Annotation of src/usr.bin/mandoc/roff.c, Revision 1.86
1.86 ! schwarze 1: /* $Id: roff.c,v 1.85 2014/06/29 22:12:54 schwarze Exp $ */
1.1 schwarze 2: /*
1.48 schwarze 3: * Copyright (c) 2010, 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
1.63 schwarze 4: * Copyright (c) 2010-2014 Ingo Schwarze <schwarze@openbsd.org>
1.1 schwarze 5: *
6: * Permission to use, copy, modify, and distribute this software for any
7: * purpose with or without fee is hereby granted, provided that the above
8: * copyright notice and this permission notice appear in all copies.
9: *
1.16 schwarze 10: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.1 schwarze 11: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.16 schwarze 12: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.1 schwarze 13: * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
14: * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
15: * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
16: * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
17: */
18: #include <assert.h>
1.3 schwarze 19: #include <ctype.h>
1.51 schwarze 20: #include <stdio.h>
1.1 schwarze 21: #include <stdlib.h>
22: #include <string.h>
23:
24: #include "mandoc.h"
1.74 schwarze 25: #include "mandoc_aux.h"
1.27 schwarze 26: #include "libroff.h"
1.8 schwarze 27: #include "libmandoc.h"
1.1 schwarze 28:
1.37 schwarze 29: /* Maximum number of nested if-else conditionals. */
1.2 schwarze 30: #define RSTACK_MAX 128
31:
1.43 schwarze 32: /* Maximum number of string expansions per line, to break infinite loops. */
33: #define EXPAND_LIMIT 1000
34:
1.1 schwarze 35: enum rofft {
1.20 schwarze 36: ROFF_ad,
1.2 schwarze 37: ROFF_am,
38: ROFF_ami,
39: ROFF_am1,
1.66 schwarze 40: ROFF_as,
1.48 schwarze 41: ROFF_cc,
1.67 schwarze 42: ROFF_ce,
1.1 schwarze 43: ROFF_de,
44: ROFF_dei,
1.2 schwarze 45: ROFF_de1,
46: ROFF_ds,
47: ROFF_el,
1.58 schwarze 48: ROFF_fam,
1.59 schwarze 49: ROFF_hw,
1.20 schwarze 50: ROFF_hy,
1.2 schwarze 51: ROFF_ie,
52: ROFF_if,
1.1 schwarze 53: ROFF_ig,
1.30 schwarze 54: ROFF_it,
1.20 schwarze 55: ROFF_ne,
56: ROFF_nh,
1.14 schwarze 57: ROFF_nr,
1.31 schwarze 58: ROFF_ns,
59: ROFF_ps,
1.2 schwarze 60: ROFF_rm,
1.76 schwarze 61: ROFF_rr,
1.14 schwarze 62: ROFF_so,
1.31 schwarze 63: ROFF_ta,
1.2 schwarze 64: ROFF_tr,
1.47 schwarze 65: ROFF_Dd,
66: ROFF_TH,
1.27 schwarze 67: ROFF_TS,
68: ROFF_TE,
69: ROFF_T_,
1.32 schwarze 70: ROFF_EQ,
71: ROFF_EN,
1.2 schwarze 72: ROFF_cblock,
1.16 schwarze 73: ROFF_USERDEF,
1.1 schwarze 74: ROFF_MAX
75: };
76:
1.41 schwarze 77: /*
1.42 schwarze 78: * An incredibly-simple string buffer.
79: */
1.8 schwarze 80: struct roffstr {
1.42 schwarze 81: char *p; /* nil-terminated buffer */
82: size_t sz; /* saved strlen(p) */
83: };
84:
85: /*
86: * A key-value roffstr pair as part of a singly-linked list.
87: */
88: struct roffkv {
89: struct roffstr key;
90: struct roffstr val;
91: struct roffkv *next; /* next in list */
1.8 schwarze 92: };
93:
1.52 schwarze 94: /*
95: * A single number register as part of a singly-linked list.
96: */
97: struct roffreg {
98: struct roffstr key;
1.53 schwarze 99: int val;
1.52 schwarze 100: struct roffreg *next;
101: };
102:
1.1 schwarze 103: struct roff {
1.35 schwarze 104: struct mparse *parse; /* parse point */
1.72 schwarze 105: int options; /* parse options */
1.1 schwarze 106: struct roffnode *last; /* leaf of stack */
1.71 schwarze 107: int rstack[RSTACK_MAX]; /* stack of !`ie' rules */
1.48 schwarze 108: char control; /* control character */
1.2 schwarze 109: int rstackpos; /* position in rstack */
1.52 schwarze 110: struct roffreg *regtab; /* number registers */
1.42 schwarze 111: struct roffkv *strtab; /* user-defined strings & macros */
112: struct roffkv *xmbtab; /* multi-byte trans table (`tr') */
113: struct roffstr *xtab; /* single-byte trans table (`tr') */
1.16 schwarze 114: const char *current_string; /* value of last called user macro */
1.27 schwarze 115: struct tbl_node *first_tbl; /* first table parsed */
116: struct tbl_node *last_tbl; /* last table parsed */
117: struct tbl_node *tbl; /* current table being parsed */
1.32 schwarze 118: struct eqn_node *last_eqn; /* last equation parsed */
119: struct eqn_node *first_eqn; /* first equation parsed */
120: struct eqn_node *eqn; /* current equation being parsed */
1.1 schwarze 121: };
122:
123: struct roffnode {
124: enum rofft tok; /* type of node */
125: struct roffnode *parent; /* up one in stack */
126: int line; /* parse line */
127: int col; /* parse col */
1.16 schwarze 128: char *name; /* node name, e.g. macro name */
1.2 schwarze 129: char *end; /* end-rules: custom token */
130: int endspan; /* end-rules: next-line or infty */
1.71 schwarze 131: int rule; /* current evaluation rule */
1.1 schwarze 132: };
133:
134: #define ROFF_ARGS struct roff *r, /* parse ctx */ \
135: enum rofft tok, /* tok of macro */ \
1.80 schwarze 136: char **bufp, /* input buffer */ \
1.1 schwarze 137: size_t *szp, /* size of input buffer */ \
138: int ln, /* parse line */ \
1.2 schwarze 139: int ppos, /* original pos in buffer */ \
140: int pos, /* current pos in buffer */ \
141: int *offs /* reset offset of buffer data */
1.1 schwarze 142:
143: typedef enum rofferr (*roffproc)(ROFF_ARGS);
144:
145: struct roffmac {
146: const char *name; /* macro name */
1.2 schwarze 147: roffproc proc; /* process new macro */
148: roffproc text; /* process as child text of macro */
149: roffproc sub; /* process as child of macro */
150: int flags;
151: #define ROFFMAC_STRUCT (1 << 0) /* always interpret */
1.3 schwarze 152: struct roffmac *next;
1.1 schwarze 153: };
154:
1.37 schwarze 155: struct predef {
156: const char *name; /* predefined input name */
157: const char *str; /* replacement symbol */
158: };
159:
160: #define PREDEF(__name, __str) \
161: { (__name), (__str) },
162:
1.42 schwarze 163: static enum rofft roffhash_find(const char *, size_t);
164: static void roffhash_init(void);
165: static void roffnode_cleanscope(struct roff *);
166: static void roffnode_pop(struct roff *);
167: static void roffnode_push(struct roff *, enum rofft,
168: const char *, int, int);
1.2 schwarze 169: static enum rofferr roff_block(ROFF_ARGS);
170: static enum rofferr roff_block_text(ROFF_ARGS);
171: static enum rofferr roff_block_sub(ROFF_ARGS);
172: static enum rofferr roff_cblock(ROFF_ARGS);
1.48 schwarze 173: static enum rofferr roff_cc(ROFF_ARGS);
1.68 schwarze 174: static void roff_ccond(struct roff *, int, int);
1.2 schwarze 175: static enum rofferr roff_cond(ROFF_ARGS);
176: static enum rofferr roff_cond_text(ROFF_ARGS);
177: static enum rofferr roff_cond_sub(ROFF_ARGS);
1.7 schwarze 178: static enum rofferr roff_ds(ROFF_ARGS);
1.71 schwarze 179: static int roff_evalcond(const char *, int *);
1.77 schwarze 180: static int roff_evalnum(const char *, int *, int *, int);
181: static int roff_evalpar(const char *, int *, int *);
1.71 schwarze 182: static int roff_evalstrcond(const char *, int *);
1.42 schwarze 183: static void roff_free1(struct roff *);
1.52 schwarze 184: static void roff_freereg(struct roffreg *);
1.42 schwarze 185: static void roff_freestr(struct roffkv *);
1.85 schwarze 186: static size_t roff_getname(struct roff *, char **, int, int);
1.56 schwarze 187: static int roff_getnum(const char *, int *, int *);
188: static int roff_getop(const char *, int *, char *);
1.53 schwarze 189: static int roff_getregn(const struct roff *,
190: const char *, size_t);
1.65 schwarze 191: static int roff_getregro(const char *name);
1.80 schwarze 192: static const char *roff_getstrn(const struct roff *,
1.8 schwarze 193: const char *, size_t);
1.51 schwarze 194: static enum rofferr roff_it(ROFF_ARGS);
1.21 schwarze 195: static enum rofferr roff_line_ignore(ROFF_ARGS);
1.6 schwarze 196: static enum rofferr roff_nr(ROFF_ARGS);
1.41 schwarze 197: static void roff_openeqn(struct roff *, const char *,
198: int, int, const char *);
1.42 schwarze 199: static enum rofft roff_parse(struct roff *, const char *, int *);
1.51 schwarze 200: static enum rofferr roff_parsetext(char **, size_t *, int, int *);
1.80 schwarze 201: static enum rofferr roff_res(struct roff *,
1.37 schwarze 202: char **, size_t *, int, int);
1.29 schwarze 203: static enum rofferr roff_rm(ROFF_ARGS);
1.76 schwarze 204: static enum rofferr roff_rr(ROFF_ARGS);
1.8 schwarze 205: static void roff_setstr(struct roff *,
1.16 schwarze 206: const char *, const char *, int);
1.80 schwarze 207: static void roff_setstrn(struct roffkv **, const char *,
1.42 schwarze 208: size_t, const char *, size_t, int);
1.14 schwarze 209: static enum rofferr roff_so(ROFF_ARGS);
1.42 schwarze 210: static enum rofferr roff_tr(ROFF_ARGS);
1.47 schwarze 211: static enum rofferr roff_Dd(ROFF_ARGS);
212: static enum rofferr roff_TH(ROFF_ARGS);
1.27 schwarze 213: static enum rofferr roff_TE(ROFF_ARGS);
214: static enum rofferr roff_TS(ROFF_ARGS);
1.32 schwarze 215: static enum rofferr roff_EQ(ROFF_ARGS);
216: static enum rofferr roff_EN(ROFF_ARGS);
1.27 schwarze 217: static enum rofferr roff_T_(ROFF_ARGS);
1.16 schwarze 218: static enum rofferr roff_userdef(ROFF_ARGS);
1.1 schwarze 219:
1.42 schwarze 220: /* See roffhash_find() */
1.3 schwarze 221:
222: #define ASCII_HI 126
223: #define ASCII_LO 33
224: #define HASHWIDTH (ASCII_HI - ASCII_LO + 1)
225:
226: static struct roffmac *hash[HASHWIDTH];
227:
228: static struct roffmac roffs[ROFF_MAX] = {
1.21 schwarze 229: { "ad", roff_line_ignore, NULL, NULL, 0, NULL },
1.3 schwarze 230: { "am", roff_block, roff_block_text, roff_block_sub, 0, NULL },
231: { "ami", roff_block, roff_block_text, roff_block_sub, 0, NULL },
232: { "am1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.66 schwarze 233: { "as", roff_ds, NULL, NULL, 0, NULL },
1.48 schwarze 234: { "cc", roff_cc, NULL, NULL, 0, NULL },
1.67 schwarze 235: { "ce", roff_line_ignore, NULL, NULL, 0, NULL },
1.3 schwarze 236: { "de", roff_block, roff_block_text, roff_block_sub, 0, NULL },
237: { "dei", roff_block, roff_block_text, roff_block_sub, 0, NULL },
238: { "de1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.7 schwarze 239: { "ds", roff_ds, NULL, NULL, 0, NULL },
1.3 schwarze 240: { "el", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
1.58 schwarze 241: { "fam", roff_line_ignore, NULL, NULL, 0, NULL },
1.59 schwarze 242: { "hw", roff_line_ignore, NULL, NULL, 0, NULL },
1.21 schwarze 243: { "hy", roff_line_ignore, NULL, NULL, 0, NULL },
1.3 schwarze 244: { "ie", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
245: { "if", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
246: { "ig", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.51 schwarze 247: { "it", roff_it, NULL, NULL, 0, NULL },
1.21 schwarze 248: { "ne", roff_line_ignore, NULL, NULL, 0, NULL },
249: { "nh", roff_line_ignore, NULL, NULL, 0, NULL },
1.14 schwarze 250: { "nr", roff_nr, NULL, NULL, 0, NULL },
1.31 schwarze 251: { "ns", roff_line_ignore, NULL, NULL, 0, NULL },
252: { "ps", roff_line_ignore, NULL, NULL, 0, NULL },
1.29 schwarze 253: { "rm", roff_rm, NULL, NULL, 0, NULL },
1.76 schwarze 254: { "rr", roff_rr, NULL, NULL, 0, NULL },
1.14 schwarze 255: { "so", roff_so, NULL, NULL, 0, NULL },
1.31 schwarze 256: { "ta", roff_line_ignore, NULL, NULL, 0, NULL },
1.42 schwarze 257: { "tr", roff_tr, NULL, NULL, 0, NULL },
1.47 schwarze 258: { "Dd", roff_Dd, NULL, NULL, 0, NULL },
259: { "TH", roff_TH, NULL, NULL, 0, NULL },
1.27 schwarze 260: { "TS", roff_TS, NULL, NULL, 0, NULL },
261: { "TE", roff_TE, NULL, NULL, 0, NULL },
262: { "T&", roff_T_, NULL, NULL, 0, NULL },
1.32 schwarze 263: { "EQ", roff_EQ, NULL, NULL, 0, NULL },
264: { "EN", roff_EN, NULL, NULL, 0, NULL },
1.3 schwarze 265: { ".", roff_cblock, NULL, NULL, 0, NULL },
1.16 schwarze 266: { NULL, roff_userdef, NULL, NULL, 0, NULL },
1.1 schwarze 267: };
268:
1.73 schwarze 269: /* not currently implemented: Ds em Eq LP Me PP pp Or Rd Sf SH */
1.47 schwarze 270: const char *const __mdoc_reserved[] = {
271: "Ac", "Ad", "An", "Ao", "Ap", "Aq", "Ar", "At",
272: "Bc", "Bd", "Bf", "Bk", "Bl", "Bo", "Bq",
273: "Brc", "Bro", "Brq", "Bsx", "Bt", "Bx",
274: "Cd", "Cm", "Db", "Dc", "Dd", "Dl", "Do", "Dq",
1.73 schwarze 275: "Dt", "Dv", "Dx", "D1",
276: "Ec", "Ed", "Ef", "Ek", "El", "Em",
277: "En", "Eo", "Er", "Es", "Ev", "Ex",
1.47 schwarze 278: "Fa", "Fc", "Fd", "Fl", "Fn", "Fo", "Fr", "Ft", "Fx",
1.73 schwarze 279: "Hf", "Ic", "In", "It", "Lb", "Li", "Lk", "Lp",
280: "Ms", "Mt", "Nd", "Nm", "No", "Ns", "Nx",
1.47 schwarze 281: "Oc", "Oo", "Op", "Os", "Ot", "Ox",
1.73 schwarze 282: "Pa", "Pc", "Pf", "Po", "Pp", "Pq",
283: "Qc", "Ql", "Qo", "Qq", "Re", "Rs", "Rv",
284: "Sc", "Sh", "Sm", "So", "Sq",
1.47 schwarze 285: "Ss", "St", "Sx", "Sy",
286: "Ta", "Tn", "Ud", "Ux", "Va", "Vt", "Xc", "Xo", "Xr",
1.73 schwarze 287: "%A", "%B", "%C", "%D", "%I", "%J", "%N", "%O",
1.47 schwarze 288: "%P", "%Q", "%R", "%T", "%U", "%V",
289: NULL
290: };
291:
1.73 schwarze 292: /* not currently implemented: BT DE DS ME MT PT SY TQ YS */
1.47 schwarze 293: const char *const __man_reserved[] = {
1.73 schwarze 294: "AT", "B", "BI", "BR", "DT",
295: "EE", "EN", "EQ", "EX", "HP", "I", "IB", "IP", "IR",
296: "LP", "OP", "P", "PD", "PP",
297: "R", "RB", "RE", "RI", "RS", "SB", "SH", "SM", "SS",
298: "TE", "TH", "TP", "TS", "T&", "UC", "UE", "UR",
1.47 schwarze 299: NULL
300: };
301:
1.37 schwarze 302: /* Array of injected predefined strings. */
303: #define PREDEFS_MAX 38
304: static const struct predef predefs[PREDEFS_MAX] = {
305: #include "predefs.in"
306: };
307:
1.42 schwarze 308: /* See roffhash_find() */
1.3 schwarze 309: #define ROFF_HASH(p) (p[0] - ASCII_LO)
310:
1.51 schwarze 311: static int roffit_lines; /* number of lines to delay */
312: static char *roffit_macro; /* nil-terminated macro line */
313:
1.80 schwarze 314:
1.3 schwarze 315: static void
1.42 schwarze 316: roffhash_init(void)
1.3 schwarze 317: {
318: struct roffmac *n;
319: int buc, i;
320:
1.16 schwarze 321: for (i = 0; i < (int)ROFF_USERDEF; i++) {
1.3 schwarze 322: assert(roffs[i].name[0] >= ASCII_LO);
323: assert(roffs[i].name[0] <= ASCII_HI);
324:
325: buc = ROFF_HASH(roffs[i].name);
326:
327: if (NULL != (n = hash[buc])) {
328: for ( ; n->next; n = n->next)
329: /* Do nothing. */ ;
330: n->next = &roffs[i];
331: } else
332: hash[buc] = &roffs[i];
333: }
334: }
335:
1.1 schwarze 336: /*
337: * Look up a roff token by its name. Returns ROFF_MAX if no macro by
338: * the nil-terminated string name could be found.
339: */
340: static enum rofft
1.42 schwarze 341: roffhash_find(const char *p, size_t s)
1.1 schwarze 342: {
1.3 schwarze 343: int buc;
344: struct roffmac *n;
1.1 schwarze 345:
1.3 schwarze 346: /*
347: * libroff has an extremely simple hashtable, for the time
348: * being, which simply keys on the first character, which must
349: * be printable, then walks a chain. It works well enough until
350: * optimised.
351: */
352:
353: if (p[0] < ASCII_LO || p[0] > ASCII_HI)
354: return(ROFF_MAX);
355:
356: buc = ROFF_HASH(p);
357:
358: if (NULL == (n = hash[buc]))
359: return(ROFF_MAX);
360: for ( ; n; n = n->next)
1.16 schwarze 361: if (0 == strncmp(n->name, p, s) && '\0' == n->name[(int)s])
1.3 schwarze 362: return((enum rofft)(n - roffs));
1.1 schwarze 363:
364: return(ROFF_MAX);
365: }
366:
367: /*
368: * Pop the current node off of the stack of roff instructions currently
369: * pending.
370: */
371: static void
372: roffnode_pop(struct roff *r)
373: {
374: struct roffnode *p;
375:
1.2 schwarze 376: assert(r->last);
1.80 schwarze 377: p = r->last;
1.2 schwarze 378:
379: r->last = r->last->parent;
1.16 schwarze 380: free(p->name);
381: free(p->end);
1.1 schwarze 382: free(p);
383: }
384:
385: /*
386: * Push a roff node onto the instruction stack. This must later be
387: * removed with roffnode_pop().
388: */
1.11 schwarze 389: static void
1.16 schwarze 390: roffnode_push(struct roff *r, enum rofft tok, const char *name,
391: int line, int col)
1.1 schwarze 392: {
393: struct roffnode *p;
394:
1.11 schwarze 395: p = mandoc_calloc(1, sizeof(struct roffnode));
1.1 schwarze 396: p->tok = tok;
1.16 schwarze 397: if (name)
398: p->name = mandoc_strdup(name);
1.1 schwarze 399: p->parent = r->last;
400: p->line = line;
401: p->col = col;
1.71 schwarze 402: p->rule = p->parent ? p->parent->rule : 0;
1.1 schwarze 403:
404: r->last = p;
405: }
406:
407: static void
408: roff_free1(struct roff *r)
409: {
1.49 schwarze 410: struct tbl_node *tbl;
1.32 schwarze 411: struct eqn_node *e;
1.42 schwarze 412: int i;
1.27 schwarze 413:
1.49 schwarze 414: while (NULL != (tbl = r->first_tbl)) {
415: r->first_tbl = tbl->next;
416: tbl_free(tbl);
1.27 schwarze 417: }
418:
419: r->first_tbl = r->last_tbl = r->tbl = NULL;
1.1 schwarze 420:
1.32 schwarze 421: while (NULL != (e = r->first_eqn)) {
422: r->first_eqn = e->next;
423: eqn_free(e);
424: }
425:
426: r->first_eqn = r->last_eqn = r->eqn = NULL;
427:
1.1 schwarze 428: while (r->last)
429: roffnode_pop(r);
1.27 schwarze 430:
1.42 schwarze 431: roff_freestr(r->strtab);
432: roff_freestr(r->xmbtab);
433:
434: r->strtab = r->xmbtab = NULL;
435:
1.52 schwarze 436: roff_freereg(r->regtab);
437:
438: r->regtab = NULL;
439:
1.42 schwarze 440: if (r->xtab)
441: for (i = 0; i < 128; i++)
442: free(r->xtab[i].p);
443:
444: free(r->xtab);
445: r->xtab = NULL;
1.1 schwarze 446: }
447:
448: void
449: roff_reset(struct roff *r)
450: {
451:
452: roff_free1(r);
1.48 schwarze 453: r->control = 0;
1.1 schwarze 454: }
455:
456: void
457: roff_free(struct roff *r)
458: {
459:
460: roff_free1(r);
461: free(r);
462: }
463:
464: struct roff *
1.72 schwarze 465: roff_alloc(struct mparse *parse, int options)
1.1 schwarze 466: {
467: struct roff *r;
468:
1.11 schwarze 469: r = mandoc_calloc(1, sizeof(struct roff));
1.35 schwarze 470: r->parse = parse;
1.72 schwarze 471: r->options = options;
1.2 schwarze 472: r->rstackpos = -1;
1.80 schwarze 473:
1.42 schwarze 474: roffhash_init();
1.37 schwarze 475:
1.1 schwarze 476: return(r);
477: }
478:
1.8 schwarze 479: /*
1.79 schwarze 480: * In the current line, expand escape sequences that tend to get
481: * used in numerical expressions and conditional requests.
482: * Also check the syntax of the remaining escape sequences.
1.8 schwarze 483: */
1.45 schwarze 484: static enum rofferr
1.37 schwarze 485: roff_res(struct roff *r, char **bufp, size_t *szp, int ln, int pos)
1.8 schwarze 486: {
1.81 schwarze 487: char ubuf[24]; /* buffer to print the number */
1.78 schwarze 488: const char *start; /* start of the string to process */
1.82 schwarze 489: char *stesc; /* start of an escape sequence ('\\') */
1.23 schwarze 490: const char *stnam; /* start of the name, after "[(*" */
491: const char *cp; /* end of the name, e.g. before ']' */
492: const char *res; /* the string to be substituted */
1.53 schwarze 493: char *nbuf; /* new buffer to copy bufp to */
494: size_t maxl; /* expected length of the escape name */
495: size_t naml; /* actual length of the escape name */
496: int expand_count; /* to avoid infinite loops */
1.79 schwarze 497: int npos; /* position in numeric expression */
498: int irc; /* return code from roff_evalnum() */
499: char term; /* character terminating the escape */
1.8 schwarze 500:
1.43 schwarze 501: expand_count = 0;
1.78 schwarze 502: start = *bufp + pos;
503: stesc = strchr(start, '\0') - 1;
504: while (stesc-- > start) {
1.43 schwarze 505:
1.78 schwarze 506: /* Search backwards for the next backslash. */
507:
508: if ('\\' != *stesc)
509: continue;
510:
511: /* If it is escaped, skip it. */
512:
513: for (cp = stesc - 1; cp >= start; cp--)
514: if ('\\' != *cp)
515: break;
516:
517: if (0 == (stesc - cp) % 2) {
1.82 schwarze 518: stesc = (char *)cp;
1.78 schwarze 519: continue;
520: }
1.23 schwarze 521:
1.79 schwarze 522: /* Decide whether to expand or to check only. */
1.23 schwarze 523:
1.79 schwarze 524: term = '\0';
1.78 schwarze 525: cp = stesc + 1;
1.53 schwarze 526: switch (*cp) {
1.80 schwarze 527: case '*':
1.53 schwarze 528: res = NULL;
529: break;
1.80 schwarze 530: case 'B':
1.79 schwarze 531: /* FALLTHROUGH */
1.80 schwarze 532: case 'w':
1.79 schwarze 533: term = cp[1];
534: /* FALLTHROUGH */
1.80 schwarze 535: case 'n':
1.53 schwarze 536: res = ubuf;
537: break;
538: default:
1.78 schwarze 539: if (ESCAPE_ERROR == mandoc_escape(&cp, NULL, NULL))
540: mandoc_msg(MANDOCERR_BADESCAPE, r->parse,
541: ln, (int)(stesc - *bufp), NULL);
542: continue;
1.42 schwarze 543: }
544:
1.78 schwarze 545: if (EXPAND_LIMIT < ++expand_count) {
546: mandoc_msg(MANDOCERR_ROFFLOOP, r->parse,
547: ln, (int)(stesc - *bufp), NULL);
548: return(ROFF_IGN);
549: }
1.23 schwarze 550:
551: /*
552: * The third character decides the length
1.53 schwarze 553: * of the name of the string or register.
1.23 schwarze 554: * Save a pointer to the name.
555: */
556:
1.79 schwarze 557: if ('\0' == term) {
558: switch (*++cp) {
1.80 schwarze 559: case '\0':
1.79 schwarze 560: maxl = 0;
561: break;
1.80 schwarze 562: case '(':
1.79 schwarze 563: cp++;
564: maxl = 2;
565: break;
1.80 schwarze 566: case '[':
1.79 schwarze 567: cp++;
568: term = ']';
569: maxl = 0;
570: break;
571: default:
572: maxl = 1;
573: break;
574: }
575: } else {
576: cp += 2;
1.8 schwarze 577: maxl = 0;
578: }
1.23 schwarze 579: stnam = cp;
1.8 schwarze 580:
1.23 schwarze 581: /* Advance to the end of the name. */
1.8 schwarze 582:
1.53 schwarze 583: for (naml = 0; 0 == maxl || naml < maxl; naml++, cp++) {
1.42 schwarze 584: if ('\0' == *cp) {
1.80 schwarze 585: mandoc_msg(MANDOCERR_BADESCAPE, r->parse,
586: ln, (int)(stesc - *bufp), NULL);
1.79 schwarze 587: break;
1.42 schwarze 588: }
1.79 schwarze 589: if (0 == maxl && *cp == term) {
590: cp++;
1.8 schwarze 591: break;
1.79 schwarze 592: }
1.8 schwarze 593: }
594:
1.23 schwarze 595: /*
596: * Retrieve the replacement string; if it is
597: * undefined, resume searching for escapes.
598: */
599:
1.79 schwarze 600: switch (stesc[1]) {
1.80 schwarze 601: case '*':
1.53 schwarze 602: res = roff_getstrn(r, stnam, naml);
1.79 schwarze 603: break;
1.80 schwarze 604: case 'B':
1.79 schwarze 605: npos = 0;
606: irc = roff_evalnum(stnam, &npos, NULL, 0);
607: ubuf[0] = irc && stnam + npos + 1 == cp
608: ? '1' : '0';
609: ubuf[1] = '\0';
610: break;
1.80 schwarze 611: case 'n':
1.81 schwarze 612: (void)snprintf(ubuf, sizeof(ubuf), "%d",
1.53 schwarze 613: roff_getregn(r, stnam, naml));
1.79 schwarze 614: break;
1.80 schwarze 615: case 'w':
1.81 schwarze 616: (void)snprintf(ubuf, sizeof(ubuf), "%d",
1.79 schwarze 617: 24 * (int)naml);
618: break;
619: }
1.8 schwarze 620:
621: if (NULL == res) {
1.80 schwarze 622: mandoc_msg(MANDOCERR_BADESCAPE, r->parse,
623: ln, (int)(stesc - *bufp), NULL);
1.37 schwarze 624: res = "";
1.8 schwarze 625: }
626:
1.23 schwarze 627: /* Replace the escape sequence by the string. */
628:
1.82 schwarze 629: *stesc = '\0';
630: *szp = mandoc_asprintf(&nbuf, "%s%s%s",
631: *bufp, res, cp) + 1;
1.8 schwarze 632:
1.78 schwarze 633: /* Prepare for the next replacement. */
1.8 schwarze 634:
1.78 schwarze 635: start = nbuf + pos;
1.82 schwarze 636: stesc = nbuf + (stesc - *bufp) + strlen(res);
1.8 schwarze 637: free(*bufp);
1.53 schwarze 638: *bufp = nbuf;
1.42 schwarze 639: }
1.45 schwarze 640: return(ROFF_CONT);
1.42 schwarze 641: }
642:
643: /*
1.51 schwarze 644: * Process text streams:
645: * Convert all breakable hyphens into ASCII_HYPH.
646: * Decrement and spring input line trap.
1.42 schwarze 647: */
648: static enum rofferr
1.51 schwarze 649: roff_parsetext(char **bufp, size_t *szp, int pos, int *offs)
1.42 schwarze 650: {
651: size_t sz;
652: const char *start;
1.51 schwarze 653: char *p;
654: int isz;
1.42 schwarze 655: enum mandoc_esc esc;
656:
1.51 schwarze 657: start = p = *bufp + pos;
1.42 schwarze 658:
659: while ('\0' != *p) {
660: sz = strcspn(p, "-\\");
661: p += sz;
662:
663: if ('\0' == *p)
664: break;
665:
666: if ('\\' == *p) {
667: /* Skip over escapes. */
668: p++;
1.62 schwarze 669: esc = mandoc_escape((const char **)&p, NULL, NULL);
1.42 schwarze 670: if (ESCAPE_ERROR == esc)
671: break;
672: continue;
673: } else if (p == start) {
674: p++;
675: continue;
676: }
677:
1.44 schwarze 678: if (isalpha((unsigned char)p[-1]) &&
679: isalpha((unsigned char)p[1]))
1.42 schwarze 680: *p = ASCII_HYPH;
681: p++;
1.8 schwarze 682: }
683:
1.51 schwarze 684: /* Spring the input line trap. */
685: if (1 == roffit_lines) {
1.75 schwarze 686: isz = mandoc_asprintf(&p, "%s\n.%s", *bufp, roffit_macro);
1.51 schwarze 687: free(*bufp);
688: *bufp = p;
689: *szp = isz + 1;
690: *offs = 0;
691: free(roffit_macro);
692: roffit_lines = 0;
693: return(ROFF_REPARSE);
694: } else if (1 < roffit_lines)
695: --roffit_lines;
1.42 schwarze 696: return(ROFF_CONT);
1.8 schwarze 697: }
698:
1.1 schwarze 699: enum rofferr
1.80 schwarze 700: roff_parseln(struct roff *r, int ln, char **bufp,
1.6 schwarze 701: size_t *szp, int pos, int *offs)
1.1 schwarze 702: {
703: enum rofft t;
1.27 schwarze 704: enum rofferr e;
1.35 schwarze 705: int ppos, ctl;
1.1 schwarze 706:
1.2 schwarze 707: /*
1.8 schwarze 708: * Run the reserved-word filter only if we have some reserved
709: * words to fill in.
710: */
711:
1.45 schwarze 712: e = roff_res(r, bufp, szp, ln, pos);
713: if (ROFF_IGN == e)
714: return(e);
715: assert(ROFF_CONT == e);
1.8 schwarze 716:
1.35 schwarze 717: ppos = pos;
1.48 schwarze 718: ctl = roff_getcontrol(r, *bufp, &pos);
1.35 schwarze 719:
1.8 schwarze 720: /*
1.2 schwarze 721: * First, if a scope is open and we're not a macro, pass the
722: * text through the macro's filter. If a scope isn't open and
723: * we're not a macro, just let it through.
1.32 schwarze 724: * Finally, if there's an equation scope open, divert it into it
725: * no matter our state.
1.2 schwarze 726: */
727:
1.35 schwarze 728: if (r->last && ! ctl) {
1.2 schwarze 729: t = r->last->tok;
730: assert(roffs[t].text);
1.80 schwarze 731: e = (*roffs[t].text)(r, t, bufp, szp, ln, pos, pos, offs);
1.27 schwarze 732: assert(ROFF_IGN == e || ROFF_CONT == e);
1.32 schwarze 733: if (ROFF_CONT != e)
734: return(e);
1.54 schwarze 735: }
736: if (r->eqn)
737: return(eqn_read(&r->eqn, ln, *bufp, ppos, offs));
738: if ( ! ctl) {
1.32 schwarze 739: if (r->tbl)
1.35 schwarze 740: return(tbl_read(r->tbl, ln, *bufp, pos));
1.51 schwarze 741: return(roff_parsetext(bufp, szp, pos, offs));
1.54 schwarze 742: }
1.2 schwarze 743:
744: /*
745: * If a scope is open, go to the child handler for that macro,
746: * as it may want to preprocess before doing anything with it.
1.32 schwarze 747: * Don't do so if an equation is open.
1.2 schwarze 748: */
749:
750: if (r->last) {
1.1 schwarze 751: t = r->last->tok;
752: assert(roffs[t].sub);
1.80 schwarze 753: return((*roffs[t].sub)(r, t, bufp, szp,
754: ln, ppos, pos, offs));
1.2 schwarze 755: }
756:
757: /*
758: * Lastly, as we've no scope open, try to look up and execute
759: * the new macro. If no macro is found, simply return and let
760: * the compilers handle it.
761: */
762:
1.16 schwarze 763: if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos)))
1.1 schwarze 764: return(ROFF_CONT);
765:
1.2 schwarze 766: assert(roffs[t].proc);
1.80 schwarze 767: return((*roffs[t].proc)(r, t, bufp, szp, ln, ppos, pos, offs));
1.2 schwarze 768: }
769:
1.27 schwarze 770: void
1.2 schwarze 771: roff_endparse(struct roff *r)
772: {
1.1 schwarze 773:
1.27 schwarze 774: if (r->last)
1.35 schwarze 775: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.80 schwarze 776: r->last->line, r->last->col, NULL);
1.27 schwarze 777:
1.32 schwarze 778: if (r->eqn) {
1.80 schwarze 779: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
780: r->eqn->eqn.ln, r->eqn->eqn.pos, NULL);
1.41 schwarze 781: eqn_end(&r->eqn);
1.32 schwarze 782: }
783:
1.27 schwarze 784: if (r->tbl) {
1.80 schwarze 785: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
786: r->tbl->line, r->tbl->pos, NULL);
1.41 schwarze 787: tbl_end(&r->tbl);
1.27 schwarze 788: }
1.1 schwarze 789: }
790:
791: /*
792: * Parse a roff node's type from the input buffer. This must be in the
793: * form of ".foo xxx" in the usual way.
794: */
795: static enum rofft
1.16 schwarze 796: roff_parse(struct roff *r, const char *buf, int *pos)
1.1 schwarze 797: {
1.16 schwarze 798: const char *mac;
799: size_t maclen;
1.1 schwarze 800: enum rofft t;
801:
1.80 schwarze 802: if ('\0' == buf[*pos] || '"' == buf[*pos] ||
803: '\t' == buf[*pos] || ' ' == buf[*pos])
1.1 schwarze 804: return(ROFF_MAX);
805:
1.68 schwarze 806: /* We stop the macro parse at an escape, tab, space, or nil. */
1.39 schwarze 807:
1.16 schwarze 808: mac = buf + *pos;
1.68 schwarze 809: maclen = strcspn(mac, " \\\t\0");
1.1 schwarze 810:
1.16 schwarze 811: t = (r->current_string = roff_getstrn(r, mac, maclen))
1.42 schwarze 812: ? ROFF_USERDEF : roffhash_find(mac, maclen);
1.1 schwarze 813:
1.34 schwarze 814: *pos += (int)maclen;
1.35 schwarze 815:
1.1 schwarze 816: while (buf[*pos] && ' ' == buf[*pos])
817: (*pos)++;
818:
819: return(t);
820: }
821:
822: static enum rofferr
1.2 schwarze 823: roff_cblock(ROFF_ARGS)
1.1 schwarze 824: {
825:
1.2 schwarze 826: /*
827: * A block-close `..' should only be invoked as a child of an
828: * ignore macro, otherwise raise a warning and just ignore it.
829: */
830:
831: if (NULL == r->last) {
1.35 schwarze 832: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.2 schwarze 833: return(ROFF_IGN);
834: }
1.1 schwarze 835:
1.2 schwarze 836: switch (r->last->tok) {
1.80 schwarze 837: case ROFF_am:
1.2 schwarze 838: /* FALLTHROUGH */
1.80 schwarze 839: case ROFF_ami:
1.2 schwarze 840: /* FALLTHROUGH */
1.80 schwarze 841: case ROFF_am1:
1.2 schwarze 842: /* FALLTHROUGH */
1.80 schwarze 843: case ROFF_de:
1.23 schwarze 844: /* ROFF_de1 is remapped to ROFF_de in roff_block(). */
1.2 schwarze 845: /* FALLTHROUGH */
1.80 schwarze 846: case ROFF_dei:
1.2 schwarze 847: /* FALLTHROUGH */
1.80 schwarze 848: case ROFF_ig:
1.2 schwarze 849: break;
850: default:
1.35 schwarze 851: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.1 schwarze 852: return(ROFF_IGN);
1.2 schwarze 853: }
854:
855: if ((*bufp)[pos])
1.35 schwarze 856: mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.2 schwarze 857:
858: roffnode_pop(r);
859: roffnode_cleanscope(r);
860: return(ROFF_IGN);
861:
862: }
1.1 schwarze 863:
1.2 schwarze 864: static void
865: roffnode_cleanscope(struct roff *r)
866: {
1.1 schwarze 867:
1.2 schwarze 868: while (r->last) {
1.46 schwarze 869: if (--r->last->endspan != 0)
1.2 schwarze 870: break;
871: roffnode_pop(r);
872: }
873: }
1.1 schwarze 874:
1.68 schwarze 875: static void
876: roff_ccond(struct roff *r, int ln, int ppos)
1.2 schwarze 877: {
1.1 schwarze 878:
1.2 schwarze 879: if (NULL == r->last) {
1.35 schwarze 880: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.68 schwarze 881: return;
1.2 schwarze 882: }
1.1 schwarze 883:
1.2 schwarze 884: switch (r->last->tok) {
1.80 schwarze 885: case ROFF_el:
1.2 schwarze 886: /* FALLTHROUGH */
1.80 schwarze 887: case ROFF_ie:
1.2 schwarze 888: /* FALLTHROUGH */
1.80 schwarze 889: case ROFF_if:
1.2 schwarze 890: break;
891: default:
1.35 schwarze 892: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.68 schwarze 893: return;
1.2 schwarze 894: }
1.1 schwarze 895:
1.2 schwarze 896: if (r->last->endspan > -1) {
1.35 schwarze 897: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.68 schwarze 898: return;
1.2 schwarze 899: }
900:
901: roffnode_pop(r);
902: roffnode_cleanscope(r);
1.68 schwarze 903: return;
1.1 schwarze 904: }
905:
906: static enum rofferr
1.2 schwarze 907: roff_block(ROFF_ARGS)
1.1 schwarze 908: {
1.86 ! schwarze 909: char *name, *cp;
! 910: size_t namesz;
1.16 schwarze 911:
1.86 ! schwarze 912: name = cp = *bufp + pos;
! 913: namesz = 0;
1.2 schwarze 914:
1.16 schwarze 915: if (ROFF_ig != tok) {
1.86 ! schwarze 916: if ('\0' == *cp) {
1.35 schwarze 917: mandoc_msg(MANDOCERR_NOARGS, r->parse, ln, ppos, NULL);
1.16 schwarze 918: return(ROFF_IGN);
919: }
1.22 schwarze 920:
921: /*
922: * Re-write `de1', since we don't really care about
923: * groff's strange compatibility mode, into `de'.
924: */
925:
1.18 schwarze 926: if (ROFF_de1 == tok)
927: tok = ROFF_de;
1.86 ! schwarze 928: else if (ROFF_de != tok)
1.35 schwarze 929: mandoc_msg(MANDOCERR_REQUEST, r->parse, ln, ppos,
1.21 schwarze 930: roffs[tok].name);
1.22 schwarze 931:
1.86 ! schwarze 932: namesz = roff_getname(r, &cp, ln, ppos);
! 933: name[namesz] = '\0';
! 934: } else
! 935: name = NULL;
1.2 schwarze 936:
1.16 schwarze 937: roffnode_push(r, tok, name, ln, ppos);
938:
939: /*
940: * At the beginning of a `de' macro, clear the existing string
941: * with the same name, if there is one. New content will be
1.66 schwarze 942: * appended from roff_block_text() in multiline mode.
1.16 schwarze 943: */
1.22 schwarze 944:
1.86 ! schwarze 945: if (namesz && ROFF_de == tok)
! 946: roff_setstrn(&r->strtab, name, namesz, "", 0, 0);
1.2 schwarze 947:
1.86 ! schwarze 948: if ('\0' == *cp)
1.2 schwarze 949: return(ROFF_IGN);
1.1 schwarze 950:
1.22 schwarze 951: /* If present, process the custom end-of-line marker. */
952:
1.86 ! schwarze 953: name = cp;
! 954: namesz = roff_getname(r, &cp, ln, ppos);
! 955: if (namesz)
! 956: r->last->end = mandoc_strndup(name, namesz);
1.2 schwarze 957:
1.86 ! schwarze 958: if ('\0' != *cp)
1.35 schwarze 959: mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.1 schwarze 960:
961: return(ROFF_IGN);
962: }
963:
964: static enum rofferr
1.2 schwarze 965: roff_block_sub(ROFF_ARGS)
1.1 schwarze 966: {
1.2 schwarze 967: enum rofft t;
968: int i, j;
969:
970: /*
971: * First check whether a custom macro exists at this level. If
972: * it does, then check against it. This is some of groff's
973: * stranger behaviours. If we encountered a custom end-scope
974: * tag and that tag also happens to be a "real" macro, then we
975: * need to try interpreting it again as a real macro. If it's
976: * not, then return ignore. Else continue.
977: */
978:
979: if (r->last->end) {
1.35 schwarze 980: for (i = pos, j = 0; r->last->end[j]; j++, i++)
1.2 schwarze 981: if ((*bufp)[i] != r->last->end[j])
982: break;
1.1 schwarze 983:
1.80 schwarze 984: if ('\0' == r->last->end[j] &&
985: ('\0' == (*bufp)[i] ||
986: ' ' == (*bufp)[i] ||
987: '\t' == (*bufp)[i])) {
1.2 schwarze 988: roffnode_pop(r);
989: roffnode_cleanscope(r);
1.1 schwarze 990:
1.35 schwarze 991: while (' ' == (*bufp)[i] || '\t' == (*bufp)[i])
992: i++;
993:
994: pos = i;
1.16 schwarze 995: if (ROFF_MAX != roff_parse(r, *bufp, &pos))
1.2 schwarze 996: return(ROFF_RERUN);
997: return(ROFF_IGN);
998: }
1.1 schwarze 999: }
1000:
1.2 schwarze 1001: /*
1002: * If we have no custom end-query or lookup failed, then try
1003: * pulling it out of the hashtable.
1004: */
1.1 schwarze 1005:
1.36 schwarze 1006: t = roff_parse(r, *bufp, &pos);
1.1 schwarze 1007:
1.16 schwarze 1008: /*
1009: * Macros other than block-end are only significant
1010: * in `de' blocks; elsewhere, simply throw them away.
1011: */
1012: if (ROFF_cblock != t) {
1013: if (ROFF_de == tok)
1.66 schwarze 1014: roff_setstr(r, r->last->name, *bufp + ppos, 2);
1.1 schwarze 1015: return(ROFF_IGN);
1.16 schwarze 1016: }
1.1 schwarze 1017:
1.2 schwarze 1018: assert(roffs[t].proc);
1.80 schwarze 1019: return((*roffs[t].proc)(r, t, bufp, szp, ln, ppos, pos, offs));
1.2 schwarze 1020: }
1021:
1022: static enum rofferr
1023: roff_block_text(ROFF_ARGS)
1024: {
1025:
1.16 schwarze 1026: if (ROFF_de == tok)
1.66 schwarze 1027: roff_setstr(r, r->last->name, *bufp + pos, 2);
1.16 schwarze 1028:
1.2 schwarze 1029: return(ROFF_IGN);
1030: }
1031:
1032: static enum rofferr
1033: roff_cond_sub(ROFF_ARGS)
1034: {
1035: enum rofft t;
1.37 schwarze 1036: char *ep;
1.71 schwarze 1037: int rr;
1.2 schwarze 1038:
1039: rr = r->last->rule;
1.37 schwarze 1040: roffnode_cleanscope(r);
1.50 schwarze 1041: t = roff_parse(r, *bufp, &pos);
1.2 schwarze 1042:
1.37 schwarze 1043: /*
1.50 schwarze 1044: * Fully handle known macros when they are structurally
1045: * required or when the conditional evaluated to true.
1.5 schwarze 1046: */
1047:
1.50 schwarze 1048: if ((ROFF_MAX != t) &&
1.71 schwarze 1049: (rr || ROFFMAC_STRUCT & roffs[t].flags)) {
1.50 schwarze 1050: assert(roffs[t].proc);
1051: return((*roffs[t].proc)(r, t, bufp, szp,
1.80 schwarze 1052: ln, ppos, pos, offs));
1.50 schwarze 1053: }
1.39 schwarze 1054:
1.69 schwarze 1055: /*
1056: * If `\}' occurs on a macro line without a preceding macro,
1057: * drop the line completely.
1058: */
1059:
1060: ep = *bufp + pos;
1061: if ('\\' == ep[0] && '}' == ep[1])
1.71 schwarze 1062: rr = 0;
1.69 schwarze 1063:
1.50 schwarze 1064: /* Always check for the closing delimiter `\}'. */
1.39 schwarze 1065:
1.50 schwarze 1066: while (NULL != (ep = strchr(ep, '\\'))) {
1.70 schwarze 1067: if ('}' == *(++ep)) {
1068: *ep = '&';
1069: roff_ccond(r, ln, ep - *bufp - 1);
1070: }
1071: ++ep;
1.50 schwarze 1072: }
1.71 schwarze 1073: return(rr ? ROFF_CONT : ROFF_IGN);
1.2 schwarze 1074: }
1075:
1076: static enum rofferr
1077: roff_cond_text(ROFF_ARGS)
1078: {
1.37 schwarze 1079: char *ep;
1.71 schwarze 1080: int rr;
1.2 schwarze 1081:
1082: rr = r->last->rule;
1.37 schwarze 1083: roffnode_cleanscope(r);
1.1 schwarze 1084:
1.70 schwarze 1085: ep = *bufp + pos;
1086: while (NULL != (ep = strchr(ep, '\\'))) {
1087: if ('}' == *(++ep)) {
1088: *ep = '&';
1089: roff_ccond(r, ln, ep - *bufp - 1);
1090: }
1091: ++ep;
1.2 schwarze 1092: }
1.71 schwarze 1093: return(rr ? ROFF_CONT : ROFF_IGN);
1.2 schwarze 1094: }
1095:
1.77 schwarze 1096: /*
1097: * Parse a single signed integer number. Stop at the first non-digit.
1098: * If there is at least one digit, return success and advance the
1099: * parse point, else return failure and let the parse point unchanged.
1100: * Ignore overflows, treat them just like the C language.
1101: */
1.56 schwarze 1102: static int
1103: roff_getnum(const char *v, int *pos, int *res)
1104: {
1.79 schwarze 1105: int myres, n, p;
1106:
1107: if (NULL == res)
1108: res = &myres;
1.56 schwarze 1109:
1110: p = *pos;
1111: n = v[p] == '-';
1112: if (n)
1113: p++;
1114:
1115: for (*res = 0; isdigit((unsigned char)v[p]); p++)
1.77 schwarze 1116: *res = 10 * *res + v[p] - '0';
1.56 schwarze 1117: if (p == *pos + n)
1118: return 0;
1119:
1120: if (n)
1121: *res = -*res;
1122:
1123: *pos = p;
1124: return 1;
1125: }
1126:
1.71 schwarze 1127: /*
1128: * Evaluate a string comparison condition.
1129: * The first character is the delimiter.
1130: * Succeed if the string up to its second occurrence
1131: * matches the string up to its third occurence.
1132: * Advance the cursor after the third occurrence
1133: * or lacking that, to the end of the line.
1134: */
1135: static int
1136: roff_evalstrcond(const char *v, int *pos)
1137: {
1138: const char *s1, *s2, *s3;
1139: int match;
1140:
1141: match = 0;
1142: s1 = v + *pos; /* initial delimiter */
1143: s2 = s1 + 1; /* for scanning the first string */
1144: s3 = strchr(s2, *s1); /* for scanning the second string */
1145:
1146: if (NULL == s3) /* found no middle delimiter */
1147: goto out;
1148:
1149: while ('\0' != *++s3) {
1150: if (*s2 != *s3) { /* mismatch */
1151: s3 = strchr(s3, *s1);
1152: break;
1153: }
1154: if (*s3 == *s1) { /* found the final delimiter */
1155: match = 1;
1156: break;
1157: }
1158: s2++;
1159: }
1160:
1161: out:
1162: if (NULL == s3)
1163: s3 = strchr(s2, '\0');
1164: else
1165: s3++;
1166: *pos = s3 - v;
1167: return(match);
1168: }
1169:
1.77 schwarze 1170: /*
1171: * Evaluate an optionally negated single character, numerical,
1172: * or string condition.
1173: */
1.71 schwarze 1174: static int
1.5 schwarze 1175: roff_evalcond(const char *v, int *pos)
1176: {
1.77 schwarze 1177: int wanttrue, number;
1.5 schwarze 1178:
1.71 schwarze 1179: if ('!' == v[*pos]) {
1180: wanttrue = 0;
1181: (*pos)++;
1182: } else
1183: wanttrue = 1;
1184:
1.5 schwarze 1185: switch (v[*pos]) {
1.80 schwarze 1186: case 'n':
1.71 schwarze 1187: /* FALLTHROUGH */
1.80 schwarze 1188: case 'o':
1.5 schwarze 1189: (*pos)++;
1.71 schwarze 1190: return(wanttrue);
1.80 schwarze 1191: case 'c':
1.71 schwarze 1192: /* FALLTHROUGH */
1.80 schwarze 1193: case 'd':
1.71 schwarze 1194: /* FALLTHROUGH */
1.80 schwarze 1195: case 'e':
1.5 schwarze 1196: /* FALLTHROUGH */
1.80 schwarze 1197: case 'r':
1.5 schwarze 1198: /* FALLTHROUGH */
1.80 schwarze 1199: case 't':
1.5 schwarze 1200: (*pos)++;
1.71 schwarze 1201: return(!wanttrue);
1.5 schwarze 1202: default:
1203: break;
1204: }
1205:
1.77 schwarze 1206: if (roff_evalnum(v, pos, &number, 0))
1207: return((number > 0) == wanttrue);
1208: else
1.71 schwarze 1209: return(roff_evalstrcond(v, pos) == wanttrue);
1.5 schwarze 1210: }
1211:
1.2 schwarze 1212: static enum rofferr
1.21 schwarze 1213: roff_line_ignore(ROFF_ARGS)
1.6 schwarze 1214: {
1.30 schwarze 1215:
1.21 schwarze 1216: return(ROFF_IGN);
1217: }
1218:
1219: static enum rofferr
1.2 schwarze 1220: roff_cond(ROFF_ARGS)
1221: {
1.46 schwarze 1222:
1223: roffnode_push(r, tok, NULL, ln, ppos);
1.2 schwarze 1224:
1.80 schwarze 1225: /*
1.35 schwarze 1226: * An `.el' has no conditional body: it will consume the value
1227: * of the current rstack entry set in prior `ie' calls or
1.80 schwarze 1228: * defaults to DENY.
1.35 schwarze 1229: *
1230: * If we're not an `el', however, then evaluate the conditional.
1231: */
1.1 schwarze 1232:
1.46 schwarze 1233: r->last->rule = ROFF_el == tok ?
1.80 schwarze 1234: (r->rstackpos < 0 ? 0 : r->rstack[r->rstackpos--]) :
1235: roff_evalcond(*bufp, &pos);
1.2 schwarze 1236:
1.35 schwarze 1237: /*
1238: * An if-else will put the NEGATION of the current evaluated
1239: * conditional into the stack of rules.
1240: */
1241:
1.2 schwarze 1242: if (ROFF_ie == tok) {
1.35 schwarze 1243: if (r->rstackpos == RSTACK_MAX - 1) {
1.80 schwarze 1244: mandoc_msg(MANDOCERR_MEM,
1245: r->parse, ln, ppos, NULL);
1.35 schwarze 1246: return(ROFF_ERR);
1247: }
1.71 schwarze 1248: r->rstack[++r->rstackpos] = !r->last->rule;
1.2 schwarze 1249: }
1.5 schwarze 1250:
1251: /* If the parent has false as its rule, then so do we. */
1252:
1.71 schwarze 1253: if (r->last->parent && !r->last->parent->rule)
1254: r->last->rule = 0;
1.5 schwarze 1255:
1256: /*
1.46 schwarze 1257: * Determine scope.
1258: * If there is nothing on the line after the conditional,
1259: * not even whitespace, use next-line scope.
1.5 schwarze 1260: */
1.2 schwarze 1261:
1.46 schwarze 1262: if ('\0' == (*bufp)[pos]) {
1263: r->last->endspan = 2;
1264: goto out;
1265: }
1266:
1267: while (' ' == (*bufp)[pos])
1268: pos++;
1269:
1270: /* An opening brace requests multiline scope. */
1.2 schwarze 1271:
1272: if ('\\' == (*bufp)[pos] && '{' == (*bufp)[pos + 1]) {
1273: r->last->endspan = -1;
1274: pos += 2;
1.46 schwarze 1275: goto out;
1.80 schwarze 1276: }
1.2 schwarze 1277:
1278: /*
1.46 schwarze 1279: * Anything else following the conditional causes
1280: * single-line scope. Warn if the scope contains
1281: * nothing but trailing whitespace.
1.2 schwarze 1282: */
1283:
1284: if ('\0' == (*bufp)[pos])
1.46 schwarze 1285: mandoc_msg(MANDOCERR_NOARGS, r->parse, ln, ppos, NULL);
1.2 schwarze 1286:
1.46 schwarze 1287: r->last->endspan = 1;
1.1 schwarze 1288:
1.46 schwarze 1289: out:
1.2 schwarze 1290: *offs = pos;
1291: return(ROFF_RERUN);
1.1 schwarze 1292: }
1293:
1.2 schwarze 1294: static enum rofferr
1.7 schwarze 1295: roff_ds(ROFF_ARGS)
1296: {
1.85 schwarze 1297: char *string;
1298: const char *name;
1299: size_t namesz;
1.10 schwarze 1300:
1301: /*
1.85 schwarze 1302: * The first word is the name of the string.
1303: * If it is empty or terminated by an escape sequence,
1304: * abort the `ds' request without defining anything.
1.10 schwarze 1305: */
1.7 schwarze 1306:
1.85 schwarze 1307: name = string = *bufp + pos;
1.7 schwarze 1308: if ('\0' == *name)
1309: return(ROFF_IGN);
1310:
1.85 schwarze 1311: namesz = roff_getname(r, &string, ln, pos);
1312: if ('\\' == name[namesz])
1313: return(ROFF_IGN);
1314:
1315: /* Read past the initial double-quote, if any. */
1.28 schwarze 1316: if ('"' == *string)
1.7 schwarze 1317: string++;
1318:
1.10 schwarze 1319: /* The rest is the value. */
1.85 schwarze 1320: roff_setstrn(&r->strtab, name, namesz, string, strlen(string),
1321: ROFF_as == tok);
1.7 schwarze 1322: return(ROFF_IGN);
1323: }
1324:
1.77 schwarze 1325: /*
1326: * Parse a single operator, one or two characters long.
1327: * If the operator is recognized, return success and advance the
1328: * parse point, else return failure and let the parse point unchanged.
1329: */
1330: static int
1331: roff_getop(const char *v, int *pos, char *res)
1332: {
1333:
1334: *res = v[*pos];
1335:
1336: switch (*res) {
1.80 schwarze 1337: case '+':
1.77 schwarze 1338: /* FALLTHROUGH */
1.80 schwarze 1339: case '-':
1.77 schwarze 1340: /* FALLTHROUGH */
1.80 schwarze 1341: case '*':
1.77 schwarze 1342: /* FALLTHROUGH */
1.80 schwarze 1343: case '/':
1.77 schwarze 1344: /* FALLTHROUGH */
1.80 schwarze 1345: case '%':
1.77 schwarze 1346: /* FALLTHROUGH */
1.80 schwarze 1347: case '&':
1.77 schwarze 1348: /* FALLTHROUGH */
1.80 schwarze 1349: case ':':
1.77 schwarze 1350: break;
1351: case '<':
1352: switch (v[*pos + 1]) {
1.80 schwarze 1353: case '=':
1.77 schwarze 1354: *res = 'l';
1355: (*pos)++;
1356: break;
1.80 schwarze 1357: case '>':
1.77 schwarze 1358: *res = '!';
1359: (*pos)++;
1360: break;
1.80 schwarze 1361: case '?':
1.77 schwarze 1362: *res = 'i';
1363: (*pos)++;
1364: break;
1365: default:
1366: break;
1367: }
1368: break;
1369: case '>':
1370: switch (v[*pos + 1]) {
1.80 schwarze 1371: case '=':
1.77 schwarze 1372: *res = 'g';
1373: (*pos)++;
1374: break;
1.80 schwarze 1375: case '?':
1.77 schwarze 1376: *res = 'a';
1377: (*pos)++;
1378: break;
1379: default:
1380: break;
1381: }
1382: break;
1383: case '=':
1384: if ('=' == v[*pos + 1])
1385: (*pos)++;
1386: break;
1387: default:
1388: return(0);
1389: }
1390: (*pos)++;
1391:
1392: return(*res);
1393: }
1394:
1395: /*
1396: * Evaluate either a parenthesized numeric expression
1397: * or a single signed integer number.
1398: */
1399: static int
1400: roff_evalpar(const char *v, int *pos, int *res)
1401: {
1402:
1403: if ('(' != v[*pos])
1404: return(roff_getnum(v, pos, res));
1405:
1406: (*pos)++;
1407: if ( ! roff_evalnum(v, pos, res, 1))
1408: return(0);
1409:
1.79 schwarze 1410: /*
1411: * Omission of the closing parenthesis
1412: * is an error in validation mode,
1413: * but ignored in evaluation mode.
1414: */
1415:
1.77 schwarze 1416: if (')' == v[*pos])
1417: (*pos)++;
1.79 schwarze 1418: else if (NULL == res)
1419: return(0);
1.77 schwarze 1420:
1421: return(1);
1422: }
1423:
1424: /*
1425: * Evaluate a complete numeric expression.
1426: * Proceed left to right, there is no concept of precedence.
1427: */
1428: static int
1429: roff_evalnum(const char *v, int *pos, int *res, int skipwhite)
1430: {
1431: int mypos, operand2;
1432: char operator;
1433:
1434: if (NULL == pos) {
1435: mypos = 0;
1436: pos = &mypos;
1437: }
1438:
1439: if (skipwhite)
1440: while (isspace((unsigned char)v[*pos]))
1441: (*pos)++;
1442:
1443: if ( ! roff_evalpar(v, pos, res))
1444: return(0);
1445:
1446: while (1) {
1447: if (skipwhite)
1448: while (isspace((unsigned char)v[*pos]))
1449: (*pos)++;
1450:
1451: if ( ! roff_getop(v, pos, &operator))
1452: break;
1453:
1454: if (skipwhite)
1455: while (isspace((unsigned char)v[*pos]))
1456: (*pos)++;
1457:
1458: if ( ! roff_evalpar(v, pos, &operand2))
1459: return(0);
1460:
1461: if (skipwhite)
1462: while (isspace((unsigned char)v[*pos]))
1463: (*pos)++;
1.79 schwarze 1464:
1465: if (NULL == res)
1466: continue;
1.77 schwarze 1467:
1468: switch (operator) {
1.80 schwarze 1469: case '+':
1.77 schwarze 1470: *res += operand2;
1471: break;
1.80 schwarze 1472: case '-':
1.77 schwarze 1473: *res -= operand2;
1474: break;
1.80 schwarze 1475: case '*':
1.77 schwarze 1476: *res *= operand2;
1477: break;
1.80 schwarze 1478: case '/':
1.77 schwarze 1479: *res /= operand2;
1480: break;
1.80 schwarze 1481: case '%':
1.77 schwarze 1482: *res %= operand2;
1483: break;
1.80 schwarze 1484: case '<':
1.77 schwarze 1485: *res = *res < operand2;
1486: break;
1.80 schwarze 1487: case '>':
1.77 schwarze 1488: *res = *res > operand2;
1489: break;
1.80 schwarze 1490: case 'l':
1.77 schwarze 1491: *res = *res <= operand2;
1492: break;
1.80 schwarze 1493: case 'g':
1.77 schwarze 1494: *res = *res >= operand2;
1495: break;
1.80 schwarze 1496: case '=':
1.77 schwarze 1497: *res = *res == operand2;
1498: break;
1.80 schwarze 1499: case '!':
1.77 schwarze 1500: *res = *res != operand2;
1501: break;
1.80 schwarze 1502: case '&':
1.77 schwarze 1503: *res = *res && operand2;
1504: break;
1.80 schwarze 1505: case ':':
1.77 schwarze 1506: *res = *res || operand2;
1507: break;
1.80 schwarze 1508: case 'i':
1.77 schwarze 1509: if (operand2 < *res)
1510: *res = operand2;
1511: break;
1.80 schwarze 1512: case 'a':
1.77 schwarze 1513: if (operand2 > *res)
1514: *res = operand2;
1515: break;
1516: default:
1517: abort();
1518: }
1519: }
1520: return(1);
1521: }
1522:
1.52 schwarze 1523: void
1.60 schwarze 1524: roff_setreg(struct roff *r, const char *name, int val, char sign)
1.41 schwarze 1525: {
1.52 schwarze 1526: struct roffreg *reg;
1527:
1528: /* Search for an existing register with the same name. */
1529: reg = r->regtab;
1530:
1531: while (reg && strcmp(name, reg->key.p))
1532: reg = reg->next;
1.41 schwarze 1533:
1.52 schwarze 1534: if (NULL == reg) {
1535: /* Create a new register. */
1536: reg = mandoc_malloc(sizeof(struct roffreg));
1537: reg->key.p = mandoc_strdup(name);
1538: reg->key.sz = strlen(name);
1.60 schwarze 1539: reg->val = 0;
1.52 schwarze 1540: reg->next = r->regtab;
1541: r->regtab = reg;
1542: }
1543:
1.60 schwarze 1544: if ('+' == sign)
1545: reg->val += val;
1546: else if ('-' == sign)
1547: reg->val -= val;
1548: else
1549: reg->val = val;
1.41 schwarze 1550: }
1551:
1.65 schwarze 1552: /*
1553: * Handle some predefined read-only number registers.
1554: * For now, return -1 if the requested register is not predefined;
1555: * in case a predefined read-only register having the value -1
1556: * were to turn up, another special value would have to be chosen.
1557: */
1558: static int
1559: roff_getregro(const char *name)
1560: {
1561:
1562: switch (*name) {
1.80 schwarze 1563: case 'A': /* ASCII approximation mode is always off. */
1.65 schwarze 1564: return(0);
1.80 schwarze 1565: case 'g': /* Groff compatibility mode is always on. */
1.65 schwarze 1566: return(1);
1.80 schwarze 1567: case 'H': /* Fixed horizontal resolution. */
1.65 schwarze 1568: return (24);
1.80 schwarze 1569: case 'j': /* Always adjust left margin only. */
1.65 schwarze 1570: return(0);
1.80 schwarze 1571: case 'T': /* Some output device is always defined. */
1.65 schwarze 1572: return(1);
1.80 schwarze 1573: case 'V': /* Fixed vertical resolution. */
1.65 schwarze 1574: return (40);
1575: default:
1576: return (-1);
1577: }
1578: }
1579:
1.53 schwarze 1580: int
1.52 schwarze 1581: roff_getreg(const struct roff *r, const char *name)
1.41 schwarze 1582: {
1.52 schwarze 1583: struct roffreg *reg;
1.65 schwarze 1584: int val;
1585:
1586: if ('.' == name[0] && '\0' != name[1] && '\0' == name[2]) {
1587: val = roff_getregro(name + 1);
1588: if (-1 != val)
1589: return (val);
1590: }
1.52 schwarze 1591:
1592: for (reg = r->regtab; reg; reg = reg->next)
1593: if (0 == strcmp(name, reg->key.p))
1.53 schwarze 1594: return(reg->val);
1595:
1596: return(0);
1597: }
1598:
1599: static int
1600: roff_getregn(const struct roff *r, const char *name, size_t len)
1601: {
1602: struct roffreg *reg;
1.65 schwarze 1603: int val;
1604:
1605: if ('.' == name[0] && 2 == len) {
1606: val = roff_getregro(name + 1);
1607: if (-1 != val)
1608: return (val);
1609: }
1.53 schwarze 1610:
1611: for (reg = r->regtab; reg; reg = reg->next)
1612: if (len == reg->key.sz &&
1613: 0 == strncmp(name, reg->key.p, len))
1614: return(reg->val);
1.41 schwarze 1615:
1.52 schwarze 1616: return(0);
1.41 schwarze 1617: }
1618:
1.52 schwarze 1619: static void
1620: roff_freereg(struct roffreg *reg)
1.41 schwarze 1621: {
1.52 schwarze 1622: struct roffreg *old_reg;
1.41 schwarze 1623:
1.52 schwarze 1624: while (NULL != reg) {
1625: free(reg->key.p);
1626: old_reg = reg;
1627: reg = reg->next;
1628: free(old_reg);
1629: }
1.41 schwarze 1630: }
1.7 schwarze 1631:
1632: static enum rofferr
1.6 schwarze 1633: roff_nr(ROFF_ARGS)
1.1 schwarze 1634: {
1.85 schwarze 1635: char *key, *val;
1636: size_t keysz;
1.37 schwarze 1637: int iv;
1.60 schwarze 1638: char sign;
1.6 schwarze 1639:
1.85 schwarze 1640: key = val = *bufp + pos;
1641: if ('\0' == *key)
1642: return(ROFF_IGN);
1643:
1644: keysz = roff_getname(r, &val, ln, pos);
1645: if ('\\' == key[keysz])
1646: return(ROFF_IGN);
1647: key[keysz] = '\0';
1.6 schwarze 1648:
1.60 schwarze 1649: sign = *val;
1650: if ('+' == sign || '-' == sign)
1651: val++;
1652:
1.77 schwarze 1653: if (roff_evalnum(val, NULL, &iv, 0))
1654: roff_setreg(r, key, iv, sign);
1.1 schwarze 1655:
1.76 schwarze 1656: return(ROFF_IGN);
1657: }
1658:
1659: static enum rofferr
1660: roff_rr(ROFF_ARGS)
1661: {
1662: struct roffreg *reg, **prev;
1.85 schwarze 1663: char *name, *cp;
1664: size_t namesz;
1.76 schwarze 1665:
1.85 schwarze 1666: name = cp = *bufp + pos;
1667: if ('\0' == *name)
1668: return(ROFF_IGN);
1669: namesz = roff_getname(r, &cp, ln, pos);
1670: name[namesz] = '\0';
1.76 schwarze 1671:
1672: prev = &r->regtab;
1673: while (1) {
1674: reg = *prev;
1675: if (NULL == reg || !strcmp(name, reg->key.p))
1676: break;
1677: prev = ®->next;
1678: }
1679: if (NULL != reg) {
1680: *prev = reg->next;
1681: free(reg->key.p);
1682: free(reg);
1683: }
1.29 schwarze 1684: return(ROFF_IGN);
1685: }
1686:
1687: static enum rofferr
1688: roff_rm(ROFF_ARGS)
1689: {
1690: const char *name;
1691: char *cp;
1.85 schwarze 1692: size_t namesz;
1.29 schwarze 1693:
1694: cp = *bufp + pos;
1695: while ('\0' != *cp) {
1.85 schwarze 1696: name = cp;
1697: namesz = roff_getname(r, &cp, ln, (int)(cp - *bufp));
1698: roff_setstrn(&r->strtab, name, namesz, NULL, 0, 0);
1699: if ('\\' == name[namesz])
1700: break;
1.29 schwarze 1701: }
1.51 schwarze 1702: return(ROFF_IGN);
1703: }
1704:
1705: static enum rofferr
1706: roff_it(ROFF_ARGS)
1707: {
1708: char *cp;
1709: size_t len;
1710: int iv;
1711:
1712: /* Parse the number of lines. */
1713: cp = *bufp + pos;
1714: len = strcspn(cp, " \t");
1715: cp[len] = '\0';
1716: if ((iv = mandoc_strntoi(cp, len, 10)) <= 0) {
1717: mandoc_msg(MANDOCERR_NUMERIC, r->parse,
1.80 schwarze 1718: ln, ppos, *bufp + 1);
1.51 schwarze 1719: return(ROFF_IGN);
1720: }
1721: cp += len + 1;
1722:
1723: /* Arm the input line trap. */
1724: roffit_lines = iv;
1725: roffit_macro = mandoc_strdup(cp);
1.2 schwarze 1726: return(ROFF_IGN);
1.47 schwarze 1727: }
1728:
1729: static enum rofferr
1730: roff_Dd(ROFF_ARGS)
1731: {
1732: const char *const *cp;
1733:
1.72 schwarze 1734: if (0 == ((MPARSE_MDOC | MPARSE_QUICK) & r->options))
1.47 schwarze 1735: for (cp = __mdoc_reserved; *cp; cp++)
1736: roff_setstr(r, *cp, NULL, 0);
1737:
1738: return(ROFF_CONT);
1739: }
1740:
1741: static enum rofferr
1742: roff_TH(ROFF_ARGS)
1743: {
1744: const char *const *cp;
1745:
1.72 schwarze 1746: if (0 == (MPARSE_QUICK & r->options))
1.47 schwarze 1747: for (cp = __man_reserved; *cp; cp++)
1748: roff_setstr(r, *cp, NULL, 0);
1749:
1750: return(ROFF_CONT);
1.14 schwarze 1751: }
1752:
1753: static enum rofferr
1.27 schwarze 1754: roff_TE(ROFF_ARGS)
1755: {
1756:
1757: if (NULL == r->tbl)
1.35 schwarze 1758: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.27 schwarze 1759: else
1.41 schwarze 1760: tbl_end(&r->tbl);
1.27 schwarze 1761:
1762: return(ROFF_IGN);
1763: }
1764:
1765: static enum rofferr
1766: roff_T_(ROFF_ARGS)
1767: {
1768:
1769: if (NULL == r->tbl)
1.35 schwarze 1770: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.27 schwarze 1771: else
1772: tbl_restart(ppos, ln, r->tbl);
1773:
1774: return(ROFF_IGN);
1775: }
1776:
1.41 schwarze 1777: #if 0
1778: static int
1779: roff_closeeqn(struct roff *r)
1780: {
1781:
1782: return(r->eqn && ROFF_EQN == eqn_end(&r->eqn) ? 1 : 0);
1783: }
1784: #endif
1785:
1786: static void
1.80 schwarze 1787: roff_openeqn(struct roff *r, const char *name, int line,
1.41 schwarze 1788: int offs, const char *buf)
1.32 schwarze 1789: {
1.41 schwarze 1790: struct eqn_node *e;
1791: int poff;
1.32 schwarze 1792:
1793: assert(NULL == r->eqn);
1.41 schwarze 1794: e = eqn_alloc(name, offs, line, r->parse);
1.32 schwarze 1795:
1796: if (r->last_eqn)
1797: r->last_eqn->next = e;
1798: else
1799: r->first_eqn = r->last_eqn = e;
1800:
1801: r->eqn = r->last_eqn = e;
1.41 schwarze 1802:
1803: if (buf) {
1804: poff = 0;
1805: eqn_read(&r->eqn, line, buf, offs, &poff);
1806: }
1807: }
1808:
1809: static enum rofferr
1810: roff_EQ(ROFF_ARGS)
1811: {
1812:
1813: roff_openeqn(r, *bufp + pos, ln, ppos, NULL);
1.32 schwarze 1814: return(ROFF_IGN);
1815: }
1816:
1817: static enum rofferr
1818: roff_EN(ROFF_ARGS)
1819: {
1820:
1.35 schwarze 1821: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.32 schwarze 1822: return(ROFF_IGN);
1823: }
1824:
1825: static enum rofferr
1.27 schwarze 1826: roff_TS(ROFF_ARGS)
1827: {
1.49 schwarze 1828: struct tbl_node *tbl;
1.27 schwarze 1829:
1830: if (r->tbl) {
1.35 schwarze 1831: mandoc_msg(MANDOCERR_SCOPEBROKEN, r->parse, ln, ppos, NULL);
1.41 schwarze 1832: tbl_end(&r->tbl);
1.27 schwarze 1833: }
1834:
1.49 schwarze 1835: tbl = tbl_alloc(ppos, ln, r->parse);
1.27 schwarze 1836:
1837: if (r->last_tbl)
1.49 schwarze 1838: r->last_tbl->next = tbl;
1.27 schwarze 1839: else
1.49 schwarze 1840: r->first_tbl = r->last_tbl = tbl;
1.27 schwarze 1841:
1.49 schwarze 1842: r->tbl = r->last_tbl = tbl;
1.27 schwarze 1843: return(ROFF_IGN);
1844: }
1845:
1846: static enum rofferr
1.48 schwarze 1847: roff_cc(ROFF_ARGS)
1848: {
1849: const char *p;
1850:
1851: p = *bufp + pos;
1852:
1853: if ('\0' == *p || '.' == (r->control = *p++))
1854: r->control = 0;
1855:
1856: if ('\0' != *p)
1857: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
1858:
1859: return(ROFF_IGN);
1860: }
1861:
1862: static enum rofferr
1.42 schwarze 1863: roff_tr(ROFF_ARGS)
1864: {
1865: const char *p, *first, *second;
1866: size_t fsz, ssz;
1867: enum mandoc_esc esc;
1868:
1869: p = *bufp + pos;
1870:
1871: if ('\0' == *p) {
1872: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
1873: return(ROFF_IGN);
1874: }
1875:
1876: while ('\0' != *p) {
1877: fsz = ssz = 1;
1878:
1879: first = p++;
1880: if ('\\' == *first) {
1881: esc = mandoc_escape(&p, NULL, NULL);
1882: if (ESCAPE_ERROR == esc) {
1.80 schwarze 1883: mandoc_msg(MANDOCERR_BADESCAPE,
1884: r->parse, ln,
1885: (int)(p - *bufp), NULL);
1.42 schwarze 1886: return(ROFF_IGN);
1887: }
1888: fsz = (size_t)(p - first);
1889: }
1890:
1891: second = p++;
1892: if ('\\' == *second) {
1893: esc = mandoc_escape(&p, NULL, NULL);
1894: if (ESCAPE_ERROR == esc) {
1.80 schwarze 1895: mandoc_msg(MANDOCERR_BADESCAPE,
1896: r->parse, ln,
1897: (int)(p - *bufp), NULL);
1.42 schwarze 1898: return(ROFF_IGN);
1899: }
1900: ssz = (size_t)(p - second);
1901: } else if ('\0' == *second) {
1.80 schwarze 1902: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse,
1903: ln, (int)(p - *bufp), NULL);
1.42 schwarze 1904: second = " ";
1905: p--;
1906: }
1907:
1908: if (fsz > 1) {
1.80 schwarze 1909: roff_setstrn(&r->xmbtab, first, fsz,
1910: second, ssz, 0);
1.42 schwarze 1911: continue;
1912: }
1913:
1914: if (NULL == r->xtab)
1.80 schwarze 1915: r->xtab = mandoc_calloc(128,
1916: sizeof(struct roffstr));
1.42 schwarze 1917:
1918: free(r->xtab[(int)*first].p);
1919: r->xtab[(int)*first].p = mandoc_strndup(second, ssz);
1920: r->xtab[(int)*first].sz = ssz;
1921: }
1922:
1923: return(ROFF_IGN);
1924: }
1925:
1926: static enum rofferr
1.14 schwarze 1927: roff_so(ROFF_ARGS)
1928: {
1929: char *name;
1.15 schwarze 1930:
1.83 schwarze 1931: name = *bufp + pos;
1932: mandoc_vmsg(MANDOCERR_SO, r->parse, ln, ppos, ".so %s", name);
1.14 schwarze 1933:
1.22 schwarze 1934: /*
1935: * Handle `so'. Be EXTREMELY careful, as we shouldn't be
1936: * opening anything that's not in our cwd or anything beneath
1937: * it. Thus, explicitly disallow traversing up the file-system
1938: * or using absolute paths.
1939: */
1940:
1.14 schwarze 1941: if ('/' == *name || strstr(name, "../") || strstr(name, "/..")) {
1.83 schwarze 1942: mandoc_vmsg(MANDOCERR_SO_PATH, r->parse, ln, ppos,
1943: ".so %s", name);
1.14 schwarze 1944: return(ROFF_ERR);
1945: }
1946:
1947: *offs = pos;
1948: return(ROFF_SO);
1.7 schwarze 1949: }
1950:
1.16 schwarze 1951: static enum rofferr
1952: roff_userdef(ROFF_ARGS)
1.12 schwarze 1953: {
1.16 schwarze 1954: const char *arg[9];
1955: char *cp, *n1, *n2;
1.25 schwarze 1956: int i;
1.12 schwarze 1957:
1.16 schwarze 1958: /*
1959: * Collect pointers to macro argument strings
1.61 schwarze 1960: * and NUL-terminate them.
1.16 schwarze 1961: */
1962: cp = *bufp + pos;
1.25 schwarze 1963: for (i = 0; i < 9; i++)
1.26 schwarze 1964: arg[i] = '\0' == *cp ? "" :
1.35 schwarze 1965: mandoc_getarg(r->parse, &cp, ln, &pos);
1.16 schwarze 1966:
1967: /*
1968: * Expand macro arguments.
1.12 schwarze 1969: */
1.16 schwarze 1970: *szp = 0;
1971: n1 = cp = mandoc_strdup(r->current_string);
1972: while (NULL != (cp = strstr(cp, "\\$"))) {
1973: i = cp[2] - '1';
1974: if (0 > i || 8 < i) {
1975: /* Not an argument invocation. */
1976: cp += 2;
1977: continue;
1978: }
1.82 schwarze 1979: *cp = '\0';
1980: *szp = mandoc_asprintf(&n2, "%s%s%s",
1981: n1, arg[i], cp + 3) + 1;
1.16 schwarze 1982: cp = n2 + (cp - n1);
1983: free(n1);
1984: n1 = n2;
1.12 schwarze 1985: }
1986:
1.16 schwarze 1987: /*
1988: * Replace the macro invocation
1989: * by the expanded macro.
1990: */
1991: free(*bufp);
1992: *bufp = n1;
1993: if (0 == *szp)
1994: *szp = strlen(*bufp) + 1;
1995:
1.19 schwarze 1996: return(*szp > 1 && '\n' == (*bufp)[(int)*szp - 2] ?
1.16 schwarze 1997: ROFF_REPARSE : ROFF_APPEND);
1.12 schwarze 1998: }
1.28 schwarze 1999:
1.85 schwarze 2000: static size_t
1.28 schwarze 2001: roff_getname(struct roff *r, char **cpp, int ln, int pos)
2002: {
2003: char *name, *cp;
1.85 schwarze 2004: size_t namesz;
1.28 schwarze 2005:
2006: name = *cpp;
2007: if ('\0' == *name)
1.85 schwarze 2008: return(0);
1.28 schwarze 2009:
1.85 schwarze 2010: /* Read until end of name and terminate it with NUL. */
2011: for (cp = name; 1; cp++) {
2012: if ('\0' == *cp || ' ' == *cp) {
2013: namesz = cp - name;
2014: break;
2015: }
1.28 schwarze 2016: if ('\\' != *cp)
2017: continue;
2018: cp++;
2019: if ('\\' == *cp)
2020: continue;
1.85 schwarze 2021: namesz = cp - name - 1;
1.35 schwarze 2022: mandoc_msg(MANDOCERR_NAMESC, r->parse, ln, pos, NULL);
1.85 schwarze 2023: mandoc_escape((const char **)&cp, NULL, NULL);
2024: break;
1.28 schwarze 2025: }
2026:
2027: /* Read past spaces. */
2028: while (' ' == *cp)
2029: cp++;
2030:
2031: *cpp = cp;
1.85 schwarze 2032: return(namesz);
1.28 schwarze 2033: }
2034:
1.16 schwarze 2035: /*
2036: * Store *string into the user-defined string called *name.
2037: * To clear an existing entry, call with (*r, *name, NULL, 0).
1.66 schwarze 2038: * append == 0: replace mode
2039: * append == 1: single-line append mode
2040: * append == 2: multiline append mode, append '\n' after each call
1.16 schwarze 2041: */
1.8 schwarze 2042: static void
1.16 schwarze 2043: roff_setstr(struct roff *r, const char *name, const char *string,
1.66 schwarze 2044: int append)
1.7 schwarze 2045: {
1.42 schwarze 2046:
2047: roff_setstrn(&r->strtab, name, strlen(name), string,
1.80 schwarze 2048: string ? strlen(string) : 0, append);
1.42 schwarze 2049: }
2050:
2051: static void
2052: roff_setstrn(struct roffkv **r, const char *name, size_t namesz,
1.66 schwarze 2053: const char *string, size_t stringsz, int append)
1.42 schwarze 2054: {
2055: struct roffkv *n;
2056: char *c;
2057: int i;
2058: size_t oldch, newch;
1.7 schwarze 2059:
1.16 schwarze 2060: /* Search for an existing string with the same name. */
1.42 schwarze 2061: n = *r;
2062:
1.84 schwarze 2063: while (n && (namesz != n->key.sz ||
2064: strncmp(n->key.p, name, namesz)))
1.7 schwarze 2065: n = n->next;
1.8 schwarze 2066:
2067: if (NULL == n) {
1.16 schwarze 2068: /* Create a new string table entry. */
1.42 schwarze 2069: n = mandoc_malloc(sizeof(struct roffkv));
2070: n->key.p = mandoc_strndup(name, namesz);
2071: n->key.sz = namesz;
2072: n->val.p = NULL;
2073: n->val.sz = 0;
2074: n->next = *r;
2075: *r = n;
1.66 schwarze 2076: } else if (0 == append) {
1.42 schwarze 2077: free(n->val.p);
2078: n->val.p = NULL;
2079: n->val.sz = 0;
1.16 schwarze 2080: }
2081:
2082: if (NULL == string)
2083: return;
2084:
2085: /*
2086: * One additional byte for the '\n' in multiline mode,
2087: * and one for the terminating '\0'.
2088: */
1.66 schwarze 2089: newch = stringsz + (1 < append ? 2u : 1u);
1.42 schwarze 2090:
2091: if (NULL == n->val.p) {
2092: n->val.p = mandoc_malloc(newch);
2093: *n->val.p = '\0';
1.16 schwarze 2094: oldch = 0;
2095: } else {
1.42 schwarze 2096: oldch = n->val.sz;
2097: n->val.p = mandoc_realloc(n->val.p, oldch + newch);
1.16 schwarze 2098: }
2099:
2100: /* Skip existing content in the destination buffer. */
1.42 schwarze 2101: c = n->val.p + (int)oldch;
1.16 schwarze 2102:
2103: /* Append new content to the destination buffer. */
1.42 schwarze 2104: i = 0;
2105: while (i < (int)stringsz) {
1.16 schwarze 2106: /*
2107: * Rudimentary roff copy mode:
2108: * Handle escaped backslashes.
2109: */
1.42 schwarze 2110: if ('\\' == string[i] && '\\' == string[i + 1])
2111: i++;
2112: *c++ = string[i++];
1.16 schwarze 2113: }
1.8 schwarze 2114:
1.16 schwarze 2115: /* Append terminating bytes. */
1.66 schwarze 2116: if (1 < append)
1.16 schwarze 2117: *c++ = '\n';
1.42 schwarze 2118:
1.16 schwarze 2119: *c = '\0';
1.42 schwarze 2120: n->val.sz = (int)(c - n->val.p);
1.7 schwarze 2121: }
2122:
1.8 schwarze 2123: static const char *
2124: roff_getstrn(const struct roff *r, const char *name, size_t len)
1.7 schwarze 2125: {
1.42 schwarze 2126: const struct roffkv *n;
1.64 schwarze 2127: int i;
1.7 schwarze 2128:
1.42 schwarze 2129: for (n = r->strtab; n; n = n->next)
1.80 schwarze 2130: if (0 == strncmp(name, n->key.p, len) &&
2131: '\0' == n->key.p[(int)len])
1.42 schwarze 2132: return(n->val.p);
1.64 schwarze 2133:
2134: for (i = 0; i < PREDEFS_MAX; i++)
2135: if (0 == strncmp(name, predefs[i].name, len) &&
2136: '\0' == predefs[i].name[(int)len])
2137: return(predefs[i].str);
1.8 schwarze 2138:
1.42 schwarze 2139: return(NULL);
1.7 schwarze 2140: }
2141:
1.8 schwarze 2142: static void
1.42 schwarze 2143: roff_freestr(struct roffkv *r)
1.7 schwarze 2144: {
1.42 schwarze 2145: struct roffkv *n, *nn;
1.7 schwarze 2146:
1.42 schwarze 2147: for (n = r; n; n = nn) {
2148: free(n->key.p);
2149: free(n->val.p);
1.7 schwarze 2150: nn = n->next;
2151: free(n);
2152: }
1.27 schwarze 2153: }
2154:
2155: const struct tbl_span *
2156: roff_span(const struct roff *r)
2157: {
1.80 schwarze 2158:
1.27 schwarze 2159: return(r->tbl ? tbl_span(r->tbl) : NULL);
1.32 schwarze 2160: }
2161:
2162: const struct eqn *
2163: roff_eqn(const struct roff *r)
2164: {
1.80 schwarze 2165:
1.32 schwarze 2166: return(r->last_eqn ? &r->last_eqn->eqn : NULL);
1.42 schwarze 2167: }
2168:
2169: /*
2170: * Duplicate an input string, making the appropriate character
2171: * conversations (as stipulated by `tr') along the way.
2172: * Returns a heap-allocated string with all the replacements made.
2173: */
2174: char *
2175: roff_strdup(const struct roff *r, const char *p)
2176: {
2177: const struct roffkv *cp;
2178: char *res;
2179: const char *pp;
2180: size_t ssz, sz;
2181: enum mandoc_esc esc;
2182:
2183: if (NULL == r->xmbtab && NULL == r->xtab)
2184: return(mandoc_strdup(p));
2185: else if ('\0' == *p)
2186: return(mandoc_strdup(""));
2187:
2188: /*
2189: * Step through each character looking for term matches
2190: * (remember that a `tr' can be invoked with an escape, which is
2191: * a glyph but the escape is multi-character).
2192: * We only do this if the character hash has been initialised
2193: * and the string is >0 length.
2194: */
2195:
2196: res = NULL;
2197: ssz = 0;
2198:
2199: while ('\0' != *p) {
2200: if ('\\' != *p && r->xtab && r->xtab[(int)*p].p) {
2201: sz = r->xtab[(int)*p].sz;
2202: res = mandoc_realloc(res, ssz + sz + 1);
2203: memcpy(res + ssz, r->xtab[(int)*p].p, sz);
2204: ssz += sz;
2205: p++;
2206: continue;
2207: } else if ('\\' != *p) {
2208: res = mandoc_realloc(res, ssz + 2);
2209: res[ssz++] = *p++;
2210: continue;
2211: }
2212:
2213: /* Search for term matches. */
2214: for (cp = r->xmbtab; cp; cp = cp->next)
2215: if (0 == strncmp(p, cp->key.p, cp->key.sz))
2216: break;
2217:
2218: if (NULL != cp) {
2219: /*
2220: * A match has been found.
2221: * Append the match to the array and move
2222: * forward by its keysize.
2223: */
1.80 schwarze 2224: res = mandoc_realloc(res,
2225: ssz + cp->val.sz + 1);
1.42 schwarze 2226: memcpy(res + ssz, cp->val.p, cp->val.sz);
2227: ssz += cp->val.sz;
2228: p += (int)cp->key.sz;
2229: continue;
2230: }
2231:
2232: /*
2233: * Handle escapes carefully: we need to copy
2234: * over just the escape itself, or else we might
2235: * do replacements within the escape itself.
2236: * Make sure to pass along the bogus string.
2237: */
2238: pp = p++;
2239: esc = mandoc_escape(&p, NULL, NULL);
2240: if (ESCAPE_ERROR == esc) {
2241: sz = strlen(pp);
2242: res = mandoc_realloc(res, ssz + sz + 1);
2243: memcpy(res + ssz, pp, sz);
2244: break;
2245: }
1.80 schwarze 2246: /*
2247: * We bail out on bad escapes.
1.42 schwarze 2248: * No need to warn: we already did so when
2249: * roff_res() was called.
2250: */
2251: sz = (int)(p - pp);
2252: res = mandoc_realloc(res, ssz + sz + 1);
2253: memcpy(res + ssz, pp, sz);
2254: ssz += sz;
2255: }
2256:
2257: res[(int)ssz] = '\0';
2258: return(res);
1.48 schwarze 2259: }
2260:
2261: /*
1.80 schwarze 2262: * Find out whether a line is a macro line or not.
1.48 schwarze 2263: * If it is, adjust the current position and return one; if it isn't,
2264: * return zero and don't change the current position.
2265: * If the control character has been set with `.cc', then let that grain
2266: * precedence.
2267: * This is slighly contrary to groff, where using the non-breaking
2268: * control character when `cc' has been invoked will cause the
2269: * non-breaking macro contents to be printed verbatim.
2270: */
2271: int
2272: roff_getcontrol(const struct roff *r, const char *cp, int *ppos)
2273: {
2274: int pos;
2275:
2276: pos = *ppos;
2277:
2278: if (0 != r->control && cp[pos] == r->control)
2279: pos++;
2280: else if (0 != r->control)
2281: return(0);
2282: else if ('\\' == cp[pos] && '.' == cp[pos + 1])
2283: pos += 2;
2284: else if ('.' == cp[pos] || '\'' == cp[pos])
2285: pos++;
2286: else
2287: return(0);
2288:
2289: while (' ' == cp[pos] || '\t' == cp[pos])
2290: pos++;
2291:
2292: *ppos = pos;
2293: return(1);
1.1 schwarze 2294: }
![[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