Annotation of src/usr.bin/mandoc/roff.c, Revision 1.94
1.94 ! schwarze 1: /* $Id: roff.c,v 1.93 2014/07/07 11:34:41 schwarze Exp $ */
1.1 schwarze 2: /*
1.48 schwarze 3: * Copyright (c) 2010, 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
1.63 schwarze 4: * Copyright (c) 2010-2014 Ingo Schwarze <schwarze@openbsd.org>
1.1 schwarze 5: *
6: * Permission to use, copy, modify, and distribute this software for any
7: * purpose with or without fee is hereby granted, provided that the above
8: * copyright notice and this permission notice appear in all copies.
9: *
1.16 schwarze 10: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.1 schwarze 11: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.16 schwarze 12: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.1 schwarze 13: * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
14: * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
15: * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
16: * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
17: */
18: #include <assert.h>
1.3 schwarze 19: #include <ctype.h>
1.51 schwarze 20: #include <stdio.h>
1.1 schwarze 21: #include <stdlib.h>
22: #include <string.h>
23:
24: #include "mandoc.h"
1.74 schwarze 25: #include "mandoc_aux.h"
1.27 schwarze 26: #include "libroff.h"
1.8 schwarze 27: #include "libmandoc.h"
1.1 schwarze 28:
1.37 schwarze 29: /* Maximum number of nested if-else conditionals. */
1.2 schwarze 30: #define RSTACK_MAX 128
31:
1.43 schwarze 32: /* Maximum number of string expansions per line, to break infinite loops. */
33: #define EXPAND_LIMIT 1000
34:
1.1 schwarze 35: enum rofft {
1.20 schwarze 36: ROFF_ad,
1.2 schwarze 37: ROFF_am,
38: ROFF_ami,
39: ROFF_am1,
1.66 schwarze 40: ROFF_as,
1.48 schwarze 41: ROFF_cc,
1.67 schwarze 42: ROFF_ce,
1.1 schwarze 43: ROFF_de,
44: ROFF_dei,
1.2 schwarze 45: ROFF_de1,
46: ROFF_ds,
47: ROFF_el,
1.58 schwarze 48: ROFF_fam,
1.59 schwarze 49: ROFF_hw,
1.20 schwarze 50: ROFF_hy,
1.2 schwarze 51: ROFF_ie,
52: ROFF_if,
1.1 schwarze 53: ROFF_ig,
1.30 schwarze 54: ROFF_it,
1.20 schwarze 55: ROFF_ne,
56: ROFF_nh,
1.14 schwarze 57: ROFF_nr,
1.31 schwarze 58: ROFF_ns,
59: ROFF_ps,
1.2 schwarze 60: ROFF_rm,
1.76 schwarze 61: ROFF_rr,
1.14 schwarze 62: ROFF_so,
1.31 schwarze 63: ROFF_ta,
1.2 schwarze 64: ROFF_tr,
1.47 schwarze 65: ROFF_Dd,
66: ROFF_TH,
1.27 schwarze 67: ROFF_TS,
68: ROFF_TE,
69: ROFF_T_,
1.32 schwarze 70: ROFF_EQ,
71: ROFF_EN,
1.2 schwarze 72: ROFF_cblock,
1.16 schwarze 73: ROFF_USERDEF,
1.1 schwarze 74: ROFF_MAX
75: };
76:
1.41 schwarze 77: /*
1.42 schwarze 78: * An incredibly-simple string buffer.
79: */
1.8 schwarze 80: struct roffstr {
1.42 schwarze 81: char *p; /* nil-terminated buffer */
82: size_t sz; /* saved strlen(p) */
83: };
84:
85: /*
86: * A key-value roffstr pair as part of a singly-linked list.
87: */
88: struct roffkv {
89: struct roffstr key;
90: struct roffstr val;
91: struct roffkv *next; /* next in list */
1.8 schwarze 92: };
93:
1.52 schwarze 94: /*
95: * A single number register as part of a singly-linked list.
96: */
97: struct roffreg {
98: struct roffstr key;
1.53 schwarze 99: int val;
1.52 schwarze 100: struct roffreg *next;
101: };
102:
1.1 schwarze 103: struct roff {
1.35 schwarze 104: struct mparse *parse; /* parse point */
1.72 schwarze 105: int options; /* parse options */
1.1 schwarze 106: struct roffnode *last; /* leaf of stack */
1.71 schwarze 107: int rstack[RSTACK_MAX]; /* stack of !`ie' rules */
1.48 schwarze 108: char control; /* control character */
1.2 schwarze 109: int rstackpos; /* position in rstack */
1.52 schwarze 110: struct roffreg *regtab; /* number registers */
1.42 schwarze 111: struct roffkv *strtab; /* user-defined strings & macros */
112: struct roffkv *xmbtab; /* multi-byte trans table (`tr') */
113: struct roffstr *xtab; /* single-byte trans table (`tr') */
1.16 schwarze 114: const char *current_string; /* value of last called user macro */
1.27 schwarze 115: struct tbl_node *first_tbl; /* first table parsed */
116: struct tbl_node *last_tbl; /* last table parsed */
117: struct tbl_node *tbl; /* current table being parsed */
1.32 schwarze 118: struct eqn_node *last_eqn; /* last equation parsed */
119: struct eqn_node *first_eqn; /* first equation parsed */
120: struct eqn_node *eqn; /* current equation being parsed */
1.1 schwarze 121: };
122:
123: struct roffnode {
124: enum rofft tok; /* type of node */
125: struct roffnode *parent; /* up one in stack */
126: int line; /* parse line */
127: int col; /* parse col */
1.16 schwarze 128: char *name; /* node name, e.g. macro name */
1.2 schwarze 129: char *end; /* end-rules: custom token */
130: int endspan; /* end-rules: next-line or infty */
1.71 schwarze 131: int rule; /* current evaluation rule */
1.1 schwarze 132: };
133:
134: #define ROFF_ARGS struct roff *r, /* parse ctx */ \
135: enum rofft tok, /* tok of macro */ \
1.80 schwarze 136: char **bufp, /* input buffer */ \
1.1 schwarze 137: size_t *szp, /* size of input buffer */ \
138: int ln, /* parse line */ \
1.2 schwarze 139: int ppos, /* original pos in buffer */ \
140: int pos, /* current pos in buffer */ \
141: int *offs /* reset offset of buffer data */
1.1 schwarze 142:
143: typedef enum rofferr (*roffproc)(ROFF_ARGS);
144:
145: struct roffmac {
146: const char *name; /* macro name */
1.2 schwarze 147: roffproc proc; /* process new macro */
148: roffproc text; /* process as child text of macro */
149: roffproc sub; /* process as child of macro */
150: int flags;
151: #define ROFFMAC_STRUCT (1 << 0) /* always interpret */
1.3 schwarze 152: struct roffmac *next;
1.1 schwarze 153: };
154:
1.37 schwarze 155: struct predef {
156: const char *name; /* predefined input name */
157: const char *str; /* replacement symbol */
158: };
159:
160: #define PREDEF(__name, __str) \
161: { (__name), (__str) },
162:
1.42 schwarze 163: static enum rofft roffhash_find(const char *, size_t);
164: static void roffhash_init(void);
165: static void roffnode_cleanscope(struct roff *);
166: static void roffnode_pop(struct roff *);
167: static void roffnode_push(struct roff *, enum rofft,
168: const char *, int, int);
1.2 schwarze 169: static enum rofferr roff_block(ROFF_ARGS);
170: static enum rofferr roff_block_text(ROFF_ARGS);
171: static enum rofferr roff_block_sub(ROFF_ARGS);
172: static enum rofferr roff_cblock(ROFF_ARGS);
1.48 schwarze 173: static enum rofferr roff_cc(ROFF_ARGS);
1.68 schwarze 174: static void roff_ccond(struct roff *, int, int);
1.2 schwarze 175: static enum rofferr roff_cond(ROFF_ARGS);
176: static enum rofferr roff_cond_text(ROFF_ARGS);
177: static enum rofferr roff_cond_sub(ROFF_ARGS);
1.7 schwarze 178: static enum rofferr roff_ds(ROFF_ARGS);
1.71 schwarze 179: static int roff_evalcond(const char *, int *);
1.77 schwarze 180: static int roff_evalnum(const char *, int *, int *, int);
181: static int roff_evalpar(const char *, int *, int *);
1.71 schwarze 182: static int roff_evalstrcond(const char *, int *);
1.42 schwarze 183: static void roff_free1(struct roff *);
1.52 schwarze 184: static void roff_freereg(struct roffreg *);
1.42 schwarze 185: static void roff_freestr(struct roffkv *);
1.85 schwarze 186: static size_t roff_getname(struct roff *, char **, int, int);
1.56 schwarze 187: static int roff_getnum(const char *, int *, int *);
188: static int roff_getop(const char *, int *, char *);
1.53 schwarze 189: static int roff_getregn(const struct roff *,
190: const char *, size_t);
1.65 schwarze 191: static int roff_getregro(const char *name);
1.80 schwarze 192: static const char *roff_getstrn(const struct roff *,
1.8 schwarze 193: const char *, size_t);
1.51 schwarze 194: static enum rofferr roff_it(ROFF_ARGS);
1.21 schwarze 195: static enum rofferr roff_line_ignore(ROFF_ARGS);
1.6 schwarze 196: static enum rofferr roff_nr(ROFF_ARGS);
1.41 schwarze 197: static void roff_openeqn(struct roff *, const char *,
198: int, int, const char *);
1.87 schwarze 199: static enum rofft roff_parse(struct roff *, char *, int *,
200: int, int);
1.51 schwarze 201: static enum rofferr roff_parsetext(char **, size_t *, int, int *);
1.80 schwarze 202: static enum rofferr roff_res(struct roff *,
1.37 schwarze 203: char **, size_t *, int, int);
1.29 schwarze 204: static enum rofferr roff_rm(ROFF_ARGS);
1.76 schwarze 205: static enum rofferr roff_rr(ROFF_ARGS);
1.8 schwarze 206: static void roff_setstr(struct roff *,
1.16 schwarze 207: const char *, const char *, int);
1.80 schwarze 208: static void roff_setstrn(struct roffkv **, const char *,
1.42 schwarze 209: size_t, const char *, size_t, int);
1.14 schwarze 210: static enum rofferr roff_so(ROFF_ARGS);
1.42 schwarze 211: static enum rofferr roff_tr(ROFF_ARGS);
1.47 schwarze 212: static enum rofferr roff_Dd(ROFF_ARGS);
213: static enum rofferr roff_TH(ROFF_ARGS);
1.27 schwarze 214: static enum rofferr roff_TE(ROFF_ARGS);
215: static enum rofferr roff_TS(ROFF_ARGS);
1.32 schwarze 216: static enum rofferr roff_EQ(ROFF_ARGS);
217: static enum rofferr roff_EN(ROFF_ARGS);
1.27 schwarze 218: static enum rofferr roff_T_(ROFF_ARGS);
1.16 schwarze 219: static enum rofferr roff_userdef(ROFF_ARGS);
1.1 schwarze 220:
1.42 schwarze 221: /* See roffhash_find() */
1.3 schwarze 222:
223: #define ASCII_HI 126
224: #define ASCII_LO 33
225: #define HASHWIDTH (ASCII_HI - ASCII_LO + 1)
226:
227: static struct roffmac *hash[HASHWIDTH];
228:
229: static struct roffmac roffs[ROFF_MAX] = {
1.21 schwarze 230: { "ad", roff_line_ignore, NULL, NULL, 0, NULL },
1.3 schwarze 231: { "am", roff_block, roff_block_text, roff_block_sub, 0, NULL },
232: { "ami", roff_block, roff_block_text, roff_block_sub, 0, NULL },
233: { "am1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.66 schwarze 234: { "as", roff_ds, NULL, NULL, 0, NULL },
1.48 schwarze 235: { "cc", roff_cc, NULL, NULL, 0, NULL },
1.67 schwarze 236: { "ce", roff_line_ignore, NULL, NULL, 0, NULL },
1.3 schwarze 237: { "de", roff_block, roff_block_text, roff_block_sub, 0, NULL },
238: { "dei", roff_block, roff_block_text, roff_block_sub, 0, NULL },
239: { "de1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.7 schwarze 240: { "ds", roff_ds, NULL, NULL, 0, NULL },
1.3 schwarze 241: { "el", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
1.58 schwarze 242: { "fam", roff_line_ignore, NULL, NULL, 0, NULL },
1.59 schwarze 243: { "hw", roff_line_ignore, NULL, NULL, 0, NULL },
1.21 schwarze 244: { "hy", roff_line_ignore, NULL, NULL, 0, NULL },
1.3 schwarze 245: { "ie", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
246: { "if", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
247: { "ig", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.51 schwarze 248: { "it", roff_it, NULL, NULL, 0, NULL },
1.21 schwarze 249: { "ne", roff_line_ignore, NULL, NULL, 0, NULL },
250: { "nh", roff_line_ignore, NULL, NULL, 0, NULL },
1.14 schwarze 251: { "nr", roff_nr, NULL, NULL, 0, NULL },
1.31 schwarze 252: { "ns", roff_line_ignore, NULL, NULL, 0, NULL },
253: { "ps", roff_line_ignore, NULL, NULL, 0, NULL },
1.29 schwarze 254: { "rm", roff_rm, NULL, NULL, 0, NULL },
1.76 schwarze 255: { "rr", roff_rr, NULL, NULL, 0, NULL },
1.14 schwarze 256: { "so", roff_so, NULL, NULL, 0, NULL },
1.31 schwarze 257: { "ta", roff_line_ignore, NULL, NULL, 0, NULL },
1.42 schwarze 258: { "tr", roff_tr, NULL, NULL, 0, NULL },
1.47 schwarze 259: { "Dd", roff_Dd, NULL, NULL, 0, NULL },
260: { "TH", roff_TH, NULL, NULL, 0, NULL },
1.27 schwarze 261: { "TS", roff_TS, NULL, NULL, 0, NULL },
262: { "TE", roff_TE, NULL, NULL, 0, NULL },
263: { "T&", roff_T_, NULL, NULL, 0, NULL },
1.32 schwarze 264: { "EQ", roff_EQ, NULL, NULL, 0, NULL },
265: { "EN", roff_EN, NULL, NULL, 0, NULL },
1.3 schwarze 266: { ".", roff_cblock, NULL, NULL, 0, NULL },
1.16 schwarze 267: { NULL, roff_userdef, NULL, NULL, 0, NULL },
1.1 schwarze 268: };
269:
1.73 schwarze 270: /* not currently implemented: Ds em Eq LP Me PP pp Or Rd Sf SH */
1.47 schwarze 271: const char *const __mdoc_reserved[] = {
272: "Ac", "Ad", "An", "Ao", "Ap", "Aq", "Ar", "At",
273: "Bc", "Bd", "Bf", "Bk", "Bl", "Bo", "Bq",
274: "Brc", "Bro", "Brq", "Bsx", "Bt", "Bx",
275: "Cd", "Cm", "Db", "Dc", "Dd", "Dl", "Do", "Dq",
1.73 schwarze 276: "Dt", "Dv", "Dx", "D1",
277: "Ec", "Ed", "Ef", "Ek", "El", "Em",
278: "En", "Eo", "Er", "Es", "Ev", "Ex",
1.47 schwarze 279: "Fa", "Fc", "Fd", "Fl", "Fn", "Fo", "Fr", "Ft", "Fx",
1.73 schwarze 280: "Hf", "Ic", "In", "It", "Lb", "Li", "Lk", "Lp",
281: "Ms", "Mt", "Nd", "Nm", "No", "Ns", "Nx",
1.47 schwarze 282: "Oc", "Oo", "Op", "Os", "Ot", "Ox",
1.73 schwarze 283: "Pa", "Pc", "Pf", "Po", "Pp", "Pq",
284: "Qc", "Ql", "Qo", "Qq", "Re", "Rs", "Rv",
285: "Sc", "Sh", "Sm", "So", "Sq",
1.47 schwarze 286: "Ss", "St", "Sx", "Sy",
287: "Ta", "Tn", "Ud", "Ux", "Va", "Vt", "Xc", "Xo", "Xr",
1.73 schwarze 288: "%A", "%B", "%C", "%D", "%I", "%J", "%N", "%O",
1.47 schwarze 289: "%P", "%Q", "%R", "%T", "%U", "%V",
290: NULL
291: };
292:
1.73 schwarze 293: /* not currently implemented: BT DE DS ME MT PT SY TQ YS */
1.47 schwarze 294: const char *const __man_reserved[] = {
1.73 schwarze 295: "AT", "B", "BI", "BR", "DT",
296: "EE", "EN", "EQ", "EX", "HP", "I", "IB", "IP", "IR",
297: "LP", "OP", "P", "PD", "PP",
298: "R", "RB", "RE", "RI", "RS", "SB", "SH", "SM", "SS",
299: "TE", "TH", "TP", "TS", "T&", "UC", "UE", "UR",
1.47 schwarze 300: NULL
301: };
302:
1.37 schwarze 303: /* Array of injected predefined strings. */
304: #define PREDEFS_MAX 38
305: static const struct predef predefs[PREDEFS_MAX] = {
306: #include "predefs.in"
307: };
308:
1.42 schwarze 309: /* See roffhash_find() */
1.3 schwarze 310: #define ROFF_HASH(p) (p[0] - ASCII_LO)
311:
1.51 schwarze 312: static int roffit_lines; /* number of lines to delay */
313: static char *roffit_macro; /* nil-terminated macro line */
314:
1.80 schwarze 315:
1.3 schwarze 316: static void
1.42 schwarze 317: roffhash_init(void)
1.3 schwarze 318: {
319: struct roffmac *n;
320: int buc, i;
321:
1.16 schwarze 322: for (i = 0; i < (int)ROFF_USERDEF; i++) {
1.3 schwarze 323: assert(roffs[i].name[0] >= ASCII_LO);
324: assert(roffs[i].name[0] <= ASCII_HI);
325:
326: buc = ROFF_HASH(roffs[i].name);
327:
328: if (NULL != (n = hash[buc])) {
329: for ( ; n->next; n = n->next)
330: /* Do nothing. */ ;
331: n->next = &roffs[i];
332: } else
333: hash[buc] = &roffs[i];
334: }
335: }
336:
1.1 schwarze 337: /*
338: * Look up a roff token by its name. Returns ROFF_MAX if no macro by
339: * the nil-terminated string name could be found.
340: */
341: static enum rofft
1.42 schwarze 342: roffhash_find(const char *p, size_t s)
1.1 schwarze 343: {
1.3 schwarze 344: int buc;
345: struct roffmac *n;
1.1 schwarze 346:
1.3 schwarze 347: /*
348: * libroff has an extremely simple hashtable, for the time
349: * being, which simply keys on the first character, which must
350: * be printable, then walks a chain. It works well enough until
351: * optimised.
352: */
353:
354: if (p[0] < ASCII_LO || p[0] > ASCII_HI)
355: return(ROFF_MAX);
356:
357: buc = ROFF_HASH(p);
358:
359: if (NULL == (n = hash[buc]))
360: return(ROFF_MAX);
361: for ( ; n; n = n->next)
1.16 schwarze 362: if (0 == strncmp(n->name, p, s) && '\0' == n->name[(int)s])
1.3 schwarze 363: return((enum rofft)(n - roffs));
1.1 schwarze 364:
365: return(ROFF_MAX);
366: }
367:
368: /*
369: * Pop the current node off of the stack of roff instructions currently
370: * pending.
371: */
372: static void
373: roffnode_pop(struct roff *r)
374: {
375: struct roffnode *p;
376:
1.2 schwarze 377: assert(r->last);
1.80 schwarze 378: p = r->last;
1.2 schwarze 379:
380: r->last = r->last->parent;
1.16 schwarze 381: free(p->name);
382: free(p->end);
1.1 schwarze 383: free(p);
384: }
385:
386: /*
387: * Push a roff node onto the instruction stack. This must later be
388: * removed with roffnode_pop().
389: */
1.11 schwarze 390: static void
1.16 schwarze 391: roffnode_push(struct roff *r, enum rofft tok, const char *name,
392: int line, int col)
1.1 schwarze 393: {
394: struct roffnode *p;
395:
1.11 schwarze 396: p = mandoc_calloc(1, sizeof(struct roffnode));
1.1 schwarze 397: p->tok = tok;
1.16 schwarze 398: if (name)
399: p->name = mandoc_strdup(name);
1.1 schwarze 400: p->parent = r->last;
401: p->line = line;
402: p->col = col;
1.71 schwarze 403: p->rule = p->parent ? p->parent->rule : 0;
1.1 schwarze 404:
405: r->last = p;
406: }
407:
408: static void
409: roff_free1(struct roff *r)
410: {
1.49 schwarze 411: struct tbl_node *tbl;
1.32 schwarze 412: struct eqn_node *e;
1.42 schwarze 413: int i;
1.27 schwarze 414:
1.49 schwarze 415: while (NULL != (tbl = r->first_tbl)) {
416: r->first_tbl = tbl->next;
417: tbl_free(tbl);
1.27 schwarze 418: }
419:
420: r->first_tbl = r->last_tbl = r->tbl = NULL;
1.1 schwarze 421:
1.32 schwarze 422: while (NULL != (e = r->first_eqn)) {
423: r->first_eqn = e->next;
424: eqn_free(e);
425: }
426:
427: r->first_eqn = r->last_eqn = r->eqn = NULL;
428:
1.1 schwarze 429: while (r->last)
430: roffnode_pop(r);
1.27 schwarze 431:
1.42 schwarze 432: roff_freestr(r->strtab);
433: roff_freestr(r->xmbtab);
434:
435: r->strtab = r->xmbtab = NULL;
436:
1.52 schwarze 437: roff_freereg(r->regtab);
438:
439: r->regtab = NULL;
440:
1.42 schwarze 441: if (r->xtab)
442: for (i = 0; i < 128; i++)
443: free(r->xtab[i].p);
444:
445: free(r->xtab);
446: r->xtab = NULL;
1.1 schwarze 447: }
448:
449: void
450: roff_reset(struct roff *r)
451: {
452:
453: roff_free1(r);
1.48 schwarze 454: r->control = 0;
1.1 schwarze 455: }
456:
457: void
458: roff_free(struct roff *r)
459: {
460:
461: roff_free1(r);
462: free(r);
463: }
464:
465: struct roff *
1.72 schwarze 466: roff_alloc(struct mparse *parse, int options)
1.1 schwarze 467: {
468: struct roff *r;
469:
1.11 schwarze 470: r = mandoc_calloc(1, sizeof(struct roff));
1.35 schwarze 471: r->parse = parse;
1.72 schwarze 472: r->options = options;
1.2 schwarze 473: r->rstackpos = -1;
1.80 schwarze 474:
1.42 schwarze 475: roffhash_init();
1.37 schwarze 476:
1.1 schwarze 477: return(r);
478: }
479:
1.8 schwarze 480: /*
1.79 schwarze 481: * In the current line, expand escape sequences that tend to get
482: * used in numerical expressions and conditional requests.
483: * Also check the syntax of the remaining escape sequences.
1.8 schwarze 484: */
1.45 schwarze 485: static enum rofferr
1.37 schwarze 486: roff_res(struct roff *r, char **bufp, size_t *szp, int ln, int pos)
1.8 schwarze 487: {
1.81 schwarze 488: char ubuf[24]; /* buffer to print the number */
1.78 schwarze 489: const char *start; /* start of the string to process */
1.82 schwarze 490: char *stesc; /* start of an escape sequence ('\\') */
1.23 schwarze 491: const char *stnam; /* start of the name, after "[(*" */
492: const char *cp; /* end of the name, e.g. before ']' */
493: const char *res; /* the string to be substituted */
1.53 schwarze 494: char *nbuf; /* new buffer to copy bufp to */
495: size_t maxl; /* expected length of the escape name */
496: size_t naml; /* actual length of the escape name */
497: int expand_count; /* to avoid infinite loops */
1.79 schwarze 498: int npos; /* position in numeric expression */
1.91 schwarze 499: int arg_complete; /* argument not interrupted by eol */
1.79 schwarze 500: char term; /* character terminating the escape */
1.8 schwarze 501:
1.43 schwarze 502: expand_count = 0;
1.78 schwarze 503: start = *bufp + pos;
504: stesc = strchr(start, '\0') - 1;
505: while (stesc-- > start) {
1.43 schwarze 506:
1.78 schwarze 507: /* Search backwards for the next backslash. */
508:
509: if ('\\' != *stesc)
510: continue;
511:
512: /* If it is escaped, skip it. */
513:
514: for (cp = stesc - 1; cp >= start; cp--)
515: if ('\\' != *cp)
516: break;
517:
518: if (0 == (stesc - cp) % 2) {
1.82 schwarze 519: stesc = (char *)cp;
1.78 schwarze 520: continue;
521: }
1.23 schwarze 522:
1.79 schwarze 523: /* Decide whether to expand or to check only. */
1.23 schwarze 524:
1.79 schwarze 525: term = '\0';
1.78 schwarze 526: cp = stesc + 1;
1.53 schwarze 527: switch (*cp) {
1.80 schwarze 528: case '*':
1.53 schwarze 529: res = NULL;
530: break;
1.80 schwarze 531: case 'B':
1.79 schwarze 532: /* FALLTHROUGH */
1.80 schwarze 533: case 'w':
1.79 schwarze 534: term = cp[1];
535: /* FALLTHROUGH */
1.80 schwarze 536: case 'n':
1.53 schwarze 537: res = ubuf;
538: break;
539: default:
1.78 schwarze 540: if (ESCAPE_ERROR == mandoc_escape(&cp, NULL, NULL))
1.92 schwarze 541: mandoc_vmsg(MANDOCERR_ESC_BAD,
542: r->parse, ln, (int)(stesc - *bufp),
543: "%.*s", (int)(cp - stesc), stesc);
1.78 schwarze 544: continue;
1.42 schwarze 545: }
546:
1.78 schwarze 547: if (EXPAND_LIMIT < ++expand_count) {
548: mandoc_msg(MANDOCERR_ROFFLOOP, r->parse,
549: ln, (int)(stesc - *bufp), NULL);
550: return(ROFF_IGN);
551: }
1.23 schwarze 552:
553: /*
554: * The third character decides the length
1.53 schwarze 555: * of the name of the string or register.
1.23 schwarze 556: * Save a pointer to the name.
557: */
558:
1.79 schwarze 559: if ('\0' == term) {
560: switch (*++cp) {
1.80 schwarze 561: case '\0':
1.79 schwarze 562: maxl = 0;
563: break;
1.80 schwarze 564: case '(':
1.79 schwarze 565: cp++;
566: maxl = 2;
567: break;
1.80 schwarze 568: case '[':
1.79 schwarze 569: cp++;
570: term = ']';
571: maxl = 0;
572: break;
573: default:
574: maxl = 1;
575: break;
576: }
577: } else {
578: cp += 2;
1.8 schwarze 579: maxl = 0;
580: }
1.23 schwarze 581: stnam = cp;
1.8 schwarze 582:
1.23 schwarze 583: /* Advance to the end of the name. */
1.8 schwarze 584:
1.91 schwarze 585: arg_complete = 1;
1.53 schwarze 586: for (naml = 0; 0 == maxl || naml < maxl; naml++, cp++) {
1.42 schwarze 587: if ('\0' == *cp) {
1.92 schwarze 588: mandoc_msg(MANDOCERR_ESC_BAD, r->parse,
589: ln, (int)(stesc - *bufp), stesc);
1.91 schwarze 590: arg_complete = 0;
1.79 schwarze 591: break;
1.42 schwarze 592: }
1.79 schwarze 593: if (0 == maxl && *cp == term) {
594: cp++;
1.8 schwarze 595: break;
1.79 schwarze 596: }
1.8 schwarze 597: }
598:
1.23 schwarze 599: /*
600: * Retrieve the replacement string; if it is
601: * undefined, resume searching for escapes.
602: */
603:
1.79 schwarze 604: switch (stesc[1]) {
1.80 schwarze 605: case '*':
1.91 schwarze 606: if (arg_complete)
607: res = roff_getstrn(r, stnam, naml);
1.79 schwarze 608: break;
1.80 schwarze 609: case 'B':
1.79 schwarze 610: npos = 0;
1.91 schwarze 611: ubuf[0] = arg_complete &&
612: roff_evalnum(stnam, &npos, NULL, 0) &&
613: stnam + npos + 1 == cp ? '1' : '0';
1.79 schwarze 614: ubuf[1] = '\0';
615: break;
1.80 schwarze 616: case 'n':
1.91 schwarze 617: if (arg_complete)
618: (void)snprintf(ubuf, sizeof(ubuf), "%d",
619: roff_getregn(r, stnam, naml));
620: else
621: ubuf[0] = '\0';
1.79 schwarze 622: break;
1.80 schwarze 623: case 'w':
1.91 schwarze 624: /* use even incomplete args */
1.81 schwarze 625: (void)snprintf(ubuf, sizeof(ubuf), "%d",
1.79 schwarze 626: 24 * (int)naml);
627: break;
628: }
1.8 schwarze 629:
630: if (NULL == res) {
1.92 schwarze 631: mandoc_vmsg(MANDOCERR_STR_UNDEF,
632: r->parse, ln, (int)(stesc - *bufp),
633: "%.*s", (int)naml, stnam);
1.37 schwarze 634: res = "";
1.8 schwarze 635: }
636:
1.23 schwarze 637: /* Replace the escape sequence by the string. */
638:
1.82 schwarze 639: *stesc = '\0';
640: *szp = mandoc_asprintf(&nbuf, "%s%s%s",
641: *bufp, res, cp) + 1;
1.8 schwarze 642:
1.78 schwarze 643: /* Prepare for the next replacement. */
1.8 schwarze 644:
1.78 schwarze 645: start = nbuf + pos;
1.82 schwarze 646: stesc = nbuf + (stesc - *bufp) + strlen(res);
1.8 schwarze 647: free(*bufp);
1.53 schwarze 648: *bufp = nbuf;
1.42 schwarze 649: }
1.45 schwarze 650: return(ROFF_CONT);
1.42 schwarze 651: }
652:
653: /*
1.51 schwarze 654: * Process text streams:
655: * Convert all breakable hyphens into ASCII_HYPH.
656: * Decrement and spring input line trap.
1.42 schwarze 657: */
658: static enum rofferr
1.51 schwarze 659: roff_parsetext(char **bufp, size_t *szp, int pos, int *offs)
1.42 schwarze 660: {
661: size_t sz;
662: const char *start;
1.51 schwarze 663: char *p;
664: int isz;
1.42 schwarze 665: enum mandoc_esc esc;
666:
1.51 schwarze 667: start = p = *bufp + pos;
1.42 schwarze 668:
669: while ('\0' != *p) {
670: sz = strcspn(p, "-\\");
671: p += sz;
672:
673: if ('\0' == *p)
674: break;
675:
676: if ('\\' == *p) {
677: /* Skip over escapes. */
678: p++;
1.62 schwarze 679: esc = mandoc_escape((const char **)&p, NULL, NULL);
1.42 schwarze 680: if (ESCAPE_ERROR == esc)
681: break;
682: continue;
683: } else if (p == start) {
684: p++;
685: continue;
686: }
687:
1.44 schwarze 688: if (isalpha((unsigned char)p[-1]) &&
689: isalpha((unsigned char)p[1]))
1.42 schwarze 690: *p = ASCII_HYPH;
691: p++;
1.8 schwarze 692: }
693:
1.51 schwarze 694: /* Spring the input line trap. */
695: if (1 == roffit_lines) {
1.75 schwarze 696: isz = mandoc_asprintf(&p, "%s\n.%s", *bufp, roffit_macro);
1.51 schwarze 697: free(*bufp);
698: *bufp = p;
699: *szp = isz + 1;
700: *offs = 0;
701: free(roffit_macro);
702: roffit_lines = 0;
703: return(ROFF_REPARSE);
704: } else if (1 < roffit_lines)
705: --roffit_lines;
1.42 schwarze 706: return(ROFF_CONT);
1.8 schwarze 707: }
708:
1.1 schwarze 709: enum rofferr
1.80 schwarze 710: roff_parseln(struct roff *r, int ln, char **bufp,
1.6 schwarze 711: size_t *szp, int pos, int *offs)
1.1 schwarze 712: {
713: enum rofft t;
1.27 schwarze 714: enum rofferr e;
1.35 schwarze 715: int ppos, ctl;
1.1 schwarze 716:
1.2 schwarze 717: /*
1.8 schwarze 718: * Run the reserved-word filter only if we have some reserved
719: * words to fill in.
720: */
721:
1.45 schwarze 722: e = roff_res(r, bufp, szp, ln, pos);
723: if (ROFF_IGN == e)
724: return(e);
725: assert(ROFF_CONT == e);
1.8 schwarze 726:
1.35 schwarze 727: ppos = pos;
1.48 schwarze 728: ctl = roff_getcontrol(r, *bufp, &pos);
1.35 schwarze 729:
1.8 schwarze 730: /*
1.2 schwarze 731: * First, if a scope is open and we're not a macro, pass the
732: * text through the macro's filter. If a scope isn't open and
733: * we're not a macro, just let it through.
1.32 schwarze 734: * Finally, if there's an equation scope open, divert it into it
735: * no matter our state.
1.2 schwarze 736: */
737:
1.35 schwarze 738: if (r->last && ! ctl) {
1.2 schwarze 739: t = r->last->tok;
740: assert(roffs[t].text);
1.80 schwarze 741: e = (*roffs[t].text)(r, t, bufp, szp, ln, pos, pos, offs);
1.27 schwarze 742: assert(ROFF_IGN == e || ROFF_CONT == e);
1.32 schwarze 743: if (ROFF_CONT != e)
744: return(e);
1.54 schwarze 745: }
746: if (r->eqn)
747: return(eqn_read(&r->eqn, ln, *bufp, ppos, offs));
748: if ( ! ctl) {
1.32 schwarze 749: if (r->tbl)
1.35 schwarze 750: return(tbl_read(r->tbl, ln, *bufp, pos));
1.51 schwarze 751: return(roff_parsetext(bufp, szp, pos, offs));
1.54 schwarze 752: }
1.2 schwarze 753:
754: /*
755: * If a scope is open, go to the child handler for that macro,
756: * as it may want to preprocess before doing anything with it.
1.32 schwarze 757: * Don't do so if an equation is open.
1.2 schwarze 758: */
759:
760: if (r->last) {
1.1 schwarze 761: t = r->last->tok;
762: assert(roffs[t].sub);
1.80 schwarze 763: return((*roffs[t].sub)(r, t, bufp, szp,
764: ln, ppos, pos, offs));
1.2 schwarze 765: }
766:
767: /*
768: * Lastly, as we've no scope open, try to look up and execute
769: * the new macro. If no macro is found, simply return and let
770: * the compilers handle it.
771: */
772:
1.87 schwarze 773: if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos, ln, ppos)))
1.1 schwarze 774: return(ROFF_CONT);
775:
1.2 schwarze 776: assert(roffs[t].proc);
1.80 schwarze 777: return((*roffs[t].proc)(r, t, bufp, szp, ln, ppos, pos, offs));
1.2 schwarze 778: }
779:
1.27 schwarze 780: void
1.2 schwarze 781: roff_endparse(struct roff *r)
782: {
1.1 schwarze 783:
1.27 schwarze 784: if (r->last)
1.94 ! schwarze 785: mandoc_msg(MANDOCERR_BLK_NOEND, r->parse,
! 786: r->last->line, r->last->col,
! 787: roffs[r->last->tok].name);
1.27 schwarze 788:
1.32 schwarze 789: if (r->eqn) {
1.94 ! schwarze 790: mandoc_msg(MANDOCERR_BLK_NOEND, r->parse,
! 791: r->eqn->eqn.ln, r->eqn->eqn.pos, "EQ");
1.41 schwarze 792: eqn_end(&r->eqn);
1.32 schwarze 793: }
794:
1.27 schwarze 795: if (r->tbl) {
1.94 ! schwarze 796: mandoc_msg(MANDOCERR_BLK_NOEND, r->parse,
! 797: r->tbl->line, r->tbl->pos, "TS");
1.41 schwarze 798: tbl_end(&r->tbl);
1.27 schwarze 799: }
1.1 schwarze 800: }
801:
802: /*
803: * Parse a roff node's type from the input buffer. This must be in the
804: * form of ".foo xxx" in the usual way.
805: */
806: static enum rofft
1.87 schwarze 807: roff_parse(struct roff *r, char *buf, int *pos, int ln, int ppos)
1.1 schwarze 808: {
1.87 schwarze 809: char *cp;
1.16 schwarze 810: const char *mac;
811: size_t maclen;
1.1 schwarze 812: enum rofft t;
813:
1.87 schwarze 814: cp = buf + *pos;
815:
816: if ('\0' == *cp || '"' == *cp || '\t' == *cp || ' ' == *cp)
1.1 schwarze 817: return(ROFF_MAX);
818:
1.87 schwarze 819: mac = cp;
820: maclen = roff_getname(r, &cp, ln, ppos);
1.1 schwarze 821:
1.16 schwarze 822: t = (r->current_string = roff_getstrn(r, mac, maclen))
1.42 schwarze 823: ? ROFF_USERDEF : roffhash_find(mac, maclen);
1.1 schwarze 824:
1.87 schwarze 825: if (ROFF_MAX != t)
826: *pos = cp - buf;
1.1 schwarze 827:
828: return(t);
829: }
830:
831: static enum rofferr
1.2 schwarze 832: roff_cblock(ROFF_ARGS)
1.1 schwarze 833: {
834:
1.2 schwarze 835: /*
836: * A block-close `..' should only be invoked as a child of an
837: * ignore macro, otherwise raise a warning and just ignore it.
838: */
839:
840: if (NULL == r->last) {
1.94 ! schwarze 841: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
! 842: ln, ppos, "..");
1.2 schwarze 843: return(ROFF_IGN);
844: }
1.1 schwarze 845:
1.2 schwarze 846: switch (r->last->tok) {
1.80 schwarze 847: case ROFF_am:
1.93 schwarze 848: /* ROFF_am1 is remapped to ROFF_am in roff_block(). */
1.2 schwarze 849: /* FALLTHROUGH */
1.80 schwarze 850: case ROFF_ami:
1.2 schwarze 851: /* FALLTHROUGH */
1.80 schwarze 852: case ROFF_de:
1.23 schwarze 853: /* ROFF_de1 is remapped to ROFF_de in roff_block(). */
1.2 schwarze 854: /* FALLTHROUGH */
1.80 schwarze 855: case ROFF_dei:
1.2 schwarze 856: /* FALLTHROUGH */
1.80 schwarze 857: case ROFF_ig:
1.2 schwarze 858: break;
859: default:
1.94 ! schwarze 860: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
! 861: ln, ppos, "..");
1.1 schwarze 862: return(ROFF_IGN);
1.2 schwarze 863: }
864:
865: if ((*bufp)[pos])
1.90 schwarze 866: mandoc_vmsg(MANDOCERR_ARG_SKIP, r->parse, ln, pos,
867: ".. %s", *bufp + pos);
1.2 schwarze 868:
869: roffnode_pop(r);
870: roffnode_cleanscope(r);
871: return(ROFF_IGN);
872:
873: }
1.1 schwarze 874:
1.2 schwarze 875: static void
876: roffnode_cleanscope(struct roff *r)
877: {
1.1 schwarze 878:
1.2 schwarze 879: while (r->last) {
1.46 schwarze 880: if (--r->last->endspan != 0)
1.2 schwarze 881: break;
882: roffnode_pop(r);
883: }
884: }
1.1 schwarze 885:
1.68 schwarze 886: static void
887: roff_ccond(struct roff *r, int ln, int ppos)
1.2 schwarze 888: {
1.1 schwarze 889:
1.2 schwarze 890: if (NULL == r->last) {
1.94 ! schwarze 891: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
! 892: ln, ppos, "\\}");
1.68 schwarze 893: return;
1.2 schwarze 894: }
1.1 schwarze 895:
1.2 schwarze 896: switch (r->last->tok) {
1.80 schwarze 897: case ROFF_el:
1.2 schwarze 898: /* FALLTHROUGH */
1.80 schwarze 899: case ROFF_ie:
1.2 schwarze 900: /* FALLTHROUGH */
1.80 schwarze 901: case ROFF_if:
1.2 schwarze 902: break;
903: default:
1.94 ! schwarze 904: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
! 905: ln, ppos, "\\}");
1.68 schwarze 906: return;
1.2 schwarze 907: }
1.1 schwarze 908:
1.2 schwarze 909: if (r->last->endspan > -1) {
1.94 ! schwarze 910: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
! 911: ln, ppos, "\\}");
1.68 schwarze 912: return;
1.2 schwarze 913: }
914:
915: roffnode_pop(r);
916: roffnode_cleanscope(r);
1.68 schwarze 917: return;
1.1 schwarze 918: }
919:
920: static enum rofferr
1.2 schwarze 921: roff_block(ROFF_ARGS)
1.1 schwarze 922: {
1.93 schwarze 923: const char *name;
924: char *iname, *cp;
1.86 schwarze 925: size_t namesz;
1.16 schwarze 926:
1.93 schwarze 927: /* Ignore groff compatibility mode for now. */
1.2 schwarze 928:
1.93 schwarze 929: if (ROFF_de1 == tok)
930: tok = ROFF_de;
931: else if (ROFF_am1 == tok)
932: tok = ROFF_am;
933:
934: /* Parse the macro name argument. */
935:
936: cp = *bufp + pos;
937: if (ROFF_ig == tok) {
938: iname = NULL;
939: namesz = 0;
940: } else {
941: iname = cp;
942: namesz = roff_getname(r, &cp, ln, ppos);
943: iname[namesz] = '\0';
944: }
1.22 schwarze 945:
1.93 schwarze 946: /* Resolve the macro name argument if it is indirect. */
1.22 schwarze 947:
1.93 schwarze 948: if (namesz && (ROFF_dei == tok || ROFF_ami == tok)) {
949: if (NULL == (name = roff_getstrn(r, iname, namesz))) {
950: mandoc_vmsg(MANDOCERR_STR_UNDEF,
951: r->parse, ln, (int)(iname - *bufp),
952: "%.*s", (int)namesz, iname);
953: namesz = 0;
954: } else
955: namesz = strlen(name);
956: } else
957: name = iname;
1.22 schwarze 958:
1.93 schwarze 959: if (0 == namesz && ROFF_ig != tok) {
960: mandoc_msg(MANDOCERR_REQ_EMPTY, r->parse,
961: ln, ppos, roffs[tok].name);
962: return(ROFF_IGN);
963: }
1.2 schwarze 964:
1.16 schwarze 965: roffnode_push(r, tok, name, ln, ppos);
966:
967: /*
968: * At the beginning of a `de' macro, clear the existing string
969: * with the same name, if there is one. New content will be
1.66 schwarze 970: * appended from roff_block_text() in multiline mode.
1.16 schwarze 971: */
1.22 schwarze 972:
1.93 schwarze 973: if (ROFF_de == tok || ROFF_dei == tok)
1.86 schwarze 974: roff_setstrn(&r->strtab, name, namesz, "", 0, 0);
1.2 schwarze 975:
1.86 schwarze 976: if ('\0' == *cp)
1.2 schwarze 977: return(ROFF_IGN);
1.1 schwarze 978:
1.93 schwarze 979: /* Get the custom end marker. */
1.22 schwarze 980:
1.93 schwarze 981: iname = cp;
1.86 schwarze 982: namesz = roff_getname(r, &cp, ln, ppos);
1.93 schwarze 983:
984: /* Resolve the end marker if it is indirect. */
985:
986: if (namesz && (ROFF_dei == tok || ROFF_ami == tok)) {
987: if (NULL == (name = roff_getstrn(r, iname, namesz))) {
988: mandoc_vmsg(MANDOCERR_STR_UNDEF,
989: r->parse, ln, (int)(iname - *bufp),
990: "%.*s", (int)namesz, iname);
991: namesz = 0;
992: } else
993: namesz = strlen(name);
994: } else
995: name = iname;
996:
1.86 schwarze 997: if (namesz)
998: r->last->end = mandoc_strndup(name, namesz);
1.2 schwarze 999:
1.86 schwarze 1000: if ('\0' != *cp)
1.90 schwarze 1001: mandoc_vmsg(MANDOCERR_ARG_EXCESS, r->parse,
1002: ln, pos, ".%s ... %s", roffs[tok].name, cp);
1.1 schwarze 1003:
1004: return(ROFF_IGN);
1005: }
1006:
1007: static enum rofferr
1.2 schwarze 1008: roff_block_sub(ROFF_ARGS)
1.1 schwarze 1009: {
1.2 schwarze 1010: enum rofft t;
1011: int i, j;
1012:
1013: /*
1014: * First check whether a custom macro exists at this level. If
1015: * it does, then check against it. This is some of groff's
1016: * stranger behaviours. If we encountered a custom end-scope
1017: * tag and that tag also happens to be a "real" macro, then we
1018: * need to try interpreting it again as a real macro. If it's
1019: * not, then return ignore. Else continue.
1020: */
1021:
1022: if (r->last->end) {
1.35 schwarze 1023: for (i = pos, j = 0; r->last->end[j]; j++, i++)
1.2 schwarze 1024: if ((*bufp)[i] != r->last->end[j])
1025: break;
1.1 schwarze 1026:
1.80 schwarze 1027: if ('\0' == r->last->end[j] &&
1028: ('\0' == (*bufp)[i] ||
1029: ' ' == (*bufp)[i] ||
1030: '\t' == (*bufp)[i])) {
1.2 schwarze 1031: roffnode_pop(r);
1032: roffnode_cleanscope(r);
1.1 schwarze 1033:
1.35 schwarze 1034: while (' ' == (*bufp)[i] || '\t' == (*bufp)[i])
1035: i++;
1036:
1037: pos = i;
1.87 schwarze 1038: if (ROFF_MAX != roff_parse(r, *bufp, &pos, ln, ppos))
1.2 schwarze 1039: return(ROFF_RERUN);
1040: return(ROFF_IGN);
1041: }
1.1 schwarze 1042: }
1043:
1.2 schwarze 1044: /*
1045: * If we have no custom end-query or lookup failed, then try
1046: * pulling it out of the hashtable.
1047: */
1.1 schwarze 1048:
1.87 schwarze 1049: t = roff_parse(r, *bufp, &pos, ln, ppos);
1.1 schwarze 1050:
1.16 schwarze 1051: if (ROFF_cblock != t) {
1.93 schwarze 1052: if (ROFF_ig != tok)
1.66 schwarze 1053: roff_setstr(r, r->last->name, *bufp + ppos, 2);
1.1 schwarze 1054: return(ROFF_IGN);
1.16 schwarze 1055: }
1.1 schwarze 1056:
1.2 schwarze 1057: assert(roffs[t].proc);
1.80 schwarze 1058: return((*roffs[t].proc)(r, t, bufp, szp, ln, ppos, pos, offs));
1.2 schwarze 1059: }
1060:
1061: static enum rofferr
1062: roff_block_text(ROFF_ARGS)
1063: {
1064:
1.93 schwarze 1065: if (ROFF_ig != tok)
1.66 schwarze 1066: roff_setstr(r, r->last->name, *bufp + pos, 2);
1.16 schwarze 1067:
1.2 schwarze 1068: return(ROFF_IGN);
1069: }
1070:
1071: static enum rofferr
1072: roff_cond_sub(ROFF_ARGS)
1073: {
1074: enum rofft t;
1.37 schwarze 1075: char *ep;
1.71 schwarze 1076: int rr;
1.2 schwarze 1077:
1078: rr = r->last->rule;
1.37 schwarze 1079: roffnode_cleanscope(r);
1.87 schwarze 1080: t = roff_parse(r, *bufp, &pos, ln, ppos);
1.2 schwarze 1081:
1.37 schwarze 1082: /*
1.50 schwarze 1083: * Fully handle known macros when they are structurally
1084: * required or when the conditional evaluated to true.
1.5 schwarze 1085: */
1086:
1.50 schwarze 1087: if ((ROFF_MAX != t) &&
1.71 schwarze 1088: (rr || ROFFMAC_STRUCT & roffs[t].flags)) {
1.50 schwarze 1089: assert(roffs[t].proc);
1090: return((*roffs[t].proc)(r, t, bufp, szp,
1.80 schwarze 1091: ln, ppos, pos, offs));
1.50 schwarze 1092: }
1.39 schwarze 1093:
1.69 schwarze 1094: /*
1095: * If `\}' occurs on a macro line without a preceding macro,
1096: * drop the line completely.
1097: */
1098:
1099: ep = *bufp + pos;
1100: if ('\\' == ep[0] && '}' == ep[1])
1.71 schwarze 1101: rr = 0;
1.69 schwarze 1102:
1.50 schwarze 1103: /* Always check for the closing delimiter `\}'. */
1.39 schwarze 1104:
1.50 schwarze 1105: while (NULL != (ep = strchr(ep, '\\'))) {
1.70 schwarze 1106: if ('}' == *(++ep)) {
1107: *ep = '&';
1108: roff_ccond(r, ln, ep - *bufp - 1);
1109: }
1110: ++ep;
1.50 schwarze 1111: }
1.71 schwarze 1112: return(rr ? ROFF_CONT : ROFF_IGN);
1.2 schwarze 1113: }
1114:
1115: static enum rofferr
1116: roff_cond_text(ROFF_ARGS)
1117: {
1.37 schwarze 1118: char *ep;
1.71 schwarze 1119: int rr;
1.2 schwarze 1120:
1121: rr = r->last->rule;
1.37 schwarze 1122: roffnode_cleanscope(r);
1.1 schwarze 1123:
1.70 schwarze 1124: ep = *bufp + pos;
1125: while (NULL != (ep = strchr(ep, '\\'))) {
1126: if ('}' == *(++ep)) {
1127: *ep = '&';
1128: roff_ccond(r, ln, ep - *bufp - 1);
1129: }
1130: ++ep;
1.2 schwarze 1131: }
1.71 schwarze 1132: return(rr ? ROFF_CONT : ROFF_IGN);
1.2 schwarze 1133: }
1134:
1.77 schwarze 1135: /*
1136: * Parse a single signed integer number. Stop at the first non-digit.
1137: * If there is at least one digit, return success and advance the
1138: * parse point, else return failure and let the parse point unchanged.
1139: * Ignore overflows, treat them just like the C language.
1140: */
1.56 schwarze 1141: static int
1142: roff_getnum(const char *v, int *pos, int *res)
1143: {
1.79 schwarze 1144: int myres, n, p;
1145:
1146: if (NULL == res)
1147: res = &myres;
1.56 schwarze 1148:
1149: p = *pos;
1150: n = v[p] == '-';
1151: if (n)
1152: p++;
1153:
1154: for (*res = 0; isdigit((unsigned char)v[p]); p++)
1.77 schwarze 1155: *res = 10 * *res + v[p] - '0';
1.56 schwarze 1156: if (p == *pos + n)
1157: return 0;
1158:
1159: if (n)
1160: *res = -*res;
1161:
1162: *pos = p;
1163: return 1;
1164: }
1165:
1.71 schwarze 1166: /*
1167: * Evaluate a string comparison condition.
1168: * The first character is the delimiter.
1169: * Succeed if the string up to its second occurrence
1170: * matches the string up to its third occurence.
1171: * Advance the cursor after the third occurrence
1172: * or lacking that, to the end of the line.
1173: */
1174: static int
1175: roff_evalstrcond(const char *v, int *pos)
1176: {
1177: const char *s1, *s2, *s3;
1178: int match;
1179:
1180: match = 0;
1181: s1 = v + *pos; /* initial delimiter */
1182: s2 = s1 + 1; /* for scanning the first string */
1183: s3 = strchr(s2, *s1); /* for scanning the second string */
1184:
1185: if (NULL == s3) /* found no middle delimiter */
1186: goto out;
1187:
1188: while ('\0' != *++s3) {
1189: if (*s2 != *s3) { /* mismatch */
1190: s3 = strchr(s3, *s1);
1191: break;
1192: }
1193: if (*s3 == *s1) { /* found the final delimiter */
1194: match = 1;
1195: break;
1196: }
1197: s2++;
1198: }
1199:
1200: out:
1201: if (NULL == s3)
1202: s3 = strchr(s2, '\0');
1203: else
1204: s3++;
1205: *pos = s3 - v;
1206: return(match);
1207: }
1208:
1.77 schwarze 1209: /*
1210: * Evaluate an optionally negated single character, numerical,
1211: * or string condition.
1212: */
1.71 schwarze 1213: static int
1.5 schwarze 1214: roff_evalcond(const char *v, int *pos)
1215: {
1.77 schwarze 1216: int wanttrue, number;
1.5 schwarze 1217:
1.71 schwarze 1218: if ('!' == v[*pos]) {
1219: wanttrue = 0;
1220: (*pos)++;
1221: } else
1222: wanttrue = 1;
1223:
1.5 schwarze 1224: switch (v[*pos]) {
1.80 schwarze 1225: case 'n':
1.71 schwarze 1226: /* FALLTHROUGH */
1.80 schwarze 1227: case 'o':
1.5 schwarze 1228: (*pos)++;
1.71 schwarze 1229: return(wanttrue);
1.80 schwarze 1230: case 'c':
1.71 schwarze 1231: /* FALLTHROUGH */
1.80 schwarze 1232: case 'd':
1.71 schwarze 1233: /* FALLTHROUGH */
1.80 schwarze 1234: case 'e':
1.5 schwarze 1235: /* FALLTHROUGH */
1.80 schwarze 1236: case 'r':
1.5 schwarze 1237: /* FALLTHROUGH */
1.80 schwarze 1238: case 't':
1.5 schwarze 1239: (*pos)++;
1.71 schwarze 1240: return(!wanttrue);
1.5 schwarze 1241: default:
1242: break;
1243: }
1244:
1.77 schwarze 1245: if (roff_evalnum(v, pos, &number, 0))
1246: return((number > 0) == wanttrue);
1247: else
1.71 schwarze 1248: return(roff_evalstrcond(v, pos) == wanttrue);
1.5 schwarze 1249: }
1250:
1.2 schwarze 1251: static enum rofferr
1.21 schwarze 1252: roff_line_ignore(ROFF_ARGS)
1.6 schwarze 1253: {
1.30 schwarze 1254:
1.21 schwarze 1255: return(ROFF_IGN);
1256: }
1257:
1258: static enum rofferr
1.2 schwarze 1259: roff_cond(ROFF_ARGS)
1260: {
1.46 schwarze 1261:
1262: roffnode_push(r, tok, NULL, ln, ppos);
1.2 schwarze 1263:
1.80 schwarze 1264: /*
1.35 schwarze 1265: * An `.el' has no conditional body: it will consume the value
1266: * of the current rstack entry set in prior `ie' calls or
1.80 schwarze 1267: * defaults to DENY.
1.35 schwarze 1268: *
1269: * If we're not an `el', however, then evaluate the conditional.
1270: */
1.1 schwarze 1271:
1.46 schwarze 1272: r->last->rule = ROFF_el == tok ?
1.80 schwarze 1273: (r->rstackpos < 0 ? 0 : r->rstack[r->rstackpos--]) :
1274: roff_evalcond(*bufp, &pos);
1.2 schwarze 1275:
1.35 schwarze 1276: /*
1277: * An if-else will put the NEGATION of the current evaluated
1278: * conditional into the stack of rules.
1279: */
1280:
1.2 schwarze 1281: if (ROFF_ie == tok) {
1.35 schwarze 1282: if (r->rstackpos == RSTACK_MAX - 1) {
1.80 schwarze 1283: mandoc_msg(MANDOCERR_MEM,
1284: r->parse, ln, ppos, NULL);
1.35 schwarze 1285: return(ROFF_ERR);
1286: }
1.71 schwarze 1287: r->rstack[++r->rstackpos] = !r->last->rule;
1.2 schwarze 1288: }
1.5 schwarze 1289:
1290: /* If the parent has false as its rule, then so do we. */
1291:
1.71 schwarze 1292: if (r->last->parent && !r->last->parent->rule)
1293: r->last->rule = 0;
1.5 schwarze 1294:
1295: /*
1.46 schwarze 1296: * Determine scope.
1297: * If there is nothing on the line after the conditional,
1298: * not even whitespace, use next-line scope.
1.5 schwarze 1299: */
1.2 schwarze 1300:
1.46 schwarze 1301: if ('\0' == (*bufp)[pos]) {
1302: r->last->endspan = 2;
1303: goto out;
1304: }
1305:
1306: while (' ' == (*bufp)[pos])
1307: pos++;
1308:
1309: /* An opening brace requests multiline scope. */
1.2 schwarze 1310:
1311: if ('\\' == (*bufp)[pos] && '{' == (*bufp)[pos + 1]) {
1312: r->last->endspan = -1;
1313: pos += 2;
1.46 schwarze 1314: goto out;
1.80 schwarze 1315: }
1.2 schwarze 1316:
1317: /*
1.46 schwarze 1318: * Anything else following the conditional causes
1319: * single-line scope. Warn if the scope contains
1320: * nothing but trailing whitespace.
1.2 schwarze 1321: */
1322:
1323: if ('\0' == (*bufp)[pos])
1.89 schwarze 1324: mandoc_msg(MANDOCERR_COND_EMPTY, r->parse,
1325: ln, ppos, roffs[tok].name);
1.2 schwarze 1326:
1.46 schwarze 1327: r->last->endspan = 1;
1.1 schwarze 1328:
1.46 schwarze 1329: out:
1.2 schwarze 1330: *offs = pos;
1331: return(ROFF_RERUN);
1.1 schwarze 1332: }
1333:
1.2 schwarze 1334: static enum rofferr
1.7 schwarze 1335: roff_ds(ROFF_ARGS)
1336: {
1.85 schwarze 1337: char *string;
1338: const char *name;
1339: size_t namesz;
1.10 schwarze 1340:
1341: /*
1.85 schwarze 1342: * The first word is the name of the string.
1343: * If it is empty or terminated by an escape sequence,
1344: * abort the `ds' request without defining anything.
1.10 schwarze 1345: */
1.7 schwarze 1346:
1.85 schwarze 1347: name = string = *bufp + pos;
1.7 schwarze 1348: if ('\0' == *name)
1349: return(ROFF_IGN);
1350:
1.85 schwarze 1351: namesz = roff_getname(r, &string, ln, pos);
1352: if ('\\' == name[namesz])
1353: return(ROFF_IGN);
1354:
1355: /* Read past the initial double-quote, if any. */
1.28 schwarze 1356: if ('"' == *string)
1.7 schwarze 1357: string++;
1358:
1.10 schwarze 1359: /* The rest is the value. */
1.85 schwarze 1360: roff_setstrn(&r->strtab, name, namesz, string, strlen(string),
1361: ROFF_as == tok);
1.7 schwarze 1362: return(ROFF_IGN);
1363: }
1364:
1.77 schwarze 1365: /*
1366: * Parse a single operator, one or two characters long.
1367: * If the operator is recognized, return success and advance the
1368: * parse point, else return failure and let the parse point unchanged.
1369: */
1370: static int
1371: roff_getop(const char *v, int *pos, char *res)
1372: {
1373:
1374: *res = v[*pos];
1375:
1376: switch (*res) {
1.80 schwarze 1377: case '+':
1.77 schwarze 1378: /* FALLTHROUGH */
1.80 schwarze 1379: case '-':
1.77 schwarze 1380: /* FALLTHROUGH */
1.80 schwarze 1381: case '*':
1.77 schwarze 1382: /* FALLTHROUGH */
1.80 schwarze 1383: case '/':
1.77 schwarze 1384: /* FALLTHROUGH */
1.80 schwarze 1385: case '%':
1.77 schwarze 1386: /* FALLTHROUGH */
1.80 schwarze 1387: case '&':
1.77 schwarze 1388: /* FALLTHROUGH */
1.80 schwarze 1389: case ':':
1.77 schwarze 1390: break;
1391: case '<':
1392: switch (v[*pos + 1]) {
1.80 schwarze 1393: case '=':
1.77 schwarze 1394: *res = 'l';
1395: (*pos)++;
1396: break;
1.80 schwarze 1397: case '>':
1.77 schwarze 1398: *res = '!';
1399: (*pos)++;
1400: break;
1.80 schwarze 1401: case '?':
1.77 schwarze 1402: *res = 'i';
1403: (*pos)++;
1404: break;
1405: default:
1406: break;
1407: }
1408: break;
1409: case '>':
1410: switch (v[*pos + 1]) {
1.80 schwarze 1411: case '=':
1.77 schwarze 1412: *res = 'g';
1413: (*pos)++;
1414: break;
1.80 schwarze 1415: case '?':
1.77 schwarze 1416: *res = 'a';
1417: (*pos)++;
1418: break;
1419: default:
1420: break;
1421: }
1422: break;
1423: case '=':
1424: if ('=' == v[*pos + 1])
1425: (*pos)++;
1426: break;
1427: default:
1428: return(0);
1429: }
1430: (*pos)++;
1431:
1432: return(*res);
1433: }
1434:
1435: /*
1436: * Evaluate either a parenthesized numeric expression
1437: * or a single signed integer number.
1438: */
1439: static int
1440: roff_evalpar(const char *v, int *pos, int *res)
1441: {
1442:
1443: if ('(' != v[*pos])
1444: return(roff_getnum(v, pos, res));
1445:
1446: (*pos)++;
1447: if ( ! roff_evalnum(v, pos, res, 1))
1448: return(0);
1449:
1.79 schwarze 1450: /*
1451: * Omission of the closing parenthesis
1452: * is an error in validation mode,
1453: * but ignored in evaluation mode.
1454: */
1455:
1.77 schwarze 1456: if (')' == v[*pos])
1457: (*pos)++;
1.79 schwarze 1458: else if (NULL == res)
1459: return(0);
1.77 schwarze 1460:
1461: return(1);
1462: }
1463:
1464: /*
1465: * Evaluate a complete numeric expression.
1466: * Proceed left to right, there is no concept of precedence.
1467: */
1468: static int
1469: roff_evalnum(const char *v, int *pos, int *res, int skipwhite)
1470: {
1471: int mypos, operand2;
1472: char operator;
1473:
1474: if (NULL == pos) {
1475: mypos = 0;
1476: pos = &mypos;
1477: }
1478:
1479: if (skipwhite)
1480: while (isspace((unsigned char)v[*pos]))
1481: (*pos)++;
1482:
1483: if ( ! roff_evalpar(v, pos, res))
1484: return(0);
1485:
1486: while (1) {
1487: if (skipwhite)
1488: while (isspace((unsigned char)v[*pos]))
1489: (*pos)++;
1490:
1491: if ( ! roff_getop(v, pos, &operator))
1492: break;
1493:
1494: if (skipwhite)
1495: while (isspace((unsigned char)v[*pos]))
1496: (*pos)++;
1497:
1498: if ( ! roff_evalpar(v, pos, &operand2))
1499: return(0);
1500:
1501: if (skipwhite)
1502: while (isspace((unsigned char)v[*pos]))
1503: (*pos)++;
1.79 schwarze 1504:
1505: if (NULL == res)
1506: continue;
1.77 schwarze 1507:
1508: switch (operator) {
1.80 schwarze 1509: case '+':
1.77 schwarze 1510: *res += operand2;
1511: break;
1.80 schwarze 1512: case '-':
1.77 schwarze 1513: *res -= operand2;
1514: break;
1.80 schwarze 1515: case '*':
1.77 schwarze 1516: *res *= operand2;
1517: break;
1.80 schwarze 1518: case '/':
1.77 schwarze 1519: *res /= operand2;
1520: break;
1.80 schwarze 1521: case '%':
1.77 schwarze 1522: *res %= operand2;
1523: break;
1.80 schwarze 1524: case '<':
1.77 schwarze 1525: *res = *res < operand2;
1526: break;
1.80 schwarze 1527: case '>':
1.77 schwarze 1528: *res = *res > operand2;
1529: break;
1.80 schwarze 1530: case 'l':
1.77 schwarze 1531: *res = *res <= operand2;
1532: break;
1.80 schwarze 1533: case 'g':
1.77 schwarze 1534: *res = *res >= operand2;
1535: break;
1.80 schwarze 1536: case '=':
1.77 schwarze 1537: *res = *res == operand2;
1538: break;
1.80 schwarze 1539: case '!':
1.77 schwarze 1540: *res = *res != operand2;
1541: break;
1.80 schwarze 1542: case '&':
1.77 schwarze 1543: *res = *res && operand2;
1544: break;
1.80 schwarze 1545: case ':':
1.77 schwarze 1546: *res = *res || operand2;
1547: break;
1.80 schwarze 1548: case 'i':
1.77 schwarze 1549: if (operand2 < *res)
1550: *res = operand2;
1551: break;
1.80 schwarze 1552: case 'a':
1.77 schwarze 1553: if (operand2 > *res)
1554: *res = operand2;
1555: break;
1556: default:
1557: abort();
1558: }
1559: }
1560: return(1);
1561: }
1562:
1.52 schwarze 1563: void
1.60 schwarze 1564: roff_setreg(struct roff *r, const char *name, int val, char sign)
1.41 schwarze 1565: {
1.52 schwarze 1566: struct roffreg *reg;
1567:
1568: /* Search for an existing register with the same name. */
1569: reg = r->regtab;
1570:
1571: while (reg && strcmp(name, reg->key.p))
1572: reg = reg->next;
1.41 schwarze 1573:
1.52 schwarze 1574: if (NULL == reg) {
1575: /* Create a new register. */
1576: reg = mandoc_malloc(sizeof(struct roffreg));
1577: reg->key.p = mandoc_strdup(name);
1578: reg->key.sz = strlen(name);
1.60 schwarze 1579: reg->val = 0;
1.52 schwarze 1580: reg->next = r->regtab;
1581: r->regtab = reg;
1582: }
1583:
1.60 schwarze 1584: if ('+' == sign)
1585: reg->val += val;
1586: else if ('-' == sign)
1587: reg->val -= val;
1588: else
1589: reg->val = val;
1.41 schwarze 1590: }
1591:
1.65 schwarze 1592: /*
1593: * Handle some predefined read-only number registers.
1594: * For now, return -1 if the requested register is not predefined;
1595: * in case a predefined read-only register having the value -1
1596: * were to turn up, another special value would have to be chosen.
1597: */
1598: static int
1599: roff_getregro(const char *name)
1600: {
1601:
1602: switch (*name) {
1.80 schwarze 1603: case 'A': /* ASCII approximation mode is always off. */
1.65 schwarze 1604: return(0);
1.80 schwarze 1605: case 'g': /* Groff compatibility mode is always on. */
1.65 schwarze 1606: return(1);
1.80 schwarze 1607: case 'H': /* Fixed horizontal resolution. */
1.65 schwarze 1608: return (24);
1.80 schwarze 1609: case 'j': /* Always adjust left margin only. */
1.65 schwarze 1610: return(0);
1.80 schwarze 1611: case 'T': /* Some output device is always defined. */
1.65 schwarze 1612: return(1);
1.80 schwarze 1613: case 'V': /* Fixed vertical resolution. */
1.65 schwarze 1614: return (40);
1615: default:
1616: return (-1);
1617: }
1618: }
1619:
1.53 schwarze 1620: int
1.52 schwarze 1621: roff_getreg(const struct roff *r, const char *name)
1.41 schwarze 1622: {
1.52 schwarze 1623: struct roffreg *reg;
1.65 schwarze 1624: int val;
1625:
1626: if ('.' == name[0] && '\0' != name[1] && '\0' == name[2]) {
1627: val = roff_getregro(name + 1);
1628: if (-1 != val)
1629: return (val);
1630: }
1.52 schwarze 1631:
1632: for (reg = r->regtab; reg; reg = reg->next)
1633: if (0 == strcmp(name, reg->key.p))
1.53 schwarze 1634: return(reg->val);
1635:
1636: return(0);
1637: }
1638:
1639: static int
1640: roff_getregn(const struct roff *r, const char *name, size_t len)
1641: {
1642: struct roffreg *reg;
1.65 schwarze 1643: int val;
1644:
1645: if ('.' == name[0] && 2 == len) {
1646: val = roff_getregro(name + 1);
1647: if (-1 != val)
1648: return (val);
1649: }
1.53 schwarze 1650:
1651: for (reg = r->regtab; reg; reg = reg->next)
1652: if (len == reg->key.sz &&
1653: 0 == strncmp(name, reg->key.p, len))
1654: return(reg->val);
1.41 schwarze 1655:
1.52 schwarze 1656: return(0);
1.41 schwarze 1657: }
1658:
1.52 schwarze 1659: static void
1660: roff_freereg(struct roffreg *reg)
1.41 schwarze 1661: {
1.52 schwarze 1662: struct roffreg *old_reg;
1.41 schwarze 1663:
1.52 schwarze 1664: while (NULL != reg) {
1665: free(reg->key.p);
1666: old_reg = reg;
1667: reg = reg->next;
1668: free(old_reg);
1669: }
1.41 schwarze 1670: }
1.7 schwarze 1671:
1672: static enum rofferr
1.6 schwarze 1673: roff_nr(ROFF_ARGS)
1.1 schwarze 1674: {
1.85 schwarze 1675: char *key, *val;
1676: size_t keysz;
1.37 schwarze 1677: int iv;
1.60 schwarze 1678: char sign;
1.6 schwarze 1679:
1.85 schwarze 1680: key = val = *bufp + pos;
1681: if ('\0' == *key)
1682: return(ROFF_IGN);
1683:
1684: keysz = roff_getname(r, &val, ln, pos);
1685: if ('\\' == key[keysz])
1686: return(ROFF_IGN);
1687: key[keysz] = '\0';
1.6 schwarze 1688:
1.60 schwarze 1689: sign = *val;
1690: if ('+' == sign || '-' == sign)
1691: val++;
1692:
1.77 schwarze 1693: if (roff_evalnum(val, NULL, &iv, 0))
1694: roff_setreg(r, key, iv, sign);
1.1 schwarze 1695:
1.76 schwarze 1696: return(ROFF_IGN);
1697: }
1698:
1699: static enum rofferr
1700: roff_rr(ROFF_ARGS)
1701: {
1702: struct roffreg *reg, **prev;
1.85 schwarze 1703: char *name, *cp;
1704: size_t namesz;
1.76 schwarze 1705:
1.85 schwarze 1706: name = cp = *bufp + pos;
1707: if ('\0' == *name)
1708: return(ROFF_IGN);
1709: namesz = roff_getname(r, &cp, ln, pos);
1710: name[namesz] = '\0';
1.76 schwarze 1711:
1712: prev = &r->regtab;
1713: while (1) {
1714: reg = *prev;
1715: if (NULL == reg || !strcmp(name, reg->key.p))
1716: break;
1717: prev = ®->next;
1718: }
1719: if (NULL != reg) {
1720: *prev = reg->next;
1721: free(reg->key.p);
1722: free(reg);
1723: }
1.29 schwarze 1724: return(ROFF_IGN);
1725: }
1726:
1727: static enum rofferr
1728: roff_rm(ROFF_ARGS)
1729: {
1730: const char *name;
1731: char *cp;
1.85 schwarze 1732: size_t namesz;
1.29 schwarze 1733:
1734: cp = *bufp + pos;
1735: while ('\0' != *cp) {
1.85 schwarze 1736: name = cp;
1737: namesz = roff_getname(r, &cp, ln, (int)(cp - *bufp));
1738: roff_setstrn(&r->strtab, name, namesz, NULL, 0, 0);
1739: if ('\\' == name[namesz])
1740: break;
1.29 schwarze 1741: }
1.51 schwarze 1742: return(ROFF_IGN);
1743: }
1744:
1745: static enum rofferr
1746: roff_it(ROFF_ARGS)
1747: {
1748: char *cp;
1749: size_t len;
1750: int iv;
1751:
1752: /* Parse the number of lines. */
1753: cp = *bufp + pos;
1754: len = strcspn(cp, " \t");
1755: cp[len] = '\0';
1756: if ((iv = mandoc_strntoi(cp, len, 10)) <= 0) {
1757: mandoc_msg(MANDOCERR_NUMERIC, r->parse,
1.80 schwarze 1758: ln, ppos, *bufp + 1);
1.51 schwarze 1759: return(ROFF_IGN);
1760: }
1761: cp += len + 1;
1762:
1763: /* Arm the input line trap. */
1764: roffit_lines = iv;
1765: roffit_macro = mandoc_strdup(cp);
1.2 schwarze 1766: return(ROFF_IGN);
1.47 schwarze 1767: }
1768:
1769: static enum rofferr
1770: roff_Dd(ROFF_ARGS)
1771: {
1772: const char *const *cp;
1773:
1.72 schwarze 1774: if (0 == ((MPARSE_MDOC | MPARSE_QUICK) & r->options))
1.47 schwarze 1775: for (cp = __mdoc_reserved; *cp; cp++)
1776: roff_setstr(r, *cp, NULL, 0);
1777:
1778: return(ROFF_CONT);
1779: }
1780:
1781: static enum rofferr
1782: roff_TH(ROFF_ARGS)
1783: {
1784: const char *const *cp;
1785:
1.72 schwarze 1786: if (0 == (MPARSE_QUICK & r->options))
1.47 schwarze 1787: for (cp = __man_reserved; *cp; cp++)
1788: roff_setstr(r, *cp, NULL, 0);
1789:
1790: return(ROFF_CONT);
1.14 schwarze 1791: }
1792:
1793: static enum rofferr
1.27 schwarze 1794: roff_TE(ROFF_ARGS)
1795: {
1796:
1797: if (NULL == r->tbl)
1.94 ! schwarze 1798: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
! 1799: ln, ppos, "TE");
1.27 schwarze 1800: else
1.41 schwarze 1801: tbl_end(&r->tbl);
1.27 schwarze 1802:
1803: return(ROFF_IGN);
1804: }
1805:
1806: static enum rofferr
1807: roff_T_(ROFF_ARGS)
1808: {
1809:
1810: if (NULL == r->tbl)
1.94 ! schwarze 1811: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
! 1812: ln, ppos, "T&");
1.27 schwarze 1813: else
1814: tbl_restart(ppos, ln, r->tbl);
1815:
1816: return(ROFF_IGN);
1817: }
1818:
1.41 schwarze 1819: #if 0
1820: static int
1821: roff_closeeqn(struct roff *r)
1822: {
1823:
1824: return(r->eqn && ROFF_EQN == eqn_end(&r->eqn) ? 1 : 0);
1825: }
1826: #endif
1827:
1828: static void
1.80 schwarze 1829: roff_openeqn(struct roff *r, const char *name, int line,
1.41 schwarze 1830: int offs, const char *buf)
1.32 schwarze 1831: {
1.41 schwarze 1832: struct eqn_node *e;
1833: int poff;
1.32 schwarze 1834:
1835: assert(NULL == r->eqn);
1.41 schwarze 1836: e = eqn_alloc(name, offs, line, r->parse);
1.32 schwarze 1837:
1838: if (r->last_eqn)
1839: r->last_eqn->next = e;
1840: else
1841: r->first_eqn = r->last_eqn = e;
1842:
1843: r->eqn = r->last_eqn = e;
1.41 schwarze 1844:
1845: if (buf) {
1846: poff = 0;
1847: eqn_read(&r->eqn, line, buf, offs, &poff);
1848: }
1849: }
1850:
1851: static enum rofferr
1852: roff_EQ(ROFF_ARGS)
1853: {
1854:
1855: roff_openeqn(r, *bufp + pos, ln, ppos, NULL);
1.32 schwarze 1856: return(ROFF_IGN);
1857: }
1858:
1859: static enum rofferr
1860: roff_EN(ROFF_ARGS)
1861: {
1862:
1.94 ! schwarze 1863: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse, ln, ppos, "EN");
1.32 schwarze 1864: return(ROFF_IGN);
1865: }
1866:
1867: static enum rofferr
1.27 schwarze 1868: roff_TS(ROFF_ARGS)
1869: {
1.49 schwarze 1870: struct tbl_node *tbl;
1.27 schwarze 1871:
1872: if (r->tbl) {
1.94 ! schwarze 1873: mandoc_msg(MANDOCERR_BLK_BROKEN, r->parse,
! 1874: ln, ppos, "TS breaks TS");
1.41 schwarze 1875: tbl_end(&r->tbl);
1.27 schwarze 1876: }
1877:
1.49 schwarze 1878: tbl = tbl_alloc(ppos, ln, r->parse);
1.27 schwarze 1879:
1880: if (r->last_tbl)
1.49 schwarze 1881: r->last_tbl->next = tbl;
1.27 schwarze 1882: else
1.49 schwarze 1883: r->first_tbl = r->last_tbl = tbl;
1.27 schwarze 1884:
1.49 schwarze 1885: r->tbl = r->last_tbl = tbl;
1.27 schwarze 1886: return(ROFF_IGN);
1887: }
1888:
1889: static enum rofferr
1.48 schwarze 1890: roff_cc(ROFF_ARGS)
1891: {
1892: const char *p;
1893:
1894: p = *bufp + pos;
1895:
1896: if ('\0' == *p || '.' == (r->control = *p++))
1897: r->control = 0;
1898:
1899: if ('\0' != *p)
1900: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
1901:
1902: return(ROFF_IGN);
1903: }
1904:
1905: static enum rofferr
1.42 schwarze 1906: roff_tr(ROFF_ARGS)
1907: {
1908: const char *p, *first, *second;
1909: size_t fsz, ssz;
1910: enum mandoc_esc esc;
1911:
1912: p = *bufp + pos;
1913:
1914: if ('\0' == *p) {
1915: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
1916: return(ROFF_IGN);
1917: }
1918:
1919: while ('\0' != *p) {
1920: fsz = ssz = 1;
1921:
1922: first = p++;
1923: if ('\\' == *first) {
1924: esc = mandoc_escape(&p, NULL, NULL);
1925: if (ESCAPE_ERROR == esc) {
1.92 schwarze 1926: mandoc_msg(MANDOCERR_ESC_BAD, r->parse,
1927: ln, (int)(p - *bufp), first);
1.42 schwarze 1928: return(ROFF_IGN);
1929: }
1930: fsz = (size_t)(p - first);
1931: }
1932:
1933: second = p++;
1934: if ('\\' == *second) {
1935: esc = mandoc_escape(&p, NULL, NULL);
1936: if (ESCAPE_ERROR == esc) {
1.92 schwarze 1937: mandoc_msg(MANDOCERR_ESC_BAD, r->parse,
1938: ln, (int)(p - *bufp), second);
1.42 schwarze 1939: return(ROFF_IGN);
1940: }
1941: ssz = (size_t)(p - second);
1942: } else if ('\0' == *second) {
1.80 schwarze 1943: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse,
1944: ln, (int)(p - *bufp), NULL);
1.42 schwarze 1945: second = " ";
1946: p--;
1947: }
1948:
1949: if (fsz > 1) {
1.80 schwarze 1950: roff_setstrn(&r->xmbtab, first, fsz,
1951: second, ssz, 0);
1.42 schwarze 1952: continue;
1953: }
1954:
1955: if (NULL == r->xtab)
1.80 schwarze 1956: r->xtab = mandoc_calloc(128,
1957: sizeof(struct roffstr));
1.42 schwarze 1958:
1959: free(r->xtab[(int)*first].p);
1960: r->xtab[(int)*first].p = mandoc_strndup(second, ssz);
1961: r->xtab[(int)*first].sz = ssz;
1962: }
1963:
1964: return(ROFF_IGN);
1965: }
1966:
1967: static enum rofferr
1.14 schwarze 1968: roff_so(ROFF_ARGS)
1969: {
1970: char *name;
1.15 schwarze 1971:
1.83 schwarze 1972: name = *bufp + pos;
1973: mandoc_vmsg(MANDOCERR_SO, r->parse, ln, ppos, ".so %s", name);
1.14 schwarze 1974:
1.22 schwarze 1975: /*
1976: * Handle `so'. Be EXTREMELY careful, as we shouldn't be
1977: * opening anything that's not in our cwd or anything beneath
1978: * it. Thus, explicitly disallow traversing up the file-system
1979: * or using absolute paths.
1980: */
1981:
1.14 schwarze 1982: if ('/' == *name || strstr(name, "../") || strstr(name, "/..")) {
1.83 schwarze 1983: mandoc_vmsg(MANDOCERR_SO_PATH, r->parse, ln, ppos,
1984: ".so %s", name);
1.14 schwarze 1985: return(ROFF_ERR);
1986: }
1987:
1988: *offs = pos;
1989: return(ROFF_SO);
1.7 schwarze 1990: }
1991:
1.16 schwarze 1992: static enum rofferr
1993: roff_userdef(ROFF_ARGS)
1.12 schwarze 1994: {
1.16 schwarze 1995: const char *arg[9];
1996: char *cp, *n1, *n2;
1.25 schwarze 1997: int i;
1.12 schwarze 1998:
1.16 schwarze 1999: /*
2000: * Collect pointers to macro argument strings
1.61 schwarze 2001: * and NUL-terminate them.
1.16 schwarze 2002: */
2003: cp = *bufp + pos;
1.25 schwarze 2004: for (i = 0; i < 9; i++)
1.26 schwarze 2005: arg[i] = '\0' == *cp ? "" :
1.35 schwarze 2006: mandoc_getarg(r->parse, &cp, ln, &pos);
1.16 schwarze 2007:
2008: /*
2009: * Expand macro arguments.
1.12 schwarze 2010: */
1.16 schwarze 2011: *szp = 0;
2012: n1 = cp = mandoc_strdup(r->current_string);
2013: while (NULL != (cp = strstr(cp, "\\$"))) {
2014: i = cp[2] - '1';
2015: if (0 > i || 8 < i) {
2016: /* Not an argument invocation. */
2017: cp += 2;
2018: continue;
2019: }
1.82 schwarze 2020: *cp = '\0';
2021: *szp = mandoc_asprintf(&n2, "%s%s%s",
2022: n1, arg[i], cp + 3) + 1;
1.16 schwarze 2023: cp = n2 + (cp - n1);
2024: free(n1);
2025: n1 = n2;
1.12 schwarze 2026: }
2027:
1.16 schwarze 2028: /*
2029: * Replace the macro invocation
2030: * by the expanded macro.
2031: */
2032: free(*bufp);
2033: *bufp = n1;
2034: if (0 == *szp)
2035: *szp = strlen(*bufp) + 1;
2036:
1.19 schwarze 2037: return(*szp > 1 && '\n' == (*bufp)[(int)*szp - 2] ?
1.16 schwarze 2038: ROFF_REPARSE : ROFF_APPEND);
1.12 schwarze 2039: }
1.28 schwarze 2040:
1.85 schwarze 2041: static size_t
1.28 schwarze 2042: roff_getname(struct roff *r, char **cpp, int ln, int pos)
2043: {
2044: char *name, *cp;
1.85 schwarze 2045: size_t namesz;
1.28 schwarze 2046:
2047: name = *cpp;
2048: if ('\0' == *name)
1.85 schwarze 2049: return(0);
1.28 schwarze 2050:
1.85 schwarze 2051: /* Read until end of name and terminate it with NUL. */
2052: for (cp = name; 1; cp++) {
2053: if ('\0' == *cp || ' ' == *cp) {
2054: namesz = cp - name;
2055: break;
2056: }
1.28 schwarze 2057: if ('\\' != *cp)
2058: continue;
1.88 schwarze 2059: namesz = cp - name;
2060: if ('{' == cp[1] || '}' == cp[1])
2061: break;
1.28 schwarze 2062: cp++;
2063: if ('\\' == *cp)
2064: continue;
1.35 schwarze 2065: mandoc_msg(MANDOCERR_NAMESC, r->parse, ln, pos, NULL);
1.85 schwarze 2066: mandoc_escape((const char **)&cp, NULL, NULL);
2067: break;
1.28 schwarze 2068: }
2069:
2070: /* Read past spaces. */
2071: while (' ' == *cp)
2072: cp++;
2073:
2074: *cpp = cp;
1.85 schwarze 2075: return(namesz);
1.28 schwarze 2076: }
2077:
1.16 schwarze 2078: /*
2079: * Store *string into the user-defined string called *name.
2080: * To clear an existing entry, call with (*r, *name, NULL, 0).
1.66 schwarze 2081: * append == 0: replace mode
2082: * append == 1: single-line append mode
2083: * append == 2: multiline append mode, append '\n' after each call
1.16 schwarze 2084: */
1.8 schwarze 2085: static void
1.16 schwarze 2086: roff_setstr(struct roff *r, const char *name, const char *string,
1.66 schwarze 2087: int append)
1.7 schwarze 2088: {
1.42 schwarze 2089:
2090: roff_setstrn(&r->strtab, name, strlen(name), string,
1.80 schwarze 2091: string ? strlen(string) : 0, append);
1.42 schwarze 2092: }
2093:
2094: static void
2095: roff_setstrn(struct roffkv **r, const char *name, size_t namesz,
1.66 schwarze 2096: const char *string, size_t stringsz, int append)
1.42 schwarze 2097: {
2098: struct roffkv *n;
2099: char *c;
2100: int i;
2101: size_t oldch, newch;
1.7 schwarze 2102:
1.16 schwarze 2103: /* Search for an existing string with the same name. */
1.42 schwarze 2104: n = *r;
2105:
1.84 schwarze 2106: while (n && (namesz != n->key.sz ||
2107: strncmp(n->key.p, name, namesz)))
1.7 schwarze 2108: n = n->next;
1.8 schwarze 2109:
2110: if (NULL == n) {
1.16 schwarze 2111: /* Create a new string table entry. */
1.42 schwarze 2112: n = mandoc_malloc(sizeof(struct roffkv));
2113: n->key.p = mandoc_strndup(name, namesz);
2114: n->key.sz = namesz;
2115: n->val.p = NULL;
2116: n->val.sz = 0;
2117: n->next = *r;
2118: *r = n;
1.66 schwarze 2119: } else if (0 == append) {
1.42 schwarze 2120: free(n->val.p);
2121: n->val.p = NULL;
2122: n->val.sz = 0;
1.16 schwarze 2123: }
2124:
2125: if (NULL == string)
2126: return;
2127:
2128: /*
2129: * One additional byte for the '\n' in multiline mode,
2130: * and one for the terminating '\0'.
2131: */
1.66 schwarze 2132: newch = stringsz + (1 < append ? 2u : 1u);
1.42 schwarze 2133:
2134: if (NULL == n->val.p) {
2135: n->val.p = mandoc_malloc(newch);
2136: *n->val.p = '\0';
1.16 schwarze 2137: oldch = 0;
2138: } else {
1.42 schwarze 2139: oldch = n->val.sz;
2140: n->val.p = mandoc_realloc(n->val.p, oldch + newch);
1.16 schwarze 2141: }
2142:
2143: /* Skip existing content in the destination buffer. */
1.42 schwarze 2144: c = n->val.p + (int)oldch;
1.16 schwarze 2145:
2146: /* Append new content to the destination buffer. */
1.42 schwarze 2147: i = 0;
2148: while (i < (int)stringsz) {
1.16 schwarze 2149: /*
2150: * Rudimentary roff copy mode:
2151: * Handle escaped backslashes.
2152: */
1.42 schwarze 2153: if ('\\' == string[i] && '\\' == string[i + 1])
2154: i++;
2155: *c++ = string[i++];
1.16 schwarze 2156: }
1.8 schwarze 2157:
1.16 schwarze 2158: /* Append terminating bytes. */
1.66 schwarze 2159: if (1 < append)
1.16 schwarze 2160: *c++ = '\n';
1.42 schwarze 2161:
1.16 schwarze 2162: *c = '\0';
1.42 schwarze 2163: n->val.sz = (int)(c - n->val.p);
1.7 schwarze 2164: }
2165:
1.8 schwarze 2166: static const char *
2167: roff_getstrn(const struct roff *r, const char *name, size_t len)
1.7 schwarze 2168: {
1.42 schwarze 2169: const struct roffkv *n;
1.64 schwarze 2170: int i;
1.7 schwarze 2171:
1.42 schwarze 2172: for (n = r->strtab; n; n = n->next)
1.80 schwarze 2173: if (0 == strncmp(name, n->key.p, len) &&
2174: '\0' == n->key.p[(int)len])
1.42 schwarze 2175: return(n->val.p);
1.64 schwarze 2176:
2177: for (i = 0; i < PREDEFS_MAX; i++)
2178: if (0 == strncmp(name, predefs[i].name, len) &&
2179: '\0' == predefs[i].name[(int)len])
2180: return(predefs[i].str);
1.8 schwarze 2181:
1.42 schwarze 2182: return(NULL);
1.7 schwarze 2183: }
2184:
1.8 schwarze 2185: static void
1.42 schwarze 2186: roff_freestr(struct roffkv *r)
1.7 schwarze 2187: {
1.42 schwarze 2188: struct roffkv *n, *nn;
1.7 schwarze 2189:
1.42 schwarze 2190: for (n = r; n; n = nn) {
2191: free(n->key.p);
2192: free(n->val.p);
1.7 schwarze 2193: nn = n->next;
2194: free(n);
2195: }
1.27 schwarze 2196: }
2197:
2198: const struct tbl_span *
2199: roff_span(const struct roff *r)
2200: {
1.80 schwarze 2201:
1.27 schwarze 2202: return(r->tbl ? tbl_span(r->tbl) : NULL);
1.32 schwarze 2203: }
2204:
2205: const struct eqn *
2206: roff_eqn(const struct roff *r)
2207: {
1.80 schwarze 2208:
1.32 schwarze 2209: return(r->last_eqn ? &r->last_eqn->eqn : NULL);
1.42 schwarze 2210: }
2211:
2212: /*
2213: * Duplicate an input string, making the appropriate character
2214: * conversations (as stipulated by `tr') along the way.
2215: * Returns a heap-allocated string with all the replacements made.
2216: */
2217: char *
2218: roff_strdup(const struct roff *r, const char *p)
2219: {
2220: const struct roffkv *cp;
2221: char *res;
2222: const char *pp;
2223: size_t ssz, sz;
2224: enum mandoc_esc esc;
2225:
2226: if (NULL == r->xmbtab && NULL == r->xtab)
2227: return(mandoc_strdup(p));
2228: else if ('\0' == *p)
2229: return(mandoc_strdup(""));
2230:
2231: /*
2232: * Step through each character looking for term matches
2233: * (remember that a `tr' can be invoked with an escape, which is
2234: * a glyph but the escape is multi-character).
2235: * We only do this if the character hash has been initialised
2236: * and the string is >0 length.
2237: */
2238:
2239: res = NULL;
2240: ssz = 0;
2241:
2242: while ('\0' != *p) {
2243: if ('\\' != *p && r->xtab && r->xtab[(int)*p].p) {
2244: sz = r->xtab[(int)*p].sz;
2245: res = mandoc_realloc(res, ssz + sz + 1);
2246: memcpy(res + ssz, r->xtab[(int)*p].p, sz);
2247: ssz += sz;
2248: p++;
2249: continue;
2250: } else if ('\\' != *p) {
2251: res = mandoc_realloc(res, ssz + 2);
2252: res[ssz++] = *p++;
2253: continue;
2254: }
2255:
2256: /* Search for term matches. */
2257: for (cp = r->xmbtab; cp; cp = cp->next)
2258: if (0 == strncmp(p, cp->key.p, cp->key.sz))
2259: break;
2260:
2261: if (NULL != cp) {
2262: /*
2263: * A match has been found.
2264: * Append the match to the array and move
2265: * forward by its keysize.
2266: */
1.80 schwarze 2267: res = mandoc_realloc(res,
2268: ssz + cp->val.sz + 1);
1.42 schwarze 2269: memcpy(res + ssz, cp->val.p, cp->val.sz);
2270: ssz += cp->val.sz;
2271: p += (int)cp->key.sz;
2272: continue;
2273: }
2274:
2275: /*
2276: * Handle escapes carefully: we need to copy
2277: * over just the escape itself, or else we might
2278: * do replacements within the escape itself.
2279: * Make sure to pass along the bogus string.
2280: */
2281: pp = p++;
2282: esc = mandoc_escape(&p, NULL, NULL);
2283: if (ESCAPE_ERROR == esc) {
2284: sz = strlen(pp);
2285: res = mandoc_realloc(res, ssz + sz + 1);
2286: memcpy(res + ssz, pp, sz);
2287: break;
2288: }
1.80 schwarze 2289: /*
2290: * We bail out on bad escapes.
1.42 schwarze 2291: * No need to warn: we already did so when
2292: * roff_res() was called.
2293: */
2294: sz = (int)(p - pp);
2295: res = mandoc_realloc(res, ssz + sz + 1);
2296: memcpy(res + ssz, pp, sz);
2297: ssz += sz;
2298: }
2299:
2300: res[(int)ssz] = '\0';
2301: return(res);
1.48 schwarze 2302: }
2303:
2304: /*
1.80 schwarze 2305: * Find out whether a line is a macro line or not.
1.48 schwarze 2306: * If it is, adjust the current position and return one; if it isn't,
2307: * return zero and don't change the current position.
2308: * If the control character has been set with `.cc', then let that grain
2309: * precedence.
2310: * This is slighly contrary to groff, where using the non-breaking
2311: * control character when `cc' has been invoked will cause the
2312: * non-breaking macro contents to be printed verbatim.
2313: */
2314: int
2315: roff_getcontrol(const struct roff *r, const char *cp, int *ppos)
2316: {
2317: int pos;
2318:
2319: pos = *ppos;
2320:
2321: if (0 != r->control && cp[pos] == r->control)
2322: pos++;
2323: else if (0 != r->control)
2324: return(0);
2325: else if ('\\' == cp[pos] && '.' == cp[pos + 1])
2326: pos += 2;
2327: else if ('.' == cp[pos] || '\'' == cp[pos])
2328: pos++;
2329: else
2330: return(0);
2331:
2332: while (' ' == cp[pos] || '\t' == cp[pos])
2333: pos++;
2334:
2335: *ppos = pos;
2336: return(1);
1.1 schwarze 2337: }
![[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