Annotation of src/usr.bin/mandoc/roff.c, Revision 1.99
1.99 ! schwarze 1: /* $OpenBSD$ */
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: }
1.2 schwarze 759:
760: /*
761: * If a scope is open, go to the child handler for that macro,
762: * as it may want to preprocess before doing anything with it.
1.32 schwarze 763: * Don't do so if an equation is open.
1.2 schwarze 764: */
765:
766: if (r->last) {
1.1 schwarze 767: t = r->last->tok;
768: assert(roffs[t].sub);
1.80 schwarze 769: return((*roffs[t].sub)(r, t, bufp, szp,
770: ln, ppos, pos, offs));
1.2 schwarze 771: }
772:
773: /*
774: * Lastly, as we've no scope open, try to look up and execute
775: * the new macro. If no macro is found, simply return and let
776: * the compilers handle it.
777: */
778:
1.87 schwarze 779: if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos, ln, ppos)))
1.1 schwarze 780: return(ROFF_CONT);
781:
1.2 schwarze 782: assert(roffs[t].proc);
1.80 schwarze 783: return((*roffs[t].proc)(r, t, bufp, szp, ln, ppos, pos, offs));
1.2 schwarze 784: }
785:
1.27 schwarze 786: void
1.2 schwarze 787: roff_endparse(struct roff *r)
788: {
1.1 schwarze 789:
1.27 schwarze 790: if (r->last)
1.94 schwarze 791: mandoc_msg(MANDOCERR_BLK_NOEND, r->parse,
792: r->last->line, r->last->col,
793: roffs[r->last->tok].name);
1.27 schwarze 794:
1.32 schwarze 795: if (r->eqn) {
1.94 schwarze 796: mandoc_msg(MANDOCERR_BLK_NOEND, r->parse,
797: r->eqn->eqn.ln, r->eqn->eqn.pos, "EQ");
1.41 schwarze 798: eqn_end(&r->eqn);
1.32 schwarze 799: }
800:
1.27 schwarze 801: if (r->tbl) {
1.94 schwarze 802: mandoc_msg(MANDOCERR_BLK_NOEND, r->parse,
803: r->tbl->line, r->tbl->pos, "TS");
1.41 schwarze 804: tbl_end(&r->tbl);
1.27 schwarze 805: }
1.1 schwarze 806: }
807:
808: /*
809: * Parse a roff node's type from the input buffer. This must be in the
810: * form of ".foo xxx" in the usual way.
811: */
812: static enum rofft
1.87 schwarze 813: roff_parse(struct roff *r, char *buf, int *pos, int ln, int ppos)
1.1 schwarze 814: {
1.87 schwarze 815: char *cp;
1.16 schwarze 816: const char *mac;
817: size_t maclen;
1.1 schwarze 818: enum rofft t;
819:
1.87 schwarze 820: cp = buf + *pos;
821:
822: if ('\0' == *cp || '"' == *cp || '\t' == *cp || ' ' == *cp)
1.1 schwarze 823: return(ROFF_MAX);
824:
1.87 schwarze 825: mac = cp;
826: maclen = roff_getname(r, &cp, ln, ppos);
1.1 schwarze 827:
1.16 schwarze 828: t = (r->current_string = roff_getstrn(r, mac, maclen))
1.42 schwarze 829: ? ROFF_USERDEF : roffhash_find(mac, maclen);
1.1 schwarze 830:
1.87 schwarze 831: if (ROFF_MAX != t)
832: *pos = cp - buf;
1.1 schwarze 833:
834: return(t);
835: }
836:
837: static enum rofferr
1.2 schwarze 838: roff_cblock(ROFF_ARGS)
1.1 schwarze 839: {
840:
1.2 schwarze 841: /*
842: * A block-close `..' should only be invoked as a child of an
843: * ignore macro, otherwise raise a warning and just ignore it.
844: */
845:
846: if (NULL == r->last) {
1.94 schwarze 847: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
848: ln, ppos, "..");
1.2 schwarze 849: return(ROFF_IGN);
850: }
1.1 schwarze 851:
1.2 schwarze 852: switch (r->last->tok) {
1.80 schwarze 853: case ROFF_am:
1.93 schwarze 854: /* ROFF_am1 is remapped to ROFF_am in roff_block(). */
1.2 schwarze 855: /* FALLTHROUGH */
1.80 schwarze 856: case ROFF_ami:
1.2 schwarze 857: /* FALLTHROUGH */
1.80 schwarze 858: case ROFF_de:
1.23 schwarze 859: /* ROFF_de1 is remapped to ROFF_de in roff_block(). */
1.2 schwarze 860: /* FALLTHROUGH */
1.80 schwarze 861: case ROFF_dei:
1.2 schwarze 862: /* FALLTHROUGH */
1.80 schwarze 863: case ROFF_ig:
1.2 schwarze 864: break;
865: default:
1.94 schwarze 866: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
867: ln, ppos, "..");
1.1 schwarze 868: return(ROFF_IGN);
1.2 schwarze 869: }
870:
871: if ((*bufp)[pos])
1.90 schwarze 872: mandoc_vmsg(MANDOCERR_ARG_SKIP, r->parse, ln, pos,
873: ".. %s", *bufp + pos);
1.2 schwarze 874:
875: roffnode_pop(r);
876: roffnode_cleanscope(r);
877: return(ROFF_IGN);
878:
879: }
1.1 schwarze 880:
1.2 schwarze 881: static void
882: roffnode_cleanscope(struct roff *r)
883: {
1.1 schwarze 884:
1.2 schwarze 885: while (r->last) {
1.46 schwarze 886: if (--r->last->endspan != 0)
1.2 schwarze 887: break;
888: roffnode_pop(r);
889: }
890: }
1.1 schwarze 891:
1.68 schwarze 892: static void
893: roff_ccond(struct roff *r, int ln, int ppos)
1.2 schwarze 894: {
1.1 schwarze 895:
1.2 schwarze 896: if (NULL == r->last) {
1.94 schwarze 897: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
898: ln, ppos, "\\}");
1.68 schwarze 899: return;
1.2 schwarze 900: }
1.1 schwarze 901:
1.2 schwarze 902: switch (r->last->tok) {
1.80 schwarze 903: case ROFF_el:
1.2 schwarze 904: /* FALLTHROUGH */
1.80 schwarze 905: case ROFF_ie:
1.2 schwarze 906: /* FALLTHROUGH */
1.80 schwarze 907: case ROFF_if:
1.2 schwarze 908: break;
909: default:
1.94 schwarze 910: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
911: ln, ppos, "\\}");
1.68 schwarze 912: return;
1.2 schwarze 913: }
1.1 schwarze 914:
1.2 schwarze 915: if (r->last->endspan > -1) {
1.94 schwarze 916: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
917: ln, ppos, "\\}");
1.68 schwarze 918: return;
1.2 schwarze 919: }
920:
921: roffnode_pop(r);
922: roffnode_cleanscope(r);
1.68 schwarze 923: return;
1.1 schwarze 924: }
925:
926: static enum rofferr
1.2 schwarze 927: roff_block(ROFF_ARGS)
1.1 schwarze 928: {
1.93 schwarze 929: const char *name;
930: char *iname, *cp;
1.86 schwarze 931: size_t namesz;
1.16 schwarze 932:
1.93 schwarze 933: /* Ignore groff compatibility mode for now. */
1.2 schwarze 934:
1.93 schwarze 935: if (ROFF_de1 == tok)
936: tok = ROFF_de;
937: else if (ROFF_am1 == tok)
938: tok = ROFF_am;
939:
940: /* Parse the macro name argument. */
941:
942: cp = *bufp + pos;
943: if (ROFF_ig == tok) {
944: iname = NULL;
945: namesz = 0;
946: } else {
947: iname = cp;
948: namesz = roff_getname(r, &cp, ln, ppos);
949: iname[namesz] = '\0';
950: }
1.22 schwarze 951:
1.93 schwarze 952: /* Resolve the macro name argument if it is indirect. */
1.22 schwarze 953:
1.93 schwarze 954: if (namesz && (ROFF_dei == tok || ROFF_ami == tok)) {
955: if (NULL == (name = roff_getstrn(r, iname, namesz))) {
956: mandoc_vmsg(MANDOCERR_STR_UNDEF,
957: r->parse, ln, (int)(iname - *bufp),
958: "%.*s", (int)namesz, iname);
959: namesz = 0;
960: } else
961: namesz = strlen(name);
962: } else
963: name = iname;
1.22 schwarze 964:
1.93 schwarze 965: if (0 == namesz && ROFF_ig != tok) {
966: mandoc_msg(MANDOCERR_REQ_EMPTY, r->parse,
967: ln, ppos, roffs[tok].name);
968: return(ROFF_IGN);
969: }
1.2 schwarze 970:
1.16 schwarze 971: roffnode_push(r, tok, name, ln, ppos);
972:
973: /*
974: * At the beginning of a `de' macro, clear the existing string
975: * with the same name, if there is one. New content will be
1.66 schwarze 976: * appended from roff_block_text() in multiline mode.
1.16 schwarze 977: */
1.22 schwarze 978:
1.93 schwarze 979: if (ROFF_de == tok || ROFF_dei == tok)
1.86 schwarze 980: roff_setstrn(&r->strtab, name, namesz, "", 0, 0);
1.2 schwarze 981:
1.86 schwarze 982: if ('\0' == *cp)
1.2 schwarze 983: return(ROFF_IGN);
1.1 schwarze 984:
1.93 schwarze 985: /* Get the custom end marker. */
1.22 schwarze 986:
1.93 schwarze 987: iname = cp;
1.86 schwarze 988: namesz = roff_getname(r, &cp, ln, ppos);
1.93 schwarze 989:
990: /* Resolve the end marker if it is indirect. */
991:
992: if (namesz && (ROFF_dei == tok || ROFF_ami == tok)) {
993: if (NULL == (name = roff_getstrn(r, iname, namesz))) {
994: mandoc_vmsg(MANDOCERR_STR_UNDEF,
995: r->parse, ln, (int)(iname - *bufp),
996: "%.*s", (int)namesz, iname);
997: namesz = 0;
998: } else
999: namesz = strlen(name);
1000: } else
1001: name = iname;
1002:
1.86 schwarze 1003: if (namesz)
1004: r->last->end = mandoc_strndup(name, namesz);
1.2 schwarze 1005:
1.86 schwarze 1006: if ('\0' != *cp)
1.90 schwarze 1007: mandoc_vmsg(MANDOCERR_ARG_EXCESS, r->parse,
1008: ln, pos, ".%s ... %s", roffs[tok].name, cp);
1.1 schwarze 1009:
1010: return(ROFF_IGN);
1011: }
1012:
1013: static enum rofferr
1.2 schwarze 1014: roff_block_sub(ROFF_ARGS)
1.1 schwarze 1015: {
1.2 schwarze 1016: enum rofft t;
1017: int i, j;
1018:
1019: /*
1020: * First check whether a custom macro exists at this level. If
1021: * it does, then check against it. This is some of groff's
1022: * stranger behaviours. If we encountered a custom end-scope
1023: * tag and that tag also happens to be a "real" macro, then we
1024: * need to try interpreting it again as a real macro. If it's
1025: * not, then return ignore. Else continue.
1026: */
1027:
1028: if (r->last->end) {
1.35 schwarze 1029: for (i = pos, j = 0; r->last->end[j]; j++, i++)
1.2 schwarze 1030: if ((*bufp)[i] != r->last->end[j])
1031: break;
1.1 schwarze 1032:
1.80 schwarze 1033: if ('\0' == r->last->end[j] &&
1034: ('\0' == (*bufp)[i] ||
1035: ' ' == (*bufp)[i] ||
1036: '\t' == (*bufp)[i])) {
1.2 schwarze 1037: roffnode_pop(r);
1038: roffnode_cleanscope(r);
1.1 schwarze 1039:
1.35 schwarze 1040: while (' ' == (*bufp)[i] || '\t' == (*bufp)[i])
1041: i++;
1042:
1043: pos = i;
1.87 schwarze 1044: if (ROFF_MAX != roff_parse(r, *bufp, &pos, ln, ppos))
1.2 schwarze 1045: return(ROFF_RERUN);
1046: return(ROFF_IGN);
1047: }
1.1 schwarze 1048: }
1049:
1.2 schwarze 1050: /*
1051: * If we have no custom end-query or lookup failed, then try
1052: * pulling it out of the hashtable.
1053: */
1.1 schwarze 1054:
1.87 schwarze 1055: t = roff_parse(r, *bufp, &pos, ln, ppos);
1.1 schwarze 1056:
1.16 schwarze 1057: if (ROFF_cblock != t) {
1.93 schwarze 1058: if (ROFF_ig != tok)
1.66 schwarze 1059: roff_setstr(r, r->last->name, *bufp + ppos, 2);
1.1 schwarze 1060: return(ROFF_IGN);
1.16 schwarze 1061: }
1.1 schwarze 1062:
1.2 schwarze 1063: assert(roffs[t].proc);
1.80 schwarze 1064: return((*roffs[t].proc)(r, t, bufp, szp, ln, ppos, pos, offs));
1.2 schwarze 1065: }
1066:
1067: static enum rofferr
1068: roff_block_text(ROFF_ARGS)
1069: {
1070:
1.93 schwarze 1071: if (ROFF_ig != tok)
1.66 schwarze 1072: roff_setstr(r, r->last->name, *bufp + pos, 2);
1.16 schwarze 1073:
1.2 schwarze 1074: return(ROFF_IGN);
1075: }
1076:
1077: static enum rofferr
1078: roff_cond_sub(ROFF_ARGS)
1079: {
1080: enum rofft t;
1.37 schwarze 1081: char *ep;
1.71 schwarze 1082: int rr;
1.2 schwarze 1083:
1084: rr = r->last->rule;
1.37 schwarze 1085: roffnode_cleanscope(r);
1.87 schwarze 1086: t = roff_parse(r, *bufp, &pos, ln, ppos);
1.2 schwarze 1087:
1.37 schwarze 1088: /*
1.50 schwarze 1089: * Fully handle known macros when they are structurally
1090: * required or when the conditional evaluated to true.
1.5 schwarze 1091: */
1092:
1.50 schwarze 1093: if ((ROFF_MAX != t) &&
1.71 schwarze 1094: (rr || ROFFMAC_STRUCT & roffs[t].flags)) {
1.50 schwarze 1095: assert(roffs[t].proc);
1096: return((*roffs[t].proc)(r, t, bufp, szp,
1.80 schwarze 1097: ln, ppos, pos, offs));
1.50 schwarze 1098: }
1.39 schwarze 1099:
1.69 schwarze 1100: /*
1101: * If `\}' occurs on a macro line without a preceding macro,
1102: * drop the line completely.
1103: */
1104:
1105: ep = *bufp + pos;
1106: if ('\\' == ep[0] && '}' == ep[1])
1.71 schwarze 1107: rr = 0;
1.69 schwarze 1108:
1.50 schwarze 1109: /* Always check for the closing delimiter `\}'. */
1.39 schwarze 1110:
1.50 schwarze 1111: while (NULL != (ep = strchr(ep, '\\'))) {
1.70 schwarze 1112: if ('}' == *(++ep)) {
1113: *ep = '&';
1114: roff_ccond(r, ln, ep - *bufp - 1);
1115: }
1116: ++ep;
1.50 schwarze 1117: }
1.71 schwarze 1118: return(rr ? ROFF_CONT : ROFF_IGN);
1.2 schwarze 1119: }
1120:
1121: static enum rofferr
1122: roff_cond_text(ROFF_ARGS)
1123: {
1.37 schwarze 1124: char *ep;
1.71 schwarze 1125: int rr;
1.2 schwarze 1126:
1127: rr = r->last->rule;
1.37 schwarze 1128: roffnode_cleanscope(r);
1.1 schwarze 1129:
1.70 schwarze 1130: ep = *bufp + pos;
1131: while (NULL != (ep = strchr(ep, '\\'))) {
1132: if ('}' == *(++ep)) {
1133: *ep = '&';
1134: roff_ccond(r, ln, ep - *bufp - 1);
1135: }
1136: ++ep;
1.2 schwarze 1137: }
1.71 schwarze 1138: return(rr ? ROFF_CONT : ROFF_IGN);
1.2 schwarze 1139: }
1140:
1.77 schwarze 1141: /*
1142: * Parse a single signed integer number. Stop at the first non-digit.
1143: * If there is at least one digit, return success and advance the
1144: * parse point, else return failure and let the parse point unchanged.
1145: * Ignore overflows, treat them just like the C language.
1146: */
1.56 schwarze 1147: static int
1148: roff_getnum(const char *v, int *pos, int *res)
1149: {
1.79 schwarze 1150: int myres, n, p;
1151:
1152: if (NULL == res)
1153: res = &myres;
1.56 schwarze 1154:
1155: p = *pos;
1156: n = v[p] == '-';
1157: if (n)
1158: p++;
1159:
1160: for (*res = 0; isdigit((unsigned char)v[p]); p++)
1.77 schwarze 1161: *res = 10 * *res + v[p] - '0';
1.56 schwarze 1162: if (p == *pos + n)
1163: return 0;
1164:
1165: if (n)
1166: *res = -*res;
1167:
1168: *pos = p;
1169: return 1;
1170: }
1171:
1.71 schwarze 1172: /*
1173: * Evaluate a string comparison condition.
1174: * The first character is the delimiter.
1175: * Succeed if the string up to its second occurrence
1176: * matches the string up to its third occurence.
1177: * Advance the cursor after the third occurrence
1178: * or lacking that, to the end of the line.
1179: */
1180: static int
1181: roff_evalstrcond(const char *v, int *pos)
1182: {
1183: const char *s1, *s2, *s3;
1184: int match;
1185:
1186: match = 0;
1187: s1 = v + *pos; /* initial delimiter */
1188: s2 = s1 + 1; /* for scanning the first string */
1189: s3 = strchr(s2, *s1); /* for scanning the second string */
1190:
1191: if (NULL == s3) /* found no middle delimiter */
1192: goto out;
1193:
1194: while ('\0' != *++s3) {
1195: if (*s2 != *s3) { /* mismatch */
1196: s3 = strchr(s3, *s1);
1197: break;
1198: }
1199: if (*s3 == *s1) { /* found the final delimiter */
1200: match = 1;
1201: break;
1202: }
1203: s2++;
1204: }
1205:
1206: out:
1207: if (NULL == s3)
1208: s3 = strchr(s2, '\0');
1209: else
1210: s3++;
1211: *pos = s3 - v;
1212: return(match);
1213: }
1214:
1.77 schwarze 1215: /*
1216: * Evaluate an optionally negated single character, numerical,
1217: * or string condition.
1218: */
1.71 schwarze 1219: static int
1.5 schwarze 1220: roff_evalcond(const char *v, int *pos)
1221: {
1.77 schwarze 1222: int wanttrue, number;
1.5 schwarze 1223:
1.71 schwarze 1224: if ('!' == v[*pos]) {
1225: wanttrue = 0;
1226: (*pos)++;
1227: } else
1228: wanttrue = 1;
1229:
1.5 schwarze 1230: switch (v[*pos]) {
1.80 schwarze 1231: case 'n':
1.71 schwarze 1232: /* FALLTHROUGH */
1.80 schwarze 1233: case 'o':
1.5 schwarze 1234: (*pos)++;
1.71 schwarze 1235: return(wanttrue);
1.80 schwarze 1236: case 'c':
1.71 schwarze 1237: /* FALLTHROUGH */
1.80 schwarze 1238: case 'd':
1.71 schwarze 1239: /* FALLTHROUGH */
1.80 schwarze 1240: case 'e':
1.5 schwarze 1241: /* FALLTHROUGH */
1.80 schwarze 1242: case 'r':
1.5 schwarze 1243: /* FALLTHROUGH */
1.80 schwarze 1244: case 't':
1.5 schwarze 1245: (*pos)++;
1.71 schwarze 1246: return(!wanttrue);
1.5 schwarze 1247: default:
1248: break;
1249: }
1250:
1.77 schwarze 1251: if (roff_evalnum(v, pos, &number, 0))
1252: return((number > 0) == wanttrue);
1253: else
1.71 schwarze 1254: return(roff_evalstrcond(v, pos) == wanttrue);
1.5 schwarze 1255: }
1256:
1.2 schwarze 1257: static enum rofferr
1.21 schwarze 1258: roff_line_ignore(ROFF_ARGS)
1.6 schwarze 1259: {
1.30 schwarze 1260:
1.21 schwarze 1261: return(ROFF_IGN);
1262: }
1263:
1264: static enum rofferr
1.2 schwarze 1265: roff_cond(ROFF_ARGS)
1266: {
1.46 schwarze 1267:
1268: roffnode_push(r, tok, NULL, ln, ppos);
1.2 schwarze 1269:
1.80 schwarze 1270: /*
1.35 schwarze 1271: * An `.el' has no conditional body: it will consume the value
1272: * of the current rstack entry set in prior `ie' calls or
1.80 schwarze 1273: * defaults to DENY.
1.35 schwarze 1274: *
1275: * If we're not an `el', however, then evaluate the conditional.
1276: */
1.1 schwarze 1277:
1.46 schwarze 1278: r->last->rule = ROFF_el == tok ?
1.80 schwarze 1279: (r->rstackpos < 0 ? 0 : r->rstack[r->rstackpos--]) :
1280: roff_evalcond(*bufp, &pos);
1.2 schwarze 1281:
1.35 schwarze 1282: /*
1283: * An if-else will put the NEGATION of the current evaluated
1284: * conditional into the stack of rules.
1285: */
1286:
1.2 schwarze 1287: if (ROFF_ie == tok) {
1.96 schwarze 1288: if (r->rstackpos + 1 == r->rstacksz) {
1289: r->rstacksz += 16;
1290: r->rstack = mandoc_reallocarray(r->rstack,
1291: r->rstacksz, sizeof(int));
1.35 schwarze 1292: }
1.71 schwarze 1293: r->rstack[++r->rstackpos] = !r->last->rule;
1.2 schwarze 1294: }
1.5 schwarze 1295:
1296: /* If the parent has false as its rule, then so do we. */
1297:
1.71 schwarze 1298: if (r->last->parent && !r->last->parent->rule)
1299: r->last->rule = 0;
1.5 schwarze 1300:
1301: /*
1.46 schwarze 1302: * Determine scope.
1303: * If there is nothing on the line after the conditional,
1304: * not even whitespace, use next-line scope.
1.5 schwarze 1305: */
1.2 schwarze 1306:
1.46 schwarze 1307: if ('\0' == (*bufp)[pos]) {
1308: r->last->endspan = 2;
1309: goto out;
1310: }
1311:
1312: while (' ' == (*bufp)[pos])
1313: pos++;
1314:
1315: /* An opening brace requests multiline scope. */
1.2 schwarze 1316:
1317: if ('\\' == (*bufp)[pos] && '{' == (*bufp)[pos + 1]) {
1318: r->last->endspan = -1;
1319: pos += 2;
1.46 schwarze 1320: goto out;
1.80 schwarze 1321: }
1.2 schwarze 1322:
1323: /*
1.46 schwarze 1324: * Anything else following the conditional causes
1325: * single-line scope. Warn if the scope contains
1326: * nothing but trailing whitespace.
1.2 schwarze 1327: */
1328:
1329: if ('\0' == (*bufp)[pos])
1.89 schwarze 1330: mandoc_msg(MANDOCERR_COND_EMPTY, r->parse,
1331: ln, ppos, roffs[tok].name);
1.2 schwarze 1332:
1.46 schwarze 1333: r->last->endspan = 1;
1.1 schwarze 1334:
1.46 schwarze 1335: out:
1.2 schwarze 1336: *offs = pos;
1337: return(ROFF_RERUN);
1.1 schwarze 1338: }
1339:
1.2 schwarze 1340: static enum rofferr
1.7 schwarze 1341: roff_ds(ROFF_ARGS)
1342: {
1.85 schwarze 1343: char *string;
1344: const char *name;
1345: size_t namesz;
1.10 schwarze 1346:
1347: /*
1.85 schwarze 1348: * The first word is the name of the string.
1349: * If it is empty or terminated by an escape sequence,
1350: * abort the `ds' request without defining anything.
1.10 schwarze 1351: */
1.7 schwarze 1352:
1.85 schwarze 1353: name = string = *bufp + pos;
1.7 schwarze 1354: if ('\0' == *name)
1355: return(ROFF_IGN);
1356:
1.85 schwarze 1357: namesz = roff_getname(r, &string, ln, pos);
1358: if ('\\' == name[namesz])
1359: return(ROFF_IGN);
1360:
1361: /* Read past the initial double-quote, if any. */
1.28 schwarze 1362: if ('"' == *string)
1.7 schwarze 1363: string++;
1364:
1.10 schwarze 1365: /* The rest is the value. */
1.85 schwarze 1366: roff_setstrn(&r->strtab, name, namesz, string, strlen(string),
1367: ROFF_as == tok);
1.7 schwarze 1368: return(ROFF_IGN);
1369: }
1370:
1.77 schwarze 1371: /*
1372: * Parse a single operator, one or two characters long.
1373: * If the operator is recognized, return success and advance the
1374: * parse point, else return failure and let the parse point unchanged.
1375: */
1376: static int
1377: roff_getop(const char *v, int *pos, char *res)
1378: {
1379:
1380: *res = v[*pos];
1381:
1382: switch (*res) {
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: /* FALLTHROUGH */
1.80 schwarze 1391: case '%':
1.77 schwarze 1392: /* FALLTHROUGH */
1.80 schwarze 1393: case '&':
1.77 schwarze 1394: /* FALLTHROUGH */
1.80 schwarze 1395: case ':':
1.77 schwarze 1396: break;
1397: case '<':
1398: switch (v[*pos + 1]) {
1.80 schwarze 1399: case '=':
1.77 schwarze 1400: *res = 'l';
1401: (*pos)++;
1402: break;
1.80 schwarze 1403: case '>':
1.77 schwarze 1404: *res = '!';
1405: (*pos)++;
1406: break;
1.80 schwarze 1407: case '?':
1.77 schwarze 1408: *res = 'i';
1409: (*pos)++;
1410: break;
1411: default:
1412: break;
1413: }
1414: break;
1415: case '>':
1416: switch (v[*pos + 1]) {
1.80 schwarze 1417: case '=':
1.77 schwarze 1418: *res = 'g';
1419: (*pos)++;
1420: break;
1.80 schwarze 1421: case '?':
1.77 schwarze 1422: *res = 'a';
1423: (*pos)++;
1424: break;
1425: default:
1426: break;
1427: }
1428: break;
1429: case '=':
1430: if ('=' == v[*pos + 1])
1431: (*pos)++;
1432: break;
1433: default:
1434: return(0);
1435: }
1436: (*pos)++;
1437:
1438: return(*res);
1439: }
1440:
1441: /*
1442: * Evaluate either a parenthesized numeric expression
1443: * or a single signed integer number.
1444: */
1445: static int
1446: roff_evalpar(const char *v, int *pos, int *res)
1447: {
1448:
1449: if ('(' != v[*pos])
1450: return(roff_getnum(v, pos, res));
1451:
1452: (*pos)++;
1453: if ( ! roff_evalnum(v, pos, res, 1))
1454: return(0);
1455:
1.79 schwarze 1456: /*
1457: * Omission of the closing parenthesis
1458: * is an error in validation mode,
1459: * but ignored in evaluation mode.
1460: */
1461:
1.77 schwarze 1462: if (')' == v[*pos])
1463: (*pos)++;
1.79 schwarze 1464: else if (NULL == res)
1465: return(0);
1.77 schwarze 1466:
1467: return(1);
1468: }
1469:
1470: /*
1471: * Evaluate a complete numeric expression.
1472: * Proceed left to right, there is no concept of precedence.
1473: */
1474: static int
1475: roff_evalnum(const char *v, int *pos, int *res, int skipwhite)
1476: {
1477: int mypos, operand2;
1478: char operator;
1479:
1480: if (NULL == pos) {
1481: mypos = 0;
1482: pos = &mypos;
1483: }
1484:
1485: if (skipwhite)
1486: while (isspace((unsigned char)v[*pos]))
1487: (*pos)++;
1488:
1489: if ( ! roff_evalpar(v, pos, res))
1490: return(0);
1491:
1492: while (1) {
1493: if (skipwhite)
1494: while (isspace((unsigned char)v[*pos]))
1495: (*pos)++;
1496:
1497: if ( ! roff_getop(v, pos, &operator))
1498: break;
1499:
1500: if (skipwhite)
1501: while (isspace((unsigned char)v[*pos]))
1502: (*pos)++;
1503:
1504: if ( ! roff_evalpar(v, pos, &operand2))
1505: return(0);
1506:
1507: if (skipwhite)
1508: while (isspace((unsigned char)v[*pos]))
1509: (*pos)++;
1.79 schwarze 1510:
1511: if (NULL == res)
1512: continue;
1.77 schwarze 1513:
1514: switch (operator) {
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 /= 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 = *res < operand2;
1532: break;
1.80 schwarze 1533: case '>':
1.77 schwarze 1534: *res = *res > operand2;
1535: break;
1.80 schwarze 1536: case 'l':
1.77 schwarze 1537: *res = *res <= operand2;
1538: break;
1.80 schwarze 1539: case 'g':
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 '&':
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 'i':
1.77 schwarze 1555: if (operand2 < *res)
1556: *res = operand2;
1557: break;
1.80 schwarze 1558: case 'a':
1.77 schwarze 1559: if (operand2 > *res)
1560: *res = operand2;
1561: break;
1562: default:
1563: abort();
1564: }
1565: }
1566: return(1);
1567: }
1568:
1.52 schwarze 1569: void
1.60 schwarze 1570: roff_setreg(struct roff *r, const char *name, int val, char sign)
1.41 schwarze 1571: {
1.52 schwarze 1572: struct roffreg *reg;
1573:
1574: /* Search for an existing register with the same name. */
1575: reg = r->regtab;
1576:
1577: while (reg && strcmp(name, reg->key.p))
1578: reg = reg->next;
1.41 schwarze 1579:
1.52 schwarze 1580: if (NULL == reg) {
1581: /* Create a new register. */
1582: reg = mandoc_malloc(sizeof(struct roffreg));
1583: reg->key.p = mandoc_strdup(name);
1584: reg->key.sz = strlen(name);
1.60 schwarze 1585: reg->val = 0;
1.52 schwarze 1586: reg->next = r->regtab;
1587: r->regtab = reg;
1588: }
1589:
1.60 schwarze 1590: if ('+' == sign)
1591: reg->val += val;
1592: else if ('-' == sign)
1593: reg->val -= val;
1594: else
1595: reg->val = val;
1.41 schwarze 1596: }
1597:
1.65 schwarze 1598: /*
1599: * Handle some predefined read-only number registers.
1600: * For now, return -1 if the requested register is not predefined;
1601: * in case a predefined read-only register having the value -1
1602: * were to turn up, another special value would have to be chosen.
1603: */
1604: static int
1605: roff_getregro(const char *name)
1606: {
1607:
1608: switch (*name) {
1.80 schwarze 1609: case 'A': /* ASCII approximation mode is always off. */
1.65 schwarze 1610: return(0);
1.80 schwarze 1611: case 'g': /* Groff compatibility mode is always on. */
1.65 schwarze 1612: return(1);
1.80 schwarze 1613: case 'H': /* Fixed horizontal resolution. */
1.65 schwarze 1614: return (24);
1.80 schwarze 1615: case 'j': /* Always adjust left margin only. */
1.65 schwarze 1616: return(0);
1.80 schwarze 1617: case 'T': /* Some output device is always defined. */
1.65 schwarze 1618: return(1);
1.80 schwarze 1619: case 'V': /* Fixed vertical resolution. */
1.65 schwarze 1620: return (40);
1621: default:
1622: return (-1);
1623: }
1624: }
1625:
1.53 schwarze 1626: int
1.52 schwarze 1627: roff_getreg(const struct roff *r, const char *name)
1.41 schwarze 1628: {
1.52 schwarze 1629: struct roffreg *reg;
1.65 schwarze 1630: int val;
1631:
1632: if ('.' == name[0] && '\0' != name[1] && '\0' == name[2]) {
1633: val = roff_getregro(name + 1);
1634: if (-1 != val)
1635: return (val);
1636: }
1.52 schwarze 1637:
1638: for (reg = r->regtab; reg; reg = reg->next)
1639: if (0 == strcmp(name, reg->key.p))
1.53 schwarze 1640: return(reg->val);
1641:
1642: return(0);
1643: }
1644:
1645: static int
1646: roff_getregn(const struct roff *r, const char *name, size_t len)
1647: {
1648: struct roffreg *reg;
1.65 schwarze 1649: int val;
1650:
1651: if ('.' == name[0] && 2 == len) {
1652: val = roff_getregro(name + 1);
1653: if (-1 != val)
1654: return (val);
1655: }
1.53 schwarze 1656:
1657: for (reg = r->regtab; reg; reg = reg->next)
1658: if (len == reg->key.sz &&
1659: 0 == strncmp(name, reg->key.p, len))
1660: return(reg->val);
1.41 schwarze 1661:
1.52 schwarze 1662: return(0);
1.41 schwarze 1663: }
1664:
1.52 schwarze 1665: static void
1666: roff_freereg(struct roffreg *reg)
1.41 schwarze 1667: {
1.52 schwarze 1668: struct roffreg *old_reg;
1.41 schwarze 1669:
1.52 schwarze 1670: while (NULL != reg) {
1671: free(reg->key.p);
1672: old_reg = reg;
1673: reg = reg->next;
1674: free(old_reg);
1675: }
1.41 schwarze 1676: }
1.7 schwarze 1677:
1678: static enum rofferr
1.6 schwarze 1679: roff_nr(ROFF_ARGS)
1.1 schwarze 1680: {
1.85 schwarze 1681: char *key, *val;
1682: size_t keysz;
1.37 schwarze 1683: int iv;
1.60 schwarze 1684: char sign;
1.6 schwarze 1685:
1.85 schwarze 1686: key = val = *bufp + pos;
1687: if ('\0' == *key)
1688: return(ROFF_IGN);
1689:
1690: keysz = roff_getname(r, &val, ln, pos);
1691: if ('\\' == key[keysz])
1692: return(ROFF_IGN);
1693: key[keysz] = '\0';
1.6 schwarze 1694:
1.60 schwarze 1695: sign = *val;
1696: if ('+' == sign || '-' == sign)
1697: val++;
1698:
1.77 schwarze 1699: if (roff_evalnum(val, NULL, &iv, 0))
1700: roff_setreg(r, key, iv, sign);
1.1 schwarze 1701:
1.76 schwarze 1702: return(ROFF_IGN);
1703: }
1704:
1705: static enum rofferr
1706: roff_rr(ROFF_ARGS)
1707: {
1708: struct roffreg *reg, **prev;
1.85 schwarze 1709: char *name, *cp;
1710: size_t namesz;
1.76 schwarze 1711:
1.85 schwarze 1712: name = cp = *bufp + pos;
1713: if ('\0' == *name)
1714: return(ROFF_IGN);
1715: namesz = roff_getname(r, &cp, ln, pos);
1716: name[namesz] = '\0';
1.76 schwarze 1717:
1718: prev = &r->regtab;
1719: while (1) {
1720: reg = *prev;
1721: if (NULL == reg || !strcmp(name, reg->key.p))
1722: break;
1723: prev = ®->next;
1724: }
1725: if (NULL != reg) {
1726: *prev = reg->next;
1727: free(reg->key.p);
1728: free(reg);
1729: }
1.29 schwarze 1730: return(ROFF_IGN);
1731: }
1732:
1733: static enum rofferr
1734: roff_rm(ROFF_ARGS)
1735: {
1736: const char *name;
1737: char *cp;
1.85 schwarze 1738: size_t namesz;
1.29 schwarze 1739:
1740: cp = *bufp + pos;
1741: while ('\0' != *cp) {
1.85 schwarze 1742: name = cp;
1743: namesz = roff_getname(r, &cp, ln, (int)(cp - *bufp));
1744: roff_setstrn(&r->strtab, name, namesz, NULL, 0, 0);
1745: if ('\\' == name[namesz])
1746: break;
1.29 schwarze 1747: }
1.51 schwarze 1748: return(ROFF_IGN);
1749: }
1750:
1751: static enum rofferr
1752: roff_it(ROFF_ARGS)
1753: {
1754: char *cp;
1755: size_t len;
1756: int iv;
1757:
1758: /* Parse the number of lines. */
1759: cp = *bufp + pos;
1760: len = strcspn(cp, " \t");
1761: cp[len] = '\0';
1762: if ((iv = mandoc_strntoi(cp, len, 10)) <= 0) {
1.95 schwarze 1763: mandoc_msg(MANDOCERR_IT_NONUM, r->parse,
1.80 schwarze 1764: ln, ppos, *bufp + 1);
1.51 schwarze 1765: return(ROFF_IGN);
1766: }
1767: cp += len + 1;
1768:
1769: /* Arm the input line trap. */
1770: roffit_lines = iv;
1771: roffit_macro = mandoc_strdup(cp);
1.2 schwarze 1772: return(ROFF_IGN);
1.47 schwarze 1773: }
1774:
1775: static enum rofferr
1776: roff_Dd(ROFF_ARGS)
1777: {
1778: const char *const *cp;
1779:
1.99 ! schwarze 1780: if ((r->options & (MPARSE_MDOC | MPARSE_QUICK)) == 0)
1.47 schwarze 1781: for (cp = __mdoc_reserved; *cp; cp++)
1782: roff_setstr(r, *cp, NULL, 0);
1783:
1.99 ! schwarze 1784: if (r->format == 0)
! 1785: r->format = MPARSE_MDOC;
! 1786:
1.47 schwarze 1787: return(ROFF_CONT);
1788: }
1789:
1790: static enum rofferr
1791: roff_TH(ROFF_ARGS)
1792: {
1793: const char *const *cp;
1794:
1.99 ! schwarze 1795: if ((r->options & MPARSE_QUICK) == 0)
1.47 schwarze 1796: for (cp = __man_reserved; *cp; cp++)
1797: roff_setstr(r, *cp, NULL, 0);
1798:
1.99 ! schwarze 1799: if (r->format == 0)
! 1800: r->format = MPARSE_MAN;
! 1801:
1.47 schwarze 1802: return(ROFF_CONT);
1.14 schwarze 1803: }
1804:
1805: static enum rofferr
1.27 schwarze 1806: roff_TE(ROFF_ARGS)
1807: {
1808:
1809: if (NULL == r->tbl)
1.94 schwarze 1810: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
1811: ln, ppos, "TE");
1.27 schwarze 1812: else
1.41 schwarze 1813: tbl_end(&r->tbl);
1.27 schwarze 1814:
1815: return(ROFF_IGN);
1816: }
1817:
1818: static enum rofferr
1819: roff_T_(ROFF_ARGS)
1820: {
1821:
1822: if (NULL == r->tbl)
1.94 schwarze 1823: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
1824: ln, ppos, "T&");
1.27 schwarze 1825: else
1826: tbl_restart(ppos, ln, r->tbl);
1827:
1828: return(ROFF_IGN);
1829: }
1830:
1.41 schwarze 1831: #if 0
1832: static int
1833: roff_closeeqn(struct roff *r)
1834: {
1835:
1836: return(r->eqn && ROFF_EQN == eqn_end(&r->eqn) ? 1 : 0);
1837: }
1838: #endif
1839:
1840: static void
1.80 schwarze 1841: roff_openeqn(struct roff *r, const char *name, int line,
1.41 schwarze 1842: int offs, const char *buf)
1.32 schwarze 1843: {
1.41 schwarze 1844: struct eqn_node *e;
1845: int poff;
1.32 schwarze 1846:
1847: assert(NULL == r->eqn);
1.41 schwarze 1848: e = eqn_alloc(name, offs, line, r->parse);
1.32 schwarze 1849:
1850: if (r->last_eqn)
1851: r->last_eqn->next = e;
1852: else
1853: r->first_eqn = r->last_eqn = e;
1854:
1855: r->eqn = r->last_eqn = e;
1.41 schwarze 1856:
1857: if (buf) {
1858: poff = 0;
1859: eqn_read(&r->eqn, line, buf, offs, &poff);
1860: }
1861: }
1862:
1863: static enum rofferr
1864: roff_EQ(ROFF_ARGS)
1865: {
1866:
1867: roff_openeqn(r, *bufp + pos, ln, ppos, NULL);
1.32 schwarze 1868: return(ROFF_IGN);
1869: }
1870:
1871: static enum rofferr
1872: roff_EN(ROFF_ARGS)
1873: {
1874:
1.94 schwarze 1875: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse, ln, ppos, "EN");
1.32 schwarze 1876: return(ROFF_IGN);
1877: }
1878:
1879: static enum rofferr
1.27 schwarze 1880: roff_TS(ROFF_ARGS)
1881: {
1.49 schwarze 1882: struct tbl_node *tbl;
1.27 schwarze 1883:
1884: if (r->tbl) {
1.94 schwarze 1885: mandoc_msg(MANDOCERR_BLK_BROKEN, r->parse,
1886: ln, ppos, "TS breaks TS");
1.41 schwarze 1887: tbl_end(&r->tbl);
1.27 schwarze 1888: }
1889:
1.49 schwarze 1890: tbl = tbl_alloc(ppos, ln, r->parse);
1.27 schwarze 1891:
1892: if (r->last_tbl)
1.49 schwarze 1893: r->last_tbl->next = tbl;
1.27 schwarze 1894: else
1.49 schwarze 1895: r->first_tbl = r->last_tbl = tbl;
1.27 schwarze 1896:
1.49 schwarze 1897: r->tbl = r->last_tbl = tbl;
1.27 schwarze 1898: return(ROFF_IGN);
1899: }
1900:
1901: static enum rofferr
1.48 schwarze 1902: roff_cc(ROFF_ARGS)
1903: {
1904: const char *p;
1905:
1906: p = *bufp + pos;
1907:
1908: if ('\0' == *p || '.' == (r->control = *p++))
1909: r->control = 0;
1910:
1911: if ('\0' != *p)
1912: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
1913:
1914: return(ROFF_IGN);
1915: }
1916:
1917: static enum rofferr
1.42 schwarze 1918: roff_tr(ROFF_ARGS)
1919: {
1920: const char *p, *first, *second;
1921: size_t fsz, ssz;
1922: enum mandoc_esc esc;
1923:
1924: p = *bufp + pos;
1925:
1926: if ('\0' == *p) {
1927: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
1928: return(ROFF_IGN);
1929: }
1930:
1931: while ('\0' != *p) {
1932: fsz = ssz = 1;
1933:
1934: first = p++;
1935: if ('\\' == *first) {
1936: esc = mandoc_escape(&p, NULL, NULL);
1937: if (ESCAPE_ERROR == esc) {
1.92 schwarze 1938: mandoc_msg(MANDOCERR_ESC_BAD, r->parse,
1939: ln, (int)(p - *bufp), first);
1.42 schwarze 1940: return(ROFF_IGN);
1941: }
1942: fsz = (size_t)(p - first);
1943: }
1944:
1945: second = p++;
1946: if ('\\' == *second) {
1947: esc = mandoc_escape(&p, NULL, NULL);
1948: if (ESCAPE_ERROR == esc) {
1.92 schwarze 1949: mandoc_msg(MANDOCERR_ESC_BAD, r->parse,
1950: ln, (int)(p - *bufp), second);
1.42 schwarze 1951: return(ROFF_IGN);
1952: }
1953: ssz = (size_t)(p - second);
1954: } else if ('\0' == *second) {
1.80 schwarze 1955: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse,
1956: ln, (int)(p - *bufp), NULL);
1.42 schwarze 1957: second = " ";
1958: p--;
1959: }
1960:
1961: if (fsz > 1) {
1.80 schwarze 1962: roff_setstrn(&r->xmbtab, first, fsz,
1963: second, ssz, 0);
1.42 schwarze 1964: continue;
1965: }
1966:
1967: if (NULL == r->xtab)
1.80 schwarze 1968: r->xtab = mandoc_calloc(128,
1969: sizeof(struct roffstr));
1.42 schwarze 1970:
1971: free(r->xtab[(int)*first].p);
1972: r->xtab[(int)*first].p = mandoc_strndup(second, ssz);
1973: r->xtab[(int)*first].sz = ssz;
1974: }
1975:
1976: return(ROFF_IGN);
1977: }
1978:
1979: static enum rofferr
1.14 schwarze 1980: roff_so(ROFF_ARGS)
1981: {
1982: char *name;
1.15 schwarze 1983:
1.83 schwarze 1984: name = *bufp + pos;
1.97 schwarze 1985: mandoc_vmsg(MANDOCERR_SO, r->parse, ln, ppos, "so %s", name);
1.14 schwarze 1986:
1.22 schwarze 1987: /*
1988: * Handle `so'. Be EXTREMELY careful, as we shouldn't be
1989: * opening anything that's not in our cwd or anything beneath
1990: * it. Thus, explicitly disallow traversing up the file-system
1991: * or using absolute paths.
1992: */
1993:
1.14 schwarze 1994: if ('/' == *name || strstr(name, "../") || strstr(name, "/..")) {
1.83 schwarze 1995: mandoc_vmsg(MANDOCERR_SO_PATH, r->parse, ln, ppos,
1996: ".so %s", name);
1.14 schwarze 1997: return(ROFF_ERR);
1998: }
1999:
2000: *offs = pos;
2001: return(ROFF_SO);
1.7 schwarze 2002: }
2003:
1.16 schwarze 2004: static enum rofferr
2005: roff_userdef(ROFF_ARGS)
1.12 schwarze 2006: {
1.16 schwarze 2007: const char *arg[9];
2008: char *cp, *n1, *n2;
1.25 schwarze 2009: int i;
1.12 schwarze 2010:
1.16 schwarze 2011: /*
2012: * Collect pointers to macro argument strings
1.61 schwarze 2013: * and NUL-terminate them.
1.16 schwarze 2014: */
2015: cp = *bufp + pos;
1.25 schwarze 2016: for (i = 0; i < 9; i++)
1.26 schwarze 2017: arg[i] = '\0' == *cp ? "" :
1.35 schwarze 2018: mandoc_getarg(r->parse, &cp, ln, &pos);
1.16 schwarze 2019:
2020: /*
2021: * Expand macro arguments.
1.12 schwarze 2022: */
1.16 schwarze 2023: *szp = 0;
2024: n1 = cp = mandoc_strdup(r->current_string);
2025: while (NULL != (cp = strstr(cp, "\\$"))) {
2026: i = cp[2] - '1';
2027: if (0 > i || 8 < i) {
2028: /* Not an argument invocation. */
2029: cp += 2;
2030: continue;
2031: }
1.82 schwarze 2032: *cp = '\0';
2033: *szp = mandoc_asprintf(&n2, "%s%s%s",
2034: n1, arg[i], cp + 3) + 1;
1.16 schwarze 2035: cp = n2 + (cp - n1);
2036: free(n1);
2037: n1 = n2;
1.12 schwarze 2038: }
2039:
1.16 schwarze 2040: /*
2041: * Replace the macro invocation
2042: * by the expanded macro.
2043: */
2044: free(*bufp);
2045: *bufp = n1;
2046: if (0 == *szp)
2047: *szp = strlen(*bufp) + 1;
2048:
1.19 schwarze 2049: return(*szp > 1 && '\n' == (*bufp)[(int)*szp - 2] ?
1.16 schwarze 2050: ROFF_REPARSE : ROFF_APPEND);
1.12 schwarze 2051: }
1.28 schwarze 2052:
1.85 schwarze 2053: static size_t
1.28 schwarze 2054: roff_getname(struct roff *r, char **cpp, int ln, int pos)
2055: {
2056: char *name, *cp;
1.85 schwarze 2057: size_t namesz;
1.28 schwarze 2058:
2059: name = *cpp;
2060: if ('\0' == *name)
1.85 schwarze 2061: return(0);
1.28 schwarze 2062:
1.85 schwarze 2063: /* Read until end of name and terminate it with NUL. */
2064: for (cp = name; 1; cp++) {
2065: if ('\0' == *cp || ' ' == *cp) {
2066: namesz = cp - name;
2067: break;
2068: }
1.28 schwarze 2069: if ('\\' != *cp)
2070: continue;
1.88 schwarze 2071: namesz = cp - name;
2072: if ('{' == cp[1] || '}' == cp[1])
2073: break;
1.28 schwarze 2074: cp++;
2075: if ('\\' == *cp)
2076: continue;
1.97 schwarze 2077: mandoc_vmsg(MANDOCERR_NAMESC, r->parse, ln, pos,
2078: "%.*s", (int)(cp - name + 1), name);
1.85 schwarze 2079: mandoc_escape((const char **)&cp, NULL, NULL);
2080: break;
1.28 schwarze 2081: }
2082:
2083: /* Read past spaces. */
2084: while (' ' == *cp)
2085: cp++;
2086:
2087: *cpp = cp;
1.85 schwarze 2088: return(namesz);
1.28 schwarze 2089: }
2090:
1.16 schwarze 2091: /*
2092: * Store *string into the user-defined string called *name.
2093: * To clear an existing entry, call with (*r, *name, NULL, 0).
1.66 schwarze 2094: * append == 0: replace mode
2095: * append == 1: single-line append mode
2096: * append == 2: multiline append mode, append '\n' after each call
1.16 schwarze 2097: */
1.8 schwarze 2098: static void
1.16 schwarze 2099: roff_setstr(struct roff *r, const char *name, const char *string,
1.66 schwarze 2100: int append)
1.7 schwarze 2101: {
1.42 schwarze 2102:
2103: roff_setstrn(&r->strtab, name, strlen(name), string,
1.80 schwarze 2104: string ? strlen(string) : 0, append);
1.42 schwarze 2105: }
2106:
2107: static void
2108: roff_setstrn(struct roffkv **r, const char *name, size_t namesz,
1.66 schwarze 2109: const char *string, size_t stringsz, int append)
1.42 schwarze 2110: {
2111: struct roffkv *n;
2112: char *c;
2113: int i;
2114: size_t oldch, newch;
1.7 schwarze 2115:
1.16 schwarze 2116: /* Search for an existing string with the same name. */
1.42 schwarze 2117: n = *r;
2118:
1.84 schwarze 2119: while (n && (namesz != n->key.sz ||
2120: strncmp(n->key.p, name, namesz)))
1.7 schwarze 2121: n = n->next;
1.8 schwarze 2122:
2123: if (NULL == n) {
1.16 schwarze 2124: /* Create a new string table entry. */
1.42 schwarze 2125: n = mandoc_malloc(sizeof(struct roffkv));
2126: n->key.p = mandoc_strndup(name, namesz);
2127: n->key.sz = namesz;
2128: n->val.p = NULL;
2129: n->val.sz = 0;
2130: n->next = *r;
2131: *r = n;
1.66 schwarze 2132: } else if (0 == append) {
1.42 schwarze 2133: free(n->val.p);
2134: n->val.p = NULL;
2135: n->val.sz = 0;
1.16 schwarze 2136: }
2137:
2138: if (NULL == string)
2139: return;
2140:
2141: /*
2142: * One additional byte for the '\n' in multiline mode,
2143: * and one for the terminating '\0'.
2144: */
1.66 schwarze 2145: newch = stringsz + (1 < append ? 2u : 1u);
1.42 schwarze 2146:
2147: if (NULL == n->val.p) {
2148: n->val.p = mandoc_malloc(newch);
2149: *n->val.p = '\0';
1.16 schwarze 2150: oldch = 0;
2151: } else {
1.42 schwarze 2152: oldch = n->val.sz;
2153: n->val.p = mandoc_realloc(n->val.p, oldch + newch);
1.16 schwarze 2154: }
2155:
2156: /* Skip existing content in the destination buffer. */
1.42 schwarze 2157: c = n->val.p + (int)oldch;
1.16 schwarze 2158:
2159: /* Append new content to the destination buffer. */
1.42 schwarze 2160: i = 0;
2161: while (i < (int)stringsz) {
1.16 schwarze 2162: /*
2163: * Rudimentary roff copy mode:
2164: * Handle escaped backslashes.
2165: */
1.42 schwarze 2166: if ('\\' == string[i] && '\\' == string[i + 1])
2167: i++;
2168: *c++ = string[i++];
1.16 schwarze 2169: }
1.8 schwarze 2170:
1.16 schwarze 2171: /* Append terminating bytes. */
1.66 schwarze 2172: if (1 < append)
1.16 schwarze 2173: *c++ = '\n';
1.42 schwarze 2174:
1.16 schwarze 2175: *c = '\0';
1.42 schwarze 2176: n->val.sz = (int)(c - n->val.p);
1.7 schwarze 2177: }
2178:
1.8 schwarze 2179: static const char *
2180: roff_getstrn(const struct roff *r, const char *name, size_t len)
1.7 schwarze 2181: {
1.42 schwarze 2182: const struct roffkv *n;
1.64 schwarze 2183: int i;
1.7 schwarze 2184:
1.42 schwarze 2185: for (n = r->strtab; n; n = n->next)
1.80 schwarze 2186: if (0 == strncmp(name, n->key.p, len) &&
2187: '\0' == n->key.p[(int)len])
1.42 schwarze 2188: return(n->val.p);
1.64 schwarze 2189:
2190: for (i = 0; i < PREDEFS_MAX; i++)
2191: if (0 == strncmp(name, predefs[i].name, len) &&
2192: '\0' == predefs[i].name[(int)len])
2193: return(predefs[i].str);
1.8 schwarze 2194:
1.42 schwarze 2195: return(NULL);
1.7 schwarze 2196: }
2197:
1.8 schwarze 2198: static void
1.42 schwarze 2199: roff_freestr(struct roffkv *r)
1.7 schwarze 2200: {
1.42 schwarze 2201: struct roffkv *n, *nn;
1.7 schwarze 2202:
1.42 schwarze 2203: for (n = r; n; n = nn) {
2204: free(n->key.p);
2205: free(n->val.p);
1.7 schwarze 2206: nn = n->next;
2207: free(n);
2208: }
1.27 schwarze 2209: }
2210:
2211: const struct tbl_span *
2212: roff_span(const struct roff *r)
2213: {
1.80 schwarze 2214:
1.27 schwarze 2215: return(r->tbl ? tbl_span(r->tbl) : NULL);
1.32 schwarze 2216: }
2217:
2218: const struct eqn *
2219: roff_eqn(const struct roff *r)
2220: {
1.80 schwarze 2221:
1.32 schwarze 2222: return(r->last_eqn ? &r->last_eqn->eqn : NULL);
1.42 schwarze 2223: }
2224:
2225: /*
2226: * Duplicate an input string, making the appropriate character
2227: * conversations (as stipulated by `tr') along the way.
2228: * Returns a heap-allocated string with all the replacements made.
2229: */
2230: char *
2231: roff_strdup(const struct roff *r, const char *p)
2232: {
2233: const struct roffkv *cp;
2234: char *res;
2235: const char *pp;
2236: size_t ssz, sz;
2237: enum mandoc_esc esc;
2238:
2239: if (NULL == r->xmbtab && NULL == r->xtab)
2240: return(mandoc_strdup(p));
2241: else if ('\0' == *p)
2242: return(mandoc_strdup(""));
2243:
2244: /*
2245: * Step through each character looking for term matches
2246: * (remember that a `tr' can be invoked with an escape, which is
2247: * a glyph but the escape is multi-character).
2248: * We only do this if the character hash has been initialised
2249: * and the string is >0 length.
2250: */
2251:
2252: res = NULL;
2253: ssz = 0;
2254:
2255: while ('\0' != *p) {
2256: if ('\\' != *p && r->xtab && r->xtab[(int)*p].p) {
2257: sz = r->xtab[(int)*p].sz;
2258: res = mandoc_realloc(res, ssz + sz + 1);
2259: memcpy(res + ssz, r->xtab[(int)*p].p, sz);
2260: ssz += sz;
2261: p++;
2262: continue;
2263: } else if ('\\' != *p) {
2264: res = mandoc_realloc(res, ssz + 2);
2265: res[ssz++] = *p++;
2266: continue;
2267: }
2268:
2269: /* Search for term matches. */
2270: for (cp = r->xmbtab; cp; cp = cp->next)
2271: if (0 == strncmp(p, cp->key.p, cp->key.sz))
2272: break;
2273:
2274: if (NULL != cp) {
2275: /*
2276: * A match has been found.
2277: * Append the match to the array and move
2278: * forward by its keysize.
2279: */
1.80 schwarze 2280: res = mandoc_realloc(res,
2281: ssz + cp->val.sz + 1);
1.42 schwarze 2282: memcpy(res + ssz, cp->val.p, cp->val.sz);
2283: ssz += cp->val.sz;
2284: p += (int)cp->key.sz;
2285: continue;
2286: }
2287:
2288: /*
2289: * Handle escapes carefully: we need to copy
2290: * over just the escape itself, or else we might
2291: * do replacements within the escape itself.
2292: * Make sure to pass along the bogus string.
2293: */
2294: pp = p++;
2295: esc = mandoc_escape(&p, NULL, NULL);
2296: if (ESCAPE_ERROR == esc) {
2297: sz = strlen(pp);
2298: res = mandoc_realloc(res, ssz + sz + 1);
2299: memcpy(res + ssz, pp, sz);
2300: break;
2301: }
1.80 schwarze 2302: /*
2303: * We bail out on bad escapes.
1.42 schwarze 2304: * No need to warn: we already did so when
2305: * roff_res() was called.
2306: */
2307: sz = (int)(p - pp);
2308: res = mandoc_realloc(res, ssz + sz + 1);
2309: memcpy(res + ssz, pp, sz);
2310: ssz += sz;
2311: }
2312:
2313: res[(int)ssz] = '\0';
2314: return(res);
1.99 ! schwarze 2315: }
! 2316:
! 2317: int
! 2318: roff_getformat(const struct roff *r)
! 2319: {
! 2320:
! 2321: return(r->format);
1.48 schwarze 2322: }
2323:
2324: /*
1.80 schwarze 2325: * Find out whether a line is a macro line or not.
1.48 schwarze 2326: * If it is, adjust the current position and return one; if it isn't,
2327: * return zero and don't change the current position.
2328: * If the control character has been set with `.cc', then let that grain
2329: * precedence.
2330: * This is slighly contrary to groff, where using the non-breaking
2331: * control character when `cc' has been invoked will cause the
2332: * non-breaking macro contents to be printed verbatim.
2333: */
2334: int
2335: roff_getcontrol(const struct roff *r, const char *cp, int *ppos)
2336: {
2337: int pos;
2338:
2339: pos = *ppos;
2340:
2341: if (0 != r->control && cp[pos] == r->control)
2342: pos++;
2343: else if (0 != r->control)
2344: return(0);
2345: else if ('\\' == cp[pos] && '.' == cp[pos + 1])
2346: pos += 2;
2347: else if ('.' == cp[pos] || '\'' == cp[pos])
2348: pos++;
2349: else
2350: return(0);
2351:
2352: while (' ' == cp[pos] || '\t' == cp[pos])
2353: pos++;
2354:
2355: *ppos = pos;
2356: return(1);
1.1 schwarze 2357: }
![[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