Annotation of mandoc/roff.c, Revision 1.199
1.199 ! schwarze 1: /* $Id: roff.c,v 1.198 2014/03/08 04:43:54 schwarze Exp $ */
1.1 kristaps 2: /*
1.175 schwarze 3: * Copyright (c) 2010, 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
1.190 schwarze 4: * Copyright (c) 2010-2014 Ingo Schwarze <schwarze@openbsd.org>
1.1 kristaps 5: *
6: * Permission to use, copy, modify, and distribute this software for any
1.66 kristaps 7: * purpose with or without fee is hereby granted, provided that the above
8: * copyright notice and this permission notice appear in all copies.
1.1 kristaps 9: *
1.106 kristaps 10: * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.66 kristaps 11: * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.106 kristaps 12: * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.66 kristaps 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.
1.1 kristaps 17: */
1.66 kristaps 18: #ifdef HAVE_CONFIG_H
19: #include "config.h"
20: #endif
1.30 kristaps 21:
1.67 kristaps 22: #include <assert.h>
1.85 kristaps 23: #include <ctype.h>
1.178 schwarze 24: #include <stdio.h>
1.1 kristaps 25: #include <stdlib.h>
1.67 kristaps 26: #include <string.h>
1.1 kristaps 27:
1.67 kristaps 28: #include "mandoc.h"
1.109 kristaps 29: #include "libroff.h"
1.94 kristaps 30: #include "libmandoc.h"
1.33 kristaps 31:
1.141 kristaps 32: /* Maximum number of nested if-else conditionals. */
1.82 kristaps 33: #define RSTACK_MAX 128
34:
1.170 schwarze 35: /* Maximum number of string expansions per line, to break infinite loops. */
36: #define EXPAND_LIMIT 1000
37:
1.67 kristaps 38: enum rofft {
1.103 kristaps 39: ROFF_ad,
1.80 kristaps 40: ROFF_am,
41: ROFF_ami,
42: ROFF_am1,
1.193 schwarze 43: ROFF_as,
1.174 kristaps 44: ROFF_cc,
1.194 schwarze 45: ROFF_ce,
1.80 kristaps 46: ROFF_de,
47: ROFF_dei,
48: ROFF_de1,
1.83 schwarze 49: ROFF_ds,
1.82 kristaps 50: ROFF_el,
1.185 schwarze 51: ROFF_fam,
1.186 schwarze 52: ROFF_hw,
1.103 kristaps 53: ROFF_hy,
1.82 kristaps 54: ROFF_ie,
1.75 kristaps 55: ROFF_if,
1.76 kristaps 56: ROFF_ig,
1.123 schwarze 57: ROFF_it,
1.103 kristaps 58: ROFF_ne,
59: ROFF_nh,
1.104 kristaps 60: ROFF_nr,
1.124 schwarze 61: ROFF_ns,
62: ROFF_ps,
1.83 schwarze 63: ROFF_rm,
1.105 kristaps 64: ROFF_so,
1.124 schwarze 65: ROFF_ta,
1.83 schwarze 66: ROFF_tr,
1.175 schwarze 67: ROFF_Dd,
68: ROFF_TH,
1.109 kristaps 69: ROFF_TS,
70: ROFF_TE,
1.112 kristaps 71: ROFF_T_,
1.125 kristaps 72: ROFF_EQ,
73: ROFF_EN,
1.76 kristaps 74: ROFF_cblock,
1.106 kristaps 75: ROFF_USERDEF,
1.67 kristaps 76: ROFF_MAX
77: };
78:
1.147 kristaps 79: /*
1.167 kristaps 80: * An incredibly-simple string buffer.
81: */
1.94 kristaps 82: struct roffstr {
1.167 kristaps 83: char *p; /* nil-terminated buffer */
84: size_t sz; /* saved strlen(p) */
1.166 kristaps 85: };
86:
87: /*
1.167 kristaps 88: * A key-value roffstr pair as part of a singly-linked list.
1.166 kristaps 89: */
90: struct roffkv {
91: struct roffstr key;
92: struct roffstr val;
93: struct roffkv *next; /* next in list */
1.94 kristaps 94: };
95:
1.180 schwarze 96: /*
97: * A single number register as part of a singly-linked list.
98: */
99: struct roffreg {
100: struct roffstr key;
1.181 schwarze 101: int val;
1.180 schwarze 102: struct roffreg *next;
103: };
104:
1.67 kristaps 105: struct roff {
1.128 kristaps 106: struct mparse *parse; /* parse point */
1.199 ! schwarze 107: int options; /* parse options */
1.67 kristaps 108: struct roffnode *last; /* leaf of stack */
1.198 schwarze 109: int rstack[RSTACK_MAX]; /* stack of !`ie' rules */
1.174 kristaps 110: char control; /* control character */
1.82 kristaps 111: int rstackpos; /* position in rstack */
1.180 schwarze 112: struct roffreg *regtab; /* number registers */
1.166 kristaps 113: struct roffkv *strtab; /* user-defined strings & macros */
1.167 kristaps 114: struct roffkv *xmbtab; /* multi-byte trans table (`tr') */
115: struct roffstr *xtab; /* single-byte trans table (`tr') */
1.106 kristaps 116: const char *current_string; /* value of last called user macro */
1.118 kristaps 117: struct tbl_node *first_tbl; /* first table parsed */
118: struct tbl_node *last_tbl; /* last table parsed */
119: struct tbl_node *tbl; /* current table being parsed */
1.125 kristaps 120: struct eqn_node *last_eqn; /* last equation parsed */
121: struct eqn_node *first_eqn; /* first equation parsed */
122: struct eqn_node *eqn; /* current equation being parsed */
1.79 kristaps 123: };
124:
1.67 kristaps 125: struct roffnode {
126: enum rofft tok; /* type of node */
127: struct roffnode *parent; /* up one in stack */
128: int line; /* parse line */
129: int col; /* parse col */
1.106 kristaps 130: char *name; /* node name, e.g. macro name */
1.79 kristaps 131: char *end; /* end-rules: custom token */
132: int endspan; /* end-rules: next-line or infty */
1.198 schwarze 133: int rule; /* current evaluation rule */
1.67 kristaps 134: };
135:
136: #define ROFF_ARGS struct roff *r, /* parse ctx */ \
1.72 kristaps 137: enum rofft tok, /* tok of macro */ \
1.67 kristaps 138: char **bufp, /* input buffer */ \
139: size_t *szp, /* size of input buffer */ \
140: int ln, /* parse line */ \
1.75 kristaps 141: int ppos, /* original pos in buffer */ \
142: int pos, /* current pos in buffer */ \
1.74 kristaps 143: int *offs /* reset offset of buffer data */
1.67 kristaps 144:
145: typedef enum rofferr (*roffproc)(ROFF_ARGS);
146:
147: struct roffmac {
148: const char *name; /* macro name */
1.79 kristaps 149: roffproc proc; /* process new macro */
150: roffproc text; /* process as child text of macro */
151: roffproc sub; /* process as child of macro */
152: int flags;
153: #define ROFFMAC_STRUCT (1 << 0) /* always interpret */
1.85 kristaps 154: struct roffmac *next;
1.67 kristaps 155: };
156:
1.141 kristaps 157: struct predef {
158: const char *name; /* predefined input name */
159: const char *str; /* replacement symbol */
160: };
161:
162: #define PREDEF(__name, __str) \
163: { (__name), (__str) },
164:
1.155 kristaps 165: static enum rofft roffhash_find(const char *, size_t);
166: static void roffhash_init(void);
167: static void roffnode_cleanscope(struct roff *);
168: static void roffnode_pop(struct roff *);
169: static void roffnode_push(struct roff *, enum rofft,
170: const char *, int, int);
1.80 kristaps 171: static enum rofferr roff_block(ROFF_ARGS);
172: static enum rofferr roff_block_text(ROFF_ARGS);
173: static enum rofferr roff_block_sub(ROFF_ARGS);
174: static enum rofferr roff_cblock(ROFF_ARGS);
1.174 kristaps 175: static enum rofferr roff_cc(ROFF_ARGS);
1.195 schwarze 176: static void roff_ccond(struct roff *, int, int);
1.82 kristaps 177: static enum rofferr roff_cond(ROFF_ARGS);
178: static enum rofferr roff_cond_text(ROFF_ARGS);
179: static enum rofferr roff_cond_sub(ROFF_ARGS);
1.92 schwarze 180: static enum rofferr roff_ds(ROFF_ARGS);
1.198 schwarze 181: static int roff_evalcond(const char *, int *);
182: static int roff_evalstrcond(const char *, int *);
1.155 kristaps 183: static void roff_free1(struct roff *);
1.180 schwarze 184: static void roff_freereg(struct roffreg *);
1.167 kristaps 185: static void roff_freestr(struct roffkv *);
1.121 schwarze 186: static char *roff_getname(struct roff *, char **, int, int);
1.184 schwarze 187: static int roff_getnum(const char *, int *, int *);
188: static int roff_getop(const char *, int *, char *);
1.181 schwarze 189: static int roff_getregn(const struct roff *,
190: const char *, size_t);
1.192 schwarze 191: static int roff_getregro(const char *name);
1.94 kristaps 192: static const char *roff_getstrn(const struct roff *,
193: const char *, size_t);
1.178 schwarze 194: static enum rofferr roff_it(ROFF_ARGS);
1.103 kristaps 195: static enum rofferr roff_line_ignore(ROFF_ARGS);
1.89 kristaps 196: static enum rofferr roff_nr(ROFF_ARGS);
1.169 schwarze 197: static void roff_openeqn(struct roff *, const char *,
1.156 kristaps 198: int, int, const char *);
1.155 kristaps 199: static enum rofft roff_parse(struct roff *, const char *, int *);
1.178 schwarze 200: static enum rofferr roff_parsetext(char **, size_t *, int, int *);
1.172 schwarze 201: static enum rofferr roff_res(struct roff *,
1.142 kristaps 202: char **, size_t *, int, int);
1.122 schwarze 203: static enum rofferr roff_rm(ROFF_ARGS);
1.94 kristaps 204: static void roff_setstr(struct roff *,
1.106 kristaps 205: const char *, const char *, int);
1.166 kristaps 206: static void roff_setstrn(struct roffkv **, const char *,
1.164 kristaps 207: size_t, const char *, size_t, int);
1.105 kristaps 208: static enum rofferr roff_so(ROFF_ARGS);
1.164 kristaps 209: static enum rofferr roff_tr(ROFF_ARGS);
1.175 schwarze 210: static enum rofferr roff_Dd(ROFF_ARGS);
211: static enum rofferr roff_TH(ROFF_ARGS);
1.109 kristaps 212: static enum rofferr roff_TE(ROFF_ARGS);
213: static enum rofferr roff_TS(ROFF_ARGS);
1.125 kristaps 214: static enum rofferr roff_EQ(ROFF_ARGS);
215: static enum rofferr roff_EN(ROFF_ARGS);
1.112 kristaps 216: static enum rofferr roff_T_(ROFF_ARGS);
1.106 kristaps 217: static enum rofferr roff_userdef(ROFF_ARGS);
1.67 kristaps 218:
1.155 kristaps 219: /* See roffhash_find() */
1.85 kristaps 220:
221: #define ASCII_HI 126
222: #define ASCII_LO 33
223: #define HASHWIDTH (ASCII_HI - ASCII_LO + 1)
224:
225: static struct roffmac *hash[HASHWIDTH];
226:
227: static struct roffmac roffs[ROFF_MAX] = {
1.103 kristaps 228: { "ad", roff_line_ignore, NULL, NULL, 0, NULL },
1.85 kristaps 229: { "am", roff_block, roff_block_text, roff_block_sub, 0, NULL },
230: { "ami", roff_block, roff_block_text, roff_block_sub, 0, NULL },
231: { "am1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.193 schwarze 232: { "as", roff_ds, NULL, NULL, 0, NULL },
1.174 kristaps 233: { "cc", roff_cc, NULL, NULL, 0, NULL },
1.194 schwarze 234: { "ce", roff_line_ignore, NULL, NULL, 0, NULL },
1.85 kristaps 235: { "de", roff_block, roff_block_text, roff_block_sub, 0, NULL },
236: { "dei", roff_block, roff_block_text, roff_block_sub, 0, NULL },
237: { "de1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.92 schwarze 238: { "ds", roff_ds, NULL, NULL, 0, NULL },
1.85 kristaps 239: { "el", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
1.185 schwarze 240: { "fam", roff_line_ignore, NULL, NULL, 0, NULL },
1.186 schwarze 241: { "hw", roff_line_ignore, NULL, NULL, 0, NULL },
1.103 kristaps 242: { "hy", roff_line_ignore, NULL, NULL, 0, NULL },
1.85 kristaps 243: { "ie", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
244: { "if", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
245: { "ig", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.178 schwarze 246: { "it", roff_it, NULL, NULL, 0, NULL },
1.103 kristaps 247: { "ne", roff_line_ignore, NULL, NULL, 0, NULL },
248: { "nh", roff_line_ignore, NULL, NULL, 0, NULL },
1.104 kristaps 249: { "nr", roff_nr, NULL, NULL, 0, NULL },
1.124 schwarze 250: { "ns", roff_line_ignore, NULL, NULL, 0, NULL },
251: { "ps", roff_line_ignore, NULL, NULL, 0, NULL },
1.122 schwarze 252: { "rm", roff_rm, NULL, NULL, 0, NULL },
1.105 kristaps 253: { "so", roff_so, NULL, NULL, 0, NULL },
1.124 schwarze 254: { "ta", roff_line_ignore, NULL, NULL, 0, NULL },
1.164 kristaps 255: { "tr", roff_tr, NULL, NULL, 0, NULL },
1.175 schwarze 256: { "Dd", roff_Dd, NULL, NULL, 0, NULL },
257: { "TH", roff_TH, NULL, NULL, 0, NULL },
1.109 kristaps 258: { "TS", roff_TS, NULL, NULL, 0, NULL },
259: { "TE", roff_TE, NULL, NULL, 0, NULL },
1.112 kristaps 260: { "T&", roff_T_, NULL, NULL, 0, NULL },
1.125 kristaps 261: { "EQ", roff_EQ, NULL, NULL, 0, NULL },
262: { "EN", roff_EN, NULL, NULL, 0, NULL },
1.85 kristaps 263: { ".", roff_cblock, NULL, NULL, 0, NULL },
1.106 kristaps 264: { NULL, roff_userdef, NULL, NULL, 0, NULL },
1.67 kristaps 265: };
266:
1.175 schwarze 267: const char *const __mdoc_reserved[] = {
268: "Ac", "Ad", "An", "Ao", "Ap", "Aq", "Ar", "At",
269: "Bc", "Bd", "Bf", "Bk", "Bl", "Bo", "Bq",
270: "Brc", "Bro", "Brq", "Bsx", "Bt", "Bx",
271: "Cd", "Cm", "Db", "Dc", "Dd", "Dl", "Do", "Dq",
272: "Ds", "Dt", "Dv", "Dx", "D1",
273: "Ec", "Ed", "Ef", "Ek", "El", "Em", "em",
274: "En", "Eo", "Eq", "Er", "Es", "Ev", "Ex",
275: "Fa", "Fc", "Fd", "Fl", "Fn", "Fo", "Fr", "Ft", "Fx",
276: "Hf", "Ic", "In", "It", "Lb", "Li", "Lk", "Lp", "LP",
277: "Me", "Ms", "Mt", "Nd", "Nm", "No", "Ns", "Nx",
278: "Oc", "Oo", "Op", "Os", "Ot", "Ox",
279: "Pa", "Pc", "Pf", "Po", "Pp", "PP", "pp", "Pq",
280: "Qc", "Ql", "Qo", "Qq", "Or", "Rd", "Re", "Rs", "Rv",
281: "Sc", "Sf", "Sh", "SH", "Sm", "So", "Sq",
282: "Ss", "St", "Sx", "Sy",
283: "Ta", "Tn", "Ud", "Ux", "Va", "Vt", "Xc", "Xo", "Xr",
284: "%A", "%B", "%D", "%I", "%J", "%N", "%O",
285: "%P", "%Q", "%R", "%T", "%U", "%V",
286: NULL
287: };
288:
289: const char *const __man_reserved[] = {
290: "AT", "B", "BI", "BR", "BT", "DE", "DS", "DT",
291: "EE", "EN", "EQ", "EX", "HF", "HP", "I", "IB", "IP", "IR",
292: "LP", "ME", "MT", "OP", "P", "PD", "PP", "PT",
293: "R", "RB", "RE", "RI", "RS", "SB", "SH", "SM", "SS", "SY",
294: "TE", "TH", "TP", "TQ", "TS", "T&", "UC", "UE", "UR", "YS",
295: NULL
296: };
297:
1.141 kristaps 298: /* Array of injected predefined strings. */
299: #define PREDEFS_MAX 38
300: static const struct predef predefs[PREDEFS_MAX] = {
301: #include "predefs.in"
302: };
303:
1.155 kristaps 304: /* See roffhash_find() */
1.85 kristaps 305: #define ROFF_HASH(p) (p[0] - ASCII_LO)
306:
1.178 schwarze 307: static int roffit_lines; /* number of lines to delay */
308: static char *roffit_macro; /* nil-terminated macro line */
309:
1.85 kristaps 310: static void
1.155 kristaps 311: roffhash_init(void)
1.85 kristaps 312: {
313: struct roffmac *n;
314: int buc, i;
315:
1.106 kristaps 316: for (i = 0; i < (int)ROFF_USERDEF; i++) {
1.85 kristaps 317: assert(roffs[i].name[0] >= ASCII_LO);
318: assert(roffs[i].name[0] <= ASCII_HI);
319:
320: buc = ROFF_HASH(roffs[i].name);
321:
322: if (NULL != (n = hash[buc])) {
323: for ( ; n->next; n = n->next)
324: /* Do nothing. */ ;
325: n->next = &roffs[i];
326: } else
327: hash[buc] = &roffs[i];
328: }
329: }
330:
1.67 kristaps 331: /*
332: * Look up a roff token by its name. Returns ROFF_MAX if no macro by
333: * the nil-terminated string name could be found.
334: */
335: static enum rofft
1.155 kristaps 336: roffhash_find(const char *p, size_t s)
1.67 kristaps 337: {
1.85 kristaps 338: int buc;
339: struct roffmac *n;
1.67 kristaps 340:
1.85 kristaps 341: /*
342: * libroff has an extremely simple hashtable, for the time
343: * being, which simply keys on the first character, which must
344: * be printable, then walks a chain. It works well enough until
345: * optimised.
346: */
347:
348: if (p[0] < ASCII_LO || p[0] > ASCII_HI)
349: return(ROFF_MAX);
350:
351: buc = ROFF_HASH(p);
352:
353: if (NULL == (n = hash[buc]))
354: return(ROFF_MAX);
355: for ( ; n; n = n->next)
1.106 kristaps 356: if (0 == strncmp(n->name, p, s) && '\0' == n->name[(int)s])
1.85 kristaps 357: return((enum rofft)(n - roffs));
1.67 kristaps 358:
359: return(ROFF_MAX);
360: }
361:
362:
363: /*
364: * Pop the current node off of the stack of roff instructions currently
365: * pending.
366: */
367: static void
368: roffnode_pop(struct roff *r)
369: {
370: struct roffnode *p;
371:
1.75 kristaps 372: assert(r->last);
373: p = r->last;
1.82 kristaps 374:
1.75 kristaps 375: r->last = r->last->parent;
1.106 kristaps 376: free(p->name);
377: free(p->end);
1.67 kristaps 378: free(p);
379: }
380:
381:
382: /*
383: * Push a roff node onto the instruction stack. This must later be
384: * removed with roffnode_pop().
385: */
1.98 schwarze 386: static void
1.106 kristaps 387: roffnode_push(struct roff *r, enum rofft tok, const char *name,
388: int line, int col)
1.67 kristaps 389: {
390: struct roffnode *p;
391:
1.98 schwarze 392: p = mandoc_calloc(1, sizeof(struct roffnode));
1.67 kristaps 393: p->tok = tok;
1.106 kristaps 394: if (name)
395: p->name = mandoc_strdup(name);
1.67 kristaps 396: p->parent = r->last;
397: p->line = line;
398: p->col = col;
1.198 schwarze 399: p->rule = p->parent ? p->parent->rule : 0;
1.67 kristaps 400:
401: r->last = p;
402: }
403:
404:
405: static void
406: roff_free1(struct roff *r)
407: {
1.176 schwarze 408: struct tbl_node *tbl;
1.125 kristaps 409: struct eqn_node *e;
1.167 kristaps 410: int i;
1.67 kristaps 411:
1.176 schwarze 412: while (NULL != (tbl = r->first_tbl)) {
413: r->first_tbl = tbl->next;
414: tbl_free(tbl);
1.109 kristaps 415: }
416:
1.113 kristaps 417: r->first_tbl = r->last_tbl = r->tbl = NULL;
418:
1.125 kristaps 419: while (NULL != (e = r->first_eqn)) {
420: r->first_eqn = e->next;
421: eqn_free(e);
422: }
423:
424: r->first_eqn = r->last_eqn = r->eqn = NULL;
425:
1.67 kristaps 426: while (r->last)
427: roffnode_pop(r);
1.109 kristaps 428:
1.167 kristaps 429: roff_freestr(r->strtab);
430: roff_freestr(r->xmbtab);
431:
432: r->strtab = r->xmbtab = NULL;
433:
1.180 schwarze 434: roff_freereg(r->regtab);
435:
436: r->regtab = NULL;
437:
1.167 kristaps 438: if (r->xtab)
439: for (i = 0; i < 128; i++)
440: free(r->xtab[i].p);
441:
442: free(r->xtab);
443: r->xtab = NULL;
1.67 kristaps 444: }
445:
446: void
447: roff_reset(struct roff *r)
448: {
449:
450: roff_free1(r);
1.174 kristaps 451: r->control = 0;
1.67 kristaps 452: }
453:
454:
455: void
456: roff_free(struct roff *r)
457: {
458:
459: roff_free1(r);
460: free(r);
461: }
462:
463:
464: struct roff *
1.199 ! schwarze 465: roff_alloc(struct mparse *parse, int options)
1.67 kristaps 466: {
467: struct roff *r;
468:
1.98 schwarze 469: r = mandoc_calloc(1, sizeof(struct roff));
1.128 kristaps 470: r->parse = parse;
1.199 ! schwarze 471: r->options = options;
1.82 kristaps 472: r->rstackpos = -1;
1.85 kristaps 473:
1.155 kristaps 474: roffhash_init();
1.141 kristaps 475:
1.67 kristaps 476: return(r);
477: }
478:
1.94 kristaps 479: /*
1.181 schwarze 480: * In the current line, expand user-defined strings ("\*")
481: * and references to number registers ("\n").
482: * Also check the syntax of other escape sequences.
1.154 kristaps 483: */
1.172 schwarze 484: static enum rofferr
1.142 kristaps 485: roff_res(struct roff *r, char **bufp, size_t *szp, int ln, int pos)
1.94 kristaps 486: {
1.181 schwarze 487: char ubuf[12]; /* buffer to print the number */
1.108 schwarze 488: const char *stesc; /* start of an escape sequence ('\\') */
489: const char *stnam; /* start of the name, after "[(*" */
490: const char *cp; /* end of the name, e.g. before ']' */
491: const char *res; /* the string to be substituted */
1.181 schwarze 492: char *nbuf; /* new buffer to copy bufp to */
493: size_t nsz; /* size of the new buffer */
494: size_t maxl; /* expected length of the escape name */
495: size_t naml; /* actual length of the escape name */
496: int expand_count; /* to avoid infinite loops */
1.94 kristaps 497:
1.170 schwarze 498: expand_count = 0;
499:
1.154 kristaps 500: again:
1.108 schwarze 501: cp = *bufp + pos;
502: while (NULL != (cp = strchr(cp, '\\'))) {
503: stesc = cp++;
504:
505: /*
1.181 schwarze 506: * The second character must be an asterisk or an n.
1.108 schwarze 507: * If it isn't, skip it anyway: It is escaped,
508: * so it can't start another escape sequence.
509: */
510:
511: if ('\0' == *cp)
1.172 schwarze 512: return(ROFF_CONT);
1.152 kristaps 513:
1.181 schwarze 514: switch (*cp) {
515: case ('*'):
516: res = NULL;
517: break;
518: case ('n'):
519: res = ubuf;
520: break;
521: default:
522: if (ESCAPE_ERROR != mandoc_escape(&cp, NULL, NULL))
1.152 kristaps 523: continue;
1.153 kristaps 524: mandoc_msg
525: (MANDOCERR_BADESCAPE, r->parse,
526: ln, (int)(stesc - *bufp), NULL);
1.172 schwarze 527: return(ROFF_CONT);
1.152 kristaps 528: }
529:
530: cp++;
1.108 schwarze 531:
532: /*
533: * The third character decides the length
1.181 schwarze 534: * of the name of the string or register.
1.108 schwarze 535: * Save a pointer to the name.
536: */
537:
1.94 kristaps 538: switch (*cp) {
1.108 schwarze 539: case ('\0'):
1.172 schwarze 540: return(ROFF_CONT);
1.94 kristaps 541: case ('('):
542: cp++;
543: maxl = 2;
544: break;
545: case ('['):
546: cp++;
547: maxl = 0;
548: break;
549: default:
550: maxl = 1;
551: break;
552: }
1.108 schwarze 553: stnam = cp;
1.94 kristaps 554:
1.108 schwarze 555: /* Advance to the end of the name. */
1.94 kristaps 556:
1.181 schwarze 557: for (naml = 0; 0 == maxl || naml < maxl; naml++, cp++) {
1.153 kristaps 558: if ('\0' == *cp) {
559: mandoc_msg
560: (MANDOCERR_BADESCAPE,
561: r->parse, ln,
562: (int)(stesc - *bufp), NULL);
1.172 schwarze 563: return(ROFF_CONT);
1.153 kristaps 564: }
1.94 kristaps 565: if (0 == maxl && ']' == *cp)
566: break;
567: }
568:
1.108 schwarze 569: /*
570: * Retrieve the replacement string; if it is
571: * undefined, resume searching for escapes.
572: */
573:
1.181 schwarze 574: if (NULL == res)
575: res = roff_getstrn(r, stnam, naml);
576: else
577: snprintf(ubuf, sizeof(ubuf), "%d",
578: roff_getregn(r, stnam, naml));
1.94 kristaps 579:
580: if (NULL == res) {
1.153 kristaps 581: mandoc_msg
582: (MANDOCERR_BADESCAPE, r->parse,
583: ln, (int)(stesc - *bufp), NULL);
1.142 kristaps 584: res = "";
1.94 kristaps 585: }
586:
1.108 schwarze 587: /* Replace the escape sequence by the string. */
588:
1.161 kristaps 589: pos = stesc - *bufp;
1.154 kristaps 590:
1.94 kristaps 591: nsz = *szp + strlen(res) + 1;
1.181 schwarze 592: nbuf = mandoc_malloc(nsz);
1.94 kristaps 593:
1.181 schwarze 594: strlcpy(nbuf, *bufp, (size_t)(stesc - *bufp + 1));
595: strlcat(nbuf, res, nsz);
596: strlcat(nbuf, cp + (maxl ? 0 : 1), nsz);
1.94 kristaps 597:
598: free(*bufp);
599:
1.181 schwarze 600: *bufp = nbuf;
1.94 kristaps 601: *szp = nsz;
1.170 schwarze 602:
603: if (EXPAND_LIMIT >= ++expand_count)
604: goto again;
605:
606: /* Just leave the string unexpanded. */
607: mandoc_msg(MANDOCERR_ROFFLOOP, r->parse, ln, pos, NULL);
1.172 schwarze 608: return(ROFF_IGN);
1.154 kristaps 609: }
1.172 schwarze 610: return(ROFF_CONT);
1.154 kristaps 611: }
612:
613: /*
1.178 schwarze 614: * Process text streams:
615: * Convert all breakable hyphens into ASCII_HYPH.
616: * Decrement and spring input line trap.
1.154 kristaps 617: */
618: static enum rofferr
1.178 schwarze 619: roff_parsetext(char **bufp, size_t *szp, int pos, int *offs)
1.154 kristaps 620: {
621: size_t sz;
622: const char *start;
1.178 schwarze 623: char *p;
624: int isz;
1.154 kristaps 625: enum mandoc_esc esc;
626:
1.178 schwarze 627: start = p = *bufp + pos;
1.154 kristaps 628:
629: while ('\0' != *p) {
630: sz = strcspn(p, "-\\");
631: p += sz;
632:
1.159 kristaps 633: if ('\0' == *p)
634: break;
635:
1.154 kristaps 636: if ('\\' == *p) {
637: /* Skip over escapes. */
638: p++;
1.189 schwarze 639: esc = mandoc_escape((const char **)&p, NULL, NULL);
1.154 kristaps 640: if (ESCAPE_ERROR == esc)
641: break;
1.155 kristaps 642: continue;
1.159 kristaps 643: } else if (p == start) {
1.158 kristaps 644: p++;
1.155 kristaps 645: continue;
1.158 kristaps 646: }
1.155 kristaps 647:
1.171 schwarze 648: if (isalpha((unsigned char)p[-1]) &&
649: isalpha((unsigned char)p[1]))
1.155 kristaps 650: *p = ASCII_HYPH;
651: p++;
1.94 kristaps 652: }
653:
1.178 schwarze 654: /* Spring the input line trap. */
655: if (1 == roffit_lines) {
656: isz = asprintf(&p, "%s\n.%s", *bufp, roffit_macro);
657: if (-1 == isz) {
658: perror(NULL);
659: exit((int)MANDOCLEVEL_SYSERR);
660: }
661: free(*bufp);
662: *bufp = p;
663: *szp = isz + 1;
664: *offs = 0;
665: free(roffit_macro);
666: roffit_lines = 0;
667: return(ROFF_REPARSE);
668: } else if (1 < roffit_lines)
669: --roffit_lines;
1.154 kristaps 670: return(ROFF_CONT);
1.94 kristaps 671: }
672:
1.67 kristaps 673: enum rofferr
1.90 kristaps 674: roff_parseln(struct roff *r, int ln, char **bufp,
675: size_t *szp, int pos, int *offs)
1.67 kristaps 676: {
677: enum rofft t;
1.109 kristaps 678: enum rofferr e;
1.130 kristaps 679: int ppos, ctl;
1.79 kristaps 680:
681: /*
1.94 kristaps 682: * Run the reserved-word filter only if we have some reserved
683: * words to fill in.
684: */
685:
1.172 schwarze 686: e = roff_res(r, bufp, szp, ln, pos);
687: if (ROFF_IGN == e)
688: return(e);
689: assert(ROFF_CONT == e);
1.94 kristaps 690:
1.130 kristaps 691: ppos = pos;
1.174 kristaps 692: ctl = roff_getcontrol(r, *bufp, &pos);
1.130 kristaps 693:
1.94 kristaps 694: /*
1.79 kristaps 695: * First, if a scope is open and we're not a macro, pass the
696: * text through the macro's filter. If a scope isn't open and
697: * we're not a macro, just let it through.
1.125 kristaps 698: * Finally, if there's an equation scope open, divert it into it
699: * no matter our state.
1.79 kristaps 700: */
1.74 kristaps 701:
1.130 kristaps 702: if (r->last && ! ctl) {
1.78 kristaps 703: t = r->last->tok;
704: assert(roffs[t].text);
1.109 kristaps 705: e = (*roffs[t].text)
706: (r, t, bufp, szp, ln, pos, pos, offs);
707: assert(ROFF_IGN == e || ROFF_CONT == e);
1.125 kristaps 708: if (ROFF_CONT != e)
709: return(e);
1.182 schwarze 710: }
711: if (r->eqn)
712: return(eqn_read(&r->eqn, ln, *bufp, ppos, offs));
713: if ( ! ctl) {
1.125 kristaps 714: if (r->tbl)
1.130 kristaps 715: return(tbl_read(r->tbl, ln, *bufp, pos));
1.178 schwarze 716: return(roff_parsetext(bufp, szp, pos, offs));
1.182 schwarze 717: }
1.67 kristaps 718:
1.79 kristaps 719: /*
720: * If a scope is open, go to the child handler for that macro,
721: * as it may want to preprocess before doing anything with it.
1.125 kristaps 722: * Don't do so if an equation is open.
1.79 kristaps 723: */
1.78 kristaps 724:
1.79 kristaps 725: if (r->last) {
726: t = r->last->tok;
727: assert(roffs[t].sub);
728: return((*roffs[t].sub)
1.90 kristaps 729: (r, t, bufp, szp,
1.130 kristaps 730: ln, ppos, pos, offs));
1.79 kristaps 731: }
1.78 kristaps 732:
1.79 kristaps 733: /*
734: * Lastly, as we've no scope open, try to look up and execute
735: * the new macro. If no macro is found, simply return and let
736: * the compilers handle it.
737: */
1.67 kristaps 738:
1.106 kristaps 739: if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos)))
1.79 kristaps 740: return(ROFF_CONT);
1.67 kristaps 741:
1.75 kristaps 742: assert(roffs[t].proc);
1.78 kristaps 743: return((*roffs[t].proc)
1.90 kristaps 744: (r, t, bufp, szp,
745: ln, ppos, pos, offs));
1.74 kristaps 746: }
747:
748:
1.117 kristaps 749: void
1.74 kristaps 750: roff_endparse(struct roff *r)
751: {
752:
1.110 kristaps 753: if (r->last)
1.128 kristaps 754: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.109 kristaps 755: r->last->line, r->last->col, NULL);
1.117 kristaps 756:
1.125 kristaps 757: if (r->eqn) {
1.128 kristaps 758: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.148 kristaps 759: r->eqn->eqn.ln, r->eqn->eqn.pos, NULL);
1.151 kristaps 760: eqn_end(&r->eqn);
1.125 kristaps 761: }
762:
1.117 kristaps 763: if (r->tbl) {
1.128 kristaps 764: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.117 kristaps 765: r->tbl->line, r->tbl->pos, NULL);
1.151 kristaps 766: tbl_end(&r->tbl);
1.117 kristaps 767: }
1.67 kristaps 768: }
769:
770: /*
771: * Parse a roff node's type from the input buffer. This must be in the
772: * form of ".foo xxx" in the usual way.
773: */
774: static enum rofft
1.106 kristaps 775: roff_parse(struct roff *r, const char *buf, int *pos)
1.67 kristaps 776: {
1.106 kristaps 777: const char *mac;
778: size_t maclen;
1.67 kristaps 779: enum rofft t;
780:
1.144 kristaps 781: if ('\0' == buf[*pos] || '"' == buf[*pos] ||
782: '\t' == buf[*pos] || ' ' == buf[*pos])
1.67 kristaps 783: return(ROFF_MAX);
784:
1.195 schwarze 785: /* We stop the macro parse at an escape, tab, space, or nil. */
1.144 kristaps 786:
1.106 kristaps 787: mac = buf + *pos;
1.195 schwarze 788: maclen = strcspn(mac, " \\\t\0");
1.67 kristaps 789:
1.106 kristaps 790: t = (r->current_string = roff_getstrn(r, mac, maclen))
1.155 kristaps 791: ? ROFF_USERDEF : roffhash_find(mac, maclen);
1.67 kristaps 792:
1.127 kristaps 793: *pos += (int)maclen;
1.130 kristaps 794:
1.67 kristaps 795: while (buf[*pos] && ' ' == buf[*pos])
796: (*pos)++;
797:
798: return(t);
799: }
800:
801: /* ARGSUSED */
802: static enum rofferr
1.76 kristaps 803: roff_cblock(ROFF_ARGS)
1.67 kristaps 804: {
805:
1.79 kristaps 806: /*
807: * A block-close `..' should only be invoked as a child of an
808: * ignore macro, otherwise raise a warning and just ignore it.
809: */
810:
1.76 kristaps 811: if (NULL == r->last) {
1.128 kristaps 812: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.76 kristaps 813: return(ROFF_IGN);
814: }
1.67 kristaps 815:
1.81 kristaps 816: switch (r->last->tok) {
817: case (ROFF_am):
818: /* FALLTHROUGH */
819: case (ROFF_ami):
820: /* FALLTHROUGH */
821: case (ROFF_am1):
822: /* FALLTHROUGH */
823: case (ROFF_de):
1.108 schwarze 824: /* ROFF_de1 is remapped to ROFF_de in roff_block(). */
1.81 kristaps 825: /* FALLTHROUGH */
826: case (ROFF_dei):
827: /* FALLTHROUGH */
828: case (ROFF_ig):
829: break;
830: default:
1.128 kristaps 831: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.67 kristaps 832: return(ROFF_IGN);
1.76 kristaps 833: }
1.67 kristaps 834:
1.76 kristaps 835: if ((*bufp)[pos])
1.128 kristaps 836: mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.71 kristaps 837:
838: roffnode_pop(r);
1.76 kristaps 839: roffnode_cleanscope(r);
840: return(ROFF_IGN);
1.71 kristaps 841:
1.67 kristaps 842: }
843:
844:
1.76 kristaps 845: static void
846: roffnode_cleanscope(struct roff *r)
1.67 kristaps 847: {
848:
1.76 kristaps 849: while (r->last) {
1.173 schwarze 850: if (--r->last->endspan != 0)
1.76 kristaps 851: break;
852: roffnode_pop(r);
853: }
1.67 kristaps 854: }
855:
856:
1.195 schwarze 857: static void
858: roff_ccond(struct roff *r, int ln, int ppos)
1.74 kristaps 859: {
860:
1.76 kristaps 861: if (NULL == r->last) {
1.128 kristaps 862: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.195 schwarze 863: return;
1.76 kristaps 864: }
865:
1.82 kristaps 866: switch (r->last->tok) {
867: case (ROFF_el):
868: /* FALLTHROUGH */
869: case (ROFF_ie):
870: /* FALLTHROUGH */
871: case (ROFF_if):
872: break;
873: default:
1.128 kristaps 874: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.195 schwarze 875: return;
1.75 kristaps 876: }
877:
1.76 kristaps 878: if (r->last->endspan > -1) {
1.128 kristaps 879: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.195 schwarze 880: return;
1.76 kristaps 881: }
882:
1.75 kristaps 883: roffnode_pop(r);
1.76 kristaps 884: roffnode_cleanscope(r);
1.195 schwarze 885: return;
1.76 kristaps 886: }
887:
1.75 kristaps 888:
1.76 kristaps 889: /* ARGSUSED */
890: static enum rofferr
1.80 kristaps 891: roff_block(ROFF_ARGS)
1.76 kristaps 892: {
1.78 kristaps 893: int sv;
894: size_t sz;
1.106 kristaps 895: char *name;
896:
897: name = NULL;
1.76 kristaps 898:
1.106 kristaps 899: if (ROFF_ig != tok) {
900: if ('\0' == (*bufp)[pos]) {
1.128 kristaps 901: mandoc_msg(MANDOCERR_NOARGS, r->parse, ln, ppos, NULL);
1.106 kristaps 902: return(ROFF_IGN);
903: }
1.107 kristaps 904:
905: /*
906: * Re-write `de1', since we don't really care about
907: * groff's strange compatibility mode, into `de'.
908: */
909:
1.106 kristaps 910: if (ROFF_de1 == tok)
911: tok = ROFF_de;
912: if (ROFF_de == tok)
913: name = *bufp + pos;
914: else
1.128 kristaps 915: mandoc_msg(MANDOCERR_REQUEST, r->parse, ln, ppos,
1.106 kristaps 916: roffs[tok].name);
1.107 kristaps 917:
1.131 schwarze 918: while ((*bufp)[pos] && ! isspace((unsigned char)(*bufp)[pos]))
1.80 kristaps 919: pos++;
1.107 kristaps 920:
1.131 schwarze 921: while (isspace((unsigned char)(*bufp)[pos]))
1.106 kristaps 922: (*bufp)[pos++] = '\0';
1.80 kristaps 923: }
924:
1.106 kristaps 925: roffnode_push(r, tok, name, ln, ppos);
926:
927: /*
928: * At the beginning of a `de' macro, clear the existing string
929: * with the same name, if there is one. New content will be
1.193 schwarze 930: * appended from roff_block_text() in multiline mode.
1.106 kristaps 931: */
1.107 kristaps 932:
1.106 kristaps 933: if (ROFF_de == tok)
1.108 schwarze 934: roff_setstr(r, name, "", 0);
1.76 kristaps 935:
1.79 kristaps 936: if ('\0' == (*bufp)[pos])
1.78 kristaps 937: return(ROFF_IGN);
938:
1.107 kristaps 939: /* If present, process the custom end-of-line marker. */
940:
1.78 kristaps 941: sv = pos;
1.131 schwarze 942: while ((*bufp)[pos] && ! isspace((unsigned char)(*bufp)[pos]))
1.78 kristaps 943: pos++;
944:
945: /*
946: * Note: groff does NOT like escape characters in the input.
947: * Instead of detecting this, we're just going to let it fly and
948: * to hell with it.
949: */
950:
951: assert(pos > sv);
952: sz = (size_t)(pos - sv);
953:
1.79 kristaps 954: if (1 == sz && '.' == (*bufp)[sv])
955: return(ROFF_IGN);
956:
1.98 schwarze 957: r->last->end = mandoc_malloc(sz + 1);
1.78 kristaps 958:
959: memcpy(r->last->end, *bufp + sv, sz);
960: r->last->end[(int)sz] = '\0';
961:
1.77 kristaps 962: if ((*bufp)[pos])
1.128 kristaps 963: mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.74 kristaps 964:
1.78 kristaps 965: return(ROFF_IGN);
966: }
967:
968:
969: /* ARGSUSED */
970: static enum rofferr
1.80 kristaps 971: roff_block_sub(ROFF_ARGS)
1.79 kristaps 972: {
973: enum rofft t;
974: int i, j;
975:
976: /*
977: * First check whether a custom macro exists at this level. If
978: * it does, then check against it. This is some of groff's
979: * stranger behaviours. If we encountered a custom end-scope
980: * tag and that tag also happens to be a "real" macro, then we
981: * need to try interpreting it again as a real macro. If it's
982: * not, then return ignore. Else continue.
983: */
984:
985: if (r->last->end) {
1.130 kristaps 986: for (i = pos, j = 0; r->last->end[j]; j++, i++)
1.79 kristaps 987: if ((*bufp)[i] != r->last->end[j])
988: break;
989:
990: if ('\0' == r->last->end[j] &&
991: ('\0' == (*bufp)[i] ||
992: ' ' == (*bufp)[i] ||
993: '\t' == (*bufp)[i])) {
994: roffnode_pop(r);
995: roffnode_cleanscope(r);
996:
1.130 kristaps 997: while (' ' == (*bufp)[i] || '\t' == (*bufp)[i])
998: i++;
999:
1000: pos = i;
1.106 kristaps 1001: if (ROFF_MAX != roff_parse(r, *bufp, &pos))
1.79 kristaps 1002: return(ROFF_RERUN);
1003: return(ROFF_IGN);
1004: }
1005: }
1006:
1007: /*
1008: * If we have no custom end-query or lookup failed, then try
1009: * pulling it out of the hashtable.
1010: */
1011:
1.137 schwarze 1012: t = roff_parse(r, *bufp, &pos);
1.79 kristaps 1013:
1.106 kristaps 1014: /*
1015: * Macros other than block-end are only significant
1016: * in `de' blocks; elsewhere, simply throw them away.
1017: */
1018: if (ROFF_cblock != t) {
1019: if (ROFF_de == tok)
1.193 schwarze 1020: roff_setstr(r, r->last->name, *bufp + ppos, 2);
1.79 kristaps 1021: return(ROFF_IGN);
1.106 kristaps 1022: }
1.79 kristaps 1023:
1024: assert(roffs[t].proc);
1.90 kristaps 1025: return((*roffs[t].proc)(r, t, bufp, szp,
1026: ln, ppos, pos, offs));
1.79 kristaps 1027: }
1028:
1029:
1030: /* ARGSUSED */
1031: static enum rofferr
1.80 kristaps 1032: roff_block_text(ROFF_ARGS)
1.78 kristaps 1033: {
1034:
1.106 kristaps 1035: if (ROFF_de == tok)
1.193 schwarze 1036: roff_setstr(r, r->last->name, *bufp + pos, 2);
1.106 kristaps 1037:
1.78 kristaps 1038: return(ROFF_IGN);
1039: }
1040:
1041:
1042: /* ARGSUSED */
1043: static enum rofferr
1.82 kristaps 1044: roff_cond_sub(ROFF_ARGS)
1045: {
1046: enum rofft t;
1.139 kristaps 1047: char *ep;
1.198 schwarze 1048: int rr;
1.82 kristaps 1049:
1050: rr = r->last->rule;
1.139 kristaps 1051: roffnode_cleanscope(r);
1.177 schwarze 1052: t = roff_parse(r, *bufp, &pos);
1.82 kristaps 1053:
1.139 kristaps 1054: /*
1.177 schwarze 1055: * Fully handle known macros when they are structurally
1056: * required or when the conditional evaluated to true.
1.87 kristaps 1057: */
1058:
1.177 schwarze 1059: if ((ROFF_MAX != t) &&
1.198 schwarze 1060: (rr || ROFFMAC_STRUCT & roffs[t].flags)) {
1.177 schwarze 1061: assert(roffs[t].proc);
1062: return((*roffs[t].proc)(r, t, bufp, szp,
1063: ln, ppos, pos, offs));
1064: }
1.144 kristaps 1065:
1.196 schwarze 1066: /*
1067: * If `\}' occurs on a macro line without a preceding macro,
1068: * drop the line completely.
1069: */
1070:
1071: ep = *bufp + pos;
1072: if ('\\' == ep[0] && '}' == ep[1])
1.198 schwarze 1073: rr = 0;
1.196 schwarze 1074:
1.177 schwarze 1075: /* Always check for the closing delimiter `\}'. */
1.144 kristaps 1076:
1.177 schwarze 1077: while (NULL != (ep = strchr(ep, '\\'))) {
1.197 schwarze 1078: if ('}' == *(++ep)) {
1079: *ep = '&';
1080: roff_ccond(r, ln, ep - *bufp - 1);
1081: }
1082: ++ep;
1.177 schwarze 1083: }
1.198 schwarze 1084: return(rr ? ROFF_CONT : ROFF_IGN);
1.82 kristaps 1085: }
1086:
1087: /* ARGSUSED */
1088: static enum rofferr
1089: roff_cond_text(ROFF_ARGS)
1.78 kristaps 1090: {
1.140 kristaps 1091: char *ep;
1.198 schwarze 1092: int rr;
1.82 kristaps 1093:
1094: rr = r->last->rule;
1.140 kristaps 1095: roffnode_cleanscope(r);
1.82 kristaps 1096:
1.197 schwarze 1097: ep = *bufp + pos;
1098: while (NULL != (ep = strchr(ep, '\\'))) {
1099: if ('}' == *(++ep)) {
1100: *ep = '&';
1101: roff_ccond(r, ln, ep - *bufp - 1);
1102: }
1103: ++ep;
1.78 kristaps 1104: }
1.198 schwarze 1105: return(rr ? ROFF_CONT : ROFF_IGN);
1.74 kristaps 1106: }
1107:
1.184 schwarze 1108: static int
1109: roff_getnum(const char *v, int *pos, int *res)
1110: {
1111: int p, n;
1112:
1113: p = *pos;
1114: n = v[p] == '-';
1115: if (n)
1116: p++;
1117:
1118: for (*res = 0; isdigit((unsigned char)v[p]); p++)
1119: *res += 10 * *res + v[p] - '0';
1120: if (p == *pos + n)
1121: return 0;
1122:
1123: if (n)
1124: *res = -*res;
1125:
1126: *pos = p;
1127: return 1;
1128: }
1129:
1130: static int
1131: roff_getop(const char *v, int *pos, char *res)
1132: {
1133: int e;
1134:
1135: *res = v[*pos];
1136: e = v[*pos + 1] == '=';
1137:
1138: switch (*res) {
1139: case '=':
1140: break;
1141: case '>':
1142: if (e)
1143: *res = 'g';
1144: break;
1145: case '<':
1146: if (e)
1147: *res = 'l';
1148: break;
1149: default:
1150: return(0);
1151: }
1152:
1153: *pos += 1 + e;
1154:
1155: return(*res);
1156: }
1157:
1.198 schwarze 1158: /*
1159: * Evaluate a string comparison condition.
1160: * The first character is the delimiter.
1161: * Succeed if the string up to its second occurrence
1162: * matches the string up to its third occurence.
1163: * Advance the cursor after the third occurrence
1164: * or lacking that, to the end of the line.
1165: */
1166: static int
1167: roff_evalstrcond(const char *v, int *pos)
1168: {
1169: const char *s1, *s2, *s3;
1170: int match;
1171:
1172: match = 0;
1173: s1 = v + *pos; /* initial delimiter */
1174: s2 = s1 + 1; /* for scanning the first string */
1175: s3 = strchr(s2, *s1); /* for scanning the second string */
1176:
1177: if (NULL == s3) /* found no middle delimiter */
1178: goto out;
1179:
1180: while ('\0' != *++s3) {
1181: if (*s2 != *s3) { /* mismatch */
1182: s3 = strchr(s3, *s1);
1183: break;
1184: }
1185: if (*s3 == *s1) { /* found the final delimiter */
1186: match = 1;
1187: break;
1188: }
1189: s2++;
1190: }
1191:
1192: out:
1193: if (NULL == s3)
1194: s3 = strchr(s2, '\0');
1195: else
1196: s3++;
1197: *pos = s3 - v;
1198: return(match);
1199: }
1200:
1201: static int
1.88 kristaps 1202: roff_evalcond(const char *v, int *pos)
1203: {
1.198 schwarze 1204: int wanttrue, lh, rh;
1.184 schwarze 1205: char op;
1.88 kristaps 1206:
1.198 schwarze 1207: if ('!' == v[*pos]) {
1208: wanttrue = 0;
1209: (*pos)++;
1210: } else
1211: wanttrue = 1;
1212:
1.88 kristaps 1213: switch (v[*pos]) {
1214: case ('n'):
1.198 schwarze 1215: /* FALLTHROUGH */
1216: case ('o'):
1.88 kristaps 1217: (*pos)++;
1.198 schwarze 1218: return(wanttrue);
1219: case ('c'):
1220: /* FALLTHROUGH */
1221: case ('d'):
1222: /* FALLTHROUGH */
1.88 kristaps 1223: case ('e'):
1224: /* FALLTHROUGH */
1.198 schwarze 1225: case ('r'):
1.88 kristaps 1226: /* FALLTHROUGH */
1227: case ('t'):
1228: (*pos)++;
1.198 schwarze 1229: return(!wanttrue);
1.88 kristaps 1230: default:
1231: break;
1232: }
1233:
1.184 schwarze 1234: if (!roff_getnum(v, pos, &lh))
1.198 schwarze 1235: return(roff_evalstrcond(v, pos) == wanttrue);
1236: if (!roff_getop(v, pos, &op))
1237: return((lh > 0) == wanttrue);
1.184 schwarze 1238: if (!roff_getnum(v, pos, &rh))
1.198 schwarze 1239: return(0);
1240:
1.184 schwarze 1241: switch (op) {
1242: case 'g':
1.198 schwarze 1243: return((lh >= rh) == wanttrue);
1.184 schwarze 1244: case 'l':
1.198 schwarze 1245: return((lh <= rh) == wanttrue);
1.184 schwarze 1246: case '=':
1.198 schwarze 1247: return((lh == rh) == wanttrue);
1.184 schwarze 1248: case '>':
1.198 schwarze 1249: return((lh > rh) == wanttrue);
1.184 schwarze 1250: case '<':
1.198 schwarze 1251: return((lh < rh) == wanttrue);
1.184 schwarze 1252: default:
1.198 schwarze 1253: return(0);
1.184 schwarze 1254: }
1.88 kristaps 1255: }
1256:
1.75 kristaps 1257: /* ARGSUSED */
1.74 kristaps 1258: static enum rofferr
1.103 kristaps 1259: roff_line_ignore(ROFF_ARGS)
1.89 kristaps 1260: {
1.123 schwarze 1261:
1.89 kristaps 1262: return(ROFF_IGN);
1263: }
1264:
1.104 kristaps 1265: /* ARGSUSED */
1266: static enum rofferr
1.82 kristaps 1267: roff_cond(ROFF_ARGS)
1.74 kristaps 1268: {
1.173 schwarze 1269:
1270: roffnode_push(r, tok, NULL, ln, ppos);
1.74 kristaps 1271:
1.134 kristaps 1272: /*
1273: * An `.el' has no conditional body: it will consume the value
1274: * of the current rstack entry set in prior `ie' calls or
1275: * defaults to DENY.
1276: *
1277: * If we're not an `el', however, then evaluate the conditional.
1278: */
1.133 kristaps 1279:
1.173 schwarze 1280: r->last->rule = ROFF_el == tok ?
1.198 schwarze 1281: (r->rstackpos < 0 ? 0 : r->rstack[r->rstackpos--]) :
1.134 kristaps 1282: roff_evalcond(*bufp, &pos);
1.77 kristaps 1283:
1.134 kristaps 1284: /*
1285: * An if-else will put the NEGATION of the current evaluated
1286: * conditional into the stack of rules.
1287: */
1288:
1.84 schwarze 1289: if (ROFF_ie == tok) {
1.134 kristaps 1290: if (r->rstackpos == RSTACK_MAX - 1) {
1291: mandoc_msg(MANDOCERR_MEM,
1292: r->parse, ln, ppos, NULL);
1293: return(ROFF_ERR);
1294: }
1.198 schwarze 1295: r->rstack[++r->rstackpos] = !r->last->rule;
1.82 kristaps 1296: }
1.88 kristaps 1297:
1298: /* If the parent has false as its rule, then so do we. */
1299:
1.198 schwarze 1300: if (r->last->parent && !r->last->parent->rule)
1301: r->last->rule = 0;
1.88 kristaps 1302:
1303: /*
1.173 schwarze 1304: * Determine scope.
1305: * If there is nothing on the line after the conditional,
1306: * not even whitespace, use next-line scope.
1.88 kristaps 1307: */
1.74 kristaps 1308:
1.173 schwarze 1309: if ('\0' == (*bufp)[pos]) {
1310: r->last->endspan = 2;
1311: goto out;
1312: }
1313:
1314: while (' ' == (*bufp)[pos])
1315: pos++;
1316:
1317: /* An opening brace requests multiline scope. */
1.75 kristaps 1318:
1319: if ('\\' == (*bufp)[pos] && '{' == (*bufp)[pos + 1]) {
1320: r->last->endspan = -1;
1321: pos += 2;
1.173 schwarze 1322: goto out;
1.109 kristaps 1323: }
1.74 kristaps 1324:
1.77 kristaps 1325: /*
1.173 schwarze 1326: * Anything else following the conditional causes
1327: * single-line scope. Warn if the scope contains
1328: * nothing but trailing whitespace.
1.77 kristaps 1329: */
1330:
1.75 kristaps 1331: if ('\0' == (*bufp)[pos])
1.173 schwarze 1332: mandoc_msg(MANDOCERR_NOARGS, r->parse, ln, ppos, NULL);
1.77 kristaps 1333:
1.173 schwarze 1334: r->last->endspan = 1;
1.74 kristaps 1335:
1.173 schwarze 1336: out:
1.75 kristaps 1337: *offs = pos;
1338: return(ROFF_RERUN);
1.83 schwarze 1339: }
1340:
1341:
1342: /* ARGSUSED */
1343: static enum rofferr
1.92 schwarze 1344: roff_ds(ROFF_ARGS)
1345: {
1.96 kristaps 1346: char *name, *string;
1347:
1348: /*
1349: * A symbol is named by the first word following the macro
1350: * invocation up to a space. Its value is anything after the
1351: * name's trailing whitespace and optional double-quote. Thus,
1352: *
1353: * [.ds foo "bar " ]
1354: *
1355: * will have `bar " ' as its value.
1356: */
1.92 schwarze 1357:
1.121 schwarze 1358: string = *bufp + pos;
1359: name = roff_getname(r, &string, ln, pos);
1.92 schwarze 1360: if ('\0' == *name)
1361: return(ROFF_IGN);
1362:
1.121 schwarze 1363: /* Read past initial double-quote. */
1364: if ('"' == *string)
1.92 schwarze 1365: string++;
1366:
1.96 kristaps 1367: /* The rest is the value. */
1.193 schwarze 1368: roff_setstr(r, name, string, ROFF_as == tok);
1.92 schwarze 1369: return(ROFF_IGN);
1370: }
1371:
1.180 schwarze 1372: void
1.187 schwarze 1373: roff_setreg(struct roff *r, const char *name, int val, char sign)
1.147 kristaps 1374: {
1.180 schwarze 1375: struct roffreg *reg;
1376:
1377: /* Search for an existing register with the same name. */
1378: reg = r->regtab;
1379:
1380: while (reg && strcmp(name, reg->key.p))
1381: reg = reg->next;
1.147 kristaps 1382:
1.180 schwarze 1383: if (NULL == reg) {
1384: /* Create a new register. */
1385: reg = mandoc_malloc(sizeof(struct roffreg));
1386: reg->key.p = mandoc_strdup(name);
1387: reg->key.sz = strlen(name);
1.187 schwarze 1388: reg->val = 0;
1.180 schwarze 1389: reg->next = r->regtab;
1390: r->regtab = reg;
1391: }
1392:
1.187 schwarze 1393: if ('+' == sign)
1394: reg->val += val;
1395: else if ('-' == sign)
1396: reg->val -= val;
1397: else
1398: reg->val = val;
1.147 kristaps 1399: }
1400:
1.192 schwarze 1401: /*
1402: * Handle some predefined read-only number registers.
1403: * For now, return -1 if the requested register is not predefined;
1404: * in case a predefined read-only register having the value -1
1405: * were to turn up, another special value would have to be chosen.
1406: */
1407: static int
1408: roff_getregro(const char *name)
1409: {
1410:
1411: switch (*name) {
1412: case ('A'): /* ASCII approximation mode is always off. */
1413: return(0);
1414: case ('g'): /* Groff compatibility mode is always on. */
1415: return(1);
1416: case ('H'): /* Fixed horizontal resolution. */
1417: return (24);
1418: case ('j'): /* Always adjust left margin only. */
1419: return(0);
1420: case ('T'): /* Some output device is always defined. */
1421: return(1);
1422: case ('V'): /* Fixed vertical resolution. */
1423: return (40);
1424: default:
1425: return (-1);
1426: }
1427: }
1428:
1.181 schwarze 1429: int
1.180 schwarze 1430: roff_getreg(const struct roff *r, const char *name)
1.147 kristaps 1431: {
1.180 schwarze 1432: struct roffreg *reg;
1.192 schwarze 1433: int val;
1434:
1435: if ('.' == name[0] && '\0' != name[1] && '\0' == name[2]) {
1436: val = roff_getregro(name + 1);
1437: if (-1 != val)
1438: return (val);
1439: }
1.180 schwarze 1440:
1441: for (reg = r->regtab; reg; reg = reg->next)
1442: if (0 == strcmp(name, reg->key.p))
1.181 schwarze 1443: return(reg->val);
1444:
1445: return(0);
1446: }
1447:
1448: static int
1449: roff_getregn(const struct roff *r, const char *name, size_t len)
1450: {
1451: struct roffreg *reg;
1.192 schwarze 1452: int val;
1453:
1454: if ('.' == name[0] && 2 == len) {
1455: val = roff_getregro(name + 1);
1456: if (-1 != val)
1457: return (val);
1458: }
1.181 schwarze 1459:
1460: for (reg = r->regtab; reg; reg = reg->next)
1461: if (len == reg->key.sz &&
1462: 0 == strncmp(name, reg->key.p, len))
1463: return(reg->val);
1.147 kristaps 1464:
1.180 schwarze 1465: return(0);
1.147 kristaps 1466: }
1467:
1.180 schwarze 1468: static void
1469: roff_freereg(struct roffreg *reg)
1.147 kristaps 1470: {
1.180 schwarze 1471: struct roffreg *old_reg;
1.147 kristaps 1472:
1.180 schwarze 1473: while (NULL != reg) {
1474: free(reg->key.p);
1475: old_reg = reg;
1476: reg = reg->next;
1477: free(old_reg);
1478: }
1.147 kristaps 1479: }
1.92 schwarze 1480:
1481: /* ARGSUSED */
1482: static enum rofferr
1.89 kristaps 1483: roff_nr(ROFF_ARGS)
1.83 schwarze 1484: {
1.121 schwarze 1485: const char *key;
1486: char *val;
1.187 schwarze 1487: size_t sz;
1.138 kristaps 1488: int iv;
1.187 schwarze 1489: char sign;
1.89 kristaps 1490:
1.121 schwarze 1491: val = *bufp + pos;
1492: key = roff_getname(r, &val, ln, pos);
1.89 kristaps 1493:
1.187 schwarze 1494: sign = *val;
1495: if ('+' == sign || '-' == sign)
1496: val++;
1497:
1498: sz = strspn(val, "0123456789");
1499: iv = sz ? mandoc_strntoi(val, sz, 10) : 0;
1.180 schwarze 1500:
1.187 schwarze 1501: roff_setreg(r, key, iv, sign);
1.109 kristaps 1502:
1.122 schwarze 1503: return(ROFF_IGN);
1504: }
1505:
1506: /* ARGSUSED */
1507: static enum rofferr
1508: roff_rm(ROFF_ARGS)
1509: {
1510: const char *name;
1511: char *cp;
1512:
1513: cp = *bufp + pos;
1514: while ('\0' != *cp) {
1.127 kristaps 1515: name = roff_getname(r, &cp, ln, (int)(cp - *bufp));
1.122 schwarze 1516: if ('\0' != *name)
1517: roff_setstr(r, name, NULL, 0);
1518: }
1.178 schwarze 1519: return(ROFF_IGN);
1520: }
1521:
1522: /* ARGSUSED */
1523: static enum rofferr
1524: roff_it(ROFF_ARGS)
1525: {
1526: char *cp;
1527: size_t len;
1528: int iv;
1529:
1530: /* Parse the number of lines. */
1531: cp = *bufp + pos;
1532: len = strcspn(cp, " \t");
1533: cp[len] = '\0';
1534: if ((iv = mandoc_strntoi(cp, len, 10)) <= 0) {
1535: mandoc_msg(MANDOCERR_NUMERIC, r->parse,
1536: ln, ppos, *bufp + 1);
1537: return(ROFF_IGN);
1538: }
1539: cp += len + 1;
1540:
1541: /* Arm the input line trap. */
1542: roffit_lines = iv;
1543: roffit_macro = mandoc_strdup(cp);
1.109 kristaps 1544: return(ROFF_IGN);
1.175 schwarze 1545: }
1546:
1547: /* ARGSUSED */
1548: static enum rofferr
1549: roff_Dd(ROFF_ARGS)
1550: {
1551: const char *const *cp;
1552:
1.199 ! schwarze 1553: if (0 == ((MPARSE_MDOC | MPARSE_QUICK) & r->options))
1.175 schwarze 1554: for (cp = __mdoc_reserved; *cp; cp++)
1555: roff_setstr(r, *cp, NULL, 0);
1556:
1557: return(ROFF_CONT);
1558: }
1559:
1560: /* ARGSUSED */
1561: static enum rofferr
1562: roff_TH(ROFF_ARGS)
1563: {
1564: const char *const *cp;
1565:
1.199 ! schwarze 1566: if (0 == (MPARSE_QUICK & r->options))
1.175 schwarze 1567: for (cp = __man_reserved; *cp; cp++)
1568: roff_setstr(r, *cp, NULL, 0);
1569:
1570: return(ROFF_CONT);
1.109 kristaps 1571: }
1572:
1573: /* ARGSUSED */
1574: static enum rofferr
1575: roff_TE(ROFF_ARGS)
1576: {
1577:
1578: if (NULL == r->tbl)
1.128 kristaps 1579: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.115 kristaps 1580: else
1.151 kristaps 1581: tbl_end(&r->tbl);
1.109 kristaps 1582:
1.112 kristaps 1583: return(ROFF_IGN);
1584: }
1585:
1586: /* ARGSUSED */
1587: static enum rofferr
1588: roff_T_(ROFF_ARGS)
1589: {
1590:
1591: if (NULL == r->tbl)
1.128 kristaps 1592: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.112 kristaps 1593: else
1.116 kristaps 1594: tbl_restart(ppos, ln, r->tbl);
1.112 kristaps 1595:
1.109 kristaps 1596: return(ROFF_IGN);
1597: }
1598:
1.156 kristaps 1599: #if 0
1600: static int
1.151 kristaps 1601: roff_closeeqn(struct roff *r)
1602: {
1603:
1604: return(r->eqn && ROFF_EQN == eqn_end(&r->eqn) ? 1 : 0);
1605: }
1.156 kristaps 1606: #endif
1.151 kristaps 1607:
1.156 kristaps 1608: static void
1.151 kristaps 1609: roff_openeqn(struct roff *r, const char *name, int line,
1610: int offs, const char *buf)
1.125 kristaps 1611: {
1.151 kristaps 1612: struct eqn_node *e;
1613: int poff;
1.125 kristaps 1614:
1615: assert(NULL == r->eqn);
1.151 kristaps 1616: e = eqn_alloc(name, offs, line, r->parse);
1.125 kristaps 1617:
1618: if (r->last_eqn)
1619: r->last_eqn->next = e;
1620: else
1621: r->first_eqn = r->last_eqn = e;
1622:
1623: r->eqn = r->last_eqn = e;
1.151 kristaps 1624:
1625: if (buf) {
1626: poff = 0;
1627: eqn_read(&r->eqn, line, buf, offs, &poff);
1628: }
1629: }
1630:
1631: /* ARGSUSED */
1632: static enum rofferr
1633: roff_EQ(ROFF_ARGS)
1634: {
1635:
1636: roff_openeqn(r, *bufp + pos, ln, ppos, NULL);
1.125 kristaps 1637: return(ROFF_IGN);
1638: }
1639:
1640: /* ARGSUSED */
1641: static enum rofferr
1642: roff_EN(ROFF_ARGS)
1643: {
1644:
1.128 kristaps 1645: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.125 kristaps 1646: return(ROFF_IGN);
1647: }
1648:
1649: /* ARGSUSED */
1650: static enum rofferr
1.109 kristaps 1651: roff_TS(ROFF_ARGS)
1652: {
1.176 schwarze 1653: struct tbl_node *tbl;
1.89 kristaps 1654:
1.115 kristaps 1655: if (r->tbl) {
1.128 kristaps 1656: mandoc_msg(MANDOCERR_SCOPEBROKEN, r->parse, ln, ppos, NULL);
1.151 kristaps 1657: tbl_end(&r->tbl);
1.115 kristaps 1658: }
1.83 schwarze 1659:
1.176 schwarze 1660: tbl = tbl_alloc(ppos, ln, r->parse);
1.113 kristaps 1661:
1662: if (r->last_tbl)
1.176 schwarze 1663: r->last_tbl->next = tbl;
1.113 kristaps 1664: else
1.176 schwarze 1665: r->first_tbl = r->last_tbl = tbl;
1.113 kristaps 1666:
1.176 schwarze 1667: r->tbl = r->last_tbl = tbl;
1.83 schwarze 1668: return(ROFF_IGN);
1.92 schwarze 1669: }
1670:
1.105 kristaps 1671: /* ARGSUSED */
1672: static enum rofferr
1.174 kristaps 1673: roff_cc(ROFF_ARGS)
1674: {
1675: const char *p;
1676:
1677: p = *bufp + pos;
1678:
1679: if ('\0' == *p || '.' == (r->control = *p++))
1680: r->control = 0;
1681:
1682: if ('\0' != *p)
1683: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
1684:
1685: return(ROFF_IGN);
1686: }
1687:
1688: /* ARGSUSED */
1689: static enum rofferr
1.164 kristaps 1690: roff_tr(ROFF_ARGS)
1691: {
1692: const char *p, *first, *second;
1693: size_t fsz, ssz;
1694: enum mandoc_esc esc;
1695:
1696: p = *bufp + pos;
1697:
1698: if ('\0' == *p) {
1699: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
1700: return(ROFF_IGN);
1701: }
1702:
1703: while ('\0' != *p) {
1704: fsz = ssz = 1;
1705:
1706: first = p++;
1707: if ('\\' == *first) {
1708: esc = mandoc_escape(&p, NULL, NULL);
1709: if (ESCAPE_ERROR == esc) {
1710: mandoc_msg
1711: (MANDOCERR_BADESCAPE, r->parse,
1712: ln, (int)(p - *bufp), NULL);
1713: return(ROFF_IGN);
1714: }
1715: fsz = (size_t)(p - first);
1716: }
1717:
1718: second = p++;
1719: if ('\\' == *second) {
1720: esc = mandoc_escape(&p, NULL, NULL);
1721: if (ESCAPE_ERROR == esc) {
1722: mandoc_msg
1723: (MANDOCERR_BADESCAPE, r->parse,
1724: ln, (int)(p - *bufp), NULL);
1725: return(ROFF_IGN);
1726: }
1727: ssz = (size_t)(p - second);
1.165 kristaps 1728: } else if ('\0' == *second) {
1.164 kristaps 1729: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse,
1730: ln, (int)(p - *bufp), NULL);
1731: second = " ";
1.165 kristaps 1732: p--;
1.164 kristaps 1733: }
1734:
1.167 kristaps 1735: if (fsz > 1) {
1736: roff_setstrn(&r->xmbtab, first,
1737: fsz, second, ssz, 0);
1738: continue;
1739: }
1740:
1741: if (NULL == r->xtab)
1742: r->xtab = mandoc_calloc
1743: (128, sizeof(struct roffstr));
1744:
1745: free(r->xtab[(int)*first].p);
1746: r->xtab[(int)*first].p = mandoc_strndup(second, ssz);
1747: r->xtab[(int)*first].sz = ssz;
1.164 kristaps 1748: }
1749:
1750: return(ROFF_IGN);
1751: }
1752:
1753: /* ARGSUSED */
1754: static enum rofferr
1.105 kristaps 1755: roff_so(ROFF_ARGS)
1756: {
1757: char *name;
1758:
1.128 kristaps 1759: mandoc_msg(MANDOCERR_SO, r->parse, ln, ppos, NULL);
1.105 kristaps 1760:
1761: /*
1762: * Handle `so'. Be EXTREMELY careful, as we shouldn't be
1763: * opening anything that's not in our cwd or anything beneath
1764: * it. Thus, explicitly disallow traversing up the file-system
1765: * or using absolute paths.
1766: */
1767:
1768: name = *bufp + pos;
1769: if ('/' == *name || strstr(name, "../") || strstr(name, "/..")) {
1.128 kristaps 1770: mandoc_msg(MANDOCERR_SOPATH, r->parse, ln, pos, NULL);
1.105 kristaps 1771: return(ROFF_ERR);
1772: }
1773:
1774: *offs = pos;
1775: return(ROFF_SO);
1776: }
1.92 schwarze 1777:
1.106 kristaps 1778: /* ARGSUSED */
1779: static enum rofferr
1780: roff_userdef(ROFF_ARGS)
1.99 kristaps 1781: {
1.106 kristaps 1782: const char *arg[9];
1783: char *cp, *n1, *n2;
1.119 schwarze 1784: int i;
1.106 kristaps 1785:
1786: /*
1787: * Collect pointers to macro argument strings
1.188 schwarze 1788: * and NUL-terminate them.
1.106 kristaps 1789: */
1790: cp = *bufp + pos;
1.119 schwarze 1791: for (i = 0; i < 9; i++)
1.120 schwarze 1792: arg[i] = '\0' == *cp ? "" :
1.136 kristaps 1793: mandoc_getarg(r->parse, &cp, ln, &pos);
1.99 kristaps 1794:
1.106 kristaps 1795: /*
1796: * Expand macro arguments.
1.99 kristaps 1797: */
1.106 kristaps 1798: *szp = 0;
1799: n1 = cp = mandoc_strdup(r->current_string);
1800: while (NULL != (cp = strstr(cp, "\\$"))) {
1801: i = cp[2] - '1';
1802: if (0 > i || 8 < i) {
1803: /* Not an argument invocation. */
1804: cp += 2;
1805: continue;
1806: }
1807:
1808: *szp = strlen(n1) - 3 + strlen(arg[i]) + 1;
1809: n2 = mandoc_malloc(*szp);
1810:
1811: strlcpy(n2, n1, (size_t)(cp - n1 + 1));
1812: strlcat(n2, arg[i], *szp);
1813: strlcat(n2, cp + 3, *szp);
1814:
1815: cp = n2 + (cp - n1);
1816: free(n1);
1817: n1 = n2;
1.99 kristaps 1818: }
1819:
1.106 kristaps 1820: /*
1821: * Replace the macro invocation
1822: * by the expanded macro.
1823: */
1824: free(*bufp);
1825: *bufp = n1;
1826: if (0 == *szp)
1827: *szp = strlen(*bufp) + 1;
1828:
1829: return(*szp > 1 && '\n' == (*bufp)[(int)*szp - 2] ?
1830: ROFF_REPARSE : ROFF_APPEND);
1.99 kristaps 1831: }
1.121 schwarze 1832:
1833: static char *
1834: roff_getname(struct roff *r, char **cpp, int ln, int pos)
1835: {
1836: char *name, *cp;
1837:
1838: name = *cpp;
1839: if ('\0' == *name)
1840: return(name);
1841:
1842: /* Read until end of name. */
1843: for (cp = name; '\0' != *cp && ' ' != *cp; cp++) {
1844: if ('\\' != *cp)
1845: continue;
1846: cp++;
1847: if ('\\' == *cp)
1848: continue;
1.128 kristaps 1849: mandoc_msg(MANDOCERR_NAMESC, r->parse, ln, pos, NULL);
1.121 schwarze 1850: *cp = '\0';
1851: name = cp;
1852: }
1853:
1854: /* Nil-terminate name. */
1855: if ('\0' != *cp)
1856: *(cp++) = '\0';
1857:
1858: /* Read past spaces. */
1859: while (' ' == *cp)
1860: cp++;
1861:
1862: *cpp = cp;
1863: return(name);
1864: }
1865:
1.106 kristaps 1866: /*
1867: * Store *string into the user-defined string called *name.
1868: * To clear an existing entry, call with (*r, *name, NULL, 0).
1.193 schwarze 1869: * append == 0: replace mode
1870: * append == 1: single-line append mode
1871: * append == 2: multiline append mode, append '\n' after each call
1.106 kristaps 1872: */
1.94 kristaps 1873: static void
1.106 kristaps 1874: roff_setstr(struct roff *r, const char *name, const char *string,
1.193 schwarze 1875: int append)
1.92 schwarze 1876: {
1.164 kristaps 1877:
1878: roff_setstrn(&r->strtab, name, strlen(name), string,
1.193 schwarze 1879: string ? strlen(string) : 0, append);
1.164 kristaps 1880: }
1881:
1882: static void
1.166 kristaps 1883: roff_setstrn(struct roffkv **r, const char *name, size_t namesz,
1.193 schwarze 1884: const char *string, size_t stringsz, int append)
1.164 kristaps 1885: {
1.166 kristaps 1886: struct roffkv *n;
1.164 kristaps 1887: char *c;
1888: int i;
1889: size_t oldch, newch;
1.92 schwarze 1890:
1.106 kristaps 1891: /* Search for an existing string with the same name. */
1.164 kristaps 1892: n = *r;
1893:
1.166 kristaps 1894: while (n && strcmp(name, n->key.p))
1.92 schwarze 1895: n = n->next;
1.94 kristaps 1896:
1897: if (NULL == n) {
1.106 kristaps 1898: /* Create a new string table entry. */
1.166 kristaps 1899: n = mandoc_malloc(sizeof(struct roffkv));
1900: n->key.p = mandoc_strndup(name, namesz);
1901: n->key.sz = namesz;
1902: n->val.p = NULL;
1903: n->val.sz = 0;
1.164 kristaps 1904: n->next = *r;
1905: *r = n;
1.193 schwarze 1906: } else if (0 == append) {
1.166 kristaps 1907: free(n->val.p);
1908: n->val.p = NULL;
1909: n->val.sz = 0;
1.106 kristaps 1910: }
1911:
1912: if (NULL == string)
1913: return;
1914:
1915: /*
1916: * One additional byte for the '\n' in multiline mode,
1917: * and one for the terminating '\0'.
1918: */
1.193 schwarze 1919: newch = stringsz + (1 < append ? 2u : 1u);
1.164 kristaps 1920:
1.166 kristaps 1921: if (NULL == n->val.p) {
1922: n->val.p = mandoc_malloc(newch);
1923: *n->val.p = '\0';
1.106 kristaps 1924: oldch = 0;
1925: } else {
1.166 kristaps 1926: oldch = n->val.sz;
1927: n->val.p = mandoc_realloc(n->val.p, oldch + newch);
1.106 kristaps 1928: }
1929:
1930: /* Skip existing content in the destination buffer. */
1.166 kristaps 1931: c = n->val.p + (int)oldch;
1.106 kristaps 1932:
1933: /* Append new content to the destination buffer. */
1.164 kristaps 1934: i = 0;
1935: while (i < (int)stringsz) {
1.106 kristaps 1936: /*
1937: * Rudimentary roff copy mode:
1938: * Handle escaped backslashes.
1939: */
1.164 kristaps 1940: if ('\\' == string[i] && '\\' == string[i + 1])
1941: i++;
1942: *c++ = string[i++];
1.106 kristaps 1943: }
1.94 kristaps 1944:
1.106 kristaps 1945: /* Append terminating bytes. */
1.193 schwarze 1946: if (1 < append)
1.106 kristaps 1947: *c++ = '\n';
1.163 kristaps 1948:
1.106 kristaps 1949: *c = '\0';
1.166 kristaps 1950: n->val.sz = (int)(c - n->val.p);
1.92 schwarze 1951: }
1952:
1.94 kristaps 1953: static const char *
1954: roff_getstrn(const struct roff *r, const char *name, size_t len)
1.92 schwarze 1955: {
1.166 kristaps 1956: const struct roffkv *n;
1.191 schwarze 1957: int i;
1.92 schwarze 1958:
1.164 kristaps 1959: for (n = r->strtab; n; n = n->next)
1.166 kristaps 1960: if (0 == strncmp(name, n->key.p, len) &&
1961: '\0' == n->key.p[(int)len])
1962: return(n->val.p);
1.191 schwarze 1963:
1964: for (i = 0; i < PREDEFS_MAX; i++)
1965: if (0 == strncmp(name, predefs[i].name, len) &&
1966: '\0' == predefs[i].name[(int)len])
1967: return(predefs[i].str);
1.94 kristaps 1968:
1.157 kristaps 1969: return(NULL);
1.92 schwarze 1970: }
1971:
1.94 kristaps 1972: static void
1.167 kristaps 1973: roff_freestr(struct roffkv *r)
1.92 schwarze 1974: {
1.166 kristaps 1975: struct roffkv *n, *nn;
1.92 schwarze 1976:
1.167 kristaps 1977: for (n = r; n; n = nn) {
1.166 kristaps 1978: free(n->key.p);
1979: free(n->val.p);
1.92 schwarze 1980: nn = n->next;
1981: free(n);
1982: }
1.114 kristaps 1983: }
1984:
1985: const struct tbl_span *
1986: roff_span(const struct roff *r)
1987: {
1988:
1989: return(r->tbl ? tbl_span(r->tbl) : NULL);
1.125 kristaps 1990: }
1991:
1992: const struct eqn *
1993: roff_eqn(const struct roff *r)
1994: {
1995:
1996: return(r->last_eqn ? &r->last_eqn->eqn : NULL);
1.164 kristaps 1997: }
1998:
1999: /*
2000: * Duplicate an input string, making the appropriate character
2001: * conversations (as stipulated by `tr') along the way.
2002: * Returns a heap-allocated string with all the replacements made.
2003: */
2004: char *
2005: roff_strdup(const struct roff *r, const char *p)
2006: {
1.166 kristaps 2007: const struct roffkv *cp;
1.164 kristaps 2008: char *res;
2009: const char *pp;
2010: size_t ssz, sz;
2011: enum mandoc_esc esc;
2012:
1.167 kristaps 2013: if (NULL == r->xmbtab && NULL == r->xtab)
1.164 kristaps 2014: return(mandoc_strdup(p));
2015: else if ('\0' == *p)
2016: return(mandoc_strdup(""));
2017:
2018: /*
2019: * Step through each character looking for term matches
2020: * (remember that a `tr' can be invoked with an escape, which is
2021: * a glyph but the escape is multi-character).
2022: * We only do this if the character hash has been initialised
2023: * and the string is >0 length.
2024: */
2025:
2026: res = NULL;
2027: ssz = 0;
2028:
2029: while ('\0' != *p) {
1.167 kristaps 2030: if ('\\' != *p && r->xtab && r->xtab[(int)*p].p) {
2031: sz = r->xtab[(int)*p].sz;
2032: res = mandoc_realloc(res, ssz + sz + 1);
2033: memcpy(res + ssz, r->xtab[(int)*p].p, sz);
2034: ssz += sz;
2035: p++;
2036: continue;
2037: } else if ('\\' != *p) {
2038: res = mandoc_realloc(res, ssz + 2);
2039: res[ssz++] = *p++;
2040: continue;
2041: }
2042:
1.164 kristaps 2043: /* Search for term matches. */
1.167 kristaps 2044: for (cp = r->xmbtab; cp; cp = cp->next)
1.166 kristaps 2045: if (0 == strncmp(p, cp->key.p, cp->key.sz))
1.164 kristaps 2046: break;
2047:
2048: if (NULL != cp) {
2049: /*
2050: * A match has been found.
2051: * Append the match to the array and move
2052: * forward by its keysize.
2053: */
1.166 kristaps 2054: res = mandoc_realloc
2055: (res, ssz + cp->val.sz + 1);
2056: memcpy(res + ssz, cp->val.p, cp->val.sz);
2057: ssz += cp->val.sz;
2058: p += (int)cp->key.sz;
1.164 kristaps 2059: continue;
2060: }
2061:
1.167 kristaps 2062: /*
2063: * Handle escapes carefully: we need to copy
2064: * over just the escape itself, or else we might
2065: * do replacements within the escape itself.
2066: * Make sure to pass along the bogus string.
2067: */
2068: pp = p++;
2069: esc = mandoc_escape(&p, NULL, NULL);
2070: if (ESCAPE_ERROR == esc) {
2071: sz = strlen(pp);
1.164 kristaps 2072: res = mandoc_realloc(res, ssz + sz + 1);
2073: memcpy(res + ssz, pp, sz);
1.167 kristaps 2074: break;
1.164 kristaps 2075: }
1.167 kristaps 2076: /*
2077: * We bail out on bad escapes.
2078: * No need to warn: we already did so when
2079: * roff_res() was called.
2080: */
2081: sz = (int)(p - pp);
2082: res = mandoc_realloc(res, ssz + sz + 1);
2083: memcpy(res + ssz, pp, sz);
2084: ssz += sz;
1.164 kristaps 2085: }
2086:
2087: res[(int)ssz] = '\0';
2088: return(res);
1.174 kristaps 2089: }
2090:
2091: /*
2092: * Find out whether a line is a macro line or not.
2093: * If it is, adjust the current position and return one; if it isn't,
2094: * return zero and don't change the current position.
2095: * If the control character has been set with `.cc', then let that grain
2096: * precedence.
2097: * This is slighly contrary to groff, where using the non-breaking
2098: * control character when `cc' has been invoked will cause the
2099: * non-breaking macro contents to be printed verbatim.
2100: */
2101: int
2102: roff_getcontrol(const struct roff *r, const char *cp, int *ppos)
2103: {
2104: int pos;
2105:
2106: pos = *ppos;
2107:
2108: if (0 != r->control && cp[pos] == r->control)
2109: pos++;
2110: else if (0 != r->control)
2111: return(0);
2112: else if ('\\' == cp[pos] && '.' == cp[pos + 1])
2113: pos += 2;
2114: else if ('.' == cp[pos] || '\'' == cp[pos])
2115: pos++;
2116: else
2117: return(0);
2118:
2119: while (' ' == cp[pos] || '\t' == cp[pos])
2120: pos++;
2121:
2122: *ppos = pos;
2123: return(1);
1.74 kristaps 2124: }
CVSweb