, roff_name[tok]); 677 if (maxargs && endbody == NULL) { 678 /* 679 * Stray .Ec without previous .Eo: 680 * Break the output line, keep the arguments. 681 */ 682 roff_elem_alloc(mdoc, line, ppos, ROFF_br); 683 rew_elem(mdoc, ROFF_br); 684 } 685 } else if (endbody == NULL) { 686 rew_last(mdoc, body); 687 if (maxargs) 688 mdoc_tail_alloc(mdoc, line, ppos, atok); 689 } 690 691 if ( ! (mdoc_macros[tok].flags & MDOC_PARSED)) { 692 if (buf[*pos] != '\0') 693 mandoc_vmsg(MANDOCERR_ARG_SKIP, 694 mdoc->parse, line, ppos, 695 "%s %s", roff_name[tok], 696 buf + *pos); 697 if (endbody == NULL && n != NULL) 698 rew_pending(mdoc, n); 699 return; 700 } 701 702 if (endbody != NULL) 703 n = endbody; 704 705 ntok = TOKEN_NONE; 706 for (j = 0; ; j++) { 707 lastarg = *pos; 708 709 if (j == maxargs && n != NULL) 710 rew_last(mdoc, n); 711 712 ac = mdoc_args(mdoc, line, pos, buf, tok, &p); 713 if (ac == ARGS_PUNCT || ac == ARGS_EOLN) 714 break; 715 716 ntok = lookup(mdoc, tok, line, lastarg, p); 717 718 if (ntok == TOKEN_NONE) { 719 dword(mdoc, line, lastarg, p, DELIM_MAX, 720 MDOC_JOIN & mdoc_macros[tok].flags); 721 continue; 722 } 723 724 if (n != NULL) 725 rew_last(mdoc, n); 726 mdoc->flags &= ~MDOC_NEWLINE; 727 mdoc_macro(mdoc, ntok, line, lastarg, pos, buf); 728 break; 729 } 730 731 if (n != NULL) { 732 pending = 0; 733 if (ntok != TOKEN_NONE && n->flags & NODE_BROKEN) { 734 target = n; 735 do 736 target = target->parent; 737 while ( ! (target->flags & NODE_ENDED)); 738 pending = find_pending(mdoc, ntok, line, ppos, target); 739 } 740 if ( ! pending) 741 rew_pending(mdoc, n); 742 } 743 if (nl) 744 append_delims(mdoc, line, pos, buf); 745 } 746 747 static void 748 in_line(MACRO_PROT_ARGS) 749 { 750 int la, scope, cnt, firstarg, mayopen, nc, nl; 751 enum roff_tok ntok; 752 enum margserr ac; 753 enum mdelim d; 754 struct mdoc_arg *arg; 755 char *p; 756 757 nl = MDOC_NEWLINE & mdoc->flags; 758 759 /* 760 * Whether we allow ignored elements (those without content, 761 * usually because of reserved words) to squeak by. 762 */ 763 764 switch (tok) { 765 case MDOC_An: 766 case MDOC_Ar: 767 case MDOC_Fl: 768 case MDOC_Mt: 769 case MDOC_Nm: 770 case MDOC_Pa: 771 nc = 1; 772 break; 773 default: 774 nc = 0; 775 break; 776 } 777 778 mdoc_argv(mdoc, line, tok, &arg, pos, buf); 779 780 d = DELIM_NONE; 781 firstarg = 1; 782 mayopen = 1; 783 for (cnt = scope = 0;; ) { 784 la = *pos; 785 ac = mdoc_args(mdoc, line, pos, buf, tok, &p); 786 787 /* 788 * At the end of a macro line, 789 * opening delimiters do not suppress spacing. 790 */ 791 792 if (ac == ARGS_EOLN) { 793 if (d == DELIM_OPEN) 794 mdoc->last->flags &= ~NODE_DELIMO; 795 break; 796 } 797 798 /* 799 * The rest of the macro line is only punctuation, 800 * to be handled by append_delims(). 801 * If there were no other arguments, 802 * do not allow the first one to suppress spacing, 803 * even if it turns out to be a closing one. 804 */ 805 806 if (ac == ARGS_PUNCT) { 807 if (cnt == 0 && (nc == 0 || tok == MDOC_An)) 808 mdoc->flags |= MDOC_NODELIMC; 809 break; 810 } 811 812 ntok = (tok == MDOC_Fn && !cnt) ? 813 TOKEN_NONE : lookup(mdoc, tok, line, la, p); 814 815 /* 816 * In this case, we've located a submacro and must 817 * execute it. Close out scope, if open. If no 818 * elements have been generated, either create one (nc) 819 * or raise a warning. 820 */ 821 822 if (ntok != TOKEN_NONE) { 823 if (scope) 824 rew_elem(mdoc, tok); 825 if (nc && ! cnt) { 826 mdoc_elem_alloc(mdoc, line, ppos, tok, arg); 827 rew_last(mdoc, mdoc->last); 828 } else if ( ! nc && ! cnt) { 829 mdoc_argv_free(arg); 830 mandoc_msg(MANDOCERR_MACRO_EMPTY, 831 mdoc->parse, line, ppos, 832 roff_name[tok]); 833 } 834 mdoc_macro(mdoc, ntok, line, la, pos, buf); 835 if (nl) 836 append_delims(mdoc, line, pos, buf); 837 return; 838 } 839 840 /* 841 * Handle punctuation. Set up our scope, if a word; 842 * rewind the scope, if a delimiter; then append the word. 843 */ 844 845 if ((d = mdoc_isdelim(p)) != DELIM_NONE) { 846 /* 847 * If we encounter closing punctuation, no word 848 * has been emitted, no scope is open, and we're 849 * allowed to have an empty element, then start 850 * a new scope. 851 */ 852 if ((d == DELIM_CLOSE || 853 (d == DELIM_MIDDLE && tok == MDOC_Fl)) && 854 !cnt && !scope && nc && mayopen) { 855 mdoc_elem_alloc(mdoc, line, ppos, tok, arg); 856 scope = 1; 857 cnt++; 858 if (tok == MDOC_Nm) 859 mayopen = 0; 860 } 861 /* 862 * Close out our scope, if one is open, before 863 * any punctuation. 864 */ 865 if (scope && tok != MDOC_Lk) { 866 rew_elem(mdoc, tok); 867 scope = 0; 868 if (tok == MDOC_Fn) 869 mayopen = 0; 870 } 871 } else if (mayopen && !scope) { 872 mdoc_elem_alloc(mdoc, line, ppos, tok, arg); 873 scope = 1; 874 cnt++; 875 } 876 877 dword(mdoc, line, la, p, d, 878 mdoc_macros[tok].flags & MDOC_JOIN); 879 880 /* 881 * If the first argument is a closing delimiter, 882 * do not suppress spacing before it. 883 */ 884 885 if (firstarg && d == DELIM_CLOSE && !nc) 886 mdoc->last->flags &= ~NODE_DELIMC; 887 firstarg = 0; 888 889 /* 890 * `Fl' macros have their scope re-opened with each new 891 * word so that the `-' can be added to each one without 892 * having to parse out spaces. 893 */ 894 if (scope && tok == MDOC_Fl) { 895 rew_elem(mdoc, tok); 896 scope = 0; 897 } 898 } 899 900 if (scope && tok != MDOC_Lk) { 901 rew_elem(mdoc, tok); 902 scope = 0; 903 } 904 905 /* 906 * If no elements have been collected and we're allowed to have 907 * empties (nc), open a scope and close it out. Otherwise, 908 * raise a warning. 909 */ 910 911 if ( ! cnt) { 912 if (nc) { 913 mdoc_elem_alloc(mdoc, line, ppos, tok, arg); 914 rew_last(mdoc, mdoc->last); 915 } else { 916 mdoc_argv_free(arg); 917 mandoc_msg(MANDOCERR_MACRO_EMPTY, mdoc->parse, 918 line, ppos, roff_name[tok]); 919 } 920 } 921 if (nl) 922 append_delims(mdoc, line, pos, buf); 923 if (scope) 924 rew_elem(mdoc, tok); 925 } 926 927 static void 928 blk_full(MACRO_PROT_ARGS) 929 { 930 int la, nl, parsed; 931 struct mdoc_arg *arg; 932 struct roff_node *blk; /* Our own or a broken block. */ 933 struct roff_node *head; /* Our own head. */ 934 struct roff_node *body; /* Our own body. */ 935 struct roff_node *n; 936 enum margserr ac, lac; 937 char *p; 938 939 nl = MDOC_NEWLINE & mdoc->flags; 940 941 if (buf[*pos] == '\0' && (tok == MDOC_Sh || tok == MDOC_Ss)) { 942 mandoc_msg(MANDOCERR_MACRO_EMPTY, mdoc->parse, 943 line, ppos, roff_name[tok]); 944 return; 945 } 946 947 if ( ! (mdoc_macros[tok].flags & MDOC_EXPLICIT)) { 948 949 /* Here, tok is one of Sh Ss Nm Nd It. */ 950 951 blk = NULL; 952 for (n = mdoc->last; n != NULL; n = n->parent) { 953 if (n->flags & NODE_ENDED) { 954 if ( ! (n->flags & NODE_VALID)) 955 n->flags |= NODE_BROKEN; 956 continue; 957 } 958 if (n->type != ROFFT_BLOCK) 959 continue; 960 961 if (tok == MDOC_It && n->tok == MDOC_Bl) { 962 if (blk != NULL) { 963 mandoc_vmsg(MANDOCERR_BLK_BROKEN, 964 mdoc->parse, line, ppos, 965 "It breaks %s", 966 roff_name[blk->tok]); 967 rew_pending(mdoc, blk); 968 } 969 break; 970 } 971 972 if (mdoc_macros[n->tok].flags & MDOC_EXPLICIT) { 973 switch (tok) { 974 case MDOC_Sh: 975 case MDOC_Ss: 976 mandoc_vmsg(MANDOCERR_BLK_BROKEN, 977 mdoc->parse, line, ppos, 978 "%s breaks %s", roff_name[tok], 979 roff_name[n->tok]); 980 rew_pending(mdoc, n); 981 n = mdoc->last; 982 continue; 983 case MDOC_It: 984 /* Delay in case it's astray. */ 985 blk = n; 986 continue; 987 default: 988 break; 989 } 990 break; 991 } 992 993 /* Here, n is one of Sh Ss Nm Nd It. */ 994 995 if (tok != MDOC_Sh && (n->tok == MDOC_Sh || 996 (tok != MDOC_Ss && (n->tok == MDOC_Ss || 997 (tok != MDOC_It && n->tok == MDOC_It))))) 998 break; 999 1000 /* Item breaking an explicit block. */ 1001 1002 if (blk != NULL) { 1003 mandoc_vmsg(MANDOCERR_BLK_BROKEN, 1004 mdoc->parse, line, ppos, 1005 "It breaks %s", roff_name[blk->tok]); 1006 rew_pending(mdoc, blk); 1007 blk = NULL; 1008 } 1009 1010 /* Close out prior implicit scopes. */ 1011 1012 rew_pending(mdoc, n); 1013 } 1014 1015 /* Skip items outside lists. */ 1016 1017 if (tok == MDOC_It && (n == NULL || n->tok != MDOC_Bl)) { 1018 mandoc_vmsg(MANDOCERR_IT_STRAY, mdoc->parse, 1019 line, ppos, "It %s", buf + *pos); 1020 roff_elem_alloc(mdoc, line, ppos, ROFF_br); 1021 rew_elem(mdoc, ROFF_br); 1022 return; 1023 } 1024 } 1025 1026 /* 1027 * This routine accommodates implicitly- and explicitly-scoped 1028 * macro openings. Implicit ones first close out prior scope 1029 * (seen above). Delay opening the head until necessary to 1030 * allow leading punctuation to print. Special consideration 1031 * for `It -column', which has phrase-part syntax instead of 1032 * regular child nodes. 1033 */ 1034 1035 mdoc_argv(mdoc, line, tok, &arg, pos, buf); 1036 blk = mdoc_block_alloc(mdoc, line, ppos, tok, arg); 1037 head = body = NULL; 1038 1039 /* 1040 * Exception: Heads of `It' macros in `-diag' lists are not 1041 * parsed, even though `It' macros in general are parsed. 1042 */ 1043 1044 parsed = tok != MDOC_It || 1045 mdoc->last->parent->tok != MDOC_Bl || 1046 mdoc->last->parent->norm->Bl.type != LIST_diag; 1047 1048 /* 1049 * The `Nd' macro has all arguments in its body: it's a hybrid 1050 * of block partial-explicit and full-implicit. Stupid. 1051 */ 1052 1053 if (tok == MDOC_Nd) { 1054 head = roff_head_alloc(mdoc, line, ppos, tok); 1055 rew_last(mdoc, head); 1056 body = roff_body_alloc(mdoc, line, ppos, tok); 1057 } 1058 1059 if (tok == MDOC_Bk) 1060 mdoc->flags |= MDOC_KEEP; 1061 1062 ac = ARGS_EOLN; 1063 for (;;) { 1064 1065 /* 1066 * If we are right after a tab character, 1067 * do not parse the first word for macros. 1068 */ 1069 1070 if (mdoc->flags & MDOC_PHRASEQN) { 1071 mdoc->flags &= ~MDOC_PHRASEQN; 1072 mdoc->flags |= MDOC_PHRASEQF; 1073 } 1074 1075 la = *pos; 1076 lac = ac; 1077 ac = mdoc_args(mdoc, line, pos, buf, tok, &p); 1078 if (ac == ARGS_EOLN) { 1079 if (lac != ARGS_PHRASE || 1080 ! (mdoc->flags & MDOC_PHRASEQF)) 1081 break; 1082 1083 /* 1084 * This line ends in a tab; start the next 1085 * column now, with a leading blank. 1086 */ 1087 1088 if (body != NULL) 1089 rew_last(mdoc, body); 1090 body = roff_body_alloc(mdoc, line, ppos, tok); 1091 roff_word_alloc(mdoc, line, ppos, "\\&"); 1092 break; 1093 } 1094 1095 if (tok == MDOC_Bd || tok == MDOC_Bk) { 1096 mandoc_vmsg(MANDOCERR_ARG_EXCESS, 1097 mdoc->parse, line, la, "%s ... %s", 1098 roff_name[tok], buf + la); 1099 break; 1100 } 1101 if (tok == MDOC_Rs) { 1102 mandoc_vmsg(MANDOCERR_ARG_SKIP, mdoc->parse, 1103 line, la, "Rs %s", buf + la); 1104 break; 1105 } 1106 if (ac == ARGS_PUNCT) 1107 break; 1108 1109 /* 1110 * Emit leading punctuation (i.e., punctuation before 1111 * the ROFFT_HEAD) for non-phrase types. 1112 */ 1113 1114 if (head == NULL && 1115 ac != ARGS_PHRASE && 1116 mdoc_isdelim(p) == DELIM_OPEN) { 1117 dword(mdoc, line, la, p, DELIM_OPEN, 0); 1118 continue; 1119 } 1120 1121 /* Open a head if one hasn't been opened. */ 1122 1123 if (head == NULL) 1124 head = roff_head_alloc(mdoc, line, ppos, tok); 1125 1126 if (ac == ARGS_PHRASE) { 1127 1128 /* 1129 * If we haven't opened a body yet, rewind the 1130 * head; if we have, rewind that instead. 1131 */ 1132 1133 rew_last(mdoc, body == NULL ? head : body); 1134 body = roff_body_alloc(mdoc, line, ppos, tok); 1135 1136 /* Process to the tab or to the end of the line. */ 1137 1138 mdoc->flags |= MDOC_PHRASE; 1139 parse_rest(mdoc, TOKEN_NONE, line, &la, buf); 1140 mdoc->flags &= ~MDOC_PHRASE; 1141 1142 /* There may have been `Ta' macros. */ 1143 1144 while (body->next != NULL) 1145 body = body->next; 1146 continue; 1147 } 1148 1149 if (macro_or_word(mdoc, tok, line, la, pos, buf, parsed)) 1150 break; 1151 } 1152 1153 if (blk->flags & NODE_VALID) 1154 return; 1155 if (head == NULL) 1156 head = roff_head_alloc(mdoc, line, ppos, tok); 1157 if (nl && tok != MDOC_Bd && tok != MDOC_Bl && tok != MDOC_Rs) 1158 append_delims(mdoc, line, pos, buf); 1159 if (body != NULL) 1160 goto out; 1161 if (find_pending(mdoc, tok, line, ppos, head)) 1162 return; 1163 1164 /* Close out scopes to remain in a consistent state. */ 1165 1166 rew_last(mdoc, head); 1167 body = roff_body_alloc(mdoc, line, ppos, tok); 1168 out: 1169 if (mdoc->flags & MDOC_FREECOL) { 1170 rew_last(mdoc, body); 1171 rew_last(mdoc, blk); 1172 mdoc->flags &= ~MDOC_FREECOL; 1173 } 1174 } 1175 1176 static void 1177 blk_part_imp(MACRO_PROT_ARGS) 1178 { 1179 int la, nl; 1180 enum margserr ac; 1181 char *p; 1182 struct roff_node *blk; /* saved block context */ 1183 struct roff_node *body; /* saved body context */ 1184 struct roff_node *n; 1185 1186 nl = MDOC_NEWLINE & mdoc->flags; 1187 1188 /* 1189 * A macro that spans to the end of the line. This is generally 1190 * (but not necessarily) called as the first macro. The block 1191 * has a head as the immediate child, which is always empty, 1192 * followed by zero or more opening punctuation nodes, then the 1193 * body (which may be empty, depending on the macro), then zero 1194 * or more closing punctuation nodes. 1195 */ 1196 1197 blk = mdoc_block_alloc(mdoc, line, ppos, tok, NULL); 1198 rew_last(mdoc, roff_head_alloc(mdoc, line, ppos, tok)); 1199 1200 /* 1201 * Open the body scope "on-demand", that is, after we've 1202 * processed all our the leading delimiters (open parenthesis, 1203 * etc.). 1204 */ 1205 1206 for (body = NULL; ; ) { 1207 la = *pos; 1208 ac = mdoc_args(mdoc, line, pos, buf, tok, &p); 1209 if (ac == ARGS_EOLN || ac == ARGS_PUNCT) 1210 break; 1211 1212 if (body == NULL && mdoc_isdelim(p) == DELIM_OPEN) { 1213 dword(mdoc, line, la, p, DELIM_OPEN, 0); 1214 continue; 1215 } 1216 1217 if (body == NULL) 1218 body = roff_body_alloc(mdoc, line, ppos, tok); 1219 1220 if (macro_or_word(mdoc, tok, line, la, pos, buf, 1)) 1221 break; 1222 } 1223 if (body == NULL) 1224 body = roff_body_alloc(mdoc, line, ppos, tok); 1225 1226 if (find_pending(mdoc, tok, line, ppos, body)) 1227 return; 1228 1229 rew_last(mdoc, body); 1230 if (nl) 1231 append_delims(mdoc, line, pos, buf); 1232 rew_pending(mdoc, blk); 1233 1234 /* Move trailing .Ns out of scope. */ 1235 1236 for (n = body->child; n && n->next; n = n->next) 1237 /* Do nothing. */ ; 1238 if (n && n->tok == MDOC_Ns) 1239 mdoc_node_relink(mdoc, n); 1240 } 1241 1242 static void 1243 blk_part_exp(MACRO_PROT_ARGS) 1244 { 1245 int la, nl; 1246 enum margserr ac; 1247 struct roff_node *head; /* keep track of head */ 1248 char *p; 1249 1250 nl = MDOC_NEWLINE & mdoc->flags; 1251 1252 /* 1253 * The opening of an explicit macro having zero or more leading 1254 * punctuation nodes; a head with optional single element (the 1255 * case of `Eo'); and a body that may be empty. 1256 */ 1257 1258 roff_block_alloc(mdoc, line, ppos, tok); 1259 head = NULL; 1260 for (;;) { 1261 la = *pos; 1262 ac = mdoc_args(mdoc, line, pos, buf, tok, &p); 1263 if (ac == ARGS_PUNCT || ac == ARGS_EOLN) 1264 break; 1265 1266 /* Flush out leading punctuation. */ 1267 1268 if (head == NULL && mdoc_isdelim(p) == DELIM_OPEN) { 1269 dword(mdoc, line, la, p, DELIM_OPEN, 0); 1270 continue; 1271 } 1272 1273 if (head == NULL) { 1274 head = roff_head_alloc(mdoc, line, ppos, tok); 1275 if (tok == MDOC_Eo) /* Not parsed. */ 1276 dword(mdoc, line, la, p, DELIM_MAX, 0); 1277 rew_last(mdoc, head); 1278 roff_body_alloc(mdoc, line, ppos, tok); 1279 if (tok == MDOC_Eo) 1280 continue; 1281 } 1282 1283 if (macro_or_word(mdoc, tok, line, la, pos, buf, 1)) 1284 break; 1285 } 1286 1287 /* Clean-up to leave in a consistent state. */ 1288 1289 if (head == NULL) { 1290 rew_last(mdoc, roff_head_alloc(mdoc, line, ppos, tok)); 1291 roff_body_alloc(mdoc, line, ppos, tok); 1292 } 1293 if (nl) 1294 append_delims(mdoc, line, pos, buf); 1295 } 1296 1297 static void 1298 in_line_argn(MACRO_PROT_ARGS) 1299 { 1300 struct mdoc_arg *arg; 1301 char *p; 1302 enum margserr ac; 1303 enum roff_tok ntok; 1304 int state; /* arg#; -1: not yet open; -2: closed */ 1305 int la, maxargs, nl; 1306 1307 nl = mdoc->flags & MDOC_NEWLINE; 1308 1309 /* 1310 * A line macro that has a fixed number of arguments (maxargs). 1311 * Only open the scope once the first non-leading-punctuation is 1312 * found (unless MDOC_IGNDELIM is noted, like in `Pf'), then 1313 * keep it open until the maximum number of arguments are 1314 * exhausted. 1315 */ 1316 1317 switch (tok) { 1318 case MDOC_Ap: 1319 case MDOC_Ns: 1320 case MDOC_Ux: 1321 maxargs = 0; 1322 break; 1323 case MDOC_Bx: 1324 case MDOC_Es: 1325 case MDOC_Xr: 1326 maxargs = 2; 1327 break; 1328 default: 1329 maxargs = 1; 1330 break; 1331 } 1332 1333 mdoc_argv(mdoc, line, tok, &arg, pos, buf); 1334 1335 state = -1; 1336 p = NULL; 1337 for (;;) { 1338 la = *pos; 1339 ac = mdoc_args(mdoc, line, pos, buf, tok, &p); 1340 1341 if (ac == ARGS_WORD && state == -1 && 1342 ! (mdoc_macros[tok].flags & MDOC_IGNDELIM) && 1343 mdoc_isdelim(p) == DELIM_OPEN) { 1344 dword(mdoc, line, la, p, DELIM_OPEN, 0); 1345 continue; 1346 } 1347 1348 if (state == -1 && tok != MDOC_In && 1349 tok != MDOC_St && tok != MDOC_Xr) { 1350 mdoc_elem_alloc(mdoc, line, ppos, tok, arg); 1351 state = 0; 1352 } 1353 1354 if (ac == ARGS_PUNCT || ac == ARGS_EOLN) { 1355 if (abs(state) < 2 && tok == MDOC_Pf) 1356 mandoc_vmsg(MANDOCERR_PF_SKIP, 1357 mdoc->parse, line, ppos, "Pf %s", 1358 p == NULL ? "at eol" : p); 1359 break; 1360 } 1361 1362 if (state == maxargs) { 1363 rew_elem(mdoc, tok); 1364 state = -2; 1365 } 1366 1367 ntok = (tok == MDOC_Pf && state == 0) ? 1368 TOKEN_NONE : lookup(mdoc, tok, line, la, p); 1369 1370 if (ntok != TOKEN_NONE) { 1371 if (state >= 0) { 1372 rew_elem(mdoc, tok); 1373 state = -2; 1374 } 1375 mdoc_macro(mdoc, ntok, line, la, pos, buf); 1376 break; 1377 } 1378 1379 if (mdoc_macros[tok].flags & MDOC_IGNDELIM || 1380 mdoc_isdelim(p) == DELIM_NONE) { 1381 if (state == -1) { 1382 mdoc_elem_alloc(mdoc, line, ppos, tok, arg); 1383 state = 1; 1384 } else if (state >= 0) 1385 state++; 1386 } else if (state >= 0) { 1387 rew_elem(mdoc, tok); 1388 state = -2; 1389 } 1390 1391 dword(mdoc, line, la, p, DELIM_MAX, 1392 mdoc_macros[tok].flags & MDOC_JOIN); 1393 } 1394 1395 if (state == -1) { 1396 mandoc_msg(MANDOCERR_MACRO_EMPTY, mdoc->parse, 1397 line, ppos, roff_name[tok]); 1398 return; 1399 } 1400 1401 if (state == 0 && tok == MDOC_Pf) 1402 append_delims(mdoc, line, pos, buf); 1403 if (state >= 0) 1404 rew_elem(mdoc, tok); 1405 if (nl) 1406 append_delims(mdoc, line, pos, buf); 1407 } 1408 1409 static void 1410 in_line_eoln(MACRO_PROT_ARGS) 1411 { 1412 struct roff_node *n; 1413 struct mdoc_arg *arg; 1414 1415 if ((tok == MDOC_Pp || tok == MDOC_Lp) && 1416 ! (mdoc->flags & MDOC_SYNOPSIS)) { 1417 n = mdoc->last; 1418 if (mdoc->next == ROFF_NEXT_SIBLING) 1419 n = n->parent; 1420 if (n->tok == MDOC_Nm) 1421 rew_last(mdoc, n->parent); 1422 } 1423 1424 if (buf[*pos] == '\0' && 1425 (tok == MDOC_Fd || *roff_name[tok] == '%')) { 1426 mandoc_msg(MANDOCERR_MACRO_EMPTY, mdoc->parse, 1427 line, ppos, roff_name[tok]); 1428 return; 1429 } 1430 1431 mdoc_argv(mdoc, line, tok, &arg, pos, buf); 1432 mdoc_elem_alloc(mdoc, line, ppos, tok, arg); 1433 if (parse_rest(mdoc, tok, line, pos, buf)) 1434 return; 1435 rew_elem(mdoc, tok); 1436 } 1437 1438 /* 1439 * The simplest argument parser available: Parse the remaining 1440 * words until the end of the phrase or line and return 0 1441 * or until the next macro, call that macro, and return 1. 1442 */ 1443 static int 1444 parse_rest(struct roff_man *mdoc, enum roff_tok tok, 1445 int line, int *pos, char *buf) 1446 { 1447 int la; 1448 1449 for (;;) { 1450 la = *pos; 1451 if (mdoc_args(mdoc, line, pos, buf, tok, NULL) == ARGS_EOLN) 1452 return 0; 1453 if (macro_or_word(mdoc, tok, line, la, pos, buf, 1)) 1454 return 1; 1455 } 1456 } 1457 1458 static void 1459 ctx_synopsis(MACRO_PROT_ARGS) 1460 { 1461 1462 if (~mdoc->flags & (MDOC_SYNOPSIS | MDOC_NEWLINE)) 1463 in_line(mdoc, tok, line, ppos, pos, buf); 1464 else if (tok == MDOC_Nm) 1465 blk_full(mdoc, tok, line, ppos, pos, buf); 1466 else { 1467 assert(tok == MDOC_Vt); 1468 blk_part_imp(mdoc, tok, line, ppos, pos, buf); 1469 } 1470 } 1471 1472 /* 1473 * Phrases occur within `Bl -column' entries, separated by `Ta' or tabs. 1474 * They're unusual because they're basically free-form text until a 1475 * macro is encountered. 1476 */ 1477 static void 1478 phrase_ta(MACRO_PROT_ARGS) 1479 { 1480 struct roff_node *body, *n; 1481 1482 /* Make sure we are in a column list or ignore this macro. */ 1483 1484 body = NULL; 1485 for (n = mdoc->last; n != NULL; n = n->parent) { 1486 if (n->flags & NODE_ENDED) 1487 continue; 1488 if (n->tok == MDOC_It && n->type == ROFFT_BODY) 1489 body = n; 1490 if (n->tok == MDOC_Bl && n->end == ENDBODY_NOT) 1491 break; 1492 } 1493 1494 if (n == NULL || n->norm->Bl.type != LIST_column) { 1495 mandoc_msg(MANDOCERR_TA_STRAY, mdoc->parse, 1496 line, ppos, "Ta"); 1497 return; 1498 } 1499 1500 /* Advance to the next column. */ 1501 1502 rew_last(mdoc, body); 1503 roff_body_alloc(mdoc, line, ppos, MDOC_It); 1504 parse_rest(mdoc, TOKEN_NONE, line, pos, buf); 1505 } 1506