1 /* $Id: mdoc_macro.c,v 1.115 2012/01/05 00:43:51 schwarze Exp $ */ 2 /* 3 * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv> 4 * Copyright (c) 2010 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 AUTHOR DISCLAIMS ALL WARRANTIES 11 * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF 12 * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR 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 #ifdef HAVE_CONFIG_H 19 #include "config.h" 20 #endif 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 "mdoc.h" 30 #include "mandoc.h" 31 #include "libmdoc.h" 32 #include "libmandoc.h" 33 34 enum rew { /* see rew_dohalt() */ 35 REWIND_NONE, 36 REWIND_THIS, 37 REWIND_MORE, 38 REWIND_FORCE, 39 REWIND_LATER, 40 REWIND_ERROR 41 }; 42 43 static int blk_full(MACRO_PROT_ARGS); 44 static int blk_exp_close(MACRO_PROT_ARGS); 45 static int blk_part_exp(MACRO_PROT_ARGS); 46 static int blk_part_imp(MACRO_PROT_ARGS); 47 static int ctx_synopsis(MACRO_PROT_ARGS); 48 static int in_line_eoln(MACRO_PROT_ARGS); 49 static int in_line_argn(MACRO_PROT_ARGS); 50 static int in_line(MACRO_PROT_ARGS); 51 static int obsolete(MACRO_PROT_ARGS); 52 static int phrase_ta(MACRO_PROT_ARGS); 53 54 static int dword(struct mdoc *, int, int, 55 const char *, enum mdelim); 56 static int append_delims(struct mdoc *, 57 int, int *, char *); 58 static enum mdoct lookup(enum mdoct, const char *); 59 static enum mdoct lookup_raw(const char *); 60 static int make_pending(struct mdoc_node *, enum mdoct, 61 struct mdoc *, int, int); 62 static int phrase(struct mdoc *, int, int, char *); 63 static enum mdoct rew_alt(enum mdoct); 64 static enum rew rew_dohalt(enum mdoct, enum mdoc_type, 65 const struct mdoc_node *); 66 static int rew_elem(struct mdoc *, enum mdoct); 67 static int rew_last(struct mdoc *, 68 const struct mdoc_node *); 69 static int rew_sub(enum mdoc_type, struct mdoc *, 70 enum mdoct, int, int); 71 72 const struct mdoc_macro __mdoc_macros[MDOC_MAX] = { 73 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Ap */ 74 { in_line_eoln, MDOC_PROLOGUE }, /* Dd */ 75 { in_line_eoln, MDOC_PROLOGUE }, /* Dt */ 76 { in_line_eoln, MDOC_PROLOGUE }, /* Os */ 77 { blk_full, MDOC_PARSED }, /* Sh */ 78 { blk_full, MDOC_PARSED }, /* Ss */ 79 { in_line_eoln, 0 }, /* Pp */ 80 { blk_part_imp, MDOC_PARSED }, /* D1 */ 81 { blk_part_imp, MDOC_PARSED }, /* Dl */ 82 { blk_full, MDOC_EXPLICIT }, /* Bd */ 83 { blk_exp_close, MDOC_EXPLICIT }, /* Ed */ 84 { blk_full, MDOC_EXPLICIT }, /* Bl */ 85 { blk_exp_close, MDOC_EXPLICIT }, /* El */ 86 { blk_full, MDOC_PARSED }, /* It */ 87 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ad */ 88 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* An */ 89 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ar */ 90 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Cd */ 91 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Cm */ 92 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Dv */ 93 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Er */ 94 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ev */ 95 { in_line_eoln, 0 }, /* Ex */ 96 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fa */ 97 { in_line_eoln, 0 }, /* Fd */ 98 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fl */ 99 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fn */ 100 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ft */ 101 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ic */ 102 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* In */ 103 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Li */ 104 { blk_full, 0 }, /* Nd */ 105 { ctx_synopsis, MDOC_CALLABLE | MDOC_PARSED }, /* Nm */ 106 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Op */ 107 { obsolete, 0 }, /* Ot */ 108 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Pa */ 109 { in_line_eoln, 0 }, /* Rv */ 110 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* St */ 111 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Va */ 112 { ctx_synopsis, MDOC_CALLABLE | MDOC_PARSED }, /* Vt */ 113 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Xr */ 114 { in_line_eoln, 0 }, /* %A */ 115 { in_line_eoln, 0 }, /* %B */ 116 { in_line_eoln, 0 }, /* %D */ 117 { in_line_eoln, 0 }, /* %I */ 118 { in_line_eoln, 0 }, /* %J */ 119 { in_line_eoln, 0 }, /* %N */ 120 { in_line_eoln, 0 }, /* %O */ 121 { in_line_eoln, 0 }, /* %P */ 122 { in_line_eoln, 0 }, /* %R */ 123 { in_line_eoln, 0 }, /* %T */ 124 { in_line_eoln, 0 }, /* %V */ 125 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Ac */ 126 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Ao */ 127 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Aq */ 128 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* At */ 129 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Bc */ 130 { blk_full, MDOC_EXPLICIT }, /* Bf */ 131 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Bo */ 132 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Bq */ 133 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Bsx */ 134 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Bx */ 135 { in_line_eoln, 0 }, /* Db */ 136 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Dc */ 137 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Do */ 138 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Dq */ 139 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Ec */ 140 { blk_exp_close, MDOC_EXPLICIT }, /* Ef */ 141 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Em */ 142 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Eo */ 143 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Fx */ 144 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ms */ 145 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED | MDOC_IGNDELIM }, /* No */ 146 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED | MDOC_IGNDELIM }, /* Ns */ 147 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Nx */ 148 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Ox */ 149 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Pc */ 150 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED | MDOC_IGNDELIM }, /* Pf */ 151 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Po */ 152 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Pq */ 153 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Qc */ 154 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Ql */ 155 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Qo */ 156 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Qq */ 157 { blk_exp_close, MDOC_EXPLICIT }, /* Re */ 158 { blk_full, MDOC_EXPLICIT }, /* Rs */ 159 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Sc */ 160 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* So */ 161 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Sq */ 162 { in_line_eoln, 0 }, /* Sm */ 163 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Sx */ 164 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Sy */ 165 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Tn */ 166 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Ux */ 167 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Xc */ 168 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Xo */ 169 { blk_full, MDOC_EXPLICIT | MDOC_CALLABLE }, /* Fo */ 170 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Fc */ 171 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Oo */ 172 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Oc */ 173 { blk_full, MDOC_EXPLICIT }, /* Bk */ 174 { blk_exp_close, MDOC_EXPLICIT }, /* Ek */ 175 { in_line_eoln, 0 }, /* Bt */ 176 { in_line_eoln, 0 }, /* Hf */ 177 { obsolete, 0 }, /* Fr */ 178 { in_line_eoln, 0 }, /* Ud */ 179 { in_line, 0 }, /* Lb */ 180 { in_line_eoln, 0 }, /* Lp */ 181 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Lk */ 182 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Mt */ 183 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Brq */ 184 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Bro */ 185 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Brc */ 186 { in_line_eoln, 0 }, /* %C */ 187 { obsolete, 0 }, /* Es */ 188 { obsolete, 0 }, /* En */ 189 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Dx */ 190 { in_line_eoln, 0 }, /* %Q */ 191 { in_line_eoln, 0 }, /* br */ 192 { in_line_eoln, 0 }, /* sp */ 193 { in_line_eoln, 0 }, /* %U */ 194 { phrase_ta, MDOC_CALLABLE | MDOC_PARSED }, /* Ta */ 195 }; 196 197 const struct mdoc_macro * const mdoc_macros = __mdoc_macros; 198 199 200 /* 201 * This is called at the end of parsing. It must traverse up the tree, 202 * closing out open [implicit] scopes. Obviously, open explicit scopes 203 * are errors. 204 */ 205 int 206 mdoc_macroend(struct mdoc *m) 207 { 208 struct mdoc_node *n; 209 210 /* Scan for open explicit scopes. */ 211 212 n = MDOC_VALID & m->last->flags ? m->last->parent : m->last; 213 214 for ( ; n; n = n->parent) 215 if (MDOC_BLOCK == n->type && 216 MDOC_EXPLICIT & mdoc_macros[n->tok].flags) 217 mdoc_nmsg(m, n, MANDOCERR_SCOPEEXIT); 218 219 /* Rewind to the first. */ 220 221 return(rew_last(m, m->first)); 222 } 223 224 225 /* 226 * Look up a macro from within a subsequent context. 227 */ 228 static enum mdoct 229 lookup(enum mdoct from, const char *p) 230 { 231 232 if ( ! (MDOC_PARSED & mdoc_macros[from].flags)) 233 return(MDOC_MAX); 234 return(lookup_raw(p)); 235 } 236 237 238 /* 239 * Lookup a macro following the initial line macro. 240 */ 241 static enum mdoct 242 lookup_raw(const char *p) 243 { 244 enum mdoct res; 245 246 if (MDOC_MAX == (res = mdoc_hash_find(p))) 247 return(MDOC_MAX); 248 if (MDOC_CALLABLE & mdoc_macros[res].flags) 249 return(res); 250 return(MDOC_MAX); 251 } 252 253 254 static int 255 rew_last(struct mdoc *mdoc, const struct mdoc_node *to) 256 { 257 struct mdoc_node *n, *np; 258 259 assert(to); 260 mdoc->next = MDOC_NEXT_SIBLING; 261 262 /* LINTED */ 263 while (mdoc->last != to) { 264 /* 265 * Save the parent here, because we may delete the 266 * m->last node in the post-validation phase and reset 267 * it to m->last->parent, causing a step in the closing 268 * out to be lost. 269 */ 270 np = mdoc->last->parent; 271 if ( ! mdoc_valid_post(mdoc)) 272 return(0); 273 n = mdoc->last; 274 mdoc->last = np; 275 assert(mdoc->last); 276 mdoc->last->last = n; 277 } 278 279 return(mdoc_valid_post(mdoc)); 280 } 281 282 283 /* 284 * For a block closing macro, return the corresponding opening one. 285 * Otherwise, return the macro itself. 286 */ 287 static enum mdoct 288 rew_alt(enum mdoct tok) 289 { 290 switch (tok) { 291 case (MDOC_Ac): 292 return(MDOC_Ao); 293 case (MDOC_Bc): 294 return(MDOC_Bo); 295 case (MDOC_Brc): 296 return(MDOC_Bro); 297 case (MDOC_Dc): 298 return(MDOC_Do); 299 case (MDOC_Ec): 300 return(MDOC_Eo); 301 case (MDOC_Ed): 302 return(MDOC_Bd); 303 case (MDOC_Ef): 304 return(MDOC_Bf); 305 case (MDOC_Ek): 306 return(MDOC_Bk); 307 case (MDOC_El): 308 return(MDOC_Bl); 309 case (MDOC_Fc): 310 return(MDOC_Fo); 311 case (MDOC_Oc): 312 return(MDOC_Oo); 313 case (MDOC_Pc): 314 return(MDOC_Po); 315 case (MDOC_Qc): 316 return(MDOC_Qo); 317 case (MDOC_Re): 318 return(MDOC_Rs); 319 case (MDOC_Sc): 320 return(MDOC_So); 321 case (MDOC_Xc): 322 return(MDOC_Xo); 323 default: 324 return(tok); 325 } 326 /* NOTREACHED */ 327 } 328 329 330 /* 331 * Rewinding to tok, how do we have to handle *p? 332 * REWIND_NONE: *p would delimit tok, but no tok scope is open 333 * inside *p, so there is no need to rewind anything at all. 334 * REWIND_THIS: *p matches tok, so rewind *p and nothing else. 335 * REWIND_MORE: *p is implicit, rewind it and keep searching for tok. 336 * REWIND_FORCE: *p is explicit, but tok is full, force rewinding *p. 337 * REWIND_LATER: *p is explicit and still open, postpone rewinding. 338 * REWIND_ERROR: No tok block is open at all. 339 */ 340 static enum rew 341 rew_dohalt(enum mdoct tok, enum mdoc_type type, 342 const struct mdoc_node *p) 343 { 344 345 /* 346 * No matching token, no delimiting block, no broken block. 347 * This can happen when full implicit macros are called for 348 * the first time but try to rewind their previous 349 * instance anyway. 350 */ 351 if (MDOC_ROOT == p->type) 352 return(MDOC_BLOCK == type && 353 MDOC_EXPLICIT & mdoc_macros[tok].flags ? 354 REWIND_ERROR : REWIND_NONE); 355 356 /* 357 * When starting to rewind, skip plain text 358 * and nodes that have already been rewound. 359 */ 360 if (MDOC_TEXT == p->type || MDOC_VALID & p->flags) 361 return(REWIND_MORE); 362 363 /* 364 * The easiest case: Found a matching token. 365 * This applies to both blocks and elements. 366 */ 367 tok = rew_alt(tok); 368 if (tok == p->tok) 369 return(p->end ? REWIND_NONE : 370 type == p->type ? REWIND_THIS : REWIND_MORE); 371 372 /* 373 * While elements do require rewinding for themselves, 374 * they never affect rewinding of other nodes. 375 */ 376 if (MDOC_ELEM == p->type) 377 return(REWIND_MORE); 378 379 /* 380 * Blocks delimited by our target token get REWIND_MORE. 381 * Blocks delimiting our target token get REWIND_NONE. 382 */ 383 switch (tok) { 384 case (MDOC_Bl): 385 if (MDOC_It == p->tok) 386 return(REWIND_MORE); 387 break; 388 case (MDOC_It): 389 if (MDOC_BODY == p->type && MDOC_Bl == p->tok) 390 return(REWIND_NONE); 391 break; 392 /* 393 * XXX Badly nested block handling still fails badly 394 * when one block is breaking two blocks of the same type. 395 * This is an incomplete and extremely ugly workaround, 396 * required to let the OpenBSD tree build. 397 */ 398 case (MDOC_Oo): 399 if (MDOC_Op == p->tok) 400 return(REWIND_MORE); 401 break; 402 case (MDOC_Nm): 403 return(REWIND_NONE); 404 case (MDOC_Nd): 405 /* FALLTHROUGH */ 406 case (MDOC_Ss): 407 if (MDOC_BODY == p->type && MDOC_Sh == p->tok) 408 return(REWIND_NONE); 409 /* FALLTHROUGH */ 410 case (MDOC_Sh): 411 if (MDOC_Nd == p->tok || MDOC_Ss == p->tok || 412 MDOC_Sh == p->tok) 413 return(REWIND_MORE); 414 break; 415 default: 416 break; 417 } 418 419 /* 420 * Default block rewinding rules. 421 * In particular, always skip block end markers, 422 * and let all blocks rewind Nm children. 423 */ 424 if (ENDBODY_NOT != p->end || MDOC_Nm == p->tok || 425 (MDOC_BLOCK == p->type && 426 ! (MDOC_EXPLICIT & mdoc_macros[tok].flags))) 427 return(REWIND_MORE); 428 429 /* 430 * By default, closing out full blocks 431 * forces closing of broken explicit blocks, 432 * while closing out partial blocks 433 * allows delayed rewinding by default. 434 */ 435 return (&blk_full == mdoc_macros[tok].fp ? 436 REWIND_FORCE : REWIND_LATER); 437 } 438 439 440 static int 441 rew_elem(struct mdoc *mdoc, enum mdoct tok) 442 { 443 struct mdoc_node *n; 444 445 n = mdoc->last; 446 if (MDOC_ELEM != n->type) 447 n = n->parent; 448 assert(MDOC_ELEM == n->type); 449 assert(tok == n->tok); 450 451 return(rew_last(mdoc, n)); 452 } 453 454 455 /* 456 * We are trying to close a block identified by tok, 457 * but the child block *broken is still open. 458 * Thus, postpone closing the tok block 459 * until the rew_sub call closing *broken. 460 */ 461 static int 462 make_pending(struct mdoc_node *broken, enum mdoct tok, 463 struct mdoc *m, int line, int ppos) 464 { 465 struct mdoc_node *breaker; 466 467 /* 468 * Iterate backwards, searching for the block matching tok, 469 * that is, the block breaking the *broken block. 470 */ 471 for (breaker = broken->parent; breaker; breaker = breaker->parent) { 472 473 /* 474 * If the *broken block had already been broken before 475 * and we encounter its breaker, make the tok block 476 * pending on the inner breaker. 477 * Graphically, "[A breaker=[B broken=[C->B B] tok=A] C]" 478 * becomes "[A broken=[B [C->B B] tok=A] C]" 479 * and finally "[A [B->A [C->B B] A] C]". 480 */ 481 if (breaker == broken->pending) { 482 broken = breaker; 483 continue; 484 } 485 486 if (REWIND_THIS != rew_dohalt(tok, MDOC_BLOCK, breaker)) 487 continue; 488 if (MDOC_BODY == broken->type) 489 broken = broken->parent; 490 491 /* 492 * Found the breaker. 493 * If another, outer breaker is already pending on 494 * the *broken block, we must not clobber the link 495 * to the outer breaker, but make it pending on the 496 * new, now inner breaker. 497 * Graphically, "[A breaker=[B broken=[C->A A] tok=B] C]" 498 * becomes "[A breaker=[B->A broken=[C A] tok=B] C]" 499 * and finally "[A [B->A [C->B A] B] C]". 500 */ 501 if (broken->pending) { 502 struct mdoc_node *taker; 503 504 /* 505 * If the breaker had also been broken before, 506 * it cannot take on the outer breaker itself, 507 * but must hand it on to its own breakers. 508 * Graphically, this is the following situation: 509 * "[A [B breaker=[C->B B] broken=[D->A A] tok=C] D]" 510 * "[A taker=[B->A breaker=[C->B B] [D->C A] C] D]" 511 */ 512 taker = breaker; 513 while (taker->pending) 514 taker = taker->pending; 515 taker->pending = broken->pending; 516 } 517 broken->pending = breaker; 518 mandoc_vmsg(MANDOCERR_SCOPENEST, m->parse, line, ppos, 519 "%s breaks %s", mdoc_macronames[tok], 520 mdoc_macronames[broken->tok]); 521 return(1); 522 } 523 524 /* 525 * Found no matching block for tok. 526 * Are you trying to close a block that is not open? 527 */ 528 return(0); 529 } 530 531 532 static int 533 rew_sub(enum mdoc_type t, struct mdoc *m, 534 enum mdoct tok, int line, int ppos) 535 { 536 struct mdoc_node *n; 537 538 n = m->last; 539 while (n) { 540 switch (rew_dohalt(tok, t, n)) { 541 case (REWIND_NONE): 542 return(1); 543 case (REWIND_THIS): 544 break; 545 case (REWIND_FORCE): 546 mandoc_vmsg(MANDOCERR_SCOPEBROKEN, m->parse, 547 line, ppos, "%s breaks %s", 548 mdoc_macronames[tok], 549 mdoc_macronames[n->tok]); 550 /* FALLTHROUGH */ 551 case (REWIND_MORE): 552 n = n->parent; 553 continue; 554 case (REWIND_LATER): 555 if (make_pending(n, tok, m, line, ppos) || 556 MDOC_BLOCK != t) 557 return(1); 558 /* FALLTHROUGH */ 559 case (REWIND_ERROR): 560 mdoc_pmsg(m, line, ppos, MANDOCERR_NOSCOPE); 561 return(1); 562 } 563 break; 564 } 565 566 assert(n); 567 if ( ! rew_last(m, n)) 568 return(0); 569 570 /* 571 * The current block extends an enclosing block. 572 * Now that the current block ends, close the enclosing block, too. 573 */ 574 while (NULL != (n = n->pending)) { 575 if ( ! rew_last(m, n)) 576 return(0); 577 if (MDOC_HEAD == n->type && 578 ! mdoc_body_alloc(m, n->line, n->pos, n->tok)) 579 return(0); 580 } 581 582 return(1); 583 } 584 585 /* 586 * Allocate a word and check whether it's punctuation or not. 587 * Punctuation consists of those tokens found in mdoc_isdelim(). 588 */ 589 static int 590 dword(struct mdoc *m, int line, 591 int col, const char *p, enum mdelim d) 592 { 593 594 if (DELIM_MAX == d) 595 d = mdoc_isdelim(p); 596 597 if ( ! mdoc_word_alloc(m, line, col, p)) 598 return(0); 599 600 if (DELIM_OPEN == d) 601 m->last->flags |= MDOC_DELIMO; 602 603 /* 604 * Closing delimiters only suppress the preceding space 605 * when they follow something, not when they start a new 606 * block or element, and not when they follow `No'. 607 * 608 * XXX Explicitly special-casing MDOC_No here feels 609 * like a layering violation. Find a better way 610 * and solve this in the code related to `No'! 611 */ 612 613 else if (DELIM_CLOSE == d && m->last->prev && 614 m->last->prev->tok != MDOC_No) 615 m->last->flags |= MDOC_DELIMC; 616 617 return(1); 618 } 619 620 static int 621 append_delims(struct mdoc *m, int line, int *pos, char *buf) 622 { 623 int la; 624 enum margserr ac; 625 char *p; 626 627 if ('\0' == buf[*pos]) 628 return(1); 629 630 for (;;) { 631 la = *pos; 632 ac = mdoc_zargs(m, line, pos, buf, &p); 633 634 if (ARGS_ERROR == ac) 635 return(0); 636 else if (ARGS_EOLN == ac) 637 break; 638 639 dword(m, line, la, p, DELIM_MAX); 640 641 /* 642 * If we encounter end-of-sentence symbols, then trigger 643 * the double-space. 644 * 645 * XXX: it's easy to allow this to propagate outward to 646 * the last symbol, such that `. )' will cause the 647 * correct double-spacing. However, (1) groff isn't 648 * smart enough to do this and (2) it would require 649 * knowing which symbols break this behaviour, for 650 * example, `. ;' shouldn't propagate the double-space. 651 */ 652 if (mandoc_eos(p, strlen(p), 0)) 653 m->last->flags |= MDOC_EOS; 654 } 655 656 return(1); 657 } 658 659 660 /* 661 * Close out block partial/full explicit. 662 */ 663 static int 664 blk_exp_close(MACRO_PROT_ARGS) 665 { 666 struct mdoc_node *body; /* Our own body. */ 667 struct mdoc_node *later; /* A sub-block starting later. */ 668 struct mdoc_node *n; /* For searching backwards. */ 669 670 int j, lastarg, maxargs, flushed, nl; 671 enum margserr ac; 672 enum mdoct atok, ntok; 673 char *p; 674 675 nl = MDOC_NEWLINE & m->flags; 676 677 switch (tok) { 678 case (MDOC_Ec): 679 maxargs = 1; 680 break; 681 default: 682 maxargs = 0; 683 break; 684 } 685 686 /* 687 * Search backwards for beginnings of blocks, 688 * both of our own and of pending sub-blocks. 689 */ 690 atok = rew_alt(tok); 691 body = later = NULL; 692 for (n = m->last; n; n = n->parent) { 693 if (MDOC_VALID & n->flags) 694 continue; 695 696 /* Remember the start of our own body. */ 697 if (MDOC_BODY == n->type && atok == n->tok) { 698 if (ENDBODY_NOT == n->end) 699 body = n; 700 continue; 701 } 702 703 if (MDOC_BLOCK != n->type || MDOC_Nm == n->tok) 704 continue; 705 if (atok == n->tok) { 706 assert(body); 707 708 /* 709 * Found the start of our own block. 710 * When there is no pending sub block, 711 * just proceed to closing out. 712 */ 713 if (NULL == later) 714 break; 715 716 /* 717 * When there is a pending sub block, 718 * postpone closing out the current block 719 * until the rew_sub() closing out the sub-block. 720 */ 721 make_pending(later, tok, m, line, ppos); 722 723 /* 724 * Mark the place where the formatting - but not 725 * the scope - of the current block ends. 726 */ 727 if ( ! mdoc_endbody_alloc(m, line, ppos, 728 atok, body, ENDBODY_SPACE)) 729 return(0); 730 break; 731 } 732 733 /* 734 * When finding an open sub block, remember the last 735 * open explicit block, or, in case there are only 736 * implicit ones, the first open implicit block. 737 */ 738 if (later && 739 MDOC_EXPLICIT & mdoc_macros[later->tok].flags) 740 continue; 741 if (MDOC_CALLABLE & mdoc_macros[n->tok].flags) 742 later = n; 743 } 744 745 if ( ! (MDOC_CALLABLE & mdoc_macros[tok].flags)) { 746 /* FIXME: do this in validate */ 747 if (buf[*pos]) 748 mdoc_pmsg(m, line, ppos, MANDOCERR_ARGSLOST); 749 750 if ( ! rew_sub(MDOC_BODY, m, tok, line, ppos)) 751 return(0); 752 return(rew_sub(MDOC_BLOCK, m, tok, line, ppos)); 753 } 754 755 if ( ! rew_sub(MDOC_BODY, m, tok, line, ppos)) 756 return(0); 757 758 if (NULL == later && maxargs > 0) 759 if ( ! mdoc_tail_alloc(m, line, ppos, rew_alt(tok))) 760 return(0); 761 762 for (flushed = j = 0; ; j++) { 763 lastarg = *pos; 764 765 if (j == maxargs && ! flushed) { 766 if ( ! rew_sub(MDOC_BLOCK, m, tok, line, ppos)) 767 return(0); 768 flushed = 1; 769 } 770 771 ac = mdoc_args(m, line, pos, buf, tok, &p); 772 773 if (ARGS_ERROR == ac) 774 return(0); 775 if (ARGS_PUNCT == ac) 776 break; 777 if (ARGS_EOLN == ac) 778 break; 779 780 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p); 781 782 if (MDOC_MAX == ntok) { 783 if ( ! dword(m, line, lastarg, p, DELIM_MAX)) 784 return(0); 785 continue; 786 } 787 788 if ( ! flushed) { 789 if ( ! rew_sub(MDOC_BLOCK, m, tok, line, ppos)) 790 return(0); 791 flushed = 1; 792 } 793 if ( ! mdoc_macro(m, ntok, line, lastarg, pos, buf)) 794 return(0); 795 break; 796 } 797 798 if ( ! flushed && ! rew_sub(MDOC_BLOCK, m, tok, line, ppos)) 799 return(0); 800 801 if ( ! nl) 802 return(1); 803 return(append_delims(m, line, pos, buf)); 804 } 805 806 807 static int 808 in_line(MACRO_PROT_ARGS) 809 { 810 int la, scope, cnt, nc, nl; 811 enum margverr av; 812 enum mdoct ntok; 813 enum margserr ac; 814 enum mdelim d; 815 struct mdoc_arg *arg; 816 char *p; 817 818 nl = MDOC_NEWLINE & m->flags; 819 820 /* 821 * Whether we allow ignored elements (those without content, 822 * usually because of reserved words) to squeak by. 823 */ 824 825 switch (tok) { 826 case (MDOC_An): 827 /* FALLTHROUGH */ 828 case (MDOC_Ar): 829 /* FALLTHROUGH */ 830 case (MDOC_Fl): 831 /* FALLTHROUGH */ 832 case (MDOC_Mt): 833 /* FALLTHROUGH */ 834 case (MDOC_Nm): 835 /* FALLTHROUGH */ 836 case (MDOC_Pa): 837 nc = 1; 838 break; 839 default: 840 nc = 0; 841 break; 842 } 843 844 for (arg = NULL;; ) { 845 la = *pos; 846 av = mdoc_argv(m, line, tok, &arg, pos, buf); 847 848 if (ARGV_WORD == av) { 849 *pos = la; 850 break; 851 } 852 if (ARGV_EOLN == av) 853 break; 854 if (ARGV_ARG == av) 855 continue; 856 857 mdoc_argv_free(arg); 858 return(0); 859 } 860 861 for (cnt = scope = 0;; ) { 862 la = *pos; 863 ac = mdoc_args(m, line, pos, buf, tok, &p); 864 865 if (ARGS_ERROR == ac) 866 return(0); 867 if (ARGS_EOLN == ac) 868 break; 869 if (ARGS_PUNCT == ac) 870 break; 871 872 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p); 873 874 /* 875 * In this case, we've located a submacro and must 876 * execute it. Close out scope, if open. If no 877 * elements have been generated, either create one (nc) 878 * or raise a warning. 879 */ 880 881 if (MDOC_MAX != ntok) { 882 if (scope && ! rew_elem(m, tok)) 883 return(0); 884 if (nc && 0 == cnt) { 885 if ( ! mdoc_elem_alloc(m, line, ppos, tok, arg)) 886 return(0); 887 if ( ! rew_last(m, m->last)) 888 return(0); 889 } else if ( ! nc && 0 == cnt) { 890 mdoc_argv_free(arg); 891 mdoc_pmsg(m, line, ppos, MANDOCERR_MACROEMPTY); 892 } 893 894 if ( ! mdoc_macro(m, ntok, line, la, pos, buf)) 895 return(0); 896 if ( ! nl) 897 return(1); 898 return(append_delims(m, line, pos, buf)); 899 } 900 901 /* 902 * Non-quote-enclosed punctuation. Set up our scope, if 903 * a word; rewind the scope, if a delimiter; then append 904 * the word. 905 */ 906 907 d = ARGS_QWORD == ac ? DELIM_NONE : mdoc_isdelim(p); 908 909 if (DELIM_NONE != d) { 910 /* 911 * If we encounter closing punctuation, no word 912 * has been omitted, no scope is open, and we're 913 * allowed to have an empty element, then start 914 * a new scope. `Ar', `Fl', and `Li', only do 915 * this once per invocation. There may be more 916 * of these (all of them?). 917 */ 918 if (0 == cnt && (nc || MDOC_Li == tok) && 919 DELIM_CLOSE == d && ! scope) { 920 if ( ! mdoc_elem_alloc(m, line, ppos, tok, arg)) 921 return(0); 922 if (MDOC_Ar == tok || MDOC_Li == tok || 923 MDOC_Fl == tok) 924 cnt++; 925 scope = 1; 926 } 927 /* 928 * Close out our scope, if one is open, before 929 * any punctuation. 930 */ 931 if (scope && ! rew_elem(m, tok)) 932 return(0); 933 scope = 0; 934 } else if ( ! scope) { 935 if ( ! mdoc_elem_alloc(m, line, ppos, tok, arg)) 936 return(0); 937 scope = 1; 938 } 939 940 if (DELIM_NONE == d) 941 cnt++; 942 943 if ( ! dword(m, line, la, p, d)) 944 return(0); 945 946 /* 947 * `Fl' macros have their scope re-opened with each new 948 * word so that the `-' can be added to each one without 949 * having to parse out spaces. 950 */ 951 if (scope && MDOC_Fl == tok) { 952 if ( ! rew_elem(m, tok)) 953 return(0); 954 scope = 0; 955 } 956 } 957 958 if (scope && ! rew_elem(m, tok)) 959 return(0); 960 961 /* 962 * If no elements have been collected and we're allowed to have 963 * empties (nc), open a scope and close it out. Otherwise, 964 * raise a warning. 965 */ 966 967 if (nc && 0 == cnt) { 968 if ( ! mdoc_elem_alloc(m, line, ppos, tok, arg)) 969 return(0); 970 if ( ! rew_last(m, m->last)) 971 return(0); 972 } else if ( ! nc && 0 == cnt) { 973 mdoc_argv_free(arg); 974 mdoc_pmsg(m, line, ppos, MANDOCERR_MACROEMPTY); 975 } 976 977 if ( ! nl) 978 return(1); 979 return(append_delims(m, line, pos, buf)); 980 } 981 982 983 static int 984 blk_full(MACRO_PROT_ARGS) 985 { 986 int la, nl, nparsed; 987 struct mdoc_arg *arg; 988 struct mdoc_node *head; /* save of head macro */ 989 struct mdoc_node *body; /* save of body macro */ 990 struct mdoc_node *n; 991 enum mdoc_type mtt; 992 enum mdoct ntok; 993 enum margserr ac, lac; 994 enum margverr av; 995 char *p; 996 997 nl = MDOC_NEWLINE & m->flags; 998 999 /* Close out prior implicit scope. */ 1000 1001 if ( ! (MDOC_EXPLICIT & mdoc_macros[tok].flags)) { 1002 if ( ! rew_sub(MDOC_BODY, m, tok, line, ppos)) 1003 return(0); 1004 if ( ! rew_sub(MDOC_BLOCK, m, tok, line, ppos)) 1005 return(0); 1006 } 1007 1008 /* 1009 * This routine accommodates implicitly- and explicitly-scoped 1010 * macro openings. Implicit ones first close out prior scope 1011 * (seen above). Delay opening the head until necessary to 1012 * allow leading punctuation to print. Special consideration 1013 * for `It -column', which has phrase-part syntax instead of 1014 * regular child nodes. 1015 */ 1016 1017 for (arg = NULL;; ) { 1018 la = *pos; 1019 av = mdoc_argv(m, line, tok, &arg, pos, buf); 1020 1021 if (ARGV_WORD == av) { 1022 *pos = la; 1023 break; 1024 } 1025 1026 if (ARGV_EOLN == av) 1027 break; 1028 if (ARGV_ARG == av) 1029 continue; 1030 1031 mdoc_argv_free(arg); 1032 return(0); 1033 } 1034 1035 if ( ! mdoc_block_alloc(m, line, ppos, tok, arg)) 1036 return(0); 1037 1038 head = body = NULL; 1039 1040 /* 1041 * Exception: Heads of `It' macros in `-diag' lists are not 1042 * parsed, even though `It' macros in general are parsed. 1043 */ 1044 nparsed = MDOC_It == tok && 1045 MDOC_Bl == m->last->parent->tok && 1046 LIST_diag == m->last->parent->norm->Bl.type; 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 (MDOC_Nd == tok) { 1054 if ( ! mdoc_head_alloc(m, line, ppos, tok)) 1055 return(0); 1056 head = m->last; 1057 if ( ! rew_sub(MDOC_HEAD, m, tok, line, ppos)) 1058 return(0); 1059 if ( ! mdoc_body_alloc(m, line, ppos, tok)) 1060 return(0); 1061 body = m->last; 1062 } 1063 1064 ac = ARGS_ERROR; 1065 1066 for ( ; ; ) { 1067 la = *pos; 1068 /* Initialise last-phrase-type with ARGS_PEND. */ 1069 lac = ARGS_ERROR == ac ? ARGS_PEND : ac; 1070 ac = mdoc_args(m, line, pos, buf, tok, &p); 1071 1072 if (ARGS_PUNCT == ac) 1073 break; 1074 1075 if (ARGS_ERROR == ac) 1076 return(0); 1077 1078 if (ARGS_EOLN == ac) { 1079 if (ARGS_PPHRASE != lac && ARGS_PHRASE != lac) 1080 break; 1081 /* 1082 * This is necessary: if the last token on a 1083 * line is a `Ta' or tab, then we'll get 1084 * ARGS_EOLN, so we must be smart enough to 1085 * reopen our scope if the last parse was a 1086 * phrase or partial phrase. 1087 */ 1088 if ( ! rew_sub(MDOC_BODY, m, tok, line, ppos)) 1089 return(0); 1090 if ( ! mdoc_body_alloc(m, line, ppos, tok)) 1091 return(0); 1092 body = m->last; 1093 break; 1094 } 1095 1096 /* 1097 * Emit leading punctuation (i.e., punctuation before 1098 * the MDOC_HEAD) for non-phrase types. 1099 */ 1100 1101 if (NULL == head && 1102 ARGS_PEND != ac && 1103 ARGS_PHRASE != ac && 1104 ARGS_PPHRASE != ac && 1105 ARGS_QWORD != ac && 1106 DELIM_OPEN == mdoc_isdelim(p)) { 1107 if ( ! dword(m, line, la, p, DELIM_OPEN)) 1108 return(0); 1109 continue; 1110 } 1111 1112 /* Open a head if one hasn't been opened. */ 1113 1114 if (NULL == head) { 1115 if ( ! mdoc_head_alloc(m, line, ppos, tok)) 1116 return(0); 1117 head = m->last; 1118 } 1119 1120 if (ARGS_PHRASE == ac || 1121 ARGS_PEND == ac || 1122 ARGS_PPHRASE == ac) { 1123 /* 1124 * If we haven't opened a body yet, rewind the 1125 * head; if we have, rewind that instead. 1126 */ 1127 1128 mtt = body ? MDOC_BODY : MDOC_HEAD; 1129 if ( ! rew_sub(mtt, m, tok, line, ppos)) 1130 return(0); 1131 1132 /* Then allocate our body context. */ 1133 1134 if ( ! mdoc_body_alloc(m, line, ppos, tok)) 1135 return(0); 1136 body = m->last; 1137 1138 /* 1139 * Process phrases: set whether we're in a 1140 * partial-phrase (this effects line handling) 1141 * then call down into the phrase parser. 1142 */ 1143 1144 if (ARGS_PPHRASE == ac) 1145 m->flags |= MDOC_PPHRASE; 1146 if (ARGS_PEND == ac && ARGS_PPHRASE == lac) 1147 m->flags |= MDOC_PPHRASE; 1148 1149 if ( ! phrase(m, line, la, buf)) 1150 return(0); 1151 1152 m->flags &= ~MDOC_PPHRASE; 1153 continue; 1154 } 1155 1156 ntok = nparsed || ARGS_QWORD == ac ? 1157 MDOC_MAX : lookup(tok, p); 1158 1159 if (MDOC_MAX == ntok) { 1160 if ( ! dword(m, line, la, p, DELIM_MAX)) 1161 return(0); 1162 continue; 1163 } 1164 1165 if ( ! mdoc_macro(m, ntok, line, la, pos, buf)) 1166 return(0); 1167 break; 1168 } 1169 1170 if (NULL == head) { 1171 if ( ! mdoc_head_alloc(m, line, ppos, tok)) 1172 return(0); 1173 head = m->last; 1174 } 1175 1176 if (nl && ! append_delims(m, line, pos, buf)) 1177 return(0); 1178 1179 /* If we've already opened our body, exit now. */ 1180 1181 if (NULL != body) 1182 goto out; 1183 1184 /* 1185 * If there is an open (i.e., unvalidated) sub-block requiring 1186 * explicit close-out, postpone switching the current block from 1187 * head to body until the rew_sub() call closing out that 1188 * sub-block. 1189 */ 1190 for (n = m->last; n && n != head; n = n->parent) { 1191 if (MDOC_BLOCK == n->type && 1192 MDOC_EXPLICIT & mdoc_macros[n->tok].flags && 1193 ! (MDOC_VALID & n->flags)) { 1194 n->pending = head; 1195 return(1); 1196 } 1197 } 1198 1199 /* Close out scopes to remain in a consistent state. */ 1200 1201 if ( ! rew_sub(MDOC_HEAD, m, tok, line, ppos)) 1202 return(0); 1203 if ( ! mdoc_body_alloc(m, line, ppos, tok)) 1204 return(0); 1205 1206 out: 1207 if ( ! (MDOC_FREECOL & m->flags)) 1208 return(1); 1209 1210 if ( ! rew_sub(MDOC_BODY, m, tok, line, ppos)) 1211 return(0); 1212 if ( ! rew_sub(MDOC_BLOCK, m, tok, line, ppos)) 1213 return(0); 1214 1215 m->flags &= ~MDOC_FREECOL; 1216 return(1); 1217 } 1218 1219 1220 static int 1221 blk_part_imp(MACRO_PROT_ARGS) 1222 { 1223 int la, nl; 1224 enum mdoct ntok; 1225 enum margserr ac; 1226 char *p; 1227 struct mdoc_node *blk; /* saved block context */ 1228 struct mdoc_node *body; /* saved body context */ 1229 struct mdoc_node *n; 1230 1231 nl = MDOC_NEWLINE & m->flags; 1232 1233 /* 1234 * A macro that spans to the end of the line. This is generally 1235 * (but not necessarily) called as the first macro. The block 1236 * has a head as the immediate child, which is always empty, 1237 * followed by zero or more opening punctuation nodes, then the 1238 * body (which may be empty, depending on the macro), then zero 1239 * or more closing punctuation nodes. 1240 */ 1241 1242 if ( ! mdoc_block_alloc(m, line, ppos, tok, NULL)) 1243 return(0); 1244 1245 blk = m->last; 1246 1247 if ( ! mdoc_head_alloc(m, line, ppos, tok)) 1248 return(0); 1249 if ( ! rew_sub(MDOC_HEAD, m, tok, line, ppos)) 1250 return(0); 1251 1252 /* 1253 * Open the body scope "on-demand", that is, after we've 1254 * processed all our the leading delimiters (open parenthesis, 1255 * etc.). 1256 */ 1257 1258 for (body = NULL; ; ) { 1259 la = *pos; 1260 ac = mdoc_args(m, line, pos, buf, tok, &p); 1261 1262 if (ARGS_ERROR == ac) 1263 return(0); 1264 if (ARGS_EOLN == ac) 1265 break; 1266 if (ARGS_PUNCT == ac) 1267 break; 1268 1269 if (NULL == body && ARGS_QWORD != ac && 1270 DELIM_OPEN == mdoc_isdelim(p)) { 1271 if ( ! dword(m, line, la, p, DELIM_OPEN)) 1272 return(0); 1273 continue; 1274 } 1275 1276 if (NULL == body) { 1277 if ( ! mdoc_body_alloc(m, line, ppos, tok)) 1278 return(0); 1279 body = m->last; 1280 } 1281 1282 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p); 1283 1284 if (MDOC_MAX == ntok) { 1285 if ( ! dword(m, line, la, p, DELIM_MAX)) 1286 return(0); 1287 continue; 1288 } 1289 1290 if ( ! mdoc_macro(m, ntok, line, la, pos, buf)) 1291 return(0); 1292 break; 1293 } 1294 1295 /* Clean-ups to leave in a consistent state. */ 1296 1297 if (NULL == body) { 1298 if ( ! mdoc_body_alloc(m, line, ppos, tok)) 1299 return(0); 1300 body = m->last; 1301 } 1302 1303 for (n = body->child; n && n->next; n = n->next) 1304 /* Do nothing. */ ; 1305 1306 /* 1307 * End of sentence spacing: if the last node is a text node and 1308 * has a trailing period, then mark it as being end-of-sentence. 1309 */ 1310 1311 if (n && MDOC_TEXT == n->type && n->string) 1312 if (mandoc_eos(n->string, strlen(n->string), 1)) 1313 n->flags |= MDOC_EOS; 1314 1315 /* Up-propagate the end-of-space flag. */ 1316 1317 if (n && (MDOC_EOS & n->flags)) { 1318 body->flags |= MDOC_EOS; 1319 body->parent->flags |= MDOC_EOS; 1320 } 1321 1322 /* 1323 * If there is an open sub-block requiring explicit close-out, 1324 * postpone closing out the current block 1325 * until the rew_sub() call closing out the sub-block. 1326 */ 1327 for (n = m->last; n && n != body && n != blk->parent; n = n->parent) { 1328 if (MDOC_BLOCK == n->type && 1329 MDOC_EXPLICIT & mdoc_macros[n->tok].flags && 1330 ! (MDOC_VALID & n->flags)) { 1331 make_pending(n, tok, m, line, ppos); 1332 if ( ! mdoc_endbody_alloc(m, line, ppos, 1333 tok, body, ENDBODY_NOSPACE)) 1334 return(0); 1335 return(1); 1336 } 1337 } 1338 1339 /* 1340 * If we can't rewind to our body, then our scope has already 1341 * been closed by another macro (like `Oc' closing `Op'). This 1342 * is ugly behaviour nodding its head to OpenBSD's overwhelming 1343 * crufty use of `Op' breakage. 1344 */ 1345 if (n != body) 1346 mandoc_vmsg(MANDOCERR_SCOPENEST, m->parse, line, ppos, 1347 "%s broken", mdoc_macronames[tok]); 1348 1349 if (n && ! rew_sub(MDOC_BODY, m, tok, line, ppos)) 1350 return(0); 1351 1352 /* Standard appending of delimiters. */ 1353 1354 if (nl && ! append_delims(m, line, pos, buf)) 1355 return(0); 1356 1357 /* Rewind scope, if applicable. */ 1358 1359 if (n && ! rew_sub(MDOC_BLOCK, m, tok, line, ppos)) 1360 return(0); 1361 1362 return(1); 1363 } 1364 1365 1366 static int 1367 blk_part_exp(MACRO_PROT_ARGS) 1368 { 1369 int la, nl; 1370 enum margserr ac; 1371 struct mdoc_node *head; /* keep track of head */ 1372 struct mdoc_node *body; /* keep track of body */ 1373 char *p; 1374 enum mdoct ntok; 1375 1376 nl = MDOC_NEWLINE & m->flags; 1377 1378 /* 1379 * The opening of an explicit macro having zero or more leading 1380 * punctuation nodes; a head with optional single element (the 1381 * case of `Eo'); and a body that may be empty. 1382 */ 1383 1384 if ( ! mdoc_block_alloc(m, line, ppos, tok, NULL)) 1385 return(0); 1386 1387 for (head = body = NULL; ; ) { 1388 la = *pos; 1389 ac = mdoc_args(m, line, pos, buf, tok, &p); 1390 1391 if (ARGS_ERROR == ac) 1392 return(0); 1393 if (ARGS_PUNCT == ac) 1394 break; 1395 if (ARGS_EOLN == ac) 1396 break; 1397 1398 /* Flush out leading punctuation. */ 1399 1400 if (NULL == head && ARGS_QWORD != ac && 1401 DELIM_OPEN == mdoc_isdelim(p)) { 1402 assert(NULL == body); 1403 if ( ! dword(m, line, la, p, DELIM_OPEN)) 1404 return(0); 1405 continue; 1406 } 1407 1408 if (NULL == head) { 1409 assert(NULL == body); 1410 if ( ! mdoc_head_alloc(m, line, ppos, tok)) 1411 return(0); 1412 head = m->last; 1413 } 1414 1415 /* 1416 * `Eo' gobbles any data into the head, but most other 1417 * macros just immediately close out and begin the body. 1418 */ 1419 1420 if (NULL == body) { 1421 assert(head); 1422 /* No check whether it's a macro! */ 1423 if (MDOC_Eo == tok) 1424 if ( ! dword(m, line, la, p, DELIM_MAX)) 1425 return(0); 1426 1427 if ( ! rew_sub(MDOC_HEAD, m, tok, line, ppos)) 1428 return(0); 1429 if ( ! mdoc_body_alloc(m, line, ppos, tok)) 1430 return(0); 1431 body = m->last; 1432 1433 if (MDOC_Eo == tok) 1434 continue; 1435 } 1436 1437 assert(NULL != head && NULL != body); 1438 1439 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p); 1440 1441 if (MDOC_MAX == ntok) { 1442 if ( ! dword(m, line, la, p, DELIM_MAX)) 1443 return(0); 1444 continue; 1445 } 1446 1447 if ( ! mdoc_macro(m, ntok, line, la, pos, buf)) 1448 return(0); 1449 break; 1450 } 1451 1452 /* Clean-up to leave in a consistent state. */ 1453 1454 if (NULL == head) 1455 if ( ! mdoc_head_alloc(m, line, ppos, tok)) 1456 return(0); 1457 1458 if (NULL == body) { 1459 if ( ! rew_sub(MDOC_HEAD, m, tok, line, ppos)) 1460 return(0); 1461 if ( ! mdoc_body_alloc(m, line, ppos, tok)) 1462 return(0); 1463 } 1464 1465 /* Standard appending of delimiters. */ 1466 1467 if ( ! nl) 1468 return(1); 1469 return(append_delims(m, line, pos, buf)); 1470 } 1471 1472 1473 /* ARGSUSED */ 1474 static int 1475 in_line_argn(MACRO_PROT_ARGS) 1476 { 1477 int la, flushed, j, maxargs, nl; 1478 enum margserr ac; 1479 enum margverr av; 1480 struct mdoc_arg *arg; 1481 char *p; 1482 enum mdoct ntok; 1483 1484 nl = MDOC_NEWLINE & m->flags; 1485 1486 /* 1487 * A line macro that has a fixed number of arguments (maxargs). 1488 * Only open the scope once the first non-leading-punctuation is 1489 * found (unless MDOC_IGNDELIM is noted, like in `Pf'), then 1490 * keep it open until the maximum number of arguments are 1491 * exhausted. 1492 */ 1493 1494 switch (tok) { 1495 case (MDOC_Ap): 1496 /* FALLTHROUGH */ 1497 case (MDOC_No): 1498 /* FALLTHROUGH */ 1499 case (MDOC_Ns): 1500 /* FALLTHROUGH */ 1501 case (MDOC_Ux): 1502 maxargs = 0; 1503 break; 1504 case (MDOC_Bx): 1505 /* FALLTHROUGH */ 1506 case (MDOC_Xr): 1507 maxargs = 2; 1508 break; 1509 default: 1510 maxargs = 1; 1511 break; 1512 } 1513 1514 for (arg = NULL; ; ) { 1515 la = *pos; 1516 av = mdoc_argv(m, line, tok, &arg, pos, buf); 1517 1518 if (ARGV_WORD == av) { 1519 *pos = la; 1520 break; 1521 } 1522 1523 if (ARGV_EOLN == av) 1524 break; 1525 if (ARGV_ARG == av) 1526 continue; 1527 1528 mdoc_argv_free(arg); 1529 return(0); 1530 } 1531 1532 for (flushed = j = 0; ; ) { 1533 la = *pos; 1534 ac = mdoc_args(m, line, pos, buf, tok, &p); 1535 1536 if (ARGS_ERROR == ac) 1537 return(0); 1538 if (ARGS_PUNCT == ac) 1539 break; 1540 if (ARGS_EOLN == ac) 1541 break; 1542 1543 if ( ! (MDOC_IGNDELIM & mdoc_macros[tok].flags) && 1544 ARGS_QWORD != ac && 0 == j && 1545 DELIM_OPEN == mdoc_isdelim(p)) { 1546 if ( ! dword(m, line, la, p, DELIM_OPEN)) 1547 return(0); 1548 continue; 1549 } else if (0 == j) 1550 if ( ! mdoc_elem_alloc(m, line, la, tok, arg)) 1551 return(0); 1552 1553 if (j == maxargs && ! flushed) { 1554 if ( ! rew_elem(m, tok)) 1555 return(0); 1556 flushed = 1; 1557 } 1558 1559 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p); 1560 1561 if (MDOC_MAX != ntok) { 1562 if ( ! flushed && ! rew_elem(m, tok)) 1563 return(0); 1564 flushed = 1; 1565 if ( ! mdoc_macro(m, ntok, line, la, pos, buf)) 1566 return(0); 1567 j++; 1568 break; 1569 } 1570 1571 if ( ! (MDOC_IGNDELIM & mdoc_macros[tok].flags) && 1572 ARGS_QWORD != ac && 1573 ! flushed && 1574 DELIM_NONE != mdoc_isdelim(p)) { 1575 if ( ! rew_elem(m, tok)) 1576 return(0); 1577 flushed = 1; 1578 } 1579 1580 if ( ! dword(m, line, la, p, DELIM_MAX)) 1581 return(0); 1582 j++; 1583 } 1584 1585 if (0 == j && ! mdoc_elem_alloc(m, line, la, tok, arg)) 1586 return(0); 1587 1588 /* Close out in a consistent state. */ 1589 1590 if ( ! flushed && ! rew_elem(m, tok)) 1591 return(0); 1592 if ( ! nl) 1593 return(1); 1594 return(append_delims(m, line, pos, buf)); 1595 } 1596 1597 1598 static int 1599 in_line_eoln(MACRO_PROT_ARGS) 1600 { 1601 int la; 1602 enum margserr ac; 1603 enum margverr av; 1604 struct mdoc_arg *arg; 1605 char *p; 1606 enum mdoct ntok; 1607 1608 assert( ! (MDOC_PARSED & mdoc_macros[tok].flags)); 1609 1610 if (tok == MDOC_Pp) 1611 rew_sub(MDOC_BLOCK, m, MDOC_Nm, line, ppos); 1612 1613 /* Parse macro arguments. */ 1614 1615 for (arg = NULL; ; ) { 1616 la = *pos; 1617 av = mdoc_argv(m, line, tok, &arg, pos, buf); 1618 1619 if (ARGV_WORD == av) { 1620 *pos = la; 1621 break; 1622 } 1623 if (ARGV_EOLN == av) 1624 break; 1625 if (ARGV_ARG == av) 1626 continue; 1627 1628 mdoc_argv_free(arg); 1629 return(0); 1630 } 1631 1632 /* Open element scope. */ 1633 1634 if ( ! mdoc_elem_alloc(m, line, ppos, tok, arg)) 1635 return(0); 1636 1637 /* Parse argument terms. */ 1638 1639 for (;;) { 1640 la = *pos; 1641 ac = mdoc_args(m, line, pos, buf, tok, &p); 1642 1643 if (ARGS_ERROR == ac) 1644 return(0); 1645 if (ARGS_EOLN == ac) 1646 break; 1647 1648 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p); 1649 1650 if (MDOC_MAX == ntok) { 1651 if ( ! dword(m, line, la, p, DELIM_MAX)) 1652 return(0); 1653 continue; 1654 } 1655 1656 if ( ! rew_elem(m, tok)) 1657 return(0); 1658 return(mdoc_macro(m, ntok, line, la, pos, buf)); 1659 } 1660 1661 /* Close out (no delimiters). */ 1662 1663 return(rew_elem(m, tok)); 1664 } 1665 1666 1667 /* ARGSUSED */ 1668 static int 1669 ctx_synopsis(MACRO_PROT_ARGS) 1670 { 1671 int nl; 1672 1673 nl = MDOC_NEWLINE & m->flags; 1674 1675 /* If we're not in the SYNOPSIS, go straight to in-line. */ 1676 if ( ! (MDOC_SYNOPSIS & m->flags)) 1677 return(in_line(m, tok, line, ppos, pos, buf)); 1678 1679 /* If we're a nested call, same place. */ 1680 if ( ! nl) 1681 return(in_line(m, tok, line, ppos, pos, buf)); 1682 1683 /* 1684 * XXX: this will open a block scope; however, if later we end 1685 * up formatting the block scope, then child nodes will inherit 1686 * the formatting. Be careful. 1687 */ 1688 if (MDOC_Nm == tok) 1689 return(blk_full(m, tok, line, ppos, pos, buf)); 1690 assert(MDOC_Vt == tok); 1691 return(blk_part_imp(m, tok, line, ppos, pos, buf)); 1692 } 1693 1694 1695 /* ARGSUSED */ 1696 static int 1697 obsolete(MACRO_PROT_ARGS) 1698 { 1699 1700 mdoc_pmsg(m, line, ppos, MANDOCERR_MACROOBS); 1701 return(1); 1702 } 1703 1704 1705 /* 1706 * Phrases occur within `Bl -column' entries, separated by `Ta' or tabs. 1707 * They're unusual because they're basically free-form text until a 1708 * macro is encountered. 1709 */ 1710 static int 1711 phrase(struct mdoc *m, int line, int ppos, char *buf) 1712 { 1713 int la, pos; 1714 enum margserr ac; 1715 enum mdoct ntok; 1716 char *p; 1717 1718 for (pos = ppos; ; ) { 1719 la = pos; 1720 1721 ac = mdoc_zargs(m, line, &pos, buf, &p); 1722 1723 if (ARGS_ERROR == ac) 1724 return(0); 1725 if (ARGS_EOLN == ac) 1726 break; 1727 1728 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup_raw(p); 1729 1730 if (MDOC_MAX == ntok) { 1731 if ( ! dword(m, line, la, p, DELIM_MAX)) 1732 return(0); 1733 continue; 1734 } 1735 1736 if ( ! mdoc_macro(m, ntok, line, la, &pos, buf)) 1737 return(0); 1738 return(append_delims(m, line, &pos, buf)); 1739 } 1740 1741 return(1); 1742 } 1743 1744 1745 /* ARGSUSED */ 1746 static int 1747 phrase_ta(MACRO_PROT_ARGS) 1748 { 1749 int la; 1750 enum mdoct ntok; 1751 enum margserr ac; 1752 char *p; 1753 1754 /* 1755 * FIXME: this is overly restrictive: if the `Ta' is unexpected, 1756 * it should simply error out with ARGSLOST. 1757 */ 1758 1759 if ( ! rew_sub(MDOC_BODY, m, MDOC_It, line, ppos)) 1760 return(0); 1761 if ( ! mdoc_body_alloc(m, line, ppos, MDOC_It)) 1762 return(0); 1763 1764 for (;;) { 1765 la = *pos; 1766 ac = mdoc_zargs(m, line, pos, buf, &p); 1767 1768 if (ARGS_ERROR == ac) 1769 return(0); 1770 if (ARGS_EOLN == ac) 1771 break; 1772 1773 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup_raw(p); 1774 1775 if (MDOC_MAX == ntok) { 1776 if ( ! dword(m, line, la, p, DELIM_MAX)) 1777 return(0); 1778 continue; 1779 } 1780 1781 if ( ! mdoc_macro(m, ntok, line, la, pos, buf)) 1782 return(0); 1783 return(append_delims(m, line, pos, buf)); 1784 } 1785 1786 return(1); 1787 }