Annotation of mandoc/roff.c, Revision 1.155
1.155 ! kristaps 1: /* $Id: roff.c,v 1.154 2011/07/27 07:09:41 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);
1.155 ! kristaps 169: static enum rofft roff_parse(struct roff *, const char *, int *);
! 170: static enum rofferr roff_parsetext(char *);
1.154 kristaps 171: static void roff_res(struct roff *,
1.142 kristaps 172: char **, size_t *, int, int);
1.122 schwarze 173: static enum rofferr roff_rm(ROFF_ARGS);
1.94 kristaps 174: static void roff_setstr(struct roff *,
1.106 kristaps 175: const char *, const char *, int);
1.105 kristaps 176: static enum rofferr roff_so(ROFF_ARGS);
1.109 kristaps 177: static enum rofferr roff_TE(ROFF_ARGS);
178: static enum rofferr roff_TS(ROFF_ARGS);
1.125 kristaps 179: static enum rofferr roff_EQ(ROFF_ARGS);
180: static enum rofferr roff_EN(ROFF_ARGS);
1.112 kristaps 181: static enum rofferr roff_T_(ROFF_ARGS);
1.106 kristaps 182: static enum rofferr roff_userdef(ROFF_ARGS);
1.67 kristaps 183:
1.155 ! kristaps 184: /* See roffhash_find() */
1.85 kristaps 185:
186: #define ASCII_HI 126
187: #define ASCII_LO 33
188: #define HASHWIDTH (ASCII_HI - ASCII_LO + 1)
189:
190: static struct roffmac *hash[HASHWIDTH];
191:
192: static struct roffmac roffs[ROFF_MAX] = {
1.103 kristaps 193: { "ad", roff_line_ignore, NULL, NULL, 0, NULL },
1.85 kristaps 194: { "am", roff_block, roff_block_text, roff_block_sub, 0, NULL },
195: { "ami", roff_block, roff_block_text, roff_block_sub, 0, NULL },
196: { "am1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
197: { "de", roff_block, roff_block_text, roff_block_sub, 0, NULL },
198: { "dei", roff_block, roff_block_text, roff_block_sub, 0, NULL },
199: { "de1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.92 schwarze 200: { "ds", roff_ds, NULL, NULL, 0, NULL },
1.85 kristaps 201: { "el", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
1.103 kristaps 202: { "hy", roff_line_ignore, NULL, NULL, 0, NULL },
1.85 kristaps 203: { "ie", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
204: { "if", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
205: { "ig", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.123 schwarze 206: { "it", roff_line_ignore, NULL, NULL, 0, NULL },
1.103 kristaps 207: { "ne", roff_line_ignore, NULL, NULL, 0, NULL },
208: { "nh", roff_line_ignore, NULL, NULL, 0, NULL },
1.104 kristaps 209: { "nr", roff_nr, NULL, NULL, 0, NULL },
1.124 schwarze 210: { "ns", roff_line_ignore, NULL, NULL, 0, NULL },
211: { "ps", roff_line_ignore, NULL, NULL, 0, NULL },
1.122 schwarze 212: { "rm", roff_rm, NULL, NULL, 0, NULL },
1.105 kristaps 213: { "so", roff_so, NULL, NULL, 0, NULL },
1.124 schwarze 214: { "ta", roff_line_ignore, NULL, NULL, 0, NULL },
1.103 kristaps 215: { "tr", roff_line_ignore, NULL, NULL, 0, NULL },
1.109 kristaps 216: { "TS", roff_TS, NULL, NULL, 0, NULL },
217: { "TE", roff_TE, NULL, NULL, 0, NULL },
1.112 kristaps 218: { "T&", roff_T_, NULL, NULL, 0, NULL },
1.125 kristaps 219: { "EQ", roff_EQ, NULL, NULL, 0, NULL },
220: { "EN", roff_EN, NULL, NULL, 0, NULL },
1.85 kristaps 221: { ".", roff_cblock, NULL, NULL, 0, NULL },
222: { "\\}", roff_ccond, NULL, NULL, 0, NULL },
1.106 kristaps 223: { NULL, roff_userdef, NULL, NULL, 0, NULL },
1.67 kristaps 224: };
225:
1.141 kristaps 226: /* Array of injected predefined strings. */
227: #define PREDEFS_MAX 38
228: static const struct predef predefs[PREDEFS_MAX] = {
229: #include "predefs.in"
230: };
231:
1.155 ! kristaps 232: /* See roffhash_find() */
1.85 kristaps 233: #define ROFF_HASH(p) (p[0] - ASCII_LO)
234:
235: static void
1.155 ! kristaps 236: roffhash_init(void)
1.85 kristaps 237: {
238: struct roffmac *n;
239: int buc, i;
240:
1.106 kristaps 241: for (i = 0; i < (int)ROFF_USERDEF; i++) {
1.85 kristaps 242: assert(roffs[i].name[0] >= ASCII_LO);
243: assert(roffs[i].name[0] <= ASCII_HI);
244:
245: buc = ROFF_HASH(roffs[i].name);
246:
247: if (NULL != (n = hash[buc])) {
248: for ( ; n->next; n = n->next)
249: /* Do nothing. */ ;
250: n->next = &roffs[i];
251: } else
252: hash[buc] = &roffs[i];
253: }
254: }
255:
1.67 kristaps 256: /*
257: * Look up a roff token by its name. Returns ROFF_MAX if no macro by
258: * the nil-terminated string name could be found.
259: */
260: static enum rofft
1.155 ! kristaps 261: roffhash_find(const char *p, size_t s)
1.67 kristaps 262: {
1.85 kristaps 263: int buc;
264: struct roffmac *n;
1.67 kristaps 265:
1.85 kristaps 266: /*
267: * libroff has an extremely simple hashtable, for the time
268: * being, which simply keys on the first character, which must
269: * be printable, then walks a chain. It works well enough until
270: * optimised.
271: */
272:
273: if (p[0] < ASCII_LO || p[0] > ASCII_HI)
274: return(ROFF_MAX);
275:
276: buc = ROFF_HASH(p);
277:
278: if (NULL == (n = hash[buc]))
279: return(ROFF_MAX);
280: for ( ; n; n = n->next)
1.106 kristaps 281: if (0 == strncmp(n->name, p, s) && '\0' == n->name[(int)s])
1.85 kristaps 282: return((enum rofft)(n - roffs));
1.67 kristaps 283:
284: return(ROFF_MAX);
285: }
286:
287:
288: /*
289: * Pop the current node off of the stack of roff instructions currently
290: * pending.
291: */
292: static void
293: roffnode_pop(struct roff *r)
294: {
295: struct roffnode *p;
296:
1.75 kristaps 297: assert(r->last);
298: p = r->last;
1.82 kristaps 299:
1.75 kristaps 300: r->last = r->last->parent;
1.106 kristaps 301: free(p->name);
302: free(p->end);
1.67 kristaps 303: free(p);
304: }
305:
306:
307: /*
308: * Push a roff node onto the instruction stack. This must later be
309: * removed with roffnode_pop().
310: */
1.98 schwarze 311: static void
1.106 kristaps 312: roffnode_push(struct roff *r, enum rofft tok, const char *name,
313: int line, int col)
1.67 kristaps 314: {
315: struct roffnode *p;
316:
1.98 schwarze 317: p = mandoc_calloc(1, sizeof(struct roffnode));
1.67 kristaps 318: p->tok = tok;
1.106 kristaps 319: if (name)
320: p->name = mandoc_strdup(name);
1.67 kristaps 321: p->parent = r->last;
322: p->line = line;
323: p->col = col;
1.79 kristaps 324: p->rule = p->parent ? p->parent->rule : ROFFRULE_DENY;
1.67 kristaps 325:
326: r->last = p;
327: }
328:
329:
330: static void
331: roff_free1(struct roff *r)
332: {
1.118 kristaps 333: struct tbl_node *t;
1.125 kristaps 334: struct eqn_node *e;
1.67 kristaps 335:
1.125 kristaps 336: while (NULL != (t = r->first_tbl)) {
1.113 kristaps 337: r->first_tbl = t->next;
338: tbl_free(t);
1.109 kristaps 339: }
340:
1.113 kristaps 341: r->first_tbl = r->last_tbl = r->tbl = NULL;
342:
1.125 kristaps 343: while (NULL != (e = r->first_eqn)) {
344: r->first_eqn = e->next;
345: eqn_free(e);
346: }
347:
348: r->first_eqn = r->last_eqn = r->eqn = NULL;
349:
1.67 kristaps 350: while (r->last)
351: roffnode_pop(r);
1.109 kristaps 352:
1.94 kristaps 353: roff_freestr(r);
1.67 kristaps 354: }
355:
356:
357: void
358: roff_reset(struct roff *r)
359: {
1.143 kristaps 360: int i;
1.67 kristaps 361:
362: roff_free1(r);
1.143 kristaps 363:
1.147 kristaps 364: memset(&r->regs, 0, sizeof(struct reg) * REG__MAX);
365:
1.143 kristaps 366: for (i = 0; i < PREDEFS_MAX; i++)
367: roff_setstr(r, predefs[i].name, predefs[i].str, 0);
1.67 kristaps 368: }
369:
370:
371: void
372: roff_free(struct roff *r)
373: {
374:
375: roff_free1(r);
376: free(r);
377: }
378:
379:
380: struct roff *
1.147 kristaps 381: roff_alloc(struct mparse *parse)
1.67 kristaps 382: {
383: struct roff *r;
1.141 kristaps 384: int i;
1.67 kristaps 385:
1.98 schwarze 386: r = mandoc_calloc(1, sizeof(struct roff));
1.128 kristaps 387: r->parse = parse;
1.82 kristaps 388: r->rstackpos = -1;
1.85 kristaps 389:
1.155 ! kristaps 390: roffhash_init();
1.141 kristaps 391:
392: for (i = 0; i < PREDEFS_MAX; i++)
393: roff_setstr(r, predefs[i].name, predefs[i].str, 0);
394:
1.67 kristaps 395: return(r);
396: }
397:
1.94 kristaps 398: /*
399: * Pre-filter each and every line for reserved words (one beginning with
400: * `\*', e.g., `\*(ab'). These must be handled before the actual line
401: * is processed.
1.153 kristaps 402: * This also checks the syntax of regular escapes.
1.154 kristaps 403: */
404: static void
1.142 kristaps 405: roff_res(struct roff *r, char **bufp, size_t *szp, int ln, int pos)
1.94 kristaps 406: {
1.152 kristaps 407: enum mandoc_esc esc;
1.108 schwarze 408: const char *stesc; /* start of an escape sequence ('\\') */
409: const char *stnam; /* start of the name, after "[(*" */
410: const char *cp; /* end of the name, e.g. before ']' */
411: const char *res; /* the string to be substituted */
1.94 kristaps 412: int i, maxl;
413: size_t nsz;
414: char *n;
415:
1.154 kristaps 416: again:
1.108 schwarze 417: cp = *bufp + pos;
418: while (NULL != (cp = strchr(cp, '\\'))) {
419: stesc = cp++;
420:
421: /*
422: * The second character must be an asterisk.
423: * If it isn't, skip it anyway: It is escaped,
424: * so it can't start another escape sequence.
425: */
426:
427: if ('\0' == *cp)
1.154 kristaps 428: return;
1.152 kristaps 429:
430: if ('*' != *cp) {
431: res = cp;
432: esc = mandoc_escape(&cp, NULL, NULL);
433: if (ESCAPE_ERROR != esc)
434: continue;
435: cp = res;
1.153 kristaps 436: mandoc_msg
437: (MANDOCERR_BADESCAPE, r->parse,
438: ln, (int)(stesc - *bufp), NULL);
1.154 kristaps 439: return;
1.152 kristaps 440: }
441:
442: cp++;
1.108 schwarze 443:
444: /*
445: * The third character decides the length
446: * of the name of the string.
447: * Save a pointer to the name.
448: */
449:
1.94 kristaps 450: switch (*cp) {
1.108 schwarze 451: case ('\0'):
1.154 kristaps 452: return;
1.94 kristaps 453: case ('('):
454: cp++;
455: maxl = 2;
456: break;
457: case ('['):
458: cp++;
459: maxl = 0;
460: break;
461: default:
462: maxl = 1;
463: break;
464: }
1.108 schwarze 465: stnam = cp;
1.94 kristaps 466:
1.108 schwarze 467: /* Advance to the end of the name. */
1.94 kristaps 468:
469: for (i = 0; 0 == maxl || i < maxl; i++, cp++) {
1.153 kristaps 470: if ('\0' == *cp) {
471: mandoc_msg
472: (MANDOCERR_BADESCAPE,
473: r->parse, ln,
474: (int)(stesc - *bufp), NULL);
1.154 kristaps 475: return;
1.153 kristaps 476: }
1.94 kristaps 477: if (0 == maxl && ']' == *cp)
478: break;
479: }
480:
1.108 schwarze 481: /*
482: * Retrieve the replacement string; if it is
483: * undefined, resume searching for escapes.
484: */
485:
486: res = roff_getstrn(r, stnam, (size_t)i);
1.94 kristaps 487:
488: if (NULL == res) {
1.153 kristaps 489: mandoc_msg
490: (MANDOCERR_BADESCAPE, r->parse,
491: ln, (int)(stesc - *bufp), NULL);
1.142 kristaps 492: res = "";
1.94 kristaps 493: }
494:
1.108 schwarze 495: /* Replace the escape sequence by the string. */
496:
1.154 kristaps 497: pos += (stesc - *bufp);
498:
1.94 kristaps 499: nsz = *szp + strlen(res) + 1;
500: n = mandoc_malloc(nsz);
501:
1.108 schwarze 502: strlcpy(n, *bufp, (size_t)(stesc - *bufp + 1));
1.94 kristaps 503: strlcat(n, res, nsz);
504: strlcat(n, cp + (maxl ? 0 : 1), nsz);
505:
506: free(*bufp);
507:
508: *bufp = n;
509: *szp = nsz;
1.154 kristaps 510: goto again;
511: }
512: }
513:
514: /*
515: * Process text streams: convert all breakable hyphens into ASCII_HYPH.
516: */
517: static enum rofferr
518: roff_parsetext(char *p)
519: {
1.155 ! kristaps 520: char l, r;
1.154 kristaps 521: size_t sz;
522: const char *start;
523: enum mandoc_esc esc;
524:
525: start = p;
526:
527: while ('\0' != *p) {
528: sz = strcspn(p, "-\\");
529: p += sz;
530:
531: if ('\\' == *p) {
532: /* Skip over escapes. */
533: p++;
534: esc = mandoc_escape
535: ((const char **)&p, NULL, NULL);
536: if (ESCAPE_ERROR == esc)
537: break;
1.155 ! kristaps 538: continue;
! 539: } else if ('-' != *p || p == start)
! 540: continue;
! 541:
! 542: l = *(p - 1);
! 543: r = *(p + 1);
! 544:
! 545: if ('\\' != l &&
! 546: '\t' != r && '\t' != l &&
! 547: ' ' != r && ' ' != l &&
! 548: '-' != r && '-' != l &&
! 549: ! isdigit((unsigned char)l) &&
! 550: ! isdigit((unsigned char)r))
! 551: *p = ASCII_HYPH;
! 552: p++;
1.94 kristaps 553: }
554:
1.154 kristaps 555: return(ROFF_CONT);
1.94 kristaps 556: }
557:
1.67 kristaps 558: enum rofferr
1.90 kristaps 559: roff_parseln(struct roff *r, int ln, char **bufp,
560: size_t *szp, int pos, int *offs)
1.67 kristaps 561: {
562: enum rofft t;
1.109 kristaps 563: enum rofferr e;
1.130 kristaps 564: int ppos, ctl;
1.79 kristaps 565:
566: /*
1.94 kristaps 567: * Run the reserved-word filter only if we have some reserved
568: * words to fill in.
569: */
570:
1.154 kristaps 571: roff_res(r, bufp, szp, ln, pos);
1.94 kristaps 572:
1.130 kristaps 573: ppos = pos;
574: ctl = mandoc_getcontrol(*bufp, &pos);
575:
1.94 kristaps 576: /*
1.79 kristaps 577: * First, if a scope is open and we're not a macro, pass the
578: * text through the macro's filter. If a scope isn't open and
579: * we're not a macro, just let it through.
1.125 kristaps 580: * Finally, if there's an equation scope open, divert it into it
581: * no matter our state.
1.79 kristaps 582: */
1.74 kristaps 583:
1.130 kristaps 584: if (r->last && ! ctl) {
1.78 kristaps 585: t = r->last->tok;
586: assert(roffs[t].text);
1.109 kristaps 587: e = (*roffs[t].text)
588: (r, t, bufp, szp, ln, pos, pos, offs);
589: assert(ROFF_IGN == e || ROFF_CONT == e);
1.125 kristaps 590: if (ROFF_CONT != e)
591: return(e);
592: if (r->eqn)
1.146 kristaps 593: return(eqn_read(&r->eqn, ln, *bufp, pos, offs));
1.125 kristaps 594: if (r->tbl)
1.130 kristaps 595: return(tbl_read(r->tbl, ln, *bufp, pos));
1.154 kristaps 596: return(roff_parsetext(*bufp + pos));
1.130 kristaps 597: } else if ( ! ctl) {
1.125 kristaps 598: if (r->eqn)
1.146 kristaps 599: return(eqn_read(&r->eqn, ln, *bufp, pos, offs));
1.109 kristaps 600: if (r->tbl)
1.130 kristaps 601: return(tbl_read(r->tbl, ln, *bufp, pos));
1.154 kristaps 602: return(roff_parsetext(*bufp + pos));
1.125 kristaps 603: } else if (r->eqn)
1.146 kristaps 604: return(eqn_read(&r->eqn, ln, *bufp, ppos, offs));
1.67 kristaps 605:
1.79 kristaps 606: /*
607: * If a scope is open, go to the child handler for that macro,
608: * as it may want to preprocess before doing anything with it.
1.125 kristaps 609: * Don't do so if an equation is open.
1.79 kristaps 610: */
1.78 kristaps 611:
1.79 kristaps 612: if (r->last) {
613: t = r->last->tok;
614: assert(roffs[t].sub);
615: return((*roffs[t].sub)
1.90 kristaps 616: (r, t, bufp, szp,
1.130 kristaps 617: ln, ppos, pos, offs));
1.79 kristaps 618: }
1.78 kristaps 619:
1.79 kristaps 620: /*
621: * Lastly, as we've no scope open, try to look up and execute
622: * the new macro. If no macro is found, simply return and let
623: * the compilers handle it.
624: */
1.67 kristaps 625:
1.106 kristaps 626: if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos)))
1.79 kristaps 627: return(ROFF_CONT);
1.67 kristaps 628:
1.75 kristaps 629: assert(roffs[t].proc);
1.78 kristaps 630: return((*roffs[t].proc)
1.90 kristaps 631: (r, t, bufp, szp,
632: ln, ppos, pos, offs));
1.74 kristaps 633: }
634:
635:
1.117 kristaps 636: void
1.74 kristaps 637: roff_endparse(struct roff *r)
638: {
639:
1.110 kristaps 640: if (r->last)
1.128 kristaps 641: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.109 kristaps 642: r->last->line, r->last->col, NULL);
1.117 kristaps 643:
1.125 kristaps 644: if (r->eqn) {
1.128 kristaps 645: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.148 kristaps 646: r->eqn->eqn.ln, r->eqn->eqn.pos, NULL);
1.151 kristaps 647: eqn_end(&r->eqn);
1.125 kristaps 648: }
649:
1.117 kristaps 650: if (r->tbl) {
1.128 kristaps 651: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.117 kristaps 652: r->tbl->line, r->tbl->pos, NULL);
1.151 kristaps 653: tbl_end(&r->tbl);
1.117 kristaps 654: }
1.67 kristaps 655: }
656:
657: /*
658: * Parse a roff node's type from the input buffer. This must be in the
659: * form of ".foo xxx" in the usual way.
660: */
661: static enum rofft
1.106 kristaps 662: roff_parse(struct roff *r, const char *buf, int *pos)
1.67 kristaps 663: {
1.106 kristaps 664: const char *mac;
665: size_t maclen;
1.67 kristaps 666: enum rofft t;
667:
1.144 kristaps 668: if ('\0' == buf[*pos] || '"' == buf[*pos] ||
669: '\t' == buf[*pos] || ' ' == buf[*pos])
1.67 kristaps 670: return(ROFF_MAX);
671:
1.144 kristaps 672: /*
673: * We stop the macro parse at an escape, tab, space, or nil.
674: * However, `\}' is also a valid macro, so make sure we don't
675: * clobber it by seeing the `\' as the end of token.
676: */
677:
1.106 kristaps 678: mac = buf + *pos;
1.144 kristaps 679: maclen = strcspn(mac + 1, " \\\t\0") + 1;
1.67 kristaps 680:
1.106 kristaps 681: t = (r->current_string = roff_getstrn(r, mac, maclen))
1.155 ! kristaps 682: ? ROFF_USERDEF : roffhash_find(mac, maclen);
1.67 kristaps 683:
1.127 kristaps 684: *pos += (int)maclen;
1.130 kristaps 685:
1.67 kristaps 686: while (buf[*pos] && ' ' == buf[*pos])
687: (*pos)++;
688:
689: return(t);
690: }
691:
692: /* ARGSUSED */
693: static enum rofferr
1.76 kristaps 694: roff_cblock(ROFF_ARGS)
1.67 kristaps 695: {
696:
1.79 kristaps 697: /*
698: * A block-close `..' should only be invoked as a child of an
699: * ignore macro, otherwise raise a warning and just ignore it.
700: */
701:
1.76 kristaps 702: if (NULL == r->last) {
1.128 kristaps 703: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.76 kristaps 704: return(ROFF_IGN);
705: }
1.67 kristaps 706:
1.81 kristaps 707: switch (r->last->tok) {
708: case (ROFF_am):
709: /* FALLTHROUGH */
710: case (ROFF_ami):
711: /* FALLTHROUGH */
712: case (ROFF_am1):
713: /* FALLTHROUGH */
714: case (ROFF_de):
1.108 schwarze 715: /* ROFF_de1 is remapped to ROFF_de in roff_block(). */
1.81 kristaps 716: /* FALLTHROUGH */
717: case (ROFF_dei):
718: /* FALLTHROUGH */
719: case (ROFF_ig):
720: break;
721: default:
1.128 kristaps 722: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.67 kristaps 723: return(ROFF_IGN);
1.76 kristaps 724: }
1.67 kristaps 725:
1.76 kristaps 726: if ((*bufp)[pos])
1.128 kristaps 727: mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.71 kristaps 728:
729: roffnode_pop(r);
1.76 kristaps 730: roffnode_cleanscope(r);
731: return(ROFF_IGN);
1.71 kristaps 732:
1.67 kristaps 733: }
734:
735:
1.76 kristaps 736: static void
737: roffnode_cleanscope(struct roff *r)
1.67 kristaps 738: {
739:
1.76 kristaps 740: while (r->last) {
741: if (--r->last->endspan < 0)
742: break;
743: roffnode_pop(r);
744: }
1.67 kristaps 745: }
746:
747:
1.75 kristaps 748: /* ARGSUSED */
1.74 kristaps 749: static enum rofferr
1.75 kristaps 750: roff_ccond(ROFF_ARGS)
1.74 kristaps 751: {
752:
1.76 kristaps 753: if (NULL == r->last) {
1.128 kristaps 754: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.76 kristaps 755: return(ROFF_IGN);
756: }
757:
1.82 kristaps 758: switch (r->last->tok) {
759: case (ROFF_el):
760: /* FALLTHROUGH */
761: case (ROFF_ie):
762: /* FALLTHROUGH */
763: case (ROFF_if):
764: break;
765: default:
1.128 kristaps 766: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.75 kristaps 767: return(ROFF_IGN);
768: }
769:
1.76 kristaps 770: if (r->last->endspan > -1) {
1.128 kristaps 771: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.76 kristaps 772: return(ROFF_IGN);
773: }
774:
775: if ((*bufp)[pos])
1.128 kristaps 776: mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.76 kristaps 777:
1.75 kristaps 778: roffnode_pop(r);
1.76 kristaps 779: roffnode_cleanscope(r);
780: return(ROFF_IGN);
781: }
782:
1.75 kristaps 783:
1.76 kristaps 784: /* ARGSUSED */
785: static enum rofferr
1.80 kristaps 786: roff_block(ROFF_ARGS)
1.76 kristaps 787: {
1.78 kristaps 788: int sv;
789: size_t sz;
1.106 kristaps 790: char *name;
791:
792: name = NULL;
1.76 kristaps 793:
1.106 kristaps 794: if (ROFF_ig != tok) {
795: if ('\0' == (*bufp)[pos]) {
1.128 kristaps 796: mandoc_msg(MANDOCERR_NOARGS, r->parse, ln, ppos, NULL);
1.106 kristaps 797: return(ROFF_IGN);
798: }
1.107 kristaps 799:
800: /*
801: * Re-write `de1', since we don't really care about
802: * groff's strange compatibility mode, into `de'.
803: */
804:
1.106 kristaps 805: if (ROFF_de1 == tok)
806: tok = ROFF_de;
807: if (ROFF_de == tok)
808: name = *bufp + pos;
809: else
1.128 kristaps 810: mandoc_msg(MANDOCERR_REQUEST, r->parse, ln, ppos,
1.106 kristaps 811: roffs[tok].name);
1.107 kristaps 812:
1.131 schwarze 813: while ((*bufp)[pos] && ! isspace((unsigned char)(*bufp)[pos]))
1.80 kristaps 814: pos++;
1.107 kristaps 815:
1.131 schwarze 816: while (isspace((unsigned char)(*bufp)[pos]))
1.106 kristaps 817: (*bufp)[pos++] = '\0';
1.80 kristaps 818: }
819:
1.106 kristaps 820: roffnode_push(r, tok, name, ln, ppos);
821:
822: /*
823: * At the beginning of a `de' macro, clear the existing string
824: * with the same name, if there is one. New content will be
825: * added from roff_block_text() in multiline mode.
826: */
1.107 kristaps 827:
1.106 kristaps 828: if (ROFF_de == tok)
1.108 schwarze 829: roff_setstr(r, name, "", 0);
1.76 kristaps 830:
1.79 kristaps 831: if ('\0' == (*bufp)[pos])
1.78 kristaps 832: return(ROFF_IGN);
833:
1.107 kristaps 834: /* If present, process the custom end-of-line marker. */
835:
1.78 kristaps 836: sv = pos;
1.131 schwarze 837: while ((*bufp)[pos] && ! isspace((unsigned char)(*bufp)[pos]))
1.78 kristaps 838: pos++;
839:
840: /*
841: * Note: groff does NOT like escape characters in the input.
842: * Instead of detecting this, we're just going to let it fly and
843: * to hell with it.
844: */
845:
846: assert(pos > sv);
847: sz = (size_t)(pos - sv);
848:
1.79 kristaps 849: if (1 == sz && '.' == (*bufp)[sv])
850: return(ROFF_IGN);
851:
1.98 schwarze 852: r->last->end = mandoc_malloc(sz + 1);
1.78 kristaps 853:
854: memcpy(r->last->end, *bufp + sv, sz);
855: r->last->end[(int)sz] = '\0';
856:
1.77 kristaps 857: if ((*bufp)[pos])
1.128 kristaps 858: mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.74 kristaps 859:
1.78 kristaps 860: return(ROFF_IGN);
861: }
862:
863:
864: /* ARGSUSED */
865: static enum rofferr
1.80 kristaps 866: roff_block_sub(ROFF_ARGS)
1.79 kristaps 867: {
868: enum rofft t;
869: int i, j;
870:
871: /*
872: * First check whether a custom macro exists at this level. If
873: * it does, then check against it. This is some of groff's
874: * stranger behaviours. If we encountered a custom end-scope
875: * tag and that tag also happens to be a "real" macro, then we
876: * need to try interpreting it again as a real macro. If it's
877: * not, then return ignore. Else continue.
878: */
879:
880: if (r->last->end) {
1.130 kristaps 881: for (i = pos, j = 0; r->last->end[j]; j++, i++)
1.79 kristaps 882: if ((*bufp)[i] != r->last->end[j])
883: break;
884:
885: if ('\0' == r->last->end[j] &&
886: ('\0' == (*bufp)[i] ||
887: ' ' == (*bufp)[i] ||
888: '\t' == (*bufp)[i])) {
889: roffnode_pop(r);
890: roffnode_cleanscope(r);
891:
1.130 kristaps 892: while (' ' == (*bufp)[i] || '\t' == (*bufp)[i])
893: i++;
894:
895: pos = i;
1.106 kristaps 896: if (ROFF_MAX != roff_parse(r, *bufp, &pos))
1.79 kristaps 897: return(ROFF_RERUN);
898: return(ROFF_IGN);
899: }
900: }
901:
902: /*
903: * If we have no custom end-query or lookup failed, then try
904: * pulling it out of the hashtable.
905: */
906:
1.137 schwarze 907: t = roff_parse(r, *bufp, &pos);
1.79 kristaps 908:
1.106 kristaps 909: /*
910: * Macros other than block-end are only significant
911: * in `de' blocks; elsewhere, simply throw them away.
912: */
913: if (ROFF_cblock != t) {
914: if (ROFF_de == tok)
915: roff_setstr(r, r->last->name, *bufp + ppos, 1);
1.79 kristaps 916: return(ROFF_IGN);
1.106 kristaps 917: }
1.79 kristaps 918:
919: assert(roffs[t].proc);
1.90 kristaps 920: return((*roffs[t].proc)(r, t, bufp, szp,
921: ln, ppos, pos, offs));
1.79 kristaps 922: }
923:
924:
925: /* ARGSUSED */
926: static enum rofferr
1.80 kristaps 927: roff_block_text(ROFF_ARGS)
1.78 kristaps 928: {
929:
1.106 kristaps 930: if (ROFF_de == tok)
931: roff_setstr(r, r->last->name, *bufp + pos, 1);
932:
1.78 kristaps 933: return(ROFF_IGN);
934: }
935:
936:
937: /* ARGSUSED */
938: static enum rofferr
1.82 kristaps 939: roff_cond_sub(ROFF_ARGS)
940: {
941: enum rofft t;
942: enum roffrule rr;
1.139 kristaps 943: char *ep;
1.82 kristaps 944:
945: rr = r->last->rule;
1.139 kristaps 946: roffnode_cleanscope(r);
1.82 kristaps 947:
1.139 kristaps 948: /*
949: * If the macro is unknown, first check if it contains a closing
950: * delimiter `\}'. If it does, close out our scope and return
951: * the currently-scoped rule (ignore or continue). Else, drop
952: * into the currently-scoped rule.
1.87 kristaps 953: */
954:
1.106 kristaps 955: if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos))) {
1.139 kristaps 956: ep = &(*bufp)[pos];
957: for ( ; NULL != (ep = strchr(ep, '\\')); ep++) {
958: ep++;
959: if ('}' != *ep)
960: continue;
1.144 kristaps 961:
962: /*
963: * Make the \} go away.
964: * This is a little haphazard, as it's not quite
965: * clear how nroff does this.
966: * If we're at the end of line, then just chop
967: * off the \} and resize the buffer.
968: * If we aren't, then conver it to spaces.
969: */
970:
971: if ('\0' == *(ep + 1)) {
972: *--ep = '\0';
973: *szp -= 2;
974: } else
975: *(ep - 1) = *ep = ' ';
976:
1.139 kristaps 977: roff_ccond(r, ROFF_ccond, bufp, szp,
978: ln, pos, pos + 2, offs);
979: break;
980: }
1.82 kristaps 981: return(ROFFRULE_DENY == rr ? ROFF_IGN : ROFF_CONT);
1.100 kristaps 982: }
1.82 kristaps 983:
984: /*
985: * A denied conditional must evaluate its children if and only
986: * if they're either structurally required (such as loops and
987: * conditionals) or a closing macro.
988: */
1.139 kristaps 989:
1.82 kristaps 990: if (ROFFRULE_DENY == rr)
991: if ( ! (ROFFMAC_STRUCT & roffs[t].flags))
992: if (ROFF_ccond != t)
993: return(ROFF_IGN);
994:
995: assert(roffs[t].proc);
1.90 kristaps 996: return((*roffs[t].proc)(r, t, bufp, szp,
997: ln, ppos, pos, offs));
1.82 kristaps 998: }
999:
1000: /* ARGSUSED */
1001: static enum rofferr
1002: roff_cond_text(ROFF_ARGS)
1.78 kristaps 1003: {
1.140 kristaps 1004: char *ep;
1.82 kristaps 1005: enum roffrule rr;
1006:
1007: rr = r->last->rule;
1.140 kristaps 1008: roffnode_cleanscope(r);
1.82 kristaps 1009:
1.140 kristaps 1010: ep = &(*bufp)[pos];
1011: for ( ; NULL != (ep = strchr(ep, '\\')); ep++) {
1012: ep++;
1013: if ('}' != *ep)
1014: continue;
1015: *ep = '&';
1016: roff_ccond(r, ROFF_ccond, bufp, szp,
1017: ln, pos, pos + 2, offs);
1.78 kristaps 1018: }
1.82 kristaps 1019: return(ROFFRULE_DENY == rr ? ROFF_IGN : ROFF_CONT);
1.74 kristaps 1020: }
1021:
1.88 kristaps 1022: static enum roffrule
1023: roff_evalcond(const char *v, int *pos)
1024: {
1025:
1026: switch (v[*pos]) {
1027: case ('n'):
1028: (*pos)++;
1029: return(ROFFRULE_ALLOW);
1030: case ('e'):
1031: /* FALLTHROUGH */
1032: case ('o'):
1033: /* FALLTHROUGH */
1034: case ('t'):
1035: (*pos)++;
1036: return(ROFFRULE_DENY);
1037: default:
1038: break;
1039: }
1040:
1041: while (v[*pos] && ' ' != v[*pos])
1042: (*pos)++;
1043: return(ROFFRULE_DENY);
1044: }
1045:
1.75 kristaps 1046: /* ARGSUSED */
1.74 kristaps 1047: static enum rofferr
1.103 kristaps 1048: roff_line_ignore(ROFF_ARGS)
1.89 kristaps 1049: {
1.123 schwarze 1050:
1051: if (ROFF_it == tok)
1.128 kristaps 1052: mandoc_msg(MANDOCERR_REQUEST, r->parse, ln, ppos, "it");
1.89 kristaps 1053:
1054: return(ROFF_IGN);
1055: }
1056:
1.104 kristaps 1057: /* ARGSUSED */
1058: static enum rofferr
1.82 kristaps 1059: roff_cond(ROFF_ARGS)
1.74 kristaps 1060: {
1.77 kristaps 1061: int sv;
1.88 kristaps 1062: enum roffrule rule;
1.74 kristaps 1063:
1.134 kristaps 1064: /*
1065: * An `.el' has no conditional body: it will consume the value
1066: * of the current rstack entry set in prior `ie' calls or
1067: * defaults to DENY.
1068: *
1069: * If we're not an `el', however, then evaluate the conditional.
1070: */
1.133 kristaps 1071:
1.134 kristaps 1072: rule = ROFF_el == tok ?
1073: (r->rstackpos < 0 ?
1074: ROFFRULE_DENY : r->rstack[r->rstackpos--]) :
1075: roff_evalcond(*bufp, &pos);
1.77 kristaps 1076:
1077: sv = pos;
1.75 kristaps 1078: while (' ' == (*bufp)[pos])
1079: pos++;
1.74 kristaps 1080:
1.77 kristaps 1081: /*
1082: * Roff is weird. If we have just white-space after the
1083: * conditional, it's considered the BODY and we exit without
1084: * really doing anything. Warn about this. It's probably
1085: * wrong.
1086: */
1.88 kristaps 1087:
1.77 kristaps 1088: if ('\0' == (*bufp)[pos] && sv != pos) {
1.128 kristaps 1089: mandoc_msg(MANDOCERR_NOARGS, r->parse, ln, ppos, NULL);
1.107 kristaps 1090: return(ROFF_IGN);
1.77 kristaps 1091: }
1092:
1.106 kristaps 1093: roffnode_push(r, tok, NULL, ln, ppos);
1.77 kristaps 1094:
1.88 kristaps 1095: r->last->rule = rule;
1096:
1.134 kristaps 1097: /*
1098: * An if-else will put the NEGATION of the current evaluated
1099: * conditional into the stack of rules.
1100: */
1101:
1.84 schwarze 1102: if (ROFF_ie == tok) {
1.134 kristaps 1103: if (r->rstackpos == RSTACK_MAX - 1) {
1104: mandoc_msg(MANDOCERR_MEM,
1105: r->parse, ln, ppos, NULL);
1106: return(ROFF_ERR);
1107: }
1108: r->rstack[++r->rstackpos] =
1109: ROFFRULE_DENY == r->last->rule ?
1110: ROFFRULE_ALLOW : ROFFRULE_DENY;
1.82 kristaps 1111: }
1.88 kristaps 1112:
1113: /* If the parent has false as its rule, then so do we. */
1114:
1.109 kristaps 1115: if (r->last->parent && ROFFRULE_DENY == r->last->parent->rule)
1.84 schwarze 1116: r->last->rule = ROFFRULE_DENY;
1.88 kristaps 1117:
1118: /*
1119: * Determine scope. If we're invoked with "\{" trailing the
1120: * conditional, then we're in a multiline scope. Else our scope
1121: * expires on the next line.
1122: */
1.74 kristaps 1123:
1.75 kristaps 1124: r->last->endspan = 1;
1125:
1126: if ('\\' == (*bufp)[pos] && '{' == (*bufp)[pos + 1]) {
1127: r->last->endspan = -1;
1128: pos += 2;
1.109 kristaps 1129: }
1.74 kristaps 1130:
1.77 kristaps 1131: /*
1132: * If there are no arguments on the line, the next-line scope is
1133: * assumed.
1134: */
1135:
1.75 kristaps 1136: if ('\0' == (*bufp)[pos])
1137: return(ROFF_IGN);
1.77 kristaps 1138:
1139: /* Otherwise re-run the roff parser after recalculating. */
1.74 kristaps 1140:
1.75 kristaps 1141: *offs = pos;
1142: return(ROFF_RERUN);
1.83 schwarze 1143: }
1144:
1145:
1146: /* ARGSUSED */
1147: static enum rofferr
1.92 schwarze 1148: roff_ds(ROFF_ARGS)
1149: {
1.96 kristaps 1150: char *name, *string;
1151:
1152: /*
1153: * A symbol is named by the first word following the macro
1154: * invocation up to a space. Its value is anything after the
1155: * name's trailing whitespace and optional double-quote. Thus,
1156: *
1157: * [.ds foo "bar " ]
1158: *
1159: * will have `bar " ' as its value.
1160: */
1.92 schwarze 1161:
1.121 schwarze 1162: string = *bufp + pos;
1163: name = roff_getname(r, &string, ln, pos);
1.92 schwarze 1164: if ('\0' == *name)
1165: return(ROFF_IGN);
1166:
1.121 schwarze 1167: /* Read past initial double-quote. */
1168: if ('"' == *string)
1.92 schwarze 1169: string++;
1170:
1.96 kristaps 1171: /* The rest is the value. */
1.106 kristaps 1172: roff_setstr(r, name, string, 0);
1.92 schwarze 1173: return(ROFF_IGN);
1174: }
1175:
1.147 kristaps 1176: int
1177: roff_regisset(const struct roff *r, enum regs reg)
1178: {
1179:
1180: return(r->regs[(int)reg].set);
1181: }
1182:
1183: unsigned int
1184: roff_regget(const struct roff *r, enum regs reg)
1185: {
1186:
1187: return(r->regs[(int)reg].u);
1188: }
1189:
1190: void
1191: roff_regunset(struct roff *r, enum regs reg)
1192: {
1193:
1194: r->regs[(int)reg].set = 0;
1195: }
1.92 schwarze 1196:
1197: /* ARGSUSED */
1198: static enum rofferr
1.89 kristaps 1199: roff_nr(ROFF_ARGS)
1.83 schwarze 1200: {
1.121 schwarze 1201: const char *key;
1202: char *val;
1.138 kristaps 1203: int iv;
1.89 kristaps 1204:
1.121 schwarze 1205: val = *bufp + pos;
1206: key = roff_getname(r, &val, ln, pos);
1.89 kristaps 1207:
1208: if (0 == strcmp(key, "nS")) {
1.147 kristaps 1209: r->regs[(int)REG_nS].set = 1;
1.149 kristaps 1210: if ((iv = mandoc_strntoi(val, strlen(val), 10)) >= 0)
1.147 kristaps 1211: r->regs[(int)REG_nS].u = (unsigned)iv;
1.138 kristaps 1212: else
1.147 kristaps 1213: r->regs[(int)REG_nS].u = 0u;
1.109 kristaps 1214: }
1215:
1.122 schwarze 1216: return(ROFF_IGN);
1217: }
1218:
1219: /* ARGSUSED */
1220: static enum rofferr
1221: roff_rm(ROFF_ARGS)
1222: {
1223: const char *name;
1224: char *cp;
1225:
1226: cp = *bufp + pos;
1227: while ('\0' != *cp) {
1.127 kristaps 1228: name = roff_getname(r, &cp, ln, (int)(cp - *bufp));
1.122 schwarze 1229: if ('\0' != *name)
1230: roff_setstr(r, name, NULL, 0);
1231: }
1.109 kristaps 1232: return(ROFF_IGN);
1233: }
1234:
1235: /* ARGSUSED */
1236: static enum rofferr
1237: roff_TE(ROFF_ARGS)
1238: {
1239:
1240: if (NULL == r->tbl)
1.128 kristaps 1241: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.115 kristaps 1242: else
1.151 kristaps 1243: tbl_end(&r->tbl);
1.109 kristaps 1244:
1.112 kristaps 1245: return(ROFF_IGN);
1246: }
1247:
1248: /* ARGSUSED */
1249: static enum rofferr
1250: roff_T_(ROFF_ARGS)
1251: {
1252:
1253: if (NULL == r->tbl)
1.128 kristaps 1254: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.112 kristaps 1255: else
1.116 kristaps 1256: tbl_restart(ppos, ln, r->tbl);
1.112 kristaps 1257:
1.109 kristaps 1258: return(ROFF_IGN);
1259: }
1260:
1.151 kristaps 1261: int
1262: roff_closeeqn(struct roff *r)
1263: {
1264:
1265: return(r->eqn && ROFF_EQN == eqn_end(&r->eqn) ? 1 : 0);
1266: }
1267:
1268: void
1269: roff_openeqn(struct roff *r, const char *name, int line,
1270: int offs, const char *buf)
1.125 kristaps 1271: {
1.151 kristaps 1272: struct eqn_node *e;
1273: int poff;
1.125 kristaps 1274:
1275: assert(NULL == r->eqn);
1.151 kristaps 1276: e = eqn_alloc(name, offs, line, r->parse);
1.125 kristaps 1277:
1278: if (r->last_eqn)
1279: r->last_eqn->next = e;
1280: else
1281: r->first_eqn = r->last_eqn = e;
1282:
1283: r->eqn = r->last_eqn = e;
1.151 kristaps 1284:
1285: if (buf) {
1286: poff = 0;
1287: eqn_read(&r->eqn, line, buf, offs, &poff);
1288: }
1289: }
1290:
1291: /* ARGSUSED */
1292: static enum rofferr
1293: roff_EQ(ROFF_ARGS)
1294: {
1295:
1296: roff_openeqn(r, *bufp + pos, ln, ppos, NULL);
1.125 kristaps 1297: return(ROFF_IGN);
1298: }
1299:
1300: /* ARGSUSED */
1301: static enum rofferr
1302: roff_EN(ROFF_ARGS)
1303: {
1304:
1.128 kristaps 1305: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.125 kristaps 1306: return(ROFF_IGN);
1307: }
1308:
1309: /* ARGSUSED */
1310: static enum rofferr
1.109 kristaps 1311: roff_TS(ROFF_ARGS)
1312: {
1.118 kristaps 1313: struct tbl_node *t;
1.89 kristaps 1314:
1.115 kristaps 1315: if (r->tbl) {
1.128 kristaps 1316: mandoc_msg(MANDOCERR_SCOPEBROKEN, r->parse, ln, ppos, NULL);
1.151 kristaps 1317: tbl_end(&r->tbl);
1.115 kristaps 1318: }
1.83 schwarze 1319:
1.128 kristaps 1320: t = tbl_alloc(ppos, ln, r->parse);
1.113 kristaps 1321:
1322: if (r->last_tbl)
1323: r->last_tbl->next = t;
1324: else
1325: r->first_tbl = r->last_tbl = t;
1326:
1327: r->tbl = r->last_tbl = t;
1.83 schwarze 1328: return(ROFF_IGN);
1.92 schwarze 1329: }
1330:
1.105 kristaps 1331: /* ARGSUSED */
1332: static enum rofferr
1333: roff_so(ROFF_ARGS)
1334: {
1335: char *name;
1336:
1.128 kristaps 1337: mandoc_msg(MANDOCERR_SO, r->parse, ln, ppos, NULL);
1.105 kristaps 1338:
1339: /*
1340: * Handle `so'. Be EXTREMELY careful, as we shouldn't be
1341: * opening anything that's not in our cwd or anything beneath
1342: * it. Thus, explicitly disallow traversing up the file-system
1343: * or using absolute paths.
1344: */
1345:
1346: name = *bufp + pos;
1347: if ('/' == *name || strstr(name, "../") || strstr(name, "/..")) {
1.128 kristaps 1348: mandoc_msg(MANDOCERR_SOPATH, r->parse, ln, pos, NULL);
1.105 kristaps 1349: return(ROFF_ERR);
1350: }
1351:
1352: *offs = pos;
1353: return(ROFF_SO);
1354: }
1.92 schwarze 1355:
1.106 kristaps 1356: /* ARGSUSED */
1357: static enum rofferr
1358: roff_userdef(ROFF_ARGS)
1.99 kristaps 1359: {
1.106 kristaps 1360: const char *arg[9];
1361: char *cp, *n1, *n2;
1.119 schwarze 1362: int i;
1.106 kristaps 1363:
1364: /*
1365: * Collect pointers to macro argument strings
1366: * and null-terminate them.
1367: */
1368: cp = *bufp + pos;
1.119 schwarze 1369: for (i = 0; i < 9; i++)
1.120 schwarze 1370: arg[i] = '\0' == *cp ? "" :
1.136 kristaps 1371: mandoc_getarg(r->parse, &cp, ln, &pos);
1.99 kristaps 1372:
1.106 kristaps 1373: /*
1374: * Expand macro arguments.
1.99 kristaps 1375: */
1.106 kristaps 1376: *szp = 0;
1377: n1 = cp = mandoc_strdup(r->current_string);
1378: while (NULL != (cp = strstr(cp, "\\$"))) {
1379: i = cp[2] - '1';
1380: if (0 > i || 8 < i) {
1381: /* Not an argument invocation. */
1382: cp += 2;
1383: continue;
1384: }
1385:
1386: *szp = strlen(n1) - 3 + strlen(arg[i]) + 1;
1387: n2 = mandoc_malloc(*szp);
1388:
1389: strlcpy(n2, n1, (size_t)(cp - n1 + 1));
1390: strlcat(n2, arg[i], *szp);
1391: strlcat(n2, cp + 3, *szp);
1392:
1393: cp = n2 + (cp - n1);
1394: free(n1);
1395: n1 = n2;
1.99 kristaps 1396: }
1397:
1.106 kristaps 1398: /*
1399: * Replace the macro invocation
1400: * by the expanded macro.
1401: */
1402: free(*bufp);
1403: *bufp = n1;
1404: if (0 == *szp)
1405: *szp = strlen(*bufp) + 1;
1406:
1407: return(*szp > 1 && '\n' == (*bufp)[(int)*szp - 2] ?
1408: ROFF_REPARSE : ROFF_APPEND);
1.99 kristaps 1409: }
1.121 schwarze 1410:
1411: static char *
1412: roff_getname(struct roff *r, char **cpp, int ln, int pos)
1413: {
1414: char *name, *cp;
1415:
1416: name = *cpp;
1417: if ('\0' == *name)
1418: return(name);
1419:
1420: /* Read until end of name. */
1421: for (cp = name; '\0' != *cp && ' ' != *cp; cp++) {
1422: if ('\\' != *cp)
1423: continue;
1424: cp++;
1425: if ('\\' == *cp)
1426: continue;
1.128 kristaps 1427: mandoc_msg(MANDOCERR_NAMESC, r->parse, ln, pos, NULL);
1.121 schwarze 1428: *cp = '\0';
1429: name = cp;
1430: }
1431:
1432: /* Nil-terminate name. */
1433: if ('\0' != *cp)
1434: *(cp++) = '\0';
1435:
1436: /* Read past spaces. */
1437: while (' ' == *cp)
1438: cp++;
1439:
1440: *cpp = cp;
1441: return(name);
1442: }
1443:
1.106 kristaps 1444: /*
1445: * Store *string into the user-defined string called *name.
1446: * In multiline mode, append to an existing entry and append '\n';
1447: * else replace the existing entry, if there is one.
1448: * To clear an existing entry, call with (*r, *name, NULL, 0).
1449: */
1.94 kristaps 1450: static void
1.106 kristaps 1451: roff_setstr(struct roff *r, const char *name, const char *string,
1452: int multiline)
1.92 schwarze 1453: {
1454: struct roffstr *n;
1.106 kristaps 1455: char *c;
1456: size_t oldch, newch;
1.92 schwarze 1457:
1.106 kristaps 1458: /* Search for an existing string with the same name. */
1.94 kristaps 1459: n = r->first_string;
1.92 schwarze 1460: while (n && strcmp(name, n->name))
1461: n = n->next;
1.94 kristaps 1462:
1463: if (NULL == n) {
1.106 kristaps 1464: /* Create a new string table entry. */
1.94 kristaps 1465: n = mandoc_malloc(sizeof(struct roffstr));
1.106 kristaps 1466: n->name = mandoc_strdup(name);
1467: n->string = NULL;
1.94 kristaps 1468: n->next = r->first_string;
1469: r->first_string = n;
1.106 kristaps 1470: } else if (0 == multiline) {
1471: /* In multiline mode, append; else replace. */
1.92 schwarze 1472: free(n->string);
1.106 kristaps 1473: n->string = NULL;
1474: }
1475:
1476: if (NULL == string)
1477: return;
1478:
1479: /*
1480: * One additional byte for the '\n' in multiline mode,
1481: * and one for the terminating '\0'.
1482: */
1.127 kristaps 1483: newch = strlen(string) + (multiline ? 2u : 1u);
1.106 kristaps 1484: if (NULL == n->string) {
1485: n->string = mandoc_malloc(newch);
1486: *n->string = '\0';
1487: oldch = 0;
1488: } else {
1489: oldch = strlen(n->string);
1490: n->string = mandoc_realloc(n->string, oldch + newch);
1491: }
1492:
1493: /* Skip existing content in the destination buffer. */
1.127 kristaps 1494: c = n->string + (int)oldch;
1.106 kristaps 1495:
1496: /* Append new content to the destination buffer. */
1497: while (*string) {
1498: /*
1499: * Rudimentary roff copy mode:
1500: * Handle escaped backslashes.
1501: */
1502: if ('\\' == *string && '\\' == *(string + 1))
1503: string++;
1504: *c++ = *string++;
1505: }
1.94 kristaps 1506:
1.106 kristaps 1507: /* Append terminating bytes. */
1508: if (multiline)
1509: *c++ = '\n';
1510: *c = '\0';
1.92 schwarze 1511: }
1512:
1.94 kristaps 1513: static const char *
1514: roff_getstrn(const struct roff *r, const char *name, size_t len)
1.92 schwarze 1515: {
1.94 kristaps 1516: const struct roffstr *n;
1.92 schwarze 1517:
1.94 kristaps 1518: n = r->first_string;
1.97 kristaps 1519: while (n && (strncmp(name, n->name, len) || '\0' != n->name[(int)len]))
1.92 schwarze 1520: n = n->next;
1.94 kristaps 1521:
1522: return(n ? n->string : NULL);
1.92 schwarze 1523: }
1524:
1.94 kristaps 1525: static void
1526: roff_freestr(struct roff *r)
1.92 schwarze 1527: {
1528: struct roffstr *n, *nn;
1529:
1.94 kristaps 1530: for (n = r->first_string; n; n = nn) {
1.92 schwarze 1531: free(n->name);
1532: free(n->string);
1533: nn = n->next;
1534: free(n);
1535: }
1.94 kristaps 1536:
1537: r->first_string = NULL;
1.114 kristaps 1538: }
1539:
1540: const struct tbl_span *
1541: roff_span(const struct roff *r)
1542: {
1543:
1544: return(r->tbl ? tbl_span(r->tbl) : NULL);
1.125 kristaps 1545: }
1546:
1547: const struct eqn *
1548: roff_eqn(const struct roff *r)
1549: {
1550:
1551: return(r->last_eqn ? &r->last_eqn->eqn : NULL);
1.151 kristaps 1552: }
1553:
1554: char
1555: roff_eqndelim(const struct roff *r)
1556: {
1557:
1558: return('\0');
1.74 kristaps 1559: }
CVSweb