Annotation of src/usr.bin/mandoc/roff.c, Revision 1.90
1.90 ! schwarze 1: /* $Id: roff.c,v 1.89 2014/07/03 21:23:08 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])
1.90 ! schwarze 855: mandoc_vmsg(MANDOCERR_ARG_SKIP, r->parse, ln, pos,
! 856: ".. %s", *bufp + pos);
1.2 schwarze 857:
858: roffnode_pop(r);
859: roffnode_cleanscope(r);
860: return(ROFF_IGN);
861:
862: }
1.1 schwarze 863:
1.2 schwarze 864: static void
865: roffnode_cleanscope(struct roff *r)
866: {
1.1 schwarze 867:
1.2 schwarze 868: while (r->last) {
1.46 schwarze 869: if (--r->last->endspan != 0)
1.2 schwarze 870: break;
871: roffnode_pop(r);
872: }
873: }
1.1 schwarze 874:
1.68 schwarze 875: static void
876: roff_ccond(struct roff *r, int ln, int ppos)
1.2 schwarze 877: {
1.1 schwarze 878:
1.2 schwarze 879: if (NULL == r->last) {
1.35 schwarze 880: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.68 schwarze 881: return;
1.2 schwarze 882: }
1.1 schwarze 883:
1.2 schwarze 884: switch (r->last->tok) {
1.80 schwarze 885: case ROFF_el:
1.2 schwarze 886: /* FALLTHROUGH */
1.80 schwarze 887: case ROFF_ie:
1.2 schwarze 888: /* FALLTHROUGH */
1.80 schwarze 889: case ROFF_if:
1.2 schwarze 890: break;
891: default:
1.35 schwarze 892: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.68 schwarze 893: return;
1.2 schwarze 894: }
1.1 schwarze 895:
1.2 schwarze 896: if (r->last->endspan > -1) {
1.35 schwarze 897: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.68 schwarze 898: return;
1.2 schwarze 899: }
900:
901: roffnode_pop(r);
902: roffnode_cleanscope(r);
1.68 schwarze 903: return;
1.1 schwarze 904: }
905:
906: static enum rofferr
1.2 schwarze 907: roff_block(ROFF_ARGS)
1.1 schwarze 908: {
1.86 schwarze 909: char *name, *cp;
910: size_t namesz;
1.16 schwarze 911:
1.86 schwarze 912: name = cp = *bufp + pos;
913: namesz = 0;
1.2 schwarze 914:
1.16 schwarze 915: if (ROFF_ig != tok) {
1.86 schwarze 916: if ('\0' == *cp) {
1.89 schwarze 917: mandoc_msg(MANDOCERR_REQ_EMPTY, r->parse,
918: ln, ppos, roffs[tok].name);
1.16 schwarze 919: return(ROFF_IGN);
920: }
1.22 schwarze 921:
922: /*
923: * Re-write `de1', since we don't really care about
924: * groff's strange compatibility mode, into `de'.
925: */
926:
1.18 schwarze 927: if (ROFF_de1 == tok)
928: tok = ROFF_de;
1.86 schwarze 929: else if (ROFF_de != tok)
1.35 schwarze 930: mandoc_msg(MANDOCERR_REQUEST, r->parse, ln, ppos,
1.21 schwarze 931: roffs[tok].name);
1.22 schwarze 932:
1.86 schwarze 933: namesz = roff_getname(r, &cp, ln, ppos);
934: name[namesz] = '\0';
935: } else
936: name = NULL;
1.2 schwarze 937:
1.16 schwarze 938: roffnode_push(r, tok, name, ln, ppos);
939:
940: /*
941: * At the beginning of a `de' macro, clear the existing string
942: * with the same name, if there is one. New content will be
1.66 schwarze 943: * appended from roff_block_text() in multiline mode.
1.16 schwarze 944: */
1.22 schwarze 945:
1.86 schwarze 946: if (namesz && ROFF_de == tok)
947: roff_setstrn(&r->strtab, name, namesz, "", 0, 0);
1.2 schwarze 948:
1.86 schwarze 949: if ('\0' == *cp)
1.2 schwarze 950: return(ROFF_IGN);
1.1 schwarze 951:
1.22 schwarze 952: /* If present, process the custom end-of-line marker. */
953:
1.86 schwarze 954: name = cp;
955: namesz = roff_getname(r, &cp, ln, ppos);
956: if (namesz)
957: r->last->end = mandoc_strndup(name, namesz);
1.2 schwarze 958:
1.86 schwarze 959: if ('\0' != *cp)
1.90 ! schwarze 960: mandoc_vmsg(MANDOCERR_ARG_EXCESS, r->parse,
! 961: ln, pos, ".%s ... %s", roffs[tok].name, cp);
1.1 schwarze 962:
963: return(ROFF_IGN);
964: }
965:
966: static enum rofferr
1.2 schwarze 967: roff_block_sub(ROFF_ARGS)
1.1 schwarze 968: {
1.2 schwarze 969: enum rofft t;
970: int i, j;
971:
972: /*
973: * First check whether a custom macro exists at this level. If
974: * it does, then check against it. This is some of groff's
975: * stranger behaviours. If we encountered a custom end-scope
976: * tag and that tag also happens to be a "real" macro, then we
977: * need to try interpreting it again as a real macro. If it's
978: * not, then return ignore. Else continue.
979: */
980:
981: if (r->last->end) {
1.35 schwarze 982: for (i = pos, j = 0; r->last->end[j]; j++, i++)
1.2 schwarze 983: if ((*bufp)[i] != r->last->end[j])
984: break;
1.1 schwarze 985:
1.80 schwarze 986: if ('\0' == r->last->end[j] &&
987: ('\0' == (*bufp)[i] ||
988: ' ' == (*bufp)[i] ||
989: '\t' == (*bufp)[i])) {
1.2 schwarze 990: roffnode_pop(r);
991: roffnode_cleanscope(r);
1.1 schwarze 992:
1.35 schwarze 993: while (' ' == (*bufp)[i] || '\t' == (*bufp)[i])
994: i++;
995:
996: pos = i;
1.87 schwarze 997: if (ROFF_MAX != roff_parse(r, *bufp, &pos, ln, ppos))
1.2 schwarze 998: return(ROFF_RERUN);
999: return(ROFF_IGN);
1000: }
1.1 schwarze 1001: }
1002:
1.2 schwarze 1003: /*
1004: * If we have no custom end-query or lookup failed, then try
1005: * pulling it out of the hashtable.
1006: */
1.1 schwarze 1007:
1.87 schwarze 1008: t = roff_parse(r, *bufp, &pos, ln, ppos);
1.1 schwarze 1009:
1.16 schwarze 1010: /*
1011: * Macros other than block-end are only significant
1012: * in `de' blocks; elsewhere, simply throw them away.
1013: */
1014: if (ROFF_cblock != t) {
1015: if (ROFF_de == tok)
1.66 schwarze 1016: roff_setstr(r, r->last->name, *bufp + ppos, 2);
1.1 schwarze 1017: return(ROFF_IGN);
1.16 schwarze 1018: }
1.1 schwarze 1019:
1.2 schwarze 1020: assert(roffs[t].proc);
1.80 schwarze 1021: return((*roffs[t].proc)(r, t, bufp, szp, ln, ppos, pos, offs));
1.2 schwarze 1022: }
1023:
1024: static enum rofferr
1025: roff_block_text(ROFF_ARGS)
1026: {
1027:
1.16 schwarze 1028: if (ROFF_de == tok)
1.66 schwarze 1029: roff_setstr(r, r->last->name, *bufp + pos, 2);
1.16 schwarze 1030:
1.2 schwarze 1031: return(ROFF_IGN);
1032: }
1033:
1034: static enum rofferr
1035: roff_cond_sub(ROFF_ARGS)
1036: {
1037: enum rofft t;
1.37 schwarze 1038: char *ep;
1.71 schwarze 1039: int rr;
1.2 schwarze 1040:
1041: rr = r->last->rule;
1.37 schwarze 1042: roffnode_cleanscope(r);
1.87 schwarze 1043: t = roff_parse(r, *bufp, &pos, ln, ppos);
1.2 schwarze 1044:
1.37 schwarze 1045: /*
1.50 schwarze 1046: * Fully handle known macros when they are structurally
1047: * required or when the conditional evaluated to true.
1.5 schwarze 1048: */
1049:
1.50 schwarze 1050: if ((ROFF_MAX != t) &&
1.71 schwarze 1051: (rr || ROFFMAC_STRUCT & roffs[t].flags)) {
1.50 schwarze 1052: assert(roffs[t].proc);
1053: return((*roffs[t].proc)(r, t, bufp, szp,
1.80 schwarze 1054: ln, ppos, pos, offs));
1.50 schwarze 1055: }
1.39 schwarze 1056:
1.69 schwarze 1057: /*
1058: * If `\}' occurs on a macro line without a preceding macro,
1059: * drop the line completely.
1060: */
1061:
1062: ep = *bufp + pos;
1063: if ('\\' == ep[0] && '}' == ep[1])
1.71 schwarze 1064: rr = 0;
1.69 schwarze 1065:
1.50 schwarze 1066: /* Always check for the closing delimiter `\}'. */
1.39 schwarze 1067:
1.50 schwarze 1068: while (NULL != (ep = strchr(ep, '\\'))) {
1.70 schwarze 1069: if ('}' == *(++ep)) {
1070: *ep = '&';
1071: roff_ccond(r, ln, ep - *bufp - 1);
1072: }
1073: ++ep;
1.50 schwarze 1074: }
1.71 schwarze 1075: return(rr ? ROFF_CONT : ROFF_IGN);
1.2 schwarze 1076: }
1077:
1078: static enum rofferr
1079: roff_cond_text(ROFF_ARGS)
1080: {
1.37 schwarze 1081: char *ep;
1.71 schwarze 1082: int rr;
1.2 schwarze 1083:
1084: rr = r->last->rule;
1.37 schwarze 1085: roffnode_cleanscope(r);
1.1 schwarze 1086:
1.70 schwarze 1087: ep = *bufp + pos;
1088: while (NULL != (ep = strchr(ep, '\\'))) {
1089: if ('}' == *(++ep)) {
1090: *ep = '&';
1091: roff_ccond(r, ln, ep - *bufp - 1);
1092: }
1093: ++ep;
1.2 schwarze 1094: }
1.71 schwarze 1095: return(rr ? ROFF_CONT : ROFF_IGN);
1.2 schwarze 1096: }
1097:
1.77 schwarze 1098: /*
1099: * Parse a single signed integer number. Stop at the first non-digit.
1100: * If there is at least one digit, return success and advance the
1101: * parse point, else return failure and let the parse point unchanged.
1102: * Ignore overflows, treat them just like the C language.
1103: */
1.56 schwarze 1104: static int
1105: roff_getnum(const char *v, int *pos, int *res)
1106: {
1.79 schwarze 1107: int myres, n, p;
1108:
1109: if (NULL == res)
1110: res = &myres;
1.56 schwarze 1111:
1112: p = *pos;
1113: n = v[p] == '-';
1114: if (n)
1115: p++;
1116:
1117: for (*res = 0; isdigit((unsigned char)v[p]); p++)
1.77 schwarze 1118: *res = 10 * *res + v[p] - '0';
1.56 schwarze 1119: if (p == *pos + n)
1120: return 0;
1121:
1122: if (n)
1123: *res = -*res;
1124:
1125: *pos = p;
1126: return 1;
1127: }
1128:
1.71 schwarze 1129: /*
1130: * Evaluate a string comparison condition.
1131: * The first character is the delimiter.
1132: * Succeed if the string up to its second occurrence
1133: * matches the string up to its third occurence.
1134: * Advance the cursor after the third occurrence
1135: * or lacking that, to the end of the line.
1136: */
1137: static int
1138: roff_evalstrcond(const char *v, int *pos)
1139: {
1140: const char *s1, *s2, *s3;
1141: int match;
1142:
1143: match = 0;
1144: s1 = v + *pos; /* initial delimiter */
1145: s2 = s1 + 1; /* for scanning the first string */
1146: s3 = strchr(s2, *s1); /* for scanning the second string */
1147:
1148: if (NULL == s3) /* found no middle delimiter */
1149: goto out;
1150:
1151: while ('\0' != *++s3) {
1152: if (*s2 != *s3) { /* mismatch */
1153: s3 = strchr(s3, *s1);
1154: break;
1155: }
1156: if (*s3 == *s1) { /* found the final delimiter */
1157: match = 1;
1158: break;
1159: }
1160: s2++;
1161: }
1162:
1163: out:
1164: if (NULL == s3)
1165: s3 = strchr(s2, '\0');
1166: else
1167: s3++;
1168: *pos = s3 - v;
1169: return(match);
1170: }
1171:
1.77 schwarze 1172: /*
1173: * Evaluate an optionally negated single character, numerical,
1174: * or string condition.
1175: */
1.71 schwarze 1176: static int
1.5 schwarze 1177: roff_evalcond(const char *v, int *pos)
1178: {
1.77 schwarze 1179: int wanttrue, number;
1.5 schwarze 1180:
1.71 schwarze 1181: if ('!' == v[*pos]) {
1182: wanttrue = 0;
1183: (*pos)++;
1184: } else
1185: wanttrue = 1;
1186:
1.5 schwarze 1187: switch (v[*pos]) {
1.80 schwarze 1188: case 'n':
1.71 schwarze 1189: /* FALLTHROUGH */
1.80 schwarze 1190: case 'o':
1.5 schwarze 1191: (*pos)++;
1.71 schwarze 1192: return(wanttrue);
1.80 schwarze 1193: case 'c':
1.71 schwarze 1194: /* FALLTHROUGH */
1.80 schwarze 1195: case 'd':
1.71 schwarze 1196: /* FALLTHROUGH */
1.80 schwarze 1197: case 'e':
1.5 schwarze 1198: /* FALLTHROUGH */
1.80 schwarze 1199: case 'r':
1.5 schwarze 1200: /* FALLTHROUGH */
1.80 schwarze 1201: case 't':
1.5 schwarze 1202: (*pos)++;
1.71 schwarze 1203: return(!wanttrue);
1.5 schwarze 1204: default:
1205: break;
1206: }
1207:
1.77 schwarze 1208: if (roff_evalnum(v, pos, &number, 0))
1209: return((number > 0) == wanttrue);
1210: else
1.71 schwarze 1211: return(roff_evalstrcond(v, pos) == wanttrue);
1.5 schwarze 1212: }
1213:
1.2 schwarze 1214: static enum rofferr
1.21 schwarze 1215: roff_line_ignore(ROFF_ARGS)
1.6 schwarze 1216: {
1.30 schwarze 1217:
1.21 schwarze 1218: return(ROFF_IGN);
1219: }
1220:
1221: static enum rofferr
1.2 schwarze 1222: roff_cond(ROFF_ARGS)
1223: {
1.46 schwarze 1224:
1225: roffnode_push(r, tok, NULL, ln, ppos);
1.2 schwarze 1226:
1.80 schwarze 1227: /*
1.35 schwarze 1228: * An `.el' has no conditional body: it will consume the value
1229: * of the current rstack entry set in prior `ie' calls or
1.80 schwarze 1230: * defaults to DENY.
1.35 schwarze 1231: *
1232: * If we're not an `el', however, then evaluate the conditional.
1233: */
1.1 schwarze 1234:
1.46 schwarze 1235: r->last->rule = ROFF_el == tok ?
1.80 schwarze 1236: (r->rstackpos < 0 ? 0 : r->rstack[r->rstackpos--]) :
1237: roff_evalcond(*bufp, &pos);
1.2 schwarze 1238:
1.35 schwarze 1239: /*
1240: * An if-else will put the NEGATION of the current evaluated
1241: * conditional into the stack of rules.
1242: */
1243:
1.2 schwarze 1244: if (ROFF_ie == tok) {
1.35 schwarze 1245: if (r->rstackpos == RSTACK_MAX - 1) {
1.80 schwarze 1246: mandoc_msg(MANDOCERR_MEM,
1247: r->parse, ln, ppos, NULL);
1.35 schwarze 1248: return(ROFF_ERR);
1249: }
1.71 schwarze 1250: r->rstack[++r->rstackpos] = !r->last->rule;
1.2 schwarze 1251: }
1.5 schwarze 1252:
1253: /* If the parent has false as its rule, then so do we. */
1254:
1.71 schwarze 1255: if (r->last->parent && !r->last->parent->rule)
1256: r->last->rule = 0;
1.5 schwarze 1257:
1258: /*
1.46 schwarze 1259: * Determine scope.
1260: * If there is nothing on the line after the conditional,
1261: * not even whitespace, use next-line scope.
1.5 schwarze 1262: */
1.2 schwarze 1263:
1.46 schwarze 1264: if ('\0' == (*bufp)[pos]) {
1265: r->last->endspan = 2;
1266: goto out;
1267: }
1268:
1269: while (' ' == (*bufp)[pos])
1270: pos++;
1271:
1272: /* An opening brace requests multiline scope. */
1.2 schwarze 1273:
1274: if ('\\' == (*bufp)[pos] && '{' == (*bufp)[pos + 1]) {
1275: r->last->endspan = -1;
1276: pos += 2;
1.46 schwarze 1277: goto out;
1.80 schwarze 1278: }
1.2 schwarze 1279:
1280: /*
1.46 schwarze 1281: * Anything else following the conditional causes
1282: * single-line scope. Warn if the scope contains
1283: * nothing but trailing whitespace.
1.2 schwarze 1284: */
1285:
1286: if ('\0' == (*bufp)[pos])
1.89 schwarze 1287: mandoc_msg(MANDOCERR_COND_EMPTY, r->parse,
1288: ln, ppos, roffs[tok].name);
1.2 schwarze 1289:
1.46 schwarze 1290: r->last->endspan = 1;
1.1 schwarze 1291:
1.46 schwarze 1292: out:
1.2 schwarze 1293: *offs = pos;
1294: return(ROFF_RERUN);
1.1 schwarze 1295: }
1296:
1.2 schwarze 1297: static enum rofferr
1.7 schwarze 1298: roff_ds(ROFF_ARGS)
1299: {
1.85 schwarze 1300: char *string;
1301: const char *name;
1302: size_t namesz;
1.10 schwarze 1303:
1304: /*
1.85 schwarze 1305: * The first word is the name of the string.
1306: * If it is empty or terminated by an escape sequence,
1307: * abort the `ds' request without defining anything.
1.10 schwarze 1308: */
1.7 schwarze 1309:
1.85 schwarze 1310: name = string = *bufp + pos;
1.7 schwarze 1311: if ('\0' == *name)
1312: return(ROFF_IGN);
1313:
1.85 schwarze 1314: namesz = roff_getname(r, &string, ln, pos);
1315: if ('\\' == name[namesz])
1316: return(ROFF_IGN);
1317:
1318: /* Read past the initial double-quote, if any. */
1.28 schwarze 1319: if ('"' == *string)
1.7 schwarze 1320: string++;
1321:
1.10 schwarze 1322: /* The rest is the value. */
1.85 schwarze 1323: roff_setstrn(&r->strtab, name, namesz, string, strlen(string),
1324: ROFF_as == tok);
1.7 schwarze 1325: return(ROFF_IGN);
1326: }
1327:
1.77 schwarze 1328: /*
1329: * Parse a single operator, one or two characters long.
1330: * If the operator is recognized, return success and advance the
1331: * parse point, else return failure and let the parse point unchanged.
1332: */
1333: static int
1334: roff_getop(const char *v, int *pos, char *res)
1335: {
1336:
1337: *res = v[*pos];
1338:
1339: switch (*res) {
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: /* FALLTHROUGH */
1.80 schwarze 1350: case '&':
1.77 schwarze 1351: /* FALLTHROUGH */
1.80 schwarze 1352: case ':':
1.77 schwarze 1353: break;
1354: case '<':
1355: switch (v[*pos + 1]) {
1.80 schwarze 1356: case '=':
1.77 schwarze 1357: *res = 'l';
1358: (*pos)++;
1359: break;
1.80 schwarze 1360: case '>':
1.77 schwarze 1361: *res = '!';
1362: (*pos)++;
1363: break;
1.80 schwarze 1364: case '?':
1.77 schwarze 1365: *res = 'i';
1366: (*pos)++;
1367: break;
1368: default:
1369: break;
1370: }
1371: break;
1372: case '>':
1373: switch (v[*pos + 1]) {
1.80 schwarze 1374: case '=':
1.77 schwarze 1375: *res = 'g';
1376: (*pos)++;
1377: break;
1.80 schwarze 1378: case '?':
1.77 schwarze 1379: *res = 'a';
1380: (*pos)++;
1381: break;
1382: default:
1383: break;
1384: }
1385: break;
1386: case '=':
1387: if ('=' == v[*pos + 1])
1388: (*pos)++;
1389: break;
1390: default:
1391: return(0);
1392: }
1393: (*pos)++;
1394:
1395: return(*res);
1396: }
1397:
1398: /*
1399: * Evaluate either a parenthesized numeric expression
1400: * or a single signed integer number.
1401: */
1402: static int
1403: roff_evalpar(const char *v, int *pos, int *res)
1404: {
1405:
1406: if ('(' != v[*pos])
1407: return(roff_getnum(v, pos, res));
1408:
1409: (*pos)++;
1410: if ( ! roff_evalnum(v, pos, res, 1))
1411: return(0);
1412:
1.79 schwarze 1413: /*
1414: * Omission of the closing parenthesis
1415: * is an error in validation mode,
1416: * but ignored in evaluation mode.
1417: */
1418:
1.77 schwarze 1419: if (')' == v[*pos])
1420: (*pos)++;
1.79 schwarze 1421: else if (NULL == res)
1422: return(0);
1.77 schwarze 1423:
1424: return(1);
1425: }
1426:
1427: /*
1428: * Evaluate a complete numeric expression.
1429: * Proceed left to right, there is no concept of precedence.
1430: */
1431: static int
1432: roff_evalnum(const char *v, int *pos, int *res, int skipwhite)
1433: {
1434: int mypos, operand2;
1435: char operator;
1436:
1437: if (NULL == pos) {
1438: mypos = 0;
1439: pos = &mypos;
1440: }
1441:
1442: if (skipwhite)
1443: while (isspace((unsigned char)v[*pos]))
1444: (*pos)++;
1445:
1446: if ( ! roff_evalpar(v, pos, res))
1447: return(0);
1448:
1449: while (1) {
1450: if (skipwhite)
1451: while (isspace((unsigned char)v[*pos]))
1452: (*pos)++;
1453:
1454: if ( ! roff_getop(v, pos, &operator))
1455: break;
1456:
1457: if (skipwhite)
1458: while (isspace((unsigned char)v[*pos]))
1459: (*pos)++;
1460:
1461: if ( ! roff_evalpar(v, pos, &operand2))
1462: return(0);
1463:
1464: if (skipwhite)
1465: while (isspace((unsigned char)v[*pos]))
1466: (*pos)++;
1.79 schwarze 1467:
1468: if (NULL == res)
1469: continue;
1.77 schwarze 1470:
1471: switch (operator) {
1.80 schwarze 1472: case '+':
1.77 schwarze 1473: *res += operand2;
1474: break;
1.80 schwarze 1475: case '-':
1.77 schwarze 1476: *res -= operand2;
1477: break;
1.80 schwarze 1478: case '*':
1.77 schwarze 1479: *res *= operand2;
1480: break;
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 = *res < operand2;
1489: break;
1.80 schwarze 1490: case '>':
1.77 schwarze 1491: *res = *res > operand2;
1492: break;
1.80 schwarze 1493: case 'l':
1.77 schwarze 1494: *res = *res <= operand2;
1495: break;
1.80 schwarze 1496: case 'g':
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 '!':
1.77 schwarze 1503: *res = *res != operand2;
1504: break;
1.80 schwarze 1505: case '&':
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 'i':
1.77 schwarze 1512: if (operand2 < *res)
1513: *res = operand2;
1514: break;
1.80 schwarze 1515: case 'a':
1.77 schwarze 1516: if (operand2 > *res)
1517: *res = operand2;
1518: break;
1519: default:
1520: abort();
1521: }
1522: }
1523: return(1);
1524: }
1525:
1.52 schwarze 1526: void
1.60 schwarze 1527: roff_setreg(struct roff *r, const char *name, int val, char sign)
1.41 schwarze 1528: {
1.52 schwarze 1529: struct roffreg *reg;
1530:
1531: /* Search for an existing register with the same name. */
1532: reg = r->regtab;
1533:
1534: while (reg && strcmp(name, reg->key.p))
1535: reg = reg->next;
1.41 schwarze 1536:
1.52 schwarze 1537: if (NULL == reg) {
1538: /* Create a new register. */
1539: reg = mandoc_malloc(sizeof(struct roffreg));
1540: reg->key.p = mandoc_strdup(name);
1541: reg->key.sz = strlen(name);
1.60 schwarze 1542: reg->val = 0;
1.52 schwarze 1543: reg->next = r->regtab;
1544: r->regtab = reg;
1545: }
1546:
1.60 schwarze 1547: if ('+' == sign)
1548: reg->val += val;
1549: else if ('-' == sign)
1550: reg->val -= val;
1551: else
1552: reg->val = val;
1.41 schwarze 1553: }
1554:
1.65 schwarze 1555: /*
1556: * Handle some predefined read-only number registers.
1557: * For now, return -1 if the requested register is not predefined;
1558: * in case a predefined read-only register having the value -1
1559: * were to turn up, another special value would have to be chosen.
1560: */
1561: static int
1562: roff_getregro(const char *name)
1563: {
1564:
1565: switch (*name) {
1.80 schwarze 1566: case 'A': /* ASCII approximation mode is always off. */
1.65 schwarze 1567: return(0);
1.80 schwarze 1568: case 'g': /* Groff compatibility mode is always on. */
1.65 schwarze 1569: return(1);
1.80 schwarze 1570: case 'H': /* Fixed horizontal resolution. */
1.65 schwarze 1571: return (24);
1.80 schwarze 1572: case 'j': /* Always adjust left margin only. */
1.65 schwarze 1573: return(0);
1.80 schwarze 1574: case 'T': /* Some output device is always defined. */
1.65 schwarze 1575: return(1);
1.80 schwarze 1576: case 'V': /* Fixed vertical resolution. */
1.65 schwarze 1577: return (40);
1578: default:
1579: return (-1);
1580: }
1581: }
1582:
1.53 schwarze 1583: int
1.52 schwarze 1584: roff_getreg(const struct roff *r, const char *name)
1.41 schwarze 1585: {
1.52 schwarze 1586: struct roffreg *reg;
1.65 schwarze 1587: int val;
1588:
1589: if ('.' == name[0] && '\0' != name[1] && '\0' == name[2]) {
1590: val = roff_getregro(name + 1);
1591: if (-1 != val)
1592: return (val);
1593: }
1.52 schwarze 1594:
1595: for (reg = r->regtab; reg; reg = reg->next)
1596: if (0 == strcmp(name, reg->key.p))
1.53 schwarze 1597: return(reg->val);
1598:
1599: return(0);
1600: }
1601:
1602: static int
1603: roff_getregn(const struct roff *r, const char *name, size_t len)
1604: {
1605: struct roffreg *reg;
1.65 schwarze 1606: int val;
1607:
1608: if ('.' == name[0] && 2 == len) {
1609: val = roff_getregro(name + 1);
1610: if (-1 != val)
1611: return (val);
1612: }
1.53 schwarze 1613:
1614: for (reg = r->regtab; reg; reg = reg->next)
1615: if (len == reg->key.sz &&
1616: 0 == strncmp(name, reg->key.p, len))
1617: return(reg->val);
1.41 schwarze 1618:
1.52 schwarze 1619: return(0);
1.41 schwarze 1620: }
1621:
1.52 schwarze 1622: static void
1623: roff_freereg(struct roffreg *reg)
1.41 schwarze 1624: {
1.52 schwarze 1625: struct roffreg *old_reg;
1.41 schwarze 1626:
1.52 schwarze 1627: while (NULL != reg) {
1628: free(reg->key.p);
1629: old_reg = reg;
1630: reg = reg->next;
1631: free(old_reg);
1632: }
1.41 schwarze 1633: }
1.7 schwarze 1634:
1635: static enum rofferr
1.6 schwarze 1636: roff_nr(ROFF_ARGS)
1.1 schwarze 1637: {
1.85 schwarze 1638: char *key, *val;
1639: size_t keysz;
1.37 schwarze 1640: int iv;
1.60 schwarze 1641: char sign;
1.6 schwarze 1642:
1.85 schwarze 1643: key = val = *bufp + pos;
1644: if ('\0' == *key)
1645: return(ROFF_IGN);
1646:
1647: keysz = roff_getname(r, &val, ln, pos);
1648: if ('\\' == key[keysz])
1649: return(ROFF_IGN);
1650: key[keysz] = '\0';
1.6 schwarze 1651:
1.60 schwarze 1652: sign = *val;
1653: if ('+' == sign || '-' == sign)
1654: val++;
1655:
1.77 schwarze 1656: if (roff_evalnum(val, NULL, &iv, 0))
1657: roff_setreg(r, key, iv, sign);
1.1 schwarze 1658:
1.76 schwarze 1659: return(ROFF_IGN);
1660: }
1661:
1662: static enum rofferr
1663: roff_rr(ROFF_ARGS)
1664: {
1665: struct roffreg *reg, **prev;
1.85 schwarze 1666: char *name, *cp;
1667: size_t namesz;
1.76 schwarze 1668:
1.85 schwarze 1669: name = cp = *bufp + pos;
1670: if ('\0' == *name)
1671: return(ROFF_IGN);
1672: namesz = roff_getname(r, &cp, ln, pos);
1673: name[namesz] = '\0';
1.76 schwarze 1674:
1675: prev = &r->regtab;
1676: while (1) {
1677: reg = *prev;
1678: if (NULL == reg || !strcmp(name, reg->key.p))
1679: break;
1680: prev = ®->next;
1681: }
1682: if (NULL != reg) {
1683: *prev = reg->next;
1684: free(reg->key.p);
1685: free(reg);
1686: }
1.29 schwarze 1687: return(ROFF_IGN);
1688: }
1689:
1690: static enum rofferr
1691: roff_rm(ROFF_ARGS)
1692: {
1693: const char *name;
1694: char *cp;
1.85 schwarze 1695: size_t namesz;
1.29 schwarze 1696:
1697: cp = *bufp + pos;
1698: while ('\0' != *cp) {
1.85 schwarze 1699: name = cp;
1700: namesz = roff_getname(r, &cp, ln, (int)(cp - *bufp));
1701: roff_setstrn(&r->strtab, name, namesz, NULL, 0, 0);
1702: if ('\\' == name[namesz])
1703: break;
1.29 schwarze 1704: }
1.51 schwarze 1705: return(ROFF_IGN);
1706: }
1707:
1708: static enum rofferr
1709: roff_it(ROFF_ARGS)
1710: {
1711: char *cp;
1712: size_t len;
1713: int iv;
1714:
1715: /* Parse the number of lines. */
1716: cp = *bufp + pos;
1717: len = strcspn(cp, " \t");
1718: cp[len] = '\0';
1719: if ((iv = mandoc_strntoi(cp, len, 10)) <= 0) {
1720: mandoc_msg(MANDOCERR_NUMERIC, r->parse,
1.80 schwarze 1721: ln, ppos, *bufp + 1);
1.51 schwarze 1722: return(ROFF_IGN);
1723: }
1724: cp += len + 1;
1725:
1726: /* Arm the input line trap. */
1727: roffit_lines = iv;
1728: roffit_macro = mandoc_strdup(cp);
1.2 schwarze 1729: return(ROFF_IGN);
1.47 schwarze 1730: }
1731:
1732: static enum rofferr
1733: roff_Dd(ROFF_ARGS)
1734: {
1735: const char *const *cp;
1736:
1.72 schwarze 1737: if (0 == ((MPARSE_MDOC | MPARSE_QUICK) & r->options))
1.47 schwarze 1738: for (cp = __mdoc_reserved; *cp; cp++)
1739: roff_setstr(r, *cp, NULL, 0);
1740:
1741: return(ROFF_CONT);
1742: }
1743:
1744: static enum rofferr
1745: roff_TH(ROFF_ARGS)
1746: {
1747: const char *const *cp;
1748:
1.72 schwarze 1749: if (0 == (MPARSE_QUICK & r->options))
1.47 schwarze 1750: for (cp = __man_reserved; *cp; cp++)
1751: roff_setstr(r, *cp, NULL, 0);
1752:
1753: return(ROFF_CONT);
1.14 schwarze 1754: }
1755:
1756: static enum rofferr
1.27 schwarze 1757: roff_TE(ROFF_ARGS)
1758: {
1759:
1760: if (NULL == r->tbl)
1.35 schwarze 1761: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.27 schwarze 1762: else
1.41 schwarze 1763: tbl_end(&r->tbl);
1.27 schwarze 1764:
1765: return(ROFF_IGN);
1766: }
1767:
1768: static enum rofferr
1769: roff_T_(ROFF_ARGS)
1770: {
1771:
1772: if (NULL == r->tbl)
1.35 schwarze 1773: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.27 schwarze 1774: else
1775: tbl_restart(ppos, ln, r->tbl);
1776:
1777: return(ROFF_IGN);
1778: }
1779:
1.41 schwarze 1780: #if 0
1781: static int
1782: roff_closeeqn(struct roff *r)
1783: {
1784:
1785: return(r->eqn && ROFF_EQN == eqn_end(&r->eqn) ? 1 : 0);
1786: }
1787: #endif
1788:
1789: static void
1.80 schwarze 1790: roff_openeqn(struct roff *r, const char *name, int line,
1.41 schwarze 1791: int offs, const char *buf)
1.32 schwarze 1792: {
1.41 schwarze 1793: struct eqn_node *e;
1794: int poff;
1.32 schwarze 1795:
1796: assert(NULL == r->eqn);
1.41 schwarze 1797: e = eqn_alloc(name, offs, line, r->parse);
1.32 schwarze 1798:
1799: if (r->last_eqn)
1800: r->last_eqn->next = e;
1801: else
1802: r->first_eqn = r->last_eqn = e;
1803:
1804: r->eqn = r->last_eqn = e;
1.41 schwarze 1805:
1806: if (buf) {
1807: poff = 0;
1808: eqn_read(&r->eqn, line, buf, offs, &poff);
1809: }
1810: }
1811:
1812: static enum rofferr
1813: roff_EQ(ROFF_ARGS)
1814: {
1815:
1816: roff_openeqn(r, *bufp + pos, ln, ppos, NULL);
1.32 schwarze 1817: return(ROFF_IGN);
1818: }
1819:
1820: static enum rofferr
1821: roff_EN(ROFF_ARGS)
1822: {
1823:
1.35 schwarze 1824: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.32 schwarze 1825: return(ROFF_IGN);
1826: }
1827:
1828: static enum rofferr
1.27 schwarze 1829: roff_TS(ROFF_ARGS)
1830: {
1.49 schwarze 1831: struct tbl_node *tbl;
1.27 schwarze 1832:
1833: if (r->tbl) {
1.35 schwarze 1834: mandoc_msg(MANDOCERR_SCOPEBROKEN, r->parse, ln, ppos, NULL);
1.41 schwarze 1835: tbl_end(&r->tbl);
1.27 schwarze 1836: }
1837:
1.49 schwarze 1838: tbl = tbl_alloc(ppos, ln, r->parse);
1.27 schwarze 1839:
1840: if (r->last_tbl)
1.49 schwarze 1841: r->last_tbl->next = tbl;
1.27 schwarze 1842: else
1.49 schwarze 1843: r->first_tbl = r->last_tbl = tbl;
1.27 schwarze 1844:
1.49 schwarze 1845: r->tbl = r->last_tbl = tbl;
1.27 schwarze 1846: return(ROFF_IGN);
1847: }
1848:
1849: static enum rofferr
1.48 schwarze 1850: roff_cc(ROFF_ARGS)
1851: {
1852: const char *p;
1853:
1854: p = *bufp + pos;
1855:
1856: if ('\0' == *p || '.' == (r->control = *p++))
1857: r->control = 0;
1858:
1859: if ('\0' != *p)
1860: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
1861:
1862: return(ROFF_IGN);
1863: }
1864:
1865: static enum rofferr
1.42 schwarze 1866: roff_tr(ROFF_ARGS)
1867: {
1868: const char *p, *first, *second;
1869: size_t fsz, ssz;
1870: enum mandoc_esc esc;
1871:
1872: p = *bufp + pos;
1873:
1874: if ('\0' == *p) {
1875: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
1876: return(ROFF_IGN);
1877: }
1878:
1879: while ('\0' != *p) {
1880: fsz = ssz = 1;
1881:
1882: first = p++;
1883: if ('\\' == *first) {
1884: esc = mandoc_escape(&p, NULL, NULL);
1885: if (ESCAPE_ERROR == esc) {
1.80 schwarze 1886: mandoc_msg(MANDOCERR_BADESCAPE,
1887: r->parse, ln,
1888: (int)(p - *bufp), NULL);
1.42 schwarze 1889: return(ROFF_IGN);
1890: }
1891: fsz = (size_t)(p - first);
1892: }
1893:
1894: second = p++;
1895: if ('\\' == *second) {
1896: esc = mandoc_escape(&p, NULL, NULL);
1897: if (ESCAPE_ERROR == esc) {
1.80 schwarze 1898: mandoc_msg(MANDOCERR_BADESCAPE,
1899: r->parse, ln,
1900: (int)(p - *bufp), NULL);
1.42 schwarze 1901: return(ROFF_IGN);
1902: }
1903: ssz = (size_t)(p - second);
1904: } else if ('\0' == *second) {
1.80 schwarze 1905: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse,
1906: ln, (int)(p - *bufp), NULL);
1.42 schwarze 1907: second = " ";
1908: p--;
1909: }
1910:
1911: if (fsz > 1) {
1.80 schwarze 1912: roff_setstrn(&r->xmbtab, first, fsz,
1913: second, ssz, 0);
1.42 schwarze 1914: continue;
1915: }
1916:
1917: if (NULL == r->xtab)
1.80 schwarze 1918: r->xtab = mandoc_calloc(128,
1919: sizeof(struct roffstr));
1.42 schwarze 1920:
1921: free(r->xtab[(int)*first].p);
1922: r->xtab[(int)*first].p = mandoc_strndup(second, ssz);
1923: r->xtab[(int)*first].sz = ssz;
1924: }
1925:
1926: return(ROFF_IGN);
1927: }
1928:
1929: static enum rofferr
1.14 schwarze 1930: roff_so(ROFF_ARGS)
1931: {
1932: char *name;
1.15 schwarze 1933:
1.83 schwarze 1934: name = *bufp + pos;
1935: mandoc_vmsg(MANDOCERR_SO, r->parse, ln, ppos, ".so %s", name);
1.14 schwarze 1936:
1.22 schwarze 1937: /*
1938: * Handle `so'. Be EXTREMELY careful, as we shouldn't be
1939: * opening anything that's not in our cwd or anything beneath
1940: * it. Thus, explicitly disallow traversing up the file-system
1941: * or using absolute paths.
1942: */
1943:
1.14 schwarze 1944: if ('/' == *name || strstr(name, "../") || strstr(name, "/..")) {
1.83 schwarze 1945: mandoc_vmsg(MANDOCERR_SO_PATH, r->parse, ln, ppos,
1946: ".so %s", name);
1.14 schwarze 1947: return(ROFF_ERR);
1948: }
1949:
1950: *offs = pos;
1951: return(ROFF_SO);
1.7 schwarze 1952: }
1953:
1.16 schwarze 1954: static enum rofferr
1955: roff_userdef(ROFF_ARGS)
1.12 schwarze 1956: {
1.16 schwarze 1957: const char *arg[9];
1958: char *cp, *n1, *n2;
1.25 schwarze 1959: int i;
1.12 schwarze 1960:
1.16 schwarze 1961: /*
1962: * Collect pointers to macro argument strings
1.61 schwarze 1963: * and NUL-terminate them.
1.16 schwarze 1964: */
1965: cp = *bufp + pos;
1.25 schwarze 1966: for (i = 0; i < 9; i++)
1.26 schwarze 1967: arg[i] = '\0' == *cp ? "" :
1.35 schwarze 1968: mandoc_getarg(r->parse, &cp, ln, &pos);
1.16 schwarze 1969:
1970: /*
1971: * Expand macro arguments.
1.12 schwarze 1972: */
1.16 schwarze 1973: *szp = 0;
1974: n1 = cp = mandoc_strdup(r->current_string);
1975: while (NULL != (cp = strstr(cp, "\\$"))) {
1976: i = cp[2] - '1';
1977: if (0 > i || 8 < i) {
1978: /* Not an argument invocation. */
1979: cp += 2;
1980: continue;
1981: }
1.82 schwarze 1982: *cp = '\0';
1983: *szp = mandoc_asprintf(&n2, "%s%s%s",
1984: n1, arg[i], cp + 3) + 1;
1.16 schwarze 1985: cp = n2 + (cp - n1);
1986: free(n1);
1987: n1 = n2;
1.12 schwarze 1988: }
1989:
1.16 schwarze 1990: /*
1991: * Replace the macro invocation
1992: * by the expanded macro.
1993: */
1994: free(*bufp);
1995: *bufp = n1;
1996: if (0 == *szp)
1997: *szp = strlen(*bufp) + 1;
1998:
1.19 schwarze 1999: return(*szp > 1 && '\n' == (*bufp)[(int)*szp - 2] ?
1.16 schwarze 2000: ROFF_REPARSE : ROFF_APPEND);
1.12 schwarze 2001: }
1.28 schwarze 2002:
1.85 schwarze 2003: static size_t
1.28 schwarze 2004: roff_getname(struct roff *r, char **cpp, int ln, int pos)
2005: {
2006: char *name, *cp;
1.85 schwarze 2007: size_t namesz;
1.28 schwarze 2008:
2009: name = *cpp;
2010: if ('\0' == *name)
1.85 schwarze 2011: return(0);
1.28 schwarze 2012:
1.85 schwarze 2013: /* Read until end of name and terminate it with NUL. */
2014: for (cp = name; 1; cp++) {
2015: if ('\0' == *cp || ' ' == *cp) {
2016: namesz = cp - name;
2017: break;
2018: }
1.28 schwarze 2019: if ('\\' != *cp)
2020: continue;
1.88 schwarze 2021: namesz = cp - name;
2022: if ('{' == cp[1] || '}' == cp[1])
2023: break;
1.28 schwarze 2024: cp++;
2025: if ('\\' == *cp)
2026: continue;
1.35 schwarze 2027: mandoc_msg(MANDOCERR_NAMESC, r->parse, ln, pos, NULL);
1.85 schwarze 2028: mandoc_escape((const char **)&cp, NULL, NULL);
2029: break;
1.28 schwarze 2030: }
2031:
2032: /* Read past spaces. */
2033: while (' ' == *cp)
2034: cp++;
2035:
2036: *cpp = cp;
1.85 schwarze 2037: return(namesz);
1.28 schwarze 2038: }
2039:
1.16 schwarze 2040: /*
2041: * Store *string into the user-defined string called *name.
2042: * To clear an existing entry, call with (*r, *name, NULL, 0).
1.66 schwarze 2043: * append == 0: replace mode
2044: * append == 1: single-line append mode
2045: * append == 2: multiline append mode, append '\n' after each call
1.16 schwarze 2046: */
1.8 schwarze 2047: static void
1.16 schwarze 2048: roff_setstr(struct roff *r, const char *name, const char *string,
1.66 schwarze 2049: int append)
1.7 schwarze 2050: {
1.42 schwarze 2051:
2052: roff_setstrn(&r->strtab, name, strlen(name), string,
1.80 schwarze 2053: string ? strlen(string) : 0, append);
1.42 schwarze 2054: }
2055:
2056: static void
2057: roff_setstrn(struct roffkv **r, const char *name, size_t namesz,
1.66 schwarze 2058: const char *string, size_t stringsz, int append)
1.42 schwarze 2059: {
2060: struct roffkv *n;
2061: char *c;
2062: int i;
2063: size_t oldch, newch;
1.7 schwarze 2064:
1.16 schwarze 2065: /* Search for an existing string with the same name. */
1.42 schwarze 2066: n = *r;
2067:
1.84 schwarze 2068: while (n && (namesz != n->key.sz ||
2069: strncmp(n->key.p, name, namesz)))
1.7 schwarze 2070: n = n->next;
1.8 schwarze 2071:
2072: if (NULL == n) {
1.16 schwarze 2073: /* Create a new string table entry. */
1.42 schwarze 2074: n = mandoc_malloc(sizeof(struct roffkv));
2075: n->key.p = mandoc_strndup(name, namesz);
2076: n->key.sz = namesz;
2077: n->val.p = NULL;
2078: n->val.sz = 0;
2079: n->next = *r;
2080: *r = n;
1.66 schwarze 2081: } else if (0 == append) {
1.42 schwarze 2082: free(n->val.p);
2083: n->val.p = NULL;
2084: n->val.sz = 0;
1.16 schwarze 2085: }
2086:
2087: if (NULL == string)
2088: return;
2089:
2090: /*
2091: * One additional byte for the '\n' in multiline mode,
2092: * and one for the terminating '\0'.
2093: */
1.66 schwarze 2094: newch = stringsz + (1 < append ? 2u : 1u);
1.42 schwarze 2095:
2096: if (NULL == n->val.p) {
2097: n->val.p = mandoc_malloc(newch);
2098: *n->val.p = '\0';
1.16 schwarze 2099: oldch = 0;
2100: } else {
1.42 schwarze 2101: oldch = n->val.sz;
2102: n->val.p = mandoc_realloc(n->val.p, oldch + newch);
1.16 schwarze 2103: }
2104:
2105: /* Skip existing content in the destination buffer. */
1.42 schwarze 2106: c = n->val.p + (int)oldch;
1.16 schwarze 2107:
2108: /* Append new content to the destination buffer. */
1.42 schwarze 2109: i = 0;
2110: while (i < (int)stringsz) {
1.16 schwarze 2111: /*
2112: * Rudimentary roff copy mode:
2113: * Handle escaped backslashes.
2114: */
1.42 schwarze 2115: if ('\\' == string[i] && '\\' == string[i + 1])
2116: i++;
2117: *c++ = string[i++];
1.16 schwarze 2118: }
1.8 schwarze 2119:
1.16 schwarze 2120: /* Append terminating bytes. */
1.66 schwarze 2121: if (1 < append)
1.16 schwarze 2122: *c++ = '\n';
1.42 schwarze 2123:
1.16 schwarze 2124: *c = '\0';
1.42 schwarze 2125: n->val.sz = (int)(c - n->val.p);
1.7 schwarze 2126: }
2127:
1.8 schwarze 2128: static const char *
2129: roff_getstrn(const struct roff *r, const char *name, size_t len)
1.7 schwarze 2130: {
1.42 schwarze 2131: const struct roffkv *n;
1.64 schwarze 2132: int i;
1.7 schwarze 2133:
1.42 schwarze 2134: for (n = r->strtab; n; n = n->next)
1.80 schwarze 2135: if (0 == strncmp(name, n->key.p, len) &&
2136: '\0' == n->key.p[(int)len])
1.42 schwarze 2137: return(n->val.p);
1.64 schwarze 2138:
2139: for (i = 0; i < PREDEFS_MAX; i++)
2140: if (0 == strncmp(name, predefs[i].name, len) &&
2141: '\0' == predefs[i].name[(int)len])
2142: return(predefs[i].str);
1.8 schwarze 2143:
1.42 schwarze 2144: return(NULL);
1.7 schwarze 2145: }
2146:
1.8 schwarze 2147: static void
1.42 schwarze 2148: roff_freestr(struct roffkv *r)
1.7 schwarze 2149: {
1.42 schwarze 2150: struct roffkv *n, *nn;
1.7 schwarze 2151:
1.42 schwarze 2152: for (n = r; n; n = nn) {
2153: free(n->key.p);
2154: free(n->val.p);
1.7 schwarze 2155: nn = n->next;
2156: free(n);
2157: }
1.27 schwarze 2158: }
2159:
2160: const struct tbl_span *
2161: roff_span(const struct roff *r)
2162: {
1.80 schwarze 2163:
1.27 schwarze 2164: return(r->tbl ? tbl_span(r->tbl) : NULL);
1.32 schwarze 2165: }
2166:
2167: const struct eqn *
2168: roff_eqn(const struct roff *r)
2169: {
1.80 schwarze 2170:
1.32 schwarze 2171: return(r->last_eqn ? &r->last_eqn->eqn : NULL);
1.42 schwarze 2172: }
2173:
2174: /*
2175: * Duplicate an input string, making the appropriate character
2176: * conversations (as stipulated by `tr') along the way.
2177: * Returns a heap-allocated string with all the replacements made.
2178: */
2179: char *
2180: roff_strdup(const struct roff *r, const char *p)
2181: {
2182: const struct roffkv *cp;
2183: char *res;
2184: const char *pp;
2185: size_t ssz, sz;
2186: enum mandoc_esc esc;
2187:
2188: if (NULL == r->xmbtab && NULL == r->xtab)
2189: return(mandoc_strdup(p));
2190: else if ('\0' == *p)
2191: return(mandoc_strdup(""));
2192:
2193: /*
2194: * Step through each character looking for term matches
2195: * (remember that a `tr' can be invoked with an escape, which is
2196: * a glyph but the escape is multi-character).
2197: * We only do this if the character hash has been initialised
2198: * and the string is >0 length.
2199: */
2200:
2201: res = NULL;
2202: ssz = 0;
2203:
2204: while ('\0' != *p) {
2205: if ('\\' != *p && r->xtab && r->xtab[(int)*p].p) {
2206: sz = r->xtab[(int)*p].sz;
2207: res = mandoc_realloc(res, ssz + sz + 1);
2208: memcpy(res + ssz, r->xtab[(int)*p].p, sz);
2209: ssz += sz;
2210: p++;
2211: continue;
2212: } else if ('\\' != *p) {
2213: res = mandoc_realloc(res, ssz + 2);
2214: res[ssz++] = *p++;
2215: continue;
2216: }
2217:
2218: /* Search for term matches. */
2219: for (cp = r->xmbtab; cp; cp = cp->next)
2220: if (0 == strncmp(p, cp->key.p, cp->key.sz))
2221: break;
2222:
2223: if (NULL != cp) {
2224: /*
2225: * A match has been found.
2226: * Append the match to the array and move
2227: * forward by its keysize.
2228: */
1.80 schwarze 2229: res = mandoc_realloc(res,
2230: ssz + cp->val.sz + 1);
1.42 schwarze 2231: memcpy(res + ssz, cp->val.p, cp->val.sz);
2232: ssz += cp->val.sz;
2233: p += (int)cp->key.sz;
2234: continue;
2235: }
2236:
2237: /*
2238: * Handle escapes carefully: we need to copy
2239: * over just the escape itself, or else we might
2240: * do replacements within the escape itself.
2241: * Make sure to pass along the bogus string.
2242: */
2243: pp = p++;
2244: esc = mandoc_escape(&p, NULL, NULL);
2245: if (ESCAPE_ERROR == esc) {
2246: sz = strlen(pp);
2247: res = mandoc_realloc(res, ssz + sz + 1);
2248: memcpy(res + ssz, pp, sz);
2249: break;
2250: }
1.80 schwarze 2251: /*
2252: * We bail out on bad escapes.
1.42 schwarze 2253: * No need to warn: we already did so when
2254: * roff_res() was called.
2255: */
2256: sz = (int)(p - pp);
2257: res = mandoc_realloc(res, ssz + sz + 1);
2258: memcpy(res + ssz, pp, sz);
2259: ssz += sz;
2260: }
2261:
2262: res[(int)ssz] = '\0';
2263: return(res);
1.48 schwarze 2264: }
2265:
2266: /*
1.80 schwarze 2267: * Find out whether a line is a macro line or not.
1.48 schwarze 2268: * If it is, adjust the current position and return one; if it isn't,
2269: * return zero and don't change the current position.
2270: * If the control character has been set with `.cc', then let that grain
2271: * precedence.
2272: * This is slighly contrary to groff, where using the non-breaking
2273: * control character when `cc' has been invoked will cause the
2274: * non-breaking macro contents to be printed verbatim.
2275: */
2276: int
2277: roff_getcontrol(const struct roff *r, const char *cp, int *ppos)
2278: {
2279: int pos;
2280:
2281: pos = *ppos;
2282:
2283: if (0 != r->control && cp[pos] == r->control)
2284: pos++;
2285: else if (0 != r->control)
2286: return(0);
2287: else if ('\\' == cp[pos] && '.' == cp[pos + 1])
2288: pos += 2;
2289: else if ('.' == cp[pos] || '\'' == cp[pos])
2290: pos++;
2291: else
2292: return(0);
2293:
2294: while (' ' == cp[pos] || '\t' == cp[pos])
2295: pos++;
2296:
2297: *ppos = pos;
2298: return(1);
1.1 schwarze 2299: }
![[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