Annotation of mandoc/roff.c, Revision 1.158
1.158 ! kristaps 1: /* $Id: roff.c,v 1.157 2011/07/27 13:42:27 kristaps Exp $ */
1.1 kristaps 2: /*
1.119 schwarze 3: * Copyright (c) 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
4: * Copyright (c) 2010, 2011 Ingo Schwarze <schwarze@openbsd.org>
1.1 kristaps 5: *
6: * Permission to use, copy, modify, and distribute this software for any
1.66 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: *
1.106 kristaps 10: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.66 kristaps 11: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.106 kristaps 12: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.66 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: */
1.66 kristaps 18: #ifdef HAVE_CONFIG_H
19: #include "config.h"
20: #endif
1.30 kristaps 21:
1.67 kristaps 22: #include <assert.h>
1.85 kristaps 23: #include <ctype.h>
1.1 kristaps 24: #include <stdlib.h>
1.67 kristaps 25: #include <string.h>
1.1 kristaps 26:
1.67 kristaps 27: #include "mandoc.h"
1.109 kristaps 28: #include "libroff.h"
1.94 kristaps 29: #include "libmandoc.h"
1.33 kristaps 30:
1.141 kristaps 31: /* Maximum number of nested if-else conditionals. */
1.82 kristaps 32: #define RSTACK_MAX 128
33:
1.67 kristaps 34: enum rofft {
1.103 kristaps 35: ROFF_ad,
1.80 kristaps 36: ROFF_am,
37: ROFF_ami,
38: ROFF_am1,
39: ROFF_de,
40: ROFF_dei,
41: ROFF_de1,
1.83 schwarze 42: ROFF_ds,
1.82 kristaps 43: ROFF_el,
1.103 kristaps 44: ROFF_hy,
1.82 kristaps 45: ROFF_ie,
1.75 kristaps 46: ROFF_if,
1.76 kristaps 47: ROFF_ig,
1.123 schwarze 48: ROFF_it,
1.103 kristaps 49: ROFF_ne,
50: ROFF_nh,
1.104 kristaps 51: ROFF_nr,
1.124 schwarze 52: ROFF_ns,
53: ROFF_ps,
1.83 schwarze 54: ROFF_rm,
1.105 kristaps 55: ROFF_so,
1.124 schwarze 56: ROFF_ta,
1.83 schwarze 57: ROFF_tr,
1.109 kristaps 58: ROFF_TS,
59: ROFF_TE,
1.112 kristaps 60: ROFF_T_,
1.125 kristaps 61: ROFF_EQ,
62: ROFF_EN,
1.76 kristaps 63: ROFF_cblock,
1.141 kristaps 64: ROFF_ccond,
1.106 kristaps 65: ROFF_USERDEF,
1.67 kristaps 66: ROFF_MAX
67: };
68:
1.82 kristaps 69: enum roffrule {
70: ROFFRULE_ALLOW,
71: ROFFRULE_DENY
72: };
73:
1.147 kristaps 74: /*
75: * A single register entity. If "set" is zero, the value of the
76: * register should be the default one, which is per-register.
77: * Registers are assumed to be unsigned ints for now.
78: */
79: struct reg {
80: int set; /* whether set or not */
81: unsigned int u; /* unsigned integer */
82: };
83:
1.94 kristaps 84: struct roffstr {
85: char *name; /* key of symbol */
86: char *string; /* current value */
87: struct roffstr *next; /* next in list */
88: };
89:
1.67 kristaps 90: struct roff {
1.128 kristaps 91: struct mparse *parse; /* parse point */
1.67 kristaps 92: struct roffnode *last; /* leaf of stack */
1.82 kristaps 93: enum roffrule rstack[RSTACK_MAX]; /* stack of !`ie' rules */
94: int rstackpos; /* position in rstack */
1.147 kristaps 95: struct reg regs[REG__MAX];
1.106 kristaps 96: struct roffstr *first_string; /* user-defined strings & macros */
97: const char *current_string; /* value of last called user macro */
1.118 kristaps 98: struct tbl_node *first_tbl; /* first table parsed */
99: struct tbl_node *last_tbl; /* last table parsed */
100: struct tbl_node *tbl; /* current table being parsed */
1.125 kristaps 101: struct eqn_node *last_eqn; /* last equation parsed */
102: struct eqn_node *first_eqn; /* first equation parsed */
103: struct eqn_node *eqn; /* current equation being parsed */
1.79 kristaps 104: };
105:
1.67 kristaps 106: struct roffnode {
107: enum rofft tok; /* type of node */
108: struct roffnode *parent; /* up one in stack */
109: int line; /* parse line */
110: int col; /* parse col */
1.106 kristaps 111: char *name; /* node name, e.g. macro name */
1.79 kristaps 112: char *end; /* end-rules: custom token */
113: int endspan; /* end-rules: next-line or infty */
1.82 kristaps 114: enum roffrule rule; /* current evaluation rule */
1.67 kristaps 115: };
116:
117: #define ROFF_ARGS struct roff *r, /* parse ctx */ \
1.72 kristaps 118: enum rofft tok, /* tok of macro */ \
1.67 kristaps 119: char **bufp, /* input buffer */ \
120: size_t *szp, /* size of input buffer */ \
121: int ln, /* parse line */ \
1.75 kristaps 122: int ppos, /* original pos in buffer */ \
123: int pos, /* current pos in buffer */ \
1.74 kristaps 124: int *offs /* reset offset of buffer data */
1.67 kristaps 125:
126: typedef enum rofferr (*roffproc)(ROFF_ARGS);
127:
128: struct roffmac {
129: const char *name; /* macro name */
1.79 kristaps 130: roffproc proc; /* process new macro */
131: roffproc text; /* process as child text of macro */
132: roffproc sub; /* process as child of macro */
133: int flags;
134: #define ROFFMAC_STRUCT (1 << 0) /* always interpret */
1.85 kristaps 135: struct roffmac *next;
1.67 kristaps 136: };
137:
1.141 kristaps 138: struct predef {
139: const char *name; /* predefined input name */
140: const char *str; /* replacement symbol */
141: };
142:
143: #define PREDEF(__name, __str) \
144: { (__name), (__str) },
145:
1.155 kristaps 146: static enum rofft roffhash_find(const char *, size_t);
147: static void roffhash_init(void);
148: static void roffnode_cleanscope(struct roff *);
149: static void roffnode_pop(struct roff *);
150: static void roffnode_push(struct roff *, enum rofft,
151: const char *, int, int);
1.80 kristaps 152: static enum rofferr roff_block(ROFF_ARGS);
153: static enum rofferr roff_block_text(ROFF_ARGS);
154: static enum rofferr roff_block_sub(ROFF_ARGS);
155: static enum rofferr roff_cblock(ROFF_ARGS);
156: static enum rofferr roff_ccond(ROFF_ARGS);
1.82 kristaps 157: static enum rofferr roff_cond(ROFF_ARGS);
158: static enum rofferr roff_cond_text(ROFF_ARGS);
159: static enum rofferr roff_cond_sub(ROFF_ARGS);
1.92 schwarze 160: static enum rofferr roff_ds(ROFF_ARGS);
1.94 kristaps 161: static enum roffrule roff_evalcond(const char *, int *);
1.155 kristaps 162: static void roff_free1(struct roff *);
1.94 kristaps 163: static void roff_freestr(struct roff *);
1.121 schwarze 164: static char *roff_getname(struct roff *, char **, int, int);
1.94 kristaps 165: static const char *roff_getstrn(const struct roff *,
166: const char *, size_t);
1.103 kristaps 167: static enum rofferr roff_line_ignore(ROFF_ARGS);
1.89 kristaps 168: static enum rofferr roff_nr(ROFF_ARGS);
1.156 kristaps 169: static void roff_openeqn(struct roff *, const char *,
170: int, int, const char *);
1.155 kristaps 171: static enum rofft roff_parse(struct roff *, const char *, int *);
172: static enum rofferr roff_parsetext(char *);
1.154 kristaps 173: static void roff_res(struct roff *,
1.142 kristaps 174: char **, size_t *, int, int);
1.122 schwarze 175: static enum rofferr roff_rm(ROFF_ARGS);
1.94 kristaps 176: static void roff_setstr(struct roff *,
1.106 kristaps 177: const char *, const char *, int);
1.105 kristaps 178: static enum rofferr roff_so(ROFF_ARGS);
1.109 kristaps 179: static enum rofferr roff_TE(ROFF_ARGS);
180: static enum rofferr roff_TS(ROFF_ARGS);
1.125 kristaps 181: static enum rofferr roff_EQ(ROFF_ARGS);
182: static enum rofferr roff_EN(ROFF_ARGS);
1.112 kristaps 183: static enum rofferr roff_T_(ROFF_ARGS);
1.106 kristaps 184: static enum rofferr roff_userdef(ROFF_ARGS);
1.67 kristaps 185:
1.155 kristaps 186: /* See roffhash_find() */
1.85 kristaps 187:
188: #define ASCII_HI 126
189: #define ASCII_LO 33
190: #define HASHWIDTH (ASCII_HI - ASCII_LO + 1)
191:
192: static struct roffmac *hash[HASHWIDTH];
193:
194: static struct roffmac roffs[ROFF_MAX] = {
1.103 kristaps 195: { "ad", roff_line_ignore, NULL, NULL, 0, NULL },
1.85 kristaps 196: { "am", roff_block, roff_block_text, roff_block_sub, 0, NULL },
197: { "ami", roff_block, roff_block_text, roff_block_sub, 0, NULL },
198: { "am1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
199: { "de", roff_block, roff_block_text, roff_block_sub, 0, NULL },
200: { "dei", roff_block, roff_block_text, roff_block_sub, 0, NULL },
201: { "de1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.92 schwarze 202: { "ds", roff_ds, NULL, NULL, 0, NULL },
1.85 kristaps 203: { "el", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
1.103 kristaps 204: { "hy", roff_line_ignore, NULL, NULL, 0, NULL },
1.85 kristaps 205: { "ie", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
206: { "if", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
207: { "ig", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.123 schwarze 208: { "it", roff_line_ignore, NULL, NULL, 0, NULL },
1.103 kristaps 209: { "ne", roff_line_ignore, NULL, NULL, 0, NULL },
210: { "nh", roff_line_ignore, NULL, NULL, 0, NULL },
1.104 kristaps 211: { "nr", roff_nr, NULL, NULL, 0, NULL },
1.124 schwarze 212: { "ns", roff_line_ignore, NULL, NULL, 0, NULL },
213: { "ps", roff_line_ignore, NULL, NULL, 0, NULL },
1.122 schwarze 214: { "rm", roff_rm, NULL, NULL, 0, NULL },
1.105 kristaps 215: { "so", roff_so, NULL, NULL, 0, NULL },
1.124 schwarze 216: { "ta", roff_line_ignore, NULL, NULL, 0, NULL },
1.103 kristaps 217: { "tr", roff_line_ignore, NULL, NULL, 0, NULL },
1.109 kristaps 218: { "TS", roff_TS, NULL, NULL, 0, NULL },
219: { "TE", roff_TE, NULL, NULL, 0, NULL },
1.112 kristaps 220: { "T&", roff_T_, NULL, NULL, 0, NULL },
1.125 kristaps 221: { "EQ", roff_EQ, NULL, NULL, 0, NULL },
222: { "EN", roff_EN, NULL, NULL, 0, NULL },
1.85 kristaps 223: { ".", roff_cblock, NULL, NULL, 0, NULL },
224: { "\\}", roff_ccond, NULL, NULL, 0, NULL },
1.106 kristaps 225: { NULL, roff_userdef, NULL, NULL, 0, NULL },
1.67 kristaps 226: };
227:
1.141 kristaps 228: /* Array of injected predefined strings. */
229: #define PREDEFS_MAX 38
230: static const struct predef predefs[PREDEFS_MAX] = {
231: #include "predefs.in"
232: };
233:
1.155 kristaps 234: /* See roffhash_find() */
1.85 kristaps 235: #define ROFF_HASH(p) (p[0] - ASCII_LO)
236:
237: static void
1.155 kristaps 238: roffhash_init(void)
1.85 kristaps 239: {
240: struct roffmac *n;
241: int buc, i;
242:
1.106 kristaps 243: for (i = 0; i < (int)ROFF_USERDEF; i++) {
1.85 kristaps 244: assert(roffs[i].name[0] >= ASCII_LO);
245: assert(roffs[i].name[0] <= ASCII_HI);
246:
247: buc = ROFF_HASH(roffs[i].name);
248:
249: if (NULL != (n = hash[buc])) {
250: for ( ; n->next; n = n->next)
251: /* Do nothing. */ ;
252: n->next = &roffs[i];
253: } else
254: hash[buc] = &roffs[i];
255: }
256: }
257:
1.67 kristaps 258: /*
259: * Look up a roff token by its name. Returns ROFF_MAX if no macro by
260: * the nil-terminated string name could be found.
261: */
262: static enum rofft
1.155 kristaps 263: roffhash_find(const char *p, size_t s)
1.67 kristaps 264: {
1.85 kristaps 265: int buc;
266: struct roffmac *n;
1.67 kristaps 267:
1.85 kristaps 268: /*
269: * libroff has an extremely simple hashtable, for the time
270: * being, which simply keys on the first character, which must
271: * be printable, then walks a chain. It works well enough until
272: * optimised.
273: */
274:
275: if (p[0] < ASCII_LO || p[0] > ASCII_HI)
276: return(ROFF_MAX);
277:
278: buc = ROFF_HASH(p);
279:
280: if (NULL == (n = hash[buc]))
281: return(ROFF_MAX);
282: for ( ; n; n = n->next)
1.106 kristaps 283: if (0 == strncmp(n->name, p, s) && '\0' == n->name[(int)s])
1.85 kristaps 284: return((enum rofft)(n - roffs));
1.67 kristaps 285:
286: return(ROFF_MAX);
287: }
288:
289:
290: /*
291: * Pop the current node off of the stack of roff instructions currently
292: * pending.
293: */
294: static void
295: roffnode_pop(struct roff *r)
296: {
297: struct roffnode *p;
298:
1.75 kristaps 299: assert(r->last);
300: p = r->last;
1.82 kristaps 301:
1.75 kristaps 302: r->last = r->last->parent;
1.106 kristaps 303: free(p->name);
304: free(p->end);
1.67 kristaps 305: free(p);
306: }
307:
308:
309: /*
310: * Push a roff node onto the instruction stack. This must later be
311: * removed with roffnode_pop().
312: */
1.98 schwarze 313: static void
1.106 kristaps 314: roffnode_push(struct roff *r, enum rofft tok, const char *name,
315: int line, int col)
1.67 kristaps 316: {
317: struct roffnode *p;
318:
1.98 schwarze 319: p = mandoc_calloc(1, sizeof(struct roffnode));
1.67 kristaps 320: p->tok = tok;
1.106 kristaps 321: if (name)
322: p->name = mandoc_strdup(name);
1.67 kristaps 323: p->parent = r->last;
324: p->line = line;
325: p->col = col;
1.79 kristaps 326: p->rule = p->parent ? p->parent->rule : ROFFRULE_DENY;
1.67 kristaps 327:
328: r->last = p;
329: }
330:
331:
332: static void
333: roff_free1(struct roff *r)
334: {
1.118 kristaps 335: struct tbl_node *t;
1.125 kristaps 336: struct eqn_node *e;
1.67 kristaps 337:
1.125 kristaps 338: while (NULL != (t = r->first_tbl)) {
1.113 kristaps 339: r->first_tbl = t->next;
340: tbl_free(t);
1.109 kristaps 341: }
342:
1.113 kristaps 343: r->first_tbl = r->last_tbl = r->tbl = NULL;
344:
1.125 kristaps 345: while (NULL != (e = r->first_eqn)) {
346: r->first_eqn = e->next;
347: eqn_free(e);
348: }
349:
350: r->first_eqn = r->last_eqn = r->eqn = NULL;
351:
1.67 kristaps 352: while (r->last)
353: roffnode_pop(r);
1.109 kristaps 354:
1.94 kristaps 355: roff_freestr(r);
1.67 kristaps 356: }
357:
358:
359: void
360: roff_reset(struct roff *r)
361: {
1.143 kristaps 362: int i;
1.67 kristaps 363:
364: roff_free1(r);
1.143 kristaps 365:
1.147 kristaps 366: memset(&r->regs, 0, sizeof(struct reg) * REG__MAX);
367:
1.143 kristaps 368: for (i = 0; i < PREDEFS_MAX; i++)
369: roff_setstr(r, predefs[i].name, predefs[i].str, 0);
1.67 kristaps 370: }
371:
372:
373: void
374: roff_free(struct roff *r)
375: {
376:
377: roff_free1(r);
378: free(r);
379: }
380:
381:
382: struct roff *
1.147 kristaps 383: roff_alloc(struct mparse *parse)
1.67 kristaps 384: {
385: struct roff *r;
1.141 kristaps 386: int i;
1.67 kristaps 387:
1.98 schwarze 388: r = mandoc_calloc(1, sizeof(struct roff));
1.128 kristaps 389: r->parse = parse;
1.82 kristaps 390: r->rstackpos = -1;
1.85 kristaps 391:
1.155 kristaps 392: roffhash_init();
1.141 kristaps 393:
394: for (i = 0; i < PREDEFS_MAX; i++)
395: roff_setstr(r, predefs[i].name, predefs[i].str, 0);
396:
1.67 kristaps 397: return(r);
398: }
399:
1.94 kristaps 400: /*
401: * Pre-filter each and every line for reserved words (one beginning with
402: * `\*', e.g., `\*(ab'). These must be handled before the actual line
403: * is processed.
1.153 kristaps 404: * This also checks the syntax of regular escapes.
1.154 kristaps 405: */
406: static void
1.142 kristaps 407: roff_res(struct roff *r, char **bufp, size_t *szp, int ln, int pos)
1.94 kristaps 408: {
1.152 kristaps 409: enum mandoc_esc esc;
1.108 schwarze 410: const char *stesc; /* start of an escape sequence ('\\') */
411: const char *stnam; /* start of the name, after "[(*" */
412: const char *cp; /* end of the name, e.g. before ']' */
413: const char *res; /* the string to be substituted */
1.94 kristaps 414: int i, maxl;
415: size_t nsz;
416: char *n;
417:
1.154 kristaps 418: again:
1.108 schwarze 419: cp = *bufp + pos;
420: while (NULL != (cp = strchr(cp, '\\'))) {
421: stesc = cp++;
422:
423: /*
424: * The second character must be an asterisk.
425: * If it isn't, skip it anyway: It is escaped,
426: * so it can't start another escape sequence.
427: */
428:
429: if ('\0' == *cp)
1.154 kristaps 430: return;
1.152 kristaps 431:
432: if ('*' != *cp) {
433: res = cp;
434: esc = mandoc_escape(&cp, NULL, NULL);
435: if (ESCAPE_ERROR != esc)
436: continue;
437: cp = res;
1.153 kristaps 438: mandoc_msg
439: (MANDOCERR_BADESCAPE, r->parse,
440: ln, (int)(stesc - *bufp), NULL);
1.154 kristaps 441: return;
1.152 kristaps 442: }
443:
444: cp++;
1.108 schwarze 445:
446: /*
447: * The third character decides the length
448: * of the name of the string.
449: * Save a pointer to the name.
450: */
451:
1.94 kristaps 452: switch (*cp) {
1.108 schwarze 453: case ('\0'):
1.154 kristaps 454: return;
1.94 kristaps 455: case ('('):
456: cp++;
457: maxl = 2;
458: break;
459: case ('['):
460: cp++;
461: maxl = 0;
462: break;
463: default:
464: maxl = 1;
465: break;
466: }
1.108 schwarze 467: stnam = cp;
1.94 kristaps 468:
1.108 schwarze 469: /* Advance to the end of the name. */
1.94 kristaps 470:
471: for (i = 0; 0 == maxl || i < maxl; i++, cp++) {
1.153 kristaps 472: if ('\0' == *cp) {
473: mandoc_msg
474: (MANDOCERR_BADESCAPE,
475: r->parse, ln,
476: (int)(stesc - *bufp), NULL);
1.154 kristaps 477: return;
1.153 kristaps 478: }
1.94 kristaps 479: if (0 == maxl && ']' == *cp)
480: break;
481: }
482:
1.108 schwarze 483: /*
484: * Retrieve the replacement string; if it is
485: * undefined, resume searching for escapes.
486: */
487:
488: res = roff_getstrn(r, stnam, (size_t)i);
1.94 kristaps 489:
490: if (NULL == res) {
1.153 kristaps 491: mandoc_msg
492: (MANDOCERR_BADESCAPE, r->parse,
493: ln, (int)(stesc - *bufp), NULL);
1.142 kristaps 494: res = "";
1.94 kristaps 495: }
496:
1.108 schwarze 497: /* Replace the escape sequence by the string. */
498:
1.154 kristaps 499: pos += (stesc - *bufp);
500:
1.94 kristaps 501: nsz = *szp + strlen(res) + 1;
502: n = mandoc_malloc(nsz);
503:
1.108 schwarze 504: strlcpy(n, *bufp, (size_t)(stesc - *bufp + 1));
1.94 kristaps 505: strlcat(n, res, nsz);
506: strlcat(n, cp + (maxl ? 0 : 1), nsz);
507:
508: free(*bufp);
509:
510: *bufp = n;
511: *szp = nsz;
1.154 kristaps 512: goto again;
513: }
514: }
515:
516: /*
517: * Process text streams: convert all breakable hyphens into ASCII_HYPH.
518: */
519: static enum rofferr
520: roff_parsetext(char *p)
521: {
1.155 kristaps 522: char l, r;
1.154 kristaps 523: size_t sz;
524: const char *start;
525: enum mandoc_esc esc;
526:
527: start = p;
528:
529: while ('\0' != *p) {
530: sz = strcspn(p, "-\\");
531: p += sz;
532:
533: if ('\\' == *p) {
534: /* Skip over escapes. */
535: p++;
536: esc = mandoc_escape
537: ((const char **)&p, NULL, NULL);
538: if (ESCAPE_ERROR == esc)
539: break;
1.155 kristaps 540: continue;
1.158 ! kristaps 541: } else if ('-' != *p || p == start) {
! 542: p++;
1.155 kristaps 543: continue;
1.158 ! kristaps 544: }
1.155 kristaps 545:
546: l = *(p - 1);
547: r = *(p + 1);
548:
549: if ('\\' != l &&
550: '\t' != r && '\t' != l &&
551: ' ' != r && ' ' != l &&
552: '-' != r && '-' != l &&
553: ! isdigit((unsigned char)l) &&
554: ! isdigit((unsigned char)r))
555: *p = ASCII_HYPH;
556: p++;
1.94 kristaps 557: }
558:
1.154 kristaps 559: return(ROFF_CONT);
1.94 kristaps 560: }
561:
1.67 kristaps 562: enum rofferr
1.90 kristaps 563: roff_parseln(struct roff *r, int ln, char **bufp,
564: size_t *szp, int pos, int *offs)
1.67 kristaps 565: {
566: enum rofft t;
1.109 kristaps 567: enum rofferr e;
1.130 kristaps 568: int ppos, ctl;
1.79 kristaps 569:
570: /*
1.94 kristaps 571: * Run the reserved-word filter only if we have some reserved
572: * words to fill in.
573: */
574:
1.154 kristaps 575: roff_res(r, bufp, szp, ln, pos);
1.94 kristaps 576:
1.130 kristaps 577: ppos = pos;
578: ctl = mandoc_getcontrol(*bufp, &pos);
579:
1.94 kristaps 580: /*
1.79 kristaps 581: * First, if a scope is open and we're not a macro, pass the
582: * text through the macro's filter. If a scope isn't open and
583: * we're not a macro, just let it through.
1.125 kristaps 584: * Finally, if there's an equation scope open, divert it into it
585: * no matter our state.
1.79 kristaps 586: */
1.74 kristaps 587:
1.130 kristaps 588: if (r->last && ! ctl) {
1.78 kristaps 589: t = r->last->tok;
590: assert(roffs[t].text);
1.109 kristaps 591: e = (*roffs[t].text)
592: (r, t, bufp, szp, ln, pos, pos, offs);
593: assert(ROFF_IGN == e || ROFF_CONT == e);
1.125 kristaps 594: if (ROFF_CONT != e)
595: return(e);
596: if (r->eqn)
1.146 kristaps 597: return(eqn_read(&r->eqn, ln, *bufp, pos, offs));
1.125 kristaps 598: if (r->tbl)
1.130 kristaps 599: return(tbl_read(r->tbl, ln, *bufp, pos));
1.154 kristaps 600: return(roff_parsetext(*bufp + pos));
1.130 kristaps 601: } else if ( ! ctl) {
1.125 kristaps 602: if (r->eqn)
1.146 kristaps 603: return(eqn_read(&r->eqn, ln, *bufp, pos, offs));
1.109 kristaps 604: if (r->tbl)
1.130 kristaps 605: return(tbl_read(r->tbl, ln, *bufp, pos));
1.154 kristaps 606: return(roff_parsetext(*bufp + pos));
1.125 kristaps 607: } else if (r->eqn)
1.146 kristaps 608: return(eqn_read(&r->eqn, ln, *bufp, ppos, offs));
1.67 kristaps 609:
1.79 kristaps 610: /*
611: * If a scope is open, go to the child handler for that macro,
612: * as it may want to preprocess before doing anything with it.
1.125 kristaps 613: * Don't do so if an equation is open.
1.79 kristaps 614: */
1.78 kristaps 615:
1.79 kristaps 616: if (r->last) {
617: t = r->last->tok;
618: assert(roffs[t].sub);
619: return((*roffs[t].sub)
1.90 kristaps 620: (r, t, bufp, szp,
1.130 kristaps 621: ln, ppos, pos, offs));
1.79 kristaps 622: }
1.78 kristaps 623:
1.79 kristaps 624: /*
625: * Lastly, as we've no scope open, try to look up and execute
626: * the new macro. If no macro is found, simply return and let
627: * the compilers handle it.
628: */
1.67 kristaps 629:
1.106 kristaps 630: if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos)))
1.79 kristaps 631: return(ROFF_CONT);
1.67 kristaps 632:
1.75 kristaps 633: assert(roffs[t].proc);
1.78 kristaps 634: return((*roffs[t].proc)
1.90 kristaps 635: (r, t, bufp, szp,
636: ln, ppos, pos, offs));
1.74 kristaps 637: }
638:
639:
1.117 kristaps 640: void
1.74 kristaps 641: roff_endparse(struct roff *r)
642: {
643:
1.110 kristaps 644: if (r->last)
1.128 kristaps 645: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.109 kristaps 646: r->last->line, r->last->col, NULL);
1.117 kristaps 647:
1.125 kristaps 648: if (r->eqn) {
1.128 kristaps 649: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.148 kristaps 650: r->eqn->eqn.ln, r->eqn->eqn.pos, NULL);
1.151 kristaps 651: eqn_end(&r->eqn);
1.125 kristaps 652: }
653:
1.117 kristaps 654: if (r->tbl) {
1.128 kristaps 655: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.117 kristaps 656: r->tbl->line, r->tbl->pos, NULL);
1.151 kristaps 657: tbl_end(&r->tbl);
1.117 kristaps 658: }
1.67 kristaps 659: }
660:
661: /*
662: * Parse a roff node's type from the input buffer. This must be in the
663: * form of ".foo xxx" in the usual way.
664: */
665: static enum rofft
1.106 kristaps 666: roff_parse(struct roff *r, const char *buf, int *pos)
1.67 kristaps 667: {
1.106 kristaps 668: const char *mac;
669: size_t maclen;
1.67 kristaps 670: enum rofft t;
671:
1.144 kristaps 672: if ('\0' == buf[*pos] || '"' == buf[*pos] ||
673: '\t' == buf[*pos] || ' ' == buf[*pos])
1.67 kristaps 674: return(ROFF_MAX);
675:
1.144 kristaps 676: /*
677: * We stop the macro parse at an escape, tab, space, or nil.
678: * However, `\}' is also a valid macro, so make sure we don't
679: * clobber it by seeing the `\' as the end of token.
680: */
681:
1.106 kristaps 682: mac = buf + *pos;
1.144 kristaps 683: maclen = strcspn(mac + 1, " \\\t\0") + 1;
1.67 kristaps 684:
1.106 kristaps 685: t = (r->current_string = roff_getstrn(r, mac, maclen))
1.155 kristaps 686: ? ROFF_USERDEF : roffhash_find(mac, maclen);
1.67 kristaps 687:
1.127 kristaps 688: *pos += (int)maclen;
1.130 kristaps 689:
1.67 kristaps 690: while (buf[*pos] && ' ' == buf[*pos])
691: (*pos)++;
692:
693: return(t);
694: }
695:
696: /* ARGSUSED */
697: static enum rofferr
1.76 kristaps 698: roff_cblock(ROFF_ARGS)
1.67 kristaps 699: {
700:
1.79 kristaps 701: /*
702: * A block-close `..' should only be invoked as a child of an
703: * ignore macro, otherwise raise a warning and just ignore it.
704: */
705:
1.76 kristaps 706: if (NULL == r->last) {
1.128 kristaps 707: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.76 kristaps 708: return(ROFF_IGN);
709: }
1.67 kristaps 710:
1.81 kristaps 711: switch (r->last->tok) {
712: case (ROFF_am):
713: /* FALLTHROUGH */
714: case (ROFF_ami):
715: /* FALLTHROUGH */
716: case (ROFF_am1):
717: /* FALLTHROUGH */
718: case (ROFF_de):
1.108 schwarze 719: /* ROFF_de1 is remapped to ROFF_de in roff_block(). */
1.81 kristaps 720: /* FALLTHROUGH */
721: case (ROFF_dei):
722: /* FALLTHROUGH */
723: case (ROFF_ig):
724: break;
725: default:
1.128 kristaps 726: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.67 kristaps 727: return(ROFF_IGN);
1.76 kristaps 728: }
1.67 kristaps 729:
1.76 kristaps 730: if ((*bufp)[pos])
1.128 kristaps 731: mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.71 kristaps 732:
733: roffnode_pop(r);
1.76 kristaps 734: roffnode_cleanscope(r);
735: return(ROFF_IGN);
1.71 kristaps 736:
1.67 kristaps 737: }
738:
739:
1.76 kristaps 740: static void
741: roffnode_cleanscope(struct roff *r)
1.67 kristaps 742: {
743:
1.76 kristaps 744: while (r->last) {
745: if (--r->last->endspan < 0)
746: break;
747: roffnode_pop(r);
748: }
1.67 kristaps 749: }
750:
751:
1.75 kristaps 752: /* ARGSUSED */
1.74 kristaps 753: static enum rofferr
1.75 kristaps 754: roff_ccond(ROFF_ARGS)
1.74 kristaps 755: {
756:
1.76 kristaps 757: if (NULL == r->last) {
1.128 kristaps 758: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.76 kristaps 759: return(ROFF_IGN);
760: }
761:
1.82 kristaps 762: switch (r->last->tok) {
763: case (ROFF_el):
764: /* FALLTHROUGH */
765: case (ROFF_ie):
766: /* FALLTHROUGH */
767: case (ROFF_if):
768: break;
769: default:
1.128 kristaps 770: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.75 kristaps 771: return(ROFF_IGN);
772: }
773:
1.76 kristaps 774: if (r->last->endspan > -1) {
1.128 kristaps 775: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.76 kristaps 776: return(ROFF_IGN);
777: }
778:
779: if ((*bufp)[pos])
1.128 kristaps 780: mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.76 kristaps 781:
1.75 kristaps 782: roffnode_pop(r);
1.76 kristaps 783: roffnode_cleanscope(r);
784: return(ROFF_IGN);
785: }
786:
1.75 kristaps 787:
1.76 kristaps 788: /* ARGSUSED */
789: static enum rofferr
1.80 kristaps 790: roff_block(ROFF_ARGS)
1.76 kristaps 791: {
1.78 kristaps 792: int sv;
793: size_t sz;
1.106 kristaps 794: char *name;
795:
796: name = NULL;
1.76 kristaps 797:
1.106 kristaps 798: if (ROFF_ig != tok) {
799: if ('\0' == (*bufp)[pos]) {
1.128 kristaps 800: mandoc_msg(MANDOCERR_NOARGS, r->parse, ln, ppos, NULL);
1.106 kristaps 801: return(ROFF_IGN);
802: }
1.107 kristaps 803:
804: /*
805: * Re-write `de1', since we don't really care about
806: * groff's strange compatibility mode, into `de'.
807: */
808:
1.106 kristaps 809: if (ROFF_de1 == tok)
810: tok = ROFF_de;
811: if (ROFF_de == tok)
812: name = *bufp + pos;
813: else
1.128 kristaps 814: mandoc_msg(MANDOCERR_REQUEST, r->parse, ln, ppos,
1.106 kristaps 815: roffs[tok].name);
1.107 kristaps 816:
1.131 schwarze 817: while ((*bufp)[pos] && ! isspace((unsigned char)(*bufp)[pos]))
1.80 kristaps 818: pos++;
1.107 kristaps 819:
1.131 schwarze 820: while (isspace((unsigned char)(*bufp)[pos]))
1.106 kristaps 821: (*bufp)[pos++] = '\0';
1.80 kristaps 822: }
823:
1.106 kristaps 824: roffnode_push(r, tok, name, ln, ppos);
825:
826: /*
827: * At the beginning of a `de' macro, clear the existing string
828: * with the same name, if there is one. New content will be
829: * added from roff_block_text() in multiline mode.
830: */
1.107 kristaps 831:
1.106 kristaps 832: if (ROFF_de == tok)
1.108 schwarze 833: roff_setstr(r, name, "", 0);
1.76 kristaps 834:
1.79 kristaps 835: if ('\0' == (*bufp)[pos])
1.78 kristaps 836: return(ROFF_IGN);
837:
1.107 kristaps 838: /* If present, process the custom end-of-line marker. */
839:
1.78 kristaps 840: sv = pos;
1.131 schwarze 841: while ((*bufp)[pos] && ! isspace((unsigned char)(*bufp)[pos]))
1.78 kristaps 842: pos++;
843:
844: /*
845: * Note: groff does NOT like escape characters in the input.
846: * Instead of detecting this, we're just going to let it fly and
847: * to hell with it.
848: */
849:
850: assert(pos > sv);
851: sz = (size_t)(pos - sv);
852:
1.79 kristaps 853: if (1 == sz && '.' == (*bufp)[sv])
854: return(ROFF_IGN);
855:
1.98 schwarze 856: r->last->end = mandoc_malloc(sz + 1);
1.78 kristaps 857:
858: memcpy(r->last->end, *bufp + sv, sz);
859: r->last->end[(int)sz] = '\0';
860:
1.77 kristaps 861: if ((*bufp)[pos])
1.128 kristaps 862: mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.74 kristaps 863:
1.78 kristaps 864: return(ROFF_IGN);
865: }
866:
867:
868: /* ARGSUSED */
869: static enum rofferr
1.80 kristaps 870: roff_block_sub(ROFF_ARGS)
1.79 kristaps 871: {
872: enum rofft t;
873: int i, j;
874:
875: /*
876: * First check whether a custom macro exists at this level. If
877: * it does, then check against it. This is some of groff's
878: * stranger behaviours. If we encountered a custom end-scope
879: * tag and that tag also happens to be a "real" macro, then we
880: * need to try interpreting it again as a real macro. If it's
881: * not, then return ignore. Else continue.
882: */
883:
884: if (r->last->end) {
1.130 kristaps 885: for (i = pos, j = 0; r->last->end[j]; j++, i++)
1.79 kristaps 886: if ((*bufp)[i] != r->last->end[j])
887: break;
888:
889: if ('\0' == r->last->end[j] &&
890: ('\0' == (*bufp)[i] ||
891: ' ' == (*bufp)[i] ||
892: '\t' == (*bufp)[i])) {
893: roffnode_pop(r);
894: roffnode_cleanscope(r);
895:
1.130 kristaps 896: while (' ' == (*bufp)[i] || '\t' == (*bufp)[i])
897: i++;
898:
899: pos = i;
1.106 kristaps 900: if (ROFF_MAX != roff_parse(r, *bufp, &pos))
1.79 kristaps 901: return(ROFF_RERUN);
902: return(ROFF_IGN);
903: }
904: }
905:
906: /*
907: * If we have no custom end-query or lookup failed, then try
908: * pulling it out of the hashtable.
909: */
910:
1.137 schwarze 911: t = roff_parse(r, *bufp, &pos);
1.79 kristaps 912:
1.106 kristaps 913: /*
914: * Macros other than block-end are only significant
915: * in `de' blocks; elsewhere, simply throw them away.
916: */
917: if (ROFF_cblock != t) {
918: if (ROFF_de == tok)
919: roff_setstr(r, r->last->name, *bufp + ppos, 1);
1.79 kristaps 920: return(ROFF_IGN);
1.106 kristaps 921: }
1.79 kristaps 922:
923: assert(roffs[t].proc);
1.90 kristaps 924: return((*roffs[t].proc)(r, t, bufp, szp,
925: ln, ppos, pos, offs));
1.79 kristaps 926: }
927:
928:
929: /* ARGSUSED */
930: static enum rofferr
1.80 kristaps 931: roff_block_text(ROFF_ARGS)
1.78 kristaps 932: {
933:
1.106 kristaps 934: if (ROFF_de == tok)
935: roff_setstr(r, r->last->name, *bufp + pos, 1);
936:
1.78 kristaps 937: return(ROFF_IGN);
938: }
939:
940:
941: /* ARGSUSED */
942: static enum rofferr
1.82 kristaps 943: roff_cond_sub(ROFF_ARGS)
944: {
945: enum rofft t;
946: enum roffrule rr;
1.139 kristaps 947: char *ep;
1.82 kristaps 948:
949: rr = r->last->rule;
1.139 kristaps 950: roffnode_cleanscope(r);
1.82 kristaps 951:
1.139 kristaps 952: /*
953: * If the macro is unknown, first check if it contains a closing
954: * delimiter `\}'. If it does, close out our scope and return
955: * the currently-scoped rule (ignore or continue). Else, drop
956: * into the currently-scoped rule.
1.87 kristaps 957: */
958:
1.106 kristaps 959: if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos))) {
1.139 kristaps 960: ep = &(*bufp)[pos];
961: for ( ; NULL != (ep = strchr(ep, '\\')); ep++) {
962: ep++;
963: if ('}' != *ep)
964: continue;
1.144 kristaps 965:
966: /*
967: * Make the \} go away.
968: * This is a little haphazard, as it's not quite
969: * clear how nroff does this.
970: * If we're at the end of line, then just chop
971: * off the \} and resize the buffer.
972: * If we aren't, then conver it to spaces.
973: */
974:
975: if ('\0' == *(ep + 1)) {
976: *--ep = '\0';
977: *szp -= 2;
978: } else
979: *(ep - 1) = *ep = ' ';
980:
1.139 kristaps 981: roff_ccond(r, ROFF_ccond, bufp, szp,
982: ln, pos, pos + 2, offs);
983: break;
984: }
1.82 kristaps 985: return(ROFFRULE_DENY == rr ? ROFF_IGN : ROFF_CONT);
1.100 kristaps 986: }
1.82 kristaps 987:
988: /*
989: * A denied conditional must evaluate its children if and only
990: * if they're either structurally required (such as loops and
991: * conditionals) or a closing macro.
992: */
1.139 kristaps 993:
1.82 kristaps 994: if (ROFFRULE_DENY == rr)
995: if ( ! (ROFFMAC_STRUCT & roffs[t].flags))
996: if (ROFF_ccond != t)
997: return(ROFF_IGN);
998:
999: assert(roffs[t].proc);
1.90 kristaps 1000: return((*roffs[t].proc)(r, t, bufp, szp,
1001: ln, ppos, pos, offs));
1.82 kristaps 1002: }
1003:
1004: /* ARGSUSED */
1005: static enum rofferr
1006: roff_cond_text(ROFF_ARGS)
1.78 kristaps 1007: {
1.140 kristaps 1008: char *ep;
1.82 kristaps 1009: enum roffrule rr;
1010:
1011: rr = r->last->rule;
1.140 kristaps 1012: roffnode_cleanscope(r);
1.82 kristaps 1013:
1.140 kristaps 1014: ep = &(*bufp)[pos];
1015: for ( ; NULL != (ep = strchr(ep, '\\')); ep++) {
1016: ep++;
1017: if ('}' != *ep)
1018: continue;
1019: *ep = '&';
1020: roff_ccond(r, ROFF_ccond, bufp, szp,
1021: ln, pos, pos + 2, offs);
1.78 kristaps 1022: }
1.82 kristaps 1023: return(ROFFRULE_DENY == rr ? ROFF_IGN : ROFF_CONT);
1.74 kristaps 1024: }
1025:
1.88 kristaps 1026: static enum roffrule
1027: roff_evalcond(const char *v, int *pos)
1028: {
1029:
1030: switch (v[*pos]) {
1031: case ('n'):
1032: (*pos)++;
1033: return(ROFFRULE_ALLOW);
1034: case ('e'):
1035: /* FALLTHROUGH */
1036: case ('o'):
1037: /* FALLTHROUGH */
1038: case ('t'):
1039: (*pos)++;
1040: return(ROFFRULE_DENY);
1041: default:
1042: break;
1043: }
1044:
1045: while (v[*pos] && ' ' != v[*pos])
1046: (*pos)++;
1047: return(ROFFRULE_DENY);
1048: }
1049:
1.75 kristaps 1050: /* ARGSUSED */
1.74 kristaps 1051: static enum rofferr
1.103 kristaps 1052: roff_line_ignore(ROFF_ARGS)
1.89 kristaps 1053: {
1.123 schwarze 1054:
1055: if (ROFF_it == tok)
1.128 kristaps 1056: mandoc_msg(MANDOCERR_REQUEST, r->parse, ln, ppos, "it");
1.89 kristaps 1057:
1058: return(ROFF_IGN);
1059: }
1060:
1.104 kristaps 1061: /* ARGSUSED */
1062: static enum rofferr
1.82 kristaps 1063: roff_cond(ROFF_ARGS)
1.74 kristaps 1064: {
1.77 kristaps 1065: int sv;
1.88 kristaps 1066: enum roffrule rule;
1.74 kristaps 1067:
1.134 kristaps 1068: /*
1069: * An `.el' has no conditional body: it will consume the value
1070: * of the current rstack entry set in prior `ie' calls or
1071: * defaults to DENY.
1072: *
1073: * If we're not an `el', however, then evaluate the conditional.
1074: */
1.133 kristaps 1075:
1.134 kristaps 1076: rule = ROFF_el == tok ?
1077: (r->rstackpos < 0 ?
1078: ROFFRULE_DENY : r->rstack[r->rstackpos--]) :
1079: roff_evalcond(*bufp, &pos);
1.77 kristaps 1080:
1081: sv = pos;
1.75 kristaps 1082: while (' ' == (*bufp)[pos])
1083: pos++;
1.74 kristaps 1084:
1.77 kristaps 1085: /*
1086: * Roff is weird. If we have just white-space after the
1087: * conditional, it's considered the BODY and we exit without
1088: * really doing anything. Warn about this. It's probably
1089: * wrong.
1090: */
1.88 kristaps 1091:
1.77 kristaps 1092: if ('\0' == (*bufp)[pos] && sv != pos) {
1.128 kristaps 1093: mandoc_msg(MANDOCERR_NOARGS, r->parse, ln, ppos, NULL);
1.107 kristaps 1094: return(ROFF_IGN);
1.77 kristaps 1095: }
1096:
1.106 kristaps 1097: roffnode_push(r, tok, NULL, ln, ppos);
1.77 kristaps 1098:
1.88 kristaps 1099: r->last->rule = rule;
1100:
1.134 kristaps 1101: /*
1102: * An if-else will put the NEGATION of the current evaluated
1103: * conditional into the stack of rules.
1104: */
1105:
1.84 schwarze 1106: if (ROFF_ie == tok) {
1.134 kristaps 1107: if (r->rstackpos == RSTACK_MAX - 1) {
1108: mandoc_msg(MANDOCERR_MEM,
1109: r->parse, ln, ppos, NULL);
1110: return(ROFF_ERR);
1111: }
1112: r->rstack[++r->rstackpos] =
1113: ROFFRULE_DENY == r->last->rule ?
1114: ROFFRULE_ALLOW : ROFFRULE_DENY;
1.82 kristaps 1115: }
1.88 kristaps 1116:
1117: /* If the parent has false as its rule, then so do we. */
1118:
1.109 kristaps 1119: if (r->last->parent && ROFFRULE_DENY == r->last->parent->rule)
1.84 schwarze 1120: r->last->rule = ROFFRULE_DENY;
1.88 kristaps 1121:
1122: /*
1123: * Determine scope. If we're invoked with "\{" trailing the
1124: * conditional, then we're in a multiline scope. Else our scope
1125: * expires on the next line.
1126: */
1.74 kristaps 1127:
1.75 kristaps 1128: r->last->endspan = 1;
1129:
1130: if ('\\' == (*bufp)[pos] && '{' == (*bufp)[pos + 1]) {
1131: r->last->endspan = -1;
1132: pos += 2;
1.109 kristaps 1133: }
1.74 kristaps 1134:
1.77 kristaps 1135: /*
1136: * If there are no arguments on the line, the next-line scope is
1137: * assumed.
1138: */
1139:
1.75 kristaps 1140: if ('\0' == (*bufp)[pos])
1141: return(ROFF_IGN);
1.77 kristaps 1142:
1143: /* Otherwise re-run the roff parser after recalculating. */
1.74 kristaps 1144:
1.75 kristaps 1145: *offs = pos;
1146: return(ROFF_RERUN);
1.83 schwarze 1147: }
1148:
1149:
1150: /* ARGSUSED */
1151: static enum rofferr
1.92 schwarze 1152: roff_ds(ROFF_ARGS)
1153: {
1.96 kristaps 1154: char *name, *string;
1155:
1156: /*
1157: * A symbol is named by the first word following the macro
1158: * invocation up to a space. Its value is anything after the
1159: * name's trailing whitespace and optional double-quote. Thus,
1160: *
1161: * [.ds foo "bar " ]
1162: *
1163: * will have `bar " ' as its value.
1164: */
1.92 schwarze 1165:
1.121 schwarze 1166: string = *bufp + pos;
1167: name = roff_getname(r, &string, ln, pos);
1.92 schwarze 1168: if ('\0' == *name)
1169: return(ROFF_IGN);
1170:
1.121 schwarze 1171: /* Read past initial double-quote. */
1172: if ('"' == *string)
1.92 schwarze 1173: string++;
1174:
1.96 kristaps 1175: /* The rest is the value. */
1.106 kristaps 1176: roff_setstr(r, name, string, 0);
1.92 schwarze 1177: return(ROFF_IGN);
1178: }
1179:
1.147 kristaps 1180: int
1181: roff_regisset(const struct roff *r, enum regs reg)
1182: {
1183:
1184: return(r->regs[(int)reg].set);
1185: }
1186:
1187: unsigned int
1188: roff_regget(const struct roff *r, enum regs reg)
1189: {
1190:
1191: return(r->regs[(int)reg].u);
1192: }
1193:
1194: void
1195: roff_regunset(struct roff *r, enum regs reg)
1196: {
1197:
1198: r->regs[(int)reg].set = 0;
1199: }
1.92 schwarze 1200:
1201: /* ARGSUSED */
1202: static enum rofferr
1.89 kristaps 1203: roff_nr(ROFF_ARGS)
1.83 schwarze 1204: {
1.121 schwarze 1205: const char *key;
1206: char *val;
1.138 kristaps 1207: int iv;
1.89 kristaps 1208:
1.121 schwarze 1209: val = *bufp + pos;
1210: key = roff_getname(r, &val, ln, pos);
1.89 kristaps 1211:
1212: if (0 == strcmp(key, "nS")) {
1.147 kristaps 1213: r->regs[(int)REG_nS].set = 1;
1.149 kristaps 1214: if ((iv = mandoc_strntoi(val, strlen(val), 10)) >= 0)
1.147 kristaps 1215: r->regs[(int)REG_nS].u = (unsigned)iv;
1.138 kristaps 1216: else
1.147 kristaps 1217: r->regs[(int)REG_nS].u = 0u;
1.109 kristaps 1218: }
1219:
1.122 schwarze 1220: return(ROFF_IGN);
1221: }
1222:
1223: /* ARGSUSED */
1224: static enum rofferr
1225: roff_rm(ROFF_ARGS)
1226: {
1227: const char *name;
1228: char *cp;
1229:
1230: cp = *bufp + pos;
1231: while ('\0' != *cp) {
1.127 kristaps 1232: name = roff_getname(r, &cp, ln, (int)(cp - *bufp));
1.122 schwarze 1233: if ('\0' != *name)
1234: roff_setstr(r, name, NULL, 0);
1235: }
1.109 kristaps 1236: return(ROFF_IGN);
1237: }
1238:
1239: /* ARGSUSED */
1240: static enum rofferr
1241: roff_TE(ROFF_ARGS)
1242: {
1243:
1244: if (NULL == r->tbl)
1.128 kristaps 1245: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.115 kristaps 1246: else
1.151 kristaps 1247: tbl_end(&r->tbl);
1.109 kristaps 1248:
1.112 kristaps 1249: return(ROFF_IGN);
1250: }
1251:
1252: /* ARGSUSED */
1253: static enum rofferr
1254: roff_T_(ROFF_ARGS)
1255: {
1256:
1257: if (NULL == r->tbl)
1.128 kristaps 1258: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.112 kristaps 1259: else
1.116 kristaps 1260: tbl_restart(ppos, ln, r->tbl);
1.112 kristaps 1261:
1.109 kristaps 1262: return(ROFF_IGN);
1263: }
1264:
1.156 kristaps 1265: #if 0
1266: static int
1.151 kristaps 1267: roff_closeeqn(struct roff *r)
1268: {
1269:
1270: return(r->eqn && ROFF_EQN == eqn_end(&r->eqn) ? 1 : 0);
1271: }
1.156 kristaps 1272: #endif
1.151 kristaps 1273:
1.156 kristaps 1274: static void
1.151 kristaps 1275: roff_openeqn(struct roff *r, const char *name, int line,
1276: int offs, const char *buf)
1.125 kristaps 1277: {
1.151 kristaps 1278: struct eqn_node *e;
1279: int poff;
1.125 kristaps 1280:
1281: assert(NULL == r->eqn);
1.151 kristaps 1282: e = eqn_alloc(name, offs, line, r->parse);
1.125 kristaps 1283:
1284: if (r->last_eqn)
1285: r->last_eqn->next = e;
1286: else
1287: r->first_eqn = r->last_eqn = e;
1288:
1289: r->eqn = r->last_eqn = e;
1.151 kristaps 1290:
1291: if (buf) {
1292: poff = 0;
1293: eqn_read(&r->eqn, line, buf, offs, &poff);
1294: }
1295: }
1296:
1297: /* ARGSUSED */
1298: static enum rofferr
1299: roff_EQ(ROFF_ARGS)
1300: {
1301:
1302: roff_openeqn(r, *bufp + pos, ln, ppos, NULL);
1.125 kristaps 1303: return(ROFF_IGN);
1304: }
1305:
1306: /* ARGSUSED */
1307: static enum rofferr
1308: roff_EN(ROFF_ARGS)
1309: {
1310:
1.128 kristaps 1311: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.125 kristaps 1312: return(ROFF_IGN);
1313: }
1314:
1315: /* ARGSUSED */
1316: static enum rofferr
1.109 kristaps 1317: roff_TS(ROFF_ARGS)
1318: {
1.118 kristaps 1319: struct tbl_node *t;
1.89 kristaps 1320:
1.115 kristaps 1321: if (r->tbl) {
1.128 kristaps 1322: mandoc_msg(MANDOCERR_SCOPEBROKEN, r->parse, ln, ppos, NULL);
1.151 kristaps 1323: tbl_end(&r->tbl);
1.115 kristaps 1324: }
1.83 schwarze 1325:
1.128 kristaps 1326: t = tbl_alloc(ppos, ln, r->parse);
1.113 kristaps 1327:
1328: if (r->last_tbl)
1329: r->last_tbl->next = t;
1330: else
1331: r->first_tbl = r->last_tbl = t;
1332:
1333: r->tbl = r->last_tbl = t;
1.83 schwarze 1334: return(ROFF_IGN);
1.92 schwarze 1335: }
1336:
1.105 kristaps 1337: /* ARGSUSED */
1338: static enum rofferr
1339: roff_so(ROFF_ARGS)
1340: {
1341: char *name;
1342:
1.128 kristaps 1343: mandoc_msg(MANDOCERR_SO, r->parse, ln, ppos, NULL);
1.105 kristaps 1344:
1345: /*
1346: * Handle `so'. Be EXTREMELY careful, as we shouldn't be
1347: * opening anything that's not in our cwd or anything beneath
1348: * it. Thus, explicitly disallow traversing up the file-system
1349: * or using absolute paths.
1350: */
1351:
1352: name = *bufp + pos;
1353: if ('/' == *name || strstr(name, "../") || strstr(name, "/..")) {
1.128 kristaps 1354: mandoc_msg(MANDOCERR_SOPATH, r->parse, ln, pos, NULL);
1.105 kristaps 1355: return(ROFF_ERR);
1356: }
1357:
1358: *offs = pos;
1359: return(ROFF_SO);
1360: }
1.92 schwarze 1361:
1.106 kristaps 1362: /* ARGSUSED */
1363: static enum rofferr
1364: roff_userdef(ROFF_ARGS)
1.99 kristaps 1365: {
1.106 kristaps 1366: const char *arg[9];
1367: char *cp, *n1, *n2;
1.119 schwarze 1368: int i;
1.106 kristaps 1369:
1370: /*
1371: * Collect pointers to macro argument strings
1372: * and null-terminate them.
1373: */
1374: cp = *bufp + pos;
1.119 schwarze 1375: for (i = 0; i < 9; i++)
1.120 schwarze 1376: arg[i] = '\0' == *cp ? "" :
1.136 kristaps 1377: mandoc_getarg(r->parse, &cp, ln, &pos);
1.99 kristaps 1378:
1.106 kristaps 1379: /*
1380: * Expand macro arguments.
1.99 kristaps 1381: */
1.106 kristaps 1382: *szp = 0;
1383: n1 = cp = mandoc_strdup(r->current_string);
1384: while (NULL != (cp = strstr(cp, "\\$"))) {
1385: i = cp[2] - '1';
1386: if (0 > i || 8 < i) {
1387: /* Not an argument invocation. */
1388: cp += 2;
1389: continue;
1390: }
1391:
1392: *szp = strlen(n1) - 3 + strlen(arg[i]) + 1;
1393: n2 = mandoc_malloc(*szp);
1394:
1395: strlcpy(n2, n1, (size_t)(cp - n1 + 1));
1396: strlcat(n2, arg[i], *szp);
1397: strlcat(n2, cp + 3, *szp);
1398:
1399: cp = n2 + (cp - n1);
1400: free(n1);
1401: n1 = n2;
1.99 kristaps 1402: }
1403:
1.106 kristaps 1404: /*
1405: * Replace the macro invocation
1406: * by the expanded macro.
1407: */
1408: free(*bufp);
1409: *bufp = n1;
1410: if (0 == *szp)
1411: *szp = strlen(*bufp) + 1;
1412:
1413: return(*szp > 1 && '\n' == (*bufp)[(int)*szp - 2] ?
1414: ROFF_REPARSE : ROFF_APPEND);
1.99 kristaps 1415: }
1.121 schwarze 1416:
1417: static char *
1418: roff_getname(struct roff *r, char **cpp, int ln, int pos)
1419: {
1420: char *name, *cp;
1421:
1422: name = *cpp;
1423: if ('\0' == *name)
1424: return(name);
1425:
1426: /* Read until end of name. */
1427: for (cp = name; '\0' != *cp && ' ' != *cp; cp++) {
1428: if ('\\' != *cp)
1429: continue;
1430: cp++;
1431: if ('\\' == *cp)
1432: continue;
1.128 kristaps 1433: mandoc_msg(MANDOCERR_NAMESC, r->parse, ln, pos, NULL);
1.121 schwarze 1434: *cp = '\0';
1435: name = cp;
1436: }
1437:
1438: /* Nil-terminate name. */
1439: if ('\0' != *cp)
1440: *(cp++) = '\0';
1441:
1442: /* Read past spaces. */
1443: while (' ' == *cp)
1444: cp++;
1445:
1446: *cpp = cp;
1447: return(name);
1448: }
1449:
1.106 kristaps 1450: /*
1451: * Store *string into the user-defined string called *name.
1452: * In multiline mode, append to an existing entry and append '\n';
1453: * else replace the existing entry, if there is one.
1454: * To clear an existing entry, call with (*r, *name, NULL, 0).
1455: */
1.94 kristaps 1456: static void
1.106 kristaps 1457: roff_setstr(struct roff *r, const char *name, const char *string,
1458: int multiline)
1.92 schwarze 1459: {
1460: struct roffstr *n;
1.106 kristaps 1461: char *c;
1462: size_t oldch, newch;
1.92 schwarze 1463:
1.106 kristaps 1464: /* Search for an existing string with the same name. */
1.94 kristaps 1465: n = r->first_string;
1.92 schwarze 1466: while (n && strcmp(name, n->name))
1467: n = n->next;
1.94 kristaps 1468:
1469: if (NULL == n) {
1.106 kristaps 1470: /* Create a new string table entry. */
1.94 kristaps 1471: n = mandoc_malloc(sizeof(struct roffstr));
1.106 kristaps 1472: n->name = mandoc_strdup(name);
1473: n->string = NULL;
1.94 kristaps 1474: n->next = r->first_string;
1475: r->first_string = n;
1.106 kristaps 1476: } else if (0 == multiline) {
1477: /* In multiline mode, append; else replace. */
1.92 schwarze 1478: free(n->string);
1.106 kristaps 1479: n->string = NULL;
1480: }
1481:
1482: if (NULL == string)
1483: return;
1484:
1485: /*
1486: * One additional byte for the '\n' in multiline mode,
1487: * and one for the terminating '\0'.
1488: */
1.127 kristaps 1489: newch = strlen(string) + (multiline ? 2u : 1u);
1.106 kristaps 1490: if (NULL == n->string) {
1491: n->string = mandoc_malloc(newch);
1492: *n->string = '\0';
1493: oldch = 0;
1494: } else {
1495: oldch = strlen(n->string);
1496: n->string = mandoc_realloc(n->string, oldch + newch);
1497: }
1498:
1499: /* Skip existing content in the destination buffer. */
1.127 kristaps 1500: c = n->string + (int)oldch;
1.106 kristaps 1501:
1502: /* Append new content to the destination buffer. */
1503: while (*string) {
1504: /*
1505: * Rudimentary roff copy mode:
1506: * Handle escaped backslashes.
1507: */
1508: if ('\\' == *string && '\\' == *(string + 1))
1509: string++;
1510: *c++ = *string++;
1511: }
1.94 kristaps 1512:
1.106 kristaps 1513: /* Append terminating bytes. */
1514: if (multiline)
1515: *c++ = '\n';
1516: *c = '\0';
1.92 schwarze 1517: }
1518:
1.94 kristaps 1519: static const char *
1520: roff_getstrn(const struct roff *r, const char *name, size_t len)
1.92 schwarze 1521: {
1.94 kristaps 1522: const struct roffstr *n;
1.92 schwarze 1523:
1.157 kristaps 1524: for (n = r->first_string; n; n = n->next)
1525: if (0 == strncmp(name, n->name, len) &&
1526: '\0' == n->name[(int)len])
1527: return(n->string);
1.94 kristaps 1528:
1.157 kristaps 1529: return(NULL);
1.92 schwarze 1530: }
1531:
1.94 kristaps 1532: static void
1533: roff_freestr(struct roff *r)
1.92 schwarze 1534: {
1535: struct roffstr *n, *nn;
1536:
1.94 kristaps 1537: for (n = r->first_string; n; n = nn) {
1.92 schwarze 1538: free(n->name);
1539: free(n->string);
1540: nn = n->next;
1541: free(n);
1542: }
1.94 kristaps 1543:
1544: r->first_string = NULL;
1.114 kristaps 1545: }
1546:
1547: const struct tbl_span *
1548: roff_span(const struct roff *r)
1549: {
1550:
1551: return(r->tbl ? tbl_span(r->tbl) : NULL);
1.125 kristaps 1552: }
1553:
1554: const struct eqn *
1555: roff_eqn(const struct roff *r)
1556: {
1557:
1558: return(r->last_eqn ? &r->last_eqn->eqn : NULL);
1.151 kristaps 1559: }
1560:
1561: char
1562: roff_eqndelim(const struct roff *r)
1563: {
1564:
1565: return('\0');
1.74 kristaps 1566: }
CVSweb