Annotation of src/usr.bin/mandoc/roff.c, Revision 1.66
1.66 ! schwarze 1: /* $Id: roff.c,v 1.65 2014/02/14 22:27:32 schwarze Exp $ */
1.1 schwarze 2: /*
1.48 schwarze 3: * Copyright (c) 2010, 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
1.63 schwarze 4: * Copyright (c) 2010-2014 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,
1.66 ! schwarze 39: ROFF_as,
1.48 schwarze 40: ROFF_cc,
1.1 schwarze 41: ROFF_de,
42: ROFF_dei,
1.2 schwarze 43: ROFF_de1,
44: ROFF_ds,
45: ROFF_el,
1.58 schwarze 46: ROFF_fam,
1.59 schwarze 47: ROFF_hw,
1.20 schwarze 48: ROFF_hy,
1.2 schwarze 49: ROFF_ie,
50: ROFF_if,
1.1 schwarze 51: ROFF_ig,
1.30 schwarze 52: ROFF_it,
1.20 schwarze 53: ROFF_ne,
54: ROFF_nh,
1.14 schwarze 55: ROFF_nr,
1.31 schwarze 56: ROFF_ns,
57: ROFF_ps,
1.2 schwarze 58: ROFF_rm,
1.14 schwarze 59: ROFF_so,
1.31 schwarze 60: ROFF_ta,
1.2 schwarze 61: ROFF_tr,
1.47 schwarze 62: ROFF_Dd,
63: ROFF_TH,
1.27 schwarze 64: ROFF_TS,
65: ROFF_TE,
66: ROFF_T_,
1.32 schwarze 67: ROFF_EQ,
68: ROFF_EN,
1.2 schwarze 69: ROFF_cblock,
1.37 schwarze 70: ROFF_ccond,
1.16 schwarze 71: ROFF_USERDEF,
1.1 schwarze 72: ROFF_MAX
73: };
74:
1.2 schwarze 75: enum roffrule {
1.55 schwarze 76: ROFFRULE_DENY,
77: ROFFRULE_ALLOW
1.2 schwarze 78: };
79:
1.41 schwarze 80: /*
1.42 schwarze 81: * An incredibly-simple string buffer.
82: */
1.8 schwarze 83: struct roffstr {
1.42 schwarze 84: char *p; /* nil-terminated buffer */
85: size_t sz; /* saved strlen(p) */
86: };
87:
88: /*
89: * A key-value roffstr pair as part of a singly-linked list.
90: */
91: struct roffkv {
92: struct roffstr key;
93: struct roffstr val;
94: struct roffkv *next; /* next in list */
1.8 schwarze 95: };
96:
1.52 schwarze 97: /*
98: * A single number register as part of a singly-linked list.
99: */
100: struct roffreg {
101: struct roffstr key;
1.53 schwarze 102: int val;
1.52 schwarze 103: struct roffreg *next;
104: };
105:
1.1 schwarze 106: struct roff {
1.47 schwarze 107: enum mparset parsetype; /* requested parse type */
1.35 schwarze 108: struct mparse *parse; /* parse point */
1.63 schwarze 109: int quick; /* skip standard macro deletion */
1.1 schwarze 110: struct roffnode *last; /* leaf of stack */
1.2 schwarze 111: enum roffrule rstack[RSTACK_MAX]; /* stack of !`ie' rules */
1.48 schwarze 112: char control; /* control character */
1.2 schwarze 113: int rstackpos; /* position in rstack */
1.52 schwarze 114: struct roffreg *regtab; /* number registers */
1.42 schwarze 115: struct roffkv *strtab; /* user-defined strings & macros */
116: struct roffkv *xmbtab; /* multi-byte trans table (`tr') */
117: struct roffstr *xtab; /* single-byte trans table (`tr') */
1.16 schwarze 118: const char *current_string; /* value of last called user macro */
1.27 schwarze 119: struct tbl_node *first_tbl; /* first table parsed */
120: struct tbl_node *last_tbl; /* last table parsed */
121: struct tbl_node *tbl; /* current table being parsed */
1.32 schwarze 122: struct eqn_node *last_eqn; /* last equation parsed */
123: struct eqn_node *first_eqn; /* first equation parsed */
124: struct eqn_node *eqn; /* current equation being parsed */
1.1 schwarze 125: };
126:
127: struct roffnode {
128: enum rofft tok; /* type of node */
129: struct roffnode *parent; /* up one in stack */
130: int line; /* parse line */
131: int col; /* parse col */
1.16 schwarze 132: char *name; /* node name, e.g. macro name */
1.2 schwarze 133: char *end; /* end-rules: custom token */
134: int endspan; /* end-rules: next-line or infty */
135: enum roffrule rule; /* current evaluation rule */
1.1 schwarze 136: };
137:
138: #define ROFF_ARGS struct roff *r, /* parse ctx */ \
139: enum rofft tok, /* tok of macro */ \
140: char **bufp, /* input buffer */ \
141: size_t *szp, /* size of input buffer */ \
142: int ln, /* parse line */ \
1.2 schwarze 143: int ppos, /* original pos in buffer */ \
144: int pos, /* current pos in buffer */ \
145: int *offs /* reset offset of buffer data */
1.1 schwarze 146:
147: typedef enum rofferr (*roffproc)(ROFF_ARGS);
148:
149: struct roffmac {
150: const char *name; /* macro name */
1.2 schwarze 151: roffproc proc; /* process new macro */
152: roffproc text; /* process as child text of macro */
153: roffproc sub; /* process as child of macro */
154: int flags;
155: #define ROFFMAC_STRUCT (1 << 0) /* always interpret */
1.3 schwarze 156: struct roffmac *next;
1.1 schwarze 157: };
158:
1.37 schwarze 159: struct predef {
160: const char *name; /* predefined input name */
161: const char *str; /* replacement symbol */
162: };
163:
164: #define PREDEF(__name, __str) \
165: { (__name), (__str) },
166:
1.42 schwarze 167: static enum rofft roffhash_find(const char *, size_t);
168: static void roffhash_init(void);
169: static void roffnode_cleanscope(struct roff *);
170: static void roffnode_pop(struct roff *);
171: static void roffnode_push(struct roff *, enum rofft,
172: const char *, int, int);
1.2 schwarze 173: static enum rofferr roff_block(ROFF_ARGS);
174: static enum rofferr roff_block_text(ROFF_ARGS);
175: static enum rofferr roff_block_sub(ROFF_ARGS);
176: static enum rofferr roff_cblock(ROFF_ARGS);
1.48 schwarze 177: static enum rofferr roff_cc(ROFF_ARGS);
1.2 schwarze 178: static enum rofferr roff_ccond(ROFF_ARGS);
179: static enum rofferr roff_cond(ROFF_ARGS);
180: static enum rofferr roff_cond_text(ROFF_ARGS);
181: static enum rofferr roff_cond_sub(ROFF_ARGS);
1.7 schwarze 182: static enum rofferr roff_ds(ROFF_ARGS);
1.8 schwarze 183: static enum roffrule roff_evalcond(const char *, int *);
1.42 schwarze 184: static void roff_free1(struct roff *);
1.52 schwarze 185: static void roff_freereg(struct roffreg *);
1.42 schwarze 186: static void roff_freestr(struct roffkv *);
1.28 schwarze 187: static char *roff_getname(struct roff *, char **, int, int);
1.56 schwarze 188: static int roff_getnum(const char *, int *, int *);
189: static int roff_getop(const char *, int *, char *);
1.53 schwarze 190: static int roff_getregn(const struct roff *,
191: const char *, size_t);
1.65 schwarze 192: static int roff_getregro(const char *name);
1.8 schwarze 193: static const char *roff_getstrn(const struct roff *,
194: const char *, size_t);
1.51 schwarze 195: static enum rofferr roff_it(ROFF_ARGS);
1.21 schwarze 196: static enum rofferr roff_line_ignore(ROFF_ARGS);
1.6 schwarze 197: static enum rofferr roff_nr(ROFF_ARGS);
1.41 schwarze 198: static void roff_openeqn(struct roff *, const char *,
199: int, int, const char *);
1.42 schwarze 200: static enum rofft roff_parse(struct roff *, const char *, int *);
1.51 schwarze 201: static enum rofferr roff_parsetext(char **, size_t *, int, int *);
1.45 schwarze 202: static enum rofferr roff_res(struct roff *,
1.37 schwarze 203: char **, size_t *, int, int);
1.29 schwarze 204: static enum rofferr roff_rm(ROFF_ARGS);
1.8 schwarze 205: static void roff_setstr(struct roff *,
1.16 schwarze 206: const char *, const char *, int);
1.42 schwarze 207: static void roff_setstrn(struct roffkv **, const char *,
208: size_t, const char *, size_t, int);
1.14 schwarze 209: static enum rofferr roff_so(ROFF_ARGS);
1.42 schwarze 210: static enum rofferr roff_tr(ROFF_ARGS);
1.47 schwarze 211: static enum rofferr roff_Dd(ROFF_ARGS);
212: static enum rofferr roff_TH(ROFF_ARGS);
1.27 schwarze 213: static enum rofferr roff_TE(ROFF_ARGS);
214: static enum rofferr roff_TS(ROFF_ARGS);
1.32 schwarze 215: static enum rofferr roff_EQ(ROFF_ARGS);
216: static enum rofferr roff_EN(ROFF_ARGS);
1.27 schwarze 217: static enum rofferr roff_T_(ROFF_ARGS);
1.16 schwarze 218: static enum rofferr roff_userdef(ROFF_ARGS);
1.1 schwarze 219:
1.42 schwarze 220: /* See roffhash_find() */
1.3 schwarze 221:
222: #define ASCII_HI 126
223: #define ASCII_LO 33
224: #define HASHWIDTH (ASCII_HI - ASCII_LO + 1)
225:
226: static struct roffmac *hash[HASHWIDTH];
227:
228: static struct roffmac roffs[ROFF_MAX] = {
1.21 schwarze 229: { "ad", roff_line_ignore, NULL, NULL, 0, NULL },
1.3 schwarze 230: { "am", roff_block, roff_block_text, roff_block_sub, 0, NULL },
231: { "ami", roff_block, roff_block_text, roff_block_sub, 0, NULL },
232: { "am1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.66 ! schwarze 233: { "as", roff_ds, NULL, NULL, 0, NULL },
1.48 schwarze 234: { "cc", roff_cc, NULL, NULL, 0, NULL },
1.3 schwarze 235: { "de", roff_block, roff_block_text, roff_block_sub, 0, NULL },
236: { "dei", roff_block, roff_block_text, roff_block_sub, 0, NULL },
237: { "de1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.7 schwarze 238: { "ds", roff_ds, NULL, NULL, 0, NULL },
1.3 schwarze 239: { "el", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
1.58 schwarze 240: { "fam", roff_line_ignore, NULL, NULL, 0, NULL },
1.59 schwarze 241: { "hw", roff_line_ignore, NULL, NULL, 0, NULL },
1.21 schwarze 242: { "hy", roff_line_ignore, NULL, NULL, 0, NULL },
1.3 schwarze 243: { "ie", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
244: { "if", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
245: { "ig", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.51 schwarze 246: { "it", roff_it, NULL, NULL, 0, NULL },
1.21 schwarze 247: { "ne", roff_line_ignore, NULL, NULL, 0, NULL },
248: { "nh", roff_line_ignore, NULL, NULL, 0, NULL },
1.14 schwarze 249: { "nr", roff_nr, NULL, NULL, 0, NULL },
1.31 schwarze 250: { "ns", roff_line_ignore, NULL, NULL, 0, NULL },
251: { "ps", roff_line_ignore, NULL, NULL, 0, NULL },
1.29 schwarze 252: { "rm", roff_rm, NULL, NULL, 0, NULL },
1.14 schwarze 253: { "so", roff_so, NULL, NULL, 0, NULL },
1.31 schwarze 254: { "ta", roff_line_ignore, NULL, NULL, 0, NULL },
1.42 schwarze 255: { "tr", roff_tr, NULL, NULL, 0, NULL },
1.47 schwarze 256: { "Dd", roff_Dd, NULL, NULL, 0, NULL },
257: { "TH", roff_TH, NULL, NULL, 0, NULL },
1.27 schwarze 258: { "TS", roff_TS, NULL, NULL, 0, NULL },
259: { "TE", roff_TE, NULL, NULL, 0, NULL },
260: { "T&", roff_T_, NULL, NULL, 0, NULL },
1.32 schwarze 261: { "EQ", roff_EQ, NULL, NULL, 0, NULL },
262: { "EN", roff_EN, NULL, NULL, 0, NULL },
1.3 schwarze 263: { ".", roff_cblock, NULL, NULL, 0, NULL },
264: { "\\}", roff_ccond, NULL, NULL, 0, NULL },
1.16 schwarze 265: { NULL, roff_userdef, NULL, NULL, 0, NULL },
1.1 schwarze 266: };
267:
1.47 schwarze 268: const char *const __mdoc_reserved[] = {
269: "Ac", "Ad", "An", "Ao", "Ap", "Aq", "Ar", "At",
270: "Bc", "Bd", "Bf", "Bk", "Bl", "Bo", "Bq",
271: "Brc", "Bro", "Brq", "Bsx", "Bt", "Bx",
272: "Cd", "Cm", "Db", "Dc", "Dd", "Dl", "Do", "Dq",
273: "Ds", "Dt", "Dv", "Dx", "D1",
274: "Ec", "Ed", "Ef", "Ek", "El", "Em", "em",
275: "En", "Eo", "Eq", "Er", "Es", "Ev", "Ex",
276: "Fa", "Fc", "Fd", "Fl", "Fn", "Fo", "Fr", "Ft", "Fx",
277: "Hf", "Ic", "In", "It", "Lb", "Li", "Lk", "Lp", "LP",
278: "Me", "Ms", "Mt", "Nd", "Nm", "No", "Ns", "Nx",
279: "Oc", "Oo", "Op", "Os", "Ot", "Ox",
280: "Pa", "Pc", "Pf", "Po", "Pp", "PP", "pp", "Pq",
281: "Qc", "Ql", "Qo", "Qq", "Or", "Rd", "Re", "Rs", "Rv",
282: "Sc", "Sf", "Sh", "SH", "Sm", "So", "Sq",
283: "Ss", "St", "Sx", "Sy",
284: "Ta", "Tn", "Ud", "Ux", "Va", "Vt", "Xc", "Xo", "Xr",
285: "%A", "%B", "%D", "%I", "%J", "%N", "%O",
286: "%P", "%Q", "%R", "%T", "%U", "%V",
287: NULL
288: };
289:
290: const char *const __man_reserved[] = {
291: "AT", "B", "BI", "BR", "BT", "DE", "DS", "DT",
292: "EE", "EN", "EQ", "EX", "HF", "HP", "I", "IB", "IP", "IR",
293: "LP", "ME", "MT", "OP", "P", "PD", "PP", "PT",
294: "R", "RB", "RE", "RI", "RS", "SB", "SH", "SM", "SS", "SY",
295: "TE", "TH", "TP", "TQ", "TS", "T&", "UC", "UE", "UR", "YS",
296: NULL
297: };
298:
1.37 schwarze 299: /* Array of injected predefined strings. */
300: #define PREDEFS_MAX 38
301: static const struct predef predefs[PREDEFS_MAX] = {
302: #include "predefs.in"
303: };
304:
1.42 schwarze 305: /* See roffhash_find() */
1.3 schwarze 306: #define ROFF_HASH(p) (p[0] - ASCII_LO)
307:
1.51 schwarze 308: static int roffit_lines; /* number of lines to delay */
309: static char *roffit_macro; /* nil-terminated macro line */
310:
1.3 schwarze 311: static void
1.42 schwarze 312: roffhash_init(void)
1.3 schwarze 313: {
314: struct roffmac *n;
315: int buc, i;
316:
1.16 schwarze 317: for (i = 0; i < (int)ROFF_USERDEF; i++) {
1.3 schwarze 318: assert(roffs[i].name[0] >= ASCII_LO);
319: assert(roffs[i].name[0] <= ASCII_HI);
320:
321: buc = ROFF_HASH(roffs[i].name);
322:
323: if (NULL != (n = hash[buc])) {
324: for ( ; n->next; n = n->next)
325: /* Do nothing. */ ;
326: n->next = &roffs[i];
327: } else
328: hash[buc] = &roffs[i];
329: }
330: }
331:
1.1 schwarze 332: /*
333: * Look up a roff token by its name. Returns ROFF_MAX if no macro by
334: * the nil-terminated string name could be found.
335: */
336: static enum rofft
1.42 schwarze 337: roffhash_find(const char *p, size_t s)
1.1 schwarze 338: {
1.3 schwarze 339: int buc;
340: struct roffmac *n;
1.1 schwarze 341:
1.3 schwarze 342: /*
343: * libroff has an extremely simple hashtable, for the time
344: * being, which simply keys on the first character, which must
345: * be printable, then walks a chain. It works well enough until
346: * optimised.
347: */
348:
349: if (p[0] < ASCII_LO || p[0] > ASCII_HI)
350: return(ROFF_MAX);
351:
352: buc = ROFF_HASH(p);
353:
354: if (NULL == (n = hash[buc]))
355: return(ROFF_MAX);
356: for ( ; n; n = n->next)
1.16 schwarze 357: if (0 == strncmp(n->name, p, s) && '\0' == n->name[(int)s])
1.3 schwarze 358: return((enum rofft)(n - roffs));
1.1 schwarze 359:
360: return(ROFF_MAX);
361: }
362:
363:
364: /*
365: * Pop the current node off of the stack of roff instructions currently
366: * pending.
367: */
368: static void
369: roffnode_pop(struct roff *r)
370: {
371: struct roffnode *p;
372:
1.2 schwarze 373: assert(r->last);
374: p = r->last;
375:
376: r->last = r->last->parent;
1.16 schwarze 377: free(p->name);
378: free(p->end);
1.1 schwarze 379: free(p);
380: }
381:
382:
383: /*
384: * Push a roff node onto the instruction stack. This must later be
385: * removed with roffnode_pop().
386: */
1.11 schwarze 387: static void
1.16 schwarze 388: roffnode_push(struct roff *r, enum rofft tok, const char *name,
389: int line, int col)
1.1 schwarze 390: {
391: struct roffnode *p;
392:
1.11 schwarze 393: p = mandoc_calloc(1, sizeof(struct roffnode));
1.1 schwarze 394: p->tok = tok;
1.16 schwarze 395: if (name)
396: p->name = mandoc_strdup(name);
1.1 schwarze 397: p->parent = r->last;
398: p->line = line;
399: p->col = col;
1.2 schwarze 400: p->rule = p->parent ? p->parent->rule : ROFFRULE_DENY;
1.1 schwarze 401:
402: r->last = p;
403: }
404:
405:
406: static void
407: roff_free1(struct roff *r)
408: {
1.49 schwarze 409: struct tbl_node *tbl;
1.32 schwarze 410: struct eqn_node *e;
1.42 schwarze 411: int i;
1.27 schwarze 412:
1.49 schwarze 413: while (NULL != (tbl = r->first_tbl)) {
414: r->first_tbl = tbl->next;
415: tbl_free(tbl);
1.27 schwarze 416: }
417:
418: r->first_tbl = r->last_tbl = r->tbl = NULL;
1.1 schwarze 419:
1.32 schwarze 420: while (NULL != (e = r->first_eqn)) {
421: r->first_eqn = e->next;
422: eqn_free(e);
423: }
424:
425: r->first_eqn = r->last_eqn = r->eqn = NULL;
426:
1.1 schwarze 427: while (r->last)
428: roffnode_pop(r);
1.27 schwarze 429:
1.42 schwarze 430: roff_freestr(r->strtab);
431: roff_freestr(r->xmbtab);
432:
433: r->strtab = r->xmbtab = NULL;
434:
1.52 schwarze 435: roff_freereg(r->regtab);
436:
437: r->regtab = NULL;
438:
1.42 schwarze 439: if (r->xtab)
440: for (i = 0; i < 128; i++)
441: free(r->xtab[i].p);
442:
443: free(r->xtab);
444: r->xtab = NULL;
1.1 schwarze 445: }
446:
447: void
448: roff_reset(struct roff *r)
449: {
450:
451: roff_free1(r);
1.48 schwarze 452: r->control = 0;
1.1 schwarze 453: }
454:
455:
456: void
457: roff_free(struct roff *r)
458: {
459:
460: roff_free1(r);
461: free(r);
462: }
463:
464:
465: struct roff *
1.63 schwarze 466: roff_alloc(enum mparset type, struct mparse *parse, int quick)
1.1 schwarze 467: {
468: struct roff *r;
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;
1.63 schwarze 473: r->quick = quick;
1.2 schwarze 474: r->rstackpos = -1;
1.3 schwarze 475:
1.42 schwarze 476: roffhash_init();
1.37 schwarze 477:
1.1 schwarze 478: return(r);
479: }
480:
1.8 schwarze 481: /*
1.53 schwarze 482: * In the current line, expand user-defined strings ("\*")
483: * and references to number registers ("\n").
484: * Also check the syntax of other escape sequences.
1.8 schwarze 485: */
1.45 schwarze 486: static enum rofferr
1.37 schwarze 487: roff_res(struct roff *r, char **bufp, size_t *szp, int ln, int pos)
1.8 schwarze 488: {
1.53 schwarze 489: char ubuf[12]; /* buffer to print the number */
1.23 schwarze 490: const char *stesc; /* start of an escape sequence ('\\') */
491: const char *stnam; /* start of the name, after "[(*" */
492: const char *cp; /* end of the name, e.g. before ']' */
493: const char *res; /* the string to be substituted */
1.53 schwarze 494: char *nbuf; /* new buffer to copy bufp to */
495: size_t nsz; /* size of the new buffer */
496: size_t maxl; /* expected length of the escape name */
497: size_t naml; /* actual length of the escape name */
498: int expand_count; /* to avoid infinite loops */
1.8 schwarze 499:
1.43 schwarze 500: expand_count = 0;
501:
1.42 schwarze 502: again:
1.24 schwarze 503: cp = *bufp + pos;
504: while (NULL != (cp = strchr(cp, '\\'))) {
505: stesc = cp++;
1.23 schwarze 506:
507: /*
1.53 schwarze 508: * The second character must be an asterisk or an n.
1.23 schwarze 509: * If it isn't, skip it anyway: It is escaped,
510: * so it can't start another escape sequence.
511: */
512:
1.24 schwarze 513: if ('\0' == *cp)
1.45 schwarze 514: return(ROFF_CONT);
1.42 schwarze 515:
1.53 schwarze 516: switch (*cp) {
517: case ('*'):
518: res = NULL;
519: break;
520: case ('n'):
521: res = ubuf;
522: break;
523: default:
524: if (ESCAPE_ERROR != mandoc_escape(&cp, NULL, NULL))
1.42 schwarze 525: continue;
526: mandoc_msg
527: (MANDOCERR_BADESCAPE, r->parse,
528: ln, (int)(stesc - *bufp), NULL);
1.45 schwarze 529: return(ROFF_CONT);
1.42 schwarze 530: }
531:
532: cp++;
1.23 schwarze 533:
534: /*
535: * The third character decides the length
1.53 schwarze 536: * of the name of the string or register.
1.23 schwarze 537: * Save a pointer to the name.
538: */
539:
1.24 schwarze 540: switch (*cp) {
541: case ('\0'):
1.45 schwarze 542: return(ROFF_CONT);
1.8 schwarze 543: case ('('):
544: cp++;
545: maxl = 2;
546: break;
547: case ('['):
548: cp++;
549: maxl = 0;
550: break;
551: default:
552: maxl = 1;
553: break;
554: }
1.23 schwarze 555: stnam = cp;
1.8 schwarze 556:
1.23 schwarze 557: /* Advance to the end of the name. */
1.8 schwarze 558:
1.53 schwarze 559: for (naml = 0; 0 == maxl || naml < maxl; naml++, cp++) {
1.42 schwarze 560: if ('\0' == *cp) {
561: mandoc_msg
562: (MANDOCERR_BADESCAPE,
563: r->parse, ln,
564: (int)(stesc - *bufp), NULL);
1.45 schwarze 565: return(ROFF_CONT);
1.42 schwarze 566: }
1.8 schwarze 567: if (0 == maxl && ']' == *cp)
568: break;
569: }
570:
1.23 schwarze 571: /*
572: * Retrieve the replacement string; if it is
573: * undefined, resume searching for escapes.
574: */
575:
1.53 schwarze 576: if (NULL == res)
577: res = roff_getstrn(r, stnam, naml);
578: else
579: snprintf(ubuf, sizeof(ubuf), "%d",
580: roff_getregn(r, stnam, naml));
1.8 schwarze 581:
582: if (NULL == res) {
1.42 schwarze 583: mandoc_msg
584: (MANDOCERR_BADESCAPE, r->parse,
585: ln, (int)(stesc - *bufp), NULL);
1.37 schwarze 586: res = "";
1.8 schwarze 587: }
588:
1.23 schwarze 589: /* Replace the escape sequence by the string. */
590:
1.42 schwarze 591: pos = stesc - *bufp;
592:
1.8 schwarze 593: nsz = *szp + strlen(res) + 1;
1.53 schwarze 594: nbuf = mandoc_malloc(nsz);
1.8 schwarze 595:
1.53 schwarze 596: strlcpy(nbuf, *bufp, (size_t)(stesc - *bufp + 1));
597: strlcat(nbuf, res, nsz);
598: strlcat(nbuf, cp + (maxl ? 0 : 1), nsz);
1.8 schwarze 599:
600: free(*bufp);
601:
1.53 schwarze 602: *bufp = nbuf;
1.8 schwarze 603: *szp = nsz;
1.43 schwarze 604:
605: if (EXPAND_LIMIT >= ++expand_count)
606: goto again;
607:
608: /* Just leave the string unexpanded. */
609: mandoc_msg(MANDOCERR_ROFFLOOP, r->parse, ln, pos, NULL);
1.45 schwarze 610: return(ROFF_IGN);
1.42 schwarze 611: }
1.45 schwarze 612: return(ROFF_CONT);
1.42 schwarze 613: }
614:
615: /*
1.51 schwarze 616: * Process text streams:
617: * Convert all breakable hyphens into ASCII_HYPH.
618: * Decrement and spring input line trap.
1.42 schwarze 619: */
620: static enum rofferr
1.51 schwarze 621: roff_parsetext(char **bufp, size_t *szp, int pos, int *offs)
1.42 schwarze 622: {
623: size_t sz;
624: const char *start;
1.51 schwarze 625: char *p;
626: int isz;
1.42 schwarze 627: enum mandoc_esc esc;
628:
1.51 schwarze 629: start = p = *bufp + pos;
1.42 schwarze 630:
631: while ('\0' != *p) {
632: sz = strcspn(p, "-\\");
633: p += sz;
634:
635: if ('\0' == *p)
636: break;
637:
638: if ('\\' == *p) {
639: /* Skip over escapes. */
640: p++;
1.62 schwarze 641: esc = mandoc_escape((const char **)&p, NULL, NULL);
1.42 schwarze 642: if (ESCAPE_ERROR == esc)
643: break;
644: continue;
645: } else if (p == start) {
646: p++;
647: continue;
648: }
649:
1.44 schwarze 650: if (isalpha((unsigned char)p[-1]) &&
651: isalpha((unsigned char)p[1]))
1.42 schwarze 652: *p = ASCII_HYPH;
653: p++;
1.8 schwarze 654: }
655:
1.51 schwarze 656: /* Spring the input line trap. */
657: if (1 == roffit_lines) {
658: isz = asprintf(&p, "%s\n.%s", *bufp, roffit_macro);
659: if (-1 == isz) {
660: perror(NULL);
661: exit((int)MANDOCLEVEL_SYSERR);
662: }
663: free(*bufp);
664: *bufp = p;
665: *szp = isz + 1;
666: *offs = 0;
667: free(roffit_macro);
668: roffit_lines = 0;
669: return(ROFF_REPARSE);
670: } else if (1 < roffit_lines)
671: --roffit_lines;
1.42 schwarze 672: return(ROFF_CONT);
1.8 schwarze 673: }
674:
1.1 schwarze 675: enum rofferr
1.6 schwarze 676: roff_parseln(struct roff *r, int ln, char **bufp,
677: size_t *szp, int pos, int *offs)
1.1 schwarze 678: {
679: enum rofft t;
1.27 schwarze 680: enum rofferr e;
1.35 schwarze 681: int ppos, ctl;
1.1 schwarze 682:
1.2 schwarze 683: /*
1.8 schwarze 684: * Run the reserved-word filter only if we have some reserved
685: * words to fill in.
686: */
687:
1.45 schwarze 688: e = roff_res(r, bufp, szp, ln, pos);
689: if (ROFF_IGN == e)
690: return(e);
691: assert(ROFF_CONT == e);
1.8 schwarze 692:
1.35 schwarze 693: ppos = pos;
1.48 schwarze 694: ctl = roff_getcontrol(r, *bufp, &pos);
1.35 schwarze 695:
1.8 schwarze 696: /*
1.2 schwarze 697: * First, if a scope is open and we're not a macro, pass the
698: * text through the macro's filter. If a scope isn't open and
699: * we're not a macro, just let it through.
1.32 schwarze 700: * Finally, if there's an equation scope open, divert it into it
701: * no matter our state.
1.2 schwarze 702: */
703:
1.35 schwarze 704: if (r->last && ! ctl) {
1.2 schwarze 705: t = r->last->tok;
706: assert(roffs[t].text);
1.27 schwarze 707: e = (*roffs[t].text)
708: (r, t, bufp, szp, ln, pos, pos, offs);
709: assert(ROFF_IGN == e || ROFF_CONT == e);
1.32 schwarze 710: if (ROFF_CONT != e)
711: return(e);
1.54 schwarze 712: }
713: if (r->eqn)
714: return(eqn_read(&r->eqn, ln, *bufp, ppos, offs));
715: if ( ! ctl) {
1.32 schwarze 716: if (r->tbl)
1.35 schwarze 717: return(tbl_read(r->tbl, ln, *bufp, pos));
1.51 schwarze 718: return(roff_parsetext(bufp, szp, pos, offs));
1.54 schwarze 719: }
1.2 schwarze 720:
721: /*
722: * If a scope is open, go to the child handler for that macro,
723: * as it may want to preprocess before doing anything with it.
1.32 schwarze 724: * Don't do so if an equation is open.
1.2 schwarze 725: */
726:
727: if (r->last) {
1.1 schwarze 728: t = r->last->tok;
729: assert(roffs[t].sub);
1.2 schwarze 730: return((*roffs[t].sub)
1.8 schwarze 731: (r, t, bufp, szp,
1.35 schwarze 732: ln, ppos, pos, offs));
1.2 schwarze 733: }
734:
735: /*
736: * Lastly, as we've no scope open, try to look up and execute
737: * the new macro. If no macro is found, simply return and let
738: * the compilers handle it.
739: */
740:
1.16 schwarze 741: if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos)))
1.1 schwarze 742: return(ROFF_CONT);
743:
1.2 schwarze 744: assert(roffs[t].proc);
745: return((*roffs[t].proc)
1.8 schwarze 746: (r, t, bufp, szp,
747: ln, ppos, pos, offs));
1.2 schwarze 748: }
749:
1.1 schwarze 750:
1.27 schwarze 751: void
1.2 schwarze 752: roff_endparse(struct roff *r)
753: {
1.1 schwarze 754:
1.27 schwarze 755: if (r->last)
1.35 schwarze 756: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.27 schwarze 757: r->last->line, r->last->col, NULL);
758:
1.32 schwarze 759: if (r->eqn) {
1.35 schwarze 760: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.41 schwarze 761: r->eqn->eqn.ln, r->eqn->eqn.pos, NULL);
762: eqn_end(&r->eqn);
1.32 schwarze 763: }
764:
1.27 schwarze 765: if (r->tbl) {
1.35 schwarze 766: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.27 schwarze 767: r->tbl->line, r->tbl->pos, NULL);
1.41 schwarze 768: tbl_end(&r->tbl);
1.27 schwarze 769: }
1.1 schwarze 770: }
771:
772: /*
773: * Parse a roff node's type from the input buffer. This must be in the
774: * form of ".foo xxx" in the usual way.
775: */
776: static enum rofft
1.16 schwarze 777: roff_parse(struct roff *r, const char *buf, int *pos)
1.1 schwarze 778: {
1.16 schwarze 779: const char *mac;
780: size_t maclen;
1.1 schwarze 781: enum rofft t;
782:
1.39 schwarze 783: if ('\0' == buf[*pos] || '"' == buf[*pos] ||
784: '\t' == buf[*pos] || ' ' == buf[*pos])
1.1 schwarze 785: return(ROFF_MAX);
786:
1.39 schwarze 787: /*
788: * We stop the macro parse at an escape, tab, space, or nil.
789: * However, `\}' is also a valid macro, so make sure we don't
790: * clobber it by seeing the `\' as the end of token.
791: */
792:
1.16 schwarze 793: mac = buf + *pos;
1.39 schwarze 794: maclen = strcspn(mac + 1, " \\\t\0") + 1;
1.1 schwarze 795:
1.16 schwarze 796: t = (r->current_string = roff_getstrn(r, mac, maclen))
1.42 schwarze 797: ? ROFF_USERDEF : roffhash_find(mac, maclen);
1.1 schwarze 798:
1.34 schwarze 799: *pos += (int)maclen;
1.35 schwarze 800:
1.1 schwarze 801: while (buf[*pos] && ' ' == buf[*pos])
802: (*pos)++;
803:
804: return(t);
805: }
806:
807: /* ARGSUSED */
808: static enum rofferr
1.2 schwarze 809: roff_cblock(ROFF_ARGS)
1.1 schwarze 810: {
811:
1.2 schwarze 812: /*
813: * A block-close `..' should only be invoked as a child of an
814: * ignore macro, otherwise raise a warning and just ignore it.
815: */
816:
817: if (NULL == r->last) {
1.35 schwarze 818: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.2 schwarze 819: return(ROFF_IGN);
820: }
1.1 schwarze 821:
1.2 schwarze 822: switch (r->last->tok) {
823: case (ROFF_am):
824: /* FALLTHROUGH */
825: case (ROFF_ami):
826: /* FALLTHROUGH */
827: case (ROFF_am1):
828: /* FALLTHROUGH */
829: case (ROFF_de):
1.23 schwarze 830: /* ROFF_de1 is remapped to ROFF_de in roff_block(). */
1.2 schwarze 831: /* FALLTHROUGH */
832: case (ROFF_dei):
833: /* FALLTHROUGH */
834: case (ROFF_ig):
835: break;
836: default:
1.35 schwarze 837: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.1 schwarze 838: return(ROFF_IGN);
1.2 schwarze 839: }
840:
841: if ((*bufp)[pos])
1.35 schwarze 842: mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.2 schwarze 843:
844: roffnode_pop(r);
845: roffnode_cleanscope(r);
846: return(ROFF_IGN);
847:
848: }
1.1 schwarze 849:
850:
1.2 schwarze 851: static void
852: roffnode_cleanscope(struct roff *r)
853: {
1.1 schwarze 854:
1.2 schwarze 855: while (r->last) {
1.46 schwarze 856: if (--r->last->endspan != 0)
1.2 schwarze 857: break;
858: roffnode_pop(r);
859: }
860: }
1.1 schwarze 861:
862:
1.2 schwarze 863: /* ARGSUSED */
864: static enum rofferr
865: roff_ccond(ROFF_ARGS)
866: {
1.1 schwarze 867:
1.2 schwarze 868: if (NULL == r->last) {
1.35 schwarze 869: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.1 schwarze 870: return(ROFF_IGN);
1.2 schwarze 871: }
1.1 schwarze 872:
1.2 schwarze 873: switch (r->last->tok) {
874: case (ROFF_el):
875: /* FALLTHROUGH */
876: case (ROFF_ie):
877: /* FALLTHROUGH */
878: case (ROFF_if):
879: break;
880: default:
1.35 schwarze 881: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.2 schwarze 882: return(ROFF_IGN);
883: }
1.1 schwarze 884:
1.2 schwarze 885: if (r->last->endspan > -1) {
1.35 schwarze 886: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.1 schwarze 887: return(ROFF_IGN);
1.2 schwarze 888: }
889:
890: if ((*bufp)[pos])
1.35 schwarze 891: mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.1 schwarze 892:
1.2 schwarze 893: roffnode_pop(r);
894: roffnode_cleanscope(r);
1.1 schwarze 895: return(ROFF_IGN);
896: }
897:
898:
899: /* ARGSUSED */
900: static enum rofferr
1.2 schwarze 901: roff_block(ROFF_ARGS)
1.1 schwarze 902: {
1.2 schwarze 903: int sv;
904: size_t sz;
1.16 schwarze 905: char *name;
906:
907: name = NULL;
1.2 schwarze 908:
1.16 schwarze 909: if (ROFF_ig != tok) {
910: if ('\0' == (*bufp)[pos]) {
1.35 schwarze 911: mandoc_msg(MANDOCERR_NOARGS, r->parse, ln, ppos, NULL);
1.16 schwarze 912: return(ROFF_IGN);
913: }
1.22 schwarze 914:
915: /*
916: * Re-write `de1', since we don't really care about
917: * groff's strange compatibility mode, into `de'.
918: */
919:
1.18 schwarze 920: if (ROFF_de1 == tok)
921: tok = ROFF_de;
1.16 schwarze 922: if (ROFF_de == tok)
923: name = *bufp + pos;
1.21 schwarze 924: else
1.35 schwarze 925: mandoc_msg(MANDOCERR_REQUEST, r->parse, ln, ppos,
1.21 schwarze 926: roffs[tok].name);
1.22 schwarze 927:
1.33 schwarze 928: while ((*bufp)[pos] && ! isspace((unsigned char)(*bufp)[pos]))
1.2 schwarze 929: pos++;
1.22 schwarze 930:
1.33 schwarze 931: while (isspace((unsigned char)(*bufp)[pos]))
1.16 schwarze 932: (*bufp)[pos++] = '\0';
1.2 schwarze 933: }
934:
1.16 schwarze 935: roffnode_push(r, tok, name, ln, ppos);
936:
937: /*
938: * At the beginning of a `de' macro, clear the existing string
939: * with the same name, if there is one. New content will be
1.66 ! schwarze 940: * appended from roff_block_text() in multiline mode.
1.16 schwarze 941: */
1.22 schwarze 942:
1.16 schwarze 943: if (ROFF_de == tok)
1.19 schwarze 944: roff_setstr(r, name, "", 0);
1.2 schwarze 945:
946: if ('\0' == (*bufp)[pos])
947: return(ROFF_IGN);
1.1 schwarze 948:
1.22 schwarze 949: /* If present, process the custom end-of-line marker. */
950:
1.2 schwarze 951: sv = pos;
1.33 schwarze 952: while ((*bufp)[pos] && ! isspace((unsigned char)(*bufp)[pos]))
1.2 schwarze 953: pos++;
954:
955: /*
956: * Note: groff does NOT like escape characters in the input.
957: * Instead of detecting this, we're just going to let it fly and
958: * to hell with it.
959: */
960:
961: assert(pos > sv);
962: sz = (size_t)(pos - sv);
963:
964: if (1 == sz && '.' == (*bufp)[sv])
965: return(ROFF_IGN);
966:
1.11 schwarze 967: r->last->end = mandoc_malloc(sz + 1);
1.2 schwarze 968:
969: memcpy(r->last->end, *bufp + sv, sz);
970: r->last->end[(int)sz] = '\0';
971:
972: if ((*bufp)[pos])
1.35 schwarze 973: mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.1 schwarze 974:
975: return(ROFF_IGN);
976: }
977:
978:
979: /* ARGSUSED */
980: static enum rofferr
1.2 schwarze 981: roff_block_sub(ROFF_ARGS)
1.1 schwarze 982: {
1.2 schwarze 983: enum rofft t;
984: int i, j;
985:
986: /*
987: * First check whether a custom macro exists at this level. If
988: * it does, then check against it. This is some of groff's
989: * stranger behaviours. If we encountered a custom end-scope
990: * tag and that tag also happens to be a "real" macro, then we
991: * need to try interpreting it again as a real macro. If it's
992: * not, then return ignore. Else continue.
993: */
994:
995: if (r->last->end) {
1.35 schwarze 996: for (i = pos, j = 0; r->last->end[j]; j++, i++)
1.2 schwarze 997: if ((*bufp)[i] != r->last->end[j])
998: break;
1.1 schwarze 999:
1.2 schwarze 1000: if ('\0' == r->last->end[j] &&
1001: ('\0' == (*bufp)[i] ||
1002: ' ' == (*bufp)[i] ||
1003: '\t' == (*bufp)[i])) {
1004: roffnode_pop(r);
1005: roffnode_cleanscope(r);
1.1 schwarze 1006:
1.35 schwarze 1007: while (' ' == (*bufp)[i] || '\t' == (*bufp)[i])
1008: i++;
1009:
1010: pos = i;
1.16 schwarze 1011: if (ROFF_MAX != roff_parse(r, *bufp, &pos))
1.2 schwarze 1012: return(ROFF_RERUN);
1013: return(ROFF_IGN);
1014: }
1.1 schwarze 1015: }
1016:
1.2 schwarze 1017: /*
1018: * If we have no custom end-query or lookup failed, then try
1019: * pulling it out of the hashtable.
1020: */
1.1 schwarze 1021:
1.36 schwarze 1022: t = roff_parse(r, *bufp, &pos);
1.1 schwarze 1023:
1.16 schwarze 1024: /*
1025: * Macros other than block-end are only significant
1026: * in `de' blocks; elsewhere, simply throw them away.
1027: */
1028: if (ROFF_cblock != t) {
1029: if (ROFF_de == tok)
1.66 ! schwarze 1030: roff_setstr(r, r->last->name, *bufp + ppos, 2);
1.1 schwarze 1031: return(ROFF_IGN);
1.16 schwarze 1032: }
1.1 schwarze 1033:
1.2 schwarze 1034: assert(roffs[t].proc);
1.6 schwarze 1035: return((*roffs[t].proc)(r, t, bufp, szp,
1036: ln, ppos, pos, offs));
1.2 schwarze 1037: }
1038:
1039:
1040: /* ARGSUSED */
1041: static enum rofferr
1042: roff_block_text(ROFF_ARGS)
1043: {
1044:
1.16 schwarze 1045: if (ROFF_de == tok)
1.66 ! schwarze 1046: roff_setstr(r, r->last->name, *bufp + pos, 2);
1.16 schwarze 1047:
1.2 schwarze 1048: return(ROFF_IGN);
1049: }
1050:
1051:
1052: /* ARGSUSED */
1053: static enum rofferr
1054: roff_cond_sub(ROFF_ARGS)
1055: {
1056: enum rofft t;
1057: enum roffrule rr;
1.37 schwarze 1058: char *ep;
1.2 schwarze 1059:
1060: rr = r->last->rule;
1.37 schwarze 1061: roffnode_cleanscope(r);
1.50 schwarze 1062: t = roff_parse(r, *bufp, &pos);
1.2 schwarze 1063:
1.37 schwarze 1064: /*
1.50 schwarze 1065: * Fully handle known macros when they are structurally
1066: * required or when the conditional evaluated to true.
1.5 schwarze 1067: */
1068:
1.50 schwarze 1069: if ((ROFF_MAX != t) &&
1070: (ROFF_ccond == t || ROFFRULE_ALLOW == rr ||
1071: ROFFMAC_STRUCT & roffs[t].flags)) {
1072: assert(roffs[t].proc);
1073: return((*roffs[t].proc)(r, t, bufp, szp,
1074: ln, ppos, pos, offs));
1075: }
1.39 schwarze 1076:
1.50 schwarze 1077: /* Always check for the closing delimiter `\}'. */
1.39 schwarze 1078:
1.50 schwarze 1079: ep = &(*bufp)[pos];
1080: while (NULL != (ep = strchr(ep, '\\'))) {
1081: if ('}' != *(++ep))
1082: continue;
1.2 schwarze 1083:
1.50 schwarze 1084: /*
1085: * If we're at the end of line, then just chop
1086: * off the \} and resize the buffer.
1087: * If we aren't, then convert it to spaces.
1088: */
1.37 schwarze 1089:
1.50 schwarze 1090: if ('\0' == *(ep + 1)) {
1091: *--ep = '\0';
1092: *szp -= 2;
1093: } else
1094: *(ep - 1) = *ep = ' ';
1.2 schwarze 1095:
1.50 schwarze 1096: roff_ccond(r, ROFF_ccond, bufp, szp,
1097: ln, pos, pos + 2, offs);
1098: break;
1099: }
1100: return(ROFFRULE_DENY == rr ? ROFF_IGN : ROFF_CONT);
1.2 schwarze 1101: }
1102:
1103: /* ARGSUSED */
1104: static enum rofferr
1105: roff_cond_text(ROFF_ARGS)
1106: {
1.37 schwarze 1107: char *ep;
1.2 schwarze 1108: enum roffrule rr;
1109:
1110: rr = r->last->rule;
1.37 schwarze 1111: roffnode_cleanscope(r);
1.1 schwarze 1112:
1.37 schwarze 1113: ep = &(*bufp)[pos];
1114: for ( ; NULL != (ep = strchr(ep, '\\')); ep++) {
1115: ep++;
1116: if ('}' != *ep)
1117: continue;
1118: *ep = '&';
1119: roff_ccond(r, ROFF_ccond, bufp, szp,
1120: ln, pos, pos + 2, offs);
1.2 schwarze 1121: }
1122: return(ROFFRULE_DENY == rr ? ROFF_IGN : ROFF_CONT);
1123: }
1124:
1.56 schwarze 1125: static int
1126: roff_getnum(const char *v, int *pos, int *res)
1127: {
1128: int p, n;
1129:
1130: p = *pos;
1131: n = v[p] == '-';
1132: if (n)
1133: p++;
1134:
1135: for (*res = 0; isdigit((unsigned char)v[p]); p++)
1136: *res += 10 * *res + v[p] - '0';
1137: if (p == *pos + n)
1138: return 0;
1139:
1140: if (n)
1141: *res = -*res;
1142:
1143: *pos = p;
1144: return 1;
1145: }
1146:
1147: static int
1148: roff_getop(const char *v, int *pos, char *res)
1149: {
1150: int e;
1151:
1152: *res = v[*pos];
1153: e = v[*pos + 1] == '=';
1154:
1155: switch (*res) {
1156: case '=':
1157: break;
1158: case '>':
1159: if (e)
1160: *res = 'g';
1161: break;
1162: case '<':
1163: if (e)
1164: *res = 'l';
1165: break;
1166: default:
1167: return(0);
1168: }
1169:
1170: *pos += 1 + e;
1171:
1172: return(*res);
1173: }
1174:
1.5 schwarze 1175: static enum roffrule
1176: roff_evalcond(const char *v, int *pos)
1177: {
1.56 schwarze 1178: int not, lh, rh;
1179: char op;
1.5 schwarze 1180:
1181: switch (v[*pos]) {
1182: case ('n'):
1183: (*pos)++;
1184: return(ROFFRULE_ALLOW);
1185: case ('e'):
1186: /* FALLTHROUGH */
1187: case ('o'):
1188: /* FALLTHROUGH */
1189: case ('t'):
1190: (*pos)++;
1191: return(ROFFRULE_DENY);
1.56 schwarze 1192: case ('!'):
1193: (*pos)++;
1194: not = 1;
1195: break;
1.5 schwarze 1196: default:
1.56 schwarze 1197: not = 0;
1.5 schwarze 1198: break;
1199: }
1200:
1.56 schwarze 1201: if (!roff_getnum(v, pos, &lh))
1202: return ROFFRULE_DENY;
1203: if (!roff_getop(v, pos, &op)) {
1204: if (lh < 0)
1205: lh = 0;
1206: goto out;
1207: }
1208: if (!roff_getnum(v, pos, &rh))
1209: return ROFFRULE_DENY;
1210: switch (op) {
1211: case 'g':
1212: lh = lh >= rh;
1213: break;
1214: case 'l':
1215: lh = lh <= rh;
1216: break;
1217: case '=':
1218: lh = lh == rh;
1219: break;
1220: case '>':
1221: lh = lh > rh;
1222: break;
1223: case '<':
1224: lh = lh < rh;
1225: break;
1226: default:
1227: return ROFFRULE_DENY;
1228: }
1229: out:
1230: if (not)
1231: lh = !lh;
1232: return lh ? ROFFRULE_ALLOW : ROFFRULE_DENY;
1.5 schwarze 1233: }
1234:
1.2 schwarze 1235: /* ARGSUSED */
1236: static enum rofferr
1.21 schwarze 1237: roff_line_ignore(ROFF_ARGS)
1.6 schwarze 1238: {
1.30 schwarze 1239:
1.21 schwarze 1240: return(ROFF_IGN);
1241: }
1242:
1243: /* ARGSUSED */
1244: static enum rofferr
1.2 schwarze 1245: roff_cond(ROFF_ARGS)
1246: {
1.46 schwarze 1247:
1248: roffnode_push(r, tok, NULL, ln, ppos);
1.2 schwarze 1249:
1.35 schwarze 1250: /*
1251: * An `.el' has no conditional body: it will consume the value
1252: * of the current rstack entry set in prior `ie' calls or
1253: * defaults to DENY.
1254: *
1255: * If we're not an `el', however, then evaluate the conditional.
1256: */
1.1 schwarze 1257:
1.46 schwarze 1258: r->last->rule = ROFF_el == tok ?
1.35 schwarze 1259: (r->rstackpos < 0 ?
1260: ROFFRULE_DENY : r->rstack[r->rstackpos--]) :
1261: roff_evalcond(*bufp, &pos);
1.2 schwarze 1262:
1.35 schwarze 1263: /*
1264: * An if-else will put the NEGATION of the current evaluated
1265: * conditional into the stack of rules.
1266: */
1267:
1.2 schwarze 1268: if (ROFF_ie == tok) {
1.35 schwarze 1269: if (r->rstackpos == RSTACK_MAX - 1) {
1270: mandoc_msg(MANDOCERR_MEM,
1271: r->parse, ln, ppos, NULL);
1272: return(ROFF_ERR);
1273: }
1274: r->rstack[++r->rstackpos] =
1275: ROFFRULE_DENY == r->last->rule ?
1276: ROFFRULE_ALLOW : ROFFRULE_DENY;
1.2 schwarze 1277: }
1.5 schwarze 1278:
1279: /* If the parent has false as its rule, then so do we. */
1280:
1.2 schwarze 1281: if (r->last->parent && ROFFRULE_DENY == r->last->parent->rule)
1282: r->last->rule = ROFFRULE_DENY;
1.5 schwarze 1283:
1284: /*
1.46 schwarze 1285: * Determine scope.
1286: * If there is nothing on the line after the conditional,
1287: * not even whitespace, use next-line scope.
1.5 schwarze 1288: */
1.2 schwarze 1289:
1.46 schwarze 1290: if ('\0' == (*bufp)[pos]) {
1291: r->last->endspan = 2;
1292: goto out;
1293: }
1294:
1295: while (' ' == (*bufp)[pos])
1296: pos++;
1297:
1298: /* An opening brace requests multiline scope. */
1.2 schwarze 1299:
1300: if ('\\' == (*bufp)[pos] && '{' == (*bufp)[pos + 1]) {
1301: r->last->endspan = -1;
1302: pos += 2;
1.46 schwarze 1303: goto out;
1.2 schwarze 1304: }
1305:
1306: /*
1.46 schwarze 1307: * Anything else following the conditional causes
1308: * single-line scope. Warn if the scope contains
1309: * nothing but trailing whitespace.
1.2 schwarze 1310: */
1311:
1312: if ('\0' == (*bufp)[pos])
1.46 schwarze 1313: mandoc_msg(MANDOCERR_NOARGS, r->parse, ln, ppos, NULL);
1.2 schwarze 1314:
1.46 schwarze 1315: r->last->endspan = 1;
1.1 schwarze 1316:
1.46 schwarze 1317: out:
1.2 schwarze 1318: *offs = pos;
1319: return(ROFF_RERUN);
1.1 schwarze 1320: }
1321:
1322:
1.2 schwarze 1323: /* ARGSUSED */
1324: static enum rofferr
1.7 schwarze 1325: roff_ds(ROFF_ARGS)
1326: {
1.10 schwarze 1327: char *name, *string;
1328:
1329: /*
1330: * A symbol is named by the first word following the macro
1331: * invocation up to a space. Its value is anything after the
1332: * name's trailing whitespace and optional double-quote. Thus,
1333: *
1334: * [.ds foo "bar " ]
1335: *
1336: * will have `bar " ' as its value.
1337: */
1.7 schwarze 1338:
1.28 schwarze 1339: string = *bufp + pos;
1340: name = roff_getname(r, &string, ln, pos);
1.7 schwarze 1341: if ('\0' == *name)
1342: return(ROFF_IGN);
1343:
1.28 schwarze 1344: /* Read past initial double-quote. */
1345: if ('"' == *string)
1.7 schwarze 1346: string++;
1347:
1.10 schwarze 1348: /* The rest is the value. */
1.66 ! schwarze 1349: roff_setstr(r, name, string, ROFF_as == tok);
1.7 schwarze 1350: return(ROFF_IGN);
1351: }
1352:
1.52 schwarze 1353: void
1.60 schwarze 1354: roff_setreg(struct roff *r, const char *name, int val, char sign)
1.41 schwarze 1355: {
1.52 schwarze 1356: struct roffreg *reg;
1357:
1358: /* Search for an existing register with the same name. */
1359: reg = r->regtab;
1360:
1361: while (reg && strcmp(name, reg->key.p))
1362: reg = reg->next;
1.41 schwarze 1363:
1.52 schwarze 1364: if (NULL == reg) {
1365: /* Create a new register. */
1366: reg = mandoc_malloc(sizeof(struct roffreg));
1367: reg->key.p = mandoc_strdup(name);
1368: reg->key.sz = strlen(name);
1.60 schwarze 1369: reg->val = 0;
1.52 schwarze 1370: reg->next = r->regtab;
1371: r->regtab = reg;
1372: }
1373:
1.60 schwarze 1374: if ('+' == sign)
1375: reg->val += val;
1376: else if ('-' == sign)
1377: reg->val -= val;
1378: else
1379: reg->val = val;
1.41 schwarze 1380: }
1381:
1.65 schwarze 1382: /*
1383: * Handle some predefined read-only number registers.
1384: * For now, return -1 if the requested register is not predefined;
1385: * in case a predefined read-only register having the value -1
1386: * were to turn up, another special value would have to be chosen.
1387: */
1388: static int
1389: roff_getregro(const char *name)
1390: {
1391:
1392: switch (*name) {
1393: case ('A'): /* ASCII approximation mode is always off. */
1394: return(0);
1395: case ('g'): /* Groff compatibility mode is always on. */
1396: return(1);
1397: case ('H'): /* Fixed horizontal resolution. */
1398: return (24);
1399: case ('j'): /* Always adjust left margin only. */
1400: return(0);
1401: case ('T'): /* Some output device is always defined. */
1402: return(1);
1403: case ('V'): /* Fixed vertical resolution. */
1404: return (40);
1405: default:
1406: return (-1);
1407: }
1408: }
1409:
1.53 schwarze 1410: int
1.52 schwarze 1411: roff_getreg(const struct roff *r, const char *name)
1.41 schwarze 1412: {
1.52 schwarze 1413: struct roffreg *reg;
1.65 schwarze 1414: int val;
1415:
1416: if ('.' == name[0] && '\0' != name[1] && '\0' == name[2]) {
1417: val = roff_getregro(name + 1);
1418: if (-1 != val)
1419: return (val);
1420: }
1.52 schwarze 1421:
1422: for (reg = r->regtab; reg; reg = reg->next)
1423: if (0 == strcmp(name, reg->key.p))
1.53 schwarze 1424: return(reg->val);
1425:
1426: return(0);
1427: }
1428:
1429: static int
1430: roff_getregn(const struct roff *r, const char *name, size_t len)
1431: {
1432: struct roffreg *reg;
1.65 schwarze 1433: int val;
1434:
1435: if ('.' == name[0] && 2 == len) {
1436: val = roff_getregro(name + 1);
1437: if (-1 != val)
1438: return (val);
1439: }
1.53 schwarze 1440:
1441: for (reg = r->regtab; reg; reg = reg->next)
1442: if (len == reg->key.sz &&
1443: 0 == strncmp(name, reg->key.p, len))
1444: return(reg->val);
1.41 schwarze 1445:
1.52 schwarze 1446: return(0);
1.41 schwarze 1447: }
1448:
1.52 schwarze 1449: static void
1450: roff_freereg(struct roffreg *reg)
1.41 schwarze 1451: {
1.52 schwarze 1452: struct roffreg *old_reg;
1.41 schwarze 1453:
1.52 schwarze 1454: while (NULL != reg) {
1455: free(reg->key.p);
1456: old_reg = reg;
1457: reg = reg->next;
1458: free(old_reg);
1459: }
1.41 schwarze 1460: }
1.7 schwarze 1461:
1462: /* ARGSUSED */
1463: static enum rofferr
1.6 schwarze 1464: roff_nr(ROFF_ARGS)
1.1 schwarze 1465: {
1.28 schwarze 1466: const char *key;
1467: char *val;
1.60 schwarze 1468: size_t sz;
1.37 schwarze 1469: int iv;
1.60 schwarze 1470: char sign;
1.6 schwarze 1471:
1.28 schwarze 1472: val = *bufp + pos;
1473: key = roff_getname(r, &val, ln, pos);
1.6 schwarze 1474:
1.60 schwarze 1475: sign = *val;
1476: if ('+' == sign || '-' == sign)
1477: val++;
1478:
1479: sz = strspn(val, "0123456789");
1480: iv = sz ? mandoc_strntoi(val, sz, 10) : 0;
1.52 schwarze 1481:
1.60 schwarze 1482: roff_setreg(r, key, iv, sign);
1.1 schwarze 1483:
1.29 schwarze 1484: return(ROFF_IGN);
1485: }
1486:
1487: /* ARGSUSED */
1488: static enum rofferr
1489: roff_rm(ROFF_ARGS)
1490: {
1491: const char *name;
1492: char *cp;
1493:
1494: cp = *bufp + pos;
1495: while ('\0' != *cp) {
1.34 schwarze 1496: name = roff_getname(r, &cp, ln, (int)(cp - *bufp));
1.29 schwarze 1497: if ('\0' != *name)
1498: roff_setstr(r, name, NULL, 0);
1499: }
1.51 schwarze 1500: return(ROFF_IGN);
1501: }
1502:
1503: /* ARGSUSED */
1504: static enum rofferr
1505: roff_it(ROFF_ARGS)
1506: {
1507: char *cp;
1508: size_t len;
1509: int iv;
1510:
1511: /* Parse the number of lines. */
1512: cp = *bufp + pos;
1513: len = strcspn(cp, " \t");
1514: cp[len] = '\0';
1515: if ((iv = mandoc_strntoi(cp, len, 10)) <= 0) {
1516: mandoc_msg(MANDOCERR_NUMERIC, r->parse,
1517: ln, ppos, *bufp + 1);
1518: return(ROFF_IGN);
1519: }
1520: cp += len + 1;
1521:
1522: /* Arm the input line trap. */
1523: roffit_lines = iv;
1524: roffit_macro = mandoc_strdup(cp);
1.2 schwarze 1525: return(ROFF_IGN);
1.47 schwarze 1526: }
1527:
1528: /* ARGSUSED */
1529: static enum rofferr
1530: roff_Dd(ROFF_ARGS)
1531: {
1532: const char *const *cp;
1533:
1.63 schwarze 1534: if (0 == r->quick && MPARSE_MDOC != r->parsetype)
1.47 schwarze 1535: for (cp = __mdoc_reserved; *cp; cp++)
1536: roff_setstr(r, *cp, NULL, 0);
1537:
1538: return(ROFF_CONT);
1539: }
1540:
1541: /* ARGSUSED */
1542: static enum rofferr
1543: roff_TH(ROFF_ARGS)
1544: {
1545: const char *const *cp;
1546:
1.63 schwarze 1547: if (0 == r->quick && MPARSE_MDOC != r->parsetype)
1.47 schwarze 1548: for (cp = __man_reserved; *cp; cp++)
1549: roff_setstr(r, *cp, NULL, 0);
1550:
1551: return(ROFF_CONT);
1.14 schwarze 1552: }
1553:
1554: /* ARGSUSED */
1555: static enum rofferr
1.27 schwarze 1556: roff_TE(ROFF_ARGS)
1557: {
1558:
1559: if (NULL == r->tbl)
1.35 schwarze 1560: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.27 schwarze 1561: else
1.41 schwarze 1562: tbl_end(&r->tbl);
1.27 schwarze 1563:
1564: return(ROFF_IGN);
1565: }
1566:
1567: /* ARGSUSED */
1568: static enum rofferr
1569: roff_T_(ROFF_ARGS)
1570: {
1571:
1572: if (NULL == r->tbl)
1.35 schwarze 1573: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.27 schwarze 1574: else
1575: tbl_restart(ppos, ln, r->tbl);
1576:
1577: return(ROFF_IGN);
1578: }
1579:
1.41 schwarze 1580: #if 0
1581: static int
1582: roff_closeeqn(struct roff *r)
1583: {
1584:
1585: return(r->eqn && ROFF_EQN == eqn_end(&r->eqn) ? 1 : 0);
1586: }
1587: #endif
1588:
1589: static void
1590: roff_openeqn(struct roff *r, const char *name, int line,
1591: int offs, const char *buf)
1.32 schwarze 1592: {
1.41 schwarze 1593: struct eqn_node *e;
1594: int poff;
1.32 schwarze 1595:
1596: assert(NULL == r->eqn);
1.41 schwarze 1597: e = eqn_alloc(name, offs, line, r->parse);
1.32 schwarze 1598:
1599: if (r->last_eqn)
1600: r->last_eqn->next = e;
1601: else
1602: r->first_eqn = r->last_eqn = e;
1603:
1604: r->eqn = r->last_eqn = e;
1.41 schwarze 1605:
1606: if (buf) {
1607: poff = 0;
1608: eqn_read(&r->eqn, line, buf, offs, &poff);
1609: }
1610: }
1611:
1612: /* ARGSUSED */
1613: static enum rofferr
1614: roff_EQ(ROFF_ARGS)
1615: {
1616:
1617: roff_openeqn(r, *bufp + pos, ln, ppos, NULL);
1.32 schwarze 1618: return(ROFF_IGN);
1619: }
1620:
1621: /* ARGSUSED */
1622: static enum rofferr
1623: roff_EN(ROFF_ARGS)
1624: {
1625:
1.35 schwarze 1626: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.32 schwarze 1627: return(ROFF_IGN);
1628: }
1629:
1630: /* ARGSUSED */
1631: static enum rofferr
1.27 schwarze 1632: roff_TS(ROFF_ARGS)
1633: {
1.49 schwarze 1634: struct tbl_node *tbl;
1.27 schwarze 1635:
1636: if (r->tbl) {
1.35 schwarze 1637: mandoc_msg(MANDOCERR_SCOPEBROKEN, r->parse, ln, ppos, NULL);
1.41 schwarze 1638: tbl_end(&r->tbl);
1.27 schwarze 1639: }
1640:
1.49 schwarze 1641: tbl = tbl_alloc(ppos, ln, r->parse);
1.27 schwarze 1642:
1643: if (r->last_tbl)
1.49 schwarze 1644: r->last_tbl->next = tbl;
1.27 schwarze 1645: else
1.49 schwarze 1646: r->first_tbl = r->last_tbl = tbl;
1.27 schwarze 1647:
1.49 schwarze 1648: r->tbl = r->last_tbl = tbl;
1.27 schwarze 1649: return(ROFF_IGN);
1650: }
1651:
1652: /* ARGSUSED */
1653: static enum rofferr
1.48 schwarze 1654: roff_cc(ROFF_ARGS)
1655: {
1656: const char *p;
1657:
1658: p = *bufp + pos;
1659:
1660: if ('\0' == *p || '.' == (r->control = *p++))
1661: r->control = 0;
1662:
1663: if ('\0' != *p)
1664: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
1665:
1666: return(ROFF_IGN);
1667: }
1668:
1669: /* ARGSUSED */
1670: static enum rofferr
1.42 schwarze 1671: roff_tr(ROFF_ARGS)
1672: {
1673: const char *p, *first, *second;
1674: size_t fsz, ssz;
1675: enum mandoc_esc esc;
1676:
1677: p = *bufp + pos;
1678:
1679: if ('\0' == *p) {
1680: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
1681: return(ROFF_IGN);
1682: }
1683:
1684: while ('\0' != *p) {
1685: fsz = ssz = 1;
1686:
1687: first = p++;
1688: if ('\\' == *first) {
1689: esc = mandoc_escape(&p, NULL, NULL);
1690: if (ESCAPE_ERROR == esc) {
1691: mandoc_msg
1692: (MANDOCERR_BADESCAPE, r->parse,
1693: ln, (int)(p - *bufp), NULL);
1694: return(ROFF_IGN);
1695: }
1696: fsz = (size_t)(p - first);
1697: }
1698:
1699: second = p++;
1700: if ('\\' == *second) {
1701: esc = mandoc_escape(&p, NULL, NULL);
1702: if (ESCAPE_ERROR == esc) {
1703: mandoc_msg
1704: (MANDOCERR_BADESCAPE, r->parse,
1705: ln, (int)(p - *bufp), NULL);
1706: return(ROFF_IGN);
1707: }
1708: ssz = (size_t)(p - second);
1709: } else if ('\0' == *second) {
1710: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse,
1711: ln, (int)(p - *bufp), NULL);
1712: second = " ";
1713: p--;
1714: }
1715:
1716: if (fsz > 1) {
1717: roff_setstrn(&r->xmbtab, first,
1718: fsz, second, ssz, 0);
1719: continue;
1720: }
1721:
1722: if (NULL == r->xtab)
1723: r->xtab = mandoc_calloc
1724: (128, sizeof(struct roffstr));
1725:
1726: free(r->xtab[(int)*first].p);
1727: r->xtab[(int)*first].p = mandoc_strndup(second, ssz);
1728: r->xtab[(int)*first].sz = ssz;
1729: }
1730:
1731: return(ROFF_IGN);
1732: }
1733:
1734: /* ARGSUSED */
1735: static enum rofferr
1.14 schwarze 1736: roff_so(ROFF_ARGS)
1737: {
1738: char *name;
1.15 schwarze 1739:
1.35 schwarze 1740: mandoc_msg(MANDOCERR_SO, r->parse, ln, ppos, NULL);
1.14 schwarze 1741:
1.22 schwarze 1742: /*
1743: * Handle `so'. Be EXTREMELY careful, as we shouldn't be
1744: * opening anything that's not in our cwd or anything beneath
1745: * it. Thus, explicitly disallow traversing up the file-system
1746: * or using absolute paths.
1747: */
1748:
1.14 schwarze 1749: name = *bufp + pos;
1750: if ('/' == *name || strstr(name, "../") || strstr(name, "/..")) {
1.35 schwarze 1751: mandoc_msg(MANDOCERR_SOPATH, r->parse, ln, pos, NULL);
1.14 schwarze 1752: return(ROFF_ERR);
1753: }
1754:
1755: *offs = pos;
1756: return(ROFF_SO);
1.7 schwarze 1757: }
1758:
1.16 schwarze 1759: /* ARGSUSED */
1760: static enum rofferr
1761: roff_userdef(ROFF_ARGS)
1.12 schwarze 1762: {
1.16 schwarze 1763: const char *arg[9];
1764: char *cp, *n1, *n2;
1.25 schwarze 1765: int i;
1.12 schwarze 1766:
1.16 schwarze 1767: /*
1768: * Collect pointers to macro argument strings
1.61 schwarze 1769: * and NUL-terminate them.
1.16 schwarze 1770: */
1771: cp = *bufp + pos;
1.25 schwarze 1772: for (i = 0; i < 9; i++)
1.26 schwarze 1773: arg[i] = '\0' == *cp ? "" :
1.35 schwarze 1774: mandoc_getarg(r->parse, &cp, ln, &pos);
1.16 schwarze 1775:
1776: /*
1777: * Expand macro arguments.
1.12 schwarze 1778: */
1.16 schwarze 1779: *szp = 0;
1780: n1 = cp = mandoc_strdup(r->current_string);
1781: while (NULL != (cp = strstr(cp, "\\$"))) {
1782: i = cp[2] - '1';
1783: if (0 > i || 8 < i) {
1784: /* Not an argument invocation. */
1785: cp += 2;
1786: continue;
1787: }
1788:
1789: *szp = strlen(n1) - 3 + strlen(arg[i]) + 1;
1790: n2 = mandoc_malloc(*szp);
1791:
1792: strlcpy(n2, n1, (size_t)(cp - n1 + 1));
1793: strlcat(n2, arg[i], *szp);
1794: strlcat(n2, cp + 3, *szp);
1795:
1796: cp = n2 + (cp - n1);
1797: free(n1);
1798: n1 = n2;
1.12 schwarze 1799: }
1800:
1.16 schwarze 1801: /*
1802: * Replace the macro invocation
1803: * by the expanded macro.
1804: */
1805: free(*bufp);
1806: *bufp = n1;
1807: if (0 == *szp)
1808: *szp = strlen(*bufp) + 1;
1809:
1.19 schwarze 1810: return(*szp > 1 && '\n' == (*bufp)[(int)*szp - 2] ?
1.16 schwarze 1811: ROFF_REPARSE : ROFF_APPEND);
1.12 schwarze 1812: }
1.28 schwarze 1813:
1814: static char *
1815: roff_getname(struct roff *r, char **cpp, int ln, int pos)
1816: {
1817: char *name, *cp;
1818:
1819: name = *cpp;
1820: if ('\0' == *name)
1821: return(name);
1822:
1823: /* Read until end of name. */
1824: for (cp = name; '\0' != *cp && ' ' != *cp; cp++) {
1825: if ('\\' != *cp)
1826: continue;
1827: cp++;
1828: if ('\\' == *cp)
1829: continue;
1.35 schwarze 1830: mandoc_msg(MANDOCERR_NAMESC, r->parse, ln, pos, NULL);
1.28 schwarze 1831: *cp = '\0';
1832: name = cp;
1833: }
1834:
1835: /* Nil-terminate name. */
1836: if ('\0' != *cp)
1837: *(cp++) = '\0';
1838:
1839: /* Read past spaces. */
1840: while (' ' == *cp)
1841: cp++;
1842:
1843: *cpp = cp;
1844: return(name);
1845: }
1846:
1.16 schwarze 1847: /*
1848: * Store *string into the user-defined string called *name.
1849: * To clear an existing entry, call with (*r, *name, NULL, 0).
1.66 ! schwarze 1850: * append == 0: replace mode
! 1851: * append == 1: single-line append mode
! 1852: * append == 2: multiline append mode, append '\n' after each call
1.16 schwarze 1853: */
1.8 schwarze 1854: static void
1.16 schwarze 1855: roff_setstr(struct roff *r, const char *name, const char *string,
1.66 ! schwarze 1856: int append)
1.7 schwarze 1857: {
1.42 schwarze 1858:
1859: roff_setstrn(&r->strtab, name, strlen(name), string,
1.66 ! schwarze 1860: string ? strlen(string) : 0, append);
1.42 schwarze 1861: }
1862:
1863: static void
1864: roff_setstrn(struct roffkv **r, const char *name, size_t namesz,
1.66 ! schwarze 1865: const char *string, size_t stringsz, int append)
1.42 schwarze 1866: {
1867: struct roffkv *n;
1868: char *c;
1869: int i;
1870: size_t oldch, newch;
1.7 schwarze 1871:
1.16 schwarze 1872: /* Search for an existing string with the same name. */
1.42 schwarze 1873: n = *r;
1874:
1875: while (n && strcmp(name, n->key.p))
1.7 schwarze 1876: n = n->next;
1.8 schwarze 1877:
1878: if (NULL == n) {
1.16 schwarze 1879: /* Create a new string table entry. */
1.42 schwarze 1880: n = mandoc_malloc(sizeof(struct roffkv));
1881: n->key.p = mandoc_strndup(name, namesz);
1882: n->key.sz = namesz;
1883: n->val.p = NULL;
1884: n->val.sz = 0;
1885: n->next = *r;
1886: *r = n;
1.66 ! schwarze 1887: } else if (0 == append) {
1.42 schwarze 1888: free(n->val.p);
1889: n->val.p = NULL;
1890: n->val.sz = 0;
1.16 schwarze 1891: }
1892:
1893: if (NULL == string)
1894: return;
1895:
1896: /*
1897: * One additional byte for the '\n' in multiline mode,
1898: * and one for the terminating '\0'.
1899: */
1.66 ! schwarze 1900: newch = stringsz + (1 < append ? 2u : 1u);
1.42 schwarze 1901:
1902: if (NULL == n->val.p) {
1903: n->val.p = mandoc_malloc(newch);
1904: *n->val.p = '\0';
1.16 schwarze 1905: oldch = 0;
1906: } else {
1.42 schwarze 1907: oldch = n->val.sz;
1908: n->val.p = mandoc_realloc(n->val.p, oldch + newch);
1.16 schwarze 1909: }
1910:
1911: /* Skip existing content in the destination buffer. */
1.42 schwarze 1912: c = n->val.p + (int)oldch;
1.16 schwarze 1913:
1914: /* Append new content to the destination buffer. */
1.42 schwarze 1915: i = 0;
1916: while (i < (int)stringsz) {
1.16 schwarze 1917: /*
1918: * Rudimentary roff copy mode:
1919: * Handle escaped backslashes.
1920: */
1.42 schwarze 1921: if ('\\' == string[i] && '\\' == string[i + 1])
1922: i++;
1923: *c++ = string[i++];
1.16 schwarze 1924: }
1.8 schwarze 1925:
1.16 schwarze 1926: /* Append terminating bytes. */
1.66 ! schwarze 1927: if (1 < append)
1.16 schwarze 1928: *c++ = '\n';
1.42 schwarze 1929:
1.16 schwarze 1930: *c = '\0';
1.42 schwarze 1931: n->val.sz = (int)(c - n->val.p);
1.7 schwarze 1932: }
1933:
1.8 schwarze 1934: static const char *
1935: roff_getstrn(const struct roff *r, const char *name, size_t len)
1.7 schwarze 1936: {
1.42 schwarze 1937: const struct roffkv *n;
1.64 schwarze 1938: int i;
1.7 schwarze 1939:
1.42 schwarze 1940: for (n = r->strtab; n; n = n->next)
1941: if (0 == strncmp(name, n->key.p, len) &&
1942: '\0' == n->key.p[(int)len])
1943: return(n->val.p);
1.64 schwarze 1944:
1945: for (i = 0; i < PREDEFS_MAX; i++)
1946: if (0 == strncmp(name, predefs[i].name, len) &&
1947: '\0' == predefs[i].name[(int)len])
1948: return(predefs[i].str);
1.8 schwarze 1949:
1.42 schwarze 1950: return(NULL);
1.7 schwarze 1951: }
1952:
1.8 schwarze 1953: static void
1.42 schwarze 1954: roff_freestr(struct roffkv *r)
1.7 schwarze 1955: {
1.42 schwarze 1956: struct roffkv *n, *nn;
1.7 schwarze 1957:
1.42 schwarze 1958: for (n = r; n; n = nn) {
1959: free(n->key.p);
1960: free(n->val.p);
1.7 schwarze 1961: nn = n->next;
1962: free(n);
1963: }
1.27 schwarze 1964: }
1965:
1966: const struct tbl_span *
1967: roff_span(const struct roff *r)
1968: {
1969:
1970: return(r->tbl ? tbl_span(r->tbl) : NULL);
1.32 schwarze 1971: }
1972:
1973: const struct eqn *
1974: roff_eqn(const struct roff *r)
1975: {
1976:
1977: return(r->last_eqn ? &r->last_eqn->eqn : NULL);
1.42 schwarze 1978: }
1979:
1980: /*
1981: * Duplicate an input string, making the appropriate character
1982: * conversations (as stipulated by `tr') along the way.
1983: * Returns a heap-allocated string with all the replacements made.
1984: */
1985: char *
1986: roff_strdup(const struct roff *r, const char *p)
1987: {
1988: const struct roffkv *cp;
1989: char *res;
1990: const char *pp;
1991: size_t ssz, sz;
1992: enum mandoc_esc esc;
1993:
1994: if (NULL == r->xmbtab && NULL == r->xtab)
1995: return(mandoc_strdup(p));
1996: else if ('\0' == *p)
1997: return(mandoc_strdup(""));
1998:
1999: /*
2000: * Step through each character looking for term matches
2001: * (remember that a `tr' can be invoked with an escape, which is
2002: * a glyph but the escape is multi-character).
2003: * We only do this if the character hash has been initialised
2004: * and the string is >0 length.
2005: */
2006:
2007: res = NULL;
2008: ssz = 0;
2009:
2010: while ('\0' != *p) {
2011: if ('\\' != *p && r->xtab && r->xtab[(int)*p].p) {
2012: sz = r->xtab[(int)*p].sz;
2013: res = mandoc_realloc(res, ssz + sz + 1);
2014: memcpy(res + ssz, r->xtab[(int)*p].p, sz);
2015: ssz += sz;
2016: p++;
2017: continue;
2018: } else if ('\\' != *p) {
2019: res = mandoc_realloc(res, ssz + 2);
2020: res[ssz++] = *p++;
2021: continue;
2022: }
2023:
2024: /* Search for term matches. */
2025: for (cp = r->xmbtab; cp; cp = cp->next)
2026: if (0 == strncmp(p, cp->key.p, cp->key.sz))
2027: break;
2028:
2029: if (NULL != cp) {
2030: /*
2031: * A match has been found.
2032: * Append the match to the array and move
2033: * forward by its keysize.
2034: */
2035: res = mandoc_realloc
2036: (res, ssz + cp->val.sz + 1);
2037: memcpy(res + ssz, cp->val.p, cp->val.sz);
2038: ssz += cp->val.sz;
2039: p += (int)cp->key.sz;
2040: continue;
2041: }
2042:
2043: /*
2044: * Handle escapes carefully: we need to copy
2045: * over just the escape itself, or else we might
2046: * do replacements within the escape itself.
2047: * Make sure to pass along the bogus string.
2048: */
2049: pp = p++;
2050: esc = mandoc_escape(&p, NULL, NULL);
2051: if (ESCAPE_ERROR == esc) {
2052: sz = strlen(pp);
2053: res = mandoc_realloc(res, ssz + sz + 1);
2054: memcpy(res + ssz, pp, sz);
2055: break;
2056: }
2057: /*
2058: * We bail out on bad escapes.
2059: * No need to warn: we already did so when
2060: * roff_res() was called.
2061: */
2062: sz = (int)(p - pp);
2063: res = mandoc_realloc(res, ssz + sz + 1);
2064: memcpy(res + ssz, pp, sz);
2065: ssz += sz;
2066: }
2067:
2068: res[(int)ssz] = '\0';
2069: return(res);
1.48 schwarze 2070: }
2071:
2072: /*
2073: * Find out whether a line is a macro line or not.
2074: * If it is, adjust the current position and return one; if it isn't,
2075: * return zero and don't change the current position.
2076: * If the control character has been set with `.cc', then let that grain
2077: * precedence.
2078: * This is slighly contrary to groff, where using the non-breaking
2079: * control character when `cc' has been invoked will cause the
2080: * non-breaking macro contents to be printed verbatim.
2081: */
2082: int
2083: roff_getcontrol(const struct roff *r, const char *cp, int *ppos)
2084: {
2085: int pos;
2086:
2087: pos = *ppos;
2088:
2089: if (0 != r->control && cp[pos] == r->control)
2090: pos++;
2091: else if (0 != r->control)
2092: return(0);
2093: else if ('\\' == cp[pos] && '.' == cp[pos + 1])
2094: pos += 2;
2095: else if ('.' == cp[pos] || '\'' == cp[pos])
2096: pos++;
2097: else
2098: return(0);
2099:
2100: while (' ' == cp[pos] || '\t' == cp[pos])
2101: pos++;
2102:
2103: *ppos = pos;
2104: return(1);
1.1 schwarze 2105: }
![[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