Annotation of mandoc/roff.c, Revision 1.235
1.235 ! schwarze 1: /* $Id: roff.c,v 1.234 2014/10/20 19:04:45 kristaps Exp $ */
1.1 kristaps 2: /*
1.175 schwarze 3: * Copyright (c) 2010, 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
1.190 schwarze 4: * Copyright (c) 2010-2014 Ingo Schwarze <schwarze@openbsd.org>
1.1 kristaps 5: *
6: * Permission to use, copy, modify, and distribute this software for any
1.66 kristaps 7: * purpose with or without fee is hereby granted, provided that the above
8: * copyright notice and this permission notice appear in all copies.
1.1 kristaps 9: *
1.106 kristaps 10: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.66 kristaps 11: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.106 kristaps 12: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.66 kristaps 13: * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
14: * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
15: * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
16: * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
1.1 kristaps 17: */
1.66 kristaps 18: #include "config.h"
1.225 schwarze 19:
20: #include <sys/types.h>
1.30 kristaps 21:
1.67 kristaps 22: #include <assert.h>
1.85 kristaps 23: #include <ctype.h>
1.178 schwarze 24: #include <stdio.h>
1.1 kristaps 25: #include <stdlib.h>
1.67 kristaps 26: #include <string.h>
1.1 kristaps 27:
1.67 kristaps 28: #include "mandoc.h"
1.201 schwarze 29: #include "mandoc_aux.h"
1.226 schwarze 30: #include "libmandoc.h"
1.109 kristaps 31: #include "libroff.h"
1.33 kristaps 32:
1.141 kristaps 33: /* Maximum number of nested if-else conditionals. */
1.82 kristaps 34: #define RSTACK_MAX 128
35:
1.170 schwarze 36: /* Maximum number of string expansions per line, to break infinite loops. */
37: #define EXPAND_LIMIT 1000
38:
1.67 kristaps 39: enum rofft {
1.103 kristaps 40: ROFF_ad,
1.80 kristaps 41: ROFF_am,
42: ROFF_ami,
43: ROFF_am1,
1.193 schwarze 44: ROFF_as,
1.174 kristaps 45: ROFF_cc,
1.194 schwarze 46: ROFF_ce,
1.80 kristaps 47: ROFF_de,
48: ROFF_dei,
49: ROFF_de1,
1.83 schwarze 50: ROFF_ds,
1.82 kristaps 51: ROFF_el,
1.185 schwarze 52: ROFF_fam,
1.186 schwarze 53: ROFF_hw,
1.103 kristaps 54: ROFF_hy,
1.82 kristaps 55: ROFF_ie,
1.75 kristaps 56: ROFF_if,
1.76 kristaps 57: ROFF_ig,
1.123 schwarze 58: ROFF_it,
1.103 kristaps 59: ROFF_ne,
60: ROFF_nh,
1.104 kristaps 61: ROFF_nr,
1.124 schwarze 62: ROFF_ns,
1.229 schwarze 63: ROFF_pl,
1.124 schwarze 64: ROFF_ps,
1.83 schwarze 65: ROFF_rm,
1.203 schwarze 66: ROFF_rr,
1.105 kristaps 67: ROFF_so,
1.124 schwarze 68: ROFF_ta,
1.83 schwarze 69: ROFF_tr,
1.175 schwarze 70: ROFF_Dd,
71: ROFF_TH,
1.109 kristaps 72: ROFF_TS,
73: ROFF_TE,
1.112 kristaps 74: ROFF_T_,
1.125 kristaps 75: ROFF_EQ,
76: ROFF_EN,
1.76 kristaps 77: ROFF_cblock,
1.106 kristaps 78: ROFF_USERDEF,
1.67 kristaps 79: ROFF_MAX
80: };
81:
1.147 kristaps 82: /*
1.167 kristaps 83: * An incredibly-simple string buffer.
84: */
1.94 kristaps 85: struct roffstr {
1.167 kristaps 86: char *p; /* nil-terminated buffer */
87: size_t sz; /* saved strlen(p) */
1.166 kristaps 88: };
89:
90: /*
1.167 kristaps 91: * A key-value roffstr pair as part of a singly-linked list.
1.166 kristaps 92: */
93: struct roffkv {
94: struct roffstr key;
95: struct roffstr val;
96: struct roffkv *next; /* next in list */
1.94 kristaps 97: };
98:
1.180 schwarze 99: /*
100: * A single number register as part of a singly-linked list.
101: */
102: struct roffreg {
103: struct roffstr key;
1.181 schwarze 104: int val;
1.180 schwarze 105: struct roffreg *next;
106: };
107:
1.67 kristaps 108: struct roff {
1.128 kristaps 109: struct mparse *parse; /* parse point */
1.67 kristaps 110: struct roffnode *last; /* leaf of stack */
1.223 schwarze 111: int *rstack; /* stack of inverted `ie' values */
1.180 schwarze 112: struct roffreg *regtab; /* number registers */
1.166 kristaps 113: struct roffkv *strtab; /* user-defined strings & macros */
1.167 kristaps 114: struct roffkv *xmbtab; /* multi-byte trans table (`tr') */
115: struct roffstr *xtab; /* single-byte trans table (`tr') */
1.106 kristaps 116: const char *current_string; /* value of last called user macro */
1.118 kristaps 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 */
1.125 kristaps 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 */
1.230 schwarze 123: int eqn_inline; /* current equation is inline */
1.223 schwarze 124: int options; /* parse options */
125: int rstacksz; /* current size limit of rstack */
126: int rstackpos; /* position in rstack */
1.227 schwarze 127: int format; /* current file in mdoc or man format */
1.223 schwarze 128: char control; /* control character */
1.79 kristaps 129: };
130:
1.67 kristaps 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 */
1.106 kristaps 136: char *name; /* node name, e.g. macro name */
1.79 kristaps 137: char *end; /* end-rules: custom token */
138: int endspan; /* end-rules: next-line or infty */
1.198 schwarze 139: int rule; /* current evaluation rule */
1.67 kristaps 140: };
141:
142: #define ROFF_ARGS struct roff *r, /* parse ctx */ \
1.72 kristaps 143: enum rofft tok, /* tok of macro */ \
1.207 schwarze 144: char **bufp, /* input buffer */ \
1.67 kristaps 145: size_t *szp, /* size of input buffer */ \
146: int ln, /* parse line */ \
1.75 kristaps 147: int ppos, /* original pos in buffer */ \
148: int pos, /* current pos in buffer */ \
1.74 kristaps 149: int *offs /* reset offset of buffer data */
1.67 kristaps 150:
151: typedef enum rofferr (*roffproc)(ROFF_ARGS);
152:
153: struct roffmac {
154: const char *name; /* macro name */
1.79 kristaps 155: roffproc proc; /* process new macro */
156: roffproc text; /* process as child text of macro */
157: roffproc sub; /* process as child of macro */
158: int flags;
159: #define ROFFMAC_STRUCT (1 << 0) /* always interpret */
1.85 kristaps 160: struct roffmac *next;
1.67 kristaps 161: };
162:
1.141 kristaps 163: struct predef {
164: const char *name; /* predefined input name */
165: const char *str; /* replacement symbol */
166: };
167:
168: #define PREDEF(__name, __str) \
169: { (__name), (__str) },
170:
1.155 kristaps 171: static enum rofft roffhash_find(const char *, size_t);
172: static void roffhash_init(void);
173: static void roffnode_cleanscope(struct roff *);
174: static void roffnode_pop(struct roff *);
175: static void roffnode_push(struct roff *, enum rofft,
176: const char *, int, int);
1.80 kristaps 177: static enum rofferr roff_block(ROFF_ARGS);
178: static enum rofferr roff_block_text(ROFF_ARGS);
179: static enum rofferr roff_block_sub(ROFF_ARGS);
180: static enum rofferr roff_cblock(ROFF_ARGS);
1.174 kristaps 181: static enum rofferr roff_cc(ROFF_ARGS);
1.195 schwarze 182: static void roff_ccond(struct roff *, int, int);
1.82 kristaps 183: static enum rofferr roff_cond(ROFF_ARGS);
184: static enum rofferr roff_cond_text(ROFF_ARGS);
185: static enum rofferr roff_cond_sub(ROFF_ARGS);
1.92 schwarze 186: static enum rofferr roff_ds(ROFF_ARGS);
1.230 schwarze 187: static enum rofferr roff_eqndelim(struct roff *,
188: char **, size_t *, int);
1.234 kristaps 189: static int roff_evalcond(struct roff *r, int,
190: const char *, int *);
191: static int roff_evalnum(struct roff *, int,
192: const char *, int *, int *, int);
193: static int roff_evalpar(struct roff *, int,
194: const char *, int *, int *);
1.198 schwarze 195: static int roff_evalstrcond(const char *, int *);
1.155 kristaps 196: static void roff_free1(struct roff *);
1.180 schwarze 197: static void roff_freereg(struct roffreg *);
1.167 kristaps 198: static void roff_freestr(struct roffkv *);
1.212 schwarze 199: static size_t roff_getname(struct roff *, char **, int, int);
1.184 schwarze 200: static int roff_getnum(const char *, int *, int *);
201: static int roff_getop(const char *, int *, char *);
1.181 schwarze 202: static int roff_getregn(const struct roff *,
203: const char *, size_t);
1.192 schwarze 204: static int roff_getregro(const char *name);
1.207 schwarze 205: static const char *roff_getstrn(const struct roff *,
1.94 kristaps 206: const char *, size_t);
1.178 schwarze 207: static enum rofferr roff_it(ROFF_ARGS);
1.103 kristaps 208: static enum rofferr roff_line_ignore(ROFF_ARGS);
1.89 kristaps 209: static enum rofferr roff_nr(ROFF_ARGS);
1.214 schwarze 210: static enum rofft roff_parse(struct roff *, char *, int *,
211: int, int);
1.178 schwarze 212: static enum rofferr roff_parsetext(char **, size_t *, int, int *);
1.207 schwarze 213: static enum rofferr roff_res(struct roff *,
1.142 kristaps 214: char **, size_t *, int, int);
1.122 schwarze 215: static enum rofferr roff_rm(ROFF_ARGS);
1.203 schwarze 216: static enum rofferr roff_rr(ROFF_ARGS);
1.94 kristaps 217: static void roff_setstr(struct roff *,
1.106 kristaps 218: const char *, const char *, int);
1.207 schwarze 219: static void roff_setstrn(struct roffkv **, const char *,
1.164 kristaps 220: size_t, const char *, size_t, int);
1.105 kristaps 221: static enum rofferr roff_so(ROFF_ARGS);
1.164 kristaps 222: static enum rofferr roff_tr(ROFF_ARGS);
1.175 schwarze 223: static enum rofferr roff_Dd(ROFF_ARGS);
224: static enum rofferr roff_TH(ROFF_ARGS);
1.109 kristaps 225: static enum rofferr roff_TE(ROFF_ARGS);
226: static enum rofferr roff_TS(ROFF_ARGS);
1.125 kristaps 227: static enum rofferr roff_EQ(ROFF_ARGS);
228: static enum rofferr roff_EN(ROFF_ARGS);
1.112 kristaps 229: static enum rofferr roff_T_(ROFF_ARGS);
1.106 kristaps 230: static enum rofferr roff_userdef(ROFF_ARGS);
1.67 kristaps 231:
1.155 kristaps 232: /* See roffhash_find() */
1.85 kristaps 233:
234: #define ASCII_HI 126
235: #define ASCII_LO 33
236: #define HASHWIDTH (ASCII_HI - ASCII_LO + 1)
237:
238: static struct roffmac *hash[HASHWIDTH];
239:
240: static struct roffmac roffs[ROFF_MAX] = {
1.103 kristaps 241: { "ad", roff_line_ignore, NULL, NULL, 0, NULL },
1.85 kristaps 242: { "am", roff_block, roff_block_text, roff_block_sub, 0, NULL },
243: { "ami", roff_block, roff_block_text, roff_block_sub, 0, NULL },
244: { "am1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.193 schwarze 245: { "as", roff_ds, NULL, NULL, 0, NULL },
1.174 kristaps 246: { "cc", roff_cc, NULL, NULL, 0, NULL },
1.194 schwarze 247: { "ce", roff_line_ignore, NULL, NULL, 0, NULL },
1.85 kristaps 248: { "de", roff_block, roff_block_text, roff_block_sub, 0, NULL },
249: { "dei", roff_block, roff_block_text, roff_block_sub, 0, NULL },
250: { "de1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.92 schwarze 251: { "ds", roff_ds, NULL, NULL, 0, NULL },
1.85 kristaps 252: { "el", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
1.185 schwarze 253: { "fam", roff_line_ignore, NULL, NULL, 0, NULL },
1.186 schwarze 254: { "hw", roff_line_ignore, NULL, NULL, 0, NULL },
1.103 kristaps 255: { "hy", roff_line_ignore, NULL, NULL, 0, NULL },
1.85 kristaps 256: { "ie", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
257: { "if", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
258: { "ig", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.178 schwarze 259: { "it", roff_it, NULL, NULL, 0, NULL },
1.103 kristaps 260: { "ne", roff_line_ignore, NULL, NULL, 0, NULL },
261: { "nh", roff_line_ignore, NULL, NULL, 0, NULL },
1.104 kristaps 262: { "nr", roff_nr, NULL, NULL, 0, NULL },
1.124 schwarze 263: { "ns", roff_line_ignore, NULL, NULL, 0, NULL },
1.229 schwarze 264: { "pl", roff_line_ignore, NULL, NULL, 0, NULL },
1.124 schwarze 265: { "ps", roff_line_ignore, NULL, NULL, 0, NULL },
1.122 schwarze 266: { "rm", roff_rm, NULL, NULL, 0, NULL },
1.203 schwarze 267: { "rr", roff_rr, NULL, NULL, 0, NULL },
1.105 kristaps 268: { "so", roff_so, NULL, NULL, 0, NULL },
1.124 schwarze 269: { "ta", roff_line_ignore, NULL, NULL, 0, NULL },
1.164 kristaps 270: { "tr", roff_tr, NULL, NULL, 0, NULL },
1.175 schwarze 271: { "Dd", roff_Dd, NULL, NULL, 0, NULL },
272: { "TH", roff_TH, NULL, NULL, 0, NULL },
1.109 kristaps 273: { "TS", roff_TS, NULL, NULL, 0, NULL },
274: { "TE", roff_TE, NULL, NULL, 0, NULL },
1.112 kristaps 275: { "T&", roff_T_, NULL, NULL, 0, NULL },
1.125 kristaps 276: { "EQ", roff_EQ, NULL, NULL, 0, NULL },
277: { "EN", roff_EN, NULL, NULL, 0, NULL },
1.85 kristaps 278: { ".", roff_cblock, NULL, NULL, 0, NULL },
1.106 kristaps 279: { NULL, roff_userdef, NULL, NULL, 0, NULL },
1.67 kristaps 280: };
281:
1.200 schwarze 282: /* not currently implemented: Ds em Eq LP Me PP pp Or Rd Sf SH */
1.175 schwarze 283: const char *const __mdoc_reserved[] = {
284: "Ac", "Ad", "An", "Ao", "Ap", "Aq", "Ar", "At",
285: "Bc", "Bd", "Bf", "Bk", "Bl", "Bo", "Bq",
286: "Brc", "Bro", "Brq", "Bsx", "Bt", "Bx",
287: "Cd", "Cm", "Db", "Dc", "Dd", "Dl", "Do", "Dq",
1.200 schwarze 288: "Dt", "Dv", "Dx", "D1",
289: "Ec", "Ed", "Ef", "Ek", "El", "Em",
290: "En", "Eo", "Er", "Es", "Ev", "Ex",
1.175 schwarze 291: "Fa", "Fc", "Fd", "Fl", "Fn", "Fo", "Fr", "Ft", "Fx",
1.200 schwarze 292: "Hf", "Ic", "In", "It", "Lb", "Li", "Lk", "Lp",
293: "Ms", "Mt", "Nd", "Nm", "No", "Ns", "Nx",
1.175 schwarze 294: "Oc", "Oo", "Op", "Os", "Ot", "Ox",
1.200 schwarze 295: "Pa", "Pc", "Pf", "Po", "Pp", "Pq",
296: "Qc", "Ql", "Qo", "Qq", "Re", "Rs", "Rv",
297: "Sc", "Sh", "Sm", "So", "Sq",
1.175 schwarze 298: "Ss", "St", "Sx", "Sy",
299: "Ta", "Tn", "Ud", "Ux", "Va", "Vt", "Xc", "Xo", "Xr",
1.200 schwarze 300: "%A", "%B", "%C", "%D", "%I", "%J", "%N", "%O",
1.175 schwarze 301: "%P", "%Q", "%R", "%T", "%U", "%V",
302: NULL
303: };
304:
1.200 schwarze 305: /* not currently implemented: BT DE DS ME MT PT SY TQ YS */
1.175 schwarze 306: const char *const __man_reserved[] = {
1.200 schwarze 307: "AT", "B", "BI", "BR", "DT",
308: "EE", "EN", "EQ", "EX", "HP", "I", "IB", "IP", "IR",
309: "LP", "OP", "P", "PD", "PP",
310: "R", "RB", "RE", "RI", "RS", "SB", "SH", "SM", "SS",
311: "TE", "TH", "TP", "TS", "T&", "UC", "UE", "UR",
1.175 schwarze 312: NULL
313: };
314:
1.141 kristaps 315: /* Array of injected predefined strings. */
316: #define PREDEFS_MAX 38
317: static const struct predef predefs[PREDEFS_MAX] = {
318: #include "predefs.in"
319: };
320:
1.155 kristaps 321: /* See roffhash_find() */
1.85 kristaps 322: #define ROFF_HASH(p) (p[0] - ASCII_LO)
323:
1.178 schwarze 324: static int roffit_lines; /* number of lines to delay */
325: static char *roffit_macro; /* nil-terminated macro line */
326:
1.207 schwarze 327:
1.85 kristaps 328: static void
1.155 kristaps 329: roffhash_init(void)
1.85 kristaps 330: {
331: struct roffmac *n;
332: int buc, i;
333:
1.106 kristaps 334: for (i = 0; i < (int)ROFF_USERDEF; i++) {
1.85 kristaps 335: assert(roffs[i].name[0] >= ASCII_LO);
336: assert(roffs[i].name[0] <= ASCII_HI);
337:
338: buc = ROFF_HASH(roffs[i].name);
339:
340: if (NULL != (n = hash[buc])) {
341: for ( ; n->next; n = n->next)
342: /* Do nothing. */ ;
343: n->next = &roffs[i];
344: } else
345: hash[buc] = &roffs[i];
346: }
347: }
348:
1.67 kristaps 349: /*
350: * Look up a roff token by its name. Returns ROFF_MAX if no macro by
351: * the nil-terminated string name could be found.
352: */
353: static enum rofft
1.155 kristaps 354: roffhash_find(const char *p, size_t s)
1.67 kristaps 355: {
1.85 kristaps 356: int buc;
357: struct roffmac *n;
1.67 kristaps 358:
1.85 kristaps 359: /*
360: * libroff has an extremely simple hashtable, for the time
361: * being, which simply keys on the first character, which must
362: * be printable, then walks a chain. It works well enough until
363: * optimised.
364: */
365:
366: if (p[0] < ASCII_LO || p[0] > ASCII_HI)
367: return(ROFF_MAX);
368:
369: buc = ROFF_HASH(p);
370:
371: if (NULL == (n = hash[buc]))
372: return(ROFF_MAX);
373: for ( ; n; n = n->next)
1.106 kristaps 374: if (0 == strncmp(n->name, p, s) && '\0' == n->name[(int)s])
1.85 kristaps 375: return((enum rofft)(n - roffs));
1.67 kristaps 376:
377: return(ROFF_MAX);
378: }
379:
380: /*
381: * Pop the current node off of the stack of roff instructions currently
382: * pending.
383: */
384: static void
385: roffnode_pop(struct roff *r)
386: {
387: struct roffnode *p;
388:
1.75 kristaps 389: assert(r->last);
1.207 schwarze 390: p = r->last;
1.82 kristaps 391:
1.75 kristaps 392: r->last = r->last->parent;
1.106 kristaps 393: free(p->name);
394: free(p->end);
1.67 kristaps 395: free(p);
396: }
397:
398: /*
399: * Push a roff node onto the instruction stack. This must later be
400: * removed with roffnode_pop().
401: */
1.98 schwarze 402: static void
1.106 kristaps 403: roffnode_push(struct roff *r, enum rofft tok, const char *name,
404: int line, int col)
1.67 kristaps 405: {
406: struct roffnode *p;
407:
1.98 schwarze 408: p = mandoc_calloc(1, sizeof(struct roffnode));
1.67 kristaps 409: p->tok = tok;
1.106 kristaps 410: if (name)
411: p->name = mandoc_strdup(name);
1.67 kristaps 412: p->parent = r->last;
413: p->line = line;
414: p->col = col;
1.198 schwarze 415: p->rule = p->parent ? p->parent->rule : 0;
1.67 kristaps 416:
417: r->last = p;
418: }
419:
420: static void
421: roff_free1(struct roff *r)
422: {
1.176 schwarze 423: struct tbl_node *tbl;
1.125 kristaps 424: struct eqn_node *e;
1.167 kristaps 425: int i;
1.67 kristaps 426:
1.176 schwarze 427: while (NULL != (tbl = r->first_tbl)) {
428: r->first_tbl = tbl->next;
429: tbl_free(tbl);
1.109 kristaps 430: }
1.113 kristaps 431: r->first_tbl = r->last_tbl = r->tbl = NULL;
432:
1.125 kristaps 433: while (NULL != (e = r->first_eqn)) {
434: r->first_eqn = e->next;
435: eqn_free(e);
436: }
437: r->first_eqn = r->last_eqn = r->eqn = NULL;
438:
1.67 kristaps 439: while (r->last)
440: roffnode_pop(r);
1.109 kristaps 441:
1.223 schwarze 442: free (r->rstack);
443: r->rstack = NULL;
444: r->rstacksz = 0;
445: r->rstackpos = -1;
446:
447: roff_freereg(r->regtab);
448: r->regtab = NULL;
449:
1.167 kristaps 450: roff_freestr(r->strtab);
451: roff_freestr(r->xmbtab);
452: r->strtab = r->xmbtab = NULL;
453:
454: if (r->xtab)
455: for (i = 0; i < 128; i++)
456: free(r->xtab[i].p);
457: free(r->xtab);
458: r->xtab = NULL;
1.67 kristaps 459: }
460:
461: void
462: roff_reset(struct roff *r)
463: {
464:
465: roff_free1(r);
1.227 schwarze 466: r->format = r->options & (MPARSE_MDOC | MPARSE_MAN);
1.174 kristaps 467: r->control = 0;
1.67 kristaps 468: }
469:
470: void
471: roff_free(struct roff *r)
472: {
473:
474: roff_free1(r);
475: free(r);
476: }
477:
478: struct roff *
1.199 schwarze 479: roff_alloc(struct mparse *parse, int options)
1.67 kristaps 480: {
481: struct roff *r;
482:
1.98 schwarze 483: r = mandoc_calloc(1, sizeof(struct roff));
1.128 kristaps 484: r->parse = parse;
1.199 schwarze 485: r->options = options;
1.227 schwarze 486: r->format = options & (MPARSE_MDOC | MPARSE_MAN);
1.82 kristaps 487: r->rstackpos = -1;
1.207 schwarze 488:
1.155 kristaps 489: roffhash_init();
1.141 kristaps 490:
1.67 kristaps 491: return(r);
492: }
493:
1.94 kristaps 494: /*
1.206 schwarze 495: * In the current line, expand escape sequences that tend to get
496: * used in numerical expressions and conditional requests.
497: * Also check the syntax of the remaining escape sequences.
1.154 kristaps 498: */
1.172 schwarze 499: static enum rofferr
1.142 kristaps 500: roff_res(struct roff *r, char **bufp, size_t *szp, int ln, int pos)
1.94 kristaps 501: {
1.208 schwarze 502: char ubuf[24]; /* buffer to print the number */
1.205 schwarze 503: const char *start; /* start of the string to process */
1.209 schwarze 504: char *stesc; /* start of an escape sequence ('\\') */
1.108 schwarze 505: const char *stnam; /* start of the name, after "[(*" */
506: const char *cp; /* end of the name, e.g. before ']' */
507: const char *res; /* the string to be substituted */
1.181 schwarze 508: char *nbuf; /* new buffer to copy bufp to */
509: size_t maxl; /* expected length of the escape name */
510: size_t naml; /* actual length of the escape name */
511: int expand_count; /* to avoid infinite loops */
1.206 schwarze 512: int npos; /* position in numeric expression */
1.218 schwarze 513: int arg_complete; /* argument not interrupted by eol */
1.206 schwarze 514: char term; /* character terminating the escape */
1.94 kristaps 515:
1.170 schwarze 516: expand_count = 0;
1.205 schwarze 517: start = *bufp + pos;
518: stesc = strchr(start, '\0') - 1;
519: while (stesc-- > start) {
1.170 schwarze 520:
1.205 schwarze 521: /* Search backwards for the next backslash. */
522:
523: if ('\\' != *stesc)
524: continue;
525:
526: /* If it is escaped, skip it. */
527:
528: for (cp = stesc - 1; cp >= start; cp--)
529: if ('\\' != *cp)
530: break;
531:
532: if (0 == (stesc - cp) % 2) {
1.209 schwarze 533: stesc = (char *)cp;
1.205 schwarze 534: continue;
535: }
1.108 schwarze 536:
1.206 schwarze 537: /* Decide whether to expand or to check only. */
1.108 schwarze 538:
1.206 schwarze 539: term = '\0';
1.205 schwarze 540: cp = stesc + 1;
1.181 schwarze 541: switch (*cp) {
1.207 schwarze 542: case '*':
1.181 schwarze 543: res = NULL;
544: break;
1.207 schwarze 545: case 'B':
1.206 schwarze 546: /* FALLTHROUGH */
1.207 schwarze 547: case 'w':
1.206 schwarze 548: term = cp[1];
549: /* FALLTHROUGH */
1.207 schwarze 550: case 'n':
1.181 schwarze 551: res = ubuf;
552: break;
553: default:
1.205 schwarze 554: if (ESCAPE_ERROR == mandoc_escape(&cp, NULL, NULL))
1.219 schwarze 555: mandoc_vmsg(MANDOCERR_ESC_BAD,
556: r->parse, ln, (int)(stesc - *bufp),
557: "%.*s", (int)(cp - stesc), stesc);
1.205 schwarze 558: continue;
1.152 kristaps 559: }
560:
1.205 schwarze 561: if (EXPAND_LIMIT < ++expand_count) {
562: mandoc_msg(MANDOCERR_ROFFLOOP, r->parse,
563: ln, (int)(stesc - *bufp), NULL);
564: return(ROFF_IGN);
565: }
1.108 schwarze 566:
567: /*
568: * The third character decides the length
1.181 schwarze 569: * of the name of the string or register.
1.108 schwarze 570: * Save a pointer to the name.
571: */
572:
1.206 schwarze 573: if ('\0' == term) {
574: switch (*++cp) {
1.207 schwarze 575: case '\0':
1.206 schwarze 576: maxl = 0;
577: break;
1.207 schwarze 578: case '(':
1.206 schwarze 579: cp++;
580: maxl = 2;
581: break;
1.207 schwarze 582: case '[':
1.206 schwarze 583: cp++;
584: term = ']';
585: maxl = 0;
586: break;
587: default:
588: maxl = 1;
589: break;
590: }
591: } else {
592: cp += 2;
1.94 kristaps 593: maxl = 0;
594: }
1.108 schwarze 595: stnam = cp;
1.94 kristaps 596:
1.108 schwarze 597: /* Advance to the end of the name. */
1.94 kristaps 598:
1.218 schwarze 599: arg_complete = 1;
1.181 schwarze 600: for (naml = 0; 0 == maxl || naml < maxl; naml++, cp++) {
1.153 kristaps 601: if ('\0' == *cp) {
1.219 schwarze 602: mandoc_msg(MANDOCERR_ESC_BAD, r->parse,
603: ln, (int)(stesc - *bufp), stesc);
1.218 schwarze 604: arg_complete = 0;
1.206 schwarze 605: break;
1.153 kristaps 606: }
1.206 schwarze 607: if (0 == maxl && *cp == term) {
608: cp++;
1.94 kristaps 609: break;
1.206 schwarze 610: }
1.94 kristaps 611: }
612:
1.108 schwarze 613: /*
614: * Retrieve the replacement string; if it is
615: * undefined, resume searching for escapes.
616: */
617:
1.206 schwarze 618: switch (stesc[1]) {
1.207 schwarze 619: case '*':
1.218 schwarze 620: if (arg_complete)
621: res = roff_getstrn(r, stnam, naml);
1.206 schwarze 622: break;
1.207 schwarze 623: case 'B':
1.206 schwarze 624: npos = 0;
1.218 schwarze 625: ubuf[0] = arg_complete &&
1.234 kristaps 626: roff_evalnum(r, ln, stnam, &npos, NULL, 0) &&
1.218 schwarze 627: stnam + npos + 1 == cp ? '1' : '0';
1.206 schwarze 628: ubuf[1] = '\0';
629: break;
1.207 schwarze 630: case 'n':
1.218 schwarze 631: if (arg_complete)
632: (void)snprintf(ubuf, sizeof(ubuf), "%d",
633: roff_getregn(r, stnam, naml));
634: else
635: ubuf[0] = '\0';
1.206 schwarze 636: break;
1.207 schwarze 637: case 'w':
1.218 schwarze 638: /* use even incomplete args */
1.208 schwarze 639: (void)snprintf(ubuf, sizeof(ubuf), "%d",
1.206 schwarze 640: 24 * (int)naml);
641: break;
642: }
1.94 kristaps 643:
644: if (NULL == res) {
1.219 schwarze 645: mandoc_vmsg(MANDOCERR_STR_UNDEF,
646: r->parse, ln, (int)(stesc - *bufp),
647: "%.*s", (int)naml, stnam);
1.142 kristaps 648: res = "";
1.94 kristaps 649: }
650:
1.108 schwarze 651: /* Replace the escape sequence by the string. */
652:
1.209 schwarze 653: *stesc = '\0';
654: *szp = mandoc_asprintf(&nbuf, "%s%s%s",
655: *bufp, res, cp) + 1;
1.94 kristaps 656:
1.205 schwarze 657: /* Prepare for the next replacement. */
1.94 kristaps 658:
1.205 schwarze 659: start = nbuf + pos;
1.209 schwarze 660: stesc = nbuf + (stesc - *bufp) + strlen(res);
1.94 kristaps 661: free(*bufp);
1.181 schwarze 662: *bufp = nbuf;
1.154 kristaps 663: }
1.172 schwarze 664: return(ROFF_CONT);
1.154 kristaps 665: }
666:
667: /*
1.178 schwarze 668: * Process text streams:
669: * Convert all breakable hyphens into ASCII_HYPH.
670: * Decrement and spring input line trap.
1.154 kristaps 671: */
672: static enum rofferr
1.178 schwarze 673: roff_parsetext(char **bufp, size_t *szp, int pos, int *offs)
1.154 kristaps 674: {
675: size_t sz;
676: const char *start;
1.178 schwarze 677: char *p;
678: int isz;
1.154 kristaps 679: enum mandoc_esc esc;
680:
1.178 schwarze 681: start = p = *bufp + pos;
1.154 kristaps 682:
683: while ('\0' != *p) {
684: sz = strcspn(p, "-\\");
685: p += sz;
686:
1.159 kristaps 687: if ('\0' == *p)
688: break;
689:
1.154 kristaps 690: if ('\\' == *p) {
691: /* Skip over escapes. */
692: p++;
1.189 schwarze 693: esc = mandoc_escape((const char **)&p, NULL, NULL);
1.154 kristaps 694: if (ESCAPE_ERROR == esc)
695: break;
1.155 kristaps 696: continue;
1.159 kristaps 697: } else if (p == start) {
1.158 kristaps 698: p++;
1.155 kristaps 699: continue;
1.158 kristaps 700: }
1.155 kristaps 701:
1.171 schwarze 702: if (isalpha((unsigned char)p[-1]) &&
703: isalpha((unsigned char)p[1]))
1.155 kristaps 704: *p = ASCII_HYPH;
705: p++;
1.94 kristaps 706: }
707:
1.178 schwarze 708: /* Spring the input line trap. */
709: if (1 == roffit_lines) {
1.202 schwarze 710: isz = mandoc_asprintf(&p, "%s\n.%s", *bufp, roffit_macro);
1.178 schwarze 711: free(*bufp);
712: *bufp = p;
713: *szp = isz + 1;
714: *offs = 0;
715: free(roffit_macro);
716: roffit_lines = 0;
717: return(ROFF_REPARSE);
718: } else if (1 < roffit_lines)
719: --roffit_lines;
1.154 kristaps 720: return(ROFF_CONT);
1.94 kristaps 721: }
722:
1.67 kristaps 723: enum rofferr
1.207 schwarze 724: roff_parseln(struct roff *r, int ln, char **bufp,
1.90 kristaps 725: size_t *szp, int pos, int *offs)
1.67 kristaps 726: {
727: enum rofft t;
1.109 kristaps 728: enum rofferr e;
1.130 kristaps 729: int ppos, ctl;
1.79 kristaps 730:
1.230 schwarze 731: /* Handle in-line equation delimiters. */
732:
733: if (r->last_eqn != NULL && r->last_eqn->delim &&
734: (r->eqn == NULL || r->eqn_inline)) {
735: e = roff_eqndelim(r, bufp, szp, pos);
736: if (e == ROFF_REPARSE)
737: return(e);
738: assert(e == ROFF_CONT);
739: }
740:
741: /* Expand some escape sequences. */
1.94 kristaps 742:
1.172 schwarze 743: e = roff_res(r, bufp, szp, ln, pos);
744: if (ROFF_IGN == e)
745: return(e);
746: assert(ROFF_CONT == e);
1.94 kristaps 747:
1.130 kristaps 748: ppos = pos;
1.174 kristaps 749: ctl = roff_getcontrol(r, *bufp, &pos);
1.130 kristaps 750:
1.94 kristaps 751: /*
1.79 kristaps 752: * First, if a scope is open and we're not a macro, pass the
753: * text through the macro's filter. If a scope isn't open and
754: * we're not a macro, just let it through.
1.125 kristaps 755: * Finally, if there's an equation scope open, divert it into it
756: * no matter our state.
1.79 kristaps 757: */
1.74 kristaps 758:
1.130 kristaps 759: if (r->last && ! ctl) {
1.78 kristaps 760: t = r->last->tok;
761: assert(roffs[t].text);
1.207 schwarze 762: e = (*roffs[t].text)(r, t, bufp, szp, ln, pos, pos, offs);
1.109 kristaps 763: assert(ROFF_IGN == e || ROFF_CONT == e);
1.125 kristaps 764: if (ROFF_CONT != e)
765: return(e);
1.182 schwarze 766: }
767: if (r->eqn)
768: return(eqn_read(&r->eqn, ln, *bufp, ppos, offs));
769: if ( ! ctl) {
1.125 kristaps 770: if (r->tbl)
1.130 kristaps 771: return(tbl_read(r->tbl, ln, *bufp, pos));
1.178 schwarze 772: return(roff_parsetext(bufp, szp, pos, offs));
1.182 schwarze 773: }
1.228 schwarze 774:
775: /* Skip empty request lines. */
776:
777: if ((*bufp)[pos] == '"') {
778: mandoc_msg(MANDOCERR_COMMENT_BAD, r->parse,
779: ln, pos, NULL);
780: return(ROFF_IGN);
781: } else if ((*bufp)[pos] == '\0')
782: return(ROFF_IGN);
1.67 kristaps 783:
1.79 kristaps 784: /*
785: * If a scope is open, go to the child handler for that macro,
786: * as it may want to preprocess before doing anything with it.
1.125 kristaps 787: * Don't do so if an equation is open.
1.79 kristaps 788: */
1.78 kristaps 789:
1.79 kristaps 790: if (r->last) {
791: t = r->last->tok;
792: assert(roffs[t].sub);
1.207 schwarze 793: return((*roffs[t].sub)(r, t, bufp, szp,
794: ln, ppos, pos, offs));
1.79 kristaps 795: }
1.78 kristaps 796:
1.79 kristaps 797: /*
798: * Lastly, as we've no scope open, try to look up and execute
799: * the new macro. If no macro is found, simply return and let
800: * the compilers handle it.
801: */
1.67 kristaps 802:
1.214 schwarze 803: if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos, ln, ppos)))
1.79 kristaps 804: return(ROFF_CONT);
1.67 kristaps 805:
1.75 kristaps 806: assert(roffs[t].proc);
1.207 schwarze 807: return((*roffs[t].proc)(r, t, bufp, szp, ln, ppos, pos, offs));
1.74 kristaps 808: }
809:
1.117 kristaps 810: void
1.74 kristaps 811: roff_endparse(struct roff *r)
812: {
813:
1.110 kristaps 814: if (r->last)
1.221 schwarze 815: mandoc_msg(MANDOCERR_BLK_NOEND, r->parse,
816: r->last->line, r->last->col,
817: roffs[r->last->tok].name);
1.117 kristaps 818:
1.125 kristaps 819: if (r->eqn) {
1.221 schwarze 820: mandoc_msg(MANDOCERR_BLK_NOEND, r->parse,
821: r->eqn->eqn.ln, r->eqn->eqn.pos, "EQ");
1.151 kristaps 822: eqn_end(&r->eqn);
1.125 kristaps 823: }
824:
1.117 kristaps 825: if (r->tbl) {
1.221 schwarze 826: mandoc_msg(MANDOCERR_BLK_NOEND, r->parse,
827: r->tbl->line, r->tbl->pos, "TS");
1.151 kristaps 828: tbl_end(&r->tbl);
1.117 kristaps 829: }
1.67 kristaps 830: }
831:
832: /*
833: * Parse a roff node's type from the input buffer. This must be in the
834: * form of ".foo xxx" in the usual way.
835: */
836: static enum rofft
1.214 schwarze 837: roff_parse(struct roff *r, char *buf, int *pos, int ln, int ppos)
1.67 kristaps 838: {
1.214 schwarze 839: char *cp;
1.106 kristaps 840: const char *mac;
841: size_t maclen;
1.67 kristaps 842: enum rofft t;
843:
1.214 schwarze 844: cp = buf + *pos;
845:
846: if ('\0' == *cp || '"' == *cp || '\t' == *cp || ' ' == *cp)
1.67 kristaps 847: return(ROFF_MAX);
848:
1.214 schwarze 849: mac = cp;
850: maclen = roff_getname(r, &cp, ln, ppos);
1.67 kristaps 851:
1.106 kristaps 852: t = (r->current_string = roff_getstrn(r, mac, maclen))
1.155 kristaps 853: ? ROFF_USERDEF : roffhash_find(mac, maclen);
1.67 kristaps 854:
1.214 schwarze 855: if (ROFF_MAX != t)
856: *pos = cp - buf;
1.67 kristaps 857:
858: return(t);
859: }
860:
861: static enum rofferr
1.76 kristaps 862: roff_cblock(ROFF_ARGS)
1.67 kristaps 863: {
864:
1.79 kristaps 865: /*
866: * A block-close `..' should only be invoked as a child of an
867: * ignore macro, otherwise raise a warning and just ignore it.
868: */
869:
1.76 kristaps 870: if (NULL == r->last) {
1.221 schwarze 871: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
872: ln, ppos, "..");
1.76 kristaps 873: return(ROFF_IGN);
874: }
1.67 kristaps 875:
1.81 kristaps 876: switch (r->last->tok) {
1.207 schwarze 877: case ROFF_am:
1.220 schwarze 878: /* ROFF_am1 is remapped to ROFF_am in roff_block(). */
1.81 kristaps 879: /* FALLTHROUGH */
1.207 schwarze 880: case ROFF_ami:
1.81 kristaps 881: /* FALLTHROUGH */
1.207 schwarze 882: case ROFF_de:
1.108 schwarze 883: /* ROFF_de1 is remapped to ROFF_de in roff_block(). */
1.81 kristaps 884: /* FALLTHROUGH */
1.207 schwarze 885: case ROFF_dei:
1.81 kristaps 886: /* FALLTHROUGH */
1.207 schwarze 887: case ROFF_ig:
1.81 kristaps 888: break;
889: default:
1.221 schwarze 890: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
891: ln, ppos, "..");
1.67 kristaps 892: return(ROFF_IGN);
1.76 kristaps 893: }
1.67 kristaps 894:
1.76 kristaps 895: if ((*bufp)[pos])
1.217 schwarze 896: mandoc_vmsg(MANDOCERR_ARG_SKIP, r->parse, ln, pos,
897: ".. %s", *bufp + pos);
1.71 kristaps 898:
899: roffnode_pop(r);
1.76 kristaps 900: roffnode_cleanscope(r);
901: return(ROFF_IGN);
1.71 kristaps 902:
1.67 kristaps 903: }
904:
1.76 kristaps 905: static void
906: roffnode_cleanscope(struct roff *r)
1.67 kristaps 907: {
908:
1.76 kristaps 909: while (r->last) {
1.173 schwarze 910: if (--r->last->endspan != 0)
1.76 kristaps 911: break;
912: roffnode_pop(r);
913: }
1.67 kristaps 914: }
915:
1.195 schwarze 916: static void
917: roff_ccond(struct roff *r, int ln, int ppos)
1.74 kristaps 918: {
919:
1.76 kristaps 920: if (NULL == r->last) {
1.221 schwarze 921: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
922: ln, ppos, "\\}");
1.195 schwarze 923: return;
1.76 kristaps 924: }
925:
1.82 kristaps 926: switch (r->last->tok) {
1.207 schwarze 927: case ROFF_el:
1.82 kristaps 928: /* FALLTHROUGH */
1.207 schwarze 929: case ROFF_ie:
1.82 kristaps 930: /* FALLTHROUGH */
1.207 schwarze 931: case ROFF_if:
1.82 kristaps 932: break;
933: default:
1.221 schwarze 934: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
935: ln, ppos, "\\}");
1.195 schwarze 936: return;
1.75 kristaps 937: }
938:
1.76 kristaps 939: if (r->last->endspan > -1) {
1.221 schwarze 940: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
941: ln, ppos, "\\}");
1.195 schwarze 942: return;
1.76 kristaps 943: }
944:
1.75 kristaps 945: roffnode_pop(r);
1.76 kristaps 946: roffnode_cleanscope(r);
1.195 schwarze 947: return;
1.76 kristaps 948: }
949:
950: static enum rofferr
1.80 kristaps 951: roff_block(ROFF_ARGS)
1.76 kristaps 952: {
1.220 schwarze 953: const char *name;
954: char *iname, *cp;
1.213 schwarze 955: size_t namesz;
1.106 kristaps 956:
1.220 schwarze 957: /* Ignore groff compatibility mode for now. */
1.76 kristaps 958:
1.220 schwarze 959: if (ROFF_de1 == tok)
960: tok = ROFF_de;
961: else if (ROFF_am1 == tok)
962: tok = ROFF_am;
963:
964: /* Parse the macro name argument. */
965:
966: cp = *bufp + pos;
967: if (ROFF_ig == tok) {
968: iname = NULL;
969: namesz = 0;
970: } else {
971: iname = cp;
972: namesz = roff_getname(r, &cp, ln, ppos);
973: iname[namesz] = '\0';
974: }
1.107 kristaps 975:
1.220 schwarze 976: /* Resolve the macro name argument if it is indirect. */
1.107 kristaps 977:
1.220 schwarze 978: if (namesz && (ROFF_dei == tok || ROFF_ami == tok)) {
979: if (NULL == (name = roff_getstrn(r, iname, namesz))) {
980: mandoc_vmsg(MANDOCERR_STR_UNDEF,
981: r->parse, ln, (int)(iname - *bufp),
982: "%.*s", (int)namesz, iname);
983: namesz = 0;
984: } else
985: namesz = strlen(name);
986: } else
987: name = iname;
1.107 kristaps 988:
1.220 schwarze 989: if (0 == namesz && ROFF_ig != tok) {
990: mandoc_msg(MANDOCERR_REQ_EMPTY, r->parse,
991: ln, ppos, roffs[tok].name);
992: return(ROFF_IGN);
993: }
1.80 kristaps 994:
1.106 kristaps 995: roffnode_push(r, tok, name, ln, ppos);
996:
997: /*
998: * At the beginning of a `de' macro, clear the existing string
999: * with the same name, if there is one. New content will be
1.193 schwarze 1000: * appended from roff_block_text() in multiline mode.
1.106 kristaps 1001: */
1.107 kristaps 1002:
1.220 schwarze 1003: if (ROFF_de == tok || ROFF_dei == tok)
1.213 schwarze 1004: roff_setstrn(&r->strtab, name, namesz, "", 0, 0);
1.76 kristaps 1005:
1.213 schwarze 1006: if ('\0' == *cp)
1.78 kristaps 1007: return(ROFF_IGN);
1008:
1.220 schwarze 1009: /* Get the custom end marker. */
1.107 kristaps 1010:
1.220 schwarze 1011: iname = cp;
1.213 schwarze 1012: namesz = roff_getname(r, &cp, ln, ppos);
1.220 schwarze 1013:
1014: /* Resolve the end marker if it is indirect. */
1015:
1016: if (namesz && (ROFF_dei == tok || ROFF_ami == tok)) {
1017: if (NULL == (name = roff_getstrn(r, iname, namesz))) {
1018: mandoc_vmsg(MANDOCERR_STR_UNDEF,
1019: r->parse, ln, (int)(iname - *bufp),
1020: "%.*s", (int)namesz, iname);
1021: namesz = 0;
1022: } else
1023: namesz = strlen(name);
1024: } else
1025: name = iname;
1026:
1.213 schwarze 1027: if (namesz)
1028: r->last->end = mandoc_strndup(name, namesz);
1.78 kristaps 1029:
1.213 schwarze 1030: if ('\0' != *cp)
1.217 schwarze 1031: mandoc_vmsg(MANDOCERR_ARG_EXCESS, r->parse,
1032: ln, pos, ".%s ... %s", roffs[tok].name, cp);
1.74 kristaps 1033:
1.78 kristaps 1034: return(ROFF_IGN);
1035: }
1036:
1037: static enum rofferr
1.80 kristaps 1038: roff_block_sub(ROFF_ARGS)
1.79 kristaps 1039: {
1040: enum rofft t;
1041: int i, j;
1042:
1043: /*
1044: * First check whether a custom macro exists at this level. If
1045: * it does, then check against it. This is some of groff's
1046: * stranger behaviours. If we encountered a custom end-scope
1047: * tag and that tag also happens to be a "real" macro, then we
1048: * need to try interpreting it again as a real macro. If it's
1049: * not, then return ignore. Else continue.
1050: */
1051:
1052: if (r->last->end) {
1.130 kristaps 1053: for (i = pos, j = 0; r->last->end[j]; j++, i++)
1.79 kristaps 1054: if ((*bufp)[i] != r->last->end[j])
1055: break;
1056:
1.207 schwarze 1057: if ('\0' == r->last->end[j] &&
1058: ('\0' == (*bufp)[i] ||
1059: ' ' == (*bufp)[i] ||
1060: '\t' == (*bufp)[i])) {
1.79 kristaps 1061: roffnode_pop(r);
1062: roffnode_cleanscope(r);
1063:
1.130 kristaps 1064: while (' ' == (*bufp)[i] || '\t' == (*bufp)[i])
1065: i++;
1066:
1067: pos = i;
1.214 schwarze 1068: if (ROFF_MAX != roff_parse(r, *bufp, &pos, ln, ppos))
1.79 kristaps 1069: return(ROFF_RERUN);
1070: return(ROFF_IGN);
1071: }
1072: }
1073:
1074: /*
1075: * If we have no custom end-query or lookup failed, then try
1076: * pulling it out of the hashtable.
1077: */
1078:
1.214 schwarze 1079: t = roff_parse(r, *bufp, &pos, ln, ppos);
1.79 kristaps 1080:
1.106 kristaps 1081: if (ROFF_cblock != t) {
1.220 schwarze 1082: if (ROFF_ig != tok)
1.193 schwarze 1083: roff_setstr(r, r->last->name, *bufp + ppos, 2);
1.79 kristaps 1084: return(ROFF_IGN);
1.106 kristaps 1085: }
1.79 kristaps 1086:
1087: assert(roffs[t].proc);
1.207 schwarze 1088: return((*roffs[t].proc)(r, t, bufp, szp, ln, ppos, pos, offs));
1.79 kristaps 1089: }
1090:
1091: static enum rofferr
1.80 kristaps 1092: roff_block_text(ROFF_ARGS)
1.78 kristaps 1093: {
1094:
1.220 schwarze 1095: if (ROFF_ig != tok)
1.193 schwarze 1096: roff_setstr(r, r->last->name, *bufp + pos, 2);
1.106 kristaps 1097:
1.78 kristaps 1098: return(ROFF_IGN);
1099: }
1100:
1101: static enum rofferr
1.82 kristaps 1102: roff_cond_sub(ROFF_ARGS)
1103: {
1104: enum rofft t;
1.139 kristaps 1105: char *ep;
1.198 schwarze 1106: int rr;
1.82 kristaps 1107:
1108: rr = r->last->rule;
1.139 kristaps 1109: roffnode_cleanscope(r);
1.214 schwarze 1110: t = roff_parse(r, *bufp, &pos, ln, ppos);
1.82 kristaps 1111:
1.139 kristaps 1112: /*
1.177 schwarze 1113: * Fully handle known macros when they are structurally
1114: * required or when the conditional evaluated to true.
1.87 kristaps 1115: */
1116:
1.177 schwarze 1117: if ((ROFF_MAX != t) &&
1.198 schwarze 1118: (rr || ROFFMAC_STRUCT & roffs[t].flags)) {
1.177 schwarze 1119: assert(roffs[t].proc);
1120: return((*roffs[t].proc)(r, t, bufp, szp,
1.207 schwarze 1121: ln, ppos, pos, offs));
1.177 schwarze 1122: }
1.144 kristaps 1123:
1.196 schwarze 1124: /*
1125: * If `\}' occurs on a macro line without a preceding macro,
1126: * drop the line completely.
1127: */
1128:
1129: ep = *bufp + pos;
1130: if ('\\' == ep[0] && '}' == ep[1])
1.198 schwarze 1131: rr = 0;
1.196 schwarze 1132:
1.177 schwarze 1133: /* Always check for the closing delimiter `\}'. */
1.144 kristaps 1134:
1.177 schwarze 1135: while (NULL != (ep = strchr(ep, '\\'))) {
1.197 schwarze 1136: if ('}' == *(++ep)) {
1137: *ep = '&';
1138: roff_ccond(r, ln, ep - *bufp - 1);
1139: }
1140: ++ep;
1.177 schwarze 1141: }
1.198 schwarze 1142: return(rr ? ROFF_CONT : ROFF_IGN);
1.82 kristaps 1143: }
1144:
1145: static enum rofferr
1146: roff_cond_text(ROFF_ARGS)
1.78 kristaps 1147: {
1.140 kristaps 1148: char *ep;
1.198 schwarze 1149: int rr;
1.82 kristaps 1150:
1151: rr = r->last->rule;
1.140 kristaps 1152: roffnode_cleanscope(r);
1.82 kristaps 1153:
1.197 schwarze 1154: ep = *bufp + pos;
1155: while (NULL != (ep = strchr(ep, '\\'))) {
1156: if ('}' == *(++ep)) {
1157: *ep = '&';
1158: roff_ccond(r, ln, ep - *bufp - 1);
1159: }
1160: ++ep;
1.78 kristaps 1161: }
1.198 schwarze 1162: return(rr ? ROFF_CONT : ROFF_IGN);
1.74 kristaps 1163: }
1164:
1.204 schwarze 1165: /*
1166: * Parse a single signed integer number. Stop at the first non-digit.
1167: * If there is at least one digit, return success and advance the
1168: * parse point, else return failure and let the parse point unchanged.
1169: * Ignore overflows, treat them just like the C language.
1170: */
1.184 schwarze 1171: static int
1172: roff_getnum(const char *v, int *pos, int *res)
1173: {
1.206 schwarze 1174: int myres, n, p;
1175:
1176: if (NULL == res)
1177: res = &myres;
1.184 schwarze 1178:
1179: p = *pos;
1180: n = v[p] == '-';
1181: if (n)
1182: p++;
1183:
1184: for (*res = 0; isdigit((unsigned char)v[p]); p++)
1.204 schwarze 1185: *res = 10 * *res + v[p] - '0';
1.184 schwarze 1186: if (p == *pos + n)
1187: return 0;
1188:
1189: if (n)
1190: *res = -*res;
1191:
1192: *pos = p;
1193: return 1;
1194: }
1195:
1.198 schwarze 1196: /*
1197: * Evaluate a string comparison condition.
1198: * The first character is the delimiter.
1199: * Succeed if the string up to its second occurrence
1200: * matches the string up to its third occurence.
1201: * Advance the cursor after the third occurrence
1202: * or lacking that, to the end of the line.
1203: */
1204: static int
1205: roff_evalstrcond(const char *v, int *pos)
1206: {
1207: const char *s1, *s2, *s3;
1208: int match;
1209:
1210: match = 0;
1211: s1 = v + *pos; /* initial delimiter */
1212: s2 = s1 + 1; /* for scanning the first string */
1213: s3 = strchr(s2, *s1); /* for scanning the second string */
1214:
1215: if (NULL == s3) /* found no middle delimiter */
1216: goto out;
1217:
1218: while ('\0' != *++s3) {
1219: if (*s2 != *s3) { /* mismatch */
1220: s3 = strchr(s3, *s1);
1221: break;
1222: }
1223: if (*s3 == *s1) { /* found the final delimiter */
1224: match = 1;
1225: break;
1226: }
1227: s2++;
1228: }
1229:
1230: out:
1231: if (NULL == s3)
1232: s3 = strchr(s2, '\0');
1233: else
1234: s3++;
1235: *pos = s3 - v;
1236: return(match);
1237: }
1238:
1.204 schwarze 1239: /*
1240: * Evaluate an optionally negated single character, numerical,
1241: * or string condition.
1242: */
1.198 schwarze 1243: static int
1.234 kristaps 1244: roff_evalcond(struct roff *r, int ln, const char *v, int *pos)
1.88 kristaps 1245: {
1.204 schwarze 1246: int wanttrue, number;
1.88 kristaps 1247:
1.198 schwarze 1248: if ('!' == v[*pos]) {
1249: wanttrue = 0;
1250: (*pos)++;
1251: } else
1252: wanttrue = 1;
1253:
1.88 kristaps 1254: switch (v[*pos]) {
1.207 schwarze 1255: case 'n':
1.198 schwarze 1256: /* FALLTHROUGH */
1.207 schwarze 1257: case 'o':
1.88 kristaps 1258: (*pos)++;
1.198 schwarze 1259: return(wanttrue);
1.207 schwarze 1260: case 'c':
1.198 schwarze 1261: /* FALLTHROUGH */
1.207 schwarze 1262: case 'd':
1.198 schwarze 1263: /* FALLTHROUGH */
1.207 schwarze 1264: case 'e':
1.88 kristaps 1265: /* FALLTHROUGH */
1.207 schwarze 1266: case 'r':
1.88 kristaps 1267: /* FALLTHROUGH */
1.207 schwarze 1268: case 't':
1.88 kristaps 1269: (*pos)++;
1.198 schwarze 1270: return(!wanttrue);
1.88 kristaps 1271: default:
1272: break;
1273: }
1274:
1.234 kristaps 1275: if (roff_evalnum(r, ln, v, pos, &number, 0))
1.204 schwarze 1276: return((number > 0) == wanttrue);
1277: else
1.198 schwarze 1278: return(roff_evalstrcond(v, pos) == wanttrue);
1.88 kristaps 1279: }
1280:
1.74 kristaps 1281: static enum rofferr
1.103 kristaps 1282: roff_line_ignore(ROFF_ARGS)
1.89 kristaps 1283: {
1.123 schwarze 1284:
1.89 kristaps 1285: return(ROFF_IGN);
1286: }
1287:
1.104 kristaps 1288: static enum rofferr
1.82 kristaps 1289: roff_cond(ROFF_ARGS)
1.74 kristaps 1290: {
1.173 schwarze 1291:
1292: roffnode_push(r, tok, NULL, ln, ppos);
1.74 kristaps 1293:
1.207 schwarze 1294: /*
1.134 kristaps 1295: * An `.el' has no conditional body: it will consume the value
1296: * of the current rstack entry set in prior `ie' calls or
1.207 schwarze 1297: * defaults to DENY.
1.134 kristaps 1298: *
1299: * If we're not an `el', however, then evaluate the conditional.
1300: */
1.133 kristaps 1301:
1.173 schwarze 1302: r->last->rule = ROFF_el == tok ?
1.207 schwarze 1303: (r->rstackpos < 0 ? 0 : r->rstack[r->rstackpos--]) :
1.234 kristaps 1304: roff_evalcond(r, ln, *bufp, &pos);
1.77 kristaps 1305:
1.134 kristaps 1306: /*
1307: * An if-else will put the NEGATION of the current evaluated
1308: * conditional into the stack of rules.
1309: */
1310:
1.84 schwarze 1311: if (ROFF_ie == tok) {
1.223 schwarze 1312: if (r->rstackpos + 1 == r->rstacksz) {
1313: r->rstacksz += 16;
1314: r->rstack = mandoc_reallocarray(r->rstack,
1315: r->rstacksz, sizeof(int));
1.134 kristaps 1316: }
1.198 schwarze 1317: r->rstack[++r->rstackpos] = !r->last->rule;
1.82 kristaps 1318: }
1.88 kristaps 1319:
1320: /* If the parent has false as its rule, then so do we. */
1321:
1.198 schwarze 1322: if (r->last->parent && !r->last->parent->rule)
1323: r->last->rule = 0;
1.88 kristaps 1324:
1325: /*
1.173 schwarze 1326: * Determine scope.
1327: * If there is nothing on the line after the conditional,
1328: * not even whitespace, use next-line scope.
1.88 kristaps 1329: */
1.74 kristaps 1330:
1.173 schwarze 1331: if ('\0' == (*bufp)[pos]) {
1332: r->last->endspan = 2;
1333: goto out;
1334: }
1335:
1336: while (' ' == (*bufp)[pos])
1337: pos++;
1338:
1339: /* An opening brace requests multiline scope. */
1.75 kristaps 1340:
1341: if ('\\' == (*bufp)[pos] && '{' == (*bufp)[pos + 1]) {
1342: r->last->endspan = -1;
1343: pos += 2;
1.173 schwarze 1344: goto out;
1.207 schwarze 1345: }
1.74 kristaps 1346:
1.77 kristaps 1347: /*
1.173 schwarze 1348: * Anything else following the conditional causes
1349: * single-line scope. Warn if the scope contains
1350: * nothing but trailing whitespace.
1.77 kristaps 1351: */
1352:
1.75 kristaps 1353: if ('\0' == (*bufp)[pos])
1.216 schwarze 1354: mandoc_msg(MANDOCERR_COND_EMPTY, r->parse,
1355: ln, ppos, roffs[tok].name);
1.77 kristaps 1356:
1.173 schwarze 1357: r->last->endspan = 1;
1.74 kristaps 1358:
1.173 schwarze 1359: out:
1.75 kristaps 1360: *offs = pos;
1361: return(ROFF_RERUN);
1.83 schwarze 1362: }
1363:
1364: static enum rofferr
1.92 schwarze 1365: roff_ds(ROFF_ARGS)
1366: {
1.212 schwarze 1367: char *string;
1368: const char *name;
1369: size_t namesz;
1.96 kristaps 1370:
1371: /*
1.212 schwarze 1372: * The first word is the name of the string.
1373: * If it is empty or terminated by an escape sequence,
1374: * abort the `ds' request without defining anything.
1.96 kristaps 1375: */
1.92 schwarze 1376:
1.212 schwarze 1377: name = string = *bufp + pos;
1.92 schwarze 1378: if ('\0' == *name)
1379: return(ROFF_IGN);
1380:
1.212 schwarze 1381: namesz = roff_getname(r, &string, ln, pos);
1382: if ('\\' == name[namesz])
1383: return(ROFF_IGN);
1384:
1385: /* Read past the initial double-quote, if any. */
1.121 schwarze 1386: if ('"' == *string)
1.92 schwarze 1387: string++;
1388:
1.96 kristaps 1389: /* The rest is the value. */
1.212 schwarze 1390: roff_setstrn(&r->strtab, name, namesz, string, strlen(string),
1391: ROFF_as == tok);
1.92 schwarze 1392: return(ROFF_IGN);
1393: }
1394:
1.204 schwarze 1395: /*
1396: * Parse a single operator, one or two characters long.
1397: * If the operator is recognized, return success and advance the
1398: * parse point, else return failure and let the parse point unchanged.
1399: */
1400: static int
1401: roff_getop(const char *v, int *pos, char *res)
1402: {
1403:
1404: *res = v[*pos];
1405:
1406: switch (*res) {
1.207 schwarze 1407: case '+':
1.204 schwarze 1408: /* FALLTHROUGH */
1.207 schwarze 1409: case '-':
1.204 schwarze 1410: /* FALLTHROUGH */
1.207 schwarze 1411: case '*':
1.204 schwarze 1412: /* FALLTHROUGH */
1.207 schwarze 1413: case '/':
1.204 schwarze 1414: /* FALLTHROUGH */
1.207 schwarze 1415: case '%':
1.204 schwarze 1416: /* FALLTHROUGH */
1.207 schwarze 1417: case '&':
1.204 schwarze 1418: /* FALLTHROUGH */
1.207 schwarze 1419: case ':':
1.204 schwarze 1420: break;
1421: case '<':
1422: switch (v[*pos + 1]) {
1.207 schwarze 1423: case '=':
1.204 schwarze 1424: *res = 'l';
1425: (*pos)++;
1426: break;
1.207 schwarze 1427: case '>':
1.204 schwarze 1428: *res = '!';
1429: (*pos)++;
1430: break;
1.207 schwarze 1431: case '?':
1.204 schwarze 1432: *res = 'i';
1433: (*pos)++;
1434: break;
1435: default:
1436: break;
1437: }
1438: break;
1439: case '>':
1440: switch (v[*pos + 1]) {
1.207 schwarze 1441: case '=':
1.204 schwarze 1442: *res = 'g';
1443: (*pos)++;
1444: break;
1.207 schwarze 1445: case '?':
1.204 schwarze 1446: *res = 'a';
1447: (*pos)++;
1448: break;
1449: default:
1450: break;
1451: }
1452: break;
1453: case '=':
1454: if ('=' == v[*pos + 1])
1455: (*pos)++;
1456: break;
1457: default:
1458: return(0);
1459: }
1460: (*pos)++;
1461:
1462: return(*res);
1463: }
1464:
1465: /*
1466: * Evaluate either a parenthesized numeric expression
1467: * or a single signed integer number.
1468: */
1469: static int
1.235 ! schwarze 1470: roff_evalpar(struct roff *r, int ln,
1.234 kristaps 1471: const char *v, int *pos, int *res)
1.204 schwarze 1472: {
1473:
1474: if ('(' != v[*pos])
1475: return(roff_getnum(v, pos, res));
1476:
1477: (*pos)++;
1.234 kristaps 1478: if ( ! roff_evalnum(r, ln, v, pos, res, 1))
1.204 schwarze 1479: return(0);
1480:
1.206 schwarze 1481: /*
1482: * Omission of the closing parenthesis
1483: * is an error in validation mode,
1484: * but ignored in evaluation mode.
1485: */
1486:
1.204 schwarze 1487: if (')' == v[*pos])
1488: (*pos)++;
1.206 schwarze 1489: else if (NULL == res)
1490: return(0);
1.204 schwarze 1491:
1492: return(1);
1493: }
1494:
1495: /*
1496: * Evaluate a complete numeric expression.
1497: * Proceed left to right, there is no concept of precedence.
1498: */
1499: static int
1.235 ! schwarze 1500: roff_evalnum(struct roff *r, int ln, const char *v,
1.234 kristaps 1501: int *pos, int *res, int skipwhite)
1.204 schwarze 1502: {
1503: int mypos, operand2;
1504: char operator;
1505:
1506: if (NULL == pos) {
1507: mypos = 0;
1508: pos = &mypos;
1509: }
1510:
1511: if (skipwhite)
1512: while (isspace((unsigned char)v[*pos]))
1513: (*pos)++;
1514:
1.234 kristaps 1515: if ( ! roff_evalpar(r, ln, v, pos, res))
1.204 schwarze 1516: return(0);
1517:
1518: while (1) {
1519: if (skipwhite)
1520: while (isspace((unsigned char)v[*pos]))
1521: (*pos)++;
1522:
1523: if ( ! roff_getop(v, pos, &operator))
1524: break;
1525:
1526: if (skipwhite)
1527: while (isspace((unsigned char)v[*pos]))
1528: (*pos)++;
1529:
1.234 kristaps 1530: if ( ! roff_evalpar(r, ln, v, pos, &operand2))
1.204 schwarze 1531: return(0);
1532:
1533: if (skipwhite)
1534: while (isspace((unsigned char)v[*pos]))
1535: (*pos)++;
1.206 schwarze 1536:
1537: if (NULL == res)
1538: continue;
1.204 schwarze 1539:
1540: switch (operator) {
1.207 schwarze 1541: case '+':
1.204 schwarze 1542: *res += operand2;
1543: break;
1.207 schwarze 1544: case '-':
1.204 schwarze 1545: *res -= operand2;
1546: break;
1.207 schwarze 1547: case '*':
1.204 schwarze 1548: *res *= operand2;
1549: break;
1.207 schwarze 1550: case '/':
1.234 kristaps 1551: if (0 == operand2) {
1.235 ! schwarze 1552: mandoc_msg(MANDOCERR_DIVZERO,
1.234 kristaps 1553: r->parse, ln, *pos, v);
1554: *res = 0;
1555: break;
1556: }
1.204 schwarze 1557: *res /= operand2;
1558: break;
1.207 schwarze 1559: case '%':
1.204 schwarze 1560: *res %= operand2;
1561: break;
1.207 schwarze 1562: case '<':
1.204 schwarze 1563: *res = *res < operand2;
1564: break;
1.207 schwarze 1565: case '>':
1.204 schwarze 1566: *res = *res > operand2;
1567: break;
1.207 schwarze 1568: case 'l':
1.204 schwarze 1569: *res = *res <= operand2;
1570: break;
1.207 schwarze 1571: case 'g':
1.204 schwarze 1572: *res = *res >= operand2;
1573: break;
1.207 schwarze 1574: case '=':
1.204 schwarze 1575: *res = *res == operand2;
1576: break;
1.207 schwarze 1577: case '!':
1.204 schwarze 1578: *res = *res != operand2;
1579: break;
1.207 schwarze 1580: case '&':
1.204 schwarze 1581: *res = *res && operand2;
1582: break;
1.207 schwarze 1583: case ':':
1.204 schwarze 1584: *res = *res || operand2;
1585: break;
1.207 schwarze 1586: case 'i':
1.204 schwarze 1587: if (operand2 < *res)
1588: *res = operand2;
1589: break;
1.207 schwarze 1590: case 'a':
1.204 schwarze 1591: if (operand2 > *res)
1592: *res = operand2;
1593: break;
1594: default:
1595: abort();
1596: }
1597: }
1598: return(1);
1599: }
1600:
1.180 schwarze 1601: void
1.187 schwarze 1602: roff_setreg(struct roff *r, const char *name, int val, char sign)
1.147 kristaps 1603: {
1.180 schwarze 1604: struct roffreg *reg;
1605:
1606: /* Search for an existing register with the same name. */
1607: reg = r->regtab;
1608:
1609: while (reg && strcmp(name, reg->key.p))
1610: reg = reg->next;
1.147 kristaps 1611:
1.180 schwarze 1612: if (NULL == reg) {
1613: /* Create a new register. */
1614: reg = mandoc_malloc(sizeof(struct roffreg));
1615: reg->key.p = mandoc_strdup(name);
1616: reg->key.sz = strlen(name);
1.187 schwarze 1617: reg->val = 0;
1.180 schwarze 1618: reg->next = r->regtab;
1619: r->regtab = reg;
1620: }
1621:
1.187 schwarze 1622: if ('+' == sign)
1623: reg->val += val;
1624: else if ('-' == sign)
1625: reg->val -= val;
1626: else
1627: reg->val = val;
1.147 kristaps 1628: }
1629:
1.192 schwarze 1630: /*
1631: * Handle some predefined read-only number registers.
1632: * For now, return -1 if the requested register is not predefined;
1633: * in case a predefined read-only register having the value -1
1634: * were to turn up, another special value would have to be chosen.
1635: */
1636: static int
1637: roff_getregro(const char *name)
1638: {
1639:
1640: switch (*name) {
1.207 schwarze 1641: case 'A': /* ASCII approximation mode is always off. */
1.192 schwarze 1642: return(0);
1.207 schwarze 1643: case 'g': /* Groff compatibility mode is always on. */
1.192 schwarze 1644: return(1);
1.207 schwarze 1645: case 'H': /* Fixed horizontal resolution. */
1.192 schwarze 1646: return (24);
1.207 schwarze 1647: case 'j': /* Always adjust left margin only. */
1.192 schwarze 1648: return(0);
1.207 schwarze 1649: case 'T': /* Some output device is always defined. */
1.192 schwarze 1650: return(1);
1.207 schwarze 1651: case 'V': /* Fixed vertical resolution. */
1.192 schwarze 1652: return (40);
1653: default:
1654: return (-1);
1655: }
1656: }
1657:
1.181 schwarze 1658: int
1.180 schwarze 1659: roff_getreg(const struct roff *r, const char *name)
1.147 kristaps 1660: {
1.180 schwarze 1661: struct roffreg *reg;
1.192 schwarze 1662: int val;
1663:
1664: if ('.' == name[0] && '\0' != name[1] && '\0' == name[2]) {
1665: val = roff_getregro(name + 1);
1666: if (-1 != val)
1667: return (val);
1668: }
1.180 schwarze 1669:
1670: for (reg = r->regtab; reg; reg = reg->next)
1671: if (0 == strcmp(name, reg->key.p))
1.181 schwarze 1672: return(reg->val);
1673:
1674: return(0);
1675: }
1676:
1677: static int
1678: roff_getregn(const struct roff *r, const char *name, size_t len)
1679: {
1680: struct roffreg *reg;
1.192 schwarze 1681: int val;
1682:
1683: if ('.' == name[0] && 2 == len) {
1684: val = roff_getregro(name + 1);
1685: if (-1 != val)
1686: return (val);
1687: }
1.181 schwarze 1688:
1689: for (reg = r->regtab; reg; reg = reg->next)
1690: if (len == reg->key.sz &&
1691: 0 == strncmp(name, reg->key.p, len))
1692: return(reg->val);
1.147 kristaps 1693:
1.180 schwarze 1694: return(0);
1.147 kristaps 1695: }
1696:
1.180 schwarze 1697: static void
1698: roff_freereg(struct roffreg *reg)
1.147 kristaps 1699: {
1.180 schwarze 1700: struct roffreg *old_reg;
1.147 kristaps 1701:
1.180 schwarze 1702: while (NULL != reg) {
1703: free(reg->key.p);
1704: old_reg = reg;
1705: reg = reg->next;
1706: free(old_reg);
1707: }
1.147 kristaps 1708: }
1.92 schwarze 1709:
1710: static enum rofferr
1.89 kristaps 1711: roff_nr(ROFF_ARGS)
1.83 schwarze 1712: {
1.212 schwarze 1713: char *key, *val;
1714: size_t keysz;
1.138 kristaps 1715: int iv;
1.187 schwarze 1716: char sign;
1.89 kristaps 1717:
1.212 schwarze 1718: key = val = *bufp + pos;
1719: if ('\0' == *key)
1720: return(ROFF_IGN);
1721:
1722: keysz = roff_getname(r, &val, ln, pos);
1723: if ('\\' == key[keysz])
1724: return(ROFF_IGN);
1725: key[keysz] = '\0';
1.89 kristaps 1726:
1.187 schwarze 1727: sign = *val;
1728: if ('+' == sign || '-' == sign)
1729: val++;
1730:
1.234 kristaps 1731: if (roff_evalnum(r, ln, val, NULL, &iv, 0))
1.204 schwarze 1732: roff_setreg(r, key, iv, sign);
1.109 kristaps 1733:
1.203 schwarze 1734: return(ROFF_IGN);
1735: }
1736:
1737: static enum rofferr
1738: roff_rr(ROFF_ARGS)
1739: {
1740: struct roffreg *reg, **prev;
1.212 schwarze 1741: char *name, *cp;
1742: size_t namesz;
1.203 schwarze 1743:
1.212 schwarze 1744: name = cp = *bufp + pos;
1745: if ('\0' == *name)
1746: return(ROFF_IGN);
1747: namesz = roff_getname(r, &cp, ln, pos);
1748: name[namesz] = '\0';
1.203 schwarze 1749:
1750: prev = &r->regtab;
1751: while (1) {
1752: reg = *prev;
1753: if (NULL == reg || !strcmp(name, reg->key.p))
1754: break;
1755: prev = ®->next;
1756: }
1757: if (NULL != reg) {
1758: *prev = reg->next;
1759: free(reg->key.p);
1760: free(reg);
1761: }
1.122 schwarze 1762: return(ROFF_IGN);
1763: }
1764:
1765: static enum rofferr
1766: roff_rm(ROFF_ARGS)
1767: {
1768: const char *name;
1769: char *cp;
1.212 schwarze 1770: size_t namesz;
1.122 schwarze 1771:
1772: cp = *bufp + pos;
1773: while ('\0' != *cp) {
1.212 schwarze 1774: name = cp;
1775: namesz = roff_getname(r, &cp, ln, (int)(cp - *bufp));
1776: roff_setstrn(&r->strtab, name, namesz, NULL, 0, 0);
1777: if ('\\' == name[namesz])
1778: break;
1.122 schwarze 1779: }
1.178 schwarze 1780: return(ROFF_IGN);
1781: }
1782:
1783: static enum rofferr
1784: roff_it(ROFF_ARGS)
1785: {
1786: char *cp;
1787: size_t len;
1788: int iv;
1789:
1790: /* Parse the number of lines. */
1791: cp = *bufp + pos;
1792: len = strcspn(cp, " \t");
1793: cp[len] = '\0';
1794: if ((iv = mandoc_strntoi(cp, len, 10)) <= 0) {
1.222 schwarze 1795: mandoc_msg(MANDOCERR_IT_NONUM, r->parse,
1.207 schwarze 1796: ln, ppos, *bufp + 1);
1.178 schwarze 1797: return(ROFF_IGN);
1798: }
1799: cp += len + 1;
1800:
1801: /* Arm the input line trap. */
1802: roffit_lines = iv;
1803: roffit_macro = mandoc_strdup(cp);
1.109 kristaps 1804: return(ROFF_IGN);
1.175 schwarze 1805: }
1806:
1807: static enum rofferr
1808: roff_Dd(ROFF_ARGS)
1809: {
1810: const char *const *cp;
1811:
1.227 schwarze 1812: if ((r->options & (MPARSE_MDOC | MPARSE_QUICK)) == 0)
1.175 schwarze 1813: for (cp = __mdoc_reserved; *cp; cp++)
1814: roff_setstr(r, *cp, NULL, 0);
1815:
1.227 schwarze 1816: if (r->format == 0)
1817: r->format = MPARSE_MDOC;
1818:
1.175 schwarze 1819: return(ROFF_CONT);
1820: }
1821:
1822: static enum rofferr
1823: roff_TH(ROFF_ARGS)
1824: {
1825: const char *const *cp;
1826:
1.227 schwarze 1827: if ((r->options & MPARSE_QUICK) == 0)
1.175 schwarze 1828: for (cp = __man_reserved; *cp; cp++)
1829: roff_setstr(r, *cp, NULL, 0);
1830:
1.227 schwarze 1831: if (r->format == 0)
1832: r->format = MPARSE_MAN;
1833:
1.175 schwarze 1834: return(ROFF_CONT);
1.109 kristaps 1835: }
1836:
1837: static enum rofferr
1838: roff_TE(ROFF_ARGS)
1839: {
1840:
1841: if (NULL == r->tbl)
1.221 schwarze 1842: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
1843: ln, ppos, "TE");
1.115 kristaps 1844: else
1.151 kristaps 1845: tbl_end(&r->tbl);
1.109 kristaps 1846:
1.112 kristaps 1847: return(ROFF_IGN);
1848: }
1849:
1850: static enum rofferr
1851: roff_T_(ROFF_ARGS)
1852: {
1853:
1854: if (NULL == r->tbl)
1.221 schwarze 1855: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
1856: ln, ppos, "T&");
1.112 kristaps 1857: else
1.116 kristaps 1858: tbl_restart(ppos, ln, r->tbl);
1.112 kristaps 1859:
1.109 kristaps 1860: return(ROFF_IGN);
1861: }
1862:
1.230 schwarze 1863: /*
1864: * Handle in-line equation delimiters.
1865: */
1866: static enum rofferr
1867: roff_eqndelim(struct roff *r, char **bufp, size_t *szp, int pos)
1.151 kristaps 1868: {
1.233 schwarze 1869: char *cp1, *cp2;
1870: const char *bef_pr, *bef_nl, *mac, *aft_nl, *aft_pr;
1.151 kristaps 1871:
1.230 schwarze 1872: /*
1873: * Outside equations, look for an opening delimiter.
1874: * If we are inside an equation, we already know it is
1875: * in-line, or this function wouldn't have been called;
1876: * so look for a closing delimiter.
1877: */
1878:
1879: cp1 = *bufp + pos;
1880: cp2 = strchr(cp1, r->eqn == NULL ?
1881: r->last_eqn->odelim : r->last_eqn->cdelim);
1882: if (cp2 == NULL)
1883: return(ROFF_CONT);
1884:
1.233 schwarze 1885: *cp2++ = '\0';
1886: bef_pr = bef_nl = aft_nl = aft_pr = "";
1887:
1888: /* Handle preceding text, protecting whitespace. */
1889:
1890: if (**bufp != '\0') {
1891: if (r->eqn == NULL)
1892: bef_pr = "\\&";
1893: bef_nl = "\n";
1894: }
1895:
1896: /*
1897: * Prepare replacing the delimiter with an equation macro
1898: * and drop leading white space from the equation.
1899: */
1.230 schwarze 1900:
1.233 schwarze 1901: if (r->eqn == NULL) {
1902: while (*cp2 == ' ')
1903: cp2++;
1904: mac = ".EQ";
1905: } else
1906: mac = ".EN";
1907:
1908: /* Handle following text, protecting whitespace. */
1909:
1910: if (*cp2 != '\0') {
1911: aft_nl = "\n";
1912: if (r->eqn != NULL)
1913: aft_pr = "\\&";
1914: }
1915:
1916: /* Do the actual replacement. */
1917:
1918: *szp = mandoc_asprintf(&cp1, "%s%s%s%s%s%s%s", *bufp,
1919: bef_pr, bef_nl, mac, aft_nl, aft_pr, cp2) + 1;
1.230 schwarze 1920: free(*bufp);
1921: *bufp = cp1;
1922:
1923: /* Toggle the in-line state of the eqn subsystem. */
1924:
1925: r->eqn_inline = r->eqn == NULL;
1926: return(ROFF_REPARSE);
1.151 kristaps 1927: }
1928:
1.235 ! schwarze 1929: static enum rofferr
! 1930: roff_EQ(ROFF_ARGS)
1.125 kristaps 1931: {
1.151 kristaps 1932: struct eqn_node *e;
1.125 kristaps 1933:
1934: assert(NULL == r->eqn);
1.235 ! schwarze 1935: e = eqn_alloc(ppos, ln, r->parse);
1.125 kristaps 1936:
1.230 schwarze 1937: if (r->last_eqn) {
1.125 kristaps 1938: r->last_eqn->next = e;
1.230 schwarze 1939: e->delim = r->last_eqn->delim;
1940: e->odelim = r->last_eqn->odelim;
1941: e->cdelim = r->last_eqn->cdelim;
1942: } else
1.125 kristaps 1943: r->first_eqn = r->last_eqn = e;
1944:
1945: r->eqn = r->last_eqn = e;
1.151 kristaps 1946:
1.235 ! schwarze 1947: if ((*bufp)[pos])
! 1948: mandoc_vmsg(MANDOCERR_ARG_SKIP, r->parse, ln, pos,
! 1949: ".EQ %s", *bufp + pos);
1.151 kristaps 1950:
1.125 kristaps 1951: return(ROFF_IGN);
1952: }
1953:
1954: static enum rofferr
1955: roff_EN(ROFF_ARGS)
1956: {
1957:
1.221 schwarze 1958: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse, ln, ppos, "EN");
1.125 kristaps 1959: return(ROFF_IGN);
1960: }
1961:
1962: static enum rofferr
1.109 kristaps 1963: roff_TS(ROFF_ARGS)
1964: {
1.176 schwarze 1965: struct tbl_node *tbl;
1.89 kristaps 1966:
1.115 kristaps 1967: if (r->tbl) {
1.221 schwarze 1968: mandoc_msg(MANDOCERR_BLK_BROKEN, r->parse,
1969: ln, ppos, "TS breaks TS");
1.151 kristaps 1970: tbl_end(&r->tbl);
1.115 kristaps 1971: }
1.83 schwarze 1972:
1.176 schwarze 1973: tbl = tbl_alloc(ppos, ln, r->parse);
1.113 kristaps 1974:
1975: if (r->last_tbl)
1.176 schwarze 1976: r->last_tbl->next = tbl;
1.113 kristaps 1977: else
1.176 schwarze 1978: r->first_tbl = r->last_tbl = tbl;
1.113 kristaps 1979:
1.176 schwarze 1980: r->tbl = r->last_tbl = tbl;
1.83 schwarze 1981: return(ROFF_IGN);
1.92 schwarze 1982: }
1983:
1.105 kristaps 1984: static enum rofferr
1.174 kristaps 1985: roff_cc(ROFF_ARGS)
1986: {
1987: const char *p;
1988:
1989: p = *bufp + pos;
1990:
1991: if ('\0' == *p || '.' == (r->control = *p++))
1992: r->control = 0;
1993:
1994: if ('\0' != *p)
1995: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
1996:
1997: return(ROFF_IGN);
1998: }
1999:
2000: static enum rofferr
1.164 kristaps 2001: roff_tr(ROFF_ARGS)
2002: {
2003: const char *p, *first, *second;
2004: size_t fsz, ssz;
2005: enum mandoc_esc esc;
2006:
2007: p = *bufp + pos;
2008:
2009: if ('\0' == *p) {
2010: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
2011: return(ROFF_IGN);
2012: }
2013:
2014: while ('\0' != *p) {
2015: fsz = ssz = 1;
2016:
2017: first = p++;
2018: if ('\\' == *first) {
2019: esc = mandoc_escape(&p, NULL, NULL);
2020: if (ESCAPE_ERROR == esc) {
1.219 schwarze 2021: mandoc_msg(MANDOCERR_ESC_BAD, r->parse,
2022: ln, (int)(p - *bufp), first);
1.164 kristaps 2023: return(ROFF_IGN);
2024: }
2025: fsz = (size_t)(p - first);
2026: }
2027:
2028: second = p++;
2029: if ('\\' == *second) {
2030: esc = mandoc_escape(&p, NULL, NULL);
2031: if (ESCAPE_ERROR == esc) {
1.219 schwarze 2032: mandoc_msg(MANDOCERR_ESC_BAD, r->parse,
2033: ln, (int)(p - *bufp), second);
1.164 kristaps 2034: return(ROFF_IGN);
2035: }
2036: ssz = (size_t)(p - second);
1.165 kristaps 2037: } else if ('\0' == *second) {
1.207 schwarze 2038: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse,
2039: ln, (int)(p - *bufp), NULL);
1.164 kristaps 2040: second = " ";
1.165 kristaps 2041: p--;
1.164 kristaps 2042: }
2043:
1.167 kristaps 2044: if (fsz > 1) {
1.207 schwarze 2045: roff_setstrn(&r->xmbtab, first, fsz,
2046: second, ssz, 0);
1.167 kristaps 2047: continue;
2048: }
2049:
2050: if (NULL == r->xtab)
1.207 schwarze 2051: r->xtab = mandoc_calloc(128,
2052: sizeof(struct roffstr));
1.167 kristaps 2053:
2054: free(r->xtab[(int)*first].p);
2055: r->xtab[(int)*first].p = mandoc_strndup(second, ssz);
2056: r->xtab[(int)*first].sz = ssz;
1.164 kristaps 2057: }
2058:
2059: return(ROFF_IGN);
2060: }
2061:
2062: static enum rofferr
1.105 kristaps 2063: roff_so(ROFF_ARGS)
2064: {
2065: char *name;
2066:
1.210 schwarze 2067: name = *bufp + pos;
1.224 schwarze 2068: mandoc_vmsg(MANDOCERR_SO, r->parse, ln, ppos, "so %s", name);
1.105 kristaps 2069:
2070: /*
2071: * Handle `so'. Be EXTREMELY careful, as we shouldn't be
2072: * opening anything that's not in our cwd or anything beneath
2073: * it. Thus, explicitly disallow traversing up the file-system
2074: * or using absolute paths.
2075: */
2076:
2077: if ('/' == *name || strstr(name, "../") || strstr(name, "/..")) {
1.210 schwarze 2078: mandoc_vmsg(MANDOCERR_SO_PATH, r->parse, ln, ppos,
2079: ".so %s", name);
1.105 kristaps 2080: return(ROFF_ERR);
2081: }
2082:
2083: *offs = pos;
2084: return(ROFF_SO);
2085: }
1.92 schwarze 2086:
1.106 kristaps 2087: static enum rofferr
2088: roff_userdef(ROFF_ARGS)
1.99 kristaps 2089: {
1.106 kristaps 2090: const char *arg[9];
2091: char *cp, *n1, *n2;
1.119 schwarze 2092: int i;
1.106 kristaps 2093:
2094: /*
2095: * Collect pointers to macro argument strings
1.188 schwarze 2096: * and NUL-terminate them.
1.106 kristaps 2097: */
2098: cp = *bufp + pos;
1.119 schwarze 2099: for (i = 0; i < 9; i++)
1.120 schwarze 2100: arg[i] = '\0' == *cp ? "" :
1.136 kristaps 2101: mandoc_getarg(r->parse, &cp, ln, &pos);
1.99 kristaps 2102:
1.106 kristaps 2103: /*
2104: * Expand macro arguments.
1.99 kristaps 2105: */
1.106 kristaps 2106: *szp = 0;
2107: n1 = cp = mandoc_strdup(r->current_string);
2108: while (NULL != (cp = strstr(cp, "\\$"))) {
2109: i = cp[2] - '1';
2110: if (0 > i || 8 < i) {
2111: /* Not an argument invocation. */
2112: cp += 2;
2113: continue;
2114: }
1.209 schwarze 2115: *cp = '\0';
2116: *szp = mandoc_asprintf(&n2, "%s%s%s",
2117: n1, arg[i], cp + 3) + 1;
1.106 kristaps 2118: cp = n2 + (cp - n1);
2119: free(n1);
2120: n1 = n2;
1.99 kristaps 2121: }
2122:
1.106 kristaps 2123: /*
2124: * Replace the macro invocation
2125: * by the expanded macro.
2126: */
2127: free(*bufp);
2128: *bufp = n1;
2129: if (0 == *szp)
2130: *szp = strlen(*bufp) + 1;
2131:
2132: return(*szp > 1 && '\n' == (*bufp)[(int)*szp - 2] ?
2133: ROFF_REPARSE : ROFF_APPEND);
1.99 kristaps 2134: }
1.121 schwarze 2135:
1.212 schwarze 2136: static size_t
1.121 schwarze 2137: roff_getname(struct roff *r, char **cpp, int ln, int pos)
2138: {
2139: char *name, *cp;
1.212 schwarze 2140: size_t namesz;
1.121 schwarze 2141:
2142: name = *cpp;
2143: if ('\0' == *name)
1.212 schwarze 2144: return(0);
1.121 schwarze 2145:
1.212 schwarze 2146: /* Read until end of name and terminate it with NUL. */
2147: for (cp = name; 1; cp++) {
2148: if ('\0' == *cp || ' ' == *cp) {
2149: namesz = cp - name;
2150: break;
2151: }
1.121 schwarze 2152: if ('\\' != *cp)
2153: continue;
1.215 schwarze 2154: namesz = cp - name;
2155: if ('{' == cp[1] || '}' == cp[1])
2156: break;
1.121 schwarze 2157: cp++;
2158: if ('\\' == *cp)
2159: continue;
1.224 schwarze 2160: mandoc_vmsg(MANDOCERR_NAMESC, r->parse, ln, pos,
2161: "%.*s", (int)(cp - name + 1), name);
1.212 schwarze 2162: mandoc_escape((const char **)&cp, NULL, NULL);
2163: break;
1.121 schwarze 2164: }
2165:
2166: /* Read past spaces. */
2167: while (' ' == *cp)
2168: cp++;
2169:
2170: *cpp = cp;
1.212 schwarze 2171: return(namesz);
1.121 schwarze 2172: }
2173:
1.106 kristaps 2174: /*
2175: * Store *string into the user-defined string called *name.
2176: * To clear an existing entry, call with (*r, *name, NULL, 0).
1.193 schwarze 2177: * append == 0: replace mode
2178: * append == 1: single-line append mode
2179: * append == 2: multiline append mode, append '\n' after each call
1.106 kristaps 2180: */
1.94 kristaps 2181: static void
1.106 kristaps 2182: roff_setstr(struct roff *r, const char *name, const char *string,
1.193 schwarze 2183: int append)
1.92 schwarze 2184: {
1.164 kristaps 2185:
2186: roff_setstrn(&r->strtab, name, strlen(name), string,
1.207 schwarze 2187: string ? strlen(string) : 0, append);
1.164 kristaps 2188: }
2189:
2190: static void
1.166 kristaps 2191: roff_setstrn(struct roffkv **r, const char *name, size_t namesz,
1.193 schwarze 2192: const char *string, size_t stringsz, int append)
1.164 kristaps 2193: {
1.166 kristaps 2194: struct roffkv *n;
1.164 kristaps 2195: char *c;
2196: int i;
2197: size_t oldch, newch;
1.92 schwarze 2198:
1.106 kristaps 2199: /* Search for an existing string with the same name. */
1.164 kristaps 2200: n = *r;
2201:
1.211 schwarze 2202: while (n && (namesz != n->key.sz ||
2203: strncmp(n->key.p, name, namesz)))
1.92 schwarze 2204: n = n->next;
1.94 kristaps 2205:
2206: if (NULL == n) {
1.106 kristaps 2207: /* Create a new string table entry. */
1.166 kristaps 2208: n = mandoc_malloc(sizeof(struct roffkv));
2209: n->key.p = mandoc_strndup(name, namesz);
2210: n->key.sz = namesz;
2211: n->val.p = NULL;
2212: n->val.sz = 0;
1.164 kristaps 2213: n->next = *r;
2214: *r = n;
1.193 schwarze 2215: } else if (0 == append) {
1.166 kristaps 2216: free(n->val.p);
2217: n->val.p = NULL;
2218: n->val.sz = 0;
1.106 kristaps 2219: }
2220:
2221: if (NULL == string)
2222: return;
2223:
2224: /*
2225: * One additional byte for the '\n' in multiline mode,
2226: * and one for the terminating '\0'.
2227: */
1.193 schwarze 2228: newch = stringsz + (1 < append ? 2u : 1u);
1.164 kristaps 2229:
1.166 kristaps 2230: if (NULL == n->val.p) {
2231: n->val.p = mandoc_malloc(newch);
2232: *n->val.p = '\0';
1.106 kristaps 2233: oldch = 0;
2234: } else {
1.166 kristaps 2235: oldch = n->val.sz;
2236: n->val.p = mandoc_realloc(n->val.p, oldch + newch);
1.106 kristaps 2237: }
2238:
2239: /* Skip existing content in the destination buffer. */
1.166 kristaps 2240: c = n->val.p + (int)oldch;
1.106 kristaps 2241:
2242: /* Append new content to the destination buffer. */
1.164 kristaps 2243: i = 0;
2244: while (i < (int)stringsz) {
1.106 kristaps 2245: /*
2246: * Rudimentary roff copy mode:
2247: * Handle escaped backslashes.
2248: */
1.164 kristaps 2249: if ('\\' == string[i] && '\\' == string[i + 1])
2250: i++;
2251: *c++ = string[i++];
1.106 kristaps 2252: }
1.94 kristaps 2253:
1.106 kristaps 2254: /* Append terminating bytes. */
1.193 schwarze 2255: if (1 < append)
1.106 kristaps 2256: *c++ = '\n';
1.163 kristaps 2257:
1.106 kristaps 2258: *c = '\0';
1.166 kristaps 2259: n->val.sz = (int)(c - n->val.p);
1.92 schwarze 2260: }
2261:
1.94 kristaps 2262: static const char *
2263: roff_getstrn(const struct roff *r, const char *name, size_t len)
1.92 schwarze 2264: {
1.166 kristaps 2265: const struct roffkv *n;
1.191 schwarze 2266: int i;
1.92 schwarze 2267:
1.164 kristaps 2268: for (n = r->strtab; n; n = n->next)
1.207 schwarze 2269: if (0 == strncmp(name, n->key.p, len) &&
2270: '\0' == n->key.p[(int)len])
1.166 kristaps 2271: return(n->val.p);
1.191 schwarze 2272:
2273: for (i = 0; i < PREDEFS_MAX; i++)
2274: if (0 == strncmp(name, predefs[i].name, len) &&
2275: '\0' == predefs[i].name[(int)len])
2276: return(predefs[i].str);
1.94 kristaps 2277:
1.157 kristaps 2278: return(NULL);
1.92 schwarze 2279: }
2280:
1.94 kristaps 2281: static void
1.167 kristaps 2282: roff_freestr(struct roffkv *r)
1.92 schwarze 2283: {
1.166 kristaps 2284: struct roffkv *n, *nn;
1.92 schwarze 2285:
1.167 kristaps 2286: for (n = r; n; n = nn) {
1.166 kristaps 2287: free(n->key.p);
2288: free(n->val.p);
1.92 schwarze 2289: nn = n->next;
2290: free(n);
2291: }
1.114 kristaps 2292: }
2293:
2294: const struct tbl_span *
2295: roff_span(const struct roff *r)
2296: {
1.207 schwarze 2297:
1.114 kristaps 2298: return(r->tbl ? tbl_span(r->tbl) : NULL);
1.125 kristaps 2299: }
2300:
2301: const struct eqn *
2302: roff_eqn(const struct roff *r)
2303: {
1.207 schwarze 2304:
1.125 kristaps 2305: return(r->last_eqn ? &r->last_eqn->eqn : NULL);
1.164 kristaps 2306: }
2307:
2308: /*
2309: * Duplicate an input string, making the appropriate character
2310: * conversations (as stipulated by `tr') along the way.
2311: * Returns a heap-allocated string with all the replacements made.
2312: */
2313: char *
2314: roff_strdup(const struct roff *r, const char *p)
2315: {
1.166 kristaps 2316: const struct roffkv *cp;
1.164 kristaps 2317: char *res;
2318: const char *pp;
2319: size_t ssz, sz;
2320: enum mandoc_esc esc;
2321:
1.167 kristaps 2322: if (NULL == r->xmbtab && NULL == r->xtab)
1.164 kristaps 2323: return(mandoc_strdup(p));
2324: else if ('\0' == *p)
2325: return(mandoc_strdup(""));
2326:
2327: /*
2328: * Step through each character looking for term matches
2329: * (remember that a `tr' can be invoked with an escape, which is
2330: * a glyph but the escape is multi-character).
2331: * We only do this if the character hash has been initialised
2332: * and the string is >0 length.
2333: */
2334:
2335: res = NULL;
2336: ssz = 0;
2337:
2338: while ('\0' != *p) {
1.167 kristaps 2339: if ('\\' != *p && r->xtab && r->xtab[(int)*p].p) {
2340: sz = r->xtab[(int)*p].sz;
2341: res = mandoc_realloc(res, ssz + sz + 1);
2342: memcpy(res + ssz, r->xtab[(int)*p].p, sz);
2343: ssz += sz;
2344: p++;
2345: continue;
2346: } else if ('\\' != *p) {
2347: res = mandoc_realloc(res, ssz + 2);
2348: res[ssz++] = *p++;
2349: continue;
2350: }
2351:
1.164 kristaps 2352: /* Search for term matches. */
1.167 kristaps 2353: for (cp = r->xmbtab; cp; cp = cp->next)
1.166 kristaps 2354: if (0 == strncmp(p, cp->key.p, cp->key.sz))
1.164 kristaps 2355: break;
2356:
2357: if (NULL != cp) {
2358: /*
2359: * A match has been found.
2360: * Append the match to the array and move
2361: * forward by its keysize.
2362: */
1.207 schwarze 2363: res = mandoc_realloc(res,
2364: ssz + cp->val.sz + 1);
1.166 kristaps 2365: memcpy(res + ssz, cp->val.p, cp->val.sz);
2366: ssz += cp->val.sz;
2367: p += (int)cp->key.sz;
1.164 kristaps 2368: continue;
2369: }
2370:
1.167 kristaps 2371: /*
2372: * Handle escapes carefully: we need to copy
2373: * over just the escape itself, or else we might
2374: * do replacements within the escape itself.
2375: * Make sure to pass along the bogus string.
2376: */
2377: pp = p++;
2378: esc = mandoc_escape(&p, NULL, NULL);
2379: if (ESCAPE_ERROR == esc) {
2380: sz = strlen(pp);
1.164 kristaps 2381: res = mandoc_realloc(res, ssz + sz + 1);
2382: memcpy(res + ssz, pp, sz);
1.167 kristaps 2383: break;
1.164 kristaps 2384: }
1.207 schwarze 2385: /*
2386: * We bail out on bad escapes.
1.167 kristaps 2387: * No need to warn: we already did so when
2388: * roff_res() was called.
2389: */
2390: sz = (int)(p - pp);
2391: res = mandoc_realloc(res, ssz + sz + 1);
2392: memcpy(res + ssz, pp, sz);
2393: ssz += sz;
1.164 kristaps 2394: }
2395:
2396: res[(int)ssz] = '\0';
2397: return(res);
1.227 schwarze 2398: }
2399:
2400: int
2401: roff_getformat(const struct roff *r)
2402: {
2403:
2404: return(r->format);
1.174 kristaps 2405: }
2406:
2407: /*
1.207 schwarze 2408: * Find out whether a line is a macro line or not.
1.174 kristaps 2409: * If it is, adjust the current position and return one; if it isn't,
2410: * return zero and don't change the current position.
2411: * If the control character has been set with `.cc', then let that grain
2412: * precedence.
2413: * This is slighly contrary to groff, where using the non-breaking
2414: * control character when `cc' has been invoked will cause the
2415: * non-breaking macro contents to be printed verbatim.
2416: */
2417: int
2418: roff_getcontrol(const struct roff *r, const char *cp, int *ppos)
2419: {
2420: int pos;
2421:
2422: pos = *ppos;
2423:
2424: if (0 != r->control && cp[pos] == r->control)
2425: pos++;
2426: else if (0 != r->control)
2427: return(0);
2428: else if ('\\' == cp[pos] && '.' == cp[pos + 1])
2429: pos += 2;
2430: else if ('.' == cp[pos] || '\'' == cp[pos])
2431: pos++;
2432: else
2433: return(0);
2434:
2435: while (' ' == cp[pos] || '\t' == cp[pos])
2436: pos++;
2437:
2438: *ppos = pos;
2439: return(1);
1.74 kristaps 2440: }
CVSweb