Annotation of mandoc/roff.c, Revision 1.180
1.180 ! schwarze 1: /* $Id: roff.c,v 1.179 2013/10/05 20:30:05 schwarze Exp $ */
1.1 kristaps 2: /*
1.175 schwarze 3: * Copyright (c) 2010, 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
1.177 schwarze 4: * Copyright (c) 2010, 2011, 2012, 2013 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"
1.109 kristaps 29: #include "libroff.h"
1.94 kristaps 30: #include "libmandoc.h"
1.33 kristaps 31:
1.141 kristaps 32: /* Maximum number of nested if-else conditionals. */
1.82 kristaps 33: #define RSTACK_MAX 128
34:
1.170 schwarze 35: /* Maximum number of string expansions per line, to break infinite loops. */
36: #define EXPAND_LIMIT 1000
37:
1.67 kristaps 38: enum rofft {
1.103 kristaps 39: ROFF_ad,
1.80 kristaps 40: ROFF_am,
41: ROFF_ami,
42: ROFF_am1,
1.174 kristaps 43: ROFF_cc,
1.80 kristaps 44: ROFF_de,
45: ROFF_dei,
46: ROFF_de1,
1.83 schwarze 47: ROFF_ds,
1.82 kristaps 48: ROFF_el,
1.103 kristaps 49: ROFF_hy,
1.82 kristaps 50: ROFF_ie,
1.75 kristaps 51: ROFF_if,
1.76 kristaps 52: ROFF_ig,
1.123 schwarze 53: ROFF_it,
1.103 kristaps 54: ROFF_ne,
55: ROFF_nh,
1.104 kristaps 56: ROFF_nr,
1.124 schwarze 57: ROFF_ns,
58: ROFF_ps,
1.83 schwarze 59: ROFF_rm,
1.105 kristaps 60: ROFF_so,
1.124 schwarze 61: ROFF_ta,
1.83 schwarze 62: ROFF_tr,
1.175 schwarze 63: ROFF_Dd,
64: ROFF_TH,
1.109 kristaps 65: ROFF_TS,
66: ROFF_TE,
1.112 kristaps 67: ROFF_T_,
1.125 kristaps 68: ROFF_EQ,
69: ROFF_EN,
1.76 kristaps 70: ROFF_cblock,
1.141 kristaps 71: ROFF_ccond,
1.106 kristaps 72: ROFF_USERDEF,
1.67 kristaps 73: ROFF_MAX
74: };
75:
1.82 kristaps 76: enum roffrule {
77: ROFFRULE_ALLOW,
78: ROFFRULE_DENY
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: * Registers are assumed to be unsigned ints for now.
! 101: */
! 102: struct roffreg {
! 103: struct roffstr key;
! 104: unsigned int u;
! 105: struct roffreg *next;
! 106: };
! 107:
1.67 kristaps 108: struct roff {
1.175 schwarze 109: enum mparset parsetype; /* requested parse type */
1.128 kristaps 110: struct mparse *parse; /* parse point */
1.67 kristaps 111: struct roffnode *last; /* leaf of stack */
1.82 kristaps 112: enum roffrule rstack[RSTACK_MAX]; /* stack of !`ie' rules */
1.174 kristaps 113: char control; /* control character */
1.82 kristaps 114: int rstackpos; /* position in rstack */
1.180 ! schwarze 115: struct roffreg *regtab; /* number registers */
1.166 kristaps 116: struct roffkv *strtab; /* user-defined strings & macros */
1.167 kristaps 117: struct roffkv *xmbtab; /* multi-byte trans table (`tr') */
118: struct roffstr *xtab; /* single-byte trans table (`tr') */
1.106 kristaps 119: const char *current_string; /* value of last called user macro */
1.118 kristaps 120: struct tbl_node *first_tbl; /* first table parsed */
121: struct tbl_node *last_tbl; /* last table parsed */
122: struct tbl_node *tbl; /* current table being parsed */
1.125 kristaps 123: struct eqn_node *last_eqn; /* last equation parsed */
124: struct eqn_node *first_eqn; /* first equation parsed */
125: struct eqn_node *eqn; /* current equation being parsed */
1.79 kristaps 126: };
127:
1.67 kristaps 128: struct roffnode {
129: enum rofft tok; /* type of node */
130: struct roffnode *parent; /* up one in stack */
131: int line; /* parse line */
132: int col; /* parse col */
1.106 kristaps 133: char *name; /* node name, e.g. macro name */
1.79 kristaps 134: char *end; /* end-rules: custom token */
135: int endspan; /* end-rules: next-line or infty */
1.82 kristaps 136: enum roffrule rule; /* current evaluation rule */
1.67 kristaps 137: };
138:
139: #define ROFF_ARGS struct roff *r, /* parse ctx */ \
1.72 kristaps 140: enum rofft tok, /* tok of macro */ \
1.67 kristaps 141: char **bufp, /* input buffer */ \
142: size_t *szp, /* size of input buffer */ \
143: int ln, /* parse line */ \
1.75 kristaps 144: int ppos, /* original pos in buffer */ \
145: int pos, /* current pos in buffer */ \
1.74 kristaps 146: int *offs /* reset offset of buffer data */
1.67 kristaps 147:
148: typedef enum rofferr (*roffproc)(ROFF_ARGS);
149:
150: struct roffmac {
151: const char *name; /* macro name */
1.79 kristaps 152: roffproc proc; /* process new macro */
153: roffproc text; /* process as child text of macro */
154: roffproc sub; /* process as child of macro */
155: int flags;
156: #define ROFFMAC_STRUCT (1 << 0) /* always interpret */
1.85 kristaps 157: struct roffmac *next;
1.67 kristaps 158: };
159:
1.141 kristaps 160: struct predef {
161: const char *name; /* predefined input name */
162: const char *str; /* replacement symbol */
163: };
164:
165: #define PREDEF(__name, __str) \
166: { (__name), (__str) },
167:
1.155 kristaps 168: static enum rofft roffhash_find(const char *, size_t);
169: static void roffhash_init(void);
170: static void roffnode_cleanscope(struct roff *);
171: static void roffnode_pop(struct roff *);
172: static void roffnode_push(struct roff *, enum rofft,
173: const char *, int, int);
1.80 kristaps 174: static enum rofferr roff_block(ROFF_ARGS);
175: static enum rofferr roff_block_text(ROFF_ARGS);
176: static enum rofferr roff_block_sub(ROFF_ARGS);
177: static enum rofferr roff_cblock(ROFF_ARGS);
1.174 kristaps 178: static enum rofferr roff_cc(ROFF_ARGS);
1.80 kristaps 179: static enum rofferr roff_ccond(ROFF_ARGS);
1.82 kristaps 180: static enum rofferr roff_cond(ROFF_ARGS);
181: static enum rofferr roff_cond_text(ROFF_ARGS);
182: static enum rofferr roff_cond_sub(ROFF_ARGS);
1.92 schwarze 183: static enum rofferr roff_ds(ROFF_ARGS);
1.94 kristaps 184: static enum roffrule roff_evalcond(const char *, int *);
1.155 kristaps 185: static void roff_free1(struct roff *);
1.180 ! schwarze 186: static void roff_freereg(struct roffreg *);
1.167 kristaps 187: static void roff_freestr(struct roffkv *);
1.121 schwarze 188: static char *roff_getname(struct roff *, char **, int, int);
1.94 kristaps 189: static const char *roff_getstrn(const struct roff *,
190: const char *, size_t);
1.178 schwarze 191: static enum rofferr roff_it(ROFF_ARGS);
1.103 kristaps 192: static enum rofferr roff_line_ignore(ROFF_ARGS);
1.89 kristaps 193: static enum rofferr roff_nr(ROFF_ARGS);
1.169 schwarze 194: static void roff_openeqn(struct roff *, const char *,
1.156 kristaps 195: int, int, const char *);
1.155 kristaps 196: static enum rofft roff_parse(struct roff *, const char *, int *);
1.178 schwarze 197: static enum rofferr roff_parsetext(char **, size_t *, int, int *);
1.172 schwarze 198: static enum rofferr roff_res(struct roff *,
1.142 kristaps 199: char **, size_t *, int, int);
1.122 schwarze 200: static enum rofferr roff_rm(ROFF_ARGS);
1.94 kristaps 201: static void roff_setstr(struct roff *,
1.106 kristaps 202: const char *, const char *, int);
1.166 kristaps 203: static void roff_setstrn(struct roffkv **, const char *,
1.164 kristaps 204: size_t, const char *, size_t, int);
1.105 kristaps 205: static enum rofferr roff_so(ROFF_ARGS);
1.164 kristaps 206: static enum rofferr roff_tr(ROFF_ARGS);
1.175 schwarze 207: static enum rofferr roff_Dd(ROFF_ARGS);
208: static enum rofferr roff_TH(ROFF_ARGS);
1.109 kristaps 209: static enum rofferr roff_TE(ROFF_ARGS);
210: static enum rofferr roff_TS(ROFF_ARGS);
1.125 kristaps 211: static enum rofferr roff_EQ(ROFF_ARGS);
212: static enum rofferr roff_EN(ROFF_ARGS);
1.112 kristaps 213: static enum rofferr roff_T_(ROFF_ARGS);
1.106 kristaps 214: static enum rofferr roff_userdef(ROFF_ARGS);
1.67 kristaps 215:
1.155 kristaps 216: /* See roffhash_find() */
1.85 kristaps 217:
218: #define ASCII_HI 126
219: #define ASCII_LO 33
220: #define HASHWIDTH (ASCII_HI - ASCII_LO + 1)
221:
222: static struct roffmac *hash[HASHWIDTH];
223:
224: static struct roffmac roffs[ROFF_MAX] = {
1.103 kristaps 225: { "ad", roff_line_ignore, NULL, NULL, 0, NULL },
1.85 kristaps 226: { "am", roff_block, roff_block_text, roff_block_sub, 0, NULL },
227: { "ami", roff_block, roff_block_text, roff_block_sub, 0, NULL },
228: { "am1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.174 kristaps 229: { "cc", roff_cc, NULL, NULL, 0, NULL },
1.85 kristaps 230: { "de", roff_block, roff_block_text, roff_block_sub, 0, NULL },
231: { "dei", roff_block, roff_block_text, roff_block_sub, 0, NULL },
232: { "de1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.92 schwarze 233: { "ds", roff_ds, NULL, NULL, 0, NULL },
1.85 kristaps 234: { "el", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
1.103 kristaps 235: { "hy", roff_line_ignore, NULL, NULL, 0, NULL },
1.85 kristaps 236: { "ie", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
237: { "if", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
238: { "ig", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.178 schwarze 239: { "it", roff_it, NULL, NULL, 0, NULL },
1.103 kristaps 240: { "ne", roff_line_ignore, NULL, NULL, 0, NULL },
241: { "nh", roff_line_ignore, NULL, NULL, 0, NULL },
1.104 kristaps 242: { "nr", roff_nr, NULL, NULL, 0, NULL },
1.124 schwarze 243: { "ns", roff_line_ignore, NULL, NULL, 0, NULL },
244: { "ps", roff_line_ignore, NULL, NULL, 0, NULL },
1.122 schwarze 245: { "rm", roff_rm, NULL, NULL, 0, NULL },
1.105 kristaps 246: { "so", roff_so, NULL, NULL, 0, NULL },
1.124 schwarze 247: { "ta", roff_line_ignore, NULL, NULL, 0, NULL },
1.164 kristaps 248: { "tr", roff_tr, NULL, NULL, 0, NULL },
1.175 schwarze 249: { "Dd", roff_Dd, NULL, NULL, 0, NULL },
250: { "TH", roff_TH, NULL, NULL, 0, NULL },
1.109 kristaps 251: { "TS", roff_TS, NULL, NULL, 0, NULL },
252: { "TE", roff_TE, NULL, NULL, 0, NULL },
1.112 kristaps 253: { "T&", roff_T_, NULL, NULL, 0, NULL },
1.125 kristaps 254: { "EQ", roff_EQ, NULL, NULL, 0, NULL },
255: { "EN", roff_EN, NULL, NULL, 0, NULL },
1.85 kristaps 256: { ".", roff_cblock, NULL, NULL, 0, NULL },
257: { "\\}", roff_ccond, NULL, NULL, 0, NULL },
1.106 kristaps 258: { NULL, roff_userdef, NULL, NULL, 0, NULL },
1.67 kristaps 259: };
260:
1.175 schwarze 261: const char *const __mdoc_reserved[] = {
262: "Ac", "Ad", "An", "Ao", "Ap", "Aq", "Ar", "At",
263: "Bc", "Bd", "Bf", "Bk", "Bl", "Bo", "Bq",
264: "Brc", "Bro", "Brq", "Bsx", "Bt", "Bx",
265: "Cd", "Cm", "Db", "Dc", "Dd", "Dl", "Do", "Dq",
266: "Ds", "Dt", "Dv", "Dx", "D1",
267: "Ec", "Ed", "Ef", "Ek", "El", "Em", "em",
268: "En", "Eo", "Eq", "Er", "Es", "Ev", "Ex",
269: "Fa", "Fc", "Fd", "Fl", "Fn", "Fo", "Fr", "Ft", "Fx",
270: "Hf", "Ic", "In", "It", "Lb", "Li", "Lk", "Lp", "LP",
271: "Me", "Ms", "Mt", "Nd", "Nm", "No", "Ns", "Nx",
272: "Oc", "Oo", "Op", "Os", "Ot", "Ox",
273: "Pa", "Pc", "Pf", "Po", "Pp", "PP", "pp", "Pq",
274: "Qc", "Ql", "Qo", "Qq", "Or", "Rd", "Re", "Rs", "Rv",
275: "Sc", "Sf", "Sh", "SH", "Sm", "So", "Sq",
276: "Ss", "St", "Sx", "Sy",
277: "Ta", "Tn", "Ud", "Ux", "Va", "Vt", "Xc", "Xo", "Xr",
278: "%A", "%B", "%D", "%I", "%J", "%N", "%O",
279: "%P", "%Q", "%R", "%T", "%U", "%V",
280: NULL
281: };
282:
283: const char *const __man_reserved[] = {
284: "AT", "B", "BI", "BR", "BT", "DE", "DS", "DT",
285: "EE", "EN", "EQ", "EX", "HF", "HP", "I", "IB", "IP", "IR",
286: "LP", "ME", "MT", "OP", "P", "PD", "PP", "PT",
287: "R", "RB", "RE", "RI", "RS", "SB", "SH", "SM", "SS", "SY",
288: "TE", "TH", "TP", "TQ", "TS", "T&", "UC", "UE", "UR", "YS",
289: NULL
290: };
291:
1.141 kristaps 292: /* Array of injected predefined strings. */
293: #define PREDEFS_MAX 38
294: static const struct predef predefs[PREDEFS_MAX] = {
295: #include "predefs.in"
296: };
297:
1.155 kristaps 298: /* See roffhash_find() */
1.85 kristaps 299: #define ROFF_HASH(p) (p[0] - ASCII_LO)
300:
1.178 schwarze 301: static int roffit_lines; /* number of lines to delay */
302: static char *roffit_macro; /* nil-terminated macro line */
303:
1.85 kristaps 304: static void
1.155 kristaps 305: roffhash_init(void)
1.85 kristaps 306: {
307: struct roffmac *n;
308: int buc, i;
309:
1.106 kristaps 310: for (i = 0; i < (int)ROFF_USERDEF; i++) {
1.85 kristaps 311: assert(roffs[i].name[0] >= ASCII_LO);
312: assert(roffs[i].name[0] <= ASCII_HI);
313:
314: buc = ROFF_HASH(roffs[i].name);
315:
316: if (NULL != (n = hash[buc])) {
317: for ( ; n->next; n = n->next)
318: /* Do nothing. */ ;
319: n->next = &roffs[i];
320: } else
321: hash[buc] = &roffs[i];
322: }
323: }
324:
1.67 kristaps 325: /*
326: * Look up a roff token by its name. Returns ROFF_MAX if no macro by
327: * the nil-terminated string name could be found.
328: */
329: static enum rofft
1.155 kristaps 330: roffhash_find(const char *p, size_t s)
1.67 kristaps 331: {
1.85 kristaps 332: int buc;
333: struct roffmac *n;
1.67 kristaps 334:
1.85 kristaps 335: /*
336: * libroff has an extremely simple hashtable, for the time
337: * being, which simply keys on the first character, which must
338: * be printable, then walks a chain. It works well enough until
339: * optimised.
340: */
341:
342: if (p[0] < ASCII_LO || p[0] > ASCII_HI)
343: return(ROFF_MAX);
344:
345: buc = ROFF_HASH(p);
346:
347: if (NULL == (n = hash[buc]))
348: return(ROFF_MAX);
349: for ( ; n; n = n->next)
1.106 kristaps 350: if (0 == strncmp(n->name, p, s) && '\0' == n->name[(int)s])
1.85 kristaps 351: return((enum rofft)(n - roffs));
1.67 kristaps 352:
353: return(ROFF_MAX);
354: }
355:
356:
357: /*
358: * Pop the current node off of the stack of roff instructions currently
359: * pending.
360: */
361: static void
362: roffnode_pop(struct roff *r)
363: {
364: struct roffnode *p;
365:
1.75 kristaps 366: assert(r->last);
367: p = r->last;
1.82 kristaps 368:
1.75 kristaps 369: r->last = r->last->parent;
1.106 kristaps 370: free(p->name);
371: free(p->end);
1.67 kristaps 372: free(p);
373: }
374:
375:
376: /*
377: * Push a roff node onto the instruction stack. This must later be
378: * removed with roffnode_pop().
379: */
1.98 schwarze 380: static void
1.106 kristaps 381: roffnode_push(struct roff *r, enum rofft tok, const char *name,
382: int line, int col)
1.67 kristaps 383: {
384: struct roffnode *p;
385:
1.98 schwarze 386: p = mandoc_calloc(1, sizeof(struct roffnode));
1.67 kristaps 387: p->tok = tok;
1.106 kristaps 388: if (name)
389: p->name = mandoc_strdup(name);
1.67 kristaps 390: p->parent = r->last;
391: p->line = line;
392: p->col = col;
1.79 kristaps 393: p->rule = p->parent ? p->parent->rule : ROFFRULE_DENY;
1.67 kristaps 394:
395: r->last = p;
396: }
397:
398:
399: static void
400: roff_free1(struct roff *r)
401: {
1.176 schwarze 402: struct tbl_node *tbl;
1.125 kristaps 403: struct eqn_node *e;
1.167 kristaps 404: int i;
1.67 kristaps 405:
1.176 schwarze 406: while (NULL != (tbl = r->first_tbl)) {
407: r->first_tbl = tbl->next;
408: tbl_free(tbl);
1.109 kristaps 409: }
410:
1.113 kristaps 411: r->first_tbl = r->last_tbl = r->tbl = NULL;
412:
1.125 kristaps 413: while (NULL != (e = r->first_eqn)) {
414: r->first_eqn = e->next;
415: eqn_free(e);
416: }
417:
418: r->first_eqn = r->last_eqn = r->eqn = NULL;
419:
1.67 kristaps 420: while (r->last)
421: roffnode_pop(r);
1.109 kristaps 422:
1.167 kristaps 423: roff_freestr(r->strtab);
424: roff_freestr(r->xmbtab);
425:
426: r->strtab = r->xmbtab = NULL;
427:
1.180 ! schwarze 428: roff_freereg(r->regtab);
! 429:
! 430: r->regtab = NULL;
! 431:
1.167 kristaps 432: if (r->xtab)
433: for (i = 0; i < 128; i++)
434: free(r->xtab[i].p);
435:
436: free(r->xtab);
437: r->xtab = NULL;
1.67 kristaps 438: }
439:
440: void
441: roff_reset(struct roff *r)
442: {
1.143 kristaps 443: int i;
1.67 kristaps 444:
445: roff_free1(r);
1.143 kristaps 446:
1.174 kristaps 447: r->control = 0;
1.147 kristaps 448:
1.143 kristaps 449: for (i = 0; i < PREDEFS_MAX; i++)
450: roff_setstr(r, predefs[i].name, predefs[i].str, 0);
1.67 kristaps 451: }
452:
453:
454: void
455: roff_free(struct roff *r)
456: {
457:
458: roff_free1(r);
459: free(r);
460: }
461:
462:
463: struct roff *
1.175 schwarze 464: roff_alloc(enum mparset type, struct mparse *parse)
1.67 kristaps 465: {
466: struct roff *r;
1.141 kristaps 467: int i;
1.67 kristaps 468:
1.98 schwarze 469: r = mandoc_calloc(1, sizeof(struct roff));
1.175 schwarze 470: r->parsetype = type;
1.128 kristaps 471: r->parse = parse;
1.82 kristaps 472: r->rstackpos = -1;
1.85 kristaps 473:
1.155 kristaps 474: roffhash_init();
1.141 kristaps 475:
476: for (i = 0; i < PREDEFS_MAX; i++)
477: roff_setstr(r, predefs[i].name, predefs[i].str, 0);
478:
1.67 kristaps 479: return(r);
480: }
481:
1.94 kristaps 482: /*
483: * Pre-filter each and every line for reserved words (one beginning with
484: * `\*', e.g., `\*(ab'). These must be handled before the actual line
485: * is processed.
1.153 kristaps 486: * This also checks the syntax of regular escapes.
1.154 kristaps 487: */
1.172 schwarze 488: static enum rofferr
1.142 kristaps 489: roff_res(struct roff *r, char **bufp, size_t *szp, int ln, int pos)
1.94 kristaps 490: {
1.152 kristaps 491: enum mandoc_esc esc;
1.108 schwarze 492: const char *stesc; /* start of an escape sequence ('\\') */
493: const char *stnam; /* start of the name, after "[(*" */
494: const char *cp; /* end of the name, e.g. before ']' */
495: const char *res; /* the string to be substituted */
1.170 schwarze 496: int i, maxl, expand_count;
1.94 kristaps 497: size_t nsz;
498: char *n;
499:
1.170 schwarze 500: expand_count = 0;
501:
1.154 kristaps 502: again:
1.108 schwarze 503: cp = *bufp + pos;
504: while (NULL != (cp = strchr(cp, '\\'))) {
505: stesc = cp++;
506:
507: /*
508: * The second character must be an asterisk.
509: * If it isn't, skip it anyway: It is escaped,
510: * so it can't start another escape sequence.
511: */
512:
513: if ('\0' == *cp)
1.172 schwarze 514: return(ROFF_CONT);
1.152 kristaps 515:
516: if ('*' != *cp) {
517: res = cp;
518: esc = mandoc_escape(&cp, NULL, NULL);
519: if (ESCAPE_ERROR != esc)
520: continue;
521: cp = res;
1.153 kristaps 522: mandoc_msg
523: (MANDOCERR_BADESCAPE, r->parse,
524: ln, (int)(stesc - *bufp), NULL);
1.172 schwarze 525: return(ROFF_CONT);
1.152 kristaps 526: }
527:
528: cp++;
1.108 schwarze 529:
530: /*
531: * The third character decides the length
532: * of the name of the string.
533: * Save a pointer to the name.
534: */
535:
1.94 kristaps 536: switch (*cp) {
1.108 schwarze 537: case ('\0'):
1.172 schwarze 538: return(ROFF_CONT);
1.94 kristaps 539: case ('('):
540: cp++;
541: maxl = 2;
542: break;
543: case ('['):
544: cp++;
545: maxl = 0;
546: break;
547: default:
548: maxl = 1;
549: break;
550: }
1.108 schwarze 551: stnam = cp;
1.94 kristaps 552:
1.108 schwarze 553: /* Advance to the end of the name. */
1.94 kristaps 554:
555: for (i = 0; 0 == maxl || i < maxl; i++, cp++) {
1.153 kristaps 556: if ('\0' == *cp) {
557: mandoc_msg
558: (MANDOCERR_BADESCAPE,
559: r->parse, ln,
560: (int)(stesc - *bufp), NULL);
1.172 schwarze 561: return(ROFF_CONT);
1.153 kristaps 562: }
1.94 kristaps 563: if (0 == maxl && ']' == *cp)
564: break;
565: }
566:
1.108 schwarze 567: /*
568: * Retrieve the replacement string; if it is
569: * undefined, resume searching for escapes.
570: */
571:
572: res = roff_getstrn(r, stnam, (size_t)i);
1.94 kristaps 573:
574: if (NULL == res) {
1.153 kristaps 575: mandoc_msg
576: (MANDOCERR_BADESCAPE, r->parse,
577: ln, (int)(stesc - *bufp), NULL);
1.142 kristaps 578: res = "";
1.94 kristaps 579: }
580:
1.108 schwarze 581: /* Replace the escape sequence by the string. */
582:
1.161 kristaps 583: pos = stesc - *bufp;
1.154 kristaps 584:
1.94 kristaps 585: nsz = *szp + strlen(res) + 1;
586: n = mandoc_malloc(nsz);
587:
1.108 schwarze 588: strlcpy(n, *bufp, (size_t)(stesc - *bufp + 1));
1.94 kristaps 589: strlcat(n, res, nsz);
590: strlcat(n, cp + (maxl ? 0 : 1), nsz);
591:
592: free(*bufp);
593:
594: *bufp = n;
595: *szp = nsz;
1.170 schwarze 596:
597: if (EXPAND_LIMIT >= ++expand_count)
598: goto again;
599:
600: /* Just leave the string unexpanded. */
601: mandoc_msg(MANDOCERR_ROFFLOOP, r->parse, ln, pos, NULL);
1.172 schwarze 602: return(ROFF_IGN);
1.154 kristaps 603: }
1.172 schwarze 604: return(ROFF_CONT);
1.154 kristaps 605: }
606:
607: /*
1.178 schwarze 608: * Process text streams:
609: * Convert all breakable hyphens into ASCII_HYPH.
610: * Decrement and spring input line trap.
1.154 kristaps 611: */
612: static enum rofferr
1.178 schwarze 613: roff_parsetext(char **bufp, size_t *szp, int pos, int *offs)
1.154 kristaps 614: {
615: size_t sz;
616: const char *start;
1.178 schwarze 617: char *p;
618: int isz;
1.154 kristaps 619: enum mandoc_esc esc;
620:
1.178 schwarze 621: start = p = *bufp + pos;
1.154 kristaps 622:
623: while ('\0' != *p) {
624: sz = strcspn(p, "-\\");
625: p += sz;
626:
1.159 kristaps 627: if ('\0' == *p)
628: break;
629:
1.154 kristaps 630: if ('\\' == *p) {
631: /* Skip over escapes. */
632: p++;
633: esc = mandoc_escape
1.179 schwarze 634: ((const char const **)&p, NULL, NULL);
1.154 kristaps 635: if (ESCAPE_ERROR == esc)
636: break;
1.155 kristaps 637: continue;
1.159 kristaps 638: } else if (p == start) {
1.158 kristaps 639: p++;
1.155 kristaps 640: continue;
1.158 kristaps 641: }
1.155 kristaps 642:
1.171 schwarze 643: if (isalpha((unsigned char)p[-1]) &&
644: isalpha((unsigned char)p[1]))
1.155 kristaps 645: *p = ASCII_HYPH;
646: p++;
1.94 kristaps 647: }
648:
1.178 schwarze 649: /* Spring the input line trap. */
650: if (1 == roffit_lines) {
651: isz = asprintf(&p, "%s\n.%s", *bufp, roffit_macro);
652: if (-1 == isz) {
653: perror(NULL);
654: exit((int)MANDOCLEVEL_SYSERR);
655: }
656: free(*bufp);
657: *bufp = p;
658: *szp = isz + 1;
659: *offs = 0;
660: free(roffit_macro);
661: roffit_lines = 0;
662: return(ROFF_REPARSE);
663: } else if (1 < roffit_lines)
664: --roffit_lines;
1.154 kristaps 665: return(ROFF_CONT);
1.94 kristaps 666: }
667:
1.67 kristaps 668: enum rofferr
1.90 kristaps 669: roff_parseln(struct roff *r, int ln, char **bufp,
670: size_t *szp, int pos, int *offs)
1.67 kristaps 671: {
672: enum rofft t;
1.109 kristaps 673: enum rofferr e;
1.130 kristaps 674: int ppos, ctl;
1.79 kristaps 675:
676: /*
1.94 kristaps 677: * Run the reserved-word filter only if we have some reserved
678: * words to fill in.
679: */
680:
1.172 schwarze 681: e = roff_res(r, bufp, szp, ln, pos);
682: if (ROFF_IGN == e)
683: return(e);
684: assert(ROFF_CONT == e);
1.94 kristaps 685:
1.130 kristaps 686: ppos = pos;
1.174 kristaps 687: ctl = roff_getcontrol(r, *bufp, &pos);
1.130 kristaps 688:
1.94 kristaps 689: /*
1.79 kristaps 690: * First, if a scope is open and we're not a macro, pass the
691: * text through the macro's filter. If a scope isn't open and
692: * we're not a macro, just let it through.
1.125 kristaps 693: * Finally, if there's an equation scope open, divert it into it
694: * no matter our state.
1.79 kristaps 695: */
1.74 kristaps 696:
1.130 kristaps 697: if (r->last && ! ctl) {
1.78 kristaps 698: t = r->last->tok;
699: assert(roffs[t].text);
1.109 kristaps 700: e = (*roffs[t].text)
701: (r, t, bufp, szp, ln, pos, pos, offs);
702: assert(ROFF_IGN == e || ROFF_CONT == e);
1.125 kristaps 703: if (ROFF_CONT != e)
704: return(e);
705: if (r->eqn)
1.146 kristaps 706: return(eqn_read(&r->eqn, ln, *bufp, pos, offs));
1.125 kristaps 707: if (r->tbl)
1.130 kristaps 708: return(tbl_read(r->tbl, ln, *bufp, pos));
1.178 schwarze 709: return(roff_parsetext(bufp, szp, pos, offs));
1.130 kristaps 710: } else if ( ! ctl) {
1.125 kristaps 711: if (r->eqn)
1.146 kristaps 712: return(eqn_read(&r->eqn, ln, *bufp, pos, offs));
1.109 kristaps 713: if (r->tbl)
1.130 kristaps 714: return(tbl_read(r->tbl, ln, *bufp, pos));
1.178 schwarze 715: return(roff_parsetext(bufp, szp, pos, offs));
1.125 kristaps 716: } else if (r->eqn)
1.146 kristaps 717: return(eqn_read(&r->eqn, ln, *bufp, ppos, offs));
1.67 kristaps 718:
1.79 kristaps 719: /*
720: * If a scope is open, go to the child handler for that macro,
721: * as it may want to preprocess before doing anything with it.
1.125 kristaps 722: * Don't do so if an equation is open.
1.79 kristaps 723: */
1.78 kristaps 724:
1.79 kristaps 725: if (r->last) {
726: t = r->last->tok;
727: assert(roffs[t].sub);
728: return((*roffs[t].sub)
1.90 kristaps 729: (r, t, bufp, szp,
1.130 kristaps 730: ln, ppos, pos, offs));
1.79 kristaps 731: }
1.78 kristaps 732:
1.79 kristaps 733: /*
734: * Lastly, as we've no scope open, try to look up and execute
735: * the new macro. If no macro is found, simply return and let
736: * the compilers handle it.
737: */
1.67 kristaps 738:
1.106 kristaps 739: if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos)))
1.79 kristaps 740: return(ROFF_CONT);
1.67 kristaps 741:
1.75 kristaps 742: assert(roffs[t].proc);
1.78 kristaps 743: return((*roffs[t].proc)
1.90 kristaps 744: (r, t, bufp, szp,
745: ln, ppos, pos, offs));
1.74 kristaps 746: }
747:
748:
1.117 kristaps 749: void
1.74 kristaps 750: roff_endparse(struct roff *r)
751: {
752:
1.110 kristaps 753: if (r->last)
1.128 kristaps 754: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.109 kristaps 755: r->last->line, r->last->col, NULL);
1.117 kristaps 756:
1.125 kristaps 757: if (r->eqn) {
1.128 kristaps 758: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.148 kristaps 759: r->eqn->eqn.ln, r->eqn->eqn.pos, NULL);
1.151 kristaps 760: eqn_end(&r->eqn);
1.125 kristaps 761: }
762:
1.117 kristaps 763: if (r->tbl) {
1.128 kristaps 764: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.117 kristaps 765: r->tbl->line, r->tbl->pos, NULL);
1.151 kristaps 766: tbl_end(&r->tbl);
1.117 kristaps 767: }
1.67 kristaps 768: }
769:
770: /*
771: * Parse a roff node's type from the input buffer. This must be in the
772: * form of ".foo xxx" in the usual way.
773: */
774: static enum rofft
1.106 kristaps 775: roff_parse(struct roff *r, const char *buf, int *pos)
1.67 kristaps 776: {
1.106 kristaps 777: const char *mac;
778: size_t maclen;
1.67 kristaps 779: enum rofft t;
780:
1.144 kristaps 781: if ('\0' == buf[*pos] || '"' == buf[*pos] ||
782: '\t' == buf[*pos] || ' ' == buf[*pos])
1.67 kristaps 783: return(ROFF_MAX);
784:
1.144 kristaps 785: /*
786: * We stop the macro parse at an escape, tab, space, or nil.
787: * However, `\}' is also a valid macro, so make sure we don't
788: * clobber it by seeing the `\' as the end of token.
789: */
790:
1.106 kristaps 791: mac = buf + *pos;
1.144 kristaps 792: maclen = strcspn(mac + 1, " \\\t\0") + 1;
1.67 kristaps 793:
1.106 kristaps 794: t = (r->current_string = roff_getstrn(r, mac, maclen))
1.155 kristaps 795: ? ROFF_USERDEF : roffhash_find(mac, maclen);
1.67 kristaps 796:
1.127 kristaps 797: *pos += (int)maclen;
1.130 kristaps 798:
1.67 kristaps 799: while (buf[*pos] && ' ' == buf[*pos])
800: (*pos)++;
801:
802: return(t);
803: }
804:
805: /* ARGSUSED */
806: static enum rofferr
1.76 kristaps 807: roff_cblock(ROFF_ARGS)
1.67 kristaps 808: {
809:
1.79 kristaps 810: /*
811: * A block-close `..' should only be invoked as a child of an
812: * ignore macro, otherwise raise a warning and just ignore it.
813: */
814:
1.76 kristaps 815: if (NULL == r->last) {
1.128 kristaps 816: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.76 kristaps 817: return(ROFF_IGN);
818: }
1.67 kristaps 819:
1.81 kristaps 820: switch (r->last->tok) {
821: case (ROFF_am):
822: /* FALLTHROUGH */
823: case (ROFF_ami):
824: /* FALLTHROUGH */
825: case (ROFF_am1):
826: /* FALLTHROUGH */
827: case (ROFF_de):
1.108 schwarze 828: /* ROFF_de1 is remapped to ROFF_de in roff_block(). */
1.81 kristaps 829: /* FALLTHROUGH */
830: case (ROFF_dei):
831: /* FALLTHROUGH */
832: case (ROFF_ig):
833: break;
834: default:
1.128 kristaps 835: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.67 kristaps 836: return(ROFF_IGN);
1.76 kristaps 837: }
1.67 kristaps 838:
1.76 kristaps 839: if ((*bufp)[pos])
1.128 kristaps 840: mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.71 kristaps 841:
842: roffnode_pop(r);
1.76 kristaps 843: roffnode_cleanscope(r);
844: return(ROFF_IGN);
1.71 kristaps 845:
1.67 kristaps 846: }
847:
848:
1.76 kristaps 849: static void
850: roffnode_cleanscope(struct roff *r)
1.67 kristaps 851: {
852:
1.76 kristaps 853: while (r->last) {
1.173 schwarze 854: if (--r->last->endspan != 0)
1.76 kristaps 855: break;
856: roffnode_pop(r);
857: }
1.67 kristaps 858: }
859:
860:
1.75 kristaps 861: /* ARGSUSED */
1.74 kristaps 862: static enum rofferr
1.75 kristaps 863: roff_ccond(ROFF_ARGS)
1.74 kristaps 864: {
865:
1.76 kristaps 866: if (NULL == r->last) {
1.128 kristaps 867: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.76 kristaps 868: return(ROFF_IGN);
869: }
870:
1.82 kristaps 871: switch (r->last->tok) {
872: case (ROFF_el):
873: /* FALLTHROUGH */
874: case (ROFF_ie):
875: /* FALLTHROUGH */
876: case (ROFF_if):
877: break;
878: default:
1.128 kristaps 879: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.75 kristaps 880: return(ROFF_IGN);
881: }
882:
1.76 kristaps 883: if (r->last->endspan > -1) {
1.128 kristaps 884: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.76 kristaps 885: return(ROFF_IGN);
886: }
887:
888: if ((*bufp)[pos])
1.128 kristaps 889: mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.76 kristaps 890:
1.75 kristaps 891: roffnode_pop(r);
1.76 kristaps 892: roffnode_cleanscope(r);
893: return(ROFF_IGN);
894: }
895:
1.75 kristaps 896:
1.76 kristaps 897: /* ARGSUSED */
898: static enum rofferr
1.80 kristaps 899: roff_block(ROFF_ARGS)
1.76 kristaps 900: {
1.78 kristaps 901: int sv;
902: size_t sz;
1.106 kristaps 903: char *name;
904:
905: name = NULL;
1.76 kristaps 906:
1.106 kristaps 907: if (ROFF_ig != tok) {
908: if ('\0' == (*bufp)[pos]) {
1.128 kristaps 909: mandoc_msg(MANDOCERR_NOARGS, r->parse, ln, ppos, NULL);
1.106 kristaps 910: return(ROFF_IGN);
911: }
1.107 kristaps 912:
913: /*
914: * Re-write `de1', since we don't really care about
915: * groff's strange compatibility mode, into `de'.
916: */
917:
1.106 kristaps 918: if (ROFF_de1 == tok)
919: tok = ROFF_de;
920: if (ROFF_de == tok)
921: name = *bufp + pos;
922: else
1.128 kristaps 923: mandoc_msg(MANDOCERR_REQUEST, r->parse, ln, ppos,
1.106 kristaps 924: roffs[tok].name);
1.107 kristaps 925:
1.131 schwarze 926: while ((*bufp)[pos] && ! isspace((unsigned char)(*bufp)[pos]))
1.80 kristaps 927: pos++;
1.107 kristaps 928:
1.131 schwarze 929: while (isspace((unsigned char)(*bufp)[pos]))
1.106 kristaps 930: (*bufp)[pos++] = '\0';
1.80 kristaps 931: }
932:
1.106 kristaps 933: roffnode_push(r, tok, name, ln, ppos);
934:
935: /*
936: * At the beginning of a `de' macro, clear the existing string
937: * with the same name, if there is one. New content will be
938: * added from roff_block_text() in multiline mode.
939: */
1.107 kristaps 940:
1.106 kristaps 941: if (ROFF_de == tok)
1.108 schwarze 942: roff_setstr(r, name, "", 0);
1.76 kristaps 943:
1.79 kristaps 944: if ('\0' == (*bufp)[pos])
1.78 kristaps 945: return(ROFF_IGN);
946:
1.107 kristaps 947: /* If present, process the custom end-of-line marker. */
948:
1.78 kristaps 949: sv = pos;
1.131 schwarze 950: while ((*bufp)[pos] && ! isspace((unsigned char)(*bufp)[pos]))
1.78 kristaps 951: pos++;
952:
953: /*
954: * Note: groff does NOT like escape characters in the input.
955: * Instead of detecting this, we're just going to let it fly and
956: * to hell with it.
957: */
958:
959: assert(pos > sv);
960: sz = (size_t)(pos - sv);
961:
1.79 kristaps 962: if (1 == sz && '.' == (*bufp)[sv])
963: return(ROFF_IGN);
964:
1.98 schwarze 965: r->last->end = mandoc_malloc(sz + 1);
1.78 kristaps 966:
967: memcpy(r->last->end, *bufp + sv, sz);
968: r->last->end[(int)sz] = '\0';
969:
1.77 kristaps 970: if ((*bufp)[pos])
1.128 kristaps 971: mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.74 kristaps 972:
1.78 kristaps 973: return(ROFF_IGN);
974: }
975:
976:
977: /* ARGSUSED */
978: static enum rofferr
1.80 kristaps 979: roff_block_sub(ROFF_ARGS)
1.79 kristaps 980: {
981: enum rofft t;
982: int i, j;
983:
984: /*
985: * First check whether a custom macro exists at this level. If
986: * it does, then check against it. This is some of groff's
987: * stranger behaviours. If we encountered a custom end-scope
988: * tag and that tag also happens to be a "real" macro, then we
989: * need to try interpreting it again as a real macro. If it's
990: * not, then return ignore. Else continue.
991: */
992:
993: if (r->last->end) {
1.130 kristaps 994: for (i = pos, j = 0; r->last->end[j]; j++, i++)
1.79 kristaps 995: if ((*bufp)[i] != r->last->end[j])
996: break;
997:
998: if ('\0' == r->last->end[j] &&
999: ('\0' == (*bufp)[i] ||
1000: ' ' == (*bufp)[i] ||
1001: '\t' == (*bufp)[i])) {
1002: roffnode_pop(r);
1003: roffnode_cleanscope(r);
1004:
1.130 kristaps 1005: while (' ' == (*bufp)[i] || '\t' == (*bufp)[i])
1006: i++;
1007:
1008: pos = i;
1.106 kristaps 1009: if (ROFF_MAX != roff_parse(r, *bufp, &pos))
1.79 kristaps 1010: return(ROFF_RERUN);
1011: return(ROFF_IGN);
1012: }
1013: }
1014:
1015: /*
1016: * If we have no custom end-query or lookup failed, then try
1017: * pulling it out of the hashtable.
1018: */
1019:
1.137 schwarze 1020: t = roff_parse(r, *bufp, &pos);
1.79 kristaps 1021:
1.106 kristaps 1022: /*
1023: * Macros other than block-end are only significant
1024: * in `de' blocks; elsewhere, simply throw them away.
1025: */
1026: if (ROFF_cblock != t) {
1027: if (ROFF_de == tok)
1028: roff_setstr(r, r->last->name, *bufp + ppos, 1);
1.79 kristaps 1029: return(ROFF_IGN);
1.106 kristaps 1030: }
1.79 kristaps 1031:
1032: assert(roffs[t].proc);
1.90 kristaps 1033: return((*roffs[t].proc)(r, t, bufp, szp,
1034: ln, ppos, pos, offs));
1.79 kristaps 1035: }
1036:
1037:
1038: /* ARGSUSED */
1039: static enum rofferr
1.80 kristaps 1040: roff_block_text(ROFF_ARGS)
1.78 kristaps 1041: {
1042:
1.106 kristaps 1043: if (ROFF_de == tok)
1044: roff_setstr(r, r->last->name, *bufp + pos, 1);
1045:
1.78 kristaps 1046: return(ROFF_IGN);
1047: }
1048:
1049:
1050: /* ARGSUSED */
1051: static enum rofferr
1.82 kristaps 1052: roff_cond_sub(ROFF_ARGS)
1053: {
1054: enum rofft t;
1055: enum roffrule rr;
1.139 kristaps 1056: char *ep;
1.82 kristaps 1057:
1058: rr = r->last->rule;
1.139 kristaps 1059: roffnode_cleanscope(r);
1.177 schwarze 1060: t = roff_parse(r, *bufp, &pos);
1.82 kristaps 1061:
1.139 kristaps 1062: /*
1.177 schwarze 1063: * Fully handle known macros when they are structurally
1064: * required or when the conditional evaluated to true.
1.87 kristaps 1065: */
1066:
1.177 schwarze 1067: if ((ROFF_MAX != t) &&
1068: (ROFF_ccond == t || ROFFRULE_ALLOW == rr ||
1069: ROFFMAC_STRUCT & roffs[t].flags)) {
1070: assert(roffs[t].proc);
1071: return((*roffs[t].proc)(r, t, bufp, szp,
1072: ln, ppos, pos, offs));
1073: }
1.144 kristaps 1074:
1.177 schwarze 1075: /* Always check for the closing delimiter `\}'. */
1.144 kristaps 1076:
1.177 schwarze 1077: ep = &(*bufp)[pos];
1078: while (NULL != (ep = strchr(ep, '\\'))) {
1079: if ('}' != *(++ep))
1080: continue;
1.82 kristaps 1081:
1.177 schwarze 1082: /*
1083: * If we're at the end of line, then just chop
1084: * off the \} and resize the buffer.
1085: * If we aren't, then convert it to spaces.
1086: */
1.139 kristaps 1087:
1.177 schwarze 1088: if ('\0' == *(ep + 1)) {
1089: *--ep = '\0';
1090: *szp -= 2;
1091: } else
1092: *(ep - 1) = *ep = ' ';
1.82 kristaps 1093:
1.177 schwarze 1094: roff_ccond(r, ROFF_ccond, bufp, szp,
1095: ln, pos, pos + 2, offs);
1096: break;
1097: }
1098: return(ROFFRULE_DENY == rr ? ROFF_IGN : ROFF_CONT);
1.82 kristaps 1099: }
1100:
1101: /* ARGSUSED */
1102: static enum rofferr
1103: roff_cond_text(ROFF_ARGS)
1.78 kristaps 1104: {
1.140 kristaps 1105: char *ep;
1.82 kristaps 1106: enum roffrule rr;
1107:
1108: rr = r->last->rule;
1.140 kristaps 1109: roffnode_cleanscope(r);
1.82 kristaps 1110:
1.140 kristaps 1111: ep = &(*bufp)[pos];
1112: for ( ; NULL != (ep = strchr(ep, '\\')); ep++) {
1113: ep++;
1114: if ('}' != *ep)
1115: continue;
1116: *ep = '&';
1117: roff_ccond(r, ROFF_ccond, bufp, szp,
1118: ln, pos, pos + 2, offs);
1.78 kristaps 1119: }
1.82 kristaps 1120: return(ROFFRULE_DENY == rr ? ROFF_IGN : ROFF_CONT);
1.74 kristaps 1121: }
1122:
1.88 kristaps 1123: static enum roffrule
1124: roff_evalcond(const char *v, int *pos)
1125: {
1126:
1127: switch (v[*pos]) {
1128: case ('n'):
1129: (*pos)++;
1130: return(ROFFRULE_ALLOW);
1131: case ('e'):
1132: /* FALLTHROUGH */
1133: case ('o'):
1134: /* FALLTHROUGH */
1135: case ('t'):
1136: (*pos)++;
1137: return(ROFFRULE_DENY);
1138: default:
1139: break;
1140: }
1141:
1142: while (v[*pos] && ' ' != v[*pos])
1143: (*pos)++;
1144: return(ROFFRULE_DENY);
1145: }
1146:
1.75 kristaps 1147: /* ARGSUSED */
1.74 kristaps 1148: static enum rofferr
1.103 kristaps 1149: roff_line_ignore(ROFF_ARGS)
1.89 kristaps 1150: {
1.123 schwarze 1151:
1.89 kristaps 1152: return(ROFF_IGN);
1153: }
1154:
1.104 kristaps 1155: /* ARGSUSED */
1156: static enum rofferr
1.82 kristaps 1157: roff_cond(ROFF_ARGS)
1.74 kristaps 1158: {
1.173 schwarze 1159:
1160: roffnode_push(r, tok, NULL, ln, ppos);
1.74 kristaps 1161:
1.134 kristaps 1162: /*
1163: * An `.el' has no conditional body: it will consume the value
1164: * of the current rstack entry set in prior `ie' calls or
1165: * defaults to DENY.
1166: *
1167: * If we're not an `el', however, then evaluate the conditional.
1168: */
1.133 kristaps 1169:
1.173 schwarze 1170: r->last->rule = ROFF_el == tok ?
1.134 kristaps 1171: (r->rstackpos < 0 ?
1172: ROFFRULE_DENY : r->rstack[r->rstackpos--]) :
1173: roff_evalcond(*bufp, &pos);
1.77 kristaps 1174:
1.134 kristaps 1175: /*
1176: * An if-else will put the NEGATION of the current evaluated
1177: * conditional into the stack of rules.
1178: */
1179:
1.84 schwarze 1180: if (ROFF_ie == tok) {
1.134 kristaps 1181: if (r->rstackpos == RSTACK_MAX - 1) {
1182: mandoc_msg(MANDOCERR_MEM,
1183: r->parse, ln, ppos, NULL);
1184: return(ROFF_ERR);
1185: }
1186: r->rstack[++r->rstackpos] =
1187: ROFFRULE_DENY == r->last->rule ?
1188: ROFFRULE_ALLOW : ROFFRULE_DENY;
1.82 kristaps 1189: }
1.88 kristaps 1190:
1191: /* If the parent has false as its rule, then so do we. */
1192:
1.109 kristaps 1193: if (r->last->parent && ROFFRULE_DENY == r->last->parent->rule)
1.84 schwarze 1194: r->last->rule = ROFFRULE_DENY;
1.88 kristaps 1195:
1196: /*
1.173 schwarze 1197: * Determine scope.
1198: * If there is nothing on the line after the conditional,
1199: * not even whitespace, use next-line scope.
1.88 kristaps 1200: */
1.74 kristaps 1201:
1.173 schwarze 1202: if ('\0' == (*bufp)[pos]) {
1203: r->last->endspan = 2;
1204: goto out;
1205: }
1206:
1207: while (' ' == (*bufp)[pos])
1208: pos++;
1209:
1210: /* An opening brace requests multiline scope. */
1.75 kristaps 1211:
1212: if ('\\' == (*bufp)[pos] && '{' == (*bufp)[pos + 1]) {
1213: r->last->endspan = -1;
1214: pos += 2;
1.173 schwarze 1215: goto out;
1.109 kristaps 1216: }
1.74 kristaps 1217:
1.77 kristaps 1218: /*
1.173 schwarze 1219: * Anything else following the conditional causes
1220: * single-line scope. Warn if the scope contains
1221: * nothing but trailing whitespace.
1.77 kristaps 1222: */
1223:
1.75 kristaps 1224: if ('\0' == (*bufp)[pos])
1.173 schwarze 1225: mandoc_msg(MANDOCERR_NOARGS, r->parse, ln, ppos, NULL);
1.77 kristaps 1226:
1.173 schwarze 1227: r->last->endspan = 1;
1.74 kristaps 1228:
1.173 schwarze 1229: out:
1.75 kristaps 1230: *offs = pos;
1231: return(ROFF_RERUN);
1.83 schwarze 1232: }
1233:
1234:
1235: /* ARGSUSED */
1236: static enum rofferr
1.92 schwarze 1237: roff_ds(ROFF_ARGS)
1238: {
1.96 kristaps 1239: char *name, *string;
1240:
1241: /*
1242: * A symbol is named by the first word following the macro
1243: * invocation up to a space. Its value is anything after the
1244: * name's trailing whitespace and optional double-quote. Thus,
1245: *
1246: * [.ds foo "bar " ]
1247: *
1248: * will have `bar " ' as its value.
1249: */
1.92 schwarze 1250:
1.121 schwarze 1251: string = *bufp + pos;
1252: name = roff_getname(r, &string, ln, pos);
1.92 schwarze 1253: if ('\0' == *name)
1254: return(ROFF_IGN);
1255:
1.121 schwarze 1256: /* Read past initial double-quote. */
1257: if ('"' == *string)
1.92 schwarze 1258: string++;
1259:
1.96 kristaps 1260: /* The rest is the value. */
1.106 kristaps 1261: roff_setstr(r, name, string, 0);
1.92 schwarze 1262: return(ROFF_IGN);
1263: }
1264:
1.180 ! schwarze 1265: void
! 1266: roff_setreg(struct roff *r, const char *name, unsigned int val)
1.147 kristaps 1267: {
1.180 ! schwarze 1268: struct roffreg *reg;
! 1269:
! 1270: /* Search for an existing register with the same name. */
! 1271: reg = r->regtab;
! 1272:
! 1273: while (reg && strcmp(name, reg->key.p))
! 1274: reg = reg->next;
1.147 kristaps 1275:
1.180 ! schwarze 1276: if (NULL == reg) {
! 1277: /* Create a new register. */
! 1278: reg = mandoc_malloc(sizeof(struct roffreg));
! 1279: reg->key.p = mandoc_strdup(name);
! 1280: reg->key.sz = strlen(name);
! 1281: reg->next = r->regtab;
! 1282: r->regtab = reg;
! 1283: }
! 1284:
! 1285: reg->u = val;
1.147 kristaps 1286: }
1287:
1288: unsigned int
1.180 ! schwarze 1289: roff_getreg(const struct roff *r, const char *name)
1.147 kristaps 1290: {
1.180 ! schwarze 1291: struct roffreg *reg;
! 1292:
! 1293: for (reg = r->regtab; reg; reg = reg->next)
! 1294: if (0 == strcmp(name, reg->key.p))
! 1295: return(reg->u);
1.147 kristaps 1296:
1.180 ! schwarze 1297: return(0);
1.147 kristaps 1298: }
1299:
1.180 ! schwarze 1300: static void
! 1301: roff_freereg(struct roffreg *reg)
1.147 kristaps 1302: {
1.180 ! schwarze 1303: struct roffreg *old_reg;
1.147 kristaps 1304:
1.180 ! schwarze 1305: while (NULL != reg) {
! 1306: free(reg->key.p);
! 1307: old_reg = reg;
! 1308: reg = reg->next;
! 1309: free(old_reg);
! 1310: }
1.147 kristaps 1311: }
1.92 schwarze 1312:
1313: /* ARGSUSED */
1314: static enum rofferr
1.89 kristaps 1315: roff_nr(ROFF_ARGS)
1.83 schwarze 1316: {
1.121 schwarze 1317: const char *key;
1318: char *val;
1.138 kristaps 1319: int iv;
1.89 kristaps 1320:
1.121 schwarze 1321: val = *bufp + pos;
1322: key = roff_getname(r, &val, ln, pos);
1.89 kristaps 1323:
1.180 ! schwarze 1324: iv = mandoc_strntoi(val, strlen(val), 10);
! 1325: if (0 > iv)
! 1326: iv = 0;
! 1327:
! 1328: roff_setreg(r, key, (unsigned)iv);
1.109 kristaps 1329:
1.122 schwarze 1330: return(ROFF_IGN);
1331: }
1332:
1333: /* ARGSUSED */
1334: static enum rofferr
1335: roff_rm(ROFF_ARGS)
1336: {
1337: const char *name;
1338: char *cp;
1339:
1340: cp = *bufp + pos;
1341: while ('\0' != *cp) {
1.127 kristaps 1342: name = roff_getname(r, &cp, ln, (int)(cp - *bufp));
1.122 schwarze 1343: if ('\0' != *name)
1344: roff_setstr(r, name, NULL, 0);
1345: }
1.178 schwarze 1346: return(ROFF_IGN);
1347: }
1348:
1349: /* ARGSUSED */
1350: static enum rofferr
1351: roff_it(ROFF_ARGS)
1352: {
1353: char *cp;
1354: size_t len;
1355: int iv;
1356:
1357: /* Parse the number of lines. */
1358: cp = *bufp + pos;
1359: len = strcspn(cp, " \t");
1360: cp[len] = '\0';
1361: if ((iv = mandoc_strntoi(cp, len, 10)) <= 0) {
1362: mandoc_msg(MANDOCERR_NUMERIC, r->parse,
1363: ln, ppos, *bufp + 1);
1364: return(ROFF_IGN);
1365: }
1366: cp += len + 1;
1367:
1368: /* Arm the input line trap. */
1369: roffit_lines = iv;
1370: roffit_macro = mandoc_strdup(cp);
1.109 kristaps 1371: return(ROFF_IGN);
1.175 schwarze 1372: }
1373:
1374: /* ARGSUSED */
1375: static enum rofferr
1376: roff_Dd(ROFF_ARGS)
1377: {
1378: const char *const *cp;
1379:
1380: if (MPARSE_MDOC != r->parsetype)
1381: for (cp = __mdoc_reserved; *cp; cp++)
1382: roff_setstr(r, *cp, NULL, 0);
1383:
1384: return(ROFF_CONT);
1385: }
1386:
1387: /* ARGSUSED */
1388: static enum rofferr
1389: roff_TH(ROFF_ARGS)
1390: {
1391: const char *const *cp;
1392:
1393: if (MPARSE_MDOC != r->parsetype)
1394: for (cp = __man_reserved; *cp; cp++)
1395: roff_setstr(r, *cp, NULL, 0);
1396:
1397: return(ROFF_CONT);
1.109 kristaps 1398: }
1399:
1400: /* ARGSUSED */
1401: static enum rofferr
1402: roff_TE(ROFF_ARGS)
1403: {
1404:
1405: if (NULL == r->tbl)
1.128 kristaps 1406: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.115 kristaps 1407: else
1.151 kristaps 1408: tbl_end(&r->tbl);
1.109 kristaps 1409:
1.112 kristaps 1410: return(ROFF_IGN);
1411: }
1412:
1413: /* ARGSUSED */
1414: static enum rofferr
1415: roff_T_(ROFF_ARGS)
1416: {
1417:
1418: if (NULL == r->tbl)
1.128 kristaps 1419: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.112 kristaps 1420: else
1.116 kristaps 1421: tbl_restart(ppos, ln, r->tbl);
1.112 kristaps 1422:
1.109 kristaps 1423: return(ROFF_IGN);
1424: }
1425:
1.156 kristaps 1426: #if 0
1427: static int
1.151 kristaps 1428: roff_closeeqn(struct roff *r)
1429: {
1430:
1431: return(r->eqn && ROFF_EQN == eqn_end(&r->eqn) ? 1 : 0);
1432: }
1.156 kristaps 1433: #endif
1.151 kristaps 1434:
1.156 kristaps 1435: static void
1.151 kristaps 1436: roff_openeqn(struct roff *r, const char *name, int line,
1437: int offs, const char *buf)
1.125 kristaps 1438: {
1.151 kristaps 1439: struct eqn_node *e;
1440: int poff;
1.125 kristaps 1441:
1442: assert(NULL == r->eqn);
1.151 kristaps 1443: e = eqn_alloc(name, offs, line, r->parse);
1.125 kristaps 1444:
1445: if (r->last_eqn)
1446: r->last_eqn->next = e;
1447: else
1448: r->first_eqn = r->last_eqn = e;
1449:
1450: r->eqn = r->last_eqn = e;
1.151 kristaps 1451:
1452: if (buf) {
1453: poff = 0;
1454: eqn_read(&r->eqn, line, buf, offs, &poff);
1455: }
1456: }
1457:
1458: /* ARGSUSED */
1459: static enum rofferr
1460: roff_EQ(ROFF_ARGS)
1461: {
1462:
1463: roff_openeqn(r, *bufp + pos, ln, ppos, NULL);
1.125 kristaps 1464: return(ROFF_IGN);
1465: }
1466:
1467: /* ARGSUSED */
1468: static enum rofferr
1469: roff_EN(ROFF_ARGS)
1470: {
1471:
1.128 kristaps 1472: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.125 kristaps 1473: return(ROFF_IGN);
1474: }
1475:
1476: /* ARGSUSED */
1477: static enum rofferr
1.109 kristaps 1478: roff_TS(ROFF_ARGS)
1479: {
1.176 schwarze 1480: struct tbl_node *tbl;
1.89 kristaps 1481:
1.115 kristaps 1482: if (r->tbl) {
1.128 kristaps 1483: mandoc_msg(MANDOCERR_SCOPEBROKEN, r->parse, ln, ppos, NULL);
1.151 kristaps 1484: tbl_end(&r->tbl);
1.115 kristaps 1485: }
1.83 schwarze 1486:
1.176 schwarze 1487: tbl = tbl_alloc(ppos, ln, r->parse);
1.113 kristaps 1488:
1489: if (r->last_tbl)
1.176 schwarze 1490: r->last_tbl->next = tbl;
1.113 kristaps 1491: else
1.176 schwarze 1492: r->first_tbl = r->last_tbl = tbl;
1.113 kristaps 1493:
1.176 schwarze 1494: r->tbl = r->last_tbl = tbl;
1.83 schwarze 1495: return(ROFF_IGN);
1.92 schwarze 1496: }
1497:
1.105 kristaps 1498: /* ARGSUSED */
1499: static enum rofferr
1.174 kristaps 1500: roff_cc(ROFF_ARGS)
1501: {
1502: const char *p;
1503:
1504: p = *bufp + pos;
1505:
1506: if ('\0' == *p || '.' == (r->control = *p++))
1507: r->control = 0;
1508:
1509: if ('\0' != *p)
1510: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
1511:
1512: return(ROFF_IGN);
1513: }
1514:
1515: /* ARGSUSED */
1516: static enum rofferr
1.164 kristaps 1517: roff_tr(ROFF_ARGS)
1518: {
1519: const char *p, *first, *second;
1520: size_t fsz, ssz;
1521: enum mandoc_esc esc;
1522:
1523: p = *bufp + pos;
1524:
1525: if ('\0' == *p) {
1526: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
1527: return(ROFF_IGN);
1528: }
1529:
1530: while ('\0' != *p) {
1531: fsz = ssz = 1;
1532:
1533: first = p++;
1534: if ('\\' == *first) {
1535: esc = mandoc_escape(&p, NULL, NULL);
1536: if (ESCAPE_ERROR == esc) {
1537: mandoc_msg
1538: (MANDOCERR_BADESCAPE, r->parse,
1539: ln, (int)(p - *bufp), NULL);
1540: return(ROFF_IGN);
1541: }
1542: fsz = (size_t)(p - first);
1543: }
1544:
1545: second = p++;
1546: if ('\\' == *second) {
1547: esc = mandoc_escape(&p, NULL, NULL);
1548: if (ESCAPE_ERROR == esc) {
1549: mandoc_msg
1550: (MANDOCERR_BADESCAPE, r->parse,
1551: ln, (int)(p - *bufp), NULL);
1552: return(ROFF_IGN);
1553: }
1554: ssz = (size_t)(p - second);
1.165 kristaps 1555: } else if ('\0' == *second) {
1.164 kristaps 1556: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse,
1557: ln, (int)(p - *bufp), NULL);
1558: second = " ";
1.165 kristaps 1559: p--;
1.164 kristaps 1560: }
1561:
1.167 kristaps 1562: if (fsz > 1) {
1563: roff_setstrn(&r->xmbtab, first,
1564: fsz, second, ssz, 0);
1565: continue;
1566: }
1567:
1568: if (NULL == r->xtab)
1569: r->xtab = mandoc_calloc
1570: (128, sizeof(struct roffstr));
1571:
1572: free(r->xtab[(int)*first].p);
1573: r->xtab[(int)*first].p = mandoc_strndup(second, ssz);
1574: r->xtab[(int)*first].sz = ssz;
1.164 kristaps 1575: }
1576:
1577: return(ROFF_IGN);
1578: }
1579:
1580: /* ARGSUSED */
1581: static enum rofferr
1.105 kristaps 1582: roff_so(ROFF_ARGS)
1583: {
1584: char *name;
1585:
1.128 kristaps 1586: mandoc_msg(MANDOCERR_SO, r->parse, ln, ppos, NULL);
1.105 kristaps 1587:
1588: /*
1589: * Handle `so'. Be EXTREMELY careful, as we shouldn't be
1590: * opening anything that's not in our cwd or anything beneath
1591: * it. Thus, explicitly disallow traversing up the file-system
1592: * or using absolute paths.
1593: */
1594:
1595: name = *bufp + pos;
1596: if ('/' == *name || strstr(name, "../") || strstr(name, "/..")) {
1.128 kristaps 1597: mandoc_msg(MANDOCERR_SOPATH, r->parse, ln, pos, NULL);
1.105 kristaps 1598: return(ROFF_ERR);
1599: }
1600:
1601: *offs = pos;
1602: return(ROFF_SO);
1603: }
1.92 schwarze 1604:
1.106 kristaps 1605: /* ARGSUSED */
1606: static enum rofferr
1607: roff_userdef(ROFF_ARGS)
1.99 kristaps 1608: {
1.106 kristaps 1609: const char *arg[9];
1610: char *cp, *n1, *n2;
1.119 schwarze 1611: int i;
1.106 kristaps 1612:
1613: /*
1614: * Collect pointers to macro argument strings
1615: * and null-terminate them.
1616: */
1617: cp = *bufp + pos;
1.119 schwarze 1618: for (i = 0; i < 9; i++)
1.120 schwarze 1619: arg[i] = '\0' == *cp ? "" :
1.136 kristaps 1620: mandoc_getarg(r->parse, &cp, ln, &pos);
1.99 kristaps 1621:
1.106 kristaps 1622: /*
1623: * Expand macro arguments.
1.99 kristaps 1624: */
1.106 kristaps 1625: *szp = 0;
1626: n1 = cp = mandoc_strdup(r->current_string);
1627: while (NULL != (cp = strstr(cp, "\\$"))) {
1628: i = cp[2] - '1';
1629: if (0 > i || 8 < i) {
1630: /* Not an argument invocation. */
1631: cp += 2;
1632: continue;
1633: }
1634:
1635: *szp = strlen(n1) - 3 + strlen(arg[i]) + 1;
1636: n2 = mandoc_malloc(*szp);
1637:
1638: strlcpy(n2, n1, (size_t)(cp - n1 + 1));
1639: strlcat(n2, arg[i], *szp);
1640: strlcat(n2, cp + 3, *szp);
1641:
1642: cp = n2 + (cp - n1);
1643: free(n1);
1644: n1 = n2;
1.99 kristaps 1645: }
1646:
1.106 kristaps 1647: /*
1648: * Replace the macro invocation
1649: * by the expanded macro.
1650: */
1651: free(*bufp);
1652: *bufp = n1;
1653: if (0 == *szp)
1654: *szp = strlen(*bufp) + 1;
1655:
1656: return(*szp > 1 && '\n' == (*bufp)[(int)*szp - 2] ?
1657: ROFF_REPARSE : ROFF_APPEND);
1.99 kristaps 1658: }
1.121 schwarze 1659:
1660: static char *
1661: roff_getname(struct roff *r, char **cpp, int ln, int pos)
1662: {
1663: char *name, *cp;
1664:
1665: name = *cpp;
1666: if ('\0' == *name)
1667: return(name);
1668:
1669: /* Read until end of name. */
1670: for (cp = name; '\0' != *cp && ' ' != *cp; cp++) {
1671: if ('\\' != *cp)
1672: continue;
1673: cp++;
1674: if ('\\' == *cp)
1675: continue;
1.128 kristaps 1676: mandoc_msg(MANDOCERR_NAMESC, r->parse, ln, pos, NULL);
1.121 schwarze 1677: *cp = '\0';
1678: name = cp;
1679: }
1680:
1681: /* Nil-terminate name. */
1682: if ('\0' != *cp)
1683: *(cp++) = '\0';
1684:
1685: /* Read past spaces. */
1686: while (' ' == *cp)
1687: cp++;
1688:
1689: *cpp = cp;
1690: return(name);
1691: }
1692:
1.106 kristaps 1693: /*
1694: * Store *string into the user-defined string called *name.
1695: * In multiline mode, append to an existing entry and append '\n';
1696: * else replace the existing entry, if there is one.
1697: * To clear an existing entry, call with (*r, *name, NULL, 0).
1698: */
1.94 kristaps 1699: static void
1.106 kristaps 1700: roff_setstr(struct roff *r, const char *name, const char *string,
1701: int multiline)
1.92 schwarze 1702: {
1.164 kristaps 1703:
1704: roff_setstrn(&r->strtab, name, strlen(name), string,
1705: string ? strlen(string) : 0, multiline);
1706: }
1707:
1708: static void
1.166 kristaps 1709: roff_setstrn(struct roffkv **r, const char *name, size_t namesz,
1.164 kristaps 1710: const char *string, size_t stringsz, int multiline)
1711: {
1.166 kristaps 1712: struct roffkv *n;
1.164 kristaps 1713: char *c;
1714: int i;
1715: size_t oldch, newch;
1.92 schwarze 1716:
1.106 kristaps 1717: /* Search for an existing string with the same name. */
1.164 kristaps 1718: n = *r;
1719:
1.166 kristaps 1720: while (n && strcmp(name, n->key.p))
1.92 schwarze 1721: n = n->next;
1.94 kristaps 1722:
1723: if (NULL == n) {
1.106 kristaps 1724: /* Create a new string table entry. */
1.166 kristaps 1725: n = mandoc_malloc(sizeof(struct roffkv));
1726: n->key.p = mandoc_strndup(name, namesz);
1727: n->key.sz = namesz;
1728: n->val.p = NULL;
1729: n->val.sz = 0;
1.164 kristaps 1730: n->next = *r;
1731: *r = n;
1.106 kristaps 1732: } else if (0 == multiline) {
1733: /* In multiline mode, append; else replace. */
1.166 kristaps 1734: free(n->val.p);
1735: n->val.p = NULL;
1736: n->val.sz = 0;
1.106 kristaps 1737: }
1738:
1739: if (NULL == string)
1740: return;
1741:
1742: /*
1743: * One additional byte for the '\n' in multiline mode,
1744: * and one for the terminating '\0'.
1745: */
1.164 kristaps 1746: newch = stringsz + (multiline ? 2u : 1u);
1747:
1.166 kristaps 1748: if (NULL == n->val.p) {
1749: n->val.p = mandoc_malloc(newch);
1750: *n->val.p = '\0';
1.106 kristaps 1751: oldch = 0;
1752: } else {
1.166 kristaps 1753: oldch = n->val.sz;
1754: n->val.p = mandoc_realloc(n->val.p, oldch + newch);
1.106 kristaps 1755: }
1756:
1757: /* Skip existing content in the destination buffer. */
1.166 kristaps 1758: c = n->val.p + (int)oldch;
1.106 kristaps 1759:
1760: /* Append new content to the destination buffer. */
1.164 kristaps 1761: i = 0;
1762: while (i < (int)stringsz) {
1.106 kristaps 1763: /*
1764: * Rudimentary roff copy mode:
1765: * Handle escaped backslashes.
1766: */
1.164 kristaps 1767: if ('\\' == string[i] && '\\' == string[i + 1])
1768: i++;
1769: *c++ = string[i++];
1.106 kristaps 1770: }
1.94 kristaps 1771:
1.106 kristaps 1772: /* Append terminating bytes. */
1773: if (multiline)
1774: *c++ = '\n';
1.163 kristaps 1775:
1.106 kristaps 1776: *c = '\0';
1.166 kristaps 1777: n->val.sz = (int)(c - n->val.p);
1.92 schwarze 1778: }
1779:
1.94 kristaps 1780: static const char *
1781: roff_getstrn(const struct roff *r, const char *name, size_t len)
1.92 schwarze 1782: {
1.166 kristaps 1783: const struct roffkv *n;
1.92 schwarze 1784:
1.164 kristaps 1785: for (n = r->strtab; n; n = n->next)
1.166 kristaps 1786: if (0 == strncmp(name, n->key.p, len) &&
1787: '\0' == n->key.p[(int)len])
1788: return(n->val.p);
1.94 kristaps 1789:
1.157 kristaps 1790: return(NULL);
1.92 schwarze 1791: }
1792:
1.94 kristaps 1793: static void
1.167 kristaps 1794: roff_freestr(struct roffkv *r)
1.92 schwarze 1795: {
1.166 kristaps 1796: struct roffkv *n, *nn;
1.92 schwarze 1797:
1.167 kristaps 1798: for (n = r; n; n = nn) {
1.166 kristaps 1799: free(n->key.p);
1800: free(n->val.p);
1.92 schwarze 1801: nn = n->next;
1802: free(n);
1803: }
1.114 kristaps 1804: }
1805:
1806: const struct tbl_span *
1807: roff_span(const struct roff *r)
1808: {
1809:
1810: return(r->tbl ? tbl_span(r->tbl) : NULL);
1.125 kristaps 1811: }
1812:
1813: const struct eqn *
1814: roff_eqn(const struct roff *r)
1815: {
1816:
1817: return(r->last_eqn ? &r->last_eqn->eqn : NULL);
1.164 kristaps 1818: }
1819:
1820: /*
1821: * Duplicate an input string, making the appropriate character
1822: * conversations (as stipulated by `tr') along the way.
1823: * Returns a heap-allocated string with all the replacements made.
1824: */
1825: char *
1826: roff_strdup(const struct roff *r, const char *p)
1827: {
1.166 kristaps 1828: const struct roffkv *cp;
1.164 kristaps 1829: char *res;
1830: const char *pp;
1831: size_t ssz, sz;
1832: enum mandoc_esc esc;
1833:
1.167 kristaps 1834: if (NULL == r->xmbtab && NULL == r->xtab)
1.164 kristaps 1835: return(mandoc_strdup(p));
1836: else if ('\0' == *p)
1837: return(mandoc_strdup(""));
1838:
1839: /*
1840: * Step through each character looking for term matches
1841: * (remember that a `tr' can be invoked with an escape, which is
1842: * a glyph but the escape is multi-character).
1843: * We only do this if the character hash has been initialised
1844: * and the string is >0 length.
1845: */
1846:
1847: res = NULL;
1848: ssz = 0;
1849:
1850: while ('\0' != *p) {
1.167 kristaps 1851: if ('\\' != *p && r->xtab && r->xtab[(int)*p].p) {
1852: sz = r->xtab[(int)*p].sz;
1853: res = mandoc_realloc(res, ssz + sz + 1);
1854: memcpy(res + ssz, r->xtab[(int)*p].p, sz);
1855: ssz += sz;
1856: p++;
1857: continue;
1858: } else if ('\\' != *p) {
1859: res = mandoc_realloc(res, ssz + 2);
1860: res[ssz++] = *p++;
1861: continue;
1862: }
1863:
1.164 kristaps 1864: /* Search for term matches. */
1.167 kristaps 1865: for (cp = r->xmbtab; cp; cp = cp->next)
1.166 kristaps 1866: if (0 == strncmp(p, cp->key.p, cp->key.sz))
1.164 kristaps 1867: break;
1868:
1869: if (NULL != cp) {
1870: /*
1871: * A match has been found.
1872: * Append the match to the array and move
1873: * forward by its keysize.
1874: */
1.166 kristaps 1875: res = mandoc_realloc
1876: (res, ssz + cp->val.sz + 1);
1877: memcpy(res + ssz, cp->val.p, cp->val.sz);
1878: ssz += cp->val.sz;
1879: p += (int)cp->key.sz;
1.164 kristaps 1880: continue;
1881: }
1882:
1.167 kristaps 1883: /*
1884: * Handle escapes carefully: we need to copy
1885: * over just the escape itself, or else we might
1886: * do replacements within the escape itself.
1887: * Make sure to pass along the bogus string.
1888: */
1889: pp = p++;
1890: esc = mandoc_escape(&p, NULL, NULL);
1891: if (ESCAPE_ERROR == esc) {
1892: sz = strlen(pp);
1.164 kristaps 1893: res = mandoc_realloc(res, ssz + sz + 1);
1894: memcpy(res + ssz, pp, sz);
1.167 kristaps 1895: break;
1.164 kristaps 1896: }
1.167 kristaps 1897: /*
1898: * We bail out on bad escapes.
1899: * No need to warn: we already did so when
1900: * roff_res() was called.
1901: */
1902: sz = (int)(p - pp);
1903: res = mandoc_realloc(res, ssz + sz + 1);
1904: memcpy(res + ssz, pp, sz);
1905: ssz += sz;
1.164 kristaps 1906: }
1907:
1908: res[(int)ssz] = '\0';
1909: return(res);
1.174 kristaps 1910: }
1911:
1912: /*
1913: * Find out whether a line is a macro line or not.
1914: * If it is, adjust the current position and return one; if it isn't,
1915: * return zero and don't change the current position.
1916: * If the control character has been set with `.cc', then let that grain
1917: * precedence.
1918: * This is slighly contrary to groff, where using the non-breaking
1919: * control character when `cc' has been invoked will cause the
1920: * non-breaking macro contents to be printed verbatim.
1921: */
1922: int
1923: roff_getcontrol(const struct roff *r, const char *cp, int *ppos)
1924: {
1925: int pos;
1926:
1927: pos = *ppos;
1928:
1929: if (0 != r->control && cp[pos] == r->control)
1930: pos++;
1931: else if (0 != r->control)
1932: return(0);
1933: else if ('\\' == cp[pos] && '.' == cp[pos + 1])
1934: pos += 2;
1935: else if ('.' == cp[pos] || '\'' == cp[pos])
1936: pos++;
1937: else
1938: return(0);
1939:
1940: while (' ' == cp[pos] || '\t' == cp[pos])
1941: pos++;
1942:
1943: *ppos = pos;
1944: return(1);
1.74 kristaps 1945: }
CVSweb