Annotation of src/usr.bin/mandoc/roff.c, Revision 1.58
1.58 ! schwarze 1: /* $Id: roff.c,v 1.57 2013/10/05 21:17:29 schwarze Exp $ */
1.1 schwarze 2: /*
1.48 schwarze 3: * Copyright (c) 2010, 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
1.50 schwarze 4: * Copyright (c) 2010, 2011, 2012, 2013 Ingo Schwarze <schwarze@openbsd.org>
1.1 schwarze 5: *
6: * Permission to use, copy, modify, and distribute this software for any
7: * purpose with or without fee is hereby granted, provided that the above
8: * copyright notice and this permission notice appear in all copies.
9: *
1.16 schwarze 10: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.1 schwarze 11: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.16 schwarze 12: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.1 schwarze 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.
17: */
18: #include <assert.h>
1.3 schwarze 19: #include <ctype.h>
1.51 schwarze 20: #include <stdio.h>
1.1 schwarze 21: #include <stdlib.h>
22: #include <string.h>
23:
24: #include "mandoc.h"
1.27 schwarze 25: #include "libroff.h"
1.8 schwarze 26: #include "libmandoc.h"
1.1 schwarze 27:
1.37 schwarze 28: /* Maximum number of nested if-else conditionals. */
1.2 schwarze 29: #define RSTACK_MAX 128
30:
1.43 schwarze 31: /* Maximum number of string expansions per line, to break infinite loops. */
32: #define EXPAND_LIMIT 1000
33:
1.1 schwarze 34: enum rofft {
1.20 schwarze 35: ROFF_ad,
1.2 schwarze 36: ROFF_am,
37: ROFF_ami,
38: ROFF_am1,
1.48 schwarze 39: ROFF_cc,
1.1 schwarze 40: ROFF_de,
41: ROFF_dei,
1.2 schwarze 42: ROFF_de1,
43: ROFF_ds,
44: ROFF_el,
1.58 ! schwarze 45: ROFF_fam,
1.20 schwarze 46: ROFF_hy,
1.2 schwarze 47: ROFF_ie,
48: ROFF_if,
1.1 schwarze 49: ROFF_ig,
1.30 schwarze 50: ROFF_it,
1.20 schwarze 51: ROFF_ne,
52: ROFF_nh,
1.14 schwarze 53: ROFF_nr,
1.31 schwarze 54: ROFF_ns,
55: ROFF_ps,
1.2 schwarze 56: ROFF_rm,
1.14 schwarze 57: ROFF_so,
1.31 schwarze 58: ROFF_ta,
1.2 schwarze 59: ROFF_tr,
1.47 schwarze 60: ROFF_Dd,
61: ROFF_TH,
1.27 schwarze 62: ROFF_TS,
63: ROFF_TE,
64: ROFF_T_,
1.32 schwarze 65: ROFF_EQ,
66: ROFF_EN,
1.2 schwarze 67: ROFF_cblock,
1.37 schwarze 68: ROFF_ccond,
1.16 schwarze 69: ROFF_USERDEF,
1.1 schwarze 70: ROFF_MAX
71: };
72:
1.2 schwarze 73: enum roffrule {
1.55 schwarze 74: ROFFRULE_DENY,
75: ROFFRULE_ALLOW
1.2 schwarze 76: };
77:
1.41 schwarze 78: /*
1.42 schwarze 79: * An incredibly-simple string buffer.
80: */
1.8 schwarze 81: struct roffstr {
1.42 schwarze 82: char *p; /* nil-terminated buffer */
83: size_t sz; /* saved strlen(p) */
84: };
85:
86: /*
87: * A key-value roffstr pair as part of a singly-linked list.
88: */
89: struct roffkv {
90: struct roffstr key;
91: struct roffstr val;
92: struct roffkv *next; /* next in list */
1.8 schwarze 93: };
94:
1.52 schwarze 95: /*
96: * A single number register as part of a singly-linked list.
97: */
98: struct roffreg {
99: struct roffstr key;
1.53 schwarze 100: int val;
1.52 schwarze 101: struct roffreg *next;
102: };
103:
1.1 schwarze 104: struct roff {
1.47 schwarze 105: enum mparset parsetype; /* requested parse type */
1.35 schwarze 106: struct mparse *parse; /* parse point */
1.1 schwarze 107: struct roffnode *last; /* leaf of stack */
1.2 schwarze 108: enum roffrule rstack[RSTACK_MAX]; /* stack of !`ie' rules */
1.48 schwarze 109: char control; /* control character */
1.2 schwarze 110: int rstackpos; /* position in rstack */
1.52 schwarze 111: struct roffreg *regtab; /* number registers */
1.42 schwarze 112: struct roffkv *strtab; /* user-defined strings & macros */
113: struct roffkv *xmbtab; /* multi-byte trans table (`tr') */
114: struct roffstr *xtab; /* single-byte trans table (`tr') */
1.16 schwarze 115: const char *current_string; /* value of last called user macro */
1.27 schwarze 116: struct tbl_node *first_tbl; /* first table parsed */
117: struct tbl_node *last_tbl; /* last table parsed */
118: struct tbl_node *tbl; /* current table being parsed */
1.32 schwarze 119: struct eqn_node *last_eqn; /* last equation parsed */
120: struct eqn_node *first_eqn; /* first equation parsed */
121: struct eqn_node *eqn; /* current equation being parsed */
1.1 schwarze 122: };
123:
124: struct roffnode {
125: enum rofft tok; /* type of node */
126: struct roffnode *parent; /* up one in stack */
127: int line; /* parse line */
128: int col; /* parse col */
1.16 schwarze 129: char *name; /* node name, e.g. macro name */
1.2 schwarze 130: char *end; /* end-rules: custom token */
131: int endspan; /* end-rules: next-line or infty */
132: enum roffrule rule; /* current evaluation rule */
1.1 schwarze 133: };
134:
135: #define ROFF_ARGS struct roff *r, /* parse ctx */ \
136: enum rofft tok, /* tok of macro */ \
137: char **bufp, /* input buffer */ \
138: size_t *szp, /* size of input buffer */ \
139: int ln, /* parse line */ \
1.2 schwarze 140: int ppos, /* original pos in buffer */ \
141: int pos, /* current pos in buffer */ \
142: int *offs /* reset offset of buffer data */
1.1 schwarze 143:
144: typedef enum rofferr (*roffproc)(ROFF_ARGS);
145:
146: struct roffmac {
147: const char *name; /* macro name */
1.2 schwarze 148: roffproc proc; /* process new macro */
149: roffproc text; /* process as child text of macro */
150: roffproc sub; /* process as child of macro */
151: int flags;
152: #define ROFFMAC_STRUCT (1 << 0) /* always interpret */
1.3 schwarze 153: struct roffmac *next;
1.1 schwarze 154: };
155:
1.37 schwarze 156: struct predef {
157: const char *name; /* predefined input name */
158: const char *str; /* replacement symbol */
159: };
160:
161: #define PREDEF(__name, __str) \
162: { (__name), (__str) },
163:
1.42 schwarze 164: static enum rofft roffhash_find(const char *, size_t);
165: static void roffhash_init(void);
166: static void roffnode_cleanscope(struct roff *);
167: static void roffnode_pop(struct roff *);
168: static void roffnode_push(struct roff *, enum rofft,
169: const char *, int, int);
1.2 schwarze 170: static enum rofferr roff_block(ROFF_ARGS);
171: static enum rofferr roff_block_text(ROFF_ARGS);
172: static enum rofferr roff_block_sub(ROFF_ARGS);
173: static enum rofferr roff_cblock(ROFF_ARGS);
1.48 schwarze 174: static enum rofferr roff_cc(ROFF_ARGS);
1.2 schwarze 175: static enum rofferr roff_ccond(ROFF_ARGS);
176: static enum rofferr roff_cond(ROFF_ARGS);
177: static enum rofferr roff_cond_text(ROFF_ARGS);
178: static enum rofferr roff_cond_sub(ROFF_ARGS);
1.7 schwarze 179: static enum rofferr roff_ds(ROFF_ARGS);
1.8 schwarze 180: static enum roffrule roff_evalcond(const char *, int *);
1.42 schwarze 181: static void roff_free1(struct roff *);
1.52 schwarze 182: static void roff_freereg(struct roffreg *);
1.42 schwarze 183: static void roff_freestr(struct roffkv *);
1.28 schwarze 184: static char *roff_getname(struct roff *, char **, int, int);
1.56 schwarze 185: static int roff_getnum(const char *, int *, int *);
186: static int roff_getop(const char *, int *, char *);
1.53 schwarze 187: static int roff_getregn(const struct roff *,
188: const char *, size_t);
1.8 schwarze 189: static const char *roff_getstrn(const struct roff *,
190: const char *, size_t);
1.51 schwarze 191: static enum rofferr roff_it(ROFF_ARGS);
1.21 schwarze 192: static enum rofferr roff_line_ignore(ROFF_ARGS);
1.6 schwarze 193: static enum rofferr roff_nr(ROFF_ARGS);
1.41 schwarze 194: static void roff_openeqn(struct roff *, const char *,
195: int, int, const char *);
1.42 schwarze 196: static enum rofft roff_parse(struct roff *, const char *, int *);
1.51 schwarze 197: static enum rofferr roff_parsetext(char **, size_t *, int, int *);
1.45 schwarze 198: static enum rofferr roff_res(struct roff *,
1.37 schwarze 199: char **, size_t *, int, int);
1.29 schwarze 200: static enum rofferr roff_rm(ROFF_ARGS);
1.8 schwarze 201: static void roff_setstr(struct roff *,
1.16 schwarze 202: const char *, const char *, int);
1.42 schwarze 203: static void roff_setstrn(struct roffkv **, const char *,
204: size_t, const char *, size_t, int);
1.14 schwarze 205: static enum rofferr roff_so(ROFF_ARGS);
1.42 schwarze 206: static enum rofferr roff_tr(ROFF_ARGS);
1.47 schwarze 207: static enum rofferr roff_Dd(ROFF_ARGS);
208: static enum rofferr roff_TH(ROFF_ARGS);
1.27 schwarze 209: static enum rofferr roff_TE(ROFF_ARGS);
210: static enum rofferr roff_TS(ROFF_ARGS);
1.32 schwarze 211: static enum rofferr roff_EQ(ROFF_ARGS);
212: static enum rofferr roff_EN(ROFF_ARGS);
1.27 schwarze 213: static enum rofferr roff_T_(ROFF_ARGS);
1.16 schwarze 214: static enum rofferr roff_userdef(ROFF_ARGS);
1.1 schwarze 215:
1.42 schwarze 216: /* See roffhash_find() */
1.3 schwarze 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] = {
1.21 schwarze 225: { "ad", roff_line_ignore, NULL, NULL, 0, NULL },
1.3 schwarze 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 },
1.48 schwarze 229: { "cc", roff_cc, NULL, NULL, 0, NULL },
1.3 schwarze 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 },
1.7 schwarze 233: { "ds", roff_ds, NULL, NULL, 0, NULL },
1.3 schwarze 234: { "el", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
1.58 ! schwarze 235: { "fam", roff_line_ignore, NULL, NULL, 0, NULL },
1.21 schwarze 236: { "hy", roff_line_ignore, NULL, NULL, 0, NULL },
1.3 schwarze 237: { "ie", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
238: { "if", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
239: { "ig", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.51 schwarze 240: { "it", roff_it, NULL, NULL, 0, NULL },
1.21 schwarze 241: { "ne", roff_line_ignore, NULL, NULL, 0, NULL },
242: { "nh", roff_line_ignore, NULL, NULL, 0, NULL },
1.14 schwarze 243: { "nr", roff_nr, NULL, NULL, 0, NULL },
1.31 schwarze 244: { "ns", roff_line_ignore, NULL, NULL, 0, NULL },
245: { "ps", roff_line_ignore, NULL, NULL, 0, NULL },
1.29 schwarze 246: { "rm", roff_rm, NULL, NULL, 0, NULL },
1.14 schwarze 247: { "so", roff_so, NULL, NULL, 0, NULL },
1.31 schwarze 248: { "ta", roff_line_ignore, NULL, NULL, 0, NULL },
1.42 schwarze 249: { "tr", roff_tr, NULL, NULL, 0, NULL },
1.47 schwarze 250: { "Dd", roff_Dd, NULL, NULL, 0, NULL },
251: { "TH", roff_TH, NULL, NULL, 0, NULL },
1.27 schwarze 252: { "TS", roff_TS, NULL, NULL, 0, NULL },
253: { "TE", roff_TE, NULL, NULL, 0, NULL },
254: { "T&", roff_T_, NULL, NULL, 0, NULL },
1.32 schwarze 255: { "EQ", roff_EQ, NULL, NULL, 0, NULL },
256: { "EN", roff_EN, NULL, NULL, 0, NULL },
1.3 schwarze 257: { ".", roff_cblock, NULL, NULL, 0, NULL },
258: { "\\}", roff_ccond, NULL, NULL, 0, NULL },
1.16 schwarze 259: { NULL, roff_userdef, NULL, NULL, 0, NULL },
1.1 schwarze 260: };
261:
1.47 schwarze 262: const char *const __mdoc_reserved[] = {
263: "Ac", "Ad", "An", "Ao", "Ap", "Aq", "Ar", "At",
264: "Bc", "Bd", "Bf", "Bk", "Bl", "Bo", "Bq",
265: "Brc", "Bro", "Brq", "Bsx", "Bt", "Bx",
266: "Cd", "Cm", "Db", "Dc", "Dd", "Dl", "Do", "Dq",
267: "Ds", "Dt", "Dv", "Dx", "D1",
268: "Ec", "Ed", "Ef", "Ek", "El", "Em", "em",
269: "En", "Eo", "Eq", "Er", "Es", "Ev", "Ex",
270: "Fa", "Fc", "Fd", "Fl", "Fn", "Fo", "Fr", "Ft", "Fx",
271: "Hf", "Ic", "In", "It", "Lb", "Li", "Lk", "Lp", "LP",
272: "Me", "Ms", "Mt", "Nd", "Nm", "No", "Ns", "Nx",
273: "Oc", "Oo", "Op", "Os", "Ot", "Ox",
274: "Pa", "Pc", "Pf", "Po", "Pp", "PP", "pp", "Pq",
275: "Qc", "Ql", "Qo", "Qq", "Or", "Rd", "Re", "Rs", "Rv",
276: "Sc", "Sf", "Sh", "SH", "Sm", "So", "Sq",
277: "Ss", "St", "Sx", "Sy",
278: "Ta", "Tn", "Ud", "Ux", "Va", "Vt", "Xc", "Xo", "Xr",
279: "%A", "%B", "%D", "%I", "%J", "%N", "%O",
280: "%P", "%Q", "%R", "%T", "%U", "%V",
281: NULL
282: };
283:
284: const char *const __man_reserved[] = {
285: "AT", "B", "BI", "BR", "BT", "DE", "DS", "DT",
286: "EE", "EN", "EQ", "EX", "HF", "HP", "I", "IB", "IP", "IR",
287: "LP", "ME", "MT", "OP", "P", "PD", "PP", "PT",
288: "R", "RB", "RE", "RI", "RS", "SB", "SH", "SM", "SS", "SY",
289: "TE", "TH", "TP", "TQ", "TS", "T&", "UC", "UE", "UR", "YS",
290: NULL
291: };
292:
1.37 schwarze 293: /* Array of injected predefined strings. */
294: #define PREDEFS_MAX 38
295: static const struct predef predefs[PREDEFS_MAX] = {
296: #include "predefs.in"
297: };
298:
1.42 schwarze 299: /* See roffhash_find() */
1.3 schwarze 300: #define ROFF_HASH(p) (p[0] - ASCII_LO)
301:
1.51 schwarze 302: static int roffit_lines; /* number of lines to delay */
303: static char *roffit_macro; /* nil-terminated macro line */
304:
1.3 schwarze 305: static void
1.42 schwarze 306: roffhash_init(void)
1.3 schwarze 307: {
308: struct roffmac *n;
309: int buc, i;
310:
1.16 schwarze 311: for (i = 0; i < (int)ROFF_USERDEF; i++) {
1.3 schwarze 312: assert(roffs[i].name[0] >= ASCII_LO);
313: assert(roffs[i].name[0] <= ASCII_HI);
314:
315: buc = ROFF_HASH(roffs[i].name);
316:
317: if (NULL != (n = hash[buc])) {
318: for ( ; n->next; n = n->next)
319: /* Do nothing. */ ;
320: n->next = &roffs[i];
321: } else
322: hash[buc] = &roffs[i];
323: }
324: }
325:
1.1 schwarze 326: /*
327: * Look up a roff token by its name. Returns ROFF_MAX if no macro by
328: * the nil-terminated string name could be found.
329: */
330: static enum rofft
1.42 schwarze 331: roffhash_find(const char *p, size_t s)
1.1 schwarze 332: {
1.3 schwarze 333: int buc;
334: struct roffmac *n;
1.1 schwarze 335:
1.3 schwarze 336: /*
337: * libroff has an extremely simple hashtable, for the time
338: * being, which simply keys on the first character, which must
339: * be printable, then walks a chain. It works well enough until
340: * optimised.
341: */
342:
343: if (p[0] < ASCII_LO || p[0] > ASCII_HI)
344: return(ROFF_MAX);
345:
346: buc = ROFF_HASH(p);
347:
348: if (NULL == (n = hash[buc]))
349: return(ROFF_MAX);
350: for ( ; n; n = n->next)
1.16 schwarze 351: if (0 == strncmp(n->name, p, s) && '\0' == n->name[(int)s])
1.3 schwarze 352: return((enum rofft)(n - roffs));
1.1 schwarze 353:
354: return(ROFF_MAX);
355: }
356:
357:
358: /*
359: * Pop the current node off of the stack of roff instructions currently
360: * pending.
361: */
362: static void
363: roffnode_pop(struct roff *r)
364: {
365: struct roffnode *p;
366:
1.2 schwarze 367: assert(r->last);
368: p = r->last;
369:
370: r->last = r->last->parent;
1.16 schwarze 371: free(p->name);
372: free(p->end);
1.1 schwarze 373: free(p);
374: }
375:
376:
377: /*
378: * Push a roff node onto the instruction stack. This must later be
379: * removed with roffnode_pop().
380: */
1.11 schwarze 381: static void
1.16 schwarze 382: roffnode_push(struct roff *r, enum rofft tok, const char *name,
383: int line, int col)
1.1 schwarze 384: {
385: struct roffnode *p;
386:
1.11 schwarze 387: p = mandoc_calloc(1, sizeof(struct roffnode));
1.1 schwarze 388: p->tok = tok;
1.16 schwarze 389: if (name)
390: p->name = mandoc_strdup(name);
1.1 schwarze 391: p->parent = r->last;
392: p->line = line;
393: p->col = col;
1.2 schwarze 394: p->rule = p->parent ? p->parent->rule : ROFFRULE_DENY;
1.1 schwarze 395:
396: r->last = p;
397: }
398:
399:
400: static void
401: roff_free1(struct roff *r)
402: {
1.49 schwarze 403: struct tbl_node *tbl;
1.32 schwarze 404: struct eqn_node *e;
1.42 schwarze 405: int i;
1.27 schwarze 406:
1.49 schwarze 407: while (NULL != (tbl = r->first_tbl)) {
408: r->first_tbl = tbl->next;
409: tbl_free(tbl);
1.27 schwarze 410: }
411:
412: r->first_tbl = r->last_tbl = r->tbl = NULL;
1.1 schwarze 413:
1.32 schwarze 414: while (NULL != (e = r->first_eqn)) {
415: r->first_eqn = e->next;
416: eqn_free(e);
417: }
418:
419: r->first_eqn = r->last_eqn = r->eqn = NULL;
420:
1.1 schwarze 421: while (r->last)
422: roffnode_pop(r);
1.27 schwarze 423:
1.42 schwarze 424: roff_freestr(r->strtab);
425: roff_freestr(r->xmbtab);
426:
427: r->strtab = r->xmbtab = NULL;
428:
1.52 schwarze 429: roff_freereg(r->regtab);
430:
431: r->regtab = NULL;
432:
1.42 schwarze 433: if (r->xtab)
434: for (i = 0; i < 128; i++)
435: free(r->xtab[i].p);
436:
437: free(r->xtab);
438: r->xtab = NULL;
1.1 schwarze 439: }
440:
441: void
442: roff_reset(struct roff *r)
443: {
1.38 schwarze 444: int i;
1.1 schwarze 445:
446: roff_free1(r);
1.38 schwarze 447:
1.48 schwarze 448: r->control = 0;
1.41 schwarze 449:
1.38 schwarze 450: for (i = 0; i < PREDEFS_MAX; i++)
451: roff_setstr(r, predefs[i].name, predefs[i].str, 0);
1.1 schwarze 452: }
453:
454:
455: void
456: roff_free(struct roff *r)
457: {
458:
459: roff_free1(r);
460: free(r);
461: }
462:
463:
464: struct roff *
1.47 schwarze 465: roff_alloc(enum mparset type, struct mparse *parse)
1.1 schwarze 466: {
467: struct roff *r;
1.37 schwarze 468: int i;
1.1 schwarze 469:
1.11 schwarze 470: r = mandoc_calloc(1, sizeof(struct roff));
1.47 schwarze 471: r->parsetype = type;
1.35 schwarze 472: r->parse = parse;
1.2 schwarze 473: r->rstackpos = -1;
1.3 schwarze 474:
1.42 schwarze 475: roffhash_init();
1.37 schwarze 476:
477: for (i = 0; i < PREDEFS_MAX; i++)
478: roff_setstr(r, predefs[i].name, predefs[i].str, 0);
479:
1.1 schwarze 480: return(r);
481: }
482:
1.8 schwarze 483: /*
1.53 schwarze 484: * In the current line, expand user-defined strings ("\*")
485: * and references to number registers ("\n").
486: * Also check the syntax of other escape sequences.
1.8 schwarze 487: */
1.45 schwarze 488: static enum rofferr
1.37 schwarze 489: roff_res(struct roff *r, char **bufp, size_t *szp, int ln, int pos)
1.8 schwarze 490: {
1.53 schwarze 491: char ubuf[12]; /* buffer to print the number */
1.23 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 */
1.53 schwarze 496: char *nbuf; /* new buffer to copy bufp to */
497: size_t nsz; /* size of the new buffer */
498: size_t maxl; /* expected length of the escape name */
499: size_t naml; /* actual length of the escape name */
500: int expand_count; /* to avoid infinite loops */
1.8 schwarze 501:
1.43 schwarze 502: expand_count = 0;
503:
1.42 schwarze 504: again:
1.24 schwarze 505: cp = *bufp + pos;
506: while (NULL != (cp = strchr(cp, '\\'))) {
507: stesc = cp++;
1.23 schwarze 508:
509: /*
1.53 schwarze 510: * The second character must be an asterisk or an n.
1.23 schwarze 511: * If it isn't, skip it anyway: It is escaped,
512: * so it can't start another escape sequence.
513: */
514:
1.24 schwarze 515: if ('\0' == *cp)
1.45 schwarze 516: return(ROFF_CONT);
1.42 schwarze 517:
1.53 schwarze 518: switch (*cp) {
519: case ('*'):
520: res = NULL;
521: break;
522: case ('n'):
523: res = ubuf;
524: break;
525: default:
526: if (ESCAPE_ERROR != mandoc_escape(&cp, NULL, NULL))
1.42 schwarze 527: continue;
528: mandoc_msg
529: (MANDOCERR_BADESCAPE, r->parse,
530: ln, (int)(stesc - *bufp), NULL);
1.45 schwarze 531: return(ROFF_CONT);
1.42 schwarze 532: }
533:
534: cp++;
1.23 schwarze 535:
536: /*
537: * The third character decides the length
1.53 schwarze 538: * of the name of the string or register.
1.23 schwarze 539: * Save a pointer to the name.
540: */
541:
1.24 schwarze 542: switch (*cp) {
543: case ('\0'):
1.45 schwarze 544: return(ROFF_CONT);
1.8 schwarze 545: case ('('):
546: cp++;
547: maxl = 2;
548: break;
549: case ('['):
550: cp++;
551: maxl = 0;
552: break;
553: default:
554: maxl = 1;
555: break;
556: }
1.23 schwarze 557: stnam = cp;
1.8 schwarze 558:
1.23 schwarze 559: /* Advance to the end of the name. */
1.8 schwarze 560:
1.53 schwarze 561: for (naml = 0; 0 == maxl || naml < maxl; naml++, cp++) {
1.42 schwarze 562: if ('\0' == *cp) {
563: mandoc_msg
564: (MANDOCERR_BADESCAPE,
565: r->parse, ln,
566: (int)(stesc - *bufp), NULL);
1.45 schwarze 567: return(ROFF_CONT);
1.42 schwarze 568: }
1.8 schwarze 569: if (0 == maxl && ']' == *cp)
570: break;
571: }
572:
1.23 schwarze 573: /*
574: * Retrieve the replacement string; if it is
575: * undefined, resume searching for escapes.
576: */
577:
1.53 schwarze 578: if (NULL == res)
579: res = roff_getstrn(r, stnam, naml);
580: else
581: snprintf(ubuf, sizeof(ubuf), "%d",
582: roff_getregn(r, stnam, naml));
1.8 schwarze 583:
584: if (NULL == res) {
1.42 schwarze 585: mandoc_msg
586: (MANDOCERR_BADESCAPE, r->parse,
587: ln, (int)(stesc - *bufp), NULL);
1.37 schwarze 588: res = "";
1.8 schwarze 589: }
590:
1.23 schwarze 591: /* Replace the escape sequence by the string. */
592:
1.42 schwarze 593: pos = stesc - *bufp;
594:
1.8 schwarze 595: nsz = *szp + strlen(res) + 1;
1.53 schwarze 596: nbuf = mandoc_malloc(nsz);
1.8 schwarze 597:
1.53 schwarze 598: strlcpy(nbuf, *bufp, (size_t)(stesc - *bufp + 1));
599: strlcat(nbuf, res, nsz);
600: strlcat(nbuf, cp + (maxl ? 0 : 1), nsz);
1.8 schwarze 601:
602: free(*bufp);
603:
1.53 schwarze 604: *bufp = nbuf;
1.8 schwarze 605: *szp = nsz;
1.43 schwarze 606:
607: if (EXPAND_LIMIT >= ++expand_count)
608: goto again;
609:
610: /* Just leave the string unexpanded. */
611: mandoc_msg(MANDOCERR_ROFFLOOP, r->parse, ln, pos, NULL);
1.45 schwarze 612: return(ROFF_IGN);
1.42 schwarze 613: }
1.45 schwarze 614: return(ROFF_CONT);
1.42 schwarze 615: }
616:
617: /*
1.51 schwarze 618: * Process text streams:
619: * Convert all breakable hyphens into ASCII_HYPH.
620: * Decrement and spring input line trap.
1.42 schwarze 621: */
622: static enum rofferr
1.51 schwarze 623: roff_parsetext(char **bufp, size_t *szp, int pos, int *offs)
1.42 schwarze 624: {
625: size_t sz;
626: const char *start;
1.51 schwarze 627: char *p;
628: int isz;
1.42 schwarze 629: enum mandoc_esc esc;
630:
1.51 schwarze 631: start = p = *bufp + pos;
1.42 schwarze 632:
633: while ('\0' != *p) {
634: sz = strcspn(p, "-\\");
635: p += sz;
636:
637: if ('\0' == *p)
638: break;
639:
640: if ('\\' == *p) {
641: /* Skip over escapes. */
642: p++;
643: esc = mandoc_escape
1.57 schwarze 644: ((const char const **)&p, NULL, NULL);
1.42 schwarze 645: if (ESCAPE_ERROR == esc)
646: break;
647: continue;
648: } else if (p == start) {
649: p++;
650: continue;
651: }
652:
1.44 schwarze 653: if (isalpha((unsigned char)p[-1]) &&
654: isalpha((unsigned char)p[1]))
1.42 schwarze 655: *p = ASCII_HYPH;
656: p++;
1.8 schwarze 657: }
658:
1.51 schwarze 659: /* Spring the input line trap. */
660: if (1 == roffit_lines) {
661: isz = asprintf(&p, "%s\n.%s", *bufp, roffit_macro);
662: if (-1 == isz) {
663: perror(NULL);
664: exit((int)MANDOCLEVEL_SYSERR);
665: }
666: free(*bufp);
667: *bufp = p;
668: *szp = isz + 1;
669: *offs = 0;
670: free(roffit_macro);
671: roffit_lines = 0;
672: return(ROFF_REPARSE);
673: } else if (1 < roffit_lines)
674: --roffit_lines;
1.42 schwarze 675: return(ROFF_CONT);
1.8 schwarze 676: }
677:
1.1 schwarze 678: enum rofferr
1.6 schwarze 679: roff_parseln(struct roff *r, int ln, char **bufp,
680: size_t *szp, int pos, int *offs)
1.1 schwarze 681: {
682: enum rofft t;
1.27 schwarze 683: enum rofferr e;
1.35 schwarze 684: int ppos, ctl;
1.1 schwarze 685:
1.2 schwarze 686: /*
1.8 schwarze 687: * Run the reserved-word filter only if we have some reserved
688: * words to fill in.
689: */
690:
1.45 schwarze 691: e = roff_res(r, bufp, szp, ln, pos);
692: if (ROFF_IGN == e)
693: return(e);
694: assert(ROFF_CONT == e);
1.8 schwarze 695:
1.35 schwarze 696: ppos = pos;
1.48 schwarze 697: ctl = roff_getcontrol(r, *bufp, &pos);
1.35 schwarze 698:
1.8 schwarze 699: /*
1.2 schwarze 700: * First, if a scope is open and we're not a macro, pass the
701: * text through the macro's filter. If a scope isn't open and
702: * we're not a macro, just let it through.
1.32 schwarze 703: * Finally, if there's an equation scope open, divert it into it
704: * no matter our state.
1.2 schwarze 705: */
706:
1.35 schwarze 707: if (r->last && ! ctl) {
1.2 schwarze 708: t = r->last->tok;
709: assert(roffs[t].text);
1.27 schwarze 710: e = (*roffs[t].text)
711: (r, t, bufp, szp, ln, pos, pos, offs);
712: assert(ROFF_IGN == e || ROFF_CONT == e);
1.32 schwarze 713: if (ROFF_CONT != e)
714: return(e);
1.54 schwarze 715: }
716: if (r->eqn)
717: return(eqn_read(&r->eqn, ln, *bufp, ppos, offs));
718: if ( ! ctl) {
1.32 schwarze 719: if (r->tbl)
1.35 schwarze 720: return(tbl_read(r->tbl, ln, *bufp, pos));
1.51 schwarze 721: return(roff_parsetext(bufp, szp, pos, offs));
1.54 schwarze 722: }
1.2 schwarze 723:
724: /*
725: * If a scope is open, go to the child handler for that macro,
726: * as it may want to preprocess before doing anything with it.
1.32 schwarze 727: * Don't do so if an equation is open.
1.2 schwarze 728: */
729:
730: if (r->last) {
1.1 schwarze 731: t = r->last->tok;
732: assert(roffs[t].sub);
1.2 schwarze 733: return((*roffs[t].sub)
1.8 schwarze 734: (r, t, bufp, szp,
1.35 schwarze 735: ln, ppos, pos, offs));
1.2 schwarze 736: }
737:
738: /*
739: * Lastly, as we've no scope open, try to look up and execute
740: * the new macro. If no macro is found, simply return and let
741: * the compilers handle it.
742: */
743:
1.16 schwarze 744: if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos)))
1.1 schwarze 745: return(ROFF_CONT);
746:
1.2 schwarze 747: assert(roffs[t].proc);
748: return((*roffs[t].proc)
1.8 schwarze 749: (r, t, bufp, szp,
750: ln, ppos, pos, offs));
1.2 schwarze 751: }
752:
1.1 schwarze 753:
1.27 schwarze 754: void
1.2 schwarze 755: roff_endparse(struct roff *r)
756: {
1.1 schwarze 757:
1.27 schwarze 758: if (r->last)
1.35 schwarze 759: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.27 schwarze 760: r->last->line, r->last->col, NULL);
761:
1.32 schwarze 762: if (r->eqn) {
1.35 schwarze 763: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.41 schwarze 764: r->eqn->eqn.ln, r->eqn->eqn.pos, NULL);
765: eqn_end(&r->eqn);
1.32 schwarze 766: }
767:
1.27 schwarze 768: if (r->tbl) {
1.35 schwarze 769: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.27 schwarze 770: r->tbl->line, r->tbl->pos, NULL);
1.41 schwarze 771: tbl_end(&r->tbl);
1.27 schwarze 772: }
1.1 schwarze 773: }
774:
775: /*
776: * Parse a roff node's type from the input buffer. This must be in the
777: * form of ".foo xxx" in the usual way.
778: */
779: static enum rofft
1.16 schwarze 780: roff_parse(struct roff *r, const char *buf, int *pos)
1.1 schwarze 781: {
1.16 schwarze 782: const char *mac;
783: size_t maclen;
1.1 schwarze 784: enum rofft t;
785:
1.39 schwarze 786: if ('\0' == buf[*pos] || '"' == buf[*pos] ||
787: '\t' == buf[*pos] || ' ' == buf[*pos])
1.1 schwarze 788: return(ROFF_MAX);
789:
1.39 schwarze 790: /*
791: * We stop the macro parse at an escape, tab, space, or nil.
792: * However, `\}' is also a valid macro, so make sure we don't
793: * clobber it by seeing the `\' as the end of token.
794: */
795:
1.16 schwarze 796: mac = buf + *pos;
1.39 schwarze 797: maclen = strcspn(mac + 1, " \\\t\0") + 1;
1.1 schwarze 798:
1.16 schwarze 799: t = (r->current_string = roff_getstrn(r, mac, maclen))
1.42 schwarze 800: ? ROFF_USERDEF : roffhash_find(mac, maclen);
1.1 schwarze 801:
1.34 schwarze 802: *pos += (int)maclen;
1.35 schwarze 803:
1.1 schwarze 804: while (buf[*pos] && ' ' == buf[*pos])
805: (*pos)++;
806:
807: return(t);
808: }
809:
810: /* ARGSUSED */
811: static enum rofferr
1.2 schwarze 812: roff_cblock(ROFF_ARGS)
1.1 schwarze 813: {
814:
1.2 schwarze 815: /*
816: * A block-close `..' should only be invoked as a child of an
817: * ignore macro, otherwise raise a warning and just ignore it.
818: */
819:
820: if (NULL == r->last) {
1.35 schwarze 821: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.2 schwarze 822: return(ROFF_IGN);
823: }
1.1 schwarze 824:
1.2 schwarze 825: switch (r->last->tok) {
826: case (ROFF_am):
827: /* FALLTHROUGH */
828: case (ROFF_ami):
829: /* FALLTHROUGH */
830: case (ROFF_am1):
831: /* FALLTHROUGH */
832: case (ROFF_de):
1.23 schwarze 833: /* ROFF_de1 is remapped to ROFF_de in roff_block(). */
1.2 schwarze 834: /* FALLTHROUGH */
835: case (ROFF_dei):
836: /* FALLTHROUGH */
837: case (ROFF_ig):
838: break;
839: default:
1.35 schwarze 840: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.1 schwarze 841: return(ROFF_IGN);
1.2 schwarze 842: }
843:
844: if ((*bufp)[pos])
1.35 schwarze 845: mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.2 schwarze 846:
847: roffnode_pop(r);
848: roffnode_cleanscope(r);
849: return(ROFF_IGN);
850:
851: }
1.1 schwarze 852:
853:
1.2 schwarze 854: static void
855: roffnode_cleanscope(struct roff *r)
856: {
1.1 schwarze 857:
1.2 schwarze 858: while (r->last) {
1.46 schwarze 859: if (--r->last->endspan != 0)
1.2 schwarze 860: break;
861: roffnode_pop(r);
862: }
863: }
1.1 schwarze 864:
865:
1.2 schwarze 866: /* ARGSUSED */
867: static enum rofferr
868: roff_ccond(ROFF_ARGS)
869: {
1.1 schwarze 870:
1.2 schwarze 871: if (NULL == r->last) {
1.35 schwarze 872: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.1 schwarze 873: return(ROFF_IGN);
1.2 schwarze 874: }
1.1 schwarze 875:
1.2 schwarze 876: switch (r->last->tok) {
877: case (ROFF_el):
878: /* FALLTHROUGH */
879: case (ROFF_ie):
880: /* FALLTHROUGH */
881: case (ROFF_if):
882: break;
883: default:
1.35 schwarze 884: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.2 schwarze 885: return(ROFF_IGN);
886: }
1.1 schwarze 887:
1.2 schwarze 888: if (r->last->endspan > -1) {
1.35 schwarze 889: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.1 schwarze 890: return(ROFF_IGN);
1.2 schwarze 891: }
892:
893: if ((*bufp)[pos])
1.35 schwarze 894: mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.1 schwarze 895:
1.2 schwarze 896: roffnode_pop(r);
897: roffnode_cleanscope(r);
1.1 schwarze 898: return(ROFF_IGN);
899: }
900:
901:
902: /* ARGSUSED */
903: static enum rofferr
1.2 schwarze 904: roff_block(ROFF_ARGS)
1.1 schwarze 905: {
1.2 schwarze 906: int sv;
907: size_t sz;
1.16 schwarze 908: char *name;
909:
910: name = NULL;
1.2 schwarze 911:
1.16 schwarze 912: if (ROFF_ig != tok) {
913: if ('\0' == (*bufp)[pos]) {
1.35 schwarze 914: mandoc_msg(MANDOCERR_NOARGS, r->parse, ln, ppos, NULL);
1.16 schwarze 915: return(ROFF_IGN);
916: }
1.22 schwarze 917:
918: /*
919: * Re-write `de1', since we don't really care about
920: * groff's strange compatibility mode, into `de'.
921: */
922:
1.18 schwarze 923: if (ROFF_de1 == tok)
924: tok = ROFF_de;
1.16 schwarze 925: if (ROFF_de == tok)
926: name = *bufp + pos;
1.21 schwarze 927: else
1.35 schwarze 928: mandoc_msg(MANDOCERR_REQUEST, r->parse, ln, ppos,
1.21 schwarze 929: roffs[tok].name);
1.22 schwarze 930:
1.33 schwarze 931: while ((*bufp)[pos] && ! isspace((unsigned char)(*bufp)[pos]))
1.2 schwarze 932: pos++;
1.22 schwarze 933:
1.33 schwarze 934: while (isspace((unsigned char)(*bufp)[pos]))
1.16 schwarze 935: (*bufp)[pos++] = '\0';
1.2 schwarze 936: }
937:
1.16 schwarze 938: roffnode_push(r, tok, name, ln, ppos);
939:
940: /*
941: * At the beginning of a `de' macro, clear the existing string
942: * with the same name, if there is one. New content will be
943: * added from roff_block_text() in multiline mode.
944: */
1.22 schwarze 945:
1.16 schwarze 946: if (ROFF_de == tok)
1.19 schwarze 947: roff_setstr(r, name, "", 0);
1.2 schwarze 948:
949: if ('\0' == (*bufp)[pos])
950: return(ROFF_IGN);
1.1 schwarze 951:
1.22 schwarze 952: /* If present, process the custom end-of-line marker. */
953:
1.2 schwarze 954: sv = pos;
1.33 schwarze 955: while ((*bufp)[pos] && ! isspace((unsigned char)(*bufp)[pos]))
1.2 schwarze 956: pos++;
957:
958: /*
959: * Note: groff does NOT like escape characters in the input.
960: * Instead of detecting this, we're just going to let it fly and
961: * to hell with it.
962: */
963:
964: assert(pos > sv);
965: sz = (size_t)(pos - sv);
966:
967: if (1 == sz && '.' == (*bufp)[sv])
968: return(ROFF_IGN);
969:
1.11 schwarze 970: r->last->end = mandoc_malloc(sz + 1);
1.2 schwarze 971:
972: memcpy(r->last->end, *bufp + sv, sz);
973: r->last->end[(int)sz] = '\0';
974:
975: if ((*bufp)[pos])
1.35 schwarze 976: mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.1 schwarze 977:
978: return(ROFF_IGN);
979: }
980:
981:
982: /* ARGSUSED */
983: static enum rofferr
1.2 schwarze 984: roff_block_sub(ROFF_ARGS)
1.1 schwarze 985: {
1.2 schwarze 986: enum rofft t;
987: int i, j;
988:
989: /*
990: * First check whether a custom macro exists at this level. If
991: * it does, then check against it. This is some of groff's
992: * stranger behaviours. If we encountered a custom end-scope
993: * tag and that tag also happens to be a "real" macro, then we
994: * need to try interpreting it again as a real macro. If it's
995: * not, then return ignore. Else continue.
996: */
997:
998: if (r->last->end) {
1.35 schwarze 999: for (i = pos, j = 0; r->last->end[j]; j++, i++)
1.2 schwarze 1000: if ((*bufp)[i] != r->last->end[j])
1001: break;
1.1 schwarze 1002:
1.2 schwarze 1003: if ('\0' == r->last->end[j] &&
1004: ('\0' == (*bufp)[i] ||
1005: ' ' == (*bufp)[i] ||
1006: '\t' == (*bufp)[i])) {
1007: roffnode_pop(r);
1008: roffnode_cleanscope(r);
1.1 schwarze 1009:
1.35 schwarze 1010: while (' ' == (*bufp)[i] || '\t' == (*bufp)[i])
1011: i++;
1012:
1013: pos = i;
1.16 schwarze 1014: if (ROFF_MAX != roff_parse(r, *bufp, &pos))
1.2 schwarze 1015: return(ROFF_RERUN);
1016: return(ROFF_IGN);
1017: }
1.1 schwarze 1018: }
1019:
1.2 schwarze 1020: /*
1021: * If we have no custom end-query or lookup failed, then try
1022: * pulling it out of the hashtable.
1023: */
1.1 schwarze 1024:
1.36 schwarze 1025: t = roff_parse(r, *bufp, &pos);
1.1 schwarze 1026:
1.16 schwarze 1027: /*
1028: * Macros other than block-end are only significant
1029: * in `de' blocks; elsewhere, simply throw them away.
1030: */
1031: if (ROFF_cblock != t) {
1032: if (ROFF_de == tok)
1033: roff_setstr(r, r->last->name, *bufp + ppos, 1);
1.1 schwarze 1034: return(ROFF_IGN);
1.16 schwarze 1035: }
1.1 schwarze 1036:
1.2 schwarze 1037: assert(roffs[t].proc);
1.6 schwarze 1038: return((*roffs[t].proc)(r, t, bufp, szp,
1039: ln, ppos, pos, offs));
1.2 schwarze 1040: }
1041:
1042:
1043: /* ARGSUSED */
1044: static enum rofferr
1045: roff_block_text(ROFF_ARGS)
1046: {
1047:
1.16 schwarze 1048: if (ROFF_de == tok)
1049: roff_setstr(r, r->last->name, *bufp + pos, 1);
1050:
1.2 schwarze 1051: return(ROFF_IGN);
1052: }
1053:
1054:
1055: /* ARGSUSED */
1056: static enum rofferr
1057: roff_cond_sub(ROFF_ARGS)
1058: {
1059: enum rofft t;
1060: enum roffrule rr;
1.37 schwarze 1061: char *ep;
1.2 schwarze 1062:
1063: rr = r->last->rule;
1.37 schwarze 1064: roffnode_cleanscope(r);
1.50 schwarze 1065: t = roff_parse(r, *bufp, &pos);
1.2 schwarze 1066:
1.37 schwarze 1067: /*
1.50 schwarze 1068: * Fully handle known macros when they are structurally
1069: * required or when the conditional evaluated to true.
1.5 schwarze 1070: */
1071:
1.50 schwarze 1072: if ((ROFF_MAX != t) &&
1073: (ROFF_ccond == t || ROFFRULE_ALLOW == rr ||
1074: ROFFMAC_STRUCT & roffs[t].flags)) {
1075: assert(roffs[t].proc);
1076: return((*roffs[t].proc)(r, t, bufp, szp,
1077: ln, ppos, pos, offs));
1078: }
1.39 schwarze 1079:
1.50 schwarze 1080: /* Always check for the closing delimiter `\}'. */
1.39 schwarze 1081:
1.50 schwarze 1082: ep = &(*bufp)[pos];
1083: while (NULL != (ep = strchr(ep, '\\'))) {
1084: if ('}' != *(++ep))
1085: continue;
1.2 schwarze 1086:
1.50 schwarze 1087: /*
1088: * If we're at the end of line, then just chop
1089: * off the \} and resize the buffer.
1090: * If we aren't, then convert it to spaces.
1091: */
1.37 schwarze 1092:
1.50 schwarze 1093: if ('\0' == *(ep + 1)) {
1094: *--ep = '\0';
1095: *szp -= 2;
1096: } else
1097: *(ep - 1) = *ep = ' ';
1.2 schwarze 1098:
1.50 schwarze 1099: roff_ccond(r, ROFF_ccond, bufp, szp,
1100: ln, pos, pos + 2, offs);
1101: break;
1102: }
1103: return(ROFFRULE_DENY == rr ? ROFF_IGN : ROFF_CONT);
1.2 schwarze 1104: }
1105:
1106: /* ARGSUSED */
1107: static enum rofferr
1108: roff_cond_text(ROFF_ARGS)
1109: {
1.37 schwarze 1110: char *ep;
1.2 schwarze 1111: enum roffrule rr;
1112:
1113: rr = r->last->rule;
1.37 schwarze 1114: roffnode_cleanscope(r);
1.1 schwarze 1115:
1.37 schwarze 1116: ep = &(*bufp)[pos];
1117: for ( ; NULL != (ep = strchr(ep, '\\')); ep++) {
1118: ep++;
1119: if ('}' != *ep)
1120: continue;
1121: *ep = '&';
1122: roff_ccond(r, ROFF_ccond, bufp, szp,
1123: ln, pos, pos + 2, offs);
1.2 schwarze 1124: }
1125: return(ROFFRULE_DENY == rr ? ROFF_IGN : ROFF_CONT);
1126: }
1127:
1.56 schwarze 1128: static int
1129: roff_getnum(const char *v, int *pos, int *res)
1130: {
1131: int p, n;
1132:
1133: p = *pos;
1134: n = v[p] == '-';
1135: if (n)
1136: p++;
1137:
1138: for (*res = 0; isdigit((unsigned char)v[p]); p++)
1139: *res += 10 * *res + v[p] - '0';
1140: if (p == *pos + n)
1141: return 0;
1142:
1143: if (n)
1144: *res = -*res;
1145:
1146: *pos = p;
1147: return 1;
1148: }
1149:
1150: static int
1151: roff_getop(const char *v, int *pos, char *res)
1152: {
1153: int e;
1154:
1155: *res = v[*pos];
1156: e = v[*pos + 1] == '=';
1157:
1158: switch (*res) {
1159: case '=':
1160: break;
1161: case '>':
1162: if (e)
1163: *res = 'g';
1164: break;
1165: case '<':
1166: if (e)
1167: *res = 'l';
1168: break;
1169: default:
1170: return(0);
1171: }
1172:
1173: *pos += 1 + e;
1174:
1175: return(*res);
1176: }
1177:
1.5 schwarze 1178: static enum roffrule
1179: roff_evalcond(const char *v, int *pos)
1180: {
1.56 schwarze 1181: int not, lh, rh;
1182: char op;
1.5 schwarze 1183:
1184: switch (v[*pos]) {
1185: case ('n'):
1186: (*pos)++;
1187: return(ROFFRULE_ALLOW);
1188: case ('e'):
1189: /* FALLTHROUGH */
1190: case ('o'):
1191: /* FALLTHROUGH */
1192: case ('t'):
1193: (*pos)++;
1194: return(ROFFRULE_DENY);
1.56 schwarze 1195: case ('!'):
1196: (*pos)++;
1197: not = 1;
1198: break;
1.5 schwarze 1199: default:
1.56 schwarze 1200: not = 0;
1.5 schwarze 1201: break;
1202: }
1203:
1.56 schwarze 1204: if (!roff_getnum(v, pos, &lh))
1205: return ROFFRULE_DENY;
1206: if (!roff_getop(v, pos, &op)) {
1207: if (lh < 0)
1208: lh = 0;
1209: goto out;
1210: }
1211: if (!roff_getnum(v, pos, &rh))
1212: return ROFFRULE_DENY;
1213: switch (op) {
1214: case 'g':
1215: lh = lh >= rh;
1216: break;
1217: case 'l':
1218: lh = lh <= rh;
1219: break;
1220: case '=':
1221: lh = lh == rh;
1222: break;
1223: case '>':
1224: lh = lh > rh;
1225: break;
1226: case '<':
1227: lh = lh < rh;
1228: break;
1229: default:
1230: return ROFFRULE_DENY;
1231: }
1232: out:
1233: if (not)
1234: lh = !lh;
1235: return lh ? ROFFRULE_ALLOW : ROFFRULE_DENY;
1.5 schwarze 1236: }
1237:
1.2 schwarze 1238: /* ARGSUSED */
1239: static enum rofferr
1.21 schwarze 1240: roff_line_ignore(ROFF_ARGS)
1.6 schwarze 1241: {
1.30 schwarze 1242:
1.21 schwarze 1243: return(ROFF_IGN);
1244: }
1245:
1246: /* ARGSUSED */
1247: static enum rofferr
1.2 schwarze 1248: roff_cond(ROFF_ARGS)
1249: {
1.46 schwarze 1250:
1251: roffnode_push(r, tok, NULL, ln, ppos);
1.2 schwarze 1252:
1.35 schwarze 1253: /*
1254: * An `.el' has no conditional body: it will consume the value
1255: * of the current rstack entry set in prior `ie' calls or
1256: * defaults to DENY.
1257: *
1258: * If we're not an `el', however, then evaluate the conditional.
1259: */
1.1 schwarze 1260:
1.46 schwarze 1261: r->last->rule = ROFF_el == tok ?
1.35 schwarze 1262: (r->rstackpos < 0 ?
1263: ROFFRULE_DENY : r->rstack[r->rstackpos--]) :
1264: roff_evalcond(*bufp, &pos);
1.2 schwarze 1265:
1.35 schwarze 1266: /*
1267: * An if-else will put the NEGATION of the current evaluated
1268: * conditional into the stack of rules.
1269: */
1270:
1.2 schwarze 1271: if (ROFF_ie == tok) {
1.35 schwarze 1272: if (r->rstackpos == RSTACK_MAX - 1) {
1273: mandoc_msg(MANDOCERR_MEM,
1274: r->parse, ln, ppos, NULL);
1275: return(ROFF_ERR);
1276: }
1277: r->rstack[++r->rstackpos] =
1278: ROFFRULE_DENY == r->last->rule ?
1279: ROFFRULE_ALLOW : ROFFRULE_DENY;
1.2 schwarze 1280: }
1.5 schwarze 1281:
1282: /* If the parent has false as its rule, then so do we. */
1283:
1.2 schwarze 1284: if (r->last->parent && ROFFRULE_DENY == r->last->parent->rule)
1285: r->last->rule = ROFFRULE_DENY;
1.5 schwarze 1286:
1287: /*
1.46 schwarze 1288: * Determine scope.
1289: * If there is nothing on the line after the conditional,
1290: * not even whitespace, use next-line scope.
1.5 schwarze 1291: */
1.2 schwarze 1292:
1.46 schwarze 1293: if ('\0' == (*bufp)[pos]) {
1294: r->last->endspan = 2;
1295: goto out;
1296: }
1297:
1298: while (' ' == (*bufp)[pos])
1299: pos++;
1300:
1301: /* An opening brace requests multiline scope. */
1.2 schwarze 1302:
1303: if ('\\' == (*bufp)[pos] && '{' == (*bufp)[pos + 1]) {
1304: r->last->endspan = -1;
1305: pos += 2;
1.46 schwarze 1306: goto out;
1.2 schwarze 1307: }
1308:
1309: /*
1.46 schwarze 1310: * Anything else following the conditional causes
1311: * single-line scope. Warn if the scope contains
1312: * nothing but trailing whitespace.
1.2 schwarze 1313: */
1314:
1315: if ('\0' == (*bufp)[pos])
1.46 schwarze 1316: mandoc_msg(MANDOCERR_NOARGS, r->parse, ln, ppos, NULL);
1.2 schwarze 1317:
1.46 schwarze 1318: r->last->endspan = 1;
1.1 schwarze 1319:
1.46 schwarze 1320: out:
1.2 schwarze 1321: *offs = pos;
1322: return(ROFF_RERUN);
1.1 schwarze 1323: }
1324:
1325:
1.2 schwarze 1326: /* ARGSUSED */
1327: static enum rofferr
1.7 schwarze 1328: roff_ds(ROFF_ARGS)
1329: {
1.10 schwarze 1330: char *name, *string;
1331:
1332: /*
1333: * A symbol is named by the first word following the macro
1334: * invocation up to a space. Its value is anything after the
1335: * name's trailing whitespace and optional double-quote. Thus,
1336: *
1337: * [.ds foo "bar " ]
1338: *
1339: * will have `bar " ' as its value.
1340: */
1.7 schwarze 1341:
1.28 schwarze 1342: string = *bufp + pos;
1343: name = roff_getname(r, &string, ln, pos);
1.7 schwarze 1344: if ('\0' == *name)
1345: return(ROFF_IGN);
1346:
1.28 schwarze 1347: /* Read past initial double-quote. */
1348: if ('"' == *string)
1.7 schwarze 1349: string++;
1350:
1.10 schwarze 1351: /* The rest is the value. */
1.16 schwarze 1352: roff_setstr(r, name, string, 0);
1.7 schwarze 1353: return(ROFF_IGN);
1354: }
1355:
1.52 schwarze 1356: void
1.53 schwarze 1357: roff_setreg(struct roff *r, const char *name, int val)
1.41 schwarze 1358: {
1.52 schwarze 1359: struct roffreg *reg;
1360:
1361: /* Search for an existing register with the same name. */
1362: reg = r->regtab;
1363:
1364: while (reg && strcmp(name, reg->key.p))
1365: reg = reg->next;
1.41 schwarze 1366:
1.52 schwarze 1367: if (NULL == reg) {
1368: /* Create a new register. */
1369: reg = mandoc_malloc(sizeof(struct roffreg));
1370: reg->key.p = mandoc_strdup(name);
1371: reg->key.sz = strlen(name);
1372: reg->next = r->regtab;
1373: r->regtab = reg;
1374: }
1375:
1.53 schwarze 1376: reg->val = val;
1.41 schwarze 1377: }
1378:
1.53 schwarze 1379: int
1.52 schwarze 1380: roff_getreg(const struct roff *r, const char *name)
1.41 schwarze 1381: {
1.52 schwarze 1382: struct roffreg *reg;
1383:
1384: for (reg = r->regtab; reg; reg = reg->next)
1385: if (0 == strcmp(name, reg->key.p))
1.53 schwarze 1386: return(reg->val);
1387:
1388: return(0);
1389: }
1390:
1391: static int
1392: roff_getregn(const struct roff *r, const char *name, size_t len)
1393: {
1394: struct roffreg *reg;
1395:
1396: for (reg = r->regtab; reg; reg = reg->next)
1397: if (len == reg->key.sz &&
1398: 0 == strncmp(name, reg->key.p, len))
1399: return(reg->val);
1.41 schwarze 1400:
1.52 schwarze 1401: return(0);
1.41 schwarze 1402: }
1403:
1.52 schwarze 1404: static void
1405: roff_freereg(struct roffreg *reg)
1.41 schwarze 1406: {
1.52 schwarze 1407: struct roffreg *old_reg;
1.41 schwarze 1408:
1.52 schwarze 1409: while (NULL != reg) {
1410: free(reg->key.p);
1411: old_reg = reg;
1412: reg = reg->next;
1413: free(old_reg);
1414: }
1.41 schwarze 1415: }
1.7 schwarze 1416:
1417: /* ARGSUSED */
1418: static enum rofferr
1.6 schwarze 1419: roff_nr(ROFF_ARGS)
1.1 schwarze 1420: {
1.28 schwarze 1421: const char *key;
1422: char *val;
1.37 schwarze 1423: int iv;
1.6 schwarze 1424:
1.28 schwarze 1425: val = *bufp + pos;
1426: key = roff_getname(r, &val, ln, pos);
1.6 schwarze 1427:
1.52 schwarze 1428: iv = mandoc_strntoi(val, strlen(val), 10);
1429:
1.53 schwarze 1430: roff_setreg(r, key, iv);
1.1 schwarze 1431:
1.29 schwarze 1432: return(ROFF_IGN);
1433: }
1434:
1435: /* ARGSUSED */
1436: static enum rofferr
1437: roff_rm(ROFF_ARGS)
1438: {
1439: const char *name;
1440: char *cp;
1441:
1442: cp = *bufp + pos;
1443: while ('\0' != *cp) {
1.34 schwarze 1444: name = roff_getname(r, &cp, ln, (int)(cp - *bufp));
1.29 schwarze 1445: if ('\0' != *name)
1446: roff_setstr(r, name, NULL, 0);
1447: }
1.51 schwarze 1448: return(ROFF_IGN);
1449: }
1450:
1451: /* ARGSUSED */
1452: static enum rofferr
1453: roff_it(ROFF_ARGS)
1454: {
1455: char *cp;
1456: size_t len;
1457: int iv;
1458:
1459: /* Parse the number of lines. */
1460: cp = *bufp + pos;
1461: len = strcspn(cp, " \t");
1462: cp[len] = '\0';
1463: if ((iv = mandoc_strntoi(cp, len, 10)) <= 0) {
1464: mandoc_msg(MANDOCERR_NUMERIC, r->parse,
1465: ln, ppos, *bufp + 1);
1466: return(ROFF_IGN);
1467: }
1468: cp += len + 1;
1469:
1470: /* Arm the input line trap. */
1471: roffit_lines = iv;
1472: roffit_macro = mandoc_strdup(cp);
1.2 schwarze 1473: return(ROFF_IGN);
1.47 schwarze 1474: }
1475:
1476: /* ARGSUSED */
1477: static enum rofferr
1478: roff_Dd(ROFF_ARGS)
1479: {
1480: const char *const *cp;
1481:
1482: if (MPARSE_MDOC != r->parsetype)
1483: for (cp = __mdoc_reserved; *cp; cp++)
1484: roff_setstr(r, *cp, NULL, 0);
1485:
1486: return(ROFF_CONT);
1487: }
1488:
1489: /* ARGSUSED */
1490: static enum rofferr
1491: roff_TH(ROFF_ARGS)
1492: {
1493: const char *const *cp;
1494:
1495: if (MPARSE_MDOC != r->parsetype)
1496: for (cp = __man_reserved; *cp; cp++)
1497: roff_setstr(r, *cp, NULL, 0);
1498:
1499: return(ROFF_CONT);
1.14 schwarze 1500: }
1501:
1502: /* ARGSUSED */
1503: static enum rofferr
1.27 schwarze 1504: roff_TE(ROFF_ARGS)
1505: {
1506:
1507: if (NULL == r->tbl)
1.35 schwarze 1508: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.27 schwarze 1509: else
1.41 schwarze 1510: tbl_end(&r->tbl);
1.27 schwarze 1511:
1512: return(ROFF_IGN);
1513: }
1514:
1515: /* ARGSUSED */
1516: static enum rofferr
1517: roff_T_(ROFF_ARGS)
1518: {
1519:
1520: if (NULL == r->tbl)
1.35 schwarze 1521: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.27 schwarze 1522: else
1523: tbl_restart(ppos, ln, r->tbl);
1524:
1525: return(ROFF_IGN);
1526: }
1527:
1.41 schwarze 1528: #if 0
1529: static int
1530: roff_closeeqn(struct roff *r)
1531: {
1532:
1533: return(r->eqn && ROFF_EQN == eqn_end(&r->eqn) ? 1 : 0);
1534: }
1535: #endif
1536:
1537: static void
1538: roff_openeqn(struct roff *r, const char *name, int line,
1539: int offs, const char *buf)
1.32 schwarze 1540: {
1.41 schwarze 1541: struct eqn_node *e;
1542: int poff;
1.32 schwarze 1543:
1544: assert(NULL == r->eqn);
1.41 schwarze 1545: e = eqn_alloc(name, offs, line, r->parse);
1.32 schwarze 1546:
1547: if (r->last_eqn)
1548: r->last_eqn->next = e;
1549: else
1550: r->first_eqn = r->last_eqn = e;
1551:
1552: r->eqn = r->last_eqn = e;
1.41 schwarze 1553:
1554: if (buf) {
1555: poff = 0;
1556: eqn_read(&r->eqn, line, buf, offs, &poff);
1557: }
1558: }
1559:
1560: /* ARGSUSED */
1561: static enum rofferr
1562: roff_EQ(ROFF_ARGS)
1563: {
1564:
1565: roff_openeqn(r, *bufp + pos, ln, ppos, NULL);
1.32 schwarze 1566: return(ROFF_IGN);
1567: }
1568:
1569: /* ARGSUSED */
1570: static enum rofferr
1571: roff_EN(ROFF_ARGS)
1572: {
1573:
1.35 schwarze 1574: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.32 schwarze 1575: return(ROFF_IGN);
1576: }
1577:
1578: /* ARGSUSED */
1579: static enum rofferr
1.27 schwarze 1580: roff_TS(ROFF_ARGS)
1581: {
1.49 schwarze 1582: struct tbl_node *tbl;
1.27 schwarze 1583:
1584: if (r->tbl) {
1.35 schwarze 1585: mandoc_msg(MANDOCERR_SCOPEBROKEN, r->parse, ln, ppos, NULL);
1.41 schwarze 1586: tbl_end(&r->tbl);
1.27 schwarze 1587: }
1588:
1.49 schwarze 1589: tbl = tbl_alloc(ppos, ln, r->parse);
1.27 schwarze 1590:
1591: if (r->last_tbl)
1.49 schwarze 1592: r->last_tbl->next = tbl;
1.27 schwarze 1593: else
1.49 schwarze 1594: r->first_tbl = r->last_tbl = tbl;
1.27 schwarze 1595:
1.49 schwarze 1596: r->tbl = r->last_tbl = tbl;
1.27 schwarze 1597: return(ROFF_IGN);
1598: }
1599:
1600: /* ARGSUSED */
1601: static enum rofferr
1.48 schwarze 1602: roff_cc(ROFF_ARGS)
1603: {
1604: const char *p;
1605:
1606: p = *bufp + pos;
1607:
1608: if ('\0' == *p || '.' == (r->control = *p++))
1609: r->control = 0;
1610:
1611: if ('\0' != *p)
1612: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
1613:
1614: return(ROFF_IGN);
1615: }
1616:
1617: /* ARGSUSED */
1618: static enum rofferr
1.42 schwarze 1619: roff_tr(ROFF_ARGS)
1620: {
1621: const char *p, *first, *second;
1622: size_t fsz, ssz;
1623: enum mandoc_esc esc;
1624:
1625: p = *bufp + pos;
1626:
1627: if ('\0' == *p) {
1628: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
1629: return(ROFF_IGN);
1630: }
1631:
1632: while ('\0' != *p) {
1633: fsz = ssz = 1;
1634:
1635: first = p++;
1636: if ('\\' == *first) {
1637: esc = mandoc_escape(&p, NULL, NULL);
1638: if (ESCAPE_ERROR == esc) {
1639: mandoc_msg
1640: (MANDOCERR_BADESCAPE, r->parse,
1641: ln, (int)(p - *bufp), NULL);
1642: return(ROFF_IGN);
1643: }
1644: fsz = (size_t)(p - first);
1645: }
1646:
1647: second = p++;
1648: if ('\\' == *second) {
1649: esc = mandoc_escape(&p, NULL, NULL);
1650: if (ESCAPE_ERROR == esc) {
1651: mandoc_msg
1652: (MANDOCERR_BADESCAPE, r->parse,
1653: ln, (int)(p - *bufp), NULL);
1654: return(ROFF_IGN);
1655: }
1656: ssz = (size_t)(p - second);
1657: } else if ('\0' == *second) {
1658: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse,
1659: ln, (int)(p - *bufp), NULL);
1660: second = " ";
1661: p--;
1662: }
1663:
1664: if (fsz > 1) {
1665: roff_setstrn(&r->xmbtab, first,
1666: fsz, second, ssz, 0);
1667: continue;
1668: }
1669:
1670: if (NULL == r->xtab)
1671: r->xtab = mandoc_calloc
1672: (128, sizeof(struct roffstr));
1673:
1674: free(r->xtab[(int)*first].p);
1675: r->xtab[(int)*first].p = mandoc_strndup(second, ssz);
1676: r->xtab[(int)*first].sz = ssz;
1677: }
1678:
1679: return(ROFF_IGN);
1680: }
1681:
1682: /* ARGSUSED */
1683: static enum rofferr
1.14 schwarze 1684: roff_so(ROFF_ARGS)
1685: {
1686: char *name;
1.15 schwarze 1687:
1.35 schwarze 1688: mandoc_msg(MANDOCERR_SO, r->parse, ln, ppos, NULL);
1.14 schwarze 1689:
1.22 schwarze 1690: /*
1691: * Handle `so'. Be EXTREMELY careful, as we shouldn't be
1692: * opening anything that's not in our cwd or anything beneath
1693: * it. Thus, explicitly disallow traversing up the file-system
1694: * or using absolute paths.
1695: */
1696:
1.14 schwarze 1697: name = *bufp + pos;
1698: if ('/' == *name || strstr(name, "../") || strstr(name, "/..")) {
1.35 schwarze 1699: mandoc_msg(MANDOCERR_SOPATH, r->parse, ln, pos, NULL);
1.14 schwarze 1700: return(ROFF_ERR);
1701: }
1702:
1703: *offs = pos;
1704: return(ROFF_SO);
1.7 schwarze 1705: }
1706:
1.16 schwarze 1707: /* ARGSUSED */
1708: static enum rofferr
1709: roff_userdef(ROFF_ARGS)
1.12 schwarze 1710: {
1.16 schwarze 1711: const char *arg[9];
1712: char *cp, *n1, *n2;
1.25 schwarze 1713: int i;
1.12 schwarze 1714:
1.16 schwarze 1715: /*
1716: * Collect pointers to macro argument strings
1717: * and null-terminate them.
1718: */
1719: cp = *bufp + pos;
1.25 schwarze 1720: for (i = 0; i < 9; i++)
1.26 schwarze 1721: arg[i] = '\0' == *cp ? "" :
1.35 schwarze 1722: mandoc_getarg(r->parse, &cp, ln, &pos);
1.16 schwarze 1723:
1724: /*
1725: * Expand macro arguments.
1.12 schwarze 1726: */
1.16 schwarze 1727: *szp = 0;
1728: n1 = cp = mandoc_strdup(r->current_string);
1729: while (NULL != (cp = strstr(cp, "\\$"))) {
1730: i = cp[2] - '1';
1731: if (0 > i || 8 < i) {
1732: /* Not an argument invocation. */
1733: cp += 2;
1734: continue;
1735: }
1736:
1737: *szp = strlen(n1) - 3 + strlen(arg[i]) + 1;
1738: n2 = mandoc_malloc(*szp);
1739:
1740: strlcpy(n2, n1, (size_t)(cp - n1 + 1));
1741: strlcat(n2, arg[i], *szp);
1742: strlcat(n2, cp + 3, *szp);
1743:
1744: cp = n2 + (cp - n1);
1745: free(n1);
1746: n1 = n2;
1.12 schwarze 1747: }
1748:
1.16 schwarze 1749: /*
1750: * Replace the macro invocation
1751: * by the expanded macro.
1752: */
1753: free(*bufp);
1754: *bufp = n1;
1755: if (0 == *szp)
1756: *szp = strlen(*bufp) + 1;
1757:
1.19 schwarze 1758: return(*szp > 1 && '\n' == (*bufp)[(int)*szp - 2] ?
1.16 schwarze 1759: ROFF_REPARSE : ROFF_APPEND);
1.12 schwarze 1760: }
1.28 schwarze 1761:
1762: static char *
1763: roff_getname(struct roff *r, char **cpp, int ln, int pos)
1764: {
1765: char *name, *cp;
1766:
1767: name = *cpp;
1768: if ('\0' == *name)
1769: return(name);
1770:
1771: /* Read until end of name. */
1772: for (cp = name; '\0' != *cp && ' ' != *cp; cp++) {
1773: if ('\\' != *cp)
1774: continue;
1775: cp++;
1776: if ('\\' == *cp)
1777: continue;
1.35 schwarze 1778: mandoc_msg(MANDOCERR_NAMESC, r->parse, ln, pos, NULL);
1.28 schwarze 1779: *cp = '\0';
1780: name = cp;
1781: }
1782:
1783: /* Nil-terminate name. */
1784: if ('\0' != *cp)
1785: *(cp++) = '\0';
1786:
1787: /* Read past spaces. */
1788: while (' ' == *cp)
1789: cp++;
1790:
1791: *cpp = cp;
1792: return(name);
1793: }
1794:
1.16 schwarze 1795: /*
1796: * Store *string into the user-defined string called *name.
1797: * In multiline mode, append to an existing entry and append '\n';
1798: * else replace the existing entry, if there is one.
1799: * To clear an existing entry, call with (*r, *name, NULL, 0).
1800: */
1.8 schwarze 1801: static void
1.16 schwarze 1802: roff_setstr(struct roff *r, const char *name, const char *string,
1803: int multiline)
1.7 schwarze 1804: {
1.42 schwarze 1805:
1806: roff_setstrn(&r->strtab, name, strlen(name), string,
1807: string ? strlen(string) : 0, multiline);
1808: }
1809:
1810: static void
1811: roff_setstrn(struct roffkv **r, const char *name, size_t namesz,
1812: const char *string, size_t stringsz, int multiline)
1813: {
1814: struct roffkv *n;
1815: char *c;
1816: int i;
1817: size_t oldch, newch;
1.7 schwarze 1818:
1.16 schwarze 1819: /* Search for an existing string with the same name. */
1.42 schwarze 1820: n = *r;
1821:
1822: while (n && strcmp(name, n->key.p))
1.7 schwarze 1823: n = n->next;
1.8 schwarze 1824:
1825: if (NULL == n) {
1.16 schwarze 1826: /* Create a new string table entry. */
1.42 schwarze 1827: n = mandoc_malloc(sizeof(struct roffkv));
1828: n->key.p = mandoc_strndup(name, namesz);
1829: n->key.sz = namesz;
1830: n->val.p = NULL;
1831: n->val.sz = 0;
1832: n->next = *r;
1833: *r = n;
1.16 schwarze 1834: } else if (0 == multiline) {
1835: /* In multiline mode, append; else replace. */
1.42 schwarze 1836: free(n->val.p);
1837: n->val.p = NULL;
1838: n->val.sz = 0;
1.16 schwarze 1839: }
1840:
1841: if (NULL == string)
1842: return;
1843:
1844: /*
1845: * One additional byte for the '\n' in multiline mode,
1846: * and one for the terminating '\0'.
1847: */
1.42 schwarze 1848: newch = stringsz + (multiline ? 2u : 1u);
1849:
1850: if (NULL == n->val.p) {
1851: n->val.p = mandoc_malloc(newch);
1852: *n->val.p = '\0';
1.16 schwarze 1853: oldch = 0;
1854: } else {
1.42 schwarze 1855: oldch = n->val.sz;
1856: n->val.p = mandoc_realloc(n->val.p, oldch + newch);
1.16 schwarze 1857: }
1858:
1859: /* Skip existing content in the destination buffer. */
1.42 schwarze 1860: c = n->val.p + (int)oldch;
1.16 schwarze 1861:
1862: /* Append new content to the destination buffer. */
1.42 schwarze 1863: i = 0;
1864: while (i < (int)stringsz) {
1.16 schwarze 1865: /*
1866: * Rudimentary roff copy mode:
1867: * Handle escaped backslashes.
1868: */
1.42 schwarze 1869: if ('\\' == string[i] && '\\' == string[i + 1])
1870: i++;
1871: *c++ = string[i++];
1.16 schwarze 1872: }
1.8 schwarze 1873:
1.16 schwarze 1874: /* Append terminating bytes. */
1875: if (multiline)
1876: *c++ = '\n';
1.42 schwarze 1877:
1.16 schwarze 1878: *c = '\0';
1.42 schwarze 1879: n->val.sz = (int)(c - n->val.p);
1.7 schwarze 1880: }
1881:
1.8 schwarze 1882: static const char *
1883: roff_getstrn(const struct roff *r, const char *name, size_t len)
1.7 schwarze 1884: {
1.42 schwarze 1885: const struct roffkv *n;
1.7 schwarze 1886:
1.42 schwarze 1887: for (n = r->strtab; n; n = n->next)
1888: if (0 == strncmp(name, n->key.p, len) &&
1889: '\0' == n->key.p[(int)len])
1890: return(n->val.p);
1.8 schwarze 1891:
1.42 schwarze 1892: return(NULL);
1.7 schwarze 1893: }
1894:
1.8 schwarze 1895: static void
1.42 schwarze 1896: roff_freestr(struct roffkv *r)
1.7 schwarze 1897: {
1.42 schwarze 1898: struct roffkv *n, *nn;
1.7 schwarze 1899:
1.42 schwarze 1900: for (n = r; n; n = nn) {
1901: free(n->key.p);
1902: free(n->val.p);
1.7 schwarze 1903: nn = n->next;
1904: free(n);
1905: }
1.27 schwarze 1906: }
1907:
1908: const struct tbl_span *
1909: roff_span(const struct roff *r)
1910: {
1911:
1912: return(r->tbl ? tbl_span(r->tbl) : NULL);
1.32 schwarze 1913: }
1914:
1915: const struct eqn *
1916: roff_eqn(const struct roff *r)
1917: {
1918:
1919: return(r->last_eqn ? &r->last_eqn->eqn : NULL);
1.42 schwarze 1920: }
1921:
1922: /*
1923: * Duplicate an input string, making the appropriate character
1924: * conversations (as stipulated by `tr') along the way.
1925: * Returns a heap-allocated string with all the replacements made.
1926: */
1927: char *
1928: roff_strdup(const struct roff *r, const char *p)
1929: {
1930: const struct roffkv *cp;
1931: char *res;
1932: const char *pp;
1933: size_t ssz, sz;
1934: enum mandoc_esc esc;
1935:
1936: if (NULL == r->xmbtab && NULL == r->xtab)
1937: return(mandoc_strdup(p));
1938: else if ('\0' == *p)
1939: return(mandoc_strdup(""));
1940:
1941: /*
1942: * Step through each character looking for term matches
1943: * (remember that a `tr' can be invoked with an escape, which is
1944: * a glyph but the escape is multi-character).
1945: * We only do this if the character hash has been initialised
1946: * and the string is >0 length.
1947: */
1948:
1949: res = NULL;
1950: ssz = 0;
1951:
1952: while ('\0' != *p) {
1953: if ('\\' != *p && r->xtab && r->xtab[(int)*p].p) {
1954: sz = r->xtab[(int)*p].sz;
1955: res = mandoc_realloc(res, ssz + sz + 1);
1956: memcpy(res + ssz, r->xtab[(int)*p].p, sz);
1957: ssz += sz;
1958: p++;
1959: continue;
1960: } else if ('\\' != *p) {
1961: res = mandoc_realloc(res, ssz + 2);
1962: res[ssz++] = *p++;
1963: continue;
1964: }
1965:
1966: /* Search for term matches. */
1967: for (cp = r->xmbtab; cp; cp = cp->next)
1968: if (0 == strncmp(p, cp->key.p, cp->key.sz))
1969: break;
1970:
1971: if (NULL != cp) {
1972: /*
1973: * A match has been found.
1974: * Append the match to the array and move
1975: * forward by its keysize.
1976: */
1977: res = mandoc_realloc
1978: (res, ssz + cp->val.sz + 1);
1979: memcpy(res + ssz, cp->val.p, cp->val.sz);
1980: ssz += cp->val.sz;
1981: p += (int)cp->key.sz;
1982: continue;
1983: }
1984:
1985: /*
1986: * Handle escapes carefully: we need to copy
1987: * over just the escape itself, or else we might
1988: * do replacements within the escape itself.
1989: * Make sure to pass along the bogus string.
1990: */
1991: pp = p++;
1992: esc = mandoc_escape(&p, NULL, NULL);
1993: if (ESCAPE_ERROR == esc) {
1994: sz = strlen(pp);
1995: res = mandoc_realloc(res, ssz + sz + 1);
1996: memcpy(res + ssz, pp, sz);
1997: break;
1998: }
1999: /*
2000: * We bail out on bad escapes.
2001: * No need to warn: we already did so when
2002: * roff_res() was called.
2003: */
2004: sz = (int)(p - pp);
2005: res = mandoc_realloc(res, ssz + sz + 1);
2006: memcpy(res + ssz, pp, sz);
2007: ssz += sz;
2008: }
2009:
2010: res[(int)ssz] = '\0';
2011: return(res);
1.48 schwarze 2012: }
2013:
2014: /*
2015: * Find out whether a line is a macro line or not.
2016: * If it is, adjust the current position and return one; if it isn't,
2017: * return zero and don't change the current position.
2018: * If the control character has been set with `.cc', then let that grain
2019: * precedence.
2020: * This is slighly contrary to groff, where using the non-breaking
2021: * control character when `cc' has been invoked will cause the
2022: * non-breaking macro contents to be printed verbatim.
2023: */
2024: int
2025: roff_getcontrol(const struct roff *r, const char *cp, int *ppos)
2026: {
2027: int pos;
2028:
2029: pos = *ppos;
2030:
2031: if (0 != r->control && cp[pos] == r->control)
2032: pos++;
2033: else if (0 != r->control)
2034: return(0);
2035: else if ('\\' == cp[pos] && '.' == cp[pos + 1])
2036: pos += 2;
2037: else if ('.' == cp[pos] || '\'' == cp[pos])
2038: pos++;
2039: else
2040: return(0);
2041:
2042: while (' ' == cp[pos] || '\t' == cp[pos])
2043: pos++;
2044:
2045: *ppos = pos;
2046: return(1);
1.1 schwarze 2047: }
![[BACK]](/icons/back.gif){kind=icon}
Return to roff.c CVS log ![[TXT]](/icons/text.gif){kind=icon}
![[DIR]](/icons/dir.gif){kind=icon}
Up to [local] / src / usr.bin / mandoc