Annotation of mandoc/roff.c, Revision 1.217
1.217 ! schwarze 1: /* $Id: roff.c,v 1.216 2014/07/03 21:23:54 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 *);
1.212 schwarze 190: static size_t 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 *);
1.214 schwarze 203: static enum rofft roff_parse(struct roff *, char *, int *,
204: int, int);
1.178 schwarze 205: static enum rofferr roff_parsetext(char **, size_t *, int, int *);
1.207 schwarze 206: static enum rofferr roff_res(struct roff *,
1.142 kristaps 207: char **, size_t *, int, int);
1.122 schwarze 208: static enum rofferr roff_rm(ROFF_ARGS);
1.203 schwarze 209: static enum rofferr roff_rr(ROFF_ARGS);
1.94 kristaps 210: static void roff_setstr(struct roff *,
1.106 kristaps 211: const char *, const char *, int);
1.207 schwarze 212: static void roff_setstrn(struct roffkv **, const char *,
1.164 kristaps 213: size_t, const char *, size_t, int);
1.105 kristaps 214: static enum rofferr roff_so(ROFF_ARGS);
1.164 kristaps 215: static enum rofferr roff_tr(ROFF_ARGS);
1.175 schwarze 216: static enum rofferr roff_Dd(ROFF_ARGS);
217: static enum rofferr roff_TH(ROFF_ARGS);
1.109 kristaps 218: static enum rofferr roff_TE(ROFF_ARGS);
219: static enum rofferr roff_TS(ROFF_ARGS);
1.125 kristaps 220: static enum rofferr roff_EQ(ROFF_ARGS);
221: static enum rofferr roff_EN(ROFF_ARGS);
1.112 kristaps 222: static enum rofferr roff_T_(ROFF_ARGS);
1.106 kristaps 223: static enum rofferr roff_userdef(ROFF_ARGS);
1.67 kristaps 224:
1.155 kristaps 225: /* See roffhash_find() */
1.85 kristaps 226:
227: #define ASCII_HI 126
228: #define ASCII_LO 33
229: #define HASHWIDTH (ASCII_HI - ASCII_LO + 1)
230:
231: static struct roffmac *hash[HASHWIDTH];
232:
233: static struct roffmac roffs[ROFF_MAX] = {
1.103 kristaps 234: { "ad", roff_line_ignore, NULL, NULL, 0, NULL },
1.85 kristaps 235: { "am", roff_block, roff_block_text, roff_block_sub, 0, NULL },
236: { "ami", roff_block, roff_block_text, roff_block_sub, 0, NULL },
237: { "am1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.193 schwarze 238: { "as", roff_ds, NULL, NULL, 0, NULL },
1.174 kristaps 239: { "cc", roff_cc, NULL, NULL, 0, NULL },
1.194 schwarze 240: { "ce", roff_line_ignore, NULL, NULL, 0, NULL },
1.85 kristaps 241: { "de", roff_block, roff_block_text, roff_block_sub, 0, NULL },
242: { "dei", roff_block, roff_block_text, roff_block_sub, 0, NULL },
243: { "de1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.92 schwarze 244: { "ds", roff_ds, NULL, NULL, 0, NULL },
1.85 kristaps 245: { "el", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
1.185 schwarze 246: { "fam", roff_line_ignore, NULL, NULL, 0, NULL },
1.186 schwarze 247: { "hw", roff_line_ignore, NULL, NULL, 0, NULL },
1.103 kristaps 248: { "hy", roff_line_ignore, NULL, NULL, 0, NULL },
1.85 kristaps 249: { "ie", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
250: { "if", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
251: { "ig", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.178 schwarze 252: { "it", roff_it, NULL, NULL, 0, NULL },
1.103 kristaps 253: { "ne", roff_line_ignore, NULL, NULL, 0, NULL },
254: { "nh", roff_line_ignore, NULL, NULL, 0, NULL },
1.104 kristaps 255: { "nr", roff_nr, NULL, NULL, 0, NULL },
1.124 schwarze 256: { "ns", roff_line_ignore, NULL, NULL, 0, NULL },
257: { "ps", roff_line_ignore, NULL, NULL, 0, NULL },
1.122 schwarze 258: { "rm", roff_rm, NULL, NULL, 0, NULL },
1.203 schwarze 259: { "rr", roff_rr, NULL, NULL, 0, NULL },
1.105 kristaps 260: { "so", roff_so, NULL, NULL, 0, NULL },
1.124 schwarze 261: { "ta", roff_line_ignore, NULL, NULL, 0, NULL },
1.164 kristaps 262: { "tr", roff_tr, NULL, NULL, 0, NULL },
1.175 schwarze 263: { "Dd", roff_Dd, NULL, NULL, 0, NULL },
264: { "TH", roff_TH, NULL, NULL, 0, NULL },
1.109 kristaps 265: { "TS", roff_TS, NULL, NULL, 0, NULL },
266: { "TE", roff_TE, NULL, NULL, 0, NULL },
1.112 kristaps 267: { "T&", roff_T_, NULL, NULL, 0, NULL },
1.125 kristaps 268: { "EQ", roff_EQ, NULL, NULL, 0, NULL },
269: { "EN", roff_EN, NULL, NULL, 0, NULL },
1.85 kristaps 270: { ".", roff_cblock, NULL, NULL, 0, NULL },
1.106 kristaps 271: { NULL, roff_userdef, NULL, NULL, 0, NULL },
1.67 kristaps 272: };
273:
1.200 schwarze 274: /* not currently implemented: Ds em Eq LP Me PP pp Or Rd Sf SH */
1.175 schwarze 275: const char *const __mdoc_reserved[] = {
276: "Ac", "Ad", "An", "Ao", "Ap", "Aq", "Ar", "At",
277: "Bc", "Bd", "Bf", "Bk", "Bl", "Bo", "Bq",
278: "Brc", "Bro", "Brq", "Bsx", "Bt", "Bx",
279: "Cd", "Cm", "Db", "Dc", "Dd", "Dl", "Do", "Dq",
1.200 schwarze 280: "Dt", "Dv", "Dx", "D1",
281: "Ec", "Ed", "Ef", "Ek", "El", "Em",
282: "En", "Eo", "Er", "Es", "Ev", "Ex",
1.175 schwarze 283: "Fa", "Fc", "Fd", "Fl", "Fn", "Fo", "Fr", "Ft", "Fx",
1.200 schwarze 284: "Hf", "Ic", "In", "It", "Lb", "Li", "Lk", "Lp",
285: "Ms", "Mt", "Nd", "Nm", "No", "Ns", "Nx",
1.175 schwarze 286: "Oc", "Oo", "Op", "Os", "Ot", "Ox",
1.200 schwarze 287: "Pa", "Pc", "Pf", "Po", "Pp", "Pq",
288: "Qc", "Ql", "Qo", "Qq", "Re", "Rs", "Rv",
289: "Sc", "Sh", "Sm", "So", "Sq",
1.175 schwarze 290: "Ss", "St", "Sx", "Sy",
291: "Ta", "Tn", "Ud", "Ux", "Va", "Vt", "Xc", "Xo", "Xr",
1.200 schwarze 292: "%A", "%B", "%C", "%D", "%I", "%J", "%N", "%O",
1.175 schwarze 293: "%P", "%Q", "%R", "%T", "%U", "%V",
294: NULL
295: };
296:
1.200 schwarze 297: /* not currently implemented: BT DE DS ME MT PT SY TQ YS */
1.175 schwarze 298: const char *const __man_reserved[] = {
1.200 schwarze 299: "AT", "B", "BI", "BR", "DT",
300: "EE", "EN", "EQ", "EX", "HP", "I", "IB", "IP", "IR",
301: "LP", "OP", "P", "PD", "PP",
302: "R", "RB", "RE", "RI", "RS", "SB", "SH", "SM", "SS",
303: "TE", "TH", "TP", "TS", "T&", "UC", "UE", "UR",
1.175 schwarze 304: NULL
305: };
306:
1.141 kristaps 307: /* Array of injected predefined strings. */
308: #define PREDEFS_MAX 38
309: static const struct predef predefs[PREDEFS_MAX] = {
310: #include "predefs.in"
311: };
312:
1.155 kristaps 313: /* See roffhash_find() */
1.85 kristaps 314: #define ROFF_HASH(p) (p[0] - ASCII_LO)
315:
1.178 schwarze 316: static int roffit_lines; /* number of lines to delay */
317: static char *roffit_macro; /* nil-terminated macro line */
318:
1.207 schwarze 319:
1.85 kristaps 320: static void
1.155 kristaps 321: roffhash_init(void)
1.85 kristaps 322: {
323: struct roffmac *n;
324: int buc, i;
325:
1.106 kristaps 326: for (i = 0; i < (int)ROFF_USERDEF; i++) {
1.85 kristaps 327: assert(roffs[i].name[0] >= ASCII_LO);
328: assert(roffs[i].name[0] <= ASCII_HI);
329:
330: buc = ROFF_HASH(roffs[i].name);
331:
332: if (NULL != (n = hash[buc])) {
333: for ( ; n->next; n = n->next)
334: /* Do nothing. */ ;
335: n->next = &roffs[i];
336: } else
337: hash[buc] = &roffs[i];
338: }
339: }
340:
1.67 kristaps 341: /*
342: * Look up a roff token by its name. Returns ROFF_MAX if no macro by
343: * the nil-terminated string name could be found.
344: */
345: static enum rofft
1.155 kristaps 346: roffhash_find(const char *p, size_t s)
1.67 kristaps 347: {
1.85 kristaps 348: int buc;
349: struct roffmac *n;
1.67 kristaps 350:
1.85 kristaps 351: /*
352: * libroff has an extremely simple hashtable, for the time
353: * being, which simply keys on the first character, which must
354: * be printable, then walks a chain. It works well enough until
355: * optimised.
356: */
357:
358: if (p[0] < ASCII_LO || p[0] > ASCII_HI)
359: return(ROFF_MAX);
360:
361: buc = ROFF_HASH(p);
362:
363: if (NULL == (n = hash[buc]))
364: return(ROFF_MAX);
365: for ( ; n; n = n->next)
1.106 kristaps 366: if (0 == strncmp(n->name, p, s) && '\0' == n->name[(int)s])
1.85 kristaps 367: return((enum rofft)(n - roffs));
1.67 kristaps 368:
369: return(ROFF_MAX);
370: }
371:
372: /*
373: * Pop the current node off of the stack of roff instructions currently
374: * pending.
375: */
376: static void
377: roffnode_pop(struct roff *r)
378: {
379: struct roffnode *p;
380:
1.75 kristaps 381: assert(r->last);
1.207 schwarze 382: p = r->last;
1.82 kristaps 383:
1.75 kristaps 384: r->last = r->last->parent;
1.106 kristaps 385: free(p->name);
386: free(p->end);
1.67 kristaps 387: free(p);
388: }
389:
390: /*
391: * Push a roff node onto the instruction stack. This must later be
392: * removed with roffnode_pop().
393: */
1.98 schwarze 394: static void
1.106 kristaps 395: roffnode_push(struct roff *r, enum rofft tok, const char *name,
396: int line, int col)
1.67 kristaps 397: {
398: struct roffnode *p;
399:
1.98 schwarze 400: p = mandoc_calloc(1, sizeof(struct roffnode));
1.67 kristaps 401: p->tok = tok;
1.106 kristaps 402: if (name)
403: p->name = mandoc_strdup(name);
1.67 kristaps 404: p->parent = r->last;
405: p->line = line;
406: p->col = col;
1.198 schwarze 407: p->rule = p->parent ? p->parent->rule : 0;
1.67 kristaps 408:
409: r->last = p;
410: }
411:
412: static void
413: roff_free1(struct roff *r)
414: {
1.176 schwarze 415: struct tbl_node *tbl;
1.125 kristaps 416: struct eqn_node *e;
1.167 kristaps 417: int i;
1.67 kristaps 418:
1.176 schwarze 419: while (NULL != (tbl = r->first_tbl)) {
420: r->first_tbl = tbl->next;
421: tbl_free(tbl);
1.109 kristaps 422: }
423:
1.113 kristaps 424: r->first_tbl = r->last_tbl = r->tbl = NULL;
425:
1.125 kristaps 426: while (NULL != (e = r->first_eqn)) {
427: r->first_eqn = e->next;
428: eqn_free(e);
429: }
430:
431: r->first_eqn = r->last_eqn = r->eqn = NULL;
432:
1.67 kristaps 433: while (r->last)
434: roffnode_pop(r);
1.109 kristaps 435:
1.167 kristaps 436: roff_freestr(r->strtab);
437: roff_freestr(r->xmbtab);
438:
439: r->strtab = r->xmbtab = NULL;
440:
1.180 schwarze 441: roff_freereg(r->regtab);
442:
443: r->regtab = NULL;
444:
1.167 kristaps 445: if (r->xtab)
446: for (i = 0; i < 128; i++)
447: free(r->xtab[i].p);
448:
449: free(r->xtab);
450: r->xtab = NULL;
1.67 kristaps 451: }
452:
453: void
454: roff_reset(struct roff *r)
455: {
456:
457: roff_free1(r);
1.174 kristaps 458: r->control = 0;
1.67 kristaps 459: }
460:
461: void
462: roff_free(struct roff *r)
463: {
464:
465: roff_free1(r);
466: free(r);
467: }
468:
469: struct roff *
1.199 schwarze 470: roff_alloc(struct mparse *parse, int options)
1.67 kristaps 471: {
472: struct roff *r;
473:
1.98 schwarze 474: r = mandoc_calloc(1, sizeof(struct roff));
1.128 kristaps 475: r->parse = parse;
1.199 schwarze 476: r->options = options;
1.82 kristaps 477: r->rstackpos = -1;
1.207 schwarze 478:
1.155 kristaps 479: roffhash_init();
1.141 kristaps 480:
1.67 kristaps 481: return(r);
482: }
483:
1.94 kristaps 484: /*
1.206 schwarze 485: * In the current line, expand escape sequences that tend to get
486: * used in numerical expressions and conditional requests.
487: * Also check the syntax of the remaining escape sequences.
1.154 kristaps 488: */
1.172 schwarze 489: static enum rofferr
1.142 kristaps 490: roff_res(struct roff *r, char **bufp, size_t *szp, int ln, int pos)
1.94 kristaps 491: {
1.208 schwarze 492: char ubuf[24]; /* buffer to print the number */
1.205 schwarze 493: const char *start; /* start of the string to process */
1.209 schwarze 494: char *stesc; /* start of an escape sequence ('\\') */
1.108 schwarze 495: const char *stnam; /* start of the name, after "[(*" */
496: const char *cp; /* end of the name, e.g. before ']' */
497: const char *res; /* the string to be substituted */
1.181 schwarze 498: char *nbuf; /* new buffer to copy bufp to */
499: size_t maxl; /* expected length of the escape name */
500: size_t naml; /* actual length of the escape name */
501: int expand_count; /* to avoid infinite loops */
1.206 schwarze 502: int npos; /* position in numeric expression */
503: int irc; /* return code from roff_evalnum() */
504: char term; /* character terminating the escape */
1.94 kristaps 505:
1.170 schwarze 506: expand_count = 0;
1.205 schwarze 507: start = *bufp + pos;
508: stesc = strchr(start, '\0') - 1;
509: while (stesc-- > start) {
1.170 schwarze 510:
1.205 schwarze 511: /* Search backwards for the next backslash. */
512:
513: if ('\\' != *stesc)
514: continue;
515:
516: /* If it is escaped, skip it. */
517:
518: for (cp = stesc - 1; cp >= start; cp--)
519: if ('\\' != *cp)
520: break;
521:
522: if (0 == (stesc - cp) % 2) {
1.209 schwarze 523: stesc = (char *)cp;
1.205 schwarze 524: continue;
525: }
1.108 schwarze 526:
1.206 schwarze 527: /* Decide whether to expand or to check only. */
1.108 schwarze 528:
1.206 schwarze 529: term = '\0';
1.205 schwarze 530: cp = stesc + 1;
1.181 schwarze 531: switch (*cp) {
1.207 schwarze 532: case '*':
1.181 schwarze 533: res = NULL;
534: break;
1.207 schwarze 535: case 'B':
1.206 schwarze 536: /* FALLTHROUGH */
1.207 schwarze 537: case 'w':
1.206 schwarze 538: term = cp[1];
539: /* FALLTHROUGH */
1.207 schwarze 540: case 'n':
1.181 schwarze 541: res = ubuf;
542: break;
543: default:
1.205 schwarze 544: if (ESCAPE_ERROR == mandoc_escape(&cp, NULL, NULL))
545: mandoc_msg(MANDOCERR_BADESCAPE, r->parse,
546: ln, (int)(stesc - *bufp), NULL);
547: continue;
1.152 kristaps 548: }
549:
1.205 schwarze 550: if (EXPAND_LIMIT < ++expand_count) {
551: mandoc_msg(MANDOCERR_ROFFLOOP, r->parse,
552: ln, (int)(stesc - *bufp), NULL);
553: return(ROFF_IGN);
554: }
1.108 schwarze 555:
556: /*
557: * The third character decides the length
1.181 schwarze 558: * of the name of the string or register.
1.108 schwarze 559: * Save a pointer to the name.
560: */
561:
1.206 schwarze 562: if ('\0' == term) {
563: switch (*++cp) {
1.207 schwarze 564: case '\0':
1.206 schwarze 565: maxl = 0;
566: break;
1.207 schwarze 567: case '(':
1.206 schwarze 568: cp++;
569: maxl = 2;
570: break;
1.207 schwarze 571: case '[':
1.206 schwarze 572: cp++;
573: term = ']';
574: maxl = 0;
575: break;
576: default:
577: maxl = 1;
578: break;
579: }
580: } else {
581: cp += 2;
1.94 kristaps 582: maxl = 0;
583: }
1.108 schwarze 584: stnam = cp;
1.94 kristaps 585:
1.108 schwarze 586: /* Advance to the end of the name. */
1.94 kristaps 587:
1.181 schwarze 588: for (naml = 0; 0 == maxl || naml < maxl; naml++, cp++) {
1.153 kristaps 589: if ('\0' == *cp) {
1.207 schwarze 590: mandoc_msg(MANDOCERR_BADESCAPE, r->parse,
591: ln, (int)(stesc - *bufp), NULL);
1.206 schwarze 592: break;
1.153 kristaps 593: }
1.206 schwarze 594: if (0 == maxl && *cp == term) {
595: cp++;
1.94 kristaps 596: break;
1.206 schwarze 597: }
1.94 kristaps 598: }
599:
1.108 schwarze 600: /*
601: * Retrieve the replacement string; if it is
602: * undefined, resume searching for escapes.
603: */
604:
1.206 schwarze 605: switch (stesc[1]) {
1.207 schwarze 606: case '*':
1.181 schwarze 607: res = roff_getstrn(r, stnam, naml);
1.206 schwarze 608: break;
1.207 schwarze 609: case 'B':
1.206 schwarze 610: npos = 0;
611: irc = roff_evalnum(stnam, &npos, NULL, 0);
612: ubuf[0] = irc && stnam + npos + 1 == cp
613: ? '1' : '0';
614: ubuf[1] = '\0';
615: break;
1.207 schwarze 616: case 'n':
1.208 schwarze 617: (void)snprintf(ubuf, sizeof(ubuf), "%d",
1.181 schwarze 618: roff_getregn(r, stnam, naml));
1.206 schwarze 619: break;
1.207 schwarze 620: case 'w':
1.208 schwarze 621: (void)snprintf(ubuf, sizeof(ubuf), "%d",
1.206 schwarze 622: 24 * (int)naml);
623: break;
624: }
1.94 kristaps 625:
626: if (NULL == res) {
1.207 schwarze 627: mandoc_msg(MANDOCERR_BADESCAPE, r->parse,
628: ln, (int)(stesc - *bufp), NULL);
1.142 kristaps 629: res = "";
1.94 kristaps 630: }
631:
1.108 schwarze 632: /* Replace the escape sequence by the string. */
633:
1.209 schwarze 634: *stesc = '\0';
635: *szp = mandoc_asprintf(&nbuf, "%s%s%s",
636: *bufp, res, cp) + 1;
1.94 kristaps 637:
1.205 schwarze 638: /* Prepare for the next replacement. */
1.94 kristaps 639:
1.205 schwarze 640: start = nbuf + pos;
1.209 schwarze 641: stesc = nbuf + (stesc - *bufp) + strlen(res);
1.94 kristaps 642: free(*bufp);
1.181 schwarze 643: *bufp = nbuf;
1.154 kristaps 644: }
1.172 schwarze 645: return(ROFF_CONT);
1.154 kristaps 646: }
647:
648: /*
1.178 schwarze 649: * Process text streams:
650: * Convert all breakable hyphens into ASCII_HYPH.
651: * Decrement and spring input line trap.
1.154 kristaps 652: */
653: static enum rofferr
1.178 schwarze 654: roff_parsetext(char **bufp, size_t *szp, int pos, int *offs)
1.154 kristaps 655: {
656: size_t sz;
657: const char *start;
1.178 schwarze 658: char *p;
659: int isz;
1.154 kristaps 660: enum mandoc_esc esc;
661:
1.178 schwarze 662: start = p = *bufp + pos;
1.154 kristaps 663:
664: while ('\0' != *p) {
665: sz = strcspn(p, "-\\");
666: p += sz;
667:
1.159 kristaps 668: if ('\0' == *p)
669: break;
670:
1.154 kristaps 671: if ('\\' == *p) {
672: /* Skip over escapes. */
673: p++;
1.189 schwarze 674: esc = mandoc_escape((const char **)&p, NULL, NULL);
1.154 kristaps 675: if (ESCAPE_ERROR == esc)
676: break;
1.155 kristaps 677: continue;
1.159 kristaps 678: } else if (p == start) {
1.158 kristaps 679: p++;
1.155 kristaps 680: continue;
1.158 kristaps 681: }
1.155 kristaps 682:
1.171 schwarze 683: if (isalpha((unsigned char)p[-1]) &&
684: isalpha((unsigned char)p[1]))
1.155 kristaps 685: *p = ASCII_HYPH;
686: p++;
1.94 kristaps 687: }
688:
1.178 schwarze 689: /* Spring the input line trap. */
690: if (1 == roffit_lines) {
1.202 schwarze 691: isz = mandoc_asprintf(&p, "%s\n.%s", *bufp, roffit_macro);
1.178 schwarze 692: free(*bufp);
693: *bufp = p;
694: *szp = isz + 1;
695: *offs = 0;
696: free(roffit_macro);
697: roffit_lines = 0;
698: return(ROFF_REPARSE);
699: } else if (1 < roffit_lines)
700: --roffit_lines;
1.154 kristaps 701: return(ROFF_CONT);
1.94 kristaps 702: }
703:
1.67 kristaps 704: enum rofferr
1.207 schwarze 705: roff_parseln(struct roff *r, int ln, char **bufp,
1.90 kristaps 706: size_t *szp, int pos, int *offs)
1.67 kristaps 707: {
708: enum rofft t;
1.109 kristaps 709: enum rofferr e;
1.130 kristaps 710: int ppos, ctl;
1.79 kristaps 711:
712: /*
1.94 kristaps 713: * Run the reserved-word filter only if we have some reserved
714: * words to fill in.
715: */
716:
1.172 schwarze 717: e = roff_res(r, bufp, szp, ln, pos);
718: if (ROFF_IGN == e)
719: return(e);
720: assert(ROFF_CONT == e);
1.94 kristaps 721:
1.130 kristaps 722: ppos = pos;
1.174 kristaps 723: ctl = roff_getcontrol(r, *bufp, &pos);
1.130 kristaps 724:
1.94 kristaps 725: /*
1.79 kristaps 726: * First, if a scope is open and we're not a macro, pass the
727: * text through the macro's filter. If a scope isn't open and
728: * we're not a macro, just let it through.
1.125 kristaps 729: * Finally, if there's an equation scope open, divert it into it
730: * no matter our state.
1.79 kristaps 731: */
1.74 kristaps 732:
1.130 kristaps 733: if (r->last && ! ctl) {
1.78 kristaps 734: t = r->last->tok;
735: assert(roffs[t].text);
1.207 schwarze 736: e = (*roffs[t].text)(r, t, bufp, szp, ln, pos, pos, offs);
1.109 kristaps 737: assert(ROFF_IGN == e || ROFF_CONT == e);
1.125 kristaps 738: if (ROFF_CONT != e)
739: return(e);
1.182 schwarze 740: }
741: if (r->eqn)
742: return(eqn_read(&r->eqn, ln, *bufp, ppos, offs));
743: if ( ! ctl) {
1.125 kristaps 744: if (r->tbl)
1.130 kristaps 745: return(tbl_read(r->tbl, ln, *bufp, pos));
1.178 schwarze 746: return(roff_parsetext(bufp, szp, pos, offs));
1.182 schwarze 747: }
1.67 kristaps 748:
1.79 kristaps 749: /*
750: * If a scope is open, go to the child handler for that macro,
751: * as it may want to preprocess before doing anything with it.
1.125 kristaps 752: * Don't do so if an equation is open.
1.79 kristaps 753: */
1.78 kristaps 754:
1.79 kristaps 755: if (r->last) {
756: t = r->last->tok;
757: assert(roffs[t].sub);
1.207 schwarze 758: return((*roffs[t].sub)(r, t, bufp, szp,
759: ln, ppos, pos, offs));
1.79 kristaps 760: }
1.78 kristaps 761:
1.79 kristaps 762: /*
763: * Lastly, as we've no scope open, try to look up and execute
764: * the new macro. If no macro is found, simply return and let
765: * the compilers handle it.
766: */
1.67 kristaps 767:
1.214 schwarze 768: if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos, ln, ppos)))
1.79 kristaps 769: return(ROFF_CONT);
1.67 kristaps 770:
1.75 kristaps 771: assert(roffs[t].proc);
1.207 schwarze 772: return((*roffs[t].proc)(r, t, bufp, szp, ln, ppos, pos, offs));
1.74 kristaps 773: }
774:
1.117 kristaps 775: void
1.74 kristaps 776: roff_endparse(struct roff *r)
777: {
778:
1.110 kristaps 779: if (r->last)
1.128 kristaps 780: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.207 schwarze 781: r->last->line, r->last->col, NULL);
1.117 kristaps 782:
1.125 kristaps 783: if (r->eqn) {
1.207 schwarze 784: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
785: r->eqn->eqn.ln, r->eqn->eqn.pos, NULL);
1.151 kristaps 786: eqn_end(&r->eqn);
1.125 kristaps 787: }
788:
1.117 kristaps 789: if (r->tbl) {
1.207 schwarze 790: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
791: r->tbl->line, r->tbl->pos, NULL);
1.151 kristaps 792: tbl_end(&r->tbl);
1.117 kristaps 793: }
1.67 kristaps 794: }
795:
796: /*
797: * Parse a roff node's type from the input buffer. This must be in the
798: * form of ".foo xxx" in the usual way.
799: */
800: static enum rofft
1.214 schwarze 801: roff_parse(struct roff *r, char *buf, int *pos, int ln, int ppos)
1.67 kristaps 802: {
1.214 schwarze 803: char *cp;
1.106 kristaps 804: const char *mac;
805: size_t maclen;
1.67 kristaps 806: enum rofft t;
807:
1.214 schwarze 808: cp = buf + *pos;
809:
810: if ('\0' == *cp || '"' == *cp || '\t' == *cp || ' ' == *cp)
1.67 kristaps 811: return(ROFF_MAX);
812:
1.214 schwarze 813: mac = cp;
814: maclen = roff_getname(r, &cp, ln, ppos);
1.67 kristaps 815:
1.106 kristaps 816: t = (r->current_string = roff_getstrn(r, mac, maclen))
1.155 kristaps 817: ? ROFF_USERDEF : roffhash_find(mac, maclen);
1.67 kristaps 818:
1.214 schwarze 819: if (ROFF_MAX != t)
820: *pos = cp - buf;
1.67 kristaps 821:
822: return(t);
823: }
824:
825: static enum rofferr
1.76 kristaps 826: roff_cblock(ROFF_ARGS)
1.67 kristaps 827: {
828:
1.79 kristaps 829: /*
830: * A block-close `..' should only be invoked as a child of an
831: * ignore macro, otherwise raise a warning and just ignore it.
832: */
833:
1.76 kristaps 834: if (NULL == r->last) {
1.128 kristaps 835: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.76 kristaps 836: return(ROFF_IGN);
837: }
1.67 kristaps 838:
1.81 kristaps 839: switch (r->last->tok) {
1.207 schwarze 840: case ROFF_am:
1.81 kristaps 841: /* FALLTHROUGH */
1.207 schwarze 842: case ROFF_ami:
1.81 kristaps 843: /* FALLTHROUGH */
1.207 schwarze 844: case ROFF_am1:
1.81 kristaps 845: /* FALLTHROUGH */
1.207 schwarze 846: case ROFF_de:
1.108 schwarze 847: /* ROFF_de1 is remapped to ROFF_de in roff_block(). */
1.81 kristaps 848: /* FALLTHROUGH */
1.207 schwarze 849: case ROFF_dei:
1.81 kristaps 850: /* FALLTHROUGH */
1.207 schwarze 851: case ROFF_ig:
1.81 kristaps 852: break;
853: default:
1.128 kristaps 854: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.67 kristaps 855: return(ROFF_IGN);
1.76 kristaps 856: }
1.67 kristaps 857:
1.76 kristaps 858: if ((*bufp)[pos])
1.217 ! schwarze 859: mandoc_vmsg(MANDOCERR_ARG_SKIP, r->parse, ln, pos,
! 860: ".. %s", *bufp + pos);
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: {
1.213 schwarze 913: char *name, *cp;
914: size_t namesz;
1.106 kristaps 915:
1.213 schwarze 916: name = cp = *bufp + pos;
917: namesz = 0;
1.76 kristaps 918:
1.106 kristaps 919: if (ROFF_ig != tok) {
1.213 schwarze 920: if ('\0' == *cp) {
1.216 schwarze 921: mandoc_msg(MANDOCERR_REQ_EMPTY, r->parse,
922: ln, ppos, roffs[tok].name);
1.106 kristaps 923: return(ROFF_IGN);
924: }
1.107 kristaps 925:
926: /*
927: * Re-write `de1', since we don't really care about
928: * groff's strange compatibility mode, into `de'.
929: */
930:
1.106 kristaps 931: if (ROFF_de1 == tok)
932: tok = ROFF_de;
1.213 schwarze 933: else if (ROFF_de != tok)
1.128 kristaps 934: mandoc_msg(MANDOCERR_REQUEST, r->parse, ln, ppos,
1.106 kristaps 935: roffs[tok].name);
1.107 kristaps 936:
1.213 schwarze 937: namesz = roff_getname(r, &cp, ln, ppos);
938: name[namesz] = '\0';
939: } else
940: name = NULL;
1.80 kristaps 941:
1.106 kristaps 942: roffnode_push(r, tok, name, ln, ppos);
943:
944: /*
945: * At the beginning of a `de' macro, clear the existing string
946: * with the same name, if there is one. New content will be
1.193 schwarze 947: * appended from roff_block_text() in multiline mode.
1.106 kristaps 948: */
1.107 kristaps 949:
1.213 schwarze 950: if (namesz && ROFF_de == tok)
951: roff_setstrn(&r->strtab, name, namesz, "", 0, 0);
1.76 kristaps 952:
1.213 schwarze 953: if ('\0' == *cp)
1.78 kristaps 954: return(ROFF_IGN);
955:
1.107 kristaps 956: /* If present, process the custom end-of-line marker. */
957:
1.213 schwarze 958: name = cp;
959: namesz = roff_getname(r, &cp, ln, ppos);
960: if (namesz)
961: r->last->end = mandoc_strndup(name, namesz);
1.78 kristaps 962:
1.213 schwarze 963: if ('\0' != *cp)
1.217 ! schwarze 964: mandoc_vmsg(MANDOCERR_ARG_EXCESS, r->parse,
! 965: ln, pos, ".%s ... %s", roffs[tok].name, cp);
1.74 kristaps 966:
1.78 kristaps 967: return(ROFF_IGN);
968: }
969:
970: static enum rofferr
1.80 kristaps 971: roff_block_sub(ROFF_ARGS)
1.79 kristaps 972: {
973: enum rofft t;
974: int i, j;
975:
976: /*
977: * First check whether a custom macro exists at this level. If
978: * it does, then check against it. This is some of groff's
979: * stranger behaviours. If we encountered a custom end-scope
980: * tag and that tag also happens to be a "real" macro, then we
981: * need to try interpreting it again as a real macro. If it's
982: * not, then return ignore. Else continue.
983: */
984:
985: if (r->last->end) {
1.130 kristaps 986: for (i = pos, j = 0; r->last->end[j]; j++, i++)
1.79 kristaps 987: if ((*bufp)[i] != r->last->end[j])
988: break;
989:
1.207 schwarze 990: if ('\0' == r->last->end[j] &&
991: ('\0' == (*bufp)[i] ||
992: ' ' == (*bufp)[i] ||
993: '\t' == (*bufp)[i])) {
1.79 kristaps 994: roffnode_pop(r);
995: roffnode_cleanscope(r);
996:
1.130 kristaps 997: while (' ' == (*bufp)[i] || '\t' == (*bufp)[i])
998: i++;
999:
1000: pos = i;
1.214 schwarze 1001: if (ROFF_MAX != roff_parse(r, *bufp, &pos, ln, ppos))
1.79 kristaps 1002: return(ROFF_RERUN);
1003: return(ROFF_IGN);
1004: }
1005: }
1006:
1007: /*
1008: * If we have no custom end-query or lookup failed, then try
1009: * pulling it out of the hashtable.
1010: */
1011:
1.214 schwarze 1012: t = roff_parse(r, *bufp, &pos, ln, ppos);
1.79 kristaps 1013:
1.106 kristaps 1014: /*
1015: * Macros other than block-end are only significant
1016: * in `de' blocks; elsewhere, simply throw them away.
1017: */
1018: if (ROFF_cblock != t) {
1019: if (ROFF_de == tok)
1.193 schwarze 1020: roff_setstr(r, r->last->name, *bufp + ppos, 2);
1.79 kristaps 1021: return(ROFF_IGN);
1.106 kristaps 1022: }
1.79 kristaps 1023:
1024: assert(roffs[t].proc);
1.207 schwarze 1025: return((*roffs[t].proc)(r, t, bufp, szp, ln, ppos, pos, offs));
1.79 kristaps 1026: }
1027:
1028: static enum rofferr
1.80 kristaps 1029: roff_block_text(ROFF_ARGS)
1.78 kristaps 1030: {
1031:
1.106 kristaps 1032: if (ROFF_de == tok)
1.193 schwarze 1033: roff_setstr(r, r->last->name, *bufp + pos, 2);
1.106 kristaps 1034:
1.78 kristaps 1035: return(ROFF_IGN);
1036: }
1037:
1038: static enum rofferr
1.82 kristaps 1039: roff_cond_sub(ROFF_ARGS)
1040: {
1041: enum rofft t;
1.139 kristaps 1042: char *ep;
1.198 schwarze 1043: int rr;
1.82 kristaps 1044:
1045: rr = r->last->rule;
1.139 kristaps 1046: roffnode_cleanscope(r);
1.214 schwarze 1047: t = roff_parse(r, *bufp, &pos, ln, ppos);
1.82 kristaps 1048:
1.139 kristaps 1049: /*
1.177 schwarze 1050: * Fully handle known macros when they are structurally
1051: * required or when the conditional evaluated to true.
1.87 kristaps 1052: */
1053:
1.177 schwarze 1054: if ((ROFF_MAX != t) &&
1.198 schwarze 1055: (rr || ROFFMAC_STRUCT & roffs[t].flags)) {
1.177 schwarze 1056: assert(roffs[t].proc);
1057: return((*roffs[t].proc)(r, t, bufp, szp,
1.207 schwarze 1058: ln, ppos, pos, offs));
1.177 schwarze 1059: }
1.144 kristaps 1060:
1.196 schwarze 1061: /*
1062: * If `\}' occurs on a macro line without a preceding macro,
1063: * drop the line completely.
1064: */
1065:
1066: ep = *bufp + pos;
1067: if ('\\' == ep[0] && '}' == ep[1])
1.198 schwarze 1068: rr = 0;
1.196 schwarze 1069:
1.177 schwarze 1070: /* Always check for the closing delimiter `\}'. */
1.144 kristaps 1071:
1.177 schwarze 1072: while (NULL != (ep = strchr(ep, '\\'))) {
1.197 schwarze 1073: if ('}' == *(++ep)) {
1074: *ep = '&';
1075: roff_ccond(r, ln, ep - *bufp - 1);
1076: }
1077: ++ep;
1.177 schwarze 1078: }
1.198 schwarze 1079: return(rr ? ROFF_CONT : ROFF_IGN);
1.82 kristaps 1080: }
1081:
1082: static enum rofferr
1083: roff_cond_text(ROFF_ARGS)
1.78 kristaps 1084: {
1.140 kristaps 1085: char *ep;
1.198 schwarze 1086: int rr;
1.82 kristaps 1087:
1088: rr = r->last->rule;
1.140 kristaps 1089: roffnode_cleanscope(r);
1.82 kristaps 1090:
1.197 schwarze 1091: ep = *bufp + pos;
1092: while (NULL != (ep = strchr(ep, '\\'))) {
1093: if ('}' == *(++ep)) {
1094: *ep = '&';
1095: roff_ccond(r, ln, ep - *bufp - 1);
1096: }
1097: ++ep;
1.78 kristaps 1098: }
1.198 schwarze 1099: return(rr ? ROFF_CONT : ROFF_IGN);
1.74 kristaps 1100: }
1101:
1.204 schwarze 1102: /*
1103: * Parse a single signed integer number. Stop at the first non-digit.
1104: * If there is at least one digit, return success and advance the
1105: * parse point, else return failure and let the parse point unchanged.
1106: * Ignore overflows, treat them just like the C language.
1107: */
1.184 schwarze 1108: static int
1109: roff_getnum(const char *v, int *pos, int *res)
1110: {
1.206 schwarze 1111: int myres, n, p;
1112:
1113: if (NULL == res)
1114: res = &myres;
1.184 schwarze 1115:
1116: p = *pos;
1117: n = v[p] == '-';
1118: if (n)
1119: p++;
1120:
1121: for (*res = 0; isdigit((unsigned char)v[p]); p++)
1.204 schwarze 1122: *res = 10 * *res + v[p] - '0';
1.184 schwarze 1123: if (p == *pos + n)
1124: return 0;
1125:
1126: if (n)
1127: *res = -*res;
1128:
1129: *pos = p;
1130: return 1;
1131: }
1132:
1.198 schwarze 1133: /*
1134: * Evaluate a string comparison condition.
1135: * The first character is the delimiter.
1136: * Succeed if the string up to its second occurrence
1137: * matches the string up to its third occurence.
1138: * Advance the cursor after the third occurrence
1139: * or lacking that, to the end of the line.
1140: */
1141: static int
1142: roff_evalstrcond(const char *v, int *pos)
1143: {
1144: const char *s1, *s2, *s3;
1145: int match;
1146:
1147: match = 0;
1148: s1 = v + *pos; /* initial delimiter */
1149: s2 = s1 + 1; /* for scanning the first string */
1150: s3 = strchr(s2, *s1); /* for scanning the second string */
1151:
1152: if (NULL == s3) /* found no middle delimiter */
1153: goto out;
1154:
1155: while ('\0' != *++s3) {
1156: if (*s2 != *s3) { /* mismatch */
1157: s3 = strchr(s3, *s1);
1158: break;
1159: }
1160: if (*s3 == *s1) { /* found the final delimiter */
1161: match = 1;
1162: break;
1163: }
1164: s2++;
1165: }
1166:
1167: out:
1168: if (NULL == s3)
1169: s3 = strchr(s2, '\0');
1170: else
1171: s3++;
1172: *pos = s3 - v;
1173: return(match);
1174: }
1175:
1.204 schwarze 1176: /*
1177: * Evaluate an optionally negated single character, numerical,
1178: * or string condition.
1179: */
1.198 schwarze 1180: static int
1.88 kristaps 1181: roff_evalcond(const char *v, int *pos)
1182: {
1.204 schwarze 1183: int wanttrue, number;
1.88 kristaps 1184:
1.198 schwarze 1185: if ('!' == v[*pos]) {
1186: wanttrue = 0;
1187: (*pos)++;
1188: } else
1189: wanttrue = 1;
1190:
1.88 kristaps 1191: switch (v[*pos]) {
1.207 schwarze 1192: case 'n':
1.198 schwarze 1193: /* FALLTHROUGH */
1.207 schwarze 1194: case 'o':
1.88 kristaps 1195: (*pos)++;
1.198 schwarze 1196: return(wanttrue);
1.207 schwarze 1197: case 'c':
1.198 schwarze 1198: /* FALLTHROUGH */
1.207 schwarze 1199: case 'd':
1.198 schwarze 1200: /* FALLTHROUGH */
1.207 schwarze 1201: case 'e':
1.88 kristaps 1202: /* FALLTHROUGH */
1.207 schwarze 1203: case 'r':
1.88 kristaps 1204: /* FALLTHROUGH */
1.207 schwarze 1205: case 't':
1.88 kristaps 1206: (*pos)++;
1.198 schwarze 1207: return(!wanttrue);
1.88 kristaps 1208: default:
1209: break;
1210: }
1211:
1.204 schwarze 1212: if (roff_evalnum(v, pos, &number, 0))
1213: return((number > 0) == wanttrue);
1214: else
1.198 schwarze 1215: return(roff_evalstrcond(v, pos) == wanttrue);
1.88 kristaps 1216: }
1217:
1.74 kristaps 1218: static enum rofferr
1.103 kristaps 1219: roff_line_ignore(ROFF_ARGS)
1.89 kristaps 1220: {
1.123 schwarze 1221:
1.89 kristaps 1222: return(ROFF_IGN);
1223: }
1224:
1.104 kristaps 1225: static enum rofferr
1.82 kristaps 1226: roff_cond(ROFF_ARGS)
1.74 kristaps 1227: {
1.173 schwarze 1228:
1229: roffnode_push(r, tok, NULL, ln, ppos);
1.74 kristaps 1230:
1.207 schwarze 1231: /*
1.134 kristaps 1232: * An `.el' has no conditional body: it will consume the value
1233: * of the current rstack entry set in prior `ie' calls or
1.207 schwarze 1234: * defaults to DENY.
1.134 kristaps 1235: *
1236: * If we're not an `el', however, then evaluate the conditional.
1237: */
1.133 kristaps 1238:
1.173 schwarze 1239: r->last->rule = ROFF_el == tok ?
1.207 schwarze 1240: (r->rstackpos < 0 ? 0 : r->rstack[r->rstackpos--]) :
1241: roff_evalcond(*bufp, &pos);
1.77 kristaps 1242:
1.134 kristaps 1243: /*
1244: * An if-else will put the NEGATION of the current evaluated
1245: * conditional into the stack of rules.
1246: */
1247:
1.84 schwarze 1248: if (ROFF_ie == tok) {
1.134 kristaps 1249: if (r->rstackpos == RSTACK_MAX - 1) {
1.207 schwarze 1250: mandoc_msg(MANDOCERR_MEM,
1251: r->parse, ln, ppos, NULL);
1.134 kristaps 1252: return(ROFF_ERR);
1253: }
1.198 schwarze 1254: r->rstack[++r->rstackpos] = !r->last->rule;
1.82 kristaps 1255: }
1.88 kristaps 1256:
1257: /* If the parent has false as its rule, then so do we. */
1258:
1.198 schwarze 1259: if (r->last->parent && !r->last->parent->rule)
1260: r->last->rule = 0;
1.88 kristaps 1261:
1262: /*
1.173 schwarze 1263: * Determine scope.
1264: * If there is nothing on the line after the conditional,
1265: * not even whitespace, use next-line scope.
1.88 kristaps 1266: */
1.74 kristaps 1267:
1.173 schwarze 1268: if ('\0' == (*bufp)[pos]) {
1269: r->last->endspan = 2;
1270: goto out;
1271: }
1272:
1273: while (' ' == (*bufp)[pos])
1274: pos++;
1275:
1276: /* An opening brace requests multiline scope. */
1.75 kristaps 1277:
1278: if ('\\' == (*bufp)[pos] && '{' == (*bufp)[pos + 1]) {
1279: r->last->endspan = -1;
1280: pos += 2;
1.173 schwarze 1281: goto out;
1.207 schwarze 1282: }
1.74 kristaps 1283:
1.77 kristaps 1284: /*
1.173 schwarze 1285: * Anything else following the conditional causes
1286: * single-line scope. Warn if the scope contains
1287: * nothing but trailing whitespace.
1.77 kristaps 1288: */
1289:
1.75 kristaps 1290: if ('\0' == (*bufp)[pos])
1.216 schwarze 1291: mandoc_msg(MANDOCERR_COND_EMPTY, r->parse,
1292: ln, ppos, roffs[tok].name);
1.77 kristaps 1293:
1.173 schwarze 1294: r->last->endspan = 1;
1.74 kristaps 1295:
1.173 schwarze 1296: out:
1.75 kristaps 1297: *offs = pos;
1298: return(ROFF_RERUN);
1.83 schwarze 1299: }
1300:
1301: static enum rofferr
1.92 schwarze 1302: roff_ds(ROFF_ARGS)
1303: {
1.212 schwarze 1304: char *string;
1305: const char *name;
1306: size_t namesz;
1.96 kristaps 1307:
1308: /*
1.212 schwarze 1309: * The first word is the name of the string.
1310: * If it is empty or terminated by an escape sequence,
1311: * abort the `ds' request without defining anything.
1.96 kristaps 1312: */
1.92 schwarze 1313:
1.212 schwarze 1314: name = string = *bufp + pos;
1.92 schwarze 1315: if ('\0' == *name)
1316: return(ROFF_IGN);
1317:
1.212 schwarze 1318: namesz = roff_getname(r, &string, ln, pos);
1319: if ('\\' == name[namesz])
1320: return(ROFF_IGN);
1321:
1322: /* Read past the initial double-quote, if any. */
1.121 schwarze 1323: if ('"' == *string)
1.92 schwarze 1324: string++;
1325:
1.96 kristaps 1326: /* The rest is the value. */
1.212 schwarze 1327: roff_setstrn(&r->strtab, name, namesz, string, strlen(string),
1328: ROFF_as == tok);
1.92 schwarze 1329: return(ROFF_IGN);
1330: }
1331:
1.204 schwarze 1332: /*
1333: * Parse a single operator, one or two characters long.
1334: * If the operator is recognized, return success and advance the
1335: * parse point, else return failure and let the parse point unchanged.
1336: */
1337: static int
1338: roff_getop(const char *v, int *pos, char *res)
1339: {
1340:
1341: *res = v[*pos];
1342:
1343: switch (*res) {
1.207 schwarze 1344: case '+':
1.204 schwarze 1345: /* FALLTHROUGH */
1.207 schwarze 1346: case '-':
1.204 schwarze 1347: /* FALLTHROUGH */
1.207 schwarze 1348: case '*':
1.204 schwarze 1349: /* FALLTHROUGH */
1.207 schwarze 1350: case '/':
1.204 schwarze 1351: /* FALLTHROUGH */
1.207 schwarze 1352: case '%':
1.204 schwarze 1353: /* FALLTHROUGH */
1.207 schwarze 1354: case '&':
1.204 schwarze 1355: /* FALLTHROUGH */
1.207 schwarze 1356: case ':':
1.204 schwarze 1357: break;
1358: case '<':
1359: switch (v[*pos + 1]) {
1.207 schwarze 1360: case '=':
1.204 schwarze 1361: *res = 'l';
1362: (*pos)++;
1363: break;
1.207 schwarze 1364: case '>':
1.204 schwarze 1365: *res = '!';
1366: (*pos)++;
1367: break;
1.207 schwarze 1368: case '?':
1.204 schwarze 1369: *res = 'i';
1370: (*pos)++;
1371: break;
1372: default:
1373: break;
1374: }
1375: break;
1376: case '>':
1377: switch (v[*pos + 1]) {
1.207 schwarze 1378: case '=':
1.204 schwarze 1379: *res = 'g';
1380: (*pos)++;
1381: break;
1.207 schwarze 1382: case '?':
1.204 schwarze 1383: *res = 'a';
1384: (*pos)++;
1385: break;
1386: default:
1387: break;
1388: }
1389: break;
1390: case '=':
1391: if ('=' == v[*pos + 1])
1392: (*pos)++;
1393: break;
1394: default:
1395: return(0);
1396: }
1397: (*pos)++;
1398:
1399: return(*res);
1400: }
1401:
1402: /*
1403: * Evaluate either a parenthesized numeric expression
1404: * or a single signed integer number.
1405: */
1406: static int
1407: roff_evalpar(const char *v, int *pos, int *res)
1408: {
1409:
1410: if ('(' != v[*pos])
1411: return(roff_getnum(v, pos, res));
1412:
1413: (*pos)++;
1414: if ( ! roff_evalnum(v, pos, res, 1))
1415: return(0);
1416:
1.206 schwarze 1417: /*
1418: * Omission of the closing parenthesis
1419: * is an error in validation mode,
1420: * but ignored in evaluation mode.
1421: */
1422:
1.204 schwarze 1423: if (')' == v[*pos])
1424: (*pos)++;
1.206 schwarze 1425: else if (NULL == res)
1426: return(0);
1.204 schwarze 1427:
1428: return(1);
1429: }
1430:
1431: /*
1432: * Evaluate a complete numeric expression.
1433: * Proceed left to right, there is no concept of precedence.
1434: */
1435: static int
1436: roff_evalnum(const char *v, int *pos, int *res, int skipwhite)
1437: {
1438: int mypos, operand2;
1439: char operator;
1440:
1441: if (NULL == pos) {
1442: mypos = 0;
1443: pos = &mypos;
1444: }
1445:
1446: if (skipwhite)
1447: while (isspace((unsigned char)v[*pos]))
1448: (*pos)++;
1449:
1450: if ( ! roff_evalpar(v, pos, res))
1451: return(0);
1452:
1453: while (1) {
1454: if (skipwhite)
1455: while (isspace((unsigned char)v[*pos]))
1456: (*pos)++;
1457:
1458: if ( ! roff_getop(v, pos, &operator))
1459: break;
1460:
1461: if (skipwhite)
1462: while (isspace((unsigned char)v[*pos]))
1463: (*pos)++;
1464:
1465: if ( ! roff_evalpar(v, pos, &operand2))
1466: return(0);
1467:
1468: if (skipwhite)
1469: while (isspace((unsigned char)v[*pos]))
1470: (*pos)++;
1.206 schwarze 1471:
1472: if (NULL == res)
1473: continue;
1.204 schwarze 1474:
1475: switch (operator) {
1.207 schwarze 1476: case '+':
1.204 schwarze 1477: *res += operand2;
1478: break;
1.207 schwarze 1479: case '-':
1.204 schwarze 1480: *res -= operand2;
1481: break;
1.207 schwarze 1482: case '*':
1.204 schwarze 1483: *res *= operand2;
1484: break;
1.207 schwarze 1485: case '/':
1.204 schwarze 1486: *res /= operand2;
1487: break;
1.207 schwarze 1488: case '%':
1.204 schwarze 1489: *res %= operand2;
1490: break;
1.207 schwarze 1491: case '<':
1.204 schwarze 1492: *res = *res < operand2;
1493: break;
1.207 schwarze 1494: case '>':
1.204 schwarze 1495: *res = *res > operand2;
1496: break;
1.207 schwarze 1497: case 'l':
1.204 schwarze 1498: *res = *res <= operand2;
1499: break;
1.207 schwarze 1500: case 'g':
1.204 schwarze 1501: *res = *res >= operand2;
1502: break;
1.207 schwarze 1503: case '=':
1.204 schwarze 1504: *res = *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 ':':
1.204 schwarze 1513: *res = *res || operand2;
1514: break;
1.207 schwarze 1515: case 'i':
1.204 schwarze 1516: if (operand2 < *res)
1517: *res = operand2;
1518: break;
1.207 schwarze 1519: case 'a':
1.204 schwarze 1520: if (operand2 > *res)
1521: *res = operand2;
1522: break;
1523: default:
1524: abort();
1525: }
1526: }
1527: return(1);
1528: }
1529:
1.180 schwarze 1530: void
1.187 schwarze 1531: roff_setreg(struct roff *r, const char *name, int val, char sign)
1.147 kristaps 1532: {
1.180 schwarze 1533: struct roffreg *reg;
1534:
1535: /* Search for an existing register with the same name. */
1536: reg = r->regtab;
1537:
1538: while (reg && strcmp(name, reg->key.p))
1539: reg = reg->next;
1.147 kristaps 1540:
1.180 schwarze 1541: if (NULL == reg) {
1542: /* Create a new register. */
1543: reg = mandoc_malloc(sizeof(struct roffreg));
1544: reg->key.p = mandoc_strdup(name);
1545: reg->key.sz = strlen(name);
1.187 schwarze 1546: reg->val = 0;
1.180 schwarze 1547: reg->next = r->regtab;
1548: r->regtab = reg;
1549: }
1550:
1.187 schwarze 1551: if ('+' == sign)
1552: reg->val += val;
1553: else if ('-' == sign)
1554: reg->val -= val;
1555: else
1556: reg->val = val;
1.147 kristaps 1557: }
1558:
1.192 schwarze 1559: /*
1560: * Handle some predefined read-only number registers.
1561: * For now, return -1 if the requested register is not predefined;
1562: * in case a predefined read-only register having the value -1
1563: * were to turn up, another special value would have to be chosen.
1564: */
1565: static int
1566: roff_getregro(const char *name)
1567: {
1568:
1569: switch (*name) {
1.207 schwarze 1570: case 'A': /* ASCII approximation mode is always off. */
1.192 schwarze 1571: return(0);
1.207 schwarze 1572: case 'g': /* Groff compatibility mode is always on. */
1.192 schwarze 1573: return(1);
1.207 schwarze 1574: case 'H': /* Fixed horizontal resolution. */
1.192 schwarze 1575: return (24);
1.207 schwarze 1576: case 'j': /* Always adjust left margin only. */
1.192 schwarze 1577: return(0);
1.207 schwarze 1578: case 'T': /* Some output device is always defined. */
1.192 schwarze 1579: return(1);
1.207 schwarze 1580: case 'V': /* Fixed vertical resolution. */
1.192 schwarze 1581: return (40);
1582: default:
1583: return (-1);
1584: }
1585: }
1586:
1.181 schwarze 1587: int
1.180 schwarze 1588: roff_getreg(const struct roff *r, const char *name)
1.147 kristaps 1589: {
1.180 schwarze 1590: struct roffreg *reg;
1.192 schwarze 1591: int val;
1592:
1593: if ('.' == name[0] && '\0' != name[1] && '\0' == name[2]) {
1594: val = roff_getregro(name + 1);
1595: if (-1 != val)
1596: return (val);
1597: }
1.180 schwarze 1598:
1599: for (reg = r->regtab; reg; reg = reg->next)
1600: if (0 == strcmp(name, reg->key.p))
1.181 schwarze 1601: return(reg->val);
1602:
1603: return(0);
1604: }
1605:
1606: static int
1607: roff_getregn(const struct roff *r, const char *name, size_t len)
1608: {
1609: struct roffreg *reg;
1.192 schwarze 1610: int val;
1611:
1612: if ('.' == name[0] && 2 == len) {
1613: val = roff_getregro(name + 1);
1614: if (-1 != val)
1615: return (val);
1616: }
1.181 schwarze 1617:
1618: for (reg = r->regtab; reg; reg = reg->next)
1619: if (len == reg->key.sz &&
1620: 0 == strncmp(name, reg->key.p, len))
1621: return(reg->val);
1.147 kristaps 1622:
1.180 schwarze 1623: return(0);
1.147 kristaps 1624: }
1625:
1.180 schwarze 1626: static void
1627: roff_freereg(struct roffreg *reg)
1.147 kristaps 1628: {
1.180 schwarze 1629: struct roffreg *old_reg;
1.147 kristaps 1630:
1.180 schwarze 1631: while (NULL != reg) {
1632: free(reg->key.p);
1633: old_reg = reg;
1634: reg = reg->next;
1635: free(old_reg);
1636: }
1.147 kristaps 1637: }
1.92 schwarze 1638:
1639: static enum rofferr
1.89 kristaps 1640: roff_nr(ROFF_ARGS)
1.83 schwarze 1641: {
1.212 schwarze 1642: char *key, *val;
1643: size_t keysz;
1.138 kristaps 1644: int iv;
1.187 schwarze 1645: char sign;
1.89 kristaps 1646:
1.212 schwarze 1647: key = val = *bufp + pos;
1648: if ('\0' == *key)
1649: return(ROFF_IGN);
1650:
1651: keysz = roff_getname(r, &val, ln, pos);
1652: if ('\\' == key[keysz])
1653: return(ROFF_IGN);
1654: key[keysz] = '\0';
1.89 kristaps 1655:
1.187 schwarze 1656: sign = *val;
1657: if ('+' == sign || '-' == sign)
1658: val++;
1659:
1.204 schwarze 1660: if (roff_evalnum(val, NULL, &iv, 0))
1661: roff_setreg(r, key, iv, sign);
1.109 kristaps 1662:
1.203 schwarze 1663: return(ROFF_IGN);
1664: }
1665:
1666: static enum rofferr
1667: roff_rr(ROFF_ARGS)
1668: {
1669: struct roffreg *reg, **prev;
1.212 schwarze 1670: char *name, *cp;
1671: size_t namesz;
1.203 schwarze 1672:
1.212 schwarze 1673: name = cp = *bufp + pos;
1674: if ('\0' == *name)
1675: return(ROFF_IGN);
1676: namesz = roff_getname(r, &cp, ln, pos);
1677: name[namesz] = '\0';
1.203 schwarze 1678:
1679: prev = &r->regtab;
1680: while (1) {
1681: reg = *prev;
1682: if (NULL == reg || !strcmp(name, reg->key.p))
1683: break;
1684: prev = ®->next;
1685: }
1686: if (NULL != reg) {
1687: *prev = reg->next;
1688: free(reg->key.p);
1689: free(reg);
1690: }
1.122 schwarze 1691: return(ROFF_IGN);
1692: }
1693:
1694: static enum rofferr
1695: roff_rm(ROFF_ARGS)
1696: {
1697: const char *name;
1698: char *cp;
1.212 schwarze 1699: size_t namesz;
1.122 schwarze 1700:
1701: cp = *bufp + pos;
1702: while ('\0' != *cp) {
1.212 schwarze 1703: name = cp;
1704: namesz = roff_getname(r, &cp, ln, (int)(cp - *bufp));
1705: roff_setstrn(&r->strtab, name, namesz, NULL, 0, 0);
1706: if ('\\' == name[namesz])
1707: break;
1.122 schwarze 1708: }
1.178 schwarze 1709: return(ROFF_IGN);
1710: }
1711:
1712: static enum rofferr
1713: roff_it(ROFF_ARGS)
1714: {
1715: char *cp;
1716: size_t len;
1717: int iv;
1718:
1719: /* Parse the number of lines. */
1720: cp = *bufp + pos;
1721: len = strcspn(cp, " \t");
1722: cp[len] = '\0';
1723: if ((iv = mandoc_strntoi(cp, len, 10)) <= 0) {
1724: mandoc_msg(MANDOCERR_NUMERIC, r->parse,
1.207 schwarze 1725: ln, ppos, *bufp + 1);
1.178 schwarze 1726: return(ROFF_IGN);
1727: }
1728: cp += len + 1;
1729:
1730: /* Arm the input line trap. */
1731: roffit_lines = iv;
1732: roffit_macro = mandoc_strdup(cp);
1.109 kristaps 1733: return(ROFF_IGN);
1.175 schwarze 1734: }
1735:
1736: static enum rofferr
1737: roff_Dd(ROFF_ARGS)
1738: {
1739: const char *const *cp;
1740:
1.199 schwarze 1741: if (0 == ((MPARSE_MDOC | MPARSE_QUICK) & r->options))
1.175 schwarze 1742: for (cp = __mdoc_reserved; *cp; cp++)
1743: roff_setstr(r, *cp, NULL, 0);
1744:
1745: return(ROFF_CONT);
1746: }
1747:
1748: static enum rofferr
1749: roff_TH(ROFF_ARGS)
1750: {
1751: const char *const *cp;
1752:
1.199 schwarze 1753: if (0 == (MPARSE_QUICK & r->options))
1.175 schwarze 1754: for (cp = __man_reserved; *cp; cp++)
1755: roff_setstr(r, *cp, NULL, 0);
1756:
1757: return(ROFF_CONT);
1.109 kristaps 1758: }
1759:
1760: static enum rofferr
1761: roff_TE(ROFF_ARGS)
1762: {
1763:
1764: if (NULL == r->tbl)
1.128 kristaps 1765: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.115 kristaps 1766: else
1.151 kristaps 1767: tbl_end(&r->tbl);
1.109 kristaps 1768:
1.112 kristaps 1769: return(ROFF_IGN);
1770: }
1771:
1772: static enum rofferr
1773: roff_T_(ROFF_ARGS)
1774: {
1775:
1776: if (NULL == r->tbl)
1.128 kristaps 1777: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.112 kristaps 1778: else
1.116 kristaps 1779: tbl_restart(ppos, ln, r->tbl);
1.112 kristaps 1780:
1.109 kristaps 1781: return(ROFF_IGN);
1782: }
1783:
1.156 kristaps 1784: #if 0
1785: static int
1.151 kristaps 1786: roff_closeeqn(struct roff *r)
1787: {
1788:
1789: return(r->eqn && ROFF_EQN == eqn_end(&r->eqn) ? 1 : 0);
1790: }
1.156 kristaps 1791: #endif
1.151 kristaps 1792:
1.156 kristaps 1793: static void
1.207 schwarze 1794: roff_openeqn(struct roff *r, const char *name, int line,
1.151 kristaps 1795: int offs, const char *buf)
1.125 kristaps 1796: {
1.151 kristaps 1797: struct eqn_node *e;
1798: int poff;
1.125 kristaps 1799:
1800: assert(NULL == r->eqn);
1.151 kristaps 1801: e = eqn_alloc(name, offs, line, r->parse);
1.125 kristaps 1802:
1803: if (r->last_eqn)
1804: r->last_eqn->next = e;
1805: else
1806: r->first_eqn = r->last_eqn = e;
1807:
1808: r->eqn = r->last_eqn = e;
1.151 kristaps 1809:
1810: if (buf) {
1811: poff = 0;
1812: eqn_read(&r->eqn, line, buf, offs, &poff);
1813: }
1814: }
1815:
1816: static enum rofferr
1817: roff_EQ(ROFF_ARGS)
1818: {
1819:
1820: roff_openeqn(r, *bufp + pos, ln, ppos, NULL);
1.125 kristaps 1821: return(ROFF_IGN);
1822: }
1823:
1824: static enum rofferr
1825: roff_EN(ROFF_ARGS)
1826: {
1827:
1.128 kristaps 1828: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.125 kristaps 1829: return(ROFF_IGN);
1830: }
1831:
1832: static enum rofferr
1.109 kristaps 1833: roff_TS(ROFF_ARGS)
1834: {
1.176 schwarze 1835: struct tbl_node *tbl;
1.89 kristaps 1836:
1.115 kristaps 1837: if (r->tbl) {
1.128 kristaps 1838: mandoc_msg(MANDOCERR_SCOPEBROKEN, r->parse, ln, ppos, NULL);
1.151 kristaps 1839: tbl_end(&r->tbl);
1.115 kristaps 1840: }
1.83 schwarze 1841:
1.176 schwarze 1842: tbl = tbl_alloc(ppos, ln, r->parse);
1.113 kristaps 1843:
1844: if (r->last_tbl)
1.176 schwarze 1845: r->last_tbl->next = tbl;
1.113 kristaps 1846: else
1.176 schwarze 1847: r->first_tbl = r->last_tbl = tbl;
1.113 kristaps 1848:
1.176 schwarze 1849: r->tbl = r->last_tbl = tbl;
1.83 schwarze 1850: return(ROFF_IGN);
1.92 schwarze 1851: }
1852:
1.105 kristaps 1853: static enum rofferr
1.174 kristaps 1854: roff_cc(ROFF_ARGS)
1855: {
1856: const char *p;
1857:
1858: p = *bufp + pos;
1859:
1860: if ('\0' == *p || '.' == (r->control = *p++))
1861: r->control = 0;
1862:
1863: if ('\0' != *p)
1864: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
1865:
1866: return(ROFF_IGN);
1867: }
1868:
1869: static enum rofferr
1.164 kristaps 1870: roff_tr(ROFF_ARGS)
1871: {
1872: const char *p, *first, *second;
1873: size_t fsz, ssz;
1874: enum mandoc_esc esc;
1875:
1876: p = *bufp + pos;
1877:
1878: if ('\0' == *p) {
1879: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
1880: return(ROFF_IGN);
1881: }
1882:
1883: while ('\0' != *p) {
1884: fsz = ssz = 1;
1885:
1886: first = p++;
1887: if ('\\' == *first) {
1888: esc = mandoc_escape(&p, NULL, NULL);
1889: if (ESCAPE_ERROR == esc) {
1.207 schwarze 1890: mandoc_msg(MANDOCERR_BADESCAPE,
1891: r->parse, ln,
1892: (int)(p - *bufp), NULL);
1.164 kristaps 1893: return(ROFF_IGN);
1894: }
1895: fsz = (size_t)(p - first);
1896: }
1897:
1898: second = p++;
1899: if ('\\' == *second) {
1900: esc = mandoc_escape(&p, NULL, NULL);
1901: if (ESCAPE_ERROR == esc) {
1.207 schwarze 1902: mandoc_msg(MANDOCERR_BADESCAPE,
1903: r->parse, ln,
1904: (int)(p - *bufp), NULL);
1.164 kristaps 1905: return(ROFF_IGN);
1906: }
1907: ssz = (size_t)(p - second);
1.165 kristaps 1908: } else if ('\0' == *second) {
1.207 schwarze 1909: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse,
1910: ln, (int)(p - *bufp), NULL);
1.164 kristaps 1911: second = " ";
1.165 kristaps 1912: p--;
1.164 kristaps 1913: }
1914:
1.167 kristaps 1915: if (fsz > 1) {
1.207 schwarze 1916: roff_setstrn(&r->xmbtab, first, fsz,
1917: second, ssz, 0);
1.167 kristaps 1918: continue;
1919: }
1920:
1921: if (NULL == r->xtab)
1.207 schwarze 1922: r->xtab = mandoc_calloc(128,
1923: sizeof(struct roffstr));
1.167 kristaps 1924:
1925: free(r->xtab[(int)*first].p);
1926: r->xtab[(int)*first].p = mandoc_strndup(second, ssz);
1927: r->xtab[(int)*first].sz = ssz;
1.164 kristaps 1928: }
1929:
1930: return(ROFF_IGN);
1931: }
1932:
1933: static enum rofferr
1.105 kristaps 1934: roff_so(ROFF_ARGS)
1935: {
1936: char *name;
1937:
1.210 schwarze 1938: name = *bufp + pos;
1939: mandoc_vmsg(MANDOCERR_SO, r->parse, ln, ppos, ".so %s", name);
1.105 kristaps 1940:
1941: /*
1942: * Handle `so'. Be EXTREMELY careful, as we shouldn't be
1943: * opening anything that's not in our cwd or anything beneath
1944: * it. Thus, explicitly disallow traversing up the file-system
1945: * or using absolute paths.
1946: */
1947:
1948: if ('/' == *name || strstr(name, "../") || strstr(name, "/..")) {
1.210 schwarze 1949: mandoc_vmsg(MANDOCERR_SO_PATH, r->parse, ln, ppos,
1950: ".so %s", name);
1.105 kristaps 1951: return(ROFF_ERR);
1952: }
1953:
1954: *offs = pos;
1955: return(ROFF_SO);
1956: }
1.92 schwarze 1957:
1.106 kristaps 1958: static enum rofferr
1959: roff_userdef(ROFF_ARGS)
1.99 kristaps 1960: {
1.106 kristaps 1961: const char *arg[9];
1962: char *cp, *n1, *n2;
1.119 schwarze 1963: int i;
1.106 kristaps 1964:
1965: /*
1966: * Collect pointers to macro argument strings
1.188 schwarze 1967: * and NUL-terminate them.
1.106 kristaps 1968: */
1969: cp = *bufp + pos;
1.119 schwarze 1970: for (i = 0; i < 9; i++)
1.120 schwarze 1971: arg[i] = '\0' == *cp ? "" :
1.136 kristaps 1972: mandoc_getarg(r->parse, &cp, ln, &pos);
1.99 kristaps 1973:
1.106 kristaps 1974: /*
1975: * Expand macro arguments.
1.99 kristaps 1976: */
1.106 kristaps 1977: *szp = 0;
1978: n1 = cp = mandoc_strdup(r->current_string);
1979: while (NULL != (cp = strstr(cp, "\\$"))) {
1980: i = cp[2] - '1';
1981: if (0 > i || 8 < i) {
1982: /* Not an argument invocation. */
1983: cp += 2;
1984: continue;
1985: }
1.209 schwarze 1986: *cp = '\0';
1987: *szp = mandoc_asprintf(&n2, "%s%s%s",
1988: n1, arg[i], cp + 3) + 1;
1.106 kristaps 1989: cp = n2 + (cp - n1);
1990: free(n1);
1991: n1 = n2;
1.99 kristaps 1992: }
1993:
1.106 kristaps 1994: /*
1995: * Replace the macro invocation
1996: * by the expanded macro.
1997: */
1998: free(*bufp);
1999: *bufp = n1;
2000: if (0 == *szp)
2001: *szp = strlen(*bufp) + 1;
2002:
2003: return(*szp > 1 && '\n' == (*bufp)[(int)*szp - 2] ?
2004: ROFF_REPARSE : ROFF_APPEND);
1.99 kristaps 2005: }
1.121 schwarze 2006:
1.212 schwarze 2007: static size_t
1.121 schwarze 2008: roff_getname(struct roff *r, char **cpp, int ln, int pos)
2009: {
2010: char *name, *cp;
1.212 schwarze 2011: size_t namesz;
1.121 schwarze 2012:
2013: name = *cpp;
2014: if ('\0' == *name)
1.212 schwarze 2015: return(0);
1.121 schwarze 2016:
1.212 schwarze 2017: /* Read until end of name and terminate it with NUL. */
2018: for (cp = name; 1; cp++) {
2019: if ('\0' == *cp || ' ' == *cp) {
2020: namesz = cp - name;
2021: break;
2022: }
1.121 schwarze 2023: if ('\\' != *cp)
2024: continue;
1.215 schwarze 2025: namesz = cp - name;
2026: if ('{' == cp[1] || '}' == cp[1])
2027: break;
1.121 schwarze 2028: cp++;
2029: if ('\\' == *cp)
2030: continue;
1.128 kristaps 2031: mandoc_msg(MANDOCERR_NAMESC, r->parse, ln, pos, NULL);
1.212 schwarze 2032: mandoc_escape((const char **)&cp, NULL, NULL);
2033: break;
1.121 schwarze 2034: }
2035:
2036: /* Read past spaces. */
2037: while (' ' == *cp)
2038: cp++;
2039:
2040: *cpp = cp;
1.212 schwarze 2041: return(namesz);
1.121 schwarze 2042: }
2043:
1.106 kristaps 2044: /*
2045: * Store *string into the user-defined string called *name.
2046: * To clear an existing entry, call with (*r, *name, NULL, 0).
1.193 schwarze 2047: * append == 0: replace mode
2048: * append == 1: single-line append mode
2049: * append == 2: multiline append mode, append '\n' after each call
1.106 kristaps 2050: */
1.94 kristaps 2051: static void
1.106 kristaps 2052: roff_setstr(struct roff *r, const char *name, const char *string,
1.193 schwarze 2053: int append)
1.92 schwarze 2054: {
1.164 kristaps 2055:
2056: roff_setstrn(&r->strtab, name, strlen(name), string,
1.207 schwarze 2057: string ? strlen(string) : 0, append);
1.164 kristaps 2058: }
2059:
2060: static void
1.166 kristaps 2061: roff_setstrn(struct roffkv **r, const char *name, size_t namesz,
1.193 schwarze 2062: const char *string, size_t stringsz, int append)
1.164 kristaps 2063: {
1.166 kristaps 2064: struct roffkv *n;
1.164 kristaps 2065: char *c;
2066: int i;
2067: size_t oldch, newch;
1.92 schwarze 2068:
1.106 kristaps 2069: /* Search for an existing string with the same name. */
1.164 kristaps 2070: n = *r;
2071:
1.211 schwarze 2072: while (n && (namesz != n->key.sz ||
2073: strncmp(n->key.p, name, namesz)))
1.92 schwarze 2074: n = n->next;
1.94 kristaps 2075:
2076: if (NULL == n) {
1.106 kristaps 2077: /* Create a new string table entry. */
1.166 kristaps 2078: n = mandoc_malloc(sizeof(struct roffkv));
2079: n->key.p = mandoc_strndup(name, namesz);
2080: n->key.sz = namesz;
2081: n->val.p = NULL;
2082: n->val.sz = 0;
1.164 kristaps 2083: n->next = *r;
2084: *r = n;
1.193 schwarze 2085: } else if (0 == append) {
1.166 kristaps 2086: free(n->val.p);
2087: n->val.p = NULL;
2088: n->val.sz = 0;
1.106 kristaps 2089: }
2090:
2091: if (NULL == string)
2092: return;
2093:
2094: /*
2095: * One additional byte for the '\n' in multiline mode,
2096: * and one for the terminating '\0'.
2097: */
1.193 schwarze 2098: newch = stringsz + (1 < append ? 2u : 1u);
1.164 kristaps 2099:
1.166 kristaps 2100: if (NULL == n->val.p) {
2101: n->val.p = mandoc_malloc(newch);
2102: *n->val.p = '\0';
1.106 kristaps 2103: oldch = 0;
2104: } else {
1.166 kristaps 2105: oldch = n->val.sz;
2106: n->val.p = mandoc_realloc(n->val.p, oldch + newch);
1.106 kristaps 2107: }
2108:
2109: /* Skip existing content in the destination buffer. */
1.166 kristaps 2110: c = n->val.p + (int)oldch;
1.106 kristaps 2111:
2112: /* Append new content to the destination buffer. */
1.164 kristaps 2113: i = 0;
2114: while (i < (int)stringsz) {
1.106 kristaps 2115: /*
2116: * Rudimentary roff copy mode:
2117: * Handle escaped backslashes.
2118: */
1.164 kristaps 2119: if ('\\' == string[i] && '\\' == string[i + 1])
2120: i++;
2121: *c++ = string[i++];
1.106 kristaps 2122: }
1.94 kristaps 2123:
1.106 kristaps 2124: /* Append terminating bytes. */
1.193 schwarze 2125: if (1 < append)
1.106 kristaps 2126: *c++ = '\n';
1.163 kristaps 2127:
1.106 kristaps 2128: *c = '\0';
1.166 kristaps 2129: n->val.sz = (int)(c - n->val.p);
1.92 schwarze 2130: }
2131:
1.94 kristaps 2132: static const char *
2133: roff_getstrn(const struct roff *r, const char *name, size_t len)
1.92 schwarze 2134: {
1.166 kristaps 2135: const struct roffkv *n;
1.191 schwarze 2136: int i;
1.92 schwarze 2137:
1.164 kristaps 2138: for (n = r->strtab; n; n = n->next)
1.207 schwarze 2139: if (0 == strncmp(name, n->key.p, len) &&
2140: '\0' == n->key.p[(int)len])
1.166 kristaps 2141: return(n->val.p);
1.191 schwarze 2142:
2143: for (i = 0; i < PREDEFS_MAX; i++)
2144: if (0 == strncmp(name, predefs[i].name, len) &&
2145: '\0' == predefs[i].name[(int)len])
2146: return(predefs[i].str);
1.94 kristaps 2147:
1.157 kristaps 2148: return(NULL);
1.92 schwarze 2149: }
2150:
1.94 kristaps 2151: static void
1.167 kristaps 2152: roff_freestr(struct roffkv *r)
1.92 schwarze 2153: {
1.166 kristaps 2154: struct roffkv *n, *nn;
1.92 schwarze 2155:
1.167 kristaps 2156: for (n = r; n; n = nn) {
1.166 kristaps 2157: free(n->key.p);
2158: free(n->val.p);
1.92 schwarze 2159: nn = n->next;
2160: free(n);
2161: }
1.114 kristaps 2162: }
2163:
2164: const struct tbl_span *
2165: roff_span(const struct roff *r)
2166: {
1.207 schwarze 2167:
1.114 kristaps 2168: return(r->tbl ? tbl_span(r->tbl) : NULL);
1.125 kristaps 2169: }
2170:
2171: const struct eqn *
2172: roff_eqn(const struct roff *r)
2173: {
1.207 schwarze 2174:
1.125 kristaps 2175: return(r->last_eqn ? &r->last_eqn->eqn : NULL);
1.164 kristaps 2176: }
2177:
2178: /*
2179: * Duplicate an input string, making the appropriate character
2180: * conversations (as stipulated by `tr') along the way.
2181: * Returns a heap-allocated string with all the replacements made.
2182: */
2183: char *
2184: roff_strdup(const struct roff *r, const char *p)
2185: {
1.166 kristaps 2186: const struct roffkv *cp;
1.164 kristaps 2187: char *res;
2188: const char *pp;
2189: size_t ssz, sz;
2190: enum mandoc_esc esc;
2191:
1.167 kristaps 2192: if (NULL == r->xmbtab && NULL == r->xtab)
1.164 kristaps 2193: return(mandoc_strdup(p));
2194: else if ('\0' == *p)
2195: return(mandoc_strdup(""));
2196:
2197: /*
2198: * Step through each character looking for term matches
2199: * (remember that a `tr' can be invoked with an escape, which is
2200: * a glyph but the escape is multi-character).
2201: * We only do this if the character hash has been initialised
2202: * and the string is >0 length.
2203: */
2204:
2205: res = NULL;
2206: ssz = 0;
2207:
2208: while ('\0' != *p) {
1.167 kristaps 2209: if ('\\' != *p && r->xtab && r->xtab[(int)*p].p) {
2210: sz = r->xtab[(int)*p].sz;
2211: res = mandoc_realloc(res, ssz + sz + 1);
2212: memcpy(res + ssz, r->xtab[(int)*p].p, sz);
2213: ssz += sz;
2214: p++;
2215: continue;
2216: } else if ('\\' != *p) {
2217: res = mandoc_realloc(res, ssz + 2);
2218: res[ssz++] = *p++;
2219: continue;
2220: }
2221:
1.164 kristaps 2222: /* Search for term matches. */
1.167 kristaps 2223: for (cp = r->xmbtab; cp; cp = cp->next)
1.166 kristaps 2224: if (0 == strncmp(p, cp->key.p, cp->key.sz))
1.164 kristaps 2225: break;
2226:
2227: if (NULL != cp) {
2228: /*
2229: * A match has been found.
2230: * Append the match to the array and move
2231: * forward by its keysize.
2232: */
1.207 schwarze 2233: res = mandoc_realloc(res,
2234: ssz + cp->val.sz + 1);
1.166 kristaps 2235: memcpy(res + ssz, cp->val.p, cp->val.sz);
2236: ssz += cp->val.sz;
2237: p += (int)cp->key.sz;
1.164 kristaps 2238: continue;
2239: }
2240:
1.167 kristaps 2241: /*
2242: * Handle escapes carefully: we need to copy
2243: * over just the escape itself, or else we might
2244: * do replacements within the escape itself.
2245: * Make sure to pass along the bogus string.
2246: */
2247: pp = p++;
2248: esc = mandoc_escape(&p, NULL, NULL);
2249: if (ESCAPE_ERROR == esc) {
2250: sz = strlen(pp);
1.164 kristaps 2251: res = mandoc_realloc(res, ssz + sz + 1);
2252: memcpy(res + ssz, pp, sz);
1.167 kristaps 2253: break;
1.164 kristaps 2254: }
1.207 schwarze 2255: /*
2256: * We bail out on bad escapes.
1.167 kristaps 2257: * No need to warn: we already did so when
2258: * roff_res() was called.
2259: */
2260: sz = (int)(p - pp);
2261: res = mandoc_realloc(res, ssz + sz + 1);
2262: memcpy(res + ssz, pp, sz);
2263: ssz += sz;
1.164 kristaps 2264: }
2265:
2266: res[(int)ssz] = '\0';
2267: return(res);
1.174 kristaps 2268: }
2269:
2270: /*
1.207 schwarze 2271: * Find out whether a line is a macro line or not.
1.174 kristaps 2272: * If it is, adjust the current position and return one; if it isn't,
2273: * return zero and don't change the current position.
2274: * If the control character has been set with `.cc', then let that grain
2275: * precedence.
2276: * This is slighly contrary to groff, where using the non-breaking
2277: * control character when `cc' has been invoked will cause the
2278: * non-breaking macro contents to be printed verbatim.
2279: */
2280: int
2281: roff_getcontrol(const struct roff *r, const char *cp, int *ppos)
2282: {
2283: int pos;
2284:
2285: pos = *ppos;
2286:
2287: if (0 != r->control && cp[pos] == r->control)
2288: pos++;
2289: else if (0 != r->control)
2290: return(0);
2291: else if ('\\' == cp[pos] && '.' == cp[pos + 1])
2292: pos += 2;
2293: else if ('.' == cp[pos] || '\'' == cp[pos])
2294: pos++;
2295: else
2296: return(0);
2297:
2298: while (' ' == cp[pos] || '\t' == cp[pos])
2299: pos++;
2300:
2301: *ppos = pos;
2302: return(1);
1.74 kristaps 2303: }
CVSweb