Annotation of src/usr.bin/mandoc/roff.c, Revision 1.88
1.88 ! schwarze 1: /* $Id: roff.c,v 1.87 2014/06/29 23:23:16 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 */
499: int irc; /* return code from roff_evalnum() */
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))
541: mandoc_msg(MANDOCERR_BADESCAPE, r->parse,
542: ln, (int)(stesc - *bufp), NULL);
543: continue;
1.42 schwarze 544: }
545:
1.78 schwarze 546: if (EXPAND_LIMIT < ++expand_count) {
547: mandoc_msg(MANDOCERR_ROFFLOOP, r->parse,
548: ln, (int)(stesc - *bufp), NULL);
549: return(ROFF_IGN);
550: }
1.23 schwarze 551:
552: /*
553: * The third character decides the length
1.53 schwarze 554: * of the name of the string or register.
1.23 schwarze 555: * Save a pointer to the name.
556: */
557:
1.79 schwarze 558: if ('\0' == term) {
559: switch (*++cp) {
1.80 schwarze 560: case '\0':
1.79 schwarze 561: maxl = 0;
562: break;
1.80 schwarze 563: case '(':
1.79 schwarze 564: cp++;
565: maxl = 2;
566: break;
1.80 schwarze 567: case '[':
1.79 schwarze 568: cp++;
569: term = ']';
570: maxl = 0;
571: break;
572: default:
573: maxl = 1;
574: break;
575: }
576: } else {
577: cp += 2;
1.8 schwarze 578: maxl = 0;
579: }
1.23 schwarze 580: stnam = cp;
1.8 schwarze 581:
1.23 schwarze 582: /* Advance to the end of the name. */
1.8 schwarze 583:
1.53 schwarze 584: for (naml = 0; 0 == maxl || naml < maxl; naml++, cp++) {
1.42 schwarze 585: if ('\0' == *cp) {
1.80 schwarze 586: mandoc_msg(MANDOCERR_BADESCAPE, r->parse,
587: ln, (int)(stesc - *bufp), NULL);
1.79 schwarze 588: break;
1.42 schwarze 589: }
1.79 schwarze 590: if (0 == maxl && *cp == term) {
591: cp++;
1.8 schwarze 592: break;
1.79 schwarze 593: }
1.8 schwarze 594: }
595:
1.23 schwarze 596: /*
597: * Retrieve the replacement string; if it is
598: * undefined, resume searching for escapes.
599: */
600:
1.79 schwarze 601: switch (stesc[1]) {
1.80 schwarze 602: case '*':
1.53 schwarze 603: res = roff_getstrn(r, stnam, naml);
1.79 schwarze 604: break;
1.80 schwarze 605: case 'B':
1.79 schwarze 606: npos = 0;
607: irc = roff_evalnum(stnam, &npos, NULL, 0);
608: ubuf[0] = irc && stnam + npos + 1 == cp
609: ? '1' : '0';
610: ubuf[1] = '\0';
611: break;
1.80 schwarze 612: case 'n':
1.81 schwarze 613: (void)snprintf(ubuf, sizeof(ubuf), "%d",
1.53 schwarze 614: roff_getregn(r, stnam, naml));
1.79 schwarze 615: break;
1.80 schwarze 616: case 'w':
1.81 schwarze 617: (void)snprintf(ubuf, sizeof(ubuf), "%d",
1.79 schwarze 618: 24 * (int)naml);
619: break;
620: }
1.8 schwarze 621:
622: if (NULL == res) {
1.80 schwarze 623: mandoc_msg(MANDOCERR_BADESCAPE, r->parse,
624: ln, (int)(stesc - *bufp), NULL);
1.37 schwarze 625: res = "";
1.8 schwarze 626: }
627:
1.23 schwarze 628: /* Replace the escape sequence by the string. */
629:
1.82 schwarze 630: *stesc = '\0';
631: *szp = mandoc_asprintf(&nbuf, "%s%s%s",
632: *bufp, res, cp) + 1;
1.8 schwarze 633:
1.78 schwarze 634: /* Prepare for the next replacement. */
1.8 schwarze 635:
1.78 schwarze 636: start = nbuf + pos;
1.82 schwarze 637: stesc = nbuf + (stesc - *bufp) + strlen(res);
1.8 schwarze 638: free(*bufp);
1.53 schwarze 639: *bufp = nbuf;
1.42 schwarze 640: }
1.45 schwarze 641: return(ROFF_CONT);
1.42 schwarze 642: }
643:
644: /*
1.51 schwarze 645: * Process text streams:
646: * Convert all breakable hyphens into ASCII_HYPH.
647: * Decrement and spring input line trap.
1.42 schwarze 648: */
649: static enum rofferr
1.51 schwarze 650: roff_parsetext(char **bufp, size_t *szp, int pos, int *offs)
1.42 schwarze 651: {
652: size_t sz;
653: const char *start;
1.51 schwarze 654: char *p;
655: int isz;
1.42 schwarze 656: enum mandoc_esc esc;
657:
1.51 schwarze 658: start = p = *bufp + pos;
1.42 schwarze 659:
660: while ('\0' != *p) {
661: sz = strcspn(p, "-\\");
662: p += sz;
663:
664: if ('\0' == *p)
665: break;
666:
667: if ('\\' == *p) {
668: /* Skip over escapes. */
669: p++;
1.62 schwarze 670: esc = mandoc_escape((const char **)&p, NULL, NULL);
1.42 schwarze 671: if (ESCAPE_ERROR == esc)
672: break;
673: continue;
674: } else if (p == start) {
675: p++;
676: continue;
677: }
678:
1.44 schwarze 679: if (isalpha((unsigned char)p[-1]) &&
680: isalpha((unsigned char)p[1]))
1.42 schwarze 681: *p = ASCII_HYPH;
682: p++;
1.8 schwarze 683: }
684:
1.51 schwarze 685: /* Spring the input line trap. */
686: if (1 == roffit_lines) {
1.75 schwarze 687: isz = mandoc_asprintf(&p, "%s\n.%s", *bufp, roffit_macro);
1.51 schwarze 688: free(*bufp);
689: *bufp = p;
690: *szp = isz + 1;
691: *offs = 0;
692: free(roffit_macro);
693: roffit_lines = 0;
694: return(ROFF_REPARSE);
695: } else if (1 < roffit_lines)
696: --roffit_lines;
1.42 schwarze 697: return(ROFF_CONT);
1.8 schwarze 698: }
699:
1.1 schwarze 700: enum rofferr
1.80 schwarze 701: roff_parseln(struct roff *r, int ln, char **bufp,
1.6 schwarze 702: size_t *szp, int pos, int *offs)
1.1 schwarze 703: {
704: enum rofft t;
1.27 schwarze 705: enum rofferr e;
1.35 schwarze 706: int ppos, ctl;
1.1 schwarze 707:
1.2 schwarze 708: /*
1.8 schwarze 709: * Run the reserved-word filter only if we have some reserved
710: * words to fill in.
711: */
712:
1.45 schwarze 713: e = roff_res(r, bufp, szp, ln, pos);
714: if (ROFF_IGN == e)
715: return(e);
716: assert(ROFF_CONT == e);
1.8 schwarze 717:
1.35 schwarze 718: ppos = pos;
1.48 schwarze 719: ctl = roff_getcontrol(r, *bufp, &pos);
1.35 schwarze 720:
1.8 schwarze 721: /*
1.2 schwarze 722: * First, if a scope is open and we're not a macro, pass the
723: * text through the macro's filter. If a scope isn't open and
724: * we're not a macro, just let it through.
1.32 schwarze 725: * Finally, if there's an equation scope open, divert it into it
726: * no matter our state.
1.2 schwarze 727: */
728:
1.35 schwarze 729: if (r->last && ! ctl) {
1.2 schwarze 730: t = r->last->tok;
731: assert(roffs[t].text);
1.80 schwarze 732: e = (*roffs[t].text)(r, t, bufp, szp, ln, pos, pos, offs);
1.27 schwarze 733: assert(ROFF_IGN == e || ROFF_CONT == e);
1.32 schwarze 734: if (ROFF_CONT != e)
735: return(e);
1.54 schwarze 736: }
737: if (r->eqn)
738: return(eqn_read(&r->eqn, ln, *bufp, ppos, offs));
739: if ( ! ctl) {
1.32 schwarze 740: if (r->tbl)
1.35 schwarze 741: return(tbl_read(r->tbl, ln, *bufp, pos));
1.51 schwarze 742: return(roff_parsetext(bufp, szp, pos, offs));
1.54 schwarze 743: }
1.2 schwarze 744:
745: /*
746: * If a scope is open, go to the child handler for that macro,
747: * as it may want to preprocess before doing anything with it.
1.32 schwarze 748: * Don't do so if an equation is open.
1.2 schwarze 749: */
750:
751: if (r->last) {
1.1 schwarze 752: t = r->last->tok;
753: assert(roffs[t].sub);
1.80 schwarze 754: return((*roffs[t].sub)(r, t, bufp, szp,
755: ln, ppos, pos, offs));
1.2 schwarze 756: }
757:
758: /*
759: * Lastly, as we've no scope open, try to look up and execute
760: * the new macro. If no macro is found, simply return and let
761: * the compilers handle it.
762: */
763:
1.87 schwarze 764: if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos, ln, ppos)))
1.1 schwarze 765: return(ROFF_CONT);
766:
1.2 schwarze 767: assert(roffs[t].proc);
1.80 schwarze 768: return((*roffs[t].proc)(r, t, bufp, szp, ln, ppos, pos, offs));
1.2 schwarze 769: }
770:
1.27 schwarze 771: void
1.2 schwarze 772: roff_endparse(struct roff *r)
773: {
1.1 schwarze 774:
1.27 schwarze 775: if (r->last)
1.35 schwarze 776: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.80 schwarze 777: r->last->line, r->last->col, NULL);
1.27 schwarze 778:
1.32 schwarze 779: if (r->eqn) {
1.80 schwarze 780: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
781: r->eqn->eqn.ln, r->eqn->eqn.pos, NULL);
1.41 schwarze 782: eqn_end(&r->eqn);
1.32 schwarze 783: }
784:
1.27 schwarze 785: if (r->tbl) {
1.80 schwarze 786: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
787: r->tbl->line, r->tbl->pos, NULL);
1.41 schwarze 788: tbl_end(&r->tbl);
1.27 schwarze 789: }
1.1 schwarze 790: }
791:
792: /*
793: * Parse a roff node's type from the input buffer. This must be in the
794: * form of ".foo xxx" in the usual way.
795: */
796: static enum rofft
1.87 schwarze 797: roff_parse(struct roff *r, char *buf, int *pos, int ln, int ppos)
1.1 schwarze 798: {
1.87 schwarze 799: char *cp;
1.16 schwarze 800: const char *mac;
801: size_t maclen;
1.1 schwarze 802: enum rofft t;
803:
1.87 schwarze 804: cp = buf + *pos;
805:
806: if ('\0' == *cp || '"' == *cp || '\t' == *cp || ' ' == *cp)
1.1 schwarze 807: return(ROFF_MAX);
808:
1.87 schwarze 809: mac = cp;
810: maclen = roff_getname(r, &cp, ln, ppos);
1.1 schwarze 811:
1.16 schwarze 812: t = (r->current_string = roff_getstrn(r, mac, maclen))
1.42 schwarze 813: ? ROFF_USERDEF : roffhash_find(mac, maclen);
1.1 schwarze 814:
1.87 schwarze 815: if (ROFF_MAX != t)
816: *pos = cp - buf;
1.1 schwarze 817:
818: return(t);
819: }
820:
821: static enum rofferr
1.2 schwarze 822: roff_cblock(ROFF_ARGS)
1.1 schwarze 823: {
824:
1.2 schwarze 825: /*
826: * A block-close `..' should only be invoked as a child of an
827: * ignore macro, otherwise raise a warning and just ignore it.
828: */
829:
830: if (NULL == r->last) {
1.35 schwarze 831: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.2 schwarze 832: return(ROFF_IGN);
833: }
1.1 schwarze 834:
1.2 schwarze 835: switch (r->last->tok) {
1.80 schwarze 836: case ROFF_am:
1.2 schwarze 837: /* FALLTHROUGH */
1.80 schwarze 838: case ROFF_ami:
1.2 schwarze 839: /* FALLTHROUGH */
1.80 schwarze 840: case ROFF_am1:
1.2 schwarze 841: /* FALLTHROUGH */
1.80 schwarze 842: case ROFF_de:
1.23 schwarze 843: /* ROFF_de1 is remapped to ROFF_de in roff_block(). */
1.2 schwarze 844: /* FALLTHROUGH */
1.80 schwarze 845: case ROFF_dei:
1.2 schwarze 846: /* FALLTHROUGH */
1.80 schwarze 847: case ROFF_ig:
1.2 schwarze 848: break;
849: default:
1.35 schwarze 850: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.1 schwarze 851: return(ROFF_IGN);
1.2 schwarze 852: }
853:
854: if ((*bufp)[pos])
1.35 schwarze 855: mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.2 schwarze 856:
857: roffnode_pop(r);
858: roffnode_cleanscope(r);
859: return(ROFF_IGN);
860:
861: }
1.1 schwarze 862:
1.2 schwarze 863: static void
864: roffnode_cleanscope(struct roff *r)
865: {
1.1 schwarze 866:
1.2 schwarze 867: while (r->last) {
1.46 schwarze 868: if (--r->last->endspan != 0)
1.2 schwarze 869: break;
870: roffnode_pop(r);
871: }
872: }
1.1 schwarze 873:
1.68 schwarze 874: static void
875: roff_ccond(struct roff *r, int ln, int ppos)
1.2 schwarze 876: {
1.1 schwarze 877:
1.2 schwarze 878: if (NULL == r->last) {
1.35 schwarze 879: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.68 schwarze 880: return;
1.2 schwarze 881: }
1.1 schwarze 882:
1.2 schwarze 883: switch (r->last->tok) {
1.80 schwarze 884: case ROFF_el:
1.2 schwarze 885: /* FALLTHROUGH */
1.80 schwarze 886: case ROFF_ie:
1.2 schwarze 887: /* FALLTHROUGH */
1.80 schwarze 888: case ROFF_if:
1.2 schwarze 889: break;
890: default:
1.35 schwarze 891: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.68 schwarze 892: return;
1.2 schwarze 893: }
1.1 schwarze 894:
1.2 schwarze 895: if (r->last->endspan > -1) {
1.35 schwarze 896: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.68 schwarze 897: return;
1.2 schwarze 898: }
899:
900: roffnode_pop(r);
901: roffnode_cleanscope(r);
1.68 schwarze 902: return;
1.1 schwarze 903: }
904:
905: static enum rofferr
1.2 schwarze 906: roff_block(ROFF_ARGS)
1.1 schwarze 907: {
1.86 schwarze 908: char *name, *cp;
909: size_t namesz;
1.16 schwarze 910:
1.86 schwarze 911: name = cp = *bufp + pos;
912: namesz = 0;
1.2 schwarze 913:
1.16 schwarze 914: if (ROFF_ig != tok) {
1.86 schwarze 915: if ('\0' == *cp) {
1.35 schwarze 916: mandoc_msg(MANDOCERR_NOARGS, r->parse, ln, ppos, NULL);
1.16 schwarze 917: return(ROFF_IGN);
918: }
1.22 schwarze 919:
920: /*
921: * Re-write `de1', since we don't really care about
922: * groff's strange compatibility mode, into `de'.
923: */
924:
1.18 schwarze 925: if (ROFF_de1 == tok)
926: tok = ROFF_de;
1.86 schwarze 927: else if (ROFF_de != tok)
1.35 schwarze 928: mandoc_msg(MANDOCERR_REQUEST, r->parse, ln, ppos,
1.21 schwarze 929: roffs[tok].name);
1.22 schwarze 930:
1.86 schwarze 931: namesz = roff_getname(r, &cp, ln, ppos);
932: name[namesz] = '\0';
933: } else
934: name = NULL;
1.2 schwarze 935:
1.16 schwarze 936: roffnode_push(r, tok, name, ln, ppos);
937:
938: /*
939: * At the beginning of a `de' macro, clear the existing string
940: * with the same name, if there is one. New content will be
1.66 schwarze 941: * appended from roff_block_text() in multiline mode.
1.16 schwarze 942: */
1.22 schwarze 943:
1.86 schwarze 944: if (namesz && ROFF_de == tok)
945: roff_setstrn(&r->strtab, name, namesz, "", 0, 0);
1.2 schwarze 946:
1.86 schwarze 947: if ('\0' == *cp)
1.2 schwarze 948: return(ROFF_IGN);
1.1 schwarze 949:
1.22 schwarze 950: /* If present, process the custom end-of-line marker. */
951:
1.86 schwarze 952: name = cp;
953: namesz = roff_getname(r, &cp, ln, ppos);
954: if (namesz)
955: r->last->end = mandoc_strndup(name, namesz);
1.2 schwarze 956:
1.86 schwarze 957: if ('\0' != *cp)
1.35 schwarze 958: mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.1 schwarze 959:
960: return(ROFF_IGN);
961: }
962:
963: static enum rofferr
1.2 schwarze 964: roff_block_sub(ROFF_ARGS)
1.1 schwarze 965: {
1.2 schwarze 966: enum rofft t;
967: int i, j;
968:
969: /*
970: * First check whether a custom macro exists at this level. If
971: * it does, then check against it. This is some of groff's
972: * stranger behaviours. If we encountered a custom end-scope
973: * tag and that tag also happens to be a "real" macro, then we
974: * need to try interpreting it again as a real macro. If it's
975: * not, then return ignore. Else continue.
976: */
977:
978: if (r->last->end) {
1.35 schwarze 979: for (i = pos, j = 0; r->last->end[j]; j++, i++)
1.2 schwarze 980: if ((*bufp)[i] != r->last->end[j])
981: break;
1.1 schwarze 982:
1.80 schwarze 983: if ('\0' == r->last->end[j] &&
984: ('\0' == (*bufp)[i] ||
985: ' ' == (*bufp)[i] ||
986: '\t' == (*bufp)[i])) {
1.2 schwarze 987: roffnode_pop(r);
988: roffnode_cleanscope(r);
1.1 schwarze 989:
1.35 schwarze 990: while (' ' == (*bufp)[i] || '\t' == (*bufp)[i])
991: i++;
992:
993: pos = i;
1.87 schwarze 994: if (ROFF_MAX != roff_parse(r, *bufp, &pos, ln, ppos))
1.2 schwarze 995: return(ROFF_RERUN);
996: return(ROFF_IGN);
997: }
1.1 schwarze 998: }
999:
1.2 schwarze 1000: /*
1001: * If we have no custom end-query or lookup failed, then try
1002: * pulling it out of the hashtable.
1003: */
1.1 schwarze 1004:
1.87 schwarze 1005: t = roff_parse(r, *bufp, &pos, ln, ppos);
1.1 schwarze 1006:
1.16 schwarze 1007: /*
1008: * Macros other than block-end are only significant
1009: * in `de' blocks; elsewhere, simply throw them away.
1010: */
1011: if (ROFF_cblock != t) {
1012: if (ROFF_de == tok)
1.66 schwarze 1013: roff_setstr(r, r->last->name, *bufp + ppos, 2);
1.1 schwarze 1014: return(ROFF_IGN);
1.16 schwarze 1015: }
1.1 schwarze 1016:
1.2 schwarze 1017: assert(roffs[t].proc);
1.80 schwarze 1018: return((*roffs[t].proc)(r, t, bufp, szp, ln, ppos, pos, offs));
1.2 schwarze 1019: }
1020:
1021: static enum rofferr
1022: roff_block_text(ROFF_ARGS)
1023: {
1024:
1.16 schwarze 1025: if (ROFF_de == tok)
1.66 schwarze 1026: roff_setstr(r, r->last->name, *bufp + pos, 2);
1.16 schwarze 1027:
1.2 schwarze 1028: return(ROFF_IGN);
1029: }
1030:
1031: static enum rofferr
1032: roff_cond_sub(ROFF_ARGS)
1033: {
1034: enum rofft t;
1.37 schwarze 1035: char *ep;
1.71 schwarze 1036: int rr;
1.2 schwarze 1037:
1038: rr = r->last->rule;
1.37 schwarze 1039: roffnode_cleanscope(r);
1.87 schwarze 1040: t = roff_parse(r, *bufp, &pos, ln, ppos);
1.2 schwarze 1041:
1.37 schwarze 1042: /*
1.50 schwarze 1043: * Fully handle known macros when they are structurally
1044: * required or when the conditional evaluated to true.
1.5 schwarze 1045: */
1046:
1.50 schwarze 1047: if ((ROFF_MAX != t) &&
1.71 schwarze 1048: (rr || ROFFMAC_STRUCT & roffs[t].flags)) {
1.50 schwarze 1049: assert(roffs[t].proc);
1050: return((*roffs[t].proc)(r, t, bufp, szp,
1.80 schwarze 1051: ln, ppos, pos, offs));
1.50 schwarze 1052: }
1.39 schwarze 1053:
1.69 schwarze 1054: /*
1055: * If `\}' occurs on a macro line without a preceding macro,
1056: * drop the line completely.
1057: */
1058:
1059: ep = *bufp + pos;
1060: if ('\\' == ep[0] && '}' == ep[1])
1.71 schwarze 1061: rr = 0;
1.69 schwarze 1062:
1.50 schwarze 1063: /* Always check for the closing delimiter `\}'. */
1.39 schwarze 1064:
1.50 schwarze 1065: while (NULL != (ep = strchr(ep, '\\'))) {
1.70 schwarze 1066: if ('}' == *(++ep)) {
1067: *ep = '&';
1068: roff_ccond(r, ln, ep - *bufp - 1);
1069: }
1070: ++ep;
1.50 schwarze 1071: }
1.71 schwarze 1072: return(rr ? ROFF_CONT : ROFF_IGN);
1.2 schwarze 1073: }
1074:
1075: static enum rofferr
1076: roff_cond_text(ROFF_ARGS)
1077: {
1.37 schwarze 1078: char *ep;
1.71 schwarze 1079: int rr;
1.2 schwarze 1080:
1081: rr = r->last->rule;
1.37 schwarze 1082: roffnode_cleanscope(r);
1.1 schwarze 1083:
1.70 schwarze 1084: ep = *bufp + pos;
1085: while (NULL != (ep = strchr(ep, '\\'))) {
1086: if ('}' == *(++ep)) {
1087: *ep = '&';
1088: roff_ccond(r, ln, ep - *bufp - 1);
1089: }
1090: ++ep;
1.2 schwarze 1091: }
1.71 schwarze 1092: return(rr ? ROFF_CONT : ROFF_IGN);
1.2 schwarze 1093: }
1094:
1.77 schwarze 1095: /*
1096: * Parse a single signed integer number. Stop at the first non-digit.
1097: * If there is at least one digit, return success and advance the
1098: * parse point, else return failure and let the parse point unchanged.
1099: * Ignore overflows, treat them just like the C language.
1100: */
1.56 schwarze 1101: static int
1102: roff_getnum(const char *v, int *pos, int *res)
1103: {
1.79 schwarze 1104: int myres, n, p;
1105:
1106: if (NULL == res)
1107: res = &myres;
1.56 schwarze 1108:
1109: p = *pos;
1110: n = v[p] == '-';
1111: if (n)
1112: p++;
1113:
1114: for (*res = 0; isdigit((unsigned char)v[p]); p++)
1.77 schwarze 1115: *res = 10 * *res + v[p] - '0';
1.56 schwarze 1116: if (p == *pos + n)
1117: return 0;
1118:
1119: if (n)
1120: *res = -*res;
1121:
1122: *pos = p;
1123: return 1;
1124: }
1125:
1.71 schwarze 1126: /*
1127: * Evaluate a string comparison condition.
1128: * The first character is the delimiter.
1129: * Succeed if the string up to its second occurrence
1130: * matches the string up to its third occurence.
1131: * Advance the cursor after the third occurrence
1132: * or lacking that, to the end of the line.
1133: */
1134: static int
1135: roff_evalstrcond(const char *v, int *pos)
1136: {
1137: const char *s1, *s2, *s3;
1138: int match;
1139:
1140: match = 0;
1141: s1 = v + *pos; /* initial delimiter */
1142: s2 = s1 + 1; /* for scanning the first string */
1143: s3 = strchr(s2, *s1); /* for scanning the second string */
1144:
1145: if (NULL == s3) /* found no middle delimiter */
1146: goto out;
1147:
1148: while ('\0' != *++s3) {
1149: if (*s2 != *s3) { /* mismatch */
1150: s3 = strchr(s3, *s1);
1151: break;
1152: }
1153: if (*s3 == *s1) { /* found the final delimiter */
1154: match = 1;
1155: break;
1156: }
1157: s2++;
1158: }
1159:
1160: out:
1161: if (NULL == s3)
1162: s3 = strchr(s2, '\0');
1163: else
1164: s3++;
1165: *pos = s3 - v;
1166: return(match);
1167: }
1168:
1.77 schwarze 1169: /*
1170: * Evaluate an optionally negated single character, numerical,
1171: * or string condition.
1172: */
1.71 schwarze 1173: static int
1.5 schwarze 1174: roff_evalcond(const char *v, int *pos)
1175: {
1.77 schwarze 1176: int wanttrue, number;
1.5 schwarze 1177:
1.71 schwarze 1178: if ('!' == v[*pos]) {
1179: wanttrue = 0;
1180: (*pos)++;
1181: } else
1182: wanttrue = 1;
1183:
1.5 schwarze 1184: switch (v[*pos]) {
1.80 schwarze 1185: case 'n':
1.71 schwarze 1186: /* FALLTHROUGH */
1.80 schwarze 1187: case 'o':
1.5 schwarze 1188: (*pos)++;
1.71 schwarze 1189: return(wanttrue);
1.80 schwarze 1190: case 'c':
1.71 schwarze 1191: /* FALLTHROUGH */
1.80 schwarze 1192: case 'd':
1.71 schwarze 1193: /* FALLTHROUGH */
1.80 schwarze 1194: case 'e':
1.5 schwarze 1195: /* FALLTHROUGH */
1.80 schwarze 1196: case 'r':
1.5 schwarze 1197: /* FALLTHROUGH */
1.80 schwarze 1198: case 't':
1.5 schwarze 1199: (*pos)++;
1.71 schwarze 1200: return(!wanttrue);
1.5 schwarze 1201: default:
1202: break;
1203: }
1204:
1.77 schwarze 1205: if (roff_evalnum(v, pos, &number, 0))
1206: return((number > 0) == wanttrue);
1207: else
1.71 schwarze 1208: return(roff_evalstrcond(v, pos) == wanttrue);
1.5 schwarze 1209: }
1210:
1.2 schwarze 1211: static enum rofferr
1.21 schwarze 1212: roff_line_ignore(ROFF_ARGS)
1.6 schwarze 1213: {
1.30 schwarze 1214:
1.21 schwarze 1215: return(ROFF_IGN);
1216: }
1217:
1218: static enum rofferr
1.2 schwarze 1219: roff_cond(ROFF_ARGS)
1220: {
1.46 schwarze 1221:
1222: roffnode_push(r, tok, NULL, ln, ppos);
1.2 schwarze 1223:
1.80 schwarze 1224: /*
1.35 schwarze 1225: * An `.el' has no conditional body: it will consume the value
1226: * of the current rstack entry set in prior `ie' calls or
1.80 schwarze 1227: * defaults to DENY.
1.35 schwarze 1228: *
1229: * If we're not an `el', however, then evaluate the conditional.
1230: */
1.1 schwarze 1231:
1.46 schwarze 1232: r->last->rule = ROFF_el == tok ?
1.80 schwarze 1233: (r->rstackpos < 0 ? 0 : r->rstack[r->rstackpos--]) :
1234: roff_evalcond(*bufp, &pos);
1.2 schwarze 1235:
1.35 schwarze 1236: /*
1237: * An if-else will put the NEGATION of the current evaluated
1238: * conditional into the stack of rules.
1239: */
1240:
1.2 schwarze 1241: if (ROFF_ie == tok) {
1.35 schwarze 1242: if (r->rstackpos == RSTACK_MAX - 1) {
1.80 schwarze 1243: mandoc_msg(MANDOCERR_MEM,
1244: r->parse, ln, ppos, NULL);
1.35 schwarze 1245: return(ROFF_ERR);
1246: }
1.71 schwarze 1247: r->rstack[++r->rstackpos] = !r->last->rule;
1.2 schwarze 1248: }
1.5 schwarze 1249:
1250: /* If the parent has false as its rule, then so do we. */
1251:
1.71 schwarze 1252: if (r->last->parent && !r->last->parent->rule)
1253: r->last->rule = 0;
1.5 schwarze 1254:
1255: /*
1.46 schwarze 1256: * Determine scope.
1257: * If there is nothing on the line after the conditional,
1258: * not even whitespace, use next-line scope.
1.5 schwarze 1259: */
1.2 schwarze 1260:
1.46 schwarze 1261: if ('\0' == (*bufp)[pos]) {
1262: r->last->endspan = 2;
1263: goto out;
1264: }
1265:
1266: while (' ' == (*bufp)[pos])
1267: pos++;
1268:
1269: /* An opening brace requests multiline scope. */
1.2 schwarze 1270:
1271: if ('\\' == (*bufp)[pos] && '{' == (*bufp)[pos + 1]) {
1272: r->last->endspan = -1;
1273: pos += 2;
1.46 schwarze 1274: goto out;
1.80 schwarze 1275: }
1.2 schwarze 1276:
1277: /*
1.46 schwarze 1278: * Anything else following the conditional causes
1279: * single-line scope. Warn if the scope contains
1280: * nothing but trailing whitespace.
1.2 schwarze 1281: */
1282:
1283: if ('\0' == (*bufp)[pos])
1.46 schwarze 1284: mandoc_msg(MANDOCERR_NOARGS, r->parse, ln, ppos, NULL);
1.2 schwarze 1285:
1.46 schwarze 1286: r->last->endspan = 1;
1.1 schwarze 1287:
1.46 schwarze 1288: out:
1.2 schwarze 1289: *offs = pos;
1290: return(ROFF_RERUN);
1.1 schwarze 1291: }
1292:
1.2 schwarze 1293: static enum rofferr
1.7 schwarze 1294: roff_ds(ROFF_ARGS)
1295: {
1.85 schwarze 1296: char *string;
1297: const char *name;
1298: size_t namesz;
1.10 schwarze 1299:
1300: /*
1.85 schwarze 1301: * The first word is the name of the string.
1302: * If it is empty or terminated by an escape sequence,
1303: * abort the `ds' request without defining anything.
1.10 schwarze 1304: */
1.7 schwarze 1305:
1.85 schwarze 1306: name = string = *bufp + pos;
1.7 schwarze 1307: if ('\0' == *name)
1308: return(ROFF_IGN);
1309:
1.85 schwarze 1310: namesz = roff_getname(r, &string, ln, pos);
1311: if ('\\' == name[namesz])
1312: return(ROFF_IGN);
1313:
1314: /* Read past the initial double-quote, if any. */
1.28 schwarze 1315: if ('"' == *string)
1.7 schwarze 1316: string++;
1317:
1.10 schwarze 1318: /* The rest is the value. */
1.85 schwarze 1319: roff_setstrn(&r->strtab, name, namesz, string, strlen(string),
1320: ROFF_as == tok);
1.7 schwarze 1321: return(ROFF_IGN);
1322: }
1323:
1.77 schwarze 1324: /*
1325: * Parse a single operator, one or two characters long.
1326: * If the operator is recognized, return success and advance the
1327: * parse point, else return failure and let the parse point unchanged.
1328: */
1329: static int
1330: roff_getop(const char *v, int *pos, char *res)
1331: {
1332:
1333: *res = v[*pos];
1334:
1335: switch (*res) {
1.80 schwarze 1336: case '+':
1.77 schwarze 1337: /* FALLTHROUGH */
1.80 schwarze 1338: case '-':
1.77 schwarze 1339: /* FALLTHROUGH */
1.80 schwarze 1340: case '*':
1.77 schwarze 1341: /* FALLTHROUGH */
1.80 schwarze 1342: case '/':
1.77 schwarze 1343: /* FALLTHROUGH */
1.80 schwarze 1344: case '%':
1.77 schwarze 1345: /* FALLTHROUGH */
1.80 schwarze 1346: case '&':
1.77 schwarze 1347: /* FALLTHROUGH */
1.80 schwarze 1348: case ':':
1.77 schwarze 1349: break;
1350: case '<':
1351: switch (v[*pos + 1]) {
1.80 schwarze 1352: case '=':
1.77 schwarze 1353: *res = 'l';
1354: (*pos)++;
1355: break;
1.80 schwarze 1356: case '>':
1.77 schwarze 1357: *res = '!';
1358: (*pos)++;
1359: break;
1.80 schwarze 1360: case '?':
1.77 schwarze 1361: *res = 'i';
1362: (*pos)++;
1363: break;
1364: default:
1365: break;
1366: }
1367: break;
1368: case '>':
1369: switch (v[*pos + 1]) {
1.80 schwarze 1370: case '=':
1.77 schwarze 1371: *res = 'g';
1372: (*pos)++;
1373: break;
1.80 schwarze 1374: case '?':
1.77 schwarze 1375: *res = 'a';
1376: (*pos)++;
1377: break;
1378: default:
1379: break;
1380: }
1381: break;
1382: case '=':
1383: if ('=' == v[*pos + 1])
1384: (*pos)++;
1385: break;
1386: default:
1387: return(0);
1388: }
1389: (*pos)++;
1390:
1391: return(*res);
1392: }
1393:
1394: /*
1395: * Evaluate either a parenthesized numeric expression
1396: * or a single signed integer number.
1397: */
1398: static int
1399: roff_evalpar(const char *v, int *pos, int *res)
1400: {
1401:
1402: if ('(' != v[*pos])
1403: return(roff_getnum(v, pos, res));
1404:
1405: (*pos)++;
1406: if ( ! roff_evalnum(v, pos, res, 1))
1407: return(0);
1408:
1.79 schwarze 1409: /*
1410: * Omission of the closing parenthesis
1411: * is an error in validation mode,
1412: * but ignored in evaluation mode.
1413: */
1414:
1.77 schwarze 1415: if (')' == v[*pos])
1416: (*pos)++;
1.79 schwarze 1417: else if (NULL == res)
1418: return(0);
1.77 schwarze 1419:
1420: return(1);
1421: }
1422:
1423: /*
1424: * Evaluate a complete numeric expression.
1425: * Proceed left to right, there is no concept of precedence.
1426: */
1427: static int
1428: roff_evalnum(const char *v, int *pos, int *res, int skipwhite)
1429: {
1430: int mypos, operand2;
1431: char operator;
1432:
1433: if (NULL == pos) {
1434: mypos = 0;
1435: pos = &mypos;
1436: }
1437:
1438: if (skipwhite)
1439: while (isspace((unsigned char)v[*pos]))
1440: (*pos)++;
1441:
1442: if ( ! roff_evalpar(v, pos, res))
1443: return(0);
1444:
1445: while (1) {
1446: if (skipwhite)
1447: while (isspace((unsigned char)v[*pos]))
1448: (*pos)++;
1449:
1450: if ( ! roff_getop(v, pos, &operator))
1451: break;
1452:
1453: if (skipwhite)
1454: while (isspace((unsigned char)v[*pos]))
1455: (*pos)++;
1456:
1457: if ( ! roff_evalpar(v, pos, &operand2))
1458: return(0);
1459:
1460: if (skipwhite)
1461: while (isspace((unsigned char)v[*pos]))
1462: (*pos)++;
1.79 schwarze 1463:
1464: if (NULL == res)
1465: continue;
1.77 schwarze 1466:
1467: switch (operator) {
1.80 schwarze 1468: case '+':
1.77 schwarze 1469: *res += operand2;
1470: break;
1.80 schwarze 1471: case '-':
1.77 schwarze 1472: *res -= operand2;
1473: break;
1.80 schwarze 1474: case '*':
1.77 schwarze 1475: *res *= operand2;
1476: break;
1.80 schwarze 1477: case '/':
1.77 schwarze 1478: *res /= operand2;
1479: break;
1.80 schwarze 1480: case '%':
1.77 schwarze 1481: *res %= operand2;
1482: break;
1.80 schwarze 1483: case '<':
1.77 schwarze 1484: *res = *res < operand2;
1485: break;
1.80 schwarze 1486: case '>':
1.77 schwarze 1487: *res = *res > operand2;
1488: break;
1.80 schwarze 1489: case 'l':
1.77 schwarze 1490: *res = *res <= operand2;
1491: break;
1.80 schwarze 1492: case 'g':
1.77 schwarze 1493: *res = *res >= operand2;
1494: break;
1.80 schwarze 1495: case '=':
1.77 schwarze 1496: *res = *res == operand2;
1497: break;
1.80 schwarze 1498: case '!':
1.77 schwarze 1499: *res = *res != operand2;
1500: break;
1.80 schwarze 1501: case '&':
1.77 schwarze 1502: *res = *res && operand2;
1503: break;
1.80 schwarze 1504: case ':':
1.77 schwarze 1505: *res = *res || operand2;
1506: break;
1.80 schwarze 1507: case 'i':
1.77 schwarze 1508: if (operand2 < *res)
1509: *res = operand2;
1510: break;
1.80 schwarze 1511: case 'a':
1.77 schwarze 1512: if (operand2 > *res)
1513: *res = operand2;
1514: break;
1515: default:
1516: abort();
1517: }
1518: }
1519: return(1);
1520: }
1521:
1.52 schwarze 1522: void
1.60 schwarze 1523: roff_setreg(struct roff *r, const char *name, int val, char sign)
1.41 schwarze 1524: {
1.52 schwarze 1525: struct roffreg *reg;
1526:
1527: /* Search for an existing register with the same name. */
1528: reg = r->regtab;
1529:
1530: while (reg && strcmp(name, reg->key.p))
1531: reg = reg->next;
1.41 schwarze 1532:
1.52 schwarze 1533: if (NULL == reg) {
1534: /* Create a new register. */
1535: reg = mandoc_malloc(sizeof(struct roffreg));
1536: reg->key.p = mandoc_strdup(name);
1537: reg->key.sz = strlen(name);
1.60 schwarze 1538: reg->val = 0;
1.52 schwarze 1539: reg->next = r->regtab;
1540: r->regtab = reg;
1541: }
1542:
1.60 schwarze 1543: if ('+' == sign)
1544: reg->val += val;
1545: else if ('-' == sign)
1546: reg->val -= val;
1547: else
1548: reg->val = val;
1.41 schwarze 1549: }
1550:
1.65 schwarze 1551: /*
1552: * Handle some predefined read-only number registers.
1553: * For now, return -1 if the requested register is not predefined;
1554: * in case a predefined read-only register having the value -1
1555: * were to turn up, another special value would have to be chosen.
1556: */
1557: static int
1558: roff_getregro(const char *name)
1559: {
1560:
1561: switch (*name) {
1.80 schwarze 1562: case 'A': /* ASCII approximation mode is always off. */
1.65 schwarze 1563: return(0);
1.80 schwarze 1564: case 'g': /* Groff compatibility mode is always on. */
1.65 schwarze 1565: return(1);
1.80 schwarze 1566: case 'H': /* Fixed horizontal resolution. */
1.65 schwarze 1567: return (24);
1.80 schwarze 1568: case 'j': /* Always adjust left margin only. */
1.65 schwarze 1569: return(0);
1.80 schwarze 1570: case 'T': /* Some output device is always defined. */
1.65 schwarze 1571: return(1);
1.80 schwarze 1572: case 'V': /* Fixed vertical resolution. */
1.65 schwarze 1573: return (40);
1574: default:
1575: return (-1);
1576: }
1577: }
1578:
1.53 schwarze 1579: int
1.52 schwarze 1580: roff_getreg(const struct roff *r, const char *name)
1.41 schwarze 1581: {
1.52 schwarze 1582: struct roffreg *reg;
1.65 schwarze 1583: int val;
1584:
1585: if ('.' == name[0] && '\0' != name[1] && '\0' == name[2]) {
1586: val = roff_getregro(name + 1);
1587: if (-1 != val)
1588: return (val);
1589: }
1.52 schwarze 1590:
1591: for (reg = r->regtab; reg; reg = reg->next)
1592: if (0 == strcmp(name, reg->key.p))
1.53 schwarze 1593: return(reg->val);
1594:
1595: return(0);
1596: }
1597:
1598: static int
1599: roff_getregn(const struct roff *r, const char *name, size_t len)
1600: {
1601: struct roffreg *reg;
1.65 schwarze 1602: int val;
1603:
1604: if ('.' == name[0] && 2 == len) {
1605: val = roff_getregro(name + 1);
1606: if (-1 != val)
1607: return (val);
1608: }
1.53 schwarze 1609:
1610: for (reg = r->regtab; reg; reg = reg->next)
1611: if (len == reg->key.sz &&
1612: 0 == strncmp(name, reg->key.p, len))
1613: return(reg->val);
1.41 schwarze 1614:
1.52 schwarze 1615: return(0);
1.41 schwarze 1616: }
1617:
1.52 schwarze 1618: static void
1619: roff_freereg(struct roffreg *reg)
1.41 schwarze 1620: {
1.52 schwarze 1621: struct roffreg *old_reg;
1.41 schwarze 1622:
1.52 schwarze 1623: while (NULL != reg) {
1624: free(reg->key.p);
1625: old_reg = reg;
1626: reg = reg->next;
1627: free(old_reg);
1628: }
1.41 schwarze 1629: }
1.7 schwarze 1630:
1631: static enum rofferr
1.6 schwarze 1632: roff_nr(ROFF_ARGS)
1.1 schwarze 1633: {
1.85 schwarze 1634: char *key, *val;
1635: size_t keysz;
1.37 schwarze 1636: int iv;
1.60 schwarze 1637: char sign;
1.6 schwarze 1638:
1.85 schwarze 1639: key = val = *bufp + pos;
1640: if ('\0' == *key)
1641: return(ROFF_IGN);
1642:
1643: keysz = roff_getname(r, &val, ln, pos);
1644: if ('\\' == key[keysz])
1645: return(ROFF_IGN);
1646: key[keysz] = '\0';
1.6 schwarze 1647:
1.60 schwarze 1648: sign = *val;
1649: if ('+' == sign || '-' == sign)
1650: val++;
1651:
1.77 schwarze 1652: if (roff_evalnum(val, NULL, &iv, 0))
1653: roff_setreg(r, key, iv, sign);
1.1 schwarze 1654:
1.76 schwarze 1655: return(ROFF_IGN);
1656: }
1657:
1658: static enum rofferr
1659: roff_rr(ROFF_ARGS)
1660: {
1661: struct roffreg *reg, **prev;
1.85 schwarze 1662: char *name, *cp;
1663: size_t namesz;
1.76 schwarze 1664:
1.85 schwarze 1665: name = cp = *bufp + pos;
1666: if ('\0' == *name)
1667: return(ROFF_IGN);
1668: namesz = roff_getname(r, &cp, ln, pos);
1669: name[namesz] = '\0';
1.76 schwarze 1670:
1671: prev = &r->regtab;
1672: while (1) {
1673: reg = *prev;
1674: if (NULL == reg || !strcmp(name, reg->key.p))
1675: break;
1676: prev = ®->next;
1677: }
1678: if (NULL != reg) {
1679: *prev = reg->next;
1680: free(reg->key.p);
1681: free(reg);
1682: }
1.29 schwarze 1683: return(ROFF_IGN);
1684: }
1685:
1686: static enum rofferr
1687: roff_rm(ROFF_ARGS)
1688: {
1689: const char *name;
1690: char *cp;
1.85 schwarze 1691: size_t namesz;
1.29 schwarze 1692:
1693: cp = *bufp + pos;
1694: while ('\0' != *cp) {
1.85 schwarze 1695: name = cp;
1696: namesz = roff_getname(r, &cp, ln, (int)(cp - *bufp));
1697: roff_setstrn(&r->strtab, name, namesz, NULL, 0, 0);
1698: if ('\\' == name[namesz])
1699: break;
1.29 schwarze 1700: }
1.51 schwarze 1701: return(ROFF_IGN);
1702: }
1703:
1704: static enum rofferr
1705: roff_it(ROFF_ARGS)
1706: {
1707: char *cp;
1708: size_t len;
1709: int iv;
1710:
1711: /* Parse the number of lines. */
1712: cp = *bufp + pos;
1713: len = strcspn(cp, " \t");
1714: cp[len] = '\0';
1715: if ((iv = mandoc_strntoi(cp, len, 10)) <= 0) {
1716: mandoc_msg(MANDOCERR_NUMERIC, r->parse,
1.80 schwarze 1717: ln, ppos, *bufp + 1);
1.51 schwarze 1718: return(ROFF_IGN);
1719: }
1720: cp += len + 1;
1721:
1722: /* Arm the input line trap. */
1723: roffit_lines = iv;
1724: roffit_macro = mandoc_strdup(cp);
1.2 schwarze 1725: return(ROFF_IGN);
1.47 schwarze 1726: }
1727:
1728: static enum rofferr
1729: roff_Dd(ROFF_ARGS)
1730: {
1731: const char *const *cp;
1732:
1.72 schwarze 1733: if (0 == ((MPARSE_MDOC | MPARSE_QUICK) & r->options))
1.47 schwarze 1734: for (cp = __mdoc_reserved; *cp; cp++)
1735: roff_setstr(r, *cp, NULL, 0);
1736:
1737: return(ROFF_CONT);
1738: }
1739:
1740: static enum rofferr
1741: roff_TH(ROFF_ARGS)
1742: {
1743: const char *const *cp;
1744:
1.72 schwarze 1745: if (0 == (MPARSE_QUICK & r->options))
1.47 schwarze 1746: for (cp = __man_reserved; *cp; cp++)
1747: roff_setstr(r, *cp, NULL, 0);
1748:
1749: return(ROFF_CONT);
1.14 schwarze 1750: }
1751:
1752: static enum rofferr
1.27 schwarze 1753: roff_TE(ROFF_ARGS)
1754: {
1755:
1756: if (NULL == r->tbl)
1.35 schwarze 1757: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.27 schwarze 1758: else
1.41 schwarze 1759: tbl_end(&r->tbl);
1.27 schwarze 1760:
1761: return(ROFF_IGN);
1762: }
1763:
1764: static enum rofferr
1765: roff_T_(ROFF_ARGS)
1766: {
1767:
1768: if (NULL == r->tbl)
1.35 schwarze 1769: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.27 schwarze 1770: else
1771: tbl_restart(ppos, ln, r->tbl);
1772:
1773: return(ROFF_IGN);
1774: }
1775:
1.41 schwarze 1776: #if 0
1777: static int
1778: roff_closeeqn(struct roff *r)
1779: {
1780:
1781: return(r->eqn && ROFF_EQN == eqn_end(&r->eqn) ? 1 : 0);
1782: }
1783: #endif
1784:
1785: static void
1.80 schwarze 1786: roff_openeqn(struct roff *r, const char *name, int line,
1.41 schwarze 1787: int offs, const char *buf)
1.32 schwarze 1788: {
1.41 schwarze 1789: struct eqn_node *e;
1790: int poff;
1.32 schwarze 1791:
1792: assert(NULL == r->eqn);
1.41 schwarze 1793: e = eqn_alloc(name, offs, line, r->parse);
1.32 schwarze 1794:
1795: if (r->last_eqn)
1796: r->last_eqn->next = e;
1797: else
1798: r->first_eqn = r->last_eqn = e;
1799:
1800: r->eqn = r->last_eqn = e;
1.41 schwarze 1801:
1802: if (buf) {
1803: poff = 0;
1804: eqn_read(&r->eqn, line, buf, offs, &poff);
1805: }
1806: }
1807:
1808: static enum rofferr
1809: roff_EQ(ROFF_ARGS)
1810: {
1811:
1812: roff_openeqn(r, *bufp + pos, ln, ppos, NULL);
1.32 schwarze 1813: return(ROFF_IGN);
1814: }
1815:
1816: static enum rofferr
1817: roff_EN(ROFF_ARGS)
1818: {
1819:
1.35 schwarze 1820: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.32 schwarze 1821: return(ROFF_IGN);
1822: }
1823:
1824: static enum rofferr
1.27 schwarze 1825: roff_TS(ROFF_ARGS)
1826: {
1.49 schwarze 1827: struct tbl_node *tbl;
1.27 schwarze 1828:
1829: if (r->tbl) {
1.35 schwarze 1830: mandoc_msg(MANDOCERR_SCOPEBROKEN, r->parse, ln, ppos, NULL);
1.41 schwarze 1831: tbl_end(&r->tbl);
1.27 schwarze 1832: }
1833:
1.49 schwarze 1834: tbl = tbl_alloc(ppos, ln, r->parse);
1.27 schwarze 1835:
1836: if (r->last_tbl)
1.49 schwarze 1837: r->last_tbl->next = tbl;
1.27 schwarze 1838: else
1.49 schwarze 1839: r->first_tbl = r->last_tbl = tbl;
1.27 schwarze 1840:
1.49 schwarze 1841: r->tbl = r->last_tbl = tbl;
1.27 schwarze 1842: return(ROFF_IGN);
1843: }
1844:
1845: static enum rofferr
1.48 schwarze 1846: roff_cc(ROFF_ARGS)
1847: {
1848: const char *p;
1849:
1850: p = *bufp + pos;
1851:
1852: if ('\0' == *p || '.' == (r->control = *p++))
1853: r->control = 0;
1854:
1855: if ('\0' != *p)
1856: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
1857:
1858: return(ROFF_IGN);
1859: }
1860:
1861: static enum rofferr
1.42 schwarze 1862: roff_tr(ROFF_ARGS)
1863: {
1864: const char *p, *first, *second;
1865: size_t fsz, ssz;
1866: enum mandoc_esc esc;
1867:
1868: p = *bufp + pos;
1869:
1870: if ('\0' == *p) {
1871: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
1872: return(ROFF_IGN);
1873: }
1874:
1875: while ('\0' != *p) {
1876: fsz = ssz = 1;
1877:
1878: first = p++;
1879: if ('\\' == *first) {
1880: esc = mandoc_escape(&p, NULL, NULL);
1881: if (ESCAPE_ERROR == esc) {
1.80 schwarze 1882: mandoc_msg(MANDOCERR_BADESCAPE,
1883: r->parse, ln,
1884: (int)(p - *bufp), NULL);
1.42 schwarze 1885: return(ROFF_IGN);
1886: }
1887: fsz = (size_t)(p - first);
1888: }
1889:
1890: second = p++;
1891: if ('\\' == *second) {
1892: esc = mandoc_escape(&p, NULL, NULL);
1893: if (ESCAPE_ERROR == esc) {
1.80 schwarze 1894: mandoc_msg(MANDOCERR_BADESCAPE,
1895: r->parse, ln,
1896: (int)(p - *bufp), NULL);
1.42 schwarze 1897: return(ROFF_IGN);
1898: }
1899: ssz = (size_t)(p - second);
1900: } else if ('\0' == *second) {
1.80 schwarze 1901: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse,
1902: ln, (int)(p - *bufp), NULL);
1.42 schwarze 1903: second = " ";
1904: p--;
1905: }
1906:
1907: if (fsz > 1) {
1.80 schwarze 1908: roff_setstrn(&r->xmbtab, first, fsz,
1909: second, ssz, 0);
1.42 schwarze 1910: continue;
1911: }
1912:
1913: if (NULL == r->xtab)
1.80 schwarze 1914: r->xtab = mandoc_calloc(128,
1915: sizeof(struct roffstr));
1.42 schwarze 1916:
1917: free(r->xtab[(int)*first].p);
1918: r->xtab[(int)*first].p = mandoc_strndup(second, ssz);
1919: r->xtab[(int)*first].sz = ssz;
1920: }
1921:
1922: return(ROFF_IGN);
1923: }
1924:
1925: static enum rofferr
1.14 schwarze 1926: roff_so(ROFF_ARGS)
1927: {
1928: char *name;
1.15 schwarze 1929:
1.83 schwarze 1930: name = *bufp + pos;
1931: mandoc_vmsg(MANDOCERR_SO, r->parse, ln, ppos, ".so %s", name);
1.14 schwarze 1932:
1.22 schwarze 1933: /*
1934: * Handle `so'. Be EXTREMELY careful, as we shouldn't be
1935: * opening anything that's not in our cwd or anything beneath
1936: * it. Thus, explicitly disallow traversing up the file-system
1937: * or using absolute paths.
1938: */
1939:
1.14 schwarze 1940: if ('/' == *name || strstr(name, "../") || strstr(name, "/..")) {
1.83 schwarze 1941: mandoc_vmsg(MANDOCERR_SO_PATH, r->parse, ln, ppos,
1942: ".so %s", name);
1.14 schwarze 1943: return(ROFF_ERR);
1944: }
1945:
1946: *offs = pos;
1947: return(ROFF_SO);
1.7 schwarze 1948: }
1949:
1.16 schwarze 1950: static enum rofferr
1951: roff_userdef(ROFF_ARGS)
1.12 schwarze 1952: {
1.16 schwarze 1953: const char *arg[9];
1954: char *cp, *n1, *n2;
1.25 schwarze 1955: int i;
1.12 schwarze 1956:
1.16 schwarze 1957: /*
1958: * Collect pointers to macro argument strings
1.61 schwarze 1959: * and NUL-terminate them.
1.16 schwarze 1960: */
1961: cp = *bufp + pos;
1.25 schwarze 1962: for (i = 0; i < 9; i++)
1.26 schwarze 1963: arg[i] = '\0' == *cp ? "" :
1.35 schwarze 1964: mandoc_getarg(r->parse, &cp, ln, &pos);
1.16 schwarze 1965:
1966: /*
1967: * Expand macro arguments.
1.12 schwarze 1968: */
1.16 schwarze 1969: *szp = 0;
1970: n1 = cp = mandoc_strdup(r->current_string);
1971: while (NULL != (cp = strstr(cp, "\\$"))) {
1972: i = cp[2] - '1';
1973: if (0 > i || 8 < i) {
1974: /* Not an argument invocation. */
1975: cp += 2;
1976: continue;
1977: }
1.82 schwarze 1978: *cp = '\0';
1979: *szp = mandoc_asprintf(&n2, "%s%s%s",
1980: n1, arg[i], cp + 3) + 1;
1.16 schwarze 1981: cp = n2 + (cp - n1);
1982: free(n1);
1983: n1 = n2;
1.12 schwarze 1984: }
1985:
1.16 schwarze 1986: /*
1987: * Replace the macro invocation
1988: * by the expanded macro.
1989: */
1990: free(*bufp);
1991: *bufp = n1;
1992: if (0 == *szp)
1993: *szp = strlen(*bufp) + 1;
1994:
1.19 schwarze 1995: return(*szp > 1 && '\n' == (*bufp)[(int)*szp - 2] ?
1.16 schwarze 1996: ROFF_REPARSE : ROFF_APPEND);
1.12 schwarze 1997: }
1.28 schwarze 1998:
1.85 schwarze 1999: static size_t
1.28 schwarze 2000: roff_getname(struct roff *r, char **cpp, int ln, int pos)
2001: {
2002: char *name, *cp;
1.85 schwarze 2003: size_t namesz;
1.28 schwarze 2004:
2005: name = *cpp;
2006: if ('\0' == *name)
1.85 schwarze 2007: return(0);
1.28 schwarze 2008:
1.85 schwarze 2009: /* Read until end of name and terminate it with NUL. */
2010: for (cp = name; 1; cp++) {
2011: if ('\0' == *cp || ' ' == *cp) {
2012: namesz = cp - name;
2013: break;
2014: }
1.28 schwarze 2015: if ('\\' != *cp)
2016: continue;
1.88 ! schwarze 2017: namesz = cp - name;
! 2018: if ('{' == cp[1] || '}' == cp[1])
! 2019: break;
1.28 schwarze 2020: cp++;
2021: if ('\\' == *cp)
2022: continue;
1.35 schwarze 2023: mandoc_msg(MANDOCERR_NAMESC, r->parse, ln, pos, NULL);
1.85 schwarze 2024: mandoc_escape((const char **)&cp, NULL, NULL);
2025: break;
1.28 schwarze 2026: }
2027:
2028: /* Read past spaces. */
2029: while (' ' == *cp)
2030: cp++;
2031:
2032: *cpp = cp;
1.85 schwarze 2033: return(namesz);
1.28 schwarze 2034: }
2035:
1.16 schwarze 2036: /*
2037: * Store *string into the user-defined string called *name.
2038: * To clear an existing entry, call with (*r, *name, NULL, 0).
1.66 schwarze 2039: * append == 0: replace mode
2040: * append == 1: single-line append mode
2041: * append == 2: multiline append mode, append '\n' after each call
1.16 schwarze 2042: */
1.8 schwarze 2043: static void
1.16 schwarze 2044: roff_setstr(struct roff *r, const char *name, const char *string,
1.66 schwarze 2045: int append)
1.7 schwarze 2046: {
1.42 schwarze 2047:
2048: roff_setstrn(&r->strtab, name, strlen(name), string,
1.80 schwarze 2049: string ? strlen(string) : 0, append);
1.42 schwarze 2050: }
2051:
2052: static void
2053: roff_setstrn(struct roffkv **r, const char *name, size_t namesz,
1.66 schwarze 2054: const char *string, size_t stringsz, int append)
1.42 schwarze 2055: {
2056: struct roffkv *n;
2057: char *c;
2058: int i;
2059: size_t oldch, newch;
1.7 schwarze 2060:
1.16 schwarze 2061: /* Search for an existing string with the same name. */
1.42 schwarze 2062: n = *r;
2063:
1.84 schwarze 2064: while (n && (namesz != n->key.sz ||
2065: strncmp(n->key.p, name, namesz)))
1.7 schwarze 2066: n = n->next;
1.8 schwarze 2067:
2068: if (NULL == n) {
1.16 schwarze 2069: /* Create a new string table entry. */
1.42 schwarze 2070: n = mandoc_malloc(sizeof(struct roffkv));
2071: n->key.p = mandoc_strndup(name, namesz);
2072: n->key.sz = namesz;
2073: n->val.p = NULL;
2074: n->val.sz = 0;
2075: n->next = *r;
2076: *r = n;
1.66 schwarze 2077: } else if (0 == append) {
1.42 schwarze 2078: free(n->val.p);
2079: n->val.p = NULL;
2080: n->val.sz = 0;
1.16 schwarze 2081: }
2082:
2083: if (NULL == string)
2084: return;
2085:
2086: /*
2087: * One additional byte for the '\n' in multiline mode,
2088: * and one for the terminating '\0'.
2089: */
1.66 schwarze 2090: newch = stringsz + (1 < append ? 2u : 1u);
1.42 schwarze 2091:
2092: if (NULL == n->val.p) {
2093: n->val.p = mandoc_malloc(newch);
2094: *n->val.p = '\0';
1.16 schwarze 2095: oldch = 0;
2096: } else {
1.42 schwarze 2097: oldch = n->val.sz;
2098: n->val.p = mandoc_realloc(n->val.p, oldch + newch);
1.16 schwarze 2099: }
2100:
2101: /* Skip existing content in the destination buffer. */
1.42 schwarze 2102: c = n->val.p + (int)oldch;
1.16 schwarze 2103:
2104: /* Append new content to the destination buffer. */
1.42 schwarze 2105: i = 0;
2106: while (i < (int)stringsz) {
1.16 schwarze 2107: /*
2108: * Rudimentary roff copy mode:
2109: * Handle escaped backslashes.
2110: */
1.42 schwarze 2111: if ('\\' == string[i] && '\\' == string[i + 1])
2112: i++;
2113: *c++ = string[i++];
1.16 schwarze 2114: }
1.8 schwarze 2115:
1.16 schwarze 2116: /* Append terminating bytes. */
1.66 schwarze 2117: if (1 < append)
1.16 schwarze 2118: *c++ = '\n';
1.42 schwarze 2119:
1.16 schwarze 2120: *c = '\0';
1.42 schwarze 2121: n->val.sz = (int)(c - n->val.p);
1.7 schwarze 2122: }
2123:
1.8 schwarze 2124: static const char *
2125: roff_getstrn(const struct roff *r, const char *name, size_t len)
1.7 schwarze 2126: {
1.42 schwarze 2127: const struct roffkv *n;
1.64 schwarze 2128: int i;
1.7 schwarze 2129:
1.42 schwarze 2130: for (n = r->strtab; n; n = n->next)
1.80 schwarze 2131: if (0 == strncmp(name, n->key.p, len) &&
2132: '\0' == n->key.p[(int)len])
1.42 schwarze 2133: return(n->val.p);
1.64 schwarze 2134:
2135: for (i = 0; i < PREDEFS_MAX; i++)
2136: if (0 == strncmp(name, predefs[i].name, len) &&
2137: '\0' == predefs[i].name[(int)len])
2138: return(predefs[i].str);
1.8 schwarze 2139:
1.42 schwarze 2140: return(NULL);
1.7 schwarze 2141: }
2142:
1.8 schwarze 2143: static void
1.42 schwarze 2144: roff_freestr(struct roffkv *r)
1.7 schwarze 2145: {
1.42 schwarze 2146: struct roffkv *n, *nn;
1.7 schwarze 2147:
1.42 schwarze 2148: for (n = r; n; n = nn) {
2149: free(n->key.p);
2150: free(n->val.p);
1.7 schwarze 2151: nn = n->next;
2152: free(n);
2153: }
1.27 schwarze 2154: }
2155:
2156: const struct tbl_span *
2157: roff_span(const struct roff *r)
2158: {
1.80 schwarze 2159:
1.27 schwarze 2160: return(r->tbl ? tbl_span(r->tbl) : NULL);
1.32 schwarze 2161: }
2162:
2163: const struct eqn *
2164: roff_eqn(const struct roff *r)
2165: {
1.80 schwarze 2166:
1.32 schwarze 2167: return(r->last_eqn ? &r->last_eqn->eqn : NULL);
1.42 schwarze 2168: }
2169:
2170: /*
2171: * Duplicate an input string, making the appropriate character
2172: * conversations (as stipulated by `tr') along the way.
2173: * Returns a heap-allocated string with all the replacements made.
2174: */
2175: char *
2176: roff_strdup(const struct roff *r, const char *p)
2177: {
2178: const struct roffkv *cp;
2179: char *res;
2180: const char *pp;
2181: size_t ssz, sz;
2182: enum mandoc_esc esc;
2183:
2184: if (NULL == r->xmbtab && NULL == r->xtab)
2185: return(mandoc_strdup(p));
2186: else if ('\0' == *p)
2187: return(mandoc_strdup(""));
2188:
2189: /*
2190: * Step through each character looking for term matches
2191: * (remember that a `tr' can be invoked with an escape, which is
2192: * a glyph but the escape is multi-character).
2193: * We only do this if the character hash has been initialised
2194: * and the string is >0 length.
2195: */
2196:
2197: res = NULL;
2198: ssz = 0;
2199:
2200: while ('\0' != *p) {
2201: if ('\\' != *p && r->xtab && r->xtab[(int)*p].p) {
2202: sz = r->xtab[(int)*p].sz;
2203: res = mandoc_realloc(res, ssz + sz + 1);
2204: memcpy(res + ssz, r->xtab[(int)*p].p, sz);
2205: ssz += sz;
2206: p++;
2207: continue;
2208: } else if ('\\' != *p) {
2209: res = mandoc_realloc(res, ssz + 2);
2210: res[ssz++] = *p++;
2211: continue;
2212: }
2213:
2214: /* Search for term matches. */
2215: for (cp = r->xmbtab; cp; cp = cp->next)
2216: if (0 == strncmp(p, cp->key.p, cp->key.sz))
2217: break;
2218:
2219: if (NULL != cp) {
2220: /*
2221: * A match has been found.
2222: * Append the match to the array and move
2223: * forward by its keysize.
2224: */
1.80 schwarze 2225: res = mandoc_realloc(res,
2226: ssz + cp->val.sz + 1);
1.42 schwarze 2227: memcpy(res + ssz, cp->val.p, cp->val.sz);
2228: ssz += cp->val.sz;
2229: p += (int)cp->key.sz;
2230: continue;
2231: }
2232:
2233: /*
2234: * Handle escapes carefully: we need to copy
2235: * over just the escape itself, or else we might
2236: * do replacements within the escape itself.
2237: * Make sure to pass along the bogus string.
2238: */
2239: pp = p++;
2240: esc = mandoc_escape(&p, NULL, NULL);
2241: if (ESCAPE_ERROR == esc) {
2242: sz = strlen(pp);
2243: res = mandoc_realloc(res, ssz + sz + 1);
2244: memcpy(res + ssz, pp, sz);
2245: break;
2246: }
1.80 schwarze 2247: /*
2248: * We bail out on bad escapes.
1.42 schwarze 2249: * No need to warn: we already did so when
2250: * roff_res() was called.
2251: */
2252: sz = (int)(p - pp);
2253: res = mandoc_realloc(res, ssz + sz + 1);
2254: memcpy(res + ssz, pp, sz);
2255: ssz += sz;
2256: }
2257:
2258: res[(int)ssz] = '\0';
2259: return(res);
1.48 schwarze 2260: }
2261:
2262: /*
1.80 schwarze 2263: * Find out whether a line is a macro line or not.
1.48 schwarze 2264: * If it is, adjust the current position and return one; if it isn't,
2265: * return zero and don't change the current position.
2266: * If the control character has been set with `.cc', then let that grain
2267: * precedence.
2268: * This is slighly contrary to groff, where using the non-breaking
2269: * control character when `cc' has been invoked will cause the
2270: * non-breaking macro contents to be printed verbatim.
2271: */
2272: int
2273: roff_getcontrol(const struct roff *r, const char *cp, int *ppos)
2274: {
2275: int pos;
2276:
2277: pos = *ppos;
2278:
2279: if (0 != r->control && cp[pos] == r->control)
2280: pos++;
2281: else if (0 != r->control)
2282: return(0);
2283: else if ('\\' == cp[pos] && '.' == cp[pos + 1])
2284: pos += 2;
2285: else if ('.' == cp[pos] || '\'' == cp[pos])
2286: pos++;
2287: else
2288: return(0);
2289:
2290: while (' ' == cp[pos] || '\t' == cp[pos])
2291: pos++;
2292:
2293: *ppos = pos;
2294: return(1);
1.1 schwarze 2295: }
![[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