Annotation of mandoc/roff.c, Revision 1.218
1.218 ! schwarze 1: /* $Id: roff.c,v 1.217 2014/07/04 16:12:08 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 */
1.218 ! schwarze 503: int arg_complete; /* argument not interrupted by eol */
1.206 schwarze 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:
1.218 ! schwarze 588: arg_complete = 1;
1.181 schwarze 589: for (naml = 0; 0 == maxl || naml < maxl; naml++, cp++) {
1.153 kristaps 590: if ('\0' == *cp) {
1.207 schwarze 591: mandoc_msg(MANDOCERR_BADESCAPE, r->parse,
592: ln, (int)(stesc - *bufp), NULL);
1.218 ! schwarze 593: arg_complete = 0;
1.206 schwarze 594: break;
1.153 kristaps 595: }
1.206 schwarze 596: if (0 == maxl && *cp == term) {
597: cp++;
1.94 kristaps 598: break;
1.206 schwarze 599: }
1.94 kristaps 600: }
601:
1.108 schwarze 602: /*
603: * Retrieve the replacement string; if it is
604: * undefined, resume searching for escapes.
605: */
606:
1.206 schwarze 607: switch (stesc[1]) {
1.207 schwarze 608: case '*':
1.218 ! schwarze 609: if (arg_complete)
! 610: res = roff_getstrn(r, stnam, naml);
1.206 schwarze 611: break;
1.207 schwarze 612: case 'B':
1.206 schwarze 613: npos = 0;
1.218 ! schwarze 614: ubuf[0] = arg_complete &&
! 615: roff_evalnum(stnam, &npos, NULL, 0) &&
! 616: stnam + npos + 1 == cp ? '1' : '0';
1.206 schwarze 617: ubuf[1] = '\0';
618: break;
1.207 schwarze 619: case 'n':
1.218 ! schwarze 620: if (arg_complete)
! 621: (void)snprintf(ubuf, sizeof(ubuf), "%d",
! 622: roff_getregn(r, stnam, naml));
! 623: else
! 624: ubuf[0] = '\0';
1.206 schwarze 625: break;
1.207 schwarze 626: case 'w':
1.218 ! schwarze 627: /* use even incomplete args */
1.208 schwarze 628: (void)snprintf(ubuf, sizeof(ubuf), "%d",
1.206 schwarze 629: 24 * (int)naml);
630: break;
631: }
1.94 kristaps 632:
633: if (NULL == res) {
1.207 schwarze 634: mandoc_msg(MANDOCERR_BADESCAPE, r->parse,
635: ln, (int)(stesc - *bufp), NULL);
1.142 kristaps 636: res = "";
1.94 kristaps 637: }
638:
1.108 schwarze 639: /* Replace the escape sequence by the string. */
640:
1.209 schwarze 641: *stesc = '\0';
642: *szp = mandoc_asprintf(&nbuf, "%s%s%s",
643: *bufp, res, cp) + 1;
1.94 kristaps 644:
1.205 schwarze 645: /* Prepare for the next replacement. */
1.94 kristaps 646:
1.205 schwarze 647: start = nbuf + pos;
1.209 schwarze 648: stesc = nbuf + (stesc - *bufp) + strlen(res);
1.94 kristaps 649: free(*bufp);
1.181 schwarze 650: *bufp = nbuf;
1.154 kristaps 651: }
1.172 schwarze 652: return(ROFF_CONT);
1.154 kristaps 653: }
654:
655: /*
1.178 schwarze 656: * Process text streams:
657: * Convert all breakable hyphens into ASCII_HYPH.
658: * Decrement and spring input line trap.
1.154 kristaps 659: */
660: static enum rofferr
1.178 schwarze 661: roff_parsetext(char **bufp, size_t *szp, int pos, int *offs)
1.154 kristaps 662: {
663: size_t sz;
664: const char *start;
1.178 schwarze 665: char *p;
666: int isz;
1.154 kristaps 667: enum mandoc_esc esc;
668:
1.178 schwarze 669: start = p = *bufp + pos;
1.154 kristaps 670:
671: while ('\0' != *p) {
672: sz = strcspn(p, "-\\");
673: p += sz;
674:
1.159 kristaps 675: if ('\0' == *p)
676: break;
677:
1.154 kristaps 678: if ('\\' == *p) {
679: /* Skip over escapes. */
680: p++;
1.189 schwarze 681: esc = mandoc_escape((const char **)&p, NULL, NULL);
1.154 kristaps 682: if (ESCAPE_ERROR == esc)
683: break;
1.155 kristaps 684: continue;
1.159 kristaps 685: } else if (p == start) {
1.158 kristaps 686: p++;
1.155 kristaps 687: continue;
1.158 kristaps 688: }
1.155 kristaps 689:
1.171 schwarze 690: if (isalpha((unsigned char)p[-1]) &&
691: isalpha((unsigned char)p[1]))
1.155 kristaps 692: *p = ASCII_HYPH;
693: p++;
1.94 kristaps 694: }
695:
1.178 schwarze 696: /* Spring the input line trap. */
697: if (1 == roffit_lines) {
1.202 schwarze 698: isz = mandoc_asprintf(&p, "%s\n.%s", *bufp, roffit_macro);
1.178 schwarze 699: free(*bufp);
700: *bufp = p;
701: *szp = isz + 1;
702: *offs = 0;
703: free(roffit_macro);
704: roffit_lines = 0;
705: return(ROFF_REPARSE);
706: } else if (1 < roffit_lines)
707: --roffit_lines;
1.154 kristaps 708: return(ROFF_CONT);
1.94 kristaps 709: }
710:
1.67 kristaps 711: enum rofferr
1.207 schwarze 712: roff_parseln(struct roff *r, int ln, char **bufp,
1.90 kristaps 713: size_t *szp, int pos, int *offs)
1.67 kristaps 714: {
715: enum rofft t;
1.109 kristaps 716: enum rofferr e;
1.130 kristaps 717: int ppos, ctl;
1.79 kristaps 718:
719: /*
1.94 kristaps 720: * Run the reserved-word filter only if we have some reserved
721: * words to fill in.
722: */
723:
1.172 schwarze 724: e = roff_res(r, bufp, szp, ln, pos);
725: if (ROFF_IGN == e)
726: return(e);
727: assert(ROFF_CONT == e);
1.94 kristaps 728:
1.130 kristaps 729: ppos = pos;
1.174 kristaps 730: ctl = roff_getcontrol(r, *bufp, &pos);
1.130 kristaps 731:
1.94 kristaps 732: /*
1.79 kristaps 733: * First, if a scope is open and we're not a macro, pass the
734: * text through the macro's filter. If a scope isn't open and
735: * we're not a macro, just let it through.
1.125 kristaps 736: * Finally, if there's an equation scope open, divert it into it
737: * no matter our state.
1.79 kristaps 738: */
1.74 kristaps 739:
1.130 kristaps 740: if (r->last && ! ctl) {
1.78 kristaps 741: t = r->last->tok;
742: assert(roffs[t].text);
1.207 schwarze 743: e = (*roffs[t].text)(r, t, bufp, szp, ln, pos, pos, offs);
1.109 kristaps 744: assert(ROFF_IGN == e || ROFF_CONT == e);
1.125 kristaps 745: if (ROFF_CONT != e)
746: return(e);
1.182 schwarze 747: }
748: if (r->eqn)
749: return(eqn_read(&r->eqn, ln, *bufp, ppos, offs));
750: if ( ! ctl) {
1.125 kristaps 751: if (r->tbl)
1.130 kristaps 752: return(tbl_read(r->tbl, ln, *bufp, pos));
1.178 schwarze 753: return(roff_parsetext(bufp, szp, pos, offs));
1.182 schwarze 754: }
1.67 kristaps 755:
1.79 kristaps 756: /*
757: * If a scope is open, go to the child handler for that macro,
758: * as it may want to preprocess before doing anything with it.
1.125 kristaps 759: * Don't do so if an equation is open.
1.79 kristaps 760: */
1.78 kristaps 761:
1.79 kristaps 762: if (r->last) {
763: t = r->last->tok;
764: assert(roffs[t].sub);
1.207 schwarze 765: return((*roffs[t].sub)(r, t, bufp, szp,
766: ln, ppos, pos, offs));
1.79 kristaps 767: }
1.78 kristaps 768:
1.79 kristaps 769: /*
770: * Lastly, as we've no scope open, try to look up and execute
771: * the new macro. If no macro is found, simply return and let
772: * the compilers handle it.
773: */
1.67 kristaps 774:
1.214 schwarze 775: if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos, ln, ppos)))
1.79 kristaps 776: return(ROFF_CONT);
1.67 kristaps 777:
1.75 kristaps 778: assert(roffs[t].proc);
1.207 schwarze 779: return((*roffs[t].proc)(r, t, bufp, szp, ln, ppos, pos, offs));
1.74 kristaps 780: }
781:
1.117 kristaps 782: void
1.74 kristaps 783: roff_endparse(struct roff *r)
784: {
785:
1.110 kristaps 786: if (r->last)
1.128 kristaps 787: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.207 schwarze 788: r->last->line, r->last->col, NULL);
1.117 kristaps 789:
1.125 kristaps 790: if (r->eqn) {
1.207 schwarze 791: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
792: r->eqn->eqn.ln, r->eqn->eqn.pos, NULL);
1.151 kristaps 793: eqn_end(&r->eqn);
1.125 kristaps 794: }
795:
1.117 kristaps 796: if (r->tbl) {
1.207 schwarze 797: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
798: r->tbl->line, r->tbl->pos, NULL);
1.151 kristaps 799: tbl_end(&r->tbl);
1.117 kristaps 800: }
1.67 kristaps 801: }
802:
803: /*
804: * Parse a roff node's type from the input buffer. This must be in the
805: * form of ".foo xxx" in the usual way.
806: */
807: static enum rofft
1.214 schwarze 808: roff_parse(struct roff *r, char *buf, int *pos, int ln, int ppos)
1.67 kristaps 809: {
1.214 schwarze 810: char *cp;
1.106 kristaps 811: const char *mac;
812: size_t maclen;
1.67 kristaps 813: enum rofft t;
814:
1.214 schwarze 815: cp = buf + *pos;
816:
817: if ('\0' == *cp || '"' == *cp || '\t' == *cp || ' ' == *cp)
1.67 kristaps 818: return(ROFF_MAX);
819:
1.214 schwarze 820: mac = cp;
821: maclen = roff_getname(r, &cp, ln, ppos);
1.67 kristaps 822:
1.106 kristaps 823: t = (r->current_string = roff_getstrn(r, mac, maclen))
1.155 kristaps 824: ? ROFF_USERDEF : roffhash_find(mac, maclen);
1.67 kristaps 825:
1.214 schwarze 826: if (ROFF_MAX != t)
827: *pos = cp - buf;
1.67 kristaps 828:
829: return(t);
830: }
831:
832: static enum rofferr
1.76 kristaps 833: roff_cblock(ROFF_ARGS)
1.67 kristaps 834: {
835:
1.79 kristaps 836: /*
837: * A block-close `..' should only be invoked as a child of an
838: * ignore macro, otherwise raise a warning and just ignore it.
839: */
840:
1.76 kristaps 841: if (NULL == r->last) {
1.128 kristaps 842: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.76 kristaps 843: return(ROFF_IGN);
844: }
1.67 kristaps 845:
1.81 kristaps 846: switch (r->last->tok) {
1.207 schwarze 847: case ROFF_am:
1.81 kristaps 848: /* FALLTHROUGH */
1.207 schwarze 849: case ROFF_ami:
1.81 kristaps 850: /* FALLTHROUGH */
1.207 schwarze 851: case ROFF_am1:
1.81 kristaps 852: /* FALLTHROUGH */
1.207 schwarze 853: case ROFF_de:
1.108 schwarze 854: /* ROFF_de1 is remapped to ROFF_de in roff_block(). */
1.81 kristaps 855: /* FALLTHROUGH */
1.207 schwarze 856: case ROFF_dei:
1.81 kristaps 857: /* FALLTHROUGH */
1.207 schwarze 858: case ROFF_ig:
1.81 kristaps 859: break;
860: default:
1.128 kristaps 861: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.67 kristaps 862: return(ROFF_IGN);
1.76 kristaps 863: }
1.67 kristaps 864:
1.76 kristaps 865: if ((*bufp)[pos])
1.217 schwarze 866: mandoc_vmsg(MANDOCERR_ARG_SKIP, r->parse, ln, pos,
867: ".. %s", *bufp + pos);
1.71 kristaps 868:
869: roffnode_pop(r);
1.76 kristaps 870: roffnode_cleanscope(r);
871: return(ROFF_IGN);
1.71 kristaps 872:
1.67 kristaps 873: }
874:
1.76 kristaps 875: static void
876: roffnode_cleanscope(struct roff *r)
1.67 kristaps 877: {
878:
1.76 kristaps 879: while (r->last) {
1.173 schwarze 880: if (--r->last->endspan != 0)
1.76 kristaps 881: break;
882: roffnode_pop(r);
883: }
1.67 kristaps 884: }
885:
1.195 schwarze 886: static void
887: roff_ccond(struct roff *r, int ln, int ppos)
1.74 kristaps 888: {
889:
1.76 kristaps 890: if (NULL == r->last) {
1.128 kristaps 891: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.195 schwarze 892: return;
1.76 kristaps 893: }
894:
1.82 kristaps 895: switch (r->last->tok) {
1.207 schwarze 896: case ROFF_el:
1.82 kristaps 897: /* FALLTHROUGH */
1.207 schwarze 898: case ROFF_ie:
1.82 kristaps 899: /* FALLTHROUGH */
1.207 schwarze 900: case ROFF_if:
1.82 kristaps 901: break;
902: default:
1.128 kristaps 903: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.195 schwarze 904: return;
1.75 kristaps 905: }
906:
1.76 kristaps 907: if (r->last->endspan > -1) {
1.128 kristaps 908: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.195 schwarze 909: return;
1.76 kristaps 910: }
911:
1.75 kristaps 912: roffnode_pop(r);
1.76 kristaps 913: roffnode_cleanscope(r);
1.195 schwarze 914: return;
1.76 kristaps 915: }
916:
917: static enum rofferr
1.80 kristaps 918: roff_block(ROFF_ARGS)
1.76 kristaps 919: {
1.213 schwarze 920: char *name, *cp;
921: size_t namesz;
1.106 kristaps 922:
1.213 schwarze 923: name = cp = *bufp + pos;
924: namesz = 0;
1.76 kristaps 925:
1.106 kristaps 926: if (ROFF_ig != tok) {
1.213 schwarze 927: if ('\0' == *cp) {
1.216 schwarze 928: mandoc_msg(MANDOCERR_REQ_EMPTY, r->parse,
929: ln, ppos, roffs[tok].name);
1.106 kristaps 930: return(ROFF_IGN);
931: }
1.107 kristaps 932:
933: /*
934: * Re-write `de1', since we don't really care about
935: * groff's strange compatibility mode, into `de'.
936: */
937:
1.106 kristaps 938: if (ROFF_de1 == tok)
939: tok = ROFF_de;
1.213 schwarze 940: else if (ROFF_de != tok)
1.128 kristaps 941: mandoc_msg(MANDOCERR_REQUEST, r->parse, ln, ppos,
1.106 kristaps 942: roffs[tok].name);
1.107 kristaps 943:
1.213 schwarze 944: namesz = roff_getname(r, &cp, ln, ppos);
945: name[namesz] = '\0';
946: } else
947: name = NULL;
1.80 kristaps 948:
1.106 kristaps 949: roffnode_push(r, tok, name, ln, ppos);
950:
951: /*
952: * At the beginning of a `de' macro, clear the existing string
953: * with the same name, if there is one. New content will be
1.193 schwarze 954: * appended from roff_block_text() in multiline mode.
1.106 kristaps 955: */
1.107 kristaps 956:
1.213 schwarze 957: if (namesz && ROFF_de == tok)
958: roff_setstrn(&r->strtab, name, namesz, "", 0, 0);
1.76 kristaps 959:
1.213 schwarze 960: if ('\0' == *cp)
1.78 kristaps 961: return(ROFF_IGN);
962:
1.107 kristaps 963: /* If present, process the custom end-of-line marker. */
964:
1.213 schwarze 965: name = cp;
966: namesz = roff_getname(r, &cp, ln, ppos);
967: if (namesz)
968: r->last->end = mandoc_strndup(name, namesz);
1.78 kristaps 969:
1.213 schwarze 970: if ('\0' != *cp)
1.217 schwarze 971: mandoc_vmsg(MANDOCERR_ARG_EXCESS, r->parse,
972: ln, pos, ".%s ... %s", roffs[tok].name, cp);
1.74 kristaps 973:
1.78 kristaps 974: return(ROFF_IGN);
975: }
976:
977: static enum rofferr
1.80 kristaps 978: roff_block_sub(ROFF_ARGS)
1.79 kristaps 979: {
980: enum rofft t;
981: int i, j;
982:
983: /*
984: * First check whether a custom macro exists at this level. If
985: * it does, then check against it. This is some of groff's
986: * stranger behaviours. If we encountered a custom end-scope
987: * tag and that tag also happens to be a "real" macro, then we
988: * need to try interpreting it again as a real macro. If it's
989: * not, then return ignore. Else continue.
990: */
991:
992: if (r->last->end) {
1.130 kristaps 993: for (i = pos, j = 0; r->last->end[j]; j++, i++)
1.79 kristaps 994: if ((*bufp)[i] != r->last->end[j])
995: break;
996:
1.207 schwarze 997: if ('\0' == r->last->end[j] &&
998: ('\0' == (*bufp)[i] ||
999: ' ' == (*bufp)[i] ||
1000: '\t' == (*bufp)[i])) {
1.79 kristaps 1001: roffnode_pop(r);
1002: roffnode_cleanscope(r);
1003:
1.130 kristaps 1004: while (' ' == (*bufp)[i] || '\t' == (*bufp)[i])
1005: i++;
1006:
1007: pos = i;
1.214 schwarze 1008: if (ROFF_MAX != roff_parse(r, *bufp, &pos, ln, ppos))
1.79 kristaps 1009: return(ROFF_RERUN);
1010: return(ROFF_IGN);
1011: }
1012: }
1013:
1014: /*
1015: * If we have no custom end-query or lookup failed, then try
1016: * pulling it out of the hashtable.
1017: */
1018:
1.214 schwarze 1019: t = roff_parse(r, *bufp, &pos, ln, ppos);
1.79 kristaps 1020:
1.106 kristaps 1021: /*
1022: * Macros other than block-end are only significant
1023: * in `de' blocks; elsewhere, simply throw them away.
1024: */
1025: if (ROFF_cblock != t) {
1026: if (ROFF_de == tok)
1.193 schwarze 1027: roff_setstr(r, r->last->name, *bufp + ppos, 2);
1.79 kristaps 1028: return(ROFF_IGN);
1.106 kristaps 1029: }
1.79 kristaps 1030:
1031: assert(roffs[t].proc);
1.207 schwarze 1032: return((*roffs[t].proc)(r, t, bufp, szp, ln, ppos, pos, offs));
1.79 kristaps 1033: }
1034:
1035: static enum rofferr
1.80 kristaps 1036: roff_block_text(ROFF_ARGS)
1.78 kristaps 1037: {
1038:
1.106 kristaps 1039: if (ROFF_de == tok)
1.193 schwarze 1040: roff_setstr(r, r->last->name, *bufp + pos, 2);
1.106 kristaps 1041:
1.78 kristaps 1042: return(ROFF_IGN);
1043: }
1044:
1045: static enum rofferr
1.82 kristaps 1046: roff_cond_sub(ROFF_ARGS)
1047: {
1048: enum rofft t;
1.139 kristaps 1049: char *ep;
1.198 schwarze 1050: int rr;
1.82 kristaps 1051:
1052: rr = r->last->rule;
1.139 kristaps 1053: roffnode_cleanscope(r);
1.214 schwarze 1054: t = roff_parse(r, *bufp, &pos, ln, ppos);
1.82 kristaps 1055:
1.139 kristaps 1056: /*
1.177 schwarze 1057: * Fully handle known macros when they are structurally
1058: * required or when the conditional evaluated to true.
1.87 kristaps 1059: */
1060:
1.177 schwarze 1061: if ((ROFF_MAX != t) &&
1.198 schwarze 1062: (rr || ROFFMAC_STRUCT & roffs[t].flags)) {
1.177 schwarze 1063: assert(roffs[t].proc);
1064: return((*roffs[t].proc)(r, t, bufp, szp,
1.207 schwarze 1065: ln, ppos, pos, offs));
1.177 schwarze 1066: }
1.144 kristaps 1067:
1.196 schwarze 1068: /*
1069: * If `\}' occurs on a macro line without a preceding macro,
1070: * drop the line completely.
1071: */
1072:
1073: ep = *bufp + pos;
1074: if ('\\' == ep[0] && '}' == ep[1])
1.198 schwarze 1075: rr = 0;
1.196 schwarze 1076:
1.177 schwarze 1077: /* Always check for the closing delimiter `\}'. */
1.144 kristaps 1078:
1.177 schwarze 1079: while (NULL != (ep = strchr(ep, '\\'))) {
1.197 schwarze 1080: if ('}' == *(++ep)) {
1081: *ep = '&';
1082: roff_ccond(r, ln, ep - *bufp - 1);
1083: }
1084: ++ep;
1.177 schwarze 1085: }
1.198 schwarze 1086: return(rr ? ROFF_CONT : ROFF_IGN);
1.82 kristaps 1087: }
1088:
1089: static enum rofferr
1090: roff_cond_text(ROFF_ARGS)
1.78 kristaps 1091: {
1.140 kristaps 1092: char *ep;
1.198 schwarze 1093: int rr;
1.82 kristaps 1094:
1095: rr = r->last->rule;
1.140 kristaps 1096: roffnode_cleanscope(r);
1.82 kristaps 1097:
1.197 schwarze 1098: ep = *bufp + pos;
1099: while (NULL != (ep = strchr(ep, '\\'))) {
1100: if ('}' == *(++ep)) {
1101: *ep = '&';
1102: roff_ccond(r, ln, ep - *bufp - 1);
1103: }
1104: ++ep;
1.78 kristaps 1105: }
1.198 schwarze 1106: return(rr ? ROFF_CONT : ROFF_IGN);
1.74 kristaps 1107: }
1108:
1.204 schwarze 1109: /*
1110: * Parse a single signed integer number. Stop at the first non-digit.
1111: * If there is at least one digit, return success and advance the
1112: * parse point, else return failure and let the parse point unchanged.
1113: * Ignore overflows, treat them just like the C language.
1114: */
1.184 schwarze 1115: static int
1116: roff_getnum(const char *v, int *pos, int *res)
1117: {
1.206 schwarze 1118: int myres, n, p;
1119:
1120: if (NULL == res)
1121: res = &myres;
1.184 schwarze 1122:
1123: p = *pos;
1124: n = v[p] == '-';
1125: if (n)
1126: p++;
1127:
1128: for (*res = 0; isdigit((unsigned char)v[p]); p++)
1.204 schwarze 1129: *res = 10 * *res + v[p] - '0';
1.184 schwarze 1130: if (p == *pos + n)
1131: return 0;
1132:
1133: if (n)
1134: *res = -*res;
1135:
1136: *pos = p;
1137: return 1;
1138: }
1139:
1.198 schwarze 1140: /*
1141: * Evaluate a string comparison condition.
1142: * The first character is the delimiter.
1143: * Succeed if the string up to its second occurrence
1144: * matches the string up to its third occurence.
1145: * Advance the cursor after the third occurrence
1146: * or lacking that, to the end of the line.
1147: */
1148: static int
1149: roff_evalstrcond(const char *v, int *pos)
1150: {
1151: const char *s1, *s2, *s3;
1152: int match;
1153:
1154: match = 0;
1155: s1 = v + *pos; /* initial delimiter */
1156: s2 = s1 + 1; /* for scanning the first string */
1157: s3 = strchr(s2, *s1); /* for scanning the second string */
1158:
1159: if (NULL == s3) /* found no middle delimiter */
1160: goto out;
1161:
1162: while ('\0' != *++s3) {
1163: if (*s2 != *s3) { /* mismatch */
1164: s3 = strchr(s3, *s1);
1165: break;
1166: }
1167: if (*s3 == *s1) { /* found the final delimiter */
1168: match = 1;
1169: break;
1170: }
1171: s2++;
1172: }
1173:
1174: out:
1175: if (NULL == s3)
1176: s3 = strchr(s2, '\0');
1177: else
1178: s3++;
1179: *pos = s3 - v;
1180: return(match);
1181: }
1182:
1.204 schwarze 1183: /*
1184: * Evaluate an optionally negated single character, numerical,
1185: * or string condition.
1186: */
1.198 schwarze 1187: static int
1.88 kristaps 1188: roff_evalcond(const char *v, int *pos)
1189: {
1.204 schwarze 1190: int wanttrue, number;
1.88 kristaps 1191:
1.198 schwarze 1192: if ('!' == v[*pos]) {
1193: wanttrue = 0;
1194: (*pos)++;
1195: } else
1196: wanttrue = 1;
1197:
1.88 kristaps 1198: switch (v[*pos]) {
1.207 schwarze 1199: case 'n':
1.198 schwarze 1200: /* FALLTHROUGH */
1.207 schwarze 1201: case 'o':
1.88 kristaps 1202: (*pos)++;
1.198 schwarze 1203: return(wanttrue);
1.207 schwarze 1204: case 'c':
1.198 schwarze 1205: /* FALLTHROUGH */
1.207 schwarze 1206: case 'd':
1.198 schwarze 1207: /* FALLTHROUGH */
1.207 schwarze 1208: case 'e':
1.88 kristaps 1209: /* FALLTHROUGH */
1.207 schwarze 1210: case 'r':
1.88 kristaps 1211: /* FALLTHROUGH */
1.207 schwarze 1212: case 't':
1.88 kristaps 1213: (*pos)++;
1.198 schwarze 1214: return(!wanttrue);
1.88 kristaps 1215: default:
1216: break;
1217: }
1218:
1.204 schwarze 1219: if (roff_evalnum(v, pos, &number, 0))
1220: return((number > 0) == wanttrue);
1221: else
1.198 schwarze 1222: return(roff_evalstrcond(v, pos) == wanttrue);
1.88 kristaps 1223: }
1224:
1.74 kristaps 1225: static enum rofferr
1.103 kristaps 1226: roff_line_ignore(ROFF_ARGS)
1.89 kristaps 1227: {
1.123 schwarze 1228:
1.89 kristaps 1229: return(ROFF_IGN);
1230: }
1231:
1.104 kristaps 1232: static enum rofferr
1.82 kristaps 1233: roff_cond(ROFF_ARGS)
1.74 kristaps 1234: {
1.173 schwarze 1235:
1236: roffnode_push(r, tok, NULL, ln, ppos);
1.74 kristaps 1237:
1.207 schwarze 1238: /*
1.134 kristaps 1239: * An `.el' has no conditional body: it will consume the value
1240: * of the current rstack entry set in prior `ie' calls or
1.207 schwarze 1241: * defaults to DENY.
1.134 kristaps 1242: *
1243: * If we're not an `el', however, then evaluate the conditional.
1244: */
1.133 kristaps 1245:
1.173 schwarze 1246: r->last->rule = ROFF_el == tok ?
1.207 schwarze 1247: (r->rstackpos < 0 ? 0 : r->rstack[r->rstackpos--]) :
1248: roff_evalcond(*bufp, &pos);
1.77 kristaps 1249:
1.134 kristaps 1250: /*
1251: * An if-else will put the NEGATION of the current evaluated
1252: * conditional into the stack of rules.
1253: */
1254:
1.84 schwarze 1255: if (ROFF_ie == tok) {
1.134 kristaps 1256: if (r->rstackpos == RSTACK_MAX - 1) {
1.207 schwarze 1257: mandoc_msg(MANDOCERR_MEM,
1258: r->parse, ln, ppos, NULL);
1.134 kristaps 1259: return(ROFF_ERR);
1260: }
1.198 schwarze 1261: r->rstack[++r->rstackpos] = !r->last->rule;
1.82 kristaps 1262: }
1.88 kristaps 1263:
1264: /* If the parent has false as its rule, then so do we. */
1265:
1.198 schwarze 1266: if (r->last->parent && !r->last->parent->rule)
1267: r->last->rule = 0;
1.88 kristaps 1268:
1269: /*
1.173 schwarze 1270: * Determine scope.
1271: * If there is nothing on the line after the conditional,
1272: * not even whitespace, use next-line scope.
1.88 kristaps 1273: */
1.74 kristaps 1274:
1.173 schwarze 1275: if ('\0' == (*bufp)[pos]) {
1276: r->last->endspan = 2;
1277: goto out;
1278: }
1279:
1280: while (' ' == (*bufp)[pos])
1281: pos++;
1282:
1283: /* An opening brace requests multiline scope. */
1.75 kristaps 1284:
1285: if ('\\' == (*bufp)[pos] && '{' == (*bufp)[pos + 1]) {
1286: r->last->endspan = -1;
1287: pos += 2;
1.173 schwarze 1288: goto out;
1.207 schwarze 1289: }
1.74 kristaps 1290:
1.77 kristaps 1291: /*
1.173 schwarze 1292: * Anything else following the conditional causes
1293: * single-line scope. Warn if the scope contains
1294: * nothing but trailing whitespace.
1.77 kristaps 1295: */
1296:
1.75 kristaps 1297: if ('\0' == (*bufp)[pos])
1.216 schwarze 1298: mandoc_msg(MANDOCERR_COND_EMPTY, r->parse,
1299: ln, ppos, roffs[tok].name);
1.77 kristaps 1300:
1.173 schwarze 1301: r->last->endspan = 1;
1.74 kristaps 1302:
1.173 schwarze 1303: out:
1.75 kristaps 1304: *offs = pos;
1305: return(ROFF_RERUN);
1.83 schwarze 1306: }
1307:
1308: static enum rofferr
1.92 schwarze 1309: roff_ds(ROFF_ARGS)
1310: {
1.212 schwarze 1311: char *string;
1312: const char *name;
1313: size_t namesz;
1.96 kristaps 1314:
1315: /*
1.212 schwarze 1316: * The first word is the name of the string.
1317: * If it is empty or terminated by an escape sequence,
1318: * abort the `ds' request without defining anything.
1.96 kristaps 1319: */
1.92 schwarze 1320:
1.212 schwarze 1321: name = string = *bufp + pos;
1.92 schwarze 1322: if ('\0' == *name)
1323: return(ROFF_IGN);
1324:
1.212 schwarze 1325: namesz = roff_getname(r, &string, ln, pos);
1326: if ('\\' == name[namesz])
1327: return(ROFF_IGN);
1328:
1329: /* Read past the initial double-quote, if any. */
1.121 schwarze 1330: if ('"' == *string)
1.92 schwarze 1331: string++;
1332:
1.96 kristaps 1333: /* The rest is the value. */
1.212 schwarze 1334: roff_setstrn(&r->strtab, name, namesz, string, strlen(string),
1335: ROFF_as == tok);
1.92 schwarze 1336: return(ROFF_IGN);
1337: }
1338:
1.204 schwarze 1339: /*
1340: * Parse a single operator, one or two characters long.
1341: * If the operator is recognized, return success and advance the
1342: * parse point, else return failure and let the parse point unchanged.
1343: */
1344: static int
1345: roff_getop(const char *v, int *pos, char *res)
1346: {
1347:
1348: *res = v[*pos];
1349:
1350: switch (*res) {
1.207 schwarze 1351: case '+':
1.204 schwarze 1352: /* FALLTHROUGH */
1.207 schwarze 1353: case '-':
1.204 schwarze 1354: /* FALLTHROUGH */
1.207 schwarze 1355: case '*':
1.204 schwarze 1356: /* FALLTHROUGH */
1.207 schwarze 1357: case '/':
1.204 schwarze 1358: /* FALLTHROUGH */
1.207 schwarze 1359: case '%':
1.204 schwarze 1360: /* FALLTHROUGH */
1.207 schwarze 1361: case '&':
1.204 schwarze 1362: /* FALLTHROUGH */
1.207 schwarze 1363: case ':':
1.204 schwarze 1364: break;
1365: case '<':
1366: switch (v[*pos + 1]) {
1.207 schwarze 1367: case '=':
1.204 schwarze 1368: *res = 'l';
1369: (*pos)++;
1370: break;
1.207 schwarze 1371: case '>':
1.204 schwarze 1372: *res = '!';
1373: (*pos)++;
1374: break;
1.207 schwarze 1375: case '?':
1.204 schwarze 1376: *res = 'i';
1377: (*pos)++;
1378: break;
1379: default:
1380: break;
1381: }
1382: break;
1383: case '>':
1384: switch (v[*pos + 1]) {
1.207 schwarze 1385: case '=':
1.204 schwarze 1386: *res = 'g';
1387: (*pos)++;
1388: break;
1.207 schwarze 1389: case '?':
1.204 schwarze 1390: *res = 'a';
1391: (*pos)++;
1392: break;
1393: default:
1394: break;
1395: }
1396: break;
1397: case '=':
1398: if ('=' == v[*pos + 1])
1399: (*pos)++;
1400: break;
1401: default:
1402: return(0);
1403: }
1404: (*pos)++;
1405:
1406: return(*res);
1407: }
1408:
1409: /*
1410: * Evaluate either a parenthesized numeric expression
1411: * or a single signed integer number.
1412: */
1413: static int
1414: roff_evalpar(const char *v, int *pos, int *res)
1415: {
1416:
1417: if ('(' != v[*pos])
1418: return(roff_getnum(v, pos, res));
1419:
1420: (*pos)++;
1421: if ( ! roff_evalnum(v, pos, res, 1))
1422: return(0);
1423:
1.206 schwarze 1424: /*
1425: * Omission of the closing parenthesis
1426: * is an error in validation mode,
1427: * but ignored in evaluation mode.
1428: */
1429:
1.204 schwarze 1430: if (')' == v[*pos])
1431: (*pos)++;
1.206 schwarze 1432: else if (NULL == res)
1433: return(0);
1.204 schwarze 1434:
1435: return(1);
1436: }
1437:
1438: /*
1439: * Evaluate a complete numeric expression.
1440: * Proceed left to right, there is no concept of precedence.
1441: */
1442: static int
1443: roff_evalnum(const char *v, int *pos, int *res, int skipwhite)
1444: {
1445: int mypos, operand2;
1446: char operator;
1447:
1448: if (NULL == pos) {
1449: mypos = 0;
1450: pos = &mypos;
1451: }
1452:
1453: if (skipwhite)
1454: while (isspace((unsigned char)v[*pos]))
1455: (*pos)++;
1456:
1457: if ( ! roff_evalpar(v, pos, res))
1458: return(0);
1459:
1460: while (1) {
1461: if (skipwhite)
1462: while (isspace((unsigned char)v[*pos]))
1463: (*pos)++;
1464:
1465: if ( ! roff_getop(v, pos, &operator))
1466: break;
1467:
1468: if (skipwhite)
1469: while (isspace((unsigned char)v[*pos]))
1470: (*pos)++;
1471:
1472: if ( ! roff_evalpar(v, pos, &operand2))
1473: return(0);
1474:
1475: if (skipwhite)
1476: while (isspace((unsigned char)v[*pos]))
1477: (*pos)++;
1.206 schwarze 1478:
1479: if (NULL == res)
1480: continue;
1.204 schwarze 1481:
1482: switch (operator) {
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 *= operand2;
1491: break;
1.207 schwarze 1492: case '/':
1.204 schwarze 1493: *res /= operand2;
1494: break;
1.207 schwarze 1495: case '%':
1.204 schwarze 1496: *res %= operand2;
1497: break;
1.207 schwarze 1498: case '<':
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 'l':
1.204 schwarze 1505: *res = *res <= operand2;
1506: break;
1.207 schwarze 1507: case 'g':
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 '!':
1.204 schwarze 1514: *res = *res != operand2;
1515: break;
1.207 schwarze 1516: case '&':
1.204 schwarze 1517: *res = *res && operand2;
1518: break;
1.207 schwarze 1519: case ':':
1.204 schwarze 1520: *res = *res || operand2;
1521: break;
1.207 schwarze 1522: case 'i':
1.204 schwarze 1523: if (operand2 < *res)
1524: *res = operand2;
1525: break;
1.207 schwarze 1526: case 'a':
1.204 schwarze 1527: if (operand2 > *res)
1528: *res = operand2;
1529: break;
1530: default:
1531: abort();
1532: }
1533: }
1534: return(1);
1535: }
1536:
1.180 schwarze 1537: void
1.187 schwarze 1538: roff_setreg(struct roff *r, const char *name, int val, char sign)
1.147 kristaps 1539: {
1.180 schwarze 1540: struct roffreg *reg;
1541:
1542: /* Search for an existing register with the same name. */
1543: reg = r->regtab;
1544:
1545: while (reg && strcmp(name, reg->key.p))
1546: reg = reg->next;
1.147 kristaps 1547:
1.180 schwarze 1548: if (NULL == reg) {
1549: /* Create a new register. */
1550: reg = mandoc_malloc(sizeof(struct roffreg));
1551: reg->key.p = mandoc_strdup(name);
1552: reg->key.sz = strlen(name);
1.187 schwarze 1553: reg->val = 0;
1.180 schwarze 1554: reg->next = r->regtab;
1555: r->regtab = reg;
1556: }
1557:
1.187 schwarze 1558: if ('+' == sign)
1559: reg->val += val;
1560: else if ('-' == sign)
1561: reg->val -= val;
1562: else
1563: reg->val = val;
1.147 kristaps 1564: }
1565:
1.192 schwarze 1566: /*
1567: * Handle some predefined read-only number registers.
1568: * For now, return -1 if the requested register is not predefined;
1569: * in case a predefined read-only register having the value -1
1570: * were to turn up, another special value would have to be chosen.
1571: */
1572: static int
1573: roff_getregro(const char *name)
1574: {
1575:
1576: switch (*name) {
1.207 schwarze 1577: case 'A': /* ASCII approximation mode is always off. */
1.192 schwarze 1578: return(0);
1.207 schwarze 1579: case 'g': /* Groff compatibility mode is always on. */
1.192 schwarze 1580: return(1);
1.207 schwarze 1581: case 'H': /* Fixed horizontal resolution. */
1.192 schwarze 1582: return (24);
1.207 schwarze 1583: case 'j': /* Always adjust left margin only. */
1.192 schwarze 1584: return(0);
1.207 schwarze 1585: case 'T': /* Some output device is always defined. */
1.192 schwarze 1586: return(1);
1.207 schwarze 1587: case 'V': /* Fixed vertical resolution. */
1.192 schwarze 1588: return (40);
1589: default:
1590: return (-1);
1591: }
1592: }
1593:
1.181 schwarze 1594: int
1.180 schwarze 1595: roff_getreg(const struct roff *r, const char *name)
1.147 kristaps 1596: {
1.180 schwarze 1597: struct roffreg *reg;
1.192 schwarze 1598: int val;
1599:
1600: if ('.' == name[0] && '\0' != name[1] && '\0' == name[2]) {
1601: val = roff_getregro(name + 1);
1602: if (-1 != val)
1603: return (val);
1604: }
1.180 schwarze 1605:
1606: for (reg = r->regtab; reg; reg = reg->next)
1607: if (0 == strcmp(name, reg->key.p))
1.181 schwarze 1608: return(reg->val);
1609:
1610: return(0);
1611: }
1612:
1613: static int
1614: roff_getregn(const struct roff *r, const char *name, size_t len)
1615: {
1616: struct roffreg *reg;
1.192 schwarze 1617: int val;
1618:
1619: if ('.' == name[0] && 2 == len) {
1620: val = roff_getregro(name + 1);
1621: if (-1 != val)
1622: return (val);
1623: }
1.181 schwarze 1624:
1625: for (reg = r->regtab; reg; reg = reg->next)
1626: if (len == reg->key.sz &&
1627: 0 == strncmp(name, reg->key.p, len))
1628: return(reg->val);
1.147 kristaps 1629:
1.180 schwarze 1630: return(0);
1.147 kristaps 1631: }
1632:
1.180 schwarze 1633: static void
1634: roff_freereg(struct roffreg *reg)
1.147 kristaps 1635: {
1.180 schwarze 1636: struct roffreg *old_reg;
1.147 kristaps 1637:
1.180 schwarze 1638: while (NULL != reg) {
1639: free(reg->key.p);
1640: old_reg = reg;
1641: reg = reg->next;
1642: free(old_reg);
1643: }
1.147 kristaps 1644: }
1.92 schwarze 1645:
1646: static enum rofferr
1.89 kristaps 1647: roff_nr(ROFF_ARGS)
1.83 schwarze 1648: {
1.212 schwarze 1649: char *key, *val;
1650: size_t keysz;
1.138 kristaps 1651: int iv;
1.187 schwarze 1652: char sign;
1.89 kristaps 1653:
1.212 schwarze 1654: key = val = *bufp + pos;
1655: if ('\0' == *key)
1656: return(ROFF_IGN);
1657:
1658: keysz = roff_getname(r, &val, ln, pos);
1659: if ('\\' == key[keysz])
1660: return(ROFF_IGN);
1661: key[keysz] = '\0';
1.89 kristaps 1662:
1.187 schwarze 1663: sign = *val;
1664: if ('+' == sign || '-' == sign)
1665: val++;
1666:
1.204 schwarze 1667: if (roff_evalnum(val, NULL, &iv, 0))
1668: roff_setreg(r, key, iv, sign);
1.109 kristaps 1669:
1.203 schwarze 1670: return(ROFF_IGN);
1671: }
1672:
1673: static enum rofferr
1674: roff_rr(ROFF_ARGS)
1675: {
1676: struct roffreg *reg, **prev;
1.212 schwarze 1677: char *name, *cp;
1678: size_t namesz;
1.203 schwarze 1679:
1.212 schwarze 1680: name = cp = *bufp + pos;
1681: if ('\0' == *name)
1682: return(ROFF_IGN);
1683: namesz = roff_getname(r, &cp, ln, pos);
1684: name[namesz] = '\0';
1.203 schwarze 1685:
1686: prev = &r->regtab;
1687: while (1) {
1688: reg = *prev;
1689: if (NULL == reg || !strcmp(name, reg->key.p))
1690: break;
1691: prev = ®->next;
1692: }
1693: if (NULL != reg) {
1694: *prev = reg->next;
1695: free(reg->key.p);
1696: free(reg);
1697: }
1.122 schwarze 1698: return(ROFF_IGN);
1699: }
1700:
1701: static enum rofferr
1702: roff_rm(ROFF_ARGS)
1703: {
1704: const char *name;
1705: char *cp;
1.212 schwarze 1706: size_t namesz;
1.122 schwarze 1707:
1708: cp = *bufp + pos;
1709: while ('\0' != *cp) {
1.212 schwarze 1710: name = cp;
1711: namesz = roff_getname(r, &cp, ln, (int)(cp - *bufp));
1712: roff_setstrn(&r->strtab, name, namesz, NULL, 0, 0);
1713: if ('\\' == name[namesz])
1714: break;
1.122 schwarze 1715: }
1.178 schwarze 1716: return(ROFF_IGN);
1717: }
1718:
1719: static enum rofferr
1720: roff_it(ROFF_ARGS)
1721: {
1722: char *cp;
1723: size_t len;
1724: int iv;
1725:
1726: /* Parse the number of lines. */
1727: cp = *bufp + pos;
1728: len = strcspn(cp, " \t");
1729: cp[len] = '\0';
1730: if ((iv = mandoc_strntoi(cp, len, 10)) <= 0) {
1731: mandoc_msg(MANDOCERR_NUMERIC, r->parse,
1.207 schwarze 1732: ln, ppos, *bufp + 1);
1.178 schwarze 1733: return(ROFF_IGN);
1734: }
1735: cp += len + 1;
1736:
1737: /* Arm the input line trap. */
1738: roffit_lines = iv;
1739: roffit_macro = mandoc_strdup(cp);
1.109 kristaps 1740: return(ROFF_IGN);
1.175 schwarze 1741: }
1742:
1743: static enum rofferr
1744: roff_Dd(ROFF_ARGS)
1745: {
1746: const char *const *cp;
1747:
1.199 schwarze 1748: if (0 == ((MPARSE_MDOC | MPARSE_QUICK) & r->options))
1.175 schwarze 1749: for (cp = __mdoc_reserved; *cp; cp++)
1750: roff_setstr(r, *cp, NULL, 0);
1751:
1752: return(ROFF_CONT);
1753: }
1754:
1755: static enum rofferr
1756: roff_TH(ROFF_ARGS)
1757: {
1758: const char *const *cp;
1759:
1.199 schwarze 1760: if (0 == (MPARSE_QUICK & r->options))
1.175 schwarze 1761: for (cp = __man_reserved; *cp; cp++)
1762: roff_setstr(r, *cp, NULL, 0);
1763:
1764: return(ROFF_CONT);
1.109 kristaps 1765: }
1766:
1767: static enum rofferr
1768: roff_TE(ROFF_ARGS)
1769: {
1770:
1771: if (NULL == r->tbl)
1.128 kristaps 1772: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.115 kristaps 1773: else
1.151 kristaps 1774: tbl_end(&r->tbl);
1.109 kristaps 1775:
1.112 kristaps 1776: return(ROFF_IGN);
1777: }
1778:
1779: static enum rofferr
1780: roff_T_(ROFF_ARGS)
1781: {
1782:
1783: if (NULL == r->tbl)
1.128 kristaps 1784: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.112 kristaps 1785: else
1.116 kristaps 1786: tbl_restart(ppos, ln, r->tbl);
1.112 kristaps 1787:
1.109 kristaps 1788: return(ROFF_IGN);
1789: }
1790:
1.156 kristaps 1791: #if 0
1792: static int
1.151 kristaps 1793: roff_closeeqn(struct roff *r)
1794: {
1795:
1796: return(r->eqn && ROFF_EQN == eqn_end(&r->eqn) ? 1 : 0);
1797: }
1.156 kristaps 1798: #endif
1.151 kristaps 1799:
1.156 kristaps 1800: static void
1.207 schwarze 1801: roff_openeqn(struct roff *r, const char *name, int line,
1.151 kristaps 1802: int offs, const char *buf)
1.125 kristaps 1803: {
1.151 kristaps 1804: struct eqn_node *e;
1805: int poff;
1.125 kristaps 1806:
1807: assert(NULL == r->eqn);
1.151 kristaps 1808: e = eqn_alloc(name, offs, line, r->parse);
1.125 kristaps 1809:
1810: if (r->last_eqn)
1811: r->last_eqn->next = e;
1812: else
1813: r->first_eqn = r->last_eqn = e;
1814:
1815: r->eqn = r->last_eqn = e;
1.151 kristaps 1816:
1817: if (buf) {
1818: poff = 0;
1819: eqn_read(&r->eqn, line, buf, offs, &poff);
1820: }
1821: }
1822:
1823: static enum rofferr
1824: roff_EQ(ROFF_ARGS)
1825: {
1826:
1827: roff_openeqn(r, *bufp + pos, ln, ppos, NULL);
1.125 kristaps 1828: return(ROFF_IGN);
1829: }
1830:
1831: static enum rofferr
1832: roff_EN(ROFF_ARGS)
1833: {
1834:
1.128 kristaps 1835: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.125 kristaps 1836: return(ROFF_IGN);
1837: }
1838:
1839: static enum rofferr
1.109 kristaps 1840: roff_TS(ROFF_ARGS)
1841: {
1.176 schwarze 1842: struct tbl_node *tbl;
1.89 kristaps 1843:
1.115 kristaps 1844: if (r->tbl) {
1.128 kristaps 1845: mandoc_msg(MANDOCERR_SCOPEBROKEN, r->parse, ln, ppos, NULL);
1.151 kristaps 1846: tbl_end(&r->tbl);
1.115 kristaps 1847: }
1.83 schwarze 1848:
1.176 schwarze 1849: tbl = tbl_alloc(ppos, ln, r->parse);
1.113 kristaps 1850:
1851: if (r->last_tbl)
1.176 schwarze 1852: r->last_tbl->next = tbl;
1.113 kristaps 1853: else
1.176 schwarze 1854: r->first_tbl = r->last_tbl = tbl;
1.113 kristaps 1855:
1.176 schwarze 1856: r->tbl = r->last_tbl = tbl;
1.83 schwarze 1857: return(ROFF_IGN);
1.92 schwarze 1858: }
1859:
1.105 kristaps 1860: static enum rofferr
1.174 kristaps 1861: roff_cc(ROFF_ARGS)
1862: {
1863: const char *p;
1864:
1865: p = *bufp + pos;
1866:
1867: if ('\0' == *p || '.' == (r->control = *p++))
1868: r->control = 0;
1869:
1870: if ('\0' != *p)
1871: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
1872:
1873: return(ROFF_IGN);
1874: }
1875:
1876: static enum rofferr
1.164 kristaps 1877: roff_tr(ROFF_ARGS)
1878: {
1879: const char *p, *first, *second;
1880: size_t fsz, ssz;
1881: enum mandoc_esc esc;
1882:
1883: p = *bufp + pos;
1884:
1885: if ('\0' == *p) {
1886: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
1887: return(ROFF_IGN);
1888: }
1889:
1890: while ('\0' != *p) {
1891: fsz = ssz = 1;
1892:
1893: first = p++;
1894: if ('\\' == *first) {
1895: esc = mandoc_escape(&p, NULL, NULL);
1896: if (ESCAPE_ERROR == esc) {
1.207 schwarze 1897: mandoc_msg(MANDOCERR_BADESCAPE,
1898: r->parse, ln,
1899: (int)(p - *bufp), NULL);
1.164 kristaps 1900: return(ROFF_IGN);
1901: }
1902: fsz = (size_t)(p - first);
1903: }
1904:
1905: second = p++;
1906: if ('\\' == *second) {
1907: esc = mandoc_escape(&p, NULL, NULL);
1908: if (ESCAPE_ERROR == esc) {
1.207 schwarze 1909: mandoc_msg(MANDOCERR_BADESCAPE,
1910: r->parse, ln,
1911: (int)(p - *bufp), NULL);
1.164 kristaps 1912: return(ROFF_IGN);
1913: }
1914: ssz = (size_t)(p - second);
1.165 kristaps 1915: } else if ('\0' == *second) {
1.207 schwarze 1916: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse,
1917: ln, (int)(p - *bufp), NULL);
1.164 kristaps 1918: second = " ";
1.165 kristaps 1919: p--;
1.164 kristaps 1920: }
1921:
1.167 kristaps 1922: if (fsz > 1) {
1.207 schwarze 1923: roff_setstrn(&r->xmbtab, first, fsz,
1924: second, ssz, 0);
1.167 kristaps 1925: continue;
1926: }
1927:
1928: if (NULL == r->xtab)
1.207 schwarze 1929: r->xtab = mandoc_calloc(128,
1930: sizeof(struct roffstr));
1.167 kristaps 1931:
1932: free(r->xtab[(int)*first].p);
1933: r->xtab[(int)*first].p = mandoc_strndup(second, ssz);
1934: r->xtab[(int)*first].sz = ssz;
1.164 kristaps 1935: }
1936:
1937: return(ROFF_IGN);
1938: }
1939:
1940: static enum rofferr
1.105 kristaps 1941: roff_so(ROFF_ARGS)
1942: {
1943: char *name;
1944:
1.210 schwarze 1945: name = *bufp + pos;
1946: mandoc_vmsg(MANDOCERR_SO, r->parse, ln, ppos, ".so %s", name);
1.105 kristaps 1947:
1948: /*
1949: * Handle `so'. Be EXTREMELY careful, as we shouldn't be
1950: * opening anything that's not in our cwd or anything beneath
1951: * it. Thus, explicitly disallow traversing up the file-system
1952: * or using absolute paths.
1953: */
1954:
1955: if ('/' == *name || strstr(name, "../") || strstr(name, "/..")) {
1.210 schwarze 1956: mandoc_vmsg(MANDOCERR_SO_PATH, r->parse, ln, ppos,
1957: ".so %s", name);
1.105 kristaps 1958: return(ROFF_ERR);
1959: }
1960:
1961: *offs = pos;
1962: return(ROFF_SO);
1963: }
1.92 schwarze 1964:
1.106 kristaps 1965: static enum rofferr
1966: roff_userdef(ROFF_ARGS)
1.99 kristaps 1967: {
1.106 kristaps 1968: const char *arg[9];
1969: char *cp, *n1, *n2;
1.119 schwarze 1970: int i;
1.106 kristaps 1971:
1972: /*
1973: * Collect pointers to macro argument strings
1.188 schwarze 1974: * and NUL-terminate them.
1.106 kristaps 1975: */
1976: cp = *bufp + pos;
1.119 schwarze 1977: for (i = 0; i < 9; i++)
1.120 schwarze 1978: arg[i] = '\0' == *cp ? "" :
1.136 kristaps 1979: mandoc_getarg(r->parse, &cp, ln, &pos);
1.99 kristaps 1980:
1.106 kristaps 1981: /*
1982: * Expand macro arguments.
1.99 kristaps 1983: */
1.106 kristaps 1984: *szp = 0;
1985: n1 = cp = mandoc_strdup(r->current_string);
1986: while (NULL != (cp = strstr(cp, "\\$"))) {
1987: i = cp[2] - '1';
1988: if (0 > i || 8 < i) {
1989: /* Not an argument invocation. */
1990: cp += 2;
1991: continue;
1992: }
1.209 schwarze 1993: *cp = '\0';
1994: *szp = mandoc_asprintf(&n2, "%s%s%s",
1995: n1, arg[i], cp + 3) + 1;
1.106 kristaps 1996: cp = n2 + (cp - n1);
1997: free(n1);
1998: n1 = n2;
1.99 kristaps 1999: }
2000:
1.106 kristaps 2001: /*
2002: * Replace the macro invocation
2003: * by the expanded macro.
2004: */
2005: free(*bufp);
2006: *bufp = n1;
2007: if (0 == *szp)
2008: *szp = strlen(*bufp) + 1;
2009:
2010: return(*szp > 1 && '\n' == (*bufp)[(int)*szp - 2] ?
2011: ROFF_REPARSE : ROFF_APPEND);
1.99 kristaps 2012: }
1.121 schwarze 2013:
1.212 schwarze 2014: static size_t
1.121 schwarze 2015: roff_getname(struct roff *r, char **cpp, int ln, int pos)
2016: {
2017: char *name, *cp;
1.212 schwarze 2018: size_t namesz;
1.121 schwarze 2019:
2020: name = *cpp;
2021: if ('\0' == *name)
1.212 schwarze 2022: return(0);
1.121 schwarze 2023:
1.212 schwarze 2024: /* Read until end of name and terminate it with NUL. */
2025: for (cp = name; 1; cp++) {
2026: if ('\0' == *cp || ' ' == *cp) {
2027: namesz = cp - name;
2028: break;
2029: }
1.121 schwarze 2030: if ('\\' != *cp)
2031: continue;
1.215 schwarze 2032: namesz = cp - name;
2033: if ('{' == cp[1] || '}' == cp[1])
2034: break;
1.121 schwarze 2035: cp++;
2036: if ('\\' == *cp)
2037: continue;
1.128 kristaps 2038: mandoc_msg(MANDOCERR_NAMESC, r->parse, ln, pos, NULL);
1.212 schwarze 2039: mandoc_escape((const char **)&cp, NULL, NULL);
2040: break;
1.121 schwarze 2041: }
2042:
2043: /* Read past spaces. */
2044: while (' ' == *cp)
2045: cp++;
2046:
2047: *cpp = cp;
1.212 schwarze 2048: return(namesz);
1.121 schwarze 2049: }
2050:
1.106 kristaps 2051: /*
2052: * Store *string into the user-defined string called *name.
2053: * To clear an existing entry, call with (*r, *name, NULL, 0).
1.193 schwarze 2054: * append == 0: replace mode
2055: * append == 1: single-line append mode
2056: * append == 2: multiline append mode, append '\n' after each call
1.106 kristaps 2057: */
1.94 kristaps 2058: static void
1.106 kristaps 2059: roff_setstr(struct roff *r, const char *name, const char *string,
1.193 schwarze 2060: int append)
1.92 schwarze 2061: {
1.164 kristaps 2062:
2063: roff_setstrn(&r->strtab, name, strlen(name), string,
1.207 schwarze 2064: string ? strlen(string) : 0, append);
1.164 kristaps 2065: }
2066:
2067: static void
1.166 kristaps 2068: roff_setstrn(struct roffkv **r, const char *name, size_t namesz,
1.193 schwarze 2069: const char *string, size_t stringsz, int append)
1.164 kristaps 2070: {
1.166 kristaps 2071: struct roffkv *n;
1.164 kristaps 2072: char *c;
2073: int i;
2074: size_t oldch, newch;
1.92 schwarze 2075:
1.106 kristaps 2076: /* Search for an existing string with the same name. */
1.164 kristaps 2077: n = *r;
2078:
1.211 schwarze 2079: while (n && (namesz != n->key.sz ||
2080: strncmp(n->key.p, name, namesz)))
1.92 schwarze 2081: n = n->next;
1.94 kristaps 2082:
2083: if (NULL == n) {
1.106 kristaps 2084: /* Create a new string table entry. */
1.166 kristaps 2085: n = mandoc_malloc(sizeof(struct roffkv));
2086: n->key.p = mandoc_strndup(name, namesz);
2087: n->key.sz = namesz;
2088: n->val.p = NULL;
2089: n->val.sz = 0;
1.164 kristaps 2090: n->next = *r;
2091: *r = n;
1.193 schwarze 2092: } else if (0 == append) {
1.166 kristaps 2093: free(n->val.p);
2094: n->val.p = NULL;
2095: n->val.sz = 0;
1.106 kristaps 2096: }
2097:
2098: if (NULL == string)
2099: return;
2100:
2101: /*
2102: * One additional byte for the '\n' in multiline mode,
2103: * and one for the terminating '\0'.
2104: */
1.193 schwarze 2105: newch = stringsz + (1 < append ? 2u : 1u);
1.164 kristaps 2106:
1.166 kristaps 2107: if (NULL == n->val.p) {
2108: n->val.p = mandoc_malloc(newch);
2109: *n->val.p = '\0';
1.106 kristaps 2110: oldch = 0;
2111: } else {
1.166 kristaps 2112: oldch = n->val.sz;
2113: n->val.p = mandoc_realloc(n->val.p, oldch + newch);
1.106 kristaps 2114: }
2115:
2116: /* Skip existing content in the destination buffer. */
1.166 kristaps 2117: c = n->val.p + (int)oldch;
1.106 kristaps 2118:
2119: /* Append new content to the destination buffer. */
1.164 kristaps 2120: i = 0;
2121: while (i < (int)stringsz) {
1.106 kristaps 2122: /*
2123: * Rudimentary roff copy mode:
2124: * Handle escaped backslashes.
2125: */
1.164 kristaps 2126: if ('\\' == string[i] && '\\' == string[i + 1])
2127: i++;
2128: *c++ = string[i++];
1.106 kristaps 2129: }
1.94 kristaps 2130:
1.106 kristaps 2131: /* Append terminating bytes. */
1.193 schwarze 2132: if (1 < append)
1.106 kristaps 2133: *c++ = '\n';
1.163 kristaps 2134:
1.106 kristaps 2135: *c = '\0';
1.166 kristaps 2136: n->val.sz = (int)(c - n->val.p);
1.92 schwarze 2137: }
2138:
1.94 kristaps 2139: static const char *
2140: roff_getstrn(const struct roff *r, const char *name, size_t len)
1.92 schwarze 2141: {
1.166 kristaps 2142: const struct roffkv *n;
1.191 schwarze 2143: int i;
1.92 schwarze 2144:
1.164 kristaps 2145: for (n = r->strtab; n; n = n->next)
1.207 schwarze 2146: if (0 == strncmp(name, n->key.p, len) &&
2147: '\0' == n->key.p[(int)len])
1.166 kristaps 2148: return(n->val.p);
1.191 schwarze 2149:
2150: for (i = 0; i < PREDEFS_MAX; i++)
2151: if (0 == strncmp(name, predefs[i].name, len) &&
2152: '\0' == predefs[i].name[(int)len])
2153: return(predefs[i].str);
1.94 kristaps 2154:
1.157 kristaps 2155: return(NULL);
1.92 schwarze 2156: }
2157:
1.94 kristaps 2158: static void
1.167 kristaps 2159: roff_freestr(struct roffkv *r)
1.92 schwarze 2160: {
1.166 kristaps 2161: struct roffkv *n, *nn;
1.92 schwarze 2162:
1.167 kristaps 2163: for (n = r; n; n = nn) {
1.166 kristaps 2164: free(n->key.p);
2165: free(n->val.p);
1.92 schwarze 2166: nn = n->next;
2167: free(n);
2168: }
1.114 kristaps 2169: }
2170:
2171: const struct tbl_span *
2172: roff_span(const struct roff *r)
2173: {
1.207 schwarze 2174:
1.114 kristaps 2175: return(r->tbl ? tbl_span(r->tbl) : NULL);
1.125 kristaps 2176: }
2177:
2178: const struct eqn *
2179: roff_eqn(const struct roff *r)
2180: {
1.207 schwarze 2181:
1.125 kristaps 2182: return(r->last_eqn ? &r->last_eqn->eqn : NULL);
1.164 kristaps 2183: }
2184:
2185: /*
2186: * Duplicate an input string, making the appropriate character
2187: * conversations (as stipulated by `tr') along the way.
2188: * Returns a heap-allocated string with all the replacements made.
2189: */
2190: char *
2191: roff_strdup(const struct roff *r, const char *p)
2192: {
1.166 kristaps 2193: const struct roffkv *cp;
1.164 kristaps 2194: char *res;
2195: const char *pp;
2196: size_t ssz, sz;
2197: enum mandoc_esc esc;
2198:
1.167 kristaps 2199: if (NULL == r->xmbtab && NULL == r->xtab)
1.164 kristaps 2200: return(mandoc_strdup(p));
2201: else if ('\0' == *p)
2202: return(mandoc_strdup(""));
2203:
2204: /*
2205: * Step through each character looking for term matches
2206: * (remember that a `tr' can be invoked with an escape, which is
2207: * a glyph but the escape is multi-character).
2208: * We only do this if the character hash has been initialised
2209: * and the string is >0 length.
2210: */
2211:
2212: res = NULL;
2213: ssz = 0;
2214:
2215: while ('\0' != *p) {
1.167 kristaps 2216: if ('\\' != *p && r->xtab && r->xtab[(int)*p].p) {
2217: sz = r->xtab[(int)*p].sz;
2218: res = mandoc_realloc(res, ssz + sz + 1);
2219: memcpy(res + ssz, r->xtab[(int)*p].p, sz);
2220: ssz += sz;
2221: p++;
2222: continue;
2223: } else if ('\\' != *p) {
2224: res = mandoc_realloc(res, ssz + 2);
2225: res[ssz++] = *p++;
2226: continue;
2227: }
2228:
1.164 kristaps 2229: /* Search for term matches. */
1.167 kristaps 2230: for (cp = r->xmbtab; cp; cp = cp->next)
1.166 kristaps 2231: if (0 == strncmp(p, cp->key.p, cp->key.sz))
1.164 kristaps 2232: break;
2233:
2234: if (NULL != cp) {
2235: /*
2236: * A match has been found.
2237: * Append the match to the array and move
2238: * forward by its keysize.
2239: */
1.207 schwarze 2240: res = mandoc_realloc(res,
2241: ssz + cp->val.sz + 1);
1.166 kristaps 2242: memcpy(res + ssz, cp->val.p, cp->val.sz);
2243: ssz += cp->val.sz;
2244: p += (int)cp->key.sz;
1.164 kristaps 2245: continue;
2246: }
2247:
1.167 kristaps 2248: /*
2249: * Handle escapes carefully: we need to copy
2250: * over just the escape itself, or else we might
2251: * do replacements within the escape itself.
2252: * Make sure to pass along the bogus string.
2253: */
2254: pp = p++;
2255: esc = mandoc_escape(&p, NULL, NULL);
2256: if (ESCAPE_ERROR == esc) {
2257: sz = strlen(pp);
1.164 kristaps 2258: res = mandoc_realloc(res, ssz + sz + 1);
2259: memcpy(res + ssz, pp, sz);
1.167 kristaps 2260: break;
1.164 kristaps 2261: }
1.207 schwarze 2262: /*
2263: * We bail out on bad escapes.
1.167 kristaps 2264: * No need to warn: we already did so when
2265: * roff_res() was called.
2266: */
2267: sz = (int)(p - pp);
2268: res = mandoc_realloc(res, ssz + sz + 1);
2269: memcpy(res + ssz, pp, sz);
2270: ssz += sz;
1.164 kristaps 2271: }
2272:
2273: res[(int)ssz] = '\0';
2274: return(res);
1.174 kristaps 2275: }
2276:
2277: /*
1.207 schwarze 2278: * Find out whether a line is a macro line or not.
1.174 kristaps 2279: * If it is, adjust the current position and return one; if it isn't,
2280: * return zero and don't change the current position.
2281: * If the control character has been set with `.cc', then let that grain
2282: * precedence.
2283: * This is slighly contrary to groff, where using the non-breaking
2284: * control character when `cc' has been invoked will cause the
2285: * non-breaking macro contents to be printed verbatim.
2286: */
2287: int
2288: roff_getcontrol(const struct roff *r, const char *cp, int *ppos)
2289: {
2290: int pos;
2291:
2292: pos = *ppos;
2293:
2294: if (0 != r->control && cp[pos] == r->control)
2295: pos++;
2296: else if (0 != r->control)
2297: return(0);
2298: else if ('\\' == cp[pos] && '.' == cp[pos + 1])
2299: pos += 2;
2300: else if ('.' == cp[pos] || '\'' == cp[pos])
2301: pos++;
2302: else
2303: return(0);
2304:
2305: while (' ' == cp[pos] || '\t' == cp[pos])
2306: pos++;
2307:
2308: *ppos = pos;
2309: return(1);
1.74 kristaps 2310: }
CVSweb