Annotation of src/usr.bin/mandoc/mdoc_argv.c, Revision 1.54
1.54 ! schwarze 1: /* $OpenBSD: mdoc_argv.c,v 1.53 2014/11/28 03:13:58 schwarze Exp $ */
1.1 kristaps 2: /*
1.36 schwarze 3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.54 ! schwarze 4: * Copyright (c) 2012, 2014 Ingo Schwarze <schwarze@openbsd.org>
1.1 kristaps 5: *
6: * Permission to use, copy, modify, and distribute this software for any
1.2 schwarze 7: * purpose with or without fee is hereby granted, provided that the above
8: * copyright notice and this permission notice appear in all copies.
1.1 kristaps 9: *
1.2 schwarze 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.
1.1 kristaps 17: */
18: #include <sys/types.h>
19:
20: #include <assert.h>
21: #include <stdlib.h>
22: #include <stdio.h>
23: #include <string.h>
24:
1.37 schwarze 25: #include "mdoc.h"
1.29 schwarze 26: #include "mandoc.h"
1.47 schwarze 27: #include "mandoc_aux.h"
1.1 kristaps 28: #include "libmdoc.h"
1.19 schwarze 29: #include "libmandoc.h"
1.1 kristaps 30:
1.36 schwarze 31: #define MULTI_STEP 5 /* pre-allocate argument values */
1.49 schwarze 32: #define DELIMSZ 6 /* max possible size of a delimiter */
1.38 schwarze 33:
34: enum argsflag {
35: ARGSFL_NONE = 0,
36: ARGSFL_DELIM, /* handle delimiters of [[::delim::][ ]+]+ */
37: ARGSFL_TABSEP /* handle tab/`Ta' separated phrases */
38: };
39:
40: enum argvflag {
41: ARGV_NONE, /* no args to flag (e.g., -split) */
42: ARGV_SINGLE, /* one arg to flag (e.g., -file xxx) */
1.40 schwarze 43: ARGV_MULTI /* multiple args (e.g., -column xxx yyy) */
1.38 schwarze 44: };
1.1 kristaps 45:
1.39 schwarze 46: struct mdocarg {
47: enum argsflag flags;
48: const enum mdocargt *argvs;
49: };
50:
51: static void argn_free(struct mdoc_arg *, int);
1.49 schwarze 52: static enum margserr args(struct mdoc *, int, int *,
1.38 schwarze 53: char *, enum argsflag, char **);
54: static int args_checkpunct(const char *, int);
1.53 schwarze 55: static void argv_multi(struct mdoc *, int,
1.1 kristaps 56: struct mdoc_argv *, int *, char *);
1.53 schwarze 57: static void argv_single(struct mdoc *, int,
1.1 kristaps 58: struct mdoc_argv *, int *, char *);
59:
1.36 schwarze 60: static const enum argvflag argvflags[MDOC_ARG_MAX] = {
1.1 kristaps 61: ARGV_NONE, /* MDOC_Split */
62: ARGV_NONE, /* MDOC_Nosplit */
63: ARGV_NONE, /* MDOC_Ragged */
64: ARGV_NONE, /* MDOC_Unfilled */
65: ARGV_NONE, /* MDOC_Literal */
1.17 schwarze 66: ARGV_SINGLE, /* MDOC_File */
1.40 schwarze 67: ARGV_SINGLE, /* MDOC_Offset */
1.1 kristaps 68: ARGV_NONE, /* MDOC_Bullet */
69: ARGV_NONE, /* MDOC_Dash */
70: ARGV_NONE, /* MDOC_Hyphen */
71: ARGV_NONE, /* MDOC_Item */
72: ARGV_NONE, /* MDOC_Enum */
73: ARGV_NONE, /* MDOC_Tag */
74: ARGV_NONE, /* MDOC_Diag */
75: ARGV_NONE, /* MDOC_Hang */
76: ARGV_NONE, /* MDOC_Ohang */
77: ARGV_NONE, /* MDOC_Inset */
78: ARGV_MULTI, /* MDOC_Column */
79: ARGV_SINGLE, /* MDOC_Width */
80: ARGV_NONE, /* MDOC_Compact */
1.14 schwarze 81: ARGV_NONE, /* MDOC_Std */
1.1 kristaps 82: ARGV_NONE, /* MDOC_Filled */
83: ARGV_NONE, /* MDOC_Words */
84: ARGV_NONE, /* MDOC_Emphasis */
85: ARGV_NONE, /* MDOC_Symbolic */
86: ARGV_NONE /* MDOC_Symbolic */
87: };
88:
1.37 schwarze 89: static const enum mdocargt args_Ex[] = {
90: MDOC_Std,
91: MDOC_ARG_MAX
92: };
93:
94: static const enum mdocargt args_An[] = {
95: MDOC_Split,
96: MDOC_Nosplit,
97: MDOC_ARG_MAX
98: };
99:
100: static const enum mdocargt args_Bd[] = {
101: MDOC_Ragged,
102: MDOC_Unfilled,
103: MDOC_Filled,
104: MDOC_Literal,
105: MDOC_File,
106: MDOC_Offset,
107: MDOC_Compact,
108: MDOC_Centred,
109: MDOC_ARG_MAX
110: };
111:
112: static const enum mdocargt args_Bf[] = {
113: MDOC_Emphasis,
114: MDOC_Literal,
115: MDOC_Symbolic,
116: MDOC_ARG_MAX
117: };
118:
119: static const enum mdocargt args_Bk[] = {
120: MDOC_Words,
121: MDOC_ARG_MAX
122: };
123:
124: static const enum mdocargt args_Bl[] = {
125: MDOC_Bullet,
126: MDOC_Dash,
127: MDOC_Hyphen,
128: MDOC_Item,
129: MDOC_Enum,
130: MDOC_Tag,
131: MDOC_Diag,
132: MDOC_Hang,
133: MDOC_Ohang,
134: MDOC_Inset,
135: MDOC_Column,
136: MDOC_Width,
137: MDOC_Offset,
138: MDOC_Compact,
139: MDOC_Nested,
140: MDOC_ARG_MAX
141: };
142:
1.39 schwarze 143: static const struct mdocarg mdocargs[MDOC_MAX] = {
1.41 schwarze 144: { ARGSFL_DELIM, NULL }, /* Ap */
1.39 schwarze 145: { ARGSFL_NONE, NULL }, /* Dd */
146: { ARGSFL_NONE, NULL }, /* Dt */
147: { ARGSFL_NONE, NULL }, /* Os */
148: { ARGSFL_NONE, NULL }, /* Sh */
1.49 schwarze 149: { ARGSFL_NONE, NULL }, /* Ss */
150: { ARGSFL_NONE, NULL }, /* Pp */
1.39 schwarze 151: { ARGSFL_DELIM, NULL }, /* D1 */
152: { ARGSFL_DELIM, NULL }, /* Dl */
153: { ARGSFL_NONE, args_Bd }, /* Bd */
154: { ARGSFL_NONE, NULL }, /* Ed */
155: { ARGSFL_NONE, args_Bl }, /* Bl */
156: { ARGSFL_NONE, NULL }, /* El */
157: { ARGSFL_NONE, NULL }, /* It */
1.49 schwarze 158: { ARGSFL_DELIM, NULL }, /* Ad */
1.39 schwarze 159: { ARGSFL_DELIM, args_An }, /* An */
160: { ARGSFL_DELIM, NULL }, /* Ar */
1.43 schwarze 161: { ARGSFL_DELIM, NULL }, /* Cd */
1.39 schwarze 162: { ARGSFL_DELIM, NULL }, /* Cm */
1.49 schwarze 163: { ARGSFL_DELIM, NULL }, /* Dv */
164: { ARGSFL_DELIM, NULL }, /* Er */
165: { ARGSFL_DELIM, NULL }, /* Ev */
1.39 schwarze 166: { ARGSFL_NONE, args_Ex }, /* Ex */
1.49 schwarze 167: { ARGSFL_DELIM, NULL }, /* Fa */
168: { ARGSFL_NONE, NULL }, /* Fd */
1.39 schwarze 169: { ARGSFL_DELIM, NULL }, /* Fl */
1.49 schwarze 170: { ARGSFL_DELIM, NULL }, /* Fn */
171: { ARGSFL_DELIM, NULL }, /* Ft */
172: { ARGSFL_DELIM, NULL }, /* Ic */
173: { ARGSFL_DELIM, NULL }, /* In */
1.39 schwarze 174: { ARGSFL_DELIM, NULL }, /* Li */
1.49 schwarze 175: { ARGSFL_NONE, NULL }, /* Nd */
176: { ARGSFL_DELIM, NULL }, /* Nm */
1.39 schwarze 177: { ARGSFL_DELIM, NULL }, /* Op */
1.51 schwarze 178: { ARGSFL_DELIM, NULL }, /* Ot */
1.39 schwarze 179: { ARGSFL_DELIM, NULL }, /* Pa */
180: { ARGSFL_NONE, args_Ex }, /* Rv */
1.49 schwarze 181: { ARGSFL_DELIM, NULL }, /* St */
1.39 schwarze 182: { ARGSFL_DELIM, NULL }, /* Va */
1.49 schwarze 183: { ARGSFL_DELIM, NULL }, /* Vt */
1.39 schwarze 184: { ARGSFL_DELIM, NULL }, /* Xr */
185: { ARGSFL_NONE, NULL }, /* %A */
186: { ARGSFL_NONE, NULL }, /* %B */
187: { ARGSFL_NONE, NULL }, /* %D */
188: { ARGSFL_NONE, NULL }, /* %I */
189: { ARGSFL_NONE, NULL }, /* %J */
190: { ARGSFL_NONE, NULL }, /* %N */
191: { ARGSFL_NONE, NULL }, /* %O */
192: { ARGSFL_NONE, NULL }, /* %P */
193: { ARGSFL_NONE, NULL }, /* %R */
194: { ARGSFL_NONE, NULL }, /* %T */
195: { ARGSFL_NONE, NULL }, /* %V */
196: { ARGSFL_DELIM, NULL }, /* Ac */
197: { ARGSFL_NONE, NULL }, /* Ao */
198: { ARGSFL_DELIM, NULL }, /* Aq */
199: { ARGSFL_DELIM, NULL }, /* At */
200: { ARGSFL_DELIM, NULL }, /* Bc */
1.49 schwarze 201: { ARGSFL_NONE, args_Bf }, /* Bf */
1.39 schwarze 202: { ARGSFL_NONE, NULL }, /* Bo */
203: { ARGSFL_DELIM, NULL }, /* Bq */
204: { ARGSFL_DELIM, NULL }, /* Bsx */
205: { ARGSFL_DELIM, NULL }, /* Bx */
206: { ARGSFL_NONE, NULL }, /* Db */
207: { ARGSFL_DELIM, NULL }, /* Dc */
208: { ARGSFL_NONE, NULL }, /* Do */
209: { ARGSFL_DELIM, NULL }, /* Dq */
210: { ARGSFL_DELIM, NULL }, /* Ec */
211: { ARGSFL_NONE, NULL }, /* Ef */
1.49 schwarze 212: { ARGSFL_DELIM, NULL }, /* Em */
1.39 schwarze 213: { ARGSFL_NONE, NULL }, /* Eo */
214: { ARGSFL_DELIM, NULL }, /* Fx */
215: { ARGSFL_DELIM, NULL }, /* Ms */
216: { ARGSFL_DELIM, NULL }, /* No */
217: { ARGSFL_DELIM, NULL }, /* Ns */
218: { ARGSFL_DELIM, NULL }, /* Nx */
219: { ARGSFL_DELIM, NULL }, /* Ox */
220: { ARGSFL_DELIM, NULL }, /* Pc */
221: { ARGSFL_DELIM, NULL }, /* Pf */
222: { ARGSFL_NONE, NULL }, /* Po */
223: { ARGSFL_DELIM, NULL }, /* Pq */
224: { ARGSFL_DELIM, NULL }, /* Qc */
225: { ARGSFL_DELIM, NULL }, /* Ql */
226: { ARGSFL_NONE, NULL }, /* Qo */
227: { ARGSFL_DELIM, NULL }, /* Qq */
228: { ARGSFL_NONE, NULL }, /* Re */
229: { ARGSFL_NONE, NULL }, /* Rs */
230: { ARGSFL_DELIM, NULL }, /* Sc */
231: { ARGSFL_NONE, NULL }, /* So */
232: { ARGSFL_DELIM, NULL }, /* Sq */
233: { ARGSFL_NONE, NULL }, /* Sm */
234: { ARGSFL_DELIM, NULL }, /* Sx */
235: { ARGSFL_DELIM, NULL }, /* Sy */
236: { ARGSFL_DELIM, NULL }, /* Tn */
237: { ARGSFL_DELIM, NULL }, /* Ux */
238: { ARGSFL_DELIM, NULL }, /* Xc */
239: { ARGSFL_NONE, NULL }, /* Xo */
1.49 schwarze 240: { ARGSFL_NONE, NULL }, /* Fo */
241: { ARGSFL_DELIM, NULL }, /* Fc */
1.39 schwarze 242: { ARGSFL_NONE, NULL }, /* Oo */
243: { ARGSFL_DELIM, NULL }, /* Oc */
244: { ARGSFL_NONE, args_Bk }, /* Bk */
245: { ARGSFL_NONE, NULL }, /* Ek */
246: { ARGSFL_NONE, NULL }, /* Bt */
247: { ARGSFL_NONE, NULL }, /* Hf */
1.51 schwarze 248: { ARGSFL_DELIM, NULL }, /* Fr */
1.39 schwarze 249: { ARGSFL_NONE, NULL }, /* Ud */
1.43 schwarze 250: { ARGSFL_DELIM, NULL }, /* Lb */
1.39 schwarze 251: { ARGSFL_NONE, NULL }, /* Lp */
252: { ARGSFL_DELIM, NULL }, /* Lk */
253: { ARGSFL_DELIM, NULL }, /* Mt */
254: { ARGSFL_DELIM, NULL }, /* Brq */
255: { ARGSFL_NONE, NULL }, /* Bro */
256: { ARGSFL_DELIM, NULL }, /* Brc */
257: { ARGSFL_NONE, NULL }, /* %C */
258: { ARGSFL_NONE, NULL }, /* Es */
1.51 schwarze 259: { ARGSFL_DELIM, NULL }, /* En */
1.45 schwarze 260: { ARGSFL_DELIM, NULL }, /* Dx */
1.39 schwarze 261: { ARGSFL_NONE, NULL }, /* %Q */
262: { ARGSFL_NONE, NULL }, /* br */
263: { ARGSFL_NONE, NULL }, /* sp */
264: { ARGSFL_NONE, NULL }, /* %U */
265: { ARGSFL_NONE, NULL }, /* Ta */
1.48 schwarze 266: { ARGSFL_NONE, NULL }, /* ll */
1.39 schwarze 267: };
268:
269:
1.1 kristaps 270: /*
1.54 ! schwarze 271: * Parse flags and their arguments from the input line.
! 272: * These come in the form -flag [argument ...].
! 273: * Some flags take no argument, some one, some multiple.
1.1 kristaps 274: */
1.54 ! schwarze 275: void
1.42 schwarze 276: mdoc_argv(struct mdoc *mdoc, int line, enum mdoct tok,
1.54 ! schwarze 277: struct mdoc_arg **reta, int *pos, char *buf)
1.1 kristaps 278: {
1.54 ! schwarze 279: struct mdoc_argv tmpv;
! 280: struct mdoc_argv **retv;
! 281: const enum mdocargt *argtable;
! 282: char *argname;
! 283: int ipos, retc;
! 284: char savechar;
1.1 kristaps 285:
1.54 ! schwarze 286: *reta = NULL;
1.1 kristaps 287:
1.54 ! schwarze 288: /* Which flags does this macro support? */
1.1 kristaps 289:
1.54 ! schwarze 290: argtable = mdocargs[tok].argvs;
! 291: if (argtable == NULL)
! 292: return;
1.1 kristaps 293:
1.54 ! schwarze 294: /* Loop over the flags on the input line. */
! 295:
! 296: ipos = *pos;
! 297: while (buf[ipos] == '-') {
! 298:
! 299: /* Seek to the first unescaped space. */
1.1 kristaps 300:
1.54 ! schwarze 301: for (argname = buf + ++ipos; buf[ipos] != '\0'; ipos++)
! 302: if (buf[ipos] == ' ' && buf[ipos - 1] != '\\')
! 303: break;
1.1 kristaps 304:
1.54 ! schwarze 305: /*
! 306: * We want to nil-terminate the word to look it up.
! 307: * But we may not have a flag, in which case we need
! 308: * to restore the line as-is. So keep around the
! 309: * stray byte, which we'll reset upon exiting.
! 310: */
! 311:
! 312: if ((savechar = buf[ipos]) != '\0')
! 313: buf[ipos++] = '\0';
1.1 kristaps 314:
1.49 schwarze 315: /*
1.54 ! schwarze 316: * Now look up the word as a flag. Use temporary
! 317: * storage that we'll copy into the node's flags.
1.39 schwarze 318: */
1.1 kristaps 319:
1.54 ! schwarze 320: while ((tmpv.arg = *argtable++) != MDOC_ARG_MAX)
! 321: if ( ! strcmp(argname, mdoc_argnames[tmpv.arg]))
! 322: break;
! 323:
! 324: /* If it isn't a flag, restore the saved byte. */
! 325:
! 326: if (tmpv.arg == MDOC_ARG_MAX) {
! 327: if (savechar != '\0')
! 328: buf[ipos - 1] = savechar;
! 329: break;
! 330: }
! 331:
! 332: /* Read to the next word (the first argument). */
! 333:
! 334: while (buf[ipos] == ' ')
! 335: ipos++;
! 336:
! 337: /* Parse the arguments of the flag. */
1.39 schwarze 338:
1.54 ! schwarze 339: tmpv.line = line;
! 340: tmpv.pos = ipos;
! 341: tmpv.sz = 0;
! 342: tmpv.value = NULL;
1.1 kristaps 343:
1.54 ! schwarze 344: switch (argvflags[tmpv.arg]) {
! 345: case ARGV_SINGLE:
! 346: argv_single(mdoc, line, &tmpv, &ipos, buf);
! 347: break;
! 348: case ARGV_MULTI:
! 349: argv_multi(mdoc, line, &tmpv, &ipos, buf);
! 350: break;
! 351: case ARGV_NONE:
! 352: break;
! 353: }
! 354:
! 355: /* Append to the return values. */
1.1 kristaps 356:
1.54 ! schwarze 357: if (*reta == NULL)
! 358: *reta = mandoc_calloc(1, sizeof(**reta));
1.1 kristaps 359:
1.54 ! schwarze 360: retc = ++(*reta)->argc;
! 361: retv = &(*reta)->argv;
! 362: *retv = mandoc_reallocarray(*retv, retc, sizeof(**retv));
! 363: memcpy(*retv + retc - 1, &tmpv, sizeof(**retv));
1.1 kristaps 364:
1.54 ! schwarze 365: /* Prepare for parsing the next flag. */
1.1 kristaps 366:
1.54 ! schwarze 367: *pos = ipos;
! 368: argtable = mdocargs[tok].argvs;
! 369: }
1.1 kristaps 370: }
371:
372: void
373: mdoc_argv_free(struct mdoc_arg *p)
374: {
1.23 schwarze 375: int i;
1.1 kristaps 376:
377: if (NULL == p)
378: return;
379:
380: if (p->refcnt) {
381: --(p->refcnt);
382: if (p->refcnt)
383: return;
384: }
385: assert(p->argc);
386:
1.24 schwarze 387: for (i = (int)p->argc - 1; i >= 0; i--)
1.37 schwarze 388: argn_free(p, i);
1.23 schwarze 389:
390: free(p->argv);
391: free(p);
392: }
393:
1.37 schwarze 394: static void
395: argn_free(struct mdoc_arg *p, int iarg)
1.23 schwarze 396: {
1.33 schwarze 397: struct mdoc_argv *arg;
1.23 schwarze 398: int j;
1.1 kristaps 399:
1.33 schwarze 400: arg = &p->argv[iarg];
401:
1.23 schwarze 402: if (arg->sz && arg->value) {
1.49 schwarze 403: for (j = (int)arg->sz - 1; j >= 0; j--)
1.23 schwarze 404: free(arg->value[j]);
405: free(arg->value);
1.1 kristaps 406: }
407:
1.23 schwarze 408: for (--p->argc; iarg < (int)p->argc; iarg++)
409: p->argv[iarg] = p->argv[iarg+1];
1.1 kristaps 410: }
411:
1.24 schwarze 412: enum margserr
1.42 schwarze 413: mdoc_zargs(struct mdoc *mdoc, int line, int *pos, char *buf, char **v)
1.12 schwarze 414: {
415:
1.42 schwarze 416: return(args(mdoc, line, pos, buf, ARGSFL_NONE, v));
1.12 schwarze 417: }
418:
1.24 schwarze 419: enum margserr
1.49 schwarze 420: mdoc_args(struct mdoc *mdoc, int line, int *pos,
1.22 schwarze 421: char *buf, enum mdoct tok, char **v)
1.1 kristaps 422: {
1.38 schwarze 423: enum argsflag fl;
1.1 kristaps 424: struct mdoc_node *n;
425:
1.39 schwarze 426: fl = mdocargs[tok].flags;
1.1 kristaps 427:
1.12 schwarze 428: if (MDOC_It != tok)
1.42 schwarze 429: return(args(mdoc, line, pos, buf, fl, v));
1.12 schwarze 430:
1.30 schwarze 431: /*
432: * We know that we're in an `It', so it's reasonable to expect
433: * us to be sitting in a `Bl'. Someday this may not be the case
434: * (if we allow random `It's sitting out there), so provide a
435: * safe fall-back into the default behaviour.
1.1 kristaps 436: */
437:
1.42 schwarze 438: for (n = mdoc->last; n; n = n->parent)
1.30 schwarze 439: if (MDOC_Bl == n->tok)
1.38 schwarze 440: if (LIST_column == n->norm->Bl.type) {
441: fl = ARGSFL_TABSEP;
442: break;
443: }
1.1 kristaps 444:
1.42 schwarze 445: return(args(mdoc, line, pos, buf, fl, v));
1.1 kristaps 446: }
447:
1.24 schwarze 448: static enum margserr
1.49 schwarze 449: args(struct mdoc *mdoc, int line, int *pos,
1.38 schwarze 450: char *buf, enum argsflag fl, char **v)
1.1 kristaps 451: {
1.30 schwarze 452: char *p, *pp;
1.44 schwarze 453: int pairs;
1.30 schwarze 454: enum margserr rc;
1.1 kristaps 455:
1.28 schwarze 456: if ('\0' == buf[*pos]) {
1.42 schwarze 457: if (MDOC_PPHRASE & mdoc->flags)
1.28 schwarze 458: return(ARGS_EOLN);
459: /*
460: * If we're not in a partial phrase and the flag for
461: * being a phrase literal is still set, the punctuation
462: * is unterminated.
463: */
1.42 schwarze 464: if (MDOC_PHRASELIT & mdoc->flags)
1.52 schwarze 465: mandoc_msg(MANDOCERR_ARG_QUOTE,
466: mdoc->parse, line, *pos, NULL);
1.28 schwarze 467:
1.42 schwarze 468: mdoc->flags &= ~MDOC_PHRASELIT;
1.1 kristaps 469: return(ARGS_EOLN);
1.28 schwarze 470: }
1.1 kristaps 471:
1.36 schwarze 472: *v = &buf[*pos];
1.1 kristaps 473:
1.38 schwarze 474: if (ARGSFL_DELIM == fl)
475: if (args_checkpunct(buf, *pos))
1.36 schwarze 476: return(ARGS_PUNCT);
1.1 kristaps 477:
1.12 schwarze 478: /*
479: * First handle TABSEP items, restricted to `Bl -column'. This
480: * ignores conventional token parsing and instead uses tabs or
481: * `Ta' macros to separate phrases. Phrases are parsed again
482: * for arguments at a later phase.
483: */
484:
1.38 schwarze 485: if (ARGSFL_TABSEP == fl) {
1.12 schwarze 486: /* Scan ahead to tab (can't be escaped). */
487: p = strchr(*v, '\t');
1.28 schwarze 488: pp = NULL;
1.12 schwarze 489:
490: /* Scan ahead to unescaped `Ta'. */
1.49 schwarze 491: if ( ! (MDOC_PHRASELIT & mdoc->flags))
1.28 schwarze 492: for (pp = *v; ; pp++) {
493: if (NULL == (pp = strstr(pp, "Ta")))
494: break;
495: if (pp > *v && ' ' != *(pp - 1))
496: continue;
1.30 schwarze 497: if (' ' == *(pp + 2) || '\0' == *(pp + 2))
1.28 schwarze 498: break;
499: }
1.1 kristaps 500:
1.24 schwarze 501: /* By default, assume a phrase. */
502: rc = ARGS_PHRASE;
503:
1.49 schwarze 504: /*
1.12 schwarze 505: * Adjust new-buffer position to be beyond delimiter
506: * mark (e.g., Ta -> end + 2).
1.1 kristaps 507: */
1.12 schwarze 508: if (p && pp) {
509: *pos += pp < p ? 2 : 1;
1.24 schwarze 510: rc = pp < p ? ARGS_PHRASE : ARGS_PPHRASE;
1.12 schwarze 511: p = pp < p ? pp : p;
512: } else if (p && ! pp) {
1.24 schwarze 513: rc = ARGS_PPHRASE;
1.12 schwarze 514: *pos += 1;
515: } else if (pp && ! p) {
516: p = pp;
517: *pos += 2;
1.25 schwarze 518: } else {
519: rc = ARGS_PEND;
1.12 schwarze 520: p = strchr(*v, 0);
1.25 schwarze 521: }
1.1 kristaps 522:
1.12 schwarze 523: /* Whitespace check for eoln case... */
1.38 schwarze 524: if ('\0' == *p && ' ' == *(p - 1))
1.52 schwarze 525: mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
526: line, *pos, NULL);
1.12 schwarze 527:
528: *pos += (int)(p - *v);
529:
530: /* Strip delimiter's preceding whitespace. */
531: pp = p - 1;
532: while (pp > *v && ' ' == *pp) {
533: if (pp > *v && '\\' == *(pp - 1))
534: break;
535: pp--;
536: }
537: *(pp + 1) = 0;
1.1 kristaps 538:
1.12 schwarze 539: /* Strip delimiter's proceeding whitespace. */
540: for (pp = &buf[*pos]; ' ' == *pp; pp++, (*pos)++)
541: /* Skip ahead. */ ;
1.1 kristaps 542:
1.24 schwarze 543: return(rc);
1.46 schwarze 544: }
1.1 kristaps 545:
1.46 schwarze 546: /*
1.12 schwarze 547: * Process a quoted literal. A quote begins with a double-quote
548: * and ends with a double-quote NOT preceded by a double-quote.
1.46 schwarze 549: * NUL-terminate the literal in place.
1.44 schwarze 550: * Collapse pairs of quotes inside quoted literals.
1.12 schwarze 551: * Whitespace is NOT involved in literal termination.
552: */
1.1 kristaps 553:
1.42 schwarze 554: if (MDOC_PHRASELIT & mdoc->flags || '\"' == buf[*pos]) {
555: if ( ! (MDOC_PHRASELIT & mdoc->flags))
1.28 schwarze 556: *v = &buf[++(*pos)];
557:
1.42 schwarze 558: if (MDOC_PPHRASE & mdoc->flags)
559: mdoc->flags |= MDOC_PHRASELIT;
1.1 kristaps 560:
1.44 schwarze 561: pairs = 0;
1.12 schwarze 562: for ( ; buf[*pos]; (*pos)++) {
1.44 schwarze 563: /* Move following text left after quoted quotes. */
564: if (pairs)
565: buf[*pos - pairs] = buf[*pos];
1.12 schwarze 566: if ('\"' != buf[*pos])
567: continue;
1.44 schwarze 568: /* Unquoted quotes end quoted args. */
1.12 schwarze 569: if ('\"' != buf[*pos + 1])
570: break;
1.44 schwarze 571: /* Quoted quotes collapse. */
572: pairs++;
1.12 schwarze 573: (*pos)++;
574: }
1.44 schwarze 575: if (pairs)
576: buf[*pos - pairs] = '\0';
1.1 kristaps 577:
1.28 schwarze 578: if ('\0' == buf[*pos]) {
1.42 schwarze 579: if (MDOC_PPHRASE & mdoc->flags)
1.13 schwarze 580: return(ARGS_QWORD);
1.52 schwarze 581: mandoc_msg(MANDOCERR_ARG_QUOTE,
582: mdoc->parse, line, *pos, NULL);
1.12 schwarze 583: return(ARGS_QWORD);
584: }
1.1 kristaps 585:
1.42 schwarze 586: mdoc->flags &= ~MDOC_PHRASELIT;
1.28 schwarze 587: buf[(*pos)++] = '\0';
1.1 kristaps 588:
1.28 schwarze 589: if ('\0' == buf[*pos])
1.12 schwarze 590: return(ARGS_QWORD);
1.1 kristaps 591:
1.12 schwarze 592: while (' ' == buf[*pos])
593: (*pos)++;
1.1 kristaps 594:
1.38 schwarze 595: if ('\0' == buf[*pos])
1.52 schwarze 596: mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
597: line, *pos, NULL);
1.1 kristaps 598:
1.12 schwarze 599: return(ARGS_QWORD);
1.1 kristaps 600: }
601:
1.38 schwarze 602: p = &buf[*pos];
1.42 schwarze 603: *v = mandoc_getarg(mdoc->parse, &p, line, pos);
1.1 kristaps 604:
1.12 schwarze 605: return(ARGS_WORD);
1.1 kristaps 606: }
607:
1.49 schwarze 608: /*
1.36 schwarze 609: * Check if the string consists only of space-separated closing
610: * delimiters. This is a bit of a dance: the first must be a close
611: * delimiter, but it may be followed by middle delimiters. Arbitrary
612: * whitespace may separate these tokens.
613: */
614: static int
1.38 schwarze 615: args_checkpunct(const char *buf, int i)
1.36 schwarze 616: {
1.38 schwarze 617: int j;
618: char dbuf[DELIMSZ];
1.36 schwarze 619: enum mdelim d;
620:
621: /* First token must be a close-delimiter. */
622:
1.38 schwarze 623: for (j = 0; buf[i] && ' ' != buf[i] && j < DELIMSZ; j++, i++)
624: dbuf[j] = buf[i];
1.1 kristaps 625:
1.36 schwarze 626: if (DELIMSZ == j)
627: return(0);
628:
1.38 schwarze 629: dbuf[j] = '\0';
630: if (DELIM_CLOSE != mdoc_isdelim(dbuf))
1.36 schwarze 631: return(0);
632:
1.38 schwarze 633: while (' ' == buf[i])
1.36 schwarze 634: i++;
635:
636: /* Remaining must NOT be open/none. */
1.49 schwarze 637:
1.38 schwarze 638: while (buf[i]) {
1.36 schwarze 639: j = 0;
1.38 schwarze 640: while (buf[i] && ' ' != buf[i] && j < DELIMSZ)
641: dbuf[j++] = buf[i++];
1.36 schwarze 642:
643: if (DELIMSZ == j)
644: return(0);
645:
1.38 schwarze 646: dbuf[j] = '\0';
647: d = mdoc_isdelim(dbuf);
1.36 schwarze 648: if (DELIM_NONE == d || DELIM_OPEN == d)
649: return(0);
650:
1.38 schwarze 651: while (' ' == buf[i])
1.36 schwarze 652: i++;
653: }
654:
1.38 schwarze 655: return('\0' == buf[i]);
1.36 schwarze 656: }
657:
1.53 schwarze 658: static void
1.49 schwarze 659: argv_multi(struct mdoc *mdoc, int line,
1.1 kristaps 660: struct mdoc_argv *v, int *pos, char *buf)
661: {
1.24 schwarze 662: enum margserr ac;
1.1 kristaps 663: char *p;
664:
665: for (v->sz = 0; ; v->sz++) {
1.53 schwarze 666: if (buf[*pos] == '-')
1.1 kristaps 667: break;
1.42 schwarze 668: ac = args(mdoc, line, pos, buf, ARGSFL_NONE, &p);
1.53 schwarze 669: if (ac == ARGS_EOLN)
1.1 kristaps 670: break;
671:
1.53 schwarze 672: if (v->sz % MULTI_STEP == 0)
1.50 schwarze 673: v->value = mandoc_reallocarray(v->value,
674: v->sz + MULTI_STEP, sizeof(char *));
1.19 schwarze 675:
676: v->value[(int)v->sz] = mandoc_strdup(p);
1.1 kristaps 677: }
678: }
679:
1.53 schwarze 680: static void
1.49 schwarze 681: argv_single(struct mdoc *mdoc, int line,
1.1 kristaps 682: struct mdoc_argv *v, int *pos, char *buf)
683: {
1.24 schwarze 684: enum margserr ac;
1.1 kristaps 685: char *p;
686:
1.42 schwarze 687: ac = args(mdoc, line, pos, buf, ARGSFL_NONE, &p);
1.53 schwarze 688: if (ac == ARGS_EOLN)
689: return;
1.1 kristaps 690:
691: v->sz = 1;
1.19 schwarze 692: v->value = mandoc_malloc(sizeof(char *));
693: v->value[0] = mandoc_strdup(p);
1.1 kristaps 694: }