Annotation of src/usr.bin/mandoc/roff.c, Revision 1.121
1.121 ! schwarze 1: /* $OpenBSD: roff.c,v 1.120 2015/01/07 12:19:17 schwarze Exp $ */
1.1 schwarze 2: /*
1.119 schwarze 3: * Copyright (c) 2010, 2011, 2012, 2014 Kristaps Dzonsons <kristaps@bsd.lv>
4: * Copyright (c) 2010-2015 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>
1.117 schwarze 22: #include <limits.h>
1.51 schwarze 23: #include <stdio.h>
1.1 schwarze 24: #include <stdlib.h>
25: #include <string.h>
26:
27: #include "mandoc.h"
1.74 schwarze 28: #include "mandoc_aux.h"
1.98 daniel 29: #include "libmandoc.h"
1.27 schwarze 30: #include "libroff.h"
1.1 schwarze 31:
1.37 schwarze 32: /* Maximum number of nested if-else conditionals. */
1.2 schwarze 33: #define RSTACK_MAX 128
34:
1.43 schwarze 35: /* Maximum number of string expansions per line, to break infinite loops. */
36: #define EXPAND_LIMIT 1000
37:
1.1 schwarze 38: enum rofft {
1.20 schwarze 39: ROFF_ad,
1.2 schwarze 40: ROFF_am,
41: ROFF_ami,
42: ROFF_am1,
1.66 schwarze 43: ROFF_as,
1.48 schwarze 44: ROFF_cc,
1.67 schwarze 45: ROFF_ce,
1.1 schwarze 46: ROFF_de,
47: ROFF_dei,
1.2 schwarze 48: ROFF_de1,
49: ROFF_ds,
50: ROFF_el,
1.58 schwarze 51: ROFF_fam,
1.59 schwarze 52: ROFF_hw,
1.20 schwarze 53: ROFF_hy,
1.2 schwarze 54: ROFF_ie,
55: ROFF_if,
1.1 schwarze 56: ROFF_ig,
1.30 schwarze 57: ROFF_it,
1.20 schwarze 58: ROFF_ne,
59: ROFF_nh,
1.14 schwarze 60: ROFF_nr,
1.31 schwarze 61: ROFF_ns,
1.101 schwarze 62: ROFF_pl,
1.31 schwarze 63: ROFF_ps,
1.2 schwarze 64: ROFF_rm,
1.76 schwarze 65: ROFF_rr,
1.14 schwarze 66: ROFF_so,
1.31 schwarze 67: ROFF_ta,
1.2 schwarze 68: ROFF_tr,
1.47 schwarze 69: ROFF_Dd,
70: ROFF_TH,
1.27 schwarze 71: ROFF_TS,
72: ROFF_TE,
73: ROFF_T_,
1.32 schwarze 74: ROFF_EQ,
75: ROFF_EN,
1.2 schwarze 76: ROFF_cblock,
1.16 schwarze 77: ROFF_USERDEF,
1.1 schwarze 78: ROFF_MAX
79: };
80:
1.41 schwarze 81: /*
1.42 schwarze 82: * An incredibly-simple string buffer.
83: */
1.8 schwarze 84: struct roffstr {
1.42 schwarze 85: char *p; /* nil-terminated buffer */
86: size_t sz; /* saved strlen(p) */
87: };
88:
89: /*
90: * A key-value roffstr pair as part of a singly-linked list.
91: */
92: struct roffkv {
93: struct roffstr key;
94: struct roffstr val;
95: struct roffkv *next; /* next in list */
1.8 schwarze 96: };
97:
1.52 schwarze 98: /*
99: * A single number register as part of a singly-linked list.
100: */
101: struct roffreg {
102: struct roffstr key;
1.53 schwarze 103: int val;
1.52 schwarze 104: struct roffreg *next;
105: };
106:
1.1 schwarze 107: struct roff {
1.35 schwarze 108: struct mparse *parse; /* parse point */
1.109 schwarze 109: const struct mchars *mchars; /* character table */
1.1 schwarze 110: struct roffnode *last; /* leaf of stack */
1.96 schwarze 111: int *rstack; /* stack of inverted `ie' values */
1.52 schwarze 112: struct roffreg *regtab; /* number registers */
1.42 schwarze 113: struct roffkv *strtab; /* user-defined strings & macros */
114: struct roffkv *xmbtab; /* multi-byte trans table (`tr') */
115: struct roffstr *xtab; /* single-byte trans table (`tr') */
1.16 schwarze 116: const char *current_string; /* value of last called user macro */
1.27 schwarze 117: struct tbl_node *first_tbl; /* first table parsed */
118: struct tbl_node *last_tbl; /* last table parsed */
119: struct tbl_node *tbl; /* current table being parsed */
1.32 schwarze 120: struct eqn_node *last_eqn; /* last equation parsed */
121: struct eqn_node *first_eqn; /* first equation parsed */
122: struct eqn_node *eqn; /* current equation being parsed */
1.102 schwarze 123: int eqn_inline; /* current equation is inline */
1.96 schwarze 124: int options; /* parse options */
125: int rstacksz; /* current size limit of rstack */
126: int rstackpos; /* position in rstack */
1.99 schwarze 127: int format; /* current file in mdoc or man format */
1.96 schwarze 128: char control; /* control character */
1.1 schwarze 129: };
130:
131: struct roffnode {
132: enum rofft tok; /* type of node */
133: struct roffnode *parent; /* up one in stack */
134: int line; /* parse line */
135: int col; /* parse col */
1.16 schwarze 136: char *name; /* node name, e.g. macro name */
1.2 schwarze 137: char *end; /* end-rules: custom token */
138: int endspan; /* end-rules: next-line or infty */
1.71 schwarze 139: int rule; /* current evaluation rule */
1.1 schwarze 140: };
141:
142: #define ROFF_ARGS struct roff *r, /* parse ctx */ \
143: enum rofft tok, /* tok of macro */ \
1.110 schwarze 144: struct buf *buf, /* input buffer */ \
1.1 schwarze 145: int ln, /* parse line */ \
1.2 schwarze 146: int ppos, /* original pos in buffer */ \
147: int pos, /* current pos in buffer */ \
148: int *offs /* reset offset of buffer data */
1.1 schwarze 149:
150: typedef enum rofferr (*roffproc)(ROFF_ARGS);
151:
152: struct roffmac {
153: const char *name; /* macro name */
1.2 schwarze 154: roffproc proc; /* process new macro */
155: roffproc text; /* process as child text of macro */
156: roffproc sub; /* process as child of macro */
157: int flags;
158: #define ROFFMAC_STRUCT (1 << 0) /* always interpret */
1.3 schwarze 159: struct roffmac *next;
1.1 schwarze 160: };
161:
1.37 schwarze 162: struct predef {
163: const char *name; /* predefined input name */
164: const char *str; /* replacement symbol */
165: };
166:
167: #define PREDEF(__name, __str) \
168: { (__name), (__str) },
169:
1.42 schwarze 170: static enum rofft roffhash_find(const char *, size_t);
171: static void roffhash_init(void);
172: static void roffnode_cleanscope(struct roff *);
173: static void roffnode_pop(struct roff *);
174: static void roffnode_push(struct roff *, enum rofft,
175: const char *, int, int);
1.2 schwarze 176: static enum rofferr roff_block(ROFF_ARGS);
177: static enum rofferr roff_block_text(ROFF_ARGS);
178: static enum rofferr roff_block_sub(ROFF_ARGS);
179: static enum rofferr roff_cblock(ROFF_ARGS);
1.48 schwarze 180: static enum rofferr roff_cc(ROFF_ARGS);
1.68 schwarze 181: static void roff_ccond(struct roff *, int, int);
1.2 schwarze 182: static enum rofferr roff_cond(ROFF_ARGS);
183: static enum rofferr roff_cond_text(ROFF_ARGS);
184: static enum rofferr roff_cond_sub(ROFF_ARGS);
1.7 schwarze 185: static enum rofferr roff_ds(ROFF_ARGS);
1.110 schwarze 186: static enum rofferr roff_eqndelim(struct roff *, struct buf *, int);
1.106 schwarze 187: static int roff_evalcond(struct roff *r, int,
188: const char *, int *);
189: static int roff_evalnum(struct roff *, int,
190: const char *, int *, int *, int);
191: static int roff_evalpar(struct roff *, int,
192: const char *, int *, int *);
1.71 schwarze 193: static int roff_evalstrcond(const char *, int *);
1.42 schwarze 194: static void roff_free1(struct roff *);
1.52 schwarze 195: static void roff_freereg(struct roffreg *);
1.42 schwarze 196: static void roff_freestr(struct roffkv *);
1.85 schwarze 197: static size_t roff_getname(struct roff *, char **, int, int);
1.56 schwarze 198: static int roff_getnum(const char *, int *, int *);
199: static int roff_getop(const char *, int *, char *);
1.53 schwarze 200: static int roff_getregn(const struct roff *,
201: const char *, size_t);
1.65 schwarze 202: static int roff_getregro(const char *name);
1.80 schwarze 203: static const char *roff_getstrn(const struct roff *,
1.8 schwarze 204: const char *, size_t);
1.51 schwarze 205: static enum rofferr roff_it(ROFF_ARGS);
1.21 schwarze 206: static enum rofferr roff_line_ignore(ROFF_ARGS);
1.6 schwarze 207: static enum rofferr roff_nr(ROFF_ARGS);
1.87 schwarze 208: static enum rofft roff_parse(struct roff *, char *, int *,
209: int, int);
1.110 schwarze 210: static enum rofferr roff_parsetext(struct buf *, int, int *);
211: static enum rofferr roff_res(struct roff *, struct buf *, int, int);
1.29 schwarze 212: static enum rofferr roff_rm(ROFF_ARGS);
1.76 schwarze 213: static enum rofferr roff_rr(ROFF_ARGS);
1.8 schwarze 214: static void roff_setstr(struct roff *,
1.16 schwarze 215: const char *, const char *, int);
1.80 schwarze 216: static void roff_setstrn(struct roffkv **, const char *,
1.42 schwarze 217: size_t, const char *, size_t, int);
1.14 schwarze 218: static enum rofferr roff_so(ROFF_ARGS);
1.42 schwarze 219: static enum rofferr roff_tr(ROFF_ARGS);
1.47 schwarze 220: static enum rofferr roff_Dd(ROFF_ARGS);
221: static enum rofferr roff_TH(ROFF_ARGS);
1.27 schwarze 222: static enum rofferr roff_TE(ROFF_ARGS);
223: static enum rofferr roff_TS(ROFF_ARGS);
1.32 schwarze 224: static enum rofferr roff_EQ(ROFF_ARGS);
225: static enum rofferr roff_EN(ROFF_ARGS);
1.27 schwarze 226: static enum rofferr roff_T_(ROFF_ARGS);
1.16 schwarze 227: static enum rofferr roff_userdef(ROFF_ARGS);
1.1 schwarze 228:
1.42 schwarze 229: /* See roffhash_find() */
1.3 schwarze 230:
231: #define ASCII_HI 126
232: #define ASCII_LO 33
233: #define HASHWIDTH (ASCII_HI - ASCII_LO + 1)
234:
235: static struct roffmac *hash[HASHWIDTH];
236:
237: static struct roffmac roffs[ROFF_MAX] = {
1.21 schwarze 238: { "ad", roff_line_ignore, NULL, NULL, 0, NULL },
1.3 schwarze 239: { "am", roff_block, roff_block_text, roff_block_sub, 0, NULL },
240: { "ami", roff_block, roff_block_text, roff_block_sub, 0, NULL },
241: { "am1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.66 schwarze 242: { "as", roff_ds, NULL, NULL, 0, NULL },
1.48 schwarze 243: { "cc", roff_cc, NULL, NULL, 0, NULL },
1.67 schwarze 244: { "ce", roff_line_ignore, NULL, NULL, 0, NULL },
1.3 schwarze 245: { "de", roff_block, roff_block_text, roff_block_sub, 0, NULL },
246: { "dei", roff_block, roff_block_text, roff_block_sub, 0, NULL },
247: { "de1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.7 schwarze 248: { "ds", roff_ds, NULL, NULL, 0, NULL },
1.3 schwarze 249: { "el", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
1.58 schwarze 250: { "fam", roff_line_ignore, NULL, NULL, 0, NULL },
1.59 schwarze 251: { "hw", roff_line_ignore, NULL, NULL, 0, NULL },
1.21 schwarze 252: { "hy", roff_line_ignore, NULL, NULL, 0, NULL },
1.3 schwarze 253: { "ie", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
254: { "if", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
255: { "ig", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.51 schwarze 256: { "it", roff_it, NULL, NULL, 0, NULL },
1.21 schwarze 257: { "ne", roff_line_ignore, NULL, NULL, 0, NULL },
258: { "nh", roff_line_ignore, NULL, NULL, 0, NULL },
1.14 schwarze 259: { "nr", roff_nr, NULL, NULL, 0, NULL },
1.31 schwarze 260: { "ns", roff_line_ignore, NULL, NULL, 0, NULL },
1.101 schwarze 261: { "pl", roff_line_ignore, NULL, NULL, 0, NULL },
1.31 schwarze 262: { "ps", roff_line_ignore, NULL, NULL, 0, NULL },
1.29 schwarze 263: { "rm", roff_rm, NULL, NULL, 0, NULL },
1.76 schwarze 264: { "rr", roff_rr, NULL, NULL, 0, NULL },
1.14 schwarze 265: { "so", roff_so, NULL, NULL, 0, NULL },
1.31 schwarze 266: { "ta", roff_line_ignore, NULL, NULL, 0, NULL },
1.42 schwarze 267: { "tr", roff_tr, NULL, NULL, 0, NULL },
1.47 schwarze 268: { "Dd", roff_Dd, NULL, NULL, 0, NULL },
269: { "TH", roff_TH, NULL, NULL, 0, NULL },
1.27 schwarze 270: { "TS", roff_TS, NULL, NULL, 0, NULL },
271: { "TE", roff_TE, NULL, NULL, 0, NULL },
272: { "T&", roff_T_, NULL, NULL, 0, NULL },
1.32 schwarze 273: { "EQ", roff_EQ, NULL, NULL, 0, NULL },
274: { "EN", roff_EN, NULL, NULL, 0, NULL },
1.3 schwarze 275: { ".", roff_cblock, NULL, NULL, 0, NULL },
1.16 schwarze 276: { NULL, roff_userdef, NULL, NULL, 0, NULL },
1.1 schwarze 277: };
278:
1.73 schwarze 279: /* not currently implemented: Ds em Eq LP Me PP pp Or Rd Sf SH */
1.47 schwarze 280: const char *const __mdoc_reserved[] = {
281: "Ac", "Ad", "An", "Ao", "Ap", "Aq", "Ar", "At",
282: "Bc", "Bd", "Bf", "Bk", "Bl", "Bo", "Bq",
283: "Brc", "Bro", "Brq", "Bsx", "Bt", "Bx",
284: "Cd", "Cm", "Db", "Dc", "Dd", "Dl", "Do", "Dq",
1.73 schwarze 285: "Dt", "Dv", "Dx", "D1",
286: "Ec", "Ed", "Ef", "Ek", "El", "Em",
287: "En", "Eo", "Er", "Es", "Ev", "Ex",
1.47 schwarze 288: "Fa", "Fc", "Fd", "Fl", "Fn", "Fo", "Fr", "Ft", "Fx",
1.73 schwarze 289: "Hf", "Ic", "In", "It", "Lb", "Li", "Lk", "Lp",
290: "Ms", "Mt", "Nd", "Nm", "No", "Ns", "Nx",
1.47 schwarze 291: "Oc", "Oo", "Op", "Os", "Ot", "Ox",
1.73 schwarze 292: "Pa", "Pc", "Pf", "Po", "Pp", "Pq",
293: "Qc", "Ql", "Qo", "Qq", "Re", "Rs", "Rv",
294: "Sc", "Sh", "Sm", "So", "Sq",
1.47 schwarze 295: "Ss", "St", "Sx", "Sy",
296: "Ta", "Tn", "Ud", "Ux", "Va", "Vt", "Xc", "Xo", "Xr",
1.73 schwarze 297: "%A", "%B", "%C", "%D", "%I", "%J", "%N", "%O",
1.47 schwarze 298: "%P", "%Q", "%R", "%T", "%U", "%V",
299: NULL
300: };
301:
1.73 schwarze 302: /* not currently implemented: BT DE DS ME MT PT SY TQ YS */
1.47 schwarze 303: const char *const __man_reserved[] = {
1.73 schwarze 304: "AT", "B", "BI", "BR", "DT",
305: "EE", "EN", "EQ", "EX", "HP", "I", "IB", "IP", "IR",
306: "LP", "OP", "P", "PD", "PP",
307: "R", "RB", "RE", "RI", "RS", "SB", "SH", "SM", "SS",
308: "TE", "TH", "TP", "TS", "T&", "UC", "UE", "UR",
1.47 schwarze 309: NULL
310: };
311:
1.37 schwarze 312: /* Array of injected predefined strings. */
313: #define PREDEFS_MAX 38
314: static const struct predef predefs[PREDEFS_MAX] = {
315: #include "predefs.in"
316: };
317:
1.42 schwarze 318: /* See roffhash_find() */
1.3 schwarze 319: #define ROFF_HASH(p) (p[0] - ASCII_LO)
320:
1.51 schwarze 321: static int roffit_lines; /* number of lines to delay */
322: static char *roffit_macro; /* nil-terminated macro line */
323:
1.80 schwarze 324:
1.3 schwarze 325: static void
1.42 schwarze 326: roffhash_init(void)
1.3 schwarze 327: {
328: struct roffmac *n;
329: int buc, i;
330:
1.16 schwarze 331: for (i = 0; i < (int)ROFF_USERDEF; i++) {
1.3 schwarze 332: assert(roffs[i].name[0] >= ASCII_LO);
333: assert(roffs[i].name[0] <= ASCII_HI);
334:
335: buc = ROFF_HASH(roffs[i].name);
336:
337: if (NULL != (n = hash[buc])) {
338: for ( ; n->next; n = n->next)
339: /* Do nothing. */ ;
340: n->next = &roffs[i];
341: } else
342: hash[buc] = &roffs[i];
343: }
344: }
345:
1.1 schwarze 346: /*
347: * Look up a roff token by its name. Returns ROFF_MAX if no macro by
348: * the nil-terminated string name could be found.
349: */
350: static enum rofft
1.42 schwarze 351: roffhash_find(const char *p, size_t s)
1.1 schwarze 352: {
1.3 schwarze 353: int buc;
354: struct roffmac *n;
1.1 schwarze 355:
1.3 schwarze 356: /*
357: * libroff has an extremely simple hashtable, for the time
358: * being, which simply keys on the first character, which must
359: * be printable, then walks a chain. It works well enough until
360: * optimised.
361: */
362:
363: if (p[0] < ASCII_LO || p[0] > ASCII_HI)
364: return(ROFF_MAX);
365:
366: buc = ROFF_HASH(p);
367:
368: if (NULL == (n = hash[buc]))
369: return(ROFF_MAX);
370: for ( ; n; n = n->next)
1.16 schwarze 371: if (0 == strncmp(n->name, p, s) && '\0' == n->name[(int)s])
1.3 schwarze 372: return((enum rofft)(n - roffs));
1.1 schwarze 373:
374: return(ROFF_MAX);
375: }
376:
377: /*
378: * Pop the current node off of the stack of roff instructions currently
379: * pending.
380: */
381: static void
382: roffnode_pop(struct roff *r)
383: {
384: struct roffnode *p;
385:
1.2 schwarze 386: assert(r->last);
1.80 schwarze 387: p = r->last;
1.2 schwarze 388:
389: r->last = r->last->parent;
1.16 schwarze 390: free(p->name);
391: free(p->end);
1.1 schwarze 392: free(p);
393: }
394:
395: /*
396: * Push a roff node onto the instruction stack. This must later be
397: * removed with roffnode_pop().
398: */
1.11 schwarze 399: static void
1.16 schwarze 400: roffnode_push(struct roff *r, enum rofft tok, const char *name,
401: int line, int col)
1.1 schwarze 402: {
403: struct roffnode *p;
404:
1.11 schwarze 405: p = mandoc_calloc(1, sizeof(struct roffnode));
1.1 schwarze 406: p->tok = tok;
1.16 schwarze 407: if (name)
408: p->name = mandoc_strdup(name);
1.1 schwarze 409: p->parent = r->last;
410: p->line = line;
411: p->col = col;
1.71 schwarze 412: p->rule = p->parent ? p->parent->rule : 0;
1.1 schwarze 413:
414: r->last = p;
415: }
416:
417: static void
418: roff_free1(struct roff *r)
419: {
1.49 schwarze 420: struct tbl_node *tbl;
1.32 schwarze 421: struct eqn_node *e;
1.42 schwarze 422: int i;
1.27 schwarze 423:
1.49 schwarze 424: while (NULL != (tbl = r->first_tbl)) {
425: r->first_tbl = tbl->next;
426: tbl_free(tbl);
1.27 schwarze 427: }
428: r->first_tbl = r->last_tbl = r->tbl = NULL;
1.1 schwarze 429:
1.32 schwarze 430: while (NULL != (e = r->first_eqn)) {
431: r->first_eqn = e->next;
432: eqn_free(e);
433: }
434: r->first_eqn = r->last_eqn = r->eqn = NULL;
435:
1.1 schwarze 436: while (r->last)
437: roffnode_pop(r);
1.27 schwarze 438:
1.96 schwarze 439: free (r->rstack);
440: r->rstack = NULL;
441: r->rstacksz = 0;
442: r->rstackpos = -1;
443:
444: roff_freereg(r->regtab);
445: r->regtab = NULL;
446:
1.42 schwarze 447: roff_freestr(r->strtab);
448: roff_freestr(r->xmbtab);
449: r->strtab = r->xmbtab = NULL;
450:
451: if (r->xtab)
452: for (i = 0; i < 128; i++)
453: free(r->xtab[i].p);
454: free(r->xtab);
455: r->xtab = NULL;
1.1 schwarze 456: }
457:
458: void
459: roff_reset(struct roff *r)
460: {
461:
462: roff_free1(r);
1.99 schwarze 463: r->format = r->options & (MPARSE_MDOC | MPARSE_MAN);
1.48 schwarze 464: r->control = 0;
1.1 schwarze 465: }
466:
467: void
468: roff_free(struct roff *r)
469: {
470:
471: roff_free1(r);
472: free(r);
473: }
474:
475: struct roff *
1.109 schwarze 476: roff_alloc(struct mparse *parse, const struct mchars *mchars, int options)
1.1 schwarze 477: {
478: struct roff *r;
479:
1.11 schwarze 480: r = mandoc_calloc(1, sizeof(struct roff));
1.35 schwarze 481: r->parse = parse;
1.109 schwarze 482: r->mchars = mchars;
1.72 schwarze 483: r->options = options;
1.99 schwarze 484: r->format = options & (MPARSE_MDOC | MPARSE_MAN);
1.2 schwarze 485: r->rstackpos = -1;
1.80 schwarze 486:
1.42 schwarze 487: roffhash_init();
1.37 schwarze 488:
1.1 schwarze 489: return(r);
490: }
491:
1.8 schwarze 492: /*
1.79 schwarze 493: * In the current line, expand escape sequences that tend to get
494: * used in numerical expressions and conditional requests.
495: * Also check the syntax of the remaining escape sequences.
1.8 schwarze 496: */
1.45 schwarze 497: static enum rofferr
1.110 schwarze 498: roff_res(struct roff *r, struct buf *buf, int ln, int pos)
1.8 schwarze 499: {
1.81 schwarze 500: char ubuf[24]; /* buffer to print the number */
1.78 schwarze 501: const char *start; /* start of the string to process */
1.82 schwarze 502: char *stesc; /* start of an escape sequence ('\\') */
1.23 schwarze 503: const char *stnam; /* start of the name, after "[(*" */
504: const char *cp; /* end of the name, e.g. before ']' */
505: const char *res; /* the string to be substituted */
1.110 schwarze 506: char *nbuf; /* new buffer to copy buf->buf to */
1.53 schwarze 507: size_t maxl; /* expected length of the escape name */
508: size_t naml; /* actual length of the escape name */
1.109 schwarze 509: enum mandoc_esc esc; /* type of the escape sequence */
510: int inaml; /* length returned from mandoc_escape() */
1.53 schwarze 511: int expand_count; /* to avoid infinite loops */
1.79 schwarze 512: int npos; /* position in numeric expression */
1.91 schwarze 513: int arg_complete; /* argument not interrupted by eol */
1.79 schwarze 514: char term; /* character terminating the escape */
1.8 schwarze 515:
1.43 schwarze 516: expand_count = 0;
1.110 schwarze 517: start = buf->buf + pos;
1.78 schwarze 518: stesc = strchr(start, '\0') - 1;
519: while (stesc-- > start) {
1.43 schwarze 520:
1.78 schwarze 521: /* Search backwards for the next backslash. */
522:
1.110 schwarze 523: if (*stesc != '\\')
1.78 schwarze 524: continue;
525:
526: /* If it is escaped, skip it. */
527:
528: for (cp = stesc - 1; cp >= start; cp--)
1.110 schwarze 529: if (*cp != '\\')
1.78 schwarze 530: break;
531:
1.110 schwarze 532: if ((stesc - cp) % 2 == 0) {
1.82 schwarze 533: stesc = (char *)cp;
1.78 schwarze 534: continue;
535: }
1.23 schwarze 536:
1.79 schwarze 537: /* Decide whether to expand or to check only. */
1.23 schwarze 538:
1.79 schwarze 539: term = '\0';
1.78 schwarze 540: cp = stesc + 1;
1.53 schwarze 541: switch (*cp) {
1.80 schwarze 542: case '*':
1.53 schwarze 543: res = NULL;
544: break;
1.80 schwarze 545: case 'B':
1.79 schwarze 546: /* FALLTHROUGH */
1.80 schwarze 547: case 'w':
1.79 schwarze 548: term = cp[1];
549: /* FALLTHROUGH */
1.80 schwarze 550: case 'n':
1.53 schwarze 551: res = ubuf;
552: break;
553: default:
1.109 schwarze 554: esc = mandoc_escape(&cp, &stnam, &inaml);
555: if (esc == ESCAPE_ERROR ||
556: (esc == ESCAPE_SPECIAL &&
557: mchars_spec2cp(r->mchars, stnam, inaml) < 0))
1.92 schwarze 558: mandoc_vmsg(MANDOCERR_ESC_BAD,
1.110 schwarze 559: r->parse, ln, (int)(stesc - buf->buf),
1.92 schwarze 560: "%.*s", (int)(cp - stesc), stesc);
1.78 schwarze 561: continue;
1.42 schwarze 562: }
563:
1.78 schwarze 564: if (EXPAND_LIMIT < ++expand_count) {
565: mandoc_msg(MANDOCERR_ROFFLOOP, r->parse,
1.110 schwarze 566: ln, (int)(stesc - buf->buf), NULL);
1.78 schwarze 567: return(ROFF_IGN);
568: }
1.23 schwarze 569:
570: /*
571: * The third character decides the length
1.53 schwarze 572: * of the name of the string or register.
1.23 schwarze 573: * Save a pointer to the name.
574: */
575:
1.110 schwarze 576: if (term == '\0') {
1.79 schwarze 577: switch (*++cp) {
1.80 schwarze 578: case '\0':
1.79 schwarze 579: maxl = 0;
580: break;
1.80 schwarze 581: case '(':
1.79 schwarze 582: cp++;
583: maxl = 2;
584: break;
1.80 schwarze 585: case '[':
1.79 schwarze 586: cp++;
587: term = ']';
588: maxl = 0;
589: break;
590: default:
591: maxl = 1;
592: break;
593: }
594: } else {
595: cp += 2;
1.8 schwarze 596: maxl = 0;
597: }
1.23 schwarze 598: stnam = cp;
1.8 schwarze 599:
1.23 schwarze 600: /* Advance to the end of the name. */
1.8 schwarze 601:
1.91 schwarze 602: arg_complete = 1;
1.110 schwarze 603: for (naml = 0; maxl == 0 || naml < maxl; naml++, cp++) {
604: if (*cp == '\0') {
1.92 schwarze 605: mandoc_msg(MANDOCERR_ESC_BAD, r->parse,
1.110 schwarze 606: ln, (int)(stesc - buf->buf), stesc);
1.91 schwarze 607: arg_complete = 0;
1.79 schwarze 608: break;
1.42 schwarze 609: }
1.110 schwarze 610: if (maxl == 0 && *cp == term) {
1.79 schwarze 611: cp++;
1.8 schwarze 612: break;
1.79 schwarze 613: }
1.8 schwarze 614: }
615:
1.23 schwarze 616: /*
617: * Retrieve the replacement string; if it is
618: * undefined, resume searching for escapes.
619: */
620:
1.79 schwarze 621: switch (stesc[1]) {
1.80 schwarze 622: case '*':
1.91 schwarze 623: if (arg_complete)
624: res = roff_getstrn(r, stnam, naml);
1.79 schwarze 625: break;
1.80 schwarze 626: case 'B':
1.79 schwarze 627: npos = 0;
1.91 schwarze 628: ubuf[0] = arg_complete &&
1.106 schwarze 629: roff_evalnum(r, ln, stnam, &npos, NULL, 0) &&
1.91 schwarze 630: stnam + npos + 1 == cp ? '1' : '0';
1.79 schwarze 631: ubuf[1] = '\0';
632: break;
1.80 schwarze 633: case 'n':
1.91 schwarze 634: if (arg_complete)
635: (void)snprintf(ubuf, sizeof(ubuf), "%d",
636: roff_getregn(r, stnam, naml));
637: else
638: ubuf[0] = '\0';
1.79 schwarze 639: break;
1.80 schwarze 640: case 'w':
1.91 schwarze 641: /* use even incomplete args */
1.81 schwarze 642: (void)snprintf(ubuf, sizeof(ubuf), "%d",
1.79 schwarze 643: 24 * (int)naml);
644: break;
645: }
1.8 schwarze 646:
1.110 schwarze 647: if (res == NULL) {
1.92 schwarze 648: mandoc_vmsg(MANDOCERR_STR_UNDEF,
1.110 schwarze 649: r->parse, ln, (int)(stesc - buf->buf),
1.92 schwarze 650: "%.*s", (int)naml, stnam);
1.37 schwarze 651: res = "";
1.118 schwarze 652: } else if (buf->sz + strlen(res) > SHRT_MAX) {
653: mandoc_msg(MANDOCERR_ROFFLOOP, r->parse,
654: ln, (int)(stesc - buf->buf), NULL);
655: return(ROFF_IGN);
1.8 schwarze 656: }
657:
1.23 schwarze 658: /* Replace the escape sequence by the string. */
659:
1.82 schwarze 660: *stesc = '\0';
1.110 schwarze 661: buf->sz = mandoc_asprintf(&nbuf, "%s%s%s",
662: buf->buf, res, cp) + 1;
1.8 schwarze 663:
1.78 schwarze 664: /* Prepare for the next replacement. */
1.8 schwarze 665:
1.78 schwarze 666: start = nbuf + pos;
1.110 schwarze 667: stesc = nbuf + (stesc - buf->buf) + strlen(res);
668: free(buf->buf);
669: buf->buf = nbuf;
1.42 schwarze 670: }
1.45 schwarze 671: return(ROFF_CONT);
1.42 schwarze 672: }
673:
674: /*
1.51 schwarze 675: * Process text streams:
676: * Convert all breakable hyphens into ASCII_HYPH.
677: * Decrement and spring input line trap.
1.42 schwarze 678: */
679: static enum rofferr
1.110 schwarze 680: roff_parsetext(struct buf *buf, int pos, int *offs)
1.42 schwarze 681: {
682: size_t sz;
683: const char *start;
1.51 schwarze 684: char *p;
685: int isz;
1.42 schwarze 686: enum mandoc_esc esc;
687:
1.110 schwarze 688: start = p = buf->buf + pos;
1.42 schwarze 689:
1.110 schwarze 690: while (*p != '\0') {
1.42 schwarze 691: sz = strcspn(p, "-\\");
692: p += sz;
693:
1.110 schwarze 694: if (*p == '\0')
1.42 schwarze 695: break;
696:
1.110 schwarze 697: if (*p == '\\') {
1.42 schwarze 698: /* Skip over escapes. */
699: p++;
1.62 schwarze 700: esc = mandoc_escape((const char **)&p, NULL, NULL);
1.110 schwarze 701: if (esc == ESCAPE_ERROR)
1.42 schwarze 702: break;
703: continue;
704: } else if (p == start) {
705: p++;
706: continue;
707: }
708:
1.44 schwarze 709: if (isalpha((unsigned char)p[-1]) &&
710: isalpha((unsigned char)p[1]))
1.42 schwarze 711: *p = ASCII_HYPH;
712: p++;
1.8 schwarze 713: }
714:
1.51 schwarze 715: /* Spring the input line trap. */
1.110 schwarze 716: if (roffit_lines == 1) {
717: isz = mandoc_asprintf(&p, "%s\n.%s", buf->buf, roffit_macro);
718: free(buf->buf);
719: buf->buf = p;
720: buf->sz = isz + 1;
1.51 schwarze 721: *offs = 0;
722: free(roffit_macro);
723: roffit_lines = 0;
724: return(ROFF_REPARSE);
1.110 schwarze 725: } else if (roffit_lines > 1)
1.51 schwarze 726: --roffit_lines;
1.42 schwarze 727: return(ROFF_CONT);
1.8 schwarze 728: }
729:
1.1 schwarze 730: enum rofferr
1.110 schwarze 731: roff_parseln(struct roff *r, int ln, struct buf *buf, int *offs)
1.1 schwarze 732: {
733: enum rofft t;
1.27 schwarze 734: enum rofferr e;
1.110 schwarze 735: int pos; /* parse point */
1.115 schwarze 736: int spos; /* saved parse point for messages */
1.110 schwarze 737: int ppos; /* original offset in buf->buf */
738: int ctl; /* macro line (boolean) */
739:
740: ppos = pos = *offs;
1.1 schwarze 741:
1.102 schwarze 742: /* Handle in-line equation delimiters. */
743:
1.108 schwarze 744: if (r->tbl == NULL &&
745: r->last_eqn != NULL && r->last_eqn->delim &&
1.102 schwarze 746: (r->eqn == NULL || r->eqn_inline)) {
1.110 schwarze 747: e = roff_eqndelim(r, buf, pos);
1.102 schwarze 748: if (e == ROFF_REPARSE)
749: return(e);
750: assert(e == ROFF_CONT);
751: }
752:
753: /* Expand some escape sequences. */
1.8 schwarze 754:
1.110 schwarze 755: e = roff_res(r, buf, ln, pos);
756: if (e == ROFF_IGN)
1.45 schwarze 757: return(e);
1.110 schwarze 758: assert(e == ROFF_CONT);
1.8 schwarze 759:
1.110 schwarze 760: ctl = roff_getcontrol(r, buf->buf, &pos);
1.35 schwarze 761:
1.8 schwarze 762: /*
1.2 schwarze 763: * First, if a scope is open and we're not a macro, pass the
764: * text through the macro's filter. If a scope isn't open and
765: * we're not a macro, just let it through.
1.32 schwarze 766: * Finally, if there's an equation scope open, divert it into it
767: * no matter our state.
1.2 schwarze 768: */
769:
1.35 schwarze 770: if (r->last && ! ctl) {
1.2 schwarze 771: t = r->last->tok;
772: assert(roffs[t].text);
1.110 schwarze 773: e = (*roffs[t].text)(r, t, buf, ln, pos, pos, offs);
774: assert(e == ROFF_IGN || e == ROFF_CONT);
775: if (e != ROFF_CONT)
1.32 schwarze 776: return(e);
1.54 schwarze 777: }
778: if (r->eqn)
1.110 schwarze 779: return(eqn_read(&r->eqn, ln, buf->buf, ppos, offs));
1.54 schwarze 780: if ( ! ctl) {
1.32 schwarze 781: if (r->tbl)
1.110 schwarze 782: return(tbl_read(r->tbl, ln, buf->buf, pos));
783: return(roff_parsetext(buf, pos, offs));
1.54 schwarze 784: }
1.100 schwarze 785:
786: /* Skip empty request lines. */
787:
1.110 schwarze 788: if (buf->buf[pos] == '"') {
1.100 schwarze 789: mandoc_msg(MANDOCERR_COMMENT_BAD, r->parse,
790: ln, pos, NULL);
791: return(ROFF_IGN);
1.110 schwarze 792: } else if (buf->buf[pos] == '\0')
1.100 schwarze 793: return(ROFF_IGN);
1.2 schwarze 794:
795: /*
796: * If a scope is open, go to the child handler for that macro,
797: * as it may want to preprocess before doing anything with it.
1.32 schwarze 798: * Don't do so if an equation is open.
1.2 schwarze 799: */
800:
801: if (r->last) {
1.1 schwarze 802: t = r->last->tok;
803: assert(roffs[t].sub);
1.110 schwarze 804: return((*roffs[t].sub)(r, t, buf, ln, ppos, pos, offs));
1.2 schwarze 805: }
806:
1.115 schwarze 807: /* No scope is open. This is a new request or macro. */
808:
809: spos = pos;
810: t = roff_parse(r, buf->buf, &pos, ln, ppos);
811:
812: /* Tables ignore most macros. */
813:
814: if (r->tbl != NULL && (t == ROFF_MAX || t == ROFF_TS)) {
815: mandoc_msg(MANDOCERR_TBLMACRO, r->parse,
816: ln, pos, buf->buf + spos);
817: return(ROFF_IGN);
818: }
819:
1.2 schwarze 820: /*
1.115 schwarze 821: * This is neither a roff request nor a user-defined macro.
822: * Let the standard macro set parsers handle it.
1.2 schwarze 823: */
824:
1.115 schwarze 825: if (t == ROFF_MAX)
1.1 schwarze 826: return(ROFF_CONT);
1.115 schwarze 827:
828: /* Execute a roff request or a user defined macro. */
1.1 schwarze 829:
1.2 schwarze 830: assert(roffs[t].proc);
1.110 schwarze 831: return((*roffs[t].proc)(r, t, buf, ln, ppos, pos, offs));
1.2 schwarze 832: }
833:
1.27 schwarze 834: void
1.2 schwarze 835: roff_endparse(struct roff *r)
836: {
1.1 schwarze 837:
1.27 schwarze 838: if (r->last)
1.94 schwarze 839: mandoc_msg(MANDOCERR_BLK_NOEND, r->parse,
840: r->last->line, r->last->col,
841: roffs[r->last->tok].name);
1.27 schwarze 842:
1.32 schwarze 843: if (r->eqn) {
1.94 schwarze 844: mandoc_msg(MANDOCERR_BLK_NOEND, r->parse,
845: r->eqn->eqn.ln, r->eqn->eqn.pos, "EQ");
1.41 schwarze 846: eqn_end(&r->eqn);
1.32 schwarze 847: }
848:
1.27 schwarze 849: if (r->tbl) {
1.94 schwarze 850: mandoc_msg(MANDOCERR_BLK_NOEND, r->parse,
851: r->tbl->line, r->tbl->pos, "TS");
1.41 schwarze 852: tbl_end(&r->tbl);
1.27 schwarze 853: }
1.1 schwarze 854: }
855:
856: /*
857: * Parse a roff node's type from the input buffer. This must be in the
858: * form of ".foo xxx" in the usual way.
859: */
860: static enum rofft
1.87 schwarze 861: roff_parse(struct roff *r, char *buf, int *pos, int ln, int ppos)
1.1 schwarze 862: {
1.87 schwarze 863: char *cp;
1.16 schwarze 864: const char *mac;
865: size_t maclen;
1.1 schwarze 866: enum rofft t;
867:
1.87 schwarze 868: cp = buf + *pos;
869:
870: if ('\0' == *cp || '"' == *cp || '\t' == *cp || ' ' == *cp)
1.1 schwarze 871: return(ROFF_MAX);
872:
1.87 schwarze 873: mac = cp;
874: maclen = roff_getname(r, &cp, ln, ppos);
1.1 schwarze 875:
1.16 schwarze 876: t = (r->current_string = roff_getstrn(r, mac, maclen))
1.42 schwarze 877: ? ROFF_USERDEF : roffhash_find(mac, maclen);
1.1 schwarze 878:
1.87 schwarze 879: if (ROFF_MAX != t)
880: *pos = cp - buf;
1.1 schwarze 881:
882: return(t);
883: }
884:
885: static enum rofferr
1.2 schwarze 886: roff_cblock(ROFF_ARGS)
1.1 schwarze 887: {
888:
1.2 schwarze 889: /*
890: * A block-close `..' should only be invoked as a child of an
891: * ignore macro, otherwise raise a warning and just ignore it.
892: */
893:
1.110 schwarze 894: if (r->last == NULL) {
1.94 schwarze 895: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
896: ln, ppos, "..");
1.2 schwarze 897: return(ROFF_IGN);
898: }
1.1 schwarze 899:
1.2 schwarze 900: switch (r->last->tok) {
1.80 schwarze 901: case ROFF_am:
1.93 schwarze 902: /* ROFF_am1 is remapped to ROFF_am in roff_block(). */
1.2 schwarze 903: /* FALLTHROUGH */
1.80 schwarze 904: case ROFF_ami:
1.2 schwarze 905: /* FALLTHROUGH */
1.80 schwarze 906: case ROFF_de:
1.23 schwarze 907: /* ROFF_de1 is remapped to ROFF_de in roff_block(). */
1.2 schwarze 908: /* FALLTHROUGH */
1.80 schwarze 909: case ROFF_dei:
1.2 schwarze 910: /* FALLTHROUGH */
1.80 schwarze 911: case ROFF_ig:
1.2 schwarze 912: break;
913: default:
1.94 schwarze 914: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
915: ln, ppos, "..");
1.1 schwarze 916: return(ROFF_IGN);
1.2 schwarze 917: }
918:
1.110 schwarze 919: if (buf->buf[pos] != '\0')
1.90 schwarze 920: mandoc_vmsg(MANDOCERR_ARG_SKIP, r->parse, ln, pos,
1.110 schwarze 921: ".. %s", buf->buf + pos);
1.2 schwarze 922:
923: roffnode_pop(r);
924: roffnode_cleanscope(r);
925: return(ROFF_IGN);
926:
927: }
1.1 schwarze 928:
1.2 schwarze 929: static void
930: roffnode_cleanscope(struct roff *r)
931: {
1.1 schwarze 932:
1.2 schwarze 933: while (r->last) {
1.46 schwarze 934: if (--r->last->endspan != 0)
1.2 schwarze 935: break;
936: roffnode_pop(r);
937: }
938: }
1.1 schwarze 939:
1.68 schwarze 940: static void
941: roff_ccond(struct roff *r, int ln, int ppos)
1.2 schwarze 942: {
1.1 schwarze 943:
1.2 schwarze 944: if (NULL == r->last) {
1.94 schwarze 945: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
946: ln, ppos, "\\}");
1.68 schwarze 947: return;
1.2 schwarze 948: }
1.1 schwarze 949:
1.2 schwarze 950: switch (r->last->tok) {
1.80 schwarze 951: case ROFF_el:
1.2 schwarze 952: /* FALLTHROUGH */
1.80 schwarze 953: case ROFF_ie:
1.2 schwarze 954: /* FALLTHROUGH */
1.80 schwarze 955: case ROFF_if:
1.2 schwarze 956: break;
957: default:
1.94 schwarze 958: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
959: ln, ppos, "\\}");
1.68 schwarze 960: return;
1.2 schwarze 961: }
1.1 schwarze 962:
1.2 schwarze 963: if (r->last->endspan > -1) {
1.94 schwarze 964: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
965: ln, ppos, "\\}");
1.68 schwarze 966: return;
1.2 schwarze 967: }
968:
969: roffnode_pop(r);
970: roffnode_cleanscope(r);
1.68 schwarze 971: return;
1.1 schwarze 972: }
973:
974: static enum rofferr
1.2 schwarze 975: roff_block(ROFF_ARGS)
1.1 schwarze 976: {
1.93 schwarze 977: const char *name;
978: char *iname, *cp;
1.86 schwarze 979: size_t namesz;
1.16 schwarze 980:
1.93 schwarze 981: /* Ignore groff compatibility mode for now. */
1.2 schwarze 982:
1.110 schwarze 983: if (tok == ROFF_de1)
1.93 schwarze 984: tok = ROFF_de;
1.110 schwarze 985: else if (tok == ROFF_am1)
1.93 schwarze 986: tok = ROFF_am;
987:
988: /* Parse the macro name argument. */
989:
1.110 schwarze 990: cp = buf->buf + pos;
991: if (tok == ROFF_ig) {
1.93 schwarze 992: iname = NULL;
993: namesz = 0;
994: } else {
995: iname = cp;
996: namesz = roff_getname(r, &cp, ln, ppos);
997: iname[namesz] = '\0';
998: }
1.22 schwarze 999:
1.93 schwarze 1000: /* Resolve the macro name argument if it is indirect. */
1.22 schwarze 1001:
1.110 schwarze 1002: if (namesz && (tok == ROFF_dei || tok == ROFF_ami)) {
1003: if ((name = roff_getstrn(r, iname, namesz)) == NULL) {
1.93 schwarze 1004: mandoc_vmsg(MANDOCERR_STR_UNDEF,
1.110 schwarze 1005: r->parse, ln, (int)(iname - buf->buf),
1.93 schwarze 1006: "%.*s", (int)namesz, iname);
1007: namesz = 0;
1008: } else
1009: namesz = strlen(name);
1010: } else
1011: name = iname;
1.22 schwarze 1012:
1.110 schwarze 1013: if (namesz == 0 && tok != ROFF_ig) {
1.93 schwarze 1014: mandoc_msg(MANDOCERR_REQ_EMPTY, r->parse,
1015: ln, ppos, roffs[tok].name);
1016: return(ROFF_IGN);
1017: }
1.2 schwarze 1018:
1.16 schwarze 1019: roffnode_push(r, tok, name, ln, ppos);
1020:
1021: /*
1022: * At the beginning of a `de' macro, clear the existing string
1023: * with the same name, if there is one. New content will be
1.66 schwarze 1024: * appended from roff_block_text() in multiline mode.
1.16 schwarze 1025: */
1.22 schwarze 1026:
1.110 schwarze 1027: if (tok == ROFF_de || tok == ROFF_dei)
1.86 schwarze 1028: roff_setstrn(&r->strtab, name, namesz, "", 0, 0);
1.2 schwarze 1029:
1.110 schwarze 1030: if (*cp == '\0')
1.2 schwarze 1031: return(ROFF_IGN);
1.1 schwarze 1032:
1.93 schwarze 1033: /* Get the custom end marker. */
1.22 schwarze 1034:
1.93 schwarze 1035: iname = cp;
1.86 schwarze 1036: namesz = roff_getname(r, &cp, ln, ppos);
1.93 schwarze 1037:
1038: /* Resolve the end marker if it is indirect. */
1039:
1.110 schwarze 1040: if (namesz && (tok == ROFF_dei || tok == ROFF_ami)) {
1041: if ((name = roff_getstrn(r, iname, namesz)) == NULL) {
1.93 schwarze 1042: mandoc_vmsg(MANDOCERR_STR_UNDEF,
1.110 schwarze 1043: r->parse, ln, (int)(iname - buf->buf),
1.93 schwarze 1044: "%.*s", (int)namesz, iname);
1045: namesz = 0;
1046: } else
1047: namesz = strlen(name);
1048: } else
1049: name = iname;
1050:
1.86 schwarze 1051: if (namesz)
1052: r->last->end = mandoc_strndup(name, namesz);
1.2 schwarze 1053:
1.110 schwarze 1054: if (*cp != '\0')
1.90 schwarze 1055: mandoc_vmsg(MANDOCERR_ARG_EXCESS, r->parse,
1056: ln, pos, ".%s ... %s", roffs[tok].name, cp);
1.1 schwarze 1057:
1058: return(ROFF_IGN);
1059: }
1060:
1061: static enum rofferr
1.2 schwarze 1062: roff_block_sub(ROFF_ARGS)
1.1 schwarze 1063: {
1.2 schwarze 1064: enum rofft t;
1065: int i, j;
1066:
1067: /*
1068: * First check whether a custom macro exists at this level. If
1069: * it does, then check against it. This is some of groff's
1070: * stranger behaviours. If we encountered a custom end-scope
1071: * tag and that tag also happens to be a "real" macro, then we
1072: * need to try interpreting it again as a real macro. If it's
1073: * not, then return ignore. Else continue.
1074: */
1075:
1076: if (r->last->end) {
1.35 schwarze 1077: for (i = pos, j = 0; r->last->end[j]; j++, i++)
1.110 schwarze 1078: if (buf->buf[i] != r->last->end[j])
1.2 schwarze 1079: break;
1.1 schwarze 1080:
1.110 schwarze 1081: if (r->last->end[j] == '\0' &&
1082: (buf->buf[i] == '\0' ||
1083: buf->buf[i] == ' ' ||
1084: buf->buf[i] == '\t')) {
1.2 schwarze 1085: roffnode_pop(r);
1086: roffnode_cleanscope(r);
1.1 schwarze 1087:
1.110 schwarze 1088: while (buf->buf[i] == ' ' || buf->buf[i] == '\t')
1.35 schwarze 1089: i++;
1090:
1091: pos = i;
1.110 schwarze 1092: if (roff_parse(r, buf->buf, &pos, ln, ppos) !=
1093: ROFF_MAX)
1.2 schwarze 1094: return(ROFF_RERUN);
1095: return(ROFF_IGN);
1096: }
1.1 schwarze 1097: }
1098:
1.2 schwarze 1099: /*
1100: * If we have no custom end-query or lookup failed, then try
1101: * pulling it out of the hashtable.
1102: */
1.1 schwarze 1103:
1.110 schwarze 1104: t = roff_parse(r, buf->buf, &pos, ln, ppos);
1.1 schwarze 1105:
1.110 schwarze 1106: if (t != ROFF_cblock) {
1107: if (tok != ROFF_ig)
1108: roff_setstr(r, r->last->name, buf->buf + ppos, 2);
1.1 schwarze 1109: return(ROFF_IGN);
1.16 schwarze 1110: }
1.1 schwarze 1111:
1.2 schwarze 1112: assert(roffs[t].proc);
1.110 schwarze 1113: return((*roffs[t].proc)(r, t, buf, ln, ppos, pos, offs));
1.2 schwarze 1114: }
1115:
1116: static enum rofferr
1117: roff_block_text(ROFF_ARGS)
1118: {
1119:
1.110 schwarze 1120: if (tok != ROFF_ig)
1121: roff_setstr(r, r->last->name, buf->buf + pos, 2);
1.16 schwarze 1122:
1.2 schwarze 1123: return(ROFF_IGN);
1124: }
1125:
1126: static enum rofferr
1127: roff_cond_sub(ROFF_ARGS)
1128: {
1129: enum rofft t;
1.37 schwarze 1130: char *ep;
1.71 schwarze 1131: int rr;
1.2 schwarze 1132:
1133: rr = r->last->rule;
1.37 schwarze 1134: roffnode_cleanscope(r);
1.110 schwarze 1135: t = roff_parse(r, buf->buf, &pos, ln, ppos);
1.2 schwarze 1136:
1.37 schwarze 1137: /*
1.50 schwarze 1138: * Fully handle known macros when they are structurally
1139: * required or when the conditional evaluated to true.
1.5 schwarze 1140: */
1141:
1.110 schwarze 1142: if ((t != ROFF_MAX) &&
1143: (rr || roffs[t].flags & ROFFMAC_STRUCT)) {
1.50 schwarze 1144: assert(roffs[t].proc);
1.110 schwarze 1145: return((*roffs[t].proc)(r, t, buf, ln, ppos, pos, offs));
1.50 schwarze 1146: }
1.39 schwarze 1147:
1.69 schwarze 1148: /*
1149: * If `\}' occurs on a macro line without a preceding macro,
1150: * drop the line completely.
1151: */
1152:
1.110 schwarze 1153: ep = buf->buf + pos;
1154: if (ep[0] == '\\' && ep[1] == '}')
1.71 schwarze 1155: rr = 0;
1.69 schwarze 1156:
1.50 schwarze 1157: /* Always check for the closing delimiter `\}'. */
1.39 schwarze 1158:
1.110 schwarze 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: }
1.119 schwarze 1164: if (*ep != '\0')
1165: ++ep;
1.50 schwarze 1166: }
1.71 schwarze 1167: return(rr ? ROFF_CONT : ROFF_IGN);
1.2 schwarze 1168: }
1169:
1170: static enum rofferr
1171: roff_cond_text(ROFF_ARGS)
1172: {
1.37 schwarze 1173: char *ep;
1.71 schwarze 1174: int rr;
1.2 schwarze 1175:
1176: rr = r->last->rule;
1.37 schwarze 1177: roffnode_cleanscope(r);
1.1 schwarze 1178:
1.110 schwarze 1179: ep = buf->buf + pos;
1180: while ((ep = strchr(ep, '\\')) != NULL) {
1181: if (*(++ep) == '}') {
1.70 schwarze 1182: *ep = '&';
1.110 schwarze 1183: roff_ccond(r, ln, ep - buf->buf - 1);
1.70 schwarze 1184: }
1.119 schwarze 1185: if (*ep != '\0')
1186: ++ep;
1.2 schwarze 1187: }
1.71 schwarze 1188: return(rr ? ROFF_CONT : ROFF_IGN);
1.2 schwarze 1189: }
1190:
1.77 schwarze 1191: /*
1192: * Parse a single signed integer number. Stop at the first non-digit.
1193: * If there is at least one digit, return success and advance the
1194: * parse point, else return failure and let the parse point unchanged.
1195: * Ignore overflows, treat them just like the C language.
1196: */
1.56 schwarze 1197: static int
1198: roff_getnum(const char *v, int *pos, int *res)
1199: {
1.79 schwarze 1200: int myres, n, p;
1201:
1202: if (NULL == res)
1203: res = &myres;
1.56 schwarze 1204:
1205: p = *pos;
1206: n = v[p] == '-';
1207: if (n)
1208: p++;
1209:
1210: for (*res = 0; isdigit((unsigned char)v[p]); p++)
1.77 schwarze 1211: *res = 10 * *res + v[p] - '0';
1.56 schwarze 1212: if (p == *pos + n)
1213: return 0;
1214:
1215: if (n)
1216: *res = -*res;
1217:
1218: *pos = p;
1219: return 1;
1220: }
1221:
1.71 schwarze 1222: /*
1223: * Evaluate a string comparison condition.
1224: * The first character is the delimiter.
1225: * Succeed if the string up to its second occurrence
1226: * matches the string up to its third occurence.
1227: * Advance the cursor after the third occurrence
1228: * or lacking that, to the end of the line.
1229: */
1230: static int
1231: roff_evalstrcond(const char *v, int *pos)
1232: {
1233: const char *s1, *s2, *s3;
1234: int match;
1235:
1236: match = 0;
1237: s1 = v + *pos; /* initial delimiter */
1238: s2 = s1 + 1; /* for scanning the first string */
1239: s3 = strchr(s2, *s1); /* for scanning the second string */
1240:
1241: if (NULL == s3) /* found no middle delimiter */
1242: goto out;
1243:
1244: while ('\0' != *++s3) {
1245: if (*s2 != *s3) { /* mismatch */
1246: s3 = strchr(s3, *s1);
1247: break;
1248: }
1249: if (*s3 == *s1) { /* found the final delimiter */
1250: match = 1;
1251: break;
1252: }
1253: s2++;
1254: }
1255:
1256: out:
1257: if (NULL == s3)
1258: s3 = strchr(s2, '\0');
1.114 schwarze 1259: else if (*s3 != '\0')
1.71 schwarze 1260: s3++;
1261: *pos = s3 - v;
1262: return(match);
1263: }
1264:
1.77 schwarze 1265: /*
1266: * Evaluate an optionally negated single character, numerical,
1267: * or string condition.
1268: */
1.71 schwarze 1269: static int
1.106 schwarze 1270: roff_evalcond(struct roff *r, int ln, const char *v, int *pos)
1.5 schwarze 1271: {
1.113 schwarze 1272: int number, savepos, wanttrue;
1.5 schwarze 1273:
1.71 schwarze 1274: if ('!' == v[*pos]) {
1275: wanttrue = 0;
1276: (*pos)++;
1277: } else
1278: wanttrue = 1;
1279:
1.5 schwarze 1280: switch (v[*pos]) {
1.112 schwarze 1281: case '\0':
1282: return(0);
1.80 schwarze 1283: case 'n':
1.71 schwarze 1284: /* FALLTHROUGH */
1.80 schwarze 1285: case 'o':
1.5 schwarze 1286: (*pos)++;
1.71 schwarze 1287: return(wanttrue);
1.80 schwarze 1288: case 'c':
1.71 schwarze 1289: /* FALLTHROUGH */
1.80 schwarze 1290: case 'd':
1.71 schwarze 1291: /* FALLTHROUGH */
1.80 schwarze 1292: case 'e':
1.5 schwarze 1293: /* FALLTHROUGH */
1.80 schwarze 1294: case 'r':
1.5 schwarze 1295: /* FALLTHROUGH */
1.80 schwarze 1296: case 't':
1.111 schwarze 1297: /* FALLTHROUGH */
1298: case 'v':
1.5 schwarze 1299: (*pos)++;
1.71 schwarze 1300: return(!wanttrue);
1.5 schwarze 1301: default:
1302: break;
1303: }
1304:
1.113 schwarze 1305: savepos = *pos;
1.106 schwarze 1306: if (roff_evalnum(r, ln, v, pos, &number, 0))
1.77 schwarze 1307: return((number > 0) == wanttrue);
1.113 schwarze 1308: else if (*pos == savepos)
1309: return(roff_evalstrcond(v, pos) == wanttrue);
1.77 schwarze 1310: else
1.113 schwarze 1311: return (0);
1.5 schwarze 1312: }
1313:
1.2 schwarze 1314: static enum rofferr
1.21 schwarze 1315: roff_line_ignore(ROFF_ARGS)
1.6 schwarze 1316: {
1.30 schwarze 1317:
1.21 schwarze 1318: return(ROFF_IGN);
1319: }
1320:
1321: static enum rofferr
1.2 schwarze 1322: roff_cond(ROFF_ARGS)
1323: {
1.46 schwarze 1324:
1325: roffnode_push(r, tok, NULL, ln, ppos);
1.2 schwarze 1326:
1.80 schwarze 1327: /*
1.35 schwarze 1328: * An `.el' has no conditional body: it will consume the value
1329: * of the current rstack entry set in prior `ie' calls or
1.80 schwarze 1330: * defaults to DENY.
1.35 schwarze 1331: *
1332: * If we're not an `el', however, then evaluate the conditional.
1333: */
1.1 schwarze 1334:
1.110 schwarze 1335: r->last->rule = tok == ROFF_el ?
1.80 schwarze 1336: (r->rstackpos < 0 ? 0 : r->rstack[r->rstackpos--]) :
1.110 schwarze 1337: roff_evalcond(r, ln, buf->buf, &pos);
1.2 schwarze 1338:
1.35 schwarze 1339: /*
1340: * An if-else will put the NEGATION of the current evaluated
1341: * conditional into the stack of rules.
1342: */
1343:
1.110 schwarze 1344: if (tok == ROFF_ie) {
1.96 schwarze 1345: if (r->rstackpos + 1 == r->rstacksz) {
1346: r->rstacksz += 16;
1347: r->rstack = mandoc_reallocarray(r->rstack,
1348: r->rstacksz, sizeof(int));
1.35 schwarze 1349: }
1.71 schwarze 1350: r->rstack[++r->rstackpos] = !r->last->rule;
1.2 schwarze 1351: }
1.5 schwarze 1352:
1353: /* If the parent has false as its rule, then so do we. */
1354:
1.71 schwarze 1355: if (r->last->parent && !r->last->parent->rule)
1356: r->last->rule = 0;
1.5 schwarze 1357:
1358: /*
1.46 schwarze 1359: * Determine scope.
1360: * If there is nothing on the line after the conditional,
1361: * not even whitespace, use next-line scope.
1.5 schwarze 1362: */
1.2 schwarze 1363:
1.110 schwarze 1364: if (buf->buf[pos] == '\0') {
1.46 schwarze 1365: r->last->endspan = 2;
1366: goto out;
1367: }
1368:
1.110 schwarze 1369: while (buf->buf[pos] == ' ')
1.46 schwarze 1370: pos++;
1371:
1372: /* An opening brace requests multiline scope. */
1.2 schwarze 1373:
1.110 schwarze 1374: if (buf->buf[pos] == '\\' && buf->buf[pos + 1] == '{') {
1.2 schwarze 1375: r->last->endspan = -1;
1376: pos += 2;
1.46 schwarze 1377: goto out;
1.80 schwarze 1378: }
1.2 schwarze 1379:
1380: /*
1.46 schwarze 1381: * Anything else following the conditional causes
1382: * single-line scope. Warn if the scope contains
1383: * nothing but trailing whitespace.
1.2 schwarze 1384: */
1385:
1.110 schwarze 1386: if (buf->buf[pos] == '\0')
1.89 schwarze 1387: mandoc_msg(MANDOCERR_COND_EMPTY, r->parse,
1388: ln, ppos, roffs[tok].name);
1.2 schwarze 1389:
1.46 schwarze 1390: r->last->endspan = 1;
1.1 schwarze 1391:
1.46 schwarze 1392: out:
1.2 schwarze 1393: *offs = pos;
1394: return(ROFF_RERUN);
1.1 schwarze 1395: }
1396:
1.2 schwarze 1397: static enum rofferr
1.7 schwarze 1398: roff_ds(ROFF_ARGS)
1399: {
1.85 schwarze 1400: char *string;
1401: const char *name;
1402: size_t namesz;
1.10 schwarze 1403:
1404: /*
1.85 schwarze 1405: * The first word is the name of the string.
1406: * If it is empty or terminated by an escape sequence,
1407: * abort the `ds' request without defining anything.
1.10 schwarze 1408: */
1.7 schwarze 1409:
1.110 schwarze 1410: name = string = buf->buf + pos;
1411: if (*name == '\0')
1.7 schwarze 1412: return(ROFF_IGN);
1413:
1.85 schwarze 1414: namesz = roff_getname(r, &string, ln, pos);
1.110 schwarze 1415: if (name[namesz] == '\\')
1.85 schwarze 1416: return(ROFF_IGN);
1417:
1418: /* Read past the initial double-quote, if any. */
1.110 schwarze 1419: if (*string == '"')
1.7 schwarze 1420: string++;
1421:
1.10 schwarze 1422: /* The rest is the value. */
1.85 schwarze 1423: roff_setstrn(&r->strtab, name, namesz, string, strlen(string),
1424: ROFF_as == tok);
1.7 schwarze 1425: return(ROFF_IGN);
1426: }
1427:
1.77 schwarze 1428: /*
1429: * Parse a single operator, one or two characters long.
1430: * If the operator is recognized, return success and advance the
1431: * parse point, else return failure and let the parse point unchanged.
1432: */
1433: static int
1434: roff_getop(const char *v, int *pos, char *res)
1435: {
1436:
1437: *res = v[*pos];
1438:
1439: switch (*res) {
1.80 schwarze 1440: case '+':
1.77 schwarze 1441: /* FALLTHROUGH */
1.80 schwarze 1442: case '-':
1.77 schwarze 1443: /* FALLTHROUGH */
1.80 schwarze 1444: case '*':
1.77 schwarze 1445: /* FALLTHROUGH */
1.80 schwarze 1446: case '/':
1.77 schwarze 1447: /* FALLTHROUGH */
1.80 schwarze 1448: case '%':
1.77 schwarze 1449: /* FALLTHROUGH */
1.80 schwarze 1450: case '&':
1.77 schwarze 1451: /* FALLTHROUGH */
1.80 schwarze 1452: case ':':
1.77 schwarze 1453: break;
1454: case '<':
1455: switch (v[*pos + 1]) {
1.80 schwarze 1456: case '=':
1.77 schwarze 1457: *res = 'l';
1458: (*pos)++;
1459: break;
1.80 schwarze 1460: case '>':
1.77 schwarze 1461: *res = '!';
1462: (*pos)++;
1463: break;
1.80 schwarze 1464: case '?':
1.77 schwarze 1465: *res = 'i';
1466: (*pos)++;
1467: break;
1468: default:
1469: break;
1470: }
1471: break;
1472: case '>':
1473: switch (v[*pos + 1]) {
1.80 schwarze 1474: case '=':
1.77 schwarze 1475: *res = 'g';
1476: (*pos)++;
1477: break;
1.80 schwarze 1478: case '?':
1.77 schwarze 1479: *res = 'a';
1480: (*pos)++;
1481: break;
1482: default:
1483: break;
1484: }
1485: break;
1486: case '=':
1487: if ('=' == v[*pos + 1])
1488: (*pos)++;
1489: break;
1490: default:
1491: return(0);
1492: }
1493: (*pos)++;
1494:
1495: return(*res);
1496: }
1497:
1498: /*
1499: * Evaluate either a parenthesized numeric expression
1500: * or a single signed integer number.
1501: */
1502: static int
1.106 schwarze 1503: roff_evalpar(struct roff *r, int ln,
1504: const char *v, int *pos, int *res)
1.77 schwarze 1505: {
1506:
1507: if ('(' != v[*pos])
1508: return(roff_getnum(v, pos, res));
1509:
1510: (*pos)++;
1.106 schwarze 1511: if ( ! roff_evalnum(r, ln, v, pos, res, 1))
1.77 schwarze 1512: return(0);
1513:
1.79 schwarze 1514: /*
1515: * Omission of the closing parenthesis
1516: * is an error in validation mode,
1517: * but ignored in evaluation mode.
1518: */
1519:
1.77 schwarze 1520: if (')' == v[*pos])
1521: (*pos)++;
1.79 schwarze 1522: else if (NULL == res)
1523: return(0);
1.77 schwarze 1524:
1525: return(1);
1526: }
1527:
1528: /*
1529: * Evaluate a complete numeric expression.
1530: * Proceed left to right, there is no concept of precedence.
1531: */
1532: static int
1.106 schwarze 1533: roff_evalnum(struct roff *r, int ln, const char *v,
1534: int *pos, int *res, int skipwhite)
1.77 schwarze 1535: {
1536: int mypos, operand2;
1537: char operator;
1538:
1539: if (NULL == pos) {
1540: mypos = 0;
1541: pos = &mypos;
1542: }
1543:
1544: if (skipwhite)
1545: while (isspace((unsigned char)v[*pos]))
1546: (*pos)++;
1547:
1.106 schwarze 1548: if ( ! roff_evalpar(r, ln, v, pos, res))
1.77 schwarze 1549: return(0);
1550:
1551: while (1) {
1552: if (skipwhite)
1553: while (isspace((unsigned char)v[*pos]))
1554: (*pos)++;
1555:
1556: if ( ! roff_getop(v, pos, &operator))
1557: break;
1558:
1559: if (skipwhite)
1560: while (isspace((unsigned char)v[*pos]))
1561: (*pos)++;
1562:
1.106 schwarze 1563: if ( ! roff_evalpar(r, ln, v, pos, &operand2))
1.77 schwarze 1564: return(0);
1565:
1566: if (skipwhite)
1567: while (isspace((unsigned char)v[*pos]))
1568: (*pos)++;
1.79 schwarze 1569:
1570: if (NULL == res)
1571: continue;
1.77 schwarze 1572:
1573: switch (operator) {
1.80 schwarze 1574: case '+':
1.77 schwarze 1575: *res += operand2;
1576: break;
1.80 schwarze 1577: case '-':
1.77 schwarze 1578: *res -= operand2;
1579: break;
1.80 schwarze 1580: case '*':
1.77 schwarze 1581: *res *= operand2;
1582: break;
1.80 schwarze 1583: case '/':
1.116 schwarze 1584: if (operand2 == 0) {
1.106 schwarze 1585: mandoc_msg(MANDOCERR_DIVZERO,
1586: r->parse, ln, *pos, v);
1587: *res = 0;
1588: break;
1589: }
1.77 schwarze 1590: *res /= operand2;
1591: break;
1.80 schwarze 1592: case '%':
1.116 schwarze 1593: if (operand2 == 0) {
1594: mandoc_msg(MANDOCERR_DIVZERO,
1595: r->parse, ln, *pos, v);
1596: *res = 0;
1597: break;
1598: }
1.77 schwarze 1599: *res %= operand2;
1600: break;
1.80 schwarze 1601: case '<':
1.77 schwarze 1602: *res = *res < operand2;
1603: break;
1.80 schwarze 1604: case '>':
1.77 schwarze 1605: *res = *res > operand2;
1606: break;
1.80 schwarze 1607: case 'l':
1.77 schwarze 1608: *res = *res <= operand2;
1609: break;
1.80 schwarze 1610: case 'g':
1.77 schwarze 1611: *res = *res >= operand2;
1612: break;
1.80 schwarze 1613: case '=':
1.77 schwarze 1614: *res = *res == operand2;
1615: break;
1.80 schwarze 1616: case '!':
1.77 schwarze 1617: *res = *res != operand2;
1618: break;
1.80 schwarze 1619: case '&':
1.77 schwarze 1620: *res = *res && operand2;
1621: break;
1.80 schwarze 1622: case ':':
1.77 schwarze 1623: *res = *res || operand2;
1624: break;
1.80 schwarze 1625: case 'i':
1.77 schwarze 1626: if (operand2 < *res)
1627: *res = operand2;
1628: break;
1.80 schwarze 1629: case 'a':
1.77 schwarze 1630: if (operand2 > *res)
1631: *res = operand2;
1632: break;
1633: default:
1634: abort();
1635: }
1636: }
1637: return(1);
1638: }
1639:
1.52 schwarze 1640: void
1.60 schwarze 1641: roff_setreg(struct roff *r, const char *name, int val, char sign)
1.41 schwarze 1642: {
1.52 schwarze 1643: struct roffreg *reg;
1644:
1645: /* Search for an existing register with the same name. */
1646: reg = r->regtab;
1647:
1648: while (reg && strcmp(name, reg->key.p))
1649: reg = reg->next;
1.41 schwarze 1650:
1.52 schwarze 1651: if (NULL == reg) {
1652: /* Create a new register. */
1653: reg = mandoc_malloc(sizeof(struct roffreg));
1654: reg->key.p = mandoc_strdup(name);
1655: reg->key.sz = strlen(name);
1.60 schwarze 1656: reg->val = 0;
1.52 schwarze 1657: reg->next = r->regtab;
1658: r->regtab = reg;
1659: }
1660:
1.60 schwarze 1661: if ('+' == sign)
1662: reg->val += val;
1663: else if ('-' == sign)
1664: reg->val -= val;
1665: else
1666: reg->val = val;
1.41 schwarze 1667: }
1668:
1.65 schwarze 1669: /*
1670: * Handle some predefined read-only number registers.
1671: * For now, return -1 if the requested register is not predefined;
1672: * in case a predefined read-only register having the value -1
1673: * were to turn up, another special value would have to be chosen.
1674: */
1675: static int
1676: roff_getregro(const char *name)
1677: {
1678:
1679: switch (*name) {
1.80 schwarze 1680: case 'A': /* ASCII approximation mode is always off. */
1.65 schwarze 1681: return(0);
1.80 schwarze 1682: case 'g': /* Groff compatibility mode is always on. */
1.65 schwarze 1683: return(1);
1.80 schwarze 1684: case 'H': /* Fixed horizontal resolution. */
1.65 schwarze 1685: return (24);
1.80 schwarze 1686: case 'j': /* Always adjust left margin only. */
1.65 schwarze 1687: return(0);
1.80 schwarze 1688: case 'T': /* Some output device is always defined. */
1.65 schwarze 1689: return(1);
1.80 schwarze 1690: case 'V': /* Fixed vertical resolution. */
1.65 schwarze 1691: return (40);
1692: default:
1693: return (-1);
1694: }
1695: }
1696:
1.53 schwarze 1697: int
1.52 schwarze 1698: roff_getreg(const struct roff *r, const char *name)
1.41 schwarze 1699: {
1.52 schwarze 1700: struct roffreg *reg;
1.65 schwarze 1701: int val;
1702:
1703: if ('.' == name[0] && '\0' != name[1] && '\0' == name[2]) {
1704: val = roff_getregro(name + 1);
1705: if (-1 != val)
1706: return (val);
1707: }
1.52 schwarze 1708:
1709: for (reg = r->regtab; reg; reg = reg->next)
1710: if (0 == strcmp(name, reg->key.p))
1.53 schwarze 1711: return(reg->val);
1712:
1713: return(0);
1714: }
1715:
1716: static int
1717: roff_getregn(const struct roff *r, const char *name, size_t len)
1718: {
1719: struct roffreg *reg;
1.65 schwarze 1720: int val;
1721:
1722: if ('.' == name[0] && 2 == len) {
1723: val = roff_getregro(name + 1);
1724: if (-1 != val)
1725: return (val);
1726: }
1.53 schwarze 1727:
1728: for (reg = r->regtab; reg; reg = reg->next)
1729: if (len == reg->key.sz &&
1730: 0 == strncmp(name, reg->key.p, len))
1731: return(reg->val);
1.41 schwarze 1732:
1.52 schwarze 1733: return(0);
1.41 schwarze 1734: }
1735:
1.52 schwarze 1736: static void
1737: roff_freereg(struct roffreg *reg)
1.41 schwarze 1738: {
1.52 schwarze 1739: struct roffreg *old_reg;
1.41 schwarze 1740:
1.52 schwarze 1741: while (NULL != reg) {
1742: free(reg->key.p);
1743: old_reg = reg;
1744: reg = reg->next;
1745: free(old_reg);
1746: }
1.41 schwarze 1747: }
1.7 schwarze 1748:
1749: static enum rofferr
1.6 schwarze 1750: roff_nr(ROFF_ARGS)
1.1 schwarze 1751: {
1.85 schwarze 1752: char *key, *val;
1753: size_t keysz;
1.37 schwarze 1754: int iv;
1.60 schwarze 1755: char sign;
1.6 schwarze 1756:
1.110 schwarze 1757: key = val = buf->buf + pos;
1758: if (*key == '\0')
1.85 schwarze 1759: return(ROFF_IGN);
1760:
1761: keysz = roff_getname(r, &val, ln, pos);
1.110 schwarze 1762: if (key[keysz] == '\\')
1.85 schwarze 1763: return(ROFF_IGN);
1764: key[keysz] = '\0';
1.6 schwarze 1765:
1.60 schwarze 1766: sign = *val;
1.110 schwarze 1767: if (sign == '+' || sign == '-')
1.60 schwarze 1768: val++;
1769:
1.106 schwarze 1770: if (roff_evalnum(r, ln, val, NULL, &iv, 0))
1.77 schwarze 1771: roff_setreg(r, key, iv, sign);
1.1 schwarze 1772:
1.76 schwarze 1773: return(ROFF_IGN);
1774: }
1775:
1776: static enum rofferr
1777: roff_rr(ROFF_ARGS)
1778: {
1779: struct roffreg *reg, **prev;
1.85 schwarze 1780: char *name, *cp;
1781: size_t namesz;
1.76 schwarze 1782:
1.110 schwarze 1783: name = cp = buf->buf + pos;
1784: if (*name == '\0')
1.85 schwarze 1785: return(ROFF_IGN);
1786: namesz = roff_getname(r, &cp, ln, pos);
1787: name[namesz] = '\0';
1.76 schwarze 1788:
1789: prev = &r->regtab;
1790: while (1) {
1791: reg = *prev;
1.110 schwarze 1792: if (reg == NULL || !strcmp(name, reg->key.p))
1.76 schwarze 1793: break;
1794: prev = ®->next;
1795: }
1.110 schwarze 1796: if (reg != NULL) {
1.76 schwarze 1797: *prev = reg->next;
1798: free(reg->key.p);
1799: free(reg);
1800: }
1.29 schwarze 1801: return(ROFF_IGN);
1802: }
1803:
1804: static enum rofferr
1805: roff_rm(ROFF_ARGS)
1806: {
1807: const char *name;
1808: char *cp;
1.85 schwarze 1809: size_t namesz;
1.29 schwarze 1810:
1.110 schwarze 1811: cp = buf->buf + pos;
1812: while (*cp != '\0') {
1.85 schwarze 1813: name = cp;
1.110 schwarze 1814: namesz = roff_getname(r, &cp, ln, (int)(cp - buf->buf));
1.85 schwarze 1815: roff_setstrn(&r->strtab, name, namesz, NULL, 0, 0);
1.110 schwarze 1816: if (name[namesz] == '\\')
1.85 schwarze 1817: break;
1.29 schwarze 1818: }
1.51 schwarze 1819: return(ROFF_IGN);
1820: }
1821:
1822: static enum rofferr
1823: roff_it(ROFF_ARGS)
1824: {
1825: char *cp;
1826: size_t len;
1827: int iv;
1828:
1829: /* Parse the number of lines. */
1.110 schwarze 1830: cp = buf->buf + pos;
1.51 schwarze 1831: len = strcspn(cp, " \t");
1832: cp[len] = '\0';
1833: if ((iv = mandoc_strntoi(cp, len, 10)) <= 0) {
1.95 schwarze 1834: mandoc_msg(MANDOCERR_IT_NONUM, r->parse,
1.110 schwarze 1835: ln, ppos, buf->buf + 1);
1.51 schwarze 1836: return(ROFF_IGN);
1837: }
1838: cp += len + 1;
1839:
1840: /* Arm the input line trap. */
1841: roffit_lines = iv;
1842: roffit_macro = mandoc_strdup(cp);
1.2 schwarze 1843: return(ROFF_IGN);
1.47 schwarze 1844: }
1845:
1846: static enum rofferr
1847: roff_Dd(ROFF_ARGS)
1848: {
1849: const char *const *cp;
1850:
1.99 schwarze 1851: if ((r->options & (MPARSE_MDOC | MPARSE_QUICK)) == 0)
1.47 schwarze 1852: for (cp = __mdoc_reserved; *cp; cp++)
1853: roff_setstr(r, *cp, NULL, 0);
1854:
1.99 schwarze 1855: if (r->format == 0)
1856: r->format = MPARSE_MDOC;
1857:
1.47 schwarze 1858: return(ROFF_CONT);
1859: }
1860:
1861: static enum rofferr
1862: roff_TH(ROFF_ARGS)
1863: {
1864: const char *const *cp;
1865:
1.99 schwarze 1866: if ((r->options & MPARSE_QUICK) == 0)
1.47 schwarze 1867: for (cp = __man_reserved; *cp; cp++)
1868: roff_setstr(r, *cp, NULL, 0);
1869:
1.99 schwarze 1870: if (r->format == 0)
1871: r->format = MPARSE_MAN;
1872:
1.47 schwarze 1873: return(ROFF_CONT);
1.14 schwarze 1874: }
1875:
1876: static enum rofferr
1.27 schwarze 1877: roff_TE(ROFF_ARGS)
1878: {
1879:
1880: if (NULL == r->tbl)
1.94 schwarze 1881: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
1882: ln, ppos, "TE");
1.27 schwarze 1883: else
1.41 schwarze 1884: tbl_end(&r->tbl);
1.27 schwarze 1885:
1886: return(ROFF_IGN);
1887: }
1888:
1889: static enum rofferr
1890: roff_T_(ROFF_ARGS)
1891: {
1892:
1893: if (NULL == r->tbl)
1.94 schwarze 1894: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
1895: ln, ppos, "T&");
1.27 schwarze 1896: else
1897: tbl_restart(ppos, ln, r->tbl);
1898:
1899: return(ROFF_IGN);
1900: }
1901:
1.102 schwarze 1902: /*
1903: * Handle in-line equation delimiters.
1904: */
1905: static enum rofferr
1.110 schwarze 1906: roff_eqndelim(struct roff *r, struct buf *buf, int pos)
1.41 schwarze 1907: {
1.105 schwarze 1908: char *cp1, *cp2;
1909: const char *bef_pr, *bef_nl, *mac, *aft_nl, *aft_pr;
1.41 schwarze 1910:
1.102 schwarze 1911: /*
1912: * Outside equations, look for an opening delimiter.
1913: * If we are inside an equation, we already know it is
1914: * in-line, or this function wouldn't have been called;
1915: * so look for a closing delimiter.
1916: */
1917:
1.110 schwarze 1918: cp1 = buf->buf + pos;
1.102 schwarze 1919: cp2 = strchr(cp1, r->eqn == NULL ?
1920: r->last_eqn->odelim : r->last_eqn->cdelim);
1921: if (cp2 == NULL)
1922: return(ROFF_CONT);
1923:
1.105 schwarze 1924: *cp2++ = '\0';
1925: bef_pr = bef_nl = aft_nl = aft_pr = "";
1926:
1927: /* Handle preceding text, protecting whitespace. */
1928:
1.110 schwarze 1929: if (*buf->buf != '\0') {
1.105 schwarze 1930: if (r->eqn == NULL)
1931: bef_pr = "\\&";
1932: bef_nl = "\n";
1933: }
1934:
1935: /*
1936: * Prepare replacing the delimiter with an equation macro
1937: * and drop leading white space from the equation.
1938: */
1.102 schwarze 1939:
1.105 schwarze 1940: if (r->eqn == NULL) {
1941: while (*cp2 == ' ')
1942: cp2++;
1943: mac = ".EQ";
1944: } else
1945: mac = ".EN";
1946:
1947: /* Handle following text, protecting whitespace. */
1948:
1949: if (*cp2 != '\0') {
1950: aft_nl = "\n";
1951: if (r->eqn != NULL)
1952: aft_pr = "\\&";
1953: }
1954:
1955: /* Do the actual replacement. */
1956:
1.110 schwarze 1957: buf->sz = mandoc_asprintf(&cp1, "%s%s%s%s%s%s%s", buf->buf,
1.105 schwarze 1958: bef_pr, bef_nl, mac, aft_nl, aft_pr, cp2) + 1;
1.110 schwarze 1959: free(buf->buf);
1960: buf->buf = cp1;
1.102 schwarze 1961:
1962: /* Toggle the in-line state of the eqn subsystem. */
1963:
1964: r->eqn_inline = r->eqn == NULL;
1965: return(ROFF_REPARSE);
1.41 schwarze 1966: }
1967:
1.107 schwarze 1968: static enum rofferr
1969: roff_EQ(ROFF_ARGS)
1.32 schwarze 1970: {
1.41 schwarze 1971: struct eqn_node *e;
1.32 schwarze 1972:
1.110 schwarze 1973: assert(r->eqn == NULL);
1.107 schwarze 1974: e = eqn_alloc(ppos, ln, r->parse);
1.32 schwarze 1975:
1.102 schwarze 1976: if (r->last_eqn) {
1.32 schwarze 1977: r->last_eqn->next = e;
1.102 schwarze 1978: e->delim = r->last_eqn->delim;
1979: e->odelim = r->last_eqn->odelim;
1980: e->cdelim = r->last_eqn->cdelim;
1981: } else
1.32 schwarze 1982: r->first_eqn = r->last_eqn = e;
1983:
1984: r->eqn = r->last_eqn = e;
1.41 schwarze 1985:
1.110 schwarze 1986: if (buf->buf[pos] != '\0')
1.107 schwarze 1987: mandoc_vmsg(MANDOCERR_ARG_SKIP, r->parse, ln, pos,
1.110 schwarze 1988: ".EQ %s", buf->buf + pos);
1.41 schwarze 1989:
1.32 schwarze 1990: return(ROFF_IGN);
1991: }
1992:
1993: static enum rofferr
1994: roff_EN(ROFF_ARGS)
1995: {
1996:
1.94 schwarze 1997: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse, ln, ppos, "EN");
1.32 schwarze 1998: return(ROFF_IGN);
1999: }
2000:
2001: static enum rofferr
1.27 schwarze 2002: roff_TS(ROFF_ARGS)
2003: {
1.49 schwarze 2004: struct tbl_node *tbl;
1.27 schwarze 2005:
2006: if (r->tbl) {
1.94 schwarze 2007: mandoc_msg(MANDOCERR_BLK_BROKEN, r->parse,
2008: ln, ppos, "TS breaks TS");
1.41 schwarze 2009: tbl_end(&r->tbl);
1.27 schwarze 2010: }
2011:
1.49 schwarze 2012: tbl = tbl_alloc(ppos, ln, r->parse);
1.27 schwarze 2013:
2014: if (r->last_tbl)
1.49 schwarze 2015: r->last_tbl->next = tbl;
1.27 schwarze 2016: else
1.49 schwarze 2017: r->first_tbl = r->last_tbl = tbl;
1.27 schwarze 2018:
1.49 schwarze 2019: r->tbl = r->last_tbl = tbl;
1.27 schwarze 2020: return(ROFF_IGN);
2021: }
2022:
2023: static enum rofferr
1.48 schwarze 2024: roff_cc(ROFF_ARGS)
2025: {
2026: const char *p;
2027:
1.110 schwarze 2028: p = buf->buf + pos;
1.48 schwarze 2029:
1.110 schwarze 2030: if (*p == '\0' || (r->control = *p++) == '.')
1.48 schwarze 2031: r->control = 0;
2032:
1.110 schwarze 2033: if (*p != '\0')
1.48 schwarze 2034: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
2035:
2036: return(ROFF_IGN);
2037: }
2038:
2039: static enum rofferr
1.42 schwarze 2040: roff_tr(ROFF_ARGS)
2041: {
2042: const char *p, *first, *second;
2043: size_t fsz, ssz;
2044: enum mandoc_esc esc;
2045:
1.110 schwarze 2046: p = buf->buf + pos;
1.42 schwarze 2047:
1.110 schwarze 2048: if (*p == '\0') {
1.42 schwarze 2049: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
2050: return(ROFF_IGN);
2051: }
2052:
1.110 schwarze 2053: while (*p != '\0') {
1.42 schwarze 2054: fsz = ssz = 1;
2055:
2056: first = p++;
1.110 schwarze 2057: if (*first == '\\') {
1.42 schwarze 2058: esc = mandoc_escape(&p, NULL, NULL);
1.110 schwarze 2059: if (esc == ESCAPE_ERROR) {
1.92 schwarze 2060: mandoc_msg(MANDOCERR_ESC_BAD, r->parse,
1.110 schwarze 2061: ln, (int)(p - buf->buf), first);
1.42 schwarze 2062: return(ROFF_IGN);
2063: }
2064: fsz = (size_t)(p - first);
2065: }
2066:
2067: second = p++;
1.110 schwarze 2068: if (*second == '\\') {
1.42 schwarze 2069: esc = mandoc_escape(&p, NULL, NULL);
1.110 schwarze 2070: if (esc == ESCAPE_ERROR) {
1.92 schwarze 2071: mandoc_msg(MANDOCERR_ESC_BAD, r->parse,
1.110 schwarze 2072: ln, (int)(p - buf->buf), second);
1.42 schwarze 2073: return(ROFF_IGN);
2074: }
2075: ssz = (size_t)(p - second);
1.110 schwarze 2076: } else if (*second == '\0') {
1.80 schwarze 2077: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse,
1.110 schwarze 2078: ln, (int)(p - buf->buf), NULL);
1.42 schwarze 2079: second = " ";
2080: p--;
2081: }
2082:
2083: if (fsz > 1) {
1.80 schwarze 2084: roff_setstrn(&r->xmbtab, first, fsz,
2085: second, ssz, 0);
1.42 schwarze 2086: continue;
2087: }
2088:
1.110 schwarze 2089: if (r->xtab == NULL)
1.80 schwarze 2090: r->xtab = mandoc_calloc(128,
2091: sizeof(struct roffstr));
1.42 schwarze 2092:
2093: free(r->xtab[(int)*first].p);
2094: r->xtab[(int)*first].p = mandoc_strndup(second, ssz);
2095: r->xtab[(int)*first].sz = ssz;
2096: }
2097:
2098: return(ROFF_IGN);
2099: }
2100:
2101: static enum rofferr
1.14 schwarze 2102: roff_so(ROFF_ARGS)
2103: {
1.121 ! schwarze 2104: char *name, *cp;
1.15 schwarze 2105:
1.110 schwarze 2106: name = buf->buf + pos;
1.97 schwarze 2107: mandoc_vmsg(MANDOCERR_SO, r->parse, ln, ppos, "so %s", name);
1.14 schwarze 2108:
1.22 schwarze 2109: /*
2110: * Handle `so'. Be EXTREMELY careful, as we shouldn't be
2111: * opening anything that's not in our cwd or anything beneath
2112: * it. Thus, explicitly disallow traversing up the file-system
2113: * or using absolute paths.
2114: */
2115:
1.110 schwarze 2116: if (*name == '/' || strstr(name, "../") || strstr(name, "/..")) {
1.83 schwarze 2117: mandoc_vmsg(MANDOCERR_SO_PATH, r->parse, ln, ppos,
2118: ".so %s", name);
1.121 ! schwarze 2119: buf->sz = mandoc_asprintf(&cp,
! 2120: ".sp\nSee the file %s.\n.sp", name) + 1;
! 2121: free(buf->buf);
! 2122: buf->buf = cp;
! 2123: *offs = 0;
! 2124: return(ROFF_REPARSE);
1.14 schwarze 2125: }
2126:
2127: *offs = pos;
2128: return(ROFF_SO);
1.7 schwarze 2129: }
2130:
1.16 schwarze 2131: static enum rofferr
2132: roff_userdef(ROFF_ARGS)
1.12 schwarze 2133: {
1.16 schwarze 2134: const char *arg[9];
2135: char *cp, *n1, *n2;
1.25 schwarze 2136: int i;
1.12 schwarze 2137:
1.16 schwarze 2138: /*
2139: * Collect pointers to macro argument strings
1.61 schwarze 2140: * and NUL-terminate them.
1.16 schwarze 2141: */
1.110 schwarze 2142: cp = buf->buf + pos;
1.25 schwarze 2143: for (i = 0; i < 9; i++)
1.110 schwarze 2144: arg[i] = *cp == '\0' ? "" :
1.35 schwarze 2145: mandoc_getarg(r->parse, &cp, ln, &pos);
1.16 schwarze 2146:
2147: /*
2148: * Expand macro arguments.
1.12 schwarze 2149: */
1.110 schwarze 2150: buf->sz = 0;
1.16 schwarze 2151: n1 = cp = mandoc_strdup(r->current_string);
1.110 schwarze 2152: while ((cp = strstr(cp, "\\$")) != NULL) {
1.16 schwarze 2153: i = cp[2] - '1';
2154: if (0 > i || 8 < i) {
2155: /* Not an argument invocation. */
2156: cp += 2;
2157: continue;
2158: }
1.82 schwarze 2159: *cp = '\0';
1.110 schwarze 2160: buf->sz = mandoc_asprintf(&n2, "%s%s%s",
1.82 schwarze 2161: n1, arg[i], cp + 3) + 1;
1.16 schwarze 2162: cp = n2 + (cp - n1);
2163: free(n1);
2164: n1 = n2;
1.12 schwarze 2165: }
2166:
1.16 schwarze 2167: /*
2168: * Replace the macro invocation
2169: * by the expanded macro.
2170: */
1.110 schwarze 2171: free(buf->buf);
2172: buf->buf = n1;
2173: if (buf->sz == 0)
2174: buf->sz = strlen(buf->buf) + 1;
1.120 schwarze 2175: *offs = 0;
1.16 schwarze 2176:
1.110 schwarze 2177: return(buf->sz > 1 && buf->buf[buf->sz - 2] == '\n' ?
1.16 schwarze 2178: ROFF_REPARSE : ROFF_APPEND);
1.12 schwarze 2179: }
1.28 schwarze 2180:
1.85 schwarze 2181: static size_t
1.28 schwarze 2182: roff_getname(struct roff *r, char **cpp, int ln, int pos)
2183: {
2184: char *name, *cp;
1.85 schwarze 2185: size_t namesz;
1.28 schwarze 2186:
2187: name = *cpp;
2188: if ('\0' == *name)
1.85 schwarze 2189: return(0);
1.28 schwarze 2190:
1.85 schwarze 2191: /* Read until end of name and terminate it with NUL. */
2192: for (cp = name; 1; cp++) {
2193: if ('\0' == *cp || ' ' == *cp) {
2194: namesz = cp - name;
2195: break;
2196: }
1.28 schwarze 2197: if ('\\' != *cp)
2198: continue;
1.88 schwarze 2199: namesz = cp - name;
2200: if ('{' == cp[1] || '}' == cp[1])
2201: break;
1.28 schwarze 2202: cp++;
2203: if ('\\' == *cp)
2204: continue;
1.97 schwarze 2205: mandoc_vmsg(MANDOCERR_NAMESC, r->parse, ln, pos,
2206: "%.*s", (int)(cp - name + 1), name);
1.85 schwarze 2207: mandoc_escape((const char **)&cp, NULL, NULL);
2208: break;
1.28 schwarze 2209: }
2210:
2211: /* Read past spaces. */
2212: while (' ' == *cp)
2213: cp++;
2214:
2215: *cpp = cp;
1.85 schwarze 2216: return(namesz);
1.28 schwarze 2217: }
2218:
1.16 schwarze 2219: /*
2220: * Store *string into the user-defined string called *name.
2221: * To clear an existing entry, call with (*r, *name, NULL, 0).
1.66 schwarze 2222: * append == 0: replace mode
2223: * append == 1: single-line append mode
2224: * append == 2: multiline append mode, append '\n' after each call
1.16 schwarze 2225: */
1.8 schwarze 2226: static void
1.16 schwarze 2227: roff_setstr(struct roff *r, const char *name, const char *string,
1.66 schwarze 2228: int append)
1.7 schwarze 2229: {
1.42 schwarze 2230:
2231: roff_setstrn(&r->strtab, name, strlen(name), string,
1.80 schwarze 2232: string ? strlen(string) : 0, append);
1.42 schwarze 2233: }
2234:
2235: static void
2236: roff_setstrn(struct roffkv **r, const char *name, size_t namesz,
1.66 schwarze 2237: const char *string, size_t stringsz, int append)
1.42 schwarze 2238: {
2239: struct roffkv *n;
2240: char *c;
2241: int i;
2242: size_t oldch, newch;
1.7 schwarze 2243:
1.16 schwarze 2244: /* Search for an existing string with the same name. */
1.42 schwarze 2245: n = *r;
2246:
1.84 schwarze 2247: while (n && (namesz != n->key.sz ||
2248: strncmp(n->key.p, name, namesz)))
1.7 schwarze 2249: n = n->next;
1.8 schwarze 2250:
2251: if (NULL == n) {
1.16 schwarze 2252: /* Create a new string table entry. */
1.42 schwarze 2253: n = mandoc_malloc(sizeof(struct roffkv));
2254: n->key.p = mandoc_strndup(name, namesz);
2255: n->key.sz = namesz;
2256: n->val.p = NULL;
2257: n->val.sz = 0;
2258: n->next = *r;
2259: *r = n;
1.66 schwarze 2260: } else if (0 == append) {
1.42 schwarze 2261: free(n->val.p);
2262: n->val.p = NULL;
2263: n->val.sz = 0;
1.16 schwarze 2264: }
2265:
2266: if (NULL == string)
2267: return;
2268:
2269: /*
2270: * One additional byte for the '\n' in multiline mode,
2271: * and one for the terminating '\0'.
2272: */
1.66 schwarze 2273: newch = stringsz + (1 < append ? 2u : 1u);
1.42 schwarze 2274:
2275: if (NULL == n->val.p) {
2276: n->val.p = mandoc_malloc(newch);
2277: *n->val.p = '\0';
1.16 schwarze 2278: oldch = 0;
2279: } else {
1.42 schwarze 2280: oldch = n->val.sz;
2281: n->val.p = mandoc_realloc(n->val.p, oldch + newch);
1.16 schwarze 2282: }
2283:
2284: /* Skip existing content in the destination buffer. */
1.42 schwarze 2285: c = n->val.p + (int)oldch;
1.16 schwarze 2286:
2287: /* Append new content to the destination buffer. */
1.42 schwarze 2288: i = 0;
2289: while (i < (int)stringsz) {
1.16 schwarze 2290: /*
2291: * Rudimentary roff copy mode:
2292: * Handle escaped backslashes.
2293: */
1.42 schwarze 2294: if ('\\' == string[i] && '\\' == string[i + 1])
2295: i++;
2296: *c++ = string[i++];
1.16 schwarze 2297: }
1.8 schwarze 2298:
1.16 schwarze 2299: /* Append terminating bytes. */
1.66 schwarze 2300: if (1 < append)
1.16 schwarze 2301: *c++ = '\n';
1.42 schwarze 2302:
1.16 schwarze 2303: *c = '\0';
1.42 schwarze 2304: n->val.sz = (int)(c - n->val.p);
1.7 schwarze 2305: }
2306:
1.8 schwarze 2307: static const char *
2308: roff_getstrn(const struct roff *r, const char *name, size_t len)
1.7 schwarze 2309: {
1.42 schwarze 2310: const struct roffkv *n;
1.64 schwarze 2311: int i;
1.7 schwarze 2312:
1.42 schwarze 2313: for (n = r->strtab; n; n = n->next)
1.80 schwarze 2314: if (0 == strncmp(name, n->key.p, len) &&
2315: '\0' == n->key.p[(int)len])
1.42 schwarze 2316: return(n->val.p);
1.64 schwarze 2317:
2318: for (i = 0; i < PREDEFS_MAX; i++)
2319: if (0 == strncmp(name, predefs[i].name, len) &&
2320: '\0' == predefs[i].name[(int)len])
2321: return(predefs[i].str);
1.8 schwarze 2322:
1.42 schwarze 2323: return(NULL);
1.7 schwarze 2324: }
2325:
1.8 schwarze 2326: static void
1.42 schwarze 2327: roff_freestr(struct roffkv *r)
1.7 schwarze 2328: {
1.42 schwarze 2329: struct roffkv *n, *nn;
1.7 schwarze 2330:
1.42 schwarze 2331: for (n = r; n; n = nn) {
2332: free(n->key.p);
2333: free(n->val.p);
1.7 schwarze 2334: nn = n->next;
2335: free(n);
2336: }
1.27 schwarze 2337: }
2338:
2339: const struct tbl_span *
2340: roff_span(const struct roff *r)
2341: {
1.80 schwarze 2342:
1.27 schwarze 2343: return(r->tbl ? tbl_span(r->tbl) : NULL);
1.32 schwarze 2344: }
2345:
2346: const struct eqn *
2347: roff_eqn(const struct roff *r)
2348: {
1.80 schwarze 2349:
1.32 schwarze 2350: return(r->last_eqn ? &r->last_eqn->eqn : NULL);
1.42 schwarze 2351: }
2352:
2353: /*
2354: * Duplicate an input string, making the appropriate character
2355: * conversations (as stipulated by `tr') along the way.
2356: * Returns a heap-allocated string with all the replacements made.
2357: */
2358: char *
2359: roff_strdup(const struct roff *r, const char *p)
2360: {
2361: const struct roffkv *cp;
2362: char *res;
2363: const char *pp;
2364: size_t ssz, sz;
2365: enum mandoc_esc esc;
2366:
2367: if (NULL == r->xmbtab && NULL == r->xtab)
2368: return(mandoc_strdup(p));
2369: else if ('\0' == *p)
2370: return(mandoc_strdup(""));
2371:
2372: /*
2373: * Step through each character looking for term matches
2374: * (remember that a `tr' can be invoked with an escape, which is
2375: * a glyph but the escape is multi-character).
2376: * We only do this if the character hash has been initialised
2377: * and the string is >0 length.
2378: */
2379:
2380: res = NULL;
2381: ssz = 0;
2382:
2383: while ('\0' != *p) {
2384: if ('\\' != *p && r->xtab && r->xtab[(int)*p].p) {
2385: sz = r->xtab[(int)*p].sz;
2386: res = mandoc_realloc(res, ssz + sz + 1);
2387: memcpy(res + ssz, r->xtab[(int)*p].p, sz);
2388: ssz += sz;
2389: p++;
2390: continue;
2391: } else if ('\\' != *p) {
2392: res = mandoc_realloc(res, ssz + 2);
2393: res[ssz++] = *p++;
2394: continue;
2395: }
2396:
2397: /* Search for term matches. */
2398: for (cp = r->xmbtab; cp; cp = cp->next)
2399: if (0 == strncmp(p, cp->key.p, cp->key.sz))
2400: break;
2401:
2402: if (NULL != cp) {
2403: /*
2404: * A match has been found.
2405: * Append the match to the array and move
2406: * forward by its keysize.
2407: */
1.80 schwarze 2408: res = mandoc_realloc(res,
2409: ssz + cp->val.sz + 1);
1.42 schwarze 2410: memcpy(res + ssz, cp->val.p, cp->val.sz);
2411: ssz += cp->val.sz;
2412: p += (int)cp->key.sz;
2413: continue;
2414: }
2415:
2416: /*
2417: * Handle escapes carefully: we need to copy
2418: * over just the escape itself, or else we might
2419: * do replacements within the escape itself.
2420: * Make sure to pass along the bogus string.
2421: */
2422: pp = p++;
2423: esc = mandoc_escape(&p, NULL, NULL);
2424: if (ESCAPE_ERROR == esc) {
2425: sz = strlen(pp);
2426: res = mandoc_realloc(res, ssz + sz + 1);
2427: memcpy(res + ssz, pp, sz);
2428: break;
2429: }
1.80 schwarze 2430: /*
2431: * We bail out on bad escapes.
1.42 schwarze 2432: * No need to warn: we already did so when
2433: * roff_res() was called.
2434: */
2435: sz = (int)(p - pp);
2436: res = mandoc_realloc(res, ssz + sz + 1);
2437: memcpy(res + ssz, pp, sz);
2438: ssz += sz;
2439: }
2440:
2441: res[(int)ssz] = '\0';
2442: return(res);
1.99 schwarze 2443: }
2444:
2445: int
2446: roff_getformat(const struct roff *r)
2447: {
2448:
2449: return(r->format);
1.48 schwarze 2450: }
2451:
2452: /*
1.80 schwarze 2453: * Find out whether a line is a macro line or not.
1.48 schwarze 2454: * If it is, adjust the current position and return one; if it isn't,
2455: * return zero and don't change the current position.
2456: * If the control character has been set with `.cc', then let that grain
2457: * precedence.
2458: * This is slighly contrary to groff, where using the non-breaking
2459: * control character when `cc' has been invoked will cause the
2460: * non-breaking macro contents to be printed verbatim.
2461: */
2462: int
2463: roff_getcontrol(const struct roff *r, const char *cp, int *ppos)
2464: {
2465: int pos;
2466:
2467: pos = *ppos;
2468:
2469: if (0 != r->control && cp[pos] == r->control)
2470: pos++;
2471: else if (0 != r->control)
2472: return(0);
2473: else if ('\\' == cp[pos] && '.' == cp[pos + 1])
2474: pos += 2;
2475: else if ('.' == cp[pos] || '\'' == cp[pos])
2476: pos++;
2477: else
2478: return(0);
2479:
2480: while (' ' == cp[pos] || '\t' == cp[pos])
2481: pos++;
2482:
2483: *ppos = pos;
2484: return(1);
1.1 schwarze 2485: }
![[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