Annotation of src/usr.bin/mandoc/roff.c, Revision 1.112
1.112 ! schwarze 1: /* $OpenBSD: roff.c,v 1.111 2014/11/19 01:20:18 schwarze Exp $ */
1.1 schwarze 2: /*
1.48 schwarze 3: * Copyright (c) 2010, 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
1.63 schwarze 4: * Copyright (c) 2010-2014 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: */
1.99 schwarze 18: #include <sys/types.h>
19:
1.1 schwarze 20: #include <assert.h>
1.3 schwarze 21: #include <ctype.h>
1.51 schwarze 22: #include <stdio.h>
1.1 schwarze 23: #include <stdlib.h>
24: #include <string.h>
25:
26: #include "mandoc.h"
1.74 schwarze 27: #include "mandoc_aux.h"
1.98 daniel 28: #include "libmandoc.h"
1.27 schwarze 29: #include "libroff.h"
1.1 schwarze 30:
1.37 schwarze 31: /* Maximum number of nested if-else conditionals. */
1.2 schwarze 32: #define RSTACK_MAX 128
33:
1.43 schwarze 34: /* Maximum number of string expansions per line, to break infinite loops. */
35: #define EXPAND_LIMIT 1000
36:
1.1 schwarze 37: enum rofft {
1.20 schwarze 38: ROFF_ad,
1.2 schwarze 39: ROFF_am,
40: ROFF_ami,
41: ROFF_am1,
1.66 schwarze 42: ROFF_as,
1.48 schwarze 43: ROFF_cc,
1.67 schwarze 44: ROFF_ce,
1.1 schwarze 45: ROFF_de,
46: ROFF_dei,
1.2 schwarze 47: ROFF_de1,
48: ROFF_ds,
49: ROFF_el,
1.58 schwarze 50: ROFF_fam,
1.59 schwarze 51: ROFF_hw,
1.20 schwarze 52: ROFF_hy,
1.2 schwarze 53: ROFF_ie,
54: ROFF_if,
1.1 schwarze 55: ROFF_ig,
1.30 schwarze 56: ROFF_it,
1.20 schwarze 57: ROFF_ne,
58: ROFF_nh,
1.14 schwarze 59: ROFF_nr,
1.31 schwarze 60: ROFF_ns,
1.101 schwarze 61: ROFF_pl,
1.31 schwarze 62: ROFF_ps,
1.2 schwarze 63: ROFF_rm,
1.76 schwarze 64: ROFF_rr,
1.14 schwarze 65: ROFF_so,
1.31 schwarze 66: ROFF_ta,
1.2 schwarze 67: ROFF_tr,
1.47 schwarze 68: ROFF_Dd,
69: ROFF_TH,
1.27 schwarze 70: ROFF_TS,
71: ROFF_TE,
72: ROFF_T_,
1.32 schwarze 73: ROFF_EQ,
74: ROFF_EN,
1.2 schwarze 75: ROFF_cblock,
1.16 schwarze 76: ROFF_USERDEF,
1.1 schwarze 77: ROFF_MAX
78: };
79:
1.41 schwarze 80: /*
1.42 schwarze 81: * An incredibly-simple string buffer.
82: */
1.8 schwarze 83: struct roffstr {
1.42 schwarze 84: char *p; /* nil-terminated buffer */
85: size_t sz; /* saved strlen(p) */
86: };
87:
88: /*
89: * A key-value roffstr pair as part of a singly-linked list.
90: */
91: struct roffkv {
92: struct roffstr key;
93: struct roffstr val;
94: struct roffkv *next; /* next in list */
1.8 schwarze 95: };
96:
1.52 schwarze 97: /*
98: * A single number register as part of a singly-linked list.
99: */
100: struct roffreg {
101: struct roffstr key;
1.53 schwarze 102: int val;
1.52 schwarze 103: struct roffreg *next;
104: };
105:
1.1 schwarze 106: struct roff {
1.35 schwarze 107: struct mparse *parse; /* parse point */
1.109 schwarze 108: const struct mchars *mchars; /* character table */
1.1 schwarze 109: struct roffnode *last; /* leaf of stack */
1.96 schwarze 110: int *rstack; /* stack of inverted `ie' values */
1.52 schwarze 111: struct roffreg *regtab; /* number registers */
1.42 schwarze 112: struct roffkv *strtab; /* user-defined strings & macros */
113: struct roffkv *xmbtab; /* multi-byte trans table (`tr') */
114: struct roffstr *xtab; /* single-byte trans table (`tr') */
1.16 schwarze 115: const char *current_string; /* value of last called user macro */
1.27 schwarze 116: struct tbl_node *first_tbl; /* first table parsed */
117: struct tbl_node *last_tbl; /* last table parsed */
118: struct tbl_node *tbl; /* current table being parsed */
1.32 schwarze 119: struct eqn_node *last_eqn; /* last equation parsed */
120: struct eqn_node *first_eqn; /* first equation parsed */
121: struct eqn_node *eqn; /* current equation being parsed */
1.102 schwarze 122: int eqn_inline; /* current equation is inline */
1.96 schwarze 123: int options; /* parse options */
124: int rstacksz; /* current size limit of rstack */
125: int rstackpos; /* position in rstack */
1.99 schwarze 126: int format; /* current file in mdoc or man format */
1.96 schwarze 127: char control; /* control character */
1.1 schwarze 128: };
129:
130: struct roffnode {
131: enum rofft tok; /* type of node */
132: struct roffnode *parent; /* up one in stack */
133: int line; /* parse line */
134: int col; /* parse col */
1.16 schwarze 135: char *name; /* node name, e.g. macro name */
1.2 schwarze 136: char *end; /* end-rules: custom token */
137: int endspan; /* end-rules: next-line or infty */
1.71 schwarze 138: int rule; /* current evaluation rule */
1.1 schwarze 139: };
140:
141: #define ROFF_ARGS struct roff *r, /* parse ctx */ \
142: enum rofft tok, /* tok of macro */ \
1.110 schwarze 143: struct buf *buf, /* input buffer */ \
1.1 schwarze 144: int ln, /* parse line */ \
1.2 schwarze 145: int ppos, /* original pos in buffer */ \
146: int pos, /* current pos in buffer */ \
147: int *offs /* reset offset of buffer data */
1.1 schwarze 148:
149: typedef enum rofferr (*roffproc)(ROFF_ARGS);
150:
151: struct roffmac {
152: const char *name; /* macro name */
1.2 schwarze 153: roffproc proc; /* process new macro */
154: roffproc text; /* process as child text of macro */
155: roffproc sub; /* process as child of macro */
156: int flags;
157: #define ROFFMAC_STRUCT (1 << 0) /* always interpret */
1.3 schwarze 158: struct roffmac *next;
1.1 schwarze 159: };
160:
1.37 schwarze 161: struct predef {
162: const char *name; /* predefined input name */
163: const char *str; /* replacement symbol */
164: };
165:
166: #define PREDEF(__name, __str) \
167: { (__name), (__str) },
168:
1.42 schwarze 169: static enum rofft roffhash_find(const char *, size_t);
170: static void roffhash_init(void);
171: static void roffnode_cleanscope(struct roff *);
172: static void roffnode_pop(struct roff *);
173: static void roffnode_push(struct roff *, enum rofft,
174: const char *, int, int);
1.2 schwarze 175: static enum rofferr roff_block(ROFF_ARGS);
176: static enum rofferr roff_block_text(ROFF_ARGS);
177: static enum rofferr roff_block_sub(ROFF_ARGS);
178: static enum rofferr roff_cblock(ROFF_ARGS);
1.48 schwarze 179: static enum rofferr roff_cc(ROFF_ARGS);
1.68 schwarze 180: static void roff_ccond(struct roff *, int, int);
1.2 schwarze 181: static enum rofferr roff_cond(ROFF_ARGS);
182: static enum rofferr roff_cond_text(ROFF_ARGS);
183: static enum rofferr roff_cond_sub(ROFF_ARGS);
1.7 schwarze 184: static enum rofferr roff_ds(ROFF_ARGS);
1.110 schwarze 185: static enum rofferr roff_eqndelim(struct roff *, struct buf *, int);
1.106 schwarze 186: static int roff_evalcond(struct roff *r, int,
187: const char *, int *);
188: static int roff_evalnum(struct roff *, int,
189: const char *, int *, int *, int);
190: static int roff_evalpar(struct roff *, int,
191: const char *, int *, int *);
1.71 schwarze 192: static int roff_evalstrcond(const char *, int *);
1.42 schwarze 193: static void roff_free1(struct roff *);
1.52 schwarze 194: static void roff_freereg(struct roffreg *);
1.42 schwarze 195: static void roff_freestr(struct roffkv *);
1.85 schwarze 196: static size_t roff_getname(struct roff *, char **, int, int);
1.56 schwarze 197: static int roff_getnum(const char *, int *, int *);
198: static int roff_getop(const char *, int *, char *);
1.53 schwarze 199: static int roff_getregn(const struct roff *,
200: const char *, size_t);
1.65 schwarze 201: static int roff_getregro(const char *name);
1.80 schwarze 202: static const char *roff_getstrn(const struct roff *,
1.8 schwarze 203: const char *, size_t);
1.51 schwarze 204: static enum rofferr roff_it(ROFF_ARGS);
1.21 schwarze 205: static enum rofferr roff_line_ignore(ROFF_ARGS);
1.6 schwarze 206: static enum rofferr roff_nr(ROFF_ARGS);
1.87 schwarze 207: static enum rofft roff_parse(struct roff *, char *, int *,
208: int, int);
1.110 schwarze 209: static enum rofferr roff_parsetext(struct buf *, int, int *);
210: static enum rofferr roff_res(struct roff *, struct buf *, int, int);
1.29 schwarze 211: static enum rofferr roff_rm(ROFF_ARGS);
1.76 schwarze 212: static enum rofferr roff_rr(ROFF_ARGS);
1.8 schwarze 213: static void roff_setstr(struct roff *,
1.16 schwarze 214: const char *, const char *, int);
1.80 schwarze 215: static void roff_setstrn(struct roffkv **, const char *,
1.42 schwarze 216: size_t, const char *, size_t, int);
1.14 schwarze 217: static enum rofferr roff_so(ROFF_ARGS);
1.42 schwarze 218: static enum rofferr roff_tr(ROFF_ARGS);
1.47 schwarze 219: static enum rofferr roff_Dd(ROFF_ARGS);
220: static enum rofferr roff_TH(ROFF_ARGS);
1.27 schwarze 221: static enum rofferr roff_TE(ROFF_ARGS);
222: static enum rofferr roff_TS(ROFF_ARGS);
1.32 schwarze 223: static enum rofferr roff_EQ(ROFF_ARGS);
224: static enum rofferr roff_EN(ROFF_ARGS);
1.27 schwarze 225: static enum rofferr roff_T_(ROFF_ARGS);
1.16 schwarze 226: static enum rofferr roff_userdef(ROFF_ARGS);
1.1 schwarze 227:
1.42 schwarze 228: /* See roffhash_find() */
1.3 schwarze 229:
230: #define ASCII_HI 126
231: #define ASCII_LO 33
232: #define HASHWIDTH (ASCII_HI - ASCII_LO + 1)
233:
234: static struct roffmac *hash[HASHWIDTH];
235:
236: static struct roffmac roffs[ROFF_MAX] = {
1.21 schwarze 237: { "ad", roff_line_ignore, NULL, NULL, 0, NULL },
1.3 schwarze 238: { "am", roff_block, roff_block_text, roff_block_sub, 0, NULL },
239: { "ami", roff_block, roff_block_text, roff_block_sub, 0, NULL },
240: { "am1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.66 schwarze 241: { "as", roff_ds, NULL, NULL, 0, NULL },
1.48 schwarze 242: { "cc", roff_cc, NULL, NULL, 0, NULL },
1.67 schwarze 243: { "ce", roff_line_ignore, NULL, NULL, 0, NULL },
1.3 schwarze 244: { "de", roff_block, roff_block_text, roff_block_sub, 0, NULL },
245: { "dei", roff_block, roff_block_text, roff_block_sub, 0, NULL },
246: { "de1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.7 schwarze 247: { "ds", roff_ds, NULL, NULL, 0, NULL },
1.3 schwarze 248: { "el", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
1.58 schwarze 249: { "fam", roff_line_ignore, NULL, NULL, 0, NULL },
1.59 schwarze 250: { "hw", roff_line_ignore, NULL, NULL, 0, NULL },
1.21 schwarze 251: { "hy", roff_line_ignore, NULL, NULL, 0, NULL },
1.3 schwarze 252: { "ie", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
253: { "if", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
254: { "ig", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.51 schwarze 255: { "it", roff_it, NULL, NULL, 0, NULL },
1.21 schwarze 256: { "ne", roff_line_ignore, NULL, NULL, 0, NULL },
257: { "nh", roff_line_ignore, NULL, NULL, 0, NULL },
1.14 schwarze 258: { "nr", roff_nr, NULL, NULL, 0, NULL },
1.31 schwarze 259: { "ns", roff_line_ignore, NULL, NULL, 0, NULL },
1.101 schwarze 260: { "pl", roff_line_ignore, NULL, NULL, 0, NULL },
1.31 schwarze 261: { "ps", roff_line_ignore, NULL, NULL, 0, NULL },
1.29 schwarze 262: { "rm", roff_rm, NULL, NULL, 0, NULL },
1.76 schwarze 263: { "rr", roff_rr, NULL, NULL, 0, NULL },
1.14 schwarze 264: { "so", roff_so, NULL, NULL, 0, NULL },
1.31 schwarze 265: { "ta", roff_line_ignore, NULL, NULL, 0, NULL },
1.42 schwarze 266: { "tr", roff_tr, NULL, NULL, 0, NULL },
1.47 schwarze 267: { "Dd", roff_Dd, NULL, NULL, 0, NULL },
268: { "TH", roff_TH, NULL, NULL, 0, NULL },
1.27 schwarze 269: { "TS", roff_TS, NULL, NULL, 0, NULL },
270: { "TE", roff_TE, NULL, NULL, 0, NULL },
271: { "T&", roff_T_, NULL, NULL, 0, NULL },
1.32 schwarze 272: { "EQ", roff_EQ, NULL, NULL, 0, NULL },
273: { "EN", roff_EN, NULL, NULL, 0, NULL },
1.3 schwarze 274: { ".", roff_cblock, NULL, NULL, 0, NULL },
1.16 schwarze 275: { NULL, roff_userdef, NULL, NULL, 0, NULL },
1.1 schwarze 276: };
277:
1.73 schwarze 278: /* not currently implemented: Ds em Eq LP Me PP pp Or Rd Sf SH */
1.47 schwarze 279: const char *const __mdoc_reserved[] = {
280: "Ac", "Ad", "An", "Ao", "Ap", "Aq", "Ar", "At",
281: "Bc", "Bd", "Bf", "Bk", "Bl", "Bo", "Bq",
282: "Brc", "Bro", "Brq", "Bsx", "Bt", "Bx",
283: "Cd", "Cm", "Db", "Dc", "Dd", "Dl", "Do", "Dq",
1.73 schwarze 284: "Dt", "Dv", "Dx", "D1",
285: "Ec", "Ed", "Ef", "Ek", "El", "Em",
286: "En", "Eo", "Er", "Es", "Ev", "Ex",
1.47 schwarze 287: "Fa", "Fc", "Fd", "Fl", "Fn", "Fo", "Fr", "Ft", "Fx",
1.73 schwarze 288: "Hf", "Ic", "In", "It", "Lb", "Li", "Lk", "Lp",
289: "Ms", "Mt", "Nd", "Nm", "No", "Ns", "Nx",
1.47 schwarze 290: "Oc", "Oo", "Op", "Os", "Ot", "Ox",
1.73 schwarze 291: "Pa", "Pc", "Pf", "Po", "Pp", "Pq",
292: "Qc", "Ql", "Qo", "Qq", "Re", "Rs", "Rv",
293: "Sc", "Sh", "Sm", "So", "Sq",
1.47 schwarze 294: "Ss", "St", "Sx", "Sy",
295: "Ta", "Tn", "Ud", "Ux", "Va", "Vt", "Xc", "Xo", "Xr",
1.73 schwarze 296: "%A", "%B", "%C", "%D", "%I", "%J", "%N", "%O",
1.47 schwarze 297: "%P", "%Q", "%R", "%T", "%U", "%V",
298: NULL
299: };
300:
1.73 schwarze 301: /* not currently implemented: BT DE DS ME MT PT SY TQ YS */
1.47 schwarze 302: const char *const __man_reserved[] = {
1.73 schwarze 303: "AT", "B", "BI", "BR", "DT",
304: "EE", "EN", "EQ", "EX", "HP", "I", "IB", "IP", "IR",
305: "LP", "OP", "P", "PD", "PP",
306: "R", "RB", "RE", "RI", "RS", "SB", "SH", "SM", "SS",
307: "TE", "TH", "TP", "TS", "T&", "UC", "UE", "UR",
1.47 schwarze 308: NULL
309: };
310:
1.37 schwarze 311: /* Array of injected predefined strings. */
312: #define PREDEFS_MAX 38
313: static const struct predef predefs[PREDEFS_MAX] = {
314: #include "predefs.in"
315: };
316:
1.42 schwarze 317: /* See roffhash_find() */
1.3 schwarze 318: #define ROFF_HASH(p) (p[0] - ASCII_LO)
319:
1.51 schwarze 320: static int roffit_lines; /* number of lines to delay */
321: static char *roffit_macro; /* nil-terminated macro line */
322:
1.80 schwarze 323:
1.3 schwarze 324: static void
1.42 schwarze 325: roffhash_init(void)
1.3 schwarze 326: {
327: struct roffmac *n;
328: int buc, i;
329:
1.16 schwarze 330: for (i = 0; i < (int)ROFF_USERDEF; i++) {
1.3 schwarze 331: assert(roffs[i].name[0] >= ASCII_LO);
332: assert(roffs[i].name[0] <= ASCII_HI);
333:
334: buc = ROFF_HASH(roffs[i].name);
335:
336: if (NULL != (n = hash[buc])) {
337: for ( ; n->next; n = n->next)
338: /* Do nothing. */ ;
339: n->next = &roffs[i];
340: } else
341: hash[buc] = &roffs[i];
342: }
343: }
344:
1.1 schwarze 345: /*
346: * Look up a roff token by its name. Returns ROFF_MAX if no macro by
347: * the nil-terminated string name could be found.
348: */
349: static enum rofft
1.42 schwarze 350: roffhash_find(const char *p, size_t s)
1.1 schwarze 351: {
1.3 schwarze 352: int buc;
353: struct roffmac *n;
1.1 schwarze 354:
1.3 schwarze 355: /*
356: * libroff has an extremely simple hashtable, for the time
357: * being, which simply keys on the first character, which must
358: * be printable, then walks a chain. It works well enough until
359: * optimised.
360: */
361:
362: if (p[0] < ASCII_LO || p[0] > ASCII_HI)
363: return(ROFF_MAX);
364:
365: buc = ROFF_HASH(p);
366:
367: if (NULL == (n = hash[buc]))
368: return(ROFF_MAX);
369: for ( ; n; n = n->next)
1.16 schwarze 370: if (0 == strncmp(n->name, p, s) && '\0' == n->name[(int)s])
1.3 schwarze 371: return((enum rofft)(n - roffs));
1.1 schwarze 372:
373: return(ROFF_MAX);
374: }
375:
376: /*
377: * Pop the current node off of the stack of roff instructions currently
378: * pending.
379: */
380: static void
381: roffnode_pop(struct roff *r)
382: {
383: struct roffnode *p;
384:
1.2 schwarze 385: assert(r->last);
1.80 schwarze 386: p = r->last;
1.2 schwarze 387:
388: r->last = r->last->parent;
1.16 schwarze 389: free(p->name);
390: free(p->end);
1.1 schwarze 391: free(p);
392: }
393:
394: /*
395: * Push a roff node onto the instruction stack. This must later be
396: * removed with roffnode_pop().
397: */
1.11 schwarze 398: static void
1.16 schwarze 399: roffnode_push(struct roff *r, enum rofft tok, const char *name,
400: int line, int col)
1.1 schwarze 401: {
402: struct roffnode *p;
403:
1.11 schwarze 404: p = mandoc_calloc(1, sizeof(struct roffnode));
1.1 schwarze 405: p->tok = tok;
1.16 schwarze 406: if (name)
407: p->name = mandoc_strdup(name);
1.1 schwarze 408: p->parent = r->last;
409: p->line = line;
410: p->col = col;
1.71 schwarze 411: p->rule = p->parent ? p->parent->rule : 0;
1.1 schwarze 412:
413: r->last = p;
414: }
415:
416: static void
417: roff_free1(struct roff *r)
418: {
1.49 schwarze 419: struct tbl_node *tbl;
1.32 schwarze 420: struct eqn_node *e;
1.42 schwarze 421: int i;
1.27 schwarze 422:
1.49 schwarze 423: while (NULL != (tbl = r->first_tbl)) {
424: r->first_tbl = tbl->next;
425: tbl_free(tbl);
1.27 schwarze 426: }
427: r->first_tbl = r->last_tbl = r->tbl = NULL;
1.1 schwarze 428:
1.32 schwarze 429: while (NULL != (e = r->first_eqn)) {
430: r->first_eqn = e->next;
431: eqn_free(e);
432: }
433: r->first_eqn = r->last_eqn = r->eqn = NULL;
434:
1.1 schwarze 435: while (r->last)
436: roffnode_pop(r);
1.27 schwarze 437:
1.96 schwarze 438: free (r->rstack);
439: r->rstack = NULL;
440: r->rstacksz = 0;
441: r->rstackpos = -1;
442:
443: roff_freereg(r->regtab);
444: r->regtab = NULL;
445:
1.42 schwarze 446: roff_freestr(r->strtab);
447: roff_freestr(r->xmbtab);
448: r->strtab = r->xmbtab = NULL;
449:
450: if (r->xtab)
451: for (i = 0; i < 128; i++)
452: free(r->xtab[i].p);
453: free(r->xtab);
454: r->xtab = NULL;
1.1 schwarze 455: }
456:
457: void
458: roff_reset(struct roff *r)
459: {
460:
461: roff_free1(r);
1.99 schwarze 462: r->format = r->options & (MPARSE_MDOC | MPARSE_MAN);
1.48 schwarze 463: r->control = 0;
1.1 schwarze 464: }
465:
466: void
467: roff_free(struct roff *r)
468: {
469:
470: roff_free1(r);
471: free(r);
472: }
473:
474: struct roff *
1.109 schwarze 475: roff_alloc(struct mparse *parse, const struct mchars *mchars, int options)
1.1 schwarze 476: {
477: struct roff *r;
478:
1.11 schwarze 479: r = mandoc_calloc(1, sizeof(struct roff));
1.35 schwarze 480: r->parse = parse;
1.109 schwarze 481: r->mchars = mchars;
1.72 schwarze 482: r->options = options;
1.99 schwarze 483: r->format = options & (MPARSE_MDOC | MPARSE_MAN);
1.2 schwarze 484: r->rstackpos = -1;
1.80 schwarze 485:
1.42 schwarze 486: roffhash_init();
1.37 schwarze 487:
1.1 schwarze 488: return(r);
489: }
490:
1.8 schwarze 491: /*
1.79 schwarze 492: * In the current line, expand escape sequences that tend to get
493: * used in numerical expressions and conditional requests.
494: * Also check the syntax of the remaining escape sequences.
1.8 schwarze 495: */
1.45 schwarze 496: static enum rofferr
1.110 schwarze 497: roff_res(struct roff *r, struct buf *buf, int ln, int pos)
1.8 schwarze 498: {
1.81 schwarze 499: char ubuf[24]; /* buffer to print the number */
1.78 schwarze 500: const char *start; /* start of the string to process */
1.82 schwarze 501: char *stesc; /* start of an escape sequence ('\\') */
1.23 schwarze 502: const char *stnam; /* start of the name, after "[(*" */
503: const char *cp; /* end of the name, e.g. before ']' */
504: const char *res; /* the string to be substituted */
1.110 schwarze 505: char *nbuf; /* new buffer to copy buf->buf to */
1.53 schwarze 506: size_t maxl; /* expected length of the escape name */
507: size_t naml; /* actual length of the escape name */
1.109 schwarze 508: enum mandoc_esc esc; /* type of the escape sequence */
509: int inaml; /* length returned from mandoc_escape() */
1.53 schwarze 510: int expand_count; /* to avoid infinite loops */
1.79 schwarze 511: int npos; /* position in numeric expression */
1.91 schwarze 512: int arg_complete; /* argument not interrupted by eol */
1.79 schwarze 513: char term; /* character terminating the escape */
1.8 schwarze 514:
1.43 schwarze 515: expand_count = 0;
1.110 schwarze 516: start = buf->buf + pos;
1.78 schwarze 517: stesc = strchr(start, '\0') - 1;
518: while (stesc-- > start) {
1.43 schwarze 519:
1.78 schwarze 520: /* Search backwards for the next backslash. */
521:
1.110 schwarze 522: if (*stesc != '\\')
1.78 schwarze 523: continue;
524:
525: /* If it is escaped, skip it. */
526:
527: for (cp = stesc - 1; cp >= start; cp--)
1.110 schwarze 528: if (*cp != '\\')
1.78 schwarze 529: break;
530:
1.110 schwarze 531: if ((stesc - cp) % 2 == 0) {
1.82 schwarze 532: stesc = (char *)cp;
1.78 schwarze 533: continue;
534: }
1.23 schwarze 535:
1.79 schwarze 536: /* Decide whether to expand or to check only. */
1.23 schwarze 537:
1.79 schwarze 538: term = '\0';
1.78 schwarze 539: cp = stesc + 1;
1.53 schwarze 540: switch (*cp) {
1.80 schwarze 541: case '*':
1.53 schwarze 542: res = NULL;
543: break;
1.80 schwarze 544: case 'B':
1.79 schwarze 545: /* FALLTHROUGH */
1.80 schwarze 546: case 'w':
1.79 schwarze 547: term = cp[1];
548: /* FALLTHROUGH */
1.80 schwarze 549: case 'n':
1.53 schwarze 550: res = ubuf;
551: break;
552: default:
1.109 schwarze 553: esc = mandoc_escape(&cp, &stnam, &inaml);
554: if (esc == ESCAPE_ERROR ||
555: (esc == ESCAPE_SPECIAL &&
556: mchars_spec2cp(r->mchars, stnam, inaml) < 0))
1.92 schwarze 557: mandoc_vmsg(MANDOCERR_ESC_BAD,
1.110 schwarze 558: r->parse, ln, (int)(stesc - buf->buf),
1.92 schwarze 559: "%.*s", (int)(cp - stesc), stesc);
1.78 schwarze 560: continue;
1.42 schwarze 561: }
562:
1.78 schwarze 563: if (EXPAND_LIMIT < ++expand_count) {
564: mandoc_msg(MANDOCERR_ROFFLOOP, r->parse,
1.110 schwarze 565: ln, (int)(stesc - buf->buf), NULL);
1.78 schwarze 566: return(ROFF_IGN);
567: }
1.23 schwarze 568:
569: /*
570: * The third character decides the length
1.53 schwarze 571: * of the name of the string or register.
1.23 schwarze 572: * Save a pointer to the name.
573: */
574:
1.110 schwarze 575: if (term == '\0') {
1.79 schwarze 576: switch (*++cp) {
1.80 schwarze 577: case '\0':
1.79 schwarze 578: maxl = 0;
579: break;
1.80 schwarze 580: case '(':
1.79 schwarze 581: cp++;
582: maxl = 2;
583: break;
1.80 schwarze 584: case '[':
1.79 schwarze 585: cp++;
586: term = ']';
587: maxl = 0;
588: break;
589: default:
590: maxl = 1;
591: break;
592: }
593: } else {
594: cp += 2;
1.8 schwarze 595: maxl = 0;
596: }
1.23 schwarze 597: stnam = cp;
1.8 schwarze 598:
1.23 schwarze 599: /* Advance to the end of the name. */
1.8 schwarze 600:
1.91 schwarze 601: arg_complete = 1;
1.110 schwarze 602: for (naml = 0; maxl == 0 || naml < maxl; naml++, cp++) {
603: if (*cp == '\0') {
1.92 schwarze 604: mandoc_msg(MANDOCERR_ESC_BAD, r->parse,
1.110 schwarze 605: ln, (int)(stesc - buf->buf), stesc);
1.91 schwarze 606: arg_complete = 0;
1.79 schwarze 607: break;
1.42 schwarze 608: }
1.110 schwarze 609: if (maxl == 0 && *cp == term) {
1.79 schwarze 610: cp++;
1.8 schwarze 611: break;
1.79 schwarze 612: }
1.8 schwarze 613: }
614:
1.23 schwarze 615: /*
616: * Retrieve the replacement string; if it is
617: * undefined, resume searching for escapes.
618: */
619:
1.79 schwarze 620: switch (stesc[1]) {
1.80 schwarze 621: case '*':
1.91 schwarze 622: if (arg_complete)
623: res = roff_getstrn(r, stnam, naml);
1.79 schwarze 624: break;
1.80 schwarze 625: case 'B':
1.79 schwarze 626: npos = 0;
1.91 schwarze 627: ubuf[0] = arg_complete &&
1.106 schwarze 628: roff_evalnum(r, ln, stnam, &npos, NULL, 0) &&
1.91 schwarze 629: stnam + npos + 1 == cp ? '1' : '0';
1.79 schwarze 630: ubuf[1] = '\0';
631: break;
1.80 schwarze 632: case 'n':
1.91 schwarze 633: if (arg_complete)
634: (void)snprintf(ubuf, sizeof(ubuf), "%d",
635: roff_getregn(r, stnam, naml));
636: else
637: ubuf[0] = '\0';
1.79 schwarze 638: break;
1.80 schwarze 639: case 'w':
1.91 schwarze 640: /* use even incomplete args */
1.81 schwarze 641: (void)snprintf(ubuf, sizeof(ubuf), "%d",
1.79 schwarze 642: 24 * (int)naml);
643: break;
644: }
1.8 schwarze 645:
1.110 schwarze 646: if (res == NULL) {
1.92 schwarze 647: mandoc_vmsg(MANDOCERR_STR_UNDEF,
1.110 schwarze 648: r->parse, ln, (int)(stesc - buf->buf),
1.92 schwarze 649: "%.*s", (int)naml, stnam);
1.37 schwarze 650: res = "";
1.8 schwarze 651: }
652:
1.23 schwarze 653: /* Replace the escape sequence by the string. */
654:
1.82 schwarze 655: *stesc = '\0';
1.110 schwarze 656: buf->sz = mandoc_asprintf(&nbuf, "%s%s%s",
657: buf->buf, res, cp) + 1;
1.8 schwarze 658:
1.78 schwarze 659: /* Prepare for the next replacement. */
1.8 schwarze 660:
1.78 schwarze 661: start = nbuf + pos;
1.110 schwarze 662: stesc = nbuf + (stesc - buf->buf) + strlen(res);
663: free(buf->buf);
664: buf->buf = nbuf;
1.42 schwarze 665: }
1.45 schwarze 666: return(ROFF_CONT);
1.42 schwarze 667: }
668:
669: /*
1.51 schwarze 670: * Process text streams:
671: * Convert all breakable hyphens into ASCII_HYPH.
672: * Decrement and spring input line trap.
1.42 schwarze 673: */
674: static enum rofferr
1.110 schwarze 675: roff_parsetext(struct buf *buf, int pos, int *offs)
1.42 schwarze 676: {
677: size_t sz;
678: const char *start;
1.51 schwarze 679: char *p;
680: int isz;
1.42 schwarze 681: enum mandoc_esc esc;
682:
1.110 schwarze 683: start = p = buf->buf + pos;
1.42 schwarze 684:
1.110 schwarze 685: while (*p != '\0') {
1.42 schwarze 686: sz = strcspn(p, "-\\");
687: p += sz;
688:
1.110 schwarze 689: if (*p == '\0')
1.42 schwarze 690: break;
691:
1.110 schwarze 692: if (*p == '\\') {
1.42 schwarze 693: /* Skip over escapes. */
694: p++;
1.62 schwarze 695: esc = mandoc_escape((const char **)&p, NULL, NULL);
1.110 schwarze 696: if (esc == ESCAPE_ERROR)
1.42 schwarze 697: break;
698: continue;
699: } else if (p == start) {
700: p++;
701: continue;
702: }
703:
1.44 schwarze 704: if (isalpha((unsigned char)p[-1]) &&
705: isalpha((unsigned char)p[1]))
1.42 schwarze 706: *p = ASCII_HYPH;
707: p++;
1.8 schwarze 708: }
709:
1.51 schwarze 710: /* Spring the input line trap. */
1.110 schwarze 711: if (roffit_lines == 1) {
712: isz = mandoc_asprintf(&p, "%s\n.%s", buf->buf, roffit_macro);
713: free(buf->buf);
714: buf->buf = p;
715: buf->sz = isz + 1;
1.51 schwarze 716: *offs = 0;
717: free(roffit_macro);
718: roffit_lines = 0;
719: return(ROFF_REPARSE);
1.110 schwarze 720: } else if (roffit_lines > 1)
1.51 schwarze 721: --roffit_lines;
1.42 schwarze 722: return(ROFF_CONT);
1.8 schwarze 723: }
724:
1.1 schwarze 725: enum rofferr
1.110 schwarze 726: roff_parseln(struct roff *r, int ln, struct buf *buf, int *offs)
1.1 schwarze 727: {
728: enum rofft t;
1.27 schwarze 729: enum rofferr e;
1.110 schwarze 730: int pos; /* parse point */
731: int ppos; /* original offset in buf->buf */
732: int ctl; /* macro line (boolean) */
733:
734: ppos = pos = *offs;
1.1 schwarze 735:
1.102 schwarze 736: /* Handle in-line equation delimiters. */
737:
1.108 schwarze 738: if (r->tbl == NULL &&
739: r->last_eqn != NULL && r->last_eqn->delim &&
1.102 schwarze 740: (r->eqn == NULL || r->eqn_inline)) {
1.110 schwarze 741: e = roff_eqndelim(r, buf, pos);
1.102 schwarze 742: if (e == ROFF_REPARSE)
743: return(e);
744: assert(e == ROFF_CONT);
745: }
746:
747: /* Expand some escape sequences. */
1.8 schwarze 748:
1.110 schwarze 749: e = roff_res(r, buf, ln, pos);
750: if (e == ROFF_IGN)
1.45 schwarze 751: return(e);
1.110 schwarze 752: assert(e == ROFF_CONT);
1.8 schwarze 753:
1.110 schwarze 754: ctl = roff_getcontrol(r, buf->buf, &pos);
1.35 schwarze 755:
1.8 schwarze 756: /*
1.2 schwarze 757: * First, if a scope is open and we're not a macro, pass the
758: * text through the macro's filter. If a scope isn't open and
759: * we're not a macro, just let it through.
1.32 schwarze 760: * Finally, if there's an equation scope open, divert it into it
761: * no matter our state.
1.2 schwarze 762: */
763:
1.35 schwarze 764: if (r->last && ! ctl) {
1.2 schwarze 765: t = r->last->tok;
766: assert(roffs[t].text);
1.110 schwarze 767: e = (*roffs[t].text)(r, t, buf, ln, pos, pos, offs);
768: assert(e == ROFF_IGN || e == ROFF_CONT);
769: if (e != ROFF_CONT)
1.32 schwarze 770: return(e);
1.54 schwarze 771: }
772: if (r->eqn)
1.110 schwarze 773: return(eqn_read(&r->eqn, ln, buf->buf, ppos, offs));
1.54 schwarze 774: if ( ! ctl) {
1.32 schwarze 775: if (r->tbl)
1.110 schwarze 776: return(tbl_read(r->tbl, ln, buf->buf, pos));
777: return(roff_parsetext(buf, pos, offs));
1.54 schwarze 778: }
1.100 schwarze 779:
780: /* Skip empty request lines. */
781:
1.110 schwarze 782: if (buf->buf[pos] == '"') {
1.100 schwarze 783: mandoc_msg(MANDOCERR_COMMENT_BAD, r->parse,
784: ln, pos, NULL);
785: return(ROFF_IGN);
1.110 schwarze 786: } else if (buf->buf[pos] == '\0')
1.100 schwarze 787: return(ROFF_IGN);
1.2 schwarze 788:
789: /*
790: * If a scope is open, go to the child handler for that macro,
791: * as it may want to preprocess before doing anything with it.
1.32 schwarze 792: * Don't do so if an equation is open.
1.2 schwarze 793: */
794:
795: if (r->last) {
1.1 schwarze 796: t = r->last->tok;
797: assert(roffs[t].sub);
1.110 schwarze 798: return((*roffs[t].sub)(r, t, buf, ln, ppos, pos, offs));
1.2 schwarze 799: }
800:
801: /*
802: * Lastly, as we've no scope open, try to look up and execute
803: * the new macro. If no macro is found, simply return and let
804: * the compilers handle it.
805: */
806:
1.110 schwarze 807: if ((t = roff_parse(r, buf->buf, &pos, ln, ppos)) == ROFF_MAX)
1.1 schwarze 808: return(ROFF_CONT);
809:
1.2 schwarze 810: assert(roffs[t].proc);
1.110 schwarze 811: return((*roffs[t].proc)(r, t, buf, ln, ppos, pos, offs));
1.2 schwarze 812: }
813:
1.27 schwarze 814: void
1.2 schwarze 815: roff_endparse(struct roff *r)
816: {
1.1 schwarze 817:
1.27 schwarze 818: if (r->last)
1.94 schwarze 819: mandoc_msg(MANDOCERR_BLK_NOEND, r->parse,
820: r->last->line, r->last->col,
821: roffs[r->last->tok].name);
1.27 schwarze 822:
1.32 schwarze 823: if (r->eqn) {
1.94 schwarze 824: mandoc_msg(MANDOCERR_BLK_NOEND, r->parse,
825: r->eqn->eqn.ln, r->eqn->eqn.pos, "EQ");
1.41 schwarze 826: eqn_end(&r->eqn);
1.32 schwarze 827: }
828:
1.27 schwarze 829: if (r->tbl) {
1.94 schwarze 830: mandoc_msg(MANDOCERR_BLK_NOEND, r->parse,
831: r->tbl->line, r->tbl->pos, "TS");
1.41 schwarze 832: tbl_end(&r->tbl);
1.27 schwarze 833: }
1.1 schwarze 834: }
835:
836: /*
837: * Parse a roff node's type from the input buffer. This must be in the
838: * form of ".foo xxx" in the usual way.
839: */
840: static enum rofft
1.87 schwarze 841: roff_parse(struct roff *r, char *buf, int *pos, int ln, int ppos)
1.1 schwarze 842: {
1.87 schwarze 843: char *cp;
1.16 schwarze 844: const char *mac;
845: size_t maclen;
1.1 schwarze 846: enum rofft t;
847:
1.87 schwarze 848: cp = buf + *pos;
849:
850: if ('\0' == *cp || '"' == *cp || '\t' == *cp || ' ' == *cp)
1.1 schwarze 851: return(ROFF_MAX);
852:
1.87 schwarze 853: mac = cp;
854: maclen = roff_getname(r, &cp, ln, ppos);
1.1 schwarze 855:
1.16 schwarze 856: t = (r->current_string = roff_getstrn(r, mac, maclen))
1.42 schwarze 857: ? ROFF_USERDEF : roffhash_find(mac, maclen);
1.1 schwarze 858:
1.87 schwarze 859: if (ROFF_MAX != t)
860: *pos = cp - buf;
1.1 schwarze 861:
862: return(t);
863: }
864:
865: static enum rofferr
1.2 schwarze 866: roff_cblock(ROFF_ARGS)
1.1 schwarze 867: {
868:
1.2 schwarze 869: /*
870: * A block-close `..' should only be invoked as a child of an
871: * ignore macro, otherwise raise a warning and just ignore it.
872: */
873:
1.110 schwarze 874: if (r->last == NULL) {
1.94 schwarze 875: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
876: ln, ppos, "..");
1.2 schwarze 877: return(ROFF_IGN);
878: }
1.1 schwarze 879:
1.2 schwarze 880: switch (r->last->tok) {
1.80 schwarze 881: case ROFF_am:
1.93 schwarze 882: /* ROFF_am1 is remapped to ROFF_am in roff_block(). */
1.2 schwarze 883: /* FALLTHROUGH */
1.80 schwarze 884: case ROFF_ami:
1.2 schwarze 885: /* FALLTHROUGH */
1.80 schwarze 886: case ROFF_de:
1.23 schwarze 887: /* ROFF_de1 is remapped to ROFF_de in roff_block(). */
1.2 schwarze 888: /* FALLTHROUGH */
1.80 schwarze 889: case ROFF_dei:
1.2 schwarze 890: /* FALLTHROUGH */
1.80 schwarze 891: case ROFF_ig:
1.2 schwarze 892: break;
893: default:
1.94 schwarze 894: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
895: ln, ppos, "..");
1.1 schwarze 896: return(ROFF_IGN);
1.2 schwarze 897: }
898:
1.110 schwarze 899: if (buf->buf[pos] != '\0')
1.90 schwarze 900: mandoc_vmsg(MANDOCERR_ARG_SKIP, r->parse, ln, pos,
1.110 schwarze 901: ".. %s", buf->buf + pos);
1.2 schwarze 902:
903: roffnode_pop(r);
904: roffnode_cleanscope(r);
905: return(ROFF_IGN);
906:
907: }
1.1 schwarze 908:
1.2 schwarze 909: static void
910: roffnode_cleanscope(struct roff *r)
911: {
1.1 schwarze 912:
1.2 schwarze 913: while (r->last) {
1.46 schwarze 914: if (--r->last->endspan != 0)
1.2 schwarze 915: break;
916: roffnode_pop(r);
917: }
918: }
1.1 schwarze 919:
1.68 schwarze 920: static void
921: roff_ccond(struct roff *r, int ln, int ppos)
1.2 schwarze 922: {
1.1 schwarze 923:
1.2 schwarze 924: if (NULL == r->last) {
1.94 schwarze 925: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
926: ln, ppos, "\\}");
1.68 schwarze 927: return;
1.2 schwarze 928: }
1.1 schwarze 929:
1.2 schwarze 930: switch (r->last->tok) {
1.80 schwarze 931: case ROFF_el:
1.2 schwarze 932: /* FALLTHROUGH */
1.80 schwarze 933: case ROFF_ie:
1.2 schwarze 934: /* FALLTHROUGH */
1.80 schwarze 935: case ROFF_if:
1.2 schwarze 936: break;
937: default:
1.94 schwarze 938: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
939: ln, ppos, "\\}");
1.68 schwarze 940: return;
1.2 schwarze 941: }
1.1 schwarze 942:
1.2 schwarze 943: if (r->last->endspan > -1) {
1.94 schwarze 944: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
945: ln, ppos, "\\}");
1.68 schwarze 946: return;
1.2 schwarze 947: }
948:
949: roffnode_pop(r);
950: roffnode_cleanscope(r);
1.68 schwarze 951: return;
1.1 schwarze 952: }
953:
954: static enum rofferr
1.2 schwarze 955: roff_block(ROFF_ARGS)
1.1 schwarze 956: {
1.93 schwarze 957: const char *name;
958: char *iname, *cp;
1.86 schwarze 959: size_t namesz;
1.16 schwarze 960:
1.93 schwarze 961: /* Ignore groff compatibility mode for now. */
1.2 schwarze 962:
1.110 schwarze 963: if (tok == ROFF_de1)
1.93 schwarze 964: tok = ROFF_de;
1.110 schwarze 965: else if (tok == ROFF_am1)
1.93 schwarze 966: tok = ROFF_am;
967:
968: /* Parse the macro name argument. */
969:
1.110 schwarze 970: cp = buf->buf + pos;
971: if (tok == ROFF_ig) {
1.93 schwarze 972: iname = NULL;
973: namesz = 0;
974: } else {
975: iname = cp;
976: namesz = roff_getname(r, &cp, ln, ppos);
977: iname[namesz] = '\0';
978: }
1.22 schwarze 979:
1.93 schwarze 980: /* Resolve the macro name argument if it is indirect. */
1.22 schwarze 981:
1.110 schwarze 982: if (namesz && (tok == ROFF_dei || tok == ROFF_ami)) {
983: if ((name = roff_getstrn(r, iname, namesz)) == NULL) {
1.93 schwarze 984: mandoc_vmsg(MANDOCERR_STR_UNDEF,
1.110 schwarze 985: r->parse, ln, (int)(iname - buf->buf),
1.93 schwarze 986: "%.*s", (int)namesz, iname);
987: namesz = 0;
988: } else
989: namesz = strlen(name);
990: } else
991: name = iname;
1.22 schwarze 992:
1.110 schwarze 993: if (namesz == 0 && tok != ROFF_ig) {
1.93 schwarze 994: mandoc_msg(MANDOCERR_REQ_EMPTY, r->parse,
995: ln, ppos, roffs[tok].name);
996: return(ROFF_IGN);
997: }
1.2 schwarze 998:
1.16 schwarze 999: roffnode_push(r, tok, name, ln, ppos);
1000:
1001: /*
1002: * At the beginning of a `de' macro, clear the existing string
1003: * with the same name, if there is one. New content will be
1.66 schwarze 1004: * appended from roff_block_text() in multiline mode.
1.16 schwarze 1005: */
1.22 schwarze 1006:
1.110 schwarze 1007: if (tok == ROFF_de || tok == ROFF_dei)
1.86 schwarze 1008: roff_setstrn(&r->strtab, name, namesz, "", 0, 0);
1.2 schwarze 1009:
1.110 schwarze 1010: if (*cp == '\0')
1.2 schwarze 1011: return(ROFF_IGN);
1.1 schwarze 1012:
1.93 schwarze 1013: /* Get the custom end marker. */
1.22 schwarze 1014:
1.93 schwarze 1015: iname = cp;
1.86 schwarze 1016: namesz = roff_getname(r, &cp, ln, ppos);
1.93 schwarze 1017:
1018: /* Resolve the end marker if it is indirect. */
1019:
1.110 schwarze 1020: if (namesz && (tok == ROFF_dei || tok == ROFF_ami)) {
1021: if ((name = roff_getstrn(r, iname, namesz)) == NULL) {
1.93 schwarze 1022: mandoc_vmsg(MANDOCERR_STR_UNDEF,
1.110 schwarze 1023: r->parse, ln, (int)(iname - buf->buf),
1.93 schwarze 1024: "%.*s", (int)namesz, iname);
1025: namesz = 0;
1026: } else
1027: namesz = strlen(name);
1028: } else
1029: name = iname;
1030:
1.86 schwarze 1031: if (namesz)
1032: r->last->end = mandoc_strndup(name, namesz);
1.2 schwarze 1033:
1.110 schwarze 1034: if (*cp != '\0')
1.90 schwarze 1035: mandoc_vmsg(MANDOCERR_ARG_EXCESS, r->parse,
1036: ln, pos, ".%s ... %s", roffs[tok].name, cp);
1.1 schwarze 1037:
1038: return(ROFF_IGN);
1039: }
1040:
1041: static enum rofferr
1.2 schwarze 1042: roff_block_sub(ROFF_ARGS)
1.1 schwarze 1043: {
1.2 schwarze 1044: enum rofft t;
1045: int i, j;
1046:
1047: /*
1048: * First check whether a custom macro exists at this level. If
1049: * it does, then check against it. This is some of groff's
1050: * stranger behaviours. If we encountered a custom end-scope
1051: * tag and that tag also happens to be a "real" macro, then we
1052: * need to try interpreting it again as a real macro. If it's
1053: * not, then return ignore. Else continue.
1054: */
1055:
1056: if (r->last->end) {
1.35 schwarze 1057: for (i = pos, j = 0; r->last->end[j]; j++, i++)
1.110 schwarze 1058: if (buf->buf[i] != r->last->end[j])
1.2 schwarze 1059: break;
1.1 schwarze 1060:
1.110 schwarze 1061: if (r->last->end[j] == '\0' &&
1062: (buf->buf[i] == '\0' ||
1063: buf->buf[i] == ' ' ||
1064: buf->buf[i] == '\t')) {
1.2 schwarze 1065: roffnode_pop(r);
1066: roffnode_cleanscope(r);
1.1 schwarze 1067:
1.110 schwarze 1068: while (buf->buf[i] == ' ' || buf->buf[i] == '\t')
1.35 schwarze 1069: i++;
1070:
1071: pos = i;
1.110 schwarze 1072: if (roff_parse(r, buf->buf, &pos, ln, ppos) !=
1073: ROFF_MAX)
1.2 schwarze 1074: return(ROFF_RERUN);
1075: return(ROFF_IGN);
1076: }
1.1 schwarze 1077: }
1078:
1.2 schwarze 1079: /*
1080: * If we have no custom end-query or lookup failed, then try
1081: * pulling it out of the hashtable.
1082: */
1.1 schwarze 1083:
1.110 schwarze 1084: t = roff_parse(r, buf->buf, &pos, ln, ppos);
1.1 schwarze 1085:
1.110 schwarze 1086: if (t != ROFF_cblock) {
1087: if (tok != ROFF_ig)
1088: roff_setstr(r, r->last->name, buf->buf + ppos, 2);
1.1 schwarze 1089: return(ROFF_IGN);
1.16 schwarze 1090: }
1.1 schwarze 1091:
1.2 schwarze 1092: assert(roffs[t].proc);
1.110 schwarze 1093: return((*roffs[t].proc)(r, t, buf, ln, ppos, pos, offs));
1.2 schwarze 1094: }
1095:
1096: static enum rofferr
1097: roff_block_text(ROFF_ARGS)
1098: {
1099:
1.110 schwarze 1100: if (tok != ROFF_ig)
1101: roff_setstr(r, r->last->name, buf->buf + pos, 2);
1.16 schwarze 1102:
1.2 schwarze 1103: return(ROFF_IGN);
1104: }
1105:
1106: static enum rofferr
1107: roff_cond_sub(ROFF_ARGS)
1108: {
1109: enum rofft t;
1.37 schwarze 1110: char *ep;
1.71 schwarze 1111: int rr;
1.2 schwarze 1112:
1113: rr = r->last->rule;
1.37 schwarze 1114: roffnode_cleanscope(r);
1.110 schwarze 1115: t = roff_parse(r, buf->buf, &pos, ln, ppos);
1.2 schwarze 1116:
1.37 schwarze 1117: /*
1.50 schwarze 1118: * Fully handle known macros when they are structurally
1119: * required or when the conditional evaluated to true.
1.5 schwarze 1120: */
1121:
1.110 schwarze 1122: if ((t != ROFF_MAX) &&
1123: (rr || roffs[t].flags & ROFFMAC_STRUCT)) {
1.50 schwarze 1124: assert(roffs[t].proc);
1.110 schwarze 1125: return((*roffs[t].proc)(r, t, buf, ln, ppos, pos, offs));
1.50 schwarze 1126: }
1.39 schwarze 1127:
1.69 schwarze 1128: /*
1129: * If `\}' occurs on a macro line without a preceding macro,
1130: * drop the line completely.
1131: */
1132:
1.110 schwarze 1133: ep = buf->buf + pos;
1134: if (ep[0] == '\\' && ep[1] == '}')
1.71 schwarze 1135: rr = 0;
1.69 schwarze 1136:
1.50 schwarze 1137: /* Always check for the closing delimiter `\}'. */
1.39 schwarze 1138:
1.110 schwarze 1139: while ((ep = strchr(ep, '\\')) != NULL) {
1140: if (*(++ep) == '}') {
1.70 schwarze 1141: *ep = '&';
1.110 schwarze 1142: roff_ccond(r, ln, ep - buf->buf - 1);
1.70 schwarze 1143: }
1144: ++ep;
1.50 schwarze 1145: }
1.71 schwarze 1146: return(rr ? ROFF_CONT : ROFF_IGN);
1.2 schwarze 1147: }
1148:
1149: static enum rofferr
1150: roff_cond_text(ROFF_ARGS)
1151: {
1.37 schwarze 1152: char *ep;
1.71 schwarze 1153: int rr;
1.2 schwarze 1154:
1155: rr = r->last->rule;
1.37 schwarze 1156: roffnode_cleanscope(r);
1.1 schwarze 1157:
1.110 schwarze 1158: ep = buf->buf + pos;
1159: while ((ep = strchr(ep, '\\')) != NULL) {
1160: if (*(++ep) == '}') {
1.70 schwarze 1161: *ep = '&';
1.110 schwarze 1162: roff_ccond(r, ln, ep - buf->buf - 1);
1.70 schwarze 1163: }
1164: ++ep;
1.2 schwarze 1165: }
1.71 schwarze 1166: return(rr ? ROFF_CONT : ROFF_IGN);
1.2 schwarze 1167: }
1168:
1.77 schwarze 1169: /*
1170: * Parse a single signed integer number. Stop at the first non-digit.
1171: * If there is at least one digit, return success and advance the
1172: * parse point, else return failure and let the parse point unchanged.
1173: * Ignore overflows, treat them just like the C language.
1174: */
1.56 schwarze 1175: static int
1176: roff_getnum(const char *v, int *pos, int *res)
1177: {
1.79 schwarze 1178: int myres, n, p;
1179:
1180: if (NULL == res)
1181: res = &myres;
1.56 schwarze 1182:
1183: p = *pos;
1184: n = v[p] == '-';
1185: if (n)
1186: p++;
1187:
1188: for (*res = 0; isdigit((unsigned char)v[p]); p++)
1.77 schwarze 1189: *res = 10 * *res + v[p] - '0';
1.56 schwarze 1190: if (p == *pos + n)
1191: return 0;
1192:
1193: if (n)
1194: *res = -*res;
1195:
1196: *pos = p;
1197: return 1;
1198: }
1199:
1.71 schwarze 1200: /*
1201: * Evaluate a string comparison condition.
1202: * The first character is the delimiter.
1203: * Succeed if the string up to its second occurrence
1204: * matches the string up to its third occurence.
1205: * Advance the cursor after the third occurrence
1206: * or lacking that, to the end of the line.
1207: */
1208: static int
1209: roff_evalstrcond(const char *v, int *pos)
1210: {
1211: const char *s1, *s2, *s3;
1212: int match;
1213:
1214: match = 0;
1215: s1 = v + *pos; /* initial delimiter */
1216: s2 = s1 + 1; /* for scanning the first string */
1217: s3 = strchr(s2, *s1); /* for scanning the second string */
1218:
1219: if (NULL == s3) /* found no middle delimiter */
1220: goto out;
1221:
1222: while ('\0' != *++s3) {
1223: if (*s2 != *s3) { /* mismatch */
1224: s3 = strchr(s3, *s1);
1225: break;
1226: }
1227: if (*s3 == *s1) { /* found the final delimiter */
1228: match = 1;
1229: break;
1230: }
1231: s2++;
1232: }
1233:
1234: out:
1235: if (NULL == s3)
1236: s3 = strchr(s2, '\0');
1237: else
1238: s3++;
1239: *pos = s3 - v;
1240: return(match);
1241: }
1242:
1.77 schwarze 1243: /*
1244: * Evaluate an optionally negated single character, numerical,
1245: * or string condition.
1246: */
1.71 schwarze 1247: static int
1.106 schwarze 1248: roff_evalcond(struct roff *r, int ln, const char *v, int *pos)
1.5 schwarze 1249: {
1.77 schwarze 1250: int wanttrue, number;
1.5 schwarze 1251:
1.71 schwarze 1252: if ('!' == v[*pos]) {
1253: wanttrue = 0;
1254: (*pos)++;
1255: } else
1256: wanttrue = 1;
1257:
1.5 schwarze 1258: switch (v[*pos]) {
1.112 ! schwarze 1259: case '\0':
! 1260: return(0);
1.80 schwarze 1261: case 'n':
1.71 schwarze 1262: /* FALLTHROUGH */
1.80 schwarze 1263: case 'o':
1.5 schwarze 1264: (*pos)++;
1.71 schwarze 1265: return(wanttrue);
1.80 schwarze 1266: case 'c':
1.71 schwarze 1267: /* FALLTHROUGH */
1.80 schwarze 1268: case 'd':
1.71 schwarze 1269: /* FALLTHROUGH */
1.80 schwarze 1270: case 'e':
1.5 schwarze 1271: /* FALLTHROUGH */
1.80 schwarze 1272: case 'r':
1.5 schwarze 1273: /* FALLTHROUGH */
1.80 schwarze 1274: case 't':
1.111 schwarze 1275: /* FALLTHROUGH */
1276: case 'v':
1.5 schwarze 1277: (*pos)++;
1.71 schwarze 1278: return(!wanttrue);
1.5 schwarze 1279: default:
1280: break;
1281: }
1282:
1.106 schwarze 1283: if (roff_evalnum(r, ln, v, pos, &number, 0))
1.77 schwarze 1284: return((number > 0) == wanttrue);
1285: else
1.71 schwarze 1286: return(roff_evalstrcond(v, pos) == wanttrue);
1.5 schwarze 1287: }
1288:
1.2 schwarze 1289: static enum rofferr
1.21 schwarze 1290: roff_line_ignore(ROFF_ARGS)
1.6 schwarze 1291: {
1.30 schwarze 1292:
1.21 schwarze 1293: return(ROFF_IGN);
1294: }
1295:
1296: static enum rofferr
1.2 schwarze 1297: roff_cond(ROFF_ARGS)
1298: {
1.46 schwarze 1299:
1300: roffnode_push(r, tok, NULL, ln, ppos);
1.2 schwarze 1301:
1.80 schwarze 1302: /*
1.35 schwarze 1303: * An `.el' has no conditional body: it will consume the value
1304: * of the current rstack entry set in prior `ie' calls or
1.80 schwarze 1305: * defaults to DENY.
1.35 schwarze 1306: *
1307: * If we're not an `el', however, then evaluate the conditional.
1308: */
1.1 schwarze 1309:
1.110 schwarze 1310: r->last->rule = tok == ROFF_el ?
1.80 schwarze 1311: (r->rstackpos < 0 ? 0 : r->rstack[r->rstackpos--]) :
1.110 schwarze 1312: roff_evalcond(r, ln, buf->buf, &pos);
1.2 schwarze 1313:
1.35 schwarze 1314: /*
1315: * An if-else will put the NEGATION of the current evaluated
1316: * conditional into the stack of rules.
1317: */
1318:
1.110 schwarze 1319: if (tok == ROFF_ie) {
1.96 schwarze 1320: if (r->rstackpos + 1 == r->rstacksz) {
1321: r->rstacksz += 16;
1322: r->rstack = mandoc_reallocarray(r->rstack,
1323: r->rstacksz, sizeof(int));
1.35 schwarze 1324: }
1.71 schwarze 1325: r->rstack[++r->rstackpos] = !r->last->rule;
1.2 schwarze 1326: }
1.5 schwarze 1327:
1328: /* If the parent has false as its rule, then so do we. */
1329:
1.71 schwarze 1330: if (r->last->parent && !r->last->parent->rule)
1331: r->last->rule = 0;
1.5 schwarze 1332:
1333: /*
1.46 schwarze 1334: * Determine scope.
1335: * If there is nothing on the line after the conditional,
1336: * not even whitespace, use next-line scope.
1.5 schwarze 1337: */
1.2 schwarze 1338:
1.110 schwarze 1339: if (buf->buf[pos] == '\0') {
1.46 schwarze 1340: r->last->endspan = 2;
1341: goto out;
1342: }
1343:
1.110 schwarze 1344: while (buf->buf[pos] == ' ')
1.46 schwarze 1345: pos++;
1346:
1347: /* An opening brace requests multiline scope. */
1.2 schwarze 1348:
1.110 schwarze 1349: if (buf->buf[pos] == '\\' && buf->buf[pos + 1] == '{') {
1.2 schwarze 1350: r->last->endspan = -1;
1351: pos += 2;
1.46 schwarze 1352: goto out;
1.80 schwarze 1353: }
1.2 schwarze 1354:
1355: /*
1.46 schwarze 1356: * Anything else following the conditional causes
1357: * single-line scope. Warn if the scope contains
1358: * nothing but trailing whitespace.
1.2 schwarze 1359: */
1360:
1.110 schwarze 1361: if (buf->buf[pos] == '\0')
1.89 schwarze 1362: mandoc_msg(MANDOCERR_COND_EMPTY, r->parse,
1363: ln, ppos, roffs[tok].name);
1.2 schwarze 1364:
1.46 schwarze 1365: r->last->endspan = 1;
1.1 schwarze 1366:
1.46 schwarze 1367: out:
1.2 schwarze 1368: *offs = pos;
1369: return(ROFF_RERUN);
1.1 schwarze 1370: }
1371:
1.2 schwarze 1372: static enum rofferr
1.7 schwarze 1373: roff_ds(ROFF_ARGS)
1374: {
1.85 schwarze 1375: char *string;
1376: const char *name;
1377: size_t namesz;
1.10 schwarze 1378:
1379: /*
1.85 schwarze 1380: * The first word is the name of the string.
1381: * If it is empty or terminated by an escape sequence,
1382: * abort the `ds' request without defining anything.
1.10 schwarze 1383: */
1.7 schwarze 1384:
1.110 schwarze 1385: name = string = buf->buf + pos;
1386: if (*name == '\0')
1.7 schwarze 1387: return(ROFF_IGN);
1388:
1.85 schwarze 1389: namesz = roff_getname(r, &string, ln, pos);
1.110 schwarze 1390: if (name[namesz] == '\\')
1.85 schwarze 1391: return(ROFF_IGN);
1392:
1393: /* Read past the initial double-quote, if any. */
1.110 schwarze 1394: if (*string == '"')
1.7 schwarze 1395: string++;
1396:
1.10 schwarze 1397: /* The rest is the value. */
1.85 schwarze 1398: roff_setstrn(&r->strtab, name, namesz, string, strlen(string),
1399: ROFF_as == tok);
1.7 schwarze 1400: return(ROFF_IGN);
1401: }
1402:
1.77 schwarze 1403: /*
1404: * Parse a single operator, one or two characters long.
1405: * If the operator is recognized, return success and advance the
1406: * parse point, else return failure and let the parse point unchanged.
1407: */
1408: static int
1409: roff_getop(const char *v, int *pos, char *res)
1410: {
1411:
1412: *res = v[*pos];
1413:
1414: switch (*res) {
1.80 schwarze 1415: case '+':
1.77 schwarze 1416: /* FALLTHROUGH */
1.80 schwarze 1417: case '-':
1.77 schwarze 1418: /* FALLTHROUGH */
1.80 schwarze 1419: case '*':
1.77 schwarze 1420: /* FALLTHROUGH */
1.80 schwarze 1421: case '/':
1.77 schwarze 1422: /* FALLTHROUGH */
1.80 schwarze 1423: case '%':
1.77 schwarze 1424: /* FALLTHROUGH */
1.80 schwarze 1425: case '&':
1.77 schwarze 1426: /* FALLTHROUGH */
1.80 schwarze 1427: case ':':
1.77 schwarze 1428: break;
1429: case '<':
1430: switch (v[*pos + 1]) {
1.80 schwarze 1431: case '=':
1.77 schwarze 1432: *res = 'l';
1433: (*pos)++;
1434: break;
1.80 schwarze 1435: case '>':
1.77 schwarze 1436: *res = '!';
1437: (*pos)++;
1438: break;
1.80 schwarze 1439: case '?':
1.77 schwarze 1440: *res = 'i';
1441: (*pos)++;
1442: break;
1443: default:
1444: break;
1445: }
1446: break;
1447: case '>':
1448: switch (v[*pos + 1]) {
1.80 schwarze 1449: case '=':
1.77 schwarze 1450: *res = 'g';
1451: (*pos)++;
1452: break;
1.80 schwarze 1453: case '?':
1.77 schwarze 1454: *res = 'a';
1455: (*pos)++;
1456: break;
1457: default:
1458: break;
1459: }
1460: break;
1461: case '=':
1462: if ('=' == v[*pos + 1])
1463: (*pos)++;
1464: break;
1465: default:
1466: return(0);
1467: }
1468: (*pos)++;
1469:
1470: return(*res);
1471: }
1472:
1473: /*
1474: * Evaluate either a parenthesized numeric expression
1475: * or a single signed integer number.
1476: */
1477: static int
1.106 schwarze 1478: roff_evalpar(struct roff *r, int ln,
1479: const char *v, int *pos, int *res)
1.77 schwarze 1480: {
1481:
1482: if ('(' != v[*pos])
1483: return(roff_getnum(v, pos, res));
1484:
1485: (*pos)++;
1.106 schwarze 1486: if ( ! roff_evalnum(r, ln, v, pos, res, 1))
1.77 schwarze 1487: return(0);
1488:
1.79 schwarze 1489: /*
1490: * Omission of the closing parenthesis
1491: * is an error in validation mode,
1492: * but ignored in evaluation mode.
1493: */
1494:
1.77 schwarze 1495: if (')' == v[*pos])
1496: (*pos)++;
1.79 schwarze 1497: else if (NULL == res)
1498: return(0);
1.77 schwarze 1499:
1500: return(1);
1501: }
1502:
1503: /*
1504: * Evaluate a complete numeric expression.
1505: * Proceed left to right, there is no concept of precedence.
1506: */
1507: static int
1.106 schwarze 1508: roff_evalnum(struct roff *r, int ln, const char *v,
1509: int *pos, int *res, int skipwhite)
1.77 schwarze 1510: {
1511: int mypos, operand2;
1512: char operator;
1513:
1514: if (NULL == pos) {
1515: mypos = 0;
1516: pos = &mypos;
1517: }
1518:
1519: if (skipwhite)
1520: while (isspace((unsigned char)v[*pos]))
1521: (*pos)++;
1522:
1.106 schwarze 1523: if ( ! roff_evalpar(r, ln, v, pos, res))
1.77 schwarze 1524: return(0);
1525:
1526: while (1) {
1527: if (skipwhite)
1528: while (isspace((unsigned char)v[*pos]))
1529: (*pos)++;
1530:
1531: if ( ! roff_getop(v, pos, &operator))
1532: break;
1533:
1534: if (skipwhite)
1535: while (isspace((unsigned char)v[*pos]))
1536: (*pos)++;
1537:
1.106 schwarze 1538: if ( ! roff_evalpar(r, ln, v, pos, &operand2))
1.77 schwarze 1539: return(0);
1540:
1541: if (skipwhite)
1542: while (isspace((unsigned char)v[*pos]))
1543: (*pos)++;
1.79 schwarze 1544:
1545: if (NULL == res)
1546: continue;
1.77 schwarze 1547:
1548: switch (operator) {
1.80 schwarze 1549: case '+':
1.77 schwarze 1550: *res += operand2;
1551: break;
1.80 schwarze 1552: case '-':
1.77 schwarze 1553: *res -= operand2;
1554: break;
1.80 schwarze 1555: case '*':
1.77 schwarze 1556: *res *= operand2;
1557: break;
1.80 schwarze 1558: case '/':
1.106 schwarze 1559: if (0 == operand2) {
1560: mandoc_msg(MANDOCERR_DIVZERO,
1561: r->parse, ln, *pos, v);
1562: *res = 0;
1563: break;
1564: }
1.77 schwarze 1565: *res /= operand2;
1566: break;
1.80 schwarze 1567: case '%':
1.77 schwarze 1568: *res %= operand2;
1569: break;
1.80 schwarze 1570: case '<':
1.77 schwarze 1571: *res = *res < operand2;
1572: break;
1.80 schwarze 1573: case '>':
1.77 schwarze 1574: *res = *res > operand2;
1575: break;
1.80 schwarze 1576: case 'l':
1.77 schwarze 1577: *res = *res <= operand2;
1578: break;
1.80 schwarze 1579: case 'g':
1.77 schwarze 1580: *res = *res >= operand2;
1581: break;
1.80 schwarze 1582: case '=':
1.77 schwarze 1583: *res = *res == operand2;
1584: break;
1.80 schwarze 1585: case '!':
1.77 schwarze 1586: *res = *res != operand2;
1587: break;
1.80 schwarze 1588: case '&':
1.77 schwarze 1589: *res = *res && operand2;
1590: break;
1.80 schwarze 1591: case ':':
1.77 schwarze 1592: *res = *res || operand2;
1593: break;
1.80 schwarze 1594: case 'i':
1.77 schwarze 1595: if (operand2 < *res)
1596: *res = operand2;
1597: break;
1.80 schwarze 1598: case 'a':
1.77 schwarze 1599: if (operand2 > *res)
1600: *res = operand2;
1601: break;
1602: default:
1603: abort();
1604: }
1605: }
1606: return(1);
1607: }
1608:
1.52 schwarze 1609: void
1.60 schwarze 1610: roff_setreg(struct roff *r, const char *name, int val, char sign)
1.41 schwarze 1611: {
1.52 schwarze 1612: struct roffreg *reg;
1613:
1614: /* Search for an existing register with the same name. */
1615: reg = r->regtab;
1616:
1617: while (reg && strcmp(name, reg->key.p))
1618: reg = reg->next;
1.41 schwarze 1619:
1.52 schwarze 1620: if (NULL == reg) {
1621: /* Create a new register. */
1622: reg = mandoc_malloc(sizeof(struct roffreg));
1623: reg->key.p = mandoc_strdup(name);
1624: reg->key.sz = strlen(name);
1.60 schwarze 1625: reg->val = 0;
1.52 schwarze 1626: reg->next = r->regtab;
1627: r->regtab = reg;
1628: }
1629:
1.60 schwarze 1630: if ('+' == sign)
1631: reg->val += val;
1632: else if ('-' == sign)
1633: reg->val -= val;
1634: else
1635: reg->val = val;
1.41 schwarze 1636: }
1637:
1.65 schwarze 1638: /*
1639: * Handle some predefined read-only number registers.
1640: * For now, return -1 if the requested register is not predefined;
1641: * in case a predefined read-only register having the value -1
1642: * were to turn up, another special value would have to be chosen.
1643: */
1644: static int
1645: roff_getregro(const char *name)
1646: {
1647:
1648: switch (*name) {
1.80 schwarze 1649: case 'A': /* ASCII approximation mode is always off. */
1.65 schwarze 1650: return(0);
1.80 schwarze 1651: case 'g': /* Groff compatibility mode is always on. */
1.65 schwarze 1652: return(1);
1.80 schwarze 1653: case 'H': /* Fixed horizontal resolution. */
1.65 schwarze 1654: return (24);
1.80 schwarze 1655: case 'j': /* Always adjust left margin only. */
1.65 schwarze 1656: return(0);
1.80 schwarze 1657: case 'T': /* Some output device is always defined. */
1.65 schwarze 1658: return(1);
1.80 schwarze 1659: case 'V': /* Fixed vertical resolution. */
1.65 schwarze 1660: return (40);
1661: default:
1662: return (-1);
1663: }
1664: }
1665:
1.53 schwarze 1666: int
1.52 schwarze 1667: roff_getreg(const struct roff *r, const char *name)
1.41 schwarze 1668: {
1.52 schwarze 1669: struct roffreg *reg;
1.65 schwarze 1670: int val;
1671:
1672: if ('.' == name[0] && '\0' != name[1] && '\0' == name[2]) {
1673: val = roff_getregro(name + 1);
1674: if (-1 != val)
1675: return (val);
1676: }
1.52 schwarze 1677:
1678: for (reg = r->regtab; reg; reg = reg->next)
1679: if (0 == strcmp(name, reg->key.p))
1.53 schwarze 1680: return(reg->val);
1681:
1682: return(0);
1683: }
1684:
1685: static int
1686: roff_getregn(const struct roff *r, const char *name, size_t len)
1687: {
1688: struct roffreg *reg;
1.65 schwarze 1689: int val;
1690:
1691: if ('.' == name[0] && 2 == len) {
1692: val = roff_getregro(name + 1);
1693: if (-1 != val)
1694: return (val);
1695: }
1.53 schwarze 1696:
1697: for (reg = r->regtab; reg; reg = reg->next)
1698: if (len == reg->key.sz &&
1699: 0 == strncmp(name, reg->key.p, len))
1700: return(reg->val);
1.41 schwarze 1701:
1.52 schwarze 1702: return(0);
1.41 schwarze 1703: }
1704:
1.52 schwarze 1705: static void
1706: roff_freereg(struct roffreg *reg)
1.41 schwarze 1707: {
1.52 schwarze 1708: struct roffreg *old_reg;
1.41 schwarze 1709:
1.52 schwarze 1710: while (NULL != reg) {
1711: free(reg->key.p);
1712: old_reg = reg;
1713: reg = reg->next;
1714: free(old_reg);
1715: }
1.41 schwarze 1716: }
1.7 schwarze 1717:
1718: static enum rofferr
1.6 schwarze 1719: roff_nr(ROFF_ARGS)
1.1 schwarze 1720: {
1.85 schwarze 1721: char *key, *val;
1722: size_t keysz;
1.37 schwarze 1723: int iv;
1.60 schwarze 1724: char sign;
1.6 schwarze 1725:
1.110 schwarze 1726: key = val = buf->buf + pos;
1727: if (*key == '\0')
1.85 schwarze 1728: return(ROFF_IGN);
1729:
1730: keysz = roff_getname(r, &val, ln, pos);
1.110 schwarze 1731: if (key[keysz] == '\\')
1.85 schwarze 1732: return(ROFF_IGN);
1733: key[keysz] = '\0';
1.6 schwarze 1734:
1.60 schwarze 1735: sign = *val;
1.110 schwarze 1736: if (sign == '+' || sign == '-')
1.60 schwarze 1737: val++;
1738:
1.106 schwarze 1739: if (roff_evalnum(r, ln, val, NULL, &iv, 0))
1.77 schwarze 1740: roff_setreg(r, key, iv, sign);
1.1 schwarze 1741:
1.76 schwarze 1742: return(ROFF_IGN);
1743: }
1744:
1745: static enum rofferr
1746: roff_rr(ROFF_ARGS)
1747: {
1748: struct roffreg *reg, **prev;
1.85 schwarze 1749: char *name, *cp;
1750: size_t namesz;
1.76 schwarze 1751:
1.110 schwarze 1752: name = cp = buf->buf + pos;
1753: if (*name == '\0')
1.85 schwarze 1754: return(ROFF_IGN);
1755: namesz = roff_getname(r, &cp, ln, pos);
1756: name[namesz] = '\0';
1.76 schwarze 1757:
1758: prev = &r->regtab;
1759: while (1) {
1760: reg = *prev;
1.110 schwarze 1761: if (reg == NULL || !strcmp(name, reg->key.p))
1.76 schwarze 1762: break;
1763: prev = ®->next;
1764: }
1.110 schwarze 1765: if (reg != NULL) {
1.76 schwarze 1766: *prev = reg->next;
1767: free(reg->key.p);
1768: free(reg);
1769: }
1.29 schwarze 1770: return(ROFF_IGN);
1771: }
1772:
1773: static enum rofferr
1774: roff_rm(ROFF_ARGS)
1775: {
1776: const char *name;
1777: char *cp;
1.85 schwarze 1778: size_t namesz;
1.29 schwarze 1779:
1.110 schwarze 1780: cp = buf->buf + pos;
1781: while (*cp != '\0') {
1.85 schwarze 1782: name = cp;
1.110 schwarze 1783: namesz = roff_getname(r, &cp, ln, (int)(cp - buf->buf));
1.85 schwarze 1784: roff_setstrn(&r->strtab, name, namesz, NULL, 0, 0);
1.110 schwarze 1785: if (name[namesz] == '\\')
1.85 schwarze 1786: break;
1.29 schwarze 1787: }
1.51 schwarze 1788: return(ROFF_IGN);
1789: }
1790:
1791: static enum rofferr
1792: roff_it(ROFF_ARGS)
1793: {
1794: char *cp;
1795: size_t len;
1796: int iv;
1797:
1798: /* Parse the number of lines. */
1.110 schwarze 1799: cp = buf->buf + pos;
1.51 schwarze 1800: len = strcspn(cp, " \t");
1801: cp[len] = '\0';
1802: if ((iv = mandoc_strntoi(cp, len, 10)) <= 0) {
1.95 schwarze 1803: mandoc_msg(MANDOCERR_IT_NONUM, r->parse,
1.110 schwarze 1804: ln, ppos, buf->buf + 1);
1.51 schwarze 1805: return(ROFF_IGN);
1806: }
1807: cp += len + 1;
1808:
1809: /* Arm the input line trap. */
1810: roffit_lines = iv;
1811: roffit_macro = mandoc_strdup(cp);
1.2 schwarze 1812: return(ROFF_IGN);
1.47 schwarze 1813: }
1814:
1815: static enum rofferr
1816: roff_Dd(ROFF_ARGS)
1817: {
1818: const char *const *cp;
1819:
1.99 schwarze 1820: if ((r->options & (MPARSE_MDOC | MPARSE_QUICK)) == 0)
1.47 schwarze 1821: for (cp = __mdoc_reserved; *cp; cp++)
1822: roff_setstr(r, *cp, NULL, 0);
1823:
1.99 schwarze 1824: if (r->format == 0)
1825: r->format = MPARSE_MDOC;
1826:
1.47 schwarze 1827: return(ROFF_CONT);
1828: }
1829:
1830: static enum rofferr
1831: roff_TH(ROFF_ARGS)
1832: {
1833: const char *const *cp;
1834:
1.99 schwarze 1835: if ((r->options & MPARSE_QUICK) == 0)
1.47 schwarze 1836: for (cp = __man_reserved; *cp; cp++)
1837: roff_setstr(r, *cp, NULL, 0);
1838:
1.99 schwarze 1839: if (r->format == 0)
1840: r->format = MPARSE_MAN;
1841:
1.47 schwarze 1842: return(ROFF_CONT);
1.14 schwarze 1843: }
1844:
1845: static enum rofferr
1.27 schwarze 1846: roff_TE(ROFF_ARGS)
1847: {
1848:
1849: if (NULL == r->tbl)
1.94 schwarze 1850: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
1851: ln, ppos, "TE");
1.27 schwarze 1852: else
1.41 schwarze 1853: tbl_end(&r->tbl);
1.27 schwarze 1854:
1855: return(ROFF_IGN);
1856: }
1857:
1858: static enum rofferr
1859: roff_T_(ROFF_ARGS)
1860: {
1861:
1862: if (NULL == r->tbl)
1.94 schwarze 1863: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
1864: ln, ppos, "T&");
1.27 schwarze 1865: else
1866: tbl_restart(ppos, ln, r->tbl);
1867:
1868: return(ROFF_IGN);
1869: }
1870:
1.102 schwarze 1871: /*
1872: * Handle in-line equation delimiters.
1873: */
1874: static enum rofferr
1.110 schwarze 1875: roff_eqndelim(struct roff *r, struct buf *buf, int pos)
1.41 schwarze 1876: {
1.105 schwarze 1877: char *cp1, *cp2;
1878: const char *bef_pr, *bef_nl, *mac, *aft_nl, *aft_pr;
1.41 schwarze 1879:
1.102 schwarze 1880: /*
1881: * Outside equations, look for an opening delimiter.
1882: * If we are inside an equation, we already know it is
1883: * in-line, or this function wouldn't have been called;
1884: * so look for a closing delimiter.
1885: */
1886:
1.110 schwarze 1887: cp1 = buf->buf + pos;
1.102 schwarze 1888: cp2 = strchr(cp1, r->eqn == NULL ?
1889: r->last_eqn->odelim : r->last_eqn->cdelim);
1890: if (cp2 == NULL)
1891: return(ROFF_CONT);
1892:
1.105 schwarze 1893: *cp2++ = '\0';
1894: bef_pr = bef_nl = aft_nl = aft_pr = "";
1895:
1896: /* Handle preceding text, protecting whitespace. */
1897:
1.110 schwarze 1898: if (*buf->buf != '\0') {
1.105 schwarze 1899: if (r->eqn == NULL)
1900: bef_pr = "\\&";
1901: bef_nl = "\n";
1902: }
1903:
1904: /*
1905: * Prepare replacing the delimiter with an equation macro
1906: * and drop leading white space from the equation.
1907: */
1.102 schwarze 1908:
1.105 schwarze 1909: if (r->eqn == NULL) {
1910: while (*cp2 == ' ')
1911: cp2++;
1912: mac = ".EQ";
1913: } else
1914: mac = ".EN";
1915:
1916: /* Handle following text, protecting whitespace. */
1917:
1918: if (*cp2 != '\0') {
1919: aft_nl = "\n";
1920: if (r->eqn != NULL)
1921: aft_pr = "\\&";
1922: }
1923:
1924: /* Do the actual replacement. */
1925:
1.110 schwarze 1926: buf->sz = mandoc_asprintf(&cp1, "%s%s%s%s%s%s%s", buf->buf,
1.105 schwarze 1927: bef_pr, bef_nl, mac, aft_nl, aft_pr, cp2) + 1;
1.110 schwarze 1928: free(buf->buf);
1929: buf->buf = cp1;
1.102 schwarze 1930:
1931: /* Toggle the in-line state of the eqn subsystem. */
1932:
1933: r->eqn_inline = r->eqn == NULL;
1934: return(ROFF_REPARSE);
1.41 schwarze 1935: }
1936:
1.107 schwarze 1937: static enum rofferr
1938: roff_EQ(ROFF_ARGS)
1.32 schwarze 1939: {
1.41 schwarze 1940: struct eqn_node *e;
1.32 schwarze 1941:
1.110 schwarze 1942: assert(r->eqn == NULL);
1.107 schwarze 1943: e = eqn_alloc(ppos, ln, r->parse);
1.32 schwarze 1944:
1.102 schwarze 1945: if (r->last_eqn) {
1.32 schwarze 1946: r->last_eqn->next = e;
1.102 schwarze 1947: e->delim = r->last_eqn->delim;
1948: e->odelim = r->last_eqn->odelim;
1949: e->cdelim = r->last_eqn->cdelim;
1950: } else
1.32 schwarze 1951: r->first_eqn = r->last_eqn = e;
1952:
1953: r->eqn = r->last_eqn = e;
1.41 schwarze 1954:
1.110 schwarze 1955: if (buf->buf[pos] != '\0')
1.107 schwarze 1956: mandoc_vmsg(MANDOCERR_ARG_SKIP, r->parse, ln, pos,
1.110 schwarze 1957: ".EQ %s", buf->buf + pos);
1.41 schwarze 1958:
1.32 schwarze 1959: return(ROFF_IGN);
1960: }
1961:
1962: static enum rofferr
1963: roff_EN(ROFF_ARGS)
1964: {
1965:
1.94 schwarze 1966: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse, ln, ppos, "EN");
1.32 schwarze 1967: return(ROFF_IGN);
1968: }
1969:
1970: static enum rofferr
1.27 schwarze 1971: roff_TS(ROFF_ARGS)
1972: {
1.49 schwarze 1973: struct tbl_node *tbl;
1.27 schwarze 1974:
1975: if (r->tbl) {
1.94 schwarze 1976: mandoc_msg(MANDOCERR_BLK_BROKEN, r->parse,
1977: ln, ppos, "TS breaks TS");
1.41 schwarze 1978: tbl_end(&r->tbl);
1.27 schwarze 1979: }
1980:
1.49 schwarze 1981: tbl = tbl_alloc(ppos, ln, r->parse);
1.27 schwarze 1982:
1983: if (r->last_tbl)
1.49 schwarze 1984: r->last_tbl->next = tbl;
1.27 schwarze 1985: else
1.49 schwarze 1986: r->first_tbl = r->last_tbl = tbl;
1.27 schwarze 1987:
1.49 schwarze 1988: r->tbl = r->last_tbl = tbl;
1.27 schwarze 1989: return(ROFF_IGN);
1990: }
1991:
1992: static enum rofferr
1.48 schwarze 1993: roff_cc(ROFF_ARGS)
1994: {
1995: const char *p;
1996:
1.110 schwarze 1997: p = buf->buf + pos;
1.48 schwarze 1998:
1.110 schwarze 1999: if (*p == '\0' || (r->control = *p++) == '.')
1.48 schwarze 2000: r->control = 0;
2001:
1.110 schwarze 2002: if (*p != '\0')
1.48 schwarze 2003: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
2004:
2005: return(ROFF_IGN);
2006: }
2007:
2008: static enum rofferr
1.42 schwarze 2009: roff_tr(ROFF_ARGS)
2010: {
2011: const char *p, *first, *second;
2012: size_t fsz, ssz;
2013: enum mandoc_esc esc;
2014:
1.110 schwarze 2015: p = buf->buf + pos;
1.42 schwarze 2016:
1.110 schwarze 2017: if (*p == '\0') {
1.42 schwarze 2018: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
2019: return(ROFF_IGN);
2020: }
2021:
1.110 schwarze 2022: while (*p != '\0') {
1.42 schwarze 2023: fsz = ssz = 1;
2024:
2025: first = p++;
1.110 schwarze 2026: if (*first == '\\') {
1.42 schwarze 2027: esc = mandoc_escape(&p, NULL, NULL);
1.110 schwarze 2028: if (esc == ESCAPE_ERROR) {
1.92 schwarze 2029: mandoc_msg(MANDOCERR_ESC_BAD, r->parse,
1.110 schwarze 2030: ln, (int)(p - buf->buf), first);
1.42 schwarze 2031: return(ROFF_IGN);
2032: }
2033: fsz = (size_t)(p - first);
2034: }
2035:
2036: second = p++;
1.110 schwarze 2037: if (*second == '\\') {
1.42 schwarze 2038: esc = mandoc_escape(&p, NULL, NULL);
1.110 schwarze 2039: if (esc == ESCAPE_ERROR) {
1.92 schwarze 2040: mandoc_msg(MANDOCERR_ESC_BAD, r->parse,
1.110 schwarze 2041: ln, (int)(p - buf->buf), second);
1.42 schwarze 2042: return(ROFF_IGN);
2043: }
2044: ssz = (size_t)(p - second);
1.110 schwarze 2045: } else if (*second == '\0') {
1.80 schwarze 2046: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse,
1.110 schwarze 2047: ln, (int)(p - buf->buf), NULL);
1.42 schwarze 2048: second = " ";
2049: p--;
2050: }
2051:
2052: if (fsz > 1) {
1.80 schwarze 2053: roff_setstrn(&r->xmbtab, first, fsz,
2054: second, ssz, 0);
1.42 schwarze 2055: continue;
2056: }
2057:
1.110 schwarze 2058: if (r->xtab == NULL)
1.80 schwarze 2059: r->xtab = mandoc_calloc(128,
2060: sizeof(struct roffstr));
1.42 schwarze 2061:
2062: free(r->xtab[(int)*first].p);
2063: r->xtab[(int)*first].p = mandoc_strndup(second, ssz);
2064: r->xtab[(int)*first].sz = ssz;
2065: }
2066:
2067: return(ROFF_IGN);
2068: }
2069:
2070: static enum rofferr
1.14 schwarze 2071: roff_so(ROFF_ARGS)
2072: {
2073: char *name;
1.15 schwarze 2074:
1.110 schwarze 2075: name = buf->buf + pos;
1.97 schwarze 2076: mandoc_vmsg(MANDOCERR_SO, r->parse, ln, ppos, "so %s", name);
1.14 schwarze 2077:
1.22 schwarze 2078: /*
2079: * Handle `so'. Be EXTREMELY careful, as we shouldn't be
2080: * opening anything that's not in our cwd or anything beneath
2081: * it. Thus, explicitly disallow traversing up the file-system
2082: * or using absolute paths.
2083: */
2084:
1.110 schwarze 2085: if (*name == '/' || strstr(name, "../") || strstr(name, "/..")) {
1.83 schwarze 2086: mandoc_vmsg(MANDOCERR_SO_PATH, r->parse, ln, ppos,
2087: ".so %s", name);
1.14 schwarze 2088: return(ROFF_ERR);
2089: }
2090:
2091: *offs = pos;
2092: return(ROFF_SO);
1.7 schwarze 2093: }
2094:
1.16 schwarze 2095: static enum rofferr
2096: roff_userdef(ROFF_ARGS)
1.12 schwarze 2097: {
1.16 schwarze 2098: const char *arg[9];
2099: char *cp, *n1, *n2;
1.25 schwarze 2100: int i;
1.12 schwarze 2101:
1.16 schwarze 2102: /*
2103: * Collect pointers to macro argument strings
1.61 schwarze 2104: * and NUL-terminate them.
1.16 schwarze 2105: */
1.110 schwarze 2106: cp = buf->buf + pos;
1.25 schwarze 2107: for (i = 0; i < 9; i++)
1.110 schwarze 2108: arg[i] = *cp == '\0' ? "" :
1.35 schwarze 2109: mandoc_getarg(r->parse, &cp, ln, &pos);
1.16 schwarze 2110:
2111: /*
2112: * Expand macro arguments.
1.12 schwarze 2113: */
1.110 schwarze 2114: buf->sz = 0;
1.16 schwarze 2115: n1 = cp = mandoc_strdup(r->current_string);
1.110 schwarze 2116: while ((cp = strstr(cp, "\\$")) != NULL) {
1.16 schwarze 2117: i = cp[2] - '1';
2118: if (0 > i || 8 < i) {
2119: /* Not an argument invocation. */
2120: cp += 2;
2121: continue;
2122: }
1.82 schwarze 2123: *cp = '\0';
1.110 schwarze 2124: buf->sz = mandoc_asprintf(&n2, "%s%s%s",
1.82 schwarze 2125: n1, arg[i], cp + 3) + 1;
1.16 schwarze 2126: cp = n2 + (cp - n1);
2127: free(n1);
2128: n1 = n2;
1.12 schwarze 2129: }
2130:
1.16 schwarze 2131: /*
2132: * Replace the macro invocation
2133: * by the expanded macro.
2134: */
1.110 schwarze 2135: free(buf->buf);
2136: buf->buf = n1;
2137: if (buf->sz == 0)
2138: buf->sz = strlen(buf->buf) + 1;
1.16 schwarze 2139:
1.110 schwarze 2140: return(buf->sz > 1 && buf->buf[buf->sz - 2] == '\n' ?
1.16 schwarze 2141: ROFF_REPARSE : ROFF_APPEND);
1.12 schwarze 2142: }
1.28 schwarze 2143:
1.85 schwarze 2144: static size_t
1.28 schwarze 2145: roff_getname(struct roff *r, char **cpp, int ln, int pos)
2146: {
2147: char *name, *cp;
1.85 schwarze 2148: size_t namesz;
1.28 schwarze 2149:
2150: name = *cpp;
2151: if ('\0' == *name)
1.85 schwarze 2152: return(0);
1.28 schwarze 2153:
1.85 schwarze 2154: /* Read until end of name and terminate it with NUL. */
2155: for (cp = name; 1; cp++) {
2156: if ('\0' == *cp || ' ' == *cp) {
2157: namesz = cp - name;
2158: break;
2159: }
1.28 schwarze 2160: if ('\\' != *cp)
2161: continue;
1.88 schwarze 2162: namesz = cp - name;
2163: if ('{' == cp[1] || '}' == cp[1])
2164: break;
1.28 schwarze 2165: cp++;
2166: if ('\\' == *cp)
2167: continue;
1.97 schwarze 2168: mandoc_vmsg(MANDOCERR_NAMESC, r->parse, ln, pos,
2169: "%.*s", (int)(cp - name + 1), name);
1.85 schwarze 2170: mandoc_escape((const char **)&cp, NULL, NULL);
2171: break;
1.28 schwarze 2172: }
2173:
2174: /* Read past spaces. */
2175: while (' ' == *cp)
2176: cp++;
2177:
2178: *cpp = cp;
1.85 schwarze 2179: return(namesz);
1.28 schwarze 2180: }
2181:
1.16 schwarze 2182: /*
2183: * Store *string into the user-defined string called *name.
2184: * To clear an existing entry, call with (*r, *name, NULL, 0).
1.66 schwarze 2185: * append == 0: replace mode
2186: * append == 1: single-line append mode
2187: * append == 2: multiline append mode, append '\n' after each call
1.16 schwarze 2188: */
1.8 schwarze 2189: static void
1.16 schwarze 2190: roff_setstr(struct roff *r, const char *name, const char *string,
1.66 schwarze 2191: int append)
1.7 schwarze 2192: {
1.42 schwarze 2193:
2194: roff_setstrn(&r->strtab, name, strlen(name), string,
1.80 schwarze 2195: string ? strlen(string) : 0, append);
1.42 schwarze 2196: }
2197:
2198: static void
2199: roff_setstrn(struct roffkv **r, const char *name, size_t namesz,
1.66 schwarze 2200: const char *string, size_t stringsz, int append)
1.42 schwarze 2201: {
2202: struct roffkv *n;
2203: char *c;
2204: int i;
2205: size_t oldch, newch;
1.7 schwarze 2206:
1.16 schwarze 2207: /* Search for an existing string with the same name. */
1.42 schwarze 2208: n = *r;
2209:
1.84 schwarze 2210: while (n && (namesz != n->key.sz ||
2211: strncmp(n->key.p, name, namesz)))
1.7 schwarze 2212: n = n->next;
1.8 schwarze 2213:
2214: if (NULL == n) {
1.16 schwarze 2215: /* Create a new string table entry. */
1.42 schwarze 2216: n = mandoc_malloc(sizeof(struct roffkv));
2217: n->key.p = mandoc_strndup(name, namesz);
2218: n->key.sz = namesz;
2219: n->val.p = NULL;
2220: n->val.sz = 0;
2221: n->next = *r;
2222: *r = n;
1.66 schwarze 2223: } else if (0 == append) {
1.42 schwarze 2224: free(n->val.p);
2225: n->val.p = NULL;
2226: n->val.sz = 0;
1.16 schwarze 2227: }
2228:
2229: if (NULL == string)
2230: return;
2231:
2232: /*
2233: * One additional byte for the '\n' in multiline mode,
2234: * and one for the terminating '\0'.
2235: */
1.66 schwarze 2236: newch = stringsz + (1 < append ? 2u : 1u);
1.42 schwarze 2237:
2238: if (NULL == n->val.p) {
2239: n->val.p = mandoc_malloc(newch);
2240: *n->val.p = '\0';
1.16 schwarze 2241: oldch = 0;
2242: } else {
1.42 schwarze 2243: oldch = n->val.sz;
2244: n->val.p = mandoc_realloc(n->val.p, oldch + newch);
1.16 schwarze 2245: }
2246:
2247: /* Skip existing content in the destination buffer. */
1.42 schwarze 2248: c = n->val.p + (int)oldch;
1.16 schwarze 2249:
2250: /* Append new content to the destination buffer. */
1.42 schwarze 2251: i = 0;
2252: while (i < (int)stringsz) {
1.16 schwarze 2253: /*
2254: * Rudimentary roff copy mode:
2255: * Handle escaped backslashes.
2256: */
1.42 schwarze 2257: if ('\\' == string[i] && '\\' == string[i + 1])
2258: i++;
2259: *c++ = string[i++];
1.16 schwarze 2260: }
1.8 schwarze 2261:
1.16 schwarze 2262: /* Append terminating bytes. */
1.66 schwarze 2263: if (1 < append)
1.16 schwarze 2264: *c++ = '\n';
1.42 schwarze 2265:
1.16 schwarze 2266: *c = '\0';
1.42 schwarze 2267: n->val.sz = (int)(c - n->val.p);
1.7 schwarze 2268: }
2269:
1.8 schwarze 2270: static const char *
2271: roff_getstrn(const struct roff *r, const char *name, size_t len)
1.7 schwarze 2272: {
1.42 schwarze 2273: const struct roffkv *n;
1.64 schwarze 2274: int i;
1.7 schwarze 2275:
1.42 schwarze 2276: for (n = r->strtab; n; n = n->next)
1.80 schwarze 2277: if (0 == strncmp(name, n->key.p, len) &&
2278: '\0' == n->key.p[(int)len])
1.42 schwarze 2279: return(n->val.p);
1.64 schwarze 2280:
2281: for (i = 0; i < PREDEFS_MAX; i++)
2282: if (0 == strncmp(name, predefs[i].name, len) &&
2283: '\0' == predefs[i].name[(int)len])
2284: return(predefs[i].str);
1.8 schwarze 2285:
1.42 schwarze 2286: return(NULL);
1.7 schwarze 2287: }
2288:
1.8 schwarze 2289: static void
1.42 schwarze 2290: roff_freestr(struct roffkv *r)
1.7 schwarze 2291: {
1.42 schwarze 2292: struct roffkv *n, *nn;
1.7 schwarze 2293:
1.42 schwarze 2294: for (n = r; n; n = nn) {
2295: free(n->key.p);
2296: free(n->val.p);
1.7 schwarze 2297: nn = n->next;
2298: free(n);
2299: }
1.27 schwarze 2300: }
2301:
2302: const struct tbl_span *
2303: roff_span(const struct roff *r)
2304: {
1.80 schwarze 2305:
1.27 schwarze 2306: return(r->tbl ? tbl_span(r->tbl) : NULL);
1.32 schwarze 2307: }
2308:
2309: const struct eqn *
2310: roff_eqn(const struct roff *r)
2311: {
1.80 schwarze 2312:
1.32 schwarze 2313: return(r->last_eqn ? &r->last_eqn->eqn : NULL);
1.42 schwarze 2314: }
2315:
2316: /*
2317: * Duplicate an input string, making the appropriate character
2318: * conversations (as stipulated by `tr') along the way.
2319: * Returns a heap-allocated string with all the replacements made.
2320: */
2321: char *
2322: roff_strdup(const struct roff *r, const char *p)
2323: {
2324: const struct roffkv *cp;
2325: char *res;
2326: const char *pp;
2327: size_t ssz, sz;
2328: enum mandoc_esc esc;
2329:
2330: if (NULL == r->xmbtab && NULL == r->xtab)
2331: return(mandoc_strdup(p));
2332: else if ('\0' == *p)
2333: return(mandoc_strdup(""));
2334:
2335: /*
2336: * Step through each character looking for term matches
2337: * (remember that a `tr' can be invoked with an escape, which is
2338: * a glyph but the escape is multi-character).
2339: * We only do this if the character hash has been initialised
2340: * and the string is >0 length.
2341: */
2342:
2343: res = NULL;
2344: ssz = 0;
2345:
2346: while ('\0' != *p) {
2347: if ('\\' != *p && r->xtab && r->xtab[(int)*p].p) {
2348: sz = r->xtab[(int)*p].sz;
2349: res = mandoc_realloc(res, ssz + sz + 1);
2350: memcpy(res + ssz, r->xtab[(int)*p].p, sz);
2351: ssz += sz;
2352: p++;
2353: continue;
2354: } else if ('\\' != *p) {
2355: res = mandoc_realloc(res, ssz + 2);
2356: res[ssz++] = *p++;
2357: continue;
2358: }
2359:
2360: /* Search for term matches. */
2361: for (cp = r->xmbtab; cp; cp = cp->next)
2362: if (0 == strncmp(p, cp->key.p, cp->key.sz))
2363: break;
2364:
2365: if (NULL != cp) {
2366: /*
2367: * A match has been found.
2368: * Append the match to the array and move
2369: * forward by its keysize.
2370: */
1.80 schwarze 2371: res = mandoc_realloc(res,
2372: ssz + cp->val.sz + 1);
1.42 schwarze 2373: memcpy(res + ssz, cp->val.p, cp->val.sz);
2374: ssz += cp->val.sz;
2375: p += (int)cp->key.sz;
2376: continue;
2377: }
2378:
2379: /*
2380: * Handle escapes carefully: we need to copy
2381: * over just the escape itself, or else we might
2382: * do replacements within the escape itself.
2383: * Make sure to pass along the bogus string.
2384: */
2385: pp = p++;
2386: esc = mandoc_escape(&p, NULL, NULL);
2387: if (ESCAPE_ERROR == esc) {
2388: sz = strlen(pp);
2389: res = mandoc_realloc(res, ssz + sz + 1);
2390: memcpy(res + ssz, pp, sz);
2391: break;
2392: }
1.80 schwarze 2393: /*
2394: * We bail out on bad escapes.
1.42 schwarze 2395: * No need to warn: we already did so when
2396: * roff_res() was called.
2397: */
2398: sz = (int)(p - pp);
2399: res = mandoc_realloc(res, ssz + sz + 1);
2400: memcpy(res + ssz, pp, sz);
2401: ssz += sz;
2402: }
2403:
2404: res[(int)ssz] = '\0';
2405: return(res);
1.99 schwarze 2406: }
2407:
2408: int
2409: roff_getformat(const struct roff *r)
2410: {
2411:
2412: return(r->format);
1.48 schwarze 2413: }
2414:
2415: /*
1.80 schwarze 2416: * Find out whether a line is a macro line or not.
1.48 schwarze 2417: * If it is, adjust the current position and return one; if it isn't,
2418: * return zero and don't change the current position.
2419: * If the control character has been set with `.cc', then let that grain
2420: * precedence.
2421: * This is slighly contrary to groff, where using the non-breaking
2422: * control character when `cc' has been invoked will cause the
2423: * non-breaking macro contents to be printed verbatim.
2424: */
2425: int
2426: roff_getcontrol(const struct roff *r, const char *cp, int *ppos)
2427: {
2428: int pos;
2429:
2430: pos = *ppos;
2431:
2432: if (0 != r->control && cp[pos] == r->control)
2433: pos++;
2434: else if (0 != r->control)
2435: return(0);
2436: else if ('\\' == cp[pos] && '.' == cp[pos + 1])
2437: pos += 2;
2438: else if ('.' == cp[pos] || '\'' == cp[pos])
2439: pos++;
2440: else
2441: return(0);
2442:
2443: while (' ' == cp[pos] || '\t' == cp[pos])
2444: pos++;
2445:
2446: *ppos = pos;
2447: return(1);
1.1 schwarze 2448: }
![[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