Annotation of mandoc/roff.c, Revision 1.170
1.170 ! schwarze 1: /* $Id: roff.c,v 1.169 2011/09/18 14:14:15 schwarze 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:
1.170 ! schwarze 34: /* Maximum number of string expansions per line, to break infinite loops. */
! 35: #define EXPAND_LIMIT 1000
! 36:
1.67 kristaps 37: enum rofft {
1.103 kristaps 38: ROFF_ad,
1.80 kristaps 39: ROFF_am,
40: ROFF_ami,
41: ROFF_am1,
42: ROFF_de,
43: ROFF_dei,
44: ROFF_de1,
1.83 schwarze 45: ROFF_ds,
1.82 kristaps 46: ROFF_el,
1.103 kristaps 47: ROFF_hy,
1.82 kristaps 48: ROFF_ie,
1.75 kristaps 49: ROFF_if,
1.76 kristaps 50: ROFF_ig,
1.123 schwarze 51: ROFF_it,
1.103 kristaps 52: ROFF_ne,
53: ROFF_nh,
1.104 kristaps 54: ROFF_nr,
1.124 schwarze 55: ROFF_ns,
56: ROFF_ps,
1.83 schwarze 57: ROFF_rm,
1.105 kristaps 58: ROFF_so,
1.124 schwarze 59: ROFF_ta,
1.83 schwarze 60: ROFF_tr,
1.109 kristaps 61: ROFF_TS,
62: ROFF_TE,
1.112 kristaps 63: ROFF_T_,
1.125 kristaps 64: ROFF_EQ,
65: ROFF_EN,
1.76 kristaps 66: ROFF_cblock,
1.141 kristaps 67: ROFF_ccond,
1.106 kristaps 68: ROFF_USERDEF,
1.67 kristaps 69: ROFF_MAX
70: };
71:
1.82 kristaps 72: enum roffrule {
73: ROFFRULE_ALLOW,
74: ROFFRULE_DENY
75: };
76:
1.147 kristaps 77: /*
78: * A single register entity. If "set" is zero, the value of the
79: * register should be the default one, which is per-register.
80: * Registers are assumed to be unsigned ints for now.
81: */
82: struct reg {
1.166 kristaps 83: int set; /* whether set or not */
84: unsigned int u; /* unsigned integer */
1.147 kristaps 85: };
86:
1.167 kristaps 87: /*
88: * An incredibly-simple string buffer.
89: */
1.94 kristaps 90: struct roffstr {
1.167 kristaps 91: char *p; /* nil-terminated buffer */
92: size_t sz; /* saved strlen(p) */
1.166 kristaps 93: };
94:
95: /*
1.167 kristaps 96: * A key-value roffstr pair as part of a singly-linked list.
1.166 kristaps 97: */
98: struct roffkv {
99: struct roffstr key;
100: struct roffstr val;
101: struct roffkv *next; /* next in list */
1.94 kristaps 102: };
103:
1.67 kristaps 104: struct roff {
1.128 kristaps 105: struct mparse *parse; /* parse point */
1.67 kristaps 106: struct roffnode *last; /* leaf of stack */
1.82 kristaps 107: enum roffrule rstack[RSTACK_MAX]; /* stack of !`ie' rules */
108: int rstackpos; /* position in rstack */
1.147 kristaps 109: struct reg regs[REG__MAX];
1.166 kristaps 110: struct roffkv *strtab; /* user-defined strings & macros */
1.167 kristaps 111: struct roffkv *xmbtab; /* multi-byte trans table (`tr') */
112: struct roffstr *xtab; /* single-byte trans table (`tr') */
1.106 kristaps 113: const char *current_string; /* value of last called user macro */
1.118 kristaps 114: struct tbl_node *first_tbl; /* first table parsed */
115: struct tbl_node *last_tbl; /* last table parsed */
116: struct tbl_node *tbl; /* current table being parsed */
1.125 kristaps 117: struct eqn_node *last_eqn; /* last equation parsed */
118: struct eqn_node *first_eqn; /* first equation parsed */
119: struct eqn_node *eqn; /* current equation being parsed */
1.79 kristaps 120: };
121:
1.67 kristaps 122: struct roffnode {
123: enum rofft tok; /* type of node */
124: struct roffnode *parent; /* up one in stack */
125: int line; /* parse line */
126: int col; /* parse col */
1.106 kristaps 127: char *name; /* node name, e.g. macro name */
1.79 kristaps 128: char *end; /* end-rules: custom token */
129: int endspan; /* end-rules: next-line or infty */
1.82 kristaps 130: enum roffrule rule; /* current evaluation rule */
1.67 kristaps 131: };
132:
133: #define ROFF_ARGS struct roff *r, /* parse ctx */ \
1.72 kristaps 134: enum rofft tok, /* tok of macro */ \
1.67 kristaps 135: char **bufp, /* input buffer */ \
136: size_t *szp, /* size of input buffer */ \
137: int ln, /* parse line */ \
1.75 kristaps 138: int ppos, /* original pos in buffer */ \
139: int pos, /* current pos in buffer */ \
1.74 kristaps 140: int *offs /* reset offset of buffer data */
1.67 kristaps 141:
142: typedef enum rofferr (*roffproc)(ROFF_ARGS);
143:
144: struct roffmac {
145: const char *name; /* macro name */
1.79 kristaps 146: roffproc proc; /* process new macro */
147: roffproc text; /* process as child text of macro */
148: roffproc sub; /* process as child of macro */
149: int flags;
150: #define ROFFMAC_STRUCT (1 << 0) /* always interpret */
1.85 kristaps 151: struct roffmac *next;
1.67 kristaps 152: };
153:
1.141 kristaps 154: struct predef {
155: const char *name; /* predefined input name */
156: const char *str; /* replacement symbol */
157: };
158:
159: #define PREDEF(__name, __str) \
160: { (__name), (__str) },
161:
1.155 kristaps 162: static enum rofft roffhash_find(const char *, size_t);
163: static void roffhash_init(void);
164: static void roffnode_cleanscope(struct roff *);
165: static void roffnode_pop(struct roff *);
166: static void roffnode_push(struct roff *, enum rofft,
167: const char *, int, int);
1.80 kristaps 168: static enum rofferr roff_block(ROFF_ARGS);
169: static enum rofferr roff_block_text(ROFF_ARGS);
170: static enum rofferr roff_block_sub(ROFF_ARGS);
171: static enum rofferr roff_cblock(ROFF_ARGS);
172: static enum rofferr roff_ccond(ROFF_ARGS);
1.82 kristaps 173: static enum rofferr roff_cond(ROFF_ARGS);
174: static enum rofferr roff_cond_text(ROFF_ARGS);
175: static enum rofferr roff_cond_sub(ROFF_ARGS);
1.92 schwarze 176: static enum rofferr roff_ds(ROFF_ARGS);
1.94 kristaps 177: static enum roffrule roff_evalcond(const char *, int *);
1.155 kristaps 178: static void roff_free1(struct roff *);
1.167 kristaps 179: static void roff_freestr(struct roffkv *);
1.121 schwarze 180: static char *roff_getname(struct roff *, char **, int, int);
1.94 kristaps 181: static const char *roff_getstrn(const struct roff *,
182: const char *, size_t);
1.103 kristaps 183: static enum rofferr roff_line_ignore(ROFF_ARGS);
1.89 kristaps 184: static enum rofferr roff_nr(ROFF_ARGS);
1.169 schwarze 185: static void roff_openeqn(struct roff *, const char *,
1.156 kristaps 186: int, int, const char *);
1.155 kristaps 187: static enum rofft roff_parse(struct roff *, const char *, int *);
188: static enum rofferr roff_parsetext(char *);
1.154 kristaps 189: static void roff_res(struct roff *,
1.142 kristaps 190: char **, size_t *, int, int);
1.122 schwarze 191: static enum rofferr roff_rm(ROFF_ARGS);
1.94 kristaps 192: static void roff_setstr(struct roff *,
1.106 kristaps 193: const char *, const char *, int);
1.166 kristaps 194: static void roff_setstrn(struct roffkv **, const char *,
1.164 kristaps 195: size_t, const char *, size_t, int);
1.105 kristaps 196: static enum rofferr roff_so(ROFF_ARGS);
1.164 kristaps 197: static enum rofferr roff_tr(ROFF_ARGS);
1.109 kristaps 198: static enum rofferr roff_TE(ROFF_ARGS);
199: static enum rofferr roff_TS(ROFF_ARGS);
1.125 kristaps 200: static enum rofferr roff_EQ(ROFF_ARGS);
201: static enum rofferr roff_EN(ROFF_ARGS);
1.112 kristaps 202: static enum rofferr roff_T_(ROFF_ARGS);
1.106 kristaps 203: static enum rofferr roff_userdef(ROFF_ARGS);
1.67 kristaps 204:
1.155 kristaps 205: /* See roffhash_find() */
1.85 kristaps 206:
207: #define ASCII_HI 126
208: #define ASCII_LO 33
209: #define HASHWIDTH (ASCII_HI - ASCII_LO + 1)
210:
211: static struct roffmac *hash[HASHWIDTH];
212:
213: static struct roffmac roffs[ROFF_MAX] = {
1.103 kristaps 214: { "ad", roff_line_ignore, NULL, NULL, 0, NULL },
1.85 kristaps 215: { "am", roff_block, roff_block_text, roff_block_sub, 0, NULL },
216: { "ami", roff_block, roff_block_text, roff_block_sub, 0, NULL },
217: { "am1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
218: { "de", roff_block, roff_block_text, roff_block_sub, 0, NULL },
219: { "dei", roff_block, roff_block_text, roff_block_sub, 0, NULL },
220: { "de1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.92 schwarze 221: { "ds", roff_ds, NULL, NULL, 0, NULL },
1.85 kristaps 222: { "el", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
1.103 kristaps 223: { "hy", roff_line_ignore, NULL, NULL, 0, NULL },
1.85 kristaps 224: { "ie", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
225: { "if", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
226: { "ig", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.123 schwarze 227: { "it", roff_line_ignore, NULL, NULL, 0, NULL },
1.103 kristaps 228: { "ne", roff_line_ignore, NULL, NULL, 0, NULL },
229: { "nh", roff_line_ignore, NULL, NULL, 0, NULL },
1.104 kristaps 230: { "nr", roff_nr, NULL, NULL, 0, NULL },
1.124 schwarze 231: { "ns", roff_line_ignore, NULL, NULL, 0, NULL },
232: { "ps", roff_line_ignore, NULL, NULL, 0, NULL },
1.122 schwarze 233: { "rm", roff_rm, NULL, NULL, 0, NULL },
1.105 kristaps 234: { "so", roff_so, NULL, NULL, 0, NULL },
1.124 schwarze 235: { "ta", roff_line_ignore, NULL, NULL, 0, NULL },
1.164 kristaps 236: { "tr", roff_tr, NULL, NULL, 0, NULL },
1.109 kristaps 237: { "TS", roff_TS, NULL, NULL, 0, NULL },
238: { "TE", roff_TE, NULL, NULL, 0, NULL },
1.112 kristaps 239: { "T&", roff_T_, NULL, NULL, 0, NULL },
1.125 kristaps 240: { "EQ", roff_EQ, NULL, NULL, 0, NULL },
241: { "EN", roff_EN, NULL, NULL, 0, NULL },
1.85 kristaps 242: { ".", roff_cblock, NULL, NULL, 0, NULL },
243: { "\\}", roff_ccond, NULL, NULL, 0, NULL },
1.106 kristaps 244: { NULL, roff_userdef, NULL, NULL, 0, NULL },
1.67 kristaps 245: };
246:
1.141 kristaps 247: /* Array of injected predefined strings. */
248: #define PREDEFS_MAX 38
249: static const struct predef predefs[PREDEFS_MAX] = {
250: #include "predefs.in"
251: };
252:
1.155 kristaps 253: /* See roffhash_find() */
1.85 kristaps 254: #define ROFF_HASH(p) (p[0] - ASCII_LO)
255:
256: static void
1.155 kristaps 257: roffhash_init(void)
1.85 kristaps 258: {
259: struct roffmac *n;
260: int buc, i;
261:
1.106 kristaps 262: for (i = 0; i < (int)ROFF_USERDEF; i++) {
1.85 kristaps 263: assert(roffs[i].name[0] >= ASCII_LO);
264: assert(roffs[i].name[0] <= ASCII_HI);
265:
266: buc = ROFF_HASH(roffs[i].name);
267:
268: if (NULL != (n = hash[buc])) {
269: for ( ; n->next; n = n->next)
270: /* Do nothing. */ ;
271: n->next = &roffs[i];
272: } else
273: hash[buc] = &roffs[i];
274: }
275: }
276:
1.67 kristaps 277: /*
278: * Look up a roff token by its name. Returns ROFF_MAX if no macro by
279: * the nil-terminated string name could be found.
280: */
281: static enum rofft
1.155 kristaps 282: roffhash_find(const char *p, size_t s)
1.67 kristaps 283: {
1.85 kristaps 284: int buc;
285: struct roffmac *n;
1.67 kristaps 286:
1.85 kristaps 287: /*
288: * libroff has an extremely simple hashtable, for the time
289: * being, which simply keys on the first character, which must
290: * be printable, then walks a chain. It works well enough until
291: * optimised.
292: */
293:
294: if (p[0] < ASCII_LO || p[0] > ASCII_HI)
295: return(ROFF_MAX);
296:
297: buc = ROFF_HASH(p);
298:
299: if (NULL == (n = hash[buc]))
300: return(ROFF_MAX);
301: for ( ; n; n = n->next)
1.106 kristaps 302: if (0 == strncmp(n->name, p, s) && '\0' == n->name[(int)s])
1.85 kristaps 303: return((enum rofft)(n - roffs));
1.67 kristaps 304:
305: return(ROFF_MAX);
306: }
307:
308:
309: /*
310: * Pop the current node off of the stack of roff instructions currently
311: * pending.
312: */
313: static void
314: roffnode_pop(struct roff *r)
315: {
316: struct roffnode *p;
317:
1.75 kristaps 318: assert(r->last);
319: p = r->last;
1.82 kristaps 320:
1.75 kristaps 321: r->last = r->last->parent;
1.106 kristaps 322: free(p->name);
323: free(p->end);
1.67 kristaps 324: free(p);
325: }
326:
327:
328: /*
329: * Push a roff node onto the instruction stack. This must later be
330: * removed with roffnode_pop().
331: */
1.98 schwarze 332: static void
1.106 kristaps 333: roffnode_push(struct roff *r, enum rofft tok, const char *name,
334: int line, int col)
1.67 kristaps 335: {
336: struct roffnode *p;
337:
1.98 schwarze 338: p = mandoc_calloc(1, sizeof(struct roffnode));
1.67 kristaps 339: p->tok = tok;
1.106 kristaps 340: if (name)
341: p->name = mandoc_strdup(name);
1.67 kristaps 342: p->parent = r->last;
343: p->line = line;
344: p->col = col;
1.79 kristaps 345: p->rule = p->parent ? p->parent->rule : ROFFRULE_DENY;
1.67 kristaps 346:
347: r->last = p;
348: }
349:
350:
351: static void
352: roff_free1(struct roff *r)
353: {
1.118 kristaps 354: struct tbl_node *t;
1.125 kristaps 355: struct eqn_node *e;
1.167 kristaps 356: int i;
1.67 kristaps 357:
1.125 kristaps 358: while (NULL != (t = r->first_tbl)) {
1.113 kristaps 359: r->first_tbl = t->next;
360: tbl_free(t);
1.109 kristaps 361: }
362:
1.113 kristaps 363: r->first_tbl = r->last_tbl = r->tbl = NULL;
364:
1.125 kristaps 365: while (NULL != (e = r->first_eqn)) {
366: r->first_eqn = e->next;
367: eqn_free(e);
368: }
369:
370: r->first_eqn = r->last_eqn = r->eqn = NULL;
371:
1.67 kristaps 372: while (r->last)
373: roffnode_pop(r);
1.109 kristaps 374:
1.167 kristaps 375: roff_freestr(r->strtab);
376: roff_freestr(r->xmbtab);
377:
378: r->strtab = r->xmbtab = NULL;
379:
380: if (r->xtab)
381: for (i = 0; i < 128; i++)
382: free(r->xtab[i].p);
383:
384: free(r->xtab);
385: r->xtab = NULL;
1.67 kristaps 386: }
387:
388: void
389: roff_reset(struct roff *r)
390: {
1.143 kristaps 391: int i;
1.67 kristaps 392:
393: roff_free1(r);
1.143 kristaps 394:
1.147 kristaps 395: memset(&r->regs, 0, sizeof(struct reg) * REG__MAX);
396:
1.143 kristaps 397: for (i = 0; i < PREDEFS_MAX; i++)
398: roff_setstr(r, predefs[i].name, predefs[i].str, 0);
1.67 kristaps 399: }
400:
401:
402: void
403: roff_free(struct roff *r)
404: {
405:
406: roff_free1(r);
407: free(r);
408: }
409:
410:
411: struct roff *
1.147 kristaps 412: roff_alloc(struct mparse *parse)
1.67 kristaps 413: {
414: struct roff *r;
1.141 kristaps 415: int i;
1.67 kristaps 416:
1.98 schwarze 417: r = mandoc_calloc(1, sizeof(struct roff));
1.128 kristaps 418: r->parse = parse;
1.82 kristaps 419: r->rstackpos = -1;
1.85 kristaps 420:
1.155 kristaps 421: roffhash_init();
1.141 kristaps 422:
423: for (i = 0; i < PREDEFS_MAX; i++)
424: roff_setstr(r, predefs[i].name, predefs[i].str, 0);
425:
1.67 kristaps 426: return(r);
427: }
428:
1.94 kristaps 429: /*
430: * Pre-filter each and every line for reserved words (one beginning with
431: * `\*', e.g., `\*(ab'). These must be handled before the actual line
432: * is processed.
1.153 kristaps 433: * This also checks the syntax of regular escapes.
1.154 kristaps 434: */
435: static void
1.142 kristaps 436: roff_res(struct roff *r, char **bufp, size_t *szp, int ln, int pos)
1.94 kristaps 437: {
1.152 kristaps 438: enum mandoc_esc esc;
1.108 schwarze 439: const char *stesc; /* start of an escape sequence ('\\') */
440: const char *stnam; /* start of the name, after "[(*" */
441: const char *cp; /* end of the name, e.g. before ']' */
442: const char *res; /* the string to be substituted */
1.170 ! schwarze 443: int i, maxl, expand_count;
1.94 kristaps 444: size_t nsz;
445: char *n;
446:
1.170 ! schwarze 447: expand_count = 0;
! 448:
1.154 kristaps 449: again:
1.108 schwarze 450: cp = *bufp + pos;
451: while (NULL != (cp = strchr(cp, '\\'))) {
452: stesc = cp++;
453:
454: /*
455: * The second character must be an asterisk.
456: * If it isn't, skip it anyway: It is escaped,
457: * so it can't start another escape sequence.
458: */
459:
460: if ('\0' == *cp)
1.154 kristaps 461: return;
1.152 kristaps 462:
463: if ('*' != *cp) {
464: res = cp;
465: esc = mandoc_escape(&cp, NULL, NULL);
466: if (ESCAPE_ERROR != esc)
467: continue;
468: cp = res;
1.153 kristaps 469: mandoc_msg
470: (MANDOCERR_BADESCAPE, r->parse,
471: ln, (int)(stesc - *bufp), NULL);
1.154 kristaps 472: return;
1.152 kristaps 473: }
474:
475: cp++;
1.108 schwarze 476:
477: /*
478: * The third character decides the length
479: * of the name of the string.
480: * Save a pointer to the name.
481: */
482:
1.94 kristaps 483: switch (*cp) {
1.108 schwarze 484: case ('\0'):
1.154 kristaps 485: return;
1.94 kristaps 486: case ('('):
487: cp++;
488: maxl = 2;
489: break;
490: case ('['):
491: cp++;
492: maxl = 0;
493: break;
494: default:
495: maxl = 1;
496: break;
497: }
1.108 schwarze 498: stnam = cp;
1.94 kristaps 499:
1.108 schwarze 500: /* Advance to the end of the name. */
1.94 kristaps 501:
502: for (i = 0; 0 == maxl || i < maxl; i++, cp++) {
1.153 kristaps 503: if ('\0' == *cp) {
504: mandoc_msg
505: (MANDOCERR_BADESCAPE,
506: r->parse, ln,
507: (int)(stesc - *bufp), NULL);
1.154 kristaps 508: return;
1.153 kristaps 509: }
1.94 kristaps 510: if (0 == maxl && ']' == *cp)
511: break;
512: }
513:
1.108 schwarze 514: /*
515: * Retrieve the replacement string; if it is
516: * undefined, resume searching for escapes.
517: */
518:
519: res = roff_getstrn(r, stnam, (size_t)i);
1.94 kristaps 520:
521: if (NULL == res) {
1.153 kristaps 522: mandoc_msg
523: (MANDOCERR_BADESCAPE, r->parse,
524: ln, (int)(stesc - *bufp), NULL);
1.142 kristaps 525: res = "";
1.94 kristaps 526: }
527:
1.108 schwarze 528: /* Replace the escape sequence by the string. */
529:
1.161 kristaps 530: pos = stesc - *bufp;
1.154 kristaps 531:
1.94 kristaps 532: nsz = *szp + strlen(res) + 1;
533: n = mandoc_malloc(nsz);
534:
1.108 schwarze 535: strlcpy(n, *bufp, (size_t)(stesc - *bufp + 1));
1.94 kristaps 536: strlcat(n, res, nsz);
537: strlcat(n, cp + (maxl ? 0 : 1), nsz);
538:
539: free(*bufp);
540:
541: *bufp = n;
542: *szp = nsz;
1.170 ! schwarze 543:
! 544: if (EXPAND_LIMIT >= ++expand_count)
! 545: goto again;
! 546:
! 547: /* Just leave the string unexpanded. */
! 548: mandoc_msg(MANDOCERR_ROFFLOOP, r->parse, ln, pos, NULL);
! 549: return;
1.154 kristaps 550: }
551: }
552:
553: /*
554: * Process text streams: convert all breakable hyphens into ASCII_HYPH.
555: */
556: static enum rofferr
557: roff_parsetext(char *p)
558: {
1.155 kristaps 559: char l, r;
1.154 kristaps 560: size_t sz;
561: const char *start;
562: enum mandoc_esc esc;
563:
564: start = p;
565:
566: while ('\0' != *p) {
567: sz = strcspn(p, "-\\");
568: p += sz;
569:
1.159 kristaps 570: if ('\0' == *p)
571: break;
572:
1.154 kristaps 573: if ('\\' == *p) {
574: /* Skip over escapes. */
575: p++;
576: esc = mandoc_escape
577: ((const char **)&p, NULL, NULL);
578: if (ESCAPE_ERROR == esc)
579: break;
1.155 kristaps 580: continue;
1.159 kristaps 581: } else if (p == start) {
1.158 kristaps 582: p++;
1.155 kristaps 583: continue;
1.158 kristaps 584: }
1.155 kristaps 585:
586: l = *(p - 1);
587: r = *(p + 1);
588: if ('\\' != l &&
589: '\t' != r && '\t' != l &&
590: ' ' != r && ' ' != l &&
591: '-' != r && '-' != l &&
592: ! isdigit((unsigned char)l) &&
1.159 kristaps 593: ! isdigit((unsigned char)r))
1.155 kristaps 594: *p = ASCII_HYPH;
595: p++;
1.94 kristaps 596: }
597:
1.154 kristaps 598: return(ROFF_CONT);
1.94 kristaps 599: }
600:
1.67 kristaps 601: enum rofferr
1.90 kristaps 602: roff_parseln(struct roff *r, int ln, char **bufp,
603: size_t *szp, int pos, int *offs)
1.67 kristaps 604: {
605: enum rofft t;
1.109 kristaps 606: enum rofferr e;
1.130 kristaps 607: int ppos, ctl;
1.79 kristaps 608:
609: /*
1.94 kristaps 610: * Run the reserved-word filter only if we have some reserved
611: * words to fill in.
612: */
613:
1.154 kristaps 614: roff_res(r, bufp, szp, ln, pos);
1.94 kristaps 615:
1.130 kristaps 616: ppos = pos;
617: ctl = mandoc_getcontrol(*bufp, &pos);
618:
1.94 kristaps 619: /*
1.79 kristaps 620: * First, if a scope is open and we're not a macro, pass the
621: * text through the macro's filter. If a scope isn't open and
622: * we're not a macro, just let it through.
1.125 kristaps 623: * Finally, if there's an equation scope open, divert it into it
624: * no matter our state.
1.79 kristaps 625: */
1.74 kristaps 626:
1.130 kristaps 627: if (r->last && ! ctl) {
1.78 kristaps 628: t = r->last->tok;
629: assert(roffs[t].text);
1.109 kristaps 630: e = (*roffs[t].text)
631: (r, t, bufp, szp, ln, pos, pos, offs);
632: assert(ROFF_IGN == e || ROFF_CONT == e);
1.125 kristaps 633: if (ROFF_CONT != e)
634: return(e);
635: if (r->eqn)
1.146 kristaps 636: return(eqn_read(&r->eqn, ln, *bufp, pos, offs));
1.125 kristaps 637: if (r->tbl)
1.130 kristaps 638: return(tbl_read(r->tbl, ln, *bufp, pos));
1.154 kristaps 639: return(roff_parsetext(*bufp + pos));
1.130 kristaps 640: } else if ( ! ctl) {
1.125 kristaps 641: if (r->eqn)
1.146 kristaps 642: return(eqn_read(&r->eqn, ln, *bufp, pos, offs));
1.109 kristaps 643: if (r->tbl)
1.130 kristaps 644: return(tbl_read(r->tbl, ln, *bufp, pos));
1.154 kristaps 645: return(roff_parsetext(*bufp + pos));
1.125 kristaps 646: } else if (r->eqn)
1.146 kristaps 647: return(eqn_read(&r->eqn, ln, *bufp, ppos, offs));
1.67 kristaps 648:
1.79 kristaps 649: /*
650: * If a scope is open, go to the child handler for that macro,
651: * as it may want to preprocess before doing anything with it.
1.125 kristaps 652: * Don't do so if an equation is open.
1.79 kristaps 653: */
1.78 kristaps 654:
1.79 kristaps 655: if (r->last) {
656: t = r->last->tok;
657: assert(roffs[t].sub);
658: return((*roffs[t].sub)
1.90 kristaps 659: (r, t, bufp, szp,
1.130 kristaps 660: ln, ppos, pos, offs));
1.79 kristaps 661: }
1.78 kristaps 662:
1.79 kristaps 663: /*
664: * Lastly, as we've no scope open, try to look up and execute
665: * the new macro. If no macro is found, simply return and let
666: * the compilers handle it.
667: */
1.67 kristaps 668:
1.106 kristaps 669: if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos)))
1.79 kristaps 670: return(ROFF_CONT);
1.67 kristaps 671:
1.75 kristaps 672: assert(roffs[t].proc);
1.78 kristaps 673: return((*roffs[t].proc)
1.90 kristaps 674: (r, t, bufp, szp,
675: ln, ppos, pos, offs));
1.74 kristaps 676: }
677:
678:
1.117 kristaps 679: void
1.74 kristaps 680: roff_endparse(struct roff *r)
681: {
682:
1.110 kristaps 683: if (r->last)
1.128 kristaps 684: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.109 kristaps 685: r->last->line, r->last->col, NULL);
1.117 kristaps 686:
1.125 kristaps 687: if (r->eqn) {
1.128 kristaps 688: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.148 kristaps 689: r->eqn->eqn.ln, r->eqn->eqn.pos, NULL);
1.151 kristaps 690: eqn_end(&r->eqn);
1.125 kristaps 691: }
692:
1.117 kristaps 693: if (r->tbl) {
1.128 kristaps 694: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.117 kristaps 695: r->tbl->line, r->tbl->pos, NULL);
1.151 kristaps 696: tbl_end(&r->tbl);
1.117 kristaps 697: }
1.67 kristaps 698: }
699:
700: /*
701: * Parse a roff node's type from the input buffer. This must be in the
702: * form of ".foo xxx" in the usual way.
703: */
704: static enum rofft
1.106 kristaps 705: roff_parse(struct roff *r, const char *buf, int *pos)
1.67 kristaps 706: {
1.106 kristaps 707: const char *mac;
708: size_t maclen;
1.67 kristaps 709: enum rofft t;
710:
1.144 kristaps 711: if ('\0' == buf[*pos] || '"' == buf[*pos] ||
712: '\t' == buf[*pos] || ' ' == buf[*pos])
1.67 kristaps 713: return(ROFF_MAX);
714:
1.144 kristaps 715: /*
716: * We stop the macro parse at an escape, tab, space, or nil.
717: * However, `\}' is also a valid macro, so make sure we don't
718: * clobber it by seeing the `\' as the end of token.
719: */
720:
1.106 kristaps 721: mac = buf + *pos;
1.144 kristaps 722: maclen = strcspn(mac + 1, " \\\t\0") + 1;
1.67 kristaps 723:
1.106 kristaps 724: t = (r->current_string = roff_getstrn(r, mac, maclen))
1.155 kristaps 725: ? ROFF_USERDEF : roffhash_find(mac, maclen);
1.67 kristaps 726:
1.127 kristaps 727: *pos += (int)maclen;
1.130 kristaps 728:
1.67 kristaps 729: while (buf[*pos] && ' ' == buf[*pos])
730: (*pos)++;
731:
732: return(t);
733: }
734:
735: /* ARGSUSED */
736: static enum rofferr
1.76 kristaps 737: roff_cblock(ROFF_ARGS)
1.67 kristaps 738: {
739:
1.79 kristaps 740: /*
741: * A block-close `..' should only be invoked as a child of an
742: * ignore macro, otherwise raise a warning and just ignore it.
743: */
744:
1.76 kristaps 745: if (NULL == r->last) {
1.128 kristaps 746: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.76 kristaps 747: return(ROFF_IGN);
748: }
1.67 kristaps 749:
1.81 kristaps 750: switch (r->last->tok) {
751: case (ROFF_am):
752: /* FALLTHROUGH */
753: case (ROFF_ami):
754: /* FALLTHROUGH */
755: case (ROFF_am1):
756: /* FALLTHROUGH */
757: case (ROFF_de):
1.108 schwarze 758: /* ROFF_de1 is remapped to ROFF_de in roff_block(). */
1.81 kristaps 759: /* FALLTHROUGH */
760: case (ROFF_dei):
761: /* FALLTHROUGH */
762: case (ROFF_ig):
763: break;
764: default:
1.128 kristaps 765: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.67 kristaps 766: return(ROFF_IGN);
1.76 kristaps 767: }
1.67 kristaps 768:
1.76 kristaps 769: if ((*bufp)[pos])
1.128 kristaps 770: mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.71 kristaps 771:
772: roffnode_pop(r);
1.76 kristaps 773: roffnode_cleanscope(r);
774: return(ROFF_IGN);
1.71 kristaps 775:
1.67 kristaps 776: }
777:
778:
1.76 kristaps 779: static void
780: roffnode_cleanscope(struct roff *r)
1.67 kristaps 781: {
782:
1.76 kristaps 783: while (r->last) {
784: if (--r->last->endspan < 0)
785: break;
786: roffnode_pop(r);
787: }
1.67 kristaps 788: }
789:
790:
1.75 kristaps 791: /* ARGSUSED */
1.74 kristaps 792: static enum rofferr
1.75 kristaps 793: roff_ccond(ROFF_ARGS)
1.74 kristaps 794: {
795:
1.76 kristaps 796: if (NULL == r->last) {
1.128 kristaps 797: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.76 kristaps 798: return(ROFF_IGN);
799: }
800:
1.82 kristaps 801: switch (r->last->tok) {
802: case (ROFF_el):
803: /* FALLTHROUGH */
804: case (ROFF_ie):
805: /* FALLTHROUGH */
806: case (ROFF_if):
807: break;
808: default:
1.128 kristaps 809: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.75 kristaps 810: return(ROFF_IGN);
811: }
812:
1.76 kristaps 813: if (r->last->endspan > -1) {
1.128 kristaps 814: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.76 kristaps 815: return(ROFF_IGN);
816: }
817:
818: if ((*bufp)[pos])
1.128 kristaps 819: mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.76 kristaps 820:
1.75 kristaps 821: roffnode_pop(r);
1.76 kristaps 822: roffnode_cleanscope(r);
823: return(ROFF_IGN);
824: }
825:
1.75 kristaps 826:
1.76 kristaps 827: /* ARGSUSED */
828: static enum rofferr
1.80 kristaps 829: roff_block(ROFF_ARGS)
1.76 kristaps 830: {
1.78 kristaps 831: int sv;
832: size_t sz;
1.106 kristaps 833: char *name;
834:
835: name = NULL;
1.76 kristaps 836:
1.106 kristaps 837: if (ROFF_ig != tok) {
838: if ('\0' == (*bufp)[pos]) {
1.128 kristaps 839: mandoc_msg(MANDOCERR_NOARGS, r->parse, ln, ppos, NULL);
1.106 kristaps 840: return(ROFF_IGN);
841: }
1.107 kristaps 842:
843: /*
844: * Re-write `de1', since we don't really care about
845: * groff's strange compatibility mode, into `de'.
846: */
847:
1.106 kristaps 848: if (ROFF_de1 == tok)
849: tok = ROFF_de;
850: if (ROFF_de == tok)
851: name = *bufp + pos;
852: else
1.128 kristaps 853: mandoc_msg(MANDOCERR_REQUEST, r->parse, ln, ppos,
1.106 kristaps 854: roffs[tok].name);
1.107 kristaps 855:
1.131 schwarze 856: while ((*bufp)[pos] && ! isspace((unsigned char)(*bufp)[pos]))
1.80 kristaps 857: pos++;
1.107 kristaps 858:
1.131 schwarze 859: while (isspace((unsigned char)(*bufp)[pos]))
1.106 kristaps 860: (*bufp)[pos++] = '\0';
1.80 kristaps 861: }
862:
1.106 kristaps 863: roffnode_push(r, tok, name, ln, ppos);
864:
865: /*
866: * At the beginning of a `de' macro, clear the existing string
867: * with the same name, if there is one. New content will be
868: * added from roff_block_text() in multiline mode.
869: */
1.107 kristaps 870:
1.106 kristaps 871: if (ROFF_de == tok)
1.108 schwarze 872: roff_setstr(r, name, "", 0);
1.76 kristaps 873:
1.79 kristaps 874: if ('\0' == (*bufp)[pos])
1.78 kristaps 875: return(ROFF_IGN);
876:
1.107 kristaps 877: /* If present, process the custom end-of-line marker. */
878:
1.78 kristaps 879: sv = pos;
1.131 schwarze 880: while ((*bufp)[pos] && ! isspace((unsigned char)(*bufp)[pos]))
1.78 kristaps 881: pos++;
882:
883: /*
884: * Note: groff does NOT like escape characters in the input.
885: * Instead of detecting this, we're just going to let it fly and
886: * to hell with it.
887: */
888:
889: assert(pos > sv);
890: sz = (size_t)(pos - sv);
891:
1.79 kristaps 892: if (1 == sz && '.' == (*bufp)[sv])
893: return(ROFF_IGN);
894:
1.98 schwarze 895: r->last->end = mandoc_malloc(sz + 1);
1.78 kristaps 896:
897: memcpy(r->last->end, *bufp + sv, sz);
898: r->last->end[(int)sz] = '\0';
899:
1.77 kristaps 900: if ((*bufp)[pos])
1.128 kristaps 901: mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.74 kristaps 902:
1.78 kristaps 903: return(ROFF_IGN);
904: }
905:
906:
907: /* ARGSUSED */
908: static enum rofferr
1.80 kristaps 909: roff_block_sub(ROFF_ARGS)
1.79 kristaps 910: {
911: enum rofft t;
912: int i, j;
913:
914: /*
915: * First check whether a custom macro exists at this level. If
916: * it does, then check against it. This is some of groff's
917: * stranger behaviours. If we encountered a custom end-scope
918: * tag and that tag also happens to be a "real" macro, then we
919: * need to try interpreting it again as a real macro. If it's
920: * not, then return ignore. Else continue.
921: */
922:
923: if (r->last->end) {
1.130 kristaps 924: for (i = pos, j = 0; r->last->end[j]; j++, i++)
1.79 kristaps 925: if ((*bufp)[i] != r->last->end[j])
926: break;
927:
928: if ('\0' == r->last->end[j] &&
929: ('\0' == (*bufp)[i] ||
930: ' ' == (*bufp)[i] ||
931: '\t' == (*bufp)[i])) {
932: roffnode_pop(r);
933: roffnode_cleanscope(r);
934:
1.130 kristaps 935: while (' ' == (*bufp)[i] || '\t' == (*bufp)[i])
936: i++;
937:
938: pos = i;
1.106 kristaps 939: if (ROFF_MAX != roff_parse(r, *bufp, &pos))
1.79 kristaps 940: return(ROFF_RERUN);
941: return(ROFF_IGN);
942: }
943: }
944:
945: /*
946: * If we have no custom end-query or lookup failed, then try
947: * pulling it out of the hashtable.
948: */
949:
1.137 schwarze 950: t = roff_parse(r, *bufp, &pos);
1.79 kristaps 951:
1.106 kristaps 952: /*
953: * Macros other than block-end are only significant
954: * in `de' blocks; elsewhere, simply throw them away.
955: */
956: if (ROFF_cblock != t) {
957: if (ROFF_de == tok)
958: roff_setstr(r, r->last->name, *bufp + ppos, 1);
1.79 kristaps 959: return(ROFF_IGN);
1.106 kristaps 960: }
1.79 kristaps 961:
962: assert(roffs[t].proc);
1.90 kristaps 963: return((*roffs[t].proc)(r, t, bufp, szp,
964: ln, ppos, pos, offs));
1.79 kristaps 965: }
966:
967:
968: /* ARGSUSED */
969: static enum rofferr
1.80 kristaps 970: roff_block_text(ROFF_ARGS)
1.78 kristaps 971: {
972:
1.106 kristaps 973: if (ROFF_de == tok)
974: roff_setstr(r, r->last->name, *bufp + pos, 1);
975:
1.78 kristaps 976: return(ROFF_IGN);
977: }
978:
979:
980: /* ARGSUSED */
981: static enum rofferr
1.82 kristaps 982: roff_cond_sub(ROFF_ARGS)
983: {
984: enum rofft t;
985: enum roffrule rr;
1.139 kristaps 986: char *ep;
1.82 kristaps 987:
988: rr = r->last->rule;
1.139 kristaps 989: roffnode_cleanscope(r);
1.82 kristaps 990:
1.139 kristaps 991: /*
992: * If the macro is unknown, first check if it contains a closing
993: * delimiter `\}'. If it does, close out our scope and return
994: * the currently-scoped rule (ignore or continue). Else, drop
995: * into the currently-scoped rule.
1.87 kristaps 996: */
997:
1.106 kristaps 998: if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos))) {
1.139 kristaps 999: ep = &(*bufp)[pos];
1000: for ( ; NULL != (ep = strchr(ep, '\\')); ep++) {
1001: ep++;
1002: if ('}' != *ep)
1003: continue;
1.144 kristaps 1004:
1005: /*
1006: * Make the \} go away.
1007: * This is a little haphazard, as it's not quite
1008: * clear how nroff does this.
1009: * If we're at the end of line, then just chop
1010: * off the \} and resize the buffer.
1011: * If we aren't, then conver it to spaces.
1012: */
1013:
1014: if ('\0' == *(ep + 1)) {
1015: *--ep = '\0';
1016: *szp -= 2;
1017: } else
1018: *(ep - 1) = *ep = ' ';
1019:
1.139 kristaps 1020: roff_ccond(r, ROFF_ccond, bufp, szp,
1021: ln, pos, pos + 2, offs);
1022: break;
1023: }
1.82 kristaps 1024: return(ROFFRULE_DENY == rr ? ROFF_IGN : ROFF_CONT);
1.100 kristaps 1025: }
1.82 kristaps 1026:
1027: /*
1028: * A denied conditional must evaluate its children if and only
1029: * if they're either structurally required (such as loops and
1030: * conditionals) or a closing macro.
1031: */
1.139 kristaps 1032:
1.82 kristaps 1033: if (ROFFRULE_DENY == rr)
1034: if ( ! (ROFFMAC_STRUCT & roffs[t].flags))
1035: if (ROFF_ccond != t)
1036: return(ROFF_IGN);
1037:
1038: assert(roffs[t].proc);
1.90 kristaps 1039: return((*roffs[t].proc)(r, t, bufp, szp,
1040: ln, ppos, pos, offs));
1.82 kristaps 1041: }
1042:
1043: /* ARGSUSED */
1044: static enum rofferr
1045: roff_cond_text(ROFF_ARGS)
1.78 kristaps 1046: {
1.140 kristaps 1047: char *ep;
1.82 kristaps 1048: enum roffrule rr;
1049:
1050: rr = r->last->rule;
1.140 kristaps 1051: roffnode_cleanscope(r);
1.82 kristaps 1052:
1.140 kristaps 1053: ep = &(*bufp)[pos];
1054: for ( ; NULL != (ep = strchr(ep, '\\')); ep++) {
1055: ep++;
1056: if ('}' != *ep)
1057: continue;
1058: *ep = '&';
1059: roff_ccond(r, ROFF_ccond, bufp, szp,
1060: ln, pos, pos + 2, offs);
1.78 kristaps 1061: }
1.82 kristaps 1062: return(ROFFRULE_DENY == rr ? ROFF_IGN : ROFF_CONT);
1.74 kristaps 1063: }
1064:
1.88 kristaps 1065: static enum roffrule
1066: roff_evalcond(const char *v, int *pos)
1067: {
1068:
1069: switch (v[*pos]) {
1070: case ('n'):
1071: (*pos)++;
1072: return(ROFFRULE_ALLOW);
1073: case ('e'):
1074: /* FALLTHROUGH */
1075: case ('o'):
1076: /* FALLTHROUGH */
1077: case ('t'):
1078: (*pos)++;
1079: return(ROFFRULE_DENY);
1080: default:
1081: break;
1082: }
1083:
1084: while (v[*pos] && ' ' != v[*pos])
1085: (*pos)++;
1086: return(ROFFRULE_DENY);
1087: }
1088:
1.75 kristaps 1089: /* ARGSUSED */
1.74 kristaps 1090: static enum rofferr
1.103 kristaps 1091: roff_line_ignore(ROFF_ARGS)
1.89 kristaps 1092: {
1.123 schwarze 1093:
1094: if (ROFF_it == tok)
1.128 kristaps 1095: mandoc_msg(MANDOCERR_REQUEST, r->parse, ln, ppos, "it");
1.89 kristaps 1096:
1097: return(ROFF_IGN);
1098: }
1099:
1.104 kristaps 1100: /* ARGSUSED */
1101: static enum rofferr
1.82 kristaps 1102: roff_cond(ROFF_ARGS)
1.74 kristaps 1103: {
1.77 kristaps 1104: int sv;
1.88 kristaps 1105: enum roffrule rule;
1.74 kristaps 1106:
1.134 kristaps 1107: /*
1108: * An `.el' has no conditional body: it will consume the value
1109: * of the current rstack entry set in prior `ie' calls or
1110: * defaults to DENY.
1111: *
1112: * If we're not an `el', however, then evaluate the conditional.
1113: */
1.133 kristaps 1114:
1.134 kristaps 1115: rule = ROFF_el == tok ?
1116: (r->rstackpos < 0 ?
1117: ROFFRULE_DENY : r->rstack[r->rstackpos--]) :
1118: roff_evalcond(*bufp, &pos);
1.77 kristaps 1119:
1120: sv = pos;
1.75 kristaps 1121: while (' ' == (*bufp)[pos])
1122: pos++;
1.74 kristaps 1123:
1.77 kristaps 1124: /*
1125: * Roff is weird. If we have just white-space after the
1126: * conditional, it's considered the BODY and we exit without
1127: * really doing anything. Warn about this. It's probably
1128: * wrong.
1129: */
1.88 kristaps 1130:
1.77 kristaps 1131: if ('\0' == (*bufp)[pos] && sv != pos) {
1.128 kristaps 1132: mandoc_msg(MANDOCERR_NOARGS, r->parse, ln, ppos, NULL);
1.107 kristaps 1133: return(ROFF_IGN);
1.77 kristaps 1134: }
1135:
1.106 kristaps 1136: roffnode_push(r, tok, NULL, ln, ppos);
1.77 kristaps 1137:
1.88 kristaps 1138: r->last->rule = rule;
1139:
1.134 kristaps 1140: /*
1141: * An if-else will put the NEGATION of the current evaluated
1142: * conditional into the stack of rules.
1143: */
1144:
1.84 schwarze 1145: if (ROFF_ie == tok) {
1.134 kristaps 1146: if (r->rstackpos == RSTACK_MAX - 1) {
1147: mandoc_msg(MANDOCERR_MEM,
1148: r->parse, ln, ppos, NULL);
1149: return(ROFF_ERR);
1150: }
1151: r->rstack[++r->rstackpos] =
1152: ROFFRULE_DENY == r->last->rule ?
1153: ROFFRULE_ALLOW : ROFFRULE_DENY;
1.82 kristaps 1154: }
1.88 kristaps 1155:
1156: /* If the parent has false as its rule, then so do we. */
1157:
1.109 kristaps 1158: if (r->last->parent && ROFFRULE_DENY == r->last->parent->rule)
1.84 schwarze 1159: r->last->rule = ROFFRULE_DENY;
1.88 kristaps 1160:
1161: /*
1162: * Determine scope. If we're invoked with "\{" trailing the
1163: * conditional, then we're in a multiline scope. Else our scope
1164: * expires on the next line.
1165: */
1.74 kristaps 1166:
1.75 kristaps 1167: r->last->endspan = 1;
1168:
1169: if ('\\' == (*bufp)[pos] && '{' == (*bufp)[pos + 1]) {
1170: r->last->endspan = -1;
1171: pos += 2;
1.109 kristaps 1172: }
1.74 kristaps 1173:
1.77 kristaps 1174: /*
1175: * If there are no arguments on the line, the next-line scope is
1176: * assumed.
1177: */
1178:
1.75 kristaps 1179: if ('\0' == (*bufp)[pos])
1180: return(ROFF_IGN);
1.77 kristaps 1181:
1182: /* Otherwise re-run the roff parser after recalculating. */
1.74 kristaps 1183:
1.75 kristaps 1184: *offs = pos;
1185: return(ROFF_RERUN);
1.83 schwarze 1186: }
1187:
1188:
1189: /* ARGSUSED */
1190: static enum rofferr
1.92 schwarze 1191: roff_ds(ROFF_ARGS)
1192: {
1.96 kristaps 1193: char *name, *string;
1194:
1195: /*
1196: * A symbol is named by the first word following the macro
1197: * invocation up to a space. Its value is anything after the
1198: * name's trailing whitespace and optional double-quote. Thus,
1199: *
1200: * [.ds foo "bar " ]
1201: *
1202: * will have `bar " ' as its value.
1203: */
1.92 schwarze 1204:
1.121 schwarze 1205: string = *bufp + pos;
1206: name = roff_getname(r, &string, ln, pos);
1.92 schwarze 1207: if ('\0' == *name)
1208: return(ROFF_IGN);
1209:
1.121 schwarze 1210: /* Read past initial double-quote. */
1211: if ('"' == *string)
1.92 schwarze 1212: string++;
1213:
1.96 kristaps 1214: /* The rest is the value. */
1.106 kristaps 1215: roff_setstr(r, name, string, 0);
1.92 schwarze 1216: return(ROFF_IGN);
1217: }
1218:
1.147 kristaps 1219: int
1220: roff_regisset(const struct roff *r, enum regs reg)
1221: {
1222:
1223: return(r->regs[(int)reg].set);
1224: }
1225:
1226: unsigned int
1227: roff_regget(const struct roff *r, enum regs reg)
1228: {
1229:
1230: return(r->regs[(int)reg].u);
1231: }
1232:
1233: void
1234: roff_regunset(struct roff *r, enum regs reg)
1235: {
1236:
1237: r->regs[(int)reg].set = 0;
1238: }
1.92 schwarze 1239:
1240: /* ARGSUSED */
1241: static enum rofferr
1.89 kristaps 1242: roff_nr(ROFF_ARGS)
1.83 schwarze 1243: {
1.121 schwarze 1244: const char *key;
1245: char *val;
1.138 kristaps 1246: int iv;
1.89 kristaps 1247:
1.121 schwarze 1248: val = *bufp + pos;
1249: key = roff_getname(r, &val, ln, pos);
1.89 kristaps 1250:
1251: if (0 == strcmp(key, "nS")) {
1.147 kristaps 1252: r->regs[(int)REG_nS].set = 1;
1.149 kristaps 1253: if ((iv = mandoc_strntoi(val, strlen(val), 10)) >= 0)
1.147 kristaps 1254: r->regs[(int)REG_nS].u = (unsigned)iv;
1.138 kristaps 1255: else
1.147 kristaps 1256: r->regs[(int)REG_nS].u = 0u;
1.109 kristaps 1257: }
1258:
1.122 schwarze 1259: return(ROFF_IGN);
1260: }
1261:
1262: /* ARGSUSED */
1263: static enum rofferr
1264: roff_rm(ROFF_ARGS)
1265: {
1266: const char *name;
1267: char *cp;
1268:
1269: cp = *bufp + pos;
1270: while ('\0' != *cp) {
1.127 kristaps 1271: name = roff_getname(r, &cp, ln, (int)(cp - *bufp));
1.122 schwarze 1272: if ('\0' != *name)
1273: roff_setstr(r, name, NULL, 0);
1274: }
1.109 kristaps 1275: return(ROFF_IGN);
1276: }
1277:
1278: /* ARGSUSED */
1279: static enum rofferr
1280: roff_TE(ROFF_ARGS)
1281: {
1282:
1283: if (NULL == r->tbl)
1.128 kristaps 1284: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.115 kristaps 1285: else
1.151 kristaps 1286: tbl_end(&r->tbl);
1.109 kristaps 1287:
1.112 kristaps 1288: return(ROFF_IGN);
1289: }
1290:
1291: /* ARGSUSED */
1292: static enum rofferr
1293: roff_T_(ROFF_ARGS)
1294: {
1295:
1296: if (NULL == r->tbl)
1.128 kristaps 1297: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.112 kristaps 1298: else
1.116 kristaps 1299: tbl_restart(ppos, ln, r->tbl);
1.112 kristaps 1300:
1.109 kristaps 1301: return(ROFF_IGN);
1302: }
1303:
1.156 kristaps 1304: #if 0
1305: static int
1.151 kristaps 1306: roff_closeeqn(struct roff *r)
1307: {
1308:
1309: return(r->eqn && ROFF_EQN == eqn_end(&r->eqn) ? 1 : 0);
1310: }
1.156 kristaps 1311: #endif
1.151 kristaps 1312:
1.156 kristaps 1313: static void
1.151 kristaps 1314: roff_openeqn(struct roff *r, const char *name, int line,
1315: int offs, const char *buf)
1.125 kristaps 1316: {
1.151 kristaps 1317: struct eqn_node *e;
1318: int poff;
1.125 kristaps 1319:
1320: assert(NULL == r->eqn);
1.151 kristaps 1321: e = eqn_alloc(name, offs, line, r->parse);
1.125 kristaps 1322:
1323: if (r->last_eqn)
1324: r->last_eqn->next = e;
1325: else
1326: r->first_eqn = r->last_eqn = e;
1327:
1328: r->eqn = r->last_eqn = e;
1.151 kristaps 1329:
1330: if (buf) {
1331: poff = 0;
1332: eqn_read(&r->eqn, line, buf, offs, &poff);
1333: }
1334: }
1335:
1336: /* ARGSUSED */
1337: static enum rofferr
1338: roff_EQ(ROFF_ARGS)
1339: {
1340:
1341: roff_openeqn(r, *bufp + pos, ln, ppos, NULL);
1.125 kristaps 1342: return(ROFF_IGN);
1343: }
1344:
1345: /* ARGSUSED */
1346: static enum rofferr
1347: roff_EN(ROFF_ARGS)
1348: {
1349:
1.128 kristaps 1350: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.125 kristaps 1351: return(ROFF_IGN);
1352: }
1353:
1354: /* ARGSUSED */
1355: static enum rofferr
1.109 kristaps 1356: roff_TS(ROFF_ARGS)
1357: {
1.118 kristaps 1358: struct tbl_node *t;
1.89 kristaps 1359:
1.115 kristaps 1360: if (r->tbl) {
1.128 kristaps 1361: mandoc_msg(MANDOCERR_SCOPEBROKEN, r->parse, ln, ppos, NULL);
1.151 kristaps 1362: tbl_end(&r->tbl);
1.115 kristaps 1363: }
1.83 schwarze 1364:
1.128 kristaps 1365: t = tbl_alloc(ppos, ln, r->parse);
1.113 kristaps 1366:
1367: if (r->last_tbl)
1368: r->last_tbl->next = t;
1369: else
1370: r->first_tbl = r->last_tbl = t;
1371:
1372: r->tbl = r->last_tbl = t;
1.83 schwarze 1373: return(ROFF_IGN);
1.92 schwarze 1374: }
1375:
1.105 kristaps 1376: /* ARGSUSED */
1377: static enum rofferr
1.164 kristaps 1378: roff_tr(ROFF_ARGS)
1379: {
1380: const char *p, *first, *second;
1381: size_t fsz, ssz;
1382: enum mandoc_esc esc;
1383:
1384: p = *bufp + pos;
1385:
1386: if ('\0' == *p) {
1387: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
1388: return(ROFF_IGN);
1389: }
1390:
1391: while ('\0' != *p) {
1392: fsz = ssz = 1;
1393:
1394: first = p++;
1395: if ('\\' == *first) {
1396: esc = mandoc_escape(&p, NULL, NULL);
1397: if (ESCAPE_ERROR == esc) {
1398: mandoc_msg
1399: (MANDOCERR_BADESCAPE, r->parse,
1400: ln, (int)(p - *bufp), NULL);
1401: return(ROFF_IGN);
1402: }
1403: fsz = (size_t)(p - first);
1404: }
1405:
1406: second = p++;
1407: if ('\\' == *second) {
1408: esc = mandoc_escape(&p, NULL, NULL);
1409: if (ESCAPE_ERROR == esc) {
1410: mandoc_msg
1411: (MANDOCERR_BADESCAPE, r->parse,
1412: ln, (int)(p - *bufp), NULL);
1413: return(ROFF_IGN);
1414: }
1415: ssz = (size_t)(p - second);
1.165 kristaps 1416: } else if ('\0' == *second) {
1.164 kristaps 1417: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse,
1418: ln, (int)(p - *bufp), NULL);
1419: second = " ";
1.165 kristaps 1420: p--;
1.164 kristaps 1421: }
1422:
1.167 kristaps 1423: if (fsz > 1) {
1424: roff_setstrn(&r->xmbtab, first,
1425: fsz, second, ssz, 0);
1426: continue;
1427: }
1428:
1429: if (NULL == r->xtab)
1430: r->xtab = mandoc_calloc
1431: (128, sizeof(struct roffstr));
1432:
1433: free(r->xtab[(int)*first].p);
1434: r->xtab[(int)*first].p = mandoc_strndup(second, ssz);
1435: r->xtab[(int)*first].sz = ssz;
1.164 kristaps 1436: }
1437:
1438: return(ROFF_IGN);
1439: }
1440:
1441: /* ARGSUSED */
1442: static enum rofferr
1.105 kristaps 1443: roff_so(ROFF_ARGS)
1444: {
1445: char *name;
1446:
1.128 kristaps 1447: mandoc_msg(MANDOCERR_SO, r->parse, ln, ppos, NULL);
1.105 kristaps 1448:
1449: /*
1450: * Handle `so'. Be EXTREMELY careful, as we shouldn't be
1451: * opening anything that's not in our cwd or anything beneath
1452: * it. Thus, explicitly disallow traversing up the file-system
1453: * or using absolute paths.
1454: */
1455:
1456: name = *bufp + pos;
1457: if ('/' == *name || strstr(name, "../") || strstr(name, "/..")) {
1.128 kristaps 1458: mandoc_msg(MANDOCERR_SOPATH, r->parse, ln, pos, NULL);
1.105 kristaps 1459: return(ROFF_ERR);
1460: }
1461:
1462: *offs = pos;
1463: return(ROFF_SO);
1464: }
1.92 schwarze 1465:
1.106 kristaps 1466: /* ARGSUSED */
1467: static enum rofferr
1468: roff_userdef(ROFF_ARGS)
1.99 kristaps 1469: {
1.106 kristaps 1470: const char *arg[9];
1471: char *cp, *n1, *n2;
1.119 schwarze 1472: int i;
1.106 kristaps 1473:
1474: /*
1475: * Collect pointers to macro argument strings
1476: * and null-terminate them.
1477: */
1478: cp = *bufp + pos;
1.119 schwarze 1479: for (i = 0; i < 9; i++)
1.120 schwarze 1480: arg[i] = '\0' == *cp ? "" :
1.136 kristaps 1481: mandoc_getarg(r->parse, &cp, ln, &pos);
1.99 kristaps 1482:
1.106 kristaps 1483: /*
1484: * Expand macro arguments.
1.99 kristaps 1485: */
1.106 kristaps 1486: *szp = 0;
1487: n1 = cp = mandoc_strdup(r->current_string);
1488: while (NULL != (cp = strstr(cp, "\\$"))) {
1489: i = cp[2] - '1';
1490: if (0 > i || 8 < i) {
1491: /* Not an argument invocation. */
1492: cp += 2;
1493: continue;
1494: }
1495:
1496: *szp = strlen(n1) - 3 + strlen(arg[i]) + 1;
1497: n2 = mandoc_malloc(*szp);
1498:
1499: strlcpy(n2, n1, (size_t)(cp - n1 + 1));
1500: strlcat(n2, arg[i], *szp);
1501: strlcat(n2, cp + 3, *szp);
1502:
1503: cp = n2 + (cp - n1);
1504: free(n1);
1505: n1 = n2;
1.99 kristaps 1506: }
1507:
1.106 kristaps 1508: /*
1509: * Replace the macro invocation
1510: * by the expanded macro.
1511: */
1512: free(*bufp);
1513: *bufp = n1;
1514: if (0 == *szp)
1515: *szp = strlen(*bufp) + 1;
1516:
1517: return(*szp > 1 && '\n' == (*bufp)[(int)*szp - 2] ?
1518: ROFF_REPARSE : ROFF_APPEND);
1.99 kristaps 1519: }
1.121 schwarze 1520:
1521: static char *
1522: roff_getname(struct roff *r, char **cpp, int ln, int pos)
1523: {
1524: char *name, *cp;
1525:
1526: name = *cpp;
1527: if ('\0' == *name)
1528: return(name);
1529:
1530: /* Read until end of name. */
1531: for (cp = name; '\0' != *cp && ' ' != *cp; cp++) {
1532: if ('\\' != *cp)
1533: continue;
1534: cp++;
1535: if ('\\' == *cp)
1536: continue;
1.128 kristaps 1537: mandoc_msg(MANDOCERR_NAMESC, r->parse, ln, pos, NULL);
1.121 schwarze 1538: *cp = '\0';
1539: name = cp;
1540: }
1541:
1542: /* Nil-terminate name. */
1543: if ('\0' != *cp)
1544: *(cp++) = '\0';
1545:
1546: /* Read past spaces. */
1547: while (' ' == *cp)
1548: cp++;
1549:
1550: *cpp = cp;
1551: return(name);
1552: }
1553:
1.106 kristaps 1554: /*
1555: * Store *string into the user-defined string called *name.
1556: * In multiline mode, append to an existing entry and append '\n';
1557: * else replace the existing entry, if there is one.
1558: * To clear an existing entry, call with (*r, *name, NULL, 0).
1559: */
1.94 kristaps 1560: static void
1.106 kristaps 1561: roff_setstr(struct roff *r, const char *name, const char *string,
1562: int multiline)
1.92 schwarze 1563: {
1.164 kristaps 1564:
1565: roff_setstrn(&r->strtab, name, strlen(name), string,
1566: string ? strlen(string) : 0, multiline);
1567: }
1568:
1569: static void
1.166 kristaps 1570: roff_setstrn(struct roffkv **r, const char *name, size_t namesz,
1.164 kristaps 1571: const char *string, size_t stringsz, int multiline)
1572: {
1.166 kristaps 1573: struct roffkv *n;
1.164 kristaps 1574: char *c;
1575: int i;
1576: size_t oldch, newch;
1.92 schwarze 1577:
1.106 kristaps 1578: /* Search for an existing string with the same name. */
1.164 kristaps 1579: n = *r;
1580:
1.166 kristaps 1581: while (n && strcmp(name, n->key.p))
1.92 schwarze 1582: n = n->next;
1.94 kristaps 1583:
1584: if (NULL == n) {
1.106 kristaps 1585: /* Create a new string table entry. */
1.166 kristaps 1586: n = mandoc_malloc(sizeof(struct roffkv));
1587: n->key.p = mandoc_strndup(name, namesz);
1588: n->key.sz = namesz;
1589: n->val.p = NULL;
1590: n->val.sz = 0;
1.164 kristaps 1591: n->next = *r;
1592: *r = n;
1.106 kristaps 1593: } else if (0 == multiline) {
1594: /* In multiline mode, append; else replace. */
1.166 kristaps 1595: free(n->val.p);
1596: n->val.p = NULL;
1597: n->val.sz = 0;
1.106 kristaps 1598: }
1599:
1600: if (NULL == string)
1601: return;
1602:
1603: /*
1604: * One additional byte for the '\n' in multiline mode,
1605: * and one for the terminating '\0'.
1606: */
1.164 kristaps 1607: newch = stringsz + (multiline ? 2u : 1u);
1608:
1.166 kristaps 1609: if (NULL == n->val.p) {
1610: n->val.p = mandoc_malloc(newch);
1611: *n->val.p = '\0';
1.106 kristaps 1612: oldch = 0;
1613: } else {
1.166 kristaps 1614: oldch = n->val.sz;
1615: n->val.p = mandoc_realloc(n->val.p, oldch + newch);
1.106 kristaps 1616: }
1617:
1618: /* Skip existing content in the destination buffer. */
1.166 kristaps 1619: c = n->val.p + (int)oldch;
1.106 kristaps 1620:
1621: /* Append new content to the destination buffer. */
1.164 kristaps 1622: i = 0;
1623: while (i < (int)stringsz) {
1.106 kristaps 1624: /*
1625: * Rudimentary roff copy mode:
1626: * Handle escaped backslashes.
1627: */
1.164 kristaps 1628: if ('\\' == string[i] && '\\' == string[i + 1])
1629: i++;
1630: *c++ = string[i++];
1.106 kristaps 1631: }
1.94 kristaps 1632:
1.106 kristaps 1633: /* Append terminating bytes. */
1634: if (multiline)
1635: *c++ = '\n';
1.163 kristaps 1636:
1.106 kristaps 1637: *c = '\0';
1.166 kristaps 1638: n->val.sz = (int)(c - n->val.p);
1.92 schwarze 1639: }
1640:
1.94 kristaps 1641: static const char *
1642: roff_getstrn(const struct roff *r, const char *name, size_t len)
1.92 schwarze 1643: {
1.166 kristaps 1644: const struct roffkv *n;
1.92 schwarze 1645:
1.164 kristaps 1646: for (n = r->strtab; n; n = n->next)
1.166 kristaps 1647: if (0 == strncmp(name, n->key.p, len) &&
1648: '\0' == n->key.p[(int)len])
1649: return(n->val.p);
1.94 kristaps 1650:
1.157 kristaps 1651: return(NULL);
1.92 schwarze 1652: }
1653:
1.94 kristaps 1654: static void
1.167 kristaps 1655: roff_freestr(struct roffkv *r)
1.92 schwarze 1656: {
1.166 kristaps 1657: struct roffkv *n, *nn;
1.92 schwarze 1658:
1.167 kristaps 1659: for (n = r; n; n = nn) {
1.166 kristaps 1660: free(n->key.p);
1661: free(n->val.p);
1.92 schwarze 1662: nn = n->next;
1663: free(n);
1664: }
1.114 kristaps 1665: }
1666:
1667: const struct tbl_span *
1668: roff_span(const struct roff *r)
1669: {
1670:
1671: return(r->tbl ? tbl_span(r->tbl) : NULL);
1.125 kristaps 1672: }
1673:
1674: const struct eqn *
1675: roff_eqn(const struct roff *r)
1676: {
1677:
1678: return(r->last_eqn ? &r->last_eqn->eqn : NULL);
1.164 kristaps 1679: }
1680:
1681: /*
1682: * Duplicate an input string, making the appropriate character
1683: * conversations (as stipulated by `tr') along the way.
1684: * Returns a heap-allocated string with all the replacements made.
1685: */
1686: char *
1687: roff_strdup(const struct roff *r, const char *p)
1688: {
1.166 kristaps 1689: const struct roffkv *cp;
1.164 kristaps 1690: char *res;
1691: const char *pp;
1692: size_t ssz, sz;
1693: enum mandoc_esc esc;
1694:
1.167 kristaps 1695: if (NULL == r->xmbtab && NULL == r->xtab)
1.164 kristaps 1696: return(mandoc_strdup(p));
1697: else if ('\0' == *p)
1698: return(mandoc_strdup(""));
1699:
1700: /*
1701: * Step through each character looking for term matches
1702: * (remember that a `tr' can be invoked with an escape, which is
1703: * a glyph but the escape is multi-character).
1704: * We only do this if the character hash has been initialised
1705: * and the string is >0 length.
1706: */
1707:
1708: res = NULL;
1709: ssz = 0;
1710:
1711: while ('\0' != *p) {
1.167 kristaps 1712: if ('\\' != *p && r->xtab && r->xtab[(int)*p].p) {
1713: sz = r->xtab[(int)*p].sz;
1714: res = mandoc_realloc(res, ssz + sz + 1);
1715: memcpy(res + ssz, r->xtab[(int)*p].p, sz);
1716: ssz += sz;
1717: p++;
1718: continue;
1719: } else if ('\\' != *p) {
1720: res = mandoc_realloc(res, ssz + 2);
1721: res[ssz++] = *p++;
1722: continue;
1723: }
1724:
1.164 kristaps 1725: /* Search for term matches. */
1.167 kristaps 1726: for (cp = r->xmbtab; cp; cp = cp->next)
1.166 kristaps 1727: if (0 == strncmp(p, cp->key.p, cp->key.sz))
1.164 kristaps 1728: break;
1729:
1730: if (NULL != cp) {
1731: /*
1732: * A match has been found.
1733: * Append the match to the array and move
1734: * forward by its keysize.
1735: */
1.166 kristaps 1736: res = mandoc_realloc
1737: (res, ssz + cp->val.sz + 1);
1738: memcpy(res + ssz, cp->val.p, cp->val.sz);
1739: ssz += cp->val.sz;
1740: p += (int)cp->key.sz;
1.164 kristaps 1741: continue;
1742: }
1743:
1.167 kristaps 1744: /*
1745: * Handle escapes carefully: we need to copy
1746: * over just the escape itself, or else we might
1747: * do replacements within the escape itself.
1748: * Make sure to pass along the bogus string.
1749: */
1750: pp = p++;
1751: esc = mandoc_escape(&p, NULL, NULL);
1752: if (ESCAPE_ERROR == esc) {
1753: sz = strlen(pp);
1.164 kristaps 1754: res = mandoc_realloc(res, ssz + sz + 1);
1755: memcpy(res + ssz, pp, sz);
1.167 kristaps 1756: break;
1.164 kristaps 1757: }
1.167 kristaps 1758: /*
1759: * We bail out on bad escapes.
1760: * No need to warn: we already did so when
1761: * roff_res() was called.
1762: */
1763: sz = (int)(p - pp);
1764: res = mandoc_realloc(res, ssz + sz + 1);
1765: memcpy(res + ssz, pp, sz);
1766: ssz += sz;
1.164 kristaps 1767: }
1768:
1769: res[(int)ssz] = '\0';
1770: return(res);
1.74 kristaps 1771: }
CVSweb