Annotation of mandoc/roff.c, Revision 1.237
1.237 ! schwarze 1: /* $Id: roff.c,v 1.236 2014/10/25 15:23:56 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 */
1.237 ! schwarze 110: const struct mchars *mchars; /* character table */
1.67 kristaps 111: struct roffnode *last; /* leaf of stack */
1.223 schwarze 112: int *rstack; /* stack of inverted `ie' values */
1.180 schwarze 113: struct roffreg *regtab; /* number registers */
1.166 kristaps 114: struct roffkv *strtab; /* user-defined strings & macros */
1.167 kristaps 115: struct roffkv *xmbtab; /* multi-byte trans table (`tr') */
116: struct roffstr *xtab; /* single-byte trans table (`tr') */
1.106 kristaps 117: const char *current_string; /* value of last called user macro */
1.118 kristaps 118: struct tbl_node *first_tbl; /* first table parsed */
119: struct tbl_node *last_tbl; /* last table parsed */
120: struct tbl_node *tbl; /* current table being parsed */
1.125 kristaps 121: struct eqn_node *last_eqn; /* last equation parsed */
122: struct eqn_node *first_eqn; /* first equation parsed */
123: struct eqn_node *eqn; /* current equation being parsed */
1.230 schwarze 124: int eqn_inline; /* current equation is inline */
1.223 schwarze 125: int options; /* parse options */
126: int rstacksz; /* current size limit of rstack */
127: int rstackpos; /* position in rstack */
1.227 schwarze 128: int format; /* current file in mdoc or man format */
1.223 schwarze 129: char control; /* control character */
1.79 kristaps 130: };
131:
1.67 kristaps 132: struct roffnode {
133: enum rofft tok; /* type of node */
134: struct roffnode *parent; /* up one in stack */
135: int line; /* parse line */
136: int col; /* parse col */
1.106 kristaps 137: char *name; /* node name, e.g. macro name */
1.79 kristaps 138: char *end; /* end-rules: custom token */
139: int endspan; /* end-rules: next-line or infty */
1.198 schwarze 140: int rule; /* current evaluation rule */
1.67 kristaps 141: };
142:
143: #define ROFF_ARGS struct roff *r, /* parse ctx */ \
1.72 kristaps 144: enum rofft tok, /* tok of macro */ \
1.207 schwarze 145: char **bufp, /* input buffer */ \
1.67 kristaps 146: size_t *szp, /* size of input buffer */ \
147: int ln, /* parse line */ \
1.75 kristaps 148: int ppos, /* original pos in buffer */ \
149: int pos, /* current pos in buffer */ \
1.74 kristaps 150: int *offs /* reset offset of buffer data */
1.67 kristaps 151:
152: typedef enum rofferr (*roffproc)(ROFF_ARGS);
153:
154: struct roffmac {
155: const char *name; /* macro name */
1.79 kristaps 156: roffproc proc; /* process new macro */
157: roffproc text; /* process as child text of macro */
158: roffproc sub; /* process as child of macro */
159: int flags;
160: #define ROFFMAC_STRUCT (1 << 0) /* always interpret */
1.85 kristaps 161: struct roffmac *next;
1.67 kristaps 162: };
163:
1.141 kristaps 164: struct predef {
165: const char *name; /* predefined input name */
166: const char *str; /* replacement symbol */
167: };
168:
169: #define PREDEF(__name, __str) \
170: { (__name), (__str) },
171:
1.155 kristaps 172: static enum rofft roffhash_find(const char *, size_t);
173: static void roffhash_init(void);
174: static void roffnode_cleanscope(struct roff *);
175: static void roffnode_pop(struct roff *);
176: static void roffnode_push(struct roff *, enum rofft,
177: const char *, int, int);
1.80 kristaps 178: static enum rofferr roff_block(ROFF_ARGS);
179: static enum rofferr roff_block_text(ROFF_ARGS);
180: static enum rofferr roff_block_sub(ROFF_ARGS);
181: static enum rofferr roff_cblock(ROFF_ARGS);
1.174 kristaps 182: static enum rofferr roff_cc(ROFF_ARGS);
1.195 schwarze 183: static void roff_ccond(struct roff *, int, int);
1.82 kristaps 184: static enum rofferr roff_cond(ROFF_ARGS);
185: static enum rofferr roff_cond_text(ROFF_ARGS);
186: static enum rofferr roff_cond_sub(ROFF_ARGS);
1.92 schwarze 187: static enum rofferr roff_ds(ROFF_ARGS);
1.230 schwarze 188: static enum rofferr roff_eqndelim(struct roff *,
189: char **, size_t *, int);
1.234 kristaps 190: static int roff_evalcond(struct roff *r, int,
191: const char *, int *);
192: static int roff_evalnum(struct roff *, int,
193: const char *, int *, int *, int);
194: static int roff_evalpar(struct roff *, int,
195: const char *, int *, int *);
1.198 schwarze 196: static int roff_evalstrcond(const char *, int *);
1.155 kristaps 197: static void roff_free1(struct roff *);
1.180 schwarze 198: static void roff_freereg(struct roffreg *);
1.167 kristaps 199: static void roff_freestr(struct roffkv *);
1.212 schwarze 200: static size_t roff_getname(struct roff *, char **, int, int);
1.184 schwarze 201: static int roff_getnum(const char *, int *, int *);
202: static int roff_getop(const char *, int *, char *);
1.181 schwarze 203: static int roff_getregn(const struct roff *,
204: const char *, size_t);
1.192 schwarze 205: static int roff_getregro(const char *name);
1.207 schwarze 206: static const char *roff_getstrn(const struct roff *,
1.94 kristaps 207: const char *, size_t);
1.178 schwarze 208: static enum rofferr roff_it(ROFF_ARGS);
1.103 kristaps 209: static enum rofferr roff_line_ignore(ROFF_ARGS);
1.89 kristaps 210: static enum rofferr roff_nr(ROFF_ARGS);
1.214 schwarze 211: static enum rofft roff_parse(struct roff *, char *, int *,
212: int, int);
1.178 schwarze 213: static enum rofferr roff_parsetext(char **, size_t *, int, int *);
1.207 schwarze 214: static enum rofferr roff_res(struct roff *,
1.142 kristaps 215: char **, size_t *, int, int);
1.122 schwarze 216: static enum rofferr roff_rm(ROFF_ARGS);
1.203 schwarze 217: static enum rofferr roff_rr(ROFF_ARGS);
1.94 kristaps 218: static void roff_setstr(struct roff *,
1.106 kristaps 219: const char *, const char *, int);
1.207 schwarze 220: static void roff_setstrn(struct roffkv **, const char *,
1.164 kristaps 221: size_t, const char *, size_t, int);
1.105 kristaps 222: static enum rofferr roff_so(ROFF_ARGS);
1.164 kristaps 223: static enum rofferr roff_tr(ROFF_ARGS);
1.175 schwarze 224: static enum rofferr roff_Dd(ROFF_ARGS);
225: static enum rofferr roff_TH(ROFF_ARGS);
1.109 kristaps 226: static enum rofferr roff_TE(ROFF_ARGS);
227: static enum rofferr roff_TS(ROFF_ARGS);
1.125 kristaps 228: static enum rofferr roff_EQ(ROFF_ARGS);
229: static enum rofferr roff_EN(ROFF_ARGS);
1.112 kristaps 230: static enum rofferr roff_T_(ROFF_ARGS);
1.106 kristaps 231: static enum rofferr roff_userdef(ROFF_ARGS);
1.67 kristaps 232:
1.155 kristaps 233: /* See roffhash_find() */
1.85 kristaps 234:
235: #define ASCII_HI 126
236: #define ASCII_LO 33
237: #define HASHWIDTH (ASCII_HI - ASCII_LO + 1)
238:
239: static struct roffmac *hash[HASHWIDTH];
240:
241: static struct roffmac roffs[ROFF_MAX] = {
1.103 kristaps 242: { "ad", roff_line_ignore, NULL, NULL, 0, NULL },
1.85 kristaps 243: { "am", roff_block, roff_block_text, roff_block_sub, 0, NULL },
244: { "ami", roff_block, roff_block_text, roff_block_sub, 0, NULL },
245: { "am1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.193 schwarze 246: { "as", roff_ds, NULL, NULL, 0, NULL },
1.174 kristaps 247: { "cc", roff_cc, NULL, NULL, 0, NULL },
1.194 schwarze 248: { "ce", roff_line_ignore, NULL, NULL, 0, NULL },
1.85 kristaps 249: { "de", roff_block, roff_block_text, roff_block_sub, 0, NULL },
250: { "dei", roff_block, roff_block_text, roff_block_sub, 0, NULL },
251: { "de1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.92 schwarze 252: { "ds", roff_ds, NULL, NULL, 0, NULL },
1.85 kristaps 253: { "el", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
1.185 schwarze 254: { "fam", roff_line_ignore, NULL, NULL, 0, NULL },
1.186 schwarze 255: { "hw", roff_line_ignore, NULL, NULL, 0, NULL },
1.103 kristaps 256: { "hy", roff_line_ignore, NULL, NULL, 0, NULL },
1.85 kristaps 257: { "ie", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
258: { "if", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
259: { "ig", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.178 schwarze 260: { "it", roff_it, NULL, NULL, 0, NULL },
1.103 kristaps 261: { "ne", roff_line_ignore, NULL, NULL, 0, NULL },
262: { "nh", roff_line_ignore, NULL, NULL, 0, NULL },
1.104 kristaps 263: { "nr", roff_nr, NULL, NULL, 0, NULL },
1.124 schwarze 264: { "ns", roff_line_ignore, NULL, NULL, 0, NULL },
1.229 schwarze 265: { "pl", roff_line_ignore, NULL, NULL, 0, NULL },
1.124 schwarze 266: { "ps", roff_line_ignore, NULL, NULL, 0, NULL },
1.122 schwarze 267: { "rm", roff_rm, NULL, NULL, 0, NULL },
1.203 schwarze 268: { "rr", roff_rr, NULL, NULL, 0, NULL },
1.105 kristaps 269: { "so", roff_so, NULL, NULL, 0, NULL },
1.124 schwarze 270: { "ta", roff_line_ignore, NULL, NULL, 0, NULL },
1.164 kristaps 271: { "tr", roff_tr, NULL, NULL, 0, NULL },
1.175 schwarze 272: { "Dd", roff_Dd, NULL, NULL, 0, NULL },
273: { "TH", roff_TH, NULL, NULL, 0, NULL },
1.109 kristaps 274: { "TS", roff_TS, NULL, NULL, 0, NULL },
275: { "TE", roff_TE, NULL, NULL, 0, NULL },
1.112 kristaps 276: { "T&", roff_T_, NULL, NULL, 0, NULL },
1.125 kristaps 277: { "EQ", roff_EQ, NULL, NULL, 0, NULL },
278: { "EN", roff_EN, NULL, NULL, 0, NULL },
1.85 kristaps 279: { ".", roff_cblock, NULL, NULL, 0, NULL },
1.106 kristaps 280: { NULL, roff_userdef, NULL, NULL, 0, NULL },
1.67 kristaps 281: };
282:
1.200 schwarze 283: /* not currently implemented: Ds em Eq LP Me PP pp Or Rd Sf SH */
1.175 schwarze 284: const char *const __mdoc_reserved[] = {
285: "Ac", "Ad", "An", "Ao", "Ap", "Aq", "Ar", "At",
286: "Bc", "Bd", "Bf", "Bk", "Bl", "Bo", "Bq",
287: "Brc", "Bro", "Brq", "Bsx", "Bt", "Bx",
288: "Cd", "Cm", "Db", "Dc", "Dd", "Dl", "Do", "Dq",
1.200 schwarze 289: "Dt", "Dv", "Dx", "D1",
290: "Ec", "Ed", "Ef", "Ek", "El", "Em",
291: "En", "Eo", "Er", "Es", "Ev", "Ex",
1.175 schwarze 292: "Fa", "Fc", "Fd", "Fl", "Fn", "Fo", "Fr", "Ft", "Fx",
1.200 schwarze 293: "Hf", "Ic", "In", "It", "Lb", "Li", "Lk", "Lp",
294: "Ms", "Mt", "Nd", "Nm", "No", "Ns", "Nx",
1.175 schwarze 295: "Oc", "Oo", "Op", "Os", "Ot", "Ox",
1.200 schwarze 296: "Pa", "Pc", "Pf", "Po", "Pp", "Pq",
297: "Qc", "Ql", "Qo", "Qq", "Re", "Rs", "Rv",
298: "Sc", "Sh", "Sm", "So", "Sq",
1.175 schwarze 299: "Ss", "St", "Sx", "Sy",
300: "Ta", "Tn", "Ud", "Ux", "Va", "Vt", "Xc", "Xo", "Xr",
1.200 schwarze 301: "%A", "%B", "%C", "%D", "%I", "%J", "%N", "%O",
1.175 schwarze 302: "%P", "%Q", "%R", "%T", "%U", "%V",
303: NULL
304: };
305:
1.200 schwarze 306: /* not currently implemented: BT DE DS ME MT PT SY TQ YS */
1.175 schwarze 307: const char *const __man_reserved[] = {
1.200 schwarze 308: "AT", "B", "BI", "BR", "DT",
309: "EE", "EN", "EQ", "EX", "HP", "I", "IB", "IP", "IR",
310: "LP", "OP", "P", "PD", "PP",
311: "R", "RB", "RE", "RI", "RS", "SB", "SH", "SM", "SS",
312: "TE", "TH", "TP", "TS", "T&", "UC", "UE", "UR",
1.175 schwarze 313: NULL
314: };
315:
1.141 kristaps 316: /* Array of injected predefined strings. */
317: #define PREDEFS_MAX 38
318: static const struct predef predefs[PREDEFS_MAX] = {
319: #include "predefs.in"
320: };
321:
1.155 kristaps 322: /* See roffhash_find() */
1.85 kristaps 323: #define ROFF_HASH(p) (p[0] - ASCII_LO)
324:
1.178 schwarze 325: static int roffit_lines; /* number of lines to delay */
326: static char *roffit_macro; /* nil-terminated macro line */
327:
1.207 schwarze 328:
1.85 kristaps 329: static void
1.155 kristaps 330: roffhash_init(void)
1.85 kristaps 331: {
332: struct roffmac *n;
333: int buc, i;
334:
1.106 kristaps 335: for (i = 0; i < (int)ROFF_USERDEF; i++) {
1.85 kristaps 336: assert(roffs[i].name[0] >= ASCII_LO);
337: assert(roffs[i].name[0] <= ASCII_HI);
338:
339: buc = ROFF_HASH(roffs[i].name);
340:
341: if (NULL != (n = hash[buc])) {
342: for ( ; n->next; n = n->next)
343: /* Do nothing. */ ;
344: n->next = &roffs[i];
345: } else
346: hash[buc] = &roffs[i];
347: }
348: }
349:
1.67 kristaps 350: /*
351: * Look up a roff token by its name. Returns ROFF_MAX if no macro by
352: * the nil-terminated string name could be found.
353: */
354: static enum rofft
1.155 kristaps 355: roffhash_find(const char *p, size_t s)
1.67 kristaps 356: {
1.85 kristaps 357: int buc;
358: struct roffmac *n;
1.67 kristaps 359:
1.85 kristaps 360: /*
361: * libroff has an extremely simple hashtable, for the time
362: * being, which simply keys on the first character, which must
363: * be printable, then walks a chain. It works well enough until
364: * optimised.
365: */
366:
367: if (p[0] < ASCII_LO || p[0] > ASCII_HI)
368: return(ROFF_MAX);
369:
370: buc = ROFF_HASH(p);
371:
372: if (NULL == (n = hash[buc]))
373: return(ROFF_MAX);
374: for ( ; n; n = n->next)
1.106 kristaps 375: if (0 == strncmp(n->name, p, s) && '\0' == n->name[(int)s])
1.85 kristaps 376: return((enum rofft)(n - roffs));
1.67 kristaps 377:
378: return(ROFF_MAX);
379: }
380:
381: /*
382: * Pop the current node off of the stack of roff instructions currently
383: * pending.
384: */
385: static void
386: roffnode_pop(struct roff *r)
387: {
388: struct roffnode *p;
389:
1.75 kristaps 390: assert(r->last);
1.207 schwarze 391: p = r->last;
1.82 kristaps 392:
1.75 kristaps 393: r->last = r->last->parent;
1.106 kristaps 394: free(p->name);
395: free(p->end);
1.67 kristaps 396: free(p);
397: }
398:
399: /*
400: * Push a roff node onto the instruction stack. This must later be
401: * removed with roffnode_pop().
402: */
1.98 schwarze 403: static void
1.106 kristaps 404: roffnode_push(struct roff *r, enum rofft tok, const char *name,
405: int line, int col)
1.67 kristaps 406: {
407: struct roffnode *p;
408:
1.98 schwarze 409: p = mandoc_calloc(1, sizeof(struct roffnode));
1.67 kristaps 410: p->tok = tok;
1.106 kristaps 411: if (name)
412: p->name = mandoc_strdup(name);
1.67 kristaps 413: p->parent = r->last;
414: p->line = line;
415: p->col = col;
1.198 schwarze 416: p->rule = p->parent ? p->parent->rule : 0;
1.67 kristaps 417:
418: r->last = p;
419: }
420:
421: static void
422: roff_free1(struct roff *r)
423: {
1.176 schwarze 424: struct tbl_node *tbl;
1.125 kristaps 425: struct eqn_node *e;
1.167 kristaps 426: int i;
1.67 kristaps 427:
1.176 schwarze 428: while (NULL != (tbl = r->first_tbl)) {
429: r->first_tbl = tbl->next;
430: tbl_free(tbl);
1.109 kristaps 431: }
1.113 kristaps 432: r->first_tbl = r->last_tbl = r->tbl = NULL;
433:
1.125 kristaps 434: while (NULL != (e = r->first_eqn)) {
435: r->first_eqn = e->next;
436: eqn_free(e);
437: }
438: r->first_eqn = r->last_eqn = r->eqn = NULL;
439:
1.67 kristaps 440: while (r->last)
441: roffnode_pop(r);
1.109 kristaps 442:
1.223 schwarze 443: free (r->rstack);
444: r->rstack = NULL;
445: r->rstacksz = 0;
446: r->rstackpos = -1;
447:
448: roff_freereg(r->regtab);
449: r->regtab = NULL;
450:
1.167 kristaps 451: roff_freestr(r->strtab);
452: roff_freestr(r->xmbtab);
453: r->strtab = r->xmbtab = NULL;
454:
455: if (r->xtab)
456: for (i = 0; i < 128; i++)
457: free(r->xtab[i].p);
458: free(r->xtab);
459: r->xtab = NULL;
1.67 kristaps 460: }
461:
462: void
463: roff_reset(struct roff *r)
464: {
465:
466: roff_free1(r);
1.227 schwarze 467: r->format = r->options & (MPARSE_MDOC | MPARSE_MAN);
1.174 kristaps 468: r->control = 0;
1.67 kristaps 469: }
470:
471: void
472: roff_free(struct roff *r)
473: {
474:
475: roff_free1(r);
476: free(r);
477: }
478:
479: struct roff *
1.237 ! schwarze 480: roff_alloc(struct mparse *parse, const struct mchars *mchars, int options)
1.67 kristaps 481: {
482: struct roff *r;
483:
1.98 schwarze 484: r = mandoc_calloc(1, sizeof(struct roff));
1.128 kristaps 485: r->parse = parse;
1.237 ! schwarze 486: r->mchars = mchars;
1.199 schwarze 487: r->options = options;
1.227 schwarze 488: r->format = options & (MPARSE_MDOC | MPARSE_MAN);
1.82 kristaps 489: r->rstackpos = -1;
1.207 schwarze 490:
1.155 kristaps 491: roffhash_init();
1.141 kristaps 492:
1.67 kristaps 493: return(r);
494: }
495:
1.94 kristaps 496: /*
1.206 schwarze 497: * In the current line, expand escape sequences that tend to get
498: * used in numerical expressions and conditional requests.
499: * Also check the syntax of the remaining escape sequences.
1.154 kristaps 500: */
1.172 schwarze 501: static enum rofferr
1.142 kristaps 502: roff_res(struct roff *r, char **bufp, size_t *szp, int ln, int pos)
1.94 kristaps 503: {
1.208 schwarze 504: char ubuf[24]; /* buffer to print the number */
1.205 schwarze 505: const char *start; /* start of the string to process */
1.209 schwarze 506: char *stesc; /* start of an escape sequence ('\\') */
1.108 schwarze 507: const char *stnam; /* start of the name, after "[(*" */
508: const char *cp; /* end of the name, e.g. before ']' */
509: const char *res; /* the string to be substituted */
1.181 schwarze 510: char *nbuf; /* new buffer to copy bufp to */
511: size_t maxl; /* expected length of the escape name */
512: size_t naml; /* actual length of the escape name */
1.237 ! schwarze 513: enum mandoc_esc esc; /* type of the escape sequence */
! 514: int inaml; /* length returned from mandoc_escape() */
1.181 schwarze 515: int expand_count; /* to avoid infinite loops */
1.206 schwarze 516: int npos; /* position in numeric expression */
1.218 schwarze 517: int arg_complete; /* argument not interrupted by eol */
1.206 schwarze 518: char term; /* character terminating the escape */
1.94 kristaps 519:
1.170 schwarze 520: expand_count = 0;
1.205 schwarze 521: start = *bufp + pos;
522: stesc = strchr(start, '\0') - 1;
523: while (stesc-- > start) {
1.170 schwarze 524:
1.205 schwarze 525: /* Search backwards for the next backslash. */
526:
527: if ('\\' != *stesc)
528: continue;
529:
530: /* If it is escaped, skip it. */
531:
532: for (cp = stesc - 1; cp >= start; cp--)
533: if ('\\' != *cp)
534: break;
535:
536: if (0 == (stesc - cp) % 2) {
1.209 schwarze 537: stesc = (char *)cp;
1.205 schwarze 538: continue;
539: }
1.108 schwarze 540:
1.206 schwarze 541: /* Decide whether to expand or to check only. */
1.108 schwarze 542:
1.206 schwarze 543: term = '\0';
1.205 schwarze 544: cp = stesc + 1;
1.181 schwarze 545: switch (*cp) {
1.207 schwarze 546: case '*':
1.181 schwarze 547: res = NULL;
548: break;
1.207 schwarze 549: case 'B':
1.206 schwarze 550: /* FALLTHROUGH */
1.207 schwarze 551: case 'w':
1.206 schwarze 552: term = cp[1];
553: /* FALLTHROUGH */
1.207 schwarze 554: case 'n':
1.181 schwarze 555: res = ubuf;
556: break;
557: default:
1.237 ! schwarze 558: esc = mandoc_escape(&cp, &stnam, &inaml);
! 559: if (esc == ESCAPE_ERROR ||
! 560: (esc == ESCAPE_SPECIAL &&
! 561: mchars_spec2cp(r->mchars, stnam, inaml) < 0))
1.219 schwarze 562: mandoc_vmsg(MANDOCERR_ESC_BAD,
563: r->parse, ln, (int)(stesc - *bufp),
564: "%.*s", (int)(cp - stesc), stesc);
1.205 schwarze 565: continue;
1.152 kristaps 566: }
567:
1.205 schwarze 568: if (EXPAND_LIMIT < ++expand_count) {
569: mandoc_msg(MANDOCERR_ROFFLOOP, r->parse,
570: ln, (int)(stesc - *bufp), NULL);
571: return(ROFF_IGN);
572: }
1.108 schwarze 573:
574: /*
575: * The third character decides the length
1.181 schwarze 576: * of the name of the string or register.
1.108 schwarze 577: * Save a pointer to the name.
578: */
579:
1.206 schwarze 580: if ('\0' == term) {
581: switch (*++cp) {
1.207 schwarze 582: case '\0':
1.206 schwarze 583: maxl = 0;
584: break;
1.207 schwarze 585: case '(':
1.206 schwarze 586: cp++;
587: maxl = 2;
588: break;
1.207 schwarze 589: case '[':
1.206 schwarze 590: cp++;
591: term = ']';
592: maxl = 0;
593: break;
594: default:
595: maxl = 1;
596: break;
597: }
598: } else {
599: cp += 2;
1.94 kristaps 600: maxl = 0;
601: }
1.108 schwarze 602: stnam = cp;
1.94 kristaps 603:
1.108 schwarze 604: /* Advance to the end of the name. */
1.94 kristaps 605:
1.218 schwarze 606: arg_complete = 1;
1.181 schwarze 607: for (naml = 0; 0 == maxl || naml < maxl; naml++, cp++) {
1.153 kristaps 608: if ('\0' == *cp) {
1.219 schwarze 609: mandoc_msg(MANDOCERR_ESC_BAD, r->parse,
610: ln, (int)(stesc - *bufp), stesc);
1.218 schwarze 611: arg_complete = 0;
1.206 schwarze 612: break;
1.153 kristaps 613: }
1.206 schwarze 614: if (0 == maxl && *cp == term) {
615: cp++;
1.94 kristaps 616: break;
1.206 schwarze 617: }
1.94 kristaps 618: }
619:
1.108 schwarze 620: /*
621: * Retrieve the replacement string; if it is
622: * undefined, resume searching for escapes.
623: */
624:
1.206 schwarze 625: switch (stesc[1]) {
1.207 schwarze 626: case '*':
1.218 schwarze 627: if (arg_complete)
628: res = roff_getstrn(r, stnam, naml);
1.206 schwarze 629: break;
1.207 schwarze 630: case 'B':
1.206 schwarze 631: npos = 0;
1.218 schwarze 632: ubuf[0] = arg_complete &&
1.234 kristaps 633: roff_evalnum(r, ln, stnam, &npos, NULL, 0) &&
1.218 schwarze 634: stnam + npos + 1 == cp ? '1' : '0';
1.206 schwarze 635: ubuf[1] = '\0';
636: break;
1.207 schwarze 637: case 'n':
1.218 schwarze 638: if (arg_complete)
639: (void)snprintf(ubuf, sizeof(ubuf), "%d",
640: roff_getregn(r, stnam, naml));
641: else
642: ubuf[0] = '\0';
1.206 schwarze 643: break;
1.207 schwarze 644: case 'w':
1.218 schwarze 645: /* use even incomplete args */
1.208 schwarze 646: (void)snprintf(ubuf, sizeof(ubuf), "%d",
1.206 schwarze 647: 24 * (int)naml);
648: break;
649: }
1.94 kristaps 650:
651: if (NULL == res) {
1.219 schwarze 652: mandoc_vmsg(MANDOCERR_STR_UNDEF,
653: r->parse, ln, (int)(stesc - *bufp),
654: "%.*s", (int)naml, stnam);
1.142 kristaps 655: res = "";
1.94 kristaps 656: }
657:
1.108 schwarze 658: /* Replace the escape sequence by the string. */
659:
1.209 schwarze 660: *stesc = '\0';
661: *szp = mandoc_asprintf(&nbuf, "%s%s%s",
662: *bufp, res, cp) + 1;
1.94 kristaps 663:
1.205 schwarze 664: /* Prepare for the next replacement. */
1.94 kristaps 665:
1.205 schwarze 666: start = nbuf + pos;
1.209 schwarze 667: stesc = nbuf + (stesc - *bufp) + strlen(res);
1.94 kristaps 668: free(*bufp);
1.181 schwarze 669: *bufp = nbuf;
1.154 kristaps 670: }
1.172 schwarze 671: return(ROFF_CONT);
1.154 kristaps 672: }
673:
674: /*
1.178 schwarze 675: * Process text streams:
676: * Convert all breakable hyphens into ASCII_HYPH.
677: * Decrement and spring input line trap.
1.154 kristaps 678: */
679: static enum rofferr
1.178 schwarze 680: roff_parsetext(char **bufp, size_t *szp, int pos, int *offs)
1.154 kristaps 681: {
682: size_t sz;
683: const char *start;
1.178 schwarze 684: char *p;
685: int isz;
1.154 kristaps 686: enum mandoc_esc esc;
687:
1.178 schwarze 688: start = p = *bufp + pos;
1.154 kristaps 689:
690: while ('\0' != *p) {
691: sz = strcspn(p, "-\\");
692: p += sz;
693:
1.159 kristaps 694: if ('\0' == *p)
695: break;
696:
1.154 kristaps 697: if ('\\' == *p) {
698: /* Skip over escapes. */
699: p++;
1.189 schwarze 700: esc = mandoc_escape((const char **)&p, NULL, NULL);
1.154 kristaps 701: if (ESCAPE_ERROR == esc)
702: break;
1.155 kristaps 703: continue;
1.159 kristaps 704: } else if (p == start) {
1.158 kristaps 705: p++;
1.155 kristaps 706: continue;
1.158 kristaps 707: }
1.155 kristaps 708:
1.171 schwarze 709: if (isalpha((unsigned char)p[-1]) &&
710: isalpha((unsigned char)p[1]))
1.155 kristaps 711: *p = ASCII_HYPH;
712: p++;
1.94 kristaps 713: }
714:
1.178 schwarze 715: /* Spring the input line trap. */
716: if (1 == roffit_lines) {
1.202 schwarze 717: isz = mandoc_asprintf(&p, "%s\n.%s", *bufp, roffit_macro);
1.178 schwarze 718: free(*bufp);
719: *bufp = p;
720: *szp = isz + 1;
721: *offs = 0;
722: free(roffit_macro);
723: roffit_lines = 0;
724: return(ROFF_REPARSE);
725: } else if (1 < roffit_lines)
726: --roffit_lines;
1.154 kristaps 727: return(ROFF_CONT);
1.94 kristaps 728: }
729:
1.67 kristaps 730: enum rofferr
1.207 schwarze 731: roff_parseln(struct roff *r, int ln, char **bufp,
1.90 kristaps 732: size_t *szp, int pos, int *offs)
1.67 kristaps 733: {
734: enum rofft t;
1.109 kristaps 735: enum rofferr e;
1.130 kristaps 736: int ppos, ctl;
1.79 kristaps 737:
1.230 schwarze 738: /* Handle in-line equation delimiters. */
739:
1.236 schwarze 740: if (r->tbl == NULL &&
741: r->last_eqn != NULL && r->last_eqn->delim &&
1.230 schwarze 742: (r->eqn == NULL || r->eqn_inline)) {
743: e = roff_eqndelim(r, bufp, szp, pos);
744: if (e == ROFF_REPARSE)
745: return(e);
746: assert(e == ROFF_CONT);
747: }
748:
749: /* Expand some escape sequences. */
1.94 kristaps 750:
1.172 schwarze 751: e = roff_res(r, bufp, szp, ln, pos);
752: if (ROFF_IGN == e)
753: return(e);
754: assert(ROFF_CONT == e);
1.94 kristaps 755:
1.130 kristaps 756: ppos = pos;
1.174 kristaps 757: ctl = roff_getcontrol(r, *bufp, &pos);
1.130 kristaps 758:
1.94 kristaps 759: /*
1.79 kristaps 760: * First, if a scope is open and we're not a macro, pass the
761: * text through the macro's filter. If a scope isn't open and
762: * we're not a macro, just let it through.
1.125 kristaps 763: * Finally, if there's an equation scope open, divert it into it
764: * no matter our state.
1.79 kristaps 765: */
1.74 kristaps 766:
1.130 kristaps 767: if (r->last && ! ctl) {
1.78 kristaps 768: t = r->last->tok;
769: assert(roffs[t].text);
1.207 schwarze 770: e = (*roffs[t].text)(r, t, bufp, szp, ln, pos, pos, offs);
1.109 kristaps 771: assert(ROFF_IGN == e || ROFF_CONT == e);
1.125 kristaps 772: if (ROFF_CONT != e)
773: return(e);
1.182 schwarze 774: }
775: if (r->eqn)
776: return(eqn_read(&r->eqn, ln, *bufp, ppos, offs));
777: if ( ! ctl) {
1.125 kristaps 778: if (r->tbl)
1.130 kristaps 779: return(tbl_read(r->tbl, ln, *bufp, pos));
1.178 schwarze 780: return(roff_parsetext(bufp, szp, pos, offs));
1.182 schwarze 781: }
1.228 schwarze 782:
783: /* Skip empty request lines. */
784:
785: if ((*bufp)[pos] == '"') {
786: mandoc_msg(MANDOCERR_COMMENT_BAD, r->parse,
787: ln, pos, NULL);
788: return(ROFF_IGN);
789: } else if ((*bufp)[pos] == '\0')
790: return(ROFF_IGN);
1.67 kristaps 791:
1.79 kristaps 792: /*
793: * If a scope is open, go to the child handler for that macro,
794: * as it may want to preprocess before doing anything with it.
1.125 kristaps 795: * Don't do so if an equation is open.
1.79 kristaps 796: */
1.78 kristaps 797:
1.79 kristaps 798: if (r->last) {
799: t = r->last->tok;
800: assert(roffs[t].sub);
1.207 schwarze 801: return((*roffs[t].sub)(r, t, bufp, szp,
802: ln, ppos, pos, offs));
1.79 kristaps 803: }
1.78 kristaps 804:
1.79 kristaps 805: /*
806: * Lastly, as we've no scope open, try to look up and execute
807: * the new macro. If no macro is found, simply return and let
808: * the compilers handle it.
809: */
1.67 kristaps 810:
1.214 schwarze 811: if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos, ln, ppos)))
1.79 kristaps 812: return(ROFF_CONT);
1.67 kristaps 813:
1.75 kristaps 814: assert(roffs[t].proc);
1.207 schwarze 815: return((*roffs[t].proc)(r, t, bufp, szp, ln, ppos, pos, offs));
1.74 kristaps 816: }
817:
1.117 kristaps 818: void
1.74 kristaps 819: roff_endparse(struct roff *r)
820: {
821:
1.110 kristaps 822: if (r->last)
1.221 schwarze 823: mandoc_msg(MANDOCERR_BLK_NOEND, r->parse,
824: r->last->line, r->last->col,
825: roffs[r->last->tok].name);
1.117 kristaps 826:
1.125 kristaps 827: if (r->eqn) {
1.221 schwarze 828: mandoc_msg(MANDOCERR_BLK_NOEND, r->parse,
829: r->eqn->eqn.ln, r->eqn->eqn.pos, "EQ");
1.151 kristaps 830: eqn_end(&r->eqn);
1.125 kristaps 831: }
832:
1.117 kristaps 833: if (r->tbl) {
1.221 schwarze 834: mandoc_msg(MANDOCERR_BLK_NOEND, r->parse,
835: r->tbl->line, r->tbl->pos, "TS");
1.151 kristaps 836: tbl_end(&r->tbl);
1.117 kristaps 837: }
1.67 kristaps 838: }
839:
840: /*
841: * Parse a roff node's type from the input buffer. This must be in the
842: * form of ".foo xxx" in the usual way.
843: */
844: static enum rofft
1.214 schwarze 845: roff_parse(struct roff *r, char *buf, int *pos, int ln, int ppos)
1.67 kristaps 846: {
1.214 schwarze 847: char *cp;
1.106 kristaps 848: const char *mac;
849: size_t maclen;
1.67 kristaps 850: enum rofft t;
851:
1.214 schwarze 852: cp = buf + *pos;
853:
854: if ('\0' == *cp || '"' == *cp || '\t' == *cp || ' ' == *cp)
1.67 kristaps 855: return(ROFF_MAX);
856:
1.214 schwarze 857: mac = cp;
858: maclen = roff_getname(r, &cp, ln, ppos);
1.67 kristaps 859:
1.106 kristaps 860: t = (r->current_string = roff_getstrn(r, mac, maclen))
1.155 kristaps 861: ? ROFF_USERDEF : roffhash_find(mac, maclen);
1.67 kristaps 862:
1.214 schwarze 863: if (ROFF_MAX != t)
864: *pos = cp - buf;
1.67 kristaps 865:
866: return(t);
867: }
868:
869: static enum rofferr
1.76 kristaps 870: roff_cblock(ROFF_ARGS)
1.67 kristaps 871: {
872:
1.79 kristaps 873: /*
874: * A block-close `..' should only be invoked as a child of an
875: * ignore macro, otherwise raise a warning and just ignore it.
876: */
877:
1.76 kristaps 878: if (NULL == r->last) {
1.221 schwarze 879: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
880: ln, ppos, "..");
1.76 kristaps 881: return(ROFF_IGN);
882: }
1.67 kristaps 883:
1.81 kristaps 884: switch (r->last->tok) {
1.207 schwarze 885: case ROFF_am:
1.220 schwarze 886: /* ROFF_am1 is remapped to ROFF_am in roff_block(). */
1.81 kristaps 887: /* FALLTHROUGH */
1.207 schwarze 888: case ROFF_ami:
1.81 kristaps 889: /* FALLTHROUGH */
1.207 schwarze 890: case ROFF_de:
1.108 schwarze 891: /* ROFF_de1 is remapped to ROFF_de in roff_block(). */
1.81 kristaps 892: /* FALLTHROUGH */
1.207 schwarze 893: case ROFF_dei:
1.81 kristaps 894: /* FALLTHROUGH */
1.207 schwarze 895: case ROFF_ig:
1.81 kristaps 896: break;
897: default:
1.221 schwarze 898: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
899: ln, ppos, "..");
1.67 kristaps 900: return(ROFF_IGN);
1.76 kristaps 901: }
1.67 kristaps 902:
1.76 kristaps 903: if ((*bufp)[pos])
1.217 schwarze 904: mandoc_vmsg(MANDOCERR_ARG_SKIP, r->parse, ln, pos,
905: ".. %s", *bufp + pos);
1.71 kristaps 906:
907: roffnode_pop(r);
1.76 kristaps 908: roffnode_cleanscope(r);
909: return(ROFF_IGN);
1.71 kristaps 910:
1.67 kristaps 911: }
912:
1.76 kristaps 913: static void
914: roffnode_cleanscope(struct roff *r)
1.67 kristaps 915: {
916:
1.76 kristaps 917: while (r->last) {
1.173 schwarze 918: if (--r->last->endspan != 0)
1.76 kristaps 919: break;
920: roffnode_pop(r);
921: }
1.67 kristaps 922: }
923:
1.195 schwarze 924: static void
925: roff_ccond(struct roff *r, int ln, int ppos)
1.74 kristaps 926: {
927:
1.76 kristaps 928: if (NULL == r->last) {
1.221 schwarze 929: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
930: ln, ppos, "\\}");
1.195 schwarze 931: return;
1.76 kristaps 932: }
933:
1.82 kristaps 934: switch (r->last->tok) {
1.207 schwarze 935: case ROFF_el:
1.82 kristaps 936: /* FALLTHROUGH */
1.207 schwarze 937: case ROFF_ie:
1.82 kristaps 938: /* FALLTHROUGH */
1.207 schwarze 939: case ROFF_if:
1.82 kristaps 940: break;
941: default:
1.221 schwarze 942: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
943: ln, ppos, "\\}");
1.195 schwarze 944: return;
1.75 kristaps 945: }
946:
1.76 kristaps 947: if (r->last->endspan > -1) {
1.221 schwarze 948: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
949: ln, ppos, "\\}");
1.195 schwarze 950: return;
1.76 kristaps 951: }
952:
1.75 kristaps 953: roffnode_pop(r);
1.76 kristaps 954: roffnode_cleanscope(r);
1.195 schwarze 955: return;
1.76 kristaps 956: }
957:
958: static enum rofferr
1.80 kristaps 959: roff_block(ROFF_ARGS)
1.76 kristaps 960: {
1.220 schwarze 961: const char *name;
962: char *iname, *cp;
1.213 schwarze 963: size_t namesz;
1.106 kristaps 964:
1.220 schwarze 965: /* Ignore groff compatibility mode for now. */
1.76 kristaps 966:
1.220 schwarze 967: if (ROFF_de1 == tok)
968: tok = ROFF_de;
969: else if (ROFF_am1 == tok)
970: tok = ROFF_am;
971:
972: /* Parse the macro name argument. */
973:
974: cp = *bufp + pos;
975: if (ROFF_ig == tok) {
976: iname = NULL;
977: namesz = 0;
978: } else {
979: iname = cp;
980: namesz = roff_getname(r, &cp, ln, ppos);
981: iname[namesz] = '\0';
982: }
1.107 kristaps 983:
1.220 schwarze 984: /* Resolve the macro name argument if it is indirect. */
1.107 kristaps 985:
1.220 schwarze 986: if (namesz && (ROFF_dei == tok || ROFF_ami == tok)) {
987: if (NULL == (name = roff_getstrn(r, iname, namesz))) {
988: mandoc_vmsg(MANDOCERR_STR_UNDEF,
989: r->parse, ln, (int)(iname - *bufp),
990: "%.*s", (int)namesz, iname);
991: namesz = 0;
992: } else
993: namesz = strlen(name);
994: } else
995: name = iname;
1.107 kristaps 996:
1.220 schwarze 997: if (0 == namesz && ROFF_ig != tok) {
998: mandoc_msg(MANDOCERR_REQ_EMPTY, r->parse,
999: ln, ppos, roffs[tok].name);
1000: return(ROFF_IGN);
1001: }
1.80 kristaps 1002:
1.106 kristaps 1003: roffnode_push(r, tok, name, ln, ppos);
1004:
1005: /*
1006: * At the beginning of a `de' macro, clear the existing string
1007: * with the same name, if there is one. New content will be
1.193 schwarze 1008: * appended from roff_block_text() in multiline mode.
1.106 kristaps 1009: */
1.107 kristaps 1010:
1.220 schwarze 1011: if (ROFF_de == tok || ROFF_dei == tok)
1.213 schwarze 1012: roff_setstrn(&r->strtab, name, namesz, "", 0, 0);
1.76 kristaps 1013:
1.213 schwarze 1014: if ('\0' == *cp)
1.78 kristaps 1015: return(ROFF_IGN);
1016:
1.220 schwarze 1017: /* Get the custom end marker. */
1.107 kristaps 1018:
1.220 schwarze 1019: iname = cp;
1.213 schwarze 1020: namesz = roff_getname(r, &cp, ln, ppos);
1.220 schwarze 1021:
1022: /* Resolve the end marker if it is indirect. */
1023:
1024: if (namesz && (ROFF_dei == tok || ROFF_ami == tok)) {
1025: if (NULL == (name = roff_getstrn(r, iname, namesz))) {
1026: mandoc_vmsg(MANDOCERR_STR_UNDEF,
1027: r->parse, ln, (int)(iname - *bufp),
1028: "%.*s", (int)namesz, iname);
1029: namesz = 0;
1030: } else
1031: namesz = strlen(name);
1032: } else
1033: name = iname;
1034:
1.213 schwarze 1035: if (namesz)
1036: r->last->end = mandoc_strndup(name, namesz);
1.78 kristaps 1037:
1.213 schwarze 1038: if ('\0' != *cp)
1.217 schwarze 1039: mandoc_vmsg(MANDOCERR_ARG_EXCESS, r->parse,
1040: ln, pos, ".%s ... %s", roffs[tok].name, cp);
1.74 kristaps 1041:
1.78 kristaps 1042: return(ROFF_IGN);
1043: }
1044:
1045: static enum rofferr
1.80 kristaps 1046: roff_block_sub(ROFF_ARGS)
1.79 kristaps 1047: {
1048: enum rofft t;
1049: int i, j;
1050:
1051: /*
1052: * First check whether a custom macro exists at this level. If
1053: * it does, then check against it. This is some of groff's
1054: * stranger behaviours. If we encountered a custom end-scope
1055: * tag and that tag also happens to be a "real" macro, then we
1056: * need to try interpreting it again as a real macro. If it's
1057: * not, then return ignore. Else continue.
1058: */
1059:
1060: if (r->last->end) {
1.130 kristaps 1061: for (i = pos, j = 0; r->last->end[j]; j++, i++)
1.79 kristaps 1062: if ((*bufp)[i] != r->last->end[j])
1063: break;
1064:
1.207 schwarze 1065: if ('\0' == r->last->end[j] &&
1066: ('\0' == (*bufp)[i] ||
1067: ' ' == (*bufp)[i] ||
1068: '\t' == (*bufp)[i])) {
1.79 kristaps 1069: roffnode_pop(r);
1070: roffnode_cleanscope(r);
1071:
1.130 kristaps 1072: while (' ' == (*bufp)[i] || '\t' == (*bufp)[i])
1073: i++;
1074:
1075: pos = i;
1.214 schwarze 1076: if (ROFF_MAX != roff_parse(r, *bufp, &pos, ln, ppos))
1.79 kristaps 1077: return(ROFF_RERUN);
1078: return(ROFF_IGN);
1079: }
1080: }
1081:
1082: /*
1083: * If we have no custom end-query or lookup failed, then try
1084: * pulling it out of the hashtable.
1085: */
1086:
1.214 schwarze 1087: t = roff_parse(r, *bufp, &pos, ln, ppos);
1.79 kristaps 1088:
1.106 kristaps 1089: if (ROFF_cblock != t) {
1.220 schwarze 1090: if (ROFF_ig != tok)
1.193 schwarze 1091: roff_setstr(r, r->last->name, *bufp + ppos, 2);
1.79 kristaps 1092: return(ROFF_IGN);
1.106 kristaps 1093: }
1.79 kristaps 1094:
1095: assert(roffs[t].proc);
1.207 schwarze 1096: return((*roffs[t].proc)(r, t, bufp, szp, ln, ppos, pos, offs));
1.79 kristaps 1097: }
1098:
1099: static enum rofferr
1.80 kristaps 1100: roff_block_text(ROFF_ARGS)
1.78 kristaps 1101: {
1102:
1.220 schwarze 1103: if (ROFF_ig != tok)
1.193 schwarze 1104: roff_setstr(r, r->last->name, *bufp + pos, 2);
1.106 kristaps 1105:
1.78 kristaps 1106: return(ROFF_IGN);
1107: }
1108:
1109: static enum rofferr
1.82 kristaps 1110: roff_cond_sub(ROFF_ARGS)
1111: {
1112: enum rofft t;
1.139 kristaps 1113: char *ep;
1.198 schwarze 1114: int rr;
1.82 kristaps 1115:
1116: rr = r->last->rule;
1.139 kristaps 1117: roffnode_cleanscope(r);
1.214 schwarze 1118: t = roff_parse(r, *bufp, &pos, ln, ppos);
1.82 kristaps 1119:
1.139 kristaps 1120: /*
1.177 schwarze 1121: * Fully handle known macros when they are structurally
1122: * required or when the conditional evaluated to true.
1.87 kristaps 1123: */
1124:
1.177 schwarze 1125: if ((ROFF_MAX != t) &&
1.198 schwarze 1126: (rr || ROFFMAC_STRUCT & roffs[t].flags)) {
1.177 schwarze 1127: assert(roffs[t].proc);
1128: return((*roffs[t].proc)(r, t, bufp, szp,
1.207 schwarze 1129: ln, ppos, pos, offs));
1.177 schwarze 1130: }
1.144 kristaps 1131:
1.196 schwarze 1132: /*
1133: * If `\}' occurs on a macro line without a preceding macro,
1134: * drop the line completely.
1135: */
1136:
1137: ep = *bufp + pos;
1138: if ('\\' == ep[0] && '}' == ep[1])
1.198 schwarze 1139: rr = 0;
1.196 schwarze 1140:
1.177 schwarze 1141: /* Always check for the closing delimiter `\}'. */
1.144 kristaps 1142:
1.177 schwarze 1143: while (NULL != (ep = strchr(ep, '\\'))) {
1.197 schwarze 1144: if ('}' == *(++ep)) {
1145: *ep = '&';
1146: roff_ccond(r, ln, ep - *bufp - 1);
1147: }
1148: ++ep;
1.177 schwarze 1149: }
1.198 schwarze 1150: return(rr ? ROFF_CONT : ROFF_IGN);
1.82 kristaps 1151: }
1152:
1153: static enum rofferr
1154: roff_cond_text(ROFF_ARGS)
1.78 kristaps 1155: {
1.140 kristaps 1156: char *ep;
1.198 schwarze 1157: int rr;
1.82 kristaps 1158:
1159: rr = r->last->rule;
1.140 kristaps 1160: roffnode_cleanscope(r);
1.82 kristaps 1161:
1.197 schwarze 1162: ep = *bufp + pos;
1163: while (NULL != (ep = strchr(ep, '\\'))) {
1164: if ('}' == *(++ep)) {
1165: *ep = '&';
1166: roff_ccond(r, ln, ep - *bufp - 1);
1167: }
1168: ++ep;
1.78 kristaps 1169: }
1.198 schwarze 1170: return(rr ? ROFF_CONT : ROFF_IGN);
1.74 kristaps 1171: }
1172:
1.204 schwarze 1173: /*
1174: * Parse a single signed integer number. Stop at the first non-digit.
1175: * If there is at least one digit, return success and advance the
1176: * parse point, else return failure and let the parse point unchanged.
1177: * Ignore overflows, treat them just like the C language.
1178: */
1.184 schwarze 1179: static int
1180: roff_getnum(const char *v, int *pos, int *res)
1181: {
1.206 schwarze 1182: int myres, n, p;
1183:
1184: if (NULL == res)
1185: res = &myres;
1.184 schwarze 1186:
1187: p = *pos;
1188: n = v[p] == '-';
1189: if (n)
1190: p++;
1191:
1192: for (*res = 0; isdigit((unsigned char)v[p]); p++)
1.204 schwarze 1193: *res = 10 * *res + v[p] - '0';
1.184 schwarze 1194: if (p == *pos + n)
1195: return 0;
1196:
1197: if (n)
1198: *res = -*res;
1199:
1200: *pos = p;
1201: return 1;
1202: }
1203:
1.198 schwarze 1204: /*
1205: * Evaluate a string comparison condition.
1206: * The first character is the delimiter.
1207: * Succeed if the string up to its second occurrence
1208: * matches the string up to its third occurence.
1209: * Advance the cursor after the third occurrence
1210: * or lacking that, to the end of the line.
1211: */
1212: static int
1213: roff_evalstrcond(const char *v, int *pos)
1214: {
1215: const char *s1, *s2, *s3;
1216: int match;
1217:
1218: match = 0;
1219: s1 = v + *pos; /* initial delimiter */
1220: s2 = s1 + 1; /* for scanning the first string */
1221: s3 = strchr(s2, *s1); /* for scanning the second string */
1222:
1223: if (NULL == s3) /* found no middle delimiter */
1224: goto out;
1225:
1226: while ('\0' != *++s3) {
1227: if (*s2 != *s3) { /* mismatch */
1228: s3 = strchr(s3, *s1);
1229: break;
1230: }
1231: if (*s3 == *s1) { /* found the final delimiter */
1232: match = 1;
1233: break;
1234: }
1235: s2++;
1236: }
1237:
1238: out:
1239: if (NULL == s3)
1240: s3 = strchr(s2, '\0');
1241: else
1242: s3++;
1243: *pos = s3 - v;
1244: return(match);
1245: }
1246:
1.204 schwarze 1247: /*
1248: * Evaluate an optionally negated single character, numerical,
1249: * or string condition.
1250: */
1.198 schwarze 1251: static int
1.234 kristaps 1252: roff_evalcond(struct roff *r, int ln, const char *v, int *pos)
1.88 kristaps 1253: {
1.204 schwarze 1254: int wanttrue, number;
1.88 kristaps 1255:
1.198 schwarze 1256: if ('!' == v[*pos]) {
1257: wanttrue = 0;
1258: (*pos)++;
1259: } else
1260: wanttrue = 1;
1261:
1.88 kristaps 1262: switch (v[*pos]) {
1.207 schwarze 1263: case 'n':
1.198 schwarze 1264: /* FALLTHROUGH */
1.207 schwarze 1265: case 'o':
1.88 kristaps 1266: (*pos)++;
1.198 schwarze 1267: return(wanttrue);
1.207 schwarze 1268: case 'c':
1.198 schwarze 1269: /* FALLTHROUGH */
1.207 schwarze 1270: case 'd':
1.198 schwarze 1271: /* FALLTHROUGH */
1.207 schwarze 1272: case 'e':
1.88 kristaps 1273: /* FALLTHROUGH */
1.207 schwarze 1274: case 'r':
1.88 kristaps 1275: /* FALLTHROUGH */
1.207 schwarze 1276: case 't':
1.88 kristaps 1277: (*pos)++;
1.198 schwarze 1278: return(!wanttrue);
1.88 kristaps 1279: default:
1280: break;
1281: }
1282:
1.234 kristaps 1283: if (roff_evalnum(r, ln, v, pos, &number, 0))
1.204 schwarze 1284: return((number > 0) == wanttrue);
1285: else
1.198 schwarze 1286: return(roff_evalstrcond(v, pos) == wanttrue);
1.88 kristaps 1287: }
1288:
1.74 kristaps 1289: static enum rofferr
1.103 kristaps 1290: roff_line_ignore(ROFF_ARGS)
1.89 kristaps 1291: {
1.123 schwarze 1292:
1.89 kristaps 1293: return(ROFF_IGN);
1294: }
1295:
1.104 kristaps 1296: static enum rofferr
1.82 kristaps 1297: roff_cond(ROFF_ARGS)
1.74 kristaps 1298: {
1.173 schwarze 1299:
1300: roffnode_push(r, tok, NULL, ln, ppos);
1.74 kristaps 1301:
1.207 schwarze 1302: /*
1.134 kristaps 1303: * An `.el' has no conditional body: it will consume the value
1304: * of the current rstack entry set in prior `ie' calls or
1.207 schwarze 1305: * defaults to DENY.
1.134 kristaps 1306: *
1307: * If we're not an `el', however, then evaluate the conditional.
1308: */
1.133 kristaps 1309:
1.173 schwarze 1310: r->last->rule = ROFF_el == tok ?
1.207 schwarze 1311: (r->rstackpos < 0 ? 0 : r->rstack[r->rstackpos--]) :
1.234 kristaps 1312: roff_evalcond(r, ln, *bufp, &pos);
1.77 kristaps 1313:
1.134 kristaps 1314: /*
1315: * An if-else will put the NEGATION of the current evaluated
1316: * conditional into the stack of rules.
1317: */
1318:
1.84 schwarze 1319: if (ROFF_ie == tok) {
1.223 schwarze 1320: if (r->rstackpos + 1 == r->rstacksz) {
1321: r->rstacksz += 16;
1322: r->rstack = mandoc_reallocarray(r->rstack,
1323: r->rstacksz, sizeof(int));
1.134 kristaps 1324: }
1.198 schwarze 1325: r->rstack[++r->rstackpos] = !r->last->rule;
1.82 kristaps 1326: }
1.88 kristaps 1327:
1328: /* If the parent has false as its rule, then so do we. */
1329:
1.198 schwarze 1330: if (r->last->parent && !r->last->parent->rule)
1331: r->last->rule = 0;
1.88 kristaps 1332:
1333: /*
1.173 schwarze 1334: * Determine scope.
1335: * If there is nothing on the line after the conditional,
1336: * not even whitespace, use next-line scope.
1.88 kristaps 1337: */
1.74 kristaps 1338:
1.173 schwarze 1339: if ('\0' == (*bufp)[pos]) {
1340: r->last->endspan = 2;
1341: goto out;
1342: }
1343:
1344: while (' ' == (*bufp)[pos])
1345: pos++;
1346:
1347: /* An opening brace requests multiline scope. */
1.75 kristaps 1348:
1349: if ('\\' == (*bufp)[pos] && '{' == (*bufp)[pos + 1]) {
1350: r->last->endspan = -1;
1351: pos += 2;
1.173 schwarze 1352: goto out;
1.207 schwarze 1353: }
1.74 kristaps 1354:
1.77 kristaps 1355: /*
1.173 schwarze 1356: * Anything else following the conditional causes
1357: * single-line scope. Warn if the scope contains
1358: * nothing but trailing whitespace.
1.77 kristaps 1359: */
1360:
1.75 kristaps 1361: if ('\0' == (*bufp)[pos])
1.216 schwarze 1362: mandoc_msg(MANDOCERR_COND_EMPTY, r->parse,
1363: ln, ppos, roffs[tok].name);
1.77 kristaps 1364:
1.173 schwarze 1365: r->last->endspan = 1;
1.74 kristaps 1366:
1.173 schwarze 1367: out:
1.75 kristaps 1368: *offs = pos;
1369: return(ROFF_RERUN);
1.83 schwarze 1370: }
1371:
1372: static enum rofferr
1.92 schwarze 1373: roff_ds(ROFF_ARGS)
1374: {
1.212 schwarze 1375: char *string;
1376: const char *name;
1377: size_t namesz;
1.96 kristaps 1378:
1379: /*
1.212 schwarze 1380: * The first word is the name of the string.
1381: * If it is empty or terminated by an escape sequence,
1382: * abort the `ds' request without defining anything.
1.96 kristaps 1383: */
1.92 schwarze 1384:
1.212 schwarze 1385: name = string = *bufp + pos;
1.92 schwarze 1386: if ('\0' == *name)
1387: return(ROFF_IGN);
1388:
1.212 schwarze 1389: namesz = roff_getname(r, &string, ln, pos);
1390: if ('\\' == name[namesz])
1391: return(ROFF_IGN);
1392:
1393: /* Read past the initial double-quote, if any. */
1.121 schwarze 1394: if ('"' == *string)
1.92 schwarze 1395: string++;
1396:
1.96 kristaps 1397: /* The rest is the value. */
1.212 schwarze 1398: roff_setstrn(&r->strtab, name, namesz, string, strlen(string),
1399: ROFF_as == tok);
1.92 schwarze 1400: return(ROFF_IGN);
1401: }
1402:
1.204 schwarze 1403: /*
1404: * Parse a single operator, one or two characters long.
1405: * If the operator is recognized, return success and advance the
1406: * parse point, else return failure and let the parse point unchanged.
1407: */
1408: static int
1409: roff_getop(const char *v, int *pos, char *res)
1410: {
1411:
1412: *res = v[*pos];
1413:
1414: switch (*res) {
1.207 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: /* FALLTHROUGH */
1.207 schwarze 1421: case '/':
1.204 schwarze 1422: /* FALLTHROUGH */
1.207 schwarze 1423: case '%':
1.204 schwarze 1424: /* FALLTHROUGH */
1.207 schwarze 1425: case '&':
1.204 schwarze 1426: /* FALLTHROUGH */
1.207 schwarze 1427: case ':':
1.204 schwarze 1428: break;
1429: case '<':
1430: switch (v[*pos + 1]) {
1.207 schwarze 1431: case '=':
1.204 schwarze 1432: *res = 'l';
1433: (*pos)++;
1434: break;
1.207 schwarze 1435: case '>':
1.204 schwarze 1436: *res = '!';
1437: (*pos)++;
1438: break;
1.207 schwarze 1439: case '?':
1.204 schwarze 1440: *res = 'i';
1441: (*pos)++;
1442: break;
1443: default:
1444: break;
1445: }
1446: break;
1447: case '>':
1448: switch (v[*pos + 1]) {
1.207 schwarze 1449: case '=':
1.204 schwarze 1450: *res = 'g';
1451: (*pos)++;
1452: break;
1.207 schwarze 1453: case '?':
1.204 schwarze 1454: *res = 'a';
1455: (*pos)++;
1456: break;
1457: default:
1458: break;
1459: }
1460: break;
1461: case '=':
1462: if ('=' == v[*pos + 1])
1463: (*pos)++;
1464: break;
1465: default:
1466: return(0);
1467: }
1468: (*pos)++;
1469:
1470: return(*res);
1471: }
1472:
1473: /*
1474: * Evaluate either a parenthesized numeric expression
1475: * or a single signed integer number.
1476: */
1477: static int
1.235 schwarze 1478: roff_evalpar(struct roff *r, int ln,
1.234 kristaps 1479: const char *v, int *pos, int *res)
1.204 schwarze 1480: {
1481:
1482: if ('(' != v[*pos])
1483: return(roff_getnum(v, pos, res));
1484:
1485: (*pos)++;
1.234 kristaps 1486: if ( ! roff_evalnum(r, ln, v, pos, res, 1))
1.204 schwarze 1487: return(0);
1488:
1.206 schwarze 1489: /*
1490: * Omission of the closing parenthesis
1491: * is an error in validation mode,
1492: * but ignored in evaluation mode.
1493: */
1494:
1.204 schwarze 1495: if (')' == v[*pos])
1496: (*pos)++;
1.206 schwarze 1497: else if (NULL == res)
1498: return(0);
1.204 schwarze 1499:
1500: return(1);
1501: }
1502:
1503: /*
1504: * Evaluate a complete numeric expression.
1505: * Proceed left to right, there is no concept of precedence.
1506: */
1507: static int
1.235 schwarze 1508: roff_evalnum(struct roff *r, int ln, const char *v,
1.234 kristaps 1509: int *pos, int *res, int skipwhite)
1.204 schwarze 1510: {
1511: int mypos, operand2;
1512: char operator;
1513:
1514: if (NULL == pos) {
1515: mypos = 0;
1516: pos = &mypos;
1517: }
1518:
1519: if (skipwhite)
1520: while (isspace((unsigned char)v[*pos]))
1521: (*pos)++;
1522:
1.234 kristaps 1523: if ( ! roff_evalpar(r, ln, v, pos, res))
1.204 schwarze 1524: return(0);
1525:
1526: while (1) {
1527: if (skipwhite)
1528: while (isspace((unsigned char)v[*pos]))
1529: (*pos)++;
1530:
1531: if ( ! roff_getop(v, pos, &operator))
1532: break;
1533:
1534: if (skipwhite)
1535: while (isspace((unsigned char)v[*pos]))
1536: (*pos)++;
1537:
1.234 kristaps 1538: if ( ! roff_evalpar(r, ln, v, pos, &operand2))
1.204 schwarze 1539: return(0);
1540:
1541: if (skipwhite)
1542: while (isspace((unsigned char)v[*pos]))
1543: (*pos)++;
1.206 schwarze 1544:
1545: if (NULL == res)
1546: continue;
1.204 schwarze 1547:
1548: switch (operator) {
1.207 schwarze 1549: case '+':
1.204 schwarze 1550: *res += operand2;
1551: break;
1.207 schwarze 1552: case '-':
1.204 schwarze 1553: *res -= operand2;
1554: break;
1.207 schwarze 1555: case '*':
1.204 schwarze 1556: *res *= operand2;
1557: break;
1.207 schwarze 1558: case '/':
1.234 kristaps 1559: if (0 == operand2) {
1.235 schwarze 1560: mandoc_msg(MANDOCERR_DIVZERO,
1.234 kristaps 1561: r->parse, ln, *pos, v);
1562: *res = 0;
1563: break;
1564: }
1.204 schwarze 1565: *res /= operand2;
1566: break;
1.207 schwarze 1567: case '%':
1.204 schwarze 1568: *res %= operand2;
1569: break;
1.207 schwarze 1570: case '<':
1.204 schwarze 1571: *res = *res < operand2;
1572: break;
1.207 schwarze 1573: case '>':
1.204 schwarze 1574: *res = *res > operand2;
1575: break;
1.207 schwarze 1576: case 'l':
1.204 schwarze 1577: *res = *res <= operand2;
1578: break;
1.207 schwarze 1579: case 'g':
1.204 schwarze 1580: *res = *res >= operand2;
1581: break;
1.207 schwarze 1582: case '=':
1.204 schwarze 1583: *res = *res == operand2;
1584: break;
1.207 schwarze 1585: case '!':
1.204 schwarze 1586: *res = *res != operand2;
1587: break;
1.207 schwarze 1588: case '&':
1.204 schwarze 1589: *res = *res && operand2;
1590: break;
1.207 schwarze 1591: case ':':
1.204 schwarze 1592: *res = *res || operand2;
1593: break;
1.207 schwarze 1594: case 'i':
1.204 schwarze 1595: if (operand2 < *res)
1596: *res = operand2;
1597: break;
1.207 schwarze 1598: case 'a':
1.204 schwarze 1599: if (operand2 > *res)
1600: *res = operand2;
1601: break;
1602: default:
1603: abort();
1604: }
1605: }
1606: return(1);
1607: }
1608:
1.180 schwarze 1609: void
1.187 schwarze 1610: roff_setreg(struct roff *r, const char *name, int val, char sign)
1.147 kristaps 1611: {
1.180 schwarze 1612: struct roffreg *reg;
1613:
1614: /* Search for an existing register with the same name. */
1615: reg = r->regtab;
1616:
1617: while (reg && strcmp(name, reg->key.p))
1618: reg = reg->next;
1.147 kristaps 1619:
1.180 schwarze 1620: if (NULL == reg) {
1621: /* Create a new register. */
1622: reg = mandoc_malloc(sizeof(struct roffreg));
1623: reg->key.p = mandoc_strdup(name);
1624: reg->key.sz = strlen(name);
1.187 schwarze 1625: reg->val = 0;
1.180 schwarze 1626: reg->next = r->regtab;
1627: r->regtab = reg;
1628: }
1629:
1.187 schwarze 1630: if ('+' == sign)
1631: reg->val += val;
1632: else if ('-' == sign)
1633: reg->val -= val;
1634: else
1635: reg->val = val;
1.147 kristaps 1636: }
1637:
1.192 schwarze 1638: /*
1639: * Handle some predefined read-only number registers.
1640: * For now, return -1 if the requested register is not predefined;
1641: * in case a predefined read-only register having the value -1
1642: * were to turn up, another special value would have to be chosen.
1643: */
1644: static int
1645: roff_getregro(const char *name)
1646: {
1647:
1648: switch (*name) {
1.207 schwarze 1649: case 'A': /* ASCII approximation mode is always off. */
1.192 schwarze 1650: return(0);
1.207 schwarze 1651: case 'g': /* Groff compatibility mode is always on. */
1.192 schwarze 1652: return(1);
1.207 schwarze 1653: case 'H': /* Fixed horizontal resolution. */
1.192 schwarze 1654: return (24);
1.207 schwarze 1655: case 'j': /* Always adjust left margin only. */
1.192 schwarze 1656: return(0);
1.207 schwarze 1657: case 'T': /* Some output device is always defined. */
1.192 schwarze 1658: return(1);
1.207 schwarze 1659: case 'V': /* Fixed vertical resolution. */
1.192 schwarze 1660: return (40);
1661: default:
1662: return (-1);
1663: }
1664: }
1665:
1.181 schwarze 1666: int
1.180 schwarze 1667: roff_getreg(const struct roff *r, const char *name)
1.147 kristaps 1668: {
1.180 schwarze 1669: struct roffreg *reg;
1.192 schwarze 1670: int val;
1671:
1672: if ('.' == name[0] && '\0' != name[1] && '\0' == name[2]) {
1673: val = roff_getregro(name + 1);
1674: if (-1 != val)
1675: return (val);
1676: }
1.180 schwarze 1677:
1678: for (reg = r->regtab; reg; reg = reg->next)
1679: if (0 == strcmp(name, reg->key.p))
1.181 schwarze 1680: return(reg->val);
1681:
1682: return(0);
1683: }
1684:
1685: static int
1686: roff_getregn(const struct roff *r, const char *name, size_t len)
1687: {
1688: struct roffreg *reg;
1.192 schwarze 1689: int val;
1690:
1691: if ('.' == name[0] && 2 == len) {
1692: val = roff_getregro(name + 1);
1693: if (-1 != val)
1694: return (val);
1695: }
1.181 schwarze 1696:
1697: for (reg = r->regtab; reg; reg = reg->next)
1698: if (len == reg->key.sz &&
1699: 0 == strncmp(name, reg->key.p, len))
1700: return(reg->val);
1.147 kristaps 1701:
1.180 schwarze 1702: return(0);
1.147 kristaps 1703: }
1704:
1.180 schwarze 1705: static void
1706: roff_freereg(struct roffreg *reg)
1.147 kristaps 1707: {
1.180 schwarze 1708: struct roffreg *old_reg;
1.147 kristaps 1709:
1.180 schwarze 1710: while (NULL != reg) {
1711: free(reg->key.p);
1712: old_reg = reg;
1713: reg = reg->next;
1714: free(old_reg);
1715: }
1.147 kristaps 1716: }
1.92 schwarze 1717:
1718: static enum rofferr
1.89 kristaps 1719: roff_nr(ROFF_ARGS)
1.83 schwarze 1720: {
1.212 schwarze 1721: char *key, *val;
1722: size_t keysz;
1.138 kristaps 1723: int iv;
1.187 schwarze 1724: char sign;
1.89 kristaps 1725:
1.212 schwarze 1726: key = val = *bufp + pos;
1727: if ('\0' == *key)
1728: return(ROFF_IGN);
1729:
1730: keysz = roff_getname(r, &val, ln, pos);
1731: if ('\\' == key[keysz])
1732: return(ROFF_IGN);
1733: key[keysz] = '\0';
1.89 kristaps 1734:
1.187 schwarze 1735: sign = *val;
1736: if ('+' == sign || '-' == sign)
1737: val++;
1738:
1.234 kristaps 1739: if (roff_evalnum(r, ln, val, NULL, &iv, 0))
1.204 schwarze 1740: roff_setreg(r, key, iv, sign);
1.109 kristaps 1741:
1.203 schwarze 1742: return(ROFF_IGN);
1743: }
1744:
1745: static enum rofferr
1746: roff_rr(ROFF_ARGS)
1747: {
1748: struct roffreg *reg, **prev;
1.212 schwarze 1749: char *name, *cp;
1750: size_t namesz;
1.203 schwarze 1751:
1.212 schwarze 1752: name = cp = *bufp + pos;
1753: if ('\0' == *name)
1754: return(ROFF_IGN);
1755: namesz = roff_getname(r, &cp, ln, pos);
1756: name[namesz] = '\0';
1.203 schwarze 1757:
1758: prev = &r->regtab;
1759: while (1) {
1760: reg = *prev;
1761: if (NULL == reg || !strcmp(name, reg->key.p))
1762: break;
1763: prev = ®->next;
1764: }
1765: if (NULL != reg) {
1766: *prev = reg->next;
1767: free(reg->key.p);
1768: free(reg);
1769: }
1.122 schwarze 1770: return(ROFF_IGN);
1771: }
1772:
1773: static enum rofferr
1774: roff_rm(ROFF_ARGS)
1775: {
1776: const char *name;
1777: char *cp;
1.212 schwarze 1778: size_t namesz;
1.122 schwarze 1779:
1780: cp = *bufp + pos;
1781: while ('\0' != *cp) {
1.212 schwarze 1782: name = cp;
1783: namesz = roff_getname(r, &cp, ln, (int)(cp - *bufp));
1784: roff_setstrn(&r->strtab, name, namesz, NULL, 0, 0);
1785: if ('\\' == name[namesz])
1786: break;
1.122 schwarze 1787: }
1.178 schwarze 1788: return(ROFF_IGN);
1789: }
1790:
1791: static enum rofferr
1792: roff_it(ROFF_ARGS)
1793: {
1794: char *cp;
1795: size_t len;
1796: int iv;
1797:
1798: /* Parse the number of lines. */
1799: cp = *bufp + pos;
1800: len = strcspn(cp, " \t");
1801: cp[len] = '\0';
1802: if ((iv = mandoc_strntoi(cp, len, 10)) <= 0) {
1.222 schwarze 1803: mandoc_msg(MANDOCERR_IT_NONUM, r->parse,
1.207 schwarze 1804: ln, ppos, *bufp + 1);
1.178 schwarze 1805: return(ROFF_IGN);
1806: }
1807: cp += len + 1;
1808:
1809: /* Arm the input line trap. */
1810: roffit_lines = iv;
1811: roffit_macro = mandoc_strdup(cp);
1.109 kristaps 1812: return(ROFF_IGN);
1.175 schwarze 1813: }
1814:
1815: static enum rofferr
1816: roff_Dd(ROFF_ARGS)
1817: {
1818: const char *const *cp;
1819:
1.227 schwarze 1820: if ((r->options & (MPARSE_MDOC | MPARSE_QUICK)) == 0)
1.175 schwarze 1821: for (cp = __mdoc_reserved; *cp; cp++)
1822: roff_setstr(r, *cp, NULL, 0);
1823:
1.227 schwarze 1824: if (r->format == 0)
1825: r->format = MPARSE_MDOC;
1826:
1.175 schwarze 1827: return(ROFF_CONT);
1828: }
1829:
1830: static enum rofferr
1831: roff_TH(ROFF_ARGS)
1832: {
1833: const char *const *cp;
1834:
1.227 schwarze 1835: if ((r->options & MPARSE_QUICK) == 0)
1.175 schwarze 1836: for (cp = __man_reserved; *cp; cp++)
1837: roff_setstr(r, *cp, NULL, 0);
1838:
1.227 schwarze 1839: if (r->format == 0)
1840: r->format = MPARSE_MAN;
1841:
1.175 schwarze 1842: return(ROFF_CONT);
1.109 kristaps 1843: }
1844:
1845: static enum rofferr
1846: roff_TE(ROFF_ARGS)
1847: {
1848:
1849: if (NULL == r->tbl)
1.221 schwarze 1850: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
1851: ln, ppos, "TE");
1.115 kristaps 1852: else
1.151 kristaps 1853: tbl_end(&r->tbl);
1.109 kristaps 1854:
1.112 kristaps 1855: return(ROFF_IGN);
1856: }
1857:
1858: static enum rofferr
1859: roff_T_(ROFF_ARGS)
1860: {
1861:
1862: if (NULL == r->tbl)
1.221 schwarze 1863: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
1864: ln, ppos, "T&");
1.112 kristaps 1865: else
1.116 kristaps 1866: tbl_restart(ppos, ln, r->tbl);
1.112 kristaps 1867:
1.109 kristaps 1868: return(ROFF_IGN);
1869: }
1870:
1.230 schwarze 1871: /*
1872: * Handle in-line equation delimiters.
1873: */
1874: static enum rofferr
1875: roff_eqndelim(struct roff *r, char **bufp, size_t *szp, int pos)
1.151 kristaps 1876: {
1.233 schwarze 1877: char *cp1, *cp2;
1878: const char *bef_pr, *bef_nl, *mac, *aft_nl, *aft_pr;
1.151 kristaps 1879:
1.230 schwarze 1880: /*
1881: * Outside equations, look for an opening delimiter.
1882: * If we are inside an equation, we already know it is
1883: * in-line, or this function wouldn't have been called;
1884: * so look for a closing delimiter.
1885: */
1886:
1887: cp1 = *bufp + pos;
1888: cp2 = strchr(cp1, r->eqn == NULL ?
1889: r->last_eqn->odelim : r->last_eqn->cdelim);
1890: if (cp2 == NULL)
1891: return(ROFF_CONT);
1892:
1.233 schwarze 1893: *cp2++ = '\0';
1894: bef_pr = bef_nl = aft_nl = aft_pr = "";
1895:
1896: /* Handle preceding text, protecting whitespace. */
1897:
1898: if (**bufp != '\0') {
1899: if (r->eqn == NULL)
1900: bef_pr = "\\&";
1901: bef_nl = "\n";
1902: }
1903:
1904: /*
1905: * Prepare replacing the delimiter with an equation macro
1906: * and drop leading white space from the equation.
1907: */
1.230 schwarze 1908:
1.233 schwarze 1909: if (r->eqn == NULL) {
1910: while (*cp2 == ' ')
1911: cp2++;
1912: mac = ".EQ";
1913: } else
1914: mac = ".EN";
1915:
1916: /* Handle following text, protecting whitespace. */
1917:
1918: if (*cp2 != '\0') {
1919: aft_nl = "\n";
1920: if (r->eqn != NULL)
1921: aft_pr = "\\&";
1922: }
1923:
1924: /* Do the actual replacement. */
1925:
1926: *szp = mandoc_asprintf(&cp1, "%s%s%s%s%s%s%s", *bufp,
1927: bef_pr, bef_nl, mac, aft_nl, aft_pr, cp2) + 1;
1.230 schwarze 1928: free(*bufp);
1929: *bufp = cp1;
1930:
1931: /* Toggle the in-line state of the eqn subsystem. */
1932:
1933: r->eqn_inline = r->eqn == NULL;
1934: return(ROFF_REPARSE);
1.151 kristaps 1935: }
1936:
1.235 schwarze 1937: static enum rofferr
1938: roff_EQ(ROFF_ARGS)
1.125 kristaps 1939: {
1.151 kristaps 1940: struct eqn_node *e;
1.125 kristaps 1941:
1942: assert(NULL == r->eqn);
1.235 schwarze 1943: e = eqn_alloc(ppos, ln, r->parse);
1.125 kristaps 1944:
1.230 schwarze 1945: if (r->last_eqn) {
1.125 kristaps 1946: r->last_eqn->next = e;
1.230 schwarze 1947: e->delim = r->last_eqn->delim;
1948: e->odelim = r->last_eqn->odelim;
1949: e->cdelim = r->last_eqn->cdelim;
1950: } else
1.125 kristaps 1951: r->first_eqn = r->last_eqn = e;
1952:
1953: r->eqn = r->last_eqn = e;
1.151 kristaps 1954:
1.235 schwarze 1955: if ((*bufp)[pos])
1956: mandoc_vmsg(MANDOCERR_ARG_SKIP, r->parse, ln, pos,
1957: ".EQ %s", *bufp + pos);
1.151 kristaps 1958:
1.125 kristaps 1959: return(ROFF_IGN);
1960: }
1961:
1962: static enum rofferr
1963: roff_EN(ROFF_ARGS)
1964: {
1965:
1.221 schwarze 1966: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse, ln, ppos, "EN");
1.125 kristaps 1967: return(ROFF_IGN);
1968: }
1969:
1970: static enum rofferr
1.109 kristaps 1971: roff_TS(ROFF_ARGS)
1972: {
1.176 schwarze 1973: struct tbl_node *tbl;
1.89 kristaps 1974:
1.115 kristaps 1975: if (r->tbl) {
1.221 schwarze 1976: mandoc_msg(MANDOCERR_BLK_BROKEN, r->parse,
1977: ln, ppos, "TS breaks TS");
1.151 kristaps 1978: tbl_end(&r->tbl);
1.115 kristaps 1979: }
1.83 schwarze 1980:
1.176 schwarze 1981: tbl = tbl_alloc(ppos, ln, r->parse);
1.113 kristaps 1982:
1983: if (r->last_tbl)
1.176 schwarze 1984: r->last_tbl->next = tbl;
1.113 kristaps 1985: else
1.176 schwarze 1986: r->first_tbl = r->last_tbl = tbl;
1.113 kristaps 1987:
1.176 schwarze 1988: r->tbl = r->last_tbl = tbl;
1.83 schwarze 1989: return(ROFF_IGN);
1.92 schwarze 1990: }
1991:
1.105 kristaps 1992: static enum rofferr
1.174 kristaps 1993: roff_cc(ROFF_ARGS)
1994: {
1995: const char *p;
1996:
1997: p = *bufp + pos;
1998:
1999: if ('\0' == *p || '.' == (r->control = *p++))
2000: r->control = 0;
2001:
2002: if ('\0' != *p)
2003: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
2004:
2005: return(ROFF_IGN);
2006: }
2007:
2008: static enum rofferr
1.164 kristaps 2009: roff_tr(ROFF_ARGS)
2010: {
2011: const char *p, *first, *second;
2012: size_t fsz, ssz;
2013: enum mandoc_esc esc;
2014:
2015: p = *bufp + pos;
2016:
2017: if ('\0' == *p) {
2018: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
2019: return(ROFF_IGN);
2020: }
2021:
2022: while ('\0' != *p) {
2023: fsz = ssz = 1;
2024:
2025: first = p++;
2026: if ('\\' == *first) {
2027: esc = mandoc_escape(&p, NULL, NULL);
2028: if (ESCAPE_ERROR == esc) {
1.219 schwarze 2029: mandoc_msg(MANDOCERR_ESC_BAD, r->parse,
2030: ln, (int)(p - *bufp), first);
1.164 kristaps 2031: return(ROFF_IGN);
2032: }
2033: fsz = (size_t)(p - first);
2034: }
2035:
2036: second = p++;
2037: if ('\\' == *second) {
2038: esc = mandoc_escape(&p, NULL, NULL);
2039: if (ESCAPE_ERROR == esc) {
1.219 schwarze 2040: mandoc_msg(MANDOCERR_ESC_BAD, r->parse,
2041: ln, (int)(p - *bufp), second);
1.164 kristaps 2042: return(ROFF_IGN);
2043: }
2044: ssz = (size_t)(p - second);
1.165 kristaps 2045: } else if ('\0' == *second) {
1.207 schwarze 2046: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse,
2047: ln, (int)(p - *bufp), NULL);
1.164 kristaps 2048: second = " ";
1.165 kristaps 2049: p--;
1.164 kristaps 2050: }
2051:
1.167 kristaps 2052: if (fsz > 1) {
1.207 schwarze 2053: roff_setstrn(&r->xmbtab, first, fsz,
2054: second, ssz, 0);
1.167 kristaps 2055: continue;
2056: }
2057:
2058: if (NULL == r->xtab)
1.207 schwarze 2059: r->xtab = mandoc_calloc(128,
2060: sizeof(struct roffstr));
1.167 kristaps 2061:
2062: free(r->xtab[(int)*first].p);
2063: r->xtab[(int)*first].p = mandoc_strndup(second, ssz);
2064: r->xtab[(int)*first].sz = ssz;
1.164 kristaps 2065: }
2066:
2067: return(ROFF_IGN);
2068: }
2069:
2070: static enum rofferr
1.105 kristaps 2071: roff_so(ROFF_ARGS)
2072: {
2073: char *name;
2074:
1.210 schwarze 2075: name = *bufp + pos;
1.224 schwarze 2076: mandoc_vmsg(MANDOCERR_SO, r->parse, ln, ppos, "so %s", name);
1.105 kristaps 2077:
2078: /*
2079: * Handle `so'. Be EXTREMELY careful, as we shouldn't be
2080: * opening anything that's not in our cwd or anything beneath
2081: * it. Thus, explicitly disallow traversing up the file-system
2082: * or using absolute paths.
2083: */
2084:
2085: if ('/' == *name || strstr(name, "../") || strstr(name, "/..")) {
1.210 schwarze 2086: mandoc_vmsg(MANDOCERR_SO_PATH, r->parse, ln, ppos,
2087: ".so %s", name);
1.105 kristaps 2088: return(ROFF_ERR);
2089: }
2090:
2091: *offs = pos;
2092: return(ROFF_SO);
2093: }
1.92 schwarze 2094:
1.106 kristaps 2095: static enum rofferr
2096: roff_userdef(ROFF_ARGS)
1.99 kristaps 2097: {
1.106 kristaps 2098: const char *arg[9];
2099: char *cp, *n1, *n2;
1.119 schwarze 2100: int i;
1.106 kristaps 2101:
2102: /*
2103: * Collect pointers to macro argument strings
1.188 schwarze 2104: * and NUL-terminate them.
1.106 kristaps 2105: */
2106: cp = *bufp + pos;
1.119 schwarze 2107: for (i = 0; i < 9; i++)
1.120 schwarze 2108: arg[i] = '\0' == *cp ? "" :
1.136 kristaps 2109: mandoc_getarg(r->parse, &cp, ln, &pos);
1.99 kristaps 2110:
1.106 kristaps 2111: /*
2112: * Expand macro arguments.
1.99 kristaps 2113: */
1.106 kristaps 2114: *szp = 0;
2115: n1 = cp = mandoc_strdup(r->current_string);
2116: while (NULL != (cp = strstr(cp, "\\$"))) {
2117: i = cp[2] - '1';
2118: if (0 > i || 8 < i) {
2119: /* Not an argument invocation. */
2120: cp += 2;
2121: continue;
2122: }
1.209 schwarze 2123: *cp = '\0';
2124: *szp = mandoc_asprintf(&n2, "%s%s%s",
2125: n1, arg[i], cp + 3) + 1;
1.106 kristaps 2126: cp = n2 + (cp - n1);
2127: free(n1);
2128: n1 = n2;
1.99 kristaps 2129: }
2130:
1.106 kristaps 2131: /*
2132: * Replace the macro invocation
2133: * by the expanded macro.
2134: */
2135: free(*bufp);
2136: *bufp = n1;
2137: if (0 == *szp)
2138: *szp = strlen(*bufp) + 1;
2139:
2140: return(*szp > 1 && '\n' == (*bufp)[(int)*szp - 2] ?
2141: ROFF_REPARSE : ROFF_APPEND);
1.99 kristaps 2142: }
1.121 schwarze 2143:
1.212 schwarze 2144: static size_t
1.121 schwarze 2145: roff_getname(struct roff *r, char **cpp, int ln, int pos)
2146: {
2147: char *name, *cp;
1.212 schwarze 2148: size_t namesz;
1.121 schwarze 2149:
2150: name = *cpp;
2151: if ('\0' == *name)
1.212 schwarze 2152: return(0);
1.121 schwarze 2153:
1.212 schwarze 2154: /* Read until end of name and terminate it with NUL. */
2155: for (cp = name; 1; cp++) {
2156: if ('\0' == *cp || ' ' == *cp) {
2157: namesz = cp - name;
2158: break;
2159: }
1.121 schwarze 2160: if ('\\' != *cp)
2161: continue;
1.215 schwarze 2162: namesz = cp - name;
2163: if ('{' == cp[1] || '}' == cp[1])
2164: break;
1.121 schwarze 2165: cp++;
2166: if ('\\' == *cp)
2167: continue;
1.224 schwarze 2168: mandoc_vmsg(MANDOCERR_NAMESC, r->parse, ln, pos,
2169: "%.*s", (int)(cp - name + 1), name);
1.212 schwarze 2170: mandoc_escape((const char **)&cp, NULL, NULL);
2171: break;
1.121 schwarze 2172: }
2173:
2174: /* Read past spaces. */
2175: while (' ' == *cp)
2176: cp++;
2177:
2178: *cpp = cp;
1.212 schwarze 2179: return(namesz);
1.121 schwarze 2180: }
2181:
1.106 kristaps 2182: /*
2183: * Store *string into the user-defined string called *name.
2184: * To clear an existing entry, call with (*r, *name, NULL, 0).
1.193 schwarze 2185: * append == 0: replace mode
2186: * append == 1: single-line append mode
2187: * append == 2: multiline append mode, append '\n' after each call
1.106 kristaps 2188: */
1.94 kristaps 2189: static void
1.106 kristaps 2190: roff_setstr(struct roff *r, const char *name, const char *string,
1.193 schwarze 2191: int append)
1.92 schwarze 2192: {
1.164 kristaps 2193:
2194: roff_setstrn(&r->strtab, name, strlen(name), string,
1.207 schwarze 2195: string ? strlen(string) : 0, append);
1.164 kristaps 2196: }
2197:
2198: static void
1.166 kristaps 2199: roff_setstrn(struct roffkv **r, const char *name, size_t namesz,
1.193 schwarze 2200: const char *string, size_t stringsz, int append)
1.164 kristaps 2201: {
1.166 kristaps 2202: struct roffkv *n;
1.164 kristaps 2203: char *c;
2204: int i;
2205: size_t oldch, newch;
1.92 schwarze 2206:
1.106 kristaps 2207: /* Search for an existing string with the same name. */
1.164 kristaps 2208: n = *r;
2209:
1.211 schwarze 2210: while (n && (namesz != n->key.sz ||
2211: strncmp(n->key.p, name, namesz)))
1.92 schwarze 2212: n = n->next;
1.94 kristaps 2213:
2214: if (NULL == n) {
1.106 kristaps 2215: /* Create a new string table entry. */
1.166 kristaps 2216: n = mandoc_malloc(sizeof(struct roffkv));
2217: n->key.p = mandoc_strndup(name, namesz);
2218: n->key.sz = namesz;
2219: n->val.p = NULL;
2220: n->val.sz = 0;
1.164 kristaps 2221: n->next = *r;
2222: *r = n;
1.193 schwarze 2223: } else if (0 == append) {
1.166 kristaps 2224: free(n->val.p);
2225: n->val.p = NULL;
2226: n->val.sz = 0;
1.106 kristaps 2227: }
2228:
2229: if (NULL == string)
2230: return;
2231:
2232: /*
2233: * One additional byte for the '\n' in multiline mode,
2234: * and one for the terminating '\0'.
2235: */
1.193 schwarze 2236: newch = stringsz + (1 < append ? 2u : 1u);
1.164 kristaps 2237:
1.166 kristaps 2238: if (NULL == n->val.p) {
2239: n->val.p = mandoc_malloc(newch);
2240: *n->val.p = '\0';
1.106 kristaps 2241: oldch = 0;
2242: } else {
1.166 kristaps 2243: oldch = n->val.sz;
2244: n->val.p = mandoc_realloc(n->val.p, oldch + newch);
1.106 kristaps 2245: }
2246:
2247: /* Skip existing content in the destination buffer. */
1.166 kristaps 2248: c = n->val.p + (int)oldch;
1.106 kristaps 2249:
2250: /* Append new content to the destination buffer. */
1.164 kristaps 2251: i = 0;
2252: while (i < (int)stringsz) {
1.106 kristaps 2253: /*
2254: * Rudimentary roff copy mode:
2255: * Handle escaped backslashes.
2256: */
1.164 kristaps 2257: if ('\\' == string[i] && '\\' == string[i + 1])
2258: i++;
2259: *c++ = string[i++];
1.106 kristaps 2260: }
1.94 kristaps 2261:
1.106 kristaps 2262: /* Append terminating bytes. */
1.193 schwarze 2263: if (1 < append)
1.106 kristaps 2264: *c++ = '\n';
1.163 kristaps 2265:
1.106 kristaps 2266: *c = '\0';
1.166 kristaps 2267: n->val.sz = (int)(c - n->val.p);
1.92 schwarze 2268: }
2269:
1.94 kristaps 2270: static const char *
2271: roff_getstrn(const struct roff *r, const char *name, size_t len)
1.92 schwarze 2272: {
1.166 kristaps 2273: const struct roffkv *n;
1.191 schwarze 2274: int i;
1.92 schwarze 2275:
1.164 kristaps 2276: for (n = r->strtab; n; n = n->next)
1.207 schwarze 2277: if (0 == strncmp(name, n->key.p, len) &&
2278: '\0' == n->key.p[(int)len])
1.166 kristaps 2279: return(n->val.p);
1.191 schwarze 2280:
2281: for (i = 0; i < PREDEFS_MAX; i++)
2282: if (0 == strncmp(name, predefs[i].name, len) &&
2283: '\0' == predefs[i].name[(int)len])
2284: return(predefs[i].str);
1.94 kristaps 2285:
1.157 kristaps 2286: return(NULL);
1.92 schwarze 2287: }
2288:
1.94 kristaps 2289: static void
1.167 kristaps 2290: roff_freestr(struct roffkv *r)
1.92 schwarze 2291: {
1.166 kristaps 2292: struct roffkv *n, *nn;
1.92 schwarze 2293:
1.167 kristaps 2294: for (n = r; n; n = nn) {
1.166 kristaps 2295: free(n->key.p);
2296: free(n->val.p);
1.92 schwarze 2297: nn = n->next;
2298: free(n);
2299: }
1.114 kristaps 2300: }
2301:
2302: const struct tbl_span *
2303: roff_span(const struct roff *r)
2304: {
1.207 schwarze 2305:
1.114 kristaps 2306: return(r->tbl ? tbl_span(r->tbl) : NULL);
1.125 kristaps 2307: }
2308:
2309: const struct eqn *
2310: roff_eqn(const struct roff *r)
2311: {
1.207 schwarze 2312:
1.125 kristaps 2313: return(r->last_eqn ? &r->last_eqn->eqn : NULL);
1.164 kristaps 2314: }
2315:
2316: /*
2317: * Duplicate an input string, making the appropriate character
2318: * conversations (as stipulated by `tr') along the way.
2319: * Returns a heap-allocated string with all the replacements made.
2320: */
2321: char *
2322: roff_strdup(const struct roff *r, const char *p)
2323: {
1.166 kristaps 2324: const struct roffkv *cp;
1.164 kristaps 2325: char *res;
2326: const char *pp;
2327: size_t ssz, sz;
2328: enum mandoc_esc esc;
2329:
1.167 kristaps 2330: if (NULL == r->xmbtab && NULL == r->xtab)
1.164 kristaps 2331: return(mandoc_strdup(p));
2332: else if ('\0' == *p)
2333: return(mandoc_strdup(""));
2334:
2335: /*
2336: * Step through each character looking for term matches
2337: * (remember that a `tr' can be invoked with an escape, which is
2338: * a glyph but the escape is multi-character).
2339: * We only do this if the character hash has been initialised
2340: * and the string is >0 length.
2341: */
2342:
2343: res = NULL;
2344: ssz = 0;
2345:
2346: while ('\0' != *p) {
1.167 kristaps 2347: if ('\\' != *p && r->xtab && r->xtab[(int)*p].p) {
2348: sz = r->xtab[(int)*p].sz;
2349: res = mandoc_realloc(res, ssz + sz + 1);
2350: memcpy(res + ssz, r->xtab[(int)*p].p, sz);
2351: ssz += sz;
2352: p++;
2353: continue;
2354: } else if ('\\' != *p) {
2355: res = mandoc_realloc(res, ssz + 2);
2356: res[ssz++] = *p++;
2357: continue;
2358: }
2359:
1.164 kristaps 2360: /* Search for term matches. */
1.167 kristaps 2361: for (cp = r->xmbtab; cp; cp = cp->next)
1.166 kristaps 2362: if (0 == strncmp(p, cp->key.p, cp->key.sz))
1.164 kristaps 2363: break;
2364:
2365: if (NULL != cp) {
2366: /*
2367: * A match has been found.
2368: * Append the match to the array and move
2369: * forward by its keysize.
2370: */
1.207 schwarze 2371: res = mandoc_realloc(res,
2372: ssz + cp->val.sz + 1);
1.166 kristaps 2373: memcpy(res + ssz, cp->val.p, cp->val.sz);
2374: ssz += cp->val.sz;
2375: p += (int)cp->key.sz;
1.164 kristaps 2376: continue;
2377: }
2378:
1.167 kristaps 2379: /*
2380: * Handle escapes carefully: we need to copy
2381: * over just the escape itself, or else we might
2382: * do replacements within the escape itself.
2383: * Make sure to pass along the bogus string.
2384: */
2385: pp = p++;
2386: esc = mandoc_escape(&p, NULL, NULL);
2387: if (ESCAPE_ERROR == esc) {
2388: sz = strlen(pp);
1.164 kristaps 2389: res = mandoc_realloc(res, ssz + sz + 1);
2390: memcpy(res + ssz, pp, sz);
1.167 kristaps 2391: break;
1.164 kristaps 2392: }
1.207 schwarze 2393: /*
2394: * We bail out on bad escapes.
1.167 kristaps 2395: * No need to warn: we already did so when
2396: * roff_res() was called.
2397: */
2398: sz = (int)(p - pp);
2399: res = mandoc_realloc(res, ssz + sz + 1);
2400: memcpy(res + ssz, pp, sz);
2401: ssz += sz;
1.164 kristaps 2402: }
2403:
2404: res[(int)ssz] = '\0';
2405: return(res);
1.227 schwarze 2406: }
2407:
2408: int
2409: roff_getformat(const struct roff *r)
2410: {
2411:
2412: return(r->format);
1.174 kristaps 2413: }
2414:
2415: /*
1.207 schwarze 2416: * Find out whether a line is a macro line or not.
1.174 kristaps 2417: * If it is, adjust the current position and return one; if it isn't,
2418: * return zero and don't change the current position.
2419: * If the control character has been set with `.cc', then let that grain
2420: * precedence.
2421: * This is slighly contrary to groff, where using the non-breaking
2422: * control character when `cc' has been invoked will cause the
2423: * non-breaking macro contents to be printed verbatim.
2424: */
2425: int
2426: roff_getcontrol(const struct roff *r, const char *cp, int *ppos)
2427: {
2428: int pos;
2429:
2430: pos = *ppos;
2431:
2432: if (0 != r->control && cp[pos] == r->control)
2433: pos++;
2434: else if (0 != r->control)
2435: return(0);
2436: else if ('\\' == cp[pos] && '.' == cp[pos + 1])
2437: pos += 2;
2438: else if ('.' == cp[pos] || '\'' == cp[pos])
2439: pos++;
2440: else
2441: return(0);
2442:
2443: while (' ' == cp[pos] || '\t' == cp[pos])
2444: pos++;
2445:
2446: *ppos = pos;
2447: return(1);
1.74 kristaps 2448: }
CVSweb