Annotation of src/usr.bin/mandoc/roff.c, Revision 1.46
1.46 ! schwarze 1: /* $Id: roff.c,v 1.45 2011/10/24 21:38:56 schwarze Exp $ */
1.1 schwarze 2: /*
1.27 schwarze 3: * Copyright (c) 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
1.46 ! schwarze 4: * Copyright (c) 2010, 2011, 2012 Ingo Schwarze <schwarze@openbsd.org>
1.1 schwarze 5: *
6: * Permission to use, copy, modify, and distribute this software for any
7: * purpose with or without fee is hereby granted, provided that the above
8: * copyright notice and this permission notice appear in all copies.
9: *
1.16 schwarze 10: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.1 schwarze 11: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.16 schwarze 12: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.1 schwarze 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.
17: */
18: #include <assert.h>
1.3 schwarze 19: #include <ctype.h>
1.1 schwarze 20: #include <stdlib.h>
21: #include <string.h>
22:
23: #include "mandoc.h"
1.27 schwarze 24: #include "libroff.h"
1.8 schwarze 25: #include "libmandoc.h"
1.1 schwarze 26:
1.37 schwarze 27: /* Maximum number of nested if-else conditionals. */
1.2 schwarze 28: #define RSTACK_MAX 128
29:
1.43 schwarze 30: /* Maximum number of string expansions per line, to break infinite loops. */
31: #define EXPAND_LIMIT 1000
32:
1.1 schwarze 33: enum rofft {
1.20 schwarze 34: ROFF_ad,
1.2 schwarze 35: ROFF_am,
36: ROFF_ami,
37: ROFF_am1,
1.1 schwarze 38: ROFF_de,
39: ROFF_dei,
1.2 schwarze 40: ROFF_de1,
41: ROFF_ds,
42: ROFF_el,
1.20 schwarze 43: ROFF_hy,
1.2 schwarze 44: ROFF_ie,
45: ROFF_if,
1.1 schwarze 46: ROFF_ig,
1.30 schwarze 47: ROFF_it,
1.20 schwarze 48: ROFF_ne,
49: ROFF_nh,
1.14 schwarze 50: ROFF_nr,
1.31 schwarze 51: ROFF_ns,
52: ROFF_ps,
1.2 schwarze 53: ROFF_rm,
1.14 schwarze 54: ROFF_so,
1.31 schwarze 55: ROFF_ta,
1.2 schwarze 56: ROFF_tr,
1.27 schwarze 57: ROFF_TS,
58: ROFF_TE,
59: ROFF_T_,
1.32 schwarze 60: ROFF_EQ,
61: ROFF_EN,
1.2 schwarze 62: ROFF_cblock,
1.37 schwarze 63: ROFF_ccond,
1.16 schwarze 64: ROFF_USERDEF,
1.1 schwarze 65: ROFF_MAX
66: };
67:
1.2 schwarze 68: enum roffrule {
69: ROFFRULE_ALLOW,
70: ROFFRULE_DENY
71: };
72:
1.41 schwarze 73: /*
74: * A single register entity. If "set" is zero, the value of the
75: * register should be the default one, which is per-register.
76: * Registers are assumed to be unsigned ints for now.
77: */
78: struct reg {
1.42 schwarze 79: int set; /* whether set or not */
80: unsigned int u; /* unsigned integer */
1.41 schwarze 81: };
82:
1.42 schwarze 83: /*
84: * An incredibly-simple string buffer.
85: */
1.8 schwarze 86: struct roffstr {
1.42 schwarze 87: char *p; /* nil-terminated buffer */
88: size_t sz; /* saved strlen(p) */
89: };
90:
91: /*
92: * A key-value roffstr pair as part of a singly-linked list.
93: */
94: struct roffkv {
95: struct roffstr key;
96: struct roffstr val;
97: struct roffkv *next; /* next in list */
1.8 schwarze 98: };
99:
1.1 schwarze 100: struct roff {
1.35 schwarze 101: struct mparse *parse; /* parse point */
1.1 schwarze 102: struct roffnode *last; /* leaf of stack */
1.2 schwarze 103: enum roffrule rstack[RSTACK_MAX]; /* stack of !`ie' rules */
104: int rstackpos; /* position in rstack */
1.41 schwarze 105: struct reg regs[REG__MAX];
1.42 schwarze 106: struct roffkv *strtab; /* user-defined strings & macros */
107: struct roffkv *xmbtab; /* multi-byte trans table (`tr') */
108: struct roffstr *xtab; /* single-byte trans table (`tr') */
1.16 schwarze 109: const char *current_string; /* value of last called user macro */
1.27 schwarze 110: struct tbl_node *first_tbl; /* first table parsed */
111: struct tbl_node *last_tbl; /* last table parsed */
112: struct tbl_node *tbl; /* current table being parsed */
1.32 schwarze 113: struct eqn_node *last_eqn; /* last equation parsed */
114: struct eqn_node *first_eqn; /* first equation parsed */
115: struct eqn_node *eqn; /* current equation being parsed */
1.1 schwarze 116: };
117:
118: struct roffnode {
119: enum rofft tok; /* type of node */
120: struct roffnode *parent; /* up one in stack */
121: int line; /* parse line */
122: int col; /* parse col */
1.16 schwarze 123: char *name; /* node name, e.g. macro name */
1.2 schwarze 124: char *end; /* end-rules: custom token */
125: int endspan; /* end-rules: next-line or infty */
126: enum roffrule rule; /* current evaluation rule */
1.1 schwarze 127: };
128:
129: #define ROFF_ARGS struct roff *r, /* parse ctx */ \
130: enum rofft tok, /* tok of macro */ \
131: char **bufp, /* input buffer */ \
132: size_t *szp, /* size of input buffer */ \
133: int ln, /* parse line */ \
1.2 schwarze 134: int ppos, /* original pos in buffer */ \
135: int pos, /* current pos in buffer */ \
136: int *offs /* reset offset of buffer data */
1.1 schwarze 137:
138: typedef enum rofferr (*roffproc)(ROFF_ARGS);
139:
140: struct roffmac {
141: const char *name; /* macro name */
1.2 schwarze 142: roffproc proc; /* process new macro */
143: roffproc text; /* process as child text of macro */
144: roffproc sub; /* process as child of macro */
145: int flags;
146: #define ROFFMAC_STRUCT (1 << 0) /* always interpret */
1.3 schwarze 147: struct roffmac *next;
1.1 schwarze 148: };
149:
1.37 schwarze 150: struct predef {
151: const char *name; /* predefined input name */
152: const char *str; /* replacement symbol */
153: };
154:
155: #define PREDEF(__name, __str) \
156: { (__name), (__str) },
157:
1.42 schwarze 158: static enum rofft roffhash_find(const char *, size_t);
159: static void roffhash_init(void);
160: static void roffnode_cleanscope(struct roff *);
161: static void roffnode_pop(struct roff *);
162: static void roffnode_push(struct roff *, enum rofft,
163: const char *, int, int);
1.2 schwarze 164: static enum rofferr roff_block(ROFF_ARGS);
165: static enum rofferr roff_block_text(ROFF_ARGS);
166: static enum rofferr roff_block_sub(ROFF_ARGS);
167: static enum rofferr roff_cblock(ROFF_ARGS);
168: static enum rofferr roff_ccond(ROFF_ARGS);
169: static enum rofferr roff_cond(ROFF_ARGS);
170: static enum rofferr roff_cond_text(ROFF_ARGS);
171: static enum rofferr roff_cond_sub(ROFF_ARGS);
1.7 schwarze 172: static enum rofferr roff_ds(ROFF_ARGS);
1.8 schwarze 173: static enum roffrule roff_evalcond(const char *, int *);
1.42 schwarze 174: static void roff_free1(struct roff *);
175: static void roff_freestr(struct roffkv *);
1.28 schwarze 176: static char *roff_getname(struct roff *, char **, int, int);
1.8 schwarze 177: static const char *roff_getstrn(const struct roff *,
178: const char *, size_t);
1.21 schwarze 179: static enum rofferr roff_line_ignore(ROFF_ARGS);
1.6 schwarze 180: static enum rofferr roff_nr(ROFF_ARGS);
1.41 schwarze 181: static void roff_openeqn(struct roff *, const char *,
182: int, int, const char *);
1.42 schwarze 183: static enum rofft roff_parse(struct roff *, const char *, int *);
184: static enum rofferr roff_parsetext(char *);
1.45 schwarze 185: static enum rofferr roff_res(struct roff *,
1.37 schwarze 186: char **, size_t *, int, int);
1.29 schwarze 187: static enum rofferr roff_rm(ROFF_ARGS);
1.8 schwarze 188: static void roff_setstr(struct roff *,
1.16 schwarze 189: const char *, const char *, int);
1.42 schwarze 190: static void roff_setstrn(struct roffkv **, const char *,
191: size_t, const char *, size_t, int);
1.14 schwarze 192: static enum rofferr roff_so(ROFF_ARGS);
1.42 schwarze 193: static enum rofferr roff_tr(ROFF_ARGS);
1.27 schwarze 194: static enum rofferr roff_TE(ROFF_ARGS);
195: static enum rofferr roff_TS(ROFF_ARGS);
1.32 schwarze 196: static enum rofferr roff_EQ(ROFF_ARGS);
197: static enum rofferr roff_EN(ROFF_ARGS);
1.27 schwarze 198: static enum rofferr roff_T_(ROFF_ARGS);
1.16 schwarze 199: static enum rofferr roff_userdef(ROFF_ARGS);
1.1 schwarze 200:
1.42 schwarze 201: /* See roffhash_find() */
1.3 schwarze 202:
203: #define ASCII_HI 126
204: #define ASCII_LO 33
205: #define HASHWIDTH (ASCII_HI - ASCII_LO + 1)
206:
207: static struct roffmac *hash[HASHWIDTH];
208:
209: static struct roffmac roffs[ROFF_MAX] = {
1.21 schwarze 210: { "ad", roff_line_ignore, NULL, NULL, 0, NULL },
1.3 schwarze 211: { "am", roff_block, roff_block_text, roff_block_sub, 0, NULL },
212: { "ami", roff_block, roff_block_text, roff_block_sub, 0, NULL },
213: { "am1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
214: { "de", roff_block, roff_block_text, roff_block_sub, 0, NULL },
215: { "dei", roff_block, roff_block_text, roff_block_sub, 0, NULL },
216: { "de1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.7 schwarze 217: { "ds", roff_ds, NULL, NULL, 0, NULL },
1.3 schwarze 218: { "el", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
1.21 schwarze 219: { "hy", roff_line_ignore, NULL, NULL, 0, NULL },
1.3 schwarze 220: { "ie", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
221: { "if", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
222: { "ig", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.30 schwarze 223: { "it", roff_line_ignore, NULL, NULL, 0, NULL },
1.21 schwarze 224: { "ne", roff_line_ignore, NULL, NULL, 0, NULL },
225: { "nh", roff_line_ignore, NULL, NULL, 0, NULL },
1.14 schwarze 226: { "nr", roff_nr, NULL, NULL, 0, NULL },
1.31 schwarze 227: { "ns", roff_line_ignore, NULL, NULL, 0, NULL },
228: { "ps", roff_line_ignore, NULL, NULL, 0, NULL },
1.29 schwarze 229: { "rm", roff_rm, NULL, NULL, 0, NULL },
1.14 schwarze 230: { "so", roff_so, NULL, NULL, 0, NULL },
1.31 schwarze 231: { "ta", roff_line_ignore, NULL, NULL, 0, NULL },
1.42 schwarze 232: { "tr", roff_tr, NULL, NULL, 0, NULL },
1.27 schwarze 233: { "TS", roff_TS, NULL, NULL, 0, NULL },
234: { "TE", roff_TE, NULL, NULL, 0, NULL },
235: { "T&", roff_T_, NULL, NULL, 0, NULL },
1.32 schwarze 236: { "EQ", roff_EQ, NULL, NULL, 0, NULL },
237: { "EN", roff_EN, NULL, NULL, 0, NULL },
1.3 schwarze 238: { ".", roff_cblock, NULL, NULL, 0, NULL },
239: { "\\}", roff_ccond, NULL, NULL, 0, NULL },
1.16 schwarze 240: { NULL, roff_userdef, NULL, NULL, 0, NULL },
1.1 schwarze 241: };
242:
1.37 schwarze 243: /* Array of injected predefined strings. */
244: #define PREDEFS_MAX 38
245: static const struct predef predefs[PREDEFS_MAX] = {
246: #include "predefs.in"
247: };
248:
1.42 schwarze 249: /* See roffhash_find() */
1.3 schwarze 250: #define ROFF_HASH(p) (p[0] - ASCII_LO)
251:
252: static void
1.42 schwarze 253: roffhash_init(void)
1.3 schwarze 254: {
255: struct roffmac *n;
256: int buc, i;
257:
1.16 schwarze 258: for (i = 0; i < (int)ROFF_USERDEF; i++) {
1.3 schwarze 259: assert(roffs[i].name[0] >= ASCII_LO);
260: assert(roffs[i].name[0] <= ASCII_HI);
261:
262: buc = ROFF_HASH(roffs[i].name);
263:
264: if (NULL != (n = hash[buc])) {
265: for ( ; n->next; n = n->next)
266: /* Do nothing. */ ;
267: n->next = &roffs[i];
268: } else
269: hash[buc] = &roffs[i];
270: }
271: }
272:
1.1 schwarze 273: /*
274: * Look up a roff token by its name. Returns ROFF_MAX if no macro by
275: * the nil-terminated string name could be found.
276: */
277: static enum rofft
1.42 schwarze 278: roffhash_find(const char *p, size_t s)
1.1 schwarze 279: {
1.3 schwarze 280: int buc;
281: struct roffmac *n;
1.1 schwarze 282:
1.3 schwarze 283: /*
284: * libroff has an extremely simple hashtable, for the time
285: * being, which simply keys on the first character, which must
286: * be printable, then walks a chain. It works well enough until
287: * optimised.
288: */
289:
290: if (p[0] < ASCII_LO || p[0] > ASCII_HI)
291: return(ROFF_MAX);
292:
293: buc = ROFF_HASH(p);
294:
295: if (NULL == (n = hash[buc]))
296: return(ROFF_MAX);
297: for ( ; n; n = n->next)
1.16 schwarze 298: if (0 == strncmp(n->name, p, s) && '\0' == n->name[(int)s])
1.3 schwarze 299: return((enum rofft)(n - roffs));
1.1 schwarze 300:
301: return(ROFF_MAX);
302: }
303:
304:
305: /*
306: * Pop the current node off of the stack of roff instructions currently
307: * pending.
308: */
309: static void
310: roffnode_pop(struct roff *r)
311: {
312: struct roffnode *p;
313:
1.2 schwarze 314: assert(r->last);
315: p = r->last;
316:
317: r->last = r->last->parent;
1.16 schwarze 318: free(p->name);
319: free(p->end);
1.1 schwarze 320: free(p);
321: }
322:
323:
324: /*
325: * Push a roff node onto the instruction stack. This must later be
326: * removed with roffnode_pop().
327: */
1.11 schwarze 328: static void
1.16 schwarze 329: roffnode_push(struct roff *r, enum rofft tok, const char *name,
330: int line, int col)
1.1 schwarze 331: {
332: struct roffnode *p;
333:
1.11 schwarze 334: p = mandoc_calloc(1, sizeof(struct roffnode));
1.1 schwarze 335: p->tok = tok;
1.16 schwarze 336: if (name)
337: p->name = mandoc_strdup(name);
1.1 schwarze 338: p->parent = r->last;
339: p->line = line;
340: p->col = col;
1.2 schwarze 341: p->rule = p->parent ? p->parent->rule : ROFFRULE_DENY;
1.1 schwarze 342:
343: r->last = p;
344: }
345:
346:
347: static void
348: roff_free1(struct roff *r)
349: {
1.27 schwarze 350: struct tbl_node *t;
1.32 schwarze 351: struct eqn_node *e;
1.42 schwarze 352: int i;
1.27 schwarze 353:
1.32 schwarze 354: while (NULL != (t = r->first_tbl)) {
1.27 schwarze 355: r->first_tbl = t->next;
356: tbl_free(t);
357: }
358:
359: r->first_tbl = r->last_tbl = r->tbl = NULL;
1.1 schwarze 360:
1.32 schwarze 361: while (NULL != (e = r->first_eqn)) {
362: r->first_eqn = e->next;
363: eqn_free(e);
364: }
365:
366: r->first_eqn = r->last_eqn = r->eqn = NULL;
367:
1.1 schwarze 368: while (r->last)
369: roffnode_pop(r);
1.27 schwarze 370:
1.42 schwarze 371: roff_freestr(r->strtab);
372: roff_freestr(r->xmbtab);
373:
374: r->strtab = r->xmbtab = NULL;
375:
376: if (r->xtab)
377: for (i = 0; i < 128; i++)
378: free(r->xtab[i].p);
379:
380: free(r->xtab);
381: r->xtab = NULL;
1.1 schwarze 382: }
383:
384: void
385: roff_reset(struct roff *r)
386: {
1.38 schwarze 387: int i;
1.1 schwarze 388:
389: roff_free1(r);
1.38 schwarze 390:
1.41 schwarze 391: memset(&r->regs, 0, sizeof(struct reg) * REG__MAX);
392:
1.38 schwarze 393: for (i = 0; i < PREDEFS_MAX; i++)
394: roff_setstr(r, predefs[i].name, predefs[i].str, 0);
1.1 schwarze 395: }
396:
397:
398: void
399: roff_free(struct roff *r)
400: {
401:
402: roff_free1(r);
403: free(r);
404: }
405:
406:
407: struct roff *
1.41 schwarze 408: roff_alloc(struct mparse *parse)
1.1 schwarze 409: {
410: struct roff *r;
1.37 schwarze 411: int i;
1.1 schwarze 412:
1.11 schwarze 413: r = mandoc_calloc(1, sizeof(struct roff));
1.35 schwarze 414: r->parse = parse;
1.2 schwarze 415: r->rstackpos = -1;
1.3 schwarze 416:
1.42 schwarze 417: roffhash_init();
1.37 schwarze 418:
419: for (i = 0; i < PREDEFS_MAX; i++)
420: roff_setstr(r, predefs[i].name, predefs[i].str, 0);
421:
1.1 schwarze 422: return(r);
423: }
424:
1.8 schwarze 425: /*
426: * Pre-filter each and every line for reserved words (one beginning with
427: * `\*', e.g., `\*(ab'). These must be handled before the actual line
428: * is processed.
1.42 schwarze 429: * This also checks the syntax of regular escapes.
1.8 schwarze 430: */
1.45 schwarze 431: static enum rofferr
1.37 schwarze 432: roff_res(struct roff *r, char **bufp, size_t *szp, int ln, int pos)
1.8 schwarze 433: {
1.42 schwarze 434: enum mandoc_esc esc;
1.23 schwarze 435: const char *stesc; /* start of an escape sequence ('\\') */
436: const char *stnam; /* start of the name, after "[(*" */
437: const char *cp; /* end of the name, e.g. before ']' */
438: const char *res; /* the string to be substituted */
1.43 schwarze 439: int i, maxl, expand_count;
1.8 schwarze 440: size_t nsz;
441: char *n;
442:
1.43 schwarze 443: expand_count = 0;
444:
1.42 schwarze 445: again:
1.24 schwarze 446: cp = *bufp + pos;
447: while (NULL != (cp = strchr(cp, '\\'))) {
448: stesc = cp++;
1.23 schwarze 449:
450: /*
451: * The second character must be an asterisk.
452: * If it isn't, skip it anyway: It is escaped,
453: * so it can't start another escape sequence.
454: */
455:
1.24 schwarze 456: if ('\0' == *cp)
1.45 schwarze 457: return(ROFF_CONT);
1.42 schwarze 458:
459: if ('*' != *cp) {
460: res = cp;
461: esc = mandoc_escape(&cp, NULL, NULL);
462: if (ESCAPE_ERROR != esc)
463: continue;
464: cp = res;
465: mandoc_msg
466: (MANDOCERR_BADESCAPE, r->parse,
467: ln, (int)(stesc - *bufp), NULL);
1.45 schwarze 468: return(ROFF_CONT);
1.42 schwarze 469: }
470:
471: cp++;
1.23 schwarze 472:
473: /*
474: * The third character decides the length
475: * of the name of the string.
476: * Save a pointer to the name.
477: */
478:
1.24 schwarze 479: switch (*cp) {
480: case ('\0'):
1.45 schwarze 481: return(ROFF_CONT);
1.8 schwarze 482: case ('('):
483: cp++;
484: maxl = 2;
485: break;
486: case ('['):
487: cp++;
488: maxl = 0;
489: break;
490: default:
491: maxl = 1;
492: break;
493: }
1.23 schwarze 494: stnam = cp;
1.8 schwarze 495:
1.23 schwarze 496: /* Advance to the end of the name. */
1.8 schwarze 497:
498: for (i = 0; 0 == maxl || i < maxl; i++, cp++) {
1.42 schwarze 499: if ('\0' == *cp) {
500: mandoc_msg
501: (MANDOCERR_BADESCAPE,
502: r->parse, ln,
503: (int)(stesc - *bufp), NULL);
1.45 schwarze 504: return(ROFF_CONT);
1.42 schwarze 505: }
1.8 schwarze 506: if (0 == maxl && ']' == *cp)
507: break;
508: }
509:
1.23 schwarze 510: /*
511: * Retrieve the replacement string; if it is
512: * undefined, resume searching for escapes.
513: */
514:
515: res = roff_getstrn(r, stnam, (size_t)i);
1.8 schwarze 516:
517: if (NULL == res) {
1.42 schwarze 518: mandoc_msg
519: (MANDOCERR_BADESCAPE, r->parse,
520: ln, (int)(stesc - *bufp), NULL);
1.37 schwarze 521: res = "";
1.8 schwarze 522: }
523:
1.23 schwarze 524: /* Replace the escape sequence by the string. */
525:
1.42 schwarze 526: pos = stesc - *bufp;
527:
1.8 schwarze 528: nsz = *szp + strlen(res) + 1;
529: n = mandoc_malloc(nsz);
530:
1.23 schwarze 531: strlcpy(n, *bufp, (size_t)(stesc - *bufp + 1));
1.8 schwarze 532: strlcat(n, res, nsz);
533: strlcat(n, cp + (maxl ? 0 : 1), nsz);
534:
535: free(*bufp);
536:
537: *bufp = n;
538: *szp = nsz;
1.43 schwarze 539:
540: if (EXPAND_LIMIT >= ++expand_count)
541: goto again;
542:
543: /* Just leave the string unexpanded. */
544: mandoc_msg(MANDOCERR_ROFFLOOP, r->parse, ln, pos, NULL);
1.45 schwarze 545: return(ROFF_IGN);
1.42 schwarze 546: }
1.45 schwarze 547: return(ROFF_CONT);
1.42 schwarze 548: }
549:
550: /*
551: * Process text streams: convert all breakable hyphens into ASCII_HYPH.
552: */
553: static enum rofferr
554: roff_parsetext(char *p)
555: {
556: size_t sz;
557: const char *start;
558: enum mandoc_esc esc;
559:
560: start = p;
561:
562: while ('\0' != *p) {
563: sz = strcspn(p, "-\\");
564: p += sz;
565:
566: if ('\0' == *p)
567: break;
568:
569: if ('\\' == *p) {
570: /* Skip over escapes. */
571: p++;
572: esc = mandoc_escape
573: ((const char **)&p, NULL, NULL);
574: if (ESCAPE_ERROR == esc)
575: break;
576: continue;
577: } else if (p == start) {
578: p++;
579: continue;
580: }
581:
1.44 schwarze 582: if (isalpha((unsigned char)p[-1]) &&
583: isalpha((unsigned char)p[1]))
1.42 schwarze 584: *p = ASCII_HYPH;
585: p++;
1.8 schwarze 586: }
587:
1.42 schwarze 588: return(ROFF_CONT);
1.8 schwarze 589: }
590:
1.1 schwarze 591: enum rofferr
1.6 schwarze 592: roff_parseln(struct roff *r, int ln, char **bufp,
593: size_t *szp, int pos, int *offs)
1.1 schwarze 594: {
595: enum rofft t;
1.27 schwarze 596: enum rofferr e;
1.35 schwarze 597: int ppos, ctl;
1.1 schwarze 598:
1.2 schwarze 599: /*
1.8 schwarze 600: * Run the reserved-word filter only if we have some reserved
601: * words to fill in.
602: */
603:
1.45 schwarze 604: e = roff_res(r, bufp, szp, ln, pos);
605: if (ROFF_IGN == e)
606: return(e);
607: assert(ROFF_CONT == e);
1.8 schwarze 608:
1.35 schwarze 609: ppos = pos;
610: ctl = mandoc_getcontrol(*bufp, &pos);
611:
1.8 schwarze 612: /*
1.2 schwarze 613: * First, if a scope is open and we're not a macro, pass the
614: * text through the macro's filter. If a scope isn't open and
615: * we're not a macro, just let it through.
1.32 schwarze 616: * Finally, if there's an equation scope open, divert it into it
617: * no matter our state.
1.2 schwarze 618: */
619:
1.35 schwarze 620: if (r->last && ! ctl) {
1.2 schwarze 621: t = r->last->tok;
622: assert(roffs[t].text);
1.27 schwarze 623: e = (*roffs[t].text)
624: (r, t, bufp, szp, ln, pos, pos, offs);
625: assert(ROFF_IGN == e || ROFF_CONT == e);
1.32 schwarze 626: if (ROFF_CONT != e)
627: return(e);
628: if (r->eqn)
1.41 schwarze 629: return(eqn_read(&r->eqn, ln, *bufp, pos, offs));
1.32 schwarze 630: if (r->tbl)
1.35 schwarze 631: return(tbl_read(r->tbl, ln, *bufp, pos));
1.42 schwarze 632: return(roff_parsetext(*bufp + pos));
1.35 schwarze 633: } else if ( ! ctl) {
1.32 schwarze 634: if (r->eqn)
1.41 schwarze 635: return(eqn_read(&r->eqn, ln, *bufp, pos, offs));
1.27 schwarze 636: if (r->tbl)
1.35 schwarze 637: return(tbl_read(r->tbl, ln, *bufp, pos));
1.42 schwarze 638: return(roff_parsetext(*bufp + pos));
1.32 schwarze 639: } else if (r->eqn)
1.41 schwarze 640: return(eqn_read(&r->eqn, ln, *bufp, ppos, offs));
1.2 schwarze 641:
642: /*
643: * If a scope is open, go to the child handler for that macro,
644: * as it may want to preprocess before doing anything with it.
1.32 schwarze 645: * Don't do so if an equation is open.
1.2 schwarze 646: */
647:
648: if (r->last) {
1.1 schwarze 649: t = r->last->tok;
650: assert(roffs[t].sub);
1.2 schwarze 651: return((*roffs[t].sub)
1.8 schwarze 652: (r, t, bufp, szp,
1.35 schwarze 653: ln, ppos, pos, offs));
1.2 schwarze 654: }
655:
656: /*
657: * Lastly, as we've no scope open, try to look up and execute
658: * the new macro. If no macro is found, simply return and let
659: * the compilers handle it.
660: */
661:
1.16 schwarze 662: if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos)))
1.1 schwarze 663: return(ROFF_CONT);
664:
1.2 schwarze 665: assert(roffs[t].proc);
666: return((*roffs[t].proc)
1.8 schwarze 667: (r, t, bufp, szp,
668: ln, ppos, pos, offs));
1.2 schwarze 669: }
670:
1.1 schwarze 671:
1.27 schwarze 672: void
1.2 schwarze 673: roff_endparse(struct roff *r)
674: {
1.1 schwarze 675:
1.27 schwarze 676: if (r->last)
1.35 schwarze 677: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.27 schwarze 678: r->last->line, r->last->col, NULL);
679:
1.32 schwarze 680: if (r->eqn) {
1.35 schwarze 681: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.41 schwarze 682: r->eqn->eqn.ln, r->eqn->eqn.pos, NULL);
683: eqn_end(&r->eqn);
1.32 schwarze 684: }
685:
1.27 schwarze 686: if (r->tbl) {
1.35 schwarze 687: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.27 schwarze 688: r->tbl->line, r->tbl->pos, NULL);
1.41 schwarze 689: tbl_end(&r->tbl);
1.27 schwarze 690: }
1.1 schwarze 691: }
692:
693: /*
694: * Parse a roff node's type from the input buffer. This must be in the
695: * form of ".foo xxx" in the usual way.
696: */
697: static enum rofft
1.16 schwarze 698: roff_parse(struct roff *r, const char *buf, int *pos)
1.1 schwarze 699: {
1.16 schwarze 700: const char *mac;
701: size_t maclen;
1.1 schwarze 702: enum rofft t;
703:
1.39 schwarze 704: if ('\0' == buf[*pos] || '"' == buf[*pos] ||
705: '\t' == buf[*pos] || ' ' == buf[*pos])
1.1 schwarze 706: return(ROFF_MAX);
707:
1.39 schwarze 708: /*
709: * We stop the macro parse at an escape, tab, space, or nil.
710: * However, `\}' is also a valid macro, so make sure we don't
711: * clobber it by seeing the `\' as the end of token.
712: */
713:
1.16 schwarze 714: mac = buf + *pos;
1.39 schwarze 715: maclen = strcspn(mac + 1, " \\\t\0") + 1;
1.1 schwarze 716:
1.16 schwarze 717: t = (r->current_string = roff_getstrn(r, mac, maclen))
1.42 schwarze 718: ? ROFF_USERDEF : roffhash_find(mac, maclen);
1.1 schwarze 719:
1.34 schwarze 720: *pos += (int)maclen;
1.35 schwarze 721:
1.1 schwarze 722: while (buf[*pos] && ' ' == buf[*pos])
723: (*pos)++;
724:
725: return(t);
726: }
727:
728: /* ARGSUSED */
729: static enum rofferr
1.2 schwarze 730: roff_cblock(ROFF_ARGS)
1.1 schwarze 731: {
732:
1.2 schwarze 733: /*
734: * A block-close `..' should only be invoked as a child of an
735: * ignore macro, otherwise raise a warning and just ignore it.
736: */
737:
738: if (NULL == r->last) {
1.35 schwarze 739: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.2 schwarze 740: return(ROFF_IGN);
741: }
1.1 schwarze 742:
1.2 schwarze 743: switch (r->last->tok) {
744: case (ROFF_am):
745: /* FALLTHROUGH */
746: case (ROFF_ami):
747: /* FALLTHROUGH */
748: case (ROFF_am1):
749: /* FALLTHROUGH */
750: case (ROFF_de):
1.23 schwarze 751: /* ROFF_de1 is remapped to ROFF_de in roff_block(). */
1.2 schwarze 752: /* FALLTHROUGH */
753: case (ROFF_dei):
754: /* FALLTHROUGH */
755: case (ROFF_ig):
756: break;
757: default:
1.35 schwarze 758: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.1 schwarze 759: return(ROFF_IGN);
1.2 schwarze 760: }
761:
762: if ((*bufp)[pos])
1.35 schwarze 763: mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.2 schwarze 764:
765: roffnode_pop(r);
766: roffnode_cleanscope(r);
767: return(ROFF_IGN);
768:
769: }
1.1 schwarze 770:
771:
1.2 schwarze 772: static void
773: roffnode_cleanscope(struct roff *r)
774: {
1.1 schwarze 775:
1.2 schwarze 776: while (r->last) {
1.46 ! schwarze 777: if (--r->last->endspan != 0)
1.2 schwarze 778: break;
779: roffnode_pop(r);
780: }
781: }
1.1 schwarze 782:
783:
1.2 schwarze 784: /* ARGSUSED */
785: static enum rofferr
786: roff_ccond(ROFF_ARGS)
787: {
1.1 schwarze 788:
1.2 schwarze 789: if (NULL == r->last) {
1.35 schwarze 790: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.1 schwarze 791: return(ROFF_IGN);
1.2 schwarze 792: }
1.1 schwarze 793:
1.2 schwarze 794: switch (r->last->tok) {
795: case (ROFF_el):
796: /* FALLTHROUGH */
797: case (ROFF_ie):
798: /* FALLTHROUGH */
799: case (ROFF_if):
800: break;
801: default:
1.35 schwarze 802: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.2 schwarze 803: return(ROFF_IGN);
804: }
1.1 schwarze 805:
1.2 schwarze 806: if (r->last->endspan > -1) {
1.35 schwarze 807: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.1 schwarze 808: return(ROFF_IGN);
1.2 schwarze 809: }
810:
811: if ((*bufp)[pos])
1.35 schwarze 812: mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.1 schwarze 813:
1.2 schwarze 814: roffnode_pop(r);
815: roffnode_cleanscope(r);
1.1 schwarze 816: return(ROFF_IGN);
817: }
818:
819:
820: /* ARGSUSED */
821: static enum rofferr
1.2 schwarze 822: roff_block(ROFF_ARGS)
1.1 schwarze 823: {
1.2 schwarze 824: int sv;
825: size_t sz;
1.16 schwarze 826: char *name;
827:
828: name = NULL;
1.2 schwarze 829:
1.16 schwarze 830: if (ROFF_ig != tok) {
831: if ('\0' == (*bufp)[pos]) {
1.35 schwarze 832: mandoc_msg(MANDOCERR_NOARGS, r->parse, ln, ppos, NULL);
1.16 schwarze 833: return(ROFF_IGN);
834: }
1.22 schwarze 835:
836: /*
837: * Re-write `de1', since we don't really care about
838: * groff's strange compatibility mode, into `de'.
839: */
840:
1.18 schwarze 841: if (ROFF_de1 == tok)
842: tok = ROFF_de;
1.16 schwarze 843: if (ROFF_de == tok)
844: name = *bufp + pos;
1.21 schwarze 845: else
1.35 schwarze 846: mandoc_msg(MANDOCERR_REQUEST, r->parse, ln, ppos,
1.21 schwarze 847: roffs[tok].name);
1.22 schwarze 848:
1.33 schwarze 849: while ((*bufp)[pos] && ! isspace((unsigned char)(*bufp)[pos]))
1.2 schwarze 850: pos++;
1.22 schwarze 851:
1.33 schwarze 852: while (isspace((unsigned char)(*bufp)[pos]))
1.16 schwarze 853: (*bufp)[pos++] = '\0';
1.2 schwarze 854: }
855:
1.16 schwarze 856: roffnode_push(r, tok, name, ln, ppos);
857:
858: /*
859: * At the beginning of a `de' macro, clear the existing string
860: * with the same name, if there is one. New content will be
861: * added from roff_block_text() in multiline mode.
862: */
1.22 schwarze 863:
1.16 schwarze 864: if (ROFF_de == tok)
1.19 schwarze 865: roff_setstr(r, name, "", 0);
1.2 schwarze 866:
867: if ('\0' == (*bufp)[pos])
868: return(ROFF_IGN);
1.1 schwarze 869:
1.22 schwarze 870: /* If present, process the custom end-of-line marker. */
871:
1.2 schwarze 872: sv = pos;
1.33 schwarze 873: while ((*bufp)[pos] && ! isspace((unsigned char)(*bufp)[pos]))
1.2 schwarze 874: pos++;
875:
876: /*
877: * Note: groff does NOT like escape characters in the input.
878: * Instead of detecting this, we're just going to let it fly and
879: * to hell with it.
880: */
881:
882: assert(pos > sv);
883: sz = (size_t)(pos - sv);
884:
885: if (1 == sz && '.' == (*bufp)[sv])
886: return(ROFF_IGN);
887:
1.11 schwarze 888: r->last->end = mandoc_malloc(sz + 1);
1.2 schwarze 889:
890: memcpy(r->last->end, *bufp + sv, sz);
891: r->last->end[(int)sz] = '\0';
892:
893: if ((*bufp)[pos])
1.35 schwarze 894: mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.1 schwarze 895:
896: return(ROFF_IGN);
897: }
898:
899:
900: /* ARGSUSED */
901: static enum rofferr
1.2 schwarze 902: roff_block_sub(ROFF_ARGS)
1.1 schwarze 903: {
1.2 schwarze 904: enum rofft t;
905: int i, j;
906:
907: /*
908: * First check whether a custom macro exists at this level. If
909: * it does, then check against it. This is some of groff's
910: * stranger behaviours. If we encountered a custom end-scope
911: * tag and that tag also happens to be a "real" macro, then we
912: * need to try interpreting it again as a real macro. If it's
913: * not, then return ignore. Else continue.
914: */
915:
916: if (r->last->end) {
1.35 schwarze 917: for (i = pos, j = 0; r->last->end[j]; j++, i++)
1.2 schwarze 918: if ((*bufp)[i] != r->last->end[j])
919: break;
1.1 schwarze 920:
1.2 schwarze 921: if ('\0' == r->last->end[j] &&
922: ('\0' == (*bufp)[i] ||
923: ' ' == (*bufp)[i] ||
924: '\t' == (*bufp)[i])) {
925: roffnode_pop(r);
926: roffnode_cleanscope(r);
1.1 schwarze 927:
1.35 schwarze 928: while (' ' == (*bufp)[i] || '\t' == (*bufp)[i])
929: i++;
930:
931: pos = i;
1.16 schwarze 932: if (ROFF_MAX != roff_parse(r, *bufp, &pos))
1.2 schwarze 933: return(ROFF_RERUN);
934: return(ROFF_IGN);
935: }
1.1 schwarze 936: }
937:
1.2 schwarze 938: /*
939: * If we have no custom end-query or lookup failed, then try
940: * pulling it out of the hashtable.
941: */
1.1 schwarze 942:
1.36 schwarze 943: t = roff_parse(r, *bufp, &pos);
1.1 schwarze 944:
1.16 schwarze 945: /*
946: * Macros other than block-end are only significant
947: * in `de' blocks; elsewhere, simply throw them away.
948: */
949: if (ROFF_cblock != t) {
950: if (ROFF_de == tok)
951: roff_setstr(r, r->last->name, *bufp + ppos, 1);
1.1 schwarze 952: return(ROFF_IGN);
1.16 schwarze 953: }
1.1 schwarze 954:
1.2 schwarze 955: assert(roffs[t].proc);
1.6 schwarze 956: return((*roffs[t].proc)(r, t, bufp, szp,
957: ln, ppos, pos, offs));
1.2 schwarze 958: }
959:
960:
961: /* ARGSUSED */
962: static enum rofferr
963: roff_block_text(ROFF_ARGS)
964: {
965:
1.16 schwarze 966: if (ROFF_de == tok)
967: roff_setstr(r, r->last->name, *bufp + pos, 1);
968:
1.2 schwarze 969: return(ROFF_IGN);
970: }
971:
972:
973: /* ARGSUSED */
974: static enum rofferr
975: roff_cond_sub(ROFF_ARGS)
976: {
977: enum rofft t;
978: enum roffrule rr;
1.37 schwarze 979: char *ep;
1.2 schwarze 980:
981: rr = r->last->rule;
1.37 schwarze 982: roffnode_cleanscope(r);
1.2 schwarze 983:
1.37 schwarze 984: /*
985: * If the macro is unknown, first check if it contains a closing
986: * delimiter `\}'. If it does, close out our scope and return
987: * the currently-scoped rule (ignore or continue). Else, drop
988: * into the currently-scoped rule.
1.5 schwarze 989: */
990:
1.16 schwarze 991: if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos))) {
1.37 schwarze 992: ep = &(*bufp)[pos];
993: for ( ; NULL != (ep = strchr(ep, '\\')); ep++) {
994: ep++;
995: if ('}' != *ep)
996: continue;
1.39 schwarze 997:
998: /*
999: * Make the \} go away.
1000: * This is a little haphazard, as it's not quite
1001: * clear how nroff does this.
1002: * If we're at the end of line, then just chop
1003: * off the \} and resize the buffer.
1004: * If we aren't, then conver it to spaces.
1005: */
1006:
1007: if ('\0' == *(ep + 1)) {
1008: *--ep = '\0';
1009: *szp -= 2;
1010: } else
1011: *(ep - 1) = *ep = ' ';
1012:
1.37 schwarze 1013: roff_ccond(r, ROFF_ccond, bufp, szp,
1014: ln, pos, pos + 2, offs);
1015: break;
1016: }
1.2 schwarze 1017: return(ROFFRULE_DENY == rr ? ROFF_IGN : ROFF_CONT);
1.12 schwarze 1018: }
1.2 schwarze 1019:
1020: /*
1021: * A denied conditional must evaluate its children if and only
1022: * if they're either structurally required (such as loops and
1023: * conditionals) or a closing macro.
1024: */
1.37 schwarze 1025:
1.2 schwarze 1026: if (ROFFRULE_DENY == rr)
1027: if ( ! (ROFFMAC_STRUCT & roffs[t].flags))
1028: if (ROFF_ccond != t)
1029: return(ROFF_IGN);
1030:
1031: assert(roffs[t].proc);
1.6 schwarze 1032: return((*roffs[t].proc)(r, t, bufp, szp,
1033: ln, ppos, pos, offs));
1.2 schwarze 1034: }
1035:
1036: /* ARGSUSED */
1037: static enum rofferr
1038: roff_cond_text(ROFF_ARGS)
1039: {
1.37 schwarze 1040: char *ep;
1.2 schwarze 1041: enum roffrule rr;
1042:
1043: rr = r->last->rule;
1.37 schwarze 1044: roffnode_cleanscope(r);
1.1 schwarze 1045:
1.37 schwarze 1046: ep = &(*bufp)[pos];
1047: for ( ; NULL != (ep = strchr(ep, '\\')); ep++) {
1048: ep++;
1049: if ('}' != *ep)
1050: continue;
1051: *ep = '&';
1052: roff_ccond(r, ROFF_ccond, bufp, szp,
1053: ln, pos, pos + 2, offs);
1.2 schwarze 1054: }
1055: return(ROFFRULE_DENY == rr ? ROFF_IGN : ROFF_CONT);
1056: }
1057:
1.5 schwarze 1058: static enum roffrule
1059: roff_evalcond(const char *v, int *pos)
1060: {
1061:
1062: switch (v[*pos]) {
1063: case ('n'):
1064: (*pos)++;
1065: return(ROFFRULE_ALLOW);
1066: case ('e'):
1067: /* FALLTHROUGH */
1068: case ('o'):
1069: /* FALLTHROUGH */
1070: case ('t'):
1071: (*pos)++;
1072: return(ROFFRULE_DENY);
1073: default:
1074: break;
1075: }
1076:
1077: while (v[*pos] && ' ' != v[*pos])
1078: (*pos)++;
1079: return(ROFFRULE_DENY);
1080: }
1081:
1.2 schwarze 1082: /* ARGSUSED */
1083: static enum rofferr
1.21 schwarze 1084: roff_line_ignore(ROFF_ARGS)
1.6 schwarze 1085: {
1.30 schwarze 1086:
1087: if (ROFF_it == tok)
1.35 schwarze 1088: mandoc_msg(MANDOCERR_REQUEST, r->parse, ln, ppos, "it");
1.6 schwarze 1089:
1.21 schwarze 1090: return(ROFF_IGN);
1091: }
1092:
1093: /* ARGSUSED */
1094: static enum rofferr
1.2 schwarze 1095: roff_cond(ROFF_ARGS)
1096: {
1.46 ! schwarze 1097:
! 1098: roffnode_push(r, tok, NULL, ln, ppos);
1.2 schwarze 1099:
1.35 schwarze 1100: /*
1101: * An `.el' has no conditional body: it will consume the value
1102: * of the current rstack entry set in prior `ie' calls or
1103: * defaults to DENY.
1104: *
1105: * If we're not an `el', however, then evaluate the conditional.
1106: */
1.1 schwarze 1107:
1.46 ! schwarze 1108: r->last->rule = ROFF_el == tok ?
1.35 schwarze 1109: (r->rstackpos < 0 ?
1110: ROFFRULE_DENY : r->rstack[r->rstackpos--]) :
1111: roff_evalcond(*bufp, &pos);
1.2 schwarze 1112:
1.35 schwarze 1113: /*
1114: * An if-else will put the NEGATION of the current evaluated
1115: * conditional into the stack of rules.
1116: */
1117:
1.2 schwarze 1118: if (ROFF_ie == tok) {
1.35 schwarze 1119: if (r->rstackpos == RSTACK_MAX - 1) {
1120: mandoc_msg(MANDOCERR_MEM,
1121: r->parse, ln, ppos, NULL);
1122: return(ROFF_ERR);
1123: }
1124: r->rstack[++r->rstackpos] =
1125: ROFFRULE_DENY == r->last->rule ?
1126: ROFFRULE_ALLOW : ROFFRULE_DENY;
1.2 schwarze 1127: }
1.5 schwarze 1128:
1129: /* If the parent has false as its rule, then so do we. */
1130:
1.2 schwarze 1131: if (r->last->parent && ROFFRULE_DENY == r->last->parent->rule)
1132: r->last->rule = ROFFRULE_DENY;
1.5 schwarze 1133:
1134: /*
1.46 ! schwarze 1135: * Determine scope.
! 1136: * If there is nothing on the line after the conditional,
! 1137: * not even whitespace, use next-line scope.
1.5 schwarze 1138: */
1.2 schwarze 1139:
1.46 ! schwarze 1140: if ('\0' == (*bufp)[pos]) {
! 1141: r->last->endspan = 2;
! 1142: goto out;
! 1143: }
! 1144:
! 1145: while (' ' == (*bufp)[pos])
! 1146: pos++;
! 1147:
! 1148: /* An opening brace requests multiline scope. */
1.2 schwarze 1149:
1150: if ('\\' == (*bufp)[pos] && '{' == (*bufp)[pos + 1]) {
1151: r->last->endspan = -1;
1152: pos += 2;
1.46 ! schwarze 1153: goto out;
1.2 schwarze 1154: }
1155:
1156: /*
1.46 ! schwarze 1157: * Anything else following the conditional causes
! 1158: * single-line scope. Warn if the scope contains
! 1159: * nothing but trailing whitespace.
1.2 schwarze 1160: */
1161:
1162: if ('\0' == (*bufp)[pos])
1.46 ! schwarze 1163: mandoc_msg(MANDOCERR_NOARGS, r->parse, ln, ppos, NULL);
1.2 schwarze 1164:
1.46 ! schwarze 1165: r->last->endspan = 1;
1.1 schwarze 1166:
1.46 ! schwarze 1167: out:
1.2 schwarze 1168: *offs = pos;
1169: return(ROFF_RERUN);
1.1 schwarze 1170: }
1171:
1172:
1.2 schwarze 1173: /* ARGSUSED */
1174: static enum rofferr
1.7 schwarze 1175: roff_ds(ROFF_ARGS)
1176: {
1.10 schwarze 1177: char *name, *string;
1178:
1179: /*
1180: * A symbol is named by the first word following the macro
1181: * invocation up to a space. Its value is anything after the
1182: * name's trailing whitespace and optional double-quote. Thus,
1183: *
1184: * [.ds foo "bar " ]
1185: *
1186: * will have `bar " ' as its value.
1187: */
1.7 schwarze 1188:
1.28 schwarze 1189: string = *bufp + pos;
1190: name = roff_getname(r, &string, ln, pos);
1.7 schwarze 1191: if ('\0' == *name)
1192: return(ROFF_IGN);
1193:
1.28 schwarze 1194: /* Read past initial double-quote. */
1195: if ('"' == *string)
1.7 schwarze 1196: string++;
1197:
1.10 schwarze 1198: /* The rest is the value. */
1.16 schwarze 1199: roff_setstr(r, name, string, 0);
1.7 schwarze 1200: return(ROFF_IGN);
1201: }
1202:
1.41 schwarze 1203: int
1204: roff_regisset(const struct roff *r, enum regs reg)
1205: {
1206:
1207: return(r->regs[(int)reg].set);
1208: }
1209:
1210: unsigned int
1211: roff_regget(const struct roff *r, enum regs reg)
1212: {
1213:
1214: return(r->regs[(int)reg].u);
1215: }
1216:
1217: void
1218: roff_regunset(struct roff *r, enum regs reg)
1219: {
1220:
1221: r->regs[(int)reg].set = 0;
1222: }
1.7 schwarze 1223:
1224: /* ARGSUSED */
1225: static enum rofferr
1.6 schwarze 1226: roff_nr(ROFF_ARGS)
1.1 schwarze 1227: {
1.28 schwarze 1228: const char *key;
1229: char *val;
1.37 schwarze 1230: int iv;
1.6 schwarze 1231:
1.28 schwarze 1232: val = *bufp + pos;
1233: key = roff_getname(r, &val, ln, pos);
1.6 schwarze 1234:
1235: if (0 == strcmp(key, "nS")) {
1.41 schwarze 1236: r->regs[(int)REG_nS].set = 1;
1237: if ((iv = mandoc_strntoi(val, strlen(val), 10)) >= 0)
1238: r->regs[(int)REG_nS].u = (unsigned)iv;
1.37 schwarze 1239: else
1.41 schwarze 1240: r->regs[(int)REG_nS].u = 0u;
1.6 schwarze 1241: }
1.1 schwarze 1242:
1.29 schwarze 1243: return(ROFF_IGN);
1244: }
1245:
1246: /* ARGSUSED */
1247: static enum rofferr
1248: roff_rm(ROFF_ARGS)
1249: {
1250: const char *name;
1251: char *cp;
1252:
1253: cp = *bufp + pos;
1254: while ('\0' != *cp) {
1.34 schwarze 1255: name = roff_getname(r, &cp, ln, (int)(cp - *bufp));
1.29 schwarze 1256: if ('\0' != *name)
1257: roff_setstr(r, name, NULL, 0);
1258: }
1.2 schwarze 1259: return(ROFF_IGN);
1.14 schwarze 1260: }
1261:
1262: /* ARGSUSED */
1263: static enum rofferr
1.27 schwarze 1264: roff_TE(ROFF_ARGS)
1265: {
1266:
1267: if (NULL == r->tbl)
1.35 schwarze 1268: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.27 schwarze 1269: else
1.41 schwarze 1270: tbl_end(&r->tbl);
1.27 schwarze 1271:
1272: return(ROFF_IGN);
1273: }
1274:
1275: /* ARGSUSED */
1276: static enum rofferr
1277: roff_T_(ROFF_ARGS)
1278: {
1279:
1280: if (NULL == r->tbl)
1.35 schwarze 1281: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.27 schwarze 1282: else
1283: tbl_restart(ppos, ln, r->tbl);
1284:
1285: return(ROFF_IGN);
1286: }
1287:
1.41 schwarze 1288: #if 0
1289: static int
1290: roff_closeeqn(struct roff *r)
1291: {
1292:
1293: return(r->eqn && ROFF_EQN == eqn_end(&r->eqn) ? 1 : 0);
1294: }
1295: #endif
1296:
1297: static void
1298: roff_openeqn(struct roff *r, const char *name, int line,
1299: int offs, const char *buf)
1.32 schwarze 1300: {
1.41 schwarze 1301: struct eqn_node *e;
1302: int poff;
1.32 schwarze 1303:
1304: assert(NULL == r->eqn);
1.41 schwarze 1305: e = eqn_alloc(name, offs, line, r->parse);
1.32 schwarze 1306:
1307: if (r->last_eqn)
1308: r->last_eqn->next = e;
1309: else
1310: r->first_eqn = r->last_eqn = e;
1311:
1312: r->eqn = r->last_eqn = e;
1.41 schwarze 1313:
1314: if (buf) {
1315: poff = 0;
1316: eqn_read(&r->eqn, line, buf, offs, &poff);
1317: }
1318: }
1319:
1320: /* ARGSUSED */
1321: static enum rofferr
1322: roff_EQ(ROFF_ARGS)
1323: {
1324:
1325: roff_openeqn(r, *bufp + pos, ln, ppos, NULL);
1.32 schwarze 1326: return(ROFF_IGN);
1327: }
1328:
1329: /* ARGSUSED */
1330: static enum rofferr
1331: roff_EN(ROFF_ARGS)
1332: {
1333:
1.35 schwarze 1334: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.32 schwarze 1335: return(ROFF_IGN);
1336: }
1337:
1338: /* ARGSUSED */
1339: static enum rofferr
1.27 schwarze 1340: roff_TS(ROFF_ARGS)
1341: {
1342: struct tbl_node *t;
1343:
1344: if (r->tbl) {
1.35 schwarze 1345: mandoc_msg(MANDOCERR_SCOPEBROKEN, r->parse, ln, ppos, NULL);
1.41 schwarze 1346: tbl_end(&r->tbl);
1.27 schwarze 1347: }
1348:
1.35 schwarze 1349: t = tbl_alloc(ppos, ln, r->parse);
1.27 schwarze 1350:
1351: if (r->last_tbl)
1352: r->last_tbl->next = t;
1353: else
1354: r->first_tbl = r->last_tbl = t;
1355:
1356: r->tbl = r->last_tbl = t;
1357: return(ROFF_IGN);
1358: }
1359:
1360: /* ARGSUSED */
1361: static enum rofferr
1.42 schwarze 1362: roff_tr(ROFF_ARGS)
1363: {
1364: const char *p, *first, *second;
1365: size_t fsz, ssz;
1366: enum mandoc_esc esc;
1367:
1368: p = *bufp + pos;
1369:
1370: if ('\0' == *p) {
1371: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
1372: return(ROFF_IGN);
1373: }
1374:
1375: while ('\0' != *p) {
1376: fsz = ssz = 1;
1377:
1378: first = p++;
1379: if ('\\' == *first) {
1380: esc = mandoc_escape(&p, NULL, NULL);
1381: if (ESCAPE_ERROR == esc) {
1382: mandoc_msg
1383: (MANDOCERR_BADESCAPE, r->parse,
1384: ln, (int)(p - *bufp), NULL);
1385: return(ROFF_IGN);
1386: }
1387: fsz = (size_t)(p - first);
1388: }
1389:
1390: second = p++;
1391: if ('\\' == *second) {
1392: esc = mandoc_escape(&p, NULL, NULL);
1393: if (ESCAPE_ERROR == esc) {
1394: mandoc_msg
1395: (MANDOCERR_BADESCAPE, r->parse,
1396: ln, (int)(p - *bufp), NULL);
1397: return(ROFF_IGN);
1398: }
1399: ssz = (size_t)(p - second);
1400: } else if ('\0' == *second) {
1401: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse,
1402: ln, (int)(p - *bufp), NULL);
1403: second = " ";
1404: p--;
1405: }
1406:
1407: if (fsz > 1) {
1408: roff_setstrn(&r->xmbtab, first,
1409: fsz, second, ssz, 0);
1410: continue;
1411: }
1412:
1413: if (NULL == r->xtab)
1414: r->xtab = mandoc_calloc
1415: (128, sizeof(struct roffstr));
1416:
1417: free(r->xtab[(int)*first].p);
1418: r->xtab[(int)*first].p = mandoc_strndup(second, ssz);
1419: r->xtab[(int)*first].sz = ssz;
1420: }
1421:
1422: return(ROFF_IGN);
1423: }
1424:
1425: /* ARGSUSED */
1426: static enum rofferr
1.14 schwarze 1427: roff_so(ROFF_ARGS)
1428: {
1429: char *name;
1.15 schwarze 1430:
1.35 schwarze 1431: mandoc_msg(MANDOCERR_SO, r->parse, ln, ppos, NULL);
1.14 schwarze 1432:
1.22 schwarze 1433: /*
1434: * Handle `so'. Be EXTREMELY careful, as we shouldn't be
1435: * opening anything that's not in our cwd or anything beneath
1436: * it. Thus, explicitly disallow traversing up the file-system
1437: * or using absolute paths.
1438: */
1439:
1.14 schwarze 1440: name = *bufp + pos;
1441: if ('/' == *name || strstr(name, "../") || strstr(name, "/..")) {
1.35 schwarze 1442: mandoc_msg(MANDOCERR_SOPATH, r->parse, ln, pos, NULL);
1.14 schwarze 1443: return(ROFF_ERR);
1444: }
1445:
1446: *offs = pos;
1447: return(ROFF_SO);
1.7 schwarze 1448: }
1449:
1.16 schwarze 1450: /* ARGSUSED */
1451: static enum rofferr
1452: roff_userdef(ROFF_ARGS)
1.12 schwarze 1453: {
1.16 schwarze 1454: const char *arg[9];
1455: char *cp, *n1, *n2;
1.25 schwarze 1456: int i;
1.12 schwarze 1457:
1.16 schwarze 1458: /*
1459: * Collect pointers to macro argument strings
1460: * and null-terminate them.
1461: */
1462: cp = *bufp + pos;
1.25 schwarze 1463: for (i = 0; i < 9; i++)
1.26 schwarze 1464: arg[i] = '\0' == *cp ? "" :
1.35 schwarze 1465: mandoc_getarg(r->parse, &cp, ln, &pos);
1.16 schwarze 1466:
1467: /*
1468: * Expand macro arguments.
1.12 schwarze 1469: */
1.16 schwarze 1470: *szp = 0;
1471: n1 = cp = mandoc_strdup(r->current_string);
1472: while (NULL != (cp = strstr(cp, "\\$"))) {
1473: i = cp[2] - '1';
1474: if (0 > i || 8 < i) {
1475: /* Not an argument invocation. */
1476: cp += 2;
1477: continue;
1478: }
1479:
1480: *szp = strlen(n1) - 3 + strlen(arg[i]) + 1;
1481: n2 = mandoc_malloc(*szp);
1482:
1483: strlcpy(n2, n1, (size_t)(cp - n1 + 1));
1484: strlcat(n2, arg[i], *szp);
1485: strlcat(n2, cp + 3, *szp);
1486:
1487: cp = n2 + (cp - n1);
1488: free(n1);
1489: n1 = n2;
1.12 schwarze 1490: }
1491:
1.16 schwarze 1492: /*
1493: * Replace the macro invocation
1494: * by the expanded macro.
1495: */
1496: free(*bufp);
1497: *bufp = n1;
1498: if (0 == *szp)
1499: *szp = strlen(*bufp) + 1;
1500:
1.19 schwarze 1501: return(*szp > 1 && '\n' == (*bufp)[(int)*szp - 2] ?
1.16 schwarze 1502: ROFF_REPARSE : ROFF_APPEND);
1.12 schwarze 1503: }
1.28 schwarze 1504:
1505: static char *
1506: roff_getname(struct roff *r, char **cpp, int ln, int pos)
1507: {
1508: char *name, *cp;
1509:
1510: name = *cpp;
1511: if ('\0' == *name)
1512: return(name);
1513:
1514: /* Read until end of name. */
1515: for (cp = name; '\0' != *cp && ' ' != *cp; cp++) {
1516: if ('\\' != *cp)
1517: continue;
1518: cp++;
1519: if ('\\' == *cp)
1520: continue;
1.35 schwarze 1521: mandoc_msg(MANDOCERR_NAMESC, r->parse, ln, pos, NULL);
1.28 schwarze 1522: *cp = '\0';
1523: name = cp;
1524: }
1525:
1526: /* Nil-terminate name. */
1527: if ('\0' != *cp)
1528: *(cp++) = '\0';
1529:
1530: /* Read past spaces. */
1531: while (' ' == *cp)
1532: cp++;
1533:
1534: *cpp = cp;
1535: return(name);
1536: }
1537:
1.16 schwarze 1538: /*
1539: * Store *string into the user-defined string called *name.
1540: * In multiline mode, append to an existing entry and append '\n';
1541: * else replace the existing entry, if there is one.
1542: * To clear an existing entry, call with (*r, *name, NULL, 0).
1543: */
1.8 schwarze 1544: static void
1.16 schwarze 1545: roff_setstr(struct roff *r, const char *name, const char *string,
1546: int multiline)
1.7 schwarze 1547: {
1.42 schwarze 1548:
1549: roff_setstrn(&r->strtab, name, strlen(name), string,
1550: string ? strlen(string) : 0, multiline);
1551: }
1552:
1553: static void
1554: roff_setstrn(struct roffkv **r, const char *name, size_t namesz,
1555: const char *string, size_t stringsz, int multiline)
1556: {
1557: struct roffkv *n;
1558: char *c;
1559: int i;
1560: size_t oldch, newch;
1.7 schwarze 1561:
1.16 schwarze 1562: /* Search for an existing string with the same name. */
1.42 schwarze 1563: n = *r;
1564:
1565: while (n && strcmp(name, n->key.p))
1.7 schwarze 1566: n = n->next;
1.8 schwarze 1567:
1568: if (NULL == n) {
1.16 schwarze 1569: /* Create a new string table entry. */
1.42 schwarze 1570: n = mandoc_malloc(sizeof(struct roffkv));
1571: n->key.p = mandoc_strndup(name, namesz);
1572: n->key.sz = namesz;
1573: n->val.p = NULL;
1574: n->val.sz = 0;
1575: n->next = *r;
1576: *r = n;
1.16 schwarze 1577: } else if (0 == multiline) {
1578: /* In multiline mode, append; else replace. */
1.42 schwarze 1579: free(n->val.p);
1580: n->val.p = NULL;
1581: n->val.sz = 0;
1.16 schwarze 1582: }
1583:
1584: if (NULL == string)
1585: return;
1586:
1587: /*
1588: * One additional byte for the '\n' in multiline mode,
1589: * and one for the terminating '\0'.
1590: */
1.42 schwarze 1591: newch = stringsz + (multiline ? 2u : 1u);
1592:
1593: if (NULL == n->val.p) {
1594: n->val.p = mandoc_malloc(newch);
1595: *n->val.p = '\0';
1.16 schwarze 1596: oldch = 0;
1597: } else {
1.42 schwarze 1598: oldch = n->val.sz;
1599: n->val.p = mandoc_realloc(n->val.p, oldch + newch);
1.16 schwarze 1600: }
1601:
1602: /* Skip existing content in the destination buffer. */
1.42 schwarze 1603: c = n->val.p + (int)oldch;
1.16 schwarze 1604:
1605: /* Append new content to the destination buffer. */
1.42 schwarze 1606: i = 0;
1607: while (i < (int)stringsz) {
1.16 schwarze 1608: /*
1609: * Rudimentary roff copy mode:
1610: * Handle escaped backslashes.
1611: */
1.42 schwarze 1612: if ('\\' == string[i] && '\\' == string[i + 1])
1613: i++;
1614: *c++ = string[i++];
1.16 schwarze 1615: }
1.8 schwarze 1616:
1.16 schwarze 1617: /* Append terminating bytes. */
1618: if (multiline)
1619: *c++ = '\n';
1.42 schwarze 1620:
1.16 schwarze 1621: *c = '\0';
1.42 schwarze 1622: n->val.sz = (int)(c - n->val.p);
1.7 schwarze 1623: }
1624:
1.8 schwarze 1625: static const char *
1626: roff_getstrn(const struct roff *r, const char *name, size_t len)
1.7 schwarze 1627: {
1.42 schwarze 1628: const struct roffkv *n;
1.7 schwarze 1629:
1.42 schwarze 1630: for (n = r->strtab; n; n = n->next)
1631: if (0 == strncmp(name, n->key.p, len) &&
1632: '\0' == n->key.p[(int)len])
1633: return(n->val.p);
1.8 schwarze 1634:
1.42 schwarze 1635: return(NULL);
1.7 schwarze 1636: }
1637:
1.8 schwarze 1638: static void
1.42 schwarze 1639: roff_freestr(struct roffkv *r)
1.7 schwarze 1640: {
1.42 schwarze 1641: struct roffkv *n, *nn;
1.7 schwarze 1642:
1.42 schwarze 1643: for (n = r; n; n = nn) {
1644: free(n->key.p);
1645: free(n->val.p);
1.7 schwarze 1646: nn = n->next;
1647: free(n);
1648: }
1.27 schwarze 1649: }
1650:
1651: const struct tbl_span *
1652: roff_span(const struct roff *r)
1653: {
1654:
1655: return(r->tbl ? tbl_span(r->tbl) : NULL);
1.32 schwarze 1656: }
1657:
1658: const struct eqn *
1659: roff_eqn(const struct roff *r)
1660: {
1661:
1662: return(r->last_eqn ? &r->last_eqn->eqn : NULL);
1.42 schwarze 1663: }
1664:
1665: /*
1666: * Duplicate an input string, making the appropriate character
1667: * conversations (as stipulated by `tr') along the way.
1668: * Returns a heap-allocated string with all the replacements made.
1669: */
1670: char *
1671: roff_strdup(const struct roff *r, const char *p)
1672: {
1673: const struct roffkv *cp;
1674: char *res;
1675: const char *pp;
1676: size_t ssz, sz;
1677: enum mandoc_esc esc;
1678:
1679: if (NULL == r->xmbtab && NULL == r->xtab)
1680: return(mandoc_strdup(p));
1681: else if ('\0' == *p)
1682: return(mandoc_strdup(""));
1683:
1684: /*
1685: * Step through each character looking for term matches
1686: * (remember that a `tr' can be invoked with an escape, which is
1687: * a glyph but the escape is multi-character).
1688: * We only do this if the character hash has been initialised
1689: * and the string is >0 length.
1690: */
1691:
1692: res = NULL;
1693: ssz = 0;
1694:
1695: while ('\0' != *p) {
1696: if ('\\' != *p && r->xtab && r->xtab[(int)*p].p) {
1697: sz = r->xtab[(int)*p].sz;
1698: res = mandoc_realloc(res, ssz + sz + 1);
1699: memcpy(res + ssz, r->xtab[(int)*p].p, sz);
1700: ssz += sz;
1701: p++;
1702: continue;
1703: } else if ('\\' != *p) {
1704: res = mandoc_realloc(res, ssz + 2);
1705: res[ssz++] = *p++;
1706: continue;
1707: }
1708:
1709: /* Search for term matches. */
1710: for (cp = r->xmbtab; cp; cp = cp->next)
1711: if (0 == strncmp(p, cp->key.p, cp->key.sz))
1712: break;
1713:
1714: if (NULL != cp) {
1715: /*
1716: * A match has been found.
1717: * Append the match to the array and move
1718: * forward by its keysize.
1719: */
1720: res = mandoc_realloc
1721: (res, ssz + cp->val.sz + 1);
1722: memcpy(res + ssz, cp->val.p, cp->val.sz);
1723: ssz += cp->val.sz;
1724: p += (int)cp->key.sz;
1725: continue;
1726: }
1727:
1728: /*
1729: * Handle escapes carefully: we need to copy
1730: * over just the escape itself, or else we might
1731: * do replacements within the escape itself.
1732: * Make sure to pass along the bogus string.
1733: */
1734: pp = p++;
1735: esc = mandoc_escape(&p, NULL, NULL);
1736: if (ESCAPE_ERROR == esc) {
1737: sz = strlen(pp);
1738: res = mandoc_realloc(res, ssz + sz + 1);
1739: memcpy(res + ssz, pp, sz);
1740: break;
1741: }
1742: /*
1743: * We bail out on bad escapes.
1744: * No need to warn: we already did so when
1745: * roff_res() was called.
1746: */
1747: sz = (int)(p - pp);
1748: res = mandoc_realloc(res, ssz + sz + 1);
1749: memcpy(res + ssz, pp, sz);
1750: ssz += sz;
1751: }
1752:
1753: res[(int)ssz] = '\0';
1754: return(res);
1.1 schwarze 1755: }
![[BACK]](/icons/back.gif){kind=icon}
Return to roff.c CVS log ![[TXT]](/icons/text.gif){kind=icon}
![[DIR]](/icons/dir.gif){kind=icon}
Up to [local] / src / usr.bin / mandoc