Annotation of mandoc/roff.c, Revision 1.216
1.216 ! schwarze 1: /* $Id: roff.c,v 1.215 2014/07/01 00:32:29 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])
1.128 kristaps 859: mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.71 kristaps 860:
861: roffnode_pop(r);
1.76 kristaps 862: roffnode_cleanscope(r);
863: return(ROFF_IGN);
1.71 kristaps 864:
1.67 kristaps 865: }
866:
1.76 kristaps 867: static void
868: roffnode_cleanscope(struct roff *r)
1.67 kristaps 869: {
870:
1.76 kristaps 871: while (r->last) {
1.173 schwarze 872: if (--r->last->endspan != 0)
1.76 kristaps 873: break;
874: roffnode_pop(r);
875: }
1.67 kristaps 876: }
877:
1.195 schwarze 878: static void
879: roff_ccond(struct roff *r, int ln, int ppos)
1.74 kristaps 880: {
881:
1.76 kristaps 882: if (NULL == r->last) {
1.128 kristaps 883: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.195 schwarze 884: return;
1.76 kristaps 885: }
886:
1.82 kristaps 887: switch (r->last->tok) {
1.207 schwarze 888: case ROFF_el:
1.82 kristaps 889: /* FALLTHROUGH */
1.207 schwarze 890: case ROFF_ie:
1.82 kristaps 891: /* FALLTHROUGH */
1.207 schwarze 892: case ROFF_if:
1.82 kristaps 893: break;
894: default:
1.128 kristaps 895: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.195 schwarze 896: return;
1.75 kristaps 897: }
898:
1.76 kristaps 899: if (r->last->endspan > -1) {
1.128 kristaps 900: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.195 schwarze 901: return;
1.76 kristaps 902: }
903:
1.75 kristaps 904: roffnode_pop(r);
1.76 kristaps 905: roffnode_cleanscope(r);
1.195 schwarze 906: return;
1.76 kristaps 907: }
908:
909: static enum rofferr
1.80 kristaps 910: roff_block(ROFF_ARGS)
1.76 kristaps 911: {
1.213 schwarze 912: char *name, *cp;
913: size_t namesz;
1.106 kristaps 914:
1.213 schwarze 915: name = cp = *bufp + pos;
916: namesz = 0;
1.76 kristaps 917:
1.106 kristaps 918: if (ROFF_ig != tok) {
1.213 schwarze 919: if ('\0' == *cp) {
1.216 ! schwarze 920: mandoc_msg(MANDOCERR_REQ_EMPTY, r->parse,
! 921: ln, ppos, roffs[tok].name);
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;
1.213 schwarze 932: else if (ROFF_de != tok)
1.128 kristaps 933: mandoc_msg(MANDOCERR_REQUEST, r->parse, ln, ppos,
1.106 kristaps 934: roffs[tok].name);
1.107 kristaps 935:
1.213 schwarze 936: namesz = roff_getname(r, &cp, ln, ppos);
937: name[namesz] = '\0';
938: } else
939: name = NULL;
1.80 kristaps 940:
1.106 kristaps 941: roffnode_push(r, tok, name, ln, ppos);
942:
943: /*
944: * At the beginning of a `de' macro, clear the existing string
945: * with the same name, if there is one. New content will be
1.193 schwarze 946: * appended from roff_block_text() in multiline mode.
1.106 kristaps 947: */
1.107 kristaps 948:
1.213 schwarze 949: if (namesz && ROFF_de == tok)
950: roff_setstrn(&r->strtab, name, namesz, "", 0, 0);
1.76 kristaps 951:
1.213 schwarze 952: if ('\0' == *cp)
1.78 kristaps 953: return(ROFF_IGN);
954:
1.107 kristaps 955: /* If present, process the custom end-of-line marker. */
956:
1.213 schwarze 957: name = cp;
958: namesz = roff_getname(r, &cp, ln, ppos);
959: if (namesz)
960: r->last->end = mandoc_strndup(name, namesz);
1.78 kristaps 961:
1.213 schwarze 962: if ('\0' != *cp)
1.128 kristaps 963: mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.74 kristaps 964:
1.78 kristaps 965: return(ROFF_IGN);
966: }
967:
968: static enum rofferr
1.80 kristaps 969: roff_block_sub(ROFF_ARGS)
1.79 kristaps 970: {
971: enum rofft t;
972: int i, j;
973:
974: /*
975: * First check whether a custom macro exists at this level. If
976: * it does, then check against it. This is some of groff's
977: * stranger behaviours. If we encountered a custom end-scope
978: * tag and that tag also happens to be a "real" macro, then we
979: * need to try interpreting it again as a real macro. If it's
980: * not, then return ignore. Else continue.
981: */
982:
983: if (r->last->end) {
1.130 kristaps 984: for (i = pos, j = 0; r->last->end[j]; j++, i++)
1.79 kristaps 985: if ((*bufp)[i] != r->last->end[j])
986: break;
987:
1.207 schwarze 988: if ('\0' == r->last->end[j] &&
989: ('\0' == (*bufp)[i] ||
990: ' ' == (*bufp)[i] ||
991: '\t' == (*bufp)[i])) {
1.79 kristaps 992: roffnode_pop(r);
993: roffnode_cleanscope(r);
994:
1.130 kristaps 995: while (' ' == (*bufp)[i] || '\t' == (*bufp)[i])
996: i++;
997:
998: pos = i;
1.214 schwarze 999: if (ROFF_MAX != roff_parse(r, *bufp, &pos, ln, ppos))
1.79 kristaps 1000: return(ROFF_RERUN);
1001: return(ROFF_IGN);
1002: }
1003: }
1004:
1005: /*
1006: * If we have no custom end-query or lookup failed, then try
1007: * pulling it out of the hashtable.
1008: */
1009:
1.214 schwarze 1010: t = roff_parse(r, *bufp, &pos, ln, ppos);
1.79 kristaps 1011:
1.106 kristaps 1012: /*
1013: * Macros other than block-end are only significant
1014: * in `de' blocks; elsewhere, simply throw them away.
1015: */
1016: if (ROFF_cblock != t) {
1017: if (ROFF_de == tok)
1.193 schwarze 1018: roff_setstr(r, r->last->name, *bufp + ppos, 2);
1.79 kristaps 1019: return(ROFF_IGN);
1.106 kristaps 1020: }
1.79 kristaps 1021:
1022: assert(roffs[t].proc);
1.207 schwarze 1023: return((*roffs[t].proc)(r, t, bufp, szp, ln, ppos, pos, offs));
1.79 kristaps 1024: }
1025:
1026: static enum rofferr
1.80 kristaps 1027: roff_block_text(ROFF_ARGS)
1.78 kristaps 1028: {
1029:
1.106 kristaps 1030: if (ROFF_de == tok)
1.193 schwarze 1031: roff_setstr(r, r->last->name, *bufp + pos, 2);
1.106 kristaps 1032:
1.78 kristaps 1033: return(ROFF_IGN);
1034: }
1035:
1036: static enum rofferr
1.82 kristaps 1037: roff_cond_sub(ROFF_ARGS)
1038: {
1039: enum rofft t;
1.139 kristaps 1040: char *ep;
1.198 schwarze 1041: int rr;
1.82 kristaps 1042:
1043: rr = r->last->rule;
1.139 kristaps 1044: roffnode_cleanscope(r);
1.214 schwarze 1045: t = roff_parse(r, *bufp, &pos, ln, ppos);
1.82 kristaps 1046:
1.139 kristaps 1047: /*
1.177 schwarze 1048: * Fully handle known macros when they are structurally
1049: * required or when the conditional evaluated to true.
1.87 kristaps 1050: */
1051:
1.177 schwarze 1052: if ((ROFF_MAX != t) &&
1.198 schwarze 1053: (rr || ROFFMAC_STRUCT & roffs[t].flags)) {
1.177 schwarze 1054: assert(roffs[t].proc);
1055: return((*roffs[t].proc)(r, t, bufp, szp,
1.207 schwarze 1056: ln, ppos, pos, offs));
1.177 schwarze 1057: }
1.144 kristaps 1058:
1.196 schwarze 1059: /*
1060: * If `\}' occurs on a macro line without a preceding macro,
1061: * drop the line completely.
1062: */
1063:
1064: ep = *bufp + pos;
1065: if ('\\' == ep[0] && '}' == ep[1])
1.198 schwarze 1066: rr = 0;
1.196 schwarze 1067:
1.177 schwarze 1068: /* Always check for the closing delimiter `\}'. */
1.144 kristaps 1069:
1.177 schwarze 1070: while (NULL != (ep = strchr(ep, '\\'))) {
1.197 schwarze 1071: if ('}' == *(++ep)) {
1072: *ep = '&';
1073: roff_ccond(r, ln, ep - *bufp - 1);
1074: }
1075: ++ep;
1.177 schwarze 1076: }
1.198 schwarze 1077: return(rr ? ROFF_CONT : ROFF_IGN);
1.82 kristaps 1078: }
1079:
1080: static enum rofferr
1081: roff_cond_text(ROFF_ARGS)
1.78 kristaps 1082: {
1.140 kristaps 1083: char *ep;
1.198 schwarze 1084: int rr;
1.82 kristaps 1085:
1086: rr = r->last->rule;
1.140 kristaps 1087: roffnode_cleanscope(r);
1.82 kristaps 1088:
1.197 schwarze 1089: ep = *bufp + pos;
1090: while (NULL != (ep = strchr(ep, '\\'))) {
1091: if ('}' == *(++ep)) {
1092: *ep = '&';
1093: roff_ccond(r, ln, ep - *bufp - 1);
1094: }
1095: ++ep;
1.78 kristaps 1096: }
1.198 schwarze 1097: return(rr ? ROFF_CONT : ROFF_IGN);
1.74 kristaps 1098: }
1099:
1.204 schwarze 1100: /*
1101: * Parse a single signed integer number. Stop at the first non-digit.
1102: * If there is at least one digit, return success and advance the
1103: * parse point, else return failure and let the parse point unchanged.
1104: * Ignore overflows, treat them just like the C language.
1105: */
1.184 schwarze 1106: static int
1107: roff_getnum(const char *v, int *pos, int *res)
1108: {
1.206 schwarze 1109: int myres, n, p;
1110:
1111: if (NULL == res)
1112: res = &myres;
1.184 schwarze 1113:
1114: p = *pos;
1115: n = v[p] == '-';
1116: if (n)
1117: p++;
1118:
1119: for (*res = 0; isdigit((unsigned char)v[p]); p++)
1.204 schwarze 1120: *res = 10 * *res + v[p] - '0';
1.184 schwarze 1121: if (p == *pos + n)
1122: return 0;
1123:
1124: if (n)
1125: *res = -*res;
1126:
1127: *pos = p;
1128: return 1;
1129: }
1130:
1.198 schwarze 1131: /*
1132: * Evaluate a string comparison condition.
1133: * The first character is the delimiter.
1134: * Succeed if the string up to its second occurrence
1135: * matches the string up to its third occurence.
1136: * Advance the cursor after the third occurrence
1137: * or lacking that, to the end of the line.
1138: */
1139: static int
1140: roff_evalstrcond(const char *v, int *pos)
1141: {
1142: const char *s1, *s2, *s3;
1143: int match;
1144:
1145: match = 0;
1146: s1 = v + *pos; /* initial delimiter */
1147: s2 = s1 + 1; /* for scanning the first string */
1148: s3 = strchr(s2, *s1); /* for scanning the second string */
1149:
1150: if (NULL == s3) /* found no middle delimiter */
1151: goto out;
1152:
1153: while ('\0' != *++s3) {
1154: if (*s2 != *s3) { /* mismatch */
1155: s3 = strchr(s3, *s1);
1156: break;
1157: }
1158: if (*s3 == *s1) { /* found the final delimiter */
1159: match = 1;
1160: break;
1161: }
1162: s2++;
1163: }
1164:
1165: out:
1166: if (NULL == s3)
1167: s3 = strchr(s2, '\0');
1168: else
1169: s3++;
1170: *pos = s3 - v;
1171: return(match);
1172: }
1173:
1.204 schwarze 1174: /*
1175: * Evaluate an optionally negated single character, numerical,
1176: * or string condition.
1177: */
1.198 schwarze 1178: static int
1.88 kristaps 1179: roff_evalcond(const char *v, int *pos)
1180: {
1.204 schwarze 1181: int wanttrue, number;
1.88 kristaps 1182:
1.198 schwarze 1183: if ('!' == v[*pos]) {
1184: wanttrue = 0;
1185: (*pos)++;
1186: } else
1187: wanttrue = 1;
1188:
1.88 kristaps 1189: switch (v[*pos]) {
1.207 schwarze 1190: case 'n':
1.198 schwarze 1191: /* FALLTHROUGH */
1.207 schwarze 1192: case 'o':
1.88 kristaps 1193: (*pos)++;
1.198 schwarze 1194: return(wanttrue);
1.207 schwarze 1195: case 'c':
1.198 schwarze 1196: /* FALLTHROUGH */
1.207 schwarze 1197: case 'd':
1.198 schwarze 1198: /* FALLTHROUGH */
1.207 schwarze 1199: case 'e':
1.88 kristaps 1200: /* FALLTHROUGH */
1.207 schwarze 1201: case 'r':
1.88 kristaps 1202: /* FALLTHROUGH */
1.207 schwarze 1203: case 't':
1.88 kristaps 1204: (*pos)++;
1.198 schwarze 1205: return(!wanttrue);
1.88 kristaps 1206: default:
1207: break;
1208: }
1209:
1.204 schwarze 1210: if (roff_evalnum(v, pos, &number, 0))
1211: return((number > 0) == wanttrue);
1212: else
1.198 schwarze 1213: return(roff_evalstrcond(v, pos) == wanttrue);
1.88 kristaps 1214: }
1215:
1.74 kristaps 1216: static enum rofferr
1.103 kristaps 1217: roff_line_ignore(ROFF_ARGS)
1.89 kristaps 1218: {
1.123 schwarze 1219:
1.89 kristaps 1220: return(ROFF_IGN);
1221: }
1222:
1.104 kristaps 1223: static enum rofferr
1.82 kristaps 1224: roff_cond(ROFF_ARGS)
1.74 kristaps 1225: {
1.173 schwarze 1226:
1227: roffnode_push(r, tok, NULL, ln, ppos);
1.74 kristaps 1228:
1.207 schwarze 1229: /*
1.134 kristaps 1230: * An `.el' has no conditional body: it will consume the value
1231: * of the current rstack entry set in prior `ie' calls or
1.207 schwarze 1232: * defaults to DENY.
1.134 kristaps 1233: *
1234: * If we're not an `el', however, then evaluate the conditional.
1235: */
1.133 kristaps 1236:
1.173 schwarze 1237: r->last->rule = ROFF_el == tok ?
1.207 schwarze 1238: (r->rstackpos < 0 ? 0 : r->rstack[r->rstackpos--]) :
1239: roff_evalcond(*bufp, &pos);
1.77 kristaps 1240:
1.134 kristaps 1241: /*
1242: * An if-else will put the NEGATION of the current evaluated
1243: * conditional into the stack of rules.
1244: */
1245:
1.84 schwarze 1246: if (ROFF_ie == tok) {
1.134 kristaps 1247: if (r->rstackpos == RSTACK_MAX - 1) {
1.207 schwarze 1248: mandoc_msg(MANDOCERR_MEM,
1249: r->parse, ln, ppos, NULL);
1.134 kristaps 1250: return(ROFF_ERR);
1251: }
1.198 schwarze 1252: r->rstack[++r->rstackpos] = !r->last->rule;
1.82 kristaps 1253: }
1.88 kristaps 1254:
1255: /* If the parent has false as its rule, then so do we. */
1256:
1.198 schwarze 1257: if (r->last->parent && !r->last->parent->rule)
1258: r->last->rule = 0;
1.88 kristaps 1259:
1260: /*
1.173 schwarze 1261: * Determine scope.
1262: * If there is nothing on the line after the conditional,
1263: * not even whitespace, use next-line scope.
1.88 kristaps 1264: */
1.74 kristaps 1265:
1.173 schwarze 1266: if ('\0' == (*bufp)[pos]) {
1267: r->last->endspan = 2;
1268: goto out;
1269: }
1270:
1271: while (' ' == (*bufp)[pos])
1272: pos++;
1273:
1274: /* An opening brace requests multiline scope. */
1.75 kristaps 1275:
1276: if ('\\' == (*bufp)[pos] && '{' == (*bufp)[pos + 1]) {
1277: r->last->endspan = -1;
1278: pos += 2;
1.173 schwarze 1279: goto out;
1.207 schwarze 1280: }
1.74 kristaps 1281:
1.77 kristaps 1282: /*
1.173 schwarze 1283: * Anything else following the conditional causes
1284: * single-line scope. Warn if the scope contains
1285: * nothing but trailing whitespace.
1.77 kristaps 1286: */
1287:
1.75 kristaps 1288: if ('\0' == (*bufp)[pos])
1.216 ! schwarze 1289: mandoc_msg(MANDOCERR_COND_EMPTY, r->parse,
! 1290: ln, ppos, roffs[tok].name);
1.77 kristaps 1291:
1.173 schwarze 1292: r->last->endspan = 1;
1.74 kristaps 1293:
1.173 schwarze 1294: out:
1.75 kristaps 1295: *offs = pos;
1296: return(ROFF_RERUN);
1.83 schwarze 1297: }
1298:
1299: static enum rofferr
1.92 schwarze 1300: roff_ds(ROFF_ARGS)
1301: {
1.212 schwarze 1302: char *string;
1303: const char *name;
1304: size_t namesz;
1.96 kristaps 1305:
1306: /*
1.212 schwarze 1307: * The first word is the name of the string.
1308: * If it is empty or terminated by an escape sequence,
1309: * abort the `ds' request without defining anything.
1.96 kristaps 1310: */
1.92 schwarze 1311:
1.212 schwarze 1312: name = string = *bufp + pos;
1.92 schwarze 1313: if ('\0' == *name)
1314: return(ROFF_IGN);
1315:
1.212 schwarze 1316: namesz = roff_getname(r, &string, ln, pos);
1317: if ('\\' == name[namesz])
1318: return(ROFF_IGN);
1319:
1320: /* Read past the initial double-quote, if any. */
1.121 schwarze 1321: if ('"' == *string)
1.92 schwarze 1322: string++;
1323:
1.96 kristaps 1324: /* The rest is the value. */
1.212 schwarze 1325: roff_setstrn(&r->strtab, name, namesz, string, strlen(string),
1326: ROFF_as == tok);
1.92 schwarze 1327: return(ROFF_IGN);
1328: }
1329:
1.204 schwarze 1330: /*
1331: * Parse a single operator, one or two characters long.
1332: * If the operator is recognized, return success and advance the
1333: * parse point, else return failure and let the parse point unchanged.
1334: */
1335: static int
1336: roff_getop(const char *v, int *pos, char *res)
1337: {
1338:
1339: *res = v[*pos];
1340:
1341: switch (*res) {
1.207 schwarze 1342: case '+':
1.204 schwarze 1343: /* FALLTHROUGH */
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: break;
1356: case '<':
1357: switch (v[*pos + 1]) {
1.207 schwarze 1358: case '=':
1.204 schwarze 1359: *res = 'l';
1360: (*pos)++;
1361: break;
1.207 schwarze 1362: case '>':
1.204 schwarze 1363: *res = '!';
1364: (*pos)++;
1365: break;
1.207 schwarze 1366: case '?':
1.204 schwarze 1367: *res = 'i';
1368: (*pos)++;
1369: break;
1370: default:
1371: break;
1372: }
1373: break;
1374: case '>':
1375: switch (v[*pos + 1]) {
1.207 schwarze 1376: case '=':
1.204 schwarze 1377: *res = 'g';
1378: (*pos)++;
1379: break;
1.207 schwarze 1380: case '?':
1.204 schwarze 1381: *res = 'a';
1382: (*pos)++;
1383: break;
1384: default:
1385: break;
1386: }
1387: break;
1388: case '=':
1389: if ('=' == v[*pos + 1])
1390: (*pos)++;
1391: break;
1392: default:
1393: return(0);
1394: }
1395: (*pos)++;
1396:
1397: return(*res);
1398: }
1399:
1400: /*
1401: * Evaluate either a parenthesized numeric expression
1402: * or a single signed integer number.
1403: */
1404: static int
1405: roff_evalpar(const char *v, int *pos, int *res)
1406: {
1407:
1408: if ('(' != v[*pos])
1409: return(roff_getnum(v, pos, res));
1410:
1411: (*pos)++;
1412: if ( ! roff_evalnum(v, pos, res, 1))
1413: return(0);
1414:
1.206 schwarze 1415: /*
1416: * Omission of the closing parenthesis
1417: * is an error in validation mode,
1418: * but ignored in evaluation mode.
1419: */
1420:
1.204 schwarze 1421: if (')' == v[*pos])
1422: (*pos)++;
1.206 schwarze 1423: else if (NULL == res)
1424: return(0);
1.204 schwarze 1425:
1426: return(1);
1427: }
1428:
1429: /*
1430: * Evaluate a complete numeric expression.
1431: * Proceed left to right, there is no concept of precedence.
1432: */
1433: static int
1434: roff_evalnum(const char *v, int *pos, int *res, int skipwhite)
1435: {
1436: int mypos, operand2;
1437: char operator;
1438:
1439: if (NULL == pos) {
1440: mypos = 0;
1441: pos = &mypos;
1442: }
1443:
1444: if (skipwhite)
1445: while (isspace((unsigned char)v[*pos]))
1446: (*pos)++;
1447:
1448: if ( ! roff_evalpar(v, pos, res))
1449: return(0);
1450:
1451: while (1) {
1452: if (skipwhite)
1453: while (isspace((unsigned char)v[*pos]))
1454: (*pos)++;
1455:
1456: if ( ! roff_getop(v, pos, &operator))
1457: break;
1458:
1459: if (skipwhite)
1460: while (isspace((unsigned char)v[*pos]))
1461: (*pos)++;
1462:
1463: if ( ! roff_evalpar(v, pos, &operand2))
1464: return(0);
1465:
1466: if (skipwhite)
1467: while (isspace((unsigned char)v[*pos]))
1468: (*pos)++;
1.206 schwarze 1469:
1470: if (NULL == res)
1471: continue;
1.204 schwarze 1472:
1473: switch (operator) {
1.207 schwarze 1474: case '+':
1.204 schwarze 1475: *res += operand2;
1476: break;
1.207 schwarze 1477: case '-':
1.204 schwarze 1478: *res -= operand2;
1479: break;
1.207 schwarze 1480: case '*':
1.204 schwarze 1481: *res *= operand2;
1482: break;
1.207 schwarze 1483: case '/':
1.204 schwarze 1484: *res /= operand2;
1485: break;
1.207 schwarze 1486: case '%':
1.204 schwarze 1487: *res %= operand2;
1488: break;
1.207 schwarze 1489: case '<':
1.204 schwarze 1490: *res = *res < operand2;
1491: break;
1.207 schwarze 1492: case '>':
1.204 schwarze 1493: *res = *res > operand2;
1494: break;
1.207 schwarze 1495: case 'l':
1.204 schwarze 1496: *res = *res <= operand2;
1497: break;
1.207 schwarze 1498: case 'g':
1.204 schwarze 1499: *res = *res >= operand2;
1500: break;
1.207 schwarze 1501: case '=':
1.204 schwarze 1502: *res = *res == operand2;
1503: break;
1.207 schwarze 1504: case '!':
1.204 schwarze 1505: *res = *res != operand2;
1506: break;
1.207 schwarze 1507: case '&':
1.204 schwarze 1508: *res = *res && operand2;
1509: break;
1.207 schwarze 1510: case ':':
1.204 schwarze 1511: *res = *res || operand2;
1512: break;
1.207 schwarze 1513: case 'i':
1.204 schwarze 1514: if (operand2 < *res)
1515: *res = operand2;
1516: break;
1.207 schwarze 1517: case 'a':
1.204 schwarze 1518: if (operand2 > *res)
1519: *res = operand2;
1520: break;
1521: default:
1522: abort();
1523: }
1524: }
1525: return(1);
1526: }
1527:
1.180 schwarze 1528: void
1.187 schwarze 1529: roff_setreg(struct roff *r, const char *name, int val, char sign)
1.147 kristaps 1530: {
1.180 schwarze 1531: struct roffreg *reg;
1532:
1533: /* Search for an existing register with the same name. */
1534: reg = r->regtab;
1535:
1536: while (reg && strcmp(name, reg->key.p))
1537: reg = reg->next;
1.147 kristaps 1538:
1.180 schwarze 1539: if (NULL == reg) {
1540: /* Create a new register. */
1541: reg = mandoc_malloc(sizeof(struct roffreg));
1542: reg->key.p = mandoc_strdup(name);
1543: reg->key.sz = strlen(name);
1.187 schwarze 1544: reg->val = 0;
1.180 schwarze 1545: reg->next = r->regtab;
1546: r->regtab = reg;
1547: }
1548:
1.187 schwarze 1549: if ('+' == sign)
1550: reg->val += val;
1551: else if ('-' == sign)
1552: reg->val -= val;
1553: else
1554: reg->val = val;
1.147 kristaps 1555: }
1556:
1.192 schwarze 1557: /*
1558: * Handle some predefined read-only number registers.
1559: * For now, return -1 if the requested register is not predefined;
1560: * in case a predefined read-only register having the value -1
1561: * were to turn up, another special value would have to be chosen.
1562: */
1563: static int
1564: roff_getregro(const char *name)
1565: {
1566:
1567: switch (*name) {
1.207 schwarze 1568: case 'A': /* ASCII approximation mode is always off. */
1.192 schwarze 1569: return(0);
1.207 schwarze 1570: case 'g': /* Groff compatibility mode is always on. */
1.192 schwarze 1571: return(1);
1.207 schwarze 1572: case 'H': /* Fixed horizontal resolution. */
1.192 schwarze 1573: return (24);
1.207 schwarze 1574: case 'j': /* Always adjust left margin only. */
1.192 schwarze 1575: return(0);
1.207 schwarze 1576: case 'T': /* Some output device is always defined. */
1.192 schwarze 1577: return(1);
1.207 schwarze 1578: case 'V': /* Fixed vertical resolution. */
1.192 schwarze 1579: return (40);
1580: default:
1581: return (-1);
1582: }
1583: }
1584:
1.181 schwarze 1585: int
1.180 schwarze 1586: roff_getreg(const struct roff *r, const char *name)
1.147 kristaps 1587: {
1.180 schwarze 1588: struct roffreg *reg;
1.192 schwarze 1589: int val;
1590:
1591: if ('.' == name[0] && '\0' != name[1] && '\0' == name[2]) {
1592: val = roff_getregro(name + 1);
1593: if (-1 != val)
1594: return (val);
1595: }
1.180 schwarze 1596:
1597: for (reg = r->regtab; reg; reg = reg->next)
1598: if (0 == strcmp(name, reg->key.p))
1.181 schwarze 1599: return(reg->val);
1600:
1601: return(0);
1602: }
1603:
1604: static int
1605: roff_getregn(const struct roff *r, const char *name, size_t len)
1606: {
1607: struct roffreg *reg;
1.192 schwarze 1608: int val;
1609:
1610: if ('.' == name[0] && 2 == len) {
1611: val = roff_getregro(name + 1);
1612: if (-1 != val)
1613: return (val);
1614: }
1.181 schwarze 1615:
1616: for (reg = r->regtab; reg; reg = reg->next)
1617: if (len == reg->key.sz &&
1618: 0 == strncmp(name, reg->key.p, len))
1619: return(reg->val);
1.147 kristaps 1620:
1.180 schwarze 1621: return(0);
1.147 kristaps 1622: }
1623:
1.180 schwarze 1624: static void
1625: roff_freereg(struct roffreg *reg)
1.147 kristaps 1626: {
1.180 schwarze 1627: struct roffreg *old_reg;
1.147 kristaps 1628:
1.180 schwarze 1629: while (NULL != reg) {
1630: free(reg->key.p);
1631: old_reg = reg;
1632: reg = reg->next;
1633: free(old_reg);
1634: }
1.147 kristaps 1635: }
1.92 schwarze 1636:
1637: static enum rofferr
1.89 kristaps 1638: roff_nr(ROFF_ARGS)
1.83 schwarze 1639: {
1.212 schwarze 1640: char *key, *val;
1641: size_t keysz;
1.138 kristaps 1642: int iv;
1.187 schwarze 1643: char sign;
1.89 kristaps 1644:
1.212 schwarze 1645: key = val = *bufp + pos;
1646: if ('\0' == *key)
1647: return(ROFF_IGN);
1648:
1649: keysz = roff_getname(r, &val, ln, pos);
1650: if ('\\' == key[keysz])
1651: return(ROFF_IGN);
1652: key[keysz] = '\0';
1.89 kristaps 1653:
1.187 schwarze 1654: sign = *val;
1655: if ('+' == sign || '-' == sign)
1656: val++;
1657:
1.204 schwarze 1658: if (roff_evalnum(val, NULL, &iv, 0))
1659: roff_setreg(r, key, iv, sign);
1.109 kristaps 1660:
1.203 schwarze 1661: return(ROFF_IGN);
1662: }
1663:
1664: static enum rofferr
1665: roff_rr(ROFF_ARGS)
1666: {
1667: struct roffreg *reg, **prev;
1.212 schwarze 1668: char *name, *cp;
1669: size_t namesz;
1.203 schwarze 1670:
1.212 schwarze 1671: name = cp = *bufp + pos;
1672: if ('\0' == *name)
1673: return(ROFF_IGN);
1674: namesz = roff_getname(r, &cp, ln, pos);
1675: name[namesz] = '\0';
1.203 schwarze 1676:
1677: prev = &r->regtab;
1678: while (1) {
1679: reg = *prev;
1680: if (NULL == reg || !strcmp(name, reg->key.p))
1681: break;
1682: prev = ®->next;
1683: }
1684: if (NULL != reg) {
1685: *prev = reg->next;
1686: free(reg->key.p);
1687: free(reg);
1688: }
1.122 schwarze 1689: return(ROFF_IGN);
1690: }
1691:
1692: static enum rofferr
1693: roff_rm(ROFF_ARGS)
1694: {
1695: const char *name;
1696: char *cp;
1.212 schwarze 1697: size_t namesz;
1.122 schwarze 1698:
1699: cp = *bufp + pos;
1700: while ('\0' != *cp) {
1.212 schwarze 1701: name = cp;
1702: namesz = roff_getname(r, &cp, ln, (int)(cp - *bufp));
1703: roff_setstrn(&r->strtab, name, namesz, NULL, 0, 0);
1704: if ('\\' == name[namesz])
1705: break;
1.122 schwarze 1706: }
1.178 schwarze 1707: return(ROFF_IGN);
1708: }
1709:
1710: static enum rofferr
1711: roff_it(ROFF_ARGS)
1712: {
1713: char *cp;
1714: size_t len;
1715: int iv;
1716:
1717: /* Parse the number of lines. */
1718: cp = *bufp + pos;
1719: len = strcspn(cp, " \t");
1720: cp[len] = '\0';
1721: if ((iv = mandoc_strntoi(cp, len, 10)) <= 0) {
1722: mandoc_msg(MANDOCERR_NUMERIC, r->parse,
1.207 schwarze 1723: ln, ppos, *bufp + 1);
1.178 schwarze 1724: return(ROFF_IGN);
1725: }
1726: cp += len + 1;
1727:
1728: /* Arm the input line trap. */
1729: roffit_lines = iv;
1730: roffit_macro = mandoc_strdup(cp);
1.109 kristaps 1731: return(ROFF_IGN);
1.175 schwarze 1732: }
1733:
1734: static enum rofferr
1735: roff_Dd(ROFF_ARGS)
1736: {
1737: const char *const *cp;
1738:
1.199 schwarze 1739: if (0 == ((MPARSE_MDOC | MPARSE_QUICK) & r->options))
1.175 schwarze 1740: for (cp = __mdoc_reserved; *cp; cp++)
1741: roff_setstr(r, *cp, NULL, 0);
1742:
1743: return(ROFF_CONT);
1744: }
1745:
1746: static enum rofferr
1747: roff_TH(ROFF_ARGS)
1748: {
1749: const char *const *cp;
1750:
1.199 schwarze 1751: if (0 == (MPARSE_QUICK & r->options))
1.175 schwarze 1752: for (cp = __man_reserved; *cp; cp++)
1753: roff_setstr(r, *cp, NULL, 0);
1754:
1755: return(ROFF_CONT);
1.109 kristaps 1756: }
1757:
1758: static enum rofferr
1759: roff_TE(ROFF_ARGS)
1760: {
1761:
1762: if (NULL == r->tbl)
1.128 kristaps 1763: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.115 kristaps 1764: else
1.151 kristaps 1765: tbl_end(&r->tbl);
1.109 kristaps 1766:
1.112 kristaps 1767: return(ROFF_IGN);
1768: }
1769:
1770: static enum rofferr
1771: roff_T_(ROFF_ARGS)
1772: {
1773:
1774: if (NULL == r->tbl)
1.128 kristaps 1775: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.112 kristaps 1776: else
1.116 kristaps 1777: tbl_restart(ppos, ln, r->tbl);
1.112 kristaps 1778:
1.109 kristaps 1779: return(ROFF_IGN);
1780: }
1781:
1.156 kristaps 1782: #if 0
1783: static int
1.151 kristaps 1784: roff_closeeqn(struct roff *r)
1785: {
1786:
1787: return(r->eqn && ROFF_EQN == eqn_end(&r->eqn) ? 1 : 0);
1788: }
1.156 kristaps 1789: #endif
1.151 kristaps 1790:
1.156 kristaps 1791: static void
1.207 schwarze 1792: roff_openeqn(struct roff *r, const char *name, int line,
1.151 kristaps 1793: int offs, const char *buf)
1.125 kristaps 1794: {
1.151 kristaps 1795: struct eqn_node *e;
1796: int poff;
1.125 kristaps 1797:
1798: assert(NULL == r->eqn);
1.151 kristaps 1799: e = eqn_alloc(name, offs, line, r->parse);
1.125 kristaps 1800:
1801: if (r->last_eqn)
1802: r->last_eqn->next = e;
1803: else
1804: r->first_eqn = r->last_eqn = e;
1805:
1806: r->eqn = r->last_eqn = e;
1.151 kristaps 1807:
1808: if (buf) {
1809: poff = 0;
1810: eqn_read(&r->eqn, line, buf, offs, &poff);
1811: }
1812: }
1813:
1814: static enum rofferr
1815: roff_EQ(ROFF_ARGS)
1816: {
1817:
1818: roff_openeqn(r, *bufp + pos, ln, ppos, NULL);
1.125 kristaps 1819: return(ROFF_IGN);
1820: }
1821:
1822: static enum rofferr
1823: roff_EN(ROFF_ARGS)
1824: {
1825:
1.128 kristaps 1826: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.125 kristaps 1827: return(ROFF_IGN);
1828: }
1829:
1830: static enum rofferr
1.109 kristaps 1831: roff_TS(ROFF_ARGS)
1832: {
1.176 schwarze 1833: struct tbl_node *tbl;
1.89 kristaps 1834:
1.115 kristaps 1835: if (r->tbl) {
1.128 kristaps 1836: mandoc_msg(MANDOCERR_SCOPEBROKEN, r->parse, ln, ppos, NULL);
1.151 kristaps 1837: tbl_end(&r->tbl);
1.115 kristaps 1838: }
1.83 schwarze 1839:
1.176 schwarze 1840: tbl = tbl_alloc(ppos, ln, r->parse);
1.113 kristaps 1841:
1842: if (r->last_tbl)
1.176 schwarze 1843: r->last_tbl->next = tbl;
1.113 kristaps 1844: else
1.176 schwarze 1845: r->first_tbl = r->last_tbl = tbl;
1.113 kristaps 1846:
1.176 schwarze 1847: r->tbl = r->last_tbl = tbl;
1.83 schwarze 1848: return(ROFF_IGN);
1.92 schwarze 1849: }
1850:
1.105 kristaps 1851: static enum rofferr
1.174 kristaps 1852: roff_cc(ROFF_ARGS)
1853: {
1854: const char *p;
1855:
1856: p = *bufp + pos;
1857:
1858: if ('\0' == *p || '.' == (r->control = *p++))
1859: r->control = 0;
1860:
1861: if ('\0' != *p)
1862: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
1863:
1864: return(ROFF_IGN);
1865: }
1866:
1867: static enum rofferr
1.164 kristaps 1868: roff_tr(ROFF_ARGS)
1869: {
1870: const char *p, *first, *second;
1871: size_t fsz, ssz;
1872: enum mandoc_esc esc;
1873:
1874: p = *bufp + pos;
1875:
1876: if ('\0' == *p) {
1877: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
1878: return(ROFF_IGN);
1879: }
1880:
1881: while ('\0' != *p) {
1882: fsz = ssz = 1;
1883:
1884: first = p++;
1885: if ('\\' == *first) {
1886: esc = mandoc_escape(&p, NULL, NULL);
1887: if (ESCAPE_ERROR == esc) {
1.207 schwarze 1888: mandoc_msg(MANDOCERR_BADESCAPE,
1889: r->parse, ln,
1890: (int)(p - *bufp), NULL);
1.164 kristaps 1891: return(ROFF_IGN);
1892: }
1893: fsz = (size_t)(p - first);
1894: }
1895:
1896: second = p++;
1897: if ('\\' == *second) {
1898: esc = mandoc_escape(&p, NULL, NULL);
1899: if (ESCAPE_ERROR == esc) {
1.207 schwarze 1900: mandoc_msg(MANDOCERR_BADESCAPE,
1901: r->parse, ln,
1902: (int)(p - *bufp), NULL);
1.164 kristaps 1903: return(ROFF_IGN);
1904: }
1905: ssz = (size_t)(p - second);
1.165 kristaps 1906: } else if ('\0' == *second) {
1.207 schwarze 1907: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse,
1908: ln, (int)(p - *bufp), NULL);
1.164 kristaps 1909: second = " ";
1.165 kristaps 1910: p--;
1.164 kristaps 1911: }
1912:
1.167 kristaps 1913: if (fsz > 1) {
1.207 schwarze 1914: roff_setstrn(&r->xmbtab, first, fsz,
1915: second, ssz, 0);
1.167 kristaps 1916: continue;
1917: }
1918:
1919: if (NULL == r->xtab)
1.207 schwarze 1920: r->xtab = mandoc_calloc(128,
1921: sizeof(struct roffstr));
1.167 kristaps 1922:
1923: free(r->xtab[(int)*first].p);
1924: r->xtab[(int)*first].p = mandoc_strndup(second, ssz);
1925: r->xtab[(int)*first].sz = ssz;
1.164 kristaps 1926: }
1927:
1928: return(ROFF_IGN);
1929: }
1930:
1931: static enum rofferr
1.105 kristaps 1932: roff_so(ROFF_ARGS)
1933: {
1934: char *name;
1935:
1.210 schwarze 1936: name = *bufp + pos;
1937: mandoc_vmsg(MANDOCERR_SO, r->parse, ln, ppos, ".so %s", name);
1.105 kristaps 1938:
1939: /*
1940: * Handle `so'. Be EXTREMELY careful, as we shouldn't be
1941: * opening anything that's not in our cwd or anything beneath
1942: * it. Thus, explicitly disallow traversing up the file-system
1943: * or using absolute paths.
1944: */
1945:
1946: if ('/' == *name || strstr(name, "../") || strstr(name, "/..")) {
1.210 schwarze 1947: mandoc_vmsg(MANDOCERR_SO_PATH, r->parse, ln, ppos,
1948: ".so %s", name);
1.105 kristaps 1949: return(ROFF_ERR);
1950: }
1951:
1952: *offs = pos;
1953: return(ROFF_SO);
1954: }
1.92 schwarze 1955:
1.106 kristaps 1956: static enum rofferr
1957: roff_userdef(ROFF_ARGS)
1.99 kristaps 1958: {
1.106 kristaps 1959: const char *arg[9];
1960: char *cp, *n1, *n2;
1.119 schwarze 1961: int i;
1.106 kristaps 1962:
1963: /*
1964: * Collect pointers to macro argument strings
1.188 schwarze 1965: * and NUL-terminate them.
1.106 kristaps 1966: */
1967: cp = *bufp + pos;
1.119 schwarze 1968: for (i = 0; i < 9; i++)
1.120 schwarze 1969: arg[i] = '\0' == *cp ? "" :
1.136 kristaps 1970: mandoc_getarg(r->parse, &cp, ln, &pos);
1.99 kristaps 1971:
1.106 kristaps 1972: /*
1973: * Expand macro arguments.
1.99 kristaps 1974: */
1.106 kristaps 1975: *szp = 0;
1976: n1 = cp = mandoc_strdup(r->current_string);
1977: while (NULL != (cp = strstr(cp, "\\$"))) {
1978: i = cp[2] - '1';
1979: if (0 > i || 8 < i) {
1980: /* Not an argument invocation. */
1981: cp += 2;
1982: continue;
1983: }
1.209 schwarze 1984: *cp = '\0';
1985: *szp = mandoc_asprintf(&n2, "%s%s%s",
1986: n1, arg[i], cp + 3) + 1;
1.106 kristaps 1987: cp = n2 + (cp - n1);
1988: free(n1);
1989: n1 = n2;
1.99 kristaps 1990: }
1991:
1.106 kristaps 1992: /*
1993: * Replace the macro invocation
1994: * by the expanded macro.
1995: */
1996: free(*bufp);
1997: *bufp = n1;
1998: if (0 == *szp)
1999: *szp = strlen(*bufp) + 1;
2000:
2001: return(*szp > 1 && '\n' == (*bufp)[(int)*szp - 2] ?
2002: ROFF_REPARSE : ROFF_APPEND);
1.99 kristaps 2003: }
1.121 schwarze 2004:
1.212 schwarze 2005: static size_t
1.121 schwarze 2006: roff_getname(struct roff *r, char **cpp, int ln, int pos)
2007: {
2008: char *name, *cp;
1.212 schwarze 2009: size_t namesz;
1.121 schwarze 2010:
2011: name = *cpp;
2012: if ('\0' == *name)
1.212 schwarze 2013: return(0);
1.121 schwarze 2014:
1.212 schwarze 2015: /* Read until end of name and terminate it with NUL. */
2016: for (cp = name; 1; cp++) {
2017: if ('\0' == *cp || ' ' == *cp) {
2018: namesz = cp - name;
2019: break;
2020: }
1.121 schwarze 2021: if ('\\' != *cp)
2022: continue;
1.215 schwarze 2023: namesz = cp - name;
2024: if ('{' == cp[1] || '}' == cp[1])
2025: break;
1.121 schwarze 2026: cp++;
2027: if ('\\' == *cp)
2028: continue;
1.128 kristaps 2029: mandoc_msg(MANDOCERR_NAMESC, r->parse, ln, pos, NULL);
1.212 schwarze 2030: mandoc_escape((const char **)&cp, NULL, NULL);
2031: break;
1.121 schwarze 2032: }
2033:
2034: /* Read past spaces. */
2035: while (' ' == *cp)
2036: cp++;
2037:
2038: *cpp = cp;
1.212 schwarze 2039: return(namesz);
1.121 schwarze 2040: }
2041:
1.106 kristaps 2042: /*
2043: * Store *string into the user-defined string called *name.
2044: * To clear an existing entry, call with (*r, *name, NULL, 0).
1.193 schwarze 2045: * append == 0: replace mode
2046: * append == 1: single-line append mode
2047: * append == 2: multiline append mode, append '\n' after each call
1.106 kristaps 2048: */
1.94 kristaps 2049: static void
1.106 kristaps 2050: roff_setstr(struct roff *r, const char *name, const char *string,
1.193 schwarze 2051: int append)
1.92 schwarze 2052: {
1.164 kristaps 2053:
2054: roff_setstrn(&r->strtab, name, strlen(name), string,
1.207 schwarze 2055: string ? strlen(string) : 0, append);
1.164 kristaps 2056: }
2057:
2058: static void
1.166 kristaps 2059: roff_setstrn(struct roffkv **r, const char *name, size_t namesz,
1.193 schwarze 2060: const char *string, size_t stringsz, int append)
1.164 kristaps 2061: {
1.166 kristaps 2062: struct roffkv *n;
1.164 kristaps 2063: char *c;
2064: int i;
2065: size_t oldch, newch;
1.92 schwarze 2066:
1.106 kristaps 2067: /* Search for an existing string with the same name. */
1.164 kristaps 2068: n = *r;
2069:
1.211 schwarze 2070: while (n && (namesz != n->key.sz ||
2071: strncmp(n->key.p, name, namesz)))
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