Annotation of src/usr.bin/mandoc/roff.c, Revision 1.100
1.100 ! schwarze 1: /* $OpenBSD: roff.c,v 1.99 2014/09/06 22:38:35 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: */
1.99 schwarze 18: #include <sys/types.h>
19:
1.1 schwarze 20: #include <assert.h>
1.3 schwarze 21: #include <ctype.h>
1.51 schwarze 22: #include <stdio.h>
1.1 schwarze 23: #include <stdlib.h>
24: #include <string.h>
25:
26: #include "mandoc.h"
1.74 schwarze 27: #include "mandoc_aux.h"
1.98 daniel 28: #include "libmandoc.h"
1.27 schwarze 29: #include "libroff.h"
1.1 schwarze 30:
1.37 schwarze 31: /* Maximum number of nested if-else conditionals. */
1.2 schwarze 32: #define RSTACK_MAX 128
33:
1.43 schwarze 34: /* Maximum number of string expansions per line, to break infinite loops. */
35: #define EXPAND_LIMIT 1000
36:
1.1 schwarze 37: enum rofft {
1.20 schwarze 38: ROFF_ad,
1.2 schwarze 39: ROFF_am,
40: ROFF_ami,
41: ROFF_am1,
1.66 schwarze 42: ROFF_as,
1.48 schwarze 43: ROFF_cc,
1.67 schwarze 44: ROFF_ce,
1.1 schwarze 45: ROFF_de,
46: ROFF_dei,
1.2 schwarze 47: ROFF_de1,
48: ROFF_ds,
49: ROFF_el,
1.58 schwarze 50: ROFF_fam,
1.59 schwarze 51: ROFF_hw,
1.20 schwarze 52: ROFF_hy,
1.2 schwarze 53: ROFF_ie,
54: ROFF_if,
1.1 schwarze 55: ROFF_ig,
1.30 schwarze 56: ROFF_it,
1.20 schwarze 57: ROFF_ne,
58: ROFF_nh,
1.14 schwarze 59: ROFF_nr,
1.31 schwarze 60: ROFF_ns,
61: ROFF_ps,
1.2 schwarze 62: ROFF_rm,
1.76 schwarze 63: ROFF_rr,
1.14 schwarze 64: ROFF_so,
1.31 schwarze 65: ROFF_ta,
1.2 schwarze 66: ROFF_tr,
1.47 schwarze 67: ROFF_Dd,
68: ROFF_TH,
1.27 schwarze 69: ROFF_TS,
70: ROFF_TE,
71: ROFF_T_,
1.32 schwarze 72: ROFF_EQ,
73: ROFF_EN,
1.2 schwarze 74: ROFF_cblock,
1.16 schwarze 75: ROFF_USERDEF,
1.1 schwarze 76: ROFF_MAX
77: };
78:
1.41 schwarze 79: /*
1.42 schwarze 80: * An incredibly-simple string buffer.
81: */
1.8 schwarze 82: struct roffstr {
1.42 schwarze 83: char *p; /* nil-terminated buffer */
84: size_t sz; /* saved strlen(p) */
85: };
86:
87: /*
88: * A key-value roffstr pair as part of a singly-linked list.
89: */
90: struct roffkv {
91: struct roffstr key;
92: struct roffstr val;
93: struct roffkv *next; /* next in list */
1.8 schwarze 94: };
95:
1.52 schwarze 96: /*
97: * A single number register as part of a singly-linked list.
98: */
99: struct roffreg {
100: struct roffstr key;
1.53 schwarze 101: int val;
1.52 schwarze 102: struct roffreg *next;
103: };
104:
1.1 schwarze 105: struct roff {
1.35 schwarze 106: struct mparse *parse; /* parse point */
1.1 schwarze 107: struct roffnode *last; /* leaf of stack */
1.96 schwarze 108: int *rstack; /* stack of inverted `ie' values */
1.52 schwarze 109: struct roffreg *regtab; /* number registers */
1.42 schwarze 110: struct roffkv *strtab; /* user-defined strings & macros */
111: struct roffkv *xmbtab; /* multi-byte trans table (`tr') */
112: struct roffstr *xtab; /* single-byte trans table (`tr') */
1.16 schwarze 113: const char *current_string; /* value of last called user macro */
1.27 schwarze 114: struct tbl_node *first_tbl; /* first table parsed */
115: struct tbl_node *last_tbl; /* last table parsed */
116: struct tbl_node *tbl; /* current table being parsed */
1.32 schwarze 117: struct eqn_node *last_eqn; /* last equation parsed */
118: struct eqn_node *first_eqn; /* first equation parsed */
119: struct eqn_node *eqn; /* current equation being parsed */
1.96 schwarze 120: int options; /* parse options */
121: int rstacksz; /* current size limit of rstack */
122: int rstackpos; /* position in rstack */
1.99 schwarze 123: int format; /* current file in mdoc or man format */
1.96 schwarze 124: char control; /* control character */
1.1 schwarze 125: };
126:
127: struct roffnode {
128: enum rofft tok; /* type of node */
129: struct roffnode *parent; /* up one in stack */
130: int line; /* parse line */
131: int col; /* parse col */
1.16 schwarze 132: char *name; /* node name, e.g. macro name */
1.2 schwarze 133: char *end; /* end-rules: custom token */
134: int endspan; /* end-rules: next-line or infty */
1.71 schwarze 135: int rule; /* current evaluation rule */
1.1 schwarze 136: };
137:
138: #define ROFF_ARGS struct roff *r, /* parse ctx */ \
139: enum rofft tok, /* tok of macro */ \
1.80 schwarze 140: char **bufp, /* input buffer */ \
1.1 schwarze 141: size_t *szp, /* size of input buffer */ \
142: int ln, /* parse line */ \
1.2 schwarze 143: int ppos, /* original pos in buffer */ \
144: int pos, /* current pos in buffer */ \
145: int *offs /* reset offset of buffer data */
1.1 schwarze 146:
147: typedef enum rofferr (*roffproc)(ROFF_ARGS);
148:
149: struct roffmac {
150: const char *name; /* macro name */
1.2 schwarze 151: roffproc proc; /* process new macro */
152: roffproc text; /* process as child text of macro */
153: roffproc sub; /* process as child of macro */
154: int flags;
155: #define ROFFMAC_STRUCT (1 << 0) /* always interpret */
1.3 schwarze 156: struct roffmac *next;
1.1 schwarze 157: };
158:
1.37 schwarze 159: struct predef {
160: const char *name; /* predefined input name */
161: const char *str; /* replacement symbol */
162: };
163:
164: #define PREDEF(__name, __str) \
165: { (__name), (__str) },
166:
1.42 schwarze 167: static enum rofft roffhash_find(const char *, size_t);
168: static void roffhash_init(void);
169: static void roffnode_cleanscope(struct roff *);
170: static void roffnode_pop(struct roff *);
171: static void roffnode_push(struct roff *, enum rofft,
172: const char *, int, int);
1.2 schwarze 173: static enum rofferr roff_block(ROFF_ARGS);
174: static enum rofferr roff_block_text(ROFF_ARGS);
175: static enum rofferr roff_block_sub(ROFF_ARGS);
176: static enum rofferr roff_cblock(ROFF_ARGS);
1.48 schwarze 177: static enum rofferr roff_cc(ROFF_ARGS);
1.68 schwarze 178: static void roff_ccond(struct roff *, int, int);
1.2 schwarze 179: static enum rofferr roff_cond(ROFF_ARGS);
180: static enum rofferr roff_cond_text(ROFF_ARGS);
181: static enum rofferr roff_cond_sub(ROFF_ARGS);
1.7 schwarze 182: static enum rofferr roff_ds(ROFF_ARGS);
1.71 schwarze 183: static int roff_evalcond(const char *, int *);
1.77 schwarze 184: static int roff_evalnum(const char *, int *, int *, int);
185: static int roff_evalpar(const char *, int *, int *);
1.71 schwarze 186: static int roff_evalstrcond(const char *, int *);
1.42 schwarze 187: static void roff_free1(struct roff *);
1.52 schwarze 188: static void roff_freereg(struct roffreg *);
1.42 schwarze 189: static void roff_freestr(struct roffkv *);
1.85 schwarze 190: static size_t roff_getname(struct roff *, char **, int, int);
1.56 schwarze 191: static int roff_getnum(const char *, int *, int *);
192: static int roff_getop(const char *, int *, char *);
1.53 schwarze 193: static int roff_getregn(const struct roff *,
194: const char *, size_t);
1.65 schwarze 195: static int roff_getregro(const char *name);
1.80 schwarze 196: static const char *roff_getstrn(const struct roff *,
1.8 schwarze 197: const char *, size_t);
1.51 schwarze 198: static enum rofferr roff_it(ROFF_ARGS);
1.21 schwarze 199: static enum rofferr roff_line_ignore(ROFF_ARGS);
1.6 schwarze 200: static enum rofferr roff_nr(ROFF_ARGS);
1.41 schwarze 201: static void roff_openeqn(struct roff *, const char *,
202: int, int, const char *);
1.87 schwarze 203: static enum rofft roff_parse(struct roff *, char *, int *,
204: int, int);
1.51 schwarze 205: static enum rofferr roff_parsetext(char **, size_t *, int, int *);
1.80 schwarze 206: static enum rofferr roff_res(struct roff *,
1.37 schwarze 207: char **, size_t *, int, int);
1.29 schwarze 208: static enum rofferr roff_rm(ROFF_ARGS);
1.76 schwarze 209: static enum rofferr roff_rr(ROFF_ARGS);
1.8 schwarze 210: static void roff_setstr(struct roff *,
1.16 schwarze 211: const char *, const char *, int);
1.80 schwarze 212: static void roff_setstrn(struct roffkv **, const char *,
1.42 schwarze 213: size_t, const char *, size_t, int);
1.14 schwarze 214: static enum rofferr roff_so(ROFF_ARGS);
1.42 schwarze 215: static enum rofferr roff_tr(ROFF_ARGS);
1.47 schwarze 216: static enum rofferr roff_Dd(ROFF_ARGS);
217: static enum rofferr roff_TH(ROFF_ARGS);
1.27 schwarze 218: static enum rofferr roff_TE(ROFF_ARGS);
219: static enum rofferr roff_TS(ROFF_ARGS);
1.32 schwarze 220: static enum rofferr roff_EQ(ROFF_ARGS);
221: static enum rofferr roff_EN(ROFF_ARGS);
1.27 schwarze 222: static enum rofferr roff_T_(ROFF_ARGS);
1.16 schwarze 223: static enum rofferr roff_userdef(ROFF_ARGS);
1.1 schwarze 224:
1.42 schwarze 225: /* See roffhash_find() */
1.3 schwarze 226:
227: #define ASCII_HI 126
228: #define ASCII_LO 33
229: #define HASHWIDTH (ASCII_HI - ASCII_LO + 1)
230:
231: static struct roffmac *hash[HASHWIDTH];
232:
233: static struct roffmac roffs[ROFF_MAX] = {
1.21 schwarze 234: { "ad", roff_line_ignore, NULL, NULL, 0, NULL },
1.3 schwarze 235: { "am", roff_block, roff_block_text, roff_block_sub, 0, NULL },
236: { "ami", roff_block, roff_block_text, roff_block_sub, 0, NULL },
237: { "am1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.66 schwarze 238: { "as", roff_ds, NULL, NULL, 0, NULL },
1.48 schwarze 239: { "cc", roff_cc, NULL, NULL, 0, NULL },
1.67 schwarze 240: { "ce", roff_line_ignore, NULL, NULL, 0, NULL },
1.3 schwarze 241: { "de", roff_block, roff_block_text, roff_block_sub, 0, NULL },
242: { "dei", roff_block, roff_block_text, roff_block_sub, 0, NULL },
243: { "de1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.7 schwarze 244: { "ds", roff_ds, NULL, NULL, 0, NULL },
1.3 schwarze 245: { "el", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
1.58 schwarze 246: { "fam", roff_line_ignore, NULL, NULL, 0, NULL },
1.59 schwarze 247: { "hw", roff_line_ignore, NULL, NULL, 0, NULL },
1.21 schwarze 248: { "hy", roff_line_ignore, NULL, NULL, 0, NULL },
1.3 schwarze 249: { "ie", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
250: { "if", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
251: { "ig", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.51 schwarze 252: { "it", roff_it, NULL, NULL, 0, NULL },
1.21 schwarze 253: { "ne", roff_line_ignore, NULL, NULL, 0, NULL },
254: { "nh", roff_line_ignore, NULL, NULL, 0, NULL },
1.14 schwarze 255: { "nr", roff_nr, NULL, NULL, 0, NULL },
1.31 schwarze 256: { "ns", roff_line_ignore, NULL, NULL, 0, NULL },
257: { "ps", roff_line_ignore, NULL, NULL, 0, NULL },
1.29 schwarze 258: { "rm", roff_rm, NULL, NULL, 0, NULL },
1.76 schwarze 259: { "rr", roff_rr, NULL, NULL, 0, NULL },
1.14 schwarze 260: { "so", roff_so, NULL, NULL, 0, NULL },
1.31 schwarze 261: { "ta", roff_line_ignore, NULL, NULL, 0, NULL },
1.42 schwarze 262: { "tr", roff_tr, NULL, NULL, 0, NULL },
1.47 schwarze 263: { "Dd", roff_Dd, NULL, NULL, 0, NULL },
264: { "TH", roff_TH, NULL, NULL, 0, NULL },
1.27 schwarze 265: { "TS", roff_TS, NULL, NULL, 0, NULL },
266: { "TE", roff_TE, NULL, NULL, 0, NULL },
267: { "T&", roff_T_, NULL, NULL, 0, NULL },
1.32 schwarze 268: { "EQ", roff_EQ, NULL, NULL, 0, NULL },
269: { "EN", roff_EN, NULL, NULL, 0, NULL },
1.3 schwarze 270: { ".", roff_cblock, NULL, NULL, 0, NULL },
1.16 schwarze 271: { NULL, roff_userdef, NULL, NULL, 0, NULL },
1.1 schwarze 272: };
273:
1.73 schwarze 274: /* not currently implemented: Ds em Eq LP Me PP pp Or Rd Sf SH */
1.47 schwarze 275: const char *const __mdoc_reserved[] = {
276: "Ac", "Ad", "An", "Ao", "Ap", "Aq", "Ar", "At",
277: "Bc", "Bd", "Bf", "Bk", "Bl", "Bo", "Bq",
278: "Brc", "Bro", "Brq", "Bsx", "Bt", "Bx",
279: "Cd", "Cm", "Db", "Dc", "Dd", "Dl", "Do", "Dq",
1.73 schwarze 280: "Dt", "Dv", "Dx", "D1",
281: "Ec", "Ed", "Ef", "Ek", "El", "Em",
282: "En", "Eo", "Er", "Es", "Ev", "Ex",
1.47 schwarze 283: "Fa", "Fc", "Fd", "Fl", "Fn", "Fo", "Fr", "Ft", "Fx",
1.73 schwarze 284: "Hf", "Ic", "In", "It", "Lb", "Li", "Lk", "Lp",
285: "Ms", "Mt", "Nd", "Nm", "No", "Ns", "Nx",
1.47 schwarze 286: "Oc", "Oo", "Op", "Os", "Ot", "Ox",
1.73 schwarze 287: "Pa", "Pc", "Pf", "Po", "Pp", "Pq",
288: "Qc", "Ql", "Qo", "Qq", "Re", "Rs", "Rv",
289: "Sc", "Sh", "Sm", "So", "Sq",
1.47 schwarze 290: "Ss", "St", "Sx", "Sy",
291: "Ta", "Tn", "Ud", "Ux", "Va", "Vt", "Xc", "Xo", "Xr",
1.73 schwarze 292: "%A", "%B", "%C", "%D", "%I", "%J", "%N", "%O",
1.47 schwarze 293: "%P", "%Q", "%R", "%T", "%U", "%V",
294: NULL
295: };
296:
1.73 schwarze 297: /* not currently implemented: BT DE DS ME MT PT SY TQ YS */
1.47 schwarze 298: const char *const __man_reserved[] = {
1.73 schwarze 299: "AT", "B", "BI", "BR", "DT",
300: "EE", "EN", "EQ", "EX", "HP", "I", "IB", "IP", "IR",
301: "LP", "OP", "P", "PD", "PP",
302: "R", "RB", "RE", "RI", "RS", "SB", "SH", "SM", "SS",
303: "TE", "TH", "TP", "TS", "T&", "UC", "UE", "UR",
1.47 schwarze 304: NULL
305: };
306:
1.37 schwarze 307: /* Array of injected predefined strings. */
308: #define PREDEFS_MAX 38
309: static const struct predef predefs[PREDEFS_MAX] = {
310: #include "predefs.in"
311: };
312:
1.42 schwarze 313: /* See roffhash_find() */
1.3 schwarze 314: #define ROFF_HASH(p) (p[0] - ASCII_LO)
315:
1.51 schwarze 316: static int roffit_lines; /* number of lines to delay */
317: static char *roffit_macro; /* nil-terminated macro line */
318:
1.80 schwarze 319:
1.3 schwarze 320: static void
1.42 schwarze 321: roffhash_init(void)
1.3 schwarze 322: {
323: struct roffmac *n;
324: int buc, i;
325:
1.16 schwarze 326: for (i = 0; i < (int)ROFF_USERDEF; i++) {
1.3 schwarze 327: assert(roffs[i].name[0] >= ASCII_LO);
328: assert(roffs[i].name[0] <= ASCII_HI);
329:
330: buc = ROFF_HASH(roffs[i].name);
331:
332: if (NULL != (n = hash[buc])) {
333: for ( ; n->next; n = n->next)
334: /* Do nothing. */ ;
335: n->next = &roffs[i];
336: } else
337: hash[buc] = &roffs[i];
338: }
339: }
340:
1.1 schwarze 341: /*
342: * Look up a roff token by its name. Returns ROFF_MAX if no macro by
343: * the nil-terminated string name could be found.
344: */
345: static enum rofft
1.42 schwarze 346: roffhash_find(const char *p, size_t s)
1.1 schwarze 347: {
1.3 schwarze 348: int buc;
349: struct roffmac *n;
1.1 schwarze 350:
1.3 schwarze 351: /*
352: * libroff has an extremely simple hashtable, for the time
353: * being, which simply keys on the first character, which must
354: * be printable, then walks a chain. It works well enough until
355: * optimised.
356: */
357:
358: if (p[0] < ASCII_LO || p[0] > ASCII_HI)
359: return(ROFF_MAX);
360:
361: buc = ROFF_HASH(p);
362:
363: if (NULL == (n = hash[buc]))
364: return(ROFF_MAX);
365: for ( ; n; n = n->next)
1.16 schwarze 366: if (0 == strncmp(n->name, p, s) && '\0' == n->name[(int)s])
1.3 schwarze 367: return((enum rofft)(n - roffs));
1.1 schwarze 368:
369: return(ROFF_MAX);
370: }
371:
372: /*
373: * Pop the current node off of the stack of roff instructions currently
374: * pending.
375: */
376: static void
377: roffnode_pop(struct roff *r)
378: {
379: struct roffnode *p;
380:
1.2 schwarze 381: assert(r->last);
1.80 schwarze 382: p = r->last;
1.2 schwarze 383:
384: r->last = r->last->parent;
1.16 schwarze 385: free(p->name);
386: free(p->end);
1.1 schwarze 387: free(p);
388: }
389:
390: /*
391: * Push a roff node onto the instruction stack. This must later be
392: * removed with roffnode_pop().
393: */
1.11 schwarze 394: static void
1.16 schwarze 395: roffnode_push(struct roff *r, enum rofft tok, const char *name,
396: int line, int col)
1.1 schwarze 397: {
398: struct roffnode *p;
399:
1.11 schwarze 400: p = mandoc_calloc(1, sizeof(struct roffnode));
1.1 schwarze 401: p->tok = tok;
1.16 schwarze 402: if (name)
403: p->name = mandoc_strdup(name);
1.1 schwarze 404: p->parent = r->last;
405: p->line = line;
406: p->col = col;
1.71 schwarze 407: p->rule = p->parent ? p->parent->rule : 0;
1.1 schwarze 408:
409: r->last = p;
410: }
411:
412: static void
413: roff_free1(struct roff *r)
414: {
1.49 schwarze 415: struct tbl_node *tbl;
1.32 schwarze 416: struct eqn_node *e;
1.42 schwarze 417: int i;
1.27 schwarze 418:
1.49 schwarze 419: while (NULL != (tbl = r->first_tbl)) {
420: r->first_tbl = tbl->next;
421: tbl_free(tbl);
1.27 schwarze 422: }
423: r->first_tbl = r->last_tbl = r->tbl = NULL;
1.1 schwarze 424:
1.32 schwarze 425: while (NULL != (e = r->first_eqn)) {
426: r->first_eqn = e->next;
427: eqn_free(e);
428: }
429: r->first_eqn = r->last_eqn = r->eqn = NULL;
430:
1.1 schwarze 431: while (r->last)
432: roffnode_pop(r);
1.27 schwarze 433:
1.96 schwarze 434: free (r->rstack);
435: r->rstack = NULL;
436: r->rstacksz = 0;
437: r->rstackpos = -1;
438:
439: roff_freereg(r->regtab);
440: r->regtab = NULL;
441:
1.42 schwarze 442: roff_freestr(r->strtab);
443: roff_freestr(r->xmbtab);
444: r->strtab = r->xmbtab = NULL;
445:
446: if (r->xtab)
447: for (i = 0; i < 128; i++)
448: free(r->xtab[i].p);
449: free(r->xtab);
450: r->xtab = NULL;
1.1 schwarze 451: }
452:
453: void
454: roff_reset(struct roff *r)
455: {
456:
457: roff_free1(r);
1.99 schwarze 458: r->format = r->options & (MPARSE_MDOC | MPARSE_MAN);
1.48 schwarze 459: r->control = 0;
1.1 schwarze 460: }
461:
462: void
463: roff_free(struct roff *r)
464: {
465:
466: roff_free1(r);
467: free(r);
468: }
469:
470: struct roff *
1.72 schwarze 471: roff_alloc(struct mparse *parse, int options)
1.1 schwarze 472: {
473: struct roff *r;
474:
1.11 schwarze 475: r = mandoc_calloc(1, sizeof(struct roff));
1.35 schwarze 476: r->parse = parse;
1.72 schwarze 477: r->options = options;
1.99 schwarze 478: r->format = options & (MPARSE_MDOC | MPARSE_MAN);
1.2 schwarze 479: r->rstackpos = -1;
1.80 schwarze 480:
1.42 schwarze 481: roffhash_init();
1.37 schwarze 482:
1.1 schwarze 483: return(r);
484: }
485:
1.8 schwarze 486: /*
1.79 schwarze 487: * In the current line, expand escape sequences that tend to get
488: * used in numerical expressions and conditional requests.
489: * Also check the syntax of the remaining escape sequences.
1.8 schwarze 490: */
1.45 schwarze 491: static enum rofferr
1.37 schwarze 492: roff_res(struct roff *r, char **bufp, size_t *szp, int ln, int pos)
1.8 schwarze 493: {
1.81 schwarze 494: char ubuf[24]; /* buffer to print the number */
1.78 schwarze 495: const char *start; /* start of the string to process */
1.82 schwarze 496: char *stesc; /* start of an escape sequence ('\\') */
1.23 schwarze 497: const char *stnam; /* start of the name, after "[(*" */
498: const char *cp; /* end of the name, e.g. before ']' */
499: const char *res; /* the string to be substituted */
1.53 schwarze 500: char *nbuf; /* new buffer to copy bufp to */
501: size_t maxl; /* expected length of the escape name */
502: size_t naml; /* actual length of the escape name */
503: int expand_count; /* to avoid infinite loops */
1.79 schwarze 504: int npos; /* position in numeric expression */
1.91 schwarze 505: int arg_complete; /* argument not interrupted by eol */
1.79 schwarze 506: char term; /* character terminating the escape */
1.8 schwarze 507:
1.43 schwarze 508: expand_count = 0;
1.78 schwarze 509: start = *bufp + pos;
510: stesc = strchr(start, '\0') - 1;
511: while (stesc-- > start) {
1.43 schwarze 512:
1.78 schwarze 513: /* Search backwards for the next backslash. */
514:
515: if ('\\' != *stesc)
516: continue;
517:
518: /* If it is escaped, skip it. */
519:
520: for (cp = stesc - 1; cp >= start; cp--)
521: if ('\\' != *cp)
522: break;
523:
524: if (0 == (stesc - cp) % 2) {
1.82 schwarze 525: stesc = (char *)cp;
1.78 schwarze 526: continue;
527: }
1.23 schwarze 528:
1.79 schwarze 529: /* Decide whether to expand or to check only. */
1.23 schwarze 530:
1.79 schwarze 531: term = '\0';
1.78 schwarze 532: cp = stesc + 1;
1.53 schwarze 533: switch (*cp) {
1.80 schwarze 534: case '*':
1.53 schwarze 535: res = NULL;
536: break;
1.80 schwarze 537: case 'B':
1.79 schwarze 538: /* FALLTHROUGH */
1.80 schwarze 539: case 'w':
1.79 schwarze 540: term = cp[1];
541: /* FALLTHROUGH */
1.80 schwarze 542: case 'n':
1.53 schwarze 543: res = ubuf;
544: break;
545: default:
1.78 schwarze 546: if (ESCAPE_ERROR == mandoc_escape(&cp, NULL, NULL))
1.92 schwarze 547: mandoc_vmsg(MANDOCERR_ESC_BAD,
548: r->parse, ln, (int)(stesc - *bufp),
549: "%.*s", (int)(cp - stesc), stesc);
1.78 schwarze 550: continue;
1.42 schwarze 551: }
552:
1.78 schwarze 553: if (EXPAND_LIMIT < ++expand_count) {
554: mandoc_msg(MANDOCERR_ROFFLOOP, r->parse,
555: ln, (int)(stesc - *bufp), NULL);
556: return(ROFF_IGN);
557: }
1.23 schwarze 558:
559: /*
560: * The third character decides the length
1.53 schwarze 561: * of the name of the string or register.
1.23 schwarze 562: * Save a pointer to the name.
563: */
564:
1.79 schwarze 565: if ('\0' == term) {
566: switch (*++cp) {
1.80 schwarze 567: case '\0':
1.79 schwarze 568: maxl = 0;
569: break;
1.80 schwarze 570: case '(':
1.79 schwarze 571: cp++;
572: maxl = 2;
573: break;
1.80 schwarze 574: case '[':
1.79 schwarze 575: cp++;
576: term = ']';
577: maxl = 0;
578: break;
579: default:
580: maxl = 1;
581: break;
582: }
583: } else {
584: cp += 2;
1.8 schwarze 585: maxl = 0;
586: }
1.23 schwarze 587: stnam = cp;
1.8 schwarze 588:
1.23 schwarze 589: /* Advance to the end of the name. */
1.8 schwarze 590:
1.91 schwarze 591: arg_complete = 1;
1.53 schwarze 592: for (naml = 0; 0 == maxl || naml < maxl; naml++, cp++) {
1.42 schwarze 593: if ('\0' == *cp) {
1.92 schwarze 594: mandoc_msg(MANDOCERR_ESC_BAD, r->parse,
595: ln, (int)(stesc - *bufp), stesc);
1.91 schwarze 596: arg_complete = 0;
1.79 schwarze 597: break;
1.42 schwarze 598: }
1.79 schwarze 599: if (0 == maxl && *cp == term) {
600: cp++;
1.8 schwarze 601: break;
1.79 schwarze 602: }
1.8 schwarze 603: }
604:
1.23 schwarze 605: /*
606: * Retrieve the replacement string; if it is
607: * undefined, resume searching for escapes.
608: */
609:
1.79 schwarze 610: switch (stesc[1]) {
1.80 schwarze 611: case '*':
1.91 schwarze 612: if (arg_complete)
613: res = roff_getstrn(r, stnam, naml);
1.79 schwarze 614: break;
1.80 schwarze 615: case 'B':
1.79 schwarze 616: npos = 0;
1.91 schwarze 617: ubuf[0] = arg_complete &&
618: roff_evalnum(stnam, &npos, NULL, 0) &&
619: stnam + npos + 1 == cp ? '1' : '0';
1.79 schwarze 620: ubuf[1] = '\0';
621: break;
1.80 schwarze 622: case 'n':
1.91 schwarze 623: if (arg_complete)
624: (void)snprintf(ubuf, sizeof(ubuf), "%d",
625: roff_getregn(r, stnam, naml));
626: else
627: ubuf[0] = '\0';
1.79 schwarze 628: break;
1.80 schwarze 629: case 'w':
1.91 schwarze 630: /* use even incomplete args */
1.81 schwarze 631: (void)snprintf(ubuf, sizeof(ubuf), "%d",
1.79 schwarze 632: 24 * (int)naml);
633: break;
634: }
1.8 schwarze 635:
636: if (NULL == res) {
1.92 schwarze 637: mandoc_vmsg(MANDOCERR_STR_UNDEF,
638: r->parse, ln, (int)(stesc - *bufp),
639: "%.*s", (int)naml, stnam);
1.37 schwarze 640: res = "";
1.8 schwarze 641: }
642:
1.23 schwarze 643: /* Replace the escape sequence by the string. */
644:
1.82 schwarze 645: *stesc = '\0';
646: *szp = mandoc_asprintf(&nbuf, "%s%s%s",
647: *bufp, res, cp) + 1;
1.8 schwarze 648:
1.78 schwarze 649: /* Prepare for the next replacement. */
1.8 schwarze 650:
1.78 schwarze 651: start = nbuf + pos;
1.82 schwarze 652: stesc = nbuf + (stesc - *bufp) + strlen(res);
1.8 schwarze 653: free(*bufp);
1.53 schwarze 654: *bufp = nbuf;
1.42 schwarze 655: }
1.45 schwarze 656: return(ROFF_CONT);
1.42 schwarze 657: }
658:
659: /*
1.51 schwarze 660: * Process text streams:
661: * Convert all breakable hyphens into ASCII_HYPH.
662: * Decrement and spring input line trap.
1.42 schwarze 663: */
664: static enum rofferr
1.51 schwarze 665: roff_parsetext(char **bufp, size_t *szp, int pos, int *offs)
1.42 schwarze 666: {
667: size_t sz;
668: const char *start;
1.51 schwarze 669: char *p;
670: int isz;
1.42 schwarze 671: enum mandoc_esc esc;
672:
1.51 schwarze 673: start = p = *bufp + pos;
1.42 schwarze 674:
675: while ('\0' != *p) {
676: sz = strcspn(p, "-\\");
677: p += sz;
678:
679: if ('\0' == *p)
680: break;
681:
682: if ('\\' == *p) {
683: /* Skip over escapes. */
684: p++;
1.62 schwarze 685: esc = mandoc_escape((const char **)&p, NULL, NULL);
1.42 schwarze 686: if (ESCAPE_ERROR == esc)
687: break;
688: continue;
689: } else if (p == start) {
690: p++;
691: continue;
692: }
693:
1.44 schwarze 694: if (isalpha((unsigned char)p[-1]) &&
695: isalpha((unsigned char)p[1]))
1.42 schwarze 696: *p = ASCII_HYPH;
697: p++;
1.8 schwarze 698: }
699:
1.51 schwarze 700: /* Spring the input line trap. */
701: if (1 == roffit_lines) {
1.75 schwarze 702: isz = mandoc_asprintf(&p, "%s\n.%s", *bufp, roffit_macro);
1.51 schwarze 703: free(*bufp);
704: *bufp = p;
705: *szp = isz + 1;
706: *offs = 0;
707: free(roffit_macro);
708: roffit_lines = 0;
709: return(ROFF_REPARSE);
710: } else if (1 < roffit_lines)
711: --roffit_lines;
1.42 schwarze 712: return(ROFF_CONT);
1.8 schwarze 713: }
714:
1.1 schwarze 715: enum rofferr
1.80 schwarze 716: roff_parseln(struct roff *r, int ln, char **bufp,
1.6 schwarze 717: size_t *szp, int pos, int *offs)
1.1 schwarze 718: {
719: enum rofft t;
1.27 schwarze 720: enum rofferr e;
1.35 schwarze 721: int ppos, ctl;
1.1 schwarze 722:
1.2 schwarze 723: /*
1.8 schwarze 724: * Run the reserved-word filter only if we have some reserved
725: * words to fill in.
726: */
727:
1.45 schwarze 728: e = roff_res(r, bufp, szp, ln, pos);
729: if (ROFF_IGN == e)
730: return(e);
731: assert(ROFF_CONT == e);
1.8 schwarze 732:
1.35 schwarze 733: ppos = pos;
1.48 schwarze 734: ctl = roff_getcontrol(r, *bufp, &pos);
1.35 schwarze 735:
1.8 schwarze 736: /*
1.2 schwarze 737: * First, if a scope is open and we're not a macro, pass the
738: * text through the macro's filter. If a scope isn't open and
739: * we're not a macro, just let it through.
1.32 schwarze 740: * Finally, if there's an equation scope open, divert it into it
741: * no matter our state.
1.2 schwarze 742: */
743:
1.35 schwarze 744: if (r->last && ! ctl) {
1.2 schwarze 745: t = r->last->tok;
746: assert(roffs[t].text);
1.80 schwarze 747: e = (*roffs[t].text)(r, t, bufp, szp, ln, pos, pos, offs);
1.27 schwarze 748: assert(ROFF_IGN == e || ROFF_CONT == e);
1.32 schwarze 749: if (ROFF_CONT != e)
750: return(e);
1.54 schwarze 751: }
752: if (r->eqn)
753: return(eqn_read(&r->eqn, ln, *bufp, ppos, offs));
754: if ( ! ctl) {
1.32 schwarze 755: if (r->tbl)
1.35 schwarze 756: return(tbl_read(r->tbl, ln, *bufp, pos));
1.51 schwarze 757: return(roff_parsetext(bufp, szp, pos, offs));
1.54 schwarze 758: }
1.100 ! schwarze 759:
! 760: /* Skip empty request lines. */
! 761:
! 762: if ((*bufp)[pos] == '"') {
! 763: mandoc_msg(MANDOCERR_COMMENT_BAD, r->parse,
! 764: ln, pos, NULL);
! 765: return(ROFF_IGN);
! 766: } else if ((*bufp)[pos] == '\0')
! 767: return(ROFF_IGN);
1.2 schwarze 768:
769: /*
770: * If a scope is open, go to the child handler for that macro,
771: * as it may want to preprocess before doing anything with it.
1.32 schwarze 772: * Don't do so if an equation is open.
1.2 schwarze 773: */
774:
775: if (r->last) {
1.1 schwarze 776: t = r->last->tok;
777: assert(roffs[t].sub);
1.80 schwarze 778: return((*roffs[t].sub)(r, t, bufp, szp,
779: ln, ppos, pos, offs));
1.2 schwarze 780: }
781:
782: /*
783: * Lastly, as we've no scope open, try to look up and execute
784: * the new macro. If no macro is found, simply return and let
785: * the compilers handle it.
786: */
787:
1.87 schwarze 788: if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos, ln, ppos)))
1.1 schwarze 789: return(ROFF_CONT);
790:
1.2 schwarze 791: assert(roffs[t].proc);
1.80 schwarze 792: return((*roffs[t].proc)(r, t, bufp, szp, ln, ppos, pos, offs));
1.2 schwarze 793: }
794:
1.27 schwarze 795: void
1.2 schwarze 796: roff_endparse(struct roff *r)
797: {
1.1 schwarze 798:
1.27 schwarze 799: if (r->last)
1.94 schwarze 800: mandoc_msg(MANDOCERR_BLK_NOEND, r->parse,
801: r->last->line, r->last->col,
802: roffs[r->last->tok].name);
1.27 schwarze 803:
1.32 schwarze 804: if (r->eqn) {
1.94 schwarze 805: mandoc_msg(MANDOCERR_BLK_NOEND, r->parse,
806: r->eqn->eqn.ln, r->eqn->eqn.pos, "EQ");
1.41 schwarze 807: eqn_end(&r->eqn);
1.32 schwarze 808: }
809:
1.27 schwarze 810: if (r->tbl) {
1.94 schwarze 811: mandoc_msg(MANDOCERR_BLK_NOEND, r->parse,
812: r->tbl->line, r->tbl->pos, "TS");
1.41 schwarze 813: tbl_end(&r->tbl);
1.27 schwarze 814: }
1.1 schwarze 815: }
816:
817: /*
818: * Parse a roff node's type from the input buffer. This must be in the
819: * form of ".foo xxx" in the usual way.
820: */
821: static enum rofft
1.87 schwarze 822: roff_parse(struct roff *r, char *buf, int *pos, int ln, int ppos)
1.1 schwarze 823: {
1.87 schwarze 824: char *cp;
1.16 schwarze 825: const char *mac;
826: size_t maclen;
1.1 schwarze 827: enum rofft t;
828:
1.87 schwarze 829: cp = buf + *pos;
830:
831: if ('\0' == *cp || '"' == *cp || '\t' == *cp || ' ' == *cp)
1.1 schwarze 832: return(ROFF_MAX);
833:
1.87 schwarze 834: mac = cp;
835: maclen = roff_getname(r, &cp, ln, ppos);
1.1 schwarze 836:
1.16 schwarze 837: t = (r->current_string = roff_getstrn(r, mac, maclen))
1.42 schwarze 838: ? ROFF_USERDEF : roffhash_find(mac, maclen);
1.1 schwarze 839:
1.87 schwarze 840: if (ROFF_MAX != t)
841: *pos = cp - buf;
1.1 schwarze 842:
843: return(t);
844: }
845:
846: static enum rofferr
1.2 schwarze 847: roff_cblock(ROFF_ARGS)
1.1 schwarze 848: {
849:
1.2 schwarze 850: /*
851: * A block-close `..' should only be invoked as a child of an
852: * ignore macro, otherwise raise a warning and just ignore it.
853: */
854:
855: if (NULL == r->last) {
1.94 schwarze 856: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
857: ln, ppos, "..");
1.2 schwarze 858: return(ROFF_IGN);
859: }
1.1 schwarze 860:
1.2 schwarze 861: switch (r->last->tok) {
1.80 schwarze 862: case ROFF_am:
1.93 schwarze 863: /* ROFF_am1 is remapped to ROFF_am in roff_block(). */
1.2 schwarze 864: /* FALLTHROUGH */
1.80 schwarze 865: case ROFF_ami:
1.2 schwarze 866: /* FALLTHROUGH */
1.80 schwarze 867: case ROFF_de:
1.23 schwarze 868: /* ROFF_de1 is remapped to ROFF_de in roff_block(). */
1.2 schwarze 869: /* FALLTHROUGH */
1.80 schwarze 870: case ROFF_dei:
1.2 schwarze 871: /* FALLTHROUGH */
1.80 schwarze 872: case ROFF_ig:
1.2 schwarze 873: break;
874: default:
1.94 schwarze 875: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
876: ln, ppos, "..");
1.1 schwarze 877: return(ROFF_IGN);
1.2 schwarze 878: }
879:
880: if ((*bufp)[pos])
1.90 schwarze 881: mandoc_vmsg(MANDOCERR_ARG_SKIP, r->parse, ln, pos,
882: ".. %s", *bufp + pos);
1.2 schwarze 883:
884: roffnode_pop(r);
885: roffnode_cleanscope(r);
886: return(ROFF_IGN);
887:
888: }
1.1 schwarze 889:
1.2 schwarze 890: static void
891: roffnode_cleanscope(struct roff *r)
892: {
1.1 schwarze 893:
1.2 schwarze 894: while (r->last) {
1.46 schwarze 895: if (--r->last->endspan != 0)
1.2 schwarze 896: break;
897: roffnode_pop(r);
898: }
899: }
1.1 schwarze 900:
1.68 schwarze 901: static void
902: roff_ccond(struct roff *r, int ln, int ppos)
1.2 schwarze 903: {
1.1 schwarze 904:
1.2 schwarze 905: if (NULL == r->last) {
1.94 schwarze 906: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
907: ln, ppos, "\\}");
1.68 schwarze 908: return;
1.2 schwarze 909: }
1.1 schwarze 910:
1.2 schwarze 911: switch (r->last->tok) {
1.80 schwarze 912: case ROFF_el:
1.2 schwarze 913: /* FALLTHROUGH */
1.80 schwarze 914: case ROFF_ie:
1.2 schwarze 915: /* FALLTHROUGH */
1.80 schwarze 916: case ROFF_if:
1.2 schwarze 917: break;
918: default:
1.94 schwarze 919: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
920: ln, ppos, "\\}");
1.68 schwarze 921: return;
1.2 schwarze 922: }
1.1 schwarze 923:
1.2 schwarze 924: if (r->last->endspan > -1) {
1.94 schwarze 925: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
926: ln, ppos, "\\}");
1.68 schwarze 927: return;
1.2 schwarze 928: }
929:
930: roffnode_pop(r);
931: roffnode_cleanscope(r);
1.68 schwarze 932: return;
1.1 schwarze 933: }
934:
935: static enum rofferr
1.2 schwarze 936: roff_block(ROFF_ARGS)
1.1 schwarze 937: {
1.93 schwarze 938: const char *name;
939: char *iname, *cp;
1.86 schwarze 940: size_t namesz;
1.16 schwarze 941:
1.93 schwarze 942: /* Ignore groff compatibility mode for now. */
1.2 schwarze 943:
1.93 schwarze 944: if (ROFF_de1 == tok)
945: tok = ROFF_de;
946: else if (ROFF_am1 == tok)
947: tok = ROFF_am;
948:
949: /* Parse the macro name argument. */
950:
951: cp = *bufp + pos;
952: if (ROFF_ig == tok) {
953: iname = NULL;
954: namesz = 0;
955: } else {
956: iname = cp;
957: namesz = roff_getname(r, &cp, ln, ppos);
958: iname[namesz] = '\0';
959: }
1.22 schwarze 960:
1.93 schwarze 961: /* Resolve the macro name argument if it is indirect. */
1.22 schwarze 962:
1.93 schwarze 963: if (namesz && (ROFF_dei == tok || ROFF_ami == tok)) {
964: if (NULL == (name = roff_getstrn(r, iname, namesz))) {
965: mandoc_vmsg(MANDOCERR_STR_UNDEF,
966: r->parse, ln, (int)(iname - *bufp),
967: "%.*s", (int)namesz, iname);
968: namesz = 0;
969: } else
970: namesz = strlen(name);
971: } else
972: name = iname;
1.22 schwarze 973:
1.93 schwarze 974: if (0 == namesz && ROFF_ig != tok) {
975: mandoc_msg(MANDOCERR_REQ_EMPTY, r->parse,
976: ln, ppos, roffs[tok].name);
977: return(ROFF_IGN);
978: }
1.2 schwarze 979:
1.16 schwarze 980: roffnode_push(r, tok, name, ln, ppos);
981:
982: /*
983: * At the beginning of a `de' macro, clear the existing string
984: * with the same name, if there is one. New content will be
1.66 schwarze 985: * appended from roff_block_text() in multiline mode.
1.16 schwarze 986: */
1.22 schwarze 987:
1.93 schwarze 988: if (ROFF_de == tok || ROFF_dei == tok)
1.86 schwarze 989: roff_setstrn(&r->strtab, name, namesz, "", 0, 0);
1.2 schwarze 990:
1.86 schwarze 991: if ('\0' == *cp)
1.2 schwarze 992: return(ROFF_IGN);
1.1 schwarze 993:
1.93 schwarze 994: /* Get the custom end marker. */
1.22 schwarze 995:
1.93 schwarze 996: iname = cp;
1.86 schwarze 997: namesz = roff_getname(r, &cp, ln, ppos);
1.93 schwarze 998:
999: /* Resolve the end marker if it is indirect. */
1000:
1001: if (namesz && (ROFF_dei == tok || ROFF_ami == tok)) {
1002: if (NULL == (name = roff_getstrn(r, iname, namesz))) {
1003: mandoc_vmsg(MANDOCERR_STR_UNDEF,
1004: r->parse, ln, (int)(iname - *bufp),
1005: "%.*s", (int)namesz, iname);
1006: namesz = 0;
1007: } else
1008: namesz = strlen(name);
1009: } else
1010: name = iname;
1011:
1.86 schwarze 1012: if (namesz)
1013: r->last->end = mandoc_strndup(name, namesz);
1.2 schwarze 1014:
1.86 schwarze 1015: if ('\0' != *cp)
1.90 schwarze 1016: mandoc_vmsg(MANDOCERR_ARG_EXCESS, r->parse,
1017: ln, pos, ".%s ... %s", roffs[tok].name, cp);
1.1 schwarze 1018:
1019: return(ROFF_IGN);
1020: }
1021:
1022: static enum rofferr
1.2 schwarze 1023: roff_block_sub(ROFF_ARGS)
1.1 schwarze 1024: {
1.2 schwarze 1025: enum rofft t;
1026: int i, j;
1027:
1028: /*
1029: * First check whether a custom macro exists at this level. If
1030: * it does, then check against it. This is some of groff's
1031: * stranger behaviours. If we encountered a custom end-scope
1032: * tag and that tag also happens to be a "real" macro, then we
1033: * need to try interpreting it again as a real macro. If it's
1034: * not, then return ignore. Else continue.
1035: */
1036:
1037: if (r->last->end) {
1.35 schwarze 1038: for (i = pos, j = 0; r->last->end[j]; j++, i++)
1.2 schwarze 1039: if ((*bufp)[i] != r->last->end[j])
1040: break;
1.1 schwarze 1041:
1.80 schwarze 1042: if ('\0' == r->last->end[j] &&
1043: ('\0' == (*bufp)[i] ||
1044: ' ' == (*bufp)[i] ||
1045: '\t' == (*bufp)[i])) {
1.2 schwarze 1046: roffnode_pop(r);
1047: roffnode_cleanscope(r);
1.1 schwarze 1048:
1.35 schwarze 1049: while (' ' == (*bufp)[i] || '\t' == (*bufp)[i])
1050: i++;
1051:
1052: pos = i;
1.87 schwarze 1053: if (ROFF_MAX != roff_parse(r, *bufp, &pos, ln, ppos))
1.2 schwarze 1054: return(ROFF_RERUN);
1055: return(ROFF_IGN);
1056: }
1.1 schwarze 1057: }
1058:
1.2 schwarze 1059: /*
1060: * If we have no custom end-query or lookup failed, then try
1061: * pulling it out of the hashtable.
1062: */
1.1 schwarze 1063:
1.87 schwarze 1064: t = roff_parse(r, *bufp, &pos, ln, ppos);
1.1 schwarze 1065:
1.16 schwarze 1066: if (ROFF_cblock != t) {
1.93 schwarze 1067: if (ROFF_ig != tok)
1.66 schwarze 1068: roff_setstr(r, r->last->name, *bufp + ppos, 2);
1.1 schwarze 1069: return(ROFF_IGN);
1.16 schwarze 1070: }
1.1 schwarze 1071:
1.2 schwarze 1072: assert(roffs[t].proc);
1.80 schwarze 1073: return((*roffs[t].proc)(r, t, bufp, szp, ln, ppos, pos, offs));
1.2 schwarze 1074: }
1075:
1076: static enum rofferr
1077: roff_block_text(ROFF_ARGS)
1078: {
1079:
1.93 schwarze 1080: if (ROFF_ig != tok)
1.66 schwarze 1081: roff_setstr(r, r->last->name, *bufp + pos, 2);
1.16 schwarze 1082:
1.2 schwarze 1083: return(ROFF_IGN);
1084: }
1085:
1086: static enum rofferr
1087: roff_cond_sub(ROFF_ARGS)
1088: {
1089: enum rofft t;
1.37 schwarze 1090: char *ep;
1.71 schwarze 1091: int rr;
1.2 schwarze 1092:
1093: rr = r->last->rule;
1.37 schwarze 1094: roffnode_cleanscope(r);
1.87 schwarze 1095: t = roff_parse(r, *bufp, &pos, ln, ppos);
1.2 schwarze 1096:
1.37 schwarze 1097: /*
1.50 schwarze 1098: * Fully handle known macros when they are structurally
1099: * required or when the conditional evaluated to true.
1.5 schwarze 1100: */
1101:
1.50 schwarze 1102: if ((ROFF_MAX != t) &&
1.71 schwarze 1103: (rr || ROFFMAC_STRUCT & roffs[t].flags)) {
1.50 schwarze 1104: assert(roffs[t].proc);
1105: return((*roffs[t].proc)(r, t, bufp, szp,
1.80 schwarze 1106: ln, ppos, pos, offs));
1.50 schwarze 1107: }
1.39 schwarze 1108:
1.69 schwarze 1109: /*
1110: * If `\}' occurs on a macro line without a preceding macro,
1111: * drop the line completely.
1112: */
1113:
1114: ep = *bufp + pos;
1115: if ('\\' == ep[0] && '}' == ep[1])
1.71 schwarze 1116: rr = 0;
1.69 schwarze 1117:
1.50 schwarze 1118: /* Always check for the closing delimiter `\}'. */
1.39 schwarze 1119:
1.50 schwarze 1120: while (NULL != (ep = strchr(ep, '\\'))) {
1.70 schwarze 1121: if ('}' == *(++ep)) {
1122: *ep = '&';
1123: roff_ccond(r, ln, ep - *bufp - 1);
1124: }
1125: ++ep;
1.50 schwarze 1126: }
1.71 schwarze 1127: return(rr ? ROFF_CONT : ROFF_IGN);
1.2 schwarze 1128: }
1129:
1130: static enum rofferr
1131: roff_cond_text(ROFF_ARGS)
1132: {
1.37 schwarze 1133: char *ep;
1.71 schwarze 1134: int rr;
1.2 schwarze 1135:
1136: rr = r->last->rule;
1.37 schwarze 1137: roffnode_cleanscope(r);
1.1 schwarze 1138:
1.70 schwarze 1139: ep = *bufp + pos;
1140: while (NULL != (ep = strchr(ep, '\\'))) {
1141: if ('}' == *(++ep)) {
1142: *ep = '&';
1143: roff_ccond(r, ln, ep - *bufp - 1);
1144: }
1145: ++ep;
1.2 schwarze 1146: }
1.71 schwarze 1147: return(rr ? ROFF_CONT : ROFF_IGN);
1.2 schwarze 1148: }
1149:
1.77 schwarze 1150: /*
1151: * Parse a single signed integer number. Stop at the first non-digit.
1152: * If there is at least one digit, return success and advance the
1153: * parse point, else return failure and let the parse point unchanged.
1154: * Ignore overflows, treat them just like the C language.
1155: */
1.56 schwarze 1156: static int
1157: roff_getnum(const char *v, int *pos, int *res)
1158: {
1.79 schwarze 1159: int myres, n, p;
1160:
1161: if (NULL == res)
1162: res = &myres;
1.56 schwarze 1163:
1164: p = *pos;
1165: n = v[p] == '-';
1166: if (n)
1167: p++;
1168:
1169: for (*res = 0; isdigit((unsigned char)v[p]); p++)
1.77 schwarze 1170: *res = 10 * *res + v[p] - '0';
1.56 schwarze 1171: if (p == *pos + n)
1172: return 0;
1173:
1174: if (n)
1175: *res = -*res;
1176:
1177: *pos = p;
1178: return 1;
1179: }
1180:
1.71 schwarze 1181: /*
1182: * Evaluate a string comparison condition.
1183: * The first character is the delimiter.
1184: * Succeed if the string up to its second occurrence
1185: * matches the string up to its third occurence.
1186: * Advance the cursor after the third occurrence
1187: * or lacking that, to the end of the line.
1188: */
1189: static int
1190: roff_evalstrcond(const char *v, int *pos)
1191: {
1192: const char *s1, *s2, *s3;
1193: int match;
1194:
1195: match = 0;
1196: s1 = v + *pos; /* initial delimiter */
1197: s2 = s1 + 1; /* for scanning the first string */
1198: s3 = strchr(s2, *s1); /* for scanning the second string */
1199:
1200: if (NULL == s3) /* found no middle delimiter */
1201: goto out;
1202:
1203: while ('\0' != *++s3) {
1204: if (*s2 != *s3) { /* mismatch */
1205: s3 = strchr(s3, *s1);
1206: break;
1207: }
1208: if (*s3 == *s1) { /* found the final delimiter */
1209: match = 1;
1210: break;
1211: }
1212: s2++;
1213: }
1214:
1215: out:
1216: if (NULL == s3)
1217: s3 = strchr(s2, '\0');
1218: else
1219: s3++;
1220: *pos = s3 - v;
1221: return(match);
1222: }
1223:
1.77 schwarze 1224: /*
1225: * Evaluate an optionally negated single character, numerical,
1226: * or string condition.
1227: */
1.71 schwarze 1228: static int
1.5 schwarze 1229: roff_evalcond(const char *v, int *pos)
1230: {
1.77 schwarze 1231: int wanttrue, number;
1.5 schwarze 1232:
1.71 schwarze 1233: if ('!' == v[*pos]) {
1234: wanttrue = 0;
1235: (*pos)++;
1236: } else
1237: wanttrue = 1;
1238:
1.5 schwarze 1239: switch (v[*pos]) {
1.80 schwarze 1240: case 'n':
1.71 schwarze 1241: /* FALLTHROUGH */
1.80 schwarze 1242: case 'o':
1.5 schwarze 1243: (*pos)++;
1.71 schwarze 1244: return(wanttrue);
1.80 schwarze 1245: case 'c':
1.71 schwarze 1246: /* FALLTHROUGH */
1.80 schwarze 1247: case 'd':
1.71 schwarze 1248: /* FALLTHROUGH */
1.80 schwarze 1249: case 'e':
1.5 schwarze 1250: /* FALLTHROUGH */
1.80 schwarze 1251: case 'r':
1.5 schwarze 1252: /* FALLTHROUGH */
1.80 schwarze 1253: case 't':
1.5 schwarze 1254: (*pos)++;
1.71 schwarze 1255: return(!wanttrue);
1.5 schwarze 1256: default:
1257: break;
1258: }
1259:
1.77 schwarze 1260: if (roff_evalnum(v, pos, &number, 0))
1261: return((number > 0) == wanttrue);
1262: else
1.71 schwarze 1263: return(roff_evalstrcond(v, pos) == wanttrue);
1.5 schwarze 1264: }
1265:
1.2 schwarze 1266: static enum rofferr
1.21 schwarze 1267: roff_line_ignore(ROFF_ARGS)
1.6 schwarze 1268: {
1.30 schwarze 1269:
1.21 schwarze 1270: return(ROFF_IGN);
1271: }
1272:
1273: static enum rofferr
1.2 schwarze 1274: roff_cond(ROFF_ARGS)
1275: {
1.46 schwarze 1276:
1277: roffnode_push(r, tok, NULL, ln, ppos);
1.2 schwarze 1278:
1.80 schwarze 1279: /*
1.35 schwarze 1280: * An `.el' has no conditional body: it will consume the value
1281: * of the current rstack entry set in prior `ie' calls or
1.80 schwarze 1282: * defaults to DENY.
1.35 schwarze 1283: *
1284: * If we're not an `el', however, then evaluate the conditional.
1285: */
1.1 schwarze 1286:
1.46 schwarze 1287: r->last->rule = ROFF_el == tok ?
1.80 schwarze 1288: (r->rstackpos < 0 ? 0 : r->rstack[r->rstackpos--]) :
1289: roff_evalcond(*bufp, &pos);
1.2 schwarze 1290:
1.35 schwarze 1291: /*
1292: * An if-else will put the NEGATION of the current evaluated
1293: * conditional into the stack of rules.
1294: */
1295:
1.2 schwarze 1296: if (ROFF_ie == tok) {
1.96 schwarze 1297: if (r->rstackpos + 1 == r->rstacksz) {
1298: r->rstacksz += 16;
1299: r->rstack = mandoc_reallocarray(r->rstack,
1300: r->rstacksz, sizeof(int));
1.35 schwarze 1301: }
1.71 schwarze 1302: r->rstack[++r->rstackpos] = !r->last->rule;
1.2 schwarze 1303: }
1.5 schwarze 1304:
1305: /* If the parent has false as its rule, then so do we. */
1306:
1.71 schwarze 1307: if (r->last->parent && !r->last->parent->rule)
1308: r->last->rule = 0;
1.5 schwarze 1309:
1310: /*
1.46 schwarze 1311: * Determine scope.
1312: * If there is nothing on the line after the conditional,
1313: * not even whitespace, use next-line scope.
1.5 schwarze 1314: */
1.2 schwarze 1315:
1.46 schwarze 1316: if ('\0' == (*bufp)[pos]) {
1317: r->last->endspan = 2;
1318: goto out;
1319: }
1320:
1321: while (' ' == (*bufp)[pos])
1322: pos++;
1323:
1324: /* An opening brace requests multiline scope. */
1.2 schwarze 1325:
1326: if ('\\' == (*bufp)[pos] && '{' == (*bufp)[pos + 1]) {
1327: r->last->endspan = -1;
1328: pos += 2;
1.46 schwarze 1329: goto out;
1.80 schwarze 1330: }
1.2 schwarze 1331:
1332: /*
1.46 schwarze 1333: * Anything else following the conditional causes
1334: * single-line scope. Warn if the scope contains
1335: * nothing but trailing whitespace.
1.2 schwarze 1336: */
1337:
1338: if ('\0' == (*bufp)[pos])
1.89 schwarze 1339: mandoc_msg(MANDOCERR_COND_EMPTY, r->parse,
1340: ln, ppos, roffs[tok].name);
1.2 schwarze 1341:
1.46 schwarze 1342: r->last->endspan = 1;
1.1 schwarze 1343:
1.46 schwarze 1344: out:
1.2 schwarze 1345: *offs = pos;
1346: return(ROFF_RERUN);
1.1 schwarze 1347: }
1348:
1.2 schwarze 1349: static enum rofferr
1.7 schwarze 1350: roff_ds(ROFF_ARGS)
1351: {
1.85 schwarze 1352: char *string;
1353: const char *name;
1354: size_t namesz;
1.10 schwarze 1355:
1356: /*
1.85 schwarze 1357: * The first word is the name of the string.
1358: * If it is empty or terminated by an escape sequence,
1359: * abort the `ds' request without defining anything.
1.10 schwarze 1360: */
1.7 schwarze 1361:
1.85 schwarze 1362: name = string = *bufp + pos;
1.7 schwarze 1363: if ('\0' == *name)
1364: return(ROFF_IGN);
1365:
1.85 schwarze 1366: namesz = roff_getname(r, &string, ln, pos);
1367: if ('\\' == name[namesz])
1368: return(ROFF_IGN);
1369:
1370: /* Read past the initial double-quote, if any. */
1.28 schwarze 1371: if ('"' == *string)
1.7 schwarze 1372: string++;
1373:
1.10 schwarze 1374: /* The rest is the value. */
1.85 schwarze 1375: roff_setstrn(&r->strtab, name, namesz, string, strlen(string),
1376: ROFF_as == tok);
1.7 schwarze 1377: return(ROFF_IGN);
1378: }
1379:
1.77 schwarze 1380: /*
1381: * Parse a single operator, one or two characters long.
1382: * If the operator is recognized, return success and advance the
1383: * parse point, else return failure and let the parse point unchanged.
1384: */
1385: static int
1386: roff_getop(const char *v, int *pos, char *res)
1387: {
1388:
1389: *res = v[*pos];
1390:
1391: switch (*res) {
1.80 schwarze 1392: case '+':
1.77 schwarze 1393: /* FALLTHROUGH */
1.80 schwarze 1394: case '-':
1.77 schwarze 1395: /* FALLTHROUGH */
1.80 schwarze 1396: case '*':
1.77 schwarze 1397: /* FALLTHROUGH */
1.80 schwarze 1398: case '/':
1.77 schwarze 1399: /* FALLTHROUGH */
1.80 schwarze 1400: case '%':
1.77 schwarze 1401: /* FALLTHROUGH */
1.80 schwarze 1402: case '&':
1.77 schwarze 1403: /* FALLTHROUGH */
1.80 schwarze 1404: case ':':
1.77 schwarze 1405: break;
1406: case '<':
1407: switch (v[*pos + 1]) {
1.80 schwarze 1408: case '=':
1.77 schwarze 1409: *res = 'l';
1410: (*pos)++;
1411: break;
1.80 schwarze 1412: case '>':
1.77 schwarze 1413: *res = '!';
1414: (*pos)++;
1415: break;
1.80 schwarze 1416: case '?':
1.77 schwarze 1417: *res = 'i';
1418: (*pos)++;
1419: break;
1420: default:
1421: break;
1422: }
1423: break;
1424: case '>':
1425: switch (v[*pos + 1]) {
1.80 schwarze 1426: case '=':
1.77 schwarze 1427: *res = 'g';
1428: (*pos)++;
1429: break;
1.80 schwarze 1430: case '?':
1.77 schwarze 1431: *res = 'a';
1432: (*pos)++;
1433: break;
1434: default:
1435: break;
1436: }
1437: break;
1438: case '=':
1439: if ('=' == v[*pos + 1])
1440: (*pos)++;
1441: break;
1442: default:
1443: return(0);
1444: }
1445: (*pos)++;
1446:
1447: return(*res);
1448: }
1449:
1450: /*
1451: * Evaluate either a parenthesized numeric expression
1452: * or a single signed integer number.
1453: */
1454: static int
1455: roff_evalpar(const char *v, int *pos, int *res)
1456: {
1457:
1458: if ('(' != v[*pos])
1459: return(roff_getnum(v, pos, res));
1460:
1461: (*pos)++;
1462: if ( ! roff_evalnum(v, pos, res, 1))
1463: return(0);
1464:
1.79 schwarze 1465: /*
1466: * Omission of the closing parenthesis
1467: * is an error in validation mode,
1468: * but ignored in evaluation mode.
1469: */
1470:
1.77 schwarze 1471: if (')' == v[*pos])
1472: (*pos)++;
1.79 schwarze 1473: else if (NULL == res)
1474: return(0);
1.77 schwarze 1475:
1476: return(1);
1477: }
1478:
1479: /*
1480: * Evaluate a complete numeric expression.
1481: * Proceed left to right, there is no concept of precedence.
1482: */
1483: static int
1484: roff_evalnum(const char *v, int *pos, int *res, int skipwhite)
1485: {
1486: int mypos, operand2;
1487: char operator;
1488:
1489: if (NULL == pos) {
1490: mypos = 0;
1491: pos = &mypos;
1492: }
1493:
1494: if (skipwhite)
1495: while (isspace((unsigned char)v[*pos]))
1496: (*pos)++;
1497:
1498: if ( ! roff_evalpar(v, pos, res))
1499: return(0);
1500:
1501: while (1) {
1502: if (skipwhite)
1503: while (isspace((unsigned char)v[*pos]))
1504: (*pos)++;
1505:
1506: if ( ! roff_getop(v, pos, &operator))
1507: break;
1508:
1509: if (skipwhite)
1510: while (isspace((unsigned char)v[*pos]))
1511: (*pos)++;
1512:
1513: if ( ! roff_evalpar(v, pos, &operand2))
1514: return(0);
1515:
1516: if (skipwhite)
1517: while (isspace((unsigned char)v[*pos]))
1518: (*pos)++;
1.79 schwarze 1519:
1520: if (NULL == res)
1521: continue;
1.77 schwarze 1522:
1523: switch (operator) {
1.80 schwarze 1524: case '+':
1.77 schwarze 1525: *res += operand2;
1526: break;
1.80 schwarze 1527: case '-':
1.77 schwarze 1528: *res -= operand2;
1529: break;
1.80 schwarze 1530: case '*':
1.77 schwarze 1531: *res *= operand2;
1532: break;
1.80 schwarze 1533: case '/':
1.77 schwarze 1534: *res /= operand2;
1535: break;
1.80 schwarze 1536: case '%':
1.77 schwarze 1537: *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 'l':
1.77 schwarze 1546: *res = *res <= operand2;
1547: break;
1.80 schwarze 1548: case 'g':
1.77 schwarze 1549: *res = *res >= operand2;
1550: break;
1.80 schwarze 1551: case '=':
1.77 schwarze 1552: *res = *res == operand2;
1553: break;
1.80 schwarze 1554: case '!':
1.77 schwarze 1555: *res = *res != operand2;
1556: break;
1.80 schwarze 1557: case '&':
1.77 schwarze 1558: *res = *res && operand2;
1559: break;
1.80 schwarze 1560: case ':':
1.77 schwarze 1561: *res = *res || operand2;
1562: break;
1.80 schwarze 1563: case 'i':
1.77 schwarze 1564: if (operand2 < *res)
1565: *res = operand2;
1566: break;
1.80 schwarze 1567: case 'a':
1.77 schwarze 1568: if (operand2 > *res)
1569: *res = operand2;
1570: break;
1571: default:
1572: abort();
1573: }
1574: }
1575: return(1);
1576: }
1577:
1.52 schwarze 1578: void
1.60 schwarze 1579: roff_setreg(struct roff *r, const char *name, int val, char sign)
1.41 schwarze 1580: {
1.52 schwarze 1581: struct roffreg *reg;
1582:
1583: /* Search for an existing register with the same name. */
1584: reg = r->regtab;
1585:
1586: while (reg && strcmp(name, reg->key.p))
1587: reg = reg->next;
1.41 schwarze 1588:
1.52 schwarze 1589: if (NULL == reg) {
1590: /* Create a new register. */
1591: reg = mandoc_malloc(sizeof(struct roffreg));
1592: reg->key.p = mandoc_strdup(name);
1593: reg->key.sz = strlen(name);
1.60 schwarze 1594: reg->val = 0;
1.52 schwarze 1595: reg->next = r->regtab;
1596: r->regtab = reg;
1597: }
1598:
1.60 schwarze 1599: if ('+' == sign)
1600: reg->val += val;
1601: else if ('-' == sign)
1602: reg->val -= val;
1603: else
1604: reg->val = val;
1.41 schwarze 1605: }
1606:
1.65 schwarze 1607: /*
1608: * Handle some predefined read-only number registers.
1609: * For now, return -1 if the requested register is not predefined;
1610: * in case a predefined read-only register having the value -1
1611: * were to turn up, another special value would have to be chosen.
1612: */
1613: static int
1614: roff_getregro(const char *name)
1615: {
1616:
1617: switch (*name) {
1.80 schwarze 1618: case 'A': /* ASCII approximation mode is always off. */
1.65 schwarze 1619: return(0);
1.80 schwarze 1620: case 'g': /* Groff compatibility mode is always on. */
1.65 schwarze 1621: return(1);
1.80 schwarze 1622: case 'H': /* Fixed horizontal resolution. */
1.65 schwarze 1623: return (24);
1.80 schwarze 1624: case 'j': /* Always adjust left margin only. */
1.65 schwarze 1625: return(0);
1.80 schwarze 1626: case 'T': /* Some output device is always defined. */
1.65 schwarze 1627: return(1);
1.80 schwarze 1628: case 'V': /* Fixed vertical resolution. */
1.65 schwarze 1629: return (40);
1630: default:
1631: return (-1);
1632: }
1633: }
1634:
1.53 schwarze 1635: int
1.52 schwarze 1636: roff_getreg(const struct roff *r, const char *name)
1.41 schwarze 1637: {
1.52 schwarze 1638: struct roffreg *reg;
1.65 schwarze 1639: int val;
1640:
1641: if ('.' == name[0] && '\0' != name[1] && '\0' == name[2]) {
1642: val = roff_getregro(name + 1);
1643: if (-1 != val)
1644: return (val);
1645: }
1.52 schwarze 1646:
1647: for (reg = r->regtab; reg; reg = reg->next)
1648: if (0 == strcmp(name, reg->key.p))
1.53 schwarze 1649: return(reg->val);
1650:
1651: return(0);
1652: }
1653:
1654: static int
1655: roff_getregn(const struct roff *r, const char *name, size_t len)
1656: {
1657: struct roffreg *reg;
1.65 schwarze 1658: int val;
1659:
1660: if ('.' == name[0] && 2 == len) {
1661: val = roff_getregro(name + 1);
1662: if (-1 != val)
1663: return (val);
1664: }
1.53 schwarze 1665:
1666: for (reg = r->regtab; reg; reg = reg->next)
1667: if (len == reg->key.sz &&
1668: 0 == strncmp(name, reg->key.p, len))
1669: return(reg->val);
1.41 schwarze 1670:
1.52 schwarze 1671: return(0);
1.41 schwarze 1672: }
1673:
1.52 schwarze 1674: static void
1675: roff_freereg(struct roffreg *reg)
1.41 schwarze 1676: {
1.52 schwarze 1677: struct roffreg *old_reg;
1.41 schwarze 1678:
1.52 schwarze 1679: while (NULL != reg) {
1680: free(reg->key.p);
1681: old_reg = reg;
1682: reg = reg->next;
1683: free(old_reg);
1684: }
1.41 schwarze 1685: }
1.7 schwarze 1686:
1687: static enum rofferr
1.6 schwarze 1688: roff_nr(ROFF_ARGS)
1.1 schwarze 1689: {
1.85 schwarze 1690: char *key, *val;
1691: size_t keysz;
1.37 schwarze 1692: int iv;
1.60 schwarze 1693: char sign;
1.6 schwarze 1694:
1.85 schwarze 1695: key = val = *bufp + pos;
1696: if ('\0' == *key)
1697: return(ROFF_IGN);
1698:
1699: keysz = roff_getname(r, &val, ln, pos);
1700: if ('\\' == key[keysz])
1701: return(ROFF_IGN);
1702: key[keysz] = '\0';
1.6 schwarze 1703:
1.60 schwarze 1704: sign = *val;
1705: if ('+' == sign || '-' == sign)
1706: val++;
1707:
1.77 schwarze 1708: if (roff_evalnum(val, NULL, &iv, 0))
1709: roff_setreg(r, key, iv, sign);
1.1 schwarze 1710:
1.76 schwarze 1711: return(ROFF_IGN);
1712: }
1713:
1714: static enum rofferr
1715: roff_rr(ROFF_ARGS)
1716: {
1717: struct roffreg *reg, **prev;
1.85 schwarze 1718: char *name, *cp;
1719: size_t namesz;
1.76 schwarze 1720:
1.85 schwarze 1721: name = cp = *bufp + pos;
1722: if ('\0' == *name)
1723: return(ROFF_IGN);
1724: namesz = roff_getname(r, &cp, ln, pos);
1725: name[namesz] = '\0';
1.76 schwarze 1726:
1727: prev = &r->regtab;
1728: while (1) {
1729: reg = *prev;
1730: if (NULL == reg || !strcmp(name, reg->key.p))
1731: break;
1732: prev = ®->next;
1733: }
1734: if (NULL != reg) {
1735: *prev = reg->next;
1736: free(reg->key.p);
1737: free(reg);
1738: }
1.29 schwarze 1739: return(ROFF_IGN);
1740: }
1741:
1742: static enum rofferr
1743: roff_rm(ROFF_ARGS)
1744: {
1745: const char *name;
1746: char *cp;
1.85 schwarze 1747: size_t namesz;
1.29 schwarze 1748:
1749: cp = *bufp + pos;
1750: while ('\0' != *cp) {
1.85 schwarze 1751: name = cp;
1752: namesz = roff_getname(r, &cp, ln, (int)(cp - *bufp));
1753: roff_setstrn(&r->strtab, name, namesz, NULL, 0, 0);
1754: if ('\\' == name[namesz])
1755: break;
1.29 schwarze 1756: }
1.51 schwarze 1757: return(ROFF_IGN);
1758: }
1759:
1760: static enum rofferr
1761: roff_it(ROFF_ARGS)
1762: {
1763: char *cp;
1764: size_t len;
1765: int iv;
1766:
1767: /* Parse the number of lines. */
1768: cp = *bufp + pos;
1769: len = strcspn(cp, " \t");
1770: cp[len] = '\0';
1771: if ((iv = mandoc_strntoi(cp, len, 10)) <= 0) {
1.95 schwarze 1772: mandoc_msg(MANDOCERR_IT_NONUM, r->parse,
1.80 schwarze 1773: ln, ppos, *bufp + 1);
1.51 schwarze 1774: return(ROFF_IGN);
1775: }
1776: cp += len + 1;
1777:
1778: /* Arm the input line trap. */
1779: roffit_lines = iv;
1780: roffit_macro = mandoc_strdup(cp);
1.2 schwarze 1781: return(ROFF_IGN);
1.47 schwarze 1782: }
1783:
1784: static enum rofferr
1785: roff_Dd(ROFF_ARGS)
1786: {
1787: const char *const *cp;
1788:
1.99 schwarze 1789: if ((r->options & (MPARSE_MDOC | MPARSE_QUICK)) == 0)
1.47 schwarze 1790: for (cp = __mdoc_reserved; *cp; cp++)
1791: roff_setstr(r, *cp, NULL, 0);
1792:
1.99 schwarze 1793: if (r->format == 0)
1794: r->format = MPARSE_MDOC;
1795:
1.47 schwarze 1796: return(ROFF_CONT);
1797: }
1798:
1799: static enum rofferr
1800: roff_TH(ROFF_ARGS)
1801: {
1802: const char *const *cp;
1803:
1.99 schwarze 1804: if ((r->options & MPARSE_QUICK) == 0)
1.47 schwarze 1805: for (cp = __man_reserved; *cp; cp++)
1806: roff_setstr(r, *cp, NULL, 0);
1807:
1.99 schwarze 1808: if (r->format == 0)
1809: r->format = MPARSE_MAN;
1810:
1.47 schwarze 1811: return(ROFF_CONT);
1.14 schwarze 1812: }
1813:
1814: static enum rofferr
1.27 schwarze 1815: roff_TE(ROFF_ARGS)
1816: {
1817:
1818: if (NULL == r->tbl)
1.94 schwarze 1819: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
1820: ln, ppos, "TE");
1.27 schwarze 1821: else
1.41 schwarze 1822: tbl_end(&r->tbl);
1.27 schwarze 1823:
1824: return(ROFF_IGN);
1825: }
1826:
1827: static enum rofferr
1828: roff_T_(ROFF_ARGS)
1829: {
1830:
1831: if (NULL == r->tbl)
1.94 schwarze 1832: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
1833: ln, ppos, "T&");
1.27 schwarze 1834: else
1835: tbl_restart(ppos, ln, r->tbl);
1836:
1837: return(ROFF_IGN);
1838: }
1839:
1.41 schwarze 1840: #if 0
1841: static int
1842: roff_closeeqn(struct roff *r)
1843: {
1844:
1845: return(r->eqn && ROFF_EQN == eqn_end(&r->eqn) ? 1 : 0);
1846: }
1847: #endif
1848:
1849: static void
1.80 schwarze 1850: roff_openeqn(struct roff *r, const char *name, int line,
1.41 schwarze 1851: int offs, const char *buf)
1.32 schwarze 1852: {
1.41 schwarze 1853: struct eqn_node *e;
1854: int poff;
1.32 schwarze 1855:
1856: assert(NULL == r->eqn);
1.41 schwarze 1857: e = eqn_alloc(name, offs, line, r->parse);
1.32 schwarze 1858:
1859: if (r->last_eqn)
1860: r->last_eqn->next = e;
1861: else
1862: r->first_eqn = r->last_eqn = e;
1863:
1864: r->eqn = r->last_eqn = e;
1.41 schwarze 1865:
1866: if (buf) {
1867: poff = 0;
1868: eqn_read(&r->eqn, line, buf, offs, &poff);
1869: }
1870: }
1871:
1872: static enum rofferr
1873: roff_EQ(ROFF_ARGS)
1874: {
1875:
1876: roff_openeqn(r, *bufp + pos, ln, ppos, NULL);
1.32 schwarze 1877: return(ROFF_IGN);
1878: }
1879:
1880: static enum rofferr
1881: roff_EN(ROFF_ARGS)
1882: {
1883:
1.94 schwarze 1884: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse, ln, ppos, "EN");
1.32 schwarze 1885: return(ROFF_IGN);
1886: }
1887:
1888: static enum rofferr
1.27 schwarze 1889: roff_TS(ROFF_ARGS)
1890: {
1.49 schwarze 1891: struct tbl_node *tbl;
1.27 schwarze 1892:
1893: if (r->tbl) {
1.94 schwarze 1894: mandoc_msg(MANDOCERR_BLK_BROKEN, r->parse,
1895: ln, ppos, "TS breaks TS");
1.41 schwarze 1896: tbl_end(&r->tbl);
1.27 schwarze 1897: }
1898:
1.49 schwarze 1899: tbl = tbl_alloc(ppos, ln, r->parse);
1.27 schwarze 1900:
1901: if (r->last_tbl)
1.49 schwarze 1902: r->last_tbl->next = tbl;
1.27 schwarze 1903: else
1.49 schwarze 1904: r->first_tbl = r->last_tbl = tbl;
1.27 schwarze 1905:
1.49 schwarze 1906: r->tbl = r->last_tbl = tbl;
1.27 schwarze 1907: return(ROFF_IGN);
1908: }
1909:
1910: static enum rofferr
1.48 schwarze 1911: roff_cc(ROFF_ARGS)
1912: {
1913: const char *p;
1914:
1915: p = *bufp + pos;
1916:
1917: if ('\0' == *p || '.' == (r->control = *p++))
1918: r->control = 0;
1919:
1920: if ('\0' != *p)
1921: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
1922:
1923: return(ROFF_IGN);
1924: }
1925:
1926: static enum rofferr
1.42 schwarze 1927: roff_tr(ROFF_ARGS)
1928: {
1929: const char *p, *first, *second;
1930: size_t fsz, ssz;
1931: enum mandoc_esc esc;
1932:
1933: p = *bufp + pos;
1934:
1935: if ('\0' == *p) {
1936: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
1937: return(ROFF_IGN);
1938: }
1939:
1940: while ('\0' != *p) {
1941: fsz = ssz = 1;
1942:
1943: first = p++;
1944: if ('\\' == *first) {
1945: esc = mandoc_escape(&p, NULL, NULL);
1946: if (ESCAPE_ERROR == esc) {
1.92 schwarze 1947: mandoc_msg(MANDOCERR_ESC_BAD, r->parse,
1948: ln, (int)(p - *bufp), first);
1.42 schwarze 1949: return(ROFF_IGN);
1950: }
1951: fsz = (size_t)(p - first);
1952: }
1953:
1954: second = p++;
1955: if ('\\' == *second) {
1956: esc = mandoc_escape(&p, NULL, NULL);
1957: if (ESCAPE_ERROR == esc) {
1.92 schwarze 1958: mandoc_msg(MANDOCERR_ESC_BAD, r->parse,
1959: ln, (int)(p - *bufp), second);
1.42 schwarze 1960: return(ROFF_IGN);
1961: }
1962: ssz = (size_t)(p - second);
1963: } else if ('\0' == *second) {
1.80 schwarze 1964: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse,
1965: ln, (int)(p - *bufp), NULL);
1.42 schwarze 1966: second = " ";
1967: p--;
1968: }
1969:
1970: if (fsz > 1) {
1.80 schwarze 1971: roff_setstrn(&r->xmbtab, first, fsz,
1972: second, ssz, 0);
1.42 schwarze 1973: continue;
1974: }
1975:
1976: if (NULL == r->xtab)
1.80 schwarze 1977: r->xtab = mandoc_calloc(128,
1978: sizeof(struct roffstr));
1.42 schwarze 1979:
1980: free(r->xtab[(int)*first].p);
1981: r->xtab[(int)*first].p = mandoc_strndup(second, ssz);
1982: r->xtab[(int)*first].sz = ssz;
1983: }
1984:
1985: return(ROFF_IGN);
1986: }
1987:
1988: static enum rofferr
1.14 schwarze 1989: roff_so(ROFF_ARGS)
1990: {
1991: char *name;
1.15 schwarze 1992:
1.83 schwarze 1993: name = *bufp + pos;
1.97 schwarze 1994: mandoc_vmsg(MANDOCERR_SO, r->parse, ln, ppos, "so %s", name);
1.14 schwarze 1995:
1.22 schwarze 1996: /*
1997: * Handle `so'. Be EXTREMELY careful, as we shouldn't be
1998: * opening anything that's not in our cwd or anything beneath
1999: * it. Thus, explicitly disallow traversing up the file-system
2000: * or using absolute paths.
2001: */
2002:
1.14 schwarze 2003: if ('/' == *name || strstr(name, "../") || strstr(name, "/..")) {
1.83 schwarze 2004: mandoc_vmsg(MANDOCERR_SO_PATH, r->parse, ln, ppos,
2005: ".so %s", name);
1.14 schwarze 2006: return(ROFF_ERR);
2007: }
2008:
2009: *offs = pos;
2010: return(ROFF_SO);
1.7 schwarze 2011: }
2012:
1.16 schwarze 2013: static enum rofferr
2014: roff_userdef(ROFF_ARGS)
1.12 schwarze 2015: {
1.16 schwarze 2016: const char *arg[9];
2017: char *cp, *n1, *n2;
1.25 schwarze 2018: int i;
1.12 schwarze 2019:
1.16 schwarze 2020: /*
2021: * Collect pointers to macro argument strings
1.61 schwarze 2022: * and NUL-terminate them.
1.16 schwarze 2023: */
2024: cp = *bufp + pos;
1.25 schwarze 2025: for (i = 0; i < 9; i++)
1.26 schwarze 2026: arg[i] = '\0' == *cp ? "" :
1.35 schwarze 2027: mandoc_getarg(r->parse, &cp, ln, &pos);
1.16 schwarze 2028:
2029: /*
2030: * Expand macro arguments.
1.12 schwarze 2031: */
1.16 schwarze 2032: *szp = 0;
2033: n1 = cp = mandoc_strdup(r->current_string);
2034: while (NULL != (cp = strstr(cp, "\\$"))) {
2035: i = cp[2] - '1';
2036: if (0 > i || 8 < i) {
2037: /* Not an argument invocation. */
2038: cp += 2;
2039: continue;
2040: }
1.82 schwarze 2041: *cp = '\0';
2042: *szp = mandoc_asprintf(&n2, "%s%s%s",
2043: n1, arg[i], cp + 3) + 1;
1.16 schwarze 2044: cp = n2 + (cp - n1);
2045: free(n1);
2046: n1 = n2;
1.12 schwarze 2047: }
2048:
1.16 schwarze 2049: /*
2050: * Replace the macro invocation
2051: * by the expanded macro.
2052: */
2053: free(*bufp);
2054: *bufp = n1;
2055: if (0 == *szp)
2056: *szp = strlen(*bufp) + 1;
2057:
1.19 schwarze 2058: return(*szp > 1 && '\n' == (*bufp)[(int)*szp - 2] ?
1.16 schwarze 2059: ROFF_REPARSE : ROFF_APPEND);
1.12 schwarze 2060: }
1.28 schwarze 2061:
1.85 schwarze 2062: static size_t
1.28 schwarze 2063: roff_getname(struct roff *r, char **cpp, int ln, int pos)
2064: {
2065: char *name, *cp;
1.85 schwarze 2066: size_t namesz;
1.28 schwarze 2067:
2068: name = *cpp;
2069: if ('\0' == *name)
1.85 schwarze 2070: return(0);
1.28 schwarze 2071:
1.85 schwarze 2072: /* Read until end of name and terminate it with NUL. */
2073: for (cp = name; 1; cp++) {
2074: if ('\0' == *cp || ' ' == *cp) {
2075: namesz = cp - name;
2076: break;
2077: }
1.28 schwarze 2078: if ('\\' != *cp)
2079: continue;
1.88 schwarze 2080: namesz = cp - name;
2081: if ('{' == cp[1] || '}' == cp[1])
2082: break;
1.28 schwarze 2083: cp++;
2084: if ('\\' == *cp)
2085: continue;
1.97 schwarze 2086: mandoc_vmsg(MANDOCERR_NAMESC, r->parse, ln, pos,
2087: "%.*s", (int)(cp - name + 1), name);
1.85 schwarze 2088: mandoc_escape((const char **)&cp, NULL, NULL);
2089: break;
1.28 schwarze 2090: }
2091:
2092: /* Read past spaces. */
2093: while (' ' == *cp)
2094: cp++;
2095:
2096: *cpp = cp;
1.85 schwarze 2097: return(namesz);
1.28 schwarze 2098: }
2099:
1.16 schwarze 2100: /*
2101: * Store *string into the user-defined string called *name.
2102: * To clear an existing entry, call with (*r, *name, NULL, 0).
1.66 schwarze 2103: * append == 0: replace mode
2104: * append == 1: single-line append mode
2105: * append == 2: multiline append mode, append '\n' after each call
1.16 schwarze 2106: */
1.8 schwarze 2107: static void
1.16 schwarze 2108: roff_setstr(struct roff *r, const char *name, const char *string,
1.66 schwarze 2109: int append)
1.7 schwarze 2110: {
1.42 schwarze 2111:
2112: roff_setstrn(&r->strtab, name, strlen(name), string,
1.80 schwarze 2113: string ? strlen(string) : 0, append);
1.42 schwarze 2114: }
2115:
2116: static void
2117: roff_setstrn(struct roffkv **r, const char *name, size_t namesz,
1.66 schwarze 2118: const char *string, size_t stringsz, int append)
1.42 schwarze 2119: {
2120: struct roffkv *n;
2121: char *c;
2122: int i;
2123: size_t oldch, newch;
1.7 schwarze 2124:
1.16 schwarze 2125: /* Search for an existing string with the same name. */
1.42 schwarze 2126: n = *r;
2127:
1.84 schwarze 2128: while (n && (namesz != n->key.sz ||
2129: strncmp(n->key.p, name, namesz)))
1.7 schwarze 2130: n = n->next;
1.8 schwarze 2131:
2132: if (NULL == n) {
1.16 schwarze 2133: /* Create a new string table entry. */
1.42 schwarze 2134: n = mandoc_malloc(sizeof(struct roffkv));
2135: n->key.p = mandoc_strndup(name, namesz);
2136: n->key.sz = namesz;
2137: n->val.p = NULL;
2138: n->val.sz = 0;
2139: n->next = *r;
2140: *r = n;
1.66 schwarze 2141: } else if (0 == append) {
1.42 schwarze 2142: free(n->val.p);
2143: n->val.p = NULL;
2144: n->val.sz = 0;
1.16 schwarze 2145: }
2146:
2147: if (NULL == string)
2148: return;
2149:
2150: /*
2151: * One additional byte for the '\n' in multiline mode,
2152: * and one for the terminating '\0'.
2153: */
1.66 schwarze 2154: newch = stringsz + (1 < append ? 2u : 1u);
1.42 schwarze 2155:
2156: if (NULL == n->val.p) {
2157: n->val.p = mandoc_malloc(newch);
2158: *n->val.p = '\0';
1.16 schwarze 2159: oldch = 0;
2160: } else {
1.42 schwarze 2161: oldch = n->val.sz;
2162: n->val.p = mandoc_realloc(n->val.p, oldch + newch);
1.16 schwarze 2163: }
2164:
2165: /* Skip existing content in the destination buffer. */
1.42 schwarze 2166: c = n->val.p + (int)oldch;
1.16 schwarze 2167:
2168: /* Append new content to the destination buffer. */
1.42 schwarze 2169: i = 0;
2170: while (i < (int)stringsz) {
1.16 schwarze 2171: /*
2172: * Rudimentary roff copy mode:
2173: * Handle escaped backslashes.
2174: */
1.42 schwarze 2175: if ('\\' == string[i] && '\\' == string[i + 1])
2176: i++;
2177: *c++ = string[i++];
1.16 schwarze 2178: }
1.8 schwarze 2179:
1.16 schwarze 2180: /* Append terminating bytes. */
1.66 schwarze 2181: if (1 < append)
1.16 schwarze 2182: *c++ = '\n';
1.42 schwarze 2183:
1.16 schwarze 2184: *c = '\0';
1.42 schwarze 2185: n->val.sz = (int)(c - n->val.p);
1.7 schwarze 2186: }
2187:
1.8 schwarze 2188: static const char *
2189: roff_getstrn(const struct roff *r, const char *name, size_t len)
1.7 schwarze 2190: {
1.42 schwarze 2191: const struct roffkv *n;
1.64 schwarze 2192: int i;
1.7 schwarze 2193:
1.42 schwarze 2194: for (n = r->strtab; n; n = n->next)
1.80 schwarze 2195: if (0 == strncmp(name, n->key.p, len) &&
2196: '\0' == n->key.p[(int)len])
1.42 schwarze 2197: return(n->val.p);
1.64 schwarze 2198:
2199: for (i = 0; i < PREDEFS_MAX; i++)
2200: if (0 == strncmp(name, predefs[i].name, len) &&
2201: '\0' == predefs[i].name[(int)len])
2202: return(predefs[i].str);
1.8 schwarze 2203:
1.42 schwarze 2204: return(NULL);
1.7 schwarze 2205: }
2206:
1.8 schwarze 2207: static void
1.42 schwarze 2208: roff_freestr(struct roffkv *r)
1.7 schwarze 2209: {
1.42 schwarze 2210: struct roffkv *n, *nn;
1.7 schwarze 2211:
1.42 schwarze 2212: for (n = r; n; n = nn) {
2213: free(n->key.p);
2214: free(n->val.p);
1.7 schwarze 2215: nn = n->next;
2216: free(n);
2217: }
1.27 schwarze 2218: }
2219:
2220: const struct tbl_span *
2221: roff_span(const struct roff *r)
2222: {
1.80 schwarze 2223:
1.27 schwarze 2224: return(r->tbl ? tbl_span(r->tbl) : NULL);
1.32 schwarze 2225: }
2226:
2227: const struct eqn *
2228: roff_eqn(const struct roff *r)
2229: {
1.80 schwarze 2230:
1.32 schwarze 2231: return(r->last_eqn ? &r->last_eqn->eqn : NULL);
1.42 schwarze 2232: }
2233:
2234: /*
2235: * Duplicate an input string, making the appropriate character
2236: * conversations (as stipulated by `tr') along the way.
2237: * Returns a heap-allocated string with all the replacements made.
2238: */
2239: char *
2240: roff_strdup(const struct roff *r, const char *p)
2241: {
2242: const struct roffkv *cp;
2243: char *res;
2244: const char *pp;
2245: size_t ssz, sz;
2246: enum mandoc_esc esc;
2247:
2248: if (NULL == r->xmbtab && NULL == r->xtab)
2249: return(mandoc_strdup(p));
2250: else if ('\0' == *p)
2251: return(mandoc_strdup(""));
2252:
2253: /*
2254: * Step through each character looking for term matches
2255: * (remember that a `tr' can be invoked with an escape, which is
2256: * a glyph but the escape is multi-character).
2257: * We only do this if the character hash has been initialised
2258: * and the string is >0 length.
2259: */
2260:
2261: res = NULL;
2262: ssz = 0;
2263:
2264: while ('\0' != *p) {
2265: if ('\\' != *p && r->xtab && r->xtab[(int)*p].p) {
2266: sz = r->xtab[(int)*p].sz;
2267: res = mandoc_realloc(res, ssz + sz + 1);
2268: memcpy(res + ssz, r->xtab[(int)*p].p, sz);
2269: ssz += sz;
2270: p++;
2271: continue;
2272: } else if ('\\' != *p) {
2273: res = mandoc_realloc(res, ssz + 2);
2274: res[ssz++] = *p++;
2275: continue;
2276: }
2277:
2278: /* Search for term matches. */
2279: for (cp = r->xmbtab; cp; cp = cp->next)
2280: if (0 == strncmp(p, cp->key.p, cp->key.sz))
2281: break;
2282:
2283: if (NULL != cp) {
2284: /*
2285: * A match has been found.
2286: * Append the match to the array and move
2287: * forward by its keysize.
2288: */
1.80 schwarze 2289: res = mandoc_realloc(res,
2290: ssz + cp->val.sz + 1);
1.42 schwarze 2291: memcpy(res + ssz, cp->val.p, cp->val.sz);
2292: ssz += cp->val.sz;
2293: p += (int)cp->key.sz;
2294: continue;
2295: }
2296:
2297: /*
2298: * Handle escapes carefully: we need to copy
2299: * over just the escape itself, or else we might
2300: * do replacements within the escape itself.
2301: * Make sure to pass along the bogus string.
2302: */
2303: pp = p++;
2304: esc = mandoc_escape(&p, NULL, NULL);
2305: if (ESCAPE_ERROR == esc) {
2306: sz = strlen(pp);
2307: res = mandoc_realloc(res, ssz + sz + 1);
2308: memcpy(res + ssz, pp, sz);
2309: break;
2310: }
1.80 schwarze 2311: /*
2312: * We bail out on bad escapes.
1.42 schwarze 2313: * No need to warn: we already did so when
2314: * roff_res() was called.
2315: */
2316: sz = (int)(p - pp);
2317: res = mandoc_realloc(res, ssz + sz + 1);
2318: memcpy(res + ssz, pp, sz);
2319: ssz += sz;
2320: }
2321:
2322: res[(int)ssz] = '\0';
2323: return(res);
1.99 schwarze 2324: }
2325:
2326: int
2327: roff_getformat(const struct roff *r)
2328: {
2329:
2330: return(r->format);
1.48 schwarze 2331: }
2332:
2333: /*
1.80 schwarze 2334: * Find out whether a line is a macro line or not.
1.48 schwarze 2335: * If it is, adjust the current position and return one; if it isn't,
2336: * return zero and don't change the current position.
2337: * If the control character has been set with `.cc', then let that grain
2338: * precedence.
2339: * This is slighly contrary to groff, where using the non-breaking
2340: * control character when `cc' has been invoked will cause the
2341: * non-breaking macro contents to be printed verbatim.
2342: */
2343: int
2344: roff_getcontrol(const struct roff *r, const char *cp, int *ppos)
2345: {
2346: int pos;
2347:
2348: pos = *ppos;
2349:
2350: if (0 != r->control && cp[pos] == r->control)
2351: pos++;
2352: else if (0 != r->control)
2353: return(0);
2354: else if ('\\' == cp[pos] && '.' == cp[pos + 1])
2355: pos += 2;
2356: else if ('.' == cp[pos] || '\'' == cp[pos])
2357: pos++;
2358: else
2359: return(0);
2360:
2361: while (' ' == cp[pos] || '\t' == cp[pos])
2362: pos++;
2363:
2364: *ppos = pos;
2365: return(1);
1.1 schwarze 2366: }
![[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