1 /* $Id: mdoc_macro.c,v 1.139 2014/08/01 17:27:44 schwarze Exp $ */ 2 /* 3 * Copyright (c) 2008-2012 Kristaps Dzonsons <kristaps@bsd.lv> 4 * Copyright (c) 2010, 2012, 2013 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 phrase_ta(MACRO_PROT_ARGS); 52 53 static int dword(struct mdoc *, int, int, const char *, 54 enum mdelim, int); 55 static int append_delims(struct mdoc *, 56 int, int *, char *); 57 static enum mdoct lookup(enum mdoct, const char *); 58 static enum mdoct lookup_raw(const char *); 59 static int make_pending(struct mdoc_node *, enum mdoct, 60 struct mdoc *, int, int); 61 static int phrase(struct mdoc *, int, int, char *); 62 static enum mdoct rew_alt(enum mdoct); 63 static enum rew rew_dohalt(enum mdoct, enum mdoc_type, 64 const struct mdoc_node *); 65 static int rew_elem(struct mdoc *, enum mdoct); 66 static int rew_last(struct mdoc *, 67 const struct mdoc_node *); 68 static int rew_sub(enum mdoc_type, struct mdoc *, 69 enum mdoct, int, int); 70 71 const struct mdoc_macro __mdoc_macros[MDOC_MAX] = { 72 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Ap */ 73 { in_line_eoln, MDOC_PROLOGUE }, /* Dd */ 74 { in_line_eoln, MDOC_PROLOGUE }, /* Dt */ 75 { in_line_eoln, MDOC_PROLOGUE }, /* Os */ 76 { blk_full, MDOC_PARSED | MDOC_JOIN }, /* Sh */ 77 { blk_full, MDOC_PARSED | MDOC_JOIN }, /* Ss */ 78 { in_line_eoln, 0 }, /* Pp */ 79 { blk_part_imp, MDOC_PARSED | MDOC_JOIN }, /* D1 */ 80 { blk_part_imp, MDOC_PARSED | MDOC_JOIN }, /* Dl */ 81 { blk_full, MDOC_EXPLICIT }, /* Bd */ 82 { blk_exp_close, MDOC_EXPLICIT | MDOC_JOIN }, /* Ed */ 83 { blk_full, MDOC_EXPLICIT }, /* Bl */ 84 { blk_exp_close, MDOC_EXPLICIT | MDOC_JOIN }, /* El */ 85 { blk_full, MDOC_PARSED | MDOC_JOIN }, /* It */ 86 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ad */ 87 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* An */ 88 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ar */ 89 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Cd */ 90 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Cm */ 91 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Dv */ 92 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Er */ 93 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ev */ 94 { in_line_eoln, 0 }, /* Ex */ 95 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fa */ 96 { in_line_eoln, 0 }, /* Fd */ 97 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fl */ 98 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fn */ 99 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ft */ 100 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Ic */ 101 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* In */ 102 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Li */ 103 { blk_full, MDOC_JOIN }, /* Nd */ 104 { ctx_synopsis, MDOC_CALLABLE | MDOC_PARSED }, /* Nm */ 105 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED }, /* Op */ 106 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ot */ 107 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Pa */ 108 { in_line_eoln, 0 }, /* Rv */ 109 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* St */ 110 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Va */ 111 { ctx_synopsis, MDOC_CALLABLE | MDOC_PARSED }, /* Vt */ 112 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Xr */ 113 { in_line_eoln, MDOC_JOIN }, /* %A */ 114 { in_line_eoln, MDOC_JOIN }, /* %B */ 115 { in_line_eoln, MDOC_JOIN }, /* %D */ 116 { in_line_eoln, MDOC_JOIN }, /* %I */ 117 { in_line_eoln, MDOC_JOIN }, /* %J */ 118 { in_line_eoln, 0 }, /* %N */ 119 { in_line_eoln, MDOC_JOIN }, /* %O */ 120 { in_line_eoln, 0 }, /* %P */ 121 { in_line_eoln, MDOC_JOIN }, /* %R */ 122 { in_line_eoln, MDOC_JOIN }, /* %T */ 123 { in_line_eoln, 0 }, /* %V */ 124 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED | 125 MDOC_EXPLICIT | MDOC_JOIN }, /* Ac */ 126 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | 127 MDOC_EXPLICIT | MDOC_JOIN }, /* Ao */ 128 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Aq */ 129 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* At */ 130 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED | 131 MDOC_EXPLICIT | MDOC_JOIN }, /* Bc */ 132 { blk_full, MDOC_EXPLICIT }, /* Bf */ 133 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | 134 MDOC_EXPLICIT | MDOC_JOIN }, /* Bo */ 135 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Bq */ 136 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Bsx */ 137 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Bx */ 138 { in_line_eoln, 0 }, /* Db */ 139 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED | 140 MDOC_EXPLICIT | MDOC_JOIN }, /* Dc */ 141 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | 142 MDOC_EXPLICIT | MDOC_JOIN }, /* Do */ 143 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Dq */ 144 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Ec */ 145 { blk_exp_close, MDOC_EXPLICIT | MDOC_JOIN }, /* Ef */ 146 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Em */ 147 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Eo */ 148 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Fx */ 149 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Ms */ 150 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED | 151 MDOC_IGNDELIM | MDOC_JOIN }, /* No */ 152 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED | 153 MDOC_IGNDELIM | MDOC_JOIN }, /* Ns */ 154 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Nx */ 155 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Ox */ 156 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED | 157 MDOC_EXPLICIT | MDOC_JOIN }, /* Pc */ 158 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED | MDOC_IGNDELIM }, /* Pf */ 159 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | 160 MDOC_EXPLICIT | MDOC_JOIN }, /* Po */ 161 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Pq */ 162 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED | 163 MDOC_EXPLICIT | MDOC_JOIN }, /* Qc */ 164 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Ql */ 165 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | 166 MDOC_EXPLICIT | MDOC_JOIN }, /* Qo */ 167 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Qq */ 168 { blk_exp_close, MDOC_EXPLICIT | MDOC_JOIN }, /* Re */ 169 { blk_full, MDOC_EXPLICIT }, /* Rs */ 170 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED | 171 MDOC_EXPLICIT | MDOC_JOIN }, /* Sc */ 172 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | 173 MDOC_EXPLICIT | MDOC_JOIN }, /* So */ 174 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Sq */ 175 { in_line_eoln, 0 }, /* Sm */ 176 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Sx */ 177 { in_line, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Sy */ 178 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Tn */ 179 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Ux */ 180 { blk_exp_close, MDOC_EXPLICIT | MDOC_CALLABLE | MDOC_PARSED }, /* Xc */ 181 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | MDOC_EXPLICIT }, /* Xo */ 182 { blk_full, MDOC_EXPLICIT | MDOC_CALLABLE }, /* Fo */ 183 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED | 184 MDOC_EXPLICIT | MDOC_JOIN }, /* Fc */ 185 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | 186 MDOC_EXPLICIT | MDOC_JOIN }, /* Oo */ 187 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED | 188 MDOC_EXPLICIT | MDOC_JOIN }, /* Oc */ 189 { blk_full, MDOC_EXPLICIT }, /* Bk */ 190 { blk_exp_close, MDOC_EXPLICIT | MDOC_JOIN }, /* Ek */ 191 { in_line_eoln, 0 }, /* Bt */ 192 { in_line_eoln, 0 }, /* Hf */ 193 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Fr */ 194 { in_line_eoln, 0 }, /* Ud */ 195 { in_line, 0 }, /* Lb */ 196 { in_line_eoln, 0 }, /* Lp */ 197 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Lk */ 198 { in_line, MDOC_CALLABLE | MDOC_PARSED }, /* Mt */ 199 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Brq */ 200 { blk_part_exp, MDOC_CALLABLE | MDOC_PARSED | 201 MDOC_EXPLICIT | MDOC_JOIN }, /* Bro */ 202 { blk_exp_close, MDOC_CALLABLE | MDOC_PARSED | 203 MDOC_EXPLICIT | MDOC_JOIN }, /* Brc */ 204 { in_line_eoln, MDOC_JOIN }, /* %C */ 205 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Es */ 206 { blk_part_imp, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* En */ 207 { in_line_argn, MDOC_CALLABLE | MDOC_PARSED }, /* Dx */ 208 { in_line_eoln, MDOC_JOIN }, /* %Q */ 209 { in_line_eoln, 0 }, /* br */ 210 { in_line_eoln, 0 }, /* sp */ 211 { in_line_eoln, 0 }, /* %U */ 212 { phrase_ta, MDOC_CALLABLE | MDOC_PARSED | MDOC_JOIN }, /* Ta */ 213 { in_line_eoln, 0 }, /* ll */ 214 }; 215 216 const struct mdoc_macro * const mdoc_macros = __mdoc_macros; 217 218 219 /* 220 * This is called at the end of parsing. It must traverse up the tree, 221 * closing out open [implicit] scopes. Obviously, open explicit scopes 222 * are errors. 223 */ 224 int 225 mdoc_macroend(struct mdoc *mdoc) 226 { 227 struct mdoc_node *n; 228 229 /* Scan for open explicit scopes. */ 230 231 n = MDOC_VALID & mdoc->last->flags ? 232 mdoc->last->parent : mdoc->last; 233 234 for ( ; n; n = n->parent) 235 if (MDOC_BLOCK == n->type && 236 MDOC_EXPLICIT & mdoc_macros[n->tok].flags) 237 mandoc_msg(MANDOCERR_BLK_NOEND, mdoc->parse, 238 n->line, n->pos, mdoc_macronames[n->tok]); 239 240 /* Rewind to the first. */ 241 242 return(rew_last(mdoc, mdoc->first)); 243 } 244 245 /* 246 * Look up a macro from within a subsequent context. 247 */ 248 static enum mdoct 249 lookup(enum mdoct from, const char *p) 250 { 251 252 if ( ! (MDOC_PARSED & mdoc_macros[from].flags)) 253 return(MDOC_MAX); 254 return(lookup_raw(p)); 255 } 256 257 /* 258 * Lookup a macro following the initial line macro. 259 */ 260 static enum mdoct 261 lookup_raw(const char *p) 262 { 263 enum mdoct res; 264 265 if (MDOC_MAX == (res = mdoc_hash_find(p))) 266 return(MDOC_MAX); 267 if (MDOC_CALLABLE & mdoc_macros[res].flags) 268 return(res); 269 return(MDOC_MAX); 270 } 271 272 static int 273 rew_last(struct mdoc *mdoc, const struct mdoc_node *to) 274 { 275 struct mdoc_node *n, *np; 276 277 assert(to); 278 mdoc->next = MDOC_NEXT_SIBLING; 279 280 281 while (mdoc->last != to) { 282 /* 283 * Save the parent here, because we may delete the 284 * mdoc->last node in the post-validation phase and reset 285 * it to mdoc->last->parent, causing a step in the closing 286 * out to be lost. 287 */ 288 np = mdoc->last->parent; 289 if ( ! mdoc_valid_post(mdoc)) 290 return(0); 291 n = mdoc->last; 292 mdoc->last = np; 293 assert(mdoc->last); 294 mdoc->last->last = n; 295 } 296 297 return(mdoc_valid_post(mdoc)); 298 } 299 300 /* 301 * For a block closing macro, return the corresponding opening one. 302 * Otherwise, return the macro itself. 303 */ 304 static enum mdoct 305 rew_alt(enum mdoct tok) 306 { 307 switch (tok) { 308 case MDOC_Ac: 309 return(MDOC_Ao); 310 case MDOC_Bc: 311 return(MDOC_Bo); 312 case MDOC_Brc: 313 return(MDOC_Bro); 314 case MDOC_Dc: 315 return(MDOC_Do); 316 case MDOC_Ec: 317 return(MDOC_Eo); 318 case MDOC_Ed: 319 return(MDOC_Bd); 320 case MDOC_Ef: 321 return(MDOC_Bf); 322 case MDOC_Ek: 323 return(MDOC_Bk); 324 case MDOC_El: 325 return(MDOC_Bl); 326 case MDOC_Fc: 327 return(MDOC_Fo); 328 case MDOC_Oc: 329 return(MDOC_Oo); 330 case MDOC_Pc: 331 return(MDOC_Po); 332 case MDOC_Qc: 333 return(MDOC_Qo); 334 case MDOC_Re: 335 return(MDOC_Rs); 336 case MDOC_Sc: 337 return(MDOC_So); 338 case MDOC_Xc: 339 return(MDOC_Xo); 340 default: 341 return(tok); 342 } 343 /* NOTREACHED */ 344 } 345 346 /* 347 * Rewinding to tok, how do we have to handle *p? 348 * REWIND_NONE: *p would delimit tok, but no tok scope is open 349 * inside *p, so there is no need to rewind anything at all. 350 * REWIND_THIS: *p matches tok, so rewind *p and nothing else. 351 * REWIND_MORE: *p is implicit, rewind it and keep searching for tok. 352 * REWIND_FORCE: *p is explicit, but tok is full, force rewinding *p. 353 * REWIND_LATER: *p is explicit and still open, postpone rewinding. 354 * REWIND_ERROR: No tok block is open at all. 355 */ 356 static enum rew 357 rew_dohalt(enum mdoct tok, enum mdoc_type type, 358 const struct mdoc_node *p) 359 { 360 361 /* 362 * No matching token, no delimiting block, no broken block. 363 * This can happen when full implicit macros are called for 364 * the first time but try to rewind their previous 365 * instance anyway. 366 */ 367 if (MDOC_ROOT == p->type) 368 return(MDOC_BLOCK == type && 369 MDOC_EXPLICIT & mdoc_macros[tok].flags ? 370 REWIND_ERROR : REWIND_NONE); 371 372 /* 373 * When starting to rewind, skip plain text 374 * and nodes that have already been rewound. 375 */ 376 if (MDOC_TEXT == p->type || MDOC_VALID & p->flags) 377 return(REWIND_MORE); 378 379 /* 380 * The easiest case: Found a matching token. 381 * This applies to both blocks and elements. 382 */ 383 tok = rew_alt(tok); 384 if (tok == p->tok) 385 return(p->end ? REWIND_NONE : 386 type == p->type ? REWIND_THIS : REWIND_MORE); 387 388 /* 389 * While elements do require rewinding for themselves, 390 * they never affect rewinding of other nodes. 391 */ 392 if (MDOC_ELEM == p->type) 393 return(REWIND_MORE); 394 395 /* 396 * Blocks delimited by our target token get REWIND_MORE. 397 * Blocks delimiting our target token get REWIND_NONE. 398 */ 399 switch (tok) { 400 case MDOC_Bl: 401 if (MDOC_It == p->tok) 402 return(REWIND_MORE); 403 break; 404 case MDOC_It: 405 if (MDOC_BODY == p->type && MDOC_Bl == p->tok) 406 return(REWIND_NONE); 407 break; 408 /* 409 * XXX Badly nested block handling still fails badly 410 * when one block is breaking two blocks of the same type. 411 * This is an incomplete and extremely ugly workaround, 412 * required to let the OpenBSD tree build. 413 */ 414 case MDOC_Oo: 415 if (MDOC_Op == p->tok) 416 return(REWIND_MORE); 417 break; 418 case MDOC_Nm: 419 return(REWIND_NONE); 420 case MDOC_Nd: 421 /* FALLTHROUGH */ 422 case MDOC_Ss: 423 if (MDOC_BODY == p->type && MDOC_Sh == p->tok) 424 return(REWIND_NONE); 425 /* FALLTHROUGH */ 426 case MDOC_Sh: 427 if (MDOC_ROOT == p->parent->type) 428 return(REWIND_THIS); 429 if (MDOC_Nd == p->tok || MDOC_Ss == p->tok || 430 MDOC_Sh == p->tok) 431 return(REWIND_MORE); 432 break; 433 default: 434 break; 435 } 436 437 /* 438 * Default block rewinding rules. 439 * In particular, always skip block end markers, 440 * and let all blocks rewind Nm children. 441 */ 442 if (ENDBODY_NOT != p->end || MDOC_Nm == p->tok || 443 (MDOC_BLOCK == p->type && 444 ! (MDOC_EXPLICIT & mdoc_macros[tok].flags))) 445 return(REWIND_MORE); 446 447 /* 448 * By default, closing out full blocks 449 * forces closing of broken explicit blocks, 450 * while closing out partial blocks 451 * allows delayed rewinding by default. 452 */ 453 return (&blk_full == mdoc_macros[tok].fp ? 454 REWIND_FORCE : REWIND_LATER); 455 } 456 457 static int 458 rew_elem(struct mdoc *mdoc, enum mdoct tok) 459 { 460 struct mdoc_node *n; 461 462 n = mdoc->last; 463 if (MDOC_ELEM != n->type) 464 n = n->parent; 465 assert(MDOC_ELEM == n->type); 466 assert(tok == n->tok); 467 468 return(rew_last(mdoc, n)); 469 } 470 471 /* 472 * We are trying to close a block identified by tok, 473 * but the child block *broken is still open. 474 * Thus, postpone closing the tok block 475 * until the rew_sub call closing *broken. 476 */ 477 static int 478 make_pending(struct mdoc_node *broken, enum mdoct tok, 479 struct mdoc *mdoc, int line, int ppos) 480 { 481 struct mdoc_node *breaker; 482 483 /* 484 * Iterate backwards, searching for the block matching tok, 485 * that is, the block breaking the *broken block. 486 */ 487 for (breaker = broken->parent; breaker; breaker = breaker->parent) { 488 489 /* 490 * If the *broken block had already been broken before 491 * and we encounter its breaker, make the tok block 492 * pending on the inner breaker. 493 * Graphically, "[A breaker=[B broken=[C->B B] tok=A] C]" 494 * becomes "[A broken=[B [C->B B] tok=A] C]" 495 * and finally "[A [B->A [C->B B] A] C]". 496 */ 497 if (breaker == broken->pending) { 498 broken = breaker; 499 continue; 500 } 501 502 if (REWIND_THIS != rew_dohalt(tok, MDOC_BLOCK, breaker)) 503 continue; 504 if (MDOC_BODY == broken->type) 505 broken = broken->parent; 506 507 /* 508 * Found the breaker. 509 * If another, outer breaker is already pending on 510 * the *broken block, we must not clobber the link 511 * to the outer breaker, but make it pending on the 512 * new, now inner breaker. 513 * Graphically, "[A breaker=[B broken=[C->A A] tok=B] C]" 514 * becomes "[A breaker=[B->A broken=[C A] tok=B] C]" 515 * and finally "[A [B->A [C->B A] B] C]". 516 */ 517 if (broken->pending) { 518 struct mdoc_node *taker; 519 520 /* 521 * If the breaker had also been broken before, 522 * it cannot take on the outer breaker itself, 523 * but must hand it on to its own breakers. 524 * Graphically, this is the following situation: 525 * "[A [B breaker=[C->B B] broken=[D->A A] tok=C] D]" 526 * "[A taker=[B->A breaker=[C->B B] [D->C A] C] D]" 527 */ 528 taker = breaker; 529 while (taker->pending) 530 taker = taker->pending; 531 taker->pending = broken->pending; 532 } 533 broken->pending = breaker; 534 mandoc_vmsg(MANDOCERR_BLK_NEST, mdoc->parse, line, ppos, 535 "%s breaks %s", mdoc_macronames[tok], 536 mdoc_macronames[broken->tok]); 537 return(1); 538 } 539 540 /* 541 * Found no matching block for tok. 542 * Are you trying to close a block that is not open? 543 */ 544 return(0); 545 } 546 547 static int 548 rew_sub(enum mdoc_type t, struct mdoc *mdoc, 549 enum mdoct tok, int line, int ppos) 550 { 551 struct mdoc_node *n; 552 553 n = mdoc->last; 554 while (n) { 555 switch (rew_dohalt(tok, t, n)) { 556 case REWIND_NONE: 557 return(1); 558 case REWIND_THIS: 559 n->lastline = line - 560 (MDOC_NEWLINE & mdoc->flags && 561 ! (MDOC_EXPLICIT & mdoc_macros[tok].flags)); 562 break; 563 case REWIND_FORCE: 564 mandoc_vmsg(MANDOCERR_BLK_BROKEN, mdoc->parse, 565 line, ppos, "%s breaks %s", 566 mdoc_macronames[tok], 567 mdoc_macronames[n->tok]); 568 /* FALLTHROUGH */ 569 case REWIND_MORE: 570 n->lastline = line - 571 (MDOC_NEWLINE & mdoc->flags ? 1 : 0); 572 n = n->parent; 573 continue; 574 case REWIND_LATER: 575 if (make_pending(n, tok, mdoc, line, ppos) || 576 MDOC_BLOCK != t) 577 return(1); 578 /* FALLTHROUGH */ 579 case REWIND_ERROR: 580 mandoc_msg(MANDOCERR_BLK_NOTOPEN, 581 mdoc->parse, line, ppos, 582 mdoc_macronames[tok]); 583 return(1); 584 } 585 break; 586 } 587 588 assert(n); 589 if ( ! rew_last(mdoc, n)) 590 return(0); 591 592 /* 593 * The current block extends an enclosing block. 594 * Now that the current block ends, close the enclosing block, too. 595 */ 596 while (NULL != (n = n->pending)) { 597 if ( ! rew_last(mdoc, n)) 598 return(0); 599 if (MDOC_HEAD == n->type && 600 ! mdoc_body_alloc(mdoc, n->line, n->pos, n->tok)) 601 return(0); 602 } 603 604 return(1); 605 } 606 607 /* 608 * Allocate a word and check whether it's punctuation or not. 609 * Punctuation consists of those tokens found in mdoc_isdelim(). 610 */ 611 static int 612 dword(struct mdoc *mdoc, int line, int col, const char *p, 613 enum mdelim d, int may_append) 614 { 615 616 if (DELIM_MAX == d) 617 d = mdoc_isdelim(p); 618 619 if (may_append && 620 ! ((MDOC_SYNOPSIS | MDOC_KEEP | MDOC_SMOFF) & mdoc->flags) && 621 DELIM_NONE == d && MDOC_TEXT == mdoc->last->type && 622 DELIM_NONE == mdoc_isdelim(mdoc->last->string)) { 623 mdoc_word_append(mdoc, p); 624 return(1); 625 } 626 627 if ( ! mdoc_word_alloc(mdoc, line, col, p)) 628 return(0); 629 630 if (DELIM_OPEN == d) 631 mdoc->last->flags |= MDOC_DELIMO; 632 633 /* 634 * Closing delimiters only suppress the preceding space 635 * when they follow something, not when they start a new 636 * block or element, and not when they follow `No'. 637 * 638 * XXX Explicitly special-casing MDOC_No here feels 639 * like a layering violation. Find a better way 640 * and solve this in the code related to `No'! 641 */ 642 643 else if (DELIM_CLOSE == d && mdoc->last->prev && 644 mdoc->last->prev->tok != MDOC_No && 645 mdoc->last->parent->tok != MDOC_Fd) 646 mdoc->last->flags |= MDOC_DELIMC; 647 648 return(1); 649 } 650 651 static int 652 append_delims(struct mdoc *mdoc, int line, int *pos, char *buf) 653 { 654 int la; 655 enum margserr ac; 656 char *p; 657 658 if ('\0' == buf[*pos]) 659 return(1); 660 661 for (;;) { 662 la = *pos; 663 ac = mdoc_zargs(mdoc, line, pos, buf, &p); 664 665 if (ARGS_ERROR == ac) 666 return(0); 667 else if (ARGS_EOLN == ac) 668 break; 669 670 dword(mdoc, line, la, p, DELIM_MAX, 1); 671 672 /* 673 * If we encounter end-of-sentence symbols, then trigger 674 * the double-space. 675 * 676 * XXX: it's easy to allow this to propagate outward to 677 * the last symbol, such that `. )' will cause the 678 * correct double-spacing. However, (1) groff isn't 679 * smart enough to do this and (2) it would require 680 * knowing which symbols break this behaviour, for 681 * example, `. ;' shouldn't propagate the double-space. 682 */ 683 if (mandoc_eos(p, strlen(p))) 684 mdoc->last->flags |= MDOC_EOS; 685 } 686 687 return(1); 688 } 689 690 /* 691 * Close out block partial/full explicit. 692 */ 693 static int 694 blk_exp_close(MACRO_PROT_ARGS) 695 { 696 struct mdoc_node *body; /* Our own body. */ 697 struct mdoc_node *later; /* A sub-block starting later. */ 698 struct mdoc_node *n; /* For searching backwards. */ 699 700 int j, lastarg, maxargs, flushed, nl; 701 enum margserr ac; 702 enum mdoct atok, ntok; 703 char *p; 704 705 nl = MDOC_NEWLINE & mdoc->flags; 706 707 switch (tok) { 708 case MDOC_Ec: 709 maxargs = 1; 710 break; 711 case MDOC_Ek: 712 mdoc->flags &= ~MDOC_KEEP; 713 default: 714 maxargs = 0; 715 break; 716 } 717 718 /* 719 * Search backwards for beginnings of blocks, 720 * both of our own and of pending sub-blocks. 721 */ 722 atok = rew_alt(tok); 723 body = later = NULL; 724 for (n = mdoc->last; n; n = n->parent) { 725 if (MDOC_VALID & n->flags) 726 continue; 727 728 /* Remember the start of our own body. */ 729 if (MDOC_BODY == n->type && atok == n->tok) { 730 if (ENDBODY_NOT == n->end) 731 body = n; 732 continue; 733 } 734 735 if (MDOC_BLOCK != n->type || MDOC_Nm == n->tok) 736 continue; 737 if (atok == n->tok) { 738 assert(body); 739 740 /* 741 * Found the start of our own block. 742 * When there is no pending sub block, 743 * just proceed to closing out. 744 */ 745 if (NULL == later) 746 break; 747 748 /* 749 * When there is a pending sub block, 750 * postpone closing out the current block 751 * until the rew_sub() closing out the sub-block. 752 */ 753 make_pending(later, tok, mdoc, line, ppos); 754 755 /* 756 * Mark the place where the formatting - but not 757 * the scope - of the current block ends. 758 */ 759 if ( ! mdoc_endbody_alloc(mdoc, line, ppos, 760 atok, body, ENDBODY_SPACE)) 761 return(0); 762 break; 763 } 764 765 /* 766 * When finding an open sub block, remember the last 767 * open explicit block, or, in case there are only 768 * implicit ones, the first open implicit block. 769 */ 770 if (later && 771 MDOC_EXPLICIT & mdoc_macros[later->tok].flags) 772 continue; 773 if (MDOC_It != n->tok) 774 later = n; 775 } 776 777 if ( ! (MDOC_PARSED & mdoc_macros[tok].flags)) { 778 if ('\0' != buf[*pos]) 779 mandoc_vmsg(MANDOCERR_ARG_SKIP, 780 mdoc->parse, line, ppos, 781 "%s %s", mdoc_macronames[tok], 782 buf + *pos); 783 if ( ! rew_sub(MDOC_BODY, mdoc, tok, line, ppos)) 784 return(0); 785 return(rew_sub(MDOC_BLOCK, mdoc, tok, line, ppos)); 786 } 787 788 if ( ! rew_sub(MDOC_BODY, mdoc, tok, line, ppos)) 789 return(0); 790 791 if (NULL == later && maxargs > 0) 792 if ( ! mdoc_tail_alloc(mdoc, line, ppos, rew_alt(tok))) 793 return(0); 794 795 for (flushed = j = 0; ; j++) { 796 lastarg = *pos; 797 798 if (j == maxargs && ! flushed) { 799 if ( ! rew_sub(MDOC_BLOCK, mdoc, tok, line, ppos)) 800 return(0); 801 flushed = 1; 802 } 803 804 ac = mdoc_args(mdoc, line, pos, buf, tok, &p); 805 806 if (ARGS_ERROR == ac) 807 return(0); 808 if (ARGS_PUNCT == ac) 809 break; 810 if (ARGS_EOLN == ac) 811 break; 812 813 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p); 814 815 if (MDOC_MAX == ntok) { 816 if ( ! dword(mdoc, line, lastarg, p, DELIM_MAX, 817 MDOC_JOIN & mdoc_macros[tok].flags)) 818 return(0); 819 continue; 820 } 821 822 if ( ! flushed) { 823 if ( ! rew_sub(MDOC_BLOCK, mdoc, tok, line, ppos)) 824 return(0); 825 flushed = 1; 826 } 827 828 mdoc->flags &= ~MDOC_NEWLINE; 829 830 if ( ! mdoc_macro(mdoc, ntok, line, lastarg, pos, buf)) 831 return(0); 832 break; 833 } 834 835 if ( ! flushed && ! rew_sub(MDOC_BLOCK, mdoc, tok, line, ppos)) 836 return(0); 837 838 if ( ! nl) 839 return(1); 840 return(append_delims(mdoc, line, pos, buf)); 841 } 842 843 static int 844 in_line(MACRO_PROT_ARGS) 845 { 846 int la, scope, cnt, nc, nl; 847 enum margverr av; 848 enum mdoct ntok; 849 enum margserr ac; 850 enum mdelim d; 851 struct mdoc_arg *arg; 852 char *p; 853 854 nl = MDOC_NEWLINE & mdoc->flags; 855 856 /* 857 * Whether we allow ignored elements (those without content, 858 * usually because of reserved words) to squeak by. 859 */ 860 861 switch (tok) { 862 case MDOC_An: 863 /* FALLTHROUGH */ 864 case MDOC_Ar: 865 /* FALLTHROUGH */ 866 case MDOC_Fl: 867 /* FALLTHROUGH */ 868 case MDOC_Mt: 869 /* FALLTHROUGH */ 870 case MDOC_Nm: 871 /* FALLTHROUGH */ 872 case MDOC_Pa: 873 nc = 1; 874 break; 875 default: 876 nc = 0; 877 break; 878 } 879 880 for (arg = NULL;; ) { 881 la = *pos; 882 av = mdoc_argv(mdoc, line, tok, &arg, pos, buf); 883 884 if (ARGV_WORD == av) { 885 *pos = la; 886 break; 887 } 888 if (ARGV_EOLN == av) 889 break; 890 if (ARGV_ARG == av) 891 continue; 892 893 mdoc_argv_free(arg); 894 return(0); 895 } 896 897 for (cnt = scope = 0;; ) { 898 la = *pos; 899 ac = mdoc_args(mdoc, line, pos, buf, tok, &p); 900 901 if (ARGS_ERROR == ac) 902 return(0); 903 if (ARGS_EOLN == ac) 904 break; 905 if (ARGS_PUNCT == ac) 906 break; 907 908 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p); 909 910 /* 911 * In this case, we've located a submacro and must 912 * execute it. Close out scope, if open. If no 913 * elements have been generated, either create one (nc) 914 * or raise a warning. 915 */ 916 917 if (MDOC_MAX != ntok) { 918 if (scope && ! rew_elem(mdoc, tok)) 919 return(0); 920 if (nc && 0 == cnt) { 921 if ( ! mdoc_elem_alloc(mdoc, 922 line, ppos, tok, arg)) 923 return(0); 924 if ( ! rew_last(mdoc, mdoc->last)) 925 return(0); 926 } else if ( ! nc && 0 == cnt) { 927 mdoc_argv_free(arg); 928 mandoc_msg(MANDOCERR_MACRO_EMPTY, 929 mdoc->parse, line, ppos, 930 mdoc_macronames[tok]); 931 } 932 933 if ( ! mdoc_macro(mdoc, ntok, line, la, pos, buf)) 934 return(0); 935 if ( ! nl) 936 return(1); 937 return(append_delims(mdoc, line, pos, buf)); 938 } 939 940 /* 941 * Non-quote-enclosed punctuation. Set up our scope, if 942 * a word; rewind the scope, if a delimiter; then append 943 * the word. 944 */ 945 946 d = ARGS_QWORD == ac ? DELIM_NONE : mdoc_isdelim(p); 947 948 if (DELIM_NONE != d) { 949 /* 950 * If we encounter closing punctuation, no word 951 * has been omitted, no scope is open, and we're 952 * allowed to have an empty element, then start 953 * a new scope. `Ar', `Fl', and `Li', only do 954 * this once per invocation. There may be more 955 * of these (all of them?). 956 */ 957 if (0 == cnt && (nc || MDOC_Li == tok) && 958 DELIM_CLOSE == d && ! scope) { 959 if ( ! mdoc_elem_alloc(mdoc, 960 line, ppos, tok, arg)) 961 return(0); 962 if (MDOC_Ar == tok || MDOC_Li == tok || 963 MDOC_Fl == tok) 964 cnt++; 965 scope = 1; 966 } 967 /* 968 * Close out our scope, if one is open, before 969 * any punctuation. 970 */ 971 if (scope && ! rew_elem(mdoc, tok)) 972 return(0); 973 scope = 0; 974 } else if ( ! scope) { 975 if ( ! mdoc_elem_alloc(mdoc, line, ppos, tok, arg)) 976 return(0); 977 scope = 1; 978 } 979 980 if (DELIM_NONE == d) 981 cnt++; 982 983 if ( ! dword(mdoc, line, la, p, d, 984 MDOC_JOIN & mdoc_macros[tok].flags)) 985 return(0); 986 987 /* 988 * `Fl' macros have their scope re-opened with each new 989 * word so that the `-' can be added to each one without 990 * having to parse out spaces. 991 */ 992 if (scope && MDOC_Fl == tok) { 993 if ( ! rew_elem(mdoc, tok)) 994 return(0); 995 scope = 0; 996 } 997 } 998 999 if (scope && ! rew_elem(mdoc, tok)) 1000 return(0); 1001 1002 /* 1003 * If no elements have been collected and we're allowed to have 1004 * empties (nc), open a scope and close it out. Otherwise, 1005 * raise a warning. 1006 */ 1007 1008 if (nc && 0 == cnt) { 1009 if ( ! mdoc_elem_alloc(mdoc, line, ppos, tok, arg)) 1010 return(0); 1011 if ( ! rew_last(mdoc, mdoc->last)) 1012 return(0); 1013 } else if ( ! nc && 0 == cnt) { 1014 mdoc_argv_free(arg); 1015 mandoc_msg(MANDOCERR_MACRO_EMPTY, mdoc->parse, 1016 line, ppos, mdoc_macronames[tok]); 1017 } 1018 1019 if ( ! nl) 1020 return(1); 1021 return(append_delims(mdoc, line, pos, buf)); 1022 } 1023 1024 static int 1025 blk_full(MACRO_PROT_ARGS) 1026 { 1027 int la, nl, nparsed; 1028 struct mdoc_arg *arg; 1029 struct mdoc_node *head; /* save of head macro */ 1030 struct mdoc_node *body; /* save of body macro */ 1031 struct mdoc_node *n; 1032 enum mdoc_type mtt; 1033 enum mdoct ntok; 1034 enum margserr ac, lac; 1035 enum margverr av; 1036 char *p; 1037 1038 nl = MDOC_NEWLINE & mdoc->flags; 1039 1040 /* Skip items outside lists. */ 1041 1042 if (tok == MDOC_It) { 1043 for (n = mdoc->last; n; n = n->parent) 1044 if (n->tok == MDOC_Bl) 1045 break; 1046 if (n == NULL) { 1047 mandoc_vmsg(MANDOCERR_IT_STRAY, mdoc->parse, 1048 line, ppos, "It %s", buf + *pos); 1049 if ( ! mdoc_elem_alloc(mdoc, line, ppos, 1050 MDOC_br, NULL)) 1051 return(0); 1052 return(rew_elem(mdoc, MDOC_br)); 1053 } 1054 } 1055 1056 /* Close out prior implicit scope. */ 1057 1058 if ( ! (MDOC_EXPLICIT & mdoc_macros[tok].flags)) { 1059 if ( ! rew_sub(MDOC_BODY, mdoc, tok, line, ppos)) 1060 return(0); 1061 if ( ! rew_sub(MDOC_BLOCK, mdoc, tok, line, ppos)) 1062 return(0); 1063 } 1064 1065 /* 1066 * This routine accommodates implicitly- and explicitly-scoped 1067 * macro openings. Implicit ones first close out prior scope 1068 * (seen above). Delay opening the head until necessary to 1069 * allow leading punctuation to print. Special consideration 1070 * for `It -column', which has phrase-part syntax instead of 1071 * regular child nodes. 1072 */ 1073 1074 for (arg = NULL;; ) { 1075 la = *pos; 1076 av = mdoc_argv(mdoc, line, tok, &arg, pos, buf); 1077 1078 if (ARGV_WORD == av) { 1079 *pos = la; 1080 break; 1081 } 1082 1083 if (ARGV_EOLN == av) 1084 break; 1085 if (ARGV_ARG == av) 1086 continue; 1087 1088 mdoc_argv_free(arg); 1089 return(0); 1090 } 1091 1092 if ( ! mdoc_block_alloc(mdoc, line, ppos, tok, arg)) 1093 return(0); 1094 1095 head = body = NULL; 1096 1097 /* 1098 * Exception: Heads of `It' macros in `-diag' lists are not 1099 * parsed, even though `It' macros in general are parsed. 1100 */ 1101 nparsed = MDOC_It == tok && 1102 MDOC_Bl == mdoc->last->parent->tok && 1103 LIST_diag == mdoc->last->parent->norm->Bl.type; 1104 1105 /* 1106 * The `Nd' macro has all arguments in its body: it's a hybrid 1107 * of block partial-explicit and full-implicit. Stupid. 1108 */ 1109 1110 if (MDOC_Nd == tok) { 1111 if ( ! mdoc_head_alloc(mdoc, line, ppos, tok)) 1112 return(0); 1113 head = mdoc->last; 1114 if ( ! rew_sub(MDOC_HEAD, mdoc, tok, line, ppos)) 1115 return(0); 1116 if ( ! mdoc_body_alloc(mdoc, line, ppos, tok)) 1117 return(0); 1118 body = mdoc->last; 1119 } 1120 1121 if (MDOC_Bk == tok) 1122 mdoc->flags |= MDOC_KEEP; 1123 1124 ac = ARGS_ERROR; 1125 1126 for ( ; ; ) { 1127 la = *pos; 1128 /* Initialise last-phrase-type with ARGS_PEND. */ 1129 lac = ARGS_ERROR == ac ? ARGS_PEND : ac; 1130 ac = mdoc_args(mdoc, line, pos, buf, tok, &p); 1131 1132 if (ARGS_PUNCT == ac) 1133 break; 1134 1135 if (ARGS_ERROR == ac) 1136 return(0); 1137 1138 if (ARGS_EOLN == ac) { 1139 if (ARGS_PPHRASE != lac && ARGS_PHRASE != lac) 1140 break; 1141 /* 1142 * This is necessary: if the last token on a 1143 * line is a `Ta' or tab, then we'll get 1144 * ARGS_EOLN, so we must be smart enough to 1145 * reopen our scope if the last parse was a 1146 * phrase or partial phrase. 1147 */ 1148 if ( ! rew_sub(MDOC_BODY, mdoc, tok, line, ppos)) 1149 return(0); 1150 if ( ! mdoc_body_alloc(mdoc, line, ppos, tok)) 1151 return(0); 1152 body = mdoc->last; 1153 break; 1154 } 1155 1156 /* 1157 * Emit leading punctuation (i.e., punctuation before 1158 * the MDOC_HEAD) for non-phrase types. 1159 */ 1160 1161 if (NULL == head && 1162 ARGS_PEND != ac && 1163 ARGS_PHRASE != ac && 1164 ARGS_PPHRASE != ac && 1165 ARGS_QWORD != ac && 1166 DELIM_OPEN == mdoc_isdelim(p)) { 1167 if ( ! dword(mdoc, line, la, p, DELIM_OPEN, 0)) 1168 return(0); 1169 continue; 1170 } 1171 1172 /* Open a head if one hasn't been opened. */ 1173 1174 if (NULL == head) { 1175 if ( ! mdoc_head_alloc(mdoc, line, ppos, tok)) 1176 return(0); 1177 head = mdoc->last; 1178 } 1179 1180 if (ARGS_PHRASE == ac || 1181 ARGS_PEND == ac || 1182 ARGS_PPHRASE == ac) { 1183 /* 1184 * If we haven't opened a body yet, rewind the 1185 * head; if we have, rewind that instead. 1186 */ 1187 1188 mtt = body ? MDOC_BODY : MDOC_HEAD; 1189 if ( ! rew_sub(mtt, mdoc, tok, line, ppos)) 1190 return(0); 1191 1192 /* Then allocate our body context. */ 1193 1194 if ( ! mdoc_body_alloc(mdoc, line, ppos, tok)) 1195 return(0); 1196 body = mdoc->last; 1197 1198 /* 1199 * Process phrases: set whether we're in a 1200 * partial-phrase (this effects line handling) 1201 * then call down into the phrase parser. 1202 */ 1203 1204 if (ARGS_PPHRASE == ac) 1205 mdoc->flags |= MDOC_PPHRASE; 1206 if (ARGS_PEND == ac && ARGS_PPHRASE == lac) 1207 mdoc->flags |= MDOC_PPHRASE; 1208 1209 if ( ! phrase(mdoc, line, la, buf)) 1210 return(0); 1211 1212 mdoc->flags &= ~MDOC_PPHRASE; 1213 continue; 1214 } 1215 1216 ntok = nparsed || ARGS_QWORD == ac ? 1217 MDOC_MAX : lookup(tok, p); 1218 1219 if (MDOC_MAX == ntok) { 1220 if ( ! dword(mdoc, line, la, p, DELIM_MAX, 1221 MDOC_JOIN & mdoc_macros[tok].flags)) 1222 return(0); 1223 continue; 1224 } 1225 1226 if ( ! mdoc_macro(mdoc, ntok, line, la, pos, buf)) 1227 return(0); 1228 break; 1229 } 1230 1231 if (NULL == head) { 1232 if ( ! mdoc_head_alloc(mdoc, line, ppos, tok)) 1233 return(0); 1234 head = mdoc->last; 1235 } 1236 1237 if (nl && ! append_delims(mdoc, line, pos, buf)) 1238 return(0); 1239 1240 /* If we've already opened our body, exit now. */ 1241 1242 if (NULL != body) 1243 goto out; 1244 1245 /* 1246 * If there is an open (i.e., unvalidated) sub-block requiring 1247 * explicit close-out, postpone switching the current block from 1248 * head to body until the rew_sub() call closing out that 1249 * sub-block. 1250 */ 1251 for (n = mdoc->last; n && n != head; n = n->parent) { 1252 if (MDOC_BLOCK == n->type && 1253 MDOC_EXPLICIT & mdoc_macros[n->tok].flags && 1254 ! (MDOC_VALID & n->flags)) { 1255 n->pending = head; 1256 return(1); 1257 } 1258 } 1259 1260 /* Close out scopes to remain in a consistent state. */ 1261 1262 if ( ! rew_sub(MDOC_HEAD, mdoc, tok, line, ppos)) 1263 return(0); 1264 if ( ! mdoc_body_alloc(mdoc, line, ppos, tok)) 1265 return(0); 1266 1267 out: 1268 if ( ! (MDOC_FREECOL & mdoc->flags)) 1269 return(1); 1270 1271 if ( ! rew_sub(MDOC_BODY, mdoc, tok, line, ppos)) 1272 return(0); 1273 if ( ! rew_sub(MDOC_BLOCK, mdoc, tok, line, ppos)) 1274 return(0); 1275 1276 mdoc->flags &= ~MDOC_FREECOL; 1277 return(1); 1278 } 1279 1280 static int 1281 blk_part_imp(MACRO_PROT_ARGS) 1282 { 1283 int la, nl; 1284 enum mdoct ntok; 1285 enum margserr ac; 1286 char *p; 1287 struct mdoc_node *blk; /* saved block context */ 1288 struct mdoc_node *body; /* saved body context */ 1289 struct mdoc_node *n; 1290 1291 nl = MDOC_NEWLINE & mdoc->flags; 1292 1293 /* 1294 * A macro that spans to the end of the line. This is generally 1295 * (but not necessarily) called as the first macro. The block 1296 * has a head as the immediate child, which is always empty, 1297 * followed by zero or more opening punctuation nodes, then the 1298 * body (which may be empty, depending on the macro), then zero 1299 * or more closing punctuation nodes. 1300 */ 1301 1302 if ( ! mdoc_block_alloc(mdoc, line, ppos, tok, NULL)) 1303 return(0); 1304 1305 blk = mdoc->last; 1306 1307 if ( ! mdoc_head_alloc(mdoc, line, ppos, tok)) 1308 return(0); 1309 if ( ! rew_sub(MDOC_HEAD, mdoc, tok, line, ppos)) 1310 return(0); 1311 1312 /* 1313 * Open the body scope "on-demand", that is, after we've 1314 * processed all our the leading delimiters (open parenthesis, 1315 * etc.). 1316 */ 1317 1318 for (body = NULL; ; ) { 1319 la = *pos; 1320 ac = mdoc_args(mdoc, line, pos, buf, tok, &p); 1321 1322 if (ARGS_ERROR == ac) 1323 return(0); 1324 if (ARGS_EOLN == ac) 1325 break; 1326 if (ARGS_PUNCT == ac) 1327 break; 1328 1329 if (NULL == body && ARGS_QWORD != ac && 1330 DELIM_OPEN == mdoc_isdelim(p)) { 1331 if ( ! dword(mdoc, line, la, p, DELIM_OPEN, 0)) 1332 return(0); 1333 continue; 1334 } 1335 1336 if (NULL == body) { 1337 if ( ! mdoc_body_alloc(mdoc, line, ppos, tok)) 1338 return(0); 1339 body = mdoc->last; 1340 } 1341 1342 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p); 1343 1344 if (MDOC_MAX == ntok) { 1345 if ( ! dword(mdoc, line, la, p, DELIM_MAX, 1346 MDOC_JOIN & mdoc_macros[tok].flags)) 1347 return(0); 1348 continue; 1349 } 1350 1351 if ( ! mdoc_macro(mdoc, ntok, line, la, pos, buf)) 1352 return(0); 1353 break; 1354 } 1355 1356 /* Clean-ups to leave in a consistent state. */ 1357 1358 if (NULL == body) { 1359 if ( ! mdoc_body_alloc(mdoc, line, ppos, tok)) 1360 return(0); 1361 body = mdoc->last; 1362 } 1363 1364 /* 1365 * If there is an open sub-block requiring explicit close-out, 1366 * postpone closing out the current block 1367 * until the rew_sub() call closing out the sub-block. 1368 */ 1369 for (n = mdoc->last; n && n != body && n != blk->parent; 1370 n = n->parent) { 1371 if (MDOC_BLOCK == n->type && 1372 MDOC_EXPLICIT & mdoc_macros[n->tok].flags && 1373 ! (MDOC_VALID & n->flags)) { 1374 make_pending(n, tok, mdoc, line, ppos); 1375 if ( ! mdoc_endbody_alloc(mdoc, line, ppos, 1376 tok, body, ENDBODY_NOSPACE)) 1377 return(0); 1378 return(1); 1379 } 1380 } 1381 assert(n == body); 1382 1383 if ( ! rew_sub(MDOC_BODY, mdoc, tok, line, ppos)) 1384 return(0); 1385 1386 /* Standard appending of delimiters. */ 1387 1388 if (nl && ! append_delims(mdoc, line, pos, buf)) 1389 return(0); 1390 1391 /* Rewind scope, if applicable. */ 1392 1393 if ( ! rew_sub(MDOC_BLOCK, mdoc, tok, line, ppos)) 1394 return(0); 1395 1396 /* Move trailing .Ns out of scope. */ 1397 1398 for (n = body->child; n && n->next; n = n->next) 1399 /* Do nothing. */ ; 1400 if (n && MDOC_Ns == n->tok) 1401 mdoc_node_relink(mdoc, n); 1402 1403 return(1); 1404 } 1405 1406 static int 1407 blk_part_exp(MACRO_PROT_ARGS) 1408 { 1409 int la, nl; 1410 enum margserr ac; 1411 struct mdoc_node *head; /* keep track of head */ 1412 struct mdoc_node *body; /* keep track of body */ 1413 char *p; 1414 enum mdoct ntok; 1415 1416 nl = MDOC_NEWLINE & mdoc->flags; 1417 1418 /* 1419 * The opening of an explicit macro having zero or more leading 1420 * punctuation nodes; a head with optional single element (the 1421 * case of `Eo'); and a body that may be empty. 1422 */ 1423 1424 if ( ! mdoc_block_alloc(mdoc, line, ppos, tok, NULL)) 1425 return(0); 1426 1427 for (head = body = NULL; ; ) { 1428 la = *pos; 1429 ac = mdoc_args(mdoc, line, pos, buf, tok, &p); 1430 1431 if (ARGS_ERROR == ac) 1432 return(0); 1433 if (ARGS_PUNCT == ac) 1434 break; 1435 if (ARGS_EOLN == ac) 1436 break; 1437 1438 /* Flush out leading punctuation. */ 1439 1440 if (NULL == head && ARGS_QWORD != ac && 1441 DELIM_OPEN == mdoc_isdelim(p)) { 1442 assert(NULL == body); 1443 if ( ! dword(mdoc, line, la, p, DELIM_OPEN, 0)) 1444 return(0); 1445 continue; 1446 } 1447 1448 if (NULL == head) { 1449 assert(NULL == body); 1450 if ( ! mdoc_head_alloc(mdoc, line, ppos, tok)) 1451 return(0); 1452 head = mdoc->last; 1453 } 1454 1455 /* 1456 * `Eo' gobbles any data into the head, but most other 1457 * macros just immediately close out and begin the body. 1458 */ 1459 1460 if (NULL == body) { 1461 assert(head); 1462 /* No check whether it's a macro! */ 1463 if (MDOC_Eo == tok) 1464 if ( ! dword(mdoc, line, la, p, DELIM_MAX, 0)) 1465 return(0); 1466 1467 if ( ! rew_sub(MDOC_HEAD, mdoc, tok, line, ppos)) 1468 return(0); 1469 if ( ! mdoc_body_alloc(mdoc, line, ppos, tok)) 1470 return(0); 1471 body = mdoc->last; 1472 1473 if (MDOC_Eo == tok) 1474 continue; 1475 } 1476 1477 assert(NULL != head && NULL != body); 1478 1479 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p); 1480 1481 if (MDOC_MAX == ntok) { 1482 if ( ! dword(mdoc, line, la, p, DELIM_MAX, 1483 MDOC_JOIN & mdoc_macros[tok].flags)) 1484 return(0); 1485 continue; 1486 } 1487 1488 if ( ! mdoc_macro(mdoc, ntok, line, la, pos, buf)) 1489 return(0); 1490 break; 1491 } 1492 1493 /* Clean-up to leave in a consistent state. */ 1494 1495 if (NULL == head) 1496 if ( ! mdoc_head_alloc(mdoc, line, ppos, tok)) 1497 return(0); 1498 1499 if (NULL == body) { 1500 if ( ! rew_sub(MDOC_HEAD, mdoc, tok, line, ppos)) 1501 return(0); 1502 if ( ! mdoc_body_alloc(mdoc, line, ppos, tok)) 1503 return(0); 1504 } 1505 1506 /* Standard appending of delimiters. */ 1507 1508 if ( ! nl) 1509 return(1); 1510 return(append_delims(mdoc, line, pos, buf)); 1511 } 1512 1513 static int 1514 in_line_argn(MACRO_PROT_ARGS) 1515 { 1516 int la, flushed, j, maxargs, nl; 1517 enum margserr ac; 1518 enum margverr av; 1519 struct mdoc_arg *arg; 1520 char *p; 1521 enum mdoct ntok; 1522 1523 nl = MDOC_NEWLINE & mdoc->flags; 1524 1525 /* 1526 * A line macro that has a fixed number of arguments (maxargs). 1527 * Only open the scope once the first non-leading-punctuation is 1528 * found (unless MDOC_IGNDELIM is noted, like in `Pf'), then 1529 * keep it open until the maximum number of arguments are 1530 * exhausted. 1531 */ 1532 1533 switch (tok) { 1534 case MDOC_Ap: 1535 /* FALLTHROUGH */ 1536 case MDOC_No: 1537 /* FALLTHROUGH */ 1538 case MDOC_Ns: 1539 /* FALLTHROUGH */ 1540 case MDOC_Ux: 1541 maxargs = 0; 1542 break; 1543 case MDOC_Bx: 1544 /* FALLTHROUGH */ 1545 case MDOC_Es: 1546 /* FALLTHROUGH */ 1547 case MDOC_Xr: 1548 maxargs = 2; 1549 break; 1550 default: 1551 maxargs = 1; 1552 break; 1553 } 1554 1555 for (arg = NULL; ; ) { 1556 la = *pos; 1557 av = mdoc_argv(mdoc, line, tok, &arg, pos, buf); 1558 1559 if (ARGV_WORD == av) { 1560 *pos = la; 1561 break; 1562 } 1563 1564 if (ARGV_EOLN == av) 1565 break; 1566 if (ARGV_ARG == av) 1567 continue; 1568 1569 mdoc_argv_free(arg); 1570 return(0); 1571 } 1572 1573 for (flushed = j = 0; ; ) { 1574 la = *pos; 1575 ac = mdoc_args(mdoc, line, pos, buf, tok, &p); 1576 1577 if (ARGS_ERROR == ac) 1578 return(0); 1579 if (ARGS_PUNCT == ac) 1580 break; 1581 if (ARGS_EOLN == ac) 1582 break; 1583 1584 if ( ! (MDOC_IGNDELIM & mdoc_macros[tok].flags) && 1585 ARGS_QWORD != ac && 0 == j && 1586 DELIM_OPEN == mdoc_isdelim(p)) { 1587 if ( ! dword(mdoc, line, la, p, DELIM_OPEN, 0)) 1588 return(0); 1589 continue; 1590 } else if (0 == j) 1591 if ( ! mdoc_elem_alloc(mdoc, line, ppos, tok, arg)) 1592 return(0); 1593 1594 if (j == maxargs && ! flushed) { 1595 if ( ! rew_elem(mdoc, tok)) 1596 return(0); 1597 flushed = 1; 1598 } 1599 1600 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p); 1601 1602 if (MDOC_MAX != ntok) { 1603 if ( ! flushed && ! rew_elem(mdoc, tok)) 1604 return(0); 1605 flushed = 1; 1606 if ( ! mdoc_macro(mdoc, ntok, line, la, pos, buf)) 1607 return(0); 1608 j++; 1609 break; 1610 } 1611 1612 if ( ! (MDOC_IGNDELIM & mdoc_macros[tok].flags) && 1613 ARGS_QWORD != ac && 1614 ! flushed && 1615 DELIM_NONE != mdoc_isdelim(p)) { 1616 if ( ! rew_elem(mdoc, tok)) 1617 return(0); 1618 flushed = 1; 1619 } 1620 1621 if ( ! dword(mdoc, line, la, p, DELIM_MAX, 1622 MDOC_JOIN & mdoc_macros[tok].flags)) 1623 return(0); 1624 j++; 1625 } 1626 1627 if (0 == j && ! mdoc_elem_alloc(mdoc, line, ppos, tok, arg)) 1628 return(0); 1629 1630 /* Close out in a consistent state. */ 1631 1632 if ( ! flushed && ! rew_elem(mdoc, tok)) 1633 return(0); 1634 if ( ! nl) 1635 return(1); 1636 return(append_delims(mdoc, line, pos, buf)); 1637 } 1638 1639 static int 1640 in_line_eoln(MACRO_PROT_ARGS) 1641 { 1642 int la; 1643 enum margserr ac; 1644 enum margverr av; 1645 struct mdoc_arg *arg; 1646 char *p; 1647 enum mdoct ntok; 1648 1649 assert( ! (MDOC_PARSED & mdoc_macros[tok].flags)); 1650 1651 if (tok == MDOC_Pp) 1652 rew_sub(MDOC_BLOCK, mdoc, MDOC_Nm, line, ppos); 1653 1654 /* Parse macro arguments. */ 1655 1656 for (arg = NULL; ; ) { 1657 la = *pos; 1658 av = mdoc_argv(mdoc, line, tok, &arg, pos, buf); 1659 1660 if (ARGV_WORD == av) { 1661 *pos = la; 1662 break; 1663 } 1664 if (ARGV_EOLN == av) 1665 break; 1666 if (ARGV_ARG == av) 1667 continue; 1668 1669 mdoc_argv_free(arg); 1670 return(0); 1671 } 1672 1673 /* Open element scope. */ 1674 1675 if ( ! mdoc_elem_alloc(mdoc, line, ppos, tok, arg)) 1676 return(0); 1677 1678 /* Parse argument terms. */ 1679 1680 for (;;) { 1681 la = *pos; 1682 ac = mdoc_args(mdoc, line, pos, buf, tok, &p); 1683 1684 if (ARGS_ERROR == ac) 1685 return(0); 1686 if (ARGS_EOLN == ac) 1687 break; 1688 1689 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup(tok, p); 1690 1691 if (MDOC_MAX == ntok) { 1692 if ( ! dword(mdoc, line, la, p, DELIM_MAX, 1693 MDOC_JOIN & mdoc_macros[tok].flags)) 1694 return(0); 1695 continue; 1696 } 1697 1698 if ( ! rew_elem(mdoc, tok)) 1699 return(0); 1700 return(mdoc_macro(mdoc, ntok, line, la, pos, buf)); 1701 } 1702 1703 /* Close out (no delimiters). */ 1704 1705 return(rew_elem(mdoc, tok)); 1706 } 1707 1708 static int 1709 ctx_synopsis(MACRO_PROT_ARGS) 1710 { 1711 int nl; 1712 1713 nl = MDOC_NEWLINE & mdoc->flags; 1714 1715 /* If we're not in the SYNOPSIS, go straight to in-line. */ 1716 if ( ! (MDOC_SYNOPSIS & mdoc->flags)) 1717 return(in_line(mdoc, tok, line, ppos, pos, buf)); 1718 1719 /* If we're a nested call, same place. */ 1720 if ( ! nl) 1721 return(in_line(mdoc, tok, line, ppos, pos, buf)); 1722 1723 /* 1724 * XXX: this will open a block scope; however, if later we end 1725 * up formatting the block scope, then child nodes will inherit 1726 * the formatting. Be careful. 1727 */ 1728 if (MDOC_Nm == tok) 1729 return(blk_full(mdoc, tok, line, ppos, pos, buf)); 1730 assert(MDOC_Vt == tok); 1731 return(blk_part_imp(mdoc, tok, line, ppos, pos, buf)); 1732 } 1733 1734 /* 1735 * Phrases occur within `Bl -column' entries, separated by `Ta' or tabs. 1736 * They're unusual because they're basically free-form text until a 1737 * macro is encountered. 1738 */ 1739 static int 1740 phrase(struct mdoc *mdoc, int line, int ppos, char *buf) 1741 { 1742 int la, pos; 1743 enum margserr ac; 1744 enum mdoct ntok; 1745 char *p; 1746 1747 for (pos = ppos; ; ) { 1748 la = pos; 1749 1750 ac = mdoc_zargs(mdoc, line, &pos, buf, &p); 1751 1752 if (ARGS_ERROR == ac) 1753 return(0); 1754 if (ARGS_EOLN == ac) 1755 break; 1756 1757 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup_raw(p); 1758 1759 if (MDOC_MAX == ntok) { 1760 if ( ! dword(mdoc, line, la, p, DELIM_MAX, 1)) 1761 return(0); 1762 continue; 1763 } 1764 1765 if ( ! mdoc_macro(mdoc, ntok, line, la, &pos, buf)) 1766 return(0); 1767 return(append_delims(mdoc, line, &pos, buf)); 1768 } 1769 1770 return(1); 1771 } 1772 1773 static int 1774 phrase_ta(MACRO_PROT_ARGS) 1775 { 1776 struct mdoc_node *n; 1777 int la; 1778 enum mdoct ntok; 1779 enum margserr ac; 1780 char *p; 1781 1782 /* Make sure we are in a column list or ignore this macro. */ 1783 n = mdoc->last; 1784 while (NULL != n && MDOC_Bl != n->tok) 1785 n = n->parent; 1786 if (NULL == n || LIST_column != n->norm->Bl.type) { 1787 mandoc_msg(MANDOCERR_TA_STRAY, mdoc->parse, 1788 line, ppos, "Ta"); 1789 return(1); 1790 } 1791 1792 /* Advance to the next column. */ 1793 if ( ! rew_sub(MDOC_BODY, mdoc, MDOC_It, line, ppos)) 1794 return(0); 1795 if ( ! mdoc_body_alloc(mdoc, line, ppos, MDOC_It)) 1796 return(0); 1797 1798 for (;;) { 1799 la = *pos; 1800 ac = mdoc_zargs(mdoc, line, pos, buf, &p); 1801 1802 if (ARGS_ERROR == ac) 1803 return(0); 1804 if (ARGS_EOLN == ac) 1805 break; 1806 1807 ntok = ARGS_QWORD == ac ? MDOC_MAX : lookup_raw(p); 1808 1809 if (MDOC_MAX == ntok) { 1810 if ( ! dword(mdoc, line, la, p, DELIM_MAX, 1811 MDOC_JOIN & mdoc_macros[tok].flags)) 1812 return(0); 1813 continue; 1814 } 1815 1816 if ( ! mdoc_macro(mdoc, ntok, line, la, pos, buf)) 1817 return(0); 1818 return(append_delims(mdoc, line, pos, buf)); 1819 } 1820 1821 return(1); 1822 } 1823