1294113Sbapt/* $Id: mdoc_argv.c,v 1.107 2015/10/17 00:21:07 schwarze Exp $ */ 2241675Suqs/* 3241675Suqs * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv> 4294113Sbapt * Copyright (c) 2012, 2014, 2015 Ingo Schwarze <schwarze@openbsd.org> 5241675Suqs * 6241675Suqs * Permission to use, copy, modify, and distribute this software for any 7241675Suqs * purpose with or without fee is hereby granted, provided that the above 8241675Suqs * copyright notice and this permission notice appear in all copies. 9241675Suqs * 10294113Sbapt * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES 11241675Suqs * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF 12294113Sbapt * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR 13241675Suqs * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES 14241675Suqs * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN 15241675Suqs * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF 16241675Suqs * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. 17241675Suqs */ 18241675Suqs#include "config.h" 19241675Suqs 20241675Suqs#include <sys/types.h> 21241675Suqs 22241675Suqs#include <assert.h> 23241675Suqs#include <stdlib.h> 24241675Suqs#include <stdio.h> 25241675Suqs#include <string.h> 26241675Suqs 27294113Sbapt#include "mandoc_aux.h" 28294113Sbapt#include "mandoc.h" 29294113Sbapt#include "roff.h" 30241675Suqs#include "mdoc.h" 31294113Sbapt#include "libmandoc.h" 32241675Suqs#include "libmdoc.h" 33241675Suqs 34241675Suqs#define MULTI_STEP 5 /* pre-allocate argument values */ 35274880Sbapt#define DELIMSZ 6 /* max possible size of a delimiter */ 36241675Suqs 37241675Suqsenum argsflag { 38241675Suqs ARGSFL_NONE = 0, 39241675Suqs ARGSFL_DELIM, /* handle delimiters of [[::delim::][ ]+]+ */ 40241675Suqs ARGSFL_TABSEP /* handle tab/`Ta' separated phrases */ 41241675Suqs}; 42241675Suqs 43241675Suqsenum argvflag { 44241675Suqs ARGV_NONE, /* no args to flag (e.g., -split) */ 45241675Suqs ARGV_SINGLE, /* one arg to flag (e.g., -file xxx) */ 46261344Suqs ARGV_MULTI /* multiple args (e.g., -column xxx yyy) */ 47241675Suqs}; 48241675Suqs 49241675Suqsstruct mdocarg { 50241675Suqs enum argsflag flags; 51241675Suqs const enum mdocargt *argvs; 52241675Suqs}; 53241675Suqs 54241675Suqsstatic void argn_free(struct mdoc_arg *, int); 55294113Sbaptstatic enum margserr args(struct roff_man *, int, int *, 56241675Suqs char *, enum argsflag, char **); 57241675Suqsstatic int args_checkpunct(const char *, int); 58294113Sbaptstatic void argv_multi(struct roff_man *, int, 59241675Suqs struct mdoc_argv *, int *, char *); 60294113Sbaptstatic void argv_single(struct roff_man *, int, 61241675Suqs struct mdoc_argv *, int *, char *); 62241675Suqs 63241675Suqsstatic const enum argvflag argvflags[MDOC_ARG_MAX] = { 64241675Suqs ARGV_NONE, /* MDOC_Split */ 65241675Suqs ARGV_NONE, /* MDOC_Nosplit */ 66241675Suqs ARGV_NONE, /* MDOC_Ragged */ 67241675Suqs ARGV_NONE, /* MDOC_Unfilled */ 68241675Suqs ARGV_NONE, /* MDOC_Literal */ 69241675Suqs ARGV_SINGLE, /* MDOC_File */ 70261344Suqs ARGV_SINGLE, /* MDOC_Offset */ 71241675Suqs ARGV_NONE, /* MDOC_Bullet */ 72241675Suqs ARGV_NONE, /* MDOC_Dash */ 73241675Suqs ARGV_NONE, /* MDOC_Hyphen */ 74241675Suqs ARGV_NONE, /* MDOC_Item */ 75241675Suqs ARGV_NONE, /* MDOC_Enum */ 76241675Suqs ARGV_NONE, /* MDOC_Tag */ 77241675Suqs ARGV_NONE, /* MDOC_Diag */ 78241675Suqs ARGV_NONE, /* MDOC_Hang */ 79241675Suqs ARGV_NONE, /* MDOC_Ohang */ 80241675Suqs ARGV_NONE, /* MDOC_Inset */ 81241675Suqs ARGV_MULTI, /* MDOC_Column */ 82261344Suqs ARGV_SINGLE, /* MDOC_Width */ 83241675Suqs ARGV_NONE, /* MDOC_Compact */ 84241675Suqs ARGV_NONE, /* MDOC_Std */ 85241675Suqs ARGV_NONE, /* MDOC_Filled */ 86241675Suqs ARGV_NONE, /* MDOC_Words */ 87241675Suqs ARGV_NONE, /* MDOC_Emphasis */ 88241675Suqs ARGV_NONE, /* MDOC_Symbolic */ 89241675Suqs ARGV_NONE /* MDOC_Symbolic */ 90241675Suqs}; 91241675Suqs 92241675Suqsstatic const enum mdocargt args_Ex[] = { 93241675Suqs MDOC_Std, 94241675Suqs MDOC_ARG_MAX 95241675Suqs}; 96241675Suqs 97241675Suqsstatic const enum mdocargt args_An[] = { 98241675Suqs MDOC_Split, 99241675Suqs MDOC_Nosplit, 100241675Suqs MDOC_ARG_MAX 101241675Suqs}; 102241675Suqs 103241675Suqsstatic const enum mdocargt args_Bd[] = { 104241675Suqs MDOC_Ragged, 105241675Suqs MDOC_Unfilled, 106241675Suqs MDOC_Filled, 107241675Suqs MDOC_Literal, 108241675Suqs MDOC_File, 109241675Suqs MDOC_Offset, 110241675Suqs MDOC_Compact, 111241675Suqs MDOC_Centred, 112241675Suqs MDOC_ARG_MAX 113241675Suqs}; 114241675Suqs 115241675Suqsstatic const enum mdocargt args_Bf[] = { 116241675Suqs MDOC_Emphasis, 117241675Suqs MDOC_Literal, 118241675Suqs MDOC_Symbolic, 119241675Suqs MDOC_ARG_MAX 120241675Suqs}; 121241675Suqs 122241675Suqsstatic const enum mdocargt args_Bk[] = { 123241675Suqs MDOC_Words, 124241675Suqs MDOC_ARG_MAX 125241675Suqs}; 126241675Suqs 127241675Suqsstatic const enum mdocargt args_Bl[] = { 128241675Suqs MDOC_Bullet, 129241675Suqs MDOC_Dash, 130241675Suqs MDOC_Hyphen, 131241675Suqs MDOC_Item, 132241675Suqs MDOC_Enum, 133241675Suqs MDOC_Tag, 134241675Suqs MDOC_Diag, 135241675Suqs MDOC_Hang, 136241675Suqs MDOC_Ohang, 137241675Suqs MDOC_Inset, 138241675Suqs MDOC_Column, 139241675Suqs MDOC_Width, 140241675Suqs MDOC_Offset, 141241675Suqs MDOC_Compact, 142241675Suqs MDOC_Nested, 143241675Suqs MDOC_ARG_MAX 144241675Suqs}; 145241675Suqs 146241675Suqsstatic const struct mdocarg mdocargs[MDOC_MAX] = { 147261344Suqs { ARGSFL_DELIM, NULL }, /* Ap */ 148241675Suqs { ARGSFL_NONE, NULL }, /* Dd */ 149241675Suqs { ARGSFL_NONE, NULL }, /* Dt */ 150241675Suqs { ARGSFL_NONE, NULL }, /* Os */ 151241675Suqs { ARGSFL_NONE, NULL }, /* Sh */ 152274880Sbapt { ARGSFL_NONE, NULL }, /* Ss */ 153274880Sbapt { ARGSFL_NONE, NULL }, /* Pp */ 154241675Suqs { ARGSFL_DELIM, NULL }, /* D1 */ 155241675Suqs { ARGSFL_DELIM, NULL }, /* Dl */ 156241675Suqs { ARGSFL_NONE, args_Bd }, /* Bd */ 157241675Suqs { ARGSFL_NONE, NULL }, /* Ed */ 158241675Suqs { ARGSFL_NONE, args_Bl }, /* Bl */ 159241675Suqs { ARGSFL_NONE, NULL }, /* El */ 160241675Suqs { ARGSFL_NONE, NULL }, /* It */ 161274880Sbapt { ARGSFL_DELIM, NULL }, /* Ad */ 162241675Suqs { ARGSFL_DELIM, args_An }, /* An */ 163241675Suqs { ARGSFL_DELIM, NULL }, /* Ar */ 164261344Suqs { ARGSFL_DELIM, NULL }, /* Cd */ 165241675Suqs { ARGSFL_DELIM, NULL }, /* Cm */ 166274880Sbapt { ARGSFL_DELIM, NULL }, /* Dv */ 167274880Sbapt { ARGSFL_DELIM, NULL }, /* Er */ 168274880Sbapt { ARGSFL_DELIM, NULL }, /* Ev */ 169241675Suqs { ARGSFL_NONE, args_Ex }, /* Ex */ 170274880Sbapt { ARGSFL_DELIM, NULL }, /* Fa */ 171274880Sbapt { ARGSFL_NONE, NULL }, /* Fd */ 172241675Suqs { ARGSFL_DELIM, NULL }, /* Fl */ 173274880Sbapt { ARGSFL_DELIM, NULL }, /* Fn */ 174274880Sbapt { ARGSFL_DELIM, NULL }, /* Ft */ 175274880Sbapt { ARGSFL_DELIM, NULL }, /* Ic */ 176274880Sbapt { ARGSFL_DELIM, NULL }, /* In */ 177241675Suqs { ARGSFL_DELIM, NULL }, /* Li */ 178274880Sbapt { ARGSFL_NONE, NULL }, /* Nd */ 179274880Sbapt { ARGSFL_DELIM, NULL }, /* Nm */ 180241675Suqs { ARGSFL_DELIM, NULL }, /* Op */ 181274880Sbapt { ARGSFL_DELIM, NULL }, /* Ot */ 182241675Suqs { ARGSFL_DELIM, NULL }, /* Pa */ 183241675Suqs { ARGSFL_NONE, args_Ex }, /* Rv */ 184274880Sbapt { ARGSFL_DELIM, NULL }, /* St */ 185241675Suqs { ARGSFL_DELIM, NULL }, /* Va */ 186274880Sbapt { ARGSFL_DELIM, NULL }, /* Vt */ 187241675Suqs { ARGSFL_DELIM, NULL }, /* Xr */ 188241675Suqs { ARGSFL_NONE, NULL }, /* %A */ 189241675Suqs { ARGSFL_NONE, NULL }, /* %B */ 190241675Suqs { ARGSFL_NONE, NULL }, /* %D */ 191241675Suqs { ARGSFL_NONE, NULL }, /* %I */ 192241675Suqs { ARGSFL_NONE, NULL }, /* %J */ 193241675Suqs { ARGSFL_NONE, NULL }, /* %N */ 194241675Suqs { ARGSFL_NONE, NULL }, /* %O */ 195241675Suqs { ARGSFL_NONE, NULL }, /* %P */ 196241675Suqs { ARGSFL_NONE, NULL }, /* %R */ 197241675Suqs { ARGSFL_NONE, NULL }, /* %T */ 198241675Suqs { ARGSFL_NONE, NULL }, /* %V */ 199241675Suqs { ARGSFL_DELIM, NULL }, /* Ac */ 200241675Suqs { ARGSFL_NONE, NULL }, /* Ao */ 201241675Suqs { ARGSFL_DELIM, NULL }, /* Aq */ 202241675Suqs { ARGSFL_DELIM, NULL }, /* At */ 203241675Suqs { ARGSFL_DELIM, NULL }, /* Bc */ 204274880Sbapt { ARGSFL_NONE, args_Bf }, /* Bf */ 205241675Suqs { ARGSFL_NONE, NULL }, /* Bo */ 206241675Suqs { ARGSFL_DELIM, NULL }, /* Bq */ 207241675Suqs { ARGSFL_DELIM, NULL }, /* Bsx */ 208241675Suqs { ARGSFL_DELIM, NULL }, /* Bx */ 209241675Suqs { ARGSFL_NONE, NULL }, /* Db */ 210241675Suqs { ARGSFL_DELIM, NULL }, /* Dc */ 211241675Suqs { ARGSFL_NONE, NULL }, /* Do */ 212241675Suqs { ARGSFL_DELIM, NULL }, /* Dq */ 213241675Suqs { ARGSFL_DELIM, NULL }, /* Ec */ 214241675Suqs { ARGSFL_NONE, NULL }, /* Ef */ 215274880Sbapt { ARGSFL_DELIM, NULL }, /* Em */ 216241675Suqs { ARGSFL_NONE, NULL }, /* Eo */ 217241675Suqs { ARGSFL_DELIM, NULL }, /* Fx */ 218241675Suqs { ARGSFL_DELIM, NULL }, /* Ms */ 219241675Suqs { ARGSFL_DELIM, NULL }, /* No */ 220241675Suqs { ARGSFL_DELIM, NULL }, /* Ns */ 221241675Suqs { ARGSFL_DELIM, NULL }, /* Nx */ 222241675Suqs { ARGSFL_DELIM, NULL }, /* Ox */ 223241675Suqs { ARGSFL_DELIM, NULL }, /* Pc */ 224241675Suqs { ARGSFL_DELIM, NULL }, /* Pf */ 225241675Suqs { ARGSFL_NONE, NULL }, /* Po */ 226241675Suqs { ARGSFL_DELIM, NULL }, /* Pq */ 227241675Suqs { ARGSFL_DELIM, NULL }, /* Qc */ 228241675Suqs { ARGSFL_DELIM, NULL }, /* Ql */ 229241675Suqs { ARGSFL_NONE, NULL }, /* Qo */ 230241675Suqs { ARGSFL_DELIM, NULL }, /* Qq */ 231241675Suqs { ARGSFL_NONE, NULL }, /* Re */ 232241675Suqs { ARGSFL_NONE, NULL }, /* Rs */ 233241675Suqs { ARGSFL_DELIM, NULL }, /* Sc */ 234241675Suqs { ARGSFL_NONE, NULL }, /* So */ 235241675Suqs { ARGSFL_DELIM, NULL }, /* Sq */ 236241675Suqs { ARGSFL_NONE, NULL }, /* Sm */ 237241675Suqs { ARGSFL_DELIM, NULL }, /* Sx */ 238241675Suqs { ARGSFL_DELIM, NULL }, /* Sy */ 239241675Suqs { ARGSFL_DELIM, NULL }, /* Tn */ 240241675Suqs { ARGSFL_DELIM, NULL }, /* Ux */ 241241675Suqs { ARGSFL_DELIM, NULL }, /* Xc */ 242241675Suqs { ARGSFL_NONE, NULL }, /* Xo */ 243274880Sbapt { ARGSFL_NONE, NULL }, /* Fo */ 244274880Sbapt { ARGSFL_DELIM, NULL }, /* Fc */ 245241675Suqs { ARGSFL_NONE, NULL }, /* Oo */ 246241675Suqs { ARGSFL_DELIM, NULL }, /* Oc */ 247241675Suqs { ARGSFL_NONE, args_Bk }, /* Bk */ 248241675Suqs { ARGSFL_NONE, NULL }, /* Ek */ 249241675Suqs { ARGSFL_NONE, NULL }, /* Bt */ 250241675Suqs { ARGSFL_NONE, NULL }, /* Hf */ 251274880Sbapt { ARGSFL_DELIM, NULL }, /* Fr */ 252241675Suqs { ARGSFL_NONE, NULL }, /* Ud */ 253261344Suqs { ARGSFL_DELIM, NULL }, /* Lb */ 254241675Suqs { ARGSFL_NONE, NULL }, /* Lp */ 255241675Suqs { ARGSFL_DELIM, NULL }, /* Lk */ 256241675Suqs { ARGSFL_DELIM, NULL }, /* Mt */ 257241675Suqs { ARGSFL_DELIM, NULL }, /* Brq */ 258241675Suqs { ARGSFL_NONE, NULL }, /* Bro */ 259241675Suqs { ARGSFL_DELIM, NULL }, /* Brc */ 260241675Suqs { ARGSFL_NONE, NULL }, /* %C */ 261241675Suqs { ARGSFL_NONE, NULL }, /* Es */ 262274880Sbapt { ARGSFL_DELIM, NULL }, /* En */ 263261344Suqs { ARGSFL_DELIM, NULL }, /* Dx */ 264241675Suqs { ARGSFL_NONE, NULL }, /* %Q */ 265241675Suqs { ARGSFL_NONE, NULL }, /* br */ 266241675Suqs { ARGSFL_NONE, NULL }, /* sp */ 267241675Suqs { ARGSFL_NONE, NULL }, /* %U */ 268241675Suqs { ARGSFL_NONE, NULL }, /* Ta */ 269274880Sbapt { ARGSFL_NONE, NULL }, /* ll */ 270241675Suqs}; 271241675Suqs 272241675Suqs 273241675Suqs/* 274275432Sbapt * Parse flags and their arguments from the input line. 275275432Sbapt * These come in the form -flag [argument ...]. 276275432Sbapt * Some flags take no argument, some one, some multiple. 277241675Suqs */ 278275432Sbaptvoid 279294113Sbaptmdoc_argv(struct roff_man *mdoc, int line, int tok, 280275432Sbapt struct mdoc_arg **reta, int *pos, char *buf) 281241675Suqs{ 282275432Sbapt struct mdoc_argv tmpv; 283275432Sbapt struct mdoc_argv **retv; 284275432Sbapt const enum mdocargt *argtable; 285275432Sbapt char *argname; 286275432Sbapt int ipos, retc; 287275432Sbapt char savechar; 288241675Suqs 289275432Sbapt *reta = NULL; 290241675Suqs 291275432Sbapt /* Which flags does this macro support? */ 292241675Suqs 293275432Sbapt argtable = mdocargs[tok].argvs; 294275432Sbapt if (argtable == NULL) 295275432Sbapt return; 296241675Suqs 297275432Sbapt /* Loop over the flags on the input line. */ 298241675Suqs 299275432Sbapt ipos = *pos; 300275432Sbapt while (buf[ipos] == '-') { 301241675Suqs 302275432Sbapt /* Seek to the first unescaped space. */ 303241675Suqs 304275432Sbapt for (argname = buf + ++ipos; buf[ipos] != '\0'; ipos++) 305275432Sbapt if (buf[ipos] == ' ' && buf[ipos - 1] != '\\') 306275432Sbapt break; 307241675Suqs 308275432Sbapt /* 309275432Sbapt * We want to nil-terminate the word to look it up. 310275432Sbapt * But we may not have a flag, in which case we need 311275432Sbapt * to restore the line as-is. So keep around the 312275432Sbapt * stray byte, which we'll reset upon exiting. 313275432Sbapt */ 314241675Suqs 315275432Sbapt if ((savechar = buf[ipos]) != '\0') 316275432Sbapt buf[ipos++] = '\0'; 317241675Suqs 318275432Sbapt /* 319275432Sbapt * Now look up the word as a flag. Use temporary 320275432Sbapt * storage that we'll copy into the node's flags. 321275432Sbapt */ 322241675Suqs 323275432Sbapt while ((tmpv.arg = *argtable++) != MDOC_ARG_MAX) 324275432Sbapt if ( ! strcmp(argname, mdoc_argnames[tmpv.arg])) 325275432Sbapt break; 326275432Sbapt 327275432Sbapt /* If it isn't a flag, restore the saved byte. */ 328275432Sbapt 329275432Sbapt if (tmpv.arg == MDOC_ARG_MAX) { 330275432Sbapt if (savechar != '\0') 331275432Sbapt buf[ipos - 1] = savechar; 332241675Suqs break; 333275432Sbapt } 334241675Suqs 335275432Sbapt /* Read to the next word (the first argument). */ 336241675Suqs 337275432Sbapt while (buf[ipos] == ' ') 338275432Sbapt ipos++; 339241675Suqs 340275432Sbapt /* Parse the arguments of the flag. */ 341241675Suqs 342275432Sbapt tmpv.line = line; 343279527Sbapt tmpv.pos = *pos; 344275432Sbapt tmpv.sz = 0; 345275432Sbapt tmpv.value = NULL; 346241675Suqs 347275432Sbapt switch (argvflags[tmpv.arg]) { 348275432Sbapt case ARGV_SINGLE: 349275432Sbapt argv_single(mdoc, line, &tmpv, &ipos, buf); 350275432Sbapt break; 351275432Sbapt case ARGV_MULTI: 352275432Sbapt argv_multi(mdoc, line, &tmpv, &ipos, buf); 353275432Sbapt break; 354275432Sbapt case ARGV_NONE: 355275432Sbapt break; 356275432Sbapt } 357241675Suqs 358275432Sbapt /* Append to the return values. */ 359241675Suqs 360275432Sbapt if (*reta == NULL) 361275432Sbapt *reta = mandoc_calloc(1, sizeof(**reta)); 362241675Suqs 363275432Sbapt retc = ++(*reta)->argc; 364275432Sbapt retv = &(*reta)->argv; 365275432Sbapt *retv = mandoc_reallocarray(*retv, retc, sizeof(**retv)); 366275432Sbapt memcpy(*retv + retc - 1, &tmpv, sizeof(**retv)); 367275432Sbapt 368275432Sbapt /* Prepare for parsing the next flag. */ 369275432Sbapt 370275432Sbapt *pos = ipos; 371275432Sbapt argtable = mdocargs[tok].argvs; 372275432Sbapt } 373241675Suqs} 374241675Suqs 375241675Suqsvoid 376241675Suqsmdoc_argv_free(struct mdoc_arg *p) 377241675Suqs{ 378241675Suqs int i; 379241675Suqs 380241675Suqs if (NULL == p) 381241675Suqs return; 382241675Suqs 383241675Suqs if (p->refcnt) { 384241675Suqs --(p->refcnt); 385241675Suqs if (p->refcnt) 386241675Suqs return; 387241675Suqs } 388241675Suqs assert(p->argc); 389241675Suqs 390241675Suqs for (i = (int)p->argc - 1; i >= 0; i--) 391241675Suqs argn_free(p, i); 392241675Suqs 393241675Suqs free(p->argv); 394241675Suqs free(p); 395241675Suqs} 396241675Suqs 397241675Suqsstatic void 398241675Suqsargn_free(struct mdoc_arg *p, int iarg) 399241675Suqs{ 400241675Suqs struct mdoc_argv *arg; 401241675Suqs int j; 402241675Suqs 403241675Suqs arg = &p->argv[iarg]; 404241675Suqs 405241675Suqs if (arg->sz && arg->value) { 406274880Sbapt for (j = (int)arg->sz - 1; j >= 0; j--) 407241675Suqs free(arg->value[j]); 408241675Suqs free(arg->value); 409241675Suqs } 410241675Suqs 411241675Suqs for (--p->argc; iarg < (int)p->argc; iarg++) 412241675Suqs p->argv[iarg] = p->argv[iarg+1]; 413241675Suqs} 414241675Suqs 415241675Suqsenum margserr 416294113Sbaptmdoc_args(struct roff_man *mdoc, int line, int *pos, 417294113Sbapt char *buf, int tok, char **v) 418241675Suqs{ 419294113Sbapt struct roff_node *n; 420275432Sbapt char *v_local; 421241675Suqs enum argsflag fl; 422241675Suqs 423275432Sbapt if (v == NULL) 424275432Sbapt v = &v_local; 425294113Sbapt fl = tok == TOKEN_NONE ? ARGSFL_NONE : mdocargs[tok].flags; 426275432Sbapt if (tok != MDOC_It) 427294113Sbapt return args(mdoc, line, pos, buf, fl, v); 428241675Suqs 429241675Suqs /* 430241675Suqs * We know that we're in an `It', so it's reasonable to expect 431241675Suqs * us to be sitting in a `Bl'. Someday this may not be the case 432241675Suqs * (if we allow random `It's sitting out there), so provide a 433241675Suqs * safe fall-back into the default behaviour. 434241675Suqs */ 435241675Suqs 436261344Suqs for (n = mdoc->last; n; n = n->parent) 437241675Suqs if (MDOC_Bl == n->tok) 438241675Suqs if (LIST_column == n->norm->Bl.type) { 439241675Suqs fl = ARGSFL_TABSEP; 440241675Suqs break; 441241675Suqs } 442241675Suqs 443294113Sbapt return args(mdoc, line, pos, buf, fl, v); 444241675Suqs} 445241675Suqs 446241675Suqsstatic enum margserr 447294113Sbaptargs(struct roff_man *mdoc, int line, int *pos, 448241675Suqs char *buf, enum argsflag fl, char **v) 449241675Suqs{ 450294113Sbapt char *p; 451261344Suqs int pairs; 452241675Suqs 453294113Sbapt if (buf[*pos] == '\0') { 454294113Sbapt if (mdoc->flags & MDOC_PHRASELIT && 455294113Sbapt ! (mdoc->flags & MDOC_PHRASE)) { 456274880Sbapt mandoc_msg(MANDOCERR_ARG_QUOTE, 457274880Sbapt mdoc->parse, line, *pos, NULL); 458294113Sbapt mdoc->flags &= ~MDOC_PHRASELIT; 459294113Sbapt } 460294113Sbapt return ARGS_EOLN; 461241675Suqs } 462241675Suqs 463294113Sbapt *v = buf + *pos; 464241675Suqs 465294113Sbapt if (fl == ARGSFL_DELIM && args_checkpunct(buf, *pos)) 466294113Sbapt return ARGS_PUNCT; 467241675Suqs 468241675Suqs /* 469294113Sbapt * Tabs in `It' lines in `Bl -column' can't be escaped. 470294113Sbapt * Phrases are reparsed for `Ta' and other macros later. 471241675Suqs */ 472241675Suqs 473294113Sbapt if (fl == ARGSFL_TABSEP) { 474294113Sbapt if ((p = strchr(*v, '\t')) != NULL) { 475241675Suqs 476294113Sbapt /* 477294113Sbapt * Words right before and right after 478294113Sbapt * tab characters are not parsed, 479294113Sbapt * unless there is a blank in between. 480294113Sbapt */ 481241675Suqs 482294113Sbapt if (p[-1] != ' ') 483294113Sbapt mdoc->flags |= MDOC_PHRASEQL; 484294113Sbapt if (p[1] != ' ') 485294113Sbapt mdoc->flags |= MDOC_PHRASEQN; 486241675Suqs 487294113Sbapt /* 488294113Sbapt * One or more blanks after a tab cause 489294113Sbapt * one leading blank in the next column. 490294113Sbapt * So skip all but one of them. 491294113Sbapt */ 492241675Suqs 493294113Sbapt *pos += (int)(p - *v) + 1; 494294113Sbapt while (buf[*pos] == ' ' && buf[*pos + 1] == ' ') 495294113Sbapt (*pos)++; 496241675Suqs 497294113Sbapt /* 498294113Sbapt * A tab at the end of an input line 499294113Sbapt * switches to the next column. 500294113Sbapt */ 501241675Suqs 502294113Sbapt if (buf[*pos] == '\0' || buf[*pos + 1] == '\0') 503294113Sbapt mdoc->flags |= MDOC_PHRASEQN; 504294113Sbapt } else { 505294113Sbapt p = strchr(*v, '\0'); 506294113Sbapt if (p[-1] == ' ') 507294113Sbapt mandoc_msg(MANDOCERR_SPACE_EOL, 508294113Sbapt mdoc->parse, line, *pos, NULL); 509294113Sbapt *pos += (int)(p - *v); 510241675Suqs } 511241675Suqs 512294113Sbapt /* Skip any trailing blank characters. */ 513294113Sbapt while (p > *v && p[-1] == ' ' && 514294113Sbapt (p - 1 == *v || p[-2] != '\\')) 515294113Sbapt p--; 516294113Sbapt *p = '\0'; 517241675Suqs 518294113Sbapt return ARGS_PHRASE; 519261344Suqs } 520241675Suqs 521261344Suqs /* 522241675Suqs * Process a quoted literal. A quote begins with a double-quote 523241675Suqs * and ends with a double-quote NOT preceded by a double-quote. 524261344Suqs * NUL-terminate the literal in place. 525261344Suqs * Collapse pairs of quotes inside quoted literals. 526241675Suqs * Whitespace is NOT involved in literal termination. 527241675Suqs */ 528241675Suqs 529294113Sbapt if (mdoc->flags & MDOC_PHRASELIT || buf[*pos] == '\"') { 530294113Sbapt if ( ! (mdoc->flags & MDOC_PHRASELIT)) 531241675Suqs *v = &buf[++(*pos)]; 532241675Suqs 533294113Sbapt if (mdoc->flags & MDOC_PHRASE) 534261344Suqs mdoc->flags |= MDOC_PHRASELIT; 535241675Suqs 536261344Suqs pairs = 0; 537241675Suqs for ( ; buf[*pos]; (*pos)++) { 538261344Suqs /* Move following text left after quoted quotes. */ 539261344Suqs if (pairs) 540261344Suqs buf[*pos - pairs] = buf[*pos]; 541241675Suqs if ('\"' != buf[*pos]) 542241675Suqs continue; 543261344Suqs /* Unquoted quotes end quoted args. */ 544241675Suqs if ('\"' != buf[*pos + 1]) 545241675Suqs break; 546261344Suqs /* Quoted quotes collapse. */ 547261344Suqs pairs++; 548241675Suqs (*pos)++; 549241675Suqs } 550261344Suqs if (pairs) 551261344Suqs buf[*pos - pairs] = '\0'; 552241675Suqs 553294113Sbapt if (buf[*pos] == '\0') { 554294113Sbapt if ( ! (mdoc->flags & MDOC_PHRASE)) 555294113Sbapt mandoc_msg(MANDOCERR_ARG_QUOTE, 556294113Sbapt mdoc->parse, line, *pos, NULL); 557294113Sbapt return ARGS_QWORD; 558241675Suqs } 559241675Suqs 560261344Suqs mdoc->flags &= ~MDOC_PHRASELIT; 561241675Suqs buf[(*pos)++] = '\0'; 562241675Suqs 563241675Suqs if ('\0' == buf[*pos]) 564294113Sbapt return ARGS_QWORD; 565241675Suqs 566241675Suqs while (' ' == buf[*pos]) 567241675Suqs (*pos)++; 568241675Suqs 569241675Suqs if ('\0' == buf[*pos]) 570274880Sbapt mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse, 571274880Sbapt line, *pos, NULL); 572241675Suqs 573294113Sbapt return ARGS_QWORD; 574241675Suqs } 575241675Suqs 576241675Suqs p = &buf[*pos]; 577261344Suqs *v = mandoc_getarg(mdoc->parse, &p, line, pos); 578241675Suqs 579294113Sbapt /* 580294113Sbapt * After parsing the last word in this phrase, 581294113Sbapt * tell lookup() whether or not to interpret it. 582294113Sbapt */ 583294113Sbapt 584294113Sbapt if (*p == '\0' && mdoc->flags & MDOC_PHRASEQL) { 585294113Sbapt mdoc->flags &= ~MDOC_PHRASEQL; 586294113Sbapt mdoc->flags |= MDOC_PHRASEQF; 587294113Sbapt } 588294113Sbapt return ARGS_WORD; 589241675Suqs} 590241675Suqs 591274880Sbapt/* 592241675Suqs * Check if the string consists only of space-separated closing 593241675Suqs * delimiters. This is a bit of a dance: the first must be a close 594241675Suqs * delimiter, but it may be followed by middle delimiters. Arbitrary 595241675Suqs * whitespace may separate these tokens. 596241675Suqs */ 597241675Suqsstatic int 598241675Suqsargs_checkpunct(const char *buf, int i) 599241675Suqs{ 600241675Suqs int j; 601241675Suqs char dbuf[DELIMSZ]; 602241675Suqs enum mdelim d; 603241675Suqs 604241675Suqs /* First token must be a close-delimiter. */ 605241675Suqs 606241675Suqs for (j = 0; buf[i] && ' ' != buf[i] && j < DELIMSZ; j++, i++) 607241675Suqs dbuf[j] = buf[i]; 608241675Suqs 609241675Suqs if (DELIMSZ == j) 610294113Sbapt return 0; 611241675Suqs 612241675Suqs dbuf[j] = '\0'; 613241675Suqs if (DELIM_CLOSE != mdoc_isdelim(dbuf)) 614294113Sbapt return 0; 615241675Suqs 616241675Suqs while (' ' == buf[i]) 617241675Suqs i++; 618241675Suqs 619241675Suqs /* Remaining must NOT be open/none. */ 620274880Sbapt 621241675Suqs while (buf[i]) { 622241675Suqs j = 0; 623241675Suqs while (buf[i] && ' ' != buf[i] && j < DELIMSZ) 624241675Suqs dbuf[j++] = buf[i++]; 625241675Suqs 626241675Suqs if (DELIMSZ == j) 627294113Sbapt return 0; 628241675Suqs 629241675Suqs dbuf[j] = '\0'; 630241675Suqs d = mdoc_isdelim(dbuf); 631241675Suqs if (DELIM_NONE == d || DELIM_OPEN == d) 632294113Sbapt return 0; 633241675Suqs 634241675Suqs while (' ' == buf[i]) 635241675Suqs i++; 636241675Suqs } 637241675Suqs 638294113Sbapt return '\0' == buf[i]; 639241675Suqs} 640241675Suqs 641275432Sbaptstatic void 642294113Sbaptargv_multi(struct roff_man *mdoc, int line, 643241675Suqs struct mdoc_argv *v, int *pos, char *buf) 644241675Suqs{ 645241675Suqs enum margserr ac; 646241675Suqs char *p; 647241675Suqs 648241675Suqs for (v->sz = 0; ; v->sz++) { 649275432Sbapt if (buf[*pos] == '-') 650241675Suqs break; 651261344Suqs ac = args(mdoc, line, pos, buf, ARGSFL_NONE, &p); 652275432Sbapt if (ac == ARGS_EOLN) 653241675Suqs break; 654241675Suqs 655275432Sbapt if (v->sz % MULTI_STEP == 0) 656274880Sbapt v->value = mandoc_reallocarray(v->value, 657274880Sbapt v->sz + MULTI_STEP, sizeof(char *)); 658241675Suqs 659241675Suqs v->value[(int)v->sz] = mandoc_strdup(p); 660241675Suqs } 661241675Suqs} 662241675Suqs 663275432Sbaptstatic void 664294113Sbaptargv_single(struct roff_man *mdoc, int line, 665241675Suqs struct mdoc_argv *v, int *pos, char *buf) 666241675Suqs{ 667241675Suqs enum margserr ac; 668241675Suqs char *p; 669241675Suqs 670261344Suqs ac = args(mdoc, line, pos, buf, ARGSFL_NONE, &p); 671275432Sbapt if (ac == ARGS_EOLN) 672275432Sbapt return; 673241675Suqs 674241675Suqs v->sz = 1; 675241675Suqs v->value = mandoc_malloc(sizeof(char *)); 676241675Suqs v->value[0] = mandoc_strdup(p); 677241675Suqs} 678