Annotation of mandoc/roff.c, Revision 1.233
1.233 ! schwarze 1: /* $Id: roff.c,v 1.232 2014/10/20 02:33:06 schwarze 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);
1.198 schwarze 189: static int roff_evalcond(const char *, int *);
1.204 schwarze 190: static int roff_evalnum(const char *, int *, int *, int);
191: static int roff_evalpar(const char *, int *, int *);
1.198 schwarze 192: static int roff_evalstrcond(const char *, int *);
1.155 kristaps 193: static void roff_free1(struct roff *);
1.180 schwarze 194: static void roff_freereg(struct roffreg *);
1.167 kristaps 195: static void roff_freestr(struct roffkv *);
1.212 schwarze 196: static size_t roff_getname(struct roff *, char **, int, int);
1.184 schwarze 197: static int roff_getnum(const char *, int *, int *);
198: static int roff_getop(const char *, int *, char *);
1.181 schwarze 199: static int roff_getregn(const struct roff *,
200: const char *, size_t);
1.192 schwarze 201: static int roff_getregro(const char *name);
1.207 schwarze 202: static const char *roff_getstrn(const struct roff *,
1.94 kristaps 203: const char *, size_t);
1.178 schwarze 204: static enum rofferr roff_it(ROFF_ARGS);
1.103 kristaps 205: static enum rofferr roff_line_ignore(ROFF_ARGS);
1.89 kristaps 206: static enum rofferr roff_nr(ROFF_ARGS);
1.169 schwarze 207: static void roff_openeqn(struct roff *, const char *,
1.156 kristaps 208: int, int, const char *);
1.214 schwarze 209: static enum rofft roff_parse(struct roff *, char *, int *,
210: int, int);
1.178 schwarze 211: static enum rofferr roff_parsetext(char **, size_t *, int, int *);
1.207 schwarze 212: static enum rofferr roff_res(struct roff *,
1.142 kristaps 213: char **, size_t *, int, int);
1.122 schwarze 214: static enum rofferr roff_rm(ROFF_ARGS);
1.203 schwarze 215: static enum rofferr roff_rr(ROFF_ARGS);
1.94 kristaps 216: static void roff_setstr(struct roff *,
1.106 kristaps 217: const char *, const char *, int);
1.207 schwarze 218: static void roff_setstrn(struct roffkv **, const char *,
1.164 kristaps 219: size_t, const char *, size_t, int);
1.105 kristaps 220: static enum rofferr roff_so(ROFF_ARGS);
1.164 kristaps 221: static enum rofferr roff_tr(ROFF_ARGS);
1.175 schwarze 222: static enum rofferr roff_Dd(ROFF_ARGS);
223: static enum rofferr roff_TH(ROFF_ARGS);
1.109 kristaps 224: static enum rofferr roff_TE(ROFF_ARGS);
225: static enum rofferr roff_TS(ROFF_ARGS);
1.125 kristaps 226: static enum rofferr roff_EQ(ROFF_ARGS);
227: static enum rofferr roff_EN(ROFF_ARGS);
1.112 kristaps 228: static enum rofferr roff_T_(ROFF_ARGS);
1.106 kristaps 229: static enum rofferr roff_userdef(ROFF_ARGS);
1.67 kristaps 230:
1.155 kristaps 231: /* See roffhash_find() */
1.85 kristaps 232:
233: #define ASCII_HI 126
234: #define ASCII_LO 33
235: #define HASHWIDTH (ASCII_HI - ASCII_LO + 1)
236:
237: static struct roffmac *hash[HASHWIDTH];
238:
239: static struct roffmac roffs[ROFF_MAX] = {
1.103 kristaps 240: { "ad", roff_line_ignore, NULL, NULL, 0, NULL },
1.85 kristaps 241: { "am", roff_block, roff_block_text, roff_block_sub, 0, NULL },
242: { "ami", roff_block, roff_block_text, roff_block_sub, 0, NULL },
243: { "am1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.193 schwarze 244: { "as", roff_ds, NULL, NULL, 0, NULL },
1.174 kristaps 245: { "cc", roff_cc, NULL, NULL, 0, NULL },
1.194 schwarze 246: { "ce", roff_line_ignore, NULL, NULL, 0, NULL },
1.85 kristaps 247: { "de", roff_block, roff_block_text, roff_block_sub, 0, NULL },
248: { "dei", roff_block, roff_block_text, roff_block_sub, 0, NULL },
249: { "de1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.92 schwarze 250: { "ds", roff_ds, NULL, NULL, 0, NULL },
1.85 kristaps 251: { "el", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
1.185 schwarze 252: { "fam", roff_line_ignore, NULL, NULL, 0, NULL },
1.186 schwarze 253: { "hw", roff_line_ignore, NULL, NULL, 0, NULL },
1.103 kristaps 254: { "hy", roff_line_ignore, NULL, NULL, 0, NULL },
1.85 kristaps 255: { "ie", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
256: { "if", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
257: { "ig", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.178 schwarze 258: { "it", roff_it, NULL, NULL, 0, NULL },
1.103 kristaps 259: { "ne", roff_line_ignore, NULL, NULL, 0, NULL },
260: { "nh", roff_line_ignore, NULL, NULL, 0, NULL },
1.104 kristaps 261: { "nr", roff_nr, NULL, NULL, 0, NULL },
1.124 schwarze 262: { "ns", roff_line_ignore, NULL, NULL, 0, NULL },
1.229 schwarze 263: { "pl", roff_line_ignore, NULL, NULL, 0, NULL },
1.124 schwarze 264: { "ps", roff_line_ignore, NULL, NULL, 0, NULL },
1.122 schwarze 265: { "rm", roff_rm, NULL, NULL, 0, NULL },
1.203 schwarze 266: { "rr", roff_rr, NULL, NULL, 0, NULL },
1.105 kristaps 267: { "so", roff_so, NULL, NULL, 0, NULL },
1.124 schwarze 268: { "ta", roff_line_ignore, NULL, NULL, 0, NULL },
1.164 kristaps 269: { "tr", roff_tr, NULL, NULL, 0, NULL },
1.175 schwarze 270: { "Dd", roff_Dd, NULL, NULL, 0, NULL },
271: { "TH", roff_TH, NULL, NULL, 0, NULL },
1.109 kristaps 272: { "TS", roff_TS, NULL, NULL, 0, NULL },
273: { "TE", roff_TE, NULL, NULL, 0, NULL },
1.112 kristaps 274: { "T&", roff_T_, NULL, NULL, 0, NULL },
1.125 kristaps 275: { "EQ", roff_EQ, NULL, NULL, 0, NULL },
276: { "EN", roff_EN, NULL, NULL, 0, NULL },
1.85 kristaps 277: { ".", roff_cblock, NULL, NULL, 0, NULL },
1.106 kristaps 278: { NULL, roff_userdef, NULL, NULL, 0, NULL },
1.67 kristaps 279: };
280:
1.200 schwarze 281: /* not currently implemented: Ds em Eq LP Me PP pp Or Rd Sf SH */
1.175 schwarze 282: const char *const __mdoc_reserved[] = {
283: "Ac", "Ad", "An", "Ao", "Ap", "Aq", "Ar", "At",
284: "Bc", "Bd", "Bf", "Bk", "Bl", "Bo", "Bq",
285: "Brc", "Bro", "Brq", "Bsx", "Bt", "Bx",
286: "Cd", "Cm", "Db", "Dc", "Dd", "Dl", "Do", "Dq",
1.200 schwarze 287: "Dt", "Dv", "Dx", "D1",
288: "Ec", "Ed", "Ef", "Ek", "El", "Em",
289: "En", "Eo", "Er", "Es", "Ev", "Ex",
1.175 schwarze 290: "Fa", "Fc", "Fd", "Fl", "Fn", "Fo", "Fr", "Ft", "Fx",
1.200 schwarze 291: "Hf", "Ic", "In", "It", "Lb", "Li", "Lk", "Lp",
292: "Ms", "Mt", "Nd", "Nm", "No", "Ns", "Nx",
1.175 schwarze 293: "Oc", "Oo", "Op", "Os", "Ot", "Ox",
1.200 schwarze 294: "Pa", "Pc", "Pf", "Po", "Pp", "Pq",
295: "Qc", "Ql", "Qo", "Qq", "Re", "Rs", "Rv",
296: "Sc", "Sh", "Sm", "So", "Sq",
1.175 schwarze 297: "Ss", "St", "Sx", "Sy",
298: "Ta", "Tn", "Ud", "Ux", "Va", "Vt", "Xc", "Xo", "Xr",
1.200 schwarze 299: "%A", "%B", "%C", "%D", "%I", "%J", "%N", "%O",
1.175 schwarze 300: "%P", "%Q", "%R", "%T", "%U", "%V",
301: NULL
302: };
303:
1.200 schwarze 304: /* not currently implemented: BT DE DS ME MT PT SY TQ YS */
1.175 schwarze 305: const char *const __man_reserved[] = {
1.200 schwarze 306: "AT", "B", "BI", "BR", "DT",
307: "EE", "EN", "EQ", "EX", "HP", "I", "IB", "IP", "IR",
308: "LP", "OP", "P", "PD", "PP",
309: "R", "RB", "RE", "RI", "RS", "SB", "SH", "SM", "SS",
310: "TE", "TH", "TP", "TS", "T&", "UC", "UE", "UR",
1.175 schwarze 311: NULL
312: };
313:
1.141 kristaps 314: /* Array of injected predefined strings. */
315: #define PREDEFS_MAX 38
316: static const struct predef predefs[PREDEFS_MAX] = {
317: #include "predefs.in"
318: };
319:
1.155 kristaps 320: /* See roffhash_find() */
1.85 kristaps 321: #define ROFF_HASH(p) (p[0] - ASCII_LO)
322:
1.178 schwarze 323: static int roffit_lines; /* number of lines to delay */
324: static char *roffit_macro; /* nil-terminated macro line */
325:
1.207 schwarze 326:
1.85 kristaps 327: static void
1.155 kristaps 328: roffhash_init(void)
1.85 kristaps 329: {
330: struct roffmac *n;
331: int buc, i;
332:
1.106 kristaps 333: for (i = 0; i < (int)ROFF_USERDEF; i++) {
1.85 kristaps 334: assert(roffs[i].name[0] >= ASCII_LO);
335: assert(roffs[i].name[0] <= ASCII_HI);
336:
337: buc = ROFF_HASH(roffs[i].name);
338:
339: if (NULL != (n = hash[buc])) {
340: for ( ; n->next; n = n->next)
341: /* Do nothing. */ ;
342: n->next = &roffs[i];
343: } else
344: hash[buc] = &roffs[i];
345: }
346: }
347:
1.67 kristaps 348: /*
349: * Look up a roff token by its name. Returns ROFF_MAX if no macro by
350: * the nil-terminated string name could be found.
351: */
352: static enum rofft
1.155 kristaps 353: roffhash_find(const char *p, size_t s)
1.67 kristaps 354: {
1.85 kristaps 355: int buc;
356: struct roffmac *n;
1.67 kristaps 357:
1.85 kristaps 358: /*
359: * libroff has an extremely simple hashtable, for the time
360: * being, which simply keys on the first character, which must
361: * be printable, then walks a chain. It works well enough until
362: * optimised.
363: */
364:
365: if (p[0] < ASCII_LO || p[0] > ASCII_HI)
366: return(ROFF_MAX);
367:
368: buc = ROFF_HASH(p);
369:
370: if (NULL == (n = hash[buc]))
371: return(ROFF_MAX);
372: for ( ; n; n = n->next)
1.106 kristaps 373: if (0 == strncmp(n->name, p, s) && '\0' == n->name[(int)s])
1.85 kristaps 374: return((enum rofft)(n - roffs));
1.67 kristaps 375:
376: return(ROFF_MAX);
377: }
378:
379: /*
380: * Pop the current node off of the stack of roff instructions currently
381: * pending.
382: */
383: static void
384: roffnode_pop(struct roff *r)
385: {
386: struct roffnode *p;
387:
1.75 kristaps 388: assert(r->last);
1.207 schwarze 389: p = r->last;
1.82 kristaps 390:
1.75 kristaps 391: r->last = r->last->parent;
1.106 kristaps 392: free(p->name);
393: free(p->end);
1.67 kristaps 394: free(p);
395: }
396:
397: /*
398: * Push a roff node onto the instruction stack. This must later be
399: * removed with roffnode_pop().
400: */
1.98 schwarze 401: static void
1.106 kristaps 402: roffnode_push(struct roff *r, enum rofft tok, const char *name,
403: int line, int col)
1.67 kristaps 404: {
405: struct roffnode *p;
406:
1.98 schwarze 407: p = mandoc_calloc(1, sizeof(struct roffnode));
1.67 kristaps 408: p->tok = tok;
1.106 kristaps 409: if (name)
410: p->name = mandoc_strdup(name);
1.67 kristaps 411: p->parent = r->last;
412: p->line = line;
413: p->col = col;
1.198 schwarze 414: p->rule = p->parent ? p->parent->rule : 0;
1.67 kristaps 415:
416: r->last = p;
417: }
418:
419: static void
420: roff_free1(struct roff *r)
421: {
1.176 schwarze 422: struct tbl_node *tbl;
1.125 kristaps 423: struct eqn_node *e;
1.167 kristaps 424: int i;
1.67 kristaps 425:
1.176 schwarze 426: while (NULL != (tbl = r->first_tbl)) {
427: r->first_tbl = tbl->next;
428: tbl_free(tbl);
1.109 kristaps 429: }
1.113 kristaps 430: r->first_tbl = r->last_tbl = r->tbl = NULL;
431:
1.125 kristaps 432: while (NULL != (e = r->first_eqn)) {
433: r->first_eqn = e->next;
434: eqn_free(e);
435: }
436: r->first_eqn = r->last_eqn = r->eqn = NULL;
437:
1.67 kristaps 438: while (r->last)
439: roffnode_pop(r);
1.109 kristaps 440:
1.223 schwarze 441: free (r->rstack);
442: r->rstack = NULL;
443: r->rstacksz = 0;
444: r->rstackpos = -1;
445:
446: roff_freereg(r->regtab);
447: r->regtab = NULL;
448:
1.167 kristaps 449: roff_freestr(r->strtab);
450: roff_freestr(r->xmbtab);
451: r->strtab = r->xmbtab = NULL;
452:
453: if (r->xtab)
454: for (i = 0; i < 128; i++)
455: free(r->xtab[i].p);
456: free(r->xtab);
457: r->xtab = NULL;
1.67 kristaps 458: }
459:
460: void
461: roff_reset(struct roff *r)
462: {
463:
464: roff_free1(r);
1.227 schwarze 465: r->format = r->options & (MPARSE_MDOC | MPARSE_MAN);
1.174 kristaps 466: r->control = 0;
1.67 kristaps 467: }
468:
469: void
470: roff_free(struct roff *r)
471: {
472:
473: roff_free1(r);
474: free(r);
475: }
476:
477: struct roff *
1.199 schwarze 478: roff_alloc(struct mparse *parse, int options)
1.67 kristaps 479: {
480: struct roff *r;
481:
1.98 schwarze 482: r = mandoc_calloc(1, sizeof(struct roff));
1.128 kristaps 483: r->parse = parse;
1.199 schwarze 484: r->options = options;
1.227 schwarze 485: r->format = options & (MPARSE_MDOC | MPARSE_MAN);
1.82 kristaps 486: r->rstackpos = -1;
1.207 schwarze 487:
1.155 kristaps 488: roffhash_init();
1.141 kristaps 489:
1.67 kristaps 490: return(r);
491: }
492:
1.94 kristaps 493: /*
1.206 schwarze 494: * In the current line, expand escape sequences that tend to get
495: * used in numerical expressions and conditional requests.
496: * Also check the syntax of the remaining escape sequences.
1.154 kristaps 497: */
1.172 schwarze 498: static enum rofferr
1.142 kristaps 499: roff_res(struct roff *r, char **bufp, size_t *szp, int ln, int pos)
1.94 kristaps 500: {
1.208 schwarze 501: char ubuf[24]; /* buffer to print the number */
1.205 schwarze 502: const char *start; /* start of the string to process */
1.209 schwarze 503: char *stesc; /* start of an escape sequence ('\\') */
1.108 schwarze 504: const char *stnam; /* start of the name, after "[(*" */
505: const char *cp; /* end of the name, e.g. before ']' */
506: const char *res; /* the string to be substituted */
1.181 schwarze 507: char *nbuf; /* new buffer to copy bufp to */
508: size_t maxl; /* expected length of the escape name */
509: size_t naml; /* actual length of the escape name */
510: int expand_count; /* to avoid infinite loops */
1.206 schwarze 511: int npos; /* position in numeric expression */
1.218 schwarze 512: int arg_complete; /* argument not interrupted by eol */
1.206 schwarze 513: char term; /* character terminating the escape */
1.94 kristaps 514:
1.170 schwarze 515: expand_count = 0;
1.205 schwarze 516: start = *bufp + pos;
517: stesc = strchr(start, '\0') - 1;
518: while (stesc-- > start) {
1.170 schwarze 519:
1.205 schwarze 520: /* Search backwards for the next backslash. */
521:
522: if ('\\' != *stesc)
523: continue;
524:
525: /* If it is escaped, skip it. */
526:
527: for (cp = stesc - 1; cp >= start; cp--)
528: if ('\\' != *cp)
529: break;
530:
531: if (0 == (stesc - cp) % 2) {
1.209 schwarze 532: stesc = (char *)cp;
1.205 schwarze 533: continue;
534: }
1.108 schwarze 535:
1.206 schwarze 536: /* Decide whether to expand or to check only. */
1.108 schwarze 537:
1.206 schwarze 538: term = '\0';
1.205 schwarze 539: cp = stesc + 1;
1.181 schwarze 540: switch (*cp) {
1.207 schwarze 541: case '*':
1.181 schwarze 542: res = NULL;
543: break;
1.207 schwarze 544: case 'B':
1.206 schwarze 545: /* FALLTHROUGH */
1.207 schwarze 546: case 'w':
1.206 schwarze 547: term = cp[1];
548: /* FALLTHROUGH */
1.207 schwarze 549: case 'n':
1.181 schwarze 550: res = ubuf;
551: break;
552: default:
1.205 schwarze 553: if (ESCAPE_ERROR == mandoc_escape(&cp, NULL, NULL))
1.219 schwarze 554: mandoc_vmsg(MANDOCERR_ESC_BAD,
555: r->parse, ln, (int)(stesc - *bufp),
556: "%.*s", (int)(cp - stesc), stesc);
1.205 schwarze 557: continue;
1.152 kristaps 558: }
559:
1.205 schwarze 560: if (EXPAND_LIMIT < ++expand_count) {
561: mandoc_msg(MANDOCERR_ROFFLOOP, r->parse,
562: ln, (int)(stesc - *bufp), NULL);
563: return(ROFF_IGN);
564: }
1.108 schwarze 565:
566: /*
567: * The third character decides the length
1.181 schwarze 568: * of the name of the string or register.
1.108 schwarze 569: * Save a pointer to the name.
570: */
571:
1.206 schwarze 572: if ('\0' == term) {
573: switch (*++cp) {
1.207 schwarze 574: case '\0':
1.206 schwarze 575: maxl = 0;
576: break;
1.207 schwarze 577: case '(':
1.206 schwarze 578: cp++;
579: maxl = 2;
580: break;
1.207 schwarze 581: case '[':
1.206 schwarze 582: cp++;
583: term = ']';
584: maxl = 0;
585: break;
586: default:
587: maxl = 1;
588: break;
589: }
590: } else {
591: cp += 2;
1.94 kristaps 592: maxl = 0;
593: }
1.108 schwarze 594: stnam = cp;
1.94 kristaps 595:
1.108 schwarze 596: /* Advance to the end of the name. */
1.94 kristaps 597:
1.218 schwarze 598: arg_complete = 1;
1.181 schwarze 599: for (naml = 0; 0 == maxl || naml < maxl; naml++, cp++) {
1.153 kristaps 600: if ('\0' == *cp) {
1.219 schwarze 601: mandoc_msg(MANDOCERR_ESC_BAD, r->parse,
602: ln, (int)(stesc - *bufp), stesc);
1.218 schwarze 603: arg_complete = 0;
1.206 schwarze 604: break;
1.153 kristaps 605: }
1.206 schwarze 606: if (0 == maxl && *cp == term) {
607: cp++;
1.94 kristaps 608: break;
1.206 schwarze 609: }
1.94 kristaps 610: }
611:
1.108 schwarze 612: /*
613: * Retrieve the replacement string; if it is
614: * undefined, resume searching for escapes.
615: */
616:
1.206 schwarze 617: switch (stesc[1]) {
1.207 schwarze 618: case '*':
1.218 schwarze 619: if (arg_complete)
620: res = roff_getstrn(r, stnam, naml);
1.206 schwarze 621: break;
1.207 schwarze 622: case 'B':
1.206 schwarze 623: npos = 0;
1.218 schwarze 624: ubuf[0] = arg_complete &&
625: roff_evalnum(stnam, &npos, NULL, 0) &&
626: stnam + npos + 1 == cp ? '1' : '0';
1.206 schwarze 627: ubuf[1] = '\0';
628: break;
1.207 schwarze 629: case 'n':
1.218 schwarze 630: if (arg_complete)
631: (void)snprintf(ubuf, sizeof(ubuf), "%d",
632: roff_getregn(r, stnam, naml));
633: else
634: ubuf[0] = '\0';
1.206 schwarze 635: break;
1.207 schwarze 636: case 'w':
1.218 schwarze 637: /* use even incomplete args */
1.208 schwarze 638: (void)snprintf(ubuf, sizeof(ubuf), "%d",
1.206 schwarze 639: 24 * (int)naml);
640: break;
641: }
1.94 kristaps 642:
643: if (NULL == res) {
1.219 schwarze 644: mandoc_vmsg(MANDOCERR_STR_UNDEF,
645: r->parse, ln, (int)(stesc - *bufp),
646: "%.*s", (int)naml, stnam);
1.142 kristaps 647: res = "";
1.94 kristaps 648: }
649:
1.108 schwarze 650: /* Replace the escape sequence by the string. */
651:
1.209 schwarze 652: *stesc = '\0';
653: *szp = mandoc_asprintf(&nbuf, "%s%s%s",
654: *bufp, res, cp) + 1;
1.94 kristaps 655:
1.205 schwarze 656: /* Prepare for the next replacement. */
1.94 kristaps 657:
1.205 schwarze 658: start = nbuf + pos;
1.209 schwarze 659: stesc = nbuf + (stesc - *bufp) + strlen(res);
1.94 kristaps 660: free(*bufp);
1.181 schwarze 661: *bufp = nbuf;
1.154 kristaps 662: }
1.172 schwarze 663: return(ROFF_CONT);
1.154 kristaps 664: }
665:
666: /*
1.178 schwarze 667: * Process text streams:
668: * Convert all breakable hyphens into ASCII_HYPH.
669: * Decrement and spring input line trap.
1.154 kristaps 670: */
671: static enum rofferr
1.178 schwarze 672: roff_parsetext(char **bufp, size_t *szp, int pos, int *offs)
1.154 kristaps 673: {
674: size_t sz;
675: const char *start;
1.178 schwarze 676: char *p;
677: int isz;
1.154 kristaps 678: enum mandoc_esc esc;
679:
1.178 schwarze 680: start = p = *bufp + pos;
1.154 kristaps 681:
682: while ('\0' != *p) {
683: sz = strcspn(p, "-\\");
684: p += sz;
685:
1.159 kristaps 686: if ('\0' == *p)
687: break;
688:
1.154 kristaps 689: if ('\\' == *p) {
690: /* Skip over escapes. */
691: p++;
1.189 schwarze 692: esc = mandoc_escape((const char **)&p, NULL, NULL);
1.154 kristaps 693: if (ESCAPE_ERROR == esc)
694: break;
1.155 kristaps 695: continue;
1.159 kristaps 696: } else if (p == start) {
1.158 kristaps 697: p++;
1.155 kristaps 698: continue;
1.158 kristaps 699: }
1.155 kristaps 700:
1.171 schwarze 701: if (isalpha((unsigned char)p[-1]) &&
702: isalpha((unsigned char)p[1]))
1.155 kristaps 703: *p = ASCII_HYPH;
704: p++;
1.94 kristaps 705: }
706:
1.178 schwarze 707: /* Spring the input line trap. */
708: if (1 == roffit_lines) {
1.202 schwarze 709: isz = mandoc_asprintf(&p, "%s\n.%s", *bufp, roffit_macro);
1.178 schwarze 710: free(*bufp);
711: *bufp = p;
712: *szp = isz + 1;
713: *offs = 0;
714: free(roffit_macro);
715: roffit_lines = 0;
716: return(ROFF_REPARSE);
717: } else if (1 < roffit_lines)
718: --roffit_lines;
1.154 kristaps 719: return(ROFF_CONT);
1.94 kristaps 720: }
721:
1.67 kristaps 722: enum rofferr
1.207 schwarze 723: roff_parseln(struct roff *r, int ln, char **bufp,
1.90 kristaps 724: size_t *szp, int pos, int *offs)
1.67 kristaps 725: {
726: enum rofft t;
1.109 kristaps 727: enum rofferr e;
1.130 kristaps 728: int ppos, ctl;
1.79 kristaps 729:
1.230 schwarze 730: /* Handle in-line equation delimiters. */
731:
732: if (r->last_eqn != NULL && r->last_eqn->delim &&
733: (r->eqn == NULL || r->eqn_inline)) {
734: e = roff_eqndelim(r, bufp, szp, pos);
735: if (e == ROFF_REPARSE)
736: return(e);
737: assert(e == ROFF_CONT);
738: }
739:
740: /* Expand some escape sequences. */
1.94 kristaps 741:
1.172 schwarze 742: e = roff_res(r, bufp, szp, ln, pos);
743: if (ROFF_IGN == e)
744: return(e);
745: assert(ROFF_CONT == e);
1.94 kristaps 746:
1.130 kristaps 747: ppos = pos;
1.174 kristaps 748: ctl = roff_getcontrol(r, *bufp, &pos);
1.130 kristaps 749:
1.94 kristaps 750: /*
1.79 kristaps 751: * First, if a scope is open and we're not a macro, pass the
752: * text through the macro's filter. If a scope isn't open and
753: * we're not a macro, just let it through.
1.125 kristaps 754: * Finally, if there's an equation scope open, divert it into it
755: * no matter our state.
1.79 kristaps 756: */
1.74 kristaps 757:
1.130 kristaps 758: if (r->last && ! ctl) {
1.78 kristaps 759: t = r->last->tok;
760: assert(roffs[t].text);
1.207 schwarze 761: e = (*roffs[t].text)(r, t, bufp, szp, ln, pos, pos, offs);
1.109 kristaps 762: assert(ROFF_IGN == e || ROFF_CONT == e);
1.125 kristaps 763: if (ROFF_CONT != e)
764: return(e);
1.182 schwarze 765: }
766: if (r->eqn)
767: return(eqn_read(&r->eqn, ln, *bufp, ppos, offs));
768: if ( ! ctl) {
1.125 kristaps 769: if (r->tbl)
1.130 kristaps 770: return(tbl_read(r->tbl, ln, *bufp, pos));
1.178 schwarze 771: return(roff_parsetext(bufp, szp, pos, offs));
1.182 schwarze 772: }
1.228 schwarze 773:
774: /* Skip empty request lines. */
775:
776: if ((*bufp)[pos] == '"') {
777: mandoc_msg(MANDOCERR_COMMENT_BAD, r->parse,
778: ln, pos, NULL);
779: return(ROFF_IGN);
780: } else if ((*bufp)[pos] == '\0')
781: return(ROFF_IGN);
1.67 kristaps 782:
1.79 kristaps 783: /*
784: * If a scope is open, go to the child handler for that macro,
785: * as it may want to preprocess before doing anything with it.
1.125 kristaps 786: * Don't do so if an equation is open.
1.79 kristaps 787: */
1.78 kristaps 788:
1.79 kristaps 789: if (r->last) {
790: t = r->last->tok;
791: assert(roffs[t].sub);
1.207 schwarze 792: return((*roffs[t].sub)(r, t, bufp, szp,
793: ln, ppos, pos, offs));
1.79 kristaps 794: }
1.78 kristaps 795:
1.79 kristaps 796: /*
797: * Lastly, as we've no scope open, try to look up and execute
798: * the new macro. If no macro is found, simply return and let
799: * the compilers handle it.
800: */
1.67 kristaps 801:
1.214 schwarze 802: if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos, ln, ppos)))
1.79 kristaps 803: return(ROFF_CONT);
1.67 kristaps 804:
1.75 kristaps 805: assert(roffs[t].proc);
1.207 schwarze 806: return((*roffs[t].proc)(r, t, bufp, szp, ln, ppos, pos, offs));
1.74 kristaps 807: }
808:
1.117 kristaps 809: void
1.74 kristaps 810: roff_endparse(struct roff *r)
811: {
812:
1.110 kristaps 813: if (r->last)
1.221 schwarze 814: mandoc_msg(MANDOCERR_BLK_NOEND, r->parse,
815: r->last->line, r->last->col,
816: roffs[r->last->tok].name);
1.117 kristaps 817:
1.125 kristaps 818: if (r->eqn) {
1.221 schwarze 819: mandoc_msg(MANDOCERR_BLK_NOEND, r->parse,
820: r->eqn->eqn.ln, r->eqn->eqn.pos, "EQ");
1.151 kristaps 821: eqn_end(&r->eqn);
1.125 kristaps 822: }
823:
1.117 kristaps 824: if (r->tbl) {
1.221 schwarze 825: mandoc_msg(MANDOCERR_BLK_NOEND, r->parse,
826: r->tbl->line, r->tbl->pos, "TS");
1.151 kristaps 827: tbl_end(&r->tbl);
1.117 kristaps 828: }
1.67 kristaps 829: }
830:
831: /*
832: * Parse a roff node's type from the input buffer. This must be in the
833: * form of ".foo xxx" in the usual way.
834: */
835: static enum rofft
1.214 schwarze 836: roff_parse(struct roff *r, char *buf, int *pos, int ln, int ppos)
1.67 kristaps 837: {
1.214 schwarze 838: char *cp;
1.106 kristaps 839: const char *mac;
840: size_t maclen;
1.67 kristaps 841: enum rofft t;
842:
1.214 schwarze 843: cp = buf + *pos;
844:
845: if ('\0' == *cp || '"' == *cp || '\t' == *cp || ' ' == *cp)
1.67 kristaps 846: return(ROFF_MAX);
847:
1.214 schwarze 848: mac = cp;
849: maclen = roff_getname(r, &cp, ln, ppos);
1.67 kristaps 850:
1.106 kristaps 851: t = (r->current_string = roff_getstrn(r, mac, maclen))
1.155 kristaps 852: ? ROFF_USERDEF : roffhash_find(mac, maclen);
1.67 kristaps 853:
1.214 schwarze 854: if (ROFF_MAX != t)
855: *pos = cp - buf;
1.67 kristaps 856:
857: return(t);
858: }
859:
860: static enum rofferr
1.76 kristaps 861: roff_cblock(ROFF_ARGS)
1.67 kristaps 862: {
863:
1.79 kristaps 864: /*
865: * A block-close `..' should only be invoked as a child of an
866: * ignore macro, otherwise raise a warning and just ignore it.
867: */
868:
1.76 kristaps 869: if (NULL == r->last) {
1.221 schwarze 870: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
871: ln, ppos, "..");
1.76 kristaps 872: return(ROFF_IGN);
873: }
1.67 kristaps 874:
1.81 kristaps 875: switch (r->last->tok) {
1.207 schwarze 876: case ROFF_am:
1.220 schwarze 877: /* ROFF_am1 is remapped to ROFF_am in roff_block(). */
1.81 kristaps 878: /* FALLTHROUGH */
1.207 schwarze 879: case ROFF_ami:
1.81 kristaps 880: /* FALLTHROUGH */
1.207 schwarze 881: case ROFF_de:
1.108 schwarze 882: /* ROFF_de1 is remapped to ROFF_de in roff_block(). */
1.81 kristaps 883: /* FALLTHROUGH */
1.207 schwarze 884: case ROFF_dei:
1.81 kristaps 885: /* FALLTHROUGH */
1.207 schwarze 886: case ROFF_ig:
1.81 kristaps 887: break;
888: default:
1.221 schwarze 889: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
890: ln, ppos, "..");
1.67 kristaps 891: return(ROFF_IGN);
1.76 kristaps 892: }
1.67 kristaps 893:
1.76 kristaps 894: if ((*bufp)[pos])
1.217 schwarze 895: mandoc_vmsg(MANDOCERR_ARG_SKIP, r->parse, ln, pos,
896: ".. %s", *bufp + pos);
1.71 kristaps 897:
898: roffnode_pop(r);
1.76 kristaps 899: roffnode_cleanscope(r);
900: return(ROFF_IGN);
1.71 kristaps 901:
1.67 kristaps 902: }
903:
1.76 kristaps 904: static void
905: roffnode_cleanscope(struct roff *r)
1.67 kristaps 906: {
907:
1.76 kristaps 908: while (r->last) {
1.173 schwarze 909: if (--r->last->endspan != 0)
1.76 kristaps 910: break;
911: roffnode_pop(r);
912: }
1.67 kristaps 913: }
914:
1.195 schwarze 915: static void
916: roff_ccond(struct roff *r, int ln, int ppos)
1.74 kristaps 917: {
918:
1.76 kristaps 919: if (NULL == r->last) {
1.221 schwarze 920: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
921: ln, ppos, "\\}");
1.195 schwarze 922: return;
1.76 kristaps 923: }
924:
1.82 kristaps 925: switch (r->last->tok) {
1.207 schwarze 926: case ROFF_el:
1.82 kristaps 927: /* FALLTHROUGH */
1.207 schwarze 928: case ROFF_ie:
1.82 kristaps 929: /* FALLTHROUGH */
1.207 schwarze 930: case ROFF_if:
1.82 kristaps 931: break;
932: default:
1.221 schwarze 933: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
934: ln, ppos, "\\}");
1.195 schwarze 935: return;
1.75 kristaps 936: }
937:
1.76 kristaps 938: if (r->last->endspan > -1) {
1.221 schwarze 939: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
940: ln, ppos, "\\}");
1.195 schwarze 941: return;
1.76 kristaps 942: }
943:
1.75 kristaps 944: roffnode_pop(r);
1.76 kristaps 945: roffnode_cleanscope(r);
1.195 schwarze 946: return;
1.76 kristaps 947: }
948:
949: static enum rofferr
1.80 kristaps 950: roff_block(ROFF_ARGS)
1.76 kristaps 951: {
1.220 schwarze 952: const char *name;
953: char *iname, *cp;
1.213 schwarze 954: size_t namesz;
1.106 kristaps 955:
1.220 schwarze 956: /* Ignore groff compatibility mode for now. */
1.76 kristaps 957:
1.220 schwarze 958: if (ROFF_de1 == tok)
959: tok = ROFF_de;
960: else if (ROFF_am1 == tok)
961: tok = ROFF_am;
962:
963: /* Parse the macro name argument. */
964:
965: cp = *bufp + pos;
966: if (ROFF_ig == tok) {
967: iname = NULL;
968: namesz = 0;
969: } else {
970: iname = cp;
971: namesz = roff_getname(r, &cp, ln, ppos);
972: iname[namesz] = '\0';
973: }
1.107 kristaps 974:
1.220 schwarze 975: /* Resolve the macro name argument if it is indirect. */
1.107 kristaps 976:
1.220 schwarze 977: if (namesz && (ROFF_dei == tok || ROFF_ami == tok)) {
978: if (NULL == (name = roff_getstrn(r, iname, namesz))) {
979: mandoc_vmsg(MANDOCERR_STR_UNDEF,
980: r->parse, ln, (int)(iname - *bufp),
981: "%.*s", (int)namesz, iname);
982: namesz = 0;
983: } else
984: namesz = strlen(name);
985: } else
986: name = iname;
1.107 kristaps 987:
1.220 schwarze 988: if (0 == namesz && ROFF_ig != tok) {
989: mandoc_msg(MANDOCERR_REQ_EMPTY, r->parse,
990: ln, ppos, roffs[tok].name);
991: return(ROFF_IGN);
992: }
1.80 kristaps 993:
1.106 kristaps 994: roffnode_push(r, tok, name, ln, ppos);
995:
996: /*
997: * At the beginning of a `de' macro, clear the existing string
998: * with the same name, if there is one. New content will be
1.193 schwarze 999: * appended from roff_block_text() in multiline mode.
1.106 kristaps 1000: */
1.107 kristaps 1001:
1.220 schwarze 1002: if (ROFF_de == tok || ROFF_dei == tok)
1.213 schwarze 1003: roff_setstrn(&r->strtab, name, namesz, "", 0, 0);
1.76 kristaps 1004:
1.213 schwarze 1005: if ('\0' == *cp)
1.78 kristaps 1006: return(ROFF_IGN);
1007:
1.220 schwarze 1008: /* Get the custom end marker. */
1.107 kristaps 1009:
1.220 schwarze 1010: iname = cp;
1.213 schwarze 1011: namesz = roff_getname(r, &cp, ln, ppos);
1.220 schwarze 1012:
1013: /* Resolve the end marker if it is indirect. */
1014:
1015: if (namesz && (ROFF_dei == tok || ROFF_ami == tok)) {
1016: if (NULL == (name = roff_getstrn(r, iname, namesz))) {
1017: mandoc_vmsg(MANDOCERR_STR_UNDEF,
1018: r->parse, ln, (int)(iname - *bufp),
1019: "%.*s", (int)namesz, iname);
1020: namesz = 0;
1021: } else
1022: namesz = strlen(name);
1023: } else
1024: name = iname;
1025:
1.213 schwarze 1026: if (namesz)
1027: r->last->end = mandoc_strndup(name, namesz);
1.78 kristaps 1028:
1.213 schwarze 1029: if ('\0' != *cp)
1.217 schwarze 1030: mandoc_vmsg(MANDOCERR_ARG_EXCESS, r->parse,
1031: ln, pos, ".%s ... %s", roffs[tok].name, cp);
1.74 kristaps 1032:
1.78 kristaps 1033: return(ROFF_IGN);
1034: }
1035:
1036: static enum rofferr
1.80 kristaps 1037: roff_block_sub(ROFF_ARGS)
1.79 kristaps 1038: {
1039: enum rofft t;
1040: int i, j;
1041:
1042: /*
1043: * First check whether a custom macro exists at this level. If
1044: * it does, then check against it. This is some of groff's
1045: * stranger behaviours. If we encountered a custom end-scope
1046: * tag and that tag also happens to be a "real" macro, then we
1047: * need to try interpreting it again as a real macro. If it's
1048: * not, then return ignore. Else continue.
1049: */
1050:
1051: if (r->last->end) {
1.130 kristaps 1052: for (i = pos, j = 0; r->last->end[j]; j++, i++)
1.79 kristaps 1053: if ((*bufp)[i] != r->last->end[j])
1054: break;
1055:
1.207 schwarze 1056: if ('\0' == r->last->end[j] &&
1057: ('\0' == (*bufp)[i] ||
1058: ' ' == (*bufp)[i] ||
1059: '\t' == (*bufp)[i])) {
1.79 kristaps 1060: roffnode_pop(r);
1061: roffnode_cleanscope(r);
1062:
1.130 kristaps 1063: while (' ' == (*bufp)[i] || '\t' == (*bufp)[i])
1064: i++;
1065:
1066: pos = i;
1.214 schwarze 1067: if (ROFF_MAX != roff_parse(r, *bufp, &pos, ln, ppos))
1.79 kristaps 1068: return(ROFF_RERUN);
1069: return(ROFF_IGN);
1070: }
1071: }
1072:
1073: /*
1074: * If we have no custom end-query or lookup failed, then try
1075: * pulling it out of the hashtable.
1076: */
1077:
1.214 schwarze 1078: t = roff_parse(r, *bufp, &pos, ln, ppos);
1.79 kristaps 1079:
1.106 kristaps 1080: if (ROFF_cblock != t) {
1.220 schwarze 1081: if (ROFF_ig != tok)
1.193 schwarze 1082: roff_setstr(r, r->last->name, *bufp + ppos, 2);
1.79 kristaps 1083: return(ROFF_IGN);
1.106 kristaps 1084: }
1.79 kristaps 1085:
1086: assert(roffs[t].proc);
1.207 schwarze 1087: return((*roffs[t].proc)(r, t, bufp, szp, ln, ppos, pos, offs));
1.79 kristaps 1088: }
1089:
1090: static enum rofferr
1.80 kristaps 1091: roff_block_text(ROFF_ARGS)
1.78 kristaps 1092: {
1093:
1.220 schwarze 1094: if (ROFF_ig != tok)
1.193 schwarze 1095: roff_setstr(r, r->last->name, *bufp + pos, 2);
1.106 kristaps 1096:
1.78 kristaps 1097: return(ROFF_IGN);
1098: }
1099:
1100: static enum rofferr
1.82 kristaps 1101: roff_cond_sub(ROFF_ARGS)
1102: {
1103: enum rofft t;
1.139 kristaps 1104: char *ep;
1.198 schwarze 1105: int rr;
1.82 kristaps 1106:
1107: rr = r->last->rule;
1.139 kristaps 1108: roffnode_cleanscope(r);
1.214 schwarze 1109: t = roff_parse(r, *bufp, &pos, ln, ppos);
1.82 kristaps 1110:
1.139 kristaps 1111: /*
1.177 schwarze 1112: * Fully handle known macros when they are structurally
1113: * required or when the conditional evaluated to true.
1.87 kristaps 1114: */
1115:
1.177 schwarze 1116: if ((ROFF_MAX != t) &&
1.198 schwarze 1117: (rr || ROFFMAC_STRUCT & roffs[t].flags)) {
1.177 schwarze 1118: assert(roffs[t].proc);
1119: return((*roffs[t].proc)(r, t, bufp, szp,
1.207 schwarze 1120: ln, ppos, pos, offs));
1.177 schwarze 1121: }
1.144 kristaps 1122:
1.196 schwarze 1123: /*
1124: * If `\}' occurs on a macro line without a preceding macro,
1125: * drop the line completely.
1126: */
1127:
1128: ep = *bufp + pos;
1129: if ('\\' == ep[0] && '}' == ep[1])
1.198 schwarze 1130: rr = 0;
1.196 schwarze 1131:
1.177 schwarze 1132: /* Always check for the closing delimiter `\}'. */
1.144 kristaps 1133:
1.177 schwarze 1134: while (NULL != (ep = strchr(ep, '\\'))) {
1.197 schwarze 1135: if ('}' == *(++ep)) {
1136: *ep = '&';
1137: roff_ccond(r, ln, ep - *bufp - 1);
1138: }
1139: ++ep;
1.177 schwarze 1140: }
1.198 schwarze 1141: return(rr ? ROFF_CONT : ROFF_IGN);
1.82 kristaps 1142: }
1143:
1144: static enum rofferr
1145: roff_cond_text(ROFF_ARGS)
1.78 kristaps 1146: {
1.140 kristaps 1147: char *ep;
1.198 schwarze 1148: int rr;
1.82 kristaps 1149:
1150: rr = r->last->rule;
1.140 kristaps 1151: roffnode_cleanscope(r);
1.82 kristaps 1152:
1.197 schwarze 1153: ep = *bufp + pos;
1154: while (NULL != (ep = strchr(ep, '\\'))) {
1155: if ('}' == *(++ep)) {
1156: *ep = '&';
1157: roff_ccond(r, ln, ep - *bufp - 1);
1158: }
1159: ++ep;
1.78 kristaps 1160: }
1.198 schwarze 1161: return(rr ? ROFF_CONT : ROFF_IGN);
1.74 kristaps 1162: }
1163:
1.204 schwarze 1164: /*
1165: * Parse a single signed integer number. Stop at the first non-digit.
1166: * If there is at least one digit, return success and advance the
1167: * parse point, else return failure and let the parse point unchanged.
1168: * Ignore overflows, treat them just like the C language.
1169: */
1.184 schwarze 1170: static int
1171: roff_getnum(const char *v, int *pos, int *res)
1172: {
1.206 schwarze 1173: int myres, n, p;
1174:
1175: if (NULL == res)
1176: res = &myres;
1.184 schwarze 1177:
1178: p = *pos;
1179: n = v[p] == '-';
1180: if (n)
1181: p++;
1182:
1183: for (*res = 0; isdigit((unsigned char)v[p]); p++)
1.204 schwarze 1184: *res = 10 * *res + v[p] - '0';
1.184 schwarze 1185: if (p == *pos + n)
1186: return 0;
1187:
1188: if (n)
1189: *res = -*res;
1190:
1191: *pos = p;
1192: return 1;
1193: }
1194:
1.198 schwarze 1195: /*
1196: * Evaluate a string comparison condition.
1197: * The first character is the delimiter.
1198: * Succeed if the string up to its second occurrence
1199: * matches the string up to its third occurence.
1200: * Advance the cursor after the third occurrence
1201: * or lacking that, to the end of the line.
1202: */
1203: static int
1204: roff_evalstrcond(const char *v, int *pos)
1205: {
1206: const char *s1, *s2, *s3;
1207: int match;
1208:
1209: match = 0;
1210: s1 = v + *pos; /* initial delimiter */
1211: s2 = s1 + 1; /* for scanning the first string */
1212: s3 = strchr(s2, *s1); /* for scanning the second string */
1213:
1214: if (NULL == s3) /* found no middle delimiter */
1215: goto out;
1216:
1217: while ('\0' != *++s3) {
1218: if (*s2 != *s3) { /* mismatch */
1219: s3 = strchr(s3, *s1);
1220: break;
1221: }
1222: if (*s3 == *s1) { /* found the final delimiter */
1223: match = 1;
1224: break;
1225: }
1226: s2++;
1227: }
1228:
1229: out:
1230: if (NULL == s3)
1231: s3 = strchr(s2, '\0');
1232: else
1233: s3++;
1234: *pos = s3 - v;
1235: return(match);
1236: }
1237:
1.204 schwarze 1238: /*
1239: * Evaluate an optionally negated single character, numerical,
1240: * or string condition.
1241: */
1.198 schwarze 1242: static int
1.88 kristaps 1243: roff_evalcond(const char *v, int *pos)
1244: {
1.204 schwarze 1245: int wanttrue, number;
1.88 kristaps 1246:
1.198 schwarze 1247: if ('!' == v[*pos]) {
1248: wanttrue = 0;
1249: (*pos)++;
1250: } else
1251: wanttrue = 1;
1252:
1.88 kristaps 1253: switch (v[*pos]) {
1.207 schwarze 1254: case 'n':
1.198 schwarze 1255: /* FALLTHROUGH */
1.207 schwarze 1256: case 'o':
1.88 kristaps 1257: (*pos)++;
1.198 schwarze 1258: return(wanttrue);
1.207 schwarze 1259: case 'c':
1.198 schwarze 1260: /* FALLTHROUGH */
1.207 schwarze 1261: case 'd':
1.198 schwarze 1262: /* FALLTHROUGH */
1.207 schwarze 1263: case 'e':
1.88 kristaps 1264: /* FALLTHROUGH */
1.207 schwarze 1265: case 'r':
1.88 kristaps 1266: /* FALLTHROUGH */
1.207 schwarze 1267: case 't':
1.88 kristaps 1268: (*pos)++;
1.198 schwarze 1269: return(!wanttrue);
1.88 kristaps 1270: default:
1271: break;
1272: }
1273:
1.204 schwarze 1274: if (roff_evalnum(v, pos, &number, 0))
1275: return((number > 0) == wanttrue);
1276: else
1.198 schwarze 1277: return(roff_evalstrcond(v, pos) == wanttrue);
1.88 kristaps 1278: }
1279:
1.74 kristaps 1280: static enum rofferr
1.103 kristaps 1281: roff_line_ignore(ROFF_ARGS)
1.89 kristaps 1282: {
1.123 schwarze 1283:
1.89 kristaps 1284: return(ROFF_IGN);
1285: }
1286:
1.104 kristaps 1287: static enum rofferr
1.82 kristaps 1288: roff_cond(ROFF_ARGS)
1.74 kristaps 1289: {
1.173 schwarze 1290:
1291: roffnode_push(r, tok, NULL, ln, ppos);
1.74 kristaps 1292:
1.207 schwarze 1293: /*
1.134 kristaps 1294: * An `.el' has no conditional body: it will consume the value
1295: * of the current rstack entry set in prior `ie' calls or
1.207 schwarze 1296: * defaults to DENY.
1.134 kristaps 1297: *
1298: * If we're not an `el', however, then evaluate the conditional.
1299: */
1.133 kristaps 1300:
1.173 schwarze 1301: r->last->rule = ROFF_el == tok ?
1.207 schwarze 1302: (r->rstackpos < 0 ? 0 : r->rstack[r->rstackpos--]) :
1303: roff_evalcond(*bufp, &pos);
1.77 kristaps 1304:
1.134 kristaps 1305: /*
1306: * An if-else will put the NEGATION of the current evaluated
1307: * conditional into the stack of rules.
1308: */
1309:
1.84 schwarze 1310: if (ROFF_ie == tok) {
1.223 schwarze 1311: if (r->rstackpos + 1 == r->rstacksz) {
1312: r->rstacksz += 16;
1313: r->rstack = mandoc_reallocarray(r->rstack,
1314: r->rstacksz, sizeof(int));
1.134 kristaps 1315: }
1.198 schwarze 1316: r->rstack[++r->rstackpos] = !r->last->rule;
1.82 kristaps 1317: }
1.88 kristaps 1318:
1319: /* If the parent has false as its rule, then so do we. */
1320:
1.198 schwarze 1321: if (r->last->parent && !r->last->parent->rule)
1322: r->last->rule = 0;
1.88 kristaps 1323:
1324: /*
1.173 schwarze 1325: * Determine scope.
1326: * If there is nothing on the line after the conditional,
1327: * not even whitespace, use next-line scope.
1.88 kristaps 1328: */
1.74 kristaps 1329:
1.173 schwarze 1330: if ('\0' == (*bufp)[pos]) {
1331: r->last->endspan = 2;
1332: goto out;
1333: }
1334:
1335: while (' ' == (*bufp)[pos])
1336: pos++;
1337:
1338: /* An opening brace requests multiline scope. */
1.75 kristaps 1339:
1340: if ('\\' == (*bufp)[pos] && '{' == (*bufp)[pos + 1]) {
1341: r->last->endspan = -1;
1342: pos += 2;
1.173 schwarze 1343: goto out;
1.207 schwarze 1344: }
1.74 kristaps 1345:
1.77 kristaps 1346: /*
1.173 schwarze 1347: * Anything else following the conditional causes
1348: * single-line scope. Warn if the scope contains
1349: * nothing but trailing whitespace.
1.77 kristaps 1350: */
1351:
1.75 kristaps 1352: if ('\0' == (*bufp)[pos])
1.216 schwarze 1353: mandoc_msg(MANDOCERR_COND_EMPTY, r->parse,
1354: ln, ppos, roffs[tok].name);
1.77 kristaps 1355:
1.173 schwarze 1356: r->last->endspan = 1;
1.74 kristaps 1357:
1.173 schwarze 1358: out:
1.75 kristaps 1359: *offs = pos;
1360: return(ROFF_RERUN);
1.83 schwarze 1361: }
1362:
1363: static enum rofferr
1.92 schwarze 1364: roff_ds(ROFF_ARGS)
1365: {
1.212 schwarze 1366: char *string;
1367: const char *name;
1368: size_t namesz;
1.96 kristaps 1369:
1370: /*
1.212 schwarze 1371: * The first word is the name of the string.
1372: * If it is empty or terminated by an escape sequence,
1373: * abort the `ds' request without defining anything.
1.96 kristaps 1374: */
1.92 schwarze 1375:
1.212 schwarze 1376: name = string = *bufp + pos;
1.92 schwarze 1377: if ('\0' == *name)
1378: return(ROFF_IGN);
1379:
1.212 schwarze 1380: namesz = roff_getname(r, &string, ln, pos);
1381: if ('\\' == name[namesz])
1382: return(ROFF_IGN);
1383:
1384: /* Read past the initial double-quote, if any. */
1.121 schwarze 1385: if ('"' == *string)
1.92 schwarze 1386: string++;
1387:
1.96 kristaps 1388: /* The rest is the value. */
1.212 schwarze 1389: roff_setstrn(&r->strtab, name, namesz, string, strlen(string),
1390: ROFF_as == tok);
1.92 schwarze 1391: return(ROFF_IGN);
1392: }
1393:
1.204 schwarze 1394: /*
1395: * Parse a single operator, one or two characters long.
1396: * If the operator is recognized, return success and advance the
1397: * parse point, else return failure and let the parse point unchanged.
1398: */
1399: static int
1400: roff_getop(const char *v, int *pos, char *res)
1401: {
1402:
1403: *res = v[*pos];
1404:
1405: switch (*res) {
1.207 schwarze 1406: case '+':
1.204 schwarze 1407: /* FALLTHROUGH */
1.207 schwarze 1408: case '-':
1.204 schwarze 1409: /* FALLTHROUGH */
1.207 schwarze 1410: case '*':
1.204 schwarze 1411: /* FALLTHROUGH */
1.207 schwarze 1412: case '/':
1.204 schwarze 1413: /* FALLTHROUGH */
1.207 schwarze 1414: case '%':
1.204 schwarze 1415: /* FALLTHROUGH */
1.207 schwarze 1416: case '&':
1.204 schwarze 1417: /* FALLTHROUGH */
1.207 schwarze 1418: case ':':
1.204 schwarze 1419: break;
1420: case '<':
1421: switch (v[*pos + 1]) {
1.207 schwarze 1422: case '=':
1.204 schwarze 1423: *res = 'l';
1424: (*pos)++;
1425: break;
1.207 schwarze 1426: case '>':
1.204 schwarze 1427: *res = '!';
1428: (*pos)++;
1429: break;
1.207 schwarze 1430: case '?':
1.204 schwarze 1431: *res = 'i';
1432: (*pos)++;
1433: break;
1434: default:
1435: break;
1436: }
1437: break;
1438: case '>':
1439: switch (v[*pos + 1]) {
1.207 schwarze 1440: case '=':
1.204 schwarze 1441: *res = 'g';
1442: (*pos)++;
1443: break;
1.207 schwarze 1444: case '?':
1.204 schwarze 1445: *res = 'a';
1446: (*pos)++;
1447: break;
1448: default:
1449: break;
1450: }
1451: break;
1452: case '=':
1453: if ('=' == v[*pos + 1])
1454: (*pos)++;
1455: break;
1456: default:
1457: return(0);
1458: }
1459: (*pos)++;
1460:
1461: return(*res);
1462: }
1463:
1464: /*
1465: * Evaluate either a parenthesized numeric expression
1466: * or a single signed integer number.
1467: */
1468: static int
1469: roff_evalpar(const char *v, int *pos, int *res)
1470: {
1471:
1472: if ('(' != v[*pos])
1473: return(roff_getnum(v, pos, res));
1474:
1475: (*pos)++;
1476: if ( ! roff_evalnum(v, pos, res, 1))
1477: return(0);
1478:
1.206 schwarze 1479: /*
1480: * Omission of the closing parenthesis
1481: * is an error in validation mode,
1482: * but ignored in evaluation mode.
1483: */
1484:
1.204 schwarze 1485: if (')' == v[*pos])
1486: (*pos)++;
1.206 schwarze 1487: else if (NULL == res)
1488: return(0);
1.204 schwarze 1489:
1490: return(1);
1491: }
1492:
1493: /*
1494: * Evaluate a complete numeric expression.
1495: * Proceed left to right, there is no concept of precedence.
1496: */
1497: static int
1498: roff_evalnum(const char *v, int *pos, int *res, int skipwhite)
1499: {
1500: int mypos, operand2;
1501: char operator;
1502:
1503: if (NULL == pos) {
1504: mypos = 0;
1505: pos = &mypos;
1506: }
1507:
1508: if (skipwhite)
1509: while (isspace((unsigned char)v[*pos]))
1510: (*pos)++;
1511:
1512: if ( ! roff_evalpar(v, pos, res))
1513: return(0);
1514:
1515: while (1) {
1516: if (skipwhite)
1517: while (isspace((unsigned char)v[*pos]))
1518: (*pos)++;
1519:
1520: if ( ! roff_getop(v, pos, &operator))
1521: break;
1522:
1523: if (skipwhite)
1524: while (isspace((unsigned char)v[*pos]))
1525: (*pos)++;
1526:
1527: if ( ! roff_evalpar(v, pos, &operand2))
1528: return(0);
1529:
1530: if (skipwhite)
1531: while (isspace((unsigned char)v[*pos]))
1532: (*pos)++;
1.206 schwarze 1533:
1534: if (NULL == res)
1535: continue;
1.204 schwarze 1536:
1537: switch (operator) {
1.207 schwarze 1538: case '+':
1.204 schwarze 1539: *res += operand2;
1540: break;
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 '%':
1.204 schwarze 1551: *res %= operand2;
1552: break;
1.207 schwarze 1553: case '<':
1.204 schwarze 1554: *res = *res < operand2;
1555: break;
1.207 schwarze 1556: case '>':
1.204 schwarze 1557: *res = *res > operand2;
1558: break;
1.207 schwarze 1559: case 'l':
1.204 schwarze 1560: *res = *res <= operand2;
1561: break;
1.207 schwarze 1562: case 'g':
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 '!':
1.204 schwarze 1569: *res = *res != operand2;
1570: break;
1.207 schwarze 1571: case '&':
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 'i':
1.204 schwarze 1578: if (operand2 < *res)
1579: *res = operand2;
1580: break;
1.207 schwarze 1581: case 'a':
1.204 schwarze 1582: if (operand2 > *res)
1583: *res = operand2;
1584: break;
1585: default:
1586: abort();
1587: }
1588: }
1589: return(1);
1590: }
1591:
1.180 schwarze 1592: void
1.187 schwarze 1593: roff_setreg(struct roff *r, const char *name, int val, char sign)
1.147 kristaps 1594: {
1.180 schwarze 1595: struct roffreg *reg;
1596:
1597: /* Search for an existing register with the same name. */
1598: reg = r->regtab;
1599:
1600: while (reg && strcmp(name, reg->key.p))
1601: reg = reg->next;
1.147 kristaps 1602:
1.180 schwarze 1603: if (NULL == reg) {
1604: /* Create a new register. */
1605: reg = mandoc_malloc(sizeof(struct roffreg));
1606: reg->key.p = mandoc_strdup(name);
1607: reg->key.sz = strlen(name);
1.187 schwarze 1608: reg->val = 0;
1.180 schwarze 1609: reg->next = r->regtab;
1610: r->regtab = reg;
1611: }
1612:
1.187 schwarze 1613: if ('+' == sign)
1614: reg->val += val;
1615: else if ('-' == sign)
1616: reg->val -= val;
1617: else
1618: reg->val = val;
1.147 kristaps 1619: }
1620:
1.192 schwarze 1621: /*
1622: * Handle some predefined read-only number registers.
1623: * For now, return -1 if the requested register is not predefined;
1624: * in case a predefined read-only register having the value -1
1625: * were to turn up, another special value would have to be chosen.
1626: */
1627: static int
1628: roff_getregro(const char *name)
1629: {
1630:
1631: switch (*name) {
1.207 schwarze 1632: case 'A': /* ASCII approximation mode is always off. */
1.192 schwarze 1633: return(0);
1.207 schwarze 1634: case 'g': /* Groff compatibility mode is always on. */
1.192 schwarze 1635: return(1);
1.207 schwarze 1636: case 'H': /* Fixed horizontal resolution. */
1.192 schwarze 1637: return (24);
1.207 schwarze 1638: case 'j': /* Always adjust left margin only. */
1.192 schwarze 1639: return(0);
1.207 schwarze 1640: case 'T': /* Some output device is always defined. */
1.192 schwarze 1641: return(1);
1.207 schwarze 1642: case 'V': /* Fixed vertical resolution. */
1.192 schwarze 1643: return (40);
1644: default:
1645: return (-1);
1646: }
1647: }
1648:
1.181 schwarze 1649: int
1.180 schwarze 1650: roff_getreg(const struct roff *r, const char *name)
1.147 kristaps 1651: {
1.180 schwarze 1652: struct roffreg *reg;
1.192 schwarze 1653: int val;
1654:
1655: if ('.' == name[0] && '\0' != name[1] && '\0' == name[2]) {
1656: val = roff_getregro(name + 1);
1657: if (-1 != val)
1658: return (val);
1659: }
1.180 schwarze 1660:
1661: for (reg = r->regtab; reg; reg = reg->next)
1662: if (0 == strcmp(name, reg->key.p))
1.181 schwarze 1663: return(reg->val);
1664:
1665: return(0);
1666: }
1667:
1668: static int
1669: roff_getregn(const struct roff *r, const char *name, size_t len)
1670: {
1671: struct roffreg *reg;
1.192 schwarze 1672: int val;
1673:
1674: if ('.' == name[0] && 2 == len) {
1675: val = roff_getregro(name + 1);
1676: if (-1 != val)
1677: return (val);
1678: }
1.181 schwarze 1679:
1680: for (reg = r->regtab; reg; reg = reg->next)
1681: if (len == reg->key.sz &&
1682: 0 == strncmp(name, reg->key.p, len))
1683: return(reg->val);
1.147 kristaps 1684:
1.180 schwarze 1685: return(0);
1.147 kristaps 1686: }
1687:
1.180 schwarze 1688: static void
1689: roff_freereg(struct roffreg *reg)
1.147 kristaps 1690: {
1.180 schwarze 1691: struct roffreg *old_reg;
1.147 kristaps 1692:
1.180 schwarze 1693: while (NULL != reg) {
1694: free(reg->key.p);
1695: old_reg = reg;
1696: reg = reg->next;
1697: free(old_reg);
1698: }
1.147 kristaps 1699: }
1.92 schwarze 1700:
1701: static enum rofferr
1.89 kristaps 1702: roff_nr(ROFF_ARGS)
1.83 schwarze 1703: {
1.212 schwarze 1704: char *key, *val;
1705: size_t keysz;
1.138 kristaps 1706: int iv;
1.187 schwarze 1707: char sign;
1.89 kristaps 1708:
1.212 schwarze 1709: key = val = *bufp + pos;
1710: if ('\0' == *key)
1711: return(ROFF_IGN);
1712:
1713: keysz = roff_getname(r, &val, ln, pos);
1714: if ('\\' == key[keysz])
1715: return(ROFF_IGN);
1716: key[keysz] = '\0';
1.89 kristaps 1717:
1.187 schwarze 1718: sign = *val;
1719: if ('+' == sign || '-' == sign)
1720: val++;
1721:
1.204 schwarze 1722: if (roff_evalnum(val, NULL, &iv, 0))
1723: roff_setreg(r, key, iv, sign);
1.109 kristaps 1724:
1.203 schwarze 1725: return(ROFF_IGN);
1726: }
1727:
1728: static enum rofferr
1729: roff_rr(ROFF_ARGS)
1730: {
1731: struct roffreg *reg, **prev;
1.212 schwarze 1732: char *name, *cp;
1733: size_t namesz;
1.203 schwarze 1734:
1.212 schwarze 1735: name = cp = *bufp + pos;
1736: if ('\0' == *name)
1737: return(ROFF_IGN);
1738: namesz = roff_getname(r, &cp, ln, pos);
1739: name[namesz] = '\0';
1.203 schwarze 1740:
1741: prev = &r->regtab;
1742: while (1) {
1743: reg = *prev;
1744: if (NULL == reg || !strcmp(name, reg->key.p))
1745: break;
1746: prev = ®->next;
1747: }
1748: if (NULL != reg) {
1749: *prev = reg->next;
1750: free(reg->key.p);
1751: free(reg);
1752: }
1.122 schwarze 1753: return(ROFF_IGN);
1754: }
1755:
1756: static enum rofferr
1757: roff_rm(ROFF_ARGS)
1758: {
1759: const char *name;
1760: char *cp;
1.212 schwarze 1761: size_t namesz;
1.122 schwarze 1762:
1763: cp = *bufp + pos;
1764: while ('\0' != *cp) {
1.212 schwarze 1765: name = cp;
1766: namesz = roff_getname(r, &cp, ln, (int)(cp - *bufp));
1767: roff_setstrn(&r->strtab, name, namesz, NULL, 0, 0);
1768: if ('\\' == name[namesz])
1769: break;
1.122 schwarze 1770: }
1.178 schwarze 1771: return(ROFF_IGN);
1772: }
1773:
1774: static enum rofferr
1775: roff_it(ROFF_ARGS)
1776: {
1777: char *cp;
1778: size_t len;
1779: int iv;
1780:
1781: /* Parse the number of lines. */
1782: cp = *bufp + pos;
1783: len = strcspn(cp, " \t");
1784: cp[len] = '\0';
1785: if ((iv = mandoc_strntoi(cp, len, 10)) <= 0) {
1.222 schwarze 1786: mandoc_msg(MANDOCERR_IT_NONUM, r->parse,
1.207 schwarze 1787: ln, ppos, *bufp + 1);
1.178 schwarze 1788: return(ROFF_IGN);
1789: }
1790: cp += len + 1;
1791:
1792: /* Arm the input line trap. */
1793: roffit_lines = iv;
1794: roffit_macro = mandoc_strdup(cp);
1.109 kristaps 1795: return(ROFF_IGN);
1.175 schwarze 1796: }
1797:
1798: static enum rofferr
1799: roff_Dd(ROFF_ARGS)
1800: {
1801: const char *const *cp;
1802:
1.227 schwarze 1803: if ((r->options & (MPARSE_MDOC | MPARSE_QUICK)) == 0)
1.175 schwarze 1804: for (cp = __mdoc_reserved; *cp; cp++)
1805: roff_setstr(r, *cp, NULL, 0);
1806:
1.227 schwarze 1807: if (r->format == 0)
1808: r->format = MPARSE_MDOC;
1809:
1.175 schwarze 1810: return(ROFF_CONT);
1811: }
1812:
1813: static enum rofferr
1814: roff_TH(ROFF_ARGS)
1815: {
1816: const char *const *cp;
1817:
1.227 schwarze 1818: if ((r->options & MPARSE_QUICK) == 0)
1.175 schwarze 1819: for (cp = __man_reserved; *cp; cp++)
1820: roff_setstr(r, *cp, NULL, 0);
1821:
1.227 schwarze 1822: if (r->format == 0)
1823: r->format = MPARSE_MAN;
1824:
1.175 schwarze 1825: return(ROFF_CONT);
1.109 kristaps 1826: }
1827:
1828: static enum rofferr
1829: roff_TE(ROFF_ARGS)
1830: {
1831:
1832: if (NULL == r->tbl)
1.221 schwarze 1833: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
1834: ln, ppos, "TE");
1.115 kristaps 1835: else
1.151 kristaps 1836: tbl_end(&r->tbl);
1.109 kristaps 1837:
1.112 kristaps 1838: return(ROFF_IGN);
1839: }
1840:
1841: static enum rofferr
1842: roff_T_(ROFF_ARGS)
1843: {
1844:
1845: if (NULL == r->tbl)
1.221 schwarze 1846: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
1847: ln, ppos, "T&");
1.112 kristaps 1848: else
1.116 kristaps 1849: tbl_restart(ppos, ln, r->tbl);
1.112 kristaps 1850:
1.109 kristaps 1851: return(ROFF_IGN);
1852: }
1853:
1.230 schwarze 1854: /*
1855: * Handle in-line equation delimiters.
1856: */
1857: static enum rofferr
1858: roff_eqndelim(struct roff *r, char **bufp, size_t *szp, int pos)
1.151 kristaps 1859: {
1.233 ! schwarze 1860: char *cp1, *cp2;
! 1861: const char *bef_pr, *bef_nl, *mac, *aft_nl, *aft_pr;
1.151 kristaps 1862:
1.230 schwarze 1863: /*
1864: * Outside equations, look for an opening delimiter.
1865: * If we are inside an equation, we already know it is
1866: * in-line, or this function wouldn't have been called;
1867: * so look for a closing delimiter.
1868: */
1869:
1870: cp1 = *bufp + pos;
1871: cp2 = strchr(cp1, r->eqn == NULL ?
1872: r->last_eqn->odelim : r->last_eqn->cdelim);
1873: if (cp2 == NULL)
1874: return(ROFF_CONT);
1875:
1.233 ! schwarze 1876: *cp2++ = '\0';
! 1877: bef_pr = bef_nl = aft_nl = aft_pr = "";
! 1878:
! 1879: /* Handle preceding text, protecting whitespace. */
! 1880:
! 1881: if (**bufp != '\0') {
! 1882: if (r->eqn == NULL)
! 1883: bef_pr = "\\&";
! 1884: bef_nl = "\n";
! 1885: }
! 1886:
! 1887: /*
! 1888: * Prepare replacing the delimiter with an equation macro
! 1889: * and drop leading white space from the equation.
! 1890: */
1.230 schwarze 1891:
1.233 ! schwarze 1892: if (r->eqn == NULL) {
! 1893: while (*cp2 == ' ')
! 1894: cp2++;
! 1895: mac = ".EQ";
! 1896: } else
! 1897: mac = ".EN";
! 1898:
! 1899: /* Handle following text, protecting whitespace. */
! 1900:
! 1901: if (*cp2 != '\0') {
! 1902: aft_nl = "\n";
! 1903: if (r->eqn != NULL)
! 1904: aft_pr = "\\&";
! 1905: }
! 1906:
! 1907: /* Do the actual replacement. */
! 1908:
! 1909: *szp = mandoc_asprintf(&cp1, "%s%s%s%s%s%s%s", *bufp,
! 1910: bef_pr, bef_nl, mac, aft_nl, aft_pr, cp2) + 1;
1.230 schwarze 1911: free(*bufp);
1912: *bufp = cp1;
1913:
1914: /* Toggle the in-line state of the eqn subsystem. */
1915:
1916: r->eqn_inline = r->eqn == NULL;
1917: return(ROFF_REPARSE);
1.151 kristaps 1918: }
1919:
1.156 kristaps 1920: static void
1.207 schwarze 1921: roff_openeqn(struct roff *r, const char *name, int line,
1.151 kristaps 1922: int offs, const char *buf)
1.125 kristaps 1923: {
1.151 kristaps 1924: struct eqn_node *e;
1925: int poff;
1.125 kristaps 1926:
1927: assert(NULL == r->eqn);
1.151 kristaps 1928: e = eqn_alloc(name, offs, line, r->parse);
1.125 kristaps 1929:
1.230 schwarze 1930: if (r->last_eqn) {
1.125 kristaps 1931: r->last_eqn->next = e;
1.230 schwarze 1932: e->delim = r->last_eqn->delim;
1933: e->odelim = r->last_eqn->odelim;
1934: e->cdelim = r->last_eqn->cdelim;
1935: } else
1.125 kristaps 1936: r->first_eqn = r->last_eqn = e;
1937:
1938: r->eqn = r->last_eqn = e;
1.151 kristaps 1939:
1940: if (buf) {
1941: poff = 0;
1942: eqn_read(&r->eqn, line, buf, offs, &poff);
1943: }
1944: }
1945:
1946: static enum rofferr
1947: roff_EQ(ROFF_ARGS)
1948: {
1949:
1950: roff_openeqn(r, *bufp + pos, ln, ppos, NULL);
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