Annotation of src/usr.bin/mandoc/roff.c, Revision 1.64
1.64 ! schwarze 1: /* $Id: roff.c,v 1.63 2014/01/06 21:33:00 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"
1.27 schwarze 25: #include "libroff.h"
1.8 schwarze 26: #include "libmandoc.h"
1.1 schwarze 27:
1.37 schwarze 28: /* Maximum number of nested if-else conditionals. */
1.2 schwarze 29: #define RSTACK_MAX 128
30:
1.43 schwarze 31: /* Maximum number of string expansions per line, to break infinite loops. */
32: #define EXPAND_LIMIT 1000
33:
1.1 schwarze 34: enum rofft {
1.20 schwarze 35: ROFF_ad,
1.2 schwarze 36: ROFF_am,
37: ROFF_ami,
38: ROFF_am1,
1.48 schwarze 39: ROFF_cc,
1.1 schwarze 40: ROFF_de,
41: ROFF_dei,
1.2 schwarze 42: ROFF_de1,
43: ROFF_ds,
44: ROFF_el,
1.58 schwarze 45: ROFF_fam,
1.59 schwarze 46: ROFF_hw,
1.20 schwarze 47: ROFF_hy,
1.2 schwarze 48: ROFF_ie,
49: ROFF_if,
1.1 schwarze 50: ROFF_ig,
1.30 schwarze 51: ROFF_it,
1.20 schwarze 52: ROFF_ne,
53: ROFF_nh,
1.14 schwarze 54: ROFF_nr,
1.31 schwarze 55: ROFF_ns,
56: ROFF_ps,
1.2 schwarze 57: ROFF_rm,
1.14 schwarze 58: ROFF_so,
1.31 schwarze 59: ROFF_ta,
1.2 schwarze 60: ROFF_tr,
1.47 schwarze 61: ROFF_Dd,
62: ROFF_TH,
1.27 schwarze 63: ROFF_TS,
64: ROFF_TE,
65: ROFF_T_,
1.32 schwarze 66: ROFF_EQ,
67: ROFF_EN,
1.2 schwarze 68: ROFF_cblock,
1.37 schwarze 69: ROFF_ccond,
1.16 schwarze 70: ROFF_USERDEF,
1.1 schwarze 71: ROFF_MAX
72: };
73:
1.2 schwarze 74: enum roffrule {
1.55 schwarze 75: ROFFRULE_DENY,
76: ROFFRULE_ALLOW
1.2 schwarze 77: };
78:
1.41 schwarze 79: /*
1.42 schwarze 80: * An incredibly-simple string buffer.
81: */
1.8 schwarze 82: struct roffstr {
1.42 schwarze 83: char *p; /* nil-terminated buffer */
84: size_t sz; /* saved strlen(p) */
85: };
86:
87: /*
88: * A key-value roffstr pair as part of a singly-linked list.
89: */
90: struct roffkv {
91: struct roffstr key;
92: struct roffstr val;
93: struct roffkv *next; /* next in list */
1.8 schwarze 94: };
95:
1.52 schwarze 96: /*
97: * A single number register as part of a singly-linked list.
98: */
99: struct roffreg {
100: struct roffstr key;
1.53 schwarze 101: int val;
1.52 schwarze 102: struct roffreg *next;
103: };
104:
1.1 schwarze 105: struct roff {
1.47 schwarze 106: enum mparset parsetype; /* requested parse type */
1.35 schwarze 107: struct mparse *parse; /* parse point */
1.63 schwarze 108: int quick; /* skip standard macro deletion */
1.1 schwarze 109: struct roffnode *last; /* leaf of stack */
1.2 schwarze 110: enum roffrule rstack[RSTACK_MAX]; /* stack of !`ie' rules */
1.48 schwarze 111: char control; /* control character */
1.2 schwarze 112: int rstackpos; /* position in rstack */
1.52 schwarze 113: struct roffreg *regtab; /* number registers */
1.42 schwarze 114: struct roffkv *strtab; /* user-defined strings & macros */
115: struct roffkv *xmbtab; /* multi-byte trans table (`tr') */
116: struct roffstr *xtab; /* single-byte trans table (`tr') */
1.16 schwarze 117: const char *current_string; /* value of last called user macro */
1.27 schwarze 118: struct tbl_node *first_tbl; /* first table parsed */
119: struct tbl_node *last_tbl; /* last table parsed */
120: struct tbl_node *tbl; /* current table being parsed */
1.32 schwarze 121: struct eqn_node *last_eqn; /* last equation parsed */
122: struct eqn_node *first_eqn; /* first equation parsed */
123: struct eqn_node *eqn; /* current equation being parsed */
1.1 schwarze 124: };
125:
126: struct roffnode {
127: enum rofft tok; /* type of node */
128: struct roffnode *parent; /* up one in stack */
129: int line; /* parse line */
130: int col; /* parse col */
1.16 schwarze 131: char *name; /* node name, e.g. macro name */
1.2 schwarze 132: char *end; /* end-rules: custom token */
133: int endspan; /* end-rules: next-line or infty */
134: enum roffrule rule; /* current evaluation rule */
1.1 schwarze 135: };
136:
137: #define ROFF_ARGS struct roff *r, /* parse ctx */ \
138: enum rofft tok, /* tok of macro */ \
139: char **bufp, /* input buffer */ \
140: size_t *szp, /* size of input buffer */ \
141: int ln, /* parse line */ \
1.2 schwarze 142: int ppos, /* original pos in buffer */ \
143: int pos, /* current pos in buffer */ \
144: int *offs /* reset offset of buffer data */
1.1 schwarze 145:
146: typedef enum rofferr (*roffproc)(ROFF_ARGS);
147:
148: struct roffmac {
149: const char *name; /* macro name */
1.2 schwarze 150: roffproc proc; /* process new macro */
151: roffproc text; /* process as child text of macro */
152: roffproc sub; /* process as child of macro */
153: int flags;
154: #define ROFFMAC_STRUCT (1 << 0) /* always interpret */
1.3 schwarze 155: struct roffmac *next;
1.1 schwarze 156: };
157:
1.37 schwarze 158: struct predef {
159: const char *name; /* predefined input name */
160: const char *str; /* replacement symbol */
161: };
162:
163: #define PREDEF(__name, __str) \
164: { (__name), (__str) },
165:
1.42 schwarze 166: static enum rofft roffhash_find(const char *, size_t);
167: static void roffhash_init(void);
168: static void roffnode_cleanscope(struct roff *);
169: static void roffnode_pop(struct roff *);
170: static void roffnode_push(struct roff *, enum rofft,
171: const char *, int, int);
1.2 schwarze 172: static enum rofferr roff_block(ROFF_ARGS);
173: static enum rofferr roff_block_text(ROFF_ARGS);
174: static enum rofferr roff_block_sub(ROFF_ARGS);
175: static enum rofferr roff_cblock(ROFF_ARGS);
1.48 schwarze 176: static enum rofferr roff_cc(ROFF_ARGS);
1.2 schwarze 177: static enum rofferr roff_ccond(ROFF_ARGS);
178: static enum rofferr roff_cond(ROFF_ARGS);
179: static enum rofferr roff_cond_text(ROFF_ARGS);
180: static enum rofferr roff_cond_sub(ROFF_ARGS);
1.7 schwarze 181: static enum rofferr roff_ds(ROFF_ARGS);
1.8 schwarze 182: static enum roffrule roff_evalcond(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);
1.8 schwarze 191: static const char *roff_getstrn(const struct roff *,
192: const char *, size_t);
1.51 schwarze 193: static enum rofferr roff_it(ROFF_ARGS);
1.21 schwarze 194: static enum rofferr roff_line_ignore(ROFF_ARGS);
1.6 schwarze 195: static enum rofferr roff_nr(ROFF_ARGS);
1.41 schwarze 196: static void roff_openeqn(struct roff *, const char *,
197: int, int, const char *);
1.42 schwarze 198: static enum rofft roff_parse(struct roff *, const char *, int *);
1.51 schwarze 199: static enum rofferr roff_parsetext(char **, size_t *, int, int *);
1.45 schwarze 200: static enum rofferr roff_res(struct roff *,
1.37 schwarze 201: char **, size_t *, int, int);
1.29 schwarze 202: static enum rofferr roff_rm(ROFF_ARGS);
1.8 schwarze 203: static void roff_setstr(struct roff *,
1.16 schwarze 204: const char *, const char *, int);
1.42 schwarze 205: static void roff_setstrn(struct roffkv **, const char *,
206: size_t, const char *, size_t, int);
1.14 schwarze 207: static enum rofferr roff_so(ROFF_ARGS);
1.42 schwarze 208: static enum rofferr roff_tr(ROFF_ARGS);
1.47 schwarze 209: static enum rofferr roff_Dd(ROFF_ARGS);
210: static enum rofferr roff_TH(ROFF_ARGS);
1.27 schwarze 211: static enum rofferr roff_TE(ROFF_ARGS);
212: static enum rofferr roff_TS(ROFF_ARGS);
1.32 schwarze 213: static enum rofferr roff_EQ(ROFF_ARGS);
214: static enum rofferr roff_EN(ROFF_ARGS);
1.27 schwarze 215: static enum rofferr roff_T_(ROFF_ARGS);
1.16 schwarze 216: static enum rofferr roff_userdef(ROFF_ARGS);
1.1 schwarze 217:
1.42 schwarze 218: /* See roffhash_find() */
1.3 schwarze 219:
220: #define ASCII_HI 126
221: #define ASCII_LO 33
222: #define HASHWIDTH (ASCII_HI - ASCII_LO + 1)
223:
224: static struct roffmac *hash[HASHWIDTH];
225:
226: static struct roffmac roffs[ROFF_MAX] = {
1.21 schwarze 227: { "ad", roff_line_ignore, NULL, NULL, 0, NULL },
1.3 schwarze 228: { "am", roff_block, roff_block_text, roff_block_sub, 0, NULL },
229: { "ami", roff_block, roff_block_text, roff_block_sub, 0, NULL },
230: { "am1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.48 schwarze 231: { "cc", roff_cc, NULL, NULL, 0, NULL },
1.3 schwarze 232: { "de", roff_block, roff_block_text, roff_block_sub, 0, NULL },
233: { "dei", roff_block, roff_block_text, roff_block_sub, 0, NULL },
234: { "de1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.7 schwarze 235: { "ds", roff_ds, NULL, NULL, 0, NULL },
1.3 schwarze 236: { "el", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
1.58 schwarze 237: { "fam", roff_line_ignore, NULL, NULL, 0, NULL },
1.59 schwarze 238: { "hw", roff_line_ignore, NULL, NULL, 0, NULL },
1.21 schwarze 239: { "hy", roff_line_ignore, NULL, NULL, 0, NULL },
1.3 schwarze 240: { "ie", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
241: { "if", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
242: { "ig", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.51 schwarze 243: { "it", roff_it, NULL, NULL, 0, NULL },
1.21 schwarze 244: { "ne", roff_line_ignore, NULL, NULL, 0, NULL },
245: { "nh", roff_line_ignore, NULL, NULL, 0, NULL },
1.14 schwarze 246: { "nr", roff_nr, NULL, NULL, 0, NULL },
1.31 schwarze 247: { "ns", roff_line_ignore, NULL, NULL, 0, NULL },
248: { "ps", roff_line_ignore, NULL, NULL, 0, NULL },
1.29 schwarze 249: { "rm", roff_rm, NULL, NULL, 0, NULL },
1.14 schwarze 250: { "so", roff_so, NULL, NULL, 0, NULL },
1.31 schwarze 251: { "ta", roff_line_ignore, NULL, NULL, 0, NULL },
1.42 schwarze 252: { "tr", roff_tr, NULL, NULL, 0, NULL },
1.47 schwarze 253: { "Dd", roff_Dd, NULL, NULL, 0, NULL },
254: { "TH", roff_TH, NULL, NULL, 0, NULL },
1.27 schwarze 255: { "TS", roff_TS, NULL, NULL, 0, NULL },
256: { "TE", roff_TE, NULL, NULL, 0, NULL },
257: { "T&", roff_T_, NULL, NULL, 0, NULL },
1.32 schwarze 258: { "EQ", roff_EQ, NULL, NULL, 0, NULL },
259: { "EN", roff_EN, NULL, NULL, 0, NULL },
1.3 schwarze 260: { ".", roff_cblock, NULL, NULL, 0, NULL },
261: { "\\}", roff_ccond, NULL, NULL, 0, NULL },
1.16 schwarze 262: { NULL, roff_userdef, NULL, NULL, 0, NULL },
1.1 schwarze 263: };
264:
1.47 schwarze 265: const char *const __mdoc_reserved[] = {
266: "Ac", "Ad", "An", "Ao", "Ap", "Aq", "Ar", "At",
267: "Bc", "Bd", "Bf", "Bk", "Bl", "Bo", "Bq",
268: "Brc", "Bro", "Brq", "Bsx", "Bt", "Bx",
269: "Cd", "Cm", "Db", "Dc", "Dd", "Dl", "Do", "Dq",
270: "Ds", "Dt", "Dv", "Dx", "D1",
271: "Ec", "Ed", "Ef", "Ek", "El", "Em", "em",
272: "En", "Eo", "Eq", "Er", "Es", "Ev", "Ex",
273: "Fa", "Fc", "Fd", "Fl", "Fn", "Fo", "Fr", "Ft", "Fx",
274: "Hf", "Ic", "In", "It", "Lb", "Li", "Lk", "Lp", "LP",
275: "Me", "Ms", "Mt", "Nd", "Nm", "No", "Ns", "Nx",
276: "Oc", "Oo", "Op", "Os", "Ot", "Ox",
277: "Pa", "Pc", "Pf", "Po", "Pp", "PP", "pp", "Pq",
278: "Qc", "Ql", "Qo", "Qq", "Or", "Rd", "Re", "Rs", "Rv",
279: "Sc", "Sf", "Sh", "SH", "Sm", "So", "Sq",
280: "Ss", "St", "Sx", "Sy",
281: "Ta", "Tn", "Ud", "Ux", "Va", "Vt", "Xc", "Xo", "Xr",
282: "%A", "%B", "%D", "%I", "%J", "%N", "%O",
283: "%P", "%Q", "%R", "%T", "%U", "%V",
284: NULL
285: };
286:
287: const char *const __man_reserved[] = {
288: "AT", "B", "BI", "BR", "BT", "DE", "DS", "DT",
289: "EE", "EN", "EQ", "EX", "HF", "HP", "I", "IB", "IP", "IR",
290: "LP", "ME", "MT", "OP", "P", "PD", "PP", "PT",
291: "R", "RB", "RE", "RI", "RS", "SB", "SH", "SM", "SS", "SY",
292: "TE", "TH", "TP", "TQ", "TS", "T&", "UC", "UE", "UR", "YS",
293: NULL
294: };
295:
1.37 schwarze 296: /* Array of injected predefined strings. */
297: #define PREDEFS_MAX 38
298: static const struct predef predefs[PREDEFS_MAX] = {
299: #include "predefs.in"
300: };
301:
1.42 schwarze 302: /* See roffhash_find() */
1.3 schwarze 303: #define ROFF_HASH(p) (p[0] - ASCII_LO)
304:
1.51 schwarze 305: static int roffit_lines; /* number of lines to delay */
306: static char *roffit_macro; /* nil-terminated macro line */
307:
1.3 schwarze 308: static void
1.42 schwarze 309: roffhash_init(void)
1.3 schwarze 310: {
311: struct roffmac *n;
312: int buc, i;
313:
1.16 schwarze 314: for (i = 0; i < (int)ROFF_USERDEF; i++) {
1.3 schwarze 315: assert(roffs[i].name[0] >= ASCII_LO);
316: assert(roffs[i].name[0] <= ASCII_HI);
317:
318: buc = ROFF_HASH(roffs[i].name);
319:
320: if (NULL != (n = hash[buc])) {
321: for ( ; n->next; n = n->next)
322: /* Do nothing. */ ;
323: n->next = &roffs[i];
324: } else
325: hash[buc] = &roffs[i];
326: }
327: }
328:
1.1 schwarze 329: /*
330: * Look up a roff token by its name. Returns ROFF_MAX if no macro by
331: * the nil-terminated string name could be found.
332: */
333: static enum rofft
1.42 schwarze 334: roffhash_find(const char *p, size_t s)
1.1 schwarze 335: {
1.3 schwarze 336: int buc;
337: struct roffmac *n;
1.1 schwarze 338:
1.3 schwarze 339: /*
340: * libroff has an extremely simple hashtable, for the time
341: * being, which simply keys on the first character, which must
342: * be printable, then walks a chain. It works well enough until
343: * optimised.
344: */
345:
346: if (p[0] < ASCII_LO || p[0] > ASCII_HI)
347: return(ROFF_MAX);
348:
349: buc = ROFF_HASH(p);
350:
351: if (NULL == (n = hash[buc]))
352: return(ROFF_MAX);
353: for ( ; n; n = n->next)
1.16 schwarze 354: if (0 == strncmp(n->name, p, s) && '\0' == n->name[(int)s])
1.3 schwarze 355: return((enum rofft)(n - roffs));
1.1 schwarze 356:
357: return(ROFF_MAX);
358: }
359:
360:
361: /*
362: * Pop the current node off of the stack of roff instructions currently
363: * pending.
364: */
365: static void
366: roffnode_pop(struct roff *r)
367: {
368: struct roffnode *p;
369:
1.2 schwarze 370: assert(r->last);
371: p = r->last;
372:
373: r->last = r->last->parent;
1.16 schwarze 374: free(p->name);
375: free(p->end);
1.1 schwarze 376: free(p);
377: }
378:
379:
380: /*
381: * Push a roff node onto the instruction stack. This must later be
382: * removed with roffnode_pop().
383: */
1.11 schwarze 384: static void
1.16 schwarze 385: roffnode_push(struct roff *r, enum rofft tok, const char *name,
386: int line, int col)
1.1 schwarze 387: {
388: struct roffnode *p;
389:
1.11 schwarze 390: p = mandoc_calloc(1, sizeof(struct roffnode));
1.1 schwarze 391: p->tok = tok;
1.16 schwarze 392: if (name)
393: p->name = mandoc_strdup(name);
1.1 schwarze 394: p->parent = r->last;
395: p->line = line;
396: p->col = col;
1.2 schwarze 397: p->rule = p->parent ? p->parent->rule : ROFFRULE_DENY;
1.1 schwarze 398:
399: r->last = p;
400: }
401:
402:
403: static void
404: roff_free1(struct roff *r)
405: {
1.49 schwarze 406: struct tbl_node *tbl;
1.32 schwarze 407: struct eqn_node *e;
1.42 schwarze 408: int i;
1.27 schwarze 409:
1.49 schwarze 410: while (NULL != (tbl = r->first_tbl)) {
411: r->first_tbl = tbl->next;
412: tbl_free(tbl);
1.27 schwarze 413: }
414:
415: r->first_tbl = r->last_tbl = r->tbl = NULL;
1.1 schwarze 416:
1.32 schwarze 417: while (NULL != (e = r->first_eqn)) {
418: r->first_eqn = e->next;
419: eqn_free(e);
420: }
421:
422: r->first_eqn = r->last_eqn = r->eqn = NULL;
423:
1.1 schwarze 424: while (r->last)
425: roffnode_pop(r);
1.27 schwarze 426:
1.42 schwarze 427: roff_freestr(r->strtab);
428: roff_freestr(r->xmbtab);
429:
430: r->strtab = r->xmbtab = NULL;
431:
1.52 schwarze 432: roff_freereg(r->regtab);
433:
434: r->regtab = NULL;
435:
1.42 schwarze 436: if (r->xtab)
437: for (i = 0; i < 128; i++)
438: free(r->xtab[i].p);
439:
440: free(r->xtab);
441: r->xtab = NULL;
1.1 schwarze 442: }
443:
444: void
445: roff_reset(struct roff *r)
446: {
447:
448: roff_free1(r);
1.48 schwarze 449: r->control = 0;
1.1 schwarze 450: }
451:
452:
453: void
454: roff_free(struct roff *r)
455: {
456:
457: roff_free1(r);
458: free(r);
459: }
460:
461:
462: struct roff *
1.63 schwarze 463: roff_alloc(enum mparset type, struct mparse *parse, int quick)
1.1 schwarze 464: {
465: struct roff *r;
466:
1.11 schwarze 467: r = mandoc_calloc(1, sizeof(struct roff));
1.47 schwarze 468: r->parsetype = type;
1.35 schwarze 469: r->parse = parse;
1.63 schwarze 470: r->quick = quick;
1.2 schwarze 471: r->rstackpos = -1;
1.3 schwarze 472:
1.42 schwarze 473: roffhash_init();
1.37 schwarze 474:
1.1 schwarze 475: return(r);
476: }
477:
1.8 schwarze 478: /*
1.53 schwarze 479: * In the current line, expand user-defined strings ("\*")
480: * and references to number registers ("\n").
481: * Also check the syntax of other escape sequences.
1.8 schwarze 482: */
1.45 schwarze 483: static enum rofferr
1.37 schwarze 484: roff_res(struct roff *r, char **bufp, size_t *szp, int ln, int pos)
1.8 schwarze 485: {
1.53 schwarze 486: char ubuf[12]; /* buffer to print the number */
1.23 schwarze 487: const char *stesc; /* start of an escape sequence ('\\') */
488: const char *stnam; /* start of the name, after "[(*" */
489: const char *cp; /* end of the name, e.g. before ']' */
490: const char *res; /* the string to be substituted */
1.53 schwarze 491: char *nbuf; /* new buffer to copy bufp to */
492: size_t nsz; /* size of the new buffer */
493: size_t maxl; /* expected length of the escape name */
494: size_t naml; /* actual length of the escape name */
495: int expand_count; /* to avoid infinite loops */
1.8 schwarze 496:
1.43 schwarze 497: expand_count = 0;
498:
1.42 schwarze 499: again:
1.24 schwarze 500: cp = *bufp + pos;
501: while (NULL != (cp = strchr(cp, '\\'))) {
502: stesc = cp++;
1.23 schwarze 503:
504: /*
1.53 schwarze 505: * The second character must be an asterisk or an n.
1.23 schwarze 506: * If it isn't, skip it anyway: It is escaped,
507: * so it can't start another escape sequence.
508: */
509:
1.24 schwarze 510: if ('\0' == *cp)
1.45 schwarze 511: return(ROFF_CONT);
1.42 schwarze 512:
1.53 schwarze 513: switch (*cp) {
514: case ('*'):
515: res = NULL;
516: break;
517: case ('n'):
518: res = ubuf;
519: break;
520: default:
521: if (ESCAPE_ERROR != mandoc_escape(&cp, NULL, NULL))
1.42 schwarze 522: continue;
523: mandoc_msg
524: (MANDOCERR_BADESCAPE, r->parse,
525: ln, (int)(stesc - *bufp), NULL);
1.45 schwarze 526: return(ROFF_CONT);
1.42 schwarze 527: }
528:
529: cp++;
1.23 schwarze 530:
531: /*
532: * The third character decides the length
1.53 schwarze 533: * of the name of the string or register.
1.23 schwarze 534: * Save a pointer to the name.
535: */
536:
1.24 schwarze 537: switch (*cp) {
538: case ('\0'):
1.45 schwarze 539: return(ROFF_CONT);
1.8 schwarze 540: case ('('):
541: cp++;
542: maxl = 2;
543: break;
544: case ('['):
545: cp++;
546: maxl = 0;
547: break;
548: default:
549: maxl = 1;
550: break;
551: }
1.23 schwarze 552: stnam = cp;
1.8 schwarze 553:
1.23 schwarze 554: /* Advance to the end of the name. */
1.8 schwarze 555:
1.53 schwarze 556: for (naml = 0; 0 == maxl || naml < maxl; naml++, cp++) {
1.42 schwarze 557: if ('\0' == *cp) {
558: mandoc_msg
559: (MANDOCERR_BADESCAPE,
560: r->parse, ln,
561: (int)(stesc - *bufp), NULL);
1.45 schwarze 562: return(ROFF_CONT);
1.42 schwarze 563: }
1.8 schwarze 564: if (0 == maxl && ']' == *cp)
565: break;
566: }
567:
1.23 schwarze 568: /*
569: * Retrieve the replacement string; if it is
570: * undefined, resume searching for escapes.
571: */
572:
1.53 schwarze 573: if (NULL == res)
574: res = roff_getstrn(r, stnam, naml);
575: else
576: snprintf(ubuf, sizeof(ubuf), "%d",
577: roff_getregn(r, stnam, naml));
1.8 schwarze 578:
579: if (NULL == res) {
1.42 schwarze 580: mandoc_msg
581: (MANDOCERR_BADESCAPE, r->parse,
582: ln, (int)(stesc - *bufp), NULL);
1.37 schwarze 583: res = "";
1.8 schwarze 584: }
585:
1.23 schwarze 586: /* Replace the escape sequence by the string. */
587:
1.42 schwarze 588: pos = stesc - *bufp;
589:
1.8 schwarze 590: nsz = *szp + strlen(res) + 1;
1.53 schwarze 591: nbuf = mandoc_malloc(nsz);
1.8 schwarze 592:
1.53 schwarze 593: strlcpy(nbuf, *bufp, (size_t)(stesc - *bufp + 1));
594: strlcat(nbuf, res, nsz);
595: strlcat(nbuf, cp + (maxl ? 0 : 1), nsz);
1.8 schwarze 596:
597: free(*bufp);
598:
1.53 schwarze 599: *bufp = nbuf;
1.8 schwarze 600: *szp = nsz;
1.43 schwarze 601:
602: if (EXPAND_LIMIT >= ++expand_count)
603: goto again;
604:
605: /* Just leave the string unexpanded. */
606: mandoc_msg(MANDOCERR_ROFFLOOP, r->parse, ln, pos, NULL);
1.45 schwarze 607: return(ROFF_IGN);
1.42 schwarze 608: }
1.45 schwarze 609: return(ROFF_CONT);
1.42 schwarze 610: }
611:
612: /*
1.51 schwarze 613: * Process text streams:
614: * Convert all breakable hyphens into ASCII_HYPH.
615: * Decrement and spring input line trap.
1.42 schwarze 616: */
617: static enum rofferr
1.51 schwarze 618: roff_parsetext(char **bufp, size_t *szp, int pos, int *offs)
1.42 schwarze 619: {
620: size_t sz;
621: const char *start;
1.51 schwarze 622: char *p;
623: int isz;
1.42 schwarze 624: enum mandoc_esc esc;
625:
1.51 schwarze 626: start = p = *bufp + pos;
1.42 schwarze 627:
628: while ('\0' != *p) {
629: sz = strcspn(p, "-\\");
630: p += sz;
631:
632: if ('\0' == *p)
633: break;
634:
635: if ('\\' == *p) {
636: /* Skip over escapes. */
637: p++;
1.62 schwarze 638: esc = mandoc_escape((const char **)&p, NULL, NULL);
1.42 schwarze 639: if (ESCAPE_ERROR == esc)
640: break;
641: continue;
642: } else if (p == start) {
643: p++;
644: continue;
645: }
646:
1.44 schwarze 647: if (isalpha((unsigned char)p[-1]) &&
648: isalpha((unsigned char)p[1]))
1.42 schwarze 649: *p = ASCII_HYPH;
650: p++;
1.8 schwarze 651: }
652:
1.51 schwarze 653: /* Spring the input line trap. */
654: if (1 == roffit_lines) {
655: isz = asprintf(&p, "%s\n.%s", *bufp, roffit_macro);
656: if (-1 == isz) {
657: perror(NULL);
658: exit((int)MANDOCLEVEL_SYSERR);
659: }
660: free(*bufp);
661: *bufp = p;
662: *szp = isz + 1;
663: *offs = 0;
664: free(roffit_macro);
665: roffit_lines = 0;
666: return(ROFF_REPARSE);
667: } else if (1 < roffit_lines)
668: --roffit_lines;
1.42 schwarze 669: return(ROFF_CONT);
1.8 schwarze 670: }
671:
1.1 schwarze 672: enum rofferr
1.6 schwarze 673: roff_parseln(struct roff *r, int ln, char **bufp,
674: size_t *szp, int pos, int *offs)
1.1 schwarze 675: {
676: enum rofft t;
1.27 schwarze 677: enum rofferr e;
1.35 schwarze 678: int ppos, ctl;
1.1 schwarze 679:
1.2 schwarze 680: /*
1.8 schwarze 681: * Run the reserved-word filter only if we have some reserved
682: * words to fill in.
683: */
684:
1.45 schwarze 685: e = roff_res(r, bufp, szp, ln, pos);
686: if (ROFF_IGN == e)
687: return(e);
688: assert(ROFF_CONT == e);
1.8 schwarze 689:
1.35 schwarze 690: ppos = pos;
1.48 schwarze 691: ctl = roff_getcontrol(r, *bufp, &pos);
1.35 schwarze 692:
1.8 schwarze 693: /*
1.2 schwarze 694: * First, if a scope is open and we're not a macro, pass the
695: * text through the macro's filter. If a scope isn't open and
696: * we're not a macro, just let it through.
1.32 schwarze 697: * Finally, if there's an equation scope open, divert it into it
698: * no matter our state.
1.2 schwarze 699: */
700:
1.35 schwarze 701: if (r->last && ! ctl) {
1.2 schwarze 702: t = r->last->tok;
703: assert(roffs[t].text);
1.27 schwarze 704: e = (*roffs[t].text)
705: (r, t, bufp, szp, ln, pos, pos, offs);
706: assert(ROFF_IGN == e || ROFF_CONT == e);
1.32 schwarze 707: if (ROFF_CONT != e)
708: return(e);
1.54 schwarze 709: }
710: if (r->eqn)
711: return(eqn_read(&r->eqn, ln, *bufp, ppos, offs));
712: if ( ! ctl) {
1.32 schwarze 713: if (r->tbl)
1.35 schwarze 714: return(tbl_read(r->tbl, ln, *bufp, pos));
1.51 schwarze 715: return(roff_parsetext(bufp, szp, pos, offs));
1.54 schwarze 716: }
1.2 schwarze 717:
718: /*
719: * If a scope is open, go to the child handler for that macro,
720: * as it may want to preprocess before doing anything with it.
1.32 schwarze 721: * Don't do so if an equation is open.
1.2 schwarze 722: */
723:
724: if (r->last) {
1.1 schwarze 725: t = r->last->tok;
726: assert(roffs[t].sub);
1.2 schwarze 727: return((*roffs[t].sub)
1.8 schwarze 728: (r, t, bufp, szp,
1.35 schwarze 729: ln, ppos, pos, offs));
1.2 schwarze 730: }
731:
732: /*
733: * Lastly, as we've no scope open, try to look up and execute
734: * the new macro. If no macro is found, simply return and let
735: * the compilers handle it.
736: */
737:
1.16 schwarze 738: if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos)))
1.1 schwarze 739: return(ROFF_CONT);
740:
1.2 schwarze 741: assert(roffs[t].proc);
742: return((*roffs[t].proc)
1.8 schwarze 743: (r, t, bufp, szp,
744: ln, ppos, pos, offs));
1.2 schwarze 745: }
746:
1.1 schwarze 747:
1.27 schwarze 748: void
1.2 schwarze 749: roff_endparse(struct roff *r)
750: {
1.1 schwarze 751:
1.27 schwarze 752: if (r->last)
1.35 schwarze 753: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.27 schwarze 754: r->last->line, r->last->col, NULL);
755:
1.32 schwarze 756: if (r->eqn) {
1.35 schwarze 757: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.41 schwarze 758: r->eqn->eqn.ln, r->eqn->eqn.pos, NULL);
759: eqn_end(&r->eqn);
1.32 schwarze 760: }
761:
1.27 schwarze 762: if (r->tbl) {
1.35 schwarze 763: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.27 schwarze 764: r->tbl->line, r->tbl->pos, NULL);
1.41 schwarze 765: tbl_end(&r->tbl);
1.27 schwarze 766: }
1.1 schwarze 767: }
768:
769: /*
770: * Parse a roff node's type from the input buffer. This must be in the
771: * form of ".foo xxx" in the usual way.
772: */
773: static enum rofft
1.16 schwarze 774: roff_parse(struct roff *r, const char *buf, int *pos)
1.1 schwarze 775: {
1.16 schwarze 776: const char *mac;
777: size_t maclen;
1.1 schwarze 778: enum rofft t;
779:
1.39 schwarze 780: if ('\0' == buf[*pos] || '"' == buf[*pos] ||
781: '\t' == buf[*pos] || ' ' == buf[*pos])
1.1 schwarze 782: return(ROFF_MAX);
783:
1.39 schwarze 784: /*
785: * We stop the macro parse at an escape, tab, space, or nil.
786: * However, `\}' is also a valid macro, so make sure we don't
787: * clobber it by seeing the `\' as the end of token.
788: */
789:
1.16 schwarze 790: mac = buf + *pos;
1.39 schwarze 791: maclen = strcspn(mac + 1, " \\\t\0") + 1;
1.1 schwarze 792:
1.16 schwarze 793: t = (r->current_string = roff_getstrn(r, mac, maclen))
1.42 schwarze 794: ? ROFF_USERDEF : roffhash_find(mac, maclen);
1.1 schwarze 795:
1.34 schwarze 796: *pos += (int)maclen;
1.35 schwarze 797:
1.1 schwarze 798: while (buf[*pos] && ' ' == buf[*pos])
799: (*pos)++;
800:
801: return(t);
802: }
803:
804: /* ARGSUSED */
805: static enum rofferr
1.2 schwarze 806: roff_cblock(ROFF_ARGS)
1.1 schwarze 807: {
808:
1.2 schwarze 809: /*
810: * A block-close `..' should only be invoked as a child of an
811: * ignore macro, otherwise raise a warning and just ignore it.
812: */
813:
814: if (NULL == r->last) {
1.35 schwarze 815: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.2 schwarze 816: return(ROFF_IGN);
817: }
1.1 schwarze 818:
1.2 schwarze 819: switch (r->last->tok) {
820: case (ROFF_am):
821: /* FALLTHROUGH */
822: case (ROFF_ami):
823: /* FALLTHROUGH */
824: case (ROFF_am1):
825: /* FALLTHROUGH */
826: case (ROFF_de):
1.23 schwarze 827: /* ROFF_de1 is remapped to ROFF_de in roff_block(). */
1.2 schwarze 828: /* FALLTHROUGH */
829: case (ROFF_dei):
830: /* FALLTHROUGH */
831: case (ROFF_ig):
832: break;
833: default:
1.35 schwarze 834: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.1 schwarze 835: return(ROFF_IGN);
1.2 schwarze 836: }
837:
838: if ((*bufp)[pos])
1.35 schwarze 839: mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.2 schwarze 840:
841: roffnode_pop(r);
842: roffnode_cleanscope(r);
843: return(ROFF_IGN);
844:
845: }
1.1 schwarze 846:
847:
1.2 schwarze 848: static void
849: roffnode_cleanscope(struct roff *r)
850: {
1.1 schwarze 851:
1.2 schwarze 852: while (r->last) {
1.46 schwarze 853: if (--r->last->endspan != 0)
1.2 schwarze 854: break;
855: roffnode_pop(r);
856: }
857: }
1.1 schwarze 858:
859:
1.2 schwarze 860: /* ARGSUSED */
861: static enum rofferr
862: roff_ccond(ROFF_ARGS)
863: {
1.1 schwarze 864:
1.2 schwarze 865: if (NULL == r->last) {
1.35 schwarze 866: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.1 schwarze 867: return(ROFF_IGN);
1.2 schwarze 868: }
1.1 schwarze 869:
1.2 schwarze 870: switch (r->last->tok) {
871: case (ROFF_el):
872: /* FALLTHROUGH */
873: case (ROFF_ie):
874: /* FALLTHROUGH */
875: case (ROFF_if):
876: break;
877: default:
1.35 schwarze 878: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.2 schwarze 879: return(ROFF_IGN);
880: }
1.1 schwarze 881:
1.2 schwarze 882: if (r->last->endspan > -1) {
1.35 schwarze 883: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.1 schwarze 884: return(ROFF_IGN);
1.2 schwarze 885: }
886:
887: if ((*bufp)[pos])
1.35 schwarze 888: mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.1 schwarze 889:
1.2 schwarze 890: roffnode_pop(r);
891: roffnode_cleanscope(r);
1.1 schwarze 892: return(ROFF_IGN);
893: }
894:
895:
896: /* ARGSUSED */
897: static enum rofferr
1.2 schwarze 898: roff_block(ROFF_ARGS)
1.1 schwarze 899: {
1.2 schwarze 900: int sv;
901: size_t sz;
1.16 schwarze 902: char *name;
903:
904: name = NULL;
1.2 schwarze 905:
1.16 schwarze 906: if (ROFF_ig != tok) {
907: if ('\0' == (*bufp)[pos]) {
1.35 schwarze 908: mandoc_msg(MANDOCERR_NOARGS, r->parse, ln, ppos, NULL);
1.16 schwarze 909: return(ROFF_IGN);
910: }
1.22 schwarze 911:
912: /*
913: * Re-write `de1', since we don't really care about
914: * groff's strange compatibility mode, into `de'.
915: */
916:
1.18 schwarze 917: if (ROFF_de1 == tok)
918: tok = ROFF_de;
1.16 schwarze 919: if (ROFF_de == tok)
920: name = *bufp + pos;
1.21 schwarze 921: else
1.35 schwarze 922: mandoc_msg(MANDOCERR_REQUEST, r->parse, ln, ppos,
1.21 schwarze 923: roffs[tok].name);
1.22 schwarze 924:
1.33 schwarze 925: while ((*bufp)[pos] && ! isspace((unsigned char)(*bufp)[pos]))
1.2 schwarze 926: pos++;
1.22 schwarze 927:
1.33 schwarze 928: while (isspace((unsigned char)(*bufp)[pos]))
1.16 schwarze 929: (*bufp)[pos++] = '\0';
1.2 schwarze 930: }
931:
1.16 schwarze 932: roffnode_push(r, tok, name, ln, ppos);
933:
934: /*
935: * At the beginning of a `de' macro, clear the existing string
936: * with the same name, if there is one. New content will be
937: * added from roff_block_text() in multiline mode.
938: */
1.22 schwarze 939:
1.16 schwarze 940: if (ROFF_de == tok)
1.19 schwarze 941: roff_setstr(r, name, "", 0);
1.2 schwarze 942:
943: if ('\0' == (*bufp)[pos])
944: return(ROFF_IGN);
1.1 schwarze 945:
1.22 schwarze 946: /* If present, process the custom end-of-line marker. */
947:
1.2 schwarze 948: sv = pos;
1.33 schwarze 949: while ((*bufp)[pos] && ! isspace((unsigned char)(*bufp)[pos]))
1.2 schwarze 950: pos++;
951:
952: /*
953: * Note: groff does NOT like escape characters in the input.
954: * Instead of detecting this, we're just going to let it fly and
955: * to hell with it.
956: */
957:
958: assert(pos > sv);
959: sz = (size_t)(pos - sv);
960:
961: if (1 == sz && '.' == (*bufp)[sv])
962: return(ROFF_IGN);
963:
1.11 schwarze 964: r->last->end = mandoc_malloc(sz + 1);
1.2 schwarze 965:
966: memcpy(r->last->end, *bufp + sv, sz);
967: r->last->end[(int)sz] = '\0';
968:
969: if ((*bufp)[pos])
1.35 schwarze 970: mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.1 schwarze 971:
972: return(ROFF_IGN);
973: }
974:
975:
976: /* ARGSUSED */
977: static enum rofferr
1.2 schwarze 978: roff_block_sub(ROFF_ARGS)
1.1 schwarze 979: {
1.2 schwarze 980: enum rofft t;
981: int i, j;
982:
983: /*
984: * First check whether a custom macro exists at this level. If
985: * it does, then check against it. This is some of groff's
986: * stranger behaviours. If we encountered a custom end-scope
987: * tag and that tag also happens to be a "real" macro, then we
988: * need to try interpreting it again as a real macro. If it's
989: * not, then return ignore. Else continue.
990: */
991:
992: if (r->last->end) {
1.35 schwarze 993: for (i = pos, j = 0; r->last->end[j]; j++, i++)
1.2 schwarze 994: if ((*bufp)[i] != r->last->end[j])
995: break;
1.1 schwarze 996:
1.2 schwarze 997: if ('\0' == r->last->end[j] &&
998: ('\0' == (*bufp)[i] ||
999: ' ' == (*bufp)[i] ||
1000: '\t' == (*bufp)[i])) {
1001: roffnode_pop(r);
1002: roffnode_cleanscope(r);
1.1 schwarze 1003:
1.35 schwarze 1004: while (' ' == (*bufp)[i] || '\t' == (*bufp)[i])
1005: i++;
1006:
1007: pos = i;
1.16 schwarze 1008: if (ROFF_MAX != roff_parse(r, *bufp, &pos))
1.2 schwarze 1009: return(ROFF_RERUN);
1010: return(ROFF_IGN);
1011: }
1.1 schwarze 1012: }
1013:
1.2 schwarze 1014: /*
1015: * If we have no custom end-query or lookup failed, then try
1016: * pulling it out of the hashtable.
1017: */
1.1 schwarze 1018:
1.36 schwarze 1019: t = roff_parse(r, *bufp, &pos);
1.1 schwarze 1020:
1.16 schwarze 1021: /*
1022: * Macros other than block-end are only significant
1023: * in `de' blocks; elsewhere, simply throw them away.
1024: */
1025: if (ROFF_cblock != t) {
1026: if (ROFF_de == tok)
1027: roff_setstr(r, r->last->name, *bufp + ppos, 1);
1.1 schwarze 1028: return(ROFF_IGN);
1.16 schwarze 1029: }
1.1 schwarze 1030:
1.2 schwarze 1031: assert(roffs[t].proc);
1.6 schwarze 1032: return((*roffs[t].proc)(r, t, bufp, szp,
1033: ln, ppos, pos, offs));
1.2 schwarze 1034: }
1035:
1036:
1037: /* ARGSUSED */
1038: static enum rofferr
1039: roff_block_text(ROFF_ARGS)
1040: {
1041:
1.16 schwarze 1042: if (ROFF_de == tok)
1043: roff_setstr(r, r->last->name, *bufp + pos, 1);
1044:
1.2 schwarze 1045: return(ROFF_IGN);
1046: }
1047:
1048:
1049: /* ARGSUSED */
1050: static enum rofferr
1051: roff_cond_sub(ROFF_ARGS)
1052: {
1053: enum rofft t;
1054: enum roffrule rr;
1.37 schwarze 1055: char *ep;
1.2 schwarze 1056:
1057: rr = r->last->rule;
1.37 schwarze 1058: roffnode_cleanscope(r);
1.50 schwarze 1059: t = roff_parse(r, *bufp, &pos);
1.2 schwarze 1060:
1.37 schwarze 1061: /*
1.50 schwarze 1062: * Fully handle known macros when they are structurally
1063: * required or when the conditional evaluated to true.
1.5 schwarze 1064: */
1065:
1.50 schwarze 1066: if ((ROFF_MAX != t) &&
1067: (ROFF_ccond == t || ROFFRULE_ALLOW == rr ||
1068: ROFFMAC_STRUCT & roffs[t].flags)) {
1069: assert(roffs[t].proc);
1070: return((*roffs[t].proc)(r, t, bufp, szp,
1071: ln, ppos, pos, offs));
1072: }
1.39 schwarze 1073:
1.50 schwarze 1074: /* Always check for the closing delimiter `\}'. */
1.39 schwarze 1075:
1.50 schwarze 1076: ep = &(*bufp)[pos];
1077: while (NULL != (ep = strchr(ep, '\\'))) {
1078: if ('}' != *(++ep))
1079: continue;
1.2 schwarze 1080:
1.50 schwarze 1081: /*
1082: * If we're at the end of line, then just chop
1083: * off the \} and resize the buffer.
1084: * If we aren't, then convert it to spaces.
1085: */
1.37 schwarze 1086:
1.50 schwarze 1087: if ('\0' == *(ep + 1)) {
1088: *--ep = '\0';
1089: *szp -= 2;
1090: } else
1091: *(ep - 1) = *ep = ' ';
1.2 schwarze 1092:
1.50 schwarze 1093: roff_ccond(r, ROFF_ccond, bufp, szp,
1094: ln, pos, pos + 2, offs);
1095: break;
1096: }
1097: return(ROFFRULE_DENY == rr ? ROFF_IGN : ROFF_CONT);
1.2 schwarze 1098: }
1099:
1100: /* ARGSUSED */
1101: static enum rofferr
1102: roff_cond_text(ROFF_ARGS)
1103: {
1.37 schwarze 1104: char *ep;
1.2 schwarze 1105: enum roffrule rr;
1106:
1107: rr = r->last->rule;
1.37 schwarze 1108: roffnode_cleanscope(r);
1.1 schwarze 1109:
1.37 schwarze 1110: ep = &(*bufp)[pos];
1111: for ( ; NULL != (ep = strchr(ep, '\\')); ep++) {
1112: ep++;
1113: if ('}' != *ep)
1114: continue;
1115: *ep = '&';
1116: roff_ccond(r, ROFF_ccond, bufp, szp,
1117: ln, pos, pos + 2, offs);
1.2 schwarze 1118: }
1119: return(ROFFRULE_DENY == rr ? ROFF_IGN : ROFF_CONT);
1120: }
1121:
1.56 schwarze 1122: static int
1123: roff_getnum(const char *v, int *pos, int *res)
1124: {
1125: int p, n;
1126:
1127: p = *pos;
1128: n = v[p] == '-';
1129: if (n)
1130: p++;
1131:
1132: for (*res = 0; isdigit((unsigned char)v[p]); p++)
1133: *res += 10 * *res + v[p] - '0';
1134: if (p == *pos + n)
1135: return 0;
1136:
1137: if (n)
1138: *res = -*res;
1139:
1140: *pos = p;
1141: return 1;
1142: }
1143:
1144: static int
1145: roff_getop(const char *v, int *pos, char *res)
1146: {
1147: int e;
1148:
1149: *res = v[*pos];
1150: e = v[*pos + 1] == '=';
1151:
1152: switch (*res) {
1153: case '=':
1154: break;
1155: case '>':
1156: if (e)
1157: *res = 'g';
1158: break;
1159: case '<':
1160: if (e)
1161: *res = 'l';
1162: break;
1163: default:
1164: return(0);
1165: }
1166:
1167: *pos += 1 + e;
1168:
1169: return(*res);
1170: }
1171:
1.5 schwarze 1172: static enum roffrule
1173: roff_evalcond(const char *v, int *pos)
1174: {
1.56 schwarze 1175: int not, lh, rh;
1176: char op;
1.5 schwarze 1177:
1178: switch (v[*pos]) {
1179: case ('n'):
1180: (*pos)++;
1181: return(ROFFRULE_ALLOW);
1182: case ('e'):
1183: /* FALLTHROUGH */
1184: case ('o'):
1185: /* FALLTHROUGH */
1186: case ('t'):
1187: (*pos)++;
1188: return(ROFFRULE_DENY);
1.56 schwarze 1189: case ('!'):
1190: (*pos)++;
1191: not = 1;
1192: break;
1.5 schwarze 1193: default:
1.56 schwarze 1194: not = 0;
1.5 schwarze 1195: break;
1196: }
1197:
1.56 schwarze 1198: if (!roff_getnum(v, pos, &lh))
1199: return ROFFRULE_DENY;
1200: if (!roff_getop(v, pos, &op)) {
1201: if (lh < 0)
1202: lh = 0;
1203: goto out;
1204: }
1205: if (!roff_getnum(v, pos, &rh))
1206: return ROFFRULE_DENY;
1207: switch (op) {
1208: case 'g':
1209: lh = lh >= rh;
1210: break;
1211: case 'l':
1212: lh = lh <= rh;
1213: break;
1214: case '=':
1215: lh = lh == rh;
1216: break;
1217: case '>':
1218: lh = lh > rh;
1219: break;
1220: case '<':
1221: lh = lh < rh;
1222: break;
1223: default:
1224: return ROFFRULE_DENY;
1225: }
1226: out:
1227: if (not)
1228: lh = !lh;
1229: return lh ? ROFFRULE_ALLOW : ROFFRULE_DENY;
1.5 schwarze 1230: }
1231:
1.2 schwarze 1232: /* ARGSUSED */
1233: static enum rofferr
1.21 schwarze 1234: roff_line_ignore(ROFF_ARGS)
1.6 schwarze 1235: {
1.30 schwarze 1236:
1.21 schwarze 1237: return(ROFF_IGN);
1238: }
1239:
1240: /* ARGSUSED */
1241: static enum rofferr
1.2 schwarze 1242: roff_cond(ROFF_ARGS)
1243: {
1.46 schwarze 1244:
1245: roffnode_push(r, tok, NULL, ln, ppos);
1.2 schwarze 1246:
1.35 schwarze 1247: /*
1248: * An `.el' has no conditional body: it will consume the value
1249: * of the current rstack entry set in prior `ie' calls or
1250: * defaults to DENY.
1251: *
1252: * If we're not an `el', however, then evaluate the conditional.
1253: */
1.1 schwarze 1254:
1.46 schwarze 1255: r->last->rule = ROFF_el == tok ?
1.35 schwarze 1256: (r->rstackpos < 0 ?
1257: ROFFRULE_DENY : r->rstack[r->rstackpos--]) :
1258: roff_evalcond(*bufp, &pos);
1.2 schwarze 1259:
1.35 schwarze 1260: /*
1261: * An if-else will put the NEGATION of the current evaluated
1262: * conditional into the stack of rules.
1263: */
1264:
1.2 schwarze 1265: if (ROFF_ie == tok) {
1.35 schwarze 1266: if (r->rstackpos == RSTACK_MAX - 1) {
1267: mandoc_msg(MANDOCERR_MEM,
1268: r->parse, ln, ppos, NULL);
1269: return(ROFF_ERR);
1270: }
1271: r->rstack[++r->rstackpos] =
1272: ROFFRULE_DENY == r->last->rule ?
1273: ROFFRULE_ALLOW : ROFFRULE_DENY;
1.2 schwarze 1274: }
1.5 schwarze 1275:
1276: /* If the parent has false as its rule, then so do we. */
1277:
1.2 schwarze 1278: if (r->last->parent && ROFFRULE_DENY == r->last->parent->rule)
1279: r->last->rule = ROFFRULE_DENY;
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;
1.2 schwarze 1301: }
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:
1319:
1.2 schwarze 1320: /* ARGSUSED */
1321: static enum rofferr
1.7 schwarze 1322: roff_ds(ROFF_ARGS)
1323: {
1.10 schwarze 1324: char *name, *string;
1325:
1326: /*
1327: * A symbol is named by the first word following the macro
1328: * invocation up to a space. Its value is anything after the
1329: * name's trailing whitespace and optional double-quote. Thus,
1330: *
1331: * [.ds foo "bar " ]
1332: *
1333: * will have `bar " ' as its value.
1334: */
1.7 schwarze 1335:
1.28 schwarze 1336: string = *bufp + pos;
1337: name = roff_getname(r, &string, ln, pos);
1.7 schwarze 1338: if ('\0' == *name)
1339: return(ROFF_IGN);
1340:
1.28 schwarze 1341: /* Read past initial double-quote. */
1342: if ('"' == *string)
1.7 schwarze 1343: string++;
1344:
1.10 schwarze 1345: /* The rest is the value. */
1.16 schwarze 1346: roff_setstr(r, name, string, 0);
1.7 schwarze 1347: return(ROFF_IGN);
1348: }
1349:
1.52 schwarze 1350: void
1.60 schwarze 1351: roff_setreg(struct roff *r, const char *name, int val, char sign)
1.41 schwarze 1352: {
1.52 schwarze 1353: struct roffreg *reg;
1354:
1355: /* Search for an existing register with the same name. */
1356: reg = r->regtab;
1357:
1358: while (reg && strcmp(name, reg->key.p))
1359: reg = reg->next;
1.41 schwarze 1360:
1.52 schwarze 1361: if (NULL == reg) {
1362: /* Create a new register. */
1363: reg = mandoc_malloc(sizeof(struct roffreg));
1364: reg->key.p = mandoc_strdup(name);
1365: reg->key.sz = strlen(name);
1.60 schwarze 1366: reg->val = 0;
1.52 schwarze 1367: reg->next = r->regtab;
1368: r->regtab = reg;
1369: }
1370:
1.60 schwarze 1371: if ('+' == sign)
1372: reg->val += val;
1373: else if ('-' == sign)
1374: reg->val -= val;
1375: else
1376: reg->val = val;
1.41 schwarze 1377: }
1378:
1.53 schwarze 1379: int
1.52 schwarze 1380: roff_getreg(const struct roff *r, const char *name)
1.41 schwarze 1381: {
1.52 schwarze 1382: struct roffreg *reg;
1383:
1384: for (reg = r->regtab; reg; reg = reg->next)
1385: if (0 == strcmp(name, reg->key.p))
1.53 schwarze 1386: return(reg->val);
1387:
1388: return(0);
1389: }
1390:
1391: static int
1392: roff_getregn(const struct roff *r, const char *name, size_t len)
1393: {
1394: struct roffreg *reg;
1395:
1396: for (reg = r->regtab; reg; reg = reg->next)
1397: if (len == reg->key.sz &&
1398: 0 == strncmp(name, reg->key.p, len))
1399: return(reg->val);
1.41 schwarze 1400:
1.52 schwarze 1401: return(0);
1.41 schwarze 1402: }
1403:
1.52 schwarze 1404: static void
1405: roff_freereg(struct roffreg *reg)
1.41 schwarze 1406: {
1.52 schwarze 1407: struct roffreg *old_reg;
1.41 schwarze 1408:
1.52 schwarze 1409: while (NULL != reg) {
1410: free(reg->key.p);
1411: old_reg = reg;
1412: reg = reg->next;
1413: free(old_reg);
1414: }
1.41 schwarze 1415: }
1.7 schwarze 1416:
1417: /* ARGSUSED */
1418: static enum rofferr
1.6 schwarze 1419: roff_nr(ROFF_ARGS)
1.1 schwarze 1420: {
1.28 schwarze 1421: const char *key;
1422: char *val;
1.60 schwarze 1423: size_t sz;
1.37 schwarze 1424: int iv;
1.60 schwarze 1425: char sign;
1.6 schwarze 1426:
1.28 schwarze 1427: val = *bufp + pos;
1428: key = roff_getname(r, &val, ln, pos);
1.6 schwarze 1429:
1.60 schwarze 1430: sign = *val;
1431: if ('+' == sign || '-' == sign)
1432: val++;
1433:
1434: sz = strspn(val, "0123456789");
1435: iv = sz ? mandoc_strntoi(val, sz, 10) : 0;
1.52 schwarze 1436:
1.60 schwarze 1437: roff_setreg(r, key, iv, sign);
1.1 schwarze 1438:
1.29 schwarze 1439: return(ROFF_IGN);
1440: }
1441:
1442: /* ARGSUSED */
1443: static enum rofferr
1444: roff_rm(ROFF_ARGS)
1445: {
1446: const char *name;
1447: char *cp;
1448:
1449: cp = *bufp + pos;
1450: while ('\0' != *cp) {
1.34 schwarze 1451: name = roff_getname(r, &cp, ln, (int)(cp - *bufp));
1.29 schwarze 1452: if ('\0' != *name)
1453: roff_setstr(r, name, NULL, 0);
1454: }
1.51 schwarze 1455: return(ROFF_IGN);
1456: }
1457:
1458: /* ARGSUSED */
1459: static enum rofferr
1460: roff_it(ROFF_ARGS)
1461: {
1462: char *cp;
1463: size_t len;
1464: int iv;
1465:
1466: /* Parse the number of lines. */
1467: cp = *bufp + pos;
1468: len = strcspn(cp, " \t");
1469: cp[len] = '\0';
1470: if ((iv = mandoc_strntoi(cp, len, 10)) <= 0) {
1471: mandoc_msg(MANDOCERR_NUMERIC, r->parse,
1472: ln, ppos, *bufp + 1);
1473: return(ROFF_IGN);
1474: }
1475: cp += len + 1;
1476:
1477: /* Arm the input line trap. */
1478: roffit_lines = iv;
1479: roffit_macro = mandoc_strdup(cp);
1.2 schwarze 1480: return(ROFF_IGN);
1.47 schwarze 1481: }
1482:
1483: /* ARGSUSED */
1484: static enum rofferr
1485: roff_Dd(ROFF_ARGS)
1486: {
1487: const char *const *cp;
1488:
1.63 schwarze 1489: if (0 == r->quick && MPARSE_MDOC != r->parsetype)
1.47 schwarze 1490: for (cp = __mdoc_reserved; *cp; cp++)
1491: roff_setstr(r, *cp, NULL, 0);
1492:
1493: return(ROFF_CONT);
1494: }
1495:
1496: /* ARGSUSED */
1497: static enum rofferr
1498: roff_TH(ROFF_ARGS)
1499: {
1500: const char *const *cp;
1501:
1.63 schwarze 1502: if (0 == r->quick && MPARSE_MDOC != r->parsetype)
1.47 schwarze 1503: for (cp = __man_reserved; *cp; cp++)
1504: roff_setstr(r, *cp, NULL, 0);
1505:
1506: return(ROFF_CONT);
1.14 schwarze 1507: }
1508:
1509: /* ARGSUSED */
1510: static enum rofferr
1.27 schwarze 1511: roff_TE(ROFF_ARGS)
1512: {
1513:
1514: if (NULL == r->tbl)
1.35 schwarze 1515: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.27 schwarze 1516: else
1.41 schwarze 1517: tbl_end(&r->tbl);
1.27 schwarze 1518:
1519: return(ROFF_IGN);
1520: }
1521:
1522: /* ARGSUSED */
1523: static enum rofferr
1524: roff_T_(ROFF_ARGS)
1525: {
1526:
1527: if (NULL == r->tbl)
1.35 schwarze 1528: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.27 schwarze 1529: else
1530: tbl_restart(ppos, ln, r->tbl);
1531:
1532: return(ROFF_IGN);
1533: }
1534:
1.41 schwarze 1535: #if 0
1536: static int
1537: roff_closeeqn(struct roff *r)
1538: {
1539:
1540: return(r->eqn && ROFF_EQN == eqn_end(&r->eqn) ? 1 : 0);
1541: }
1542: #endif
1543:
1544: static void
1545: roff_openeqn(struct roff *r, const char *name, int line,
1546: int offs, const char *buf)
1.32 schwarze 1547: {
1.41 schwarze 1548: struct eqn_node *e;
1549: int poff;
1.32 schwarze 1550:
1551: assert(NULL == r->eqn);
1.41 schwarze 1552: e = eqn_alloc(name, offs, line, r->parse);
1.32 schwarze 1553:
1554: if (r->last_eqn)
1555: r->last_eqn->next = e;
1556: else
1557: r->first_eqn = r->last_eqn = e;
1558:
1559: r->eqn = r->last_eqn = e;
1.41 schwarze 1560:
1561: if (buf) {
1562: poff = 0;
1563: eqn_read(&r->eqn, line, buf, offs, &poff);
1564: }
1565: }
1566:
1567: /* ARGSUSED */
1568: static enum rofferr
1569: roff_EQ(ROFF_ARGS)
1570: {
1571:
1572: roff_openeqn(r, *bufp + pos, ln, ppos, NULL);
1.32 schwarze 1573: return(ROFF_IGN);
1574: }
1575:
1576: /* ARGSUSED */
1577: static enum rofferr
1578: roff_EN(ROFF_ARGS)
1579: {
1580:
1.35 schwarze 1581: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.32 schwarze 1582: return(ROFF_IGN);
1583: }
1584:
1585: /* ARGSUSED */
1586: static enum rofferr
1.27 schwarze 1587: roff_TS(ROFF_ARGS)
1588: {
1.49 schwarze 1589: struct tbl_node *tbl;
1.27 schwarze 1590:
1591: if (r->tbl) {
1.35 schwarze 1592: mandoc_msg(MANDOCERR_SCOPEBROKEN, r->parse, ln, ppos, NULL);
1.41 schwarze 1593: tbl_end(&r->tbl);
1.27 schwarze 1594: }
1595:
1.49 schwarze 1596: tbl = tbl_alloc(ppos, ln, r->parse);
1.27 schwarze 1597:
1598: if (r->last_tbl)
1.49 schwarze 1599: r->last_tbl->next = tbl;
1.27 schwarze 1600: else
1.49 schwarze 1601: r->first_tbl = r->last_tbl = tbl;
1.27 schwarze 1602:
1.49 schwarze 1603: r->tbl = r->last_tbl = tbl;
1.27 schwarze 1604: return(ROFF_IGN);
1605: }
1606:
1607: /* ARGSUSED */
1608: static enum rofferr
1.48 schwarze 1609: roff_cc(ROFF_ARGS)
1610: {
1611: const char *p;
1612:
1613: p = *bufp + pos;
1614:
1615: if ('\0' == *p || '.' == (r->control = *p++))
1616: r->control = 0;
1617:
1618: if ('\0' != *p)
1619: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
1620:
1621: return(ROFF_IGN);
1622: }
1623:
1624: /* ARGSUSED */
1625: static enum rofferr
1.42 schwarze 1626: roff_tr(ROFF_ARGS)
1627: {
1628: const char *p, *first, *second;
1629: size_t fsz, ssz;
1630: enum mandoc_esc esc;
1631:
1632: p = *bufp + pos;
1633:
1634: if ('\0' == *p) {
1635: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
1636: return(ROFF_IGN);
1637: }
1638:
1639: while ('\0' != *p) {
1640: fsz = ssz = 1;
1641:
1642: first = p++;
1643: if ('\\' == *first) {
1644: esc = mandoc_escape(&p, NULL, NULL);
1645: if (ESCAPE_ERROR == esc) {
1646: mandoc_msg
1647: (MANDOCERR_BADESCAPE, r->parse,
1648: ln, (int)(p - *bufp), NULL);
1649: return(ROFF_IGN);
1650: }
1651: fsz = (size_t)(p - first);
1652: }
1653:
1654: second = p++;
1655: if ('\\' == *second) {
1656: esc = mandoc_escape(&p, NULL, NULL);
1657: if (ESCAPE_ERROR == esc) {
1658: mandoc_msg
1659: (MANDOCERR_BADESCAPE, r->parse,
1660: ln, (int)(p - *bufp), NULL);
1661: return(ROFF_IGN);
1662: }
1663: ssz = (size_t)(p - second);
1664: } else if ('\0' == *second) {
1665: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse,
1666: ln, (int)(p - *bufp), NULL);
1667: second = " ";
1668: p--;
1669: }
1670:
1671: if (fsz > 1) {
1672: roff_setstrn(&r->xmbtab, first,
1673: fsz, second, ssz, 0);
1674: continue;
1675: }
1676:
1677: if (NULL == r->xtab)
1678: r->xtab = mandoc_calloc
1679: (128, sizeof(struct roffstr));
1680:
1681: free(r->xtab[(int)*first].p);
1682: r->xtab[(int)*first].p = mandoc_strndup(second, ssz);
1683: r->xtab[(int)*first].sz = ssz;
1684: }
1685:
1686: return(ROFF_IGN);
1687: }
1688:
1689: /* ARGSUSED */
1690: static enum rofferr
1.14 schwarze 1691: roff_so(ROFF_ARGS)
1692: {
1693: char *name;
1.15 schwarze 1694:
1.35 schwarze 1695: mandoc_msg(MANDOCERR_SO, r->parse, ln, ppos, NULL);
1.14 schwarze 1696:
1.22 schwarze 1697: /*
1698: * Handle `so'. Be EXTREMELY careful, as we shouldn't be
1699: * opening anything that's not in our cwd or anything beneath
1700: * it. Thus, explicitly disallow traversing up the file-system
1701: * or using absolute paths.
1702: */
1703:
1.14 schwarze 1704: name = *bufp + pos;
1705: if ('/' == *name || strstr(name, "../") || strstr(name, "/..")) {
1.35 schwarze 1706: mandoc_msg(MANDOCERR_SOPATH, r->parse, ln, pos, NULL);
1.14 schwarze 1707: return(ROFF_ERR);
1708: }
1709:
1710: *offs = pos;
1711: return(ROFF_SO);
1.7 schwarze 1712: }
1713:
1.16 schwarze 1714: /* ARGSUSED */
1715: static enum rofferr
1716: roff_userdef(ROFF_ARGS)
1.12 schwarze 1717: {
1.16 schwarze 1718: const char *arg[9];
1719: char *cp, *n1, *n2;
1.25 schwarze 1720: int i;
1.12 schwarze 1721:
1.16 schwarze 1722: /*
1723: * Collect pointers to macro argument strings
1.61 schwarze 1724: * and NUL-terminate them.
1.16 schwarze 1725: */
1726: cp = *bufp + pos;
1.25 schwarze 1727: for (i = 0; i < 9; i++)
1.26 schwarze 1728: arg[i] = '\0' == *cp ? "" :
1.35 schwarze 1729: mandoc_getarg(r->parse, &cp, ln, &pos);
1.16 schwarze 1730:
1731: /*
1732: * Expand macro arguments.
1.12 schwarze 1733: */
1.16 schwarze 1734: *szp = 0;
1735: n1 = cp = mandoc_strdup(r->current_string);
1736: while (NULL != (cp = strstr(cp, "\\$"))) {
1737: i = cp[2] - '1';
1738: if (0 > i || 8 < i) {
1739: /* Not an argument invocation. */
1740: cp += 2;
1741: continue;
1742: }
1743:
1744: *szp = strlen(n1) - 3 + strlen(arg[i]) + 1;
1745: n2 = mandoc_malloc(*szp);
1746:
1747: strlcpy(n2, n1, (size_t)(cp - n1 + 1));
1748: strlcat(n2, arg[i], *szp);
1749: strlcat(n2, cp + 3, *szp);
1750:
1751: cp = n2 + (cp - n1);
1752: free(n1);
1753: n1 = n2;
1.12 schwarze 1754: }
1755:
1.16 schwarze 1756: /*
1757: * Replace the macro invocation
1758: * by the expanded macro.
1759: */
1760: free(*bufp);
1761: *bufp = n1;
1762: if (0 == *szp)
1763: *szp = strlen(*bufp) + 1;
1764:
1.19 schwarze 1765: return(*szp > 1 && '\n' == (*bufp)[(int)*szp - 2] ?
1.16 schwarze 1766: ROFF_REPARSE : ROFF_APPEND);
1.12 schwarze 1767: }
1.28 schwarze 1768:
1769: static char *
1770: roff_getname(struct roff *r, char **cpp, int ln, int pos)
1771: {
1772: char *name, *cp;
1773:
1774: name = *cpp;
1775: if ('\0' == *name)
1776: return(name);
1777:
1778: /* Read until end of name. */
1779: for (cp = name; '\0' != *cp && ' ' != *cp; cp++) {
1780: if ('\\' != *cp)
1781: continue;
1782: cp++;
1783: if ('\\' == *cp)
1784: continue;
1.35 schwarze 1785: mandoc_msg(MANDOCERR_NAMESC, r->parse, ln, pos, NULL);
1.28 schwarze 1786: *cp = '\0';
1787: name = cp;
1788: }
1789:
1790: /* Nil-terminate name. */
1791: if ('\0' != *cp)
1792: *(cp++) = '\0';
1793:
1794: /* Read past spaces. */
1795: while (' ' == *cp)
1796: cp++;
1797:
1798: *cpp = cp;
1799: return(name);
1800: }
1801:
1.16 schwarze 1802: /*
1803: * Store *string into the user-defined string called *name.
1804: * In multiline mode, append to an existing entry and append '\n';
1805: * else replace the existing entry, if there is one.
1806: * To clear an existing entry, call with (*r, *name, NULL, 0).
1807: */
1.8 schwarze 1808: static void
1.16 schwarze 1809: roff_setstr(struct roff *r, const char *name, const char *string,
1810: int multiline)
1.7 schwarze 1811: {
1.42 schwarze 1812:
1813: roff_setstrn(&r->strtab, name, strlen(name), string,
1814: string ? strlen(string) : 0, multiline);
1815: }
1816:
1817: static void
1818: roff_setstrn(struct roffkv **r, const char *name, size_t namesz,
1819: const char *string, size_t stringsz, int multiline)
1820: {
1821: struct roffkv *n;
1822: char *c;
1823: int i;
1824: size_t oldch, newch;
1.7 schwarze 1825:
1.16 schwarze 1826: /* Search for an existing string with the same name. */
1.42 schwarze 1827: n = *r;
1828:
1829: while (n && strcmp(name, n->key.p))
1.7 schwarze 1830: n = n->next;
1.8 schwarze 1831:
1832: if (NULL == n) {
1.16 schwarze 1833: /* Create a new string table entry. */
1.42 schwarze 1834: n = mandoc_malloc(sizeof(struct roffkv));
1835: n->key.p = mandoc_strndup(name, namesz);
1836: n->key.sz = namesz;
1837: n->val.p = NULL;
1838: n->val.sz = 0;
1839: n->next = *r;
1840: *r = n;
1.16 schwarze 1841: } else if (0 == multiline) {
1842: /* In multiline mode, append; else replace. */
1.42 schwarze 1843: free(n->val.p);
1844: n->val.p = NULL;
1845: n->val.sz = 0;
1.16 schwarze 1846: }
1847:
1848: if (NULL == string)
1849: return;
1850:
1851: /*
1852: * One additional byte for the '\n' in multiline mode,
1853: * and one for the terminating '\0'.
1854: */
1.42 schwarze 1855: newch = stringsz + (multiline ? 2u : 1u);
1856:
1857: if (NULL == n->val.p) {
1858: n->val.p = mandoc_malloc(newch);
1859: *n->val.p = '\0';
1.16 schwarze 1860: oldch = 0;
1861: } else {
1.42 schwarze 1862: oldch = n->val.sz;
1863: n->val.p = mandoc_realloc(n->val.p, oldch + newch);
1.16 schwarze 1864: }
1865:
1866: /* Skip existing content in the destination buffer. */
1.42 schwarze 1867: c = n->val.p + (int)oldch;
1.16 schwarze 1868:
1869: /* Append new content to the destination buffer. */
1.42 schwarze 1870: i = 0;
1871: while (i < (int)stringsz) {
1.16 schwarze 1872: /*
1873: * Rudimentary roff copy mode:
1874: * Handle escaped backslashes.
1875: */
1.42 schwarze 1876: if ('\\' == string[i] && '\\' == string[i + 1])
1877: i++;
1878: *c++ = string[i++];
1.16 schwarze 1879: }
1.8 schwarze 1880:
1.16 schwarze 1881: /* Append terminating bytes. */
1882: if (multiline)
1883: *c++ = '\n';
1.42 schwarze 1884:
1.16 schwarze 1885: *c = '\0';
1.42 schwarze 1886: n->val.sz = (int)(c - n->val.p);
1.7 schwarze 1887: }
1888:
1.8 schwarze 1889: static const char *
1890: roff_getstrn(const struct roff *r, const char *name, size_t len)
1.7 schwarze 1891: {
1.42 schwarze 1892: const struct roffkv *n;
1.64 ! schwarze 1893: int i;
1.7 schwarze 1894:
1.42 schwarze 1895: for (n = r->strtab; n; n = n->next)
1896: if (0 == strncmp(name, n->key.p, len) &&
1897: '\0' == n->key.p[(int)len])
1898: return(n->val.p);
1.64 ! schwarze 1899:
! 1900: for (i = 0; i < PREDEFS_MAX; i++)
! 1901: if (0 == strncmp(name, predefs[i].name, len) &&
! 1902: '\0' == predefs[i].name[(int)len])
! 1903: return(predefs[i].str);
1.8 schwarze 1904:
1.42 schwarze 1905: return(NULL);
1.7 schwarze 1906: }
1907:
1.8 schwarze 1908: static void
1.42 schwarze 1909: roff_freestr(struct roffkv *r)
1.7 schwarze 1910: {
1.42 schwarze 1911: struct roffkv *n, *nn;
1.7 schwarze 1912:
1.42 schwarze 1913: for (n = r; n; n = nn) {
1914: free(n->key.p);
1915: free(n->val.p);
1.7 schwarze 1916: nn = n->next;
1917: free(n);
1918: }
1.27 schwarze 1919: }
1920:
1921: const struct tbl_span *
1922: roff_span(const struct roff *r)
1923: {
1924:
1925: return(r->tbl ? tbl_span(r->tbl) : NULL);
1.32 schwarze 1926: }
1927:
1928: const struct eqn *
1929: roff_eqn(const struct roff *r)
1930: {
1931:
1932: return(r->last_eqn ? &r->last_eqn->eqn : NULL);
1.42 schwarze 1933: }
1934:
1935: /*
1936: * Duplicate an input string, making the appropriate character
1937: * conversations (as stipulated by `tr') along the way.
1938: * Returns a heap-allocated string with all the replacements made.
1939: */
1940: char *
1941: roff_strdup(const struct roff *r, const char *p)
1942: {
1943: const struct roffkv *cp;
1944: char *res;
1945: const char *pp;
1946: size_t ssz, sz;
1947: enum mandoc_esc esc;
1948:
1949: if (NULL == r->xmbtab && NULL == r->xtab)
1950: return(mandoc_strdup(p));
1951: else if ('\0' == *p)
1952: return(mandoc_strdup(""));
1953:
1954: /*
1955: * Step through each character looking for term matches
1956: * (remember that a `tr' can be invoked with an escape, which is
1957: * a glyph but the escape is multi-character).
1958: * We only do this if the character hash has been initialised
1959: * and the string is >0 length.
1960: */
1961:
1962: res = NULL;
1963: ssz = 0;
1964:
1965: while ('\0' != *p) {
1966: if ('\\' != *p && r->xtab && r->xtab[(int)*p].p) {
1967: sz = r->xtab[(int)*p].sz;
1968: res = mandoc_realloc(res, ssz + sz + 1);
1969: memcpy(res + ssz, r->xtab[(int)*p].p, sz);
1970: ssz += sz;
1971: p++;
1972: continue;
1973: } else if ('\\' != *p) {
1974: res = mandoc_realloc(res, ssz + 2);
1975: res[ssz++] = *p++;
1976: continue;
1977: }
1978:
1979: /* Search for term matches. */
1980: for (cp = r->xmbtab; cp; cp = cp->next)
1981: if (0 == strncmp(p, cp->key.p, cp->key.sz))
1982: break;
1983:
1984: if (NULL != cp) {
1985: /*
1986: * A match has been found.
1987: * Append the match to the array and move
1988: * forward by its keysize.
1989: */
1990: res = mandoc_realloc
1991: (res, ssz + cp->val.sz + 1);
1992: memcpy(res + ssz, cp->val.p, cp->val.sz);
1993: ssz += cp->val.sz;
1994: p += (int)cp->key.sz;
1995: continue;
1996: }
1997:
1998: /*
1999: * Handle escapes carefully: we need to copy
2000: * over just the escape itself, or else we might
2001: * do replacements within the escape itself.
2002: * Make sure to pass along the bogus string.
2003: */
2004: pp = p++;
2005: esc = mandoc_escape(&p, NULL, NULL);
2006: if (ESCAPE_ERROR == esc) {
2007: sz = strlen(pp);
2008: res = mandoc_realloc(res, ssz + sz + 1);
2009: memcpy(res + ssz, pp, sz);
2010: break;
2011: }
2012: /*
2013: * We bail out on bad escapes.
2014: * No need to warn: we already did so when
2015: * roff_res() was called.
2016: */
2017: sz = (int)(p - pp);
2018: res = mandoc_realloc(res, ssz + sz + 1);
2019: memcpy(res + ssz, pp, sz);
2020: ssz += sz;
2021: }
2022:
2023: res[(int)ssz] = '\0';
2024: return(res);
1.48 schwarze 2025: }
2026:
2027: /*
2028: * Find out whether a line is a macro line or not.
2029: * If it is, adjust the current position and return one; if it isn't,
2030: * return zero and don't change the current position.
2031: * If the control character has been set with `.cc', then let that grain
2032: * precedence.
2033: * This is slighly contrary to groff, where using the non-breaking
2034: * control character when `cc' has been invoked will cause the
2035: * non-breaking macro contents to be printed verbatim.
2036: */
2037: int
2038: roff_getcontrol(const struct roff *r, const char *cp, int *ppos)
2039: {
2040: int pos;
2041:
2042: pos = *ppos;
2043:
2044: if (0 != r->control && cp[pos] == r->control)
2045: pos++;
2046: else if (0 != r->control)
2047: return(0);
2048: else if ('\\' == cp[pos] && '.' == cp[pos + 1])
2049: pos += 2;
2050: else if ('.' == cp[pos] || '\'' == cp[pos])
2051: pos++;
2052: else
2053: return(0);
2054:
2055: while (' ' == cp[pos] || '\t' == cp[pos])
2056: pos++;
2057:
2058: *ppos = pos;
2059: return(1);
1.1 schwarze 2060: }
![[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