Annotation of src/usr.bin/mandoc/roff.c, Revision 1.92
1.92 ! schwarze 1: /* $Id: roff.c,v 1.91 2014/07/06 18:46:51 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"
1.74 schwarze 25: #include "mandoc_aux.h"
1.27 schwarze 26: #include "libroff.h"
1.8 schwarze 27: #include "libmandoc.h"
1.1 schwarze 28:
1.37 schwarze 29: /* Maximum number of nested if-else conditionals. */
1.2 schwarze 30: #define RSTACK_MAX 128
31:
1.43 schwarze 32: /* Maximum number of string expansions per line, to break infinite loops. */
33: #define EXPAND_LIMIT 1000
34:
1.1 schwarze 35: enum rofft {
1.20 schwarze 36: ROFF_ad,
1.2 schwarze 37: ROFF_am,
38: ROFF_ami,
39: ROFF_am1,
1.66 schwarze 40: ROFF_as,
1.48 schwarze 41: ROFF_cc,
1.67 schwarze 42: ROFF_ce,
1.1 schwarze 43: ROFF_de,
44: ROFF_dei,
1.2 schwarze 45: ROFF_de1,
46: ROFF_ds,
47: ROFF_el,
1.58 schwarze 48: ROFF_fam,
1.59 schwarze 49: ROFF_hw,
1.20 schwarze 50: ROFF_hy,
1.2 schwarze 51: ROFF_ie,
52: ROFF_if,
1.1 schwarze 53: ROFF_ig,
1.30 schwarze 54: ROFF_it,
1.20 schwarze 55: ROFF_ne,
56: ROFF_nh,
1.14 schwarze 57: ROFF_nr,
1.31 schwarze 58: ROFF_ns,
59: ROFF_ps,
1.2 schwarze 60: ROFF_rm,
1.76 schwarze 61: ROFF_rr,
1.14 schwarze 62: ROFF_so,
1.31 schwarze 63: ROFF_ta,
1.2 schwarze 64: ROFF_tr,
1.47 schwarze 65: ROFF_Dd,
66: ROFF_TH,
1.27 schwarze 67: ROFF_TS,
68: ROFF_TE,
69: ROFF_T_,
1.32 schwarze 70: ROFF_EQ,
71: ROFF_EN,
1.2 schwarze 72: ROFF_cblock,
1.16 schwarze 73: ROFF_USERDEF,
1.1 schwarze 74: ROFF_MAX
75: };
76:
1.41 schwarze 77: /*
1.42 schwarze 78: * An incredibly-simple string buffer.
79: */
1.8 schwarze 80: struct roffstr {
1.42 schwarze 81: char *p; /* nil-terminated buffer */
82: size_t sz; /* saved strlen(p) */
83: };
84:
85: /*
86: * A key-value roffstr pair as part of a singly-linked list.
87: */
88: struct roffkv {
89: struct roffstr key;
90: struct roffstr val;
91: struct roffkv *next; /* next in list */
1.8 schwarze 92: };
93:
1.52 schwarze 94: /*
95: * A single number register as part of a singly-linked list.
96: */
97: struct roffreg {
98: struct roffstr key;
1.53 schwarze 99: int val;
1.52 schwarze 100: struct roffreg *next;
101: };
102:
1.1 schwarze 103: struct roff {
1.35 schwarze 104: struct mparse *parse; /* parse point */
1.72 schwarze 105: int options; /* parse options */
1.1 schwarze 106: struct roffnode *last; /* leaf of stack */
1.71 schwarze 107: int rstack[RSTACK_MAX]; /* stack of !`ie' rules */
1.48 schwarze 108: char control; /* control character */
1.2 schwarze 109: int rstackpos; /* position in rstack */
1.52 schwarze 110: struct roffreg *regtab; /* number registers */
1.42 schwarze 111: struct roffkv *strtab; /* user-defined strings & macros */
112: struct roffkv *xmbtab; /* multi-byte trans table (`tr') */
113: struct roffstr *xtab; /* single-byte trans table (`tr') */
1.16 schwarze 114: const char *current_string; /* value of last called user macro */
1.27 schwarze 115: struct tbl_node *first_tbl; /* first table parsed */
116: struct tbl_node *last_tbl; /* last table parsed */
117: struct tbl_node *tbl; /* current table being parsed */
1.32 schwarze 118: struct eqn_node *last_eqn; /* last equation parsed */
119: struct eqn_node *first_eqn; /* first equation parsed */
120: struct eqn_node *eqn; /* current equation being parsed */
1.1 schwarze 121: };
122:
123: struct roffnode {
124: enum rofft tok; /* type of node */
125: struct roffnode *parent; /* up one in stack */
126: int line; /* parse line */
127: int col; /* parse col */
1.16 schwarze 128: char *name; /* node name, e.g. macro name */
1.2 schwarze 129: char *end; /* end-rules: custom token */
130: int endspan; /* end-rules: next-line or infty */
1.71 schwarze 131: int rule; /* current evaluation rule */
1.1 schwarze 132: };
133:
134: #define ROFF_ARGS struct roff *r, /* parse ctx */ \
135: enum rofft tok, /* tok of macro */ \
1.80 schwarze 136: char **bufp, /* input buffer */ \
1.1 schwarze 137: size_t *szp, /* size of input buffer */ \
138: int ln, /* parse line */ \
1.2 schwarze 139: int ppos, /* original pos in buffer */ \
140: int pos, /* current pos in buffer */ \
141: int *offs /* reset offset of buffer data */
1.1 schwarze 142:
143: typedef enum rofferr (*roffproc)(ROFF_ARGS);
144:
145: struct roffmac {
146: const char *name; /* macro name */
1.2 schwarze 147: roffproc proc; /* process new macro */
148: roffproc text; /* process as child text of macro */
149: roffproc sub; /* process as child of macro */
150: int flags;
151: #define ROFFMAC_STRUCT (1 << 0) /* always interpret */
1.3 schwarze 152: struct roffmac *next;
1.1 schwarze 153: };
154:
1.37 schwarze 155: struct predef {
156: const char *name; /* predefined input name */
157: const char *str; /* replacement symbol */
158: };
159:
160: #define PREDEF(__name, __str) \
161: { (__name), (__str) },
162:
1.42 schwarze 163: static enum rofft roffhash_find(const char *, size_t);
164: static void roffhash_init(void);
165: static void roffnode_cleanscope(struct roff *);
166: static void roffnode_pop(struct roff *);
167: static void roffnode_push(struct roff *, enum rofft,
168: const char *, int, int);
1.2 schwarze 169: static enum rofferr roff_block(ROFF_ARGS);
170: static enum rofferr roff_block_text(ROFF_ARGS);
171: static enum rofferr roff_block_sub(ROFF_ARGS);
172: static enum rofferr roff_cblock(ROFF_ARGS);
1.48 schwarze 173: static enum rofferr roff_cc(ROFF_ARGS);
1.68 schwarze 174: static void roff_ccond(struct roff *, int, int);
1.2 schwarze 175: static enum rofferr roff_cond(ROFF_ARGS);
176: static enum rofferr roff_cond_text(ROFF_ARGS);
177: static enum rofferr roff_cond_sub(ROFF_ARGS);
1.7 schwarze 178: static enum rofferr roff_ds(ROFF_ARGS);
1.71 schwarze 179: static int roff_evalcond(const char *, int *);
1.77 schwarze 180: static int roff_evalnum(const char *, int *, int *, int);
181: static int roff_evalpar(const char *, int *, int *);
1.71 schwarze 182: static int roff_evalstrcond(const char *, int *);
1.42 schwarze 183: static void roff_free1(struct roff *);
1.52 schwarze 184: static void roff_freereg(struct roffreg *);
1.42 schwarze 185: static void roff_freestr(struct roffkv *);
1.85 schwarze 186: static size_t roff_getname(struct roff *, char **, int, int);
1.56 schwarze 187: static int roff_getnum(const char *, int *, int *);
188: static int roff_getop(const char *, int *, char *);
1.53 schwarze 189: static int roff_getregn(const struct roff *,
190: const char *, size_t);
1.65 schwarze 191: static int roff_getregro(const char *name);
1.80 schwarze 192: static const char *roff_getstrn(const struct roff *,
1.8 schwarze 193: const char *, size_t);
1.51 schwarze 194: static enum rofferr roff_it(ROFF_ARGS);
1.21 schwarze 195: static enum rofferr roff_line_ignore(ROFF_ARGS);
1.6 schwarze 196: static enum rofferr roff_nr(ROFF_ARGS);
1.41 schwarze 197: static void roff_openeqn(struct roff *, const char *,
198: int, int, const char *);
1.87 schwarze 199: static enum rofft roff_parse(struct roff *, char *, int *,
200: int, int);
1.51 schwarze 201: static enum rofferr roff_parsetext(char **, size_t *, int, int *);
1.80 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);
1.76 schwarze 205: static enum rofferr roff_rr(ROFF_ARGS);
1.8 schwarze 206: static void roff_setstr(struct roff *,
1.16 schwarze 207: const char *, const char *, int);
1.80 schwarze 208: static void roff_setstrn(struct roffkv **, const char *,
1.42 schwarze 209: size_t, const char *, size_t, int);
1.14 schwarze 210: static enum rofferr roff_so(ROFF_ARGS);
1.42 schwarze 211: static enum rofferr roff_tr(ROFF_ARGS);
1.47 schwarze 212: static enum rofferr roff_Dd(ROFF_ARGS);
213: static enum rofferr roff_TH(ROFF_ARGS);
1.27 schwarze 214: static enum rofferr roff_TE(ROFF_ARGS);
215: static enum rofferr roff_TS(ROFF_ARGS);
1.32 schwarze 216: static enum rofferr roff_EQ(ROFF_ARGS);
217: static enum rofferr roff_EN(ROFF_ARGS);
1.27 schwarze 218: static enum rofferr roff_T_(ROFF_ARGS);
1.16 schwarze 219: static enum rofferr roff_userdef(ROFF_ARGS);
1.1 schwarze 220:
1.42 schwarze 221: /* See roffhash_find() */
1.3 schwarze 222:
223: #define ASCII_HI 126
224: #define ASCII_LO 33
225: #define HASHWIDTH (ASCII_HI - ASCII_LO + 1)
226:
227: static struct roffmac *hash[HASHWIDTH];
228:
229: static struct roffmac roffs[ROFF_MAX] = {
1.21 schwarze 230: { "ad", roff_line_ignore, NULL, NULL, 0, NULL },
1.3 schwarze 231: { "am", roff_block, roff_block_text, roff_block_sub, 0, NULL },
232: { "ami", roff_block, roff_block_text, roff_block_sub, 0, NULL },
233: { "am1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.66 schwarze 234: { "as", roff_ds, NULL, NULL, 0, NULL },
1.48 schwarze 235: { "cc", roff_cc, NULL, NULL, 0, NULL },
1.67 schwarze 236: { "ce", roff_line_ignore, NULL, NULL, 0, NULL },
1.3 schwarze 237: { "de", roff_block, roff_block_text, roff_block_sub, 0, NULL },
238: { "dei", roff_block, roff_block_text, roff_block_sub, 0, NULL },
239: { "de1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.7 schwarze 240: { "ds", roff_ds, NULL, NULL, 0, NULL },
1.3 schwarze 241: { "el", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
1.58 schwarze 242: { "fam", roff_line_ignore, NULL, NULL, 0, NULL },
1.59 schwarze 243: { "hw", roff_line_ignore, NULL, NULL, 0, NULL },
1.21 schwarze 244: { "hy", roff_line_ignore, NULL, NULL, 0, NULL },
1.3 schwarze 245: { "ie", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
246: { "if", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
247: { "ig", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.51 schwarze 248: { "it", roff_it, NULL, NULL, 0, NULL },
1.21 schwarze 249: { "ne", roff_line_ignore, NULL, NULL, 0, NULL },
250: { "nh", roff_line_ignore, NULL, NULL, 0, NULL },
1.14 schwarze 251: { "nr", roff_nr, NULL, NULL, 0, NULL },
1.31 schwarze 252: { "ns", roff_line_ignore, NULL, NULL, 0, NULL },
253: { "ps", roff_line_ignore, NULL, NULL, 0, NULL },
1.29 schwarze 254: { "rm", roff_rm, NULL, NULL, 0, NULL },
1.76 schwarze 255: { "rr", roff_rr, NULL, NULL, 0, NULL },
1.14 schwarze 256: { "so", roff_so, NULL, NULL, 0, NULL },
1.31 schwarze 257: { "ta", roff_line_ignore, NULL, NULL, 0, NULL },
1.42 schwarze 258: { "tr", roff_tr, NULL, NULL, 0, NULL },
1.47 schwarze 259: { "Dd", roff_Dd, NULL, NULL, 0, NULL },
260: { "TH", roff_TH, NULL, NULL, 0, NULL },
1.27 schwarze 261: { "TS", roff_TS, NULL, NULL, 0, NULL },
262: { "TE", roff_TE, NULL, NULL, 0, NULL },
263: { "T&", roff_T_, NULL, NULL, 0, NULL },
1.32 schwarze 264: { "EQ", roff_EQ, NULL, NULL, 0, NULL },
265: { "EN", roff_EN, NULL, NULL, 0, NULL },
1.3 schwarze 266: { ".", roff_cblock, NULL, NULL, 0, NULL },
1.16 schwarze 267: { NULL, roff_userdef, NULL, NULL, 0, NULL },
1.1 schwarze 268: };
269:
1.73 schwarze 270: /* not currently implemented: Ds em Eq LP Me PP pp Or Rd Sf SH */
1.47 schwarze 271: const char *const __mdoc_reserved[] = {
272: "Ac", "Ad", "An", "Ao", "Ap", "Aq", "Ar", "At",
273: "Bc", "Bd", "Bf", "Bk", "Bl", "Bo", "Bq",
274: "Brc", "Bro", "Brq", "Bsx", "Bt", "Bx",
275: "Cd", "Cm", "Db", "Dc", "Dd", "Dl", "Do", "Dq",
1.73 schwarze 276: "Dt", "Dv", "Dx", "D1",
277: "Ec", "Ed", "Ef", "Ek", "El", "Em",
278: "En", "Eo", "Er", "Es", "Ev", "Ex",
1.47 schwarze 279: "Fa", "Fc", "Fd", "Fl", "Fn", "Fo", "Fr", "Ft", "Fx",
1.73 schwarze 280: "Hf", "Ic", "In", "It", "Lb", "Li", "Lk", "Lp",
281: "Ms", "Mt", "Nd", "Nm", "No", "Ns", "Nx",
1.47 schwarze 282: "Oc", "Oo", "Op", "Os", "Ot", "Ox",
1.73 schwarze 283: "Pa", "Pc", "Pf", "Po", "Pp", "Pq",
284: "Qc", "Ql", "Qo", "Qq", "Re", "Rs", "Rv",
285: "Sc", "Sh", "Sm", "So", "Sq",
1.47 schwarze 286: "Ss", "St", "Sx", "Sy",
287: "Ta", "Tn", "Ud", "Ux", "Va", "Vt", "Xc", "Xo", "Xr",
1.73 schwarze 288: "%A", "%B", "%C", "%D", "%I", "%J", "%N", "%O",
1.47 schwarze 289: "%P", "%Q", "%R", "%T", "%U", "%V",
290: NULL
291: };
292:
1.73 schwarze 293: /* not currently implemented: BT DE DS ME MT PT SY TQ YS */
1.47 schwarze 294: const char *const __man_reserved[] = {
1.73 schwarze 295: "AT", "B", "BI", "BR", "DT",
296: "EE", "EN", "EQ", "EX", "HP", "I", "IB", "IP", "IR",
297: "LP", "OP", "P", "PD", "PP",
298: "R", "RB", "RE", "RI", "RS", "SB", "SH", "SM", "SS",
299: "TE", "TH", "TP", "TS", "T&", "UC", "UE", "UR",
1.47 schwarze 300: NULL
301: };
302:
1.37 schwarze 303: /* Array of injected predefined strings. */
304: #define PREDEFS_MAX 38
305: static const struct predef predefs[PREDEFS_MAX] = {
306: #include "predefs.in"
307: };
308:
1.42 schwarze 309: /* See roffhash_find() */
1.3 schwarze 310: #define ROFF_HASH(p) (p[0] - ASCII_LO)
311:
1.51 schwarze 312: static int roffit_lines; /* number of lines to delay */
313: static char *roffit_macro; /* nil-terminated macro line */
314:
1.80 schwarze 315:
1.3 schwarze 316: static void
1.42 schwarze 317: roffhash_init(void)
1.3 schwarze 318: {
319: struct roffmac *n;
320: int buc, i;
321:
1.16 schwarze 322: for (i = 0; i < (int)ROFF_USERDEF; i++) {
1.3 schwarze 323: assert(roffs[i].name[0] >= ASCII_LO);
324: assert(roffs[i].name[0] <= ASCII_HI);
325:
326: buc = ROFF_HASH(roffs[i].name);
327:
328: if (NULL != (n = hash[buc])) {
329: for ( ; n->next; n = n->next)
330: /* Do nothing. */ ;
331: n->next = &roffs[i];
332: } else
333: hash[buc] = &roffs[i];
334: }
335: }
336:
1.1 schwarze 337: /*
338: * Look up a roff token by its name. Returns ROFF_MAX if no macro by
339: * the nil-terminated string name could be found.
340: */
341: static enum rofft
1.42 schwarze 342: roffhash_find(const char *p, size_t s)
1.1 schwarze 343: {
1.3 schwarze 344: int buc;
345: struct roffmac *n;
1.1 schwarze 346:
1.3 schwarze 347: /*
348: * libroff has an extremely simple hashtable, for the time
349: * being, which simply keys on the first character, which must
350: * be printable, then walks a chain. It works well enough until
351: * optimised.
352: */
353:
354: if (p[0] < ASCII_LO || p[0] > ASCII_HI)
355: return(ROFF_MAX);
356:
357: buc = ROFF_HASH(p);
358:
359: if (NULL == (n = hash[buc]))
360: return(ROFF_MAX);
361: for ( ; n; n = n->next)
1.16 schwarze 362: if (0 == strncmp(n->name, p, s) && '\0' == n->name[(int)s])
1.3 schwarze 363: return((enum rofft)(n - roffs));
1.1 schwarze 364:
365: return(ROFF_MAX);
366: }
367:
368: /*
369: * Pop the current node off of the stack of roff instructions currently
370: * pending.
371: */
372: static void
373: roffnode_pop(struct roff *r)
374: {
375: struct roffnode *p;
376:
1.2 schwarze 377: assert(r->last);
1.80 schwarze 378: p = r->last;
1.2 schwarze 379:
380: r->last = r->last->parent;
1.16 schwarze 381: free(p->name);
382: free(p->end);
1.1 schwarze 383: free(p);
384: }
385:
386: /*
387: * Push a roff node onto the instruction stack. This must later be
388: * removed with roffnode_pop().
389: */
1.11 schwarze 390: static void
1.16 schwarze 391: roffnode_push(struct roff *r, enum rofft tok, const char *name,
392: int line, int col)
1.1 schwarze 393: {
394: struct roffnode *p;
395:
1.11 schwarze 396: p = mandoc_calloc(1, sizeof(struct roffnode));
1.1 schwarze 397: p->tok = tok;
1.16 schwarze 398: if (name)
399: p->name = mandoc_strdup(name);
1.1 schwarze 400: p->parent = r->last;
401: p->line = line;
402: p->col = col;
1.71 schwarze 403: p->rule = p->parent ? p->parent->rule : 0;
1.1 schwarze 404:
405: r->last = p;
406: }
407:
408: static void
409: roff_free1(struct roff *r)
410: {
1.49 schwarze 411: struct tbl_node *tbl;
1.32 schwarze 412: struct eqn_node *e;
1.42 schwarze 413: int i;
1.27 schwarze 414:
1.49 schwarze 415: while (NULL != (tbl = r->first_tbl)) {
416: r->first_tbl = tbl->next;
417: tbl_free(tbl);
1.27 schwarze 418: }
419:
420: r->first_tbl = r->last_tbl = r->tbl = NULL;
1.1 schwarze 421:
1.32 schwarze 422: while (NULL != (e = r->first_eqn)) {
423: r->first_eqn = e->next;
424: eqn_free(e);
425: }
426:
427: r->first_eqn = r->last_eqn = r->eqn = NULL;
428:
1.1 schwarze 429: while (r->last)
430: roffnode_pop(r);
1.27 schwarze 431:
1.42 schwarze 432: roff_freestr(r->strtab);
433: roff_freestr(r->xmbtab);
434:
435: r->strtab = r->xmbtab = NULL;
436:
1.52 schwarze 437: roff_freereg(r->regtab);
438:
439: r->regtab = NULL;
440:
1.42 schwarze 441: if (r->xtab)
442: for (i = 0; i < 128; i++)
443: free(r->xtab[i].p);
444:
445: free(r->xtab);
446: r->xtab = NULL;
1.1 schwarze 447: }
448:
449: void
450: roff_reset(struct roff *r)
451: {
452:
453: roff_free1(r);
1.48 schwarze 454: r->control = 0;
1.1 schwarze 455: }
456:
457: void
458: roff_free(struct roff *r)
459: {
460:
461: roff_free1(r);
462: free(r);
463: }
464:
465: struct roff *
1.72 schwarze 466: roff_alloc(struct mparse *parse, int options)
1.1 schwarze 467: {
468: struct roff *r;
469:
1.11 schwarze 470: r = mandoc_calloc(1, sizeof(struct roff));
1.35 schwarze 471: r->parse = parse;
1.72 schwarze 472: r->options = options;
1.2 schwarze 473: r->rstackpos = -1;
1.80 schwarze 474:
1.42 schwarze 475: roffhash_init();
1.37 schwarze 476:
1.1 schwarze 477: return(r);
478: }
479:
1.8 schwarze 480: /*
1.79 schwarze 481: * In the current line, expand escape sequences that tend to get
482: * used in numerical expressions and conditional requests.
483: * Also check the syntax of the remaining escape sequences.
1.8 schwarze 484: */
1.45 schwarze 485: static enum rofferr
1.37 schwarze 486: roff_res(struct roff *r, char **bufp, size_t *szp, int ln, int pos)
1.8 schwarze 487: {
1.81 schwarze 488: char ubuf[24]; /* buffer to print the number */
1.78 schwarze 489: const char *start; /* start of the string to process */
1.82 schwarze 490: char *stesc; /* start of an escape sequence ('\\') */
1.23 schwarze 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 maxl; /* expected length of the escape name */
496: size_t naml; /* actual length of the escape name */
497: int expand_count; /* to avoid infinite loops */
1.79 schwarze 498: int npos; /* position in numeric expression */
1.91 schwarze 499: int arg_complete; /* argument not interrupted by eol */
1.79 schwarze 500: char term; /* character terminating the escape */
1.8 schwarze 501:
1.43 schwarze 502: expand_count = 0;
1.78 schwarze 503: start = *bufp + pos;
504: stesc = strchr(start, '\0') - 1;
505: while (stesc-- > start) {
1.43 schwarze 506:
1.78 schwarze 507: /* Search backwards for the next backslash. */
508:
509: if ('\\' != *stesc)
510: continue;
511:
512: /* If it is escaped, skip it. */
513:
514: for (cp = stesc - 1; cp >= start; cp--)
515: if ('\\' != *cp)
516: break;
517:
518: if (0 == (stesc - cp) % 2) {
1.82 schwarze 519: stesc = (char *)cp;
1.78 schwarze 520: continue;
521: }
1.23 schwarze 522:
1.79 schwarze 523: /* Decide whether to expand or to check only. */
1.23 schwarze 524:
1.79 schwarze 525: term = '\0';
1.78 schwarze 526: cp = stesc + 1;
1.53 schwarze 527: switch (*cp) {
1.80 schwarze 528: case '*':
1.53 schwarze 529: res = NULL;
530: break;
1.80 schwarze 531: case 'B':
1.79 schwarze 532: /* FALLTHROUGH */
1.80 schwarze 533: case 'w':
1.79 schwarze 534: term = cp[1];
535: /* FALLTHROUGH */
1.80 schwarze 536: case 'n':
1.53 schwarze 537: res = ubuf;
538: break;
539: default:
1.78 schwarze 540: if (ESCAPE_ERROR == mandoc_escape(&cp, NULL, NULL))
1.92 ! schwarze 541: mandoc_vmsg(MANDOCERR_ESC_BAD,
! 542: r->parse, ln, (int)(stesc - *bufp),
! 543: "%.*s", (int)(cp - stesc), stesc);
1.78 schwarze 544: continue;
1.42 schwarze 545: }
546:
1.78 schwarze 547: if (EXPAND_LIMIT < ++expand_count) {
548: mandoc_msg(MANDOCERR_ROFFLOOP, r->parse,
549: ln, (int)(stesc - *bufp), NULL);
550: return(ROFF_IGN);
551: }
1.23 schwarze 552:
553: /*
554: * The third character decides the length
1.53 schwarze 555: * of the name of the string or register.
1.23 schwarze 556: * Save a pointer to the name.
557: */
558:
1.79 schwarze 559: if ('\0' == term) {
560: switch (*++cp) {
1.80 schwarze 561: case '\0':
1.79 schwarze 562: maxl = 0;
563: break;
1.80 schwarze 564: case '(':
1.79 schwarze 565: cp++;
566: maxl = 2;
567: break;
1.80 schwarze 568: case '[':
1.79 schwarze 569: cp++;
570: term = ']';
571: maxl = 0;
572: break;
573: default:
574: maxl = 1;
575: break;
576: }
577: } else {
578: cp += 2;
1.8 schwarze 579: maxl = 0;
580: }
1.23 schwarze 581: stnam = cp;
1.8 schwarze 582:
1.23 schwarze 583: /* Advance to the end of the name. */
1.8 schwarze 584:
1.91 schwarze 585: arg_complete = 1;
1.53 schwarze 586: for (naml = 0; 0 == maxl || naml < maxl; naml++, cp++) {
1.42 schwarze 587: if ('\0' == *cp) {
1.92 ! schwarze 588: mandoc_msg(MANDOCERR_ESC_BAD, r->parse,
! 589: ln, (int)(stesc - *bufp), stesc);
1.91 schwarze 590: arg_complete = 0;
1.79 schwarze 591: break;
1.42 schwarze 592: }
1.79 schwarze 593: if (0 == maxl && *cp == term) {
594: cp++;
1.8 schwarze 595: break;
1.79 schwarze 596: }
1.8 schwarze 597: }
598:
1.23 schwarze 599: /*
600: * Retrieve the replacement string; if it is
601: * undefined, resume searching for escapes.
602: */
603:
1.79 schwarze 604: switch (stesc[1]) {
1.80 schwarze 605: case '*':
1.91 schwarze 606: if (arg_complete)
607: res = roff_getstrn(r, stnam, naml);
1.79 schwarze 608: break;
1.80 schwarze 609: case 'B':
1.79 schwarze 610: npos = 0;
1.91 schwarze 611: ubuf[0] = arg_complete &&
612: roff_evalnum(stnam, &npos, NULL, 0) &&
613: stnam + npos + 1 == cp ? '1' : '0';
1.79 schwarze 614: ubuf[1] = '\0';
615: break;
1.80 schwarze 616: case 'n':
1.91 schwarze 617: if (arg_complete)
618: (void)snprintf(ubuf, sizeof(ubuf), "%d",
619: roff_getregn(r, stnam, naml));
620: else
621: ubuf[0] = '\0';
1.79 schwarze 622: break;
1.80 schwarze 623: case 'w':
1.91 schwarze 624: /* use even incomplete args */
1.81 schwarze 625: (void)snprintf(ubuf, sizeof(ubuf), "%d",
1.79 schwarze 626: 24 * (int)naml);
627: break;
628: }
1.8 schwarze 629:
630: if (NULL == res) {
1.92 ! schwarze 631: mandoc_vmsg(MANDOCERR_STR_UNDEF,
! 632: r->parse, ln, (int)(stesc - *bufp),
! 633: "%.*s", (int)naml, stnam);
1.37 schwarze 634: res = "";
1.8 schwarze 635: }
636:
1.23 schwarze 637: /* Replace the escape sequence by the string. */
638:
1.82 schwarze 639: *stesc = '\0';
640: *szp = mandoc_asprintf(&nbuf, "%s%s%s",
641: *bufp, res, cp) + 1;
1.8 schwarze 642:
1.78 schwarze 643: /* Prepare for the next replacement. */
1.8 schwarze 644:
1.78 schwarze 645: start = nbuf + pos;
1.82 schwarze 646: stesc = nbuf + (stesc - *bufp) + strlen(res);
1.8 schwarze 647: free(*bufp);
1.53 schwarze 648: *bufp = nbuf;
1.42 schwarze 649: }
1.45 schwarze 650: return(ROFF_CONT);
1.42 schwarze 651: }
652:
653: /*
1.51 schwarze 654: * Process text streams:
655: * Convert all breakable hyphens into ASCII_HYPH.
656: * Decrement and spring input line trap.
1.42 schwarze 657: */
658: static enum rofferr
1.51 schwarze 659: roff_parsetext(char **bufp, size_t *szp, int pos, int *offs)
1.42 schwarze 660: {
661: size_t sz;
662: const char *start;
1.51 schwarze 663: char *p;
664: int isz;
1.42 schwarze 665: enum mandoc_esc esc;
666:
1.51 schwarze 667: start = p = *bufp + pos;
1.42 schwarze 668:
669: while ('\0' != *p) {
670: sz = strcspn(p, "-\\");
671: p += sz;
672:
673: if ('\0' == *p)
674: break;
675:
676: if ('\\' == *p) {
677: /* Skip over escapes. */
678: p++;
1.62 schwarze 679: esc = mandoc_escape((const char **)&p, NULL, NULL);
1.42 schwarze 680: if (ESCAPE_ERROR == esc)
681: break;
682: continue;
683: } else if (p == start) {
684: p++;
685: continue;
686: }
687:
1.44 schwarze 688: if (isalpha((unsigned char)p[-1]) &&
689: isalpha((unsigned char)p[1]))
1.42 schwarze 690: *p = ASCII_HYPH;
691: p++;
1.8 schwarze 692: }
693:
1.51 schwarze 694: /* Spring the input line trap. */
695: if (1 == roffit_lines) {
1.75 schwarze 696: isz = mandoc_asprintf(&p, "%s\n.%s", *bufp, roffit_macro);
1.51 schwarze 697: free(*bufp);
698: *bufp = p;
699: *szp = isz + 1;
700: *offs = 0;
701: free(roffit_macro);
702: roffit_lines = 0;
703: return(ROFF_REPARSE);
704: } else if (1 < roffit_lines)
705: --roffit_lines;
1.42 schwarze 706: return(ROFF_CONT);
1.8 schwarze 707: }
708:
1.1 schwarze 709: enum rofferr
1.80 schwarze 710: roff_parseln(struct roff *r, int ln, char **bufp,
1.6 schwarze 711: size_t *szp, int pos, int *offs)
1.1 schwarze 712: {
713: enum rofft t;
1.27 schwarze 714: enum rofferr e;
1.35 schwarze 715: int ppos, ctl;
1.1 schwarze 716:
1.2 schwarze 717: /*
1.8 schwarze 718: * Run the reserved-word filter only if we have some reserved
719: * words to fill in.
720: */
721:
1.45 schwarze 722: e = roff_res(r, bufp, szp, ln, pos);
723: if (ROFF_IGN == e)
724: return(e);
725: assert(ROFF_CONT == e);
1.8 schwarze 726:
1.35 schwarze 727: ppos = pos;
1.48 schwarze 728: ctl = roff_getcontrol(r, *bufp, &pos);
1.35 schwarze 729:
1.8 schwarze 730: /*
1.2 schwarze 731: * First, if a scope is open and we're not a macro, pass the
732: * text through the macro's filter. If a scope isn't open and
733: * we're not a macro, just let it through.
1.32 schwarze 734: * Finally, if there's an equation scope open, divert it into it
735: * no matter our state.
1.2 schwarze 736: */
737:
1.35 schwarze 738: if (r->last && ! ctl) {
1.2 schwarze 739: t = r->last->tok;
740: assert(roffs[t].text);
1.80 schwarze 741: e = (*roffs[t].text)(r, t, bufp, szp, ln, pos, pos, offs);
1.27 schwarze 742: assert(ROFF_IGN == e || ROFF_CONT == e);
1.32 schwarze 743: if (ROFF_CONT != e)
744: return(e);
1.54 schwarze 745: }
746: if (r->eqn)
747: return(eqn_read(&r->eqn, ln, *bufp, ppos, offs));
748: if ( ! ctl) {
1.32 schwarze 749: if (r->tbl)
1.35 schwarze 750: return(tbl_read(r->tbl, ln, *bufp, pos));
1.51 schwarze 751: return(roff_parsetext(bufp, szp, pos, offs));
1.54 schwarze 752: }
1.2 schwarze 753:
754: /*
755: * If a scope is open, go to the child handler for that macro,
756: * as it may want to preprocess before doing anything with it.
1.32 schwarze 757: * Don't do so if an equation is open.
1.2 schwarze 758: */
759:
760: if (r->last) {
1.1 schwarze 761: t = r->last->tok;
762: assert(roffs[t].sub);
1.80 schwarze 763: return((*roffs[t].sub)(r, t, bufp, szp,
764: ln, ppos, pos, offs));
1.2 schwarze 765: }
766:
767: /*
768: * Lastly, as we've no scope open, try to look up and execute
769: * the new macro. If no macro is found, simply return and let
770: * the compilers handle it.
771: */
772:
1.87 schwarze 773: if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos, ln, ppos)))
1.1 schwarze 774: return(ROFF_CONT);
775:
1.2 schwarze 776: assert(roffs[t].proc);
1.80 schwarze 777: return((*roffs[t].proc)(r, t, bufp, szp, ln, ppos, pos, offs));
1.2 schwarze 778: }
779:
1.27 schwarze 780: void
1.2 schwarze 781: roff_endparse(struct roff *r)
782: {
1.1 schwarze 783:
1.27 schwarze 784: if (r->last)
1.35 schwarze 785: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.80 schwarze 786: r->last->line, r->last->col, NULL);
1.27 schwarze 787:
1.32 schwarze 788: if (r->eqn) {
1.80 schwarze 789: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
790: r->eqn->eqn.ln, r->eqn->eqn.pos, NULL);
1.41 schwarze 791: eqn_end(&r->eqn);
1.32 schwarze 792: }
793:
1.27 schwarze 794: if (r->tbl) {
1.80 schwarze 795: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
796: r->tbl->line, r->tbl->pos, NULL);
1.41 schwarze 797: tbl_end(&r->tbl);
1.27 schwarze 798: }
1.1 schwarze 799: }
800:
801: /*
802: * Parse a roff node's type from the input buffer. This must be in the
803: * form of ".foo xxx" in the usual way.
804: */
805: static enum rofft
1.87 schwarze 806: roff_parse(struct roff *r, char *buf, int *pos, int ln, int ppos)
1.1 schwarze 807: {
1.87 schwarze 808: char *cp;
1.16 schwarze 809: const char *mac;
810: size_t maclen;
1.1 schwarze 811: enum rofft t;
812:
1.87 schwarze 813: cp = buf + *pos;
814:
815: if ('\0' == *cp || '"' == *cp || '\t' == *cp || ' ' == *cp)
1.1 schwarze 816: return(ROFF_MAX);
817:
1.87 schwarze 818: mac = cp;
819: maclen = roff_getname(r, &cp, ln, ppos);
1.1 schwarze 820:
1.16 schwarze 821: t = (r->current_string = roff_getstrn(r, mac, maclen))
1.42 schwarze 822: ? ROFF_USERDEF : roffhash_find(mac, maclen);
1.1 schwarze 823:
1.87 schwarze 824: if (ROFF_MAX != t)
825: *pos = cp - buf;
1.1 schwarze 826:
827: return(t);
828: }
829:
830: static enum rofferr
1.2 schwarze 831: roff_cblock(ROFF_ARGS)
1.1 schwarze 832: {
833:
1.2 schwarze 834: /*
835: * A block-close `..' should only be invoked as a child of an
836: * ignore macro, otherwise raise a warning and just ignore it.
837: */
838:
839: if (NULL == r->last) {
1.35 schwarze 840: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.2 schwarze 841: return(ROFF_IGN);
842: }
1.1 schwarze 843:
1.2 schwarze 844: switch (r->last->tok) {
1.80 schwarze 845: case ROFF_am:
1.2 schwarze 846: /* FALLTHROUGH */
1.80 schwarze 847: case ROFF_ami:
1.2 schwarze 848: /* FALLTHROUGH */
1.80 schwarze 849: case ROFF_am1:
1.2 schwarze 850: /* FALLTHROUGH */
1.80 schwarze 851: case ROFF_de:
1.23 schwarze 852: /* ROFF_de1 is remapped to ROFF_de in roff_block(). */
1.2 schwarze 853: /* FALLTHROUGH */
1.80 schwarze 854: case ROFF_dei:
1.2 schwarze 855: /* FALLTHROUGH */
1.80 schwarze 856: case ROFF_ig:
1.2 schwarze 857: break;
858: default:
1.35 schwarze 859: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.1 schwarze 860: return(ROFF_IGN);
1.2 schwarze 861: }
862:
863: if ((*bufp)[pos])
1.90 schwarze 864: mandoc_vmsg(MANDOCERR_ARG_SKIP, r->parse, ln, pos,
865: ".. %s", *bufp + pos);
1.2 schwarze 866:
867: roffnode_pop(r);
868: roffnode_cleanscope(r);
869: return(ROFF_IGN);
870:
871: }
1.1 schwarze 872:
1.2 schwarze 873: static void
874: roffnode_cleanscope(struct roff *r)
875: {
1.1 schwarze 876:
1.2 schwarze 877: while (r->last) {
1.46 schwarze 878: if (--r->last->endspan != 0)
1.2 schwarze 879: break;
880: roffnode_pop(r);
881: }
882: }
1.1 schwarze 883:
1.68 schwarze 884: static void
885: roff_ccond(struct roff *r, int ln, int ppos)
1.2 schwarze 886: {
1.1 schwarze 887:
1.2 schwarze 888: if (NULL == r->last) {
1.35 schwarze 889: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.68 schwarze 890: return;
1.2 schwarze 891: }
1.1 schwarze 892:
1.2 schwarze 893: switch (r->last->tok) {
1.80 schwarze 894: case ROFF_el:
1.2 schwarze 895: /* FALLTHROUGH */
1.80 schwarze 896: case ROFF_ie:
1.2 schwarze 897: /* FALLTHROUGH */
1.80 schwarze 898: case ROFF_if:
1.2 schwarze 899: break;
900: default:
1.35 schwarze 901: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.68 schwarze 902: return;
1.2 schwarze 903: }
1.1 schwarze 904:
1.2 schwarze 905: if (r->last->endspan > -1) {
1.35 schwarze 906: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.68 schwarze 907: return;
1.2 schwarze 908: }
909:
910: roffnode_pop(r);
911: roffnode_cleanscope(r);
1.68 schwarze 912: return;
1.1 schwarze 913: }
914:
915: static enum rofferr
1.2 schwarze 916: roff_block(ROFF_ARGS)
1.1 schwarze 917: {
1.86 schwarze 918: char *name, *cp;
919: size_t namesz;
1.16 schwarze 920:
1.86 schwarze 921: name = cp = *bufp + pos;
922: namesz = 0;
1.2 schwarze 923:
1.16 schwarze 924: if (ROFF_ig != tok) {
1.86 schwarze 925: if ('\0' == *cp) {
1.89 schwarze 926: mandoc_msg(MANDOCERR_REQ_EMPTY, r->parse,
927: ln, ppos, roffs[tok].name);
1.16 schwarze 928: return(ROFF_IGN);
929: }
1.22 schwarze 930:
931: /*
932: * Re-write `de1', since we don't really care about
933: * groff's strange compatibility mode, into `de'.
934: */
935:
1.18 schwarze 936: if (ROFF_de1 == tok)
937: tok = ROFF_de;
1.86 schwarze 938: else if (ROFF_de != tok)
1.35 schwarze 939: mandoc_msg(MANDOCERR_REQUEST, r->parse, ln, ppos,
1.21 schwarze 940: roffs[tok].name);
1.22 schwarze 941:
1.86 schwarze 942: namesz = roff_getname(r, &cp, ln, ppos);
943: name[namesz] = '\0';
944: } else
945: name = NULL;
1.2 schwarze 946:
1.16 schwarze 947: roffnode_push(r, tok, name, ln, ppos);
948:
949: /*
950: * At the beginning of a `de' macro, clear the existing string
951: * with the same name, if there is one. New content will be
1.66 schwarze 952: * appended from roff_block_text() in multiline mode.
1.16 schwarze 953: */
1.22 schwarze 954:
1.86 schwarze 955: if (namesz && ROFF_de == tok)
956: roff_setstrn(&r->strtab, name, namesz, "", 0, 0);
1.2 schwarze 957:
1.86 schwarze 958: if ('\0' == *cp)
1.2 schwarze 959: return(ROFF_IGN);
1.1 schwarze 960:
1.22 schwarze 961: /* If present, process the custom end-of-line marker. */
962:
1.86 schwarze 963: name = cp;
964: namesz = roff_getname(r, &cp, ln, ppos);
965: if (namesz)
966: r->last->end = mandoc_strndup(name, namesz);
1.2 schwarze 967:
1.86 schwarze 968: if ('\0' != *cp)
1.90 schwarze 969: mandoc_vmsg(MANDOCERR_ARG_EXCESS, r->parse,
970: ln, pos, ".%s ... %s", roffs[tok].name, cp);
1.1 schwarze 971:
972: return(ROFF_IGN);
973: }
974:
975: static enum rofferr
1.2 schwarze 976: roff_block_sub(ROFF_ARGS)
1.1 schwarze 977: {
1.2 schwarze 978: enum rofft t;
979: int i, j;
980:
981: /*
982: * First check whether a custom macro exists at this level. If
983: * it does, then check against it. This is some of groff's
984: * stranger behaviours. If we encountered a custom end-scope
985: * tag and that tag also happens to be a "real" macro, then we
986: * need to try interpreting it again as a real macro. If it's
987: * not, then return ignore. Else continue.
988: */
989:
990: if (r->last->end) {
1.35 schwarze 991: for (i = pos, j = 0; r->last->end[j]; j++, i++)
1.2 schwarze 992: if ((*bufp)[i] != r->last->end[j])
993: break;
1.1 schwarze 994:
1.80 schwarze 995: if ('\0' == r->last->end[j] &&
996: ('\0' == (*bufp)[i] ||
997: ' ' == (*bufp)[i] ||
998: '\t' == (*bufp)[i])) {
1.2 schwarze 999: roffnode_pop(r);
1000: roffnode_cleanscope(r);
1.1 schwarze 1001:
1.35 schwarze 1002: while (' ' == (*bufp)[i] || '\t' == (*bufp)[i])
1003: i++;
1004:
1005: pos = i;
1.87 schwarze 1006: if (ROFF_MAX != roff_parse(r, *bufp, &pos, ln, ppos))
1.2 schwarze 1007: return(ROFF_RERUN);
1008: return(ROFF_IGN);
1009: }
1.1 schwarze 1010: }
1011:
1.2 schwarze 1012: /*
1013: * If we have no custom end-query or lookup failed, then try
1014: * pulling it out of the hashtable.
1015: */
1.1 schwarze 1016:
1.87 schwarze 1017: t = roff_parse(r, *bufp, &pos, ln, ppos);
1.1 schwarze 1018:
1.16 schwarze 1019: /*
1020: * Macros other than block-end are only significant
1021: * in `de' blocks; elsewhere, simply throw them away.
1022: */
1023: if (ROFF_cblock != t) {
1024: if (ROFF_de == tok)
1.66 schwarze 1025: roff_setstr(r, r->last->name, *bufp + ppos, 2);
1.1 schwarze 1026: return(ROFF_IGN);
1.16 schwarze 1027: }
1.1 schwarze 1028:
1.2 schwarze 1029: assert(roffs[t].proc);
1.80 schwarze 1030: return((*roffs[t].proc)(r, t, bufp, szp, ln, ppos, pos, offs));
1.2 schwarze 1031: }
1032:
1033: static enum rofferr
1034: roff_block_text(ROFF_ARGS)
1035: {
1036:
1.16 schwarze 1037: if (ROFF_de == tok)
1.66 schwarze 1038: roff_setstr(r, r->last->name, *bufp + pos, 2);
1.16 schwarze 1039:
1.2 schwarze 1040: return(ROFF_IGN);
1041: }
1042:
1043: static enum rofferr
1044: roff_cond_sub(ROFF_ARGS)
1045: {
1046: enum rofft t;
1.37 schwarze 1047: char *ep;
1.71 schwarze 1048: int rr;
1.2 schwarze 1049:
1050: rr = r->last->rule;
1.37 schwarze 1051: roffnode_cleanscope(r);
1.87 schwarze 1052: t = roff_parse(r, *bufp, &pos, ln, ppos);
1.2 schwarze 1053:
1.37 schwarze 1054: /*
1.50 schwarze 1055: * Fully handle known macros when they are structurally
1056: * required or when the conditional evaluated to true.
1.5 schwarze 1057: */
1058:
1.50 schwarze 1059: if ((ROFF_MAX != t) &&
1.71 schwarze 1060: (rr || ROFFMAC_STRUCT & roffs[t].flags)) {
1.50 schwarze 1061: assert(roffs[t].proc);
1062: return((*roffs[t].proc)(r, t, bufp, szp,
1.80 schwarze 1063: ln, ppos, pos, offs));
1.50 schwarze 1064: }
1.39 schwarze 1065:
1.69 schwarze 1066: /*
1067: * If `\}' occurs on a macro line without a preceding macro,
1068: * drop the line completely.
1069: */
1070:
1071: ep = *bufp + pos;
1072: if ('\\' == ep[0] && '}' == ep[1])
1.71 schwarze 1073: rr = 0;
1.69 schwarze 1074:
1.50 schwarze 1075: /* Always check for the closing delimiter `\}'. */
1.39 schwarze 1076:
1.50 schwarze 1077: while (NULL != (ep = strchr(ep, '\\'))) {
1.70 schwarze 1078: if ('}' == *(++ep)) {
1079: *ep = '&';
1080: roff_ccond(r, ln, ep - *bufp - 1);
1081: }
1082: ++ep;
1.50 schwarze 1083: }
1.71 schwarze 1084: return(rr ? ROFF_CONT : ROFF_IGN);
1.2 schwarze 1085: }
1086:
1087: static enum rofferr
1088: roff_cond_text(ROFF_ARGS)
1089: {
1.37 schwarze 1090: char *ep;
1.71 schwarze 1091: int rr;
1.2 schwarze 1092:
1093: rr = r->last->rule;
1.37 schwarze 1094: roffnode_cleanscope(r);
1.1 schwarze 1095:
1.70 schwarze 1096: ep = *bufp + pos;
1097: while (NULL != (ep = strchr(ep, '\\'))) {
1098: if ('}' == *(++ep)) {
1099: *ep = '&';
1100: roff_ccond(r, ln, ep - *bufp - 1);
1101: }
1102: ++ep;
1.2 schwarze 1103: }
1.71 schwarze 1104: return(rr ? ROFF_CONT : ROFF_IGN);
1.2 schwarze 1105: }
1106:
1.77 schwarze 1107: /*
1108: * Parse a single signed integer number. Stop at the first non-digit.
1109: * If there is at least one digit, return success and advance the
1110: * parse point, else return failure and let the parse point unchanged.
1111: * Ignore overflows, treat them just like the C language.
1112: */
1.56 schwarze 1113: static int
1114: roff_getnum(const char *v, int *pos, int *res)
1115: {
1.79 schwarze 1116: int myres, n, p;
1117:
1118: if (NULL == res)
1119: res = &myres;
1.56 schwarze 1120:
1121: p = *pos;
1122: n = v[p] == '-';
1123: if (n)
1124: p++;
1125:
1126: for (*res = 0; isdigit((unsigned char)v[p]); p++)
1.77 schwarze 1127: *res = 10 * *res + v[p] - '0';
1.56 schwarze 1128: if (p == *pos + n)
1129: return 0;
1130:
1131: if (n)
1132: *res = -*res;
1133:
1134: *pos = p;
1135: return 1;
1136: }
1137:
1.71 schwarze 1138: /*
1139: * Evaluate a string comparison condition.
1140: * The first character is the delimiter.
1141: * Succeed if the string up to its second occurrence
1142: * matches the string up to its third occurence.
1143: * Advance the cursor after the third occurrence
1144: * or lacking that, to the end of the line.
1145: */
1146: static int
1147: roff_evalstrcond(const char *v, int *pos)
1148: {
1149: const char *s1, *s2, *s3;
1150: int match;
1151:
1152: match = 0;
1153: s1 = v + *pos; /* initial delimiter */
1154: s2 = s1 + 1; /* for scanning the first string */
1155: s3 = strchr(s2, *s1); /* for scanning the second string */
1156:
1157: if (NULL == s3) /* found no middle delimiter */
1158: goto out;
1159:
1160: while ('\0' != *++s3) {
1161: if (*s2 != *s3) { /* mismatch */
1162: s3 = strchr(s3, *s1);
1163: break;
1164: }
1165: if (*s3 == *s1) { /* found the final delimiter */
1166: match = 1;
1167: break;
1168: }
1169: s2++;
1170: }
1171:
1172: out:
1173: if (NULL == s3)
1174: s3 = strchr(s2, '\0');
1175: else
1176: s3++;
1177: *pos = s3 - v;
1178: return(match);
1179: }
1180:
1.77 schwarze 1181: /*
1182: * Evaluate an optionally negated single character, numerical,
1183: * or string condition.
1184: */
1.71 schwarze 1185: static int
1.5 schwarze 1186: roff_evalcond(const char *v, int *pos)
1187: {
1.77 schwarze 1188: int wanttrue, number;
1.5 schwarze 1189:
1.71 schwarze 1190: if ('!' == v[*pos]) {
1191: wanttrue = 0;
1192: (*pos)++;
1193: } else
1194: wanttrue = 1;
1195:
1.5 schwarze 1196: switch (v[*pos]) {
1.80 schwarze 1197: case 'n':
1.71 schwarze 1198: /* FALLTHROUGH */
1.80 schwarze 1199: case 'o':
1.5 schwarze 1200: (*pos)++;
1.71 schwarze 1201: return(wanttrue);
1.80 schwarze 1202: case 'c':
1.71 schwarze 1203: /* FALLTHROUGH */
1.80 schwarze 1204: case 'd':
1.71 schwarze 1205: /* FALLTHROUGH */
1.80 schwarze 1206: case 'e':
1.5 schwarze 1207: /* FALLTHROUGH */
1.80 schwarze 1208: case 'r':
1.5 schwarze 1209: /* FALLTHROUGH */
1.80 schwarze 1210: case 't':
1.5 schwarze 1211: (*pos)++;
1.71 schwarze 1212: return(!wanttrue);
1.5 schwarze 1213: default:
1214: break;
1215: }
1216:
1.77 schwarze 1217: if (roff_evalnum(v, pos, &number, 0))
1218: return((number > 0) == wanttrue);
1219: else
1.71 schwarze 1220: return(roff_evalstrcond(v, pos) == wanttrue);
1.5 schwarze 1221: }
1222:
1.2 schwarze 1223: static enum rofferr
1.21 schwarze 1224: roff_line_ignore(ROFF_ARGS)
1.6 schwarze 1225: {
1.30 schwarze 1226:
1.21 schwarze 1227: return(ROFF_IGN);
1228: }
1229:
1230: static enum rofferr
1.2 schwarze 1231: roff_cond(ROFF_ARGS)
1232: {
1.46 schwarze 1233:
1234: roffnode_push(r, tok, NULL, ln, ppos);
1.2 schwarze 1235:
1.80 schwarze 1236: /*
1.35 schwarze 1237: * An `.el' has no conditional body: it will consume the value
1238: * of the current rstack entry set in prior `ie' calls or
1.80 schwarze 1239: * defaults to DENY.
1.35 schwarze 1240: *
1241: * If we're not an `el', however, then evaluate the conditional.
1242: */
1.1 schwarze 1243:
1.46 schwarze 1244: r->last->rule = ROFF_el == tok ?
1.80 schwarze 1245: (r->rstackpos < 0 ? 0 : r->rstack[r->rstackpos--]) :
1246: roff_evalcond(*bufp, &pos);
1.2 schwarze 1247:
1.35 schwarze 1248: /*
1249: * An if-else will put the NEGATION of the current evaluated
1250: * conditional into the stack of rules.
1251: */
1252:
1.2 schwarze 1253: if (ROFF_ie == tok) {
1.35 schwarze 1254: if (r->rstackpos == RSTACK_MAX - 1) {
1.80 schwarze 1255: mandoc_msg(MANDOCERR_MEM,
1256: r->parse, ln, ppos, NULL);
1.35 schwarze 1257: return(ROFF_ERR);
1258: }
1.71 schwarze 1259: r->rstack[++r->rstackpos] = !r->last->rule;
1.2 schwarze 1260: }
1.5 schwarze 1261:
1262: /* If the parent has false as its rule, then so do we. */
1263:
1.71 schwarze 1264: if (r->last->parent && !r->last->parent->rule)
1265: r->last->rule = 0;
1.5 schwarze 1266:
1267: /*
1.46 schwarze 1268: * Determine scope.
1269: * If there is nothing on the line after the conditional,
1270: * not even whitespace, use next-line scope.
1.5 schwarze 1271: */
1.2 schwarze 1272:
1.46 schwarze 1273: if ('\0' == (*bufp)[pos]) {
1274: r->last->endspan = 2;
1275: goto out;
1276: }
1277:
1278: while (' ' == (*bufp)[pos])
1279: pos++;
1280:
1281: /* An opening brace requests multiline scope. */
1.2 schwarze 1282:
1283: if ('\\' == (*bufp)[pos] && '{' == (*bufp)[pos + 1]) {
1284: r->last->endspan = -1;
1285: pos += 2;
1.46 schwarze 1286: goto out;
1.80 schwarze 1287: }
1.2 schwarze 1288:
1289: /*
1.46 schwarze 1290: * Anything else following the conditional causes
1291: * single-line scope. Warn if the scope contains
1292: * nothing but trailing whitespace.
1.2 schwarze 1293: */
1294:
1295: if ('\0' == (*bufp)[pos])
1.89 schwarze 1296: mandoc_msg(MANDOCERR_COND_EMPTY, r->parse,
1297: ln, ppos, roffs[tok].name);
1.2 schwarze 1298:
1.46 schwarze 1299: r->last->endspan = 1;
1.1 schwarze 1300:
1.46 schwarze 1301: out:
1.2 schwarze 1302: *offs = pos;
1303: return(ROFF_RERUN);
1.1 schwarze 1304: }
1305:
1.2 schwarze 1306: static enum rofferr
1.7 schwarze 1307: roff_ds(ROFF_ARGS)
1308: {
1.85 schwarze 1309: char *string;
1310: const char *name;
1311: size_t namesz;
1.10 schwarze 1312:
1313: /*
1.85 schwarze 1314: * The first word is the name of the string.
1315: * If it is empty or terminated by an escape sequence,
1316: * abort the `ds' request without defining anything.
1.10 schwarze 1317: */
1.7 schwarze 1318:
1.85 schwarze 1319: name = string = *bufp + pos;
1.7 schwarze 1320: if ('\0' == *name)
1321: return(ROFF_IGN);
1322:
1.85 schwarze 1323: namesz = roff_getname(r, &string, ln, pos);
1324: if ('\\' == name[namesz])
1325: return(ROFF_IGN);
1326:
1327: /* Read past the initial double-quote, if any. */
1.28 schwarze 1328: if ('"' == *string)
1.7 schwarze 1329: string++;
1330:
1.10 schwarze 1331: /* The rest is the value. */
1.85 schwarze 1332: roff_setstrn(&r->strtab, name, namesz, string, strlen(string),
1333: ROFF_as == tok);
1.7 schwarze 1334: return(ROFF_IGN);
1335: }
1336:
1.77 schwarze 1337: /*
1338: * Parse a single operator, one or two characters long.
1339: * If the operator is recognized, return success and advance the
1340: * parse point, else return failure and let the parse point unchanged.
1341: */
1342: static int
1343: roff_getop(const char *v, int *pos, char *res)
1344: {
1345:
1346: *res = v[*pos];
1347:
1348: switch (*res) {
1.80 schwarze 1349: case '+':
1.77 schwarze 1350: /* FALLTHROUGH */
1.80 schwarze 1351: case '-':
1.77 schwarze 1352: /* FALLTHROUGH */
1.80 schwarze 1353: case '*':
1.77 schwarze 1354: /* FALLTHROUGH */
1.80 schwarze 1355: case '/':
1.77 schwarze 1356: /* FALLTHROUGH */
1.80 schwarze 1357: case '%':
1.77 schwarze 1358: /* FALLTHROUGH */
1.80 schwarze 1359: case '&':
1.77 schwarze 1360: /* FALLTHROUGH */
1.80 schwarze 1361: case ':':
1.77 schwarze 1362: break;
1363: case '<':
1364: switch (v[*pos + 1]) {
1.80 schwarze 1365: case '=':
1.77 schwarze 1366: *res = 'l';
1367: (*pos)++;
1368: break;
1.80 schwarze 1369: case '>':
1.77 schwarze 1370: *res = '!';
1371: (*pos)++;
1372: break;
1.80 schwarze 1373: case '?':
1.77 schwarze 1374: *res = 'i';
1375: (*pos)++;
1376: break;
1377: default:
1378: break;
1379: }
1380: break;
1381: case '>':
1382: switch (v[*pos + 1]) {
1.80 schwarze 1383: case '=':
1.77 schwarze 1384: *res = 'g';
1385: (*pos)++;
1386: break;
1.80 schwarze 1387: case '?':
1.77 schwarze 1388: *res = 'a';
1389: (*pos)++;
1390: break;
1391: default:
1392: break;
1393: }
1394: break;
1395: case '=':
1396: if ('=' == v[*pos + 1])
1397: (*pos)++;
1398: break;
1399: default:
1400: return(0);
1401: }
1402: (*pos)++;
1403:
1404: return(*res);
1405: }
1406:
1407: /*
1408: * Evaluate either a parenthesized numeric expression
1409: * or a single signed integer number.
1410: */
1411: static int
1412: roff_evalpar(const char *v, int *pos, int *res)
1413: {
1414:
1415: if ('(' != v[*pos])
1416: return(roff_getnum(v, pos, res));
1417:
1418: (*pos)++;
1419: if ( ! roff_evalnum(v, pos, res, 1))
1420: return(0);
1421:
1.79 schwarze 1422: /*
1423: * Omission of the closing parenthesis
1424: * is an error in validation mode,
1425: * but ignored in evaluation mode.
1426: */
1427:
1.77 schwarze 1428: if (')' == v[*pos])
1429: (*pos)++;
1.79 schwarze 1430: else if (NULL == res)
1431: return(0);
1.77 schwarze 1432:
1433: return(1);
1434: }
1435:
1436: /*
1437: * Evaluate a complete numeric expression.
1438: * Proceed left to right, there is no concept of precedence.
1439: */
1440: static int
1441: roff_evalnum(const char *v, int *pos, int *res, int skipwhite)
1442: {
1443: int mypos, operand2;
1444: char operator;
1445:
1446: if (NULL == pos) {
1447: mypos = 0;
1448: pos = &mypos;
1449: }
1450:
1451: if (skipwhite)
1452: while (isspace((unsigned char)v[*pos]))
1453: (*pos)++;
1454:
1455: if ( ! roff_evalpar(v, pos, res))
1456: return(0);
1457:
1458: while (1) {
1459: if (skipwhite)
1460: while (isspace((unsigned char)v[*pos]))
1461: (*pos)++;
1462:
1463: if ( ! roff_getop(v, pos, &operator))
1464: break;
1465:
1466: if (skipwhite)
1467: while (isspace((unsigned char)v[*pos]))
1468: (*pos)++;
1469:
1470: if ( ! roff_evalpar(v, pos, &operand2))
1471: return(0);
1472:
1473: if (skipwhite)
1474: while (isspace((unsigned char)v[*pos]))
1475: (*pos)++;
1.79 schwarze 1476:
1477: if (NULL == res)
1478: continue;
1.77 schwarze 1479:
1480: switch (operator) {
1.80 schwarze 1481: case '+':
1.77 schwarze 1482: *res += operand2;
1483: break;
1.80 schwarze 1484: case '-':
1.77 schwarze 1485: *res -= operand2;
1486: break;
1.80 schwarze 1487: case '*':
1.77 schwarze 1488: *res *= operand2;
1489: break;
1.80 schwarze 1490: case '/':
1.77 schwarze 1491: *res /= operand2;
1492: break;
1.80 schwarze 1493: case '%':
1.77 schwarze 1494: *res %= operand2;
1495: break;
1.80 schwarze 1496: case '<':
1.77 schwarze 1497: *res = *res < operand2;
1498: break;
1.80 schwarze 1499: case '>':
1.77 schwarze 1500: *res = *res > operand2;
1501: break;
1.80 schwarze 1502: case 'l':
1.77 schwarze 1503: *res = *res <= operand2;
1504: break;
1.80 schwarze 1505: case 'g':
1.77 schwarze 1506: *res = *res >= operand2;
1507: break;
1.80 schwarze 1508: case '=':
1.77 schwarze 1509: *res = *res == operand2;
1510: break;
1.80 schwarze 1511: case '!':
1.77 schwarze 1512: *res = *res != operand2;
1513: break;
1.80 schwarze 1514: case '&':
1.77 schwarze 1515: *res = *res && operand2;
1516: break;
1.80 schwarze 1517: case ':':
1.77 schwarze 1518: *res = *res || operand2;
1519: break;
1.80 schwarze 1520: case 'i':
1.77 schwarze 1521: if (operand2 < *res)
1522: *res = operand2;
1523: break;
1.80 schwarze 1524: case 'a':
1.77 schwarze 1525: if (operand2 > *res)
1526: *res = operand2;
1527: break;
1528: default:
1529: abort();
1530: }
1531: }
1532: return(1);
1533: }
1534:
1.52 schwarze 1535: void
1.60 schwarze 1536: roff_setreg(struct roff *r, const char *name, int val, char sign)
1.41 schwarze 1537: {
1.52 schwarze 1538: struct roffreg *reg;
1539:
1540: /* Search for an existing register with the same name. */
1541: reg = r->regtab;
1542:
1543: while (reg && strcmp(name, reg->key.p))
1544: reg = reg->next;
1.41 schwarze 1545:
1.52 schwarze 1546: if (NULL == reg) {
1547: /* Create a new register. */
1548: reg = mandoc_malloc(sizeof(struct roffreg));
1549: reg->key.p = mandoc_strdup(name);
1550: reg->key.sz = strlen(name);
1.60 schwarze 1551: reg->val = 0;
1.52 schwarze 1552: reg->next = r->regtab;
1553: r->regtab = reg;
1554: }
1555:
1.60 schwarze 1556: if ('+' == sign)
1557: reg->val += val;
1558: else if ('-' == sign)
1559: reg->val -= val;
1560: else
1561: reg->val = val;
1.41 schwarze 1562: }
1563:
1.65 schwarze 1564: /*
1565: * Handle some predefined read-only number registers.
1566: * For now, return -1 if the requested register is not predefined;
1567: * in case a predefined read-only register having the value -1
1568: * were to turn up, another special value would have to be chosen.
1569: */
1570: static int
1571: roff_getregro(const char *name)
1572: {
1573:
1574: switch (*name) {
1.80 schwarze 1575: case 'A': /* ASCII approximation mode is always off. */
1.65 schwarze 1576: return(0);
1.80 schwarze 1577: case 'g': /* Groff compatibility mode is always on. */
1.65 schwarze 1578: return(1);
1.80 schwarze 1579: case 'H': /* Fixed horizontal resolution. */
1.65 schwarze 1580: return (24);
1.80 schwarze 1581: case 'j': /* Always adjust left margin only. */
1.65 schwarze 1582: return(0);
1.80 schwarze 1583: case 'T': /* Some output device is always defined. */
1.65 schwarze 1584: return(1);
1.80 schwarze 1585: case 'V': /* Fixed vertical resolution. */
1.65 schwarze 1586: return (40);
1587: default:
1588: return (-1);
1589: }
1590: }
1591:
1.53 schwarze 1592: int
1.52 schwarze 1593: roff_getreg(const struct roff *r, const char *name)
1.41 schwarze 1594: {
1.52 schwarze 1595: struct roffreg *reg;
1.65 schwarze 1596: int val;
1597:
1598: if ('.' == name[0] && '\0' != name[1] && '\0' == name[2]) {
1599: val = roff_getregro(name + 1);
1600: if (-1 != val)
1601: return (val);
1602: }
1.52 schwarze 1603:
1604: for (reg = r->regtab; reg; reg = reg->next)
1605: if (0 == strcmp(name, reg->key.p))
1.53 schwarze 1606: return(reg->val);
1607:
1608: return(0);
1609: }
1610:
1611: static int
1612: roff_getregn(const struct roff *r, const char *name, size_t len)
1613: {
1614: struct roffreg *reg;
1.65 schwarze 1615: int val;
1616:
1617: if ('.' == name[0] && 2 == len) {
1618: val = roff_getregro(name + 1);
1619: if (-1 != val)
1620: return (val);
1621: }
1.53 schwarze 1622:
1623: for (reg = r->regtab; reg; reg = reg->next)
1624: if (len == reg->key.sz &&
1625: 0 == strncmp(name, reg->key.p, len))
1626: return(reg->val);
1.41 schwarze 1627:
1.52 schwarze 1628: return(0);
1.41 schwarze 1629: }
1630:
1.52 schwarze 1631: static void
1632: roff_freereg(struct roffreg *reg)
1.41 schwarze 1633: {
1.52 schwarze 1634: struct roffreg *old_reg;
1.41 schwarze 1635:
1.52 schwarze 1636: while (NULL != reg) {
1637: free(reg->key.p);
1638: old_reg = reg;
1639: reg = reg->next;
1640: free(old_reg);
1641: }
1.41 schwarze 1642: }
1.7 schwarze 1643:
1644: static enum rofferr
1.6 schwarze 1645: roff_nr(ROFF_ARGS)
1.1 schwarze 1646: {
1.85 schwarze 1647: char *key, *val;
1648: size_t keysz;
1.37 schwarze 1649: int iv;
1.60 schwarze 1650: char sign;
1.6 schwarze 1651:
1.85 schwarze 1652: key = val = *bufp + pos;
1653: if ('\0' == *key)
1654: return(ROFF_IGN);
1655:
1656: keysz = roff_getname(r, &val, ln, pos);
1657: if ('\\' == key[keysz])
1658: return(ROFF_IGN);
1659: key[keysz] = '\0';
1.6 schwarze 1660:
1.60 schwarze 1661: sign = *val;
1662: if ('+' == sign || '-' == sign)
1663: val++;
1664:
1.77 schwarze 1665: if (roff_evalnum(val, NULL, &iv, 0))
1666: roff_setreg(r, key, iv, sign);
1.1 schwarze 1667:
1.76 schwarze 1668: return(ROFF_IGN);
1669: }
1670:
1671: static enum rofferr
1672: roff_rr(ROFF_ARGS)
1673: {
1674: struct roffreg *reg, **prev;
1.85 schwarze 1675: char *name, *cp;
1676: size_t namesz;
1.76 schwarze 1677:
1.85 schwarze 1678: name = cp = *bufp + pos;
1679: if ('\0' == *name)
1680: return(ROFF_IGN);
1681: namesz = roff_getname(r, &cp, ln, pos);
1682: name[namesz] = '\0';
1.76 schwarze 1683:
1684: prev = &r->regtab;
1685: while (1) {
1686: reg = *prev;
1687: if (NULL == reg || !strcmp(name, reg->key.p))
1688: break;
1689: prev = ®->next;
1690: }
1691: if (NULL != reg) {
1692: *prev = reg->next;
1693: free(reg->key.p);
1694: free(reg);
1695: }
1.29 schwarze 1696: return(ROFF_IGN);
1697: }
1698:
1699: static enum rofferr
1700: roff_rm(ROFF_ARGS)
1701: {
1702: const char *name;
1703: char *cp;
1.85 schwarze 1704: size_t namesz;
1.29 schwarze 1705:
1706: cp = *bufp + pos;
1707: while ('\0' != *cp) {
1.85 schwarze 1708: name = cp;
1709: namesz = roff_getname(r, &cp, ln, (int)(cp - *bufp));
1710: roff_setstrn(&r->strtab, name, namesz, NULL, 0, 0);
1711: if ('\\' == name[namesz])
1712: break;
1.29 schwarze 1713: }
1.51 schwarze 1714: return(ROFF_IGN);
1715: }
1716:
1717: static enum rofferr
1718: roff_it(ROFF_ARGS)
1719: {
1720: char *cp;
1721: size_t len;
1722: int iv;
1723:
1724: /* Parse the number of lines. */
1725: cp = *bufp + pos;
1726: len = strcspn(cp, " \t");
1727: cp[len] = '\0';
1728: if ((iv = mandoc_strntoi(cp, len, 10)) <= 0) {
1729: mandoc_msg(MANDOCERR_NUMERIC, r->parse,
1.80 schwarze 1730: ln, ppos, *bufp + 1);
1.51 schwarze 1731: return(ROFF_IGN);
1732: }
1733: cp += len + 1;
1734:
1735: /* Arm the input line trap. */
1736: roffit_lines = iv;
1737: roffit_macro = mandoc_strdup(cp);
1.2 schwarze 1738: return(ROFF_IGN);
1.47 schwarze 1739: }
1740:
1741: static enum rofferr
1742: roff_Dd(ROFF_ARGS)
1743: {
1744: const char *const *cp;
1745:
1.72 schwarze 1746: if (0 == ((MPARSE_MDOC | MPARSE_QUICK) & r->options))
1.47 schwarze 1747: for (cp = __mdoc_reserved; *cp; cp++)
1748: roff_setstr(r, *cp, NULL, 0);
1749:
1750: return(ROFF_CONT);
1751: }
1752:
1753: static enum rofferr
1754: roff_TH(ROFF_ARGS)
1755: {
1756: const char *const *cp;
1757:
1.72 schwarze 1758: if (0 == (MPARSE_QUICK & r->options))
1.47 schwarze 1759: for (cp = __man_reserved; *cp; cp++)
1760: roff_setstr(r, *cp, NULL, 0);
1761:
1762: return(ROFF_CONT);
1.14 schwarze 1763: }
1764:
1765: static enum rofferr
1.27 schwarze 1766: roff_TE(ROFF_ARGS)
1767: {
1768:
1769: if (NULL == r->tbl)
1.35 schwarze 1770: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.27 schwarze 1771: else
1.41 schwarze 1772: tbl_end(&r->tbl);
1.27 schwarze 1773:
1774: return(ROFF_IGN);
1775: }
1776:
1777: static enum rofferr
1778: roff_T_(ROFF_ARGS)
1779: {
1780:
1781: if (NULL == r->tbl)
1.35 schwarze 1782: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.27 schwarze 1783: else
1784: tbl_restart(ppos, ln, r->tbl);
1785:
1786: return(ROFF_IGN);
1787: }
1788:
1.41 schwarze 1789: #if 0
1790: static int
1791: roff_closeeqn(struct roff *r)
1792: {
1793:
1794: return(r->eqn && ROFF_EQN == eqn_end(&r->eqn) ? 1 : 0);
1795: }
1796: #endif
1797:
1798: static void
1.80 schwarze 1799: roff_openeqn(struct roff *r, const char *name, int line,
1.41 schwarze 1800: int offs, const char *buf)
1.32 schwarze 1801: {
1.41 schwarze 1802: struct eqn_node *e;
1803: int poff;
1.32 schwarze 1804:
1805: assert(NULL == r->eqn);
1.41 schwarze 1806: e = eqn_alloc(name, offs, line, r->parse);
1.32 schwarze 1807:
1808: if (r->last_eqn)
1809: r->last_eqn->next = e;
1810: else
1811: r->first_eqn = r->last_eqn = e;
1812:
1813: r->eqn = r->last_eqn = e;
1.41 schwarze 1814:
1815: if (buf) {
1816: poff = 0;
1817: eqn_read(&r->eqn, line, buf, offs, &poff);
1818: }
1819: }
1820:
1821: static enum rofferr
1822: roff_EQ(ROFF_ARGS)
1823: {
1824:
1825: roff_openeqn(r, *bufp + pos, ln, ppos, NULL);
1.32 schwarze 1826: return(ROFF_IGN);
1827: }
1828:
1829: static enum rofferr
1830: roff_EN(ROFF_ARGS)
1831: {
1832:
1.35 schwarze 1833: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.32 schwarze 1834: return(ROFF_IGN);
1835: }
1836:
1837: static enum rofferr
1.27 schwarze 1838: roff_TS(ROFF_ARGS)
1839: {
1.49 schwarze 1840: struct tbl_node *tbl;
1.27 schwarze 1841:
1842: if (r->tbl) {
1.35 schwarze 1843: mandoc_msg(MANDOCERR_SCOPEBROKEN, r->parse, ln, ppos, NULL);
1.41 schwarze 1844: tbl_end(&r->tbl);
1.27 schwarze 1845: }
1846:
1.49 schwarze 1847: tbl = tbl_alloc(ppos, ln, r->parse);
1.27 schwarze 1848:
1849: if (r->last_tbl)
1.49 schwarze 1850: r->last_tbl->next = tbl;
1.27 schwarze 1851: else
1.49 schwarze 1852: r->first_tbl = r->last_tbl = tbl;
1.27 schwarze 1853:
1.49 schwarze 1854: r->tbl = r->last_tbl = tbl;
1.27 schwarze 1855: return(ROFF_IGN);
1856: }
1857:
1858: static enum rofferr
1.48 schwarze 1859: roff_cc(ROFF_ARGS)
1860: {
1861: const char *p;
1862:
1863: p = *bufp + pos;
1864:
1865: if ('\0' == *p || '.' == (r->control = *p++))
1866: r->control = 0;
1867:
1868: if ('\0' != *p)
1869: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
1870:
1871: return(ROFF_IGN);
1872: }
1873:
1874: static enum rofferr
1.42 schwarze 1875: roff_tr(ROFF_ARGS)
1876: {
1877: const char *p, *first, *second;
1878: size_t fsz, ssz;
1879: enum mandoc_esc esc;
1880:
1881: p = *bufp + pos;
1882:
1883: if ('\0' == *p) {
1884: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
1885: return(ROFF_IGN);
1886: }
1887:
1888: while ('\0' != *p) {
1889: fsz = ssz = 1;
1890:
1891: first = p++;
1892: if ('\\' == *first) {
1893: esc = mandoc_escape(&p, NULL, NULL);
1894: if (ESCAPE_ERROR == esc) {
1.92 ! schwarze 1895: mandoc_msg(MANDOCERR_ESC_BAD, r->parse,
! 1896: ln, (int)(p - *bufp), first);
1.42 schwarze 1897: return(ROFF_IGN);
1898: }
1899: fsz = (size_t)(p - first);
1900: }
1901:
1902: second = p++;
1903: if ('\\' == *second) {
1904: esc = mandoc_escape(&p, NULL, NULL);
1905: if (ESCAPE_ERROR == esc) {
1.92 ! schwarze 1906: mandoc_msg(MANDOCERR_ESC_BAD, r->parse,
! 1907: ln, (int)(p - *bufp), second);
1.42 schwarze 1908: return(ROFF_IGN);
1909: }
1910: ssz = (size_t)(p - second);
1911: } else if ('\0' == *second) {
1.80 schwarze 1912: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse,
1913: ln, (int)(p - *bufp), NULL);
1.42 schwarze 1914: second = " ";
1915: p--;
1916: }
1917:
1918: if (fsz > 1) {
1.80 schwarze 1919: roff_setstrn(&r->xmbtab, first, fsz,
1920: second, ssz, 0);
1.42 schwarze 1921: continue;
1922: }
1923:
1924: if (NULL == r->xtab)
1.80 schwarze 1925: r->xtab = mandoc_calloc(128,
1926: sizeof(struct roffstr));
1.42 schwarze 1927:
1928: free(r->xtab[(int)*first].p);
1929: r->xtab[(int)*first].p = mandoc_strndup(second, ssz);
1930: r->xtab[(int)*first].sz = ssz;
1931: }
1932:
1933: return(ROFF_IGN);
1934: }
1935:
1936: static enum rofferr
1.14 schwarze 1937: roff_so(ROFF_ARGS)
1938: {
1939: char *name;
1.15 schwarze 1940:
1.83 schwarze 1941: name = *bufp + pos;
1942: mandoc_vmsg(MANDOCERR_SO, r->parse, ln, ppos, ".so %s", name);
1.14 schwarze 1943:
1.22 schwarze 1944: /*
1945: * Handle `so'. Be EXTREMELY careful, as we shouldn't be
1946: * opening anything that's not in our cwd or anything beneath
1947: * it. Thus, explicitly disallow traversing up the file-system
1948: * or using absolute paths.
1949: */
1950:
1.14 schwarze 1951: if ('/' == *name || strstr(name, "../") || strstr(name, "/..")) {
1.83 schwarze 1952: mandoc_vmsg(MANDOCERR_SO_PATH, r->parse, ln, ppos,
1953: ".so %s", name);
1.14 schwarze 1954: return(ROFF_ERR);
1955: }
1956:
1957: *offs = pos;
1958: return(ROFF_SO);
1.7 schwarze 1959: }
1960:
1.16 schwarze 1961: static enum rofferr
1962: roff_userdef(ROFF_ARGS)
1.12 schwarze 1963: {
1.16 schwarze 1964: const char *arg[9];
1965: char *cp, *n1, *n2;
1.25 schwarze 1966: int i;
1.12 schwarze 1967:
1.16 schwarze 1968: /*
1969: * Collect pointers to macro argument strings
1.61 schwarze 1970: * and NUL-terminate them.
1.16 schwarze 1971: */
1972: cp = *bufp + pos;
1.25 schwarze 1973: for (i = 0; i < 9; i++)
1.26 schwarze 1974: arg[i] = '\0' == *cp ? "" :
1.35 schwarze 1975: mandoc_getarg(r->parse, &cp, ln, &pos);
1.16 schwarze 1976:
1977: /*
1978: * Expand macro arguments.
1.12 schwarze 1979: */
1.16 schwarze 1980: *szp = 0;
1981: n1 = cp = mandoc_strdup(r->current_string);
1982: while (NULL != (cp = strstr(cp, "\\$"))) {
1983: i = cp[2] - '1';
1984: if (0 > i || 8 < i) {
1985: /* Not an argument invocation. */
1986: cp += 2;
1987: continue;
1988: }
1.82 schwarze 1989: *cp = '\0';
1990: *szp = mandoc_asprintf(&n2, "%s%s%s",
1991: n1, arg[i], cp + 3) + 1;
1.16 schwarze 1992: cp = n2 + (cp - n1);
1993: free(n1);
1994: n1 = n2;
1.12 schwarze 1995: }
1996:
1.16 schwarze 1997: /*
1998: * Replace the macro invocation
1999: * by the expanded macro.
2000: */
2001: free(*bufp);
2002: *bufp = n1;
2003: if (0 == *szp)
2004: *szp = strlen(*bufp) + 1;
2005:
1.19 schwarze 2006: return(*szp > 1 && '\n' == (*bufp)[(int)*szp - 2] ?
1.16 schwarze 2007: ROFF_REPARSE : ROFF_APPEND);
1.12 schwarze 2008: }
1.28 schwarze 2009:
1.85 schwarze 2010: static size_t
1.28 schwarze 2011: roff_getname(struct roff *r, char **cpp, int ln, int pos)
2012: {
2013: char *name, *cp;
1.85 schwarze 2014: size_t namesz;
1.28 schwarze 2015:
2016: name = *cpp;
2017: if ('\0' == *name)
1.85 schwarze 2018: return(0);
1.28 schwarze 2019:
1.85 schwarze 2020: /* Read until end of name and terminate it with NUL. */
2021: for (cp = name; 1; cp++) {
2022: if ('\0' == *cp || ' ' == *cp) {
2023: namesz = cp - name;
2024: break;
2025: }
1.28 schwarze 2026: if ('\\' != *cp)
2027: continue;
1.88 schwarze 2028: namesz = cp - name;
2029: if ('{' == cp[1] || '}' == cp[1])
2030: break;
1.28 schwarze 2031: cp++;
2032: if ('\\' == *cp)
2033: continue;
1.35 schwarze 2034: mandoc_msg(MANDOCERR_NAMESC, r->parse, ln, pos, NULL);
1.85 schwarze 2035: mandoc_escape((const char **)&cp, NULL, NULL);
2036: break;
1.28 schwarze 2037: }
2038:
2039: /* Read past spaces. */
2040: while (' ' == *cp)
2041: cp++;
2042:
2043: *cpp = cp;
1.85 schwarze 2044: return(namesz);
1.28 schwarze 2045: }
2046:
1.16 schwarze 2047: /*
2048: * Store *string into the user-defined string called *name.
2049: * To clear an existing entry, call with (*r, *name, NULL, 0).
1.66 schwarze 2050: * append == 0: replace mode
2051: * append == 1: single-line append mode
2052: * append == 2: multiline append mode, append '\n' after each call
1.16 schwarze 2053: */
1.8 schwarze 2054: static void
1.16 schwarze 2055: roff_setstr(struct roff *r, const char *name, const char *string,
1.66 schwarze 2056: int append)
1.7 schwarze 2057: {
1.42 schwarze 2058:
2059: roff_setstrn(&r->strtab, name, strlen(name), string,
1.80 schwarze 2060: string ? strlen(string) : 0, append);
1.42 schwarze 2061: }
2062:
2063: static void
2064: roff_setstrn(struct roffkv **r, const char *name, size_t namesz,
1.66 schwarze 2065: const char *string, size_t stringsz, int append)
1.42 schwarze 2066: {
2067: struct roffkv *n;
2068: char *c;
2069: int i;
2070: size_t oldch, newch;
1.7 schwarze 2071:
1.16 schwarze 2072: /* Search for an existing string with the same name. */
1.42 schwarze 2073: n = *r;
2074:
1.84 schwarze 2075: while (n && (namesz != n->key.sz ||
2076: strncmp(n->key.p, name, namesz)))
1.7 schwarze 2077: n = n->next;
1.8 schwarze 2078:
2079: if (NULL == n) {
1.16 schwarze 2080: /* Create a new string table entry. */
1.42 schwarze 2081: n = mandoc_malloc(sizeof(struct roffkv));
2082: n->key.p = mandoc_strndup(name, namesz);
2083: n->key.sz = namesz;
2084: n->val.p = NULL;
2085: n->val.sz = 0;
2086: n->next = *r;
2087: *r = n;
1.66 schwarze 2088: } else if (0 == append) {
1.42 schwarze 2089: free(n->val.p);
2090: n->val.p = NULL;
2091: n->val.sz = 0;
1.16 schwarze 2092: }
2093:
2094: if (NULL == string)
2095: return;
2096:
2097: /*
2098: * One additional byte for the '\n' in multiline mode,
2099: * and one for the terminating '\0'.
2100: */
1.66 schwarze 2101: newch = stringsz + (1 < append ? 2u : 1u);
1.42 schwarze 2102:
2103: if (NULL == n->val.p) {
2104: n->val.p = mandoc_malloc(newch);
2105: *n->val.p = '\0';
1.16 schwarze 2106: oldch = 0;
2107: } else {
1.42 schwarze 2108: oldch = n->val.sz;
2109: n->val.p = mandoc_realloc(n->val.p, oldch + newch);
1.16 schwarze 2110: }
2111:
2112: /* Skip existing content in the destination buffer. */
1.42 schwarze 2113: c = n->val.p + (int)oldch;
1.16 schwarze 2114:
2115: /* Append new content to the destination buffer. */
1.42 schwarze 2116: i = 0;
2117: while (i < (int)stringsz) {
1.16 schwarze 2118: /*
2119: * Rudimentary roff copy mode:
2120: * Handle escaped backslashes.
2121: */
1.42 schwarze 2122: if ('\\' == string[i] && '\\' == string[i + 1])
2123: i++;
2124: *c++ = string[i++];
1.16 schwarze 2125: }
1.8 schwarze 2126:
1.16 schwarze 2127: /* Append terminating bytes. */
1.66 schwarze 2128: if (1 < append)
1.16 schwarze 2129: *c++ = '\n';
1.42 schwarze 2130:
1.16 schwarze 2131: *c = '\0';
1.42 schwarze 2132: n->val.sz = (int)(c - n->val.p);
1.7 schwarze 2133: }
2134:
1.8 schwarze 2135: static const char *
2136: roff_getstrn(const struct roff *r, const char *name, size_t len)
1.7 schwarze 2137: {
1.42 schwarze 2138: const struct roffkv *n;
1.64 schwarze 2139: int i;
1.7 schwarze 2140:
1.42 schwarze 2141: for (n = r->strtab; n; n = n->next)
1.80 schwarze 2142: if (0 == strncmp(name, n->key.p, len) &&
2143: '\0' == n->key.p[(int)len])
1.42 schwarze 2144: return(n->val.p);
1.64 schwarze 2145:
2146: for (i = 0; i < PREDEFS_MAX; i++)
2147: if (0 == strncmp(name, predefs[i].name, len) &&
2148: '\0' == predefs[i].name[(int)len])
2149: return(predefs[i].str);
1.8 schwarze 2150:
1.42 schwarze 2151: return(NULL);
1.7 schwarze 2152: }
2153:
1.8 schwarze 2154: static void
1.42 schwarze 2155: roff_freestr(struct roffkv *r)
1.7 schwarze 2156: {
1.42 schwarze 2157: struct roffkv *n, *nn;
1.7 schwarze 2158:
1.42 schwarze 2159: for (n = r; n; n = nn) {
2160: free(n->key.p);
2161: free(n->val.p);
1.7 schwarze 2162: nn = n->next;
2163: free(n);
2164: }
1.27 schwarze 2165: }
2166:
2167: const struct tbl_span *
2168: roff_span(const struct roff *r)
2169: {
1.80 schwarze 2170:
1.27 schwarze 2171: return(r->tbl ? tbl_span(r->tbl) : NULL);
1.32 schwarze 2172: }
2173:
2174: const struct eqn *
2175: roff_eqn(const struct roff *r)
2176: {
1.80 schwarze 2177:
1.32 schwarze 2178: return(r->last_eqn ? &r->last_eqn->eqn : NULL);
1.42 schwarze 2179: }
2180:
2181: /*
2182: * Duplicate an input string, making the appropriate character
2183: * conversations (as stipulated by `tr') along the way.
2184: * Returns a heap-allocated string with all the replacements made.
2185: */
2186: char *
2187: roff_strdup(const struct roff *r, const char *p)
2188: {
2189: const struct roffkv *cp;
2190: char *res;
2191: const char *pp;
2192: size_t ssz, sz;
2193: enum mandoc_esc esc;
2194:
2195: if (NULL == r->xmbtab && NULL == r->xtab)
2196: return(mandoc_strdup(p));
2197: else if ('\0' == *p)
2198: return(mandoc_strdup(""));
2199:
2200: /*
2201: * Step through each character looking for term matches
2202: * (remember that a `tr' can be invoked with an escape, which is
2203: * a glyph but the escape is multi-character).
2204: * We only do this if the character hash has been initialised
2205: * and the string is >0 length.
2206: */
2207:
2208: res = NULL;
2209: ssz = 0;
2210:
2211: while ('\0' != *p) {
2212: if ('\\' != *p && r->xtab && r->xtab[(int)*p].p) {
2213: sz = r->xtab[(int)*p].sz;
2214: res = mandoc_realloc(res, ssz + sz + 1);
2215: memcpy(res + ssz, r->xtab[(int)*p].p, sz);
2216: ssz += sz;
2217: p++;
2218: continue;
2219: } else if ('\\' != *p) {
2220: res = mandoc_realloc(res, ssz + 2);
2221: res[ssz++] = *p++;
2222: continue;
2223: }
2224:
2225: /* Search for term matches. */
2226: for (cp = r->xmbtab; cp; cp = cp->next)
2227: if (0 == strncmp(p, cp->key.p, cp->key.sz))
2228: break;
2229:
2230: if (NULL != cp) {
2231: /*
2232: * A match has been found.
2233: * Append the match to the array and move
2234: * forward by its keysize.
2235: */
1.80 schwarze 2236: res = mandoc_realloc(res,
2237: ssz + cp->val.sz + 1);
1.42 schwarze 2238: memcpy(res + ssz, cp->val.p, cp->val.sz);
2239: ssz += cp->val.sz;
2240: p += (int)cp->key.sz;
2241: continue;
2242: }
2243:
2244: /*
2245: * Handle escapes carefully: we need to copy
2246: * over just the escape itself, or else we might
2247: * do replacements within the escape itself.
2248: * Make sure to pass along the bogus string.
2249: */
2250: pp = p++;
2251: esc = mandoc_escape(&p, NULL, NULL);
2252: if (ESCAPE_ERROR == esc) {
2253: sz = strlen(pp);
2254: res = mandoc_realloc(res, ssz + sz + 1);
2255: memcpy(res + ssz, pp, sz);
2256: break;
2257: }
1.80 schwarze 2258: /*
2259: * We bail out on bad escapes.
1.42 schwarze 2260: * No need to warn: we already did so when
2261: * roff_res() was called.
2262: */
2263: sz = (int)(p - pp);
2264: res = mandoc_realloc(res, ssz + sz + 1);
2265: memcpy(res + ssz, pp, sz);
2266: ssz += sz;
2267: }
2268:
2269: res[(int)ssz] = '\0';
2270: return(res);
1.48 schwarze 2271: }
2272:
2273: /*
1.80 schwarze 2274: * Find out whether a line is a macro line or not.
1.48 schwarze 2275: * If it is, adjust the current position and return one; if it isn't,
2276: * return zero and don't change the current position.
2277: * If the control character has been set with `.cc', then let that grain
2278: * precedence.
2279: * This is slighly contrary to groff, where using the non-breaking
2280: * control character when `cc' has been invoked will cause the
2281: * non-breaking macro contents to be printed verbatim.
2282: */
2283: int
2284: roff_getcontrol(const struct roff *r, const char *cp, int *ppos)
2285: {
2286: int pos;
2287:
2288: pos = *ppos;
2289:
2290: if (0 != r->control && cp[pos] == r->control)
2291: pos++;
2292: else if (0 != r->control)
2293: return(0);
2294: else if ('\\' == cp[pos] && '.' == cp[pos + 1])
2295: pos += 2;
2296: else if ('.' == cp[pos] || '\'' == cp[pos])
2297: pos++;
2298: else
2299: return(0);
2300:
2301: while (' ' == cp[pos] || '\t' == cp[pos])
2302: pos++;
2303:
2304: *ppos = pos;
2305: return(1);
1.1 schwarze 2306: }
![[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