Annotation of mandoc/roff.c, Revision 1.202
1.202 ! schwarze 1: /* $Id: roff.c,v 1.201 2014/03/23 11:25:26 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"
1.201 schwarze 29: #include "mandoc_aux.h"
1.109 kristaps 30: #include "libroff.h"
1.94 kristaps 31: #include "libmandoc.h"
1.33 kristaps 32:
1.141 kristaps 33: /* Maximum number of nested if-else conditionals. */
1.82 kristaps 34: #define RSTACK_MAX 128
35:
1.170 schwarze 36: /* Maximum number of string expansions per line, to break infinite loops. */
37: #define EXPAND_LIMIT 1000
38:
1.67 kristaps 39: enum rofft {
1.103 kristaps 40: ROFF_ad,
1.80 kristaps 41: ROFF_am,
42: ROFF_ami,
43: ROFF_am1,
1.193 schwarze 44: ROFF_as,
1.174 kristaps 45: ROFF_cc,
1.194 schwarze 46: ROFF_ce,
1.80 kristaps 47: ROFF_de,
48: ROFF_dei,
49: ROFF_de1,
1.83 schwarze 50: ROFF_ds,
1.82 kristaps 51: ROFF_el,
1.185 schwarze 52: ROFF_fam,
1.186 schwarze 53: ROFF_hw,
1.103 kristaps 54: ROFF_hy,
1.82 kristaps 55: ROFF_ie,
1.75 kristaps 56: ROFF_if,
1.76 kristaps 57: ROFF_ig,
1.123 schwarze 58: ROFF_it,
1.103 kristaps 59: ROFF_ne,
60: ROFF_nh,
1.104 kristaps 61: ROFF_nr,
1.124 schwarze 62: ROFF_ns,
63: ROFF_ps,
1.83 schwarze 64: ROFF_rm,
1.105 kristaps 65: ROFF_so,
1.124 schwarze 66: ROFF_ta,
1.83 schwarze 67: ROFF_tr,
1.175 schwarze 68: ROFF_Dd,
69: ROFF_TH,
1.109 kristaps 70: ROFF_TS,
71: ROFF_TE,
1.112 kristaps 72: ROFF_T_,
1.125 kristaps 73: ROFF_EQ,
74: ROFF_EN,
1.76 kristaps 75: ROFF_cblock,
1.106 kristaps 76: ROFF_USERDEF,
1.67 kristaps 77: ROFF_MAX
78: };
79:
1.147 kristaps 80: /*
1.167 kristaps 81: * An incredibly-simple string buffer.
82: */
1.94 kristaps 83: struct roffstr {
1.167 kristaps 84: char *p; /* nil-terminated buffer */
85: size_t sz; /* saved strlen(p) */
1.166 kristaps 86: };
87:
88: /*
1.167 kristaps 89: * A key-value roffstr pair as part of a singly-linked list.
1.166 kristaps 90: */
91: struct roffkv {
92: struct roffstr key;
93: struct roffstr val;
94: struct roffkv *next; /* next in list */
1.94 kristaps 95: };
96:
1.180 schwarze 97: /*
98: * A single number register as part of a singly-linked list.
99: */
100: struct roffreg {
101: struct roffstr key;
1.181 schwarze 102: int val;
1.180 schwarze 103: struct roffreg *next;
104: };
105:
1.67 kristaps 106: struct roff {
1.128 kristaps 107: struct mparse *parse; /* parse point */
1.199 schwarze 108: int options; /* parse options */
1.67 kristaps 109: struct roffnode *last; /* leaf of stack */
1.198 schwarze 110: int rstack[RSTACK_MAX]; /* stack of !`ie' rules */
1.174 kristaps 111: char control; /* control character */
1.82 kristaps 112: int rstackpos; /* position in rstack */
1.180 schwarze 113: struct roffreg *regtab; /* number registers */
1.166 kristaps 114: struct roffkv *strtab; /* user-defined strings & macros */
1.167 kristaps 115: struct roffkv *xmbtab; /* multi-byte trans table (`tr') */
116: struct roffstr *xtab; /* single-byte trans table (`tr') */
1.106 kristaps 117: const char *current_string; /* value of last called user macro */
1.118 kristaps 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 */
1.125 kristaps 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 */
1.79 kristaps 124: };
125:
1.67 kristaps 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 */
1.106 kristaps 131: char *name; /* node name, e.g. macro name */
1.79 kristaps 132: char *end; /* end-rules: custom token */
133: int endspan; /* end-rules: next-line or infty */
1.198 schwarze 134: int rule; /* current evaluation rule */
1.67 kristaps 135: };
136:
137: #define ROFF_ARGS struct roff *r, /* parse ctx */ \
1.72 kristaps 138: enum rofft tok, /* tok of macro */ \
1.67 kristaps 139: char **bufp, /* input buffer */ \
140: size_t *szp, /* size of input buffer */ \
141: int ln, /* parse line */ \
1.75 kristaps 142: int ppos, /* original pos in buffer */ \
143: int pos, /* current pos in buffer */ \
1.74 kristaps 144: int *offs /* reset offset of buffer data */
1.67 kristaps 145:
146: typedef enum rofferr (*roffproc)(ROFF_ARGS);
147:
148: struct roffmac {
149: const char *name; /* macro name */
1.79 kristaps 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 */
1.85 kristaps 155: struct roffmac *next;
1.67 kristaps 156: };
157:
1.141 kristaps 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:
1.155 kristaps 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);
1.80 kristaps 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);
1.174 kristaps 176: static enum rofferr roff_cc(ROFF_ARGS);
1.195 schwarze 177: static void roff_ccond(struct roff *, int, int);
1.82 kristaps 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);
1.92 schwarze 181: static enum rofferr roff_ds(ROFF_ARGS);
1.198 schwarze 182: static int roff_evalcond(const char *, int *);
183: static int roff_evalstrcond(const char *, int *);
1.155 kristaps 184: static void roff_free1(struct roff *);
1.180 schwarze 185: static void roff_freereg(struct roffreg *);
1.167 kristaps 186: static void roff_freestr(struct roffkv *);
1.121 schwarze 187: static char *roff_getname(struct roff *, char **, int, int);
1.184 schwarze 188: static int roff_getnum(const char *, int *, int *);
189: static int roff_getop(const char *, int *, char *);
1.181 schwarze 190: static int roff_getregn(const struct roff *,
191: const char *, size_t);
1.192 schwarze 192: static int roff_getregro(const char *name);
1.94 kristaps 193: static const char *roff_getstrn(const struct roff *,
194: const char *, size_t);
1.178 schwarze 195: static enum rofferr roff_it(ROFF_ARGS);
1.103 kristaps 196: static enum rofferr roff_line_ignore(ROFF_ARGS);
1.89 kristaps 197: static enum rofferr roff_nr(ROFF_ARGS);
1.169 schwarze 198: static void roff_openeqn(struct roff *, const char *,
1.156 kristaps 199: int, int, const char *);
1.155 kristaps 200: static enum rofft roff_parse(struct roff *, const char *, int *);
1.178 schwarze 201: static enum rofferr roff_parsetext(char **, size_t *, int, int *);
1.172 schwarze 202: static enum rofferr roff_res(struct roff *,
1.142 kristaps 203: char **, size_t *, int, int);
1.122 schwarze 204: static enum rofferr roff_rm(ROFF_ARGS);
1.94 kristaps 205: static void roff_setstr(struct roff *,
1.106 kristaps 206: const char *, const char *, int);
1.166 kristaps 207: static void roff_setstrn(struct roffkv **, const char *,
1.164 kristaps 208: size_t, const char *, size_t, int);
1.105 kristaps 209: static enum rofferr roff_so(ROFF_ARGS);
1.164 kristaps 210: static enum rofferr roff_tr(ROFF_ARGS);
1.175 schwarze 211: static enum rofferr roff_Dd(ROFF_ARGS);
212: static enum rofferr roff_TH(ROFF_ARGS);
1.109 kristaps 213: static enum rofferr roff_TE(ROFF_ARGS);
214: static enum rofferr roff_TS(ROFF_ARGS);
1.125 kristaps 215: static enum rofferr roff_EQ(ROFF_ARGS);
216: static enum rofferr roff_EN(ROFF_ARGS);
1.112 kristaps 217: static enum rofferr roff_T_(ROFF_ARGS);
1.106 kristaps 218: static enum rofferr roff_userdef(ROFF_ARGS);
1.67 kristaps 219:
1.155 kristaps 220: /* See roffhash_find() */
1.85 kristaps 221:
222: #define ASCII_HI 126
223: #define ASCII_LO 33
224: #define HASHWIDTH (ASCII_HI - ASCII_LO + 1)
225:
226: static struct roffmac *hash[HASHWIDTH];
227:
228: static struct roffmac roffs[ROFF_MAX] = {
1.103 kristaps 229: { "ad", roff_line_ignore, NULL, NULL, 0, NULL },
1.85 kristaps 230: { "am", roff_block, roff_block_text, roff_block_sub, 0, NULL },
231: { "ami", roff_block, roff_block_text, roff_block_sub, 0, NULL },
232: { "am1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.193 schwarze 233: { "as", roff_ds, NULL, NULL, 0, NULL },
1.174 kristaps 234: { "cc", roff_cc, NULL, NULL, 0, NULL },
1.194 schwarze 235: { "ce", roff_line_ignore, NULL, NULL, 0, NULL },
1.85 kristaps 236: { "de", roff_block, roff_block_text, roff_block_sub, 0, NULL },
237: { "dei", roff_block, roff_block_text, roff_block_sub, 0, NULL },
238: { "de1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.92 schwarze 239: { "ds", roff_ds, NULL, NULL, 0, NULL },
1.85 kristaps 240: { "el", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
1.185 schwarze 241: { "fam", roff_line_ignore, NULL, NULL, 0, NULL },
1.186 schwarze 242: { "hw", roff_line_ignore, NULL, NULL, 0, NULL },
1.103 kristaps 243: { "hy", roff_line_ignore, NULL, NULL, 0, NULL },
1.85 kristaps 244: { "ie", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
245: { "if", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
246: { "ig", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.178 schwarze 247: { "it", roff_it, NULL, NULL, 0, NULL },
1.103 kristaps 248: { "ne", roff_line_ignore, NULL, NULL, 0, NULL },
249: { "nh", roff_line_ignore, NULL, NULL, 0, NULL },
1.104 kristaps 250: { "nr", roff_nr, NULL, NULL, 0, NULL },
1.124 schwarze 251: { "ns", roff_line_ignore, NULL, NULL, 0, NULL },
252: { "ps", roff_line_ignore, NULL, NULL, 0, NULL },
1.122 schwarze 253: { "rm", roff_rm, NULL, NULL, 0, NULL },
1.105 kristaps 254: { "so", roff_so, NULL, NULL, 0, NULL },
1.124 schwarze 255: { "ta", roff_line_ignore, NULL, NULL, 0, NULL },
1.164 kristaps 256: { "tr", roff_tr, NULL, NULL, 0, NULL },
1.175 schwarze 257: { "Dd", roff_Dd, NULL, NULL, 0, NULL },
258: { "TH", roff_TH, NULL, NULL, 0, NULL },
1.109 kristaps 259: { "TS", roff_TS, NULL, NULL, 0, NULL },
260: { "TE", roff_TE, NULL, NULL, 0, NULL },
1.112 kristaps 261: { "T&", roff_T_, NULL, NULL, 0, NULL },
1.125 kristaps 262: { "EQ", roff_EQ, NULL, NULL, 0, NULL },
263: { "EN", roff_EN, NULL, NULL, 0, NULL },
1.85 kristaps 264: { ".", roff_cblock, NULL, NULL, 0, NULL },
1.106 kristaps 265: { NULL, roff_userdef, NULL, NULL, 0, NULL },
1.67 kristaps 266: };
267:
1.200 schwarze 268: /* not currently implemented: Ds em Eq LP Me PP pp Or Rd Sf SH */
1.175 schwarze 269: const char *const __mdoc_reserved[] = {
270: "Ac", "Ad", "An", "Ao", "Ap", "Aq", "Ar", "At",
271: "Bc", "Bd", "Bf", "Bk", "Bl", "Bo", "Bq",
272: "Brc", "Bro", "Brq", "Bsx", "Bt", "Bx",
273: "Cd", "Cm", "Db", "Dc", "Dd", "Dl", "Do", "Dq",
1.200 schwarze 274: "Dt", "Dv", "Dx", "D1",
275: "Ec", "Ed", "Ef", "Ek", "El", "Em",
276: "En", "Eo", "Er", "Es", "Ev", "Ex",
1.175 schwarze 277: "Fa", "Fc", "Fd", "Fl", "Fn", "Fo", "Fr", "Ft", "Fx",
1.200 schwarze 278: "Hf", "Ic", "In", "It", "Lb", "Li", "Lk", "Lp",
279: "Ms", "Mt", "Nd", "Nm", "No", "Ns", "Nx",
1.175 schwarze 280: "Oc", "Oo", "Op", "Os", "Ot", "Ox",
1.200 schwarze 281: "Pa", "Pc", "Pf", "Po", "Pp", "Pq",
282: "Qc", "Ql", "Qo", "Qq", "Re", "Rs", "Rv",
283: "Sc", "Sh", "Sm", "So", "Sq",
1.175 schwarze 284: "Ss", "St", "Sx", "Sy",
285: "Ta", "Tn", "Ud", "Ux", "Va", "Vt", "Xc", "Xo", "Xr",
1.200 schwarze 286: "%A", "%B", "%C", "%D", "%I", "%J", "%N", "%O",
1.175 schwarze 287: "%P", "%Q", "%R", "%T", "%U", "%V",
288: NULL
289: };
290:
1.200 schwarze 291: /* not currently implemented: BT DE DS ME MT PT SY TQ YS */
1.175 schwarze 292: const char *const __man_reserved[] = {
1.200 schwarze 293: "AT", "B", "BI", "BR", "DT",
294: "EE", "EN", "EQ", "EX", "HP", "I", "IB", "IP", "IR",
295: "LP", "OP", "P", "PD", "PP",
296: "R", "RB", "RE", "RI", "RS", "SB", "SH", "SM", "SS",
297: "TE", "TH", "TP", "TS", "T&", "UC", "UE", "UR",
1.175 schwarze 298: NULL
299: };
300:
1.141 kristaps 301: /* Array of injected predefined strings. */
302: #define PREDEFS_MAX 38
303: static const struct predef predefs[PREDEFS_MAX] = {
304: #include "predefs.in"
305: };
306:
1.155 kristaps 307: /* See roffhash_find() */
1.85 kristaps 308: #define ROFF_HASH(p) (p[0] - ASCII_LO)
309:
1.178 schwarze 310: static int roffit_lines; /* number of lines to delay */
311: static char *roffit_macro; /* nil-terminated macro line */
312:
1.85 kristaps 313: static void
1.155 kristaps 314: roffhash_init(void)
1.85 kristaps 315: {
316: struct roffmac *n;
317: int buc, i;
318:
1.106 kristaps 319: for (i = 0; i < (int)ROFF_USERDEF; i++) {
1.85 kristaps 320: assert(roffs[i].name[0] >= ASCII_LO);
321: assert(roffs[i].name[0] <= ASCII_HI);
322:
323: buc = ROFF_HASH(roffs[i].name);
324:
325: if (NULL != (n = hash[buc])) {
326: for ( ; n->next; n = n->next)
327: /* Do nothing. */ ;
328: n->next = &roffs[i];
329: } else
330: hash[buc] = &roffs[i];
331: }
332: }
333:
1.67 kristaps 334: /*
335: * Look up a roff token by its name. Returns ROFF_MAX if no macro by
336: * the nil-terminated string name could be found.
337: */
338: static enum rofft
1.155 kristaps 339: roffhash_find(const char *p, size_t s)
1.67 kristaps 340: {
1.85 kristaps 341: int buc;
342: struct roffmac *n;
1.67 kristaps 343:
1.85 kristaps 344: /*
345: * libroff has an extremely simple hashtable, for the time
346: * being, which simply keys on the first character, which must
347: * be printable, then walks a chain. It works well enough until
348: * optimised.
349: */
350:
351: if (p[0] < ASCII_LO || p[0] > ASCII_HI)
352: return(ROFF_MAX);
353:
354: buc = ROFF_HASH(p);
355:
356: if (NULL == (n = hash[buc]))
357: return(ROFF_MAX);
358: for ( ; n; n = n->next)
1.106 kristaps 359: if (0 == strncmp(n->name, p, s) && '\0' == n->name[(int)s])
1.85 kristaps 360: return((enum rofft)(n - roffs));
1.67 kristaps 361:
362: return(ROFF_MAX);
363: }
364:
365:
366: /*
367: * Pop the current node off of the stack of roff instructions currently
368: * pending.
369: */
370: static void
371: roffnode_pop(struct roff *r)
372: {
373: struct roffnode *p;
374:
1.75 kristaps 375: assert(r->last);
376: p = r->last;
1.82 kristaps 377:
1.75 kristaps 378: r->last = r->last->parent;
1.106 kristaps 379: free(p->name);
380: free(p->end);
1.67 kristaps 381: free(p);
382: }
383:
384:
385: /*
386: * Push a roff node onto the instruction stack. This must later be
387: * removed with roffnode_pop().
388: */
1.98 schwarze 389: static void
1.106 kristaps 390: roffnode_push(struct roff *r, enum rofft tok, const char *name,
391: int line, int col)
1.67 kristaps 392: {
393: struct roffnode *p;
394:
1.98 schwarze 395: p = mandoc_calloc(1, sizeof(struct roffnode));
1.67 kristaps 396: p->tok = tok;
1.106 kristaps 397: if (name)
398: p->name = mandoc_strdup(name);
1.67 kristaps 399: p->parent = r->last;
400: p->line = line;
401: p->col = col;
1.198 schwarze 402: p->rule = p->parent ? p->parent->rule : 0;
1.67 kristaps 403:
404: r->last = p;
405: }
406:
407:
408: static void
409: roff_free1(struct roff *r)
410: {
1.176 schwarze 411: struct tbl_node *tbl;
1.125 kristaps 412: struct eqn_node *e;
1.167 kristaps 413: int i;
1.67 kristaps 414:
1.176 schwarze 415: while (NULL != (tbl = r->first_tbl)) {
416: r->first_tbl = tbl->next;
417: tbl_free(tbl);
1.109 kristaps 418: }
419:
1.113 kristaps 420: r->first_tbl = r->last_tbl = r->tbl = NULL;
421:
1.125 kristaps 422: while (NULL != (e = r->first_eqn)) {
423: r->first_eqn = e->next;
424: eqn_free(e);
425: }
426:
427: r->first_eqn = r->last_eqn = r->eqn = NULL;
428:
1.67 kristaps 429: while (r->last)
430: roffnode_pop(r);
1.109 kristaps 431:
1.167 kristaps 432: roff_freestr(r->strtab);
433: roff_freestr(r->xmbtab);
434:
435: r->strtab = r->xmbtab = NULL;
436:
1.180 schwarze 437: roff_freereg(r->regtab);
438:
439: r->regtab = NULL;
440:
1.167 kristaps 441: if (r->xtab)
442: for (i = 0; i < 128; i++)
443: free(r->xtab[i].p);
444:
445: free(r->xtab);
446: r->xtab = NULL;
1.67 kristaps 447: }
448:
449: void
450: roff_reset(struct roff *r)
451: {
452:
453: roff_free1(r);
1.174 kristaps 454: r->control = 0;
1.67 kristaps 455: }
456:
457:
458: void
459: roff_free(struct roff *r)
460: {
461:
462: roff_free1(r);
463: free(r);
464: }
465:
466:
467: struct roff *
1.199 schwarze 468: roff_alloc(struct mparse *parse, int options)
1.67 kristaps 469: {
470: struct roff *r;
471:
1.98 schwarze 472: r = mandoc_calloc(1, sizeof(struct roff));
1.128 kristaps 473: r->parse = parse;
1.199 schwarze 474: r->options = options;
1.82 kristaps 475: r->rstackpos = -1;
1.85 kristaps 476:
1.155 kristaps 477: roffhash_init();
1.141 kristaps 478:
1.67 kristaps 479: return(r);
480: }
481:
1.94 kristaps 482: /*
1.181 schwarze 483: * In the current line, expand user-defined strings ("\*")
484: * and references to number registers ("\n").
485: * Also check the syntax of other escape sequences.
1.154 kristaps 486: */
1.172 schwarze 487: static enum rofferr
1.142 kristaps 488: roff_res(struct roff *r, char **bufp, size_t *szp, int ln, int pos)
1.94 kristaps 489: {
1.181 schwarze 490: char ubuf[12]; /* buffer to print the number */
1.108 schwarze 491: const char *stesc; /* start of an escape sequence ('\\') */
492: const char *stnam; /* start of the name, after "[(*" */
493: const char *cp; /* end of the name, e.g. before ']' */
494: const char *res; /* the string to be substituted */
1.181 schwarze 495: char *nbuf; /* new buffer to copy bufp to */
496: size_t nsz; /* size of the new buffer */
497: size_t maxl; /* expected length of the escape name */
498: size_t naml; /* actual length of the escape name */
499: int expand_count; /* to avoid infinite loops */
1.94 kristaps 500:
1.170 schwarze 501: expand_count = 0;
502:
1.154 kristaps 503: again:
1.108 schwarze 504: cp = *bufp + pos;
505: while (NULL != (cp = strchr(cp, '\\'))) {
506: stesc = cp++;
507:
508: /*
1.181 schwarze 509: * The second character must be an asterisk or an n.
1.108 schwarze 510: * If it isn't, skip it anyway: It is escaped,
511: * so it can't start another escape sequence.
512: */
513:
514: if ('\0' == *cp)
1.172 schwarze 515: return(ROFF_CONT);
1.152 kristaps 516:
1.181 schwarze 517: switch (*cp) {
518: case ('*'):
519: res = NULL;
520: break;
521: case ('n'):
522: res = ubuf;
523: break;
524: default:
525: if (ESCAPE_ERROR != mandoc_escape(&cp, NULL, NULL))
1.152 kristaps 526: continue;
1.153 kristaps 527: mandoc_msg
528: (MANDOCERR_BADESCAPE, r->parse,
529: ln, (int)(stesc - *bufp), NULL);
1.172 schwarze 530: return(ROFF_CONT);
1.152 kristaps 531: }
532:
533: cp++;
1.108 schwarze 534:
535: /*
536: * The third character decides the length
1.181 schwarze 537: * of the name of the string or register.
1.108 schwarze 538: * Save a pointer to the name.
539: */
540:
1.94 kristaps 541: switch (*cp) {
1.108 schwarze 542: case ('\0'):
1.172 schwarze 543: return(ROFF_CONT);
1.94 kristaps 544: case ('('):
545: cp++;
546: maxl = 2;
547: break;
548: case ('['):
549: cp++;
550: maxl = 0;
551: break;
552: default:
553: maxl = 1;
554: break;
555: }
1.108 schwarze 556: stnam = cp;
1.94 kristaps 557:
1.108 schwarze 558: /* Advance to the end of the name. */
1.94 kristaps 559:
1.181 schwarze 560: for (naml = 0; 0 == maxl || naml < maxl; naml++, cp++) {
1.153 kristaps 561: if ('\0' == *cp) {
562: mandoc_msg
563: (MANDOCERR_BADESCAPE,
564: r->parse, ln,
565: (int)(stesc - *bufp), NULL);
1.172 schwarze 566: return(ROFF_CONT);
1.153 kristaps 567: }
1.94 kristaps 568: if (0 == maxl && ']' == *cp)
569: break;
570: }
571:
1.108 schwarze 572: /*
573: * Retrieve the replacement string; if it is
574: * undefined, resume searching for escapes.
575: */
576:
1.181 schwarze 577: if (NULL == res)
578: res = roff_getstrn(r, stnam, naml);
579: else
580: snprintf(ubuf, sizeof(ubuf), "%d",
581: roff_getregn(r, stnam, naml));
1.94 kristaps 582:
583: if (NULL == res) {
1.153 kristaps 584: mandoc_msg
585: (MANDOCERR_BADESCAPE, r->parse,
586: ln, (int)(stesc - *bufp), NULL);
1.142 kristaps 587: res = "";
1.94 kristaps 588: }
589:
1.108 schwarze 590: /* Replace the escape sequence by the string. */
591:
1.161 kristaps 592: pos = stesc - *bufp;
1.154 kristaps 593:
1.94 kristaps 594: nsz = *szp + strlen(res) + 1;
1.181 schwarze 595: nbuf = mandoc_malloc(nsz);
1.94 kristaps 596:
1.181 schwarze 597: strlcpy(nbuf, *bufp, (size_t)(stesc - *bufp + 1));
598: strlcat(nbuf, res, nsz);
599: strlcat(nbuf, cp + (maxl ? 0 : 1), nsz);
1.94 kristaps 600:
601: free(*bufp);
602:
1.181 schwarze 603: *bufp = nbuf;
1.94 kristaps 604: *szp = nsz;
1.170 schwarze 605:
606: if (EXPAND_LIMIT >= ++expand_count)
607: goto again;
608:
609: /* Just leave the string unexpanded. */
610: mandoc_msg(MANDOCERR_ROFFLOOP, r->parse, ln, pos, NULL);
1.172 schwarze 611: return(ROFF_IGN);
1.154 kristaps 612: }
1.172 schwarze 613: return(ROFF_CONT);
1.154 kristaps 614: }
615:
616: /*
1.178 schwarze 617: * Process text streams:
618: * Convert all breakable hyphens into ASCII_HYPH.
619: * Decrement and spring input line trap.
1.154 kristaps 620: */
621: static enum rofferr
1.178 schwarze 622: roff_parsetext(char **bufp, size_t *szp, int pos, int *offs)
1.154 kristaps 623: {
624: size_t sz;
625: const char *start;
1.178 schwarze 626: char *p;
627: int isz;
1.154 kristaps 628: enum mandoc_esc esc;
629:
1.178 schwarze 630: start = p = *bufp + pos;
1.154 kristaps 631:
632: while ('\0' != *p) {
633: sz = strcspn(p, "-\\");
634: p += sz;
635:
1.159 kristaps 636: if ('\0' == *p)
637: break;
638:
1.154 kristaps 639: if ('\\' == *p) {
640: /* Skip over escapes. */
641: p++;
1.189 schwarze 642: esc = mandoc_escape((const char **)&p, NULL, NULL);
1.154 kristaps 643: if (ESCAPE_ERROR == esc)
644: break;
1.155 kristaps 645: continue;
1.159 kristaps 646: } else if (p == start) {
1.158 kristaps 647: p++;
1.155 kristaps 648: continue;
1.158 kristaps 649: }
1.155 kristaps 650:
1.171 schwarze 651: if (isalpha((unsigned char)p[-1]) &&
652: isalpha((unsigned char)p[1]))
1.155 kristaps 653: *p = ASCII_HYPH;
654: p++;
1.94 kristaps 655: }
656:
1.178 schwarze 657: /* Spring the input line trap. */
658: if (1 == roffit_lines) {
1.202 ! schwarze 659: isz = mandoc_asprintf(&p, "%s\n.%s", *bufp, roffit_macro);
1.178 schwarze 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;
1.154 kristaps 669: return(ROFF_CONT);
1.94 kristaps 670: }
671:
1.67 kristaps 672: enum rofferr
1.90 kristaps 673: roff_parseln(struct roff *r, int ln, char **bufp,
674: size_t *szp, int pos, int *offs)
1.67 kristaps 675: {
676: enum rofft t;
1.109 kristaps 677: enum rofferr e;
1.130 kristaps 678: int ppos, ctl;
1.79 kristaps 679:
680: /*
1.94 kristaps 681: * Run the reserved-word filter only if we have some reserved
682: * words to fill in.
683: */
684:
1.172 schwarze 685: e = roff_res(r, bufp, szp, ln, pos);
686: if (ROFF_IGN == e)
687: return(e);
688: assert(ROFF_CONT == e);
1.94 kristaps 689:
1.130 kristaps 690: ppos = pos;
1.174 kristaps 691: ctl = roff_getcontrol(r, *bufp, &pos);
1.130 kristaps 692:
1.94 kristaps 693: /*
1.79 kristaps 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.
1.125 kristaps 697: * Finally, if there's an equation scope open, divert it into it
698: * no matter our state.
1.79 kristaps 699: */
1.74 kristaps 700:
1.130 kristaps 701: if (r->last && ! ctl) {
1.78 kristaps 702: t = r->last->tok;
703: assert(roffs[t].text);
1.109 kristaps 704: e = (*roffs[t].text)
705: (r, t, bufp, szp, ln, pos, pos, offs);
706: assert(ROFF_IGN == e || ROFF_CONT == e);
1.125 kristaps 707: if (ROFF_CONT != e)
708: return(e);
1.182 schwarze 709: }
710: if (r->eqn)
711: return(eqn_read(&r->eqn, ln, *bufp, ppos, offs));
712: if ( ! ctl) {
1.125 kristaps 713: if (r->tbl)
1.130 kristaps 714: return(tbl_read(r->tbl, ln, *bufp, pos));
1.178 schwarze 715: return(roff_parsetext(bufp, szp, pos, offs));
1.182 schwarze 716: }
1.67 kristaps 717:
1.79 kristaps 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.
1.125 kristaps 721: * Don't do so if an equation is open.
1.79 kristaps 722: */
1.78 kristaps 723:
1.79 kristaps 724: if (r->last) {
725: t = r->last->tok;
726: assert(roffs[t].sub);
727: return((*roffs[t].sub)
1.90 kristaps 728: (r, t, bufp, szp,
1.130 kristaps 729: ln, ppos, pos, offs));
1.79 kristaps 730: }
1.78 kristaps 731:
1.79 kristaps 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: */
1.67 kristaps 737:
1.106 kristaps 738: if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos)))
1.79 kristaps 739: return(ROFF_CONT);
1.67 kristaps 740:
1.75 kristaps 741: assert(roffs[t].proc);
1.78 kristaps 742: return((*roffs[t].proc)
1.90 kristaps 743: (r, t, bufp, szp,
744: ln, ppos, pos, offs));
1.74 kristaps 745: }
746:
747:
1.117 kristaps 748: void
1.74 kristaps 749: roff_endparse(struct roff *r)
750: {
751:
1.110 kristaps 752: if (r->last)
1.128 kristaps 753: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.109 kristaps 754: r->last->line, r->last->col, NULL);
1.117 kristaps 755:
1.125 kristaps 756: if (r->eqn) {
1.128 kristaps 757: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.148 kristaps 758: r->eqn->eqn.ln, r->eqn->eqn.pos, NULL);
1.151 kristaps 759: eqn_end(&r->eqn);
1.125 kristaps 760: }
761:
1.117 kristaps 762: if (r->tbl) {
1.128 kristaps 763: mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.117 kristaps 764: r->tbl->line, r->tbl->pos, NULL);
1.151 kristaps 765: tbl_end(&r->tbl);
1.117 kristaps 766: }
1.67 kristaps 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
1.106 kristaps 774: roff_parse(struct roff *r, const char *buf, int *pos)
1.67 kristaps 775: {
1.106 kristaps 776: const char *mac;
777: size_t maclen;
1.67 kristaps 778: enum rofft t;
779:
1.144 kristaps 780: if ('\0' == buf[*pos] || '"' == buf[*pos] ||
781: '\t' == buf[*pos] || ' ' == buf[*pos])
1.67 kristaps 782: return(ROFF_MAX);
783:
1.195 schwarze 784: /* We stop the macro parse at an escape, tab, space, or nil. */
1.144 kristaps 785:
1.106 kristaps 786: mac = buf + *pos;
1.195 schwarze 787: maclen = strcspn(mac, " \\\t\0");
1.67 kristaps 788:
1.106 kristaps 789: t = (r->current_string = roff_getstrn(r, mac, maclen))
1.155 kristaps 790: ? ROFF_USERDEF : roffhash_find(mac, maclen);
1.67 kristaps 791:
1.127 kristaps 792: *pos += (int)maclen;
1.130 kristaps 793:
1.67 kristaps 794: while (buf[*pos] && ' ' == buf[*pos])
795: (*pos)++;
796:
797: return(t);
798: }
799:
800: /* ARGSUSED */
801: static enum rofferr
1.76 kristaps 802: roff_cblock(ROFF_ARGS)
1.67 kristaps 803: {
804:
1.79 kristaps 805: /*
806: * A block-close `..' should only be invoked as a child of an
807: * ignore macro, otherwise raise a warning and just ignore it.
808: */
809:
1.76 kristaps 810: if (NULL == r->last) {
1.128 kristaps 811: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.76 kristaps 812: return(ROFF_IGN);
813: }
1.67 kristaps 814:
1.81 kristaps 815: switch (r->last->tok) {
816: case (ROFF_am):
817: /* FALLTHROUGH */
818: case (ROFF_ami):
819: /* FALLTHROUGH */
820: case (ROFF_am1):
821: /* FALLTHROUGH */
822: case (ROFF_de):
1.108 schwarze 823: /* ROFF_de1 is remapped to ROFF_de in roff_block(). */
1.81 kristaps 824: /* FALLTHROUGH */
825: case (ROFF_dei):
826: /* FALLTHROUGH */
827: case (ROFF_ig):
828: break;
829: default:
1.128 kristaps 830: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.67 kristaps 831: return(ROFF_IGN);
1.76 kristaps 832: }
1.67 kristaps 833:
1.76 kristaps 834: if ((*bufp)[pos])
1.128 kristaps 835: mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.71 kristaps 836:
837: roffnode_pop(r);
1.76 kristaps 838: roffnode_cleanscope(r);
839: return(ROFF_IGN);
1.71 kristaps 840:
1.67 kristaps 841: }
842:
843:
1.76 kristaps 844: static void
845: roffnode_cleanscope(struct roff *r)
1.67 kristaps 846: {
847:
1.76 kristaps 848: while (r->last) {
1.173 schwarze 849: if (--r->last->endspan != 0)
1.76 kristaps 850: break;
851: roffnode_pop(r);
852: }
1.67 kristaps 853: }
854:
855:
1.195 schwarze 856: static void
857: roff_ccond(struct roff *r, int ln, int ppos)
1.74 kristaps 858: {
859:
1.76 kristaps 860: if (NULL == r->last) {
1.128 kristaps 861: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.195 schwarze 862: return;
1.76 kristaps 863: }
864:
1.82 kristaps 865: switch (r->last->tok) {
866: case (ROFF_el):
867: /* FALLTHROUGH */
868: case (ROFF_ie):
869: /* FALLTHROUGH */
870: case (ROFF_if):
871: break;
872: default:
1.128 kristaps 873: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.195 schwarze 874: return;
1.75 kristaps 875: }
876:
1.76 kristaps 877: if (r->last->endspan > -1) {
1.128 kristaps 878: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.195 schwarze 879: return;
1.76 kristaps 880: }
881:
1.75 kristaps 882: roffnode_pop(r);
1.76 kristaps 883: roffnode_cleanscope(r);
1.195 schwarze 884: return;
1.76 kristaps 885: }
886:
1.75 kristaps 887:
1.76 kristaps 888: /* ARGSUSED */
889: static enum rofferr
1.80 kristaps 890: roff_block(ROFF_ARGS)
1.76 kristaps 891: {
1.78 kristaps 892: int sv;
893: size_t sz;
1.106 kristaps 894: char *name;
895:
896: name = NULL;
1.76 kristaps 897:
1.106 kristaps 898: if (ROFF_ig != tok) {
899: if ('\0' == (*bufp)[pos]) {
1.128 kristaps 900: mandoc_msg(MANDOCERR_NOARGS, r->parse, ln, ppos, NULL);
1.106 kristaps 901: return(ROFF_IGN);
902: }
1.107 kristaps 903:
904: /*
905: * Re-write `de1', since we don't really care about
906: * groff's strange compatibility mode, into `de'.
907: */
908:
1.106 kristaps 909: if (ROFF_de1 == tok)
910: tok = ROFF_de;
911: if (ROFF_de == tok)
912: name = *bufp + pos;
913: else
1.128 kristaps 914: mandoc_msg(MANDOCERR_REQUEST, r->parse, ln, ppos,
1.106 kristaps 915: roffs[tok].name);
1.107 kristaps 916:
1.131 schwarze 917: while ((*bufp)[pos] && ! isspace((unsigned char)(*bufp)[pos]))
1.80 kristaps 918: pos++;
1.107 kristaps 919:
1.131 schwarze 920: while (isspace((unsigned char)(*bufp)[pos]))
1.106 kristaps 921: (*bufp)[pos++] = '\0';
1.80 kristaps 922: }
923:
1.106 kristaps 924: roffnode_push(r, tok, name, ln, ppos);
925:
926: /*
927: * At the beginning of a `de' macro, clear the existing string
928: * with the same name, if there is one. New content will be
1.193 schwarze 929: * appended from roff_block_text() in multiline mode.
1.106 kristaps 930: */
1.107 kristaps 931:
1.106 kristaps 932: if (ROFF_de == tok)
1.108 schwarze 933: roff_setstr(r, name, "", 0);
1.76 kristaps 934:
1.79 kristaps 935: if ('\0' == (*bufp)[pos])
1.78 kristaps 936: return(ROFF_IGN);
937:
1.107 kristaps 938: /* If present, process the custom end-of-line marker. */
939:
1.78 kristaps 940: sv = pos;
1.131 schwarze 941: while ((*bufp)[pos] && ! isspace((unsigned char)(*bufp)[pos]))
1.78 kristaps 942: pos++;
943:
944: /*
945: * Note: groff does NOT like escape characters in the input.
946: * Instead of detecting this, we're just going to let it fly and
947: * to hell with it.
948: */
949:
950: assert(pos > sv);
951: sz = (size_t)(pos - sv);
952:
1.79 kristaps 953: if (1 == sz && '.' == (*bufp)[sv])
954: return(ROFF_IGN);
955:
1.98 schwarze 956: r->last->end = mandoc_malloc(sz + 1);
1.78 kristaps 957:
958: memcpy(r->last->end, *bufp + sv, sz);
959: r->last->end[(int)sz] = '\0';
960:
1.77 kristaps 961: if ((*bufp)[pos])
1.128 kristaps 962: mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.74 kristaps 963:
1.78 kristaps 964: return(ROFF_IGN);
965: }
966:
967:
968: /* ARGSUSED */
969: static enum rofferr
1.80 kristaps 970: roff_block_sub(ROFF_ARGS)
1.79 kristaps 971: {
972: enum rofft t;
973: int i, j;
974:
975: /*
976: * First check whether a custom macro exists at this level. If
977: * it does, then check against it. This is some of groff's
978: * stranger behaviours. If we encountered a custom end-scope
979: * tag and that tag also happens to be a "real" macro, then we
980: * need to try interpreting it again as a real macro. If it's
981: * not, then return ignore. Else continue.
982: */
983:
984: if (r->last->end) {
1.130 kristaps 985: for (i = pos, j = 0; r->last->end[j]; j++, i++)
1.79 kristaps 986: if ((*bufp)[i] != r->last->end[j])
987: break;
988:
989: if ('\0' == r->last->end[j] &&
990: ('\0' == (*bufp)[i] ||
991: ' ' == (*bufp)[i] ||
992: '\t' == (*bufp)[i])) {
993: roffnode_pop(r);
994: roffnode_cleanscope(r);
995:
1.130 kristaps 996: while (' ' == (*bufp)[i] || '\t' == (*bufp)[i])
997: i++;
998:
999: pos = i;
1.106 kristaps 1000: if (ROFF_MAX != roff_parse(r, *bufp, &pos))
1.79 kristaps 1001: return(ROFF_RERUN);
1002: return(ROFF_IGN);
1003: }
1004: }
1005:
1006: /*
1007: * If we have no custom end-query or lookup failed, then try
1008: * pulling it out of the hashtable.
1009: */
1010:
1.137 schwarze 1011: t = roff_parse(r, *bufp, &pos);
1.79 kristaps 1012:
1.106 kristaps 1013: /*
1014: * Macros other than block-end are only significant
1015: * in `de' blocks; elsewhere, simply throw them away.
1016: */
1017: if (ROFF_cblock != t) {
1018: if (ROFF_de == tok)
1.193 schwarze 1019: roff_setstr(r, r->last->name, *bufp + ppos, 2);
1.79 kristaps 1020: return(ROFF_IGN);
1.106 kristaps 1021: }
1.79 kristaps 1022:
1023: assert(roffs[t].proc);
1.90 kristaps 1024: return((*roffs[t].proc)(r, t, bufp, szp,
1025: ln, ppos, pos, offs));
1.79 kristaps 1026: }
1027:
1028:
1029: /* ARGSUSED */
1030: static enum rofferr
1.80 kristaps 1031: roff_block_text(ROFF_ARGS)
1.78 kristaps 1032: {
1033:
1.106 kristaps 1034: if (ROFF_de == tok)
1.193 schwarze 1035: roff_setstr(r, r->last->name, *bufp + pos, 2);
1.106 kristaps 1036:
1.78 kristaps 1037: return(ROFF_IGN);
1038: }
1039:
1040:
1041: /* ARGSUSED */
1042: static enum rofferr
1.82 kristaps 1043: roff_cond_sub(ROFF_ARGS)
1044: {
1045: enum rofft t;
1.139 kristaps 1046: char *ep;
1.198 schwarze 1047: int rr;
1.82 kristaps 1048:
1049: rr = r->last->rule;
1.139 kristaps 1050: roffnode_cleanscope(r);
1.177 schwarze 1051: t = roff_parse(r, *bufp, &pos);
1.82 kristaps 1052:
1.139 kristaps 1053: /*
1.177 schwarze 1054: * Fully handle known macros when they are structurally
1055: * required or when the conditional evaluated to true.
1.87 kristaps 1056: */
1057:
1.177 schwarze 1058: if ((ROFF_MAX != t) &&
1.198 schwarze 1059: (rr || ROFFMAC_STRUCT & roffs[t].flags)) {
1.177 schwarze 1060: assert(roffs[t].proc);
1061: return((*roffs[t].proc)(r, t, bufp, szp,
1062: ln, ppos, pos, offs));
1063: }
1.144 kristaps 1064:
1.196 schwarze 1065: /*
1066: * If `\}' occurs on a macro line without a preceding macro,
1067: * drop the line completely.
1068: */
1069:
1070: ep = *bufp + pos;
1071: if ('\\' == ep[0] && '}' == ep[1])
1.198 schwarze 1072: rr = 0;
1.196 schwarze 1073:
1.177 schwarze 1074: /* Always check for the closing delimiter `\}'. */
1.144 kristaps 1075:
1.177 schwarze 1076: while (NULL != (ep = strchr(ep, '\\'))) {
1.197 schwarze 1077: if ('}' == *(++ep)) {
1078: *ep = '&';
1079: roff_ccond(r, ln, ep - *bufp - 1);
1080: }
1081: ++ep;
1.177 schwarze 1082: }
1.198 schwarze 1083: return(rr ? ROFF_CONT : ROFF_IGN);
1.82 kristaps 1084: }
1085:
1086: /* ARGSUSED */
1087: static enum rofferr
1088: roff_cond_text(ROFF_ARGS)
1.78 kristaps 1089: {
1.140 kristaps 1090: char *ep;
1.198 schwarze 1091: int rr;
1.82 kristaps 1092:
1093: rr = r->last->rule;
1.140 kristaps 1094: roffnode_cleanscope(r);
1.82 kristaps 1095:
1.197 schwarze 1096: ep = *bufp + pos;
1097: while (NULL != (ep = strchr(ep, '\\'))) {
1098: if ('}' == *(++ep)) {
1099: *ep = '&';
1100: roff_ccond(r, ln, ep - *bufp - 1);
1101: }
1102: ++ep;
1.78 kristaps 1103: }
1.198 schwarze 1104: return(rr ? ROFF_CONT : ROFF_IGN);
1.74 kristaps 1105: }
1106:
1.184 schwarze 1107: static int
1108: roff_getnum(const char *v, int *pos, int *res)
1109: {
1110: int p, n;
1111:
1112: p = *pos;
1113: n = v[p] == '-';
1114: if (n)
1115: p++;
1116:
1117: for (*res = 0; isdigit((unsigned char)v[p]); p++)
1118: *res += 10 * *res + v[p] - '0';
1119: if (p == *pos + n)
1120: return 0;
1121:
1122: if (n)
1123: *res = -*res;
1124:
1125: *pos = p;
1126: return 1;
1127: }
1128:
1129: static int
1130: roff_getop(const char *v, int *pos, char *res)
1131: {
1132: int e;
1133:
1134: *res = v[*pos];
1135: e = v[*pos + 1] == '=';
1136:
1137: switch (*res) {
1138: case '=':
1139: break;
1140: case '>':
1141: if (e)
1142: *res = 'g';
1143: break;
1144: case '<':
1145: if (e)
1146: *res = 'l';
1147: break;
1148: default:
1149: return(0);
1150: }
1151:
1152: *pos += 1 + e;
1153:
1154: return(*res);
1155: }
1156:
1.198 schwarze 1157: /*
1158: * Evaluate a string comparison condition.
1159: * The first character is the delimiter.
1160: * Succeed if the string up to its second occurrence
1161: * matches the string up to its third occurence.
1162: * Advance the cursor after the third occurrence
1163: * or lacking that, to the end of the line.
1164: */
1165: static int
1166: roff_evalstrcond(const char *v, int *pos)
1167: {
1168: const char *s1, *s2, *s3;
1169: int match;
1170:
1171: match = 0;
1172: s1 = v + *pos; /* initial delimiter */
1173: s2 = s1 + 1; /* for scanning the first string */
1174: s3 = strchr(s2, *s1); /* for scanning the second string */
1175:
1176: if (NULL == s3) /* found no middle delimiter */
1177: goto out;
1178:
1179: while ('\0' != *++s3) {
1180: if (*s2 != *s3) { /* mismatch */
1181: s3 = strchr(s3, *s1);
1182: break;
1183: }
1184: if (*s3 == *s1) { /* found the final delimiter */
1185: match = 1;
1186: break;
1187: }
1188: s2++;
1189: }
1190:
1191: out:
1192: if (NULL == s3)
1193: s3 = strchr(s2, '\0');
1194: else
1195: s3++;
1196: *pos = s3 - v;
1197: return(match);
1198: }
1199:
1200: static int
1.88 kristaps 1201: roff_evalcond(const char *v, int *pos)
1202: {
1.198 schwarze 1203: int wanttrue, lh, rh;
1.184 schwarze 1204: char op;
1.88 kristaps 1205:
1.198 schwarze 1206: if ('!' == v[*pos]) {
1207: wanttrue = 0;
1208: (*pos)++;
1209: } else
1210: wanttrue = 1;
1211:
1.88 kristaps 1212: switch (v[*pos]) {
1213: case ('n'):
1.198 schwarze 1214: /* FALLTHROUGH */
1215: case ('o'):
1.88 kristaps 1216: (*pos)++;
1.198 schwarze 1217: return(wanttrue);
1218: case ('c'):
1219: /* FALLTHROUGH */
1220: case ('d'):
1221: /* FALLTHROUGH */
1.88 kristaps 1222: case ('e'):
1223: /* FALLTHROUGH */
1.198 schwarze 1224: case ('r'):
1.88 kristaps 1225: /* FALLTHROUGH */
1226: case ('t'):
1227: (*pos)++;
1.198 schwarze 1228: return(!wanttrue);
1.88 kristaps 1229: default:
1230: break;
1231: }
1232:
1.184 schwarze 1233: if (!roff_getnum(v, pos, &lh))
1.198 schwarze 1234: return(roff_evalstrcond(v, pos) == wanttrue);
1235: if (!roff_getop(v, pos, &op))
1236: return((lh > 0) == wanttrue);
1.184 schwarze 1237: if (!roff_getnum(v, pos, &rh))
1.198 schwarze 1238: return(0);
1239:
1.184 schwarze 1240: switch (op) {
1241: case 'g':
1.198 schwarze 1242: return((lh >= rh) == wanttrue);
1.184 schwarze 1243: case 'l':
1.198 schwarze 1244: return((lh <= rh) == wanttrue);
1.184 schwarze 1245: case '=':
1.198 schwarze 1246: return((lh == rh) == wanttrue);
1.184 schwarze 1247: case '>':
1.198 schwarze 1248: return((lh > rh) == wanttrue);
1.184 schwarze 1249: case '<':
1.198 schwarze 1250: return((lh < rh) == wanttrue);
1.184 schwarze 1251: default:
1.198 schwarze 1252: return(0);
1.184 schwarze 1253: }
1.88 kristaps 1254: }
1255:
1.75 kristaps 1256: /* ARGSUSED */
1.74 kristaps 1257: static enum rofferr
1.103 kristaps 1258: roff_line_ignore(ROFF_ARGS)
1.89 kristaps 1259: {
1.123 schwarze 1260:
1.89 kristaps 1261: return(ROFF_IGN);
1262: }
1263:
1.104 kristaps 1264: /* ARGSUSED */
1265: static enum rofferr
1.82 kristaps 1266: roff_cond(ROFF_ARGS)
1.74 kristaps 1267: {
1.173 schwarze 1268:
1269: roffnode_push(r, tok, NULL, ln, ppos);
1.74 kristaps 1270:
1.134 kristaps 1271: /*
1272: * An `.el' has no conditional body: it will consume the value
1273: * of the current rstack entry set in prior `ie' calls or
1274: * defaults to DENY.
1275: *
1276: * If we're not an `el', however, then evaluate the conditional.
1277: */
1.133 kristaps 1278:
1.173 schwarze 1279: r->last->rule = ROFF_el == tok ?
1.198 schwarze 1280: (r->rstackpos < 0 ? 0 : r->rstack[r->rstackpos--]) :
1.134 kristaps 1281: roff_evalcond(*bufp, &pos);
1.77 kristaps 1282:
1.134 kristaps 1283: /*
1284: * An if-else will put the NEGATION of the current evaluated
1285: * conditional into the stack of rules.
1286: */
1287:
1.84 schwarze 1288: if (ROFF_ie == tok) {
1.134 kristaps 1289: if (r->rstackpos == RSTACK_MAX - 1) {
1290: mandoc_msg(MANDOCERR_MEM,
1291: r->parse, ln, ppos, NULL);
1292: return(ROFF_ERR);
1293: }
1.198 schwarze 1294: r->rstack[++r->rstackpos] = !r->last->rule;
1.82 kristaps 1295: }
1.88 kristaps 1296:
1297: /* If the parent has false as its rule, then so do we. */
1298:
1.198 schwarze 1299: if (r->last->parent && !r->last->parent->rule)
1300: r->last->rule = 0;
1.88 kristaps 1301:
1302: /*
1.173 schwarze 1303: * Determine scope.
1304: * If there is nothing on the line after the conditional,
1305: * not even whitespace, use next-line scope.
1.88 kristaps 1306: */
1.74 kristaps 1307:
1.173 schwarze 1308: if ('\0' == (*bufp)[pos]) {
1309: r->last->endspan = 2;
1310: goto out;
1311: }
1312:
1313: while (' ' == (*bufp)[pos])
1314: pos++;
1315:
1316: /* An opening brace requests multiline scope. */
1.75 kristaps 1317:
1318: if ('\\' == (*bufp)[pos] && '{' == (*bufp)[pos + 1]) {
1319: r->last->endspan = -1;
1320: pos += 2;
1.173 schwarze 1321: goto out;
1.109 kristaps 1322: }
1.74 kristaps 1323:
1.77 kristaps 1324: /*
1.173 schwarze 1325: * Anything else following the conditional causes
1326: * single-line scope. Warn if the scope contains
1327: * nothing but trailing whitespace.
1.77 kristaps 1328: */
1329:
1.75 kristaps 1330: if ('\0' == (*bufp)[pos])
1.173 schwarze 1331: mandoc_msg(MANDOCERR_NOARGS, r->parse, ln, ppos, NULL);
1.77 kristaps 1332:
1.173 schwarze 1333: r->last->endspan = 1;
1.74 kristaps 1334:
1.173 schwarze 1335: out:
1.75 kristaps 1336: *offs = pos;
1337: return(ROFF_RERUN);
1.83 schwarze 1338: }
1339:
1340:
1341: /* ARGSUSED */
1342: static enum rofferr
1.92 schwarze 1343: roff_ds(ROFF_ARGS)
1344: {
1.96 kristaps 1345: char *name, *string;
1346:
1347: /*
1348: * A symbol is named by the first word following the macro
1349: * invocation up to a space. Its value is anything after the
1350: * name's trailing whitespace and optional double-quote. Thus,
1351: *
1352: * [.ds foo "bar " ]
1353: *
1354: * will have `bar " ' as its value.
1355: */
1.92 schwarze 1356:
1.121 schwarze 1357: string = *bufp + pos;
1358: name = roff_getname(r, &string, ln, pos);
1.92 schwarze 1359: if ('\0' == *name)
1360: return(ROFF_IGN);
1361:
1.121 schwarze 1362: /* Read past initial double-quote. */
1363: if ('"' == *string)
1.92 schwarze 1364: string++;
1365:
1.96 kristaps 1366: /* The rest is the value. */
1.193 schwarze 1367: roff_setstr(r, name, string, ROFF_as == tok);
1.92 schwarze 1368: return(ROFF_IGN);
1369: }
1370:
1.180 schwarze 1371: void
1.187 schwarze 1372: roff_setreg(struct roff *r, const char *name, int val, char sign)
1.147 kristaps 1373: {
1.180 schwarze 1374: struct roffreg *reg;
1375:
1376: /* Search for an existing register with the same name. */
1377: reg = r->regtab;
1378:
1379: while (reg && strcmp(name, reg->key.p))
1380: reg = reg->next;
1.147 kristaps 1381:
1.180 schwarze 1382: if (NULL == reg) {
1383: /* Create a new register. */
1384: reg = mandoc_malloc(sizeof(struct roffreg));
1385: reg->key.p = mandoc_strdup(name);
1386: reg->key.sz = strlen(name);
1.187 schwarze 1387: reg->val = 0;
1.180 schwarze 1388: reg->next = r->regtab;
1389: r->regtab = reg;
1390: }
1391:
1.187 schwarze 1392: if ('+' == sign)
1393: reg->val += val;
1394: else if ('-' == sign)
1395: reg->val -= val;
1396: else
1397: reg->val = val;
1.147 kristaps 1398: }
1399:
1.192 schwarze 1400: /*
1401: * Handle some predefined read-only number registers.
1402: * For now, return -1 if the requested register is not predefined;
1403: * in case a predefined read-only register having the value -1
1404: * were to turn up, another special value would have to be chosen.
1405: */
1406: static int
1407: roff_getregro(const char *name)
1408: {
1409:
1410: switch (*name) {
1411: case ('A'): /* ASCII approximation mode is always off. */
1412: return(0);
1413: case ('g'): /* Groff compatibility mode is always on. */
1414: return(1);
1415: case ('H'): /* Fixed horizontal resolution. */
1416: return (24);
1417: case ('j'): /* Always adjust left margin only. */
1418: return(0);
1419: case ('T'): /* Some output device is always defined. */
1420: return(1);
1421: case ('V'): /* Fixed vertical resolution. */
1422: return (40);
1423: default:
1424: return (-1);
1425: }
1426: }
1427:
1.181 schwarze 1428: int
1.180 schwarze 1429: roff_getreg(const struct roff *r, const char *name)
1.147 kristaps 1430: {
1.180 schwarze 1431: struct roffreg *reg;
1.192 schwarze 1432: int val;
1433:
1434: if ('.' == name[0] && '\0' != name[1] && '\0' == name[2]) {
1435: val = roff_getregro(name + 1);
1436: if (-1 != val)
1437: return (val);
1438: }
1.180 schwarze 1439:
1440: for (reg = r->regtab; reg; reg = reg->next)
1441: if (0 == strcmp(name, reg->key.p))
1.181 schwarze 1442: return(reg->val);
1443:
1444: return(0);
1445: }
1446:
1447: static int
1448: roff_getregn(const struct roff *r, const char *name, size_t len)
1449: {
1450: struct roffreg *reg;
1.192 schwarze 1451: int val;
1452:
1453: if ('.' == name[0] && 2 == len) {
1454: val = roff_getregro(name + 1);
1455: if (-1 != val)
1456: return (val);
1457: }
1.181 schwarze 1458:
1459: for (reg = r->regtab; reg; reg = reg->next)
1460: if (len == reg->key.sz &&
1461: 0 == strncmp(name, reg->key.p, len))
1462: return(reg->val);
1.147 kristaps 1463:
1.180 schwarze 1464: return(0);
1.147 kristaps 1465: }
1466:
1.180 schwarze 1467: static void
1468: roff_freereg(struct roffreg *reg)
1.147 kristaps 1469: {
1.180 schwarze 1470: struct roffreg *old_reg;
1.147 kristaps 1471:
1.180 schwarze 1472: while (NULL != reg) {
1473: free(reg->key.p);
1474: old_reg = reg;
1475: reg = reg->next;
1476: free(old_reg);
1477: }
1.147 kristaps 1478: }
1.92 schwarze 1479:
1480: /* ARGSUSED */
1481: static enum rofferr
1.89 kristaps 1482: roff_nr(ROFF_ARGS)
1.83 schwarze 1483: {
1.121 schwarze 1484: const char *key;
1485: char *val;
1.187 schwarze 1486: size_t sz;
1.138 kristaps 1487: int iv;
1.187 schwarze 1488: char sign;
1.89 kristaps 1489:
1.121 schwarze 1490: val = *bufp + pos;
1491: key = roff_getname(r, &val, ln, pos);
1.89 kristaps 1492:
1.187 schwarze 1493: sign = *val;
1494: if ('+' == sign || '-' == sign)
1495: val++;
1496:
1497: sz = strspn(val, "0123456789");
1498: iv = sz ? mandoc_strntoi(val, sz, 10) : 0;
1.180 schwarze 1499:
1.187 schwarze 1500: roff_setreg(r, key, iv, sign);
1.109 kristaps 1501:
1.122 schwarze 1502: return(ROFF_IGN);
1503: }
1504:
1505: /* ARGSUSED */
1506: static enum rofferr
1507: roff_rm(ROFF_ARGS)
1508: {
1509: const char *name;
1510: char *cp;
1511:
1512: cp = *bufp + pos;
1513: while ('\0' != *cp) {
1.127 kristaps 1514: name = roff_getname(r, &cp, ln, (int)(cp - *bufp));
1.122 schwarze 1515: if ('\0' != *name)
1516: roff_setstr(r, name, NULL, 0);
1517: }
1.178 schwarze 1518: return(ROFF_IGN);
1519: }
1520:
1521: /* ARGSUSED */
1522: static enum rofferr
1523: roff_it(ROFF_ARGS)
1524: {
1525: char *cp;
1526: size_t len;
1527: int iv;
1528:
1529: /* Parse the number of lines. */
1530: cp = *bufp + pos;
1531: len = strcspn(cp, " \t");
1532: cp[len] = '\0';
1533: if ((iv = mandoc_strntoi(cp, len, 10)) <= 0) {
1534: mandoc_msg(MANDOCERR_NUMERIC, r->parse,
1535: ln, ppos, *bufp + 1);
1536: return(ROFF_IGN);
1537: }
1538: cp += len + 1;
1539:
1540: /* Arm the input line trap. */
1541: roffit_lines = iv;
1542: roffit_macro = mandoc_strdup(cp);
1.109 kristaps 1543: return(ROFF_IGN);
1.175 schwarze 1544: }
1545:
1546: /* ARGSUSED */
1547: static enum rofferr
1548: roff_Dd(ROFF_ARGS)
1549: {
1550: const char *const *cp;
1551:
1.199 schwarze 1552: if (0 == ((MPARSE_MDOC | MPARSE_QUICK) & r->options))
1.175 schwarze 1553: for (cp = __mdoc_reserved; *cp; cp++)
1554: roff_setstr(r, *cp, NULL, 0);
1555:
1556: return(ROFF_CONT);
1557: }
1558:
1559: /* ARGSUSED */
1560: static enum rofferr
1561: roff_TH(ROFF_ARGS)
1562: {
1563: const char *const *cp;
1564:
1.199 schwarze 1565: if (0 == (MPARSE_QUICK & r->options))
1.175 schwarze 1566: for (cp = __man_reserved; *cp; cp++)
1567: roff_setstr(r, *cp, NULL, 0);
1568:
1569: return(ROFF_CONT);
1.109 kristaps 1570: }
1571:
1572: /* ARGSUSED */
1573: static enum rofferr
1574: roff_TE(ROFF_ARGS)
1575: {
1576:
1577: if (NULL == r->tbl)
1.128 kristaps 1578: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.115 kristaps 1579: else
1.151 kristaps 1580: tbl_end(&r->tbl);
1.109 kristaps 1581:
1.112 kristaps 1582: return(ROFF_IGN);
1583: }
1584:
1585: /* ARGSUSED */
1586: static enum rofferr
1587: roff_T_(ROFF_ARGS)
1588: {
1589:
1590: if (NULL == r->tbl)
1.128 kristaps 1591: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.112 kristaps 1592: else
1.116 kristaps 1593: tbl_restart(ppos, ln, r->tbl);
1.112 kristaps 1594:
1.109 kristaps 1595: return(ROFF_IGN);
1596: }
1597:
1.156 kristaps 1598: #if 0
1599: static int
1.151 kristaps 1600: roff_closeeqn(struct roff *r)
1601: {
1602:
1603: return(r->eqn && ROFF_EQN == eqn_end(&r->eqn) ? 1 : 0);
1604: }
1.156 kristaps 1605: #endif
1.151 kristaps 1606:
1.156 kristaps 1607: static void
1.151 kristaps 1608: roff_openeqn(struct roff *r, const char *name, int line,
1609: int offs, const char *buf)
1.125 kristaps 1610: {
1.151 kristaps 1611: struct eqn_node *e;
1612: int poff;
1.125 kristaps 1613:
1614: assert(NULL == r->eqn);
1.151 kristaps 1615: e = eqn_alloc(name, offs, line, r->parse);
1.125 kristaps 1616:
1617: if (r->last_eqn)
1618: r->last_eqn->next = e;
1619: else
1620: r->first_eqn = r->last_eqn = e;
1621:
1622: r->eqn = r->last_eqn = e;
1.151 kristaps 1623:
1624: if (buf) {
1625: poff = 0;
1626: eqn_read(&r->eqn, line, buf, offs, &poff);
1627: }
1628: }
1629:
1630: /* ARGSUSED */
1631: static enum rofferr
1632: roff_EQ(ROFF_ARGS)
1633: {
1634:
1635: roff_openeqn(r, *bufp + pos, ln, ppos, NULL);
1.125 kristaps 1636: return(ROFF_IGN);
1637: }
1638:
1639: /* ARGSUSED */
1640: static enum rofferr
1641: roff_EN(ROFF_ARGS)
1642: {
1643:
1.128 kristaps 1644: mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.125 kristaps 1645: return(ROFF_IGN);
1646: }
1647:
1648: /* ARGSUSED */
1649: static enum rofferr
1.109 kristaps 1650: roff_TS(ROFF_ARGS)
1651: {
1.176 schwarze 1652: struct tbl_node *tbl;
1.89 kristaps 1653:
1.115 kristaps 1654: if (r->tbl) {
1.128 kristaps 1655: mandoc_msg(MANDOCERR_SCOPEBROKEN, r->parse, ln, ppos, NULL);
1.151 kristaps 1656: tbl_end(&r->tbl);
1.115 kristaps 1657: }
1.83 schwarze 1658:
1.176 schwarze 1659: tbl = tbl_alloc(ppos, ln, r->parse);
1.113 kristaps 1660:
1661: if (r->last_tbl)
1.176 schwarze 1662: r->last_tbl->next = tbl;
1.113 kristaps 1663: else
1.176 schwarze 1664: r->first_tbl = r->last_tbl = tbl;
1.113 kristaps 1665:
1.176 schwarze 1666: r->tbl = r->last_tbl = tbl;
1.83 schwarze 1667: return(ROFF_IGN);
1.92 schwarze 1668: }
1669:
1.105 kristaps 1670: /* ARGSUSED */
1671: static enum rofferr
1.174 kristaps 1672: roff_cc(ROFF_ARGS)
1673: {
1674: const char *p;
1675:
1676: p = *bufp + pos;
1677:
1678: if ('\0' == *p || '.' == (r->control = *p++))
1679: r->control = 0;
1680:
1681: if ('\0' != *p)
1682: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
1683:
1684: return(ROFF_IGN);
1685: }
1686:
1687: /* ARGSUSED */
1688: static enum rofferr
1.164 kristaps 1689: roff_tr(ROFF_ARGS)
1690: {
1691: const char *p, *first, *second;
1692: size_t fsz, ssz;
1693: enum mandoc_esc esc;
1694:
1695: p = *bufp + pos;
1696:
1697: if ('\0' == *p) {
1698: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
1699: return(ROFF_IGN);
1700: }
1701:
1702: while ('\0' != *p) {
1703: fsz = ssz = 1;
1704:
1705: first = p++;
1706: if ('\\' == *first) {
1707: esc = mandoc_escape(&p, NULL, NULL);
1708: if (ESCAPE_ERROR == esc) {
1709: mandoc_msg
1710: (MANDOCERR_BADESCAPE, r->parse,
1711: ln, (int)(p - *bufp), NULL);
1712: return(ROFF_IGN);
1713: }
1714: fsz = (size_t)(p - first);
1715: }
1716:
1717: second = p++;
1718: if ('\\' == *second) {
1719: esc = mandoc_escape(&p, NULL, NULL);
1720: if (ESCAPE_ERROR == esc) {
1721: mandoc_msg
1722: (MANDOCERR_BADESCAPE, r->parse,
1723: ln, (int)(p - *bufp), NULL);
1724: return(ROFF_IGN);
1725: }
1726: ssz = (size_t)(p - second);
1.165 kristaps 1727: } else if ('\0' == *second) {
1.164 kristaps 1728: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse,
1729: ln, (int)(p - *bufp), NULL);
1730: second = " ";
1.165 kristaps 1731: p--;
1.164 kristaps 1732: }
1733:
1.167 kristaps 1734: if (fsz > 1) {
1735: roff_setstrn(&r->xmbtab, first,
1736: fsz, second, ssz, 0);
1737: continue;
1738: }
1739:
1740: if (NULL == r->xtab)
1741: r->xtab = mandoc_calloc
1742: (128, sizeof(struct roffstr));
1743:
1744: free(r->xtab[(int)*first].p);
1745: r->xtab[(int)*first].p = mandoc_strndup(second, ssz);
1746: r->xtab[(int)*first].sz = ssz;
1.164 kristaps 1747: }
1748:
1749: return(ROFF_IGN);
1750: }
1751:
1752: /* ARGSUSED */
1753: static enum rofferr
1.105 kristaps 1754: roff_so(ROFF_ARGS)
1755: {
1756: char *name;
1757:
1.128 kristaps 1758: mandoc_msg(MANDOCERR_SO, r->parse, ln, ppos, NULL);
1.105 kristaps 1759:
1760: /*
1761: * Handle `so'. Be EXTREMELY careful, as we shouldn't be
1762: * opening anything that's not in our cwd or anything beneath
1763: * it. Thus, explicitly disallow traversing up the file-system
1764: * or using absolute paths.
1765: */
1766:
1767: name = *bufp + pos;
1768: if ('/' == *name || strstr(name, "../") || strstr(name, "/..")) {
1.128 kristaps 1769: mandoc_msg(MANDOCERR_SOPATH, r->parse, ln, pos, NULL);
1.105 kristaps 1770: return(ROFF_ERR);
1771: }
1772:
1773: *offs = pos;
1774: return(ROFF_SO);
1775: }
1.92 schwarze 1776:
1.106 kristaps 1777: /* ARGSUSED */
1778: static enum rofferr
1779: roff_userdef(ROFF_ARGS)
1.99 kristaps 1780: {
1.106 kristaps 1781: const char *arg[9];
1782: char *cp, *n1, *n2;
1.119 schwarze 1783: int i;
1.106 kristaps 1784:
1785: /*
1786: * Collect pointers to macro argument strings
1.188 schwarze 1787: * and NUL-terminate them.
1.106 kristaps 1788: */
1789: cp = *bufp + pos;
1.119 schwarze 1790: for (i = 0; i < 9; i++)
1.120 schwarze 1791: arg[i] = '\0' == *cp ? "" :
1.136 kristaps 1792: mandoc_getarg(r->parse, &cp, ln, &pos);
1.99 kristaps 1793:
1.106 kristaps 1794: /*
1795: * Expand macro arguments.
1.99 kristaps 1796: */
1.106 kristaps 1797: *szp = 0;
1798: n1 = cp = mandoc_strdup(r->current_string);
1799: while (NULL != (cp = strstr(cp, "\\$"))) {
1800: i = cp[2] - '1';
1801: if (0 > i || 8 < i) {
1802: /* Not an argument invocation. */
1803: cp += 2;
1804: continue;
1805: }
1806:
1807: *szp = strlen(n1) - 3 + strlen(arg[i]) + 1;
1808: n2 = mandoc_malloc(*szp);
1809:
1810: strlcpy(n2, n1, (size_t)(cp - n1 + 1));
1811: strlcat(n2, arg[i], *szp);
1812: strlcat(n2, cp + 3, *szp);
1813:
1814: cp = n2 + (cp - n1);
1815: free(n1);
1816: n1 = n2;
1.99 kristaps 1817: }
1818:
1.106 kristaps 1819: /*
1820: * Replace the macro invocation
1821: * by the expanded macro.
1822: */
1823: free(*bufp);
1824: *bufp = n1;
1825: if (0 == *szp)
1826: *szp = strlen(*bufp) + 1;
1827:
1828: return(*szp > 1 && '\n' == (*bufp)[(int)*szp - 2] ?
1829: ROFF_REPARSE : ROFF_APPEND);
1.99 kristaps 1830: }
1.121 schwarze 1831:
1832: static char *
1833: roff_getname(struct roff *r, char **cpp, int ln, int pos)
1834: {
1835: char *name, *cp;
1836:
1837: name = *cpp;
1838: if ('\0' == *name)
1839: return(name);
1840:
1841: /* Read until end of name. */
1842: for (cp = name; '\0' != *cp && ' ' != *cp; cp++) {
1843: if ('\\' != *cp)
1844: continue;
1845: cp++;
1846: if ('\\' == *cp)
1847: continue;
1.128 kristaps 1848: mandoc_msg(MANDOCERR_NAMESC, r->parse, ln, pos, NULL);
1.121 schwarze 1849: *cp = '\0';
1850: name = cp;
1851: }
1852:
1853: /* Nil-terminate name. */
1854: if ('\0' != *cp)
1855: *(cp++) = '\0';
1856:
1857: /* Read past spaces. */
1858: while (' ' == *cp)
1859: cp++;
1860:
1861: *cpp = cp;
1862: return(name);
1863: }
1864:
1.106 kristaps 1865: /*
1866: * Store *string into the user-defined string called *name.
1867: * To clear an existing entry, call with (*r, *name, NULL, 0).
1.193 schwarze 1868: * append == 0: replace mode
1869: * append == 1: single-line append mode
1870: * append == 2: multiline append mode, append '\n' after each call
1.106 kristaps 1871: */
1.94 kristaps 1872: static void
1.106 kristaps 1873: roff_setstr(struct roff *r, const char *name, const char *string,
1.193 schwarze 1874: int append)
1.92 schwarze 1875: {
1.164 kristaps 1876:
1877: roff_setstrn(&r->strtab, name, strlen(name), string,
1.193 schwarze 1878: string ? strlen(string) : 0, append);
1.164 kristaps 1879: }
1880:
1881: static void
1.166 kristaps 1882: roff_setstrn(struct roffkv **r, const char *name, size_t namesz,
1.193 schwarze 1883: const char *string, size_t stringsz, int append)
1.164 kristaps 1884: {
1.166 kristaps 1885: struct roffkv *n;
1.164 kristaps 1886: char *c;
1887: int i;
1888: size_t oldch, newch;
1.92 schwarze 1889:
1.106 kristaps 1890: /* Search for an existing string with the same name. */
1.164 kristaps 1891: n = *r;
1892:
1.166 kristaps 1893: while (n && strcmp(name, n->key.p))
1.92 schwarze 1894: n = n->next;
1.94 kristaps 1895:
1896: if (NULL == n) {
1.106 kristaps 1897: /* Create a new string table entry. */
1.166 kristaps 1898: n = mandoc_malloc(sizeof(struct roffkv));
1899: n->key.p = mandoc_strndup(name, namesz);
1900: n->key.sz = namesz;
1901: n->val.p = NULL;
1902: n->val.sz = 0;
1.164 kristaps 1903: n->next = *r;
1904: *r = n;
1.193 schwarze 1905: } else if (0 == append) {
1.166 kristaps 1906: free(n->val.p);
1907: n->val.p = NULL;
1908: n->val.sz = 0;
1.106 kristaps 1909: }
1910:
1911: if (NULL == string)
1912: return;
1913:
1914: /*
1915: * One additional byte for the '\n' in multiline mode,
1916: * and one for the terminating '\0'.
1917: */
1.193 schwarze 1918: newch = stringsz + (1 < append ? 2u : 1u);
1.164 kristaps 1919:
1.166 kristaps 1920: if (NULL == n->val.p) {
1921: n->val.p = mandoc_malloc(newch);
1922: *n->val.p = '\0';
1.106 kristaps 1923: oldch = 0;
1924: } else {
1.166 kristaps 1925: oldch = n->val.sz;
1926: n->val.p = mandoc_realloc(n->val.p, oldch + newch);
1.106 kristaps 1927: }
1928:
1929: /* Skip existing content in the destination buffer. */
1.166 kristaps 1930: c = n->val.p + (int)oldch;
1.106 kristaps 1931:
1932: /* Append new content to the destination buffer. */
1.164 kristaps 1933: i = 0;
1934: while (i < (int)stringsz) {
1.106 kristaps 1935: /*
1936: * Rudimentary roff copy mode:
1937: * Handle escaped backslashes.
1938: */
1.164 kristaps 1939: if ('\\' == string[i] && '\\' == string[i + 1])
1940: i++;
1941: *c++ = string[i++];
1.106 kristaps 1942: }
1.94 kristaps 1943:
1.106 kristaps 1944: /* Append terminating bytes. */
1.193 schwarze 1945: if (1 < append)
1.106 kristaps 1946: *c++ = '\n';
1.163 kristaps 1947:
1.106 kristaps 1948: *c = '\0';
1.166 kristaps 1949: n->val.sz = (int)(c - n->val.p);
1.92 schwarze 1950: }
1951:
1.94 kristaps 1952: static const char *
1953: roff_getstrn(const struct roff *r, const char *name, size_t len)
1.92 schwarze 1954: {
1.166 kristaps 1955: const struct roffkv *n;
1.191 schwarze 1956: int i;
1.92 schwarze 1957:
1.164 kristaps 1958: for (n = r->strtab; n; n = n->next)
1.166 kristaps 1959: if (0 == strncmp(name, n->key.p, len) &&
1960: '\0' == n->key.p[(int)len])
1961: return(n->val.p);
1.191 schwarze 1962:
1963: for (i = 0; i < PREDEFS_MAX; i++)
1964: if (0 == strncmp(name, predefs[i].name, len) &&
1965: '\0' == predefs[i].name[(int)len])
1966: return(predefs[i].str);
1.94 kristaps 1967:
1.157 kristaps 1968: return(NULL);
1.92 schwarze 1969: }
1970:
1.94 kristaps 1971: static void
1.167 kristaps 1972: roff_freestr(struct roffkv *r)
1.92 schwarze 1973: {
1.166 kristaps 1974: struct roffkv *n, *nn;
1.92 schwarze 1975:
1.167 kristaps 1976: for (n = r; n; n = nn) {
1.166 kristaps 1977: free(n->key.p);
1978: free(n->val.p);
1.92 schwarze 1979: nn = n->next;
1980: free(n);
1981: }
1.114 kristaps 1982: }
1983:
1984: const struct tbl_span *
1985: roff_span(const struct roff *r)
1986: {
1987:
1988: return(r->tbl ? tbl_span(r->tbl) : NULL);
1.125 kristaps 1989: }
1990:
1991: const struct eqn *
1992: roff_eqn(const struct roff *r)
1993: {
1994:
1995: return(r->last_eqn ? &r->last_eqn->eqn : NULL);
1.164 kristaps 1996: }
1997:
1998: /*
1999: * Duplicate an input string, making the appropriate character
2000: * conversations (as stipulated by `tr') along the way.
2001: * Returns a heap-allocated string with all the replacements made.
2002: */
2003: char *
2004: roff_strdup(const struct roff *r, const char *p)
2005: {
1.166 kristaps 2006: const struct roffkv *cp;
1.164 kristaps 2007: char *res;
2008: const char *pp;
2009: size_t ssz, sz;
2010: enum mandoc_esc esc;
2011:
1.167 kristaps 2012: if (NULL == r->xmbtab && NULL == r->xtab)
1.164 kristaps 2013: return(mandoc_strdup(p));
2014: else if ('\0' == *p)
2015: return(mandoc_strdup(""));
2016:
2017: /*
2018: * Step through each character looking for term matches
2019: * (remember that a `tr' can be invoked with an escape, which is
2020: * a glyph but the escape is multi-character).
2021: * We only do this if the character hash has been initialised
2022: * and the string is >0 length.
2023: */
2024:
2025: res = NULL;
2026: ssz = 0;
2027:
2028: while ('\0' != *p) {
1.167 kristaps 2029: if ('\\' != *p && r->xtab && r->xtab[(int)*p].p) {
2030: sz = r->xtab[(int)*p].sz;
2031: res = mandoc_realloc(res, ssz + sz + 1);
2032: memcpy(res + ssz, r->xtab[(int)*p].p, sz);
2033: ssz += sz;
2034: p++;
2035: continue;
2036: } else if ('\\' != *p) {
2037: res = mandoc_realloc(res, ssz + 2);
2038: res[ssz++] = *p++;
2039: continue;
2040: }
2041:
1.164 kristaps 2042: /* Search for term matches. */
1.167 kristaps 2043: for (cp = r->xmbtab; cp; cp = cp->next)
1.166 kristaps 2044: if (0 == strncmp(p, cp->key.p, cp->key.sz))
1.164 kristaps 2045: break;
2046:
2047: if (NULL != cp) {
2048: /*
2049: * A match has been found.
2050: * Append the match to the array and move
2051: * forward by its keysize.
2052: */
1.166 kristaps 2053: res = mandoc_realloc
2054: (res, ssz + cp->val.sz + 1);
2055: memcpy(res + ssz, cp->val.p, cp->val.sz);
2056: ssz += cp->val.sz;
2057: p += (int)cp->key.sz;
1.164 kristaps 2058: continue;
2059: }
2060:
1.167 kristaps 2061: /*
2062: * Handle escapes carefully: we need to copy
2063: * over just the escape itself, or else we might
2064: * do replacements within the escape itself.
2065: * Make sure to pass along the bogus string.
2066: */
2067: pp = p++;
2068: esc = mandoc_escape(&p, NULL, NULL);
2069: if (ESCAPE_ERROR == esc) {
2070: sz = strlen(pp);
1.164 kristaps 2071: res = mandoc_realloc(res, ssz + sz + 1);
2072: memcpy(res + ssz, pp, sz);
1.167 kristaps 2073: break;
1.164 kristaps 2074: }
1.167 kristaps 2075: /*
2076: * We bail out on bad escapes.
2077: * No need to warn: we already did so when
2078: * roff_res() was called.
2079: */
2080: sz = (int)(p - pp);
2081: res = mandoc_realloc(res, ssz + sz + 1);
2082: memcpy(res + ssz, pp, sz);
2083: ssz += sz;
1.164 kristaps 2084: }
2085:
2086: res[(int)ssz] = '\0';
2087: return(res);
1.174 kristaps 2088: }
2089:
2090: /*
2091: * Find out whether a line is a macro line or not.
2092: * If it is, adjust the current position and return one; if it isn't,
2093: * return zero and don't change the current position.
2094: * If the control character has been set with `.cc', then let that grain
2095: * precedence.
2096: * This is slighly contrary to groff, where using the non-breaking
2097: * control character when `cc' has been invoked will cause the
2098: * non-breaking macro contents to be printed verbatim.
2099: */
2100: int
2101: roff_getcontrol(const struct roff *r, const char *cp, int *ppos)
2102: {
2103: int pos;
2104:
2105: pos = *ppos;
2106:
2107: if (0 != r->control && cp[pos] == r->control)
2108: pos++;
2109: else if (0 != r->control)
2110: return(0);
2111: else if ('\\' == cp[pos] && '.' == cp[pos + 1])
2112: pos += 2;
2113: else if ('.' == cp[pos] || '\'' == cp[pos])
2114: pos++;
2115: else
2116: return(0);
2117:
2118: while (' ' == cp[pos] || '\t' == cp[pos])
2119: pos++;
2120:
2121: *ppos = pos;
2122: return(1);
1.74 kristaps 2123: }
CVSweb