Annotation of mandoc/mdoc_argv.c, Revision 1.120
1.120 ! schwarze 1: /* $Id: mdoc_argv.c,v 1.119 2018/12/21 17:15:19 schwarze Exp $ */
1.1 kristaps 2: /*
1.81 schwarze 3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.120 ! schwarze 4: * Copyright (c) 2012, 2014-2019 Ingo Schwarze <schwarze@openbsd.org>
1.1 kristaps 5: *
6: * Permission to use, copy, modify, and distribute this software for any
1.3 kristaps 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: *
1.101 schwarze 10: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.3 kristaps 11: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.101 schwarze 12: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.3 kristaps 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: */
1.33 kristaps 18: #include "config.h"
19:
1.1 kristaps 20: #include <sys/types.h>
21:
22: #include <assert.h>
23: #include <stdlib.h>
24: #include <stdio.h>
25: #include <string.h>
26:
1.101 schwarze 27: #include "mandoc_aux.h"
28: #include "mandoc.h"
29: #include "roff.h"
1.72 kristaps 30: #include "mdoc.h"
1.101 schwarze 31: #include "libmandoc.h"
1.108 schwarze 32: #include "roff_int.h"
1.1 kristaps 33: #include "libmdoc.h"
34:
1.68 kristaps 35: #define MULTI_STEP 5 /* pre-allocate argument values */
1.92 schwarze 36: #define DELIMSZ 6 /* max possible size of a delimiter */
1.76 kristaps 37:
38: enum argsflag {
39: ARGSFL_NONE = 0,
40: ARGSFL_DELIM, /* handle delimiters of [[::delim::][ ]+]+ */
41: ARGSFL_TABSEP /* handle tab/`Ta' separated phrases */
42: };
43:
44: enum argvflag {
45: ARGV_NONE, /* no args to flag (e.g., -split) */
46: ARGV_SINGLE, /* one arg to flag (e.g., -file xxx) */
1.83 schwarze 47: ARGV_MULTI /* multiple args (e.g., -column xxx yyy) */
1.76 kristaps 48: };
1.1 kristaps 49:
1.78 kristaps 50: struct mdocarg {
51: enum argsflag flags;
52: const enum mdocargt *argvs;
53: };
54:
1.80 kristaps 55: static void argn_free(struct mdoc_arg *, int);
1.103 schwarze 56: static enum margserr args(struct roff_man *, int, int *,
1.76 kristaps 57: char *, enum argsflag, char **);
1.77 kristaps 58: static int args_checkpunct(const char *, int);
1.103 schwarze 59: static void argv_multi(struct roff_man *, int,
1.1 kristaps 60: struct mdoc_argv *, int *, char *);
1.103 schwarze 61: static void argv_single(struct roff_man *, int,
1.1 kristaps 62: struct mdoc_argv *, int *, char *);
1.2 kristaps 63:
1.68 kristaps 64: static const enum argvflag argvflags[MDOC_ARG_MAX] = {
1.1 kristaps 65: ARGV_NONE, /* MDOC_Split */
66: ARGV_NONE, /* MDOC_Nosplit */
67: ARGV_NONE, /* MDOC_Ragged */
68: ARGV_NONE, /* MDOC_Unfilled */
69: ARGV_NONE, /* MDOC_Literal */
1.29 kristaps 70: ARGV_SINGLE, /* MDOC_File */
1.83 schwarze 71: ARGV_SINGLE, /* MDOC_Offset */
1.1 kristaps 72: ARGV_NONE, /* MDOC_Bullet */
73: ARGV_NONE, /* MDOC_Dash */
74: ARGV_NONE, /* MDOC_Hyphen */
75: ARGV_NONE, /* MDOC_Item */
76: ARGV_NONE, /* MDOC_Enum */
77: ARGV_NONE, /* MDOC_Tag */
78: ARGV_NONE, /* MDOC_Diag */
79: ARGV_NONE, /* MDOC_Hang */
80: ARGV_NONE, /* MDOC_Ohang */
81: ARGV_NONE, /* MDOC_Inset */
82: ARGV_MULTI, /* MDOC_Column */
1.83 schwarze 83: ARGV_SINGLE, /* MDOC_Width */
1.1 kristaps 84: ARGV_NONE, /* MDOC_Compact */
1.25 kristaps 85: ARGV_NONE, /* MDOC_Std */
1.1 kristaps 86: ARGV_NONE, /* MDOC_Filled */
87: ARGV_NONE, /* MDOC_Words */
88: ARGV_NONE, /* MDOC_Emphasis */
89: ARGV_NONE, /* MDOC_Symbolic */
90: ARGV_NONE /* MDOC_Symbolic */
91: };
92:
1.73 kristaps 93: static const enum mdocargt args_Ex[] = {
94: MDOC_Std,
95: MDOC_ARG_MAX
96: };
97:
98: static const enum mdocargt args_An[] = {
99: MDOC_Split,
100: MDOC_Nosplit,
101: MDOC_ARG_MAX
102: };
103:
104: static const enum mdocargt args_Bd[] = {
105: MDOC_Ragged,
106: MDOC_Unfilled,
107: MDOC_Filled,
108: MDOC_Literal,
109: MDOC_File,
110: MDOC_Offset,
111: MDOC_Compact,
112: MDOC_Centred,
113: MDOC_ARG_MAX
114: };
115:
116: static const enum mdocargt args_Bf[] = {
117: MDOC_Emphasis,
118: MDOC_Literal,
119: MDOC_Symbolic,
120: MDOC_ARG_MAX
121: };
122:
123: static const enum mdocargt args_Bk[] = {
124: MDOC_Words,
125: MDOC_ARG_MAX
126: };
127:
128: static const enum mdocargt args_Bl[] = {
129: MDOC_Bullet,
130: MDOC_Dash,
131: MDOC_Hyphen,
132: MDOC_Item,
133: MDOC_Enum,
134: MDOC_Tag,
135: MDOC_Diag,
136: MDOC_Hang,
137: MDOC_Ohang,
138: MDOC_Inset,
139: MDOC_Column,
140: MDOC_Width,
141: MDOC_Offset,
142: MDOC_Compact,
143: MDOC_Nested,
144: MDOC_ARG_MAX
145: };
146:
1.116 schwarze 147: static const struct mdocarg mdocargs[MDOC_MAX - MDOC_Dd] = {
1.78 kristaps 148: { ARGSFL_NONE, NULL }, /* Dd */
149: { ARGSFL_NONE, NULL }, /* Dt */
150: { ARGSFL_NONE, NULL }, /* Os */
151: { ARGSFL_NONE, NULL }, /* Sh */
1.92 schwarze 152: { ARGSFL_NONE, NULL }, /* Ss */
153: { ARGSFL_NONE, NULL }, /* Pp */
1.78 kristaps 154: { ARGSFL_DELIM, NULL }, /* D1 */
155: { ARGSFL_DELIM, NULL }, /* Dl */
156: { ARGSFL_NONE, args_Bd }, /* Bd */
157: { ARGSFL_NONE, NULL }, /* Ed */
158: { ARGSFL_NONE, args_Bl }, /* Bl */
159: { ARGSFL_NONE, NULL }, /* El */
160: { ARGSFL_NONE, NULL }, /* It */
1.92 schwarze 161: { ARGSFL_DELIM, NULL }, /* Ad */
1.78 kristaps 162: { ARGSFL_DELIM, args_An }, /* An */
1.111 schwarze 163: { ARGSFL_DELIM, NULL }, /* Ap */
1.78 kristaps 164: { ARGSFL_DELIM, NULL }, /* Ar */
1.86 schwarze 165: { ARGSFL_DELIM, NULL }, /* Cd */
1.78 kristaps 166: { ARGSFL_DELIM, NULL }, /* Cm */
1.92 schwarze 167: { ARGSFL_DELIM, NULL }, /* Dv */
168: { ARGSFL_DELIM, NULL }, /* Er */
169: { ARGSFL_DELIM, NULL }, /* Ev */
1.78 kristaps 170: { ARGSFL_NONE, args_Ex }, /* Ex */
1.92 schwarze 171: { ARGSFL_DELIM, NULL }, /* Fa */
172: { ARGSFL_NONE, NULL }, /* Fd */
1.78 kristaps 173: { ARGSFL_DELIM, NULL }, /* Fl */
1.92 schwarze 174: { ARGSFL_DELIM, NULL }, /* Fn */
175: { ARGSFL_DELIM, NULL }, /* Ft */
176: { ARGSFL_DELIM, NULL }, /* Ic */
177: { ARGSFL_DELIM, NULL }, /* In */
1.78 kristaps 178: { ARGSFL_DELIM, NULL }, /* Li */
1.92 schwarze 179: { ARGSFL_NONE, NULL }, /* Nd */
180: { ARGSFL_DELIM, NULL }, /* Nm */
1.78 kristaps 181: { ARGSFL_DELIM, NULL }, /* Op */
1.94 schwarze 182: { ARGSFL_DELIM, NULL }, /* Ot */
1.78 kristaps 183: { ARGSFL_DELIM, NULL }, /* Pa */
184: { ARGSFL_NONE, args_Ex }, /* Rv */
1.92 schwarze 185: { ARGSFL_DELIM, NULL }, /* St */
1.78 kristaps 186: { ARGSFL_DELIM, NULL }, /* Va */
1.92 schwarze 187: { ARGSFL_DELIM, NULL }, /* Vt */
1.78 kristaps 188: { ARGSFL_DELIM, NULL }, /* Xr */
189: { ARGSFL_NONE, NULL }, /* %A */
190: { ARGSFL_NONE, NULL }, /* %B */
191: { ARGSFL_NONE, NULL }, /* %D */
192: { ARGSFL_NONE, NULL }, /* %I */
193: { ARGSFL_NONE, NULL }, /* %J */
194: { ARGSFL_NONE, NULL }, /* %N */
195: { ARGSFL_NONE, NULL }, /* %O */
196: { ARGSFL_NONE, NULL }, /* %P */
197: { ARGSFL_NONE, NULL }, /* %R */
198: { ARGSFL_NONE, NULL }, /* %T */
199: { ARGSFL_NONE, NULL }, /* %V */
200: { ARGSFL_DELIM, NULL }, /* Ac */
201: { ARGSFL_NONE, NULL }, /* Ao */
202: { ARGSFL_DELIM, NULL }, /* Aq */
203: { ARGSFL_DELIM, NULL }, /* At */
204: { ARGSFL_DELIM, NULL }, /* Bc */
1.92 schwarze 205: { ARGSFL_NONE, args_Bf }, /* Bf */
1.78 kristaps 206: { ARGSFL_NONE, NULL }, /* Bo */
207: { ARGSFL_DELIM, NULL }, /* Bq */
208: { ARGSFL_DELIM, NULL }, /* Bsx */
209: { ARGSFL_DELIM, NULL }, /* Bx */
210: { ARGSFL_NONE, NULL }, /* Db */
211: { ARGSFL_DELIM, NULL }, /* Dc */
212: { ARGSFL_NONE, NULL }, /* Do */
213: { ARGSFL_DELIM, NULL }, /* Dq */
214: { ARGSFL_DELIM, NULL }, /* Ec */
215: { ARGSFL_NONE, NULL }, /* Ef */
1.92 schwarze 216: { ARGSFL_DELIM, NULL }, /* Em */
1.78 kristaps 217: { ARGSFL_NONE, NULL }, /* Eo */
218: { ARGSFL_DELIM, NULL }, /* Fx */
219: { ARGSFL_DELIM, NULL }, /* Ms */
220: { ARGSFL_DELIM, NULL }, /* No */
221: { ARGSFL_DELIM, NULL }, /* Ns */
222: { ARGSFL_DELIM, NULL }, /* Nx */
223: { ARGSFL_DELIM, NULL }, /* Ox */
224: { ARGSFL_DELIM, NULL }, /* Pc */
225: { ARGSFL_DELIM, NULL }, /* Pf */
226: { ARGSFL_NONE, NULL }, /* Po */
227: { ARGSFL_DELIM, NULL }, /* Pq */
228: { ARGSFL_DELIM, NULL }, /* Qc */
229: { ARGSFL_DELIM, NULL }, /* Ql */
230: { ARGSFL_NONE, NULL }, /* Qo */
231: { ARGSFL_DELIM, NULL }, /* Qq */
232: { ARGSFL_NONE, NULL }, /* Re */
233: { ARGSFL_NONE, NULL }, /* Rs */
234: { ARGSFL_DELIM, NULL }, /* Sc */
235: { ARGSFL_NONE, NULL }, /* So */
236: { ARGSFL_DELIM, NULL }, /* Sq */
237: { ARGSFL_NONE, NULL }, /* Sm */
238: { ARGSFL_DELIM, NULL }, /* Sx */
239: { ARGSFL_DELIM, NULL }, /* Sy */
240: { ARGSFL_DELIM, NULL }, /* Tn */
241: { ARGSFL_DELIM, NULL }, /* Ux */
242: { ARGSFL_DELIM, NULL }, /* Xc */
243: { ARGSFL_NONE, NULL }, /* Xo */
1.92 schwarze 244: { ARGSFL_NONE, NULL }, /* Fo */
245: { ARGSFL_DELIM, NULL }, /* Fc */
1.78 kristaps 246: { ARGSFL_NONE, NULL }, /* Oo */
247: { ARGSFL_DELIM, NULL }, /* Oc */
248: { ARGSFL_NONE, args_Bk }, /* Bk */
249: { ARGSFL_NONE, NULL }, /* Ek */
250: { ARGSFL_NONE, NULL }, /* Bt */
251: { ARGSFL_NONE, NULL }, /* Hf */
1.94 schwarze 252: { ARGSFL_DELIM, NULL }, /* Fr */
1.78 kristaps 253: { ARGSFL_NONE, NULL }, /* Ud */
1.86 schwarze 254: { ARGSFL_DELIM, NULL }, /* Lb */
1.78 kristaps 255: { ARGSFL_NONE, NULL }, /* Lp */
256: { ARGSFL_DELIM, NULL }, /* Lk */
257: { ARGSFL_DELIM, NULL }, /* Mt */
258: { ARGSFL_DELIM, NULL }, /* Brq */
259: { ARGSFL_NONE, NULL }, /* Bro */
260: { ARGSFL_DELIM, NULL }, /* Brc */
261: { ARGSFL_NONE, NULL }, /* %C */
262: { ARGSFL_NONE, NULL }, /* Es */
1.94 schwarze 263: { ARGSFL_DELIM, NULL }, /* En */
1.88 schwarze 264: { ARGSFL_DELIM, NULL }, /* Dx */
1.78 kristaps 265: { ARGSFL_NONE, NULL }, /* %Q */
266: { ARGSFL_NONE, NULL }, /* %U */
267: { ARGSFL_NONE, NULL }, /* Ta */
268: };
269:
270:
1.1 kristaps 271: /*
1.98 schwarze 272: * Parse flags and their arguments from the input line.
273: * These come in the form -flag [argument ...].
274: * Some flags take no argument, some one, some multiple.
1.1 kristaps 275: */
1.98 schwarze 276: void
1.111 schwarze 277: mdoc_argv(struct roff_man *mdoc, int line, enum roff_tok tok,
1.98 schwarze 278: struct mdoc_arg **reta, int *pos, char *buf)
1.1 kristaps 279: {
1.98 schwarze 280: struct mdoc_argv tmpv;
281: struct mdoc_argv **retv;
282: const enum mdocargt *argtable;
283: char *argname;
284: int ipos, retc;
285: char savechar;
1.1 kristaps 286:
1.98 schwarze 287: *reta = NULL;
1.1 kristaps 288:
1.98 schwarze 289: /* Which flags does this macro support? */
1.1 kristaps 290:
1.111 schwarze 291: assert(tok >= MDOC_Dd && tok < MDOC_MAX);
1.116 schwarze 292: argtable = mdocargs[tok - MDOC_Dd].argvs;
1.98 schwarze 293: if (argtable == NULL)
294: return;
1.1 kristaps 295:
1.98 schwarze 296: /* Loop over the flags on the input line. */
297:
298: ipos = *pos;
299: while (buf[ipos] == '-') {
300:
301: /* Seek to the first unescaped space. */
1.1 kristaps 302:
1.98 schwarze 303: for (argname = buf + ++ipos; buf[ipos] != '\0'; ipos++)
304: if (buf[ipos] == ' ' && buf[ipos - 1] != '\\')
305: break;
1.1 kristaps 306:
1.98 schwarze 307: /*
308: * We want to nil-terminate the word to look it up.
309: * But we may not have a flag, in which case we need
310: * to restore the line as-is. So keep around the
311: * stray byte, which we'll reset upon exiting.
312: */
313:
314: if ((savechar = buf[ipos]) != '\0')
315: buf[ipos++] = '\0';
1.1 kristaps 316:
1.92 schwarze 317: /*
1.98 schwarze 318: * Now look up the word as a flag. Use temporary
319: * storage that we'll copy into the node's flags.
1.78 kristaps 320: */
1.1 kristaps 321:
1.98 schwarze 322: while ((tmpv.arg = *argtable++) != MDOC_ARG_MAX)
323: if ( ! strcmp(argname, mdoc_argnames[tmpv.arg]))
324: break;
325:
326: /* If it isn't a flag, restore the saved byte. */
327:
328: if (tmpv.arg == MDOC_ARG_MAX) {
329: if (savechar != '\0')
330: buf[ipos - 1] = savechar;
331: break;
332: }
333:
334: /* Read to the next word (the first argument). */
335:
336: while (buf[ipos] == ' ')
337: ipos++;
338:
339: /* Parse the arguments of the flag. */
1.78 kristaps 340:
1.98 schwarze 341: tmpv.line = line;
1.100 schwarze 342: tmpv.pos = *pos;
1.98 schwarze 343: tmpv.sz = 0;
344: tmpv.value = NULL;
1.1 kristaps 345:
1.98 schwarze 346: switch (argvflags[tmpv.arg]) {
347: case ARGV_SINGLE:
348: argv_single(mdoc, line, &tmpv, &ipos, buf);
349: break;
350: case ARGV_MULTI:
351: argv_multi(mdoc, line, &tmpv, &ipos, buf);
352: break;
353: case ARGV_NONE:
354: break;
355: }
356:
357: /* Append to the return values. */
1.1 kristaps 358:
1.98 schwarze 359: if (*reta == NULL)
360: *reta = mandoc_calloc(1, sizeof(**reta));
1.1 kristaps 361:
1.98 schwarze 362: retc = ++(*reta)->argc;
363: retv = &(*reta)->argv;
364: *retv = mandoc_reallocarray(*retv, retc, sizeof(**retv));
365: memcpy(*retv + retc - 1, &tmpv, sizeof(**retv));
1.2 kristaps 366:
1.98 schwarze 367: /* Prepare for parsing the next flag. */
1.1 kristaps 368:
1.98 schwarze 369: *pos = ipos;
1.116 schwarze 370: argtable = mdocargs[tok - MDOC_Dd].argvs;
1.98 schwarze 371: }
1.1 kristaps 372: }
373:
374: void
375: mdoc_argv_free(struct mdoc_arg *p)
376: {
1.37 kristaps 377: int i;
1.1 kristaps 378:
379: if (NULL == p)
380: return;
381:
382: if (p->refcnt) {
383: --(p->refcnt);
384: if (p->refcnt)
385: return;
386: }
387: assert(p->argc);
388:
1.37 kristaps 389: for (i = (int)p->argc - 1; i >= 0; i--)
1.71 kristaps 390: argn_free(p, i);
1.37 kristaps 391:
392: free(p->argv);
393: free(p);
394: }
395:
1.71 kristaps 396: static void
397: argn_free(struct mdoc_arg *p, int iarg)
1.37 kristaps 398: {
1.58 kristaps 399: struct mdoc_argv *arg;
1.37 kristaps 400: int j;
1.58 kristaps 401:
402: arg = &p->argv[iarg];
1.1 kristaps 403:
1.37 kristaps 404: if (arg->sz && arg->value) {
1.92 schwarze 405: for (j = (int)arg->sz - 1; j >= 0; j--)
1.37 kristaps 406: free(arg->value[j]);
407: free(arg->value);
1.1 kristaps 408: }
409:
1.37 kristaps 410: for (--p->argc; iarg < (int)p->argc; iarg++)
411: p->argv[iarg] = p->argv[iarg+1];
1.1 kristaps 412: }
413:
1.39 kristaps 414: enum margserr
1.103 schwarze 415: mdoc_args(struct roff_man *mdoc, int line, int *pos,
1.111 schwarze 416: char *buf, enum roff_tok tok, char **v)
1.1 kristaps 417: {
1.102 schwarze 418: struct roff_node *n;
1.76 kristaps 419: enum argsflag fl;
1.1 kristaps 420:
1.116 schwarze 421: fl = tok == TOKEN_NONE ? ARGSFL_NONE : mdocargs[tok - MDOC_Dd].flags;
1.17 kristaps 422:
1.51 kristaps 423: /*
424: * We know that we're in an `It', so it's reasonable to expect
425: * us to be sitting in a `Bl'. Someday this may not be the case
426: * (if we allow random `It's sitting out there), so provide a
427: * safe fall-back into the default behaviour.
428: */
429:
1.110 schwarze 430: if (tok == MDOC_It) {
431: for (n = mdoc->last; n != NULL; n = n->parent) {
432: if (n->tok != MDOC_Bl)
433: continue;
434: if (n->norm->Bl.type == LIST_column)
1.76 kristaps 435: fl = ARGSFL_TABSEP;
1.110 schwarze 436: break;
437: }
438: }
1.1 kristaps 439:
1.105 schwarze 440: return args(mdoc, line, pos, buf, fl, v);
1.1 kristaps 441: }
442:
1.39 kristaps 443: static enum margserr
1.103 schwarze 444: args(struct roff_man *mdoc, int line, int *pos,
1.76 kristaps 445: char *buf, enum argsflag fl, char **v)
1.1 kristaps 446: {
1.106 schwarze 447: char *p;
1.119 schwarze 448: char *v_local;
1.87 schwarze 449: int pairs;
1.1 kristaps 450:
1.106 schwarze 451: if (buf[*pos] == '\0') {
452: if (mdoc->flags & MDOC_PHRASELIT &&
1.107 schwarze 453: ! (mdoc->flags & MDOC_PHRASE)) {
1.117 schwarze 454: mandoc_msg(MANDOCERR_ARG_QUOTE, line, *pos, NULL);
1.106 schwarze 455: mdoc->flags &= ~MDOC_PHRASELIT;
456: }
1.120 ! schwarze 457: mdoc->flags &= ~MDOC_PHRASEQL;
1.105 schwarze 458: return ARGS_EOLN;
1.46 kristaps 459: }
1.1 kristaps 460:
1.119 schwarze 461: if (v == NULL)
462: v = &v_local;
1.106 schwarze 463: *v = buf + *pos;
1.17 kristaps 464:
1.106 schwarze 465: if (fl == ARGSFL_DELIM && args_checkpunct(buf, *pos))
466: return ARGS_PUNCT;
1.1 kristaps 467:
1.17 kristaps 468: /*
1.106 schwarze 469: * Tabs in `It' lines in `Bl -column' can't be escaped.
470: * Phrases are reparsed for `Ta' and other macros later.
1.17 kristaps 471: */
1.1 kristaps 472:
1.106 schwarze 473: if (fl == ARGSFL_TABSEP) {
474: if ((p = strchr(*v, '\t')) != NULL) {
1.107 schwarze 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:
1.109 schwarze 482: if (p > buf && p[-1] != ' ')
1.107 schwarze 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:
1.106 schwarze 493: *pos += (int)(p - *v) + 1;
1.107 schwarze 494: while (buf[*pos] == ' ' && buf[*pos + 1] == ' ')
1.106 schwarze 495: (*pos)++;
1.107 schwarze 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;
1.44 kristaps 504: } else {
1.106 schwarze 505: p = strchr(*v, '\0');
506: if (p[-1] == ' ')
507: mandoc_msg(MANDOCERR_SPACE_EOL,
1.117 schwarze 508: line, *pos, NULL);
1.106 schwarze 509: *pos += (int)(p - *v);
1.17 kristaps 510: }
511:
1.106 schwarze 512: /* Skip any trailing blank characters. */
513: while (p > *v && p[-1] == ' ' &&
514: (p - 1 == *v || p[-2] != '\\'))
515: p--;
516: *p = '\0';
1.1 kristaps 517:
1.107 schwarze 518: return ARGS_PHRASE;
1.89 schwarze 519: }
1.1 kristaps 520:
1.89 schwarze 521: /*
1.17 kristaps 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.
1.89 schwarze 524: * NUL-terminate the literal in place.
1.87 schwarze 525: * Collapse pairs of quotes inside quoted literals.
1.17 kristaps 526: * Whitespace is NOT involved in literal termination.
527: */
1.1 kristaps 528:
1.119 schwarze 529: if (mdoc->flags & MDOC_PHRASELIT ||
530: (mdoc->flags & MDOC_PHRASE && buf[*pos] == '\"')) {
531: if ((mdoc->flags & MDOC_PHRASELIT) == 0) {
1.46 kristaps 532: *v = &buf[++(*pos)];
1.85 schwarze 533: mdoc->flags |= MDOC_PHRASELIT;
1.119 schwarze 534: }
1.87 schwarze 535: pairs = 0;
1.17 kristaps 536: for ( ; buf[*pos]; (*pos)++) {
1.87 schwarze 537: /* Move following text left after quoted quotes. */
538: if (pairs)
539: buf[*pos - pairs] = buf[*pos];
1.17 kristaps 540: if ('\"' != buf[*pos])
541: continue;
1.87 schwarze 542: /* Unquoted quotes end quoted args. */
1.17 kristaps 543: if ('\"' != buf[*pos + 1])
544: break;
1.87 schwarze 545: /* Quoted quotes collapse. */
546: pairs++;
1.17 kristaps 547: (*pos)++;
548: }
1.87 schwarze 549: if (pairs)
550: buf[*pos - pairs] = '\0';
1.1 kristaps 551:
1.107 schwarze 552: if (buf[*pos] == '\0') {
553: if ( ! (mdoc->flags & MDOC_PHRASE))
554: mandoc_msg(MANDOCERR_ARG_QUOTE,
1.117 schwarze 555: line, *pos, NULL);
1.115 schwarze 556: return ARGS_WORD;
1.17 kristaps 557: }
1.1 kristaps 558:
1.85 schwarze 559: mdoc->flags &= ~MDOC_PHRASELIT;
1.46 kristaps 560: buf[(*pos)++] = '\0';
1.1 kristaps 561:
1.46 kristaps 562: if ('\0' == buf[*pos])
1.115 schwarze 563: return ARGS_WORD;
1.1 kristaps 564:
1.17 kristaps 565: while (' ' == buf[*pos])
566: (*pos)++;
1.1 kristaps 567:
1.76 kristaps 568: if ('\0' == buf[*pos])
1.117 schwarze 569: mandoc_msg(MANDOCERR_SPACE_EOL, line, *pos, NULL);
1.1 kristaps 570:
1.115 schwarze 571: return ARGS_WORD;
1.1 kristaps 572: }
573:
1.75 kristaps 574: p = &buf[*pos];
1.119 schwarze 575: *v = roff_getarg(mdoc->roff, &p, line, pos);
576: if (v == &v_local)
577: free(*v);
1.1 kristaps 578:
1.107 schwarze 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: }
1.119 schwarze 588: return ARGS_ALLOC;
1.64 kristaps 589: }
590:
1.92 schwarze 591: /*
1.64 kristaps 592: * Check if the string consists only of space-separated closing
1.65 kristaps 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.
1.64 kristaps 596: */
597: static int
1.77 kristaps 598: args_checkpunct(const char *buf, int i)
1.64 kristaps 599: {
1.69 kristaps 600: int j;
601: char dbuf[DELIMSZ];
1.64 kristaps 602: enum mdelim d;
603:
1.65 kristaps 604: /* First token must be a close-delimiter. */
605:
1.69 kristaps 606: for (j = 0; buf[i] && ' ' != buf[i] && j < DELIMSZ; j++, i++)
607: dbuf[j] = buf[i];
1.65 kristaps 608:
609: if (DELIMSZ == j)
1.105 schwarze 610: return 0;
1.65 kristaps 611:
1.69 kristaps 612: dbuf[j] = '\0';
1.70 kristaps 613: if (DELIM_CLOSE != mdoc_isdelim(dbuf))
1.105 schwarze 614: return 0;
1.64 kristaps 615:
1.69 kristaps 616: while (' ' == buf[i])
1.65 kristaps 617: i++;
618:
619: /* Remaining must NOT be open/none. */
1.92 schwarze 620:
1.69 kristaps 621: while (buf[i]) {
1.65 kristaps 622: j = 0;
1.69 kristaps 623: while (buf[i] && ' ' != buf[i] && j < DELIMSZ)
624: dbuf[j++] = buf[i++];
1.65 kristaps 625:
626: if (DELIMSZ == j)
1.105 schwarze 627: return 0;
1.65 kristaps 628:
1.69 kristaps 629: dbuf[j] = '\0';
1.70 kristaps 630: d = mdoc_isdelim(dbuf);
1.64 kristaps 631: if (DELIM_NONE == d || DELIM_OPEN == d)
1.105 schwarze 632: return 0;
1.65 kristaps 633:
1.69 kristaps 634: while (' ' == buf[i])
1.64 kristaps 635: i++;
636: }
1.69 kristaps 637:
1.105 schwarze 638: return '\0' == buf[i];
1.1 kristaps 639: }
640:
1.97 schwarze 641: static void
1.103 schwarze 642: argv_multi(struct roff_man *mdoc, int line,
1.1 kristaps 643: struct mdoc_argv *v, int *pos, char *buf)
644: {
1.43 kristaps 645: enum margserr ac;
1.1 kristaps 646: char *p;
647:
648: for (v->sz = 0; ; v->sz++) {
1.97 schwarze 649: if (buf[*pos] == '-')
1.1 kristaps 650: break;
1.85 schwarze 651: ac = args(mdoc, line, pos, buf, ARGSFL_NONE, &p);
1.97 schwarze 652: if (ac == ARGS_EOLN)
1.1 kristaps 653: break;
654:
1.97 schwarze 655: if (v->sz % MULTI_STEP == 0)
1.93 schwarze 656: v->value = mandoc_reallocarray(v->value,
657: v->sz + MULTI_STEP, sizeof(char *));
1.31 kristaps 658:
1.119 schwarze 659: if (ac != ARGS_ALLOC)
660: p = mandoc_strdup(p);
661: v->value[(int)v->sz] = p;
1.1 kristaps 662: }
663: }
664:
1.97 schwarze 665: static void
1.103 schwarze 666: argv_single(struct roff_man *mdoc, int line,
1.1 kristaps 667: struct mdoc_argv *v, int *pos, char *buf)
668: {
1.43 kristaps 669: enum margserr ac;
1.1 kristaps 670: char *p;
671:
1.85 schwarze 672: ac = args(mdoc, line, pos, buf, ARGSFL_NONE, &p);
1.97 schwarze 673: if (ac == ARGS_EOLN)
674: return;
1.1 kristaps 675:
1.119 schwarze 676: if (ac != ARGS_ALLOC)
677: p = mandoc_strdup(p);
678:
1.1 kristaps 679: v->sz = 1;
1.31 kristaps 680: v->value = mandoc_malloc(sizeof(char *));
1.119 schwarze 681: v->value[0] = p;
1.1 kristaps 682: }
CVSweb