Annotation of mandoc/roff.c, Revision 1.166
1.166 ! kristaps 1: /* $Id: roff.c,v 1.165 2011/07/28 14:53:22 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 {
1.166 ! kristaps 80: int set; /* whether set or not */
! 81: unsigned int u; /* unsigned integer */
1.147 kristaps 82: };
83:
1.94 kristaps 84: struct roffstr {
1.166 ! kristaps 85: char *p;
! 86: size_t sz;
! 87: };
! 88:
! 89: /*
! 90: * A key-value string pair with lengths.
! 91: */
! 92: struct roffkv {
! 93: struct roffstr key;
! 94: struct roffstr val;
! 95: struct roffkv *next; /* next in list */
1.94 kristaps 96: };
97:
1.67 kristaps 98: struct roff {
1.128 kristaps 99: struct mparse *parse; /* parse point */
1.67 kristaps 100: struct roffnode *last; /* leaf of stack */
1.82 kristaps 101: enum roffrule rstack[RSTACK_MAX]; /* stack of !`ie' rules */
102: int rstackpos; /* position in rstack */
1.147 kristaps 103: struct reg regs[REG__MAX];
1.166 ! kristaps 104: struct roffkv *strtab; /* user-defined strings & macros */
! 105: struct roffkv *chrtab; /* user-defined characters */
1.106 kristaps 106: const char *current_string; /* value of last called user macro */
1.118 kristaps 107: struct tbl_node *first_tbl; /* first table parsed */
108: struct tbl_node *last_tbl; /* last table parsed */
109: struct tbl_node *tbl; /* current table being parsed */
1.125 kristaps 110: struct eqn_node *last_eqn; /* last equation parsed */
111: struct eqn_node *first_eqn; /* first equation parsed */
112: struct eqn_node *eqn; /* current equation being parsed */
1.79 kristaps 113: };
114:
1.67 kristaps 115: struct roffnode {
116: enum rofft tok; /* type of node */
117: struct roffnode *parent; /* up one in stack */
118: int line; /* parse line */
119: int col; /* parse col */
1.106 kristaps 120: char *name; /* node name, e.g. macro name */
1.79 kristaps 121: char *end; /* end-rules: custom token */
122: int endspan; /* end-rules: next-line or infty */
1.82 kristaps 123: enum roffrule rule; /* current evaluation rule */
1.67 kristaps 124: };
125:
126: #define ROFF_ARGS struct roff *r, /* parse ctx */ \
1.72 kristaps 127: enum rofft tok, /* tok of macro */ \
1.67 kristaps 128: char **bufp, /* input buffer */ \
129: size_t *szp, /* size of input buffer */ \
130: int ln, /* parse line */ \
1.75 kristaps 131: int ppos, /* original pos in buffer */ \
132: int pos, /* current pos in buffer */ \
1.74 kristaps 133: int *offs /* reset offset of buffer data */
1.67 kristaps 134:
135: typedef enum rofferr (*roffproc)(ROFF_ARGS);
136:
137: struct roffmac {
138: const char *name; /* macro name */
1.79 kristaps 139: roffproc proc; /* process new macro */
140: roffproc text; /* process as child text of macro */
141: roffproc sub; /* process as child of macro */
142: int flags;
143: #define ROFFMAC_STRUCT (1 << 0) /* always interpret */
1.85 kristaps 144: struct roffmac *next;
1.67 kristaps 145: };
146:
1.141 kristaps 147: struct predef {
148: const char *name; /* predefined input name */
149: const char *str; /* replacement symbol */
150: };
151:
152: #define PREDEF(__name, __str) \
153: { (__name), (__str) },
154:
1.155 kristaps 155: static enum rofft roffhash_find(const char *, size_t);
156: static void roffhash_init(void);
157: static void roffnode_cleanscope(struct roff *);
158: static void roffnode_pop(struct roff *);
159: static void roffnode_push(struct roff *, enum rofft,
160: const char *, int, int);
1.80 kristaps 161: static enum rofferr roff_block(ROFF_ARGS);
162: static enum rofferr roff_block_text(ROFF_ARGS);
163: static enum rofferr roff_block_sub(ROFF_ARGS);
164: static enum rofferr roff_cblock(ROFF_ARGS);
165: static enum rofferr roff_ccond(ROFF_ARGS);
1.82 kristaps 166: static enum rofferr roff_cond(ROFF_ARGS);
167: static enum rofferr roff_cond_text(ROFF_ARGS);
168: static enum rofferr roff_cond_sub(ROFF_ARGS);
1.92 schwarze 169: static enum rofferr roff_ds(ROFF_ARGS);
1.94 kristaps 170: static enum roffrule roff_evalcond(const char *, int *);
1.155 kristaps 171: static void roff_free1(struct roff *);
1.166 ! kristaps 172: static void roff_freestr(struct roffkv **);
1.121 schwarze 173: static char *roff_getname(struct roff *, char **, int, int);
1.94 kristaps 174: static const char *roff_getstrn(const struct roff *,
175: const char *, size_t);
1.103 kristaps 176: static enum rofferr roff_line_ignore(ROFF_ARGS);
1.89 kristaps 177: static enum rofferr roff_nr(ROFF_ARGS);
1.156 kristaps 178: static void roff_openeqn(struct roff *, const char *,
179: int, int, const char *);
1.155 kristaps 180: static enum rofft roff_parse(struct roff *, const char *, int *);
181: static enum rofferr roff_parsetext(char *);
1.154 kristaps 182: static void roff_res(struct roff *,
1.142 kristaps 183: char **, size_t *, int, int);
1.122 schwarze 184: static enum rofferr roff_rm(ROFF_ARGS);
1.94 kristaps 185: static void roff_setstr(struct roff *,
1.106 kristaps 186: const char *, const char *, int);
1.166 ! kristaps 187: static void roff_setstrn(struct roffkv **, const char *,
1.164 kristaps 188: size_t, const char *, size_t, int);
1.105 kristaps 189: static enum rofferr roff_so(ROFF_ARGS);
1.164 kristaps 190: static enum rofferr roff_tr(ROFF_ARGS);
1.109 kristaps 191: static enum rofferr roff_TE(ROFF_ARGS);
192: static enum rofferr roff_TS(ROFF_ARGS);
1.125 kristaps 193: static enum rofferr roff_EQ(ROFF_ARGS);
194: static enum rofferr roff_EN(ROFF_ARGS);
1.112 kristaps 195: static enum rofferr roff_T_(ROFF_ARGS);
1.106 kristaps 196: static enum rofferr roff_userdef(ROFF_ARGS);
1.67 kristaps 197:
1.155 kristaps 198: /* See roffhash_find() */
1.85 kristaps 199:
200: #define ASCII_HI 126
201: #define ASCII_LO 33
202: #define HASHWIDTH (ASCII_HI - ASCII_LO + 1)
203:
204: static struct roffmac *hash[HASHWIDTH];
205:
206: static struct roffmac roffs[ROFF_MAX] = {
1.103 kristaps 207: { "ad", roff_line_ignore, NULL, NULL, 0, NULL },
1.85 kristaps 208: { "am", roff_block, roff_block_text, roff_block_sub, 0, NULL },
209: { "ami", roff_block, roff_block_text, roff_block_sub, 0, NULL },
210: { "am1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
211: { "de", roff_block, roff_block_text, roff_block_sub, 0, NULL },
212: { "dei", roff_block, roff_block_text, roff_block_sub, 0, NULL },
213: { "de1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.92 schwarze 214: { "ds", roff_ds, NULL, NULL, 0, NULL },
1.85 kristaps 215: { "el", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
1.103 kristaps 216: { "hy", roff_line_ignore, NULL, NULL, 0, NULL },
1.85 kristaps 217: { "ie", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
218: { "if", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
219: { "ig", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.123 schwarze 220: { "it", roff_line_ignore, NULL, NULL, 0, NULL },
1.103 kristaps 221: { "ne", roff_line_ignore, NULL, NULL, 0, NULL },
222: { "nh", roff_line_ignore, NULL, NULL, 0, NULL },
1.104 kristaps 223: { "nr", roff_nr, NULL, NULL, 0, NULL },
1.124 schwarze 224: { "ns", roff_line_ignore, NULL, NULL, 0, NULL },
225: { "ps", roff_line_ignore, NULL, NULL, 0, NULL },
1.122 schwarze 226: { "rm", roff_rm, NULL, NULL, 0, NULL },
1.105 kristaps 227: { "so", roff_so, NULL, NULL, 0, NULL },
1.124 schwarze 228: { "ta", roff_line_ignore, NULL, NULL, 0, NULL },
1.164 kristaps 229: { "tr", roff_tr, NULL, NULL, 0, NULL },
1.109 kristaps 230: { "TS", roff_TS, NULL, NULL, 0, NULL },
231: { "TE", roff_TE, NULL, NULL, 0, NULL },
1.112 kristaps 232: { "T&", roff_T_, NULL, NULL, 0, NULL },
1.125 kristaps 233: { "EQ", roff_EQ, NULL, NULL, 0, NULL },
234: { "EN", roff_EN, NULL, NULL, 0, NULL },
1.85 kristaps 235: { ".", roff_cblock, NULL, NULL, 0, NULL },
236: { "\\}", roff_ccond, NULL, NULL, 0, NULL },
1.106 kristaps 237: { NULL, roff_userdef, NULL, NULL, 0, NULL },
1.67 kristaps 238: };
239:
1.141 kristaps 240: /* Array of injected predefined strings. */
241: #define PREDEFS_MAX 38
242: static const struct predef predefs[PREDEFS_MAX] = {
243: #include "predefs.in"
244: };
245:
1.155 kristaps 246: /* See roffhash_find() */
1.85 kristaps 247: #define ROFF_HASH(p) (p[0] - ASCII_LO)
248:
249: static void
1.155 kristaps 250: roffhash_init(void)
1.85 kristaps 251: {
252: struct roffmac *n;
253: int buc, i;
254:
1.106 kristaps 255: for (i = 0; i < (int)ROFF_USERDEF; i++) {
1.85 kristaps 256: assert(roffs[i].name[0] >= ASCII_LO);
257: assert(roffs[i].name[0] <= ASCII_HI);
258:
259: buc = ROFF_HASH(roffs[i].name);
260:
261: if (NULL != (n = hash[buc])) {
262: for ( ; n->next; n = n->next)
263: /* Do nothing. */ ;
264: n->next = &roffs[i];
265: } else
266: hash[buc] = &roffs[i];
267: }
268: }
269:
1.67 kristaps 270: /*
271: * Look up a roff token by its name. Returns ROFF_MAX if no macro by
272: * the nil-terminated string name could be found.
273: */
274: static enum rofft
1.155 kristaps 275: roffhash_find(const char *p, size_t s)
1.67 kristaps 276: {
1.85 kristaps 277: int buc;
278: struct roffmac *n;
1.67 kristaps 279:
1.85 kristaps 280: /*
281: * libroff has an extremely simple hashtable, for the time
282: * being, which simply keys on the first character, which must
283: * be printable, then walks a chain. It works well enough until
284: * optimised.
285: */
286:
287: if (p[0] < ASCII_LO || p[0] > ASCII_HI)
288: return(ROFF_MAX);
289:
290: buc = ROFF_HASH(p);
291:
292: if (NULL == (n = hash[buc]))
293: return(ROFF_MAX);
294: for ( ; n; n = n->next)
1.106 kristaps 295: if (0 == strncmp(n->name, p, s) && '\0' == n->name[(int)s])
1.85 kristaps 296: return((enum rofft)(n - roffs));
1.67 kristaps 297:
298: return(ROFF_MAX);
299: }
300:
301:
302: /*
303: * Pop the current node off of the stack of roff instructions currently
304: * pending.
305: */
306: static void
307: roffnode_pop(struct roff *r)
308: {
309: struct roffnode *p;
310:
1.75 kristaps 311: assert(r->last);
312: p = r->last;
1.82 kristaps 313:
1.75 kristaps 314: r->last = r->last->parent;
1.106 kristaps 315: free(p->name);
316: free(p->end);
1.67 kristaps 317: free(p);
318: }
319:
320:
321: /*
322: * Push a roff node onto the instruction stack. This must later be
323: * removed with roffnode_pop().
324: */
1.98 schwarze 325: static void
1.106 kristaps 326: roffnode_push(struct roff *r, enum rofft tok, const char *name,
327: int line, int col)
1.67 kristaps 328: {
329: struct roffnode *p;
330:
1.98 schwarze 331: p = mandoc_calloc(1, sizeof(struct roffnode));
1.67 kristaps 332: p->tok = tok;
1.106 kristaps 333: if (name)
334: p->name = mandoc_strdup(name);
1.67 kristaps 335: p->parent = r->last;
336: p->line = line;
337: p->col = col;
1.79 kristaps 338: p->rule = p->parent ? p->parent->rule : ROFFRULE_DENY;
1.67 kristaps 339:
340: r->last = p;
341: }
342:
343:
344: static void
345: roff_free1(struct roff *r)
346: {
1.118 kristaps 347: struct tbl_node *t;
1.125 kristaps 348: struct eqn_node *e;
1.67 kristaps 349:
1.125 kristaps 350: while (NULL != (t = r->first_tbl)) {
1.113 kristaps 351: r->first_tbl = t->next;
352: tbl_free(t);
1.109 kristaps 353: }
354:
1.113 kristaps 355: r->first_tbl = r->last_tbl = r->tbl = NULL;
356:
1.125 kristaps 357: while (NULL != (e = r->first_eqn)) {
358: r->first_eqn = e->next;
359: eqn_free(e);
360: }
361:
362: r->first_eqn = r->last_eqn = r->eqn = NULL;
363:
1.67 kristaps 364: while (r->last)
365: roffnode_pop(r);
1.109 kristaps 366:
1.164 kristaps 367: roff_freestr(&r->strtab);
368: roff_freestr(&r->chrtab);
1.67 kristaps 369: }
370:
371:
372: void
373: roff_reset(struct roff *r)
374: {
1.143 kristaps 375: int i;
1.67 kristaps 376:
377: roff_free1(r);
1.143 kristaps 378:
1.147 kristaps 379: memset(&r->regs, 0, sizeof(struct reg) * REG__MAX);
380:
1.143 kristaps 381: for (i = 0; i < PREDEFS_MAX; i++)
382: roff_setstr(r, predefs[i].name, predefs[i].str, 0);
1.67 kristaps 383: }
384:
385:
386: void
387: roff_free(struct roff *r)
388: {
389:
390: roff_free1(r);
391: free(r);
392: }
393:
394:
395: struct roff *
1.147 kristaps 396: roff_alloc(struct mparse *parse)
1.67 kristaps 397: {
398: struct roff *r;
1.141 kristaps 399: int i;
1.67 kristaps 400:
1.98 schwarze 401: r = mandoc_calloc(1, sizeof(struct roff));
1.128 kristaps 402: r->parse = parse;
1.82 kristaps 403: r->rstackpos = -1;
1.85 kristaps 404:
1.155 kristaps 405: roffhash_init();
1.141 kristaps 406:
407: for (i = 0; i < PREDEFS_MAX; i++)
408: roff_setstr(r, predefs[i].name, predefs[i].str, 0);
409:
1.67 kristaps 410: return(r);
411: }
412:
1.94 kristaps 413: /*
414: * Pre-filter each and every line for reserved words (one beginning with
415: * `\*', e.g., `\*(ab'). These must be handled before the actual line
416: * is processed.
1.153 kristaps 417: * This also checks the syntax of regular escapes.
1.154 kristaps 418: */
419: static void
1.142 kristaps 420: roff_res(struct roff *r, char **bufp, size_t *szp, int ln, int pos)
1.94 kristaps 421: {
1.152 kristaps 422: enum mandoc_esc esc;
1.108 schwarze 423: const char *stesc; /* start of an escape sequence ('\\') */
424: const char *stnam; /* start of the name, after "[(*" */
425: const char *cp; /* end of the name, e.g. before ']' */
426: const char *res; /* the string to be substituted */
1.94 kristaps 427: int i, maxl;
428: size_t nsz;
429: char *n;
430:
1.154 kristaps 431: again:
1.108 schwarze 432: cp = *bufp + pos;
433: while (NULL != (cp = strchr(cp, '\\'))) {
434: stesc = cp++;
435:
436: /*
437: * The second character must be an asterisk.
438: * If it isn't, skip it anyway: It is escaped,
439: * so it can't start another escape sequence.
440: */
441:
442: if ('\0' == *cp)
1.154 kristaps 443: return;
1.152 kristaps 444:
445: if ('*' != *cp) {
446: res = cp;
447: esc = mandoc_escape(&cp, NULL, NULL);
448: if (ESCAPE_ERROR != esc)
449: continue;
450: cp = res;
1.153 kristaps 451: mandoc_msg
452: (MANDOCERR_BADESCAPE, r->parse,
453: ln, (int)(stesc - *bufp), NULL);
1.154 kristaps 454: return;
1.152 kristaps 455: }
456:
457: cp++;
1.108 schwarze 458:
459: /*
460: * The third character decides the length
461: * of the name of the string.
462: * Save a pointer to the name.
463: */
464:
1.94 kristaps 465: switch (*cp) {
1.108 schwarze 466: case ('\0'):
1.154 kristaps 467: return;
1.94 kristaps 468: case ('('):
469: cp++;
470: maxl = 2;
471: break;
472: case ('['):
473: cp++;
474: maxl = 0;
475: break;
476: default:
477: maxl = 1;
478: break;
479: }
1.108 schwarze 480: stnam = cp;
1.94 kristaps 481:
1.108 schwarze 482: /* Advance to the end of the name. */
1.94 kristaps 483:
484: for (i = 0; 0 == maxl || i < maxl; i++, cp++) {
1.153 kristaps 485: if ('\0' == *cp) {
486: mandoc_msg
487: (MANDOCERR_BADESCAPE,
488: r->parse, ln,
489: (int)(stesc - *bufp), NULL);
1.154 kristaps 490: return;
1.153 kristaps 491: }
1.94 kristaps 492: if (0 == maxl && ']' == *cp)
493: break;
494: }
495:
1.108 schwarze 496: /*
497: * Retrieve the replacement string; if it is
498: * undefined, resume searching for escapes.
499: */
500:
501: res = roff_getstrn(r, stnam, (size_t)i);
1.94 kristaps 502:
503: if (NULL == res) {
1.153 kristaps 504: mandoc_msg
505: (MANDOCERR_BADESCAPE, r->parse,
506: ln, (int)(stesc - *bufp), NULL);
1.142 kristaps 507: res = "";
1.94 kristaps 508: }
509:
1.108 schwarze 510: /* Replace the escape sequence by the string. */
511:
1.161 kristaps 512: pos = stesc - *bufp;
1.154 kristaps 513:
1.94 kristaps 514: nsz = *szp + strlen(res) + 1;
515: n = mandoc_malloc(nsz);
516:
1.108 schwarze 517: strlcpy(n, *bufp, (size_t)(stesc - *bufp + 1));
1.94 kristaps 518: strlcat(n, res, nsz);
519: strlcat(n, cp + (maxl ? 0 : 1), nsz);
520:
521: free(*bufp);
522:
523: *bufp = n;
524: *szp = nsz;
1.154 kristaps 525: goto again;
526: }
527: }
528:
529: /*
530: * Process text streams: convert all breakable hyphens into ASCII_HYPH.
531: */
532: static enum rofferr
533: roff_parsetext(char *p)
534: {
1.155 kristaps 535: char l, r;
1.154 kristaps 536: size_t sz;
537: const char *start;
538: enum mandoc_esc esc;
539:
540: start = p;
541:
542: while ('\0' != *p) {
543: sz = strcspn(p, "-\\");
544: p += sz;
545:
1.159 kristaps 546: if ('\0' == *p)
547: break;
548:
1.154 kristaps 549: if ('\\' == *p) {
550: /* Skip over escapes. */
551: p++;
552: esc = mandoc_escape
553: ((const char **)&p, NULL, NULL);
554: if (ESCAPE_ERROR == esc)
555: break;
1.155 kristaps 556: continue;
1.159 kristaps 557: } else if (p == start) {
1.158 kristaps 558: p++;
1.155 kristaps 559: continue;
1.158 kristaps 560: }
1.155 kristaps 561:
562: l = *(p - 1);
563: r = *(p + 1);
564: if ('\\' != l &&
565: '\t' != r && '\t' != l &&
566: ' ' != r && ' ' != l &&
567: '-' != r && '-' != l &&
568: ! isdigit((unsigned char)l) &&
1.159 kristaps 569: ! isdigit((unsigned char)r))
1.155 kristaps 570: *p = ASCII_HYPH;
571: p++;
1.94 kristaps 572: }
573:
1.154 kristaps 574: return(ROFF_CONT);
1.94 kristaps 575: }
576:
1.67 kristaps 577: enum rofferr
1.90 kristaps 578: roff_parseln(struct roff *r, int ln, char **bufp,
579: size_t *szp, int pos, int *offs)
1.67 kristaps 580: {
581: enum rofft t;
1.109 kristaps 582: enum rofferr e;
1.130 kristaps 583: int ppos, ctl;
1.79 kristaps 584:
585: /*
1.94 kristaps 586: * Run the reserved-word filter only if we have some reserved
587: * words to fill in.
588: */
589:
1.154 kristaps 590: roff_res(r, bufp, szp, ln, pos);
1.94 kristaps 591:
1.130 kristaps 592: ppos = pos;
593: ctl = mandoc_getcontrol(*bufp, &pos);
594:
1.94 kristaps 595: /*
1.79 kristaps 596: * First, if a scope is open and we're not a macro, pass the
597: * text through the macro's filter. If a scope isn't open and
598: * we're not a macro, just let it through.
1.125 kristaps 599: * Finally, if there's an equation scope open, divert it into it
600: * no matter our state.
1.79 kristaps 601: */
1.74 kristaps 602:
1.130 kristaps 603: if (r->last && ! ctl) {
1.78 kristaps 604: t = r->last->tok;
605: assert(roffs[t].text);
1.109 kristaps 606: e = (*roffs[t].text)
607: (r, t, bufp, szp, ln, pos, pos, offs);
608: assert(ROFF_IGN == e || ROFF_CONT == e);
1.125 kristaps 609: if (ROFF_CONT != e)
610: return(e);
611: if (r->eqn)
1.146 kristaps 612: return(eqn_read(&r->eqn, ln, *bufp, pos, offs));
1.125 kristaps 613: if (r->tbl)
1.130 kristaps 614: return(tbl_read(r->tbl, ln, *bufp, pos));
1.154 kristaps 615: return(roff_parsetext(*bufp + pos));
1.130 kristaps 616: } else if ( ! ctl) {
1.125 kristaps 617: if (r->eqn)
1.146 kristaps 618: return(eqn_read(&r->eqn, ln, *bufp, pos, offs));
1.109 kristaps 619: if (r->tbl)
1.130 kristaps 620: return(tbl_read(r->tbl, ln, *bufp, pos));
1.154 kristaps 621: return(roff_parsetext(*bufp + pos));
1.125 kristaps 622: } else if (r->eqn)
1.146 kristaps 623: return(eqn_read(&r->eqn, ln, *bufp, ppos, offs));
1.67 kristaps 624:
1.79 kristaps 625: /*
626: * If a scope is open, go to the child handler for that macro,
627: * as it may want to preprocess before doing anything with it.
1.125 kristaps 628: * Don't do so if an equation is open.
1.79 kristaps 629: */
1.78 kristaps 630:
1.79 kristaps 631: if (r->last) {
632: t = r->last->tok;
633: assert(roffs[t].sub);
634: return((*roffs[t].sub)
1.90 kristaps 635: (r, t, bufp, szp,
1.130 kristaps 636: ln, ppos, pos, offs));
1.79 kristaps 637: }
1.78 kristaps 638:
1.79 kristaps 639: /*
640: * Lastly, as we've no scope open, try to look up and execute
641: * the new macro. If no macro is found, simply return and let
642: * the compilers handle it.
643: */
1.67 kristaps 644:
1.106 kristaps 645: if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos)))
1.79 kristaps 646: return(ROFF_CONT);
1.67 kristaps 647:
1.75 kristaps 648: assert(roffs[t].proc);
1.78 kristaps 649: return((*roffs[t].proc)
1.90 kristaps 650: (r, t, bufp, szp,
651: ln, ppos, pos, offs));
1.74 kristaps 652: }
653:
654:
1.117 kristaps 655: void
1.74 kristaps 656: roff_endparse(struct roff *r)
657: {
658:
1.110 kristaps 659: if (r->last)
1.128 kristaps 660: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.109 kristaps 661: r->last->line, r->last->col, NULL);
1.117 kristaps 662:
1.125 kristaps 663: if (r->eqn) {
1.128 kristaps 664: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.148 kristaps 665: r->eqn->eqn.ln, r->eqn->eqn.pos, NULL);
1.151 kristaps 666: eqn_end(&r->eqn);
1.125 kristaps 667: }
668:
1.117 kristaps 669: if (r->tbl) {
1.128 kristaps 670: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.117 kristaps 671: r->tbl->line, r->tbl->pos, NULL);
1.151 kristaps 672: tbl_end(&r->tbl);
1.117 kristaps 673: }
1.67 kristaps 674: }
675:
676: /*
677: * Parse a roff node's type from the input buffer. This must be in the
678: * form of ".foo xxx" in the usual way.
679: */
680: static enum rofft
1.106 kristaps 681: roff_parse(struct roff *r, const char *buf, int *pos)
1.67 kristaps 682: {
1.106 kristaps 683: const char *mac;
684: size_t maclen;
1.67 kristaps 685: enum rofft t;
686:
1.144 kristaps 687: if ('\0' == buf[*pos] || '"' == buf[*pos] ||
688: '\t' == buf[*pos] || ' ' == buf[*pos])
1.67 kristaps 689: return(ROFF_MAX);
690:
1.144 kristaps 691: /*
692: * We stop the macro parse at an escape, tab, space, or nil.
693: * However, `\}' is also a valid macro, so make sure we don't
694: * clobber it by seeing the `\' as the end of token.
695: */
696:
1.106 kristaps 697: mac = buf + *pos;
1.144 kristaps 698: maclen = strcspn(mac + 1, " \\\t\0") + 1;
1.67 kristaps 699:
1.106 kristaps 700: t = (r->current_string = roff_getstrn(r, mac, maclen))
1.155 kristaps 701: ? ROFF_USERDEF : roffhash_find(mac, maclen);
1.67 kristaps 702:
1.127 kristaps 703: *pos += (int)maclen;
1.130 kristaps 704:
1.67 kristaps 705: while (buf[*pos] && ' ' == buf[*pos])
706: (*pos)++;
707:
708: return(t);
709: }
710:
711: /* ARGSUSED */
712: static enum rofferr
1.76 kristaps 713: roff_cblock(ROFF_ARGS)
1.67 kristaps 714: {
715:
1.79 kristaps 716: /*
717: * A block-close `..' should only be invoked as a child of an
718: * ignore macro, otherwise raise a warning and just ignore it.
719: */
720:
1.76 kristaps 721: if (NULL == r->last) {
1.128 kristaps 722: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.76 kristaps 723: return(ROFF_IGN);
724: }
1.67 kristaps 725:
1.81 kristaps 726: switch (r->last->tok) {
727: case (ROFF_am):
728: /* FALLTHROUGH */
729: case (ROFF_ami):
730: /* FALLTHROUGH */
731: case (ROFF_am1):
732: /* FALLTHROUGH */
733: case (ROFF_de):
1.108 schwarze 734: /* ROFF_de1 is remapped to ROFF_de in roff_block(). */
1.81 kristaps 735: /* FALLTHROUGH */
736: case (ROFF_dei):
737: /* FALLTHROUGH */
738: case (ROFF_ig):
739: break;
740: default:
1.128 kristaps 741: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.67 kristaps 742: return(ROFF_IGN);
1.76 kristaps 743: }
1.67 kristaps 744:
1.76 kristaps 745: if ((*bufp)[pos])
1.128 kristaps 746: mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.71 kristaps 747:
748: roffnode_pop(r);
1.76 kristaps 749: roffnode_cleanscope(r);
750: return(ROFF_IGN);
1.71 kristaps 751:
1.67 kristaps 752: }
753:
754:
1.76 kristaps 755: static void
756: roffnode_cleanscope(struct roff *r)
1.67 kristaps 757: {
758:
1.76 kristaps 759: while (r->last) {
760: if (--r->last->endspan < 0)
761: break;
762: roffnode_pop(r);
763: }
1.67 kristaps 764: }
765:
766:
1.75 kristaps 767: /* ARGSUSED */
1.74 kristaps 768: static enum rofferr
1.75 kristaps 769: roff_ccond(ROFF_ARGS)
1.74 kristaps 770: {
771:
1.76 kristaps 772: if (NULL == r->last) {
1.128 kristaps 773: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.76 kristaps 774: return(ROFF_IGN);
775: }
776:
1.82 kristaps 777: switch (r->last->tok) {
778: case (ROFF_el):
779: /* FALLTHROUGH */
780: case (ROFF_ie):
781: /* FALLTHROUGH */
782: case (ROFF_if):
783: break;
784: default:
1.128 kristaps 785: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.75 kristaps 786: return(ROFF_IGN);
787: }
788:
1.76 kristaps 789: if (r->last->endspan > -1) {
1.128 kristaps 790: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.76 kristaps 791: return(ROFF_IGN);
792: }
793:
794: if ((*bufp)[pos])
1.128 kristaps 795: mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.76 kristaps 796:
1.75 kristaps 797: roffnode_pop(r);
1.76 kristaps 798: roffnode_cleanscope(r);
799: return(ROFF_IGN);
800: }
801:
1.75 kristaps 802:
1.76 kristaps 803: /* ARGSUSED */
804: static enum rofferr
1.80 kristaps 805: roff_block(ROFF_ARGS)
1.76 kristaps 806: {
1.78 kristaps 807: int sv;
808: size_t sz;
1.106 kristaps 809: char *name;
810:
811: name = NULL;
1.76 kristaps 812:
1.106 kristaps 813: if (ROFF_ig != tok) {
814: if ('\0' == (*bufp)[pos]) {
1.128 kristaps 815: mandoc_msg(MANDOCERR_NOARGS, r->parse, ln, ppos, NULL);
1.106 kristaps 816: return(ROFF_IGN);
817: }
1.107 kristaps 818:
819: /*
820: * Re-write `de1', since we don't really care about
821: * groff's strange compatibility mode, into `de'.
822: */
823:
1.106 kristaps 824: if (ROFF_de1 == tok)
825: tok = ROFF_de;
826: if (ROFF_de == tok)
827: name = *bufp + pos;
828: else
1.128 kristaps 829: mandoc_msg(MANDOCERR_REQUEST, r->parse, ln, ppos,
1.106 kristaps 830: roffs[tok].name);
1.107 kristaps 831:
1.131 schwarze 832: while ((*bufp)[pos] && ! isspace((unsigned char)(*bufp)[pos]))
1.80 kristaps 833: pos++;
1.107 kristaps 834:
1.131 schwarze 835: while (isspace((unsigned char)(*bufp)[pos]))
1.106 kristaps 836: (*bufp)[pos++] = '\0';
1.80 kristaps 837: }
838:
1.106 kristaps 839: roffnode_push(r, tok, name, ln, ppos);
840:
841: /*
842: * At the beginning of a `de' macro, clear the existing string
843: * with the same name, if there is one. New content will be
844: * added from roff_block_text() in multiline mode.
845: */
1.107 kristaps 846:
1.106 kristaps 847: if (ROFF_de == tok)
1.108 schwarze 848: roff_setstr(r, name, "", 0);
1.76 kristaps 849:
1.79 kristaps 850: if ('\0' == (*bufp)[pos])
1.78 kristaps 851: return(ROFF_IGN);
852:
1.107 kristaps 853: /* If present, process the custom end-of-line marker. */
854:
1.78 kristaps 855: sv = pos;
1.131 schwarze 856: while ((*bufp)[pos] && ! isspace((unsigned char)(*bufp)[pos]))
1.78 kristaps 857: pos++;
858:
859: /*
860: * Note: groff does NOT like escape characters in the input.
861: * Instead of detecting this, we're just going to let it fly and
862: * to hell with it.
863: */
864:
865: assert(pos > sv);
866: sz = (size_t)(pos - sv);
867:
1.79 kristaps 868: if (1 == sz && '.' == (*bufp)[sv])
869: return(ROFF_IGN);
870:
1.98 schwarze 871: r->last->end = mandoc_malloc(sz + 1);
1.78 kristaps 872:
873: memcpy(r->last->end, *bufp + sv, sz);
874: r->last->end[(int)sz] = '\0';
875:
1.77 kristaps 876: if ((*bufp)[pos])
1.128 kristaps 877: mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.74 kristaps 878:
1.78 kristaps 879: return(ROFF_IGN);
880: }
881:
882:
883: /* ARGSUSED */
884: static enum rofferr
1.80 kristaps 885: roff_block_sub(ROFF_ARGS)
1.79 kristaps 886: {
887: enum rofft t;
888: int i, j;
889:
890: /*
891: * First check whether a custom macro exists at this level. If
892: * it does, then check against it. This is some of groff's
893: * stranger behaviours. If we encountered a custom end-scope
894: * tag and that tag also happens to be a "real" macro, then we
895: * need to try interpreting it again as a real macro. If it's
896: * not, then return ignore. Else continue.
897: */
898:
899: if (r->last->end) {
1.130 kristaps 900: for (i = pos, j = 0; r->last->end[j]; j++, i++)
1.79 kristaps 901: if ((*bufp)[i] != r->last->end[j])
902: break;
903:
904: if ('\0' == r->last->end[j] &&
905: ('\0' == (*bufp)[i] ||
906: ' ' == (*bufp)[i] ||
907: '\t' == (*bufp)[i])) {
908: roffnode_pop(r);
909: roffnode_cleanscope(r);
910:
1.130 kristaps 911: while (' ' == (*bufp)[i] || '\t' == (*bufp)[i])
912: i++;
913:
914: pos = i;
1.106 kristaps 915: if (ROFF_MAX != roff_parse(r, *bufp, &pos))
1.79 kristaps 916: return(ROFF_RERUN);
917: return(ROFF_IGN);
918: }
919: }
920:
921: /*
922: * If we have no custom end-query or lookup failed, then try
923: * pulling it out of the hashtable.
924: */
925:
1.137 schwarze 926: t = roff_parse(r, *bufp, &pos);
1.79 kristaps 927:
1.106 kristaps 928: /*
929: * Macros other than block-end are only significant
930: * in `de' blocks; elsewhere, simply throw them away.
931: */
932: if (ROFF_cblock != t) {
933: if (ROFF_de == tok)
934: roff_setstr(r, r->last->name, *bufp + ppos, 1);
1.79 kristaps 935: return(ROFF_IGN);
1.106 kristaps 936: }
1.79 kristaps 937:
938: assert(roffs[t].proc);
1.90 kristaps 939: return((*roffs[t].proc)(r, t, bufp, szp,
940: ln, ppos, pos, offs));
1.79 kristaps 941: }
942:
943:
944: /* ARGSUSED */
945: static enum rofferr
1.80 kristaps 946: roff_block_text(ROFF_ARGS)
1.78 kristaps 947: {
948:
1.106 kristaps 949: if (ROFF_de == tok)
950: roff_setstr(r, r->last->name, *bufp + pos, 1);
951:
1.78 kristaps 952: return(ROFF_IGN);
953: }
954:
955:
956: /* ARGSUSED */
957: static enum rofferr
1.82 kristaps 958: roff_cond_sub(ROFF_ARGS)
959: {
960: enum rofft t;
961: enum roffrule rr;
1.139 kristaps 962: char *ep;
1.82 kristaps 963:
964: rr = r->last->rule;
1.139 kristaps 965: roffnode_cleanscope(r);
1.82 kristaps 966:
1.139 kristaps 967: /*
968: * If the macro is unknown, first check if it contains a closing
969: * delimiter `\}'. If it does, close out our scope and return
970: * the currently-scoped rule (ignore or continue). Else, drop
971: * into the currently-scoped rule.
1.87 kristaps 972: */
973:
1.106 kristaps 974: if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos))) {
1.139 kristaps 975: ep = &(*bufp)[pos];
976: for ( ; NULL != (ep = strchr(ep, '\\')); ep++) {
977: ep++;
978: if ('}' != *ep)
979: continue;
1.144 kristaps 980:
981: /*
982: * Make the \} go away.
983: * This is a little haphazard, as it's not quite
984: * clear how nroff does this.
985: * If we're at the end of line, then just chop
986: * off the \} and resize the buffer.
987: * If we aren't, then conver it to spaces.
988: */
989:
990: if ('\0' == *(ep + 1)) {
991: *--ep = '\0';
992: *szp -= 2;
993: } else
994: *(ep - 1) = *ep = ' ';
995:
1.139 kristaps 996: roff_ccond(r, ROFF_ccond, bufp, szp,
997: ln, pos, pos + 2, offs);
998: break;
999: }
1.82 kristaps 1000: return(ROFFRULE_DENY == rr ? ROFF_IGN : ROFF_CONT);
1.100 kristaps 1001: }
1.82 kristaps 1002:
1003: /*
1004: * A denied conditional must evaluate its children if and only
1005: * if they're either structurally required (such as loops and
1006: * conditionals) or a closing macro.
1007: */
1.139 kristaps 1008:
1.82 kristaps 1009: if (ROFFRULE_DENY == rr)
1010: if ( ! (ROFFMAC_STRUCT & roffs[t].flags))
1011: if (ROFF_ccond != t)
1012: return(ROFF_IGN);
1013:
1014: assert(roffs[t].proc);
1.90 kristaps 1015: return((*roffs[t].proc)(r, t, bufp, szp,
1016: ln, ppos, pos, offs));
1.82 kristaps 1017: }
1018:
1019: /* ARGSUSED */
1020: static enum rofferr
1021: roff_cond_text(ROFF_ARGS)
1.78 kristaps 1022: {
1.140 kristaps 1023: char *ep;
1.82 kristaps 1024: enum roffrule rr;
1025:
1026: rr = r->last->rule;
1.140 kristaps 1027: roffnode_cleanscope(r);
1.82 kristaps 1028:
1.140 kristaps 1029: ep = &(*bufp)[pos];
1030: for ( ; NULL != (ep = strchr(ep, '\\')); ep++) {
1031: ep++;
1032: if ('}' != *ep)
1033: continue;
1034: *ep = '&';
1035: roff_ccond(r, ROFF_ccond, bufp, szp,
1036: ln, pos, pos + 2, offs);
1.78 kristaps 1037: }
1.82 kristaps 1038: return(ROFFRULE_DENY == rr ? ROFF_IGN : ROFF_CONT);
1.74 kristaps 1039: }
1040:
1.88 kristaps 1041: static enum roffrule
1042: roff_evalcond(const char *v, int *pos)
1043: {
1044:
1045: switch (v[*pos]) {
1046: case ('n'):
1047: (*pos)++;
1048: return(ROFFRULE_ALLOW);
1049: case ('e'):
1050: /* FALLTHROUGH */
1051: case ('o'):
1052: /* FALLTHROUGH */
1053: case ('t'):
1054: (*pos)++;
1055: return(ROFFRULE_DENY);
1056: default:
1057: break;
1058: }
1059:
1060: while (v[*pos] && ' ' != v[*pos])
1061: (*pos)++;
1062: return(ROFFRULE_DENY);
1063: }
1064:
1.75 kristaps 1065: /* ARGSUSED */
1.74 kristaps 1066: static enum rofferr
1.103 kristaps 1067: roff_line_ignore(ROFF_ARGS)
1.89 kristaps 1068: {
1.123 schwarze 1069:
1070: if (ROFF_it == tok)
1.128 kristaps 1071: mandoc_msg(MANDOCERR_REQUEST, r->parse, ln, ppos, "it");
1.89 kristaps 1072:
1073: return(ROFF_IGN);
1074: }
1075:
1.104 kristaps 1076: /* ARGSUSED */
1077: static enum rofferr
1.82 kristaps 1078: roff_cond(ROFF_ARGS)
1.74 kristaps 1079: {
1.77 kristaps 1080: int sv;
1.88 kristaps 1081: enum roffrule rule;
1.74 kristaps 1082:
1.134 kristaps 1083: /*
1084: * An `.el' has no conditional body: it will consume the value
1085: * of the current rstack entry set in prior `ie' calls or
1086: * defaults to DENY.
1087: *
1088: * If we're not an `el', however, then evaluate the conditional.
1089: */
1.133 kristaps 1090:
1.134 kristaps 1091: rule = ROFF_el == tok ?
1092: (r->rstackpos < 0 ?
1093: ROFFRULE_DENY : r->rstack[r->rstackpos--]) :
1094: roff_evalcond(*bufp, &pos);
1.77 kristaps 1095:
1096: sv = pos;
1.75 kristaps 1097: while (' ' == (*bufp)[pos])
1098: pos++;
1.74 kristaps 1099:
1.77 kristaps 1100: /*
1101: * Roff is weird. If we have just white-space after the
1102: * conditional, it's considered the BODY and we exit without
1103: * really doing anything. Warn about this. It's probably
1104: * wrong.
1105: */
1.88 kristaps 1106:
1.77 kristaps 1107: if ('\0' == (*bufp)[pos] && sv != pos) {
1.128 kristaps 1108: mandoc_msg(MANDOCERR_NOARGS, r->parse, ln, ppos, NULL);
1.107 kristaps 1109: return(ROFF_IGN);
1.77 kristaps 1110: }
1111:
1.106 kristaps 1112: roffnode_push(r, tok, NULL, ln, ppos);
1.77 kristaps 1113:
1.88 kristaps 1114: r->last->rule = rule;
1115:
1.134 kristaps 1116: /*
1117: * An if-else will put the NEGATION of the current evaluated
1118: * conditional into the stack of rules.
1119: */
1120:
1.84 schwarze 1121: if (ROFF_ie == tok) {
1.134 kristaps 1122: if (r->rstackpos == RSTACK_MAX - 1) {
1123: mandoc_msg(MANDOCERR_MEM,
1124: r->parse, ln, ppos, NULL);
1125: return(ROFF_ERR);
1126: }
1127: r->rstack[++r->rstackpos] =
1128: ROFFRULE_DENY == r->last->rule ?
1129: ROFFRULE_ALLOW : ROFFRULE_DENY;
1.82 kristaps 1130: }
1.88 kristaps 1131:
1132: /* If the parent has false as its rule, then so do we. */
1133:
1.109 kristaps 1134: if (r->last->parent && ROFFRULE_DENY == r->last->parent->rule)
1.84 schwarze 1135: r->last->rule = ROFFRULE_DENY;
1.88 kristaps 1136:
1137: /*
1138: * Determine scope. If we're invoked with "\{" trailing the
1139: * conditional, then we're in a multiline scope. Else our scope
1140: * expires on the next line.
1141: */
1.74 kristaps 1142:
1.75 kristaps 1143: r->last->endspan = 1;
1144:
1145: if ('\\' == (*bufp)[pos] && '{' == (*bufp)[pos + 1]) {
1146: r->last->endspan = -1;
1147: pos += 2;
1.109 kristaps 1148: }
1.74 kristaps 1149:
1.77 kristaps 1150: /*
1151: * If there are no arguments on the line, the next-line scope is
1152: * assumed.
1153: */
1154:
1.75 kristaps 1155: if ('\0' == (*bufp)[pos])
1156: return(ROFF_IGN);
1.77 kristaps 1157:
1158: /* Otherwise re-run the roff parser after recalculating. */
1.74 kristaps 1159:
1.75 kristaps 1160: *offs = pos;
1161: return(ROFF_RERUN);
1.83 schwarze 1162: }
1163:
1164:
1165: /* ARGSUSED */
1166: static enum rofferr
1.92 schwarze 1167: roff_ds(ROFF_ARGS)
1168: {
1.96 kristaps 1169: char *name, *string;
1170:
1171: /*
1172: * A symbol is named by the first word following the macro
1173: * invocation up to a space. Its value is anything after the
1174: * name's trailing whitespace and optional double-quote. Thus,
1175: *
1176: * [.ds foo "bar " ]
1177: *
1178: * will have `bar " ' as its value.
1179: */
1.92 schwarze 1180:
1.121 schwarze 1181: string = *bufp + pos;
1182: name = roff_getname(r, &string, ln, pos);
1.92 schwarze 1183: if ('\0' == *name)
1184: return(ROFF_IGN);
1185:
1.121 schwarze 1186: /* Read past initial double-quote. */
1187: if ('"' == *string)
1.92 schwarze 1188: string++;
1189:
1.96 kristaps 1190: /* The rest is the value. */
1.106 kristaps 1191: roff_setstr(r, name, string, 0);
1.92 schwarze 1192: return(ROFF_IGN);
1193: }
1194:
1.147 kristaps 1195: int
1196: roff_regisset(const struct roff *r, enum regs reg)
1197: {
1198:
1199: return(r->regs[(int)reg].set);
1200: }
1201:
1202: unsigned int
1203: roff_regget(const struct roff *r, enum regs reg)
1204: {
1205:
1206: return(r->regs[(int)reg].u);
1207: }
1208:
1209: void
1210: roff_regunset(struct roff *r, enum regs reg)
1211: {
1212:
1213: r->regs[(int)reg].set = 0;
1214: }
1.92 schwarze 1215:
1216: /* ARGSUSED */
1217: static enum rofferr
1.89 kristaps 1218: roff_nr(ROFF_ARGS)
1.83 schwarze 1219: {
1.121 schwarze 1220: const char *key;
1221: char *val;
1.138 kristaps 1222: int iv;
1.89 kristaps 1223:
1.121 schwarze 1224: val = *bufp + pos;
1225: key = roff_getname(r, &val, ln, pos);
1.89 kristaps 1226:
1227: if (0 == strcmp(key, "nS")) {
1.147 kristaps 1228: r->regs[(int)REG_nS].set = 1;
1.149 kristaps 1229: if ((iv = mandoc_strntoi(val, strlen(val), 10)) >= 0)
1.147 kristaps 1230: r->regs[(int)REG_nS].u = (unsigned)iv;
1.138 kristaps 1231: else
1.147 kristaps 1232: r->regs[(int)REG_nS].u = 0u;
1.109 kristaps 1233: }
1234:
1.122 schwarze 1235: return(ROFF_IGN);
1236: }
1237:
1238: /* ARGSUSED */
1239: static enum rofferr
1240: roff_rm(ROFF_ARGS)
1241: {
1242: const char *name;
1243: char *cp;
1244:
1245: cp = *bufp + pos;
1246: while ('\0' != *cp) {
1.127 kristaps 1247: name = roff_getname(r, &cp, ln, (int)(cp - *bufp));
1.122 schwarze 1248: if ('\0' != *name)
1249: roff_setstr(r, name, NULL, 0);
1250: }
1.109 kristaps 1251: return(ROFF_IGN);
1252: }
1253:
1254: /* ARGSUSED */
1255: static enum rofferr
1256: roff_TE(ROFF_ARGS)
1257: {
1258:
1259: if (NULL == r->tbl)
1.128 kristaps 1260: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.115 kristaps 1261: else
1.151 kristaps 1262: tbl_end(&r->tbl);
1.109 kristaps 1263:
1.112 kristaps 1264: return(ROFF_IGN);
1265: }
1266:
1267: /* ARGSUSED */
1268: static enum rofferr
1269: roff_T_(ROFF_ARGS)
1270: {
1271:
1272: if (NULL == r->tbl)
1.128 kristaps 1273: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.112 kristaps 1274: else
1.116 kristaps 1275: tbl_restart(ppos, ln, r->tbl);
1.112 kristaps 1276:
1.109 kristaps 1277: return(ROFF_IGN);
1278: }
1279:
1.156 kristaps 1280: #if 0
1281: static int
1.151 kristaps 1282: roff_closeeqn(struct roff *r)
1283: {
1284:
1285: return(r->eqn && ROFF_EQN == eqn_end(&r->eqn) ? 1 : 0);
1286: }
1.156 kristaps 1287: #endif
1.151 kristaps 1288:
1.156 kristaps 1289: static void
1.151 kristaps 1290: roff_openeqn(struct roff *r, const char *name, int line,
1291: int offs, const char *buf)
1.125 kristaps 1292: {
1.151 kristaps 1293: struct eqn_node *e;
1294: int poff;
1.125 kristaps 1295:
1296: assert(NULL == r->eqn);
1.151 kristaps 1297: e = eqn_alloc(name, offs, line, r->parse);
1.125 kristaps 1298:
1299: if (r->last_eqn)
1300: r->last_eqn->next = e;
1301: else
1302: r->first_eqn = r->last_eqn = e;
1303:
1304: r->eqn = r->last_eqn = e;
1.151 kristaps 1305:
1306: if (buf) {
1307: poff = 0;
1308: eqn_read(&r->eqn, line, buf, offs, &poff);
1309: }
1310: }
1311:
1312: /* ARGSUSED */
1313: static enum rofferr
1314: roff_EQ(ROFF_ARGS)
1315: {
1316:
1317: roff_openeqn(r, *bufp + pos, ln, ppos, NULL);
1.125 kristaps 1318: return(ROFF_IGN);
1319: }
1320:
1321: /* ARGSUSED */
1322: static enum rofferr
1323: roff_EN(ROFF_ARGS)
1324: {
1325:
1.128 kristaps 1326: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.125 kristaps 1327: return(ROFF_IGN);
1328: }
1329:
1330: /* ARGSUSED */
1331: static enum rofferr
1.109 kristaps 1332: roff_TS(ROFF_ARGS)
1333: {
1.118 kristaps 1334: struct tbl_node *t;
1.89 kristaps 1335:
1.115 kristaps 1336: if (r->tbl) {
1.128 kristaps 1337: mandoc_msg(MANDOCERR_SCOPEBROKEN, r->parse, ln, ppos, NULL);
1.151 kristaps 1338: tbl_end(&r->tbl);
1.115 kristaps 1339: }
1.83 schwarze 1340:
1.128 kristaps 1341: t = tbl_alloc(ppos, ln, r->parse);
1.113 kristaps 1342:
1343: if (r->last_tbl)
1344: r->last_tbl->next = t;
1345: else
1346: r->first_tbl = r->last_tbl = t;
1347:
1348: r->tbl = r->last_tbl = t;
1.83 schwarze 1349: return(ROFF_IGN);
1.92 schwarze 1350: }
1351:
1.105 kristaps 1352: /* ARGSUSED */
1353: static enum rofferr
1.164 kristaps 1354: roff_tr(ROFF_ARGS)
1355: {
1356: const char *p, *first, *second;
1357: size_t fsz, ssz;
1358: enum mandoc_esc esc;
1359:
1360: p = *bufp + pos;
1361:
1362: if ('\0' == *p) {
1363: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
1364: return(ROFF_IGN);
1365: }
1366:
1367: while ('\0' != *p) {
1368: fsz = ssz = 1;
1369:
1370: first = p++;
1371: if ('\\' == *first) {
1372: esc = mandoc_escape(&p, NULL, NULL);
1373: if (ESCAPE_ERROR == esc) {
1374: mandoc_msg
1375: (MANDOCERR_BADESCAPE, r->parse,
1376: ln, (int)(p - *bufp), NULL);
1377: return(ROFF_IGN);
1378: }
1379: fsz = (size_t)(p - first);
1380: }
1381:
1382: second = p++;
1383: if ('\\' == *second) {
1384: esc = mandoc_escape(&p, NULL, NULL);
1385: if (ESCAPE_ERROR == esc) {
1386: mandoc_msg
1387: (MANDOCERR_BADESCAPE, r->parse,
1388: ln, (int)(p - *bufp), NULL);
1389: return(ROFF_IGN);
1390: }
1391: ssz = (size_t)(p - second);
1.165 kristaps 1392: } else if ('\0' == *second) {
1.164 kristaps 1393: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse,
1394: ln, (int)(p - *bufp), NULL);
1395: second = " ";
1.165 kristaps 1396: p--;
1.164 kristaps 1397: }
1398:
1399: roff_setstrn(&r->chrtab, first, fsz, second, ssz, 0);
1400: }
1401:
1402: return(ROFF_IGN);
1403: }
1404:
1405: /* ARGSUSED */
1406: static enum rofferr
1.105 kristaps 1407: roff_so(ROFF_ARGS)
1408: {
1409: char *name;
1410:
1.128 kristaps 1411: mandoc_msg(MANDOCERR_SO, r->parse, ln, ppos, NULL);
1.105 kristaps 1412:
1413: /*
1414: * Handle `so'. Be EXTREMELY careful, as we shouldn't be
1415: * opening anything that's not in our cwd or anything beneath
1416: * it. Thus, explicitly disallow traversing up the file-system
1417: * or using absolute paths.
1418: */
1419:
1420: name = *bufp + pos;
1421: if ('/' == *name || strstr(name, "../") || strstr(name, "/..")) {
1.128 kristaps 1422: mandoc_msg(MANDOCERR_SOPATH, r->parse, ln, pos, NULL);
1.105 kristaps 1423: return(ROFF_ERR);
1424: }
1425:
1426: *offs = pos;
1427: return(ROFF_SO);
1428: }
1.92 schwarze 1429:
1.106 kristaps 1430: /* ARGSUSED */
1431: static enum rofferr
1432: roff_userdef(ROFF_ARGS)
1.99 kristaps 1433: {
1.106 kristaps 1434: const char *arg[9];
1435: char *cp, *n1, *n2;
1.119 schwarze 1436: int i;
1.106 kristaps 1437:
1438: /*
1439: * Collect pointers to macro argument strings
1440: * and null-terminate them.
1441: */
1442: cp = *bufp + pos;
1.119 schwarze 1443: for (i = 0; i < 9; i++)
1.120 schwarze 1444: arg[i] = '\0' == *cp ? "" :
1.136 kristaps 1445: mandoc_getarg(r->parse, &cp, ln, &pos);
1.99 kristaps 1446:
1.106 kristaps 1447: /*
1448: * Expand macro arguments.
1.99 kristaps 1449: */
1.106 kristaps 1450: *szp = 0;
1451: n1 = cp = mandoc_strdup(r->current_string);
1452: while (NULL != (cp = strstr(cp, "\\$"))) {
1453: i = cp[2] - '1';
1454: if (0 > i || 8 < i) {
1455: /* Not an argument invocation. */
1456: cp += 2;
1457: continue;
1458: }
1459:
1460: *szp = strlen(n1) - 3 + strlen(arg[i]) + 1;
1461: n2 = mandoc_malloc(*szp);
1462:
1463: strlcpy(n2, n1, (size_t)(cp - n1 + 1));
1464: strlcat(n2, arg[i], *szp);
1465: strlcat(n2, cp + 3, *szp);
1466:
1467: cp = n2 + (cp - n1);
1468: free(n1);
1469: n1 = n2;
1.99 kristaps 1470: }
1471:
1.106 kristaps 1472: /*
1473: * Replace the macro invocation
1474: * by the expanded macro.
1475: */
1476: free(*bufp);
1477: *bufp = n1;
1478: if (0 == *szp)
1479: *szp = strlen(*bufp) + 1;
1480:
1481: return(*szp > 1 && '\n' == (*bufp)[(int)*szp - 2] ?
1482: ROFF_REPARSE : ROFF_APPEND);
1.99 kristaps 1483: }
1.121 schwarze 1484:
1485: static char *
1486: roff_getname(struct roff *r, char **cpp, int ln, int pos)
1487: {
1488: char *name, *cp;
1489:
1490: name = *cpp;
1491: if ('\0' == *name)
1492: return(name);
1493:
1494: /* Read until end of name. */
1495: for (cp = name; '\0' != *cp && ' ' != *cp; cp++) {
1496: if ('\\' != *cp)
1497: continue;
1498: cp++;
1499: if ('\\' == *cp)
1500: continue;
1.128 kristaps 1501: mandoc_msg(MANDOCERR_NAMESC, r->parse, ln, pos, NULL);
1.121 schwarze 1502: *cp = '\0';
1503: name = cp;
1504: }
1505:
1506: /* Nil-terminate name. */
1507: if ('\0' != *cp)
1508: *(cp++) = '\0';
1509:
1510: /* Read past spaces. */
1511: while (' ' == *cp)
1512: cp++;
1513:
1514: *cpp = cp;
1515: return(name);
1516: }
1517:
1.106 kristaps 1518: /*
1519: * Store *string into the user-defined string called *name.
1520: * In multiline mode, append to an existing entry and append '\n';
1521: * else replace the existing entry, if there is one.
1522: * To clear an existing entry, call with (*r, *name, NULL, 0).
1523: */
1.94 kristaps 1524: static void
1.106 kristaps 1525: roff_setstr(struct roff *r, const char *name, const char *string,
1526: int multiline)
1.92 schwarze 1527: {
1.164 kristaps 1528:
1529: roff_setstrn(&r->strtab, name, strlen(name), string,
1530: string ? strlen(string) : 0, multiline);
1531: }
1532:
1533: static void
1.166 ! kristaps 1534: roff_setstrn(struct roffkv **r, const char *name, size_t namesz,
1.164 kristaps 1535: const char *string, size_t stringsz, int multiline)
1536: {
1.166 ! kristaps 1537: struct roffkv *n;
1.164 kristaps 1538: char *c;
1539: int i;
1540: size_t oldch, newch;
1.92 schwarze 1541:
1.106 kristaps 1542: /* Search for an existing string with the same name. */
1.164 kristaps 1543: n = *r;
1544:
1.166 ! kristaps 1545: while (n && strcmp(name, n->key.p))
1.92 schwarze 1546: n = n->next;
1.94 kristaps 1547:
1548: if (NULL == n) {
1.106 kristaps 1549: /* Create a new string table entry. */
1.166 ! kristaps 1550: n = mandoc_malloc(sizeof(struct roffkv));
! 1551: n->key.p = mandoc_strndup(name, namesz);
! 1552: n->key.sz = namesz;
! 1553: n->val.p = NULL;
! 1554: n->val.sz = 0;
1.164 kristaps 1555: n->next = *r;
1556: *r = n;
1.106 kristaps 1557: } else if (0 == multiline) {
1558: /* In multiline mode, append; else replace. */
1.166 ! kristaps 1559: free(n->val.p);
! 1560: n->val.p = NULL;
! 1561: n->val.sz = 0;
1.106 kristaps 1562: }
1563:
1564: if (NULL == string)
1565: return;
1566:
1567: /*
1568: * One additional byte for the '\n' in multiline mode,
1569: * and one for the terminating '\0'.
1570: */
1.164 kristaps 1571: newch = stringsz + (multiline ? 2u : 1u);
1572:
1.166 ! kristaps 1573: if (NULL == n->val.p) {
! 1574: n->val.p = mandoc_malloc(newch);
! 1575: *n->val.p = '\0';
1.106 kristaps 1576: oldch = 0;
1577: } else {
1.166 ! kristaps 1578: oldch = n->val.sz;
! 1579: n->val.p = mandoc_realloc(n->val.p, oldch + newch);
1.106 kristaps 1580: }
1581:
1582: /* Skip existing content in the destination buffer. */
1.166 ! kristaps 1583: c = n->val.p + (int)oldch;
1.106 kristaps 1584:
1585: /* Append new content to the destination buffer. */
1.164 kristaps 1586: i = 0;
1587: while (i < (int)stringsz) {
1.106 kristaps 1588: /*
1589: * Rudimentary roff copy mode:
1590: * Handle escaped backslashes.
1591: */
1.164 kristaps 1592: if ('\\' == string[i] && '\\' == string[i + 1])
1593: i++;
1594: *c++ = string[i++];
1.106 kristaps 1595: }
1.94 kristaps 1596:
1.106 kristaps 1597: /* Append terminating bytes. */
1598: if (multiline)
1599: *c++ = '\n';
1.163 kristaps 1600:
1.106 kristaps 1601: *c = '\0';
1.166 ! kristaps 1602: n->val.sz = (int)(c - n->val.p);
1.92 schwarze 1603: }
1604:
1.94 kristaps 1605: static const char *
1606: roff_getstrn(const struct roff *r, const char *name, size_t len)
1.92 schwarze 1607: {
1.166 ! kristaps 1608: const struct roffkv *n;
1.92 schwarze 1609:
1.164 kristaps 1610: for (n = r->strtab; n; n = n->next)
1.166 ! kristaps 1611: if (0 == strncmp(name, n->key.p, len) &&
! 1612: '\0' == n->key.p[(int)len])
! 1613: return(n->val.p);
1.94 kristaps 1614:
1.157 kristaps 1615: return(NULL);
1.92 schwarze 1616: }
1617:
1.94 kristaps 1618: static void
1.166 ! kristaps 1619: roff_freestr(struct roffkv **r)
1.92 schwarze 1620: {
1.166 ! kristaps 1621: struct roffkv *n, *nn;
1.92 schwarze 1622:
1.164 kristaps 1623: for (n = *r; n; n = nn) {
1.166 ! kristaps 1624: free(n->key.p);
! 1625: free(n->val.p);
1.92 schwarze 1626: nn = n->next;
1627: free(n);
1628: }
1.94 kristaps 1629:
1.164 kristaps 1630: *r = NULL;
1.114 kristaps 1631: }
1632:
1633: const struct tbl_span *
1634: roff_span(const struct roff *r)
1635: {
1636:
1637: return(r->tbl ? tbl_span(r->tbl) : NULL);
1.125 kristaps 1638: }
1639:
1640: const struct eqn *
1641: roff_eqn(const struct roff *r)
1642: {
1643:
1644: return(r->last_eqn ? &r->last_eqn->eqn : NULL);
1.151 kristaps 1645: }
1646:
1647: char
1648: roff_eqndelim(const struct roff *r)
1649: {
1650:
1651: return('\0');
1.164 kristaps 1652: }
1653:
1654: /*
1655: * Duplicate an input string, making the appropriate character
1656: * conversations (as stipulated by `tr') along the way.
1657: * Returns a heap-allocated string with all the replacements made.
1658: */
1659: char *
1660: roff_strdup(const struct roff *r, const char *p)
1661: {
1.166 ! kristaps 1662: const struct roffkv *cp;
1.164 kristaps 1663: char *res;
1664: const char *pp;
1665: size_t ssz, sz;
1666: enum mandoc_esc esc;
1667:
1668: if (NULL == r->chrtab)
1669: return(mandoc_strdup(p));
1670: else if ('\0' == *p)
1671: return(mandoc_strdup(""));
1672:
1673: /*
1674: * Step through each character looking for term matches
1675: * (remember that a `tr' can be invoked with an escape, which is
1676: * a glyph but the escape is multi-character).
1677: * We only do this if the character hash has been initialised
1678: * and the string is >0 length.
1679: */
1680:
1681: res = NULL;
1682: ssz = 0;
1683:
1684: while ('\0' != *p) {
1685: /* Search for term matches. */
1686: for (cp = r->chrtab; cp; cp = cp->next)
1.166 ! kristaps 1687: if (0 == strncmp(p, cp->key.p, cp->key.sz))
1.164 kristaps 1688: break;
1689:
1690: if (NULL != cp) {
1691: /*
1692: * A match has been found.
1693: * Append the match to the array and move
1694: * forward by its keysize.
1695: */
1.166 ! kristaps 1696: res = mandoc_realloc
! 1697: (res, ssz + cp->val.sz + 1);
! 1698: memcpy(res + ssz, cp->val.p, cp->val.sz);
! 1699: ssz += cp->val.sz;
! 1700: p += (int)cp->key.sz;
1.164 kristaps 1701: continue;
1702: }
1703:
1704: if ('\\' == *p) {
1705: /*
1706: * Handle escapes carefully: we need to copy
1707: * over just the escape itself, or else we might
1708: * do replacements within the escape itself.
1709: * Make sure to pass along the bogus string.
1710: */
1711: pp = p++;
1712: esc = mandoc_escape(&p, NULL, NULL);
1713: if (ESCAPE_ERROR == esc) {
1714: sz = strlen(pp);
1715: res = mandoc_realloc(res, ssz + sz + 1);
1716: memcpy(res + ssz, pp, sz);
1717: break;
1718: }
1719: /*
1720: * We bail out on bad escapes.
1721: * No need to warn: we already did so when
1722: * roff_res() was called.
1723: */
1724: sz = (int)(p - pp);
1725: res = mandoc_realloc(res, ssz + sz + 1);
1726: memcpy(res + ssz, pp, sz);
1727: ssz += sz;
1728: continue;
1729: }
1730:
1731: /* Just append the charater. */
1732: res = mandoc_realloc(res, ssz + 2);
1733: res[ssz++] = *p++;
1734: }
1735:
1736: res[(int)ssz] = '\0';
1737: return(res);
1.74 kristaps 1738: }
CVSweb