1205821Sedwin/* $Id: mdoc_argv.c,v 1.120 2019/07/11 17:06:17 schwarze Exp $ */ 2205872Sedwin/* 3205872Sedwin * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv> 4205821Sedwin * Copyright (c) 2012, 2014-2019 Ingo Schwarze <schwarze@openbsd.org> 5205821Sedwin * 6205821Sedwin * Permission to use, copy, modify, and distribute this software for any 7205821Sedwin * purpose with or without fee is hereby granted, provided that the above 8205821Sedwin * copyright notice and this permission notice appear in all copies. 9205821Sedwin * 10205821Sedwin * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES 11205821Sedwin * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF 12205821Sedwin * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR 13251647Sgrog * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES 14205821Sedwin * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN 15205821Sedwin * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF 16205821Sedwin * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. 17205821Sedwin */ 18205821Sedwin#include "config.h" 19205821Sedwin 20205821Sedwin#include <sys/types.h> 21205821Sedwin 22205821Sedwin#include <assert.h> 23205821Sedwin#include <stdlib.h> 24205821Sedwin#include <stdio.h> 25251647Sgrog#include <string.h> 26205821Sedwin 27205821Sedwin#include "mandoc_aux.h" 28205821Sedwin#include "mandoc.h" 29205821Sedwin#include "roff.h" 30205821Sedwin#include "mdoc.h" 31205821Sedwin#include "libmandoc.h" 32205821Sedwin#include "roff_int.h" 33205821Sedwin#include "libmdoc.h" 34205821Sedwin 35205821Sedwin#define MULTI_STEP 5 /* pre-allocate argument values */ 36205821Sedwin#define DELIMSZ 6 /* max possible size of a delimiter */ 37205821Sedwin 38205821Sedwinenum argsflag { 39205821Sedwin ARGSFL_NONE = 0, 40205821Sedwin ARGSFL_DELIM, /* handle delimiters of [[::delim::][ ]+]+ */ 41205821Sedwin ARGSFL_TABSEP /* handle tab/`Ta' separated phrases */ 42205821Sedwin}; 43205821Sedwin 44205821Sedwinenum argvflag { 45205821Sedwin ARGV_NONE, /* no args to flag (e.g., -split) */ 46205821Sedwin ARGV_SINGLE, /* one arg to flag (e.g., -file xxx) */ 47205821Sedwin ARGV_MULTI /* multiple args (e.g., -column xxx yyy) */ 48205821Sedwin}; 49205821Sedwin 50205821Sedwinstruct mdocarg { 51205821Sedwin enum argsflag flags; 52205821Sedwin const enum mdocargt *argvs; 53205821Sedwin}; 54205821Sedwin 55205821Sedwinstatic void argn_free(struct mdoc_arg *, int); 56205821Sedwinstatic enum margserr args(struct roff_man *, int, int *, 57205821Sedwin char *, enum argsflag, char **); 58205821Sedwinstatic int args_checkpunct(const char *, int); 59205821Sedwinstatic void argv_multi(struct roff_man *, int, 60205821Sedwin struct mdoc_argv *, int *, char *); 61205821Sedwinstatic void argv_single(struct roff_man *, int, 62205821Sedwin struct mdoc_argv *, int *, char *); 63205821Sedwin 64205821Sedwinstatic const enum argvflag argvflags[MDOC_ARG_MAX] = { 65205821Sedwin ARGV_NONE, /* MDOC_Split */ 66205821Sedwin ARGV_NONE, /* MDOC_Nosplit */ 67205821Sedwin ARGV_NONE, /* MDOC_Ragged */ 68205821Sedwin ARGV_NONE, /* MDOC_Unfilled */ 69205821Sedwin ARGV_NONE, /* MDOC_Literal */ 70205821Sedwin ARGV_SINGLE, /* MDOC_File */ 71205821Sedwin ARGV_SINGLE, /* MDOC_Offset */ 72205821Sedwin ARGV_NONE, /* MDOC_Bullet */ 73205821Sedwin ARGV_NONE, /* MDOC_Dash */ 74205821Sedwin ARGV_NONE, /* MDOC_Hyphen */ 75205821Sedwin ARGV_NONE, /* MDOC_Item */ 76205821Sedwin ARGV_NONE, /* MDOC_Enum */ 77205821Sedwin ARGV_NONE, /* MDOC_Tag */ 78205821Sedwin ARGV_NONE, /* MDOC_Diag */ 79205821Sedwin ARGV_NONE, /* MDOC_Hang */ 80205821Sedwin ARGV_NONE, /* MDOC_Ohang */ 81205821Sedwin ARGV_NONE, /* MDOC_Inset */ 82205821Sedwin ARGV_MULTI, /* MDOC_Column */ 83205821Sedwin ARGV_SINGLE, /* MDOC_Width */ 84205821Sedwin ARGV_NONE, /* MDOC_Compact */ 85205821Sedwin ARGV_NONE, /* MDOC_Std */ 86205821Sedwin ARGV_NONE, /* MDOC_Filled */ 87205821Sedwin ARGV_NONE, /* MDOC_Words */ 88205821Sedwin ARGV_NONE, /* MDOC_Emphasis */ 89205821Sedwin ARGV_NONE, /* MDOC_Symbolic */ 90205821Sedwin ARGV_NONE /* MDOC_Symbolic */ 91205821Sedwin}; 92205821Sedwin 93205821Sedwinstatic const enum mdocargt args_Ex[] = { 94205821Sedwin MDOC_Std, 95205821Sedwin MDOC_ARG_MAX 96205821Sedwin}; 97205821Sedwin 98205821Sedwinstatic const enum mdocargt args_An[] = { 99205821Sedwin MDOC_Split, 100205821Sedwin MDOC_Nosplit, 101205821Sedwin MDOC_ARG_MAX 102205821Sedwin}; 103205821Sedwin 104205821Sedwinstatic const enum mdocargt args_Bd[] = { 105205821Sedwin MDOC_Ragged, 106205821Sedwin MDOC_Unfilled, 107205821Sedwin MDOC_Filled, 108205821Sedwin MDOC_Literal, 109205821Sedwin MDOC_File, 110205821Sedwin MDOC_Offset, 111205821Sedwin MDOC_Compact, 112205821Sedwin MDOC_Centred, 113205821Sedwin MDOC_ARG_MAX 114205821Sedwin}; 115205821Sedwin 116205821Sedwinstatic const enum mdocargt args_Bf[] = { 117205821Sedwin MDOC_Emphasis, 118205821Sedwin MDOC_Literal, 119205821Sedwin MDOC_Symbolic, 120205821Sedwin MDOC_ARG_MAX 121205821Sedwin}; 122205821Sedwin 123205821Sedwinstatic const enum mdocargt args_Bk[] = { 124205821Sedwin MDOC_Words, 125205821Sedwin MDOC_ARG_MAX 126205821Sedwin}; 127205821Sedwin 128205821Sedwinstatic const enum mdocargt args_Bl[] = { 129205821Sedwin MDOC_Bullet, 130205821Sedwin MDOC_Dash, 131205821Sedwin MDOC_Hyphen, 132205821Sedwin MDOC_Item, 133205821Sedwin MDOC_Enum, 134205821Sedwin MDOC_Tag, 135205821Sedwin MDOC_Diag, 136205821Sedwin MDOC_Hang, 137205821Sedwin MDOC_Ohang, 138205821Sedwin MDOC_Inset, 139205821Sedwin MDOC_Column, 140205821Sedwin MDOC_Width, 141205821Sedwin MDOC_Offset, 142205821Sedwin MDOC_Compact, 143205821Sedwin MDOC_Nested, 144205821Sedwin MDOC_ARG_MAX 145205821Sedwin}; 146205821Sedwin 147205821Sedwinstatic const struct mdocarg mdocargs[MDOC_MAX - MDOC_Dd] = { 148205821Sedwin { ARGSFL_NONE, NULL }, /* Dd */ 149205821Sedwin { ARGSFL_NONE, NULL }, /* Dt */ 150205821Sedwin { ARGSFL_NONE, NULL }, /* Os */ 151205821Sedwin { ARGSFL_NONE, NULL }, /* Sh */ 152205821Sedwin { ARGSFL_NONE, NULL }, /* Ss */ 153205821Sedwin { ARGSFL_NONE, NULL }, /* Pp */ 154205821Sedwin { ARGSFL_DELIM, NULL }, /* D1 */ 155205821Sedwin { ARGSFL_DELIM, NULL }, /* Dl */ 156205821Sedwin { ARGSFL_NONE, args_Bd }, /* Bd */ 157205821Sedwin { ARGSFL_NONE, NULL }, /* Ed */ 158205821Sedwin { ARGSFL_NONE, args_Bl }, /* Bl */ 159205821Sedwin { ARGSFL_NONE, NULL }, /* El */ 160205821Sedwin { ARGSFL_NONE, NULL }, /* It */ 161205821Sedwin { ARGSFL_DELIM, NULL }, /* Ad */ 162205821Sedwin { ARGSFL_DELIM, args_An }, /* An */ 163205821Sedwin { ARGSFL_DELIM, NULL }, /* Ap */ 164205821Sedwin { ARGSFL_DELIM, NULL }, /* Ar */ 165205821Sedwin { ARGSFL_DELIM, NULL }, /* Cd */ 166205821Sedwin { ARGSFL_DELIM, NULL }, /* Cm */ 167205821Sedwin { ARGSFL_DELIM, NULL }, /* Dv */ 168205821Sedwin { ARGSFL_DELIM, NULL }, /* Er */ 169205821Sedwin { ARGSFL_DELIM, NULL }, /* Ev */ 170205821Sedwin { ARGSFL_NONE, args_Ex }, /* Ex */ 171205821Sedwin { ARGSFL_DELIM, NULL }, /* Fa */ 172205821Sedwin { ARGSFL_NONE, NULL }, /* Fd */ 173205821Sedwin { ARGSFL_DELIM, NULL }, /* Fl */ 174205821Sedwin { ARGSFL_DELIM, NULL }, /* Fn */ 175205821Sedwin { ARGSFL_DELIM, NULL }, /* Ft */ 176205821Sedwin { ARGSFL_DELIM, NULL }, /* Ic */ 177205821Sedwin { ARGSFL_DELIM, NULL }, /* In */ 178205821Sedwin { ARGSFL_DELIM, NULL }, /* Li */ 179205821Sedwin { ARGSFL_NONE, NULL }, /* Nd */ 180205821Sedwin { ARGSFL_DELIM, NULL }, /* Nm */ 181205821Sedwin { ARGSFL_DELIM, NULL }, /* Op */ 182205821Sedwin { ARGSFL_DELIM, NULL }, /* Ot */ 183205821Sedwin { ARGSFL_DELIM, NULL }, /* Pa */ 184205821Sedwin { ARGSFL_NONE, args_Ex }, /* Rv */ 185205821Sedwin { ARGSFL_DELIM, NULL }, /* St */ 186205821Sedwin { ARGSFL_DELIM, NULL }, /* Va */ 187205821Sedwin { ARGSFL_DELIM, NULL }, /* Vt */ 188205821Sedwin { ARGSFL_DELIM, NULL }, /* Xr */ 189205821Sedwin { ARGSFL_NONE, NULL }, /* %A */ 190205821Sedwin { ARGSFL_NONE, NULL }, /* %B */ 191205821Sedwin { ARGSFL_NONE, NULL }, /* %D */ 192205821Sedwin { ARGSFL_NONE, NULL }, /* %I */ 193205821Sedwin { ARGSFL_NONE, NULL }, /* %J */ 194205821Sedwin { ARGSFL_NONE, NULL }, /* %N */ 195205821Sedwin { ARGSFL_NONE, NULL }, /* %O */ 196205821Sedwin { ARGSFL_NONE, NULL }, /* %P */ 197205821Sedwin { ARGSFL_NONE, NULL }, /* %R */ 198205821Sedwin { ARGSFL_NONE, NULL }, /* %T */ 199205821Sedwin { ARGSFL_NONE, NULL }, /* %V */ 200205821Sedwin { ARGSFL_DELIM, NULL }, /* Ac */ 201205821Sedwin { ARGSFL_NONE, NULL }, /* Ao */ 202205821Sedwin { ARGSFL_DELIM, NULL }, /* Aq */ 203205821Sedwin { ARGSFL_DELIM, NULL }, /* At */ 204205821Sedwin { ARGSFL_DELIM, NULL }, /* Bc */ 205205821Sedwin { ARGSFL_NONE, args_Bf }, /* Bf */ 206205821Sedwin { ARGSFL_NONE, NULL }, /* Bo */ 207205821Sedwin { ARGSFL_DELIM, NULL }, /* Bq */ 208205821Sedwin { ARGSFL_DELIM, NULL }, /* Bsx */ 209205821Sedwin { ARGSFL_DELIM, NULL }, /* Bx */ 210205821Sedwin { ARGSFL_NONE, NULL }, /* Db */ 211205821Sedwin { ARGSFL_DELIM, NULL }, /* Dc */ 212205821Sedwin { ARGSFL_NONE, NULL }, /* Do */ 213205821Sedwin { ARGSFL_DELIM, NULL }, /* Dq */ 214205821Sedwin { ARGSFL_DELIM, NULL }, /* Ec */ 215205821Sedwin { ARGSFL_NONE, NULL }, /* Ef */ 216205821Sedwin { ARGSFL_DELIM, NULL }, /* Em */ 217205821Sedwin { ARGSFL_NONE, NULL }, /* Eo */ 218205821Sedwin { ARGSFL_DELIM, NULL }, /* Fx */ 219205821Sedwin { ARGSFL_DELIM, NULL }, /* Ms */ 220205821Sedwin { ARGSFL_DELIM, NULL }, /* No */ 221205821Sedwin { ARGSFL_DELIM, NULL }, /* Ns */ 222205821Sedwin { ARGSFL_DELIM, NULL }, /* Nx */ 223205821Sedwin { ARGSFL_DELIM, NULL }, /* Ox */ 224205821Sedwin { ARGSFL_DELIM, NULL }, /* Pc */ 225205821Sedwin { ARGSFL_DELIM, NULL }, /* Pf */ 226205821Sedwin { ARGSFL_NONE, NULL }, /* Po */ 227205821Sedwin { ARGSFL_DELIM, NULL }, /* Pq */ 228205821Sedwin { ARGSFL_DELIM, NULL }, /* Qc */ 229205821Sedwin { ARGSFL_DELIM, NULL }, /* Ql */ 230208829Sedwin { ARGSFL_NONE, NULL }, /* Qo */ 231205821Sedwin { ARGSFL_DELIM, NULL }, /* Qq */ 232205821Sedwin { ARGSFL_NONE, NULL }, /* Re */ 233205821Sedwin { ARGSFL_NONE, NULL }, /* Rs */ 234208829Sedwin { ARGSFL_DELIM, NULL }, /* Sc */ 235205821Sedwin { ARGSFL_NONE, NULL }, /* So */ 236205821Sedwin { ARGSFL_DELIM, NULL }, /* Sq */ 237205821Sedwin { ARGSFL_NONE, NULL }, /* Sm */ 238205821Sedwin { ARGSFL_DELIM, NULL }, /* Sx */ 239205821Sedwin { ARGSFL_DELIM, NULL }, /* Sy */ 240205821Sedwin { ARGSFL_DELIM, NULL }, /* Tn */ 241208829Sedwin { ARGSFL_DELIM, NULL }, /* Ux */ 242208829Sedwin { ARGSFL_DELIM, NULL }, /* Xc */ 243205821Sedwin { ARGSFL_NONE, NULL }, /* Xo */ 244205821Sedwin { ARGSFL_NONE, NULL }, /* Fo */ 245205821Sedwin { ARGSFL_DELIM, NULL }, /* Fc */ 246208829Sedwin { ARGSFL_NONE, NULL }, /* Oo */ 247205821Sedwin { ARGSFL_DELIM, NULL }, /* Oc */ 248205821Sedwin { ARGSFL_NONE, args_Bk }, /* Bk */ 249205821Sedwin { ARGSFL_NONE, NULL }, /* Ek */ 250205821Sedwin { ARGSFL_NONE, NULL }, /* Bt */ 251205821Sedwin { ARGSFL_NONE, NULL }, /* Hf */ 252205821Sedwin { ARGSFL_DELIM, NULL }, /* Fr */ 253205821Sedwin { ARGSFL_NONE, NULL }, /* Ud */ 254208829Sedwin { ARGSFL_DELIM, NULL }, /* Lb */ 255208829Sedwin { ARGSFL_NONE, NULL }, /* Lp */ 256208829Sedwin { ARGSFL_DELIM, NULL }, /* Lk */ 257205821Sedwin { ARGSFL_DELIM, NULL }, /* Mt */ 258205821Sedwin { ARGSFL_DELIM, NULL }, /* Brq */ 259205821Sedwin { ARGSFL_NONE, NULL }, /* Bro */ 260205821Sedwin { ARGSFL_DELIM, NULL }, /* Brc */ 261205821Sedwin { ARGSFL_NONE, NULL }, /* %C */ 262205821Sedwin { ARGSFL_NONE, NULL }, /* Es */ 263205821Sedwin { ARGSFL_DELIM, NULL }, /* En */ 264205821Sedwin { ARGSFL_DELIM, NULL }, /* Dx */ 265205821Sedwin { ARGSFL_NONE, NULL }, /* %Q */ 266205821Sedwin { ARGSFL_NONE, NULL }, /* %U */ 267205821Sedwin { ARGSFL_NONE, NULL }, /* Ta */ 268205821Sedwin}; 269208829Sedwin 270205821Sedwin 271205821Sedwin/* 272205821Sedwin * Parse flags and their arguments from the input line. 273208829Sedwin * These come in the form -flag [argument ...]. 274205821Sedwin * Some flags take no argument, some one, some multiple. 275205821Sedwin */ 276205821Sedwinvoid 277205821Sedwinmdoc_argv(struct roff_man *mdoc, int line, enum roff_tok tok, 278205821Sedwin struct mdoc_arg **reta, int *pos, char *buf) 279205821Sedwin{ 280208829Sedwin struct mdoc_argv tmpv; 281208829Sedwin struct mdoc_argv **retv; 282205821Sedwin const enum mdocargt *argtable; 283205821Sedwin char *argname; 284205821Sedwin int ipos, retc; 285208829Sedwin char savechar; 286205821Sedwin 287205821Sedwin *reta = NULL; 288205821Sedwin 289205821Sedwin /* Which flags does this macro support? */ 290205821Sedwin 291205821Sedwin assert(tok >= MDOC_Dd && tok < MDOC_MAX); 292205821Sedwin argtable = mdocargs[tok - MDOC_Dd].argvs; 293208829Sedwin if (argtable == NULL) 294208829Sedwin return; 295208829Sedwin 296205821Sedwin /* Loop over the flags on the input line. */ 297205821Sedwin 298205821Sedwin ipos = *pos; 299205821Sedwin while (buf[ipos] == '-') { 300205821Sedwin 301205821Sedwin /* Seek to the first unescaped space. */ 302205821Sedwin 303205821Sedwin for (argname = buf + ++ipos; buf[ipos] != '\0'; ipos++) 304205821Sedwin if (buf[ipos] == ' ' && buf[ipos - 1] != '\\') 305205821Sedwin break; 306205821Sedwin 307205821Sedwin /* 308205821Sedwin * We want to nil-terminate the word to look it up. 309205821Sedwin * But we may not have a flag, in which case we need 310205821Sedwin * to restore the line as-is. So keep around the 311205821Sedwin * stray byte, which we'll reset upon exiting. 312205821Sedwin */ 313205821Sedwin 314205821Sedwin if ((savechar = buf[ipos]) != '\0') 315205821Sedwin buf[ipos++] = '\0'; 316205821Sedwin 317205821Sedwin /* 318205821Sedwin * Now look up the word as a flag. Use temporary 319205821Sedwin * storage that we'll copy into the node's flags. 320205821Sedwin */ 321205821Sedwin 322205821Sedwin while ((tmpv.arg = *argtable++) != MDOC_ARG_MAX) 323205821Sedwin if ( ! strcmp(argname, mdoc_argnames[tmpv.arg])) 324205821Sedwin break; 325205821Sedwin 326205821Sedwin /* If it isn't a flag, restore the saved byte. */ 327205821Sedwin 328205821Sedwin if (tmpv.arg == MDOC_ARG_MAX) { 329205821Sedwin if (savechar != '\0') 330205821Sedwin buf[ipos - 1] = savechar; 331205821Sedwin break; 332205821Sedwin } 333205821Sedwin 334205821Sedwin /* Read to the next word (the first argument). */ 335205821Sedwin 336205821Sedwin while (buf[ipos] == ' ') 337205821Sedwin ipos++; 338205821Sedwin 339205821Sedwin /* Parse the arguments of the flag. */ 340205821Sedwin 341205821Sedwin tmpv.line = line; 342205821Sedwin tmpv.pos = *pos; 343205821Sedwin tmpv.sz = 0; 344205821Sedwin tmpv.value = NULL; 345205821Sedwin 346205821Sedwin switch (argvflags[tmpv.arg]) { 347205821Sedwin case ARGV_SINGLE: 348205821Sedwin argv_single(mdoc, line, &tmpv, &ipos, buf); 349205821Sedwin break; 350205821Sedwin case ARGV_MULTI: 351205821Sedwin argv_multi(mdoc, line, &tmpv, &ipos, buf); 352205821Sedwin break; 353205821Sedwin case ARGV_NONE: 354205821Sedwin break; 355205821Sedwin } 356205821Sedwin 357205821Sedwin /* Append to the return values. */ 358205821Sedwin 359205821Sedwin if (*reta == NULL) 360205821Sedwin *reta = mandoc_calloc(1, sizeof(**reta)); 361205821Sedwin 362205821Sedwin retc = ++(*reta)->argc; 363205821Sedwin retv = &(*reta)->argv; 364205821Sedwin *retv = mandoc_reallocarray(*retv, retc, sizeof(**retv)); 365205821Sedwin memcpy(*retv + retc - 1, &tmpv, sizeof(**retv)); 366205821Sedwin 367205821Sedwin /* Prepare for parsing the next flag. */ 368205821Sedwin 369205821Sedwin *pos = ipos; 370205821Sedwin argtable = mdocargs[tok - MDOC_Dd].argvs; 371205821Sedwin } 372205821Sedwin} 373205821Sedwin 374205821Sedwinvoid 375205821Sedwinmdoc_argv_free(struct mdoc_arg *p) 376251647Sgrog{ 377205821Sedwin int i; 378205821Sedwin 379205821Sedwin if (NULL == p) 380205821Sedwin return; 381205821Sedwin 382205821Sedwin if (p->refcnt) { 383205821Sedwin --(p->refcnt); 384205821Sedwin if (p->refcnt) 385205821Sedwin return; 386205821Sedwin } 387205821Sedwin assert(p->argc); 388205821Sedwin 389205821Sedwin for (i = (int)p->argc - 1; i >= 0; i--) 390205821Sedwin argn_free(p, i); 391205821Sedwin 392205821Sedwin free(p->argv); 393205821Sedwin free(p); 394205821Sedwin} 395205821Sedwin 396205821Sedwinstatic void 397205821Sedwinargn_free(struct mdoc_arg *p, int iarg) 398205821Sedwin{ 399205821Sedwin struct mdoc_argv *arg; 400205821Sedwin int j; 401205821Sedwin 402205821Sedwin arg = &p->argv[iarg]; 403205821Sedwin 404205821Sedwin if (arg->sz && arg->value) { 405205821Sedwin for (j = (int)arg->sz - 1; j >= 0; j--) 406205821Sedwin free(arg->value[j]); 407205821Sedwin free(arg->value); 408205821Sedwin } 409205821Sedwin 410205821Sedwin for (--p->argc; iarg < (int)p->argc; iarg++) 411205821Sedwin p->argv[iarg] = p->argv[iarg+1]; 412205821Sedwin} 413205821Sedwin 414205821Sedwinenum margserr 415205821Sedwinmdoc_args(struct roff_man *mdoc, int line, int *pos, 416205821Sedwin char *buf, enum roff_tok tok, char **v) 417205821Sedwin{ 418205821Sedwin struct roff_node *n; 419205821Sedwin enum argsflag fl; 420205821Sedwin 421205821Sedwin fl = tok == TOKEN_NONE ? ARGSFL_NONE : mdocargs[tok - MDOC_Dd].flags; 422205821Sedwin 423205821Sedwin /* 424205821Sedwin * We know that we're in an `It', so it's reasonable to expect 425205821Sedwin * us to be sitting in a `Bl'. Someday this may not be the case 426205821Sedwin * (if we allow random `It's sitting out there), so provide a 427205821Sedwin * safe fall-back into the default behaviour. 428205821Sedwin */ 429205821Sedwin 430205821Sedwin if (tok == MDOC_It) { 431205821Sedwin for (n = mdoc->last; n != NULL; n = n->parent) { 432205821Sedwin if (n->tok != MDOC_Bl) 433205821Sedwin continue; 434205821Sedwin if (n->norm->Bl.type == LIST_column) 435205821Sedwin fl = ARGSFL_TABSEP; 436205821Sedwin break; 437205821Sedwin } 438205821Sedwin } 439205821Sedwin 440205821Sedwin return args(mdoc, line, pos, buf, fl, v); 441205821Sedwin} 442205821Sedwin 443205821Sedwinstatic enum margserr 444205821Sedwinargs(struct roff_man *mdoc, int line, int *pos, 445205821Sedwin char *buf, enum argsflag fl, char **v) 446205821Sedwin{ 447205821Sedwin char *p; 448205821Sedwin char *v_local; 449 int pairs; 450 451 if (buf[*pos] == '\0') { 452 if (mdoc->flags & MDOC_PHRASELIT && 453 ! (mdoc->flags & MDOC_PHRASE)) { 454 mandoc_msg(MANDOCERR_ARG_QUOTE, line, *pos, NULL); 455 mdoc->flags &= ~MDOC_PHRASELIT; 456 } 457 mdoc->flags &= ~MDOC_PHRASEQL; 458 return ARGS_EOLN; 459 } 460 461 if (v == NULL) 462 v = &v_local; 463 *v = buf + *pos; 464 465 if (fl == ARGSFL_DELIM && args_checkpunct(buf, *pos)) 466 return ARGS_PUNCT; 467 468 /* 469 * Tabs in `It' lines in `Bl -column' can't be escaped. 470 * Phrases are reparsed for `Ta' and other macros later. 471 */ 472 473 if (fl == ARGSFL_TABSEP) { 474 if ((p = strchr(*v, '\t')) != NULL) { 475 476 /* 477 * Words right before and right after 478 * tab characters are not parsed, 479 * unless there is a blank in between. 480 */ 481 482 if (p > buf && p[-1] != ' ') 483 mdoc->flags |= MDOC_PHRASEQL; 484 if (p[1] != ' ') 485 mdoc->flags |= MDOC_PHRASEQN; 486 487 /* 488 * One or more blanks after a tab cause 489 * one leading blank in the next column. 490 * So skip all but one of them. 491 */ 492 493 *pos += (int)(p - *v) + 1; 494 while (buf[*pos] == ' ' && buf[*pos + 1] == ' ') 495 (*pos)++; 496 497 /* 498 * A tab at the end of an input line 499 * switches to the next column. 500 */ 501 502 if (buf[*pos] == '\0' || buf[*pos + 1] == '\0') 503 mdoc->flags |= MDOC_PHRASEQN; 504 } else { 505 p = strchr(*v, '\0'); 506 if (p[-1] == ' ') 507 mandoc_msg(MANDOCERR_SPACE_EOL, 508 line, *pos, NULL); 509 *pos += (int)(p - *v); 510 } 511 512 /* Skip any trailing blank characters. */ 513 while (p > *v && p[-1] == ' ' && 514 (p - 1 == *v || p[-2] != '\\')) 515 p--; 516 *p = '\0'; 517 518 return ARGS_PHRASE; 519 } 520 521 /* 522 * Process a quoted literal. A quote begins with a double-quote 523 * and ends with a double-quote NOT preceded by a double-quote. 524 * NUL-terminate the literal in place. 525 * Collapse pairs of quotes inside quoted literals. 526 * Whitespace is NOT involved in literal termination. 527 */ 528 529 if (mdoc->flags & MDOC_PHRASELIT || 530 (mdoc->flags & MDOC_PHRASE && buf[*pos] == '\"')) { 531 if ((mdoc->flags & MDOC_PHRASELIT) == 0) { 532 *v = &buf[++(*pos)]; 533 mdoc->flags |= MDOC_PHRASELIT; 534 } 535 pairs = 0; 536 for ( ; buf[*pos]; (*pos)++) { 537 /* Move following text left after quoted quotes. */ 538 if (pairs) 539 buf[*pos - pairs] = buf[*pos]; 540 if ('\"' != buf[*pos]) 541 continue; 542 /* Unquoted quotes end quoted args. */ 543 if ('\"' != buf[*pos + 1]) 544 break; 545 /* Quoted quotes collapse. */ 546 pairs++; 547 (*pos)++; 548 } 549 if (pairs) 550 buf[*pos - pairs] = '\0'; 551 552 if (buf[*pos] == '\0') { 553 if ( ! (mdoc->flags & MDOC_PHRASE)) 554 mandoc_msg(MANDOCERR_ARG_QUOTE, 555 line, *pos, NULL); 556 return ARGS_WORD; 557 } 558 559 mdoc->flags &= ~MDOC_PHRASELIT; 560 buf[(*pos)++] = '\0'; 561 562 if ('\0' == buf[*pos]) 563 return ARGS_WORD; 564 565 while (' ' == buf[*pos]) 566 (*pos)++; 567 568 if ('\0' == buf[*pos]) 569 mandoc_msg(MANDOCERR_SPACE_EOL, line, *pos, NULL); 570 571 return ARGS_WORD; 572 } 573 574 p = &buf[*pos]; 575 *v = roff_getarg(mdoc->roff, &p, line, pos); 576 if (v == &v_local) 577 free(*v); 578 579 /* 580 * After parsing the last word in this phrase, 581 * tell lookup() whether or not to interpret it. 582 */ 583 584 if (*p == '\0' && mdoc->flags & MDOC_PHRASEQL) { 585 mdoc->flags &= ~MDOC_PHRASEQL; 586 mdoc->flags |= MDOC_PHRASEQF; 587 } 588 return ARGS_ALLOC; 589} 590 591/* 592 * Check if the string consists only of space-separated closing 593 * delimiters. This is a bit of a dance: the first must be a close 594 * delimiter, but it may be followed by middle delimiters. Arbitrary 595 * whitespace may separate these tokens. 596 */ 597static int 598args_checkpunct(const char *buf, int i) 599{ 600 int j; 601 char dbuf[DELIMSZ]; 602 enum mdelim d; 603 604 /* First token must be a close-delimiter. */ 605 606 for (j = 0; buf[i] && ' ' != buf[i] && j < DELIMSZ; j++, i++) 607 dbuf[j] = buf[i]; 608 609 if (DELIMSZ == j) 610 return 0; 611 612 dbuf[j] = '\0'; 613 if (DELIM_CLOSE != mdoc_isdelim(dbuf)) 614 return 0; 615 616 while (' ' == buf[i]) 617 i++; 618 619 /* Remaining must NOT be open/none. */ 620 621 while (buf[i]) { 622 j = 0; 623 while (buf[i] && ' ' != buf[i] && j < DELIMSZ) 624 dbuf[j++] = buf[i++]; 625 626 if (DELIMSZ == j) 627 return 0; 628 629 dbuf[j] = '\0'; 630 d = mdoc_isdelim(dbuf); 631 if (DELIM_NONE == d || DELIM_OPEN == d) 632 return 0; 633 634 while (' ' == buf[i]) 635 i++; 636 } 637 638 return '\0' == buf[i]; 639} 640 641static void 642argv_multi(struct roff_man *mdoc, int line, 643 struct mdoc_argv *v, int *pos, char *buf) 644{ 645 enum margserr ac; 646 char *p; 647 648 for (v->sz = 0; ; v->sz++) { 649 if (buf[*pos] == '-') 650 break; 651 ac = args(mdoc, line, pos, buf, ARGSFL_NONE, &p); 652 if (ac == ARGS_EOLN) 653 break; 654 655 if (v->sz % MULTI_STEP == 0) 656 v->value = mandoc_reallocarray(v->value, 657 v->sz + MULTI_STEP, sizeof(char *)); 658 659 if (ac != ARGS_ALLOC) 660 p = mandoc_strdup(p); 661 v->value[(int)v->sz] = p; 662 } 663} 664 665static void 666argv_single(struct roff_man *mdoc, int line, 667 struct mdoc_argv *v, int *pos, char *buf) 668{ 669 enum margserr ac; 670 char *p; 671 672 ac = args(mdoc, line, pos, buf, ARGSFL_NONE, &p); 673 if (ac == ARGS_EOLN) 674 return; 675 676 if (ac != ARGS_ALLOC) 677 p = mandoc_strdup(p); 678 679 v->sz = 1; 680 v->value = mandoc_malloc(sizeof(char *)); 681 v->value[0] = p; 682} 683