Annotation of mandoc/roff.c, Revision 1.210
1.210 ! schwarze 1: /* $Id: roff.c,v 1.209 2014/04/23 16:08:33 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: #ifdef HAVE_CONFIG_H
19: #include "config.h"
20: #endif
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"
1.109 kristaps 30: #include "libroff.h"
1.94 kristaps 31: #include "libmandoc.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,
63: ROFF_ps,
1.83 schwarze 64: ROFF_rm,
1.203 schwarze 65: ROFF_rr,
1.105 kristaps 66: ROFF_so,
1.124 schwarze 67: ROFF_ta,
1.83 schwarze 68: ROFF_tr,
1.175 schwarze 69: ROFF_Dd,
70: ROFF_TH,
1.109 kristaps 71: ROFF_TS,
72: ROFF_TE,
1.112 kristaps 73: ROFF_T_,
1.125 kristaps 74: ROFF_EQ,
75: ROFF_EN,
1.76 kristaps 76: ROFF_cblock,
1.106 kristaps 77: ROFF_USERDEF,
1.67 kristaps 78: ROFF_MAX
79: };
80:
1.147 kristaps 81: /*
1.167 kristaps 82: * An incredibly-simple string buffer.
83: */
1.94 kristaps 84: struct roffstr {
1.167 kristaps 85: char *p; /* nil-terminated buffer */
86: size_t sz; /* saved strlen(p) */
1.166 kristaps 87: };
88:
89: /*
1.167 kristaps 90: * A key-value roffstr pair as part of a singly-linked list.
1.166 kristaps 91: */
92: struct roffkv {
93: struct roffstr key;
94: struct roffstr val;
95: struct roffkv *next; /* next in list */
1.94 kristaps 96: };
97:
1.180 schwarze 98: /*
99: * A single number register as part of a singly-linked list.
100: */
101: struct roffreg {
102: struct roffstr key;
1.181 schwarze 103: int val;
1.180 schwarze 104: struct roffreg *next;
105: };
106:
1.67 kristaps 107: struct roff {
1.128 kristaps 108: struct mparse *parse; /* parse point */
1.199 schwarze 109: int options; /* parse options */
1.67 kristaps 110: struct roffnode *last; /* leaf of stack */
1.198 schwarze 111: int rstack[RSTACK_MAX]; /* stack of !`ie' rules */
1.174 kristaps 112: char control; /* control character */
1.82 kristaps 113: int rstackpos; /* position in rstack */
1.180 schwarze 114: struct roffreg *regtab; /* number registers */
1.166 kristaps 115: struct roffkv *strtab; /* user-defined strings & macros */
1.167 kristaps 116: struct roffkv *xmbtab; /* multi-byte trans table (`tr') */
117: struct roffstr *xtab; /* single-byte trans table (`tr') */
1.106 kristaps 118: const char *current_string; /* value of last called user macro */
1.118 kristaps 119: struct tbl_node *first_tbl; /* first table parsed */
120: struct tbl_node *last_tbl; /* last table parsed */
121: struct tbl_node *tbl; /* current table being parsed */
1.125 kristaps 122: struct eqn_node *last_eqn; /* last equation parsed */
123: struct eqn_node *first_eqn; /* first equation parsed */
124: struct eqn_node *eqn; /* current equation being parsed */
1.79 kristaps 125: };
126:
1.67 kristaps 127: struct roffnode {
128: enum rofft tok; /* type of node */
129: struct roffnode *parent; /* up one in stack */
130: int line; /* parse line */
131: int col; /* parse col */
1.106 kristaps 132: char *name; /* node name, e.g. macro name */
1.79 kristaps 133: char *end; /* end-rules: custom token */
134: int endspan; /* end-rules: next-line or infty */
1.198 schwarze 135: int rule; /* current evaluation rule */
1.67 kristaps 136: };
137:
138: #define ROFF_ARGS struct roff *r, /* parse ctx */ \
1.72 kristaps 139: enum rofft tok, /* tok of macro */ \
1.207 schwarze 140: char **bufp, /* input buffer */ \
1.67 kristaps 141: size_t *szp, /* size of input buffer */ \
142: int ln, /* parse line */ \
1.75 kristaps 143: int ppos, /* original pos in buffer */ \
144: int pos, /* current pos in buffer */ \
1.74 kristaps 145: int *offs /* reset offset of buffer data */
1.67 kristaps 146:
147: typedef enum rofferr (*roffproc)(ROFF_ARGS);
148:
149: struct roffmac {
150: const char *name; /* macro name */
1.79 kristaps 151: roffproc proc; /* process new macro */
152: roffproc text; /* process as child text of macro */
153: roffproc sub; /* process as child of macro */
154: int flags;
155: #define ROFFMAC_STRUCT (1 << 0) /* always interpret */
1.85 kristaps 156: struct roffmac *next;
1.67 kristaps 157: };
158:
1.141 kristaps 159: struct predef {
160: const char *name; /* predefined input name */
161: const char *str; /* replacement symbol */
162: };
163:
164: #define PREDEF(__name, __str) \
165: { (__name), (__str) },
166:
1.155 kristaps 167: static enum rofft roffhash_find(const char *, size_t);
168: static void roffhash_init(void);
169: static void roffnode_cleanscope(struct roff *);
170: static void roffnode_pop(struct roff *);
171: static void roffnode_push(struct roff *, enum rofft,
172: const char *, int, int);
1.80 kristaps 173: static enum rofferr roff_block(ROFF_ARGS);
174: static enum rofferr roff_block_text(ROFF_ARGS);
175: static enum rofferr roff_block_sub(ROFF_ARGS);
176: static enum rofferr roff_cblock(ROFF_ARGS);
1.174 kristaps 177: static enum rofferr roff_cc(ROFF_ARGS);
1.195 schwarze 178: static void roff_ccond(struct roff *, int, int);
1.82 kristaps 179: static enum rofferr roff_cond(ROFF_ARGS);
180: static enum rofferr roff_cond_text(ROFF_ARGS);
181: static enum rofferr roff_cond_sub(ROFF_ARGS);
1.92 schwarze 182: static enum rofferr roff_ds(ROFF_ARGS);
1.198 schwarze 183: static int roff_evalcond(const char *, int *);
1.204 schwarze 184: static int roff_evalnum(const char *, int *, int *, int);
185: static int roff_evalpar(const char *, int *, int *);
1.198 schwarze 186: static int roff_evalstrcond(const char *, int *);
1.155 kristaps 187: static void roff_free1(struct roff *);
1.180 schwarze 188: static void roff_freereg(struct roffreg *);
1.167 kristaps 189: static void roff_freestr(struct roffkv *);
1.121 schwarze 190: static char *roff_getname(struct roff *, char **, int, int);
1.184 schwarze 191: static int roff_getnum(const char *, int *, int *);
192: static int roff_getop(const char *, int *, char *);
1.181 schwarze 193: static int roff_getregn(const struct roff *,
194: const char *, size_t);
1.192 schwarze 195: static int roff_getregro(const char *name);
1.207 schwarze 196: static const char *roff_getstrn(const struct roff *,
1.94 kristaps 197: const char *, size_t);
1.178 schwarze 198: static enum rofferr roff_it(ROFF_ARGS);
1.103 kristaps 199: static enum rofferr roff_line_ignore(ROFF_ARGS);
1.89 kristaps 200: static enum rofferr roff_nr(ROFF_ARGS);
1.169 schwarze 201: static void roff_openeqn(struct roff *, const char *,
1.156 kristaps 202: int, int, const char *);
1.155 kristaps 203: static enum rofft roff_parse(struct roff *, const char *, int *);
1.178 schwarze 204: static enum rofferr roff_parsetext(char **, size_t *, int, int *);
1.207 schwarze 205: static enum rofferr roff_res(struct roff *,
1.142 kristaps 206: char **, size_t *, int, int);
1.122 schwarze 207: static enum rofferr roff_rm(ROFF_ARGS);
1.203 schwarze 208: static enum rofferr roff_rr(ROFF_ARGS);
1.94 kristaps 209: static void roff_setstr(struct roff *,
1.106 kristaps 210: const char *, const char *, int);
1.207 schwarze 211: static void roff_setstrn(struct roffkv **, const char *,
1.164 kristaps 212: size_t, const char *, size_t, int);
1.105 kristaps 213: static enum rofferr roff_so(ROFF_ARGS);
1.164 kristaps 214: static enum rofferr roff_tr(ROFF_ARGS);
1.175 schwarze 215: static enum rofferr roff_Dd(ROFF_ARGS);
216: static enum rofferr roff_TH(ROFF_ARGS);
1.109 kristaps 217: static enum rofferr roff_TE(ROFF_ARGS);
218: static enum rofferr roff_TS(ROFF_ARGS);
1.125 kristaps 219: static enum rofferr roff_EQ(ROFF_ARGS);
220: static enum rofferr roff_EN(ROFF_ARGS);
1.112 kristaps 221: static enum rofferr roff_T_(ROFF_ARGS);
1.106 kristaps 222: static enum rofferr roff_userdef(ROFF_ARGS);
1.67 kristaps 223:
1.155 kristaps 224: /* See roffhash_find() */
1.85 kristaps 225:
226: #define ASCII_HI 126
227: #define ASCII_LO 33
228: #define HASHWIDTH (ASCII_HI - ASCII_LO + 1)
229:
230: static struct roffmac *hash[HASHWIDTH];
231:
232: static struct roffmac roffs[ROFF_MAX] = {
1.103 kristaps 233: { "ad", roff_line_ignore, NULL, NULL, 0, NULL },
1.85 kristaps 234: { "am", roff_block, roff_block_text, roff_block_sub, 0, NULL },
235: { "ami", roff_block, roff_block_text, roff_block_sub, 0, NULL },
236: { "am1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.193 schwarze 237: { "as", roff_ds, NULL, NULL, 0, NULL },
1.174 kristaps 238: { "cc", roff_cc, NULL, NULL, 0, NULL },
1.194 schwarze 239: { "ce", roff_line_ignore, NULL, NULL, 0, NULL },
1.85 kristaps 240: { "de", roff_block, roff_block_text, roff_block_sub, 0, NULL },
241: { "dei", roff_block, roff_block_text, roff_block_sub, 0, NULL },
242: { "de1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.92 schwarze 243: { "ds", roff_ds, NULL, NULL, 0, NULL },
1.85 kristaps 244: { "el", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
1.185 schwarze 245: { "fam", roff_line_ignore, NULL, NULL, 0, NULL },
1.186 schwarze 246: { "hw", roff_line_ignore, NULL, NULL, 0, NULL },
1.103 kristaps 247: { "hy", roff_line_ignore, NULL, NULL, 0, NULL },
1.85 kristaps 248: { "ie", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
249: { "if", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
250: { "ig", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.178 schwarze 251: { "it", roff_it, NULL, NULL, 0, NULL },
1.103 kristaps 252: { "ne", roff_line_ignore, NULL, NULL, 0, NULL },
253: { "nh", roff_line_ignore, NULL, NULL, 0, NULL },
1.104 kristaps 254: { "nr", roff_nr, NULL, NULL, 0, NULL },
1.124 schwarze 255: { "ns", roff_line_ignore, NULL, NULL, 0, NULL },
256: { "ps", roff_line_ignore, NULL, NULL, 0, NULL },
1.122 schwarze 257: { "rm", roff_rm, NULL, NULL, 0, NULL },
1.203 schwarze 258: { "rr", roff_rr, NULL, NULL, 0, NULL },
1.105 kristaps 259: { "so", roff_so, NULL, NULL, 0, NULL },
1.124 schwarze 260: { "ta", roff_line_ignore, NULL, NULL, 0, NULL },
1.164 kristaps 261: { "tr", roff_tr, NULL, NULL, 0, NULL },
1.175 schwarze 262: { "Dd", roff_Dd, NULL, NULL, 0, NULL },
263: { "TH", roff_TH, NULL, NULL, 0, NULL },
1.109 kristaps 264: { "TS", roff_TS, NULL, NULL, 0, NULL },
265: { "TE", roff_TE, NULL, NULL, 0, NULL },
1.112 kristaps 266: { "T&", roff_T_, NULL, NULL, 0, NULL },
1.125 kristaps 267: { "EQ", roff_EQ, NULL, NULL, 0, NULL },
268: { "EN", roff_EN, NULL, NULL, 0, NULL },
1.85 kristaps 269: { ".", roff_cblock, NULL, NULL, 0, NULL },
1.106 kristaps 270: { NULL, roff_userdef, NULL, NULL, 0, NULL },
1.67 kristaps 271: };
272:
1.200 schwarze 273: /* not currently implemented: Ds em Eq LP Me PP pp Or Rd Sf SH */
1.175 schwarze 274: const char *const __mdoc_reserved[] = {
275: "Ac", "Ad", "An", "Ao", "Ap", "Aq", "Ar", "At",
276: "Bc", "Bd", "Bf", "Bk", "Bl", "Bo", "Bq",
277: "Brc", "Bro", "Brq", "Bsx", "Bt", "Bx",
278: "Cd", "Cm", "Db", "Dc", "Dd", "Dl", "Do", "Dq",
1.200 schwarze 279: "Dt", "Dv", "Dx", "D1",
280: "Ec", "Ed", "Ef", "Ek", "El", "Em",
281: "En", "Eo", "Er", "Es", "Ev", "Ex",
1.175 schwarze 282: "Fa", "Fc", "Fd", "Fl", "Fn", "Fo", "Fr", "Ft", "Fx",
1.200 schwarze 283: "Hf", "Ic", "In", "It", "Lb", "Li", "Lk", "Lp",
284: "Ms", "Mt", "Nd", "Nm", "No", "Ns", "Nx",
1.175 schwarze 285: "Oc", "Oo", "Op", "Os", "Ot", "Ox",
1.200 schwarze 286: "Pa", "Pc", "Pf", "Po", "Pp", "Pq",
287: "Qc", "Ql", "Qo", "Qq", "Re", "Rs", "Rv",
288: "Sc", "Sh", "Sm", "So", "Sq",
1.175 schwarze 289: "Ss", "St", "Sx", "Sy",
290: "Ta", "Tn", "Ud", "Ux", "Va", "Vt", "Xc", "Xo", "Xr",
1.200 schwarze 291: "%A", "%B", "%C", "%D", "%I", "%J", "%N", "%O",
1.175 schwarze 292: "%P", "%Q", "%R", "%T", "%U", "%V",
293: NULL
294: };
295:
1.200 schwarze 296: /* not currently implemented: BT DE DS ME MT PT SY TQ YS */
1.175 schwarze 297: const char *const __man_reserved[] = {
1.200 schwarze 298: "AT", "B", "BI", "BR", "DT",
299: "EE", "EN", "EQ", "EX", "HP", "I", "IB", "IP", "IR",
300: "LP", "OP", "P", "PD", "PP",
301: "R", "RB", "RE", "RI", "RS", "SB", "SH", "SM", "SS",
302: "TE", "TH", "TP", "TS", "T&", "UC", "UE", "UR",
1.175 schwarze 303: NULL
304: };
305:
1.141 kristaps 306: /* Array of injected predefined strings. */
307: #define PREDEFS_MAX 38
308: static const struct predef predefs[PREDEFS_MAX] = {
309: #include "predefs.in"
310: };
311:
1.155 kristaps 312: /* See roffhash_find() */
1.85 kristaps 313: #define ROFF_HASH(p) (p[0] - ASCII_LO)
314:
1.178 schwarze 315: static int roffit_lines; /* number of lines to delay */
316: static char *roffit_macro; /* nil-terminated macro line */
317:
1.207 schwarze 318:
1.85 kristaps 319: static void
1.155 kristaps 320: roffhash_init(void)
1.85 kristaps 321: {
322: struct roffmac *n;
323: int buc, i;
324:
1.106 kristaps 325: for (i = 0; i < (int)ROFF_USERDEF; i++) {
1.85 kristaps 326: assert(roffs[i].name[0] >= ASCII_LO);
327: assert(roffs[i].name[0] <= ASCII_HI);
328:
329: buc = ROFF_HASH(roffs[i].name);
330:
331: if (NULL != (n = hash[buc])) {
332: for ( ; n->next; n = n->next)
333: /* Do nothing. */ ;
334: n->next = &roffs[i];
335: } else
336: hash[buc] = &roffs[i];
337: }
338: }
339:
1.67 kristaps 340: /*
341: * Look up a roff token by its name. Returns ROFF_MAX if no macro by
342: * the nil-terminated string name could be found.
343: */
344: static enum rofft
1.155 kristaps 345: roffhash_find(const char *p, size_t s)
1.67 kristaps 346: {
1.85 kristaps 347: int buc;
348: struct roffmac *n;
1.67 kristaps 349:
1.85 kristaps 350: /*
351: * libroff has an extremely simple hashtable, for the time
352: * being, which simply keys on the first character, which must
353: * be printable, then walks a chain. It works well enough until
354: * optimised.
355: */
356:
357: if (p[0] < ASCII_LO || p[0] > ASCII_HI)
358: return(ROFF_MAX);
359:
360: buc = ROFF_HASH(p);
361:
362: if (NULL == (n = hash[buc]))
363: return(ROFF_MAX);
364: for ( ; n; n = n->next)
1.106 kristaps 365: if (0 == strncmp(n->name, p, s) && '\0' == n->name[(int)s])
1.85 kristaps 366: return((enum rofft)(n - roffs));
1.67 kristaps 367:
368: return(ROFF_MAX);
369: }
370:
371: /*
372: * Pop the current node off of the stack of roff instructions currently
373: * pending.
374: */
375: static void
376: roffnode_pop(struct roff *r)
377: {
378: struct roffnode *p;
379:
1.75 kristaps 380: assert(r->last);
1.207 schwarze 381: p = r->last;
1.82 kristaps 382:
1.75 kristaps 383: r->last = r->last->parent;
1.106 kristaps 384: free(p->name);
385: free(p->end);
1.67 kristaps 386: free(p);
387: }
388:
389: /*
390: * Push a roff node onto the instruction stack. This must later be
391: * removed with roffnode_pop().
392: */
1.98 schwarze 393: static void
1.106 kristaps 394: roffnode_push(struct roff *r, enum rofft tok, const char *name,
395: int line, int col)
1.67 kristaps 396: {
397: struct roffnode *p;
398:
1.98 schwarze 399: p = mandoc_calloc(1, sizeof(struct roffnode));
1.67 kristaps 400: p->tok = tok;
1.106 kristaps 401: if (name)
402: p->name = mandoc_strdup(name);
1.67 kristaps 403: p->parent = r->last;
404: p->line = line;
405: p->col = col;
1.198 schwarze 406: p->rule = p->parent ? p->parent->rule : 0;
1.67 kristaps 407:
408: r->last = p;
409: }
410:
411: static void
412: roff_free1(struct roff *r)
413: {
1.176 schwarze 414: struct tbl_node *tbl;
1.125 kristaps 415: struct eqn_node *e;
1.167 kristaps 416: int i;
1.67 kristaps 417:
1.176 schwarze 418: while (NULL != (tbl = r->first_tbl)) {
419: r->first_tbl = tbl->next;
420: tbl_free(tbl);
1.109 kristaps 421: }
422:
1.113 kristaps 423: r->first_tbl = r->last_tbl = r->tbl = NULL;
424:
1.125 kristaps 425: while (NULL != (e = r->first_eqn)) {
426: r->first_eqn = e->next;
427: eqn_free(e);
428: }
429:
430: r->first_eqn = r->last_eqn = r->eqn = NULL;
431:
1.67 kristaps 432: while (r->last)
433: roffnode_pop(r);
1.109 kristaps 434:
1.167 kristaps 435: roff_freestr(r->strtab);
436: roff_freestr(r->xmbtab);
437:
438: r->strtab = r->xmbtab = NULL;
439:
1.180 schwarze 440: roff_freereg(r->regtab);
441:
442: r->regtab = NULL;
443:
1.167 kristaps 444: if (r->xtab)
445: for (i = 0; i < 128; i++)
446: free(r->xtab[i].p);
447:
448: free(r->xtab);
449: r->xtab = NULL;
1.67 kristaps 450: }
451:
452: void
453: roff_reset(struct roff *r)
454: {
455:
456: roff_free1(r);
1.174 kristaps 457: r->control = 0;
1.67 kristaps 458: }
459:
460: void
461: roff_free(struct roff *r)
462: {
463:
464: roff_free1(r);
465: free(r);
466: }
467:
468: struct roff *
1.199 schwarze 469: roff_alloc(struct mparse *parse, int options)
1.67 kristaps 470: {
471: struct roff *r;
472:
1.98 schwarze 473: r = mandoc_calloc(1, sizeof(struct roff));
1.128 kristaps 474: r->parse = parse;
1.199 schwarze 475: r->options = options;
1.82 kristaps 476: r->rstackpos = -1;
1.207 schwarze 477:
1.155 kristaps 478: roffhash_init();
1.141 kristaps 479:
1.67 kristaps 480: return(r);
481: }
482:
1.94 kristaps 483: /*
1.206 schwarze 484: * In the current line, expand escape sequences that tend to get
485: * used in numerical expressions and conditional requests.
486: * Also check the syntax of the remaining escape sequences.
1.154 kristaps 487: */
1.172 schwarze 488: static enum rofferr
1.142 kristaps 489: roff_res(struct roff *r, char **bufp, size_t *szp, int ln, int pos)
1.94 kristaps 490: {
1.208 schwarze 491: char ubuf[24]; /* buffer to print the number */
1.205 schwarze 492: const char *start; /* start of the string to process */
1.209 schwarze 493: char *stesc; /* start of an escape sequence ('\\') */
1.108 schwarze 494: const char *stnam; /* start of the name, after "[(*" */
495: const char *cp; /* end of the name, e.g. before ']' */
496: const char *res; /* the string to be substituted */
1.181 schwarze 497: char *nbuf; /* new buffer to copy bufp to */
498: size_t maxl; /* expected length of the escape name */
499: size_t naml; /* actual length of the escape name */
500: int expand_count; /* to avoid infinite loops */
1.206 schwarze 501: int npos; /* position in numeric expression */
502: int irc; /* return code from roff_evalnum() */
503: char term; /* character terminating the escape */
1.94 kristaps 504:
1.170 schwarze 505: expand_count = 0;
1.205 schwarze 506: start = *bufp + pos;
507: stesc = strchr(start, '\0') - 1;
508: while (stesc-- > start) {
1.170 schwarze 509:
1.205 schwarze 510: /* Search backwards for the next backslash. */
511:
512: if ('\\' != *stesc)
513: continue;
514:
515: /* If it is escaped, skip it. */
516:
517: for (cp = stesc - 1; cp >= start; cp--)
518: if ('\\' != *cp)
519: break;
520:
521: if (0 == (stesc - cp) % 2) {
1.209 schwarze 522: stesc = (char *)cp;
1.205 schwarze 523: continue;
524: }
1.108 schwarze 525:
1.206 schwarze 526: /* Decide whether to expand or to check only. */
1.108 schwarze 527:
1.206 schwarze 528: term = '\0';
1.205 schwarze 529: cp = stesc + 1;
1.181 schwarze 530: switch (*cp) {
1.207 schwarze 531: case '*':
1.181 schwarze 532: res = NULL;
533: break;
1.207 schwarze 534: case 'B':
1.206 schwarze 535: /* FALLTHROUGH */
1.207 schwarze 536: case 'w':
1.206 schwarze 537: term = cp[1];
538: /* FALLTHROUGH */
1.207 schwarze 539: case 'n':
1.181 schwarze 540: res = ubuf;
541: break;
542: default:
1.205 schwarze 543: if (ESCAPE_ERROR == mandoc_escape(&cp, NULL, NULL))
544: mandoc_msg(MANDOCERR_BADESCAPE, r->parse,
545: ln, (int)(stesc - *bufp), NULL);
546: continue;
1.152 kristaps 547: }
548:
1.205 schwarze 549: if (EXPAND_LIMIT < ++expand_count) {
550: mandoc_msg(MANDOCERR_ROFFLOOP, r->parse,
551: ln, (int)(stesc - *bufp), NULL);
552: return(ROFF_IGN);
553: }
1.108 schwarze 554:
555: /*
556: * The third character decides the length
1.181 schwarze 557: * of the name of the string or register.
1.108 schwarze 558: * Save a pointer to the name.
559: */
560:
1.206 schwarze 561: if ('\0' == term) {
562: switch (*++cp) {
1.207 schwarze 563: case '\0':
1.206 schwarze 564: maxl = 0;
565: break;
1.207 schwarze 566: case '(':
1.206 schwarze 567: cp++;
568: maxl = 2;
569: break;
1.207 schwarze 570: case '[':
1.206 schwarze 571: cp++;
572: term = ']';
573: maxl = 0;
574: break;
575: default:
576: maxl = 1;
577: break;
578: }
579: } else {
580: cp += 2;
1.94 kristaps 581: maxl = 0;
582: }
1.108 schwarze 583: stnam = cp;
1.94 kristaps 584:
1.108 schwarze 585: /* Advance to the end of the name. */
1.94 kristaps 586:
1.181 schwarze 587: for (naml = 0; 0 == maxl || naml < maxl; naml++, cp++) {
1.153 kristaps 588: if ('\0' == *cp) {
1.207 schwarze 589: mandoc_msg(MANDOCERR_BADESCAPE, r->parse,
590: ln, (int)(stesc - *bufp), NULL);
1.206 schwarze 591: break;
1.153 kristaps 592: }
1.206 schwarze 593: if (0 == maxl && *cp == term) {
594: cp++;
1.94 kristaps 595: break;
1.206 schwarze 596: }
1.94 kristaps 597: }
598:
1.108 schwarze 599: /*
600: * Retrieve the replacement string; if it is
601: * undefined, resume searching for escapes.
602: */
603:
1.206 schwarze 604: switch (stesc[1]) {
1.207 schwarze 605: case '*':
1.181 schwarze 606: res = roff_getstrn(r, stnam, naml);
1.206 schwarze 607: break;
1.207 schwarze 608: case 'B':
1.206 schwarze 609: npos = 0;
610: irc = roff_evalnum(stnam, &npos, NULL, 0);
611: ubuf[0] = irc && stnam + npos + 1 == cp
612: ? '1' : '0';
613: ubuf[1] = '\0';
614: break;
1.207 schwarze 615: case 'n':
1.208 schwarze 616: (void)snprintf(ubuf, sizeof(ubuf), "%d",
1.181 schwarze 617: roff_getregn(r, stnam, naml));
1.206 schwarze 618: break;
1.207 schwarze 619: case 'w':
1.208 schwarze 620: (void)snprintf(ubuf, sizeof(ubuf), "%d",
1.206 schwarze 621: 24 * (int)naml);
622: break;
623: }
1.94 kristaps 624:
625: if (NULL == res) {
1.207 schwarze 626: mandoc_msg(MANDOCERR_BADESCAPE, r->parse,
627: ln, (int)(stesc - *bufp), NULL);
1.142 kristaps 628: res = "";
1.94 kristaps 629: }
630:
1.108 schwarze 631: /* Replace the escape sequence by the string. */
632:
1.209 schwarze 633: *stesc = '\0';
634: *szp = mandoc_asprintf(&nbuf, "%s%s%s",
635: *bufp, res, cp) + 1;
1.94 kristaps 636:
1.205 schwarze 637: /* Prepare for the next replacement. */
1.94 kristaps 638:
1.205 schwarze 639: start = nbuf + pos;
1.209 schwarze 640: stesc = nbuf + (stesc - *bufp) + strlen(res);
1.94 kristaps 641: free(*bufp);
1.181 schwarze 642: *bufp = nbuf;
1.154 kristaps 643: }
1.172 schwarze 644: return(ROFF_CONT);
1.154 kristaps 645: }
646:
647: /*
1.178 schwarze 648: * Process text streams:
649: * Convert all breakable hyphens into ASCII_HYPH.
650: * Decrement and spring input line trap.
1.154 kristaps 651: */
652: static enum rofferr
1.178 schwarze 653: roff_parsetext(char **bufp, size_t *szp, int pos, int *offs)
1.154 kristaps 654: {
655: size_t sz;
656: const char *start;
1.178 schwarze 657: char *p;
658: int isz;
1.154 kristaps 659: enum mandoc_esc esc;
660:
1.178 schwarze 661: start = p = *bufp + pos;
1.154 kristaps 662:
663: while ('\0' != *p) {
664: sz = strcspn(p, "-\\");
665: p += sz;
666:
1.159 kristaps 667: if ('\0' == *p)
668: break;
669:
1.154 kristaps 670: if ('\\' == *p) {
671: /* Skip over escapes. */
672: p++;
1.189 schwarze 673: esc = mandoc_escape((const char **)&p, NULL, NULL);
1.154 kristaps 674: if (ESCAPE_ERROR == esc)
675: break;
1.155 kristaps 676: continue;
1.159 kristaps 677: } else if (p == start) {
1.158 kristaps 678: p++;
1.155 kristaps 679: continue;
1.158 kristaps 680: }
1.155 kristaps 681:
1.171 schwarze 682: if (isalpha((unsigned char)p[-1]) &&
683: isalpha((unsigned char)p[1]))
1.155 kristaps 684: *p = ASCII_HYPH;
685: p++;
1.94 kristaps 686: }
687:
1.178 schwarze 688: /* Spring the input line trap. */
689: if (1 == roffit_lines) {
1.202 schwarze 690: isz = mandoc_asprintf(&p, "%s\n.%s", *bufp, roffit_macro);
1.178 schwarze 691: free(*bufp);
692: *bufp = p;
693: *szp = isz + 1;
694: *offs = 0;
695: free(roffit_macro);
696: roffit_lines = 0;
697: return(ROFF_REPARSE);
698: } else if (1 < roffit_lines)
699: --roffit_lines;
1.154 kristaps 700: return(ROFF_CONT);
1.94 kristaps 701: }
702:
1.67 kristaps 703: enum rofferr
1.207 schwarze 704: roff_parseln(struct roff *r, int ln, char **bufp,
1.90 kristaps 705: size_t *szp, int pos, int *offs)
1.67 kristaps 706: {
707: enum rofft t;
1.109 kristaps 708: enum rofferr e;
1.130 kristaps 709: int ppos, ctl;
1.79 kristaps 710:
711: /*
1.94 kristaps 712: * Run the reserved-word filter only if we have some reserved
713: * words to fill in.
714: */
715:
1.172 schwarze 716: e = roff_res(r, bufp, szp, ln, pos);
717: if (ROFF_IGN == e)
718: return(e);
719: assert(ROFF_CONT == e);
1.94 kristaps 720:
1.130 kristaps 721: ppos = pos;
1.174 kristaps 722: ctl = roff_getcontrol(r, *bufp, &pos);
1.130 kristaps 723:
1.94 kristaps 724: /*
1.79 kristaps 725: * First, if a scope is open and we're not a macro, pass the
726: * text through the macro's filter. If a scope isn't open and
727: * we're not a macro, just let it through.
1.125 kristaps 728: * Finally, if there's an equation scope open, divert it into it
729: * no matter our state.
1.79 kristaps 730: */
1.74 kristaps 731:
1.130 kristaps 732: if (r->last && ! ctl) {
1.78 kristaps 733: t = r->last->tok;
734: assert(roffs[t].text);
1.207 schwarze 735: e = (*roffs[t].text)(r, t, bufp, szp, ln, pos, pos, offs);
1.109 kristaps 736: assert(ROFF_IGN == e || ROFF_CONT == e);
1.125 kristaps 737: if (ROFF_CONT != e)
738: return(e);
1.182 schwarze 739: }
740: if (r->eqn)
741: return(eqn_read(&r->eqn, ln, *bufp, ppos, offs));
742: if ( ! ctl) {
1.125 kristaps 743: if (r->tbl)
1.130 kristaps 744: return(tbl_read(r->tbl, ln, *bufp, pos));
1.178 schwarze 745: return(roff_parsetext(bufp, szp, pos, offs));
1.182 schwarze 746: }
1.67 kristaps 747:
1.79 kristaps 748: /*
749: * If a scope is open, go to the child handler for that macro,
750: * as it may want to preprocess before doing anything with it.
1.125 kristaps 751: * Don't do so if an equation is open.
1.79 kristaps 752: */
1.78 kristaps 753:
1.79 kristaps 754: if (r->last) {
755: t = r->last->tok;
756: assert(roffs[t].sub);
1.207 schwarze 757: return((*roffs[t].sub)(r, t, bufp, szp,
758: ln, ppos, pos, offs));
1.79 kristaps 759: }
1.78 kristaps 760:
1.79 kristaps 761: /*
762: * Lastly, as we've no scope open, try to look up and execute
763: * the new macro. If no macro is found, simply return and let
764: * the compilers handle it.
765: */
1.67 kristaps 766:
1.106 kristaps 767: if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos)))
1.79 kristaps 768: return(ROFF_CONT);
1.67 kristaps 769:
1.75 kristaps 770: assert(roffs[t].proc);
1.207 schwarze 771: return((*roffs[t].proc)(r, t, bufp, szp, ln, ppos, pos, offs));
1.74 kristaps 772: }
773:
1.117 kristaps 774: void
1.74 kristaps 775: roff_endparse(struct roff *r)
776: {
777:
1.110 kristaps 778: if (r->last)
1.128 kristaps 779: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.207 schwarze 780: r->last->line, r->last->col, NULL);
1.117 kristaps 781:
1.125 kristaps 782: if (r->eqn) {
1.207 schwarze 783: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
784: r->eqn->eqn.ln, r->eqn->eqn.pos, NULL);
1.151 kristaps 785: eqn_end(&r->eqn);
1.125 kristaps 786: }
787:
1.117 kristaps 788: if (r->tbl) {
1.207 schwarze 789: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
790: r->tbl->line, r->tbl->pos, NULL);
1.151 kristaps 791: tbl_end(&r->tbl);
1.117 kristaps 792: }
1.67 kristaps 793: }
794:
795: /*
796: * Parse a roff node's type from the input buffer. This must be in the
797: * form of ".foo xxx" in the usual way.
798: */
799: static enum rofft
1.106 kristaps 800: roff_parse(struct roff *r, const char *buf, int *pos)
1.67 kristaps 801: {
1.106 kristaps 802: const char *mac;
803: size_t maclen;
1.67 kristaps 804: enum rofft t;
805:
1.207 schwarze 806: if ('\0' == buf[*pos] || '"' == buf[*pos] ||
807: '\t' == buf[*pos] || ' ' == buf[*pos])
1.67 kristaps 808: return(ROFF_MAX);
809:
1.195 schwarze 810: /* We stop the macro parse at an escape, tab, space, or nil. */
1.144 kristaps 811:
1.106 kristaps 812: mac = buf + *pos;
1.195 schwarze 813: maclen = strcspn(mac, " \\\t\0");
1.67 kristaps 814:
1.106 kristaps 815: t = (r->current_string = roff_getstrn(r, mac, maclen))
1.155 kristaps 816: ? ROFF_USERDEF : roffhash_find(mac, maclen);
1.67 kristaps 817:
1.127 kristaps 818: *pos += (int)maclen;
1.130 kristaps 819:
1.67 kristaps 820: while (buf[*pos] && ' ' == buf[*pos])
821: (*pos)++;
822:
823: return(t);
824: }
825:
826: static enum rofferr
1.76 kristaps 827: roff_cblock(ROFF_ARGS)
1.67 kristaps 828: {
829:
1.79 kristaps 830: /*
831: * A block-close `..' should only be invoked as a child of an
832: * ignore macro, otherwise raise a warning and just ignore it.
833: */
834:
1.76 kristaps 835: if (NULL == r->last) {
1.128 kristaps 836: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.76 kristaps 837: return(ROFF_IGN);
838: }
1.67 kristaps 839:
1.81 kristaps 840: switch (r->last->tok) {
1.207 schwarze 841: case ROFF_am:
1.81 kristaps 842: /* FALLTHROUGH */
1.207 schwarze 843: case ROFF_ami:
1.81 kristaps 844: /* FALLTHROUGH */
1.207 schwarze 845: case ROFF_am1:
1.81 kristaps 846: /* FALLTHROUGH */
1.207 schwarze 847: case ROFF_de:
1.108 schwarze 848: /* ROFF_de1 is remapped to ROFF_de in roff_block(). */
1.81 kristaps 849: /* FALLTHROUGH */
1.207 schwarze 850: case ROFF_dei:
1.81 kristaps 851: /* FALLTHROUGH */
1.207 schwarze 852: case ROFF_ig:
1.81 kristaps 853: break;
854: default:
1.128 kristaps 855: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.67 kristaps 856: return(ROFF_IGN);
1.76 kristaps 857: }
1.67 kristaps 858:
1.76 kristaps 859: if ((*bufp)[pos])
1.128 kristaps 860: mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.71 kristaps 861:
862: roffnode_pop(r);
1.76 kristaps 863: roffnode_cleanscope(r);
864: return(ROFF_IGN);
1.71 kristaps 865:
1.67 kristaps 866: }
867:
1.76 kristaps 868: static void
869: roffnode_cleanscope(struct roff *r)
1.67 kristaps 870: {
871:
1.76 kristaps 872: while (r->last) {
1.173 schwarze 873: if (--r->last->endspan != 0)
1.76 kristaps 874: break;
875: roffnode_pop(r);
876: }
1.67 kristaps 877: }
878:
1.195 schwarze 879: static void
880: roff_ccond(struct roff *r, int ln, int ppos)
1.74 kristaps 881: {
882:
1.76 kristaps 883: if (NULL == r->last) {
1.128 kristaps 884: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.195 schwarze 885: return;
1.76 kristaps 886: }
887:
1.82 kristaps 888: switch (r->last->tok) {
1.207 schwarze 889: case ROFF_el:
1.82 kristaps 890: /* FALLTHROUGH */
1.207 schwarze 891: case ROFF_ie:
1.82 kristaps 892: /* FALLTHROUGH */
1.207 schwarze 893: case ROFF_if:
1.82 kristaps 894: break;
895: default:
1.128 kristaps 896: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.195 schwarze 897: return;
1.75 kristaps 898: }
899:
1.76 kristaps 900: if (r->last->endspan > -1) {
1.128 kristaps 901: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.195 schwarze 902: return;
1.76 kristaps 903: }
904:
1.75 kristaps 905: roffnode_pop(r);
1.76 kristaps 906: roffnode_cleanscope(r);
1.195 schwarze 907: return;
1.76 kristaps 908: }
909:
910: static enum rofferr
1.80 kristaps 911: roff_block(ROFF_ARGS)
1.76 kristaps 912: {
1.78 kristaps 913: int sv;
914: size_t sz;
1.106 kristaps 915: char *name;
916:
917: name = NULL;
1.76 kristaps 918:
1.106 kristaps 919: if (ROFF_ig != tok) {
920: if ('\0' == (*bufp)[pos]) {
1.128 kristaps 921: mandoc_msg(MANDOCERR_NOARGS, r->parse, ln, ppos, NULL);
1.106 kristaps 922: return(ROFF_IGN);
923: }
1.107 kristaps 924:
925: /*
926: * Re-write `de1', since we don't really care about
927: * groff's strange compatibility mode, into `de'.
928: */
929:
1.106 kristaps 930: if (ROFF_de1 == tok)
931: tok = ROFF_de;
932: if (ROFF_de == tok)
933: name = *bufp + pos;
934: else
1.128 kristaps 935: mandoc_msg(MANDOCERR_REQUEST, r->parse, ln, ppos,
1.106 kristaps 936: roffs[tok].name);
1.107 kristaps 937:
1.131 schwarze 938: while ((*bufp)[pos] && ! isspace((unsigned char)(*bufp)[pos]))
1.80 kristaps 939: pos++;
1.107 kristaps 940:
1.131 schwarze 941: while (isspace((unsigned char)(*bufp)[pos]))
1.106 kristaps 942: (*bufp)[pos++] = '\0';
1.80 kristaps 943: }
944:
1.106 kristaps 945: roffnode_push(r, tok, name, ln, ppos);
946:
947: /*
948: * At the beginning of a `de' macro, clear the existing string
949: * with the same name, if there is one. New content will be
1.193 schwarze 950: * appended from roff_block_text() in multiline mode.
1.106 kristaps 951: */
1.107 kristaps 952:
1.106 kristaps 953: if (ROFF_de == tok)
1.108 schwarze 954: roff_setstr(r, name, "", 0);
1.76 kristaps 955:
1.79 kristaps 956: if ('\0' == (*bufp)[pos])
1.78 kristaps 957: return(ROFF_IGN);
958:
1.107 kristaps 959: /* If present, process the custom end-of-line marker. */
960:
1.78 kristaps 961: sv = pos;
1.131 schwarze 962: while ((*bufp)[pos] && ! isspace((unsigned char)(*bufp)[pos]))
1.78 kristaps 963: pos++;
964:
965: /*
966: * Note: groff does NOT like escape characters in the input.
967: * Instead of detecting this, we're just going to let it fly and
968: * to hell with it.
969: */
970:
971: assert(pos > sv);
972: sz = (size_t)(pos - sv);
973:
1.79 kristaps 974: if (1 == sz && '.' == (*bufp)[sv])
975: return(ROFF_IGN);
976:
1.98 schwarze 977: r->last->end = mandoc_malloc(sz + 1);
1.78 kristaps 978:
979: memcpy(r->last->end, *bufp + sv, sz);
980: r->last->end[(int)sz] = '\0';
981:
1.77 kristaps 982: if ((*bufp)[pos])
1.128 kristaps 983: mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.74 kristaps 984:
1.78 kristaps 985: return(ROFF_IGN);
986: }
987:
988: static enum rofferr
1.80 kristaps 989: roff_block_sub(ROFF_ARGS)
1.79 kristaps 990: {
991: enum rofft t;
992: int i, j;
993:
994: /*
995: * First check whether a custom macro exists at this level. If
996: * it does, then check against it. This is some of groff's
997: * stranger behaviours. If we encountered a custom end-scope
998: * tag and that tag also happens to be a "real" macro, then we
999: * need to try interpreting it again as a real macro. If it's
1000: * not, then return ignore. Else continue.
1001: */
1002:
1003: if (r->last->end) {
1.130 kristaps 1004: for (i = pos, j = 0; r->last->end[j]; j++, i++)
1.79 kristaps 1005: if ((*bufp)[i] != r->last->end[j])
1006: break;
1007:
1.207 schwarze 1008: if ('\0' == r->last->end[j] &&
1009: ('\0' == (*bufp)[i] ||
1010: ' ' == (*bufp)[i] ||
1011: '\t' == (*bufp)[i])) {
1.79 kristaps 1012: roffnode_pop(r);
1013: roffnode_cleanscope(r);
1014:
1.130 kristaps 1015: while (' ' == (*bufp)[i] || '\t' == (*bufp)[i])
1016: i++;
1017:
1018: pos = i;
1.106 kristaps 1019: if (ROFF_MAX != roff_parse(r, *bufp, &pos))
1.79 kristaps 1020: return(ROFF_RERUN);
1021: return(ROFF_IGN);
1022: }
1023: }
1024:
1025: /*
1026: * If we have no custom end-query or lookup failed, then try
1027: * pulling it out of the hashtable.
1028: */
1029:
1.137 schwarze 1030: t = roff_parse(r, *bufp, &pos);
1.79 kristaps 1031:
1.106 kristaps 1032: /*
1033: * Macros other than block-end are only significant
1034: * in `de' blocks; elsewhere, simply throw them away.
1035: */
1036: if (ROFF_cblock != t) {
1037: if (ROFF_de == tok)
1.193 schwarze 1038: roff_setstr(r, r->last->name, *bufp + ppos, 2);
1.79 kristaps 1039: return(ROFF_IGN);
1.106 kristaps 1040: }
1.79 kristaps 1041:
1042: assert(roffs[t].proc);
1.207 schwarze 1043: return((*roffs[t].proc)(r, t, bufp, szp, ln, ppos, pos, offs));
1.79 kristaps 1044: }
1045:
1046: static enum rofferr
1.80 kristaps 1047: roff_block_text(ROFF_ARGS)
1.78 kristaps 1048: {
1049:
1.106 kristaps 1050: if (ROFF_de == tok)
1.193 schwarze 1051: roff_setstr(r, r->last->name, *bufp + pos, 2);
1.106 kristaps 1052:
1.78 kristaps 1053: return(ROFF_IGN);
1054: }
1055:
1056: static enum rofferr
1.82 kristaps 1057: roff_cond_sub(ROFF_ARGS)
1058: {
1059: enum rofft t;
1.139 kristaps 1060: char *ep;
1.198 schwarze 1061: int rr;
1.82 kristaps 1062:
1063: rr = r->last->rule;
1.139 kristaps 1064: roffnode_cleanscope(r);
1.177 schwarze 1065: t = roff_parse(r, *bufp, &pos);
1.82 kristaps 1066:
1.139 kristaps 1067: /*
1.177 schwarze 1068: * Fully handle known macros when they are structurally
1069: * required or when the conditional evaluated to true.
1.87 kristaps 1070: */
1071:
1.177 schwarze 1072: if ((ROFF_MAX != t) &&
1.198 schwarze 1073: (rr || ROFFMAC_STRUCT & roffs[t].flags)) {
1.177 schwarze 1074: assert(roffs[t].proc);
1075: return((*roffs[t].proc)(r, t, bufp, szp,
1.207 schwarze 1076: ln, ppos, pos, offs));
1.177 schwarze 1077: }
1.144 kristaps 1078:
1.196 schwarze 1079: /*
1080: * If `\}' occurs on a macro line without a preceding macro,
1081: * drop the line completely.
1082: */
1083:
1084: ep = *bufp + pos;
1085: if ('\\' == ep[0] && '}' == ep[1])
1.198 schwarze 1086: rr = 0;
1.196 schwarze 1087:
1.177 schwarze 1088: /* Always check for the closing delimiter `\}'. */
1.144 kristaps 1089:
1.177 schwarze 1090: while (NULL != (ep = strchr(ep, '\\'))) {
1.197 schwarze 1091: if ('}' == *(++ep)) {
1092: *ep = '&';
1093: roff_ccond(r, ln, ep - *bufp - 1);
1094: }
1095: ++ep;
1.177 schwarze 1096: }
1.198 schwarze 1097: return(rr ? ROFF_CONT : ROFF_IGN);
1.82 kristaps 1098: }
1099:
1100: static enum rofferr
1101: roff_cond_text(ROFF_ARGS)
1.78 kristaps 1102: {
1.140 kristaps 1103: char *ep;
1.198 schwarze 1104: int rr;
1.82 kristaps 1105:
1106: rr = r->last->rule;
1.140 kristaps 1107: roffnode_cleanscope(r);
1.82 kristaps 1108:
1.197 schwarze 1109: ep = *bufp + pos;
1110: while (NULL != (ep = strchr(ep, '\\'))) {
1111: if ('}' == *(++ep)) {
1112: *ep = '&';
1113: roff_ccond(r, ln, ep - *bufp - 1);
1114: }
1115: ++ep;
1.78 kristaps 1116: }
1.198 schwarze 1117: return(rr ? ROFF_CONT : ROFF_IGN);
1.74 kristaps 1118: }
1119:
1.204 schwarze 1120: /*
1121: * Parse a single signed integer number. Stop at the first non-digit.
1122: * If there is at least one digit, return success and advance the
1123: * parse point, else return failure and let the parse point unchanged.
1124: * Ignore overflows, treat them just like the C language.
1125: */
1.184 schwarze 1126: static int
1127: roff_getnum(const char *v, int *pos, int *res)
1128: {
1.206 schwarze 1129: int myres, n, p;
1130:
1131: if (NULL == res)
1132: res = &myres;
1.184 schwarze 1133:
1134: p = *pos;
1135: n = v[p] == '-';
1136: if (n)
1137: p++;
1138:
1139: for (*res = 0; isdigit((unsigned char)v[p]); p++)
1.204 schwarze 1140: *res = 10 * *res + v[p] - '0';
1.184 schwarze 1141: if (p == *pos + n)
1142: return 0;
1143:
1144: if (n)
1145: *res = -*res;
1146:
1147: *pos = p;
1148: return 1;
1149: }
1150:
1.198 schwarze 1151: /*
1152: * Evaluate a string comparison condition.
1153: * The first character is the delimiter.
1154: * Succeed if the string up to its second occurrence
1155: * matches the string up to its third occurence.
1156: * Advance the cursor after the third occurrence
1157: * or lacking that, to the end of the line.
1158: */
1159: static int
1160: roff_evalstrcond(const char *v, int *pos)
1161: {
1162: const char *s1, *s2, *s3;
1163: int match;
1164:
1165: match = 0;
1166: s1 = v + *pos; /* initial delimiter */
1167: s2 = s1 + 1; /* for scanning the first string */
1168: s3 = strchr(s2, *s1); /* for scanning the second string */
1169:
1170: if (NULL == s3) /* found no middle delimiter */
1171: goto out;
1172:
1173: while ('\0' != *++s3) {
1174: if (*s2 != *s3) { /* mismatch */
1175: s3 = strchr(s3, *s1);
1176: break;
1177: }
1178: if (*s3 == *s1) { /* found the final delimiter */
1179: match = 1;
1180: break;
1181: }
1182: s2++;
1183: }
1184:
1185: out:
1186: if (NULL == s3)
1187: s3 = strchr(s2, '\0');
1188: else
1189: s3++;
1190: *pos = s3 - v;
1191: return(match);
1192: }
1193:
1.204 schwarze 1194: /*
1195: * Evaluate an optionally negated single character, numerical,
1196: * or string condition.
1197: */
1.198 schwarze 1198: static int
1.88 kristaps 1199: roff_evalcond(const char *v, int *pos)
1200: {
1.204 schwarze 1201: int wanttrue, number;
1.88 kristaps 1202:
1.198 schwarze 1203: if ('!' == v[*pos]) {
1204: wanttrue = 0;
1205: (*pos)++;
1206: } else
1207: wanttrue = 1;
1208:
1.88 kristaps 1209: switch (v[*pos]) {
1.207 schwarze 1210: case 'n':
1.198 schwarze 1211: /* FALLTHROUGH */
1.207 schwarze 1212: case 'o':
1.88 kristaps 1213: (*pos)++;
1.198 schwarze 1214: return(wanttrue);
1.207 schwarze 1215: case 'c':
1.198 schwarze 1216: /* FALLTHROUGH */
1.207 schwarze 1217: case 'd':
1.198 schwarze 1218: /* FALLTHROUGH */
1.207 schwarze 1219: case 'e':
1.88 kristaps 1220: /* FALLTHROUGH */
1.207 schwarze 1221: case 'r':
1.88 kristaps 1222: /* FALLTHROUGH */
1.207 schwarze 1223: case 't':
1.88 kristaps 1224: (*pos)++;
1.198 schwarze 1225: return(!wanttrue);
1.88 kristaps 1226: default:
1227: break;
1228: }
1229:
1.204 schwarze 1230: if (roff_evalnum(v, pos, &number, 0))
1231: return((number > 0) == wanttrue);
1232: else
1.198 schwarze 1233: return(roff_evalstrcond(v, pos) == wanttrue);
1.88 kristaps 1234: }
1235:
1.74 kristaps 1236: static enum rofferr
1.103 kristaps 1237: roff_line_ignore(ROFF_ARGS)
1.89 kristaps 1238: {
1.123 schwarze 1239:
1.89 kristaps 1240: return(ROFF_IGN);
1241: }
1242:
1.104 kristaps 1243: static enum rofferr
1.82 kristaps 1244: roff_cond(ROFF_ARGS)
1.74 kristaps 1245: {
1.173 schwarze 1246:
1247: roffnode_push(r, tok, NULL, ln, ppos);
1.74 kristaps 1248:
1.207 schwarze 1249: /*
1.134 kristaps 1250: * An `.el' has no conditional body: it will consume the value
1251: * of the current rstack entry set in prior `ie' calls or
1.207 schwarze 1252: * defaults to DENY.
1.134 kristaps 1253: *
1254: * If we're not an `el', however, then evaluate the conditional.
1255: */
1.133 kristaps 1256:
1.173 schwarze 1257: r->last->rule = ROFF_el == tok ?
1.207 schwarze 1258: (r->rstackpos < 0 ? 0 : r->rstack[r->rstackpos--]) :
1259: roff_evalcond(*bufp, &pos);
1.77 kristaps 1260:
1.134 kristaps 1261: /*
1262: * An if-else will put the NEGATION of the current evaluated
1263: * conditional into the stack of rules.
1264: */
1265:
1.84 schwarze 1266: if (ROFF_ie == tok) {
1.134 kristaps 1267: if (r->rstackpos == RSTACK_MAX - 1) {
1.207 schwarze 1268: mandoc_msg(MANDOCERR_MEM,
1269: r->parse, ln, ppos, NULL);
1.134 kristaps 1270: return(ROFF_ERR);
1271: }
1.198 schwarze 1272: r->rstack[++r->rstackpos] = !r->last->rule;
1.82 kristaps 1273: }
1.88 kristaps 1274:
1275: /* If the parent has false as its rule, then so do we. */
1276:
1.198 schwarze 1277: if (r->last->parent && !r->last->parent->rule)
1278: r->last->rule = 0;
1.88 kristaps 1279:
1280: /*
1.173 schwarze 1281: * Determine scope.
1282: * If there is nothing on the line after the conditional,
1283: * not even whitespace, use next-line scope.
1.88 kristaps 1284: */
1.74 kristaps 1285:
1.173 schwarze 1286: if ('\0' == (*bufp)[pos]) {
1287: r->last->endspan = 2;
1288: goto out;
1289: }
1290:
1291: while (' ' == (*bufp)[pos])
1292: pos++;
1293:
1294: /* An opening brace requests multiline scope. */
1.75 kristaps 1295:
1296: if ('\\' == (*bufp)[pos] && '{' == (*bufp)[pos + 1]) {
1297: r->last->endspan = -1;
1298: pos += 2;
1.173 schwarze 1299: goto out;
1.207 schwarze 1300: }
1.74 kristaps 1301:
1.77 kristaps 1302: /*
1.173 schwarze 1303: * Anything else following the conditional causes
1304: * single-line scope. Warn if the scope contains
1305: * nothing but trailing whitespace.
1.77 kristaps 1306: */
1307:
1.75 kristaps 1308: if ('\0' == (*bufp)[pos])
1.173 schwarze 1309: mandoc_msg(MANDOCERR_NOARGS, r->parse, ln, ppos, NULL);
1.77 kristaps 1310:
1.173 schwarze 1311: r->last->endspan = 1;
1.74 kristaps 1312:
1.173 schwarze 1313: out:
1.75 kristaps 1314: *offs = pos;
1315: return(ROFF_RERUN);
1.83 schwarze 1316: }
1317:
1318: static enum rofferr
1.92 schwarze 1319: roff_ds(ROFF_ARGS)
1320: {
1.96 kristaps 1321: char *name, *string;
1322:
1323: /*
1324: * A symbol is named by the first word following the macro
1325: * invocation up to a space. Its value is anything after the
1326: * name's trailing whitespace and optional double-quote. Thus,
1327: *
1328: * [.ds foo "bar " ]
1329: *
1330: * will have `bar " ' as its value.
1331: */
1.92 schwarze 1332:
1.121 schwarze 1333: string = *bufp + pos;
1334: name = roff_getname(r, &string, ln, pos);
1.92 schwarze 1335: if ('\0' == *name)
1336: return(ROFF_IGN);
1337:
1.121 schwarze 1338: /* Read past initial double-quote. */
1339: if ('"' == *string)
1.92 schwarze 1340: string++;
1341:
1.96 kristaps 1342: /* The rest is the value. */
1.193 schwarze 1343: roff_setstr(r, name, string, ROFF_as == tok);
1.92 schwarze 1344: return(ROFF_IGN);
1345: }
1346:
1.204 schwarze 1347: /*
1348: * Parse a single operator, one or two characters long.
1349: * If the operator is recognized, return success and advance the
1350: * parse point, else return failure and let the parse point unchanged.
1351: */
1352: static int
1353: roff_getop(const char *v, int *pos, char *res)
1354: {
1355:
1356: *res = v[*pos];
1357:
1358: switch (*res) {
1.207 schwarze 1359: case '+':
1.204 schwarze 1360: /* FALLTHROUGH */
1.207 schwarze 1361: case '-':
1.204 schwarze 1362: /* FALLTHROUGH */
1.207 schwarze 1363: case '*':
1.204 schwarze 1364: /* FALLTHROUGH */
1.207 schwarze 1365: case '/':
1.204 schwarze 1366: /* FALLTHROUGH */
1.207 schwarze 1367: case '%':
1.204 schwarze 1368: /* FALLTHROUGH */
1.207 schwarze 1369: case '&':
1.204 schwarze 1370: /* FALLTHROUGH */
1.207 schwarze 1371: case ':':
1.204 schwarze 1372: break;
1373: case '<':
1374: switch (v[*pos + 1]) {
1.207 schwarze 1375: case '=':
1.204 schwarze 1376: *res = 'l';
1377: (*pos)++;
1378: break;
1.207 schwarze 1379: case '>':
1.204 schwarze 1380: *res = '!';
1381: (*pos)++;
1382: break;
1.207 schwarze 1383: case '?':
1.204 schwarze 1384: *res = 'i';
1385: (*pos)++;
1386: break;
1387: default:
1388: break;
1389: }
1390: break;
1391: case '>':
1392: switch (v[*pos + 1]) {
1.207 schwarze 1393: case '=':
1.204 schwarze 1394: *res = 'g';
1395: (*pos)++;
1396: break;
1.207 schwarze 1397: case '?':
1.204 schwarze 1398: *res = 'a';
1399: (*pos)++;
1400: break;
1401: default:
1402: break;
1403: }
1404: break;
1405: case '=':
1406: if ('=' == v[*pos + 1])
1407: (*pos)++;
1408: break;
1409: default:
1410: return(0);
1411: }
1412: (*pos)++;
1413:
1414: return(*res);
1415: }
1416:
1417: /*
1418: * Evaluate either a parenthesized numeric expression
1419: * or a single signed integer number.
1420: */
1421: static int
1422: roff_evalpar(const char *v, int *pos, int *res)
1423: {
1424:
1425: if ('(' != v[*pos])
1426: return(roff_getnum(v, pos, res));
1427:
1428: (*pos)++;
1429: if ( ! roff_evalnum(v, pos, res, 1))
1430: return(0);
1431:
1.206 schwarze 1432: /*
1433: * Omission of the closing parenthesis
1434: * is an error in validation mode,
1435: * but ignored in evaluation mode.
1436: */
1437:
1.204 schwarze 1438: if (')' == v[*pos])
1439: (*pos)++;
1.206 schwarze 1440: else if (NULL == res)
1441: return(0);
1.204 schwarze 1442:
1443: return(1);
1444: }
1445:
1446: /*
1447: * Evaluate a complete numeric expression.
1448: * Proceed left to right, there is no concept of precedence.
1449: */
1450: static int
1451: roff_evalnum(const char *v, int *pos, int *res, int skipwhite)
1452: {
1453: int mypos, operand2;
1454: char operator;
1455:
1456: if (NULL == pos) {
1457: mypos = 0;
1458: pos = &mypos;
1459: }
1460:
1461: if (skipwhite)
1462: while (isspace((unsigned char)v[*pos]))
1463: (*pos)++;
1464:
1465: if ( ! roff_evalpar(v, pos, res))
1466: return(0);
1467:
1468: while (1) {
1469: if (skipwhite)
1470: while (isspace((unsigned char)v[*pos]))
1471: (*pos)++;
1472:
1473: if ( ! roff_getop(v, pos, &operator))
1474: break;
1475:
1476: if (skipwhite)
1477: while (isspace((unsigned char)v[*pos]))
1478: (*pos)++;
1479:
1480: if ( ! roff_evalpar(v, pos, &operand2))
1481: return(0);
1482:
1483: if (skipwhite)
1484: while (isspace((unsigned char)v[*pos]))
1485: (*pos)++;
1.206 schwarze 1486:
1487: if (NULL == res)
1488: continue;
1.204 schwarze 1489:
1490: switch (operator) {
1.207 schwarze 1491: case '+':
1.204 schwarze 1492: *res += operand2;
1493: break;
1.207 schwarze 1494: case '-':
1.204 schwarze 1495: *res -= operand2;
1496: break;
1.207 schwarze 1497: case '*':
1.204 schwarze 1498: *res *= operand2;
1499: break;
1.207 schwarze 1500: case '/':
1.204 schwarze 1501: *res /= operand2;
1502: break;
1.207 schwarze 1503: case '%':
1.204 schwarze 1504: *res %= operand2;
1505: break;
1.207 schwarze 1506: case '<':
1.204 schwarze 1507: *res = *res < operand2;
1508: break;
1.207 schwarze 1509: case '>':
1.204 schwarze 1510: *res = *res > operand2;
1511: break;
1.207 schwarze 1512: case 'l':
1.204 schwarze 1513: *res = *res <= operand2;
1514: break;
1.207 schwarze 1515: case 'g':
1.204 schwarze 1516: *res = *res >= operand2;
1517: break;
1.207 schwarze 1518: case '=':
1.204 schwarze 1519: *res = *res == operand2;
1520: break;
1.207 schwarze 1521: case '!':
1.204 schwarze 1522: *res = *res != operand2;
1523: break;
1.207 schwarze 1524: case '&':
1.204 schwarze 1525: *res = *res && operand2;
1526: break;
1.207 schwarze 1527: case ':':
1.204 schwarze 1528: *res = *res || operand2;
1529: break;
1.207 schwarze 1530: case 'i':
1.204 schwarze 1531: if (operand2 < *res)
1532: *res = operand2;
1533: break;
1.207 schwarze 1534: case 'a':
1.204 schwarze 1535: if (operand2 > *res)
1536: *res = operand2;
1537: break;
1538: default:
1539: abort();
1540: }
1541: }
1542: return(1);
1543: }
1544:
1.180 schwarze 1545: void
1.187 schwarze 1546: roff_setreg(struct roff *r, const char *name, int val, char sign)
1.147 kristaps 1547: {
1.180 schwarze 1548: struct roffreg *reg;
1549:
1550: /* Search for an existing register with the same name. */
1551: reg = r->regtab;
1552:
1553: while (reg && strcmp(name, reg->key.p))
1554: reg = reg->next;
1.147 kristaps 1555:
1.180 schwarze 1556: if (NULL == reg) {
1557: /* Create a new register. */
1558: reg = mandoc_malloc(sizeof(struct roffreg));
1559: reg->key.p = mandoc_strdup(name);
1560: reg->key.sz = strlen(name);
1.187 schwarze 1561: reg->val = 0;
1.180 schwarze 1562: reg->next = r->regtab;
1563: r->regtab = reg;
1564: }
1565:
1.187 schwarze 1566: if ('+' == sign)
1567: reg->val += val;
1568: else if ('-' == sign)
1569: reg->val -= val;
1570: else
1571: reg->val = val;
1.147 kristaps 1572: }
1573:
1.192 schwarze 1574: /*
1575: * Handle some predefined read-only number registers.
1576: * For now, return -1 if the requested register is not predefined;
1577: * in case a predefined read-only register having the value -1
1578: * were to turn up, another special value would have to be chosen.
1579: */
1580: static int
1581: roff_getregro(const char *name)
1582: {
1583:
1584: switch (*name) {
1.207 schwarze 1585: case 'A': /* ASCII approximation mode is always off. */
1.192 schwarze 1586: return(0);
1.207 schwarze 1587: case 'g': /* Groff compatibility mode is always on. */
1.192 schwarze 1588: return(1);
1.207 schwarze 1589: case 'H': /* Fixed horizontal resolution. */
1.192 schwarze 1590: return (24);
1.207 schwarze 1591: case 'j': /* Always adjust left margin only. */
1.192 schwarze 1592: return(0);
1.207 schwarze 1593: case 'T': /* Some output device is always defined. */
1.192 schwarze 1594: return(1);
1.207 schwarze 1595: case 'V': /* Fixed vertical resolution. */
1.192 schwarze 1596: return (40);
1597: default:
1598: return (-1);
1599: }
1600: }
1601:
1.181 schwarze 1602: int
1.180 schwarze 1603: roff_getreg(const struct roff *r, const char *name)
1.147 kristaps 1604: {
1.180 schwarze 1605: struct roffreg *reg;
1.192 schwarze 1606: int val;
1607:
1608: if ('.' == name[0] && '\0' != name[1] && '\0' == name[2]) {
1609: val = roff_getregro(name + 1);
1610: if (-1 != val)
1611: return (val);
1612: }
1.180 schwarze 1613:
1614: for (reg = r->regtab; reg; reg = reg->next)
1615: if (0 == strcmp(name, reg->key.p))
1.181 schwarze 1616: return(reg->val);
1617:
1618: return(0);
1619: }
1620:
1621: static int
1622: roff_getregn(const struct roff *r, const char *name, size_t len)
1623: {
1624: struct roffreg *reg;
1.192 schwarze 1625: int val;
1626:
1627: if ('.' == name[0] && 2 == len) {
1628: val = roff_getregro(name + 1);
1629: if (-1 != val)
1630: return (val);
1631: }
1.181 schwarze 1632:
1633: for (reg = r->regtab; reg; reg = reg->next)
1634: if (len == reg->key.sz &&
1635: 0 == strncmp(name, reg->key.p, len))
1636: return(reg->val);
1.147 kristaps 1637:
1.180 schwarze 1638: return(0);
1.147 kristaps 1639: }
1640:
1.180 schwarze 1641: static void
1642: roff_freereg(struct roffreg *reg)
1.147 kristaps 1643: {
1.180 schwarze 1644: struct roffreg *old_reg;
1.147 kristaps 1645:
1.180 schwarze 1646: while (NULL != reg) {
1647: free(reg->key.p);
1648: old_reg = reg;
1649: reg = reg->next;
1650: free(old_reg);
1651: }
1.147 kristaps 1652: }
1.92 schwarze 1653:
1654: static enum rofferr
1.89 kristaps 1655: roff_nr(ROFF_ARGS)
1.83 schwarze 1656: {
1.121 schwarze 1657: const char *key;
1658: char *val;
1.138 kristaps 1659: int iv;
1.187 schwarze 1660: char sign;
1.89 kristaps 1661:
1.121 schwarze 1662: val = *bufp + pos;
1663: key = roff_getname(r, &val, ln, pos);
1.89 kristaps 1664:
1.187 schwarze 1665: sign = *val;
1666: if ('+' == sign || '-' == sign)
1667: val++;
1668:
1.204 schwarze 1669: if (roff_evalnum(val, NULL, &iv, 0))
1670: roff_setreg(r, key, iv, sign);
1.109 kristaps 1671:
1.203 schwarze 1672: return(ROFF_IGN);
1673: }
1674:
1675: static enum rofferr
1676: roff_rr(ROFF_ARGS)
1677: {
1678: struct roffreg *reg, **prev;
1679: const char *name;
1680: char *cp;
1681:
1682: cp = *bufp + pos;
1683: name = roff_getname(r, &cp, ln, pos);
1684:
1685: prev = &r->regtab;
1686: while (1) {
1687: reg = *prev;
1688: if (NULL == reg || !strcmp(name, reg->key.p))
1689: break;
1690: prev = ®->next;
1691: }
1692: if (NULL != reg) {
1693: *prev = reg->next;
1694: free(reg->key.p);
1695: free(reg);
1696: }
1.122 schwarze 1697: return(ROFF_IGN);
1698: }
1699:
1700: static enum rofferr
1701: roff_rm(ROFF_ARGS)
1702: {
1703: const char *name;
1704: char *cp;
1705:
1706: cp = *bufp + pos;
1707: while ('\0' != *cp) {
1.127 kristaps 1708: name = roff_getname(r, &cp, ln, (int)(cp - *bufp));
1.122 schwarze 1709: if ('\0' != *name)
1710: roff_setstr(r, name, NULL, 0);
1711: }
1.178 schwarze 1712: return(ROFF_IGN);
1713: }
1714:
1715: static enum rofferr
1716: roff_it(ROFF_ARGS)
1717: {
1718: char *cp;
1719: size_t len;
1720: int iv;
1721:
1722: /* Parse the number of lines. */
1723: cp = *bufp + pos;
1724: len = strcspn(cp, " \t");
1725: cp[len] = '\0';
1726: if ((iv = mandoc_strntoi(cp, len, 10)) <= 0) {
1727: mandoc_msg(MANDOCERR_NUMERIC, r->parse,
1.207 schwarze 1728: ln, ppos, *bufp + 1);
1.178 schwarze 1729: return(ROFF_IGN);
1730: }
1731: cp += len + 1;
1732:
1733: /* Arm the input line trap. */
1734: roffit_lines = iv;
1735: roffit_macro = mandoc_strdup(cp);
1.109 kristaps 1736: return(ROFF_IGN);
1.175 schwarze 1737: }
1738:
1739: static enum rofferr
1740: roff_Dd(ROFF_ARGS)
1741: {
1742: const char *const *cp;
1743:
1.199 schwarze 1744: if (0 == ((MPARSE_MDOC | MPARSE_QUICK) & r->options))
1.175 schwarze 1745: for (cp = __mdoc_reserved; *cp; cp++)
1746: roff_setstr(r, *cp, NULL, 0);
1747:
1748: return(ROFF_CONT);
1749: }
1750:
1751: static enum rofferr
1752: roff_TH(ROFF_ARGS)
1753: {
1754: const char *const *cp;
1755:
1.199 schwarze 1756: if (0 == (MPARSE_QUICK & r->options))
1.175 schwarze 1757: for (cp = __man_reserved; *cp; cp++)
1758: roff_setstr(r, *cp, NULL, 0);
1759:
1760: return(ROFF_CONT);
1.109 kristaps 1761: }
1762:
1763: static enum rofferr
1764: roff_TE(ROFF_ARGS)
1765: {
1766:
1767: if (NULL == r->tbl)
1.128 kristaps 1768: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.115 kristaps 1769: else
1.151 kristaps 1770: tbl_end(&r->tbl);
1.109 kristaps 1771:
1.112 kristaps 1772: return(ROFF_IGN);
1773: }
1774:
1775: static enum rofferr
1776: roff_T_(ROFF_ARGS)
1777: {
1778:
1779: if (NULL == r->tbl)
1.128 kristaps 1780: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.112 kristaps 1781: else
1.116 kristaps 1782: tbl_restart(ppos, ln, r->tbl);
1.112 kristaps 1783:
1.109 kristaps 1784: return(ROFF_IGN);
1785: }
1786:
1.156 kristaps 1787: #if 0
1788: static int
1.151 kristaps 1789: roff_closeeqn(struct roff *r)
1790: {
1791:
1792: return(r->eqn && ROFF_EQN == eqn_end(&r->eqn) ? 1 : 0);
1793: }
1.156 kristaps 1794: #endif
1.151 kristaps 1795:
1.156 kristaps 1796: static void
1.207 schwarze 1797: roff_openeqn(struct roff *r, const char *name, int line,
1.151 kristaps 1798: int offs, const char *buf)
1.125 kristaps 1799: {
1.151 kristaps 1800: struct eqn_node *e;
1801: int poff;
1.125 kristaps 1802:
1803: assert(NULL == r->eqn);
1.151 kristaps 1804: e = eqn_alloc(name, offs, line, r->parse);
1.125 kristaps 1805:
1806: if (r->last_eqn)
1807: r->last_eqn->next = e;
1808: else
1809: r->first_eqn = r->last_eqn = e;
1810:
1811: r->eqn = r->last_eqn = e;
1.151 kristaps 1812:
1813: if (buf) {
1814: poff = 0;
1815: eqn_read(&r->eqn, line, buf, offs, &poff);
1816: }
1817: }
1818:
1819: static enum rofferr
1820: roff_EQ(ROFF_ARGS)
1821: {
1822:
1823: roff_openeqn(r, *bufp + pos, ln, ppos, NULL);
1.125 kristaps 1824: return(ROFF_IGN);
1825: }
1826:
1827: static enum rofferr
1828: roff_EN(ROFF_ARGS)
1829: {
1830:
1.128 kristaps 1831: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.125 kristaps 1832: return(ROFF_IGN);
1833: }
1834:
1835: static enum rofferr
1.109 kristaps 1836: roff_TS(ROFF_ARGS)
1837: {
1.176 schwarze 1838: struct tbl_node *tbl;
1.89 kristaps 1839:
1.115 kristaps 1840: if (r->tbl) {
1.128 kristaps 1841: mandoc_msg(MANDOCERR_SCOPEBROKEN, r->parse, ln, ppos, NULL);
1.151 kristaps 1842: tbl_end(&r->tbl);
1.115 kristaps 1843: }
1.83 schwarze 1844:
1.176 schwarze 1845: tbl = tbl_alloc(ppos, ln, r->parse);
1.113 kristaps 1846:
1847: if (r->last_tbl)
1.176 schwarze 1848: r->last_tbl->next = tbl;
1.113 kristaps 1849: else
1.176 schwarze 1850: r->first_tbl = r->last_tbl = tbl;
1.113 kristaps 1851:
1.176 schwarze 1852: r->tbl = r->last_tbl = tbl;
1.83 schwarze 1853: return(ROFF_IGN);
1.92 schwarze 1854: }
1855:
1.105 kristaps 1856: static enum rofferr
1.174 kristaps 1857: roff_cc(ROFF_ARGS)
1858: {
1859: const char *p;
1860:
1861: p = *bufp + pos;
1862:
1863: if ('\0' == *p || '.' == (r->control = *p++))
1864: r->control = 0;
1865:
1866: if ('\0' != *p)
1867: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
1868:
1869: return(ROFF_IGN);
1870: }
1871:
1872: static enum rofferr
1.164 kristaps 1873: roff_tr(ROFF_ARGS)
1874: {
1875: const char *p, *first, *second;
1876: size_t fsz, ssz;
1877: enum mandoc_esc esc;
1878:
1879: p = *bufp + pos;
1880:
1881: if ('\0' == *p) {
1882: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
1883: return(ROFF_IGN);
1884: }
1885:
1886: while ('\0' != *p) {
1887: fsz = ssz = 1;
1888:
1889: first = p++;
1890: if ('\\' == *first) {
1891: esc = mandoc_escape(&p, NULL, NULL);
1892: if (ESCAPE_ERROR == esc) {
1.207 schwarze 1893: mandoc_msg(MANDOCERR_BADESCAPE,
1894: r->parse, ln,
1895: (int)(p - *bufp), NULL);
1.164 kristaps 1896: return(ROFF_IGN);
1897: }
1898: fsz = (size_t)(p - first);
1899: }
1900:
1901: second = p++;
1902: if ('\\' == *second) {
1903: esc = mandoc_escape(&p, NULL, NULL);
1904: if (ESCAPE_ERROR == esc) {
1.207 schwarze 1905: mandoc_msg(MANDOCERR_BADESCAPE,
1906: r->parse, ln,
1907: (int)(p - *bufp), NULL);
1.164 kristaps 1908: return(ROFF_IGN);
1909: }
1910: ssz = (size_t)(p - second);
1.165 kristaps 1911: } else if ('\0' == *second) {
1.207 schwarze 1912: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse,
1913: ln, (int)(p - *bufp), NULL);
1.164 kristaps 1914: second = " ";
1.165 kristaps 1915: p--;
1.164 kristaps 1916: }
1917:
1.167 kristaps 1918: if (fsz > 1) {
1.207 schwarze 1919: roff_setstrn(&r->xmbtab, first, fsz,
1920: second, ssz, 0);
1.167 kristaps 1921: continue;
1922: }
1923:
1924: if (NULL == r->xtab)
1.207 schwarze 1925: r->xtab = mandoc_calloc(128,
1926: sizeof(struct roffstr));
1.167 kristaps 1927:
1928: free(r->xtab[(int)*first].p);
1929: r->xtab[(int)*first].p = mandoc_strndup(second, ssz);
1930: r->xtab[(int)*first].sz = ssz;
1.164 kristaps 1931: }
1932:
1933: return(ROFF_IGN);
1934: }
1935:
1936: static enum rofferr
1.105 kristaps 1937: roff_so(ROFF_ARGS)
1938: {
1939: char *name;
1940:
1.210 ! schwarze 1941: name = *bufp + pos;
! 1942: mandoc_vmsg(MANDOCERR_SO, r->parse, ln, ppos, ".so %s", name);
1.105 kristaps 1943:
1944: /*
1945: * Handle `so'. Be EXTREMELY careful, as we shouldn't be
1946: * opening anything that's not in our cwd or anything beneath
1947: * it. Thus, explicitly disallow traversing up the file-system
1948: * or using absolute paths.
1949: */
1950:
1951: if ('/' == *name || strstr(name, "../") || strstr(name, "/..")) {
1.210 ! schwarze 1952: mandoc_vmsg(MANDOCERR_SO_PATH, r->parse, ln, ppos,
! 1953: ".so %s", name);
1.105 kristaps 1954: return(ROFF_ERR);
1955: }
1956:
1957: *offs = pos;
1958: return(ROFF_SO);
1959: }
1.92 schwarze 1960:
1.106 kristaps 1961: static enum rofferr
1962: roff_userdef(ROFF_ARGS)
1.99 kristaps 1963: {
1.106 kristaps 1964: const char *arg[9];
1965: char *cp, *n1, *n2;
1.119 schwarze 1966: int i;
1.106 kristaps 1967:
1968: /*
1969: * Collect pointers to macro argument strings
1.188 schwarze 1970: * and NUL-terminate them.
1.106 kristaps 1971: */
1972: cp = *bufp + pos;
1.119 schwarze 1973: for (i = 0; i < 9; i++)
1.120 schwarze 1974: arg[i] = '\0' == *cp ? "" :
1.136 kristaps 1975: mandoc_getarg(r->parse, &cp, ln, &pos);
1.99 kristaps 1976:
1.106 kristaps 1977: /*
1978: * Expand macro arguments.
1.99 kristaps 1979: */
1.106 kristaps 1980: *szp = 0;
1981: n1 = cp = mandoc_strdup(r->current_string);
1982: while (NULL != (cp = strstr(cp, "\\$"))) {
1983: i = cp[2] - '1';
1984: if (0 > i || 8 < i) {
1985: /* Not an argument invocation. */
1986: cp += 2;
1987: continue;
1988: }
1.209 schwarze 1989: *cp = '\0';
1990: *szp = mandoc_asprintf(&n2, "%s%s%s",
1991: n1, arg[i], cp + 3) + 1;
1.106 kristaps 1992: cp = n2 + (cp - n1);
1993: free(n1);
1994: n1 = n2;
1.99 kristaps 1995: }
1996:
1.106 kristaps 1997: /*
1998: * Replace the macro invocation
1999: * by the expanded macro.
2000: */
2001: free(*bufp);
2002: *bufp = n1;
2003: if (0 == *szp)
2004: *szp = strlen(*bufp) + 1;
2005:
2006: return(*szp > 1 && '\n' == (*bufp)[(int)*szp - 2] ?
2007: ROFF_REPARSE : ROFF_APPEND);
1.99 kristaps 2008: }
1.121 schwarze 2009:
2010: static char *
2011: roff_getname(struct roff *r, char **cpp, int ln, int pos)
2012: {
2013: char *name, *cp;
2014:
2015: name = *cpp;
2016: if ('\0' == *name)
2017: return(name);
2018:
2019: /* Read until end of name. */
2020: for (cp = name; '\0' != *cp && ' ' != *cp; cp++) {
2021: if ('\\' != *cp)
2022: continue;
2023: cp++;
2024: if ('\\' == *cp)
2025: continue;
1.128 kristaps 2026: mandoc_msg(MANDOCERR_NAMESC, r->parse, ln, pos, NULL);
1.121 schwarze 2027: *cp = '\0';
2028: name = cp;
2029: }
2030:
2031: /* Nil-terminate name. */
2032: if ('\0' != *cp)
2033: *(cp++) = '\0';
2034:
2035: /* Read past spaces. */
2036: while (' ' == *cp)
2037: cp++;
2038:
2039: *cpp = cp;
2040: return(name);
2041: }
2042:
1.106 kristaps 2043: /*
2044: * Store *string into the user-defined string called *name.
2045: * To clear an existing entry, call with (*r, *name, NULL, 0).
1.193 schwarze 2046: * append == 0: replace mode
2047: * append == 1: single-line append mode
2048: * append == 2: multiline append mode, append '\n' after each call
1.106 kristaps 2049: */
1.94 kristaps 2050: static void
1.106 kristaps 2051: roff_setstr(struct roff *r, const char *name, const char *string,
1.193 schwarze 2052: int append)
1.92 schwarze 2053: {
1.164 kristaps 2054:
2055: roff_setstrn(&r->strtab, name, strlen(name), string,
1.207 schwarze 2056: string ? strlen(string) : 0, append);
1.164 kristaps 2057: }
2058:
2059: static void
1.166 kristaps 2060: roff_setstrn(struct roffkv **r, const char *name, size_t namesz,
1.193 schwarze 2061: const char *string, size_t stringsz, int append)
1.164 kristaps 2062: {
1.166 kristaps 2063: struct roffkv *n;
1.164 kristaps 2064: char *c;
2065: int i;
2066: size_t oldch, newch;
1.92 schwarze 2067:
1.106 kristaps 2068: /* Search for an existing string with the same name. */
1.164 kristaps 2069: n = *r;
2070:
1.166 kristaps 2071: while (n && strcmp(name, n->key.p))
1.92 schwarze 2072: n = n->next;
1.94 kristaps 2073:
2074: if (NULL == n) {
1.106 kristaps 2075: /* Create a new string table entry. */
1.166 kristaps 2076: n = mandoc_malloc(sizeof(struct roffkv));
2077: n->key.p = mandoc_strndup(name, namesz);
2078: n->key.sz = namesz;
2079: n->val.p = NULL;
2080: n->val.sz = 0;
1.164 kristaps 2081: n->next = *r;
2082: *r = n;
1.193 schwarze 2083: } else if (0 == append) {
1.166 kristaps 2084: free(n->val.p);
2085: n->val.p = NULL;
2086: n->val.sz = 0;
1.106 kristaps 2087: }
2088:
2089: if (NULL == string)
2090: return;
2091:
2092: /*
2093: * One additional byte for the '\n' in multiline mode,
2094: * and one for the terminating '\0'.
2095: */
1.193 schwarze 2096: newch = stringsz + (1 < append ? 2u : 1u);
1.164 kristaps 2097:
1.166 kristaps 2098: if (NULL == n->val.p) {
2099: n->val.p = mandoc_malloc(newch);
2100: *n->val.p = '\0';
1.106 kristaps 2101: oldch = 0;
2102: } else {
1.166 kristaps 2103: oldch = n->val.sz;
2104: n->val.p = mandoc_realloc(n->val.p, oldch + newch);
1.106 kristaps 2105: }
2106:
2107: /* Skip existing content in the destination buffer. */
1.166 kristaps 2108: c = n->val.p + (int)oldch;
1.106 kristaps 2109:
2110: /* Append new content to the destination buffer. */
1.164 kristaps 2111: i = 0;
2112: while (i < (int)stringsz) {
1.106 kristaps 2113: /*
2114: * Rudimentary roff copy mode:
2115: * Handle escaped backslashes.
2116: */
1.164 kristaps 2117: if ('\\' == string[i] && '\\' == string[i + 1])
2118: i++;
2119: *c++ = string[i++];
1.106 kristaps 2120: }
1.94 kristaps 2121:
1.106 kristaps 2122: /* Append terminating bytes. */
1.193 schwarze 2123: if (1 < append)
1.106 kristaps 2124: *c++ = '\n';
1.163 kristaps 2125:
1.106 kristaps 2126: *c = '\0';
1.166 kristaps 2127: n->val.sz = (int)(c - n->val.p);
1.92 schwarze 2128: }
2129:
1.94 kristaps 2130: static const char *
2131: roff_getstrn(const struct roff *r, const char *name, size_t len)
1.92 schwarze 2132: {
1.166 kristaps 2133: const struct roffkv *n;
1.191 schwarze 2134: int i;
1.92 schwarze 2135:
1.164 kristaps 2136: for (n = r->strtab; n; n = n->next)
1.207 schwarze 2137: if (0 == strncmp(name, n->key.p, len) &&
2138: '\0' == n->key.p[(int)len])
1.166 kristaps 2139: return(n->val.p);
1.191 schwarze 2140:
2141: for (i = 0; i < PREDEFS_MAX; i++)
2142: if (0 == strncmp(name, predefs[i].name, len) &&
2143: '\0' == predefs[i].name[(int)len])
2144: return(predefs[i].str);
1.94 kristaps 2145:
1.157 kristaps 2146: return(NULL);
1.92 schwarze 2147: }
2148:
1.94 kristaps 2149: static void
1.167 kristaps 2150: roff_freestr(struct roffkv *r)
1.92 schwarze 2151: {
1.166 kristaps 2152: struct roffkv *n, *nn;
1.92 schwarze 2153:
1.167 kristaps 2154: for (n = r; n; n = nn) {
1.166 kristaps 2155: free(n->key.p);
2156: free(n->val.p);
1.92 schwarze 2157: nn = n->next;
2158: free(n);
2159: }
1.114 kristaps 2160: }
2161:
2162: const struct tbl_span *
2163: roff_span(const struct roff *r)
2164: {
1.207 schwarze 2165:
1.114 kristaps 2166: return(r->tbl ? tbl_span(r->tbl) : NULL);
1.125 kristaps 2167: }
2168:
2169: const struct eqn *
2170: roff_eqn(const struct roff *r)
2171: {
1.207 schwarze 2172:
1.125 kristaps 2173: return(r->last_eqn ? &r->last_eqn->eqn : NULL);
1.164 kristaps 2174: }
2175:
2176: /*
2177: * Duplicate an input string, making the appropriate character
2178: * conversations (as stipulated by `tr') along the way.
2179: * Returns a heap-allocated string with all the replacements made.
2180: */
2181: char *
2182: roff_strdup(const struct roff *r, const char *p)
2183: {
1.166 kristaps 2184: const struct roffkv *cp;
1.164 kristaps 2185: char *res;
2186: const char *pp;
2187: size_t ssz, sz;
2188: enum mandoc_esc esc;
2189:
1.167 kristaps 2190: if (NULL == r->xmbtab && NULL == r->xtab)
1.164 kristaps 2191: return(mandoc_strdup(p));
2192: else if ('\0' == *p)
2193: return(mandoc_strdup(""));
2194:
2195: /*
2196: * Step through each character looking for term matches
2197: * (remember that a `tr' can be invoked with an escape, which is
2198: * a glyph but the escape is multi-character).
2199: * We only do this if the character hash has been initialised
2200: * and the string is >0 length.
2201: */
2202:
2203: res = NULL;
2204: ssz = 0;
2205:
2206: while ('\0' != *p) {
1.167 kristaps 2207: if ('\\' != *p && r->xtab && r->xtab[(int)*p].p) {
2208: sz = r->xtab[(int)*p].sz;
2209: res = mandoc_realloc(res, ssz + sz + 1);
2210: memcpy(res + ssz, r->xtab[(int)*p].p, sz);
2211: ssz += sz;
2212: p++;
2213: continue;
2214: } else if ('\\' != *p) {
2215: res = mandoc_realloc(res, ssz + 2);
2216: res[ssz++] = *p++;
2217: continue;
2218: }
2219:
1.164 kristaps 2220: /* Search for term matches. */
1.167 kristaps 2221: for (cp = r->xmbtab; cp; cp = cp->next)
1.166 kristaps 2222: if (0 == strncmp(p, cp->key.p, cp->key.sz))
1.164 kristaps 2223: break;
2224:
2225: if (NULL != cp) {
2226: /*
2227: * A match has been found.
2228: * Append the match to the array and move
2229: * forward by its keysize.
2230: */
1.207 schwarze 2231: res = mandoc_realloc(res,
2232: ssz + cp->val.sz + 1);
1.166 kristaps 2233: memcpy(res + ssz, cp->val.p, cp->val.sz);
2234: ssz += cp->val.sz;
2235: p += (int)cp->key.sz;
1.164 kristaps 2236: continue;
2237: }
2238:
1.167 kristaps 2239: /*
2240: * Handle escapes carefully: we need to copy
2241: * over just the escape itself, or else we might
2242: * do replacements within the escape itself.
2243: * Make sure to pass along the bogus string.
2244: */
2245: pp = p++;
2246: esc = mandoc_escape(&p, NULL, NULL);
2247: if (ESCAPE_ERROR == esc) {
2248: sz = strlen(pp);
1.164 kristaps 2249: res = mandoc_realloc(res, ssz + sz + 1);
2250: memcpy(res + ssz, pp, sz);
1.167 kristaps 2251: break;
1.164 kristaps 2252: }
1.207 schwarze 2253: /*
2254: * We bail out on bad escapes.
1.167 kristaps 2255: * No need to warn: we already did so when
2256: * roff_res() was called.
2257: */
2258: sz = (int)(p - pp);
2259: res = mandoc_realloc(res, ssz + sz + 1);
2260: memcpy(res + ssz, pp, sz);
2261: ssz += sz;
1.164 kristaps 2262: }
2263:
2264: res[(int)ssz] = '\0';
2265: return(res);
1.174 kristaps 2266: }
2267:
2268: /*
1.207 schwarze 2269: * Find out whether a line is a macro line or not.
1.174 kristaps 2270: * If it is, adjust the current position and return one; if it isn't,
2271: * return zero and don't change the current position.
2272: * If the control character has been set with `.cc', then let that grain
2273: * precedence.
2274: * This is slighly contrary to groff, where using the non-breaking
2275: * control character when `cc' has been invoked will cause the
2276: * non-breaking macro contents to be printed verbatim.
2277: */
2278: int
2279: roff_getcontrol(const struct roff *r, const char *cp, int *ppos)
2280: {
2281: int pos;
2282:
2283: pos = *ppos;
2284:
2285: if (0 != r->control && cp[pos] == r->control)
2286: pos++;
2287: else if (0 != r->control)
2288: return(0);
2289: else if ('\\' == cp[pos] && '.' == cp[pos + 1])
2290: pos += 2;
2291: else if ('.' == cp[pos] || '\'' == cp[pos])
2292: pos++;
2293: else
2294: return(0);
2295:
2296: while (' ' == cp[pos] || '\t' == cp[pos])
2297: pos++;
2298:
2299: *ppos = pos;
2300: return(1);
1.74 kristaps 2301: }
CVSweb