Annotation of mandoc/mdoc_argv.c, Revision 1.108
1.108 ! schwarze 1: /* $Id: mdoc_argv.c,v 1.107 2015/10/17 00:21:07 schwarze Exp $ */
1.1 kristaps 2: /*
1.81 schwarze 3: * Copyright (c) 2008, 2009, 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.103 schwarze 4: * Copyright (c) 2012, 2014, 2015 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:
1.78 kristaps 147: static const struct mdocarg mdocargs[MDOC_MAX] = {
1.84 schwarze 148: { ARGSFL_DELIM, NULL }, /* Ap */
1.78 kristaps 149: { ARGSFL_NONE, NULL }, /* Dd */
150: { ARGSFL_NONE, NULL }, /* Dt */
151: { ARGSFL_NONE, NULL }, /* Os */
152: { ARGSFL_NONE, NULL }, /* Sh */
1.92 schwarze 153: { ARGSFL_NONE, NULL }, /* Ss */
154: { ARGSFL_NONE, NULL }, /* Pp */
1.78 kristaps 155: { ARGSFL_DELIM, NULL }, /* D1 */
156: { ARGSFL_DELIM, NULL }, /* Dl */
157: { ARGSFL_NONE, args_Bd }, /* Bd */
158: { ARGSFL_NONE, NULL }, /* Ed */
159: { ARGSFL_NONE, args_Bl }, /* Bl */
160: { ARGSFL_NONE, NULL }, /* El */
161: { ARGSFL_NONE, NULL }, /* It */
1.92 schwarze 162: { ARGSFL_DELIM, NULL }, /* Ad */
1.78 kristaps 163: { ARGSFL_DELIM, args_An }, /* An */
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 }, /* br */
267: { ARGSFL_NONE, NULL }, /* sp */
268: { ARGSFL_NONE, NULL }, /* %U */
269: { ARGSFL_NONE, NULL }, /* Ta */
1.91 schwarze 270: { ARGSFL_NONE, NULL }, /* ll */
1.78 kristaps 271: };
272:
273:
1.1 kristaps 274: /*
1.98 schwarze 275: * Parse flags and their arguments from the input line.
276: * These come in the form -flag [argument ...].
277: * Some flags take no argument, some one, some multiple.
1.1 kristaps 278: */
1.98 schwarze 279: void
1.103 schwarze 280: mdoc_argv(struct roff_man *mdoc, int line, int tok,
1.98 schwarze 281: struct mdoc_arg **reta, int *pos, char *buf)
1.1 kristaps 282: {
1.98 schwarze 283: struct mdoc_argv tmpv;
284: struct mdoc_argv **retv;
285: const enum mdocargt *argtable;
286: char *argname;
287: int ipos, retc;
288: char savechar;
1.1 kristaps 289:
1.98 schwarze 290: *reta = NULL;
1.1 kristaps 291:
1.98 schwarze 292: /* Which flags does this macro support? */
1.1 kristaps 293:
1.98 schwarze 294: argtable = mdocargs[tok].argvs;
295: if (argtable == NULL)
296: return;
1.1 kristaps 297:
1.98 schwarze 298: /* Loop over the flags on the input line. */
299:
300: ipos = *pos;
301: while (buf[ipos] == '-') {
302:
303: /* Seek to the first unescaped space. */
1.1 kristaps 304:
1.98 schwarze 305: for (argname = buf + ++ipos; buf[ipos] != '\0'; ipos++)
306: if (buf[ipos] == ' ' && buf[ipos - 1] != '\\')
307: break;
1.1 kristaps 308:
1.98 schwarze 309: /*
310: * We want to nil-terminate the word to look it up.
311: * But we may not have a flag, in which case we need
312: * to restore the line as-is. So keep around the
313: * stray byte, which we'll reset upon exiting.
314: */
315:
316: if ((savechar = buf[ipos]) != '\0')
317: buf[ipos++] = '\0';
1.1 kristaps 318:
1.92 schwarze 319: /*
1.98 schwarze 320: * Now look up the word as a flag. Use temporary
321: * storage that we'll copy into the node's flags.
1.78 kristaps 322: */
1.1 kristaps 323:
1.98 schwarze 324: while ((tmpv.arg = *argtable++) != MDOC_ARG_MAX)
325: if ( ! strcmp(argname, mdoc_argnames[tmpv.arg]))
326: break;
327:
328: /* If it isn't a flag, restore the saved byte. */
329:
330: if (tmpv.arg == MDOC_ARG_MAX) {
331: if (savechar != '\0')
332: buf[ipos - 1] = savechar;
333: break;
334: }
335:
336: /* Read to the next word (the first argument). */
337:
338: while (buf[ipos] == ' ')
339: ipos++;
340:
341: /* Parse the arguments of the flag. */
1.78 kristaps 342:
1.98 schwarze 343: tmpv.line = line;
1.100 schwarze 344: tmpv.pos = *pos;
1.98 schwarze 345: tmpv.sz = 0;
346: tmpv.value = NULL;
1.1 kristaps 347:
1.98 schwarze 348: switch (argvflags[tmpv.arg]) {
349: case ARGV_SINGLE:
350: argv_single(mdoc, line, &tmpv, &ipos, buf);
351: break;
352: case ARGV_MULTI:
353: argv_multi(mdoc, line, &tmpv, &ipos, buf);
354: break;
355: case ARGV_NONE:
356: break;
357: }
358:
359: /* Append to the return values. */
1.1 kristaps 360:
1.98 schwarze 361: if (*reta == NULL)
362: *reta = mandoc_calloc(1, sizeof(**reta));
1.1 kristaps 363:
1.98 schwarze 364: retc = ++(*reta)->argc;
365: retv = &(*reta)->argv;
366: *retv = mandoc_reallocarray(*retv, retc, sizeof(**retv));
367: memcpy(*retv + retc - 1, &tmpv, sizeof(**retv));
1.2 kristaps 368:
1.98 schwarze 369: /* Prepare for parsing the next flag. */
1.1 kristaps 370:
1.98 schwarze 371: *pos = ipos;
372: argtable = mdocargs[tok].argvs;
373: }
1.1 kristaps 374: }
375:
376: void
377: mdoc_argv_free(struct mdoc_arg *p)
378: {
1.37 kristaps 379: int i;
1.1 kristaps 380:
381: if (NULL == p)
382: return;
383:
384: if (p->refcnt) {
385: --(p->refcnt);
386: if (p->refcnt)
387: return;
388: }
389: assert(p->argc);
390:
1.37 kristaps 391: for (i = (int)p->argc - 1; i >= 0; i--)
1.71 kristaps 392: argn_free(p, i);
1.37 kristaps 393:
394: free(p->argv);
395: free(p);
396: }
397:
1.71 kristaps 398: static void
399: argn_free(struct mdoc_arg *p, int iarg)
1.37 kristaps 400: {
1.58 kristaps 401: struct mdoc_argv *arg;
1.37 kristaps 402: int j;
1.58 kristaps 403:
404: arg = &p->argv[iarg];
1.1 kristaps 405:
1.37 kristaps 406: if (arg->sz && arg->value) {
1.92 schwarze 407: for (j = (int)arg->sz - 1; j >= 0; j--)
1.37 kristaps 408: free(arg->value[j]);
409: free(arg->value);
1.1 kristaps 410: }
411:
1.37 kristaps 412: for (--p->argc; iarg < (int)p->argc; iarg++)
413: p->argv[iarg] = p->argv[iarg+1];
1.1 kristaps 414: }
415:
1.39 kristaps 416: enum margserr
1.103 schwarze 417: mdoc_args(struct roff_man *mdoc, int line, int *pos,
1.102 schwarze 418: char *buf, int tok, char **v)
1.1 kristaps 419: {
1.102 schwarze 420: struct roff_node *n;
1.99 schwarze 421: char *v_local;
1.76 kristaps 422: enum argsflag fl;
1.1 kristaps 423:
1.99 schwarze 424: if (v == NULL)
425: v = &v_local;
1.104 schwarze 426: fl = tok == TOKEN_NONE ? ARGSFL_NONE : mdocargs[tok].flags;
1.99 schwarze 427: if (tok != MDOC_It)
1.105 schwarze 428: return args(mdoc, line, pos, buf, fl, v);
1.17 kristaps 429:
1.51 kristaps 430: /*
431: * We know that we're in an `It', so it's reasonable to expect
432: * us to be sitting in a `Bl'. Someday this may not be the case
433: * (if we allow random `It's sitting out there), so provide a
434: * safe fall-back into the default behaviour.
435: */
436:
1.85 schwarze 437: for (n = mdoc->last; n; n = n->parent)
1.50 kristaps 438: if (MDOC_Bl == n->tok)
1.76 kristaps 439: if (LIST_column == n->norm->Bl.type) {
440: fl = ARGSFL_TABSEP;
441: break;
442: }
1.1 kristaps 443:
1.105 schwarze 444: return args(mdoc, line, pos, buf, fl, v);
1.1 kristaps 445: }
446:
1.39 kristaps 447: static enum margserr
1.103 schwarze 448: args(struct roff_man *mdoc, int line, int *pos,
1.76 kristaps 449: char *buf, enum argsflag fl, char **v)
1.1 kristaps 450: {
1.106 schwarze 451: char *p;
1.87 schwarze 452: int pairs;
1.1 kristaps 453:
1.106 schwarze 454: if (buf[*pos] == '\0') {
455: if (mdoc->flags & MDOC_PHRASELIT &&
1.107 schwarze 456: ! (mdoc->flags & MDOC_PHRASE)) {
1.95 schwarze 457: mandoc_msg(MANDOCERR_ARG_QUOTE,
458: mdoc->parse, line, *pos, NULL);
1.106 schwarze 459: mdoc->flags &= ~MDOC_PHRASELIT;
460: }
1.105 schwarze 461: return ARGS_EOLN;
1.46 kristaps 462: }
1.1 kristaps 463:
1.106 schwarze 464: *v = buf + *pos;
1.17 kristaps 465:
1.106 schwarze 466: if (fl == ARGSFL_DELIM && args_checkpunct(buf, *pos))
467: return ARGS_PUNCT;
1.1 kristaps 468:
1.17 kristaps 469: /*
1.106 schwarze 470: * Tabs in `It' lines in `Bl -column' can't be escaped.
471: * Phrases are reparsed for `Ta' and other macros later.
1.17 kristaps 472: */
1.1 kristaps 473:
1.106 schwarze 474: if (fl == ARGSFL_TABSEP) {
475: if ((p = strchr(*v, '\t')) != NULL) {
1.107 schwarze 476:
477: /*
478: * Words right before and right after
479: * tab characters are not parsed,
480: * unless there is a blank in between.
481: */
482:
483: if (p[-1] != ' ')
484: mdoc->flags |= MDOC_PHRASEQL;
485: if (p[1] != ' ')
486: mdoc->flags |= MDOC_PHRASEQN;
487:
488: /*
489: * One or more blanks after a tab cause
490: * one leading blank in the next column.
491: * So skip all but one of them.
492: */
493:
1.106 schwarze 494: *pos += (int)(p - *v) + 1;
1.107 schwarze 495: while (buf[*pos] == ' ' && buf[*pos + 1] == ' ')
1.106 schwarze 496: (*pos)++;
1.107 schwarze 497:
498: /*
499: * A tab at the end of an input line
500: * switches to the next column.
501: */
502:
503: if (buf[*pos] == '\0' || buf[*pos + 1] == '\0')
504: mdoc->flags |= MDOC_PHRASEQN;
1.44 kristaps 505: } else {
1.106 schwarze 506: p = strchr(*v, '\0');
507: if (p[-1] == ' ')
508: mandoc_msg(MANDOCERR_SPACE_EOL,
509: mdoc->parse, line, *pos, NULL);
510: *pos += (int)(p - *v);
1.17 kristaps 511: }
512:
1.106 schwarze 513: /* Skip any trailing blank characters. */
514: while (p > *v && p[-1] == ' ' &&
515: (p - 1 == *v || p[-2] != '\\'))
516: p--;
517: *p = '\0';
1.1 kristaps 518:
1.107 schwarze 519: return ARGS_PHRASE;
1.89 schwarze 520: }
1.1 kristaps 521:
1.89 schwarze 522: /*
1.17 kristaps 523: * Process a quoted literal. A quote begins with a double-quote
524: * and ends with a double-quote NOT preceded by a double-quote.
1.89 schwarze 525: * NUL-terminate the literal in place.
1.87 schwarze 526: * Collapse pairs of quotes inside quoted literals.
1.17 kristaps 527: * Whitespace is NOT involved in literal termination.
528: */
1.1 kristaps 529:
1.107 schwarze 530: if (mdoc->flags & MDOC_PHRASELIT || buf[*pos] == '\"') {
531: if ( ! (mdoc->flags & MDOC_PHRASELIT))
1.46 kristaps 532: *v = &buf[++(*pos)];
533:
1.107 schwarze 534: if (mdoc->flags & MDOC_PHRASE)
1.85 schwarze 535: mdoc->flags |= MDOC_PHRASELIT;
1.1 kristaps 536:
1.87 schwarze 537: pairs = 0;
1.17 kristaps 538: for ( ; buf[*pos]; (*pos)++) {
1.87 schwarze 539: /* Move following text left after quoted quotes. */
540: if (pairs)
541: buf[*pos - pairs] = buf[*pos];
1.17 kristaps 542: if ('\"' != buf[*pos])
543: continue;
1.87 schwarze 544: /* Unquoted quotes end quoted args. */
1.17 kristaps 545: if ('\"' != buf[*pos + 1])
546: break;
1.87 schwarze 547: /* Quoted quotes collapse. */
548: pairs++;
1.17 kristaps 549: (*pos)++;
550: }
1.87 schwarze 551: if (pairs)
552: buf[*pos - pairs] = '\0';
1.1 kristaps 553:
1.107 schwarze 554: if (buf[*pos] == '\0') {
555: if ( ! (mdoc->flags & MDOC_PHRASE))
556: mandoc_msg(MANDOCERR_ARG_QUOTE,
557: mdoc->parse, line, *pos, NULL);
1.105 schwarze 558: return ARGS_QWORD;
1.17 kristaps 559: }
1.1 kristaps 560:
1.85 schwarze 561: mdoc->flags &= ~MDOC_PHRASELIT;
1.46 kristaps 562: buf[(*pos)++] = '\0';
1.1 kristaps 563:
1.46 kristaps 564: if ('\0' == buf[*pos])
1.105 schwarze 565: return ARGS_QWORD;
1.1 kristaps 566:
1.17 kristaps 567: while (' ' == buf[*pos])
568: (*pos)++;
1.1 kristaps 569:
1.76 kristaps 570: if ('\0' == buf[*pos])
1.95 schwarze 571: mandoc_msg(MANDOCERR_SPACE_EOL, mdoc->parse,
572: line, *pos, NULL);
1.1 kristaps 573:
1.105 schwarze 574: return ARGS_QWORD;
1.1 kristaps 575: }
576:
1.75 kristaps 577: p = &buf[*pos];
1.85 schwarze 578: *v = mandoc_getarg(mdoc->parse, &p, line, pos);
1.1 kristaps 579:
1.107 schwarze 580: /*
581: * After parsing the last word in this phrase,
582: * tell lookup() whether or not to interpret it.
583: */
584:
585: if (*p == '\0' && mdoc->flags & MDOC_PHRASEQL) {
586: mdoc->flags &= ~MDOC_PHRASEQL;
587: mdoc->flags |= MDOC_PHRASEQF;
588: }
1.105 schwarze 589: return ARGS_WORD;
1.64 kristaps 590: }
591:
1.92 schwarze 592: /*
1.64 kristaps 593: * Check if the string consists only of space-separated closing
1.65 kristaps 594: * delimiters. This is a bit of a dance: the first must be a close
595: * delimiter, but it may be followed by middle delimiters. Arbitrary
596: * whitespace may separate these tokens.
1.64 kristaps 597: */
598: static int
1.77 kristaps 599: args_checkpunct(const char *buf, int i)
1.64 kristaps 600: {
1.69 kristaps 601: int j;
602: char dbuf[DELIMSZ];
1.64 kristaps 603: enum mdelim d;
604:
1.65 kristaps 605: /* First token must be a close-delimiter. */
606:
1.69 kristaps 607: for (j = 0; buf[i] && ' ' != buf[i] && j < DELIMSZ; j++, i++)
608: dbuf[j] = buf[i];
1.65 kristaps 609:
610: if (DELIMSZ == j)
1.105 schwarze 611: return 0;
1.65 kristaps 612:
1.69 kristaps 613: dbuf[j] = '\0';
1.70 kristaps 614: if (DELIM_CLOSE != mdoc_isdelim(dbuf))
1.105 schwarze 615: return 0;
1.64 kristaps 616:
1.69 kristaps 617: while (' ' == buf[i])
1.65 kristaps 618: i++;
619:
620: /* Remaining must NOT be open/none. */
1.92 schwarze 621:
1.69 kristaps 622: while (buf[i]) {
1.65 kristaps 623: j = 0;
1.69 kristaps 624: while (buf[i] && ' ' != buf[i] && j < DELIMSZ)
625: dbuf[j++] = buf[i++];
1.65 kristaps 626:
627: if (DELIMSZ == j)
1.105 schwarze 628: return 0;
1.65 kristaps 629:
1.69 kristaps 630: dbuf[j] = '\0';
1.70 kristaps 631: d = mdoc_isdelim(dbuf);
1.64 kristaps 632: if (DELIM_NONE == d || DELIM_OPEN == d)
1.105 schwarze 633: return 0;
1.65 kristaps 634:
1.69 kristaps 635: while (' ' == buf[i])
1.64 kristaps 636: i++;
637: }
1.69 kristaps 638:
1.105 schwarze 639: return '\0' == buf[i];
1.1 kristaps 640: }
641:
1.97 schwarze 642: static void
1.103 schwarze 643: argv_multi(struct roff_man *mdoc, int line,
1.1 kristaps 644: struct mdoc_argv *v, int *pos, char *buf)
645: {
1.43 kristaps 646: enum margserr ac;
1.1 kristaps 647: char *p;
648:
649: for (v->sz = 0; ; v->sz++) {
1.97 schwarze 650: if (buf[*pos] == '-')
1.1 kristaps 651: break;
1.85 schwarze 652: ac = args(mdoc, line, pos, buf, ARGSFL_NONE, &p);
1.97 schwarze 653: if (ac == ARGS_EOLN)
1.1 kristaps 654: break;
655:
1.97 schwarze 656: if (v->sz % MULTI_STEP == 0)
1.93 schwarze 657: v->value = mandoc_reallocarray(v->value,
658: v->sz + MULTI_STEP, sizeof(char *));
1.31 kristaps 659:
660: v->value[(int)v->sz] = mandoc_strdup(p);
1.1 kristaps 661: }
662: }
663:
1.97 schwarze 664: static void
1.103 schwarze 665: argv_single(struct roff_man *mdoc, int line,
1.1 kristaps 666: struct mdoc_argv *v, int *pos, char *buf)
667: {
1.43 kristaps 668: enum margserr ac;
1.1 kristaps 669: char *p;
670:
1.85 schwarze 671: ac = args(mdoc, line, pos, buf, ARGSFL_NONE, &p);
1.97 schwarze 672: if (ac == ARGS_EOLN)
673: return;
1.1 kristaps 674:
675: v->sz = 1;
1.31 kristaps 676: v->value = mandoc_malloc(sizeof(char *));
677: v->value[0] = mandoc_strdup(p);
1.1 kristaps 678: }
CVSweb