Annotation of src/usr.bin/mandoc/roff.c, Revision 1.80
1.80 ! schwarze 1: /* $Id: roff.c,v 1.79 2014/04/08 01:36:50 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 *);
1.28 schwarze 186: static char *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: {
1.53 schwarze 487: char ubuf[12]; /* buffer to print the number */
1.78 schwarze 488: const char *start; /* start of the string to process */
1.23 schwarze 489: const char *stesc; /* start of an escape sequence ('\\') */
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 */
1.78 schwarze 496: size_t ressz; /* size of the replacement string */
1.53 schwarze 497: int expand_count; /* to avoid infinite loops */
1.79 schwarze 498: int npos; /* position in numeric expression */
499: int irc; /* return code from roff_evalnum() */
500: char term; /* character terminating the escape */
1.8 schwarze 501:
1.43 schwarze 502: expand_count = 0;
1.78 schwarze 503: start = *bufp + pos;
504: stesc = strchr(start, '\0') - 1;
505: while (stesc-- > start) {
1.43 schwarze 506:
1.78 schwarze 507: /* Search backwards for the next backslash. */
508:
509: if ('\\' != *stesc)
510: continue;
511:
512: /* If it is escaped, skip it. */
513:
514: for (cp = stesc - 1; cp >= start; cp--)
515: if ('\\' != *cp)
516: break;
517:
518: if (0 == (stesc - cp) % 2) {
519: stesc = cp;
520: continue;
521: }
1.23 schwarze 522:
1.79 schwarze 523: /* Decide whether to expand or to check only. */
1.23 schwarze 524:
1.79 schwarze 525: term = '\0';
1.78 schwarze 526: cp = stesc + 1;
1.53 schwarze 527: switch (*cp) {
1.80 ! schwarze 528: case '*':
1.53 schwarze 529: res = NULL;
530: break;
1.80 ! schwarze 531: case 'B':
1.79 schwarze 532: /* FALLTHROUGH */
1.80 ! schwarze 533: case 'w':
1.79 schwarze 534: term = cp[1];
535: /* FALLTHROUGH */
1.80 ! schwarze 536: case 'n':
1.53 schwarze 537: res = ubuf;
538: break;
539: default:
1.78 schwarze 540: if (ESCAPE_ERROR == mandoc_escape(&cp, NULL, NULL))
541: mandoc_msg(MANDOCERR_BADESCAPE, r->parse,
542: ln, (int)(stesc - *bufp), NULL);
543: continue;
1.42 schwarze 544: }
545:
1.78 schwarze 546: if (EXPAND_LIMIT < ++expand_count) {
547: mandoc_msg(MANDOCERR_ROFFLOOP, r->parse,
548: ln, (int)(stesc - *bufp), NULL);
549: return(ROFF_IGN);
550: }
1.23 schwarze 551:
552: /*
553: * The third character decides the length
1.53 schwarze 554: * of the name of the string or register.
1.23 schwarze 555: * Save a pointer to the name.
556: */
557:
1.79 schwarze 558: if ('\0' == term) {
559: switch (*++cp) {
1.80 ! schwarze 560: case '\0':
1.79 schwarze 561: maxl = 0;
562: break;
1.80 ! schwarze 563: case '(':
1.79 schwarze 564: cp++;
565: maxl = 2;
566: break;
1.80 ! schwarze 567: case '[':
1.79 schwarze 568: cp++;
569: term = ']';
570: maxl = 0;
571: break;
572: default:
573: maxl = 1;
574: break;
575: }
576: } else {
577: cp += 2;
1.8 schwarze 578: maxl = 0;
579: }
1.23 schwarze 580: stnam = cp;
1.8 schwarze 581:
1.23 schwarze 582: /* Advance to the end of the name. */
1.8 schwarze 583:
1.53 schwarze 584: for (naml = 0; 0 == maxl || naml < maxl; naml++, cp++) {
1.42 schwarze 585: if ('\0' == *cp) {
1.80 ! schwarze 586: mandoc_msg(MANDOCERR_BADESCAPE, r->parse,
! 587: ln, (int)(stesc - *bufp), NULL);
1.79 schwarze 588: break;
1.42 schwarze 589: }
1.79 schwarze 590: if (0 == maxl && *cp == term) {
591: cp++;
1.8 schwarze 592: break;
1.79 schwarze 593: }
1.8 schwarze 594: }
595:
1.23 schwarze 596: /*
597: * Retrieve the replacement string; if it is
598: * undefined, resume searching for escapes.
599: */
600:
1.79 schwarze 601: switch (stesc[1]) {
1.80 ! schwarze 602: case '*':
1.53 schwarze 603: res = roff_getstrn(r, stnam, naml);
1.79 schwarze 604: break;
1.80 ! schwarze 605: case 'B':
1.79 schwarze 606: npos = 0;
607: irc = roff_evalnum(stnam, &npos, NULL, 0);
608: ubuf[0] = irc && stnam + npos + 1 == cp
609: ? '1' : '0';
610: ubuf[1] = '\0';
611: break;
1.80 ! schwarze 612: case 'n':
1.53 schwarze 613: snprintf(ubuf, sizeof(ubuf), "%d",
614: roff_getregn(r, stnam, naml));
1.79 schwarze 615: break;
1.80 ! schwarze 616: case 'w':
1.79 schwarze 617: snprintf(ubuf, sizeof(ubuf), "%d",
618: 24 * (int)naml);
619: break;
620: }
1.8 schwarze 621:
622: if (NULL == res) {
1.80 ! schwarze 623: mandoc_msg(MANDOCERR_BADESCAPE, r->parse,
! 624: ln, (int)(stesc - *bufp), NULL);
1.37 schwarze 625: res = "";
1.8 schwarze 626: }
1.78 schwarze 627: ressz = strlen(res);
1.8 schwarze 628:
1.23 schwarze 629: /* Replace the escape sequence by the string. */
630:
1.78 schwarze 631: *szp += ressz + 1;
632: nbuf = mandoc_malloc(*szp);
1.42 schwarze 633:
1.78 schwarze 634: strlcpy(nbuf, *bufp, (size_t)(stesc - *bufp + 1));
635: strlcat(nbuf, res, *szp);
1.79 schwarze 636: strlcat(nbuf, cp, *szp);
1.8 schwarze 637:
1.78 schwarze 638: /* Prepare for the next replacement. */
1.8 schwarze 639:
1.78 schwarze 640: start = nbuf + pos;
641: stesc = nbuf + (stesc - *bufp) + ressz;
1.8 schwarze 642: free(*bufp);
1.53 schwarze 643: *bufp = nbuf;
1.42 schwarze 644: }
1.45 schwarze 645: return(ROFF_CONT);
1.42 schwarze 646: }
647:
648: /*
1.51 schwarze 649: * Process text streams:
650: * Convert all breakable hyphens into ASCII_HYPH.
651: * Decrement and spring input line trap.
1.42 schwarze 652: */
653: static enum rofferr
1.51 schwarze 654: roff_parsetext(char **bufp, size_t *szp, int pos, int *offs)
1.42 schwarze 655: {
656: size_t sz;
657: const char *start;
1.51 schwarze 658: char *p;
659: int isz;
1.42 schwarze 660: enum mandoc_esc esc;
661:
1.51 schwarze 662: start = p = *bufp + pos;
1.42 schwarze 663:
664: while ('\0' != *p) {
665: sz = strcspn(p, "-\\");
666: p += sz;
667:
668: if ('\0' == *p)
669: break;
670:
671: if ('\\' == *p) {
672: /* Skip over escapes. */
673: p++;
1.62 schwarze 674: esc = mandoc_escape((const char **)&p, NULL, NULL);
1.42 schwarze 675: if (ESCAPE_ERROR == esc)
676: break;
677: continue;
678: } else if (p == start) {
679: p++;
680: continue;
681: }
682:
1.44 schwarze 683: if (isalpha((unsigned char)p[-1]) &&
684: isalpha((unsigned char)p[1]))
1.42 schwarze 685: *p = ASCII_HYPH;
686: p++;
1.8 schwarze 687: }
688:
1.51 schwarze 689: /* Spring the input line trap. */
690: if (1 == roffit_lines) {
1.75 schwarze 691: isz = mandoc_asprintf(&p, "%s\n.%s", *bufp, roffit_macro);
1.51 schwarze 692: free(*bufp);
693: *bufp = p;
694: *szp = isz + 1;
695: *offs = 0;
696: free(roffit_macro);
697: roffit_lines = 0;
698: return(ROFF_REPARSE);
699: } else if (1 < roffit_lines)
700: --roffit_lines;
1.42 schwarze 701: return(ROFF_CONT);
1.8 schwarze 702: }
703:
1.1 schwarze 704: enum rofferr
1.80 ! schwarze 705: roff_parseln(struct roff *r, int ln, char **bufp,
1.6 schwarze 706: size_t *szp, int pos, int *offs)
1.1 schwarze 707: {
708: enum rofft t;
1.27 schwarze 709: enum rofferr e;
1.35 schwarze 710: int ppos, ctl;
1.1 schwarze 711:
1.2 schwarze 712: /*
1.8 schwarze 713: * Run the reserved-word filter only if we have some reserved
714: * words to fill in.
715: */
716:
1.45 schwarze 717: e = roff_res(r, bufp, szp, ln, pos);
718: if (ROFF_IGN == e)
719: return(e);
720: assert(ROFF_CONT == e);
1.8 schwarze 721:
1.35 schwarze 722: ppos = pos;
1.48 schwarze 723: ctl = roff_getcontrol(r, *bufp, &pos);
1.35 schwarze 724:
1.8 schwarze 725: /*
1.2 schwarze 726: * First, if a scope is open and we're not a macro, pass the
727: * text through the macro's filter. If a scope isn't open and
728: * we're not a macro, just let it through.
1.32 schwarze 729: * Finally, if there's an equation scope open, divert it into it
730: * no matter our state.
1.2 schwarze 731: */
732:
1.35 schwarze 733: if (r->last && ! ctl) {
1.2 schwarze 734: t = r->last->tok;
735: assert(roffs[t].text);
1.80 ! schwarze 736: e = (*roffs[t].text)(r, t, bufp, szp, ln, pos, pos, offs);
1.27 schwarze 737: assert(ROFF_IGN == e || ROFF_CONT == e);
1.32 schwarze 738: if (ROFF_CONT != e)
739: return(e);
1.54 schwarze 740: }
741: if (r->eqn)
742: return(eqn_read(&r->eqn, ln, *bufp, ppos, offs));
743: if ( ! ctl) {
1.32 schwarze 744: if (r->tbl)
1.35 schwarze 745: return(tbl_read(r->tbl, ln, *bufp, pos));
1.51 schwarze 746: return(roff_parsetext(bufp, szp, pos, offs));
1.54 schwarze 747: }
1.2 schwarze 748:
749: /*
750: * If a scope is open, go to the child handler for that macro,
751: * as it may want to preprocess before doing anything with it.
1.32 schwarze 752: * Don't do so if an equation is open.
1.2 schwarze 753: */
754:
755: if (r->last) {
1.1 schwarze 756: t = r->last->tok;
757: assert(roffs[t].sub);
1.80 ! schwarze 758: return((*roffs[t].sub)(r, t, bufp, szp,
! 759: ln, ppos, pos, offs));
1.2 schwarze 760: }
761:
762: /*
763: * Lastly, as we've no scope open, try to look up and execute
764: * the new macro. If no macro is found, simply return and let
765: * the compilers handle it.
766: */
767:
1.16 schwarze 768: if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos)))
1.1 schwarze 769: return(ROFF_CONT);
770:
1.2 schwarze 771: assert(roffs[t].proc);
1.80 ! schwarze 772: return((*roffs[t].proc)(r, t, bufp, szp, ln, ppos, pos, offs));
1.2 schwarze 773: }
774:
1.27 schwarze 775: void
1.2 schwarze 776: roff_endparse(struct roff *r)
777: {
1.1 schwarze 778:
1.27 schwarze 779: if (r->last)
1.35 schwarze 780: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.80 ! schwarze 781: r->last->line, r->last->col, NULL);
1.27 schwarze 782:
1.32 schwarze 783: if (r->eqn) {
1.80 ! schwarze 784: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
! 785: r->eqn->eqn.ln, r->eqn->eqn.pos, NULL);
1.41 schwarze 786: eqn_end(&r->eqn);
1.32 schwarze 787: }
788:
1.27 schwarze 789: if (r->tbl) {
1.80 ! schwarze 790: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
! 791: r->tbl->line, r->tbl->pos, NULL);
1.41 schwarze 792: tbl_end(&r->tbl);
1.27 schwarze 793: }
1.1 schwarze 794: }
795:
796: /*
797: * Parse a roff node's type from the input buffer. This must be in the
798: * form of ".foo xxx" in the usual way.
799: */
800: static enum rofft
1.16 schwarze 801: roff_parse(struct roff *r, const char *buf, int *pos)
1.1 schwarze 802: {
1.16 schwarze 803: const char *mac;
804: size_t maclen;
1.1 schwarze 805: enum rofft t;
806:
1.80 ! schwarze 807: if ('\0' == buf[*pos] || '"' == buf[*pos] ||
! 808: '\t' == buf[*pos] || ' ' == buf[*pos])
1.1 schwarze 809: return(ROFF_MAX);
810:
1.68 schwarze 811: /* We stop the macro parse at an escape, tab, space, or nil. */
1.39 schwarze 812:
1.16 schwarze 813: mac = buf + *pos;
1.68 schwarze 814: maclen = strcspn(mac, " \\\t\0");
1.1 schwarze 815:
1.16 schwarze 816: t = (r->current_string = roff_getstrn(r, mac, maclen))
1.42 schwarze 817: ? ROFF_USERDEF : roffhash_find(mac, maclen);
1.1 schwarze 818:
1.34 schwarze 819: *pos += (int)maclen;
1.35 schwarze 820:
1.1 schwarze 821: while (buf[*pos] && ' ' == buf[*pos])
822: (*pos)++;
823:
824: return(t);
825: }
826:
827: static enum rofferr
1.2 schwarze 828: roff_cblock(ROFF_ARGS)
1.1 schwarze 829: {
830:
1.2 schwarze 831: /*
832: * A block-close `..' should only be invoked as a child of an
833: * ignore macro, otherwise raise a warning and just ignore it.
834: */
835:
836: if (NULL == r->last) {
1.35 schwarze 837: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.2 schwarze 838: return(ROFF_IGN);
839: }
1.1 schwarze 840:
1.2 schwarze 841: switch (r->last->tok) {
1.80 ! schwarze 842: case ROFF_am:
1.2 schwarze 843: /* FALLTHROUGH */
1.80 ! schwarze 844: case ROFF_ami:
1.2 schwarze 845: /* FALLTHROUGH */
1.80 ! schwarze 846: case ROFF_am1:
1.2 schwarze 847: /* FALLTHROUGH */
1.80 ! schwarze 848: case ROFF_de:
1.23 schwarze 849: /* ROFF_de1 is remapped to ROFF_de in roff_block(). */
1.2 schwarze 850: /* FALLTHROUGH */
1.80 ! schwarze 851: case ROFF_dei:
1.2 schwarze 852: /* FALLTHROUGH */
1.80 ! schwarze 853: case ROFF_ig:
1.2 schwarze 854: break;
855: default:
1.35 schwarze 856: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.1 schwarze 857: return(ROFF_IGN);
1.2 schwarze 858: }
859:
860: if ((*bufp)[pos])
1.35 schwarze 861: mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.2 schwarze 862:
863: roffnode_pop(r);
864: roffnode_cleanscope(r);
865: return(ROFF_IGN);
866:
867: }
1.1 schwarze 868:
1.2 schwarze 869: static void
870: roffnode_cleanscope(struct roff *r)
871: {
1.1 schwarze 872:
1.2 schwarze 873: while (r->last) {
1.46 schwarze 874: if (--r->last->endspan != 0)
1.2 schwarze 875: break;
876: roffnode_pop(r);
877: }
878: }
1.1 schwarze 879:
1.68 schwarze 880: static void
881: roff_ccond(struct roff *r, int ln, int ppos)
1.2 schwarze 882: {
1.1 schwarze 883:
1.2 schwarze 884: if (NULL == r->last) {
1.35 schwarze 885: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.68 schwarze 886: return;
1.2 schwarze 887: }
1.1 schwarze 888:
1.2 schwarze 889: switch (r->last->tok) {
1.80 ! schwarze 890: case ROFF_el:
1.2 schwarze 891: /* FALLTHROUGH */
1.80 ! schwarze 892: case ROFF_ie:
1.2 schwarze 893: /* FALLTHROUGH */
1.80 ! schwarze 894: case ROFF_if:
1.2 schwarze 895: break;
896: default:
1.35 schwarze 897: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.68 schwarze 898: return;
1.2 schwarze 899: }
1.1 schwarze 900:
1.2 schwarze 901: if (r->last->endspan > -1) {
1.35 schwarze 902: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.68 schwarze 903: return;
1.2 schwarze 904: }
905:
906: roffnode_pop(r);
907: roffnode_cleanscope(r);
1.68 schwarze 908: return;
1.1 schwarze 909: }
910:
911: static enum rofferr
1.2 schwarze 912: roff_block(ROFF_ARGS)
1.1 schwarze 913: {
1.2 schwarze 914: int sv;
915: size_t sz;
1.16 schwarze 916: char *name;
917:
918: name = NULL;
1.2 schwarze 919:
1.16 schwarze 920: if (ROFF_ig != tok) {
921: if ('\0' == (*bufp)[pos]) {
1.35 schwarze 922: mandoc_msg(MANDOCERR_NOARGS, r->parse, ln, ppos, NULL);
1.16 schwarze 923: return(ROFF_IGN);
924: }
1.22 schwarze 925:
926: /*
927: * Re-write `de1', since we don't really care about
928: * groff's strange compatibility mode, into `de'.
929: */
930:
1.18 schwarze 931: if (ROFF_de1 == tok)
932: tok = ROFF_de;
1.16 schwarze 933: if (ROFF_de == tok)
934: name = *bufp + pos;
1.21 schwarze 935: else
1.35 schwarze 936: mandoc_msg(MANDOCERR_REQUEST, r->parse, ln, ppos,
1.21 schwarze 937: roffs[tok].name);
1.22 schwarze 938:
1.33 schwarze 939: while ((*bufp)[pos] && ! isspace((unsigned char)(*bufp)[pos]))
1.2 schwarze 940: pos++;
1.22 schwarze 941:
1.33 schwarze 942: while (isspace((unsigned char)(*bufp)[pos]))
1.16 schwarze 943: (*bufp)[pos++] = '\0';
1.2 schwarze 944: }
945:
1.16 schwarze 946: roffnode_push(r, tok, name, ln, ppos);
947:
948: /*
949: * At the beginning of a `de' macro, clear the existing string
950: * with the same name, if there is one. New content will be
1.66 schwarze 951: * appended from roff_block_text() in multiline mode.
1.16 schwarze 952: */
1.22 schwarze 953:
1.16 schwarze 954: if (ROFF_de == tok)
1.19 schwarze 955: roff_setstr(r, name, "", 0);
1.2 schwarze 956:
957: if ('\0' == (*bufp)[pos])
958: return(ROFF_IGN);
1.1 schwarze 959:
1.22 schwarze 960: /* If present, process the custom end-of-line marker. */
961:
1.2 schwarze 962: sv = pos;
1.33 schwarze 963: while ((*bufp)[pos] && ! isspace((unsigned char)(*bufp)[pos]))
1.2 schwarze 964: pos++;
965:
966: /*
967: * Note: groff does NOT like escape characters in the input.
968: * Instead of detecting this, we're just going to let it fly and
969: * to hell with it.
970: */
971:
972: assert(pos > sv);
973: sz = (size_t)(pos - sv);
974:
975: if (1 == sz && '.' == (*bufp)[sv])
976: return(ROFF_IGN);
977:
1.11 schwarze 978: r->last->end = mandoc_malloc(sz + 1);
1.2 schwarze 979:
980: memcpy(r->last->end, *bufp + sv, sz);
981: r->last->end[(int)sz] = '\0';
982:
983: if ((*bufp)[pos])
1.35 schwarze 984: mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.1 schwarze 985:
986: return(ROFF_IGN);
987: }
988:
989: static enum rofferr
1.2 schwarze 990: roff_block_sub(ROFF_ARGS)
1.1 schwarze 991: {
1.2 schwarze 992: enum rofft t;
993: int i, j;
994:
995: /*
996: * First check whether a custom macro exists at this level. If
997: * it does, then check against it. This is some of groff's
998: * stranger behaviours. If we encountered a custom end-scope
999: * tag and that tag also happens to be a "real" macro, then we
1000: * need to try interpreting it again as a real macro. If it's
1001: * not, then return ignore. Else continue.
1002: */
1003:
1004: if (r->last->end) {
1.35 schwarze 1005: for (i = pos, j = 0; r->last->end[j]; j++, i++)
1.2 schwarze 1006: if ((*bufp)[i] != r->last->end[j])
1007: break;
1.1 schwarze 1008:
1.80 ! schwarze 1009: if ('\0' == r->last->end[j] &&
! 1010: ('\0' == (*bufp)[i] ||
! 1011: ' ' == (*bufp)[i] ||
! 1012: '\t' == (*bufp)[i])) {
1.2 schwarze 1013: roffnode_pop(r);
1014: roffnode_cleanscope(r);
1.1 schwarze 1015:
1.35 schwarze 1016: while (' ' == (*bufp)[i] || '\t' == (*bufp)[i])
1017: i++;
1018:
1019: pos = i;
1.16 schwarze 1020: if (ROFF_MAX != roff_parse(r, *bufp, &pos))
1.2 schwarze 1021: return(ROFF_RERUN);
1022: return(ROFF_IGN);
1023: }
1.1 schwarze 1024: }
1025:
1.2 schwarze 1026: /*
1027: * If we have no custom end-query or lookup failed, then try
1028: * pulling it out of the hashtable.
1029: */
1.1 schwarze 1030:
1.36 schwarze 1031: t = roff_parse(r, *bufp, &pos);
1.1 schwarze 1032:
1.16 schwarze 1033: /*
1034: * Macros other than block-end are only significant
1035: * in `de' blocks; elsewhere, simply throw them away.
1036: */
1037: if (ROFF_cblock != t) {
1038: if (ROFF_de == tok)
1.66 schwarze 1039: roff_setstr(r, r->last->name, *bufp + ppos, 2);
1.1 schwarze 1040: return(ROFF_IGN);
1.16 schwarze 1041: }
1.1 schwarze 1042:
1.2 schwarze 1043: assert(roffs[t].proc);
1.80 ! schwarze 1044: return((*roffs[t].proc)(r, t, bufp, szp, ln, ppos, pos, offs));
1.2 schwarze 1045: }
1046:
1047: static enum rofferr
1048: roff_block_text(ROFF_ARGS)
1049: {
1050:
1.16 schwarze 1051: if (ROFF_de == tok)
1.66 schwarze 1052: roff_setstr(r, r->last->name, *bufp + pos, 2);
1.16 schwarze 1053:
1.2 schwarze 1054: return(ROFF_IGN);
1055: }
1056:
1057: static enum rofferr
1058: roff_cond_sub(ROFF_ARGS)
1059: {
1060: enum rofft t;
1.37 schwarze 1061: char *ep;
1.71 schwarze 1062: int rr;
1.2 schwarze 1063:
1064: rr = r->last->rule;
1.37 schwarze 1065: roffnode_cleanscope(r);
1.50 schwarze 1066: t = roff_parse(r, *bufp, &pos);
1.2 schwarze 1067:
1.37 schwarze 1068: /*
1.50 schwarze 1069: * Fully handle known macros when they are structurally
1070: * required or when the conditional evaluated to true.
1.5 schwarze 1071: */
1072:
1.50 schwarze 1073: if ((ROFF_MAX != t) &&
1.71 schwarze 1074: (rr || ROFFMAC_STRUCT & roffs[t].flags)) {
1.50 schwarze 1075: assert(roffs[t].proc);
1076: return((*roffs[t].proc)(r, t, bufp, szp,
1.80 ! schwarze 1077: ln, ppos, pos, offs));
1.50 schwarze 1078: }
1.39 schwarze 1079:
1.69 schwarze 1080: /*
1081: * If `\}' occurs on a macro line without a preceding macro,
1082: * drop the line completely.
1083: */
1084:
1085: ep = *bufp + pos;
1086: if ('\\' == ep[0] && '}' == ep[1])
1.71 schwarze 1087: rr = 0;
1.69 schwarze 1088:
1.50 schwarze 1089: /* Always check for the closing delimiter `\}'. */
1.39 schwarze 1090:
1.50 schwarze 1091: while (NULL != (ep = strchr(ep, '\\'))) {
1.70 schwarze 1092: if ('}' == *(++ep)) {
1093: *ep = '&';
1094: roff_ccond(r, ln, ep - *bufp - 1);
1095: }
1096: ++ep;
1.50 schwarze 1097: }
1.71 schwarze 1098: return(rr ? ROFF_CONT : ROFF_IGN);
1.2 schwarze 1099: }
1100:
1101: static enum rofferr
1102: roff_cond_text(ROFF_ARGS)
1103: {
1.37 schwarze 1104: char *ep;
1.71 schwarze 1105: int rr;
1.2 schwarze 1106:
1107: rr = r->last->rule;
1.37 schwarze 1108: roffnode_cleanscope(r);
1.1 schwarze 1109:
1.70 schwarze 1110: ep = *bufp + pos;
1111: while (NULL != (ep = strchr(ep, '\\'))) {
1112: if ('}' == *(++ep)) {
1113: *ep = '&';
1114: roff_ccond(r, ln, ep - *bufp - 1);
1115: }
1116: ++ep;
1.2 schwarze 1117: }
1.71 schwarze 1118: return(rr ? ROFF_CONT : ROFF_IGN);
1.2 schwarze 1119: }
1120:
1.77 schwarze 1121: /*
1122: * Parse a single signed integer number. Stop at the first non-digit.
1123: * If there is at least one digit, return success and advance the
1124: * parse point, else return failure and let the parse point unchanged.
1125: * Ignore overflows, treat them just like the C language.
1126: */
1.56 schwarze 1127: static int
1128: roff_getnum(const char *v, int *pos, int *res)
1129: {
1.79 schwarze 1130: int myres, n, p;
1131:
1132: if (NULL == res)
1133: res = &myres;
1.56 schwarze 1134:
1135: p = *pos;
1136: n = v[p] == '-';
1137: if (n)
1138: p++;
1139:
1140: for (*res = 0; isdigit((unsigned char)v[p]); p++)
1.77 schwarze 1141: *res = 10 * *res + v[p] - '0';
1.56 schwarze 1142: if (p == *pos + n)
1143: return 0;
1144:
1145: if (n)
1146: *res = -*res;
1147:
1148: *pos = p;
1149: return 1;
1150: }
1151:
1.71 schwarze 1152: /*
1153: * Evaluate a string comparison condition.
1154: * The first character is the delimiter.
1155: * Succeed if the string up to its second occurrence
1156: * matches the string up to its third occurence.
1157: * Advance the cursor after the third occurrence
1158: * or lacking that, to the end of the line.
1159: */
1160: static int
1161: roff_evalstrcond(const char *v, int *pos)
1162: {
1163: const char *s1, *s2, *s3;
1164: int match;
1165:
1166: match = 0;
1167: s1 = v + *pos; /* initial delimiter */
1168: s2 = s1 + 1; /* for scanning the first string */
1169: s3 = strchr(s2, *s1); /* for scanning the second string */
1170:
1171: if (NULL == s3) /* found no middle delimiter */
1172: goto out;
1173:
1174: while ('\0' != *++s3) {
1175: if (*s2 != *s3) { /* mismatch */
1176: s3 = strchr(s3, *s1);
1177: break;
1178: }
1179: if (*s3 == *s1) { /* found the final delimiter */
1180: match = 1;
1181: break;
1182: }
1183: s2++;
1184: }
1185:
1186: out:
1187: if (NULL == s3)
1188: s3 = strchr(s2, '\0');
1189: else
1190: s3++;
1191: *pos = s3 - v;
1192: return(match);
1193: }
1194:
1.77 schwarze 1195: /*
1196: * Evaluate an optionally negated single character, numerical,
1197: * or string condition.
1198: */
1.71 schwarze 1199: static int
1.5 schwarze 1200: roff_evalcond(const char *v, int *pos)
1201: {
1.77 schwarze 1202: int wanttrue, number;
1.5 schwarze 1203:
1.71 schwarze 1204: if ('!' == v[*pos]) {
1205: wanttrue = 0;
1206: (*pos)++;
1207: } else
1208: wanttrue = 1;
1209:
1.5 schwarze 1210: switch (v[*pos]) {
1.80 ! schwarze 1211: case 'n':
1.71 schwarze 1212: /* FALLTHROUGH */
1.80 ! schwarze 1213: case 'o':
1.5 schwarze 1214: (*pos)++;
1.71 schwarze 1215: return(wanttrue);
1.80 ! schwarze 1216: case 'c':
1.71 schwarze 1217: /* FALLTHROUGH */
1.80 ! schwarze 1218: case 'd':
1.71 schwarze 1219: /* FALLTHROUGH */
1.80 ! schwarze 1220: case 'e':
1.5 schwarze 1221: /* FALLTHROUGH */
1.80 ! schwarze 1222: case 'r':
1.5 schwarze 1223: /* FALLTHROUGH */
1.80 ! schwarze 1224: case 't':
1.5 schwarze 1225: (*pos)++;
1.71 schwarze 1226: return(!wanttrue);
1.5 schwarze 1227: default:
1228: break;
1229: }
1230:
1.77 schwarze 1231: if (roff_evalnum(v, pos, &number, 0))
1232: return((number > 0) == wanttrue);
1233: else
1.71 schwarze 1234: return(roff_evalstrcond(v, pos) == wanttrue);
1.5 schwarze 1235: }
1236:
1.2 schwarze 1237: static enum rofferr
1.21 schwarze 1238: roff_line_ignore(ROFF_ARGS)
1.6 schwarze 1239: {
1.30 schwarze 1240:
1.21 schwarze 1241: return(ROFF_IGN);
1242: }
1243:
1244: static enum rofferr
1.2 schwarze 1245: roff_cond(ROFF_ARGS)
1246: {
1.46 schwarze 1247:
1248: roffnode_push(r, tok, NULL, ln, ppos);
1.2 schwarze 1249:
1.80 ! schwarze 1250: /*
1.35 schwarze 1251: * An `.el' has no conditional body: it will consume the value
1252: * of the current rstack entry set in prior `ie' calls or
1.80 ! schwarze 1253: * defaults to DENY.
1.35 schwarze 1254: *
1255: * If we're not an `el', however, then evaluate the conditional.
1256: */
1.1 schwarze 1257:
1.46 schwarze 1258: r->last->rule = ROFF_el == tok ?
1.80 ! schwarze 1259: (r->rstackpos < 0 ? 0 : r->rstack[r->rstackpos--]) :
! 1260: roff_evalcond(*bufp, &pos);
1.2 schwarze 1261:
1.35 schwarze 1262: /*
1263: * An if-else will put the NEGATION of the current evaluated
1264: * conditional into the stack of rules.
1265: */
1266:
1.2 schwarze 1267: if (ROFF_ie == tok) {
1.35 schwarze 1268: if (r->rstackpos == RSTACK_MAX - 1) {
1.80 ! schwarze 1269: mandoc_msg(MANDOCERR_MEM,
! 1270: r->parse, ln, ppos, NULL);
1.35 schwarze 1271: return(ROFF_ERR);
1272: }
1.71 schwarze 1273: r->rstack[++r->rstackpos] = !r->last->rule;
1.2 schwarze 1274: }
1.5 schwarze 1275:
1276: /* If the parent has false as its rule, then so do we. */
1277:
1.71 schwarze 1278: if (r->last->parent && !r->last->parent->rule)
1279: r->last->rule = 0;
1.5 schwarze 1280:
1281: /*
1.46 schwarze 1282: * Determine scope.
1283: * If there is nothing on the line after the conditional,
1284: * not even whitespace, use next-line scope.
1.5 schwarze 1285: */
1.2 schwarze 1286:
1.46 schwarze 1287: if ('\0' == (*bufp)[pos]) {
1288: r->last->endspan = 2;
1289: goto out;
1290: }
1291:
1292: while (' ' == (*bufp)[pos])
1293: pos++;
1294:
1295: /* An opening brace requests multiline scope. */
1.2 schwarze 1296:
1297: if ('\\' == (*bufp)[pos] && '{' == (*bufp)[pos + 1]) {
1298: r->last->endspan = -1;
1299: pos += 2;
1.46 schwarze 1300: goto out;
1.80 ! schwarze 1301: }
1.2 schwarze 1302:
1303: /*
1.46 schwarze 1304: * Anything else following the conditional causes
1305: * single-line scope. Warn if the scope contains
1306: * nothing but trailing whitespace.
1.2 schwarze 1307: */
1308:
1309: if ('\0' == (*bufp)[pos])
1.46 schwarze 1310: mandoc_msg(MANDOCERR_NOARGS, r->parse, ln, ppos, NULL);
1.2 schwarze 1311:
1.46 schwarze 1312: r->last->endspan = 1;
1.1 schwarze 1313:
1.46 schwarze 1314: out:
1.2 schwarze 1315: *offs = pos;
1316: return(ROFF_RERUN);
1.1 schwarze 1317: }
1318:
1.2 schwarze 1319: static enum rofferr
1.7 schwarze 1320: roff_ds(ROFF_ARGS)
1321: {
1.10 schwarze 1322: char *name, *string;
1323:
1324: /*
1325: * A symbol is named by the first word following the macro
1326: * invocation up to a space. Its value is anything after the
1327: * name's trailing whitespace and optional double-quote. Thus,
1328: *
1329: * [.ds foo "bar " ]
1330: *
1331: * will have `bar " ' as its value.
1332: */
1.7 schwarze 1333:
1.28 schwarze 1334: string = *bufp + pos;
1335: name = roff_getname(r, &string, ln, pos);
1.7 schwarze 1336: if ('\0' == *name)
1337: return(ROFF_IGN);
1338:
1.28 schwarze 1339: /* Read past initial double-quote. */
1340: if ('"' == *string)
1.7 schwarze 1341: string++;
1342:
1.10 schwarze 1343: /* The rest is the value. */
1.66 schwarze 1344: roff_setstr(r, name, string, ROFF_as == tok);
1.7 schwarze 1345: return(ROFF_IGN);
1346: }
1347:
1.77 schwarze 1348: /*
1349: * Parse a single operator, one or two characters long.
1350: * If the operator is recognized, return success and advance the
1351: * parse point, else return failure and let the parse point unchanged.
1352: */
1353: static int
1354: roff_getop(const char *v, int *pos, char *res)
1355: {
1356:
1357: *res = v[*pos];
1358:
1359: switch (*res) {
1.80 ! schwarze 1360: case '+':
1.77 schwarze 1361: /* FALLTHROUGH */
1.80 ! schwarze 1362: case '-':
1.77 schwarze 1363: /* FALLTHROUGH */
1.80 ! schwarze 1364: case '*':
1.77 schwarze 1365: /* FALLTHROUGH */
1.80 ! schwarze 1366: case '/':
1.77 schwarze 1367: /* FALLTHROUGH */
1.80 ! schwarze 1368: case '%':
1.77 schwarze 1369: /* FALLTHROUGH */
1.80 ! schwarze 1370: case '&':
1.77 schwarze 1371: /* FALLTHROUGH */
1.80 ! schwarze 1372: case ':':
1.77 schwarze 1373: break;
1374: case '<':
1375: switch (v[*pos + 1]) {
1.80 ! schwarze 1376: case '=':
1.77 schwarze 1377: *res = 'l';
1378: (*pos)++;
1379: break;
1.80 ! schwarze 1380: case '>':
1.77 schwarze 1381: *res = '!';
1382: (*pos)++;
1383: break;
1.80 ! schwarze 1384: case '?':
1.77 schwarze 1385: *res = 'i';
1386: (*pos)++;
1387: break;
1388: default:
1389: break;
1390: }
1391: break;
1392: case '>':
1393: switch (v[*pos + 1]) {
1.80 ! schwarze 1394: case '=':
1.77 schwarze 1395: *res = 'g';
1396: (*pos)++;
1397: break;
1.80 ! schwarze 1398: case '?':
1.77 schwarze 1399: *res = 'a';
1400: (*pos)++;
1401: break;
1402: default:
1403: break;
1404: }
1405: break;
1406: case '=':
1407: if ('=' == v[*pos + 1])
1408: (*pos)++;
1409: break;
1410: default:
1411: return(0);
1412: }
1413: (*pos)++;
1414:
1415: return(*res);
1416: }
1417:
1418: /*
1419: * Evaluate either a parenthesized numeric expression
1420: * or a single signed integer number.
1421: */
1422: static int
1423: roff_evalpar(const char *v, int *pos, int *res)
1424: {
1425:
1426: if ('(' != v[*pos])
1427: return(roff_getnum(v, pos, res));
1428:
1429: (*pos)++;
1430: if ( ! roff_evalnum(v, pos, res, 1))
1431: return(0);
1432:
1.79 schwarze 1433: /*
1434: * Omission of the closing parenthesis
1435: * is an error in validation mode,
1436: * but ignored in evaluation mode.
1437: */
1438:
1.77 schwarze 1439: if (')' == v[*pos])
1440: (*pos)++;
1.79 schwarze 1441: else if (NULL == res)
1442: return(0);
1.77 schwarze 1443:
1444: return(1);
1445: }
1446:
1447: /*
1448: * Evaluate a complete numeric expression.
1449: * Proceed left to right, there is no concept of precedence.
1450: */
1451: static int
1452: roff_evalnum(const char *v, int *pos, int *res, int skipwhite)
1453: {
1454: int mypos, operand2;
1455: char operator;
1456:
1457: if (NULL == pos) {
1458: mypos = 0;
1459: pos = &mypos;
1460: }
1461:
1462: if (skipwhite)
1463: while (isspace((unsigned char)v[*pos]))
1464: (*pos)++;
1465:
1466: if ( ! roff_evalpar(v, pos, res))
1467: return(0);
1468:
1469: while (1) {
1470: if (skipwhite)
1471: while (isspace((unsigned char)v[*pos]))
1472: (*pos)++;
1473:
1474: if ( ! roff_getop(v, pos, &operator))
1475: break;
1476:
1477: if (skipwhite)
1478: while (isspace((unsigned char)v[*pos]))
1479: (*pos)++;
1480:
1481: if ( ! roff_evalpar(v, pos, &operand2))
1482: return(0);
1483:
1484: if (skipwhite)
1485: while (isspace((unsigned char)v[*pos]))
1486: (*pos)++;
1.79 schwarze 1487:
1488: if (NULL == res)
1489: continue;
1.77 schwarze 1490:
1491: switch (operator) {
1.80 ! schwarze 1492: case '+':
1.77 schwarze 1493: *res += operand2;
1494: break;
1.80 ! schwarze 1495: case '-':
1.77 schwarze 1496: *res -= operand2;
1497: break;
1.80 ! schwarze 1498: case '*':
1.77 schwarze 1499: *res *= operand2;
1500: break;
1.80 ! schwarze 1501: case '/':
1.77 schwarze 1502: *res /= operand2;
1503: break;
1.80 ! schwarze 1504: case '%':
1.77 schwarze 1505: *res %= operand2;
1506: break;
1.80 ! schwarze 1507: case '<':
1.77 schwarze 1508: *res = *res < operand2;
1509: break;
1.80 ! schwarze 1510: case '>':
1.77 schwarze 1511: *res = *res > operand2;
1512: break;
1.80 ! schwarze 1513: case 'l':
1.77 schwarze 1514: *res = *res <= operand2;
1515: break;
1.80 ! schwarze 1516: case 'g':
1.77 schwarze 1517: *res = *res >= operand2;
1518: break;
1.80 ! schwarze 1519: case '=':
1.77 schwarze 1520: *res = *res == operand2;
1521: break;
1.80 ! schwarze 1522: case '!':
1.77 schwarze 1523: *res = *res != operand2;
1524: break;
1.80 ! schwarze 1525: case '&':
1.77 schwarze 1526: *res = *res && operand2;
1527: break;
1.80 ! schwarze 1528: case ':':
1.77 schwarze 1529: *res = *res || operand2;
1530: break;
1.80 ! schwarze 1531: case 'i':
1.77 schwarze 1532: if (operand2 < *res)
1533: *res = operand2;
1534: break;
1.80 ! schwarze 1535: case 'a':
1.77 schwarze 1536: if (operand2 > *res)
1537: *res = operand2;
1538: break;
1539: default:
1540: abort();
1541: }
1542: }
1543: return(1);
1544: }
1545:
1.52 schwarze 1546: void
1.60 schwarze 1547: roff_setreg(struct roff *r, const char *name, int val, char sign)
1.41 schwarze 1548: {
1.52 schwarze 1549: struct roffreg *reg;
1550:
1551: /* Search for an existing register with the same name. */
1552: reg = r->regtab;
1553:
1554: while (reg && strcmp(name, reg->key.p))
1555: reg = reg->next;
1.41 schwarze 1556:
1.52 schwarze 1557: if (NULL == reg) {
1558: /* Create a new register. */
1559: reg = mandoc_malloc(sizeof(struct roffreg));
1560: reg->key.p = mandoc_strdup(name);
1561: reg->key.sz = strlen(name);
1.60 schwarze 1562: reg->val = 0;
1.52 schwarze 1563: reg->next = r->regtab;
1564: r->regtab = reg;
1565: }
1566:
1.60 schwarze 1567: if ('+' == sign)
1568: reg->val += val;
1569: else if ('-' == sign)
1570: reg->val -= val;
1571: else
1572: reg->val = val;
1.41 schwarze 1573: }
1574:
1.65 schwarze 1575: /*
1576: * Handle some predefined read-only number registers.
1577: * For now, return -1 if the requested register is not predefined;
1578: * in case a predefined read-only register having the value -1
1579: * were to turn up, another special value would have to be chosen.
1580: */
1581: static int
1582: roff_getregro(const char *name)
1583: {
1584:
1585: switch (*name) {
1.80 ! schwarze 1586: case 'A': /* ASCII approximation mode is always off. */
1.65 schwarze 1587: return(0);
1.80 ! schwarze 1588: case 'g': /* Groff compatibility mode is always on. */
1.65 schwarze 1589: return(1);
1.80 ! schwarze 1590: case 'H': /* Fixed horizontal resolution. */
1.65 schwarze 1591: return (24);
1.80 ! schwarze 1592: case 'j': /* Always adjust left margin only. */
1.65 schwarze 1593: return(0);
1.80 ! schwarze 1594: case 'T': /* Some output device is always defined. */
1.65 schwarze 1595: return(1);
1.80 ! schwarze 1596: case 'V': /* Fixed vertical resolution. */
1.65 schwarze 1597: return (40);
1598: default:
1599: return (-1);
1600: }
1601: }
1602:
1.53 schwarze 1603: int
1.52 schwarze 1604: roff_getreg(const struct roff *r, const char *name)
1.41 schwarze 1605: {
1.52 schwarze 1606: struct roffreg *reg;
1.65 schwarze 1607: int val;
1608:
1609: if ('.' == name[0] && '\0' != name[1] && '\0' == name[2]) {
1610: val = roff_getregro(name + 1);
1611: if (-1 != val)
1612: return (val);
1613: }
1.52 schwarze 1614:
1615: for (reg = r->regtab; reg; reg = reg->next)
1616: if (0 == strcmp(name, reg->key.p))
1.53 schwarze 1617: return(reg->val);
1618:
1619: return(0);
1620: }
1621:
1622: static int
1623: roff_getregn(const struct roff *r, const char *name, size_t len)
1624: {
1625: struct roffreg *reg;
1.65 schwarze 1626: int val;
1627:
1628: if ('.' == name[0] && 2 == len) {
1629: val = roff_getregro(name + 1);
1630: if (-1 != val)
1631: return (val);
1632: }
1.53 schwarze 1633:
1634: for (reg = r->regtab; reg; reg = reg->next)
1635: if (len == reg->key.sz &&
1636: 0 == strncmp(name, reg->key.p, len))
1637: return(reg->val);
1.41 schwarze 1638:
1.52 schwarze 1639: return(0);
1.41 schwarze 1640: }
1641:
1.52 schwarze 1642: static void
1643: roff_freereg(struct roffreg *reg)
1.41 schwarze 1644: {
1.52 schwarze 1645: struct roffreg *old_reg;
1.41 schwarze 1646:
1.52 schwarze 1647: while (NULL != reg) {
1648: free(reg->key.p);
1649: old_reg = reg;
1650: reg = reg->next;
1651: free(old_reg);
1652: }
1.41 schwarze 1653: }
1.7 schwarze 1654:
1655: static enum rofferr
1.6 schwarze 1656: roff_nr(ROFF_ARGS)
1.1 schwarze 1657: {
1.28 schwarze 1658: const char *key;
1659: char *val;
1.37 schwarze 1660: int iv;
1.60 schwarze 1661: char sign;
1.6 schwarze 1662:
1.28 schwarze 1663: val = *bufp + pos;
1664: key = roff_getname(r, &val, ln, pos);
1.6 schwarze 1665:
1.60 schwarze 1666: sign = *val;
1667: if ('+' == sign || '-' == sign)
1668: val++;
1669:
1.77 schwarze 1670: if (roff_evalnum(val, NULL, &iv, 0))
1671: roff_setreg(r, key, iv, sign);
1.1 schwarze 1672:
1.76 schwarze 1673: return(ROFF_IGN);
1674: }
1675:
1676: static enum rofferr
1677: roff_rr(ROFF_ARGS)
1678: {
1679: struct roffreg *reg, **prev;
1680: const char *name;
1681: char *cp;
1682:
1683: cp = *bufp + pos;
1684: name = roff_getname(r, &cp, ln, pos);
1685:
1686: prev = &r->regtab;
1687: while (1) {
1688: reg = *prev;
1689: if (NULL == reg || !strcmp(name, reg->key.p))
1690: break;
1691: prev = ®->next;
1692: }
1693: if (NULL != reg) {
1694: *prev = reg->next;
1695: free(reg->key.p);
1696: free(reg);
1697: }
1.29 schwarze 1698: return(ROFF_IGN);
1699: }
1700:
1701: static enum rofferr
1702: roff_rm(ROFF_ARGS)
1703: {
1704: const char *name;
1705: char *cp;
1706:
1707: cp = *bufp + pos;
1708: while ('\0' != *cp) {
1.34 schwarze 1709: name = roff_getname(r, &cp, ln, (int)(cp - *bufp));
1.29 schwarze 1710: if ('\0' != *name)
1711: roff_setstr(r, name, NULL, 0);
1712: }
1.51 schwarze 1713: return(ROFF_IGN);
1714: }
1715:
1716: static enum rofferr
1717: roff_it(ROFF_ARGS)
1718: {
1719: char *cp;
1720: size_t len;
1721: int iv;
1722:
1723: /* Parse the number of lines. */
1724: cp = *bufp + pos;
1725: len = strcspn(cp, " \t");
1726: cp[len] = '\0';
1727: if ((iv = mandoc_strntoi(cp, len, 10)) <= 0) {
1728: mandoc_msg(MANDOCERR_NUMERIC, r->parse,
1.80 ! schwarze 1729: ln, ppos, *bufp + 1);
1.51 schwarze 1730: return(ROFF_IGN);
1731: }
1732: cp += len + 1;
1733:
1734: /* Arm the input line trap. */
1735: roffit_lines = iv;
1736: roffit_macro = mandoc_strdup(cp);
1.2 schwarze 1737: return(ROFF_IGN);
1.47 schwarze 1738: }
1739:
1740: static enum rofferr
1741: roff_Dd(ROFF_ARGS)
1742: {
1743: const char *const *cp;
1744:
1.72 schwarze 1745: if (0 == ((MPARSE_MDOC | MPARSE_QUICK) & r->options))
1.47 schwarze 1746: for (cp = __mdoc_reserved; *cp; cp++)
1747: roff_setstr(r, *cp, NULL, 0);
1748:
1749: return(ROFF_CONT);
1750: }
1751:
1752: static enum rofferr
1753: roff_TH(ROFF_ARGS)
1754: {
1755: const char *const *cp;
1756:
1.72 schwarze 1757: if (0 == (MPARSE_QUICK & r->options))
1.47 schwarze 1758: for (cp = __man_reserved; *cp; cp++)
1759: roff_setstr(r, *cp, NULL, 0);
1760:
1761: return(ROFF_CONT);
1.14 schwarze 1762: }
1763:
1764: static enum rofferr
1.27 schwarze 1765: roff_TE(ROFF_ARGS)
1766: {
1767:
1768: if (NULL == r->tbl)
1.35 schwarze 1769: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.27 schwarze 1770: else
1.41 schwarze 1771: tbl_end(&r->tbl);
1.27 schwarze 1772:
1773: return(ROFF_IGN);
1774: }
1775:
1776: static enum rofferr
1777: roff_T_(ROFF_ARGS)
1778: {
1779:
1780: if (NULL == r->tbl)
1.35 schwarze 1781: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.27 schwarze 1782: else
1783: tbl_restart(ppos, ln, r->tbl);
1784:
1785: return(ROFF_IGN);
1786: }
1787:
1.41 schwarze 1788: #if 0
1789: static int
1790: roff_closeeqn(struct roff *r)
1791: {
1792:
1793: return(r->eqn && ROFF_EQN == eqn_end(&r->eqn) ? 1 : 0);
1794: }
1795: #endif
1796:
1797: static void
1.80 ! schwarze 1798: roff_openeqn(struct roff *r, const char *name, int line,
1.41 schwarze 1799: int offs, const char *buf)
1.32 schwarze 1800: {
1.41 schwarze 1801: struct eqn_node *e;
1802: int poff;
1.32 schwarze 1803:
1804: assert(NULL == r->eqn);
1.41 schwarze 1805: e = eqn_alloc(name, offs, line, r->parse);
1.32 schwarze 1806:
1807: if (r->last_eqn)
1808: r->last_eqn->next = e;
1809: else
1810: r->first_eqn = r->last_eqn = e;
1811:
1812: r->eqn = r->last_eqn = e;
1.41 schwarze 1813:
1814: if (buf) {
1815: poff = 0;
1816: eqn_read(&r->eqn, line, buf, offs, &poff);
1817: }
1818: }
1819:
1820: static enum rofferr
1821: roff_EQ(ROFF_ARGS)
1822: {
1823:
1824: roff_openeqn(r, *bufp + pos, ln, ppos, NULL);
1.32 schwarze 1825: return(ROFF_IGN);
1826: }
1827:
1828: static enum rofferr
1829: roff_EN(ROFF_ARGS)
1830: {
1831:
1.35 schwarze 1832: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.32 schwarze 1833: return(ROFF_IGN);
1834: }
1835:
1836: static enum rofferr
1.27 schwarze 1837: roff_TS(ROFF_ARGS)
1838: {
1.49 schwarze 1839: struct tbl_node *tbl;
1.27 schwarze 1840:
1841: if (r->tbl) {
1.35 schwarze 1842: mandoc_msg(MANDOCERR_SCOPEBROKEN, r->parse, ln, ppos, NULL);
1.41 schwarze 1843: tbl_end(&r->tbl);
1.27 schwarze 1844: }
1845:
1.49 schwarze 1846: tbl = tbl_alloc(ppos, ln, r->parse);
1.27 schwarze 1847:
1848: if (r->last_tbl)
1.49 schwarze 1849: r->last_tbl->next = tbl;
1.27 schwarze 1850: else
1.49 schwarze 1851: r->first_tbl = r->last_tbl = tbl;
1.27 schwarze 1852:
1.49 schwarze 1853: r->tbl = r->last_tbl = tbl;
1.27 schwarze 1854: return(ROFF_IGN);
1855: }
1856:
1857: static enum rofferr
1.48 schwarze 1858: roff_cc(ROFF_ARGS)
1859: {
1860: const char *p;
1861:
1862: p = *bufp + pos;
1863:
1864: if ('\0' == *p || '.' == (r->control = *p++))
1865: r->control = 0;
1866:
1867: if ('\0' != *p)
1868: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
1869:
1870: return(ROFF_IGN);
1871: }
1872:
1873: static enum rofferr
1.42 schwarze 1874: roff_tr(ROFF_ARGS)
1875: {
1876: const char *p, *first, *second;
1877: size_t fsz, ssz;
1878: enum mandoc_esc esc;
1879:
1880: p = *bufp + pos;
1881:
1882: if ('\0' == *p) {
1883: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
1884: return(ROFF_IGN);
1885: }
1886:
1887: while ('\0' != *p) {
1888: fsz = ssz = 1;
1889:
1890: first = p++;
1891: if ('\\' == *first) {
1892: esc = mandoc_escape(&p, NULL, NULL);
1893: if (ESCAPE_ERROR == esc) {
1.80 ! schwarze 1894: mandoc_msg(MANDOCERR_BADESCAPE,
! 1895: r->parse, ln,
! 1896: (int)(p - *bufp), NULL);
1.42 schwarze 1897: return(ROFF_IGN);
1898: }
1899: fsz = (size_t)(p - first);
1900: }
1901:
1902: second = p++;
1903: if ('\\' == *second) {
1904: esc = mandoc_escape(&p, NULL, NULL);
1905: if (ESCAPE_ERROR == esc) {
1.80 ! schwarze 1906: mandoc_msg(MANDOCERR_BADESCAPE,
! 1907: r->parse, ln,
! 1908: (int)(p - *bufp), NULL);
1.42 schwarze 1909: return(ROFF_IGN);
1910: }
1911: ssz = (size_t)(p - second);
1912: } else if ('\0' == *second) {
1.80 ! schwarze 1913: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse,
! 1914: ln, (int)(p - *bufp), NULL);
1.42 schwarze 1915: second = " ";
1916: p--;
1917: }
1918:
1919: if (fsz > 1) {
1.80 ! schwarze 1920: roff_setstrn(&r->xmbtab, first, fsz,
! 1921: second, ssz, 0);
1.42 schwarze 1922: continue;
1923: }
1924:
1925: if (NULL == r->xtab)
1.80 ! schwarze 1926: r->xtab = mandoc_calloc(128,
! 1927: sizeof(struct roffstr));
1.42 schwarze 1928:
1929: free(r->xtab[(int)*first].p);
1930: r->xtab[(int)*first].p = mandoc_strndup(second, ssz);
1931: r->xtab[(int)*first].sz = ssz;
1932: }
1933:
1934: return(ROFF_IGN);
1935: }
1936:
1937: static enum rofferr
1.14 schwarze 1938: roff_so(ROFF_ARGS)
1939: {
1940: char *name;
1.15 schwarze 1941:
1.35 schwarze 1942: mandoc_msg(MANDOCERR_SO, r->parse, ln, ppos, NULL);
1.14 schwarze 1943:
1.22 schwarze 1944: /*
1945: * Handle `so'. Be EXTREMELY careful, as we shouldn't be
1946: * opening anything that's not in our cwd or anything beneath
1947: * it. Thus, explicitly disallow traversing up the file-system
1948: * or using absolute paths.
1949: */
1950:
1.14 schwarze 1951: name = *bufp + pos;
1952: if ('/' == *name || strstr(name, "../") || strstr(name, "/..")) {
1.35 schwarze 1953: mandoc_msg(MANDOCERR_SOPATH, r->parse, ln, pos, NULL);
1.14 schwarze 1954: return(ROFF_ERR);
1955: }
1956:
1957: *offs = pos;
1958: return(ROFF_SO);
1.7 schwarze 1959: }
1960:
1.16 schwarze 1961: static enum rofferr
1962: roff_userdef(ROFF_ARGS)
1.12 schwarze 1963: {
1.16 schwarze 1964: const char *arg[9];
1965: char *cp, *n1, *n2;
1.25 schwarze 1966: int i;
1.12 schwarze 1967:
1.16 schwarze 1968: /*
1969: * Collect pointers to macro argument strings
1.61 schwarze 1970: * and NUL-terminate them.
1.16 schwarze 1971: */
1972: cp = *bufp + pos;
1.25 schwarze 1973: for (i = 0; i < 9; i++)
1.26 schwarze 1974: arg[i] = '\0' == *cp ? "" :
1.35 schwarze 1975: mandoc_getarg(r->parse, &cp, ln, &pos);
1.16 schwarze 1976:
1977: /*
1978: * Expand macro arguments.
1.12 schwarze 1979: */
1.16 schwarze 1980: *szp = 0;
1981: n1 = cp = mandoc_strdup(r->current_string);
1982: while (NULL != (cp = strstr(cp, "\\$"))) {
1983: i = cp[2] - '1';
1984: if (0 > i || 8 < i) {
1985: /* Not an argument invocation. */
1986: cp += 2;
1987: continue;
1988: }
1989:
1990: *szp = strlen(n1) - 3 + strlen(arg[i]) + 1;
1991: n2 = mandoc_malloc(*szp);
1992:
1993: strlcpy(n2, n1, (size_t)(cp - n1 + 1));
1994: strlcat(n2, arg[i], *szp);
1995: strlcat(n2, cp + 3, *szp);
1996:
1997: cp = n2 + (cp - n1);
1998: free(n1);
1999: n1 = n2;
1.12 schwarze 2000: }
2001:
1.16 schwarze 2002: /*
2003: * Replace the macro invocation
2004: * by the expanded macro.
2005: */
2006: free(*bufp);
2007: *bufp = n1;
2008: if (0 == *szp)
2009: *szp = strlen(*bufp) + 1;
2010:
1.19 schwarze 2011: return(*szp > 1 && '\n' == (*bufp)[(int)*szp - 2] ?
1.16 schwarze 2012: ROFF_REPARSE : ROFF_APPEND);
1.12 schwarze 2013: }
1.28 schwarze 2014:
2015: static char *
2016: roff_getname(struct roff *r, char **cpp, int ln, int pos)
2017: {
2018: char *name, *cp;
2019:
2020: name = *cpp;
2021: if ('\0' == *name)
2022: return(name);
2023:
2024: /* Read until end of name. */
2025: for (cp = name; '\0' != *cp && ' ' != *cp; cp++) {
2026: if ('\\' != *cp)
2027: continue;
2028: cp++;
2029: if ('\\' == *cp)
2030: continue;
1.35 schwarze 2031: mandoc_msg(MANDOCERR_NAMESC, r->parse, ln, pos, NULL);
1.28 schwarze 2032: *cp = '\0';
2033: name = cp;
2034: }
2035:
2036: /* Nil-terminate name. */
2037: if ('\0' != *cp)
2038: *(cp++) = '\0';
2039:
2040: /* Read past spaces. */
2041: while (' ' == *cp)
2042: cp++;
2043:
2044: *cpp = cp;
2045: return(name);
2046: }
2047:
1.16 schwarze 2048: /*
2049: * Store *string into the user-defined string called *name.
2050: * To clear an existing entry, call with (*r, *name, NULL, 0).
1.66 schwarze 2051: * append == 0: replace mode
2052: * append == 1: single-line append mode
2053: * append == 2: multiline append mode, append '\n' after each call
1.16 schwarze 2054: */
1.8 schwarze 2055: static void
1.16 schwarze 2056: roff_setstr(struct roff *r, const char *name, const char *string,
1.66 schwarze 2057: int append)
1.7 schwarze 2058: {
1.42 schwarze 2059:
2060: roff_setstrn(&r->strtab, name, strlen(name), string,
1.80 ! schwarze 2061: string ? strlen(string) : 0, append);
1.42 schwarze 2062: }
2063:
2064: static void
2065: roff_setstrn(struct roffkv **r, const char *name, size_t namesz,
1.66 schwarze 2066: const char *string, size_t stringsz, int append)
1.42 schwarze 2067: {
2068: struct roffkv *n;
2069: char *c;
2070: int i;
2071: size_t oldch, newch;
1.7 schwarze 2072:
1.16 schwarze 2073: /* Search for an existing string with the same name. */
1.42 schwarze 2074: n = *r;
2075:
2076: while (n && strcmp(name, n->key.p))
1.7 schwarze 2077: n = n->next;
1.8 schwarze 2078:
2079: if (NULL == n) {
1.16 schwarze 2080: /* Create a new string table entry. */
1.42 schwarze 2081: n = mandoc_malloc(sizeof(struct roffkv));
2082: n->key.p = mandoc_strndup(name, namesz);
2083: n->key.sz = namesz;
2084: n->val.p = NULL;
2085: n->val.sz = 0;
2086: n->next = *r;
2087: *r = n;
1.66 schwarze 2088: } else if (0 == append) {
1.42 schwarze 2089: free(n->val.p);
2090: n->val.p = NULL;
2091: n->val.sz = 0;
1.16 schwarze 2092: }
2093:
2094: if (NULL == string)
2095: return;
2096:
2097: /*
2098: * One additional byte for the '\n' in multiline mode,
2099: * and one for the terminating '\0'.
2100: */
1.66 schwarze 2101: newch = stringsz + (1 < append ? 2u : 1u);
1.42 schwarze 2102:
2103: if (NULL == n->val.p) {
2104: n->val.p = mandoc_malloc(newch);
2105: *n->val.p = '\0';
1.16 schwarze 2106: oldch = 0;
2107: } else {
1.42 schwarze 2108: oldch = n->val.sz;
2109: n->val.p = mandoc_realloc(n->val.p, oldch + newch);
1.16 schwarze 2110: }
2111:
2112: /* Skip existing content in the destination buffer. */
1.42 schwarze 2113: c = n->val.p + (int)oldch;
1.16 schwarze 2114:
2115: /* Append new content to the destination buffer. */
1.42 schwarze 2116: i = 0;
2117: while (i < (int)stringsz) {
1.16 schwarze 2118: /*
2119: * Rudimentary roff copy mode:
2120: * Handle escaped backslashes.
2121: */
1.42 schwarze 2122: if ('\\' == string[i] && '\\' == string[i + 1])
2123: i++;
2124: *c++ = string[i++];
1.16 schwarze 2125: }
1.8 schwarze 2126:
1.16 schwarze 2127: /* Append terminating bytes. */
1.66 schwarze 2128: if (1 < append)
1.16 schwarze 2129: *c++ = '\n';
1.42 schwarze 2130:
1.16 schwarze 2131: *c = '\0';
1.42 schwarze 2132: n->val.sz = (int)(c - n->val.p);
1.7 schwarze 2133: }
2134:
1.8 schwarze 2135: static const char *
2136: roff_getstrn(const struct roff *r, const char *name, size_t len)
1.7 schwarze 2137: {
1.42 schwarze 2138: const struct roffkv *n;
1.64 schwarze 2139: int i;
1.7 schwarze 2140:
1.42 schwarze 2141: for (n = r->strtab; n; n = n->next)
1.80 ! schwarze 2142: if (0 == strncmp(name, n->key.p, len) &&
! 2143: '\0' == n->key.p[(int)len])
1.42 schwarze 2144: return(n->val.p);
1.64 schwarze 2145:
2146: for (i = 0; i < PREDEFS_MAX; i++)
2147: if (0 == strncmp(name, predefs[i].name, len) &&
2148: '\0' == predefs[i].name[(int)len])
2149: return(predefs[i].str);
1.8 schwarze 2150:
1.42 schwarze 2151: return(NULL);
1.7 schwarze 2152: }
2153:
1.8 schwarze 2154: static void
1.42 schwarze 2155: roff_freestr(struct roffkv *r)
1.7 schwarze 2156: {
1.42 schwarze 2157: struct roffkv *n, *nn;
1.7 schwarze 2158:
1.42 schwarze 2159: for (n = r; n; n = nn) {
2160: free(n->key.p);
2161: free(n->val.p);
1.7 schwarze 2162: nn = n->next;
2163: free(n);
2164: }
1.27 schwarze 2165: }
2166:
2167: const struct tbl_span *
2168: roff_span(const struct roff *r)
2169: {
1.80 ! schwarze 2170:
1.27 schwarze 2171: return(r->tbl ? tbl_span(r->tbl) : NULL);
1.32 schwarze 2172: }
2173:
2174: const struct eqn *
2175: roff_eqn(const struct roff *r)
2176: {
1.80 ! schwarze 2177:
1.32 schwarze 2178: return(r->last_eqn ? &r->last_eqn->eqn : NULL);
1.42 schwarze 2179: }
2180:
2181: /*
2182: * Duplicate an input string, making the appropriate character
2183: * conversations (as stipulated by `tr') along the way.
2184: * Returns a heap-allocated string with all the replacements made.
2185: */
2186: char *
2187: roff_strdup(const struct roff *r, const char *p)
2188: {
2189: const struct roffkv *cp;
2190: char *res;
2191: const char *pp;
2192: size_t ssz, sz;
2193: enum mandoc_esc esc;
2194:
2195: if (NULL == r->xmbtab && NULL == r->xtab)
2196: return(mandoc_strdup(p));
2197: else if ('\0' == *p)
2198: return(mandoc_strdup(""));
2199:
2200: /*
2201: * Step through each character looking for term matches
2202: * (remember that a `tr' can be invoked with an escape, which is
2203: * a glyph but the escape is multi-character).
2204: * We only do this if the character hash has been initialised
2205: * and the string is >0 length.
2206: */
2207:
2208: res = NULL;
2209: ssz = 0;
2210:
2211: while ('\0' != *p) {
2212: if ('\\' != *p && r->xtab && r->xtab[(int)*p].p) {
2213: sz = r->xtab[(int)*p].sz;
2214: res = mandoc_realloc(res, ssz + sz + 1);
2215: memcpy(res + ssz, r->xtab[(int)*p].p, sz);
2216: ssz += sz;
2217: p++;
2218: continue;
2219: } else if ('\\' != *p) {
2220: res = mandoc_realloc(res, ssz + 2);
2221: res[ssz++] = *p++;
2222: continue;
2223: }
2224:
2225: /* Search for term matches. */
2226: for (cp = r->xmbtab; cp; cp = cp->next)
2227: if (0 == strncmp(p, cp->key.p, cp->key.sz))
2228: break;
2229:
2230: if (NULL != cp) {
2231: /*
2232: * A match has been found.
2233: * Append the match to the array and move
2234: * forward by its keysize.
2235: */
1.80 ! schwarze 2236: res = mandoc_realloc(res,
! 2237: ssz + cp->val.sz + 1);
1.42 schwarze 2238: memcpy(res + ssz, cp->val.p, cp->val.sz);
2239: ssz += cp->val.sz;
2240: p += (int)cp->key.sz;
2241: continue;
2242: }
2243:
2244: /*
2245: * Handle escapes carefully: we need to copy
2246: * over just the escape itself, or else we might
2247: * do replacements within the escape itself.
2248: * Make sure to pass along the bogus string.
2249: */
2250: pp = p++;
2251: esc = mandoc_escape(&p, NULL, NULL);
2252: if (ESCAPE_ERROR == esc) {
2253: sz = strlen(pp);
2254: res = mandoc_realloc(res, ssz + sz + 1);
2255: memcpy(res + ssz, pp, sz);
2256: break;
2257: }
1.80 ! schwarze 2258: /*
! 2259: * We bail out on bad escapes.
1.42 schwarze 2260: * No need to warn: we already did so when
2261: * roff_res() was called.
2262: */
2263: sz = (int)(p - pp);
2264: res = mandoc_realloc(res, ssz + sz + 1);
2265: memcpy(res + ssz, pp, sz);
2266: ssz += sz;
2267: }
2268:
2269: res[(int)ssz] = '\0';
2270: return(res);
1.48 schwarze 2271: }
2272:
2273: /*
1.80 ! schwarze 2274: * Find out whether a line is a macro line or not.
1.48 schwarze 2275: * If it is, adjust the current position and return one; if it isn't,
2276: * return zero and don't change the current position.
2277: * If the control character has been set with `.cc', then let that grain
2278: * precedence.
2279: * This is slighly contrary to groff, where using the non-breaking
2280: * control character when `cc' has been invoked will cause the
2281: * non-breaking macro contents to be printed verbatim.
2282: */
2283: int
2284: roff_getcontrol(const struct roff *r, const char *cp, int *ppos)
2285: {
2286: int pos;
2287:
2288: pos = *ppos;
2289:
2290: if (0 != r->control && cp[pos] == r->control)
2291: pos++;
2292: else if (0 != r->control)
2293: return(0);
2294: else if ('\\' == cp[pos] && '.' == cp[pos + 1])
2295: pos += 2;
2296: else if ('.' == cp[pos] || '\'' == cp[pos])
2297: pos++;
2298: else
2299: return(0);
2300:
2301: while (' ' == cp[pos] || '\t' == cp[pos])
2302: pos++;
2303:
2304: *ppos = pos;
2305: return(1);
1.1 schwarze 2306: }
![[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