Annotation of mandoc/roff.c, Revision 1.229
1.229 ! schwarze 1: /* $Id: roff.c,v 1.228 2014/09/06 23:24:32 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: #include "config.h"
1.225 schwarze 19:
20: #include <sys/types.h>
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"
1.226 schwarze 30: #include "libmandoc.h"
1.109 kristaps 31: #include "libroff.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,
1.229 ! schwarze 63: ROFF_pl,
1.124 schwarze 64: ROFF_ps,
1.83 schwarze 65: ROFF_rm,
1.203 schwarze 66: ROFF_rr,
1.105 kristaps 67: ROFF_so,
1.124 schwarze 68: ROFF_ta,
1.83 schwarze 69: ROFF_tr,
1.175 schwarze 70: ROFF_Dd,
71: ROFF_TH,
1.109 kristaps 72: ROFF_TS,
73: ROFF_TE,
1.112 kristaps 74: ROFF_T_,
1.125 kristaps 75: ROFF_EQ,
76: ROFF_EN,
1.76 kristaps 77: ROFF_cblock,
1.106 kristaps 78: ROFF_USERDEF,
1.67 kristaps 79: ROFF_MAX
80: };
81:
1.147 kristaps 82: /*
1.167 kristaps 83: * An incredibly-simple string buffer.
84: */
1.94 kristaps 85: struct roffstr {
1.167 kristaps 86: char *p; /* nil-terminated buffer */
87: size_t sz; /* saved strlen(p) */
1.166 kristaps 88: };
89:
90: /*
1.167 kristaps 91: * A key-value roffstr pair as part of a singly-linked list.
1.166 kristaps 92: */
93: struct roffkv {
94: struct roffstr key;
95: struct roffstr val;
96: struct roffkv *next; /* next in list */
1.94 kristaps 97: };
98:
1.180 schwarze 99: /*
100: * A single number register as part of a singly-linked list.
101: */
102: struct roffreg {
103: struct roffstr key;
1.181 schwarze 104: int val;
1.180 schwarze 105: struct roffreg *next;
106: };
107:
1.67 kristaps 108: struct roff {
1.128 kristaps 109: struct mparse *parse; /* parse point */
1.67 kristaps 110: struct roffnode *last; /* leaf of stack */
1.223 schwarze 111: int *rstack; /* stack of inverted `ie' values */
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 */
1.223 schwarze 123: int options; /* parse options */
124: int rstacksz; /* current size limit of rstack */
125: int rstackpos; /* position in rstack */
1.227 schwarze 126: int format; /* current file in mdoc or man format */
1.223 schwarze 127: char control; /* control character */
1.79 kristaps 128: };
129:
1.67 kristaps 130: struct roffnode {
131: enum rofft tok; /* type of node */
132: struct roffnode *parent; /* up one in stack */
133: int line; /* parse line */
134: int col; /* parse col */
1.106 kristaps 135: char *name; /* node name, e.g. macro name */
1.79 kristaps 136: char *end; /* end-rules: custom token */
137: int endspan; /* end-rules: next-line or infty */
1.198 schwarze 138: int rule; /* current evaluation rule */
1.67 kristaps 139: };
140:
141: #define ROFF_ARGS struct roff *r, /* parse ctx */ \
1.72 kristaps 142: enum rofft tok, /* tok of macro */ \
1.207 schwarze 143: char **bufp, /* input buffer */ \
1.67 kristaps 144: size_t *szp, /* size of input buffer */ \
145: int ln, /* parse line */ \
1.75 kristaps 146: int ppos, /* original pos in buffer */ \
147: int pos, /* current pos in buffer */ \
1.74 kristaps 148: int *offs /* reset offset of buffer data */
1.67 kristaps 149:
150: typedef enum rofferr (*roffproc)(ROFF_ARGS);
151:
152: struct roffmac {
153: const char *name; /* macro name */
1.79 kristaps 154: roffproc proc; /* process new macro */
155: roffproc text; /* process as child text of macro */
156: roffproc sub; /* process as child of macro */
157: int flags;
158: #define ROFFMAC_STRUCT (1 << 0) /* always interpret */
1.85 kristaps 159: struct roffmac *next;
1.67 kristaps 160: };
161:
1.141 kristaps 162: struct predef {
163: const char *name; /* predefined input name */
164: const char *str; /* replacement symbol */
165: };
166:
167: #define PREDEF(__name, __str) \
168: { (__name), (__str) },
169:
1.155 kristaps 170: static enum rofft roffhash_find(const char *, size_t);
171: static void roffhash_init(void);
172: static void roffnode_cleanscope(struct roff *);
173: static void roffnode_pop(struct roff *);
174: static void roffnode_push(struct roff *, enum rofft,
175: const char *, int, int);
1.80 kristaps 176: static enum rofferr roff_block(ROFF_ARGS);
177: static enum rofferr roff_block_text(ROFF_ARGS);
178: static enum rofferr roff_block_sub(ROFF_ARGS);
179: static enum rofferr roff_cblock(ROFF_ARGS);
1.174 kristaps 180: static enum rofferr roff_cc(ROFF_ARGS);
1.195 schwarze 181: static void roff_ccond(struct roff *, int, int);
1.82 kristaps 182: static enum rofferr roff_cond(ROFF_ARGS);
183: static enum rofferr roff_cond_text(ROFF_ARGS);
184: static enum rofferr roff_cond_sub(ROFF_ARGS);
1.92 schwarze 185: static enum rofferr roff_ds(ROFF_ARGS);
1.198 schwarze 186: static int roff_evalcond(const char *, int *);
1.204 schwarze 187: static int roff_evalnum(const char *, int *, int *, int);
188: static int roff_evalpar(const char *, int *, int *);
1.198 schwarze 189: static int roff_evalstrcond(const char *, int *);
1.155 kristaps 190: static void roff_free1(struct roff *);
1.180 schwarze 191: static void roff_freereg(struct roffreg *);
1.167 kristaps 192: static void roff_freestr(struct roffkv *);
1.212 schwarze 193: static size_t roff_getname(struct roff *, char **, int, int);
1.184 schwarze 194: static int roff_getnum(const char *, int *, int *);
195: static int roff_getop(const char *, int *, char *);
1.181 schwarze 196: static int roff_getregn(const struct roff *,
197: const char *, size_t);
1.192 schwarze 198: static int roff_getregro(const char *name);
1.207 schwarze 199: static const char *roff_getstrn(const struct roff *,
1.94 kristaps 200: const char *, size_t);
1.178 schwarze 201: static enum rofferr roff_it(ROFF_ARGS);
1.103 kristaps 202: static enum rofferr roff_line_ignore(ROFF_ARGS);
1.89 kristaps 203: static enum rofferr roff_nr(ROFF_ARGS);
1.169 schwarze 204: static void roff_openeqn(struct roff *, const char *,
1.156 kristaps 205: int, int, const char *);
1.214 schwarze 206: static enum rofft roff_parse(struct roff *, char *, int *,
207: int, int);
1.178 schwarze 208: static enum rofferr roff_parsetext(char **, size_t *, int, int *);
1.207 schwarze 209: static enum rofferr roff_res(struct roff *,
1.142 kristaps 210: char **, size_t *, int, int);
1.122 schwarze 211: static enum rofferr roff_rm(ROFF_ARGS);
1.203 schwarze 212: static enum rofferr roff_rr(ROFF_ARGS);
1.94 kristaps 213: static void roff_setstr(struct roff *,
1.106 kristaps 214: const char *, const char *, int);
1.207 schwarze 215: static void roff_setstrn(struct roffkv **, const char *,
1.164 kristaps 216: size_t, const char *, size_t, int);
1.105 kristaps 217: static enum rofferr roff_so(ROFF_ARGS);
1.164 kristaps 218: static enum rofferr roff_tr(ROFF_ARGS);
1.175 schwarze 219: static enum rofferr roff_Dd(ROFF_ARGS);
220: static enum rofferr roff_TH(ROFF_ARGS);
1.109 kristaps 221: static enum rofferr roff_TE(ROFF_ARGS);
222: static enum rofferr roff_TS(ROFF_ARGS);
1.125 kristaps 223: static enum rofferr roff_EQ(ROFF_ARGS);
224: static enum rofferr roff_EN(ROFF_ARGS);
1.112 kristaps 225: static enum rofferr roff_T_(ROFF_ARGS);
1.106 kristaps 226: static enum rofferr roff_userdef(ROFF_ARGS);
1.67 kristaps 227:
1.155 kristaps 228: /* See roffhash_find() */
1.85 kristaps 229:
230: #define ASCII_HI 126
231: #define ASCII_LO 33
232: #define HASHWIDTH (ASCII_HI - ASCII_LO + 1)
233:
234: static struct roffmac *hash[HASHWIDTH];
235:
236: static struct roffmac roffs[ROFF_MAX] = {
1.103 kristaps 237: { "ad", roff_line_ignore, NULL, NULL, 0, NULL },
1.85 kristaps 238: { "am", roff_block, roff_block_text, roff_block_sub, 0, NULL },
239: { "ami", roff_block, roff_block_text, roff_block_sub, 0, NULL },
240: { "am1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.193 schwarze 241: { "as", roff_ds, NULL, NULL, 0, NULL },
1.174 kristaps 242: { "cc", roff_cc, NULL, NULL, 0, NULL },
1.194 schwarze 243: { "ce", roff_line_ignore, NULL, NULL, 0, NULL },
1.85 kristaps 244: { "de", roff_block, roff_block_text, roff_block_sub, 0, NULL },
245: { "dei", roff_block, roff_block_text, roff_block_sub, 0, NULL },
246: { "de1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.92 schwarze 247: { "ds", roff_ds, NULL, NULL, 0, NULL },
1.85 kristaps 248: { "el", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
1.185 schwarze 249: { "fam", roff_line_ignore, NULL, NULL, 0, NULL },
1.186 schwarze 250: { "hw", roff_line_ignore, NULL, NULL, 0, NULL },
1.103 kristaps 251: { "hy", roff_line_ignore, NULL, NULL, 0, NULL },
1.85 kristaps 252: { "ie", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
253: { "if", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
254: { "ig", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.178 schwarze 255: { "it", roff_it, NULL, NULL, 0, NULL },
1.103 kristaps 256: { "ne", roff_line_ignore, NULL, NULL, 0, NULL },
257: { "nh", roff_line_ignore, NULL, NULL, 0, NULL },
1.104 kristaps 258: { "nr", roff_nr, NULL, NULL, 0, NULL },
1.124 schwarze 259: { "ns", roff_line_ignore, NULL, NULL, 0, NULL },
1.229 ! schwarze 260: { "pl", roff_line_ignore, NULL, NULL, 0, NULL },
1.124 schwarze 261: { "ps", roff_line_ignore, NULL, NULL, 0, NULL },
1.122 schwarze 262: { "rm", roff_rm, NULL, NULL, 0, NULL },
1.203 schwarze 263: { "rr", roff_rr, NULL, NULL, 0, NULL },
1.105 kristaps 264: { "so", roff_so, NULL, NULL, 0, NULL },
1.124 schwarze 265: { "ta", roff_line_ignore, NULL, NULL, 0, NULL },
1.164 kristaps 266: { "tr", roff_tr, NULL, NULL, 0, NULL },
1.175 schwarze 267: { "Dd", roff_Dd, NULL, NULL, 0, NULL },
268: { "TH", roff_TH, NULL, NULL, 0, NULL },
1.109 kristaps 269: { "TS", roff_TS, NULL, NULL, 0, NULL },
270: { "TE", roff_TE, NULL, NULL, 0, NULL },
1.112 kristaps 271: { "T&", roff_T_, NULL, NULL, 0, NULL },
1.125 kristaps 272: { "EQ", roff_EQ, NULL, NULL, 0, NULL },
273: { "EN", roff_EN, NULL, NULL, 0, NULL },
1.85 kristaps 274: { ".", roff_cblock, NULL, NULL, 0, NULL },
1.106 kristaps 275: { NULL, roff_userdef, NULL, NULL, 0, NULL },
1.67 kristaps 276: };
277:
1.200 schwarze 278: /* not currently implemented: Ds em Eq LP Me PP pp Or Rd Sf SH */
1.175 schwarze 279: const char *const __mdoc_reserved[] = {
280: "Ac", "Ad", "An", "Ao", "Ap", "Aq", "Ar", "At",
281: "Bc", "Bd", "Bf", "Bk", "Bl", "Bo", "Bq",
282: "Brc", "Bro", "Brq", "Bsx", "Bt", "Bx",
283: "Cd", "Cm", "Db", "Dc", "Dd", "Dl", "Do", "Dq",
1.200 schwarze 284: "Dt", "Dv", "Dx", "D1",
285: "Ec", "Ed", "Ef", "Ek", "El", "Em",
286: "En", "Eo", "Er", "Es", "Ev", "Ex",
1.175 schwarze 287: "Fa", "Fc", "Fd", "Fl", "Fn", "Fo", "Fr", "Ft", "Fx",
1.200 schwarze 288: "Hf", "Ic", "In", "It", "Lb", "Li", "Lk", "Lp",
289: "Ms", "Mt", "Nd", "Nm", "No", "Ns", "Nx",
1.175 schwarze 290: "Oc", "Oo", "Op", "Os", "Ot", "Ox",
1.200 schwarze 291: "Pa", "Pc", "Pf", "Po", "Pp", "Pq",
292: "Qc", "Ql", "Qo", "Qq", "Re", "Rs", "Rv",
293: "Sc", "Sh", "Sm", "So", "Sq",
1.175 schwarze 294: "Ss", "St", "Sx", "Sy",
295: "Ta", "Tn", "Ud", "Ux", "Va", "Vt", "Xc", "Xo", "Xr",
1.200 schwarze 296: "%A", "%B", "%C", "%D", "%I", "%J", "%N", "%O",
1.175 schwarze 297: "%P", "%Q", "%R", "%T", "%U", "%V",
298: NULL
299: };
300:
1.200 schwarze 301: /* not currently implemented: BT DE DS ME MT PT SY TQ YS */
1.175 schwarze 302: const char *const __man_reserved[] = {
1.200 schwarze 303: "AT", "B", "BI", "BR", "DT",
304: "EE", "EN", "EQ", "EX", "HP", "I", "IB", "IP", "IR",
305: "LP", "OP", "P", "PD", "PP",
306: "R", "RB", "RE", "RI", "RS", "SB", "SH", "SM", "SS",
307: "TE", "TH", "TP", "TS", "T&", "UC", "UE", "UR",
1.175 schwarze 308: NULL
309: };
310:
1.141 kristaps 311: /* Array of injected predefined strings. */
312: #define PREDEFS_MAX 38
313: static const struct predef predefs[PREDEFS_MAX] = {
314: #include "predefs.in"
315: };
316:
1.155 kristaps 317: /* See roffhash_find() */
1.85 kristaps 318: #define ROFF_HASH(p) (p[0] - ASCII_LO)
319:
1.178 schwarze 320: static int roffit_lines; /* number of lines to delay */
321: static char *roffit_macro; /* nil-terminated macro line */
322:
1.207 schwarze 323:
1.85 kristaps 324: static void
1.155 kristaps 325: roffhash_init(void)
1.85 kristaps 326: {
327: struct roffmac *n;
328: int buc, i;
329:
1.106 kristaps 330: for (i = 0; i < (int)ROFF_USERDEF; i++) {
1.85 kristaps 331: assert(roffs[i].name[0] >= ASCII_LO);
332: assert(roffs[i].name[0] <= ASCII_HI);
333:
334: buc = ROFF_HASH(roffs[i].name);
335:
336: if (NULL != (n = hash[buc])) {
337: for ( ; n->next; n = n->next)
338: /* Do nothing. */ ;
339: n->next = &roffs[i];
340: } else
341: hash[buc] = &roffs[i];
342: }
343: }
344:
1.67 kristaps 345: /*
346: * Look up a roff token by its name. Returns ROFF_MAX if no macro by
347: * the nil-terminated string name could be found.
348: */
349: static enum rofft
1.155 kristaps 350: roffhash_find(const char *p, size_t s)
1.67 kristaps 351: {
1.85 kristaps 352: int buc;
353: struct roffmac *n;
1.67 kristaps 354:
1.85 kristaps 355: /*
356: * libroff has an extremely simple hashtable, for the time
357: * being, which simply keys on the first character, which must
358: * be printable, then walks a chain. It works well enough until
359: * optimised.
360: */
361:
362: if (p[0] < ASCII_LO || p[0] > ASCII_HI)
363: return(ROFF_MAX);
364:
365: buc = ROFF_HASH(p);
366:
367: if (NULL == (n = hash[buc]))
368: return(ROFF_MAX);
369: for ( ; n; n = n->next)
1.106 kristaps 370: if (0 == strncmp(n->name, p, s) && '\0' == n->name[(int)s])
1.85 kristaps 371: return((enum rofft)(n - roffs));
1.67 kristaps 372:
373: return(ROFF_MAX);
374: }
375:
376: /*
377: * Pop the current node off of the stack of roff instructions currently
378: * pending.
379: */
380: static void
381: roffnode_pop(struct roff *r)
382: {
383: struct roffnode *p;
384:
1.75 kristaps 385: assert(r->last);
1.207 schwarze 386: p = r->last;
1.82 kristaps 387:
1.75 kristaps 388: r->last = r->last->parent;
1.106 kristaps 389: free(p->name);
390: free(p->end);
1.67 kristaps 391: free(p);
392: }
393:
394: /*
395: * Push a roff node onto the instruction stack. This must later be
396: * removed with roffnode_pop().
397: */
1.98 schwarze 398: static void
1.106 kristaps 399: roffnode_push(struct roff *r, enum rofft tok, const char *name,
400: int line, int col)
1.67 kristaps 401: {
402: struct roffnode *p;
403:
1.98 schwarze 404: p = mandoc_calloc(1, sizeof(struct roffnode));
1.67 kristaps 405: p->tok = tok;
1.106 kristaps 406: if (name)
407: p->name = mandoc_strdup(name);
1.67 kristaps 408: p->parent = r->last;
409: p->line = line;
410: p->col = col;
1.198 schwarze 411: p->rule = p->parent ? p->parent->rule : 0;
1.67 kristaps 412:
413: r->last = p;
414: }
415:
416: static void
417: roff_free1(struct roff *r)
418: {
1.176 schwarze 419: struct tbl_node *tbl;
1.125 kristaps 420: struct eqn_node *e;
1.167 kristaps 421: int i;
1.67 kristaps 422:
1.176 schwarze 423: while (NULL != (tbl = r->first_tbl)) {
424: r->first_tbl = tbl->next;
425: tbl_free(tbl);
1.109 kristaps 426: }
1.113 kristaps 427: r->first_tbl = r->last_tbl = r->tbl = NULL;
428:
1.125 kristaps 429: while (NULL != (e = r->first_eqn)) {
430: r->first_eqn = e->next;
431: eqn_free(e);
432: }
433: r->first_eqn = r->last_eqn = r->eqn = NULL;
434:
1.67 kristaps 435: while (r->last)
436: roffnode_pop(r);
1.109 kristaps 437:
1.223 schwarze 438: free (r->rstack);
439: r->rstack = NULL;
440: r->rstacksz = 0;
441: r->rstackpos = -1;
442:
443: roff_freereg(r->regtab);
444: r->regtab = NULL;
445:
1.167 kristaps 446: roff_freestr(r->strtab);
447: roff_freestr(r->xmbtab);
448: r->strtab = r->xmbtab = NULL;
449:
450: if (r->xtab)
451: for (i = 0; i < 128; i++)
452: free(r->xtab[i].p);
453: free(r->xtab);
454: r->xtab = NULL;
1.67 kristaps 455: }
456:
457: void
458: roff_reset(struct roff *r)
459: {
460:
461: roff_free1(r);
1.227 schwarze 462: r->format = r->options & (MPARSE_MDOC | MPARSE_MAN);
1.174 kristaps 463: r->control = 0;
1.67 kristaps 464: }
465:
466: void
467: roff_free(struct roff *r)
468: {
469:
470: roff_free1(r);
471: free(r);
472: }
473:
474: struct roff *
1.199 schwarze 475: roff_alloc(struct mparse *parse, int options)
1.67 kristaps 476: {
477: struct roff *r;
478:
1.98 schwarze 479: r = mandoc_calloc(1, sizeof(struct roff));
1.128 kristaps 480: r->parse = parse;
1.199 schwarze 481: r->options = options;
1.227 schwarze 482: r->format = options & (MPARSE_MDOC | MPARSE_MAN);
1.82 kristaps 483: r->rstackpos = -1;
1.207 schwarze 484:
1.155 kristaps 485: roffhash_init();
1.141 kristaps 486:
1.67 kristaps 487: return(r);
488: }
489:
1.94 kristaps 490: /*
1.206 schwarze 491: * In the current line, expand escape sequences that tend to get
492: * used in numerical expressions and conditional requests.
493: * Also check the syntax of the remaining escape sequences.
1.154 kristaps 494: */
1.172 schwarze 495: static enum rofferr
1.142 kristaps 496: roff_res(struct roff *r, char **bufp, size_t *szp, int ln, int pos)
1.94 kristaps 497: {
1.208 schwarze 498: char ubuf[24]; /* buffer to print the number */
1.205 schwarze 499: const char *start; /* start of the string to process */
1.209 schwarze 500: char *stesc; /* start of an escape sequence ('\\') */
1.108 schwarze 501: const char *stnam; /* start of the name, after "[(*" */
502: const char *cp; /* end of the name, e.g. before ']' */
503: const char *res; /* the string to be substituted */
1.181 schwarze 504: char *nbuf; /* new buffer to copy bufp to */
505: size_t maxl; /* expected length of the escape name */
506: size_t naml; /* actual length of the escape name */
507: int expand_count; /* to avoid infinite loops */
1.206 schwarze 508: int npos; /* position in numeric expression */
1.218 schwarze 509: int arg_complete; /* argument not interrupted by eol */
1.206 schwarze 510: char term; /* character terminating the escape */
1.94 kristaps 511:
1.170 schwarze 512: expand_count = 0;
1.205 schwarze 513: start = *bufp + pos;
514: stesc = strchr(start, '\0') - 1;
515: while (stesc-- > start) {
1.170 schwarze 516:
1.205 schwarze 517: /* Search backwards for the next backslash. */
518:
519: if ('\\' != *stesc)
520: continue;
521:
522: /* If it is escaped, skip it. */
523:
524: for (cp = stesc - 1; cp >= start; cp--)
525: if ('\\' != *cp)
526: break;
527:
528: if (0 == (stesc - cp) % 2) {
1.209 schwarze 529: stesc = (char *)cp;
1.205 schwarze 530: continue;
531: }
1.108 schwarze 532:
1.206 schwarze 533: /* Decide whether to expand or to check only. */
1.108 schwarze 534:
1.206 schwarze 535: term = '\0';
1.205 schwarze 536: cp = stesc + 1;
1.181 schwarze 537: switch (*cp) {
1.207 schwarze 538: case '*':
1.181 schwarze 539: res = NULL;
540: break;
1.207 schwarze 541: case 'B':
1.206 schwarze 542: /* FALLTHROUGH */
1.207 schwarze 543: case 'w':
1.206 schwarze 544: term = cp[1];
545: /* FALLTHROUGH */
1.207 schwarze 546: case 'n':
1.181 schwarze 547: res = ubuf;
548: break;
549: default:
1.205 schwarze 550: if (ESCAPE_ERROR == mandoc_escape(&cp, NULL, NULL))
1.219 schwarze 551: mandoc_vmsg(MANDOCERR_ESC_BAD,
552: r->parse, ln, (int)(stesc - *bufp),
553: "%.*s", (int)(cp - stesc), stesc);
1.205 schwarze 554: continue;
1.152 kristaps 555: }
556:
1.205 schwarze 557: if (EXPAND_LIMIT < ++expand_count) {
558: mandoc_msg(MANDOCERR_ROFFLOOP, r->parse,
559: ln, (int)(stesc - *bufp), NULL);
560: return(ROFF_IGN);
561: }
1.108 schwarze 562:
563: /*
564: * The third character decides the length
1.181 schwarze 565: * of the name of the string or register.
1.108 schwarze 566: * Save a pointer to the name.
567: */
568:
1.206 schwarze 569: if ('\0' == term) {
570: switch (*++cp) {
1.207 schwarze 571: case '\0':
1.206 schwarze 572: maxl = 0;
573: break;
1.207 schwarze 574: case '(':
1.206 schwarze 575: cp++;
576: maxl = 2;
577: break;
1.207 schwarze 578: case '[':
1.206 schwarze 579: cp++;
580: term = ']';
581: maxl = 0;
582: break;
583: default:
584: maxl = 1;
585: break;
586: }
587: } else {
588: cp += 2;
1.94 kristaps 589: maxl = 0;
590: }
1.108 schwarze 591: stnam = cp;
1.94 kristaps 592:
1.108 schwarze 593: /* Advance to the end of the name. */
1.94 kristaps 594:
1.218 schwarze 595: arg_complete = 1;
1.181 schwarze 596: for (naml = 0; 0 == maxl || naml < maxl; naml++, cp++) {
1.153 kristaps 597: if ('\0' == *cp) {
1.219 schwarze 598: mandoc_msg(MANDOCERR_ESC_BAD, r->parse,
599: ln, (int)(stesc - *bufp), stesc);
1.218 schwarze 600: arg_complete = 0;
1.206 schwarze 601: break;
1.153 kristaps 602: }
1.206 schwarze 603: if (0 == maxl && *cp == term) {
604: cp++;
1.94 kristaps 605: break;
1.206 schwarze 606: }
1.94 kristaps 607: }
608:
1.108 schwarze 609: /*
610: * Retrieve the replacement string; if it is
611: * undefined, resume searching for escapes.
612: */
613:
1.206 schwarze 614: switch (stesc[1]) {
1.207 schwarze 615: case '*':
1.218 schwarze 616: if (arg_complete)
617: res = roff_getstrn(r, stnam, naml);
1.206 schwarze 618: break;
1.207 schwarze 619: case 'B':
1.206 schwarze 620: npos = 0;
1.218 schwarze 621: ubuf[0] = arg_complete &&
622: roff_evalnum(stnam, &npos, NULL, 0) &&
623: stnam + npos + 1 == cp ? '1' : '0';
1.206 schwarze 624: ubuf[1] = '\0';
625: break;
1.207 schwarze 626: case 'n':
1.218 schwarze 627: if (arg_complete)
628: (void)snprintf(ubuf, sizeof(ubuf), "%d",
629: roff_getregn(r, stnam, naml));
630: else
631: ubuf[0] = '\0';
1.206 schwarze 632: break;
1.207 schwarze 633: case 'w':
1.218 schwarze 634: /* use even incomplete args */
1.208 schwarze 635: (void)snprintf(ubuf, sizeof(ubuf), "%d",
1.206 schwarze 636: 24 * (int)naml);
637: break;
638: }
1.94 kristaps 639:
640: if (NULL == res) {
1.219 schwarze 641: mandoc_vmsg(MANDOCERR_STR_UNDEF,
642: r->parse, ln, (int)(stesc - *bufp),
643: "%.*s", (int)naml, stnam);
1.142 kristaps 644: res = "";
1.94 kristaps 645: }
646:
1.108 schwarze 647: /* Replace the escape sequence by the string. */
648:
1.209 schwarze 649: *stesc = '\0';
650: *szp = mandoc_asprintf(&nbuf, "%s%s%s",
651: *bufp, res, cp) + 1;
1.94 kristaps 652:
1.205 schwarze 653: /* Prepare for the next replacement. */
1.94 kristaps 654:
1.205 schwarze 655: start = nbuf + pos;
1.209 schwarze 656: stesc = nbuf + (stesc - *bufp) + strlen(res);
1.94 kristaps 657: free(*bufp);
1.181 schwarze 658: *bufp = nbuf;
1.154 kristaps 659: }
1.172 schwarze 660: return(ROFF_CONT);
1.154 kristaps 661: }
662:
663: /*
1.178 schwarze 664: * Process text streams:
665: * Convert all breakable hyphens into ASCII_HYPH.
666: * Decrement and spring input line trap.
1.154 kristaps 667: */
668: static enum rofferr
1.178 schwarze 669: roff_parsetext(char **bufp, size_t *szp, int pos, int *offs)
1.154 kristaps 670: {
671: size_t sz;
672: const char *start;
1.178 schwarze 673: char *p;
674: int isz;
1.154 kristaps 675: enum mandoc_esc esc;
676:
1.178 schwarze 677: start = p = *bufp + pos;
1.154 kristaps 678:
679: while ('\0' != *p) {
680: sz = strcspn(p, "-\\");
681: p += sz;
682:
1.159 kristaps 683: if ('\0' == *p)
684: break;
685:
1.154 kristaps 686: if ('\\' == *p) {
687: /* Skip over escapes. */
688: p++;
1.189 schwarze 689: esc = mandoc_escape((const char **)&p, NULL, NULL);
1.154 kristaps 690: if (ESCAPE_ERROR == esc)
691: break;
1.155 kristaps 692: continue;
1.159 kristaps 693: } else if (p == start) {
1.158 kristaps 694: p++;
1.155 kristaps 695: continue;
1.158 kristaps 696: }
1.155 kristaps 697:
1.171 schwarze 698: if (isalpha((unsigned char)p[-1]) &&
699: isalpha((unsigned char)p[1]))
1.155 kristaps 700: *p = ASCII_HYPH;
701: p++;
1.94 kristaps 702: }
703:
1.178 schwarze 704: /* Spring the input line trap. */
705: if (1 == roffit_lines) {
1.202 schwarze 706: isz = mandoc_asprintf(&p, "%s\n.%s", *bufp, roffit_macro);
1.178 schwarze 707: free(*bufp);
708: *bufp = p;
709: *szp = isz + 1;
710: *offs = 0;
711: free(roffit_macro);
712: roffit_lines = 0;
713: return(ROFF_REPARSE);
714: } else if (1 < roffit_lines)
715: --roffit_lines;
1.154 kristaps 716: return(ROFF_CONT);
1.94 kristaps 717: }
718:
1.67 kristaps 719: enum rofferr
1.207 schwarze 720: roff_parseln(struct roff *r, int ln, char **bufp,
1.90 kristaps 721: size_t *szp, int pos, int *offs)
1.67 kristaps 722: {
723: enum rofft t;
1.109 kristaps 724: enum rofferr e;
1.130 kristaps 725: int ppos, ctl;
1.79 kristaps 726:
727: /*
1.94 kristaps 728: * Run the reserved-word filter only if we have some reserved
729: * words to fill in.
730: */
731:
1.172 schwarze 732: e = roff_res(r, bufp, szp, ln, pos);
733: if (ROFF_IGN == e)
734: return(e);
735: assert(ROFF_CONT == e);
1.94 kristaps 736:
1.130 kristaps 737: ppos = pos;
1.174 kristaps 738: ctl = roff_getcontrol(r, *bufp, &pos);
1.130 kristaps 739:
1.94 kristaps 740: /*
1.79 kristaps 741: * First, if a scope is open and we're not a macro, pass the
742: * text through the macro's filter. If a scope isn't open and
743: * we're not a macro, just let it through.
1.125 kristaps 744: * Finally, if there's an equation scope open, divert it into it
745: * no matter our state.
1.79 kristaps 746: */
1.74 kristaps 747:
1.130 kristaps 748: if (r->last && ! ctl) {
1.78 kristaps 749: t = r->last->tok;
750: assert(roffs[t].text);
1.207 schwarze 751: e = (*roffs[t].text)(r, t, bufp, szp, ln, pos, pos, offs);
1.109 kristaps 752: assert(ROFF_IGN == e || ROFF_CONT == e);
1.125 kristaps 753: if (ROFF_CONT != e)
754: return(e);
1.182 schwarze 755: }
756: if (r->eqn)
757: return(eqn_read(&r->eqn, ln, *bufp, ppos, offs));
758: if ( ! ctl) {
1.125 kristaps 759: if (r->tbl)
1.130 kristaps 760: return(tbl_read(r->tbl, ln, *bufp, pos));
1.178 schwarze 761: return(roff_parsetext(bufp, szp, pos, offs));
1.182 schwarze 762: }
1.228 schwarze 763:
764: /* Skip empty request lines. */
765:
766: if ((*bufp)[pos] == '"') {
767: mandoc_msg(MANDOCERR_COMMENT_BAD, r->parse,
768: ln, pos, NULL);
769: return(ROFF_IGN);
770: } else if ((*bufp)[pos] == '\0')
771: return(ROFF_IGN);
1.67 kristaps 772:
1.79 kristaps 773: /*
774: * If a scope is open, go to the child handler for that macro,
775: * as it may want to preprocess before doing anything with it.
1.125 kristaps 776: * Don't do so if an equation is open.
1.79 kristaps 777: */
1.78 kristaps 778:
1.79 kristaps 779: if (r->last) {
780: t = r->last->tok;
781: assert(roffs[t].sub);
1.207 schwarze 782: return((*roffs[t].sub)(r, t, bufp, szp,
783: ln, ppos, pos, offs));
1.79 kristaps 784: }
1.78 kristaps 785:
1.79 kristaps 786: /*
787: * Lastly, as we've no scope open, try to look up and execute
788: * the new macro. If no macro is found, simply return and let
789: * the compilers handle it.
790: */
1.67 kristaps 791:
1.214 schwarze 792: if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos, ln, ppos)))
1.79 kristaps 793: return(ROFF_CONT);
1.67 kristaps 794:
1.75 kristaps 795: assert(roffs[t].proc);
1.207 schwarze 796: return((*roffs[t].proc)(r, t, bufp, szp, ln, ppos, pos, offs));
1.74 kristaps 797: }
798:
1.117 kristaps 799: void
1.74 kristaps 800: roff_endparse(struct roff *r)
801: {
802:
1.110 kristaps 803: if (r->last)
1.221 schwarze 804: mandoc_msg(MANDOCERR_BLK_NOEND, r->parse,
805: r->last->line, r->last->col,
806: roffs[r->last->tok].name);
1.117 kristaps 807:
1.125 kristaps 808: if (r->eqn) {
1.221 schwarze 809: mandoc_msg(MANDOCERR_BLK_NOEND, r->parse,
810: r->eqn->eqn.ln, r->eqn->eqn.pos, "EQ");
1.151 kristaps 811: eqn_end(&r->eqn);
1.125 kristaps 812: }
813:
1.117 kristaps 814: if (r->tbl) {
1.221 schwarze 815: mandoc_msg(MANDOCERR_BLK_NOEND, r->parse,
816: r->tbl->line, r->tbl->pos, "TS");
1.151 kristaps 817: tbl_end(&r->tbl);
1.117 kristaps 818: }
1.67 kristaps 819: }
820:
821: /*
822: * Parse a roff node's type from the input buffer. This must be in the
823: * form of ".foo xxx" in the usual way.
824: */
825: static enum rofft
1.214 schwarze 826: roff_parse(struct roff *r, char *buf, int *pos, int ln, int ppos)
1.67 kristaps 827: {
1.214 schwarze 828: char *cp;
1.106 kristaps 829: const char *mac;
830: size_t maclen;
1.67 kristaps 831: enum rofft t;
832:
1.214 schwarze 833: cp = buf + *pos;
834:
835: if ('\0' == *cp || '"' == *cp || '\t' == *cp || ' ' == *cp)
1.67 kristaps 836: return(ROFF_MAX);
837:
1.214 schwarze 838: mac = cp;
839: maclen = roff_getname(r, &cp, ln, ppos);
1.67 kristaps 840:
1.106 kristaps 841: t = (r->current_string = roff_getstrn(r, mac, maclen))
1.155 kristaps 842: ? ROFF_USERDEF : roffhash_find(mac, maclen);
1.67 kristaps 843:
1.214 schwarze 844: if (ROFF_MAX != t)
845: *pos = cp - buf;
1.67 kristaps 846:
847: return(t);
848: }
849:
850: static enum rofferr
1.76 kristaps 851: roff_cblock(ROFF_ARGS)
1.67 kristaps 852: {
853:
1.79 kristaps 854: /*
855: * A block-close `..' should only be invoked as a child of an
856: * ignore macro, otherwise raise a warning and just ignore it.
857: */
858:
1.76 kristaps 859: if (NULL == r->last) {
1.221 schwarze 860: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
861: ln, ppos, "..");
1.76 kristaps 862: return(ROFF_IGN);
863: }
1.67 kristaps 864:
1.81 kristaps 865: switch (r->last->tok) {
1.207 schwarze 866: case ROFF_am:
1.220 schwarze 867: /* ROFF_am1 is remapped to ROFF_am in roff_block(). */
1.81 kristaps 868: /* FALLTHROUGH */
1.207 schwarze 869: case ROFF_ami:
1.81 kristaps 870: /* FALLTHROUGH */
1.207 schwarze 871: case ROFF_de:
1.108 schwarze 872: /* ROFF_de1 is remapped to ROFF_de in roff_block(). */
1.81 kristaps 873: /* FALLTHROUGH */
1.207 schwarze 874: case ROFF_dei:
1.81 kristaps 875: /* FALLTHROUGH */
1.207 schwarze 876: case ROFF_ig:
1.81 kristaps 877: break;
878: default:
1.221 schwarze 879: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
880: ln, ppos, "..");
1.67 kristaps 881: return(ROFF_IGN);
1.76 kristaps 882: }
1.67 kristaps 883:
1.76 kristaps 884: if ((*bufp)[pos])
1.217 schwarze 885: mandoc_vmsg(MANDOCERR_ARG_SKIP, r->parse, ln, pos,
886: ".. %s", *bufp + pos);
1.71 kristaps 887:
888: roffnode_pop(r);
1.76 kristaps 889: roffnode_cleanscope(r);
890: return(ROFF_IGN);
1.71 kristaps 891:
1.67 kristaps 892: }
893:
1.76 kristaps 894: static void
895: roffnode_cleanscope(struct roff *r)
1.67 kristaps 896: {
897:
1.76 kristaps 898: while (r->last) {
1.173 schwarze 899: if (--r->last->endspan != 0)
1.76 kristaps 900: break;
901: roffnode_pop(r);
902: }
1.67 kristaps 903: }
904:
1.195 schwarze 905: static void
906: roff_ccond(struct roff *r, int ln, int ppos)
1.74 kristaps 907: {
908:
1.76 kristaps 909: if (NULL == r->last) {
1.221 schwarze 910: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
911: ln, ppos, "\\}");
1.195 schwarze 912: return;
1.76 kristaps 913: }
914:
1.82 kristaps 915: switch (r->last->tok) {
1.207 schwarze 916: case ROFF_el:
1.82 kristaps 917: /* FALLTHROUGH */
1.207 schwarze 918: case ROFF_ie:
1.82 kristaps 919: /* FALLTHROUGH */
1.207 schwarze 920: case ROFF_if:
1.82 kristaps 921: break;
922: default:
1.221 schwarze 923: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
924: ln, ppos, "\\}");
1.195 schwarze 925: return;
1.75 kristaps 926: }
927:
1.76 kristaps 928: if (r->last->endspan > -1) {
1.221 schwarze 929: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
930: ln, ppos, "\\}");
1.195 schwarze 931: return;
1.76 kristaps 932: }
933:
1.75 kristaps 934: roffnode_pop(r);
1.76 kristaps 935: roffnode_cleanscope(r);
1.195 schwarze 936: return;
1.76 kristaps 937: }
938:
939: static enum rofferr
1.80 kristaps 940: roff_block(ROFF_ARGS)
1.76 kristaps 941: {
1.220 schwarze 942: const char *name;
943: char *iname, *cp;
1.213 schwarze 944: size_t namesz;
1.106 kristaps 945:
1.220 schwarze 946: /* Ignore groff compatibility mode for now. */
1.76 kristaps 947:
1.220 schwarze 948: if (ROFF_de1 == tok)
949: tok = ROFF_de;
950: else if (ROFF_am1 == tok)
951: tok = ROFF_am;
952:
953: /* Parse the macro name argument. */
954:
955: cp = *bufp + pos;
956: if (ROFF_ig == tok) {
957: iname = NULL;
958: namesz = 0;
959: } else {
960: iname = cp;
961: namesz = roff_getname(r, &cp, ln, ppos);
962: iname[namesz] = '\0';
963: }
1.107 kristaps 964:
1.220 schwarze 965: /* Resolve the macro name argument if it is indirect. */
1.107 kristaps 966:
1.220 schwarze 967: if (namesz && (ROFF_dei == tok || ROFF_ami == tok)) {
968: if (NULL == (name = roff_getstrn(r, iname, namesz))) {
969: mandoc_vmsg(MANDOCERR_STR_UNDEF,
970: r->parse, ln, (int)(iname - *bufp),
971: "%.*s", (int)namesz, iname);
972: namesz = 0;
973: } else
974: namesz = strlen(name);
975: } else
976: name = iname;
1.107 kristaps 977:
1.220 schwarze 978: if (0 == namesz && ROFF_ig != tok) {
979: mandoc_msg(MANDOCERR_REQ_EMPTY, r->parse,
980: ln, ppos, roffs[tok].name);
981: return(ROFF_IGN);
982: }
1.80 kristaps 983:
1.106 kristaps 984: roffnode_push(r, tok, name, ln, ppos);
985:
986: /*
987: * At the beginning of a `de' macro, clear the existing string
988: * with the same name, if there is one. New content will be
1.193 schwarze 989: * appended from roff_block_text() in multiline mode.
1.106 kristaps 990: */
1.107 kristaps 991:
1.220 schwarze 992: if (ROFF_de == tok || ROFF_dei == tok)
1.213 schwarze 993: roff_setstrn(&r->strtab, name, namesz, "", 0, 0);
1.76 kristaps 994:
1.213 schwarze 995: if ('\0' == *cp)
1.78 kristaps 996: return(ROFF_IGN);
997:
1.220 schwarze 998: /* Get the custom end marker. */
1.107 kristaps 999:
1.220 schwarze 1000: iname = cp;
1.213 schwarze 1001: namesz = roff_getname(r, &cp, ln, ppos);
1.220 schwarze 1002:
1003: /* Resolve the end marker if it is indirect. */
1004:
1005: if (namesz && (ROFF_dei == tok || ROFF_ami == tok)) {
1006: if (NULL == (name = roff_getstrn(r, iname, namesz))) {
1007: mandoc_vmsg(MANDOCERR_STR_UNDEF,
1008: r->parse, ln, (int)(iname - *bufp),
1009: "%.*s", (int)namesz, iname);
1010: namesz = 0;
1011: } else
1012: namesz = strlen(name);
1013: } else
1014: name = iname;
1015:
1.213 schwarze 1016: if (namesz)
1017: r->last->end = mandoc_strndup(name, namesz);
1.78 kristaps 1018:
1.213 schwarze 1019: if ('\0' != *cp)
1.217 schwarze 1020: mandoc_vmsg(MANDOCERR_ARG_EXCESS, r->parse,
1021: ln, pos, ".%s ... %s", roffs[tok].name, cp);
1.74 kristaps 1022:
1.78 kristaps 1023: return(ROFF_IGN);
1024: }
1025:
1026: static enum rofferr
1.80 kristaps 1027: roff_block_sub(ROFF_ARGS)
1.79 kristaps 1028: {
1029: enum rofft t;
1030: int i, j;
1031:
1032: /*
1033: * First check whether a custom macro exists at this level. If
1034: * it does, then check against it. This is some of groff's
1035: * stranger behaviours. If we encountered a custom end-scope
1036: * tag and that tag also happens to be a "real" macro, then we
1037: * need to try interpreting it again as a real macro. If it's
1038: * not, then return ignore. Else continue.
1039: */
1040:
1041: if (r->last->end) {
1.130 kristaps 1042: for (i = pos, j = 0; r->last->end[j]; j++, i++)
1.79 kristaps 1043: if ((*bufp)[i] != r->last->end[j])
1044: break;
1045:
1.207 schwarze 1046: if ('\0' == r->last->end[j] &&
1047: ('\0' == (*bufp)[i] ||
1048: ' ' == (*bufp)[i] ||
1049: '\t' == (*bufp)[i])) {
1.79 kristaps 1050: roffnode_pop(r);
1051: roffnode_cleanscope(r);
1052:
1.130 kristaps 1053: while (' ' == (*bufp)[i] || '\t' == (*bufp)[i])
1054: i++;
1055:
1056: pos = i;
1.214 schwarze 1057: if (ROFF_MAX != roff_parse(r, *bufp, &pos, ln, ppos))
1.79 kristaps 1058: return(ROFF_RERUN);
1059: return(ROFF_IGN);
1060: }
1061: }
1062:
1063: /*
1064: * If we have no custom end-query or lookup failed, then try
1065: * pulling it out of the hashtable.
1066: */
1067:
1.214 schwarze 1068: t = roff_parse(r, *bufp, &pos, ln, ppos);
1.79 kristaps 1069:
1.106 kristaps 1070: if (ROFF_cblock != t) {
1.220 schwarze 1071: if (ROFF_ig != tok)
1.193 schwarze 1072: roff_setstr(r, r->last->name, *bufp + ppos, 2);
1.79 kristaps 1073: return(ROFF_IGN);
1.106 kristaps 1074: }
1.79 kristaps 1075:
1076: assert(roffs[t].proc);
1.207 schwarze 1077: return((*roffs[t].proc)(r, t, bufp, szp, ln, ppos, pos, offs));
1.79 kristaps 1078: }
1079:
1080: static enum rofferr
1.80 kristaps 1081: roff_block_text(ROFF_ARGS)
1.78 kristaps 1082: {
1083:
1.220 schwarze 1084: if (ROFF_ig != tok)
1.193 schwarze 1085: roff_setstr(r, r->last->name, *bufp + pos, 2);
1.106 kristaps 1086:
1.78 kristaps 1087: return(ROFF_IGN);
1088: }
1089:
1090: static enum rofferr
1.82 kristaps 1091: roff_cond_sub(ROFF_ARGS)
1092: {
1093: enum rofft t;
1.139 kristaps 1094: char *ep;
1.198 schwarze 1095: int rr;
1.82 kristaps 1096:
1097: rr = r->last->rule;
1.139 kristaps 1098: roffnode_cleanscope(r);
1.214 schwarze 1099: t = roff_parse(r, *bufp, &pos, ln, ppos);
1.82 kristaps 1100:
1.139 kristaps 1101: /*
1.177 schwarze 1102: * Fully handle known macros when they are structurally
1103: * required or when the conditional evaluated to true.
1.87 kristaps 1104: */
1105:
1.177 schwarze 1106: if ((ROFF_MAX != t) &&
1.198 schwarze 1107: (rr || ROFFMAC_STRUCT & roffs[t].flags)) {
1.177 schwarze 1108: assert(roffs[t].proc);
1109: return((*roffs[t].proc)(r, t, bufp, szp,
1.207 schwarze 1110: ln, ppos, pos, offs));
1.177 schwarze 1111: }
1.144 kristaps 1112:
1.196 schwarze 1113: /*
1114: * If `\}' occurs on a macro line without a preceding macro,
1115: * drop the line completely.
1116: */
1117:
1118: ep = *bufp + pos;
1119: if ('\\' == ep[0] && '}' == ep[1])
1.198 schwarze 1120: rr = 0;
1.196 schwarze 1121:
1.177 schwarze 1122: /* Always check for the closing delimiter `\}'. */
1.144 kristaps 1123:
1.177 schwarze 1124: while (NULL != (ep = strchr(ep, '\\'))) {
1.197 schwarze 1125: if ('}' == *(++ep)) {
1126: *ep = '&';
1127: roff_ccond(r, ln, ep - *bufp - 1);
1128: }
1129: ++ep;
1.177 schwarze 1130: }
1.198 schwarze 1131: return(rr ? ROFF_CONT : ROFF_IGN);
1.82 kristaps 1132: }
1133:
1134: static enum rofferr
1135: roff_cond_text(ROFF_ARGS)
1.78 kristaps 1136: {
1.140 kristaps 1137: char *ep;
1.198 schwarze 1138: int rr;
1.82 kristaps 1139:
1140: rr = r->last->rule;
1.140 kristaps 1141: roffnode_cleanscope(r);
1.82 kristaps 1142:
1.197 schwarze 1143: ep = *bufp + pos;
1144: while (NULL != (ep = strchr(ep, '\\'))) {
1145: if ('}' == *(++ep)) {
1146: *ep = '&';
1147: roff_ccond(r, ln, ep - *bufp - 1);
1148: }
1149: ++ep;
1.78 kristaps 1150: }
1.198 schwarze 1151: return(rr ? ROFF_CONT : ROFF_IGN);
1.74 kristaps 1152: }
1153:
1.204 schwarze 1154: /*
1155: * Parse a single signed integer number. Stop at the first non-digit.
1156: * If there is at least one digit, return success and advance the
1157: * parse point, else return failure and let the parse point unchanged.
1158: * Ignore overflows, treat them just like the C language.
1159: */
1.184 schwarze 1160: static int
1161: roff_getnum(const char *v, int *pos, int *res)
1162: {
1.206 schwarze 1163: int myres, n, p;
1164:
1165: if (NULL == res)
1166: res = &myres;
1.184 schwarze 1167:
1168: p = *pos;
1169: n = v[p] == '-';
1170: if (n)
1171: p++;
1172:
1173: for (*res = 0; isdigit((unsigned char)v[p]); p++)
1.204 schwarze 1174: *res = 10 * *res + v[p] - '0';
1.184 schwarze 1175: if (p == *pos + n)
1176: return 0;
1177:
1178: if (n)
1179: *res = -*res;
1180:
1181: *pos = p;
1182: return 1;
1183: }
1184:
1.198 schwarze 1185: /*
1186: * Evaluate a string comparison condition.
1187: * The first character is the delimiter.
1188: * Succeed if the string up to its second occurrence
1189: * matches the string up to its third occurence.
1190: * Advance the cursor after the third occurrence
1191: * or lacking that, to the end of the line.
1192: */
1193: static int
1194: roff_evalstrcond(const char *v, int *pos)
1195: {
1196: const char *s1, *s2, *s3;
1197: int match;
1198:
1199: match = 0;
1200: s1 = v + *pos; /* initial delimiter */
1201: s2 = s1 + 1; /* for scanning the first string */
1202: s3 = strchr(s2, *s1); /* for scanning the second string */
1203:
1204: if (NULL == s3) /* found no middle delimiter */
1205: goto out;
1206:
1207: while ('\0' != *++s3) {
1208: if (*s2 != *s3) { /* mismatch */
1209: s3 = strchr(s3, *s1);
1210: break;
1211: }
1212: if (*s3 == *s1) { /* found the final delimiter */
1213: match = 1;
1214: break;
1215: }
1216: s2++;
1217: }
1218:
1219: out:
1220: if (NULL == s3)
1221: s3 = strchr(s2, '\0');
1222: else
1223: s3++;
1224: *pos = s3 - v;
1225: return(match);
1226: }
1227:
1.204 schwarze 1228: /*
1229: * Evaluate an optionally negated single character, numerical,
1230: * or string condition.
1231: */
1.198 schwarze 1232: static int
1.88 kristaps 1233: roff_evalcond(const char *v, int *pos)
1234: {
1.204 schwarze 1235: int wanttrue, number;
1.88 kristaps 1236:
1.198 schwarze 1237: if ('!' == v[*pos]) {
1238: wanttrue = 0;
1239: (*pos)++;
1240: } else
1241: wanttrue = 1;
1242:
1.88 kristaps 1243: switch (v[*pos]) {
1.207 schwarze 1244: case 'n':
1.198 schwarze 1245: /* FALLTHROUGH */
1.207 schwarze 1246: case 'o':
1.88 kristaps 1247: (*pos)++;
1.198 schwarze 1248: return(wanttrue);
1.207 schwarze 1249: case 'c':
1.198 schwarze 1250: /* FALLTHROUGH */
1.207 schwarze 1251: case 'd':
1.198 schwarze 1252: /* FALLTHROUGH */
1.207 schwarze 1253: case 'e':
1.88 kristaps 1254: /* FALLTHROUGH */
1.207 schwarze 1255: case 'r':
1.88 kristaps 1256: /* FALLTHROUGH */
1.207 schwarze 1257: case 't':
1.88 kristaps 1258: (*pos)++;
1.198 schwarze 1259: return(!wanttrue);
1.88 kristaps 1260: default:
1261: break;
1262: }
1263:
1.204 schwarze 1264: if (roff_evalnum(v, pos, &number, 0))
1265: return((number > 0) == wanttrue);
1266: else
1.198 schwarze 1267: return(roff_evalstrcond(v, pos) == wanttrue);
1.88 kristaps 1268: }
1269:
1.74 kristaps 1270: static enum rofferr
1.103 kristaps 1271: roff_line_ignore(ROFF_ARGS)
1.89 kristaps 1272: {
1.123 schwarze 1273:
1.89 kristaps 1274: return(ROFF_IGN);
1275: }
1276:
1.104 kristaps 1277: static enum rofferr
1.82 kristaps 1278: roff_cond(ROFF_ARGS)
1.74 kristaps 1279: {
1.173 schwarze 1280:
1281: roffnode_push(r, tok, NULL, ln, ppos);
1.74 kristaps 1282:
1.207 schwarze 1283: /*
1.134 kristaps 1284: * An `.el' has no conditional body: it will consume the value
1285: * of the current rstack entry set in prior `ie' calls or
1.207 schwarze 1286: * defaults to DENY.
1.134 kristaps 1287: *
1288: * If we're not an `el', however, then evaluate the conditional.
1289: */
1.133 kristaps 1290:
1.173 schwarze 1291: r->last->rule = ROFF_el == tok ?
1.207 schwarze 1292: (r->rstackpos < 0 ? 0 : r->rstack[r->rstackpos--]) :
1293: roff_evalcond(*bufp, &pos);
1.77 kristaps 1294:
1.134 kristaps 1295: /*
1296: * An if-else will put the NEGATION of the current evaluated
1297: * conditional into the stack of rules.
1298: */
1299:
1.84 schwarze 1300: if (ROFF_ie == tok) {
1.223 schwarze 1301: if (r->rstackpos + 1 == r->rstacksz) {
1302: r->rstacksz += 16;
1303: r->rstack = mandoc_reallocarray(r->rstack,
1304: r->rstacksz, sizeof(int));
1.134 kristaps 1305: }
1.198 schwarze 1306: r->rstack[++r->rstackpos] = !r->last->rule;
1.82 kristaps 1307: }
1.88 kristaps 1308:
1309: /* If the parent has false as its rule, then so do we. */
1310:
1.198 schwarze 1311: if (r->last->parent && !r->last->parent->rule)
1312: r->last->rule = 0;
1.88 kristaps 1313:
1314: /*
1.173 schwarze 1315: * Determine scope.
1316: * If there is nothing on the line after the conditional,
1317: * not even whitespace, use next-line scope.
1.88 kristaps 1318: */
1.74 kristaps 1319:
1.173 schwarze 1320: if ('\0' == (*bufp)[pos]) {
1321: r->last->endspan = 2;
1322: goto out;
1323: }
1324:
1325: while (' ' == (*bufp)[pos])
1326: pos++;
1327:
1328: /* An opening brace requests multiline scope. */
1.75 kristaps 1329:
1330: if ('\\' == (*bufp)[pos] && '{' == (*bufp)[pos + 1]) {
1331: r->last->endspan = -1;
1332: pos += 2;
1.173 schwarze 1333: goto out;
1.207 schwarze 1334: }
1.74 kristaps 1335:
1.77 kristaps 1336: /*
1.173 schwarze 1337: * Anything else following the conditional causes
1338: * single-line scope. Warn if the scope contains
1339: * nothing but trailing whitespace.
1.77 kristaps 1340: */
1341:
1.75 kristaps 1342: if ('\0' == (*bufp)[pos])
1.216 schwarze 1343: mandoc_msg(MANDOCERR_COND_EMPTY, r->parse,
1344: ln, ppos, roffs[tok].name);
1.77 kristaps 1345:
1.173 schwarze 1346: r->last->endspan = 1;
1.74 kristaps 1347:
1.173 schwarze 1348: out:
1.75 kristaps 1349: *offs = pos;
1350: return(ROFF_RERUN);
1.83 schwarze 1351: }
1352:
1353: static enum rofferr
1.92 schwarze 1354: roff_ds(ROFF_ARGS)
1355: {
1.212 schwarze 1356: char *string;
1357: const char *name;
1358: size_t namesz;
1.96 kristaps 1359:
1360: /*
1.212 schwarze 1361: * The first word is the name of the string.
1362: * If it is empty or terminated by an escape sequence,
1363: * abort the `ds' request without defining anything.
1.96 kristaps 1364: */
1.92 schwarze 1365:
1.212 schwarze 1366: name = string = *bufp + pos;
1.92 schwarze 1367: if ('\0' == *name)
1368: return(ROFF_IGN);
1369:
1.212 schwarze 1370: namesz = roff_getname(r, &string, ln, pos);
1371: if ('\\' == name[namesz])
1372: return(ROFF_IGN);
1373:
1374: /* Read past the initial double-quote, if any. */
1.121 schwarze 1375: if ('"' == *string)
1.92 schwarze 1376: string++;
1377:
1.96 kristaps 1378: /* The rest is the value. */
1.212 schwarze 1379: roff_setstrn(&r->strtab, name, namesz, string, strlen(string),
1380: ROFF_as == tok);
1.92 schwarze 1381: return(ROFF_IGN);
1382: }
1383:
1.204 schwarze 1384: /*
1385: * Parse a single operator, one or two characters long.
1386: * If the operator is recognized, return success and advance the
1387: * parse point, else return failure and let the parse point unchanged.
1388: */
1389: static int
1390: roff_getop(const char *v, int *pos, char *res)
1391: {
1392:
1393: *res = v[*pos];
1394:
1395: switch (*res) {
1.207 schwarze 1396: case '+':
1.204 schwarze 1397: /* FALLTHROUGH */
1.207 schwarze 1398: case '-':
1.204 schwarze 1399: /* FALLTHROUGH */
1.207 schwarze 1400: case '*':
1.204 schwarze 1401: /* FALLTHROUGH */
1.207 schwarze 1402: case '/':
1.204 schwarze 1403: /* FALLTHROUGH */
1.207 schwarze 1404: case '%':
1.204 schwarze 1405: /* FALLTHROUGH */
1.207 schwarze 1406: case '&':
1.204 schwarze 1407: /* FALLTHROUGH */
1.207 schwarze 1408: case ':':
1.204 schwarze 1409: break;
1410: case '<':
1411: switch (v[*pos + 1]) {
1.207 schwarze 1412: case '=':
1.204 schwarze 1413: *res = 'l';
1414: (*pos)++;
1415: break;
1.207 schwarze 1416: case '>':
1.204 schwarze 1417: *res = '!';
1418: (*pos)++;
1419: break;
1.207 schwarze 1420: case '?':
1.204 schwarze 1421: *res = 'i';
1422: (*pos)++;
1423: break;
1424: default:
1425: break;
1426: }
1427: break;
1428: case '>':
1429: switch (v[*pos + 1]) {
1.207 schwarze 1430: case '=':
1.204 schwarze 1431: *res = 'g';
1432: (*pos)++;
1433: break;
1.207 schwarze 1434: case '?':
1.204 schwarze 1435: *res = 'a';
1436: (*pos)++;
1437: break;
1438: default:
1439: break;
1440: }
1441: break;
1442: case '=':
1443: if ('=' == v[*pos + 1])
1444: (*pos)++;
1445: break;
1446: default:
1447: return(0);
1448: }
1449: (*pos)++;
1450:
1451: return(*res);
1452: }
1453:
1454: /*
1455: * Evaluate either a parenthesized numeric expression
1456: * or a single signed integer number.
1457: */
1458: static int
1459: roff_evalpar(const char *v, int *pos, int *res)
1460: {
1461:
1462: if ('(' != v[*pos])
1463: return(roff_getnum(v, pos, res));
1464:
1465: (*pos)++;
1466: if ( ! roff_evalnum(v, pos, res, 1))
1467: return(0);
1468:
1.206 schwarze 1469: /*
1470: * Omission of the closing parenthesis
1471: * is an error in validation mode,
1472: * but ignored in evaluation mode.
1473: */
1474:
1.204 schwarze 1475: if (')' == v[*pos])
1476: (*pos)++;
1.206 schwarze 1477: else if (NULL == res)
1478: return(0);
1.204 schwarze 1479:
1480: return(1);
1481: }
1482:
1483: /*
1484: * Evaluate a complete numeric expression.
1485: * Proceed left to right, there is no concept of precedence.
1486: */
1487: static int
1488: roff_evalnum(const char *v, int *pos, int *res, int skipwhite)
1489: {
1490: int mypos, operand2;
1491: char operator;
1492:
1493: if (NULL == pos) {
1494: mypos = 0;
1495: pos = &mypos;
1496: }
1497:
1498: if (skipwhite)
1499: while (isspace((unsigned char)v[*pos]))
1500: (*pos)++;
1501:
1502: if ( ! roff_evalpar(v, pos, res))
1503: return(0);
1504:
1505: while (1) {
1506: if (skipwhite)
1507: while (isspace((unsigned char)v[*pos]))
1508: (*pos)++;
1509:
1510: if ( ! roff_getop(v, pos, &operator))
1511: break;
1512:
1513: if (skipwhite)
1514: while (isspace((unsigned char)v[*pos]))
1515: (*pos)++;
1516:
1517: if ( ! roff_evalpar(v, pos, &operand2))
1518: return(0);
1519:
1520: if (skipwhite)
1521: while (isspace((unsigned char)v[*pos]))
1522: (*pos)++;
1.206 schwarze 1523:
1524: if (NULL == res)
1525: continue;
1.204 schwarze 1526:
1527: switch (operator) {
1.207 schwarze 1528: case '+':
1.204 schwarze 1529: *res += operand2;
1530: break;
1.207 schwarze 1531: case '-':
1.204 schwarze 1532: *res -= operand2;
1533: break;
1.207 schwarze 1534: case '*':
1.204 schwarze 1535: *res *= operand2;
1536: break;
1.207 schwarze 1537: case '/':
1.204 schwarze 1538: *res /= operand2;
1539: break;
1.207 schwarze 1540: case '%':
1.204 schwarze 1541: *res %= operand2;
1542: break;
1.207 schwarze 1543: case '<':
1.204 schwarze 1544: *res = *res < operand2;
1545: break;
1.207 schwarze 1546: case '>':
1.204 schwarze 1547: *res = *res > operand2;
1548: break;
1.207 schwarze 1549: case 'l':
1.204 schwarze 1550: *res = *res <= operand2;
1551: break;
1.207 schwarze 1552: case 'g':
1.204 schwarze 1553: *res = *res >= operand2;
1554: break;
1.207 schwarze 1555: case '=':
1.204 schwarze 1556: *res = *res == operand2;
1557: break;
1.207 schwarze 1558: case '!':
1.204 schwarze 1559: *res = *res != operand2;
1560: break;
1.207 schwarze 1561: case '&':
1.204 schwarze 1562: *res = *res && operand2;
1563: break;
1.207 schwarze 1564: case ':':
1.204 schwarze 1565: *res = *res || operand2;
1566: break;
1.207 schwarze 1567: case 'i':
1.204 schwarze 1568: if (operand2 < *res)
1569: *res = operand2;
1570: break;
1.207 schwarze 1571: case 'a':
1.204 schwarze 1572: if (operand2 > *res)
1573: *res = operand2;
1574: break;
1575: default:
1576: abort();
1577: }
1578: }
1579: return(1);
1580: }
1581:
1.180 schwarze 1582: void
1.187 schwarze 1583: roff_setreg(struct roff *r, const char *name, int val, char sign)
1.147 kristaps 1584: {
1.180 schwarze 1585: struct roffreg *reg;
1586:
1587: /* Search for an existing register with the same name. */
1588: reg = r->regtab;
1589:
1590: while (reg && strcmp(name, reg->key.p))
1591: reg = reg->next;
1.147 kristaps 1592:
1.180 schwarze 1593: if (NULL == reg) {
1594: /* Create a new register. */
1595: reg = mandoc_malloc(sizeof(struct roffreg));
1596: reg->key.p = mandoc_strdup(name);
1597: reg->key.sz = strlen(name);
1.187 schwarze 1598: reg->val = 0;
1.180 schwarze 1599: reg->next = r->regtab;
1600: r->regtab = reg;
1601: }
1602:
1.187 schwarze 1603: if ('+' == sign)
1604: reg->val += val;
1605: else if ('-' == sign)
1606: reg->val -= val;
1607: else
1608: reg->val = val;
1.147 kristaps 1609: }
1610:
1.192 schwarze 1611: /*
1612: * Handle some predefined read-only number registers.
1613: * For now, return -1 if the requested register is not predefined;
1614: * in case a predefined read-only register having the value -1
1615: * were to turn up, another special value would have to be chosen.
1616: */
1617: static int
1618: roff_getregro(const char *name)
1619: {
1620:
1621: switch (*name) {
1.207 schwarze 1622: case 'A': /* ASCII approximation mode is always off. */
1.192 schwarze 1623: return(0);
1.207 schwarze 1624: case 'g': /* Groff compatibility mode is always on. */
1.192 schwarze 1625: return(1);
1.207 schwarze 1626: case 'H': /* Fixed horizontal resolution. */
1.192 schwarze 1627: return (24);
1.207 schwarze 1628: case 'j': /* Always adjust left margin only. */
1.192 schwarze 1629: return(0);
1.207 schwarze 1630: case 'T': /* Some output device is always defined. */
1.192 schwarze 1631: return(1);
1.207 schwarze 1632: case 'V': /* Fixed vertical resolution. */
1.192 schwarze 1633: return (40);
1634: default:
1635: return (-1);
1636: }
1637: }
1638:
1.181 schwarze 1639: int
1.180 schwarze 1640: roff_getreg(const struct roff *r, const char *name)
1.147 kristaps 1641: {
1.180 schwarze 1642: struct roffreg *reg;
1.192 schwarze 1643: int val;
1644:
1645: if ('.' == name[0] && '\0' != name[1] && '\0' == name[2]) {
1646: val = roff_getregro(name + 1);
1647: if (-1 != val)
1648: return (val);
1649: }
1.180 schwarze 1650:
1651: for (reg = r->regtab; reg; reg = reg->next)
1652: if (0 == strcmp(name, reg->key.p))
1.181 schwarze 1653: return(reg->val);
1654:
1655: return(0);
1656: }
1657:
1658: static int
1659: roff_getregn(const struct roff *r, const char *name, size_t len)
1660: {
1661: struct roffreg *reg;
1.192 schwarze 1662: int val;
1663:
1664: if ('.' == name[0] && 2 == len) {
1665: val = roff_getregro(name + 1);
1666: if (-1 != val)
1667: return (val);
1668: }
1.181 schwarze 1669:
1670: for (reg = r->regtab; reg; reg = reg->next)
1671: if (len == reg->key.sz &&
1672: 0 == strncmp(name, reg->key.p, len))
1673: return(reg->val);
1.147 kristaps 1674:
1.180 schwarze 1675: return(0);
1.147 kristaps 1676: }
1677:
1.180 schwarze 1678: static void
1679: roff_freereg(struct roffreg *reg)
1.147 kristaps 1680: {
1.180 schwarze 1681: struct roffreg *old_reg;
1.147 kristaps 1682:
1.180 schwarze 1683: while (NULL != reg) {
1684: free(reg->key.p);
1685: old_reg = reg;
1686: reg = reg->next;
1687: free(old_reg);
1688: }
1.147 kristaps 1689: }
1.92 schwarze 1690:
1691: static enum rofferr
1.89 kristaps 1692: roff_nr(ROFF_ARGS)
1.83 schwarze 1693: {
1.212 schwarze 1694: char *key, *val;
1695: size_t keysz;
1.138 kristaps 1696: int iv;
1.187 schwarze 1697: char sign;
1.89 kristaps 1698:
1.212 schwarze 1699: key = val = *bufp + pos;
1700: if ('\0' == *key)
1701: return(ROFF_IGN);
1702:
1703: keysz = roff_getname(r, &val, ln, pos);
1704: if ('\\' == key[keysz])
1705: return(ROFF_IGN);
1706: key[keysz] = '\0';
1.89 kristaps 1707:
1.187 schwarze 1708: sign = *val;
1709: if ('+' == sign || '-' == sign)
1710: val++;
1711:
1.204 schwarze 1712: if (roff_evalnum(val, NULL, &iv, 0))
1713: roff_setreg(r, key, iv, sign);
1.109 kristaps 1714:
1.203 schwarze 1715: return(ROFF_IGN);
1716: }
1717:
1718: static enum rofferr
1719: roff_rr(ROFF_ARGS)
1720: {
1721: struct roffreg *reg, **prev;
1.212 schwarze 1722: char *name, *cp;
1723: size_t namesz;
1.203 schwarze 1724:
1.212 schwarze 1725: name = cp = *bufp + pos;
1726: if ('\0' == *name)
1727: return(ROFF_IGN);
1728: namesz = roff_getname(r, &cp, ln, pos);
1729: name[namesz] = '\0';
1.203 schwarze 1730:
1731: prev = &r->regtab;
1732: while (1) {
1733: reg = *prev;
1734: if (NULL == reg || !strcmp(name, reg->key.p))
1735: break;
1736: prev = ®->next;
1737: }
1738: if (NULL != reg) {
1739: *prev = reg->next;
1740: free(reg->key.p);
1741: free(reg);
1742: }
1.122 schwarze 1743: return(ROFF_IGN);
1744: }
1745:
1746: static enum rofferr
1747: roff_rm(ROFF_ARGS)
1748: {
1749: const char *name;
1750: char *cp;
1.212 schwarze 1751: size_t namesz;
1.122 schwarze 1752:
1753: cp = *bufp + pos;
1754: while ('\0' != *cp) {
1.212 schwarze 1755: name = cp;
1756: namesz = roff_getname(r, &cp, ln, (int)(cp - *bufp));
1757: roff_setstrn(&r->strtab, name, namesz, NULL, 0, 0);
1758: if ('\\' == name[namesz])
1759: break;
1.122 schwarze 1760: }
1.178 schwarze 1761: return(ROFF_IGN);
1762: }
1763:
1764: static enum rofferr
1765: roff_it(ROFF_ARGS)
1766: {
1767: char *cp;
1768: size_t len;
1769: int iv;
1770:
1771: /* Parse the number of lines. */
1772: cp = *bufp + pos;
1773: len = strcspn(cp, " \t");
1774: cp[len] = '\0';
1775: if ((iv = mandoc_strntoi(cp, len, 10)) <= 0) {
1.222 schwarze 1776: mandoc_msg(MANDOCERR_IT_NONUM, r->parse,
1.207 schwarze 1777: ln, ppos, *bufp + 1);
1.178 schwarze 1778: return(ROFF_IGN);
1779: }
1780: cp += len + 1;
1781:
1782: /* Arm the input line trap. */
1783: roffit_lines = iv;
1784: roffit_macro = mandoc_strdup(cp);
1.109 kristaps 1785: return(ROFF_IGN);
1.175 schwarze 1786: }
1787:
1788: static enum rofferr
1789: roff_Dd(ROFF_ARGS)
1790: {
1791: const char *const *cp;
1792:
1.227 schwarze 1793: if ((r->options & (MPARSE_MDOC | MPARSE_QUICK)) == 0)
1.175 schwarze 1794: for (cp = __mdoc_reserved; *cp; cp++)
1795: roff_setstr(r, *cp, NULL, 0);
1796:
1.227 schwarze 1797: if (r->format == 0)
1798: r->format = MPARSE_MDOC;
1799:
1.175 schwarze 1800: return(ROFF_CONT);
1801: }
1802:
1803: static enum rofferr
1804: roff_TH(ROFF_ARGS)
1805: {
1806: const char *const *cp;
1807:
1.227 schwarze 1808: if ((r->options & MPARSE_QUICK) == 0)
1.175 schwarze 1809: for (cp = __man_reserved; *cp; cp++)
1810: roff_setstr(r, *cp, NULL, 0);
1811:
1.227 schwarze 1812: if (r->format == 0)
1813: r->format = MPARSE_MAN;
1814:
1.175 schwarze 1815: return(ROFF_CONT);
1.109 kristaps 1816: }
1817:
1818: static enum rofferr
1819: roff_TE(ROFF_ARGS)
1820: {
1821:
1822: if (NULL == r->tbl)
1.221 schwarze 1823: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
1824: ln, ppos, "TE");
1.115 kristaps 1825: else
1.151 kristaps 1826: tbl_end(&r->tbl);
1.109 kristaps 1827:
1.112 kristaps 1828: return(ROFF_IGN);
1829: }
1830:
1831: static enum rofferr
1832: roff_T_(ROFF_ARGS)
1833: {
1834:
1835: if (NULL == r->tbl)
1.221 schwarze 1836: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse,
1837: ln, ppos, "T&");
1.112 kristaps 1838: else
1.116 kristaps 1839: tbl_restart(ppos, ln, r->tbl);
1.112 kristaps 1840:
1.109 kristaps 1841: return(ROFF_IGN);
1842: }
1843:
1.156 kristaps 1844: #if 0
1845: static int
1.151 kristaps 1846: roff_closeeqn(struct roff *r)
1847: {
1848:
1849: return(r->eqn && ROFF_EQN == eqn_end(&r->eqn) ? 1 : 0);
1850: }
1.156 kristaps 1851: #endif
1.151 kristaps 1852:
1.156 kristaps 1853: static void
1.207 schwarze 1854: roff_openeqn(struct roff *r, const char *name, int line,
1.151 kristaps 1855: int offs, const char *buf)
1.125 kristaps 1856: {
1.151 kristaps 1857: struct eqn_node *e;
1858: int poff;
1.125 kristaps 1859:
1860: assert(NULL == r->eqn);
1.151 kristaps 1861: e = eqn_alloc(name, offs, line, r->parse);
1.125 kristaps 1862:
1863: if (r->last_eqn)
1864: r->last_eqn->next = e;
1865: else
1866: r->first_eqn = r->last_eqn = e;
1867:
1868: r->eqn = r->last_eqn = e;
1.151 kristaps 1869:
1870: if (buf) {
1871: poff = 0;
1872: eqn_read(&r->eqn, line, buf, offs, &poff);
1873: }
1874: }
1875:
1876: static enum rofferr
1877: roff_EQ(ROFF_ARGS)
1878: {
1879:
1880: roff_openeqn(r, *bufp + pos, ln, ppos, NULL);
1.125 kristaps 1881: return(ROFF_IGN);
1882: }
1883:
1884: static enum rofferr
1885: roff_EN(ROFF_ARGS)
1886: {
1887:
1.221 schwarze 1888: mandoc_msg(MANDOCERR_BLK_NOTOPEN, r->parse, ln, ppos, "EN");
1.125 kristaps 1889: return(ROFF_IGN);
1890: }
1891:
1892: static enum rofferr
1.109 kristaps 1893: roff_TS(ROFF_ARGS)
1894: {
1.176 schwarze 1895: struct tbl_node *tbl;
1.89 kristaps 1896:
1.115 kristaps 1897: if (r->tbl) {
1.221 schwarze 1898: mandoc_msg(MANDOCERR_BLK_BROKEN, r->parse,
1899: ln, ppos, "TS breaks TS");
1.151 kristaps 1900: tbl_end(&r->tbl);
1.115 kristaps 1901: }
1.83 schwarze 1902:
1.176 schwarze 1903: tbl = tbl_alloc(ppos, ln, r->parse);
1.113 kristaps 1904:
1905: if (r->last_tbl)
1.176 schwarze 1906: r->last_tbl->next = tbl;
1.113 kristaps 1907: else
1.176 schwarze 1908: r->first_tbl = r->last_tbl = tbl;
1.113 kristaps 1909:
1.176 schwarze 1910: r->tbl = r->last_tbl = tbl;
1.83 schwarze 1911: return(ROFF_IGN);
1.92 schwarze 1912: }
1913:
1.105 kristaps 1914: static enum rofferr
1.174 kristaps 1915: roff_cc(ROFF_ARGS)
1916: {
1917: const char *p;
1918:
1919: p = *bufp + pos;
1920:
1921: if ('\0' == *p || '.' == (r->control = *p++))
1922: r->control = 0;
1923:
1924: if ('\0' != *p)
1925: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
1926:
1927: return(ROFF_IGN);
1928: }
1929:
1930: static enum rofferr
1.164 kristaps 1931: roff_tr(ROFF_ARGS)
1932: {
1933: const char *p, *first, *second;
1934: size_t fsz, ssz;
1935: enum mandoc_esc esc;
1936:
1937: p = *bufp + pos;
1938:
1939: if ('\0' == *p) {
1940: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
1941: return(ROFF_IGN);
1942: }
1943:
1944: while ('\0' != *p) {
1945: fsz = ssz = 1;
1946:
1947: first = p++;
1948: if ('\\' == *first) {
1949: esc = mandoc_escape(&p, NULL, NULL);
1950: if (ESCAPE_ERROR == esc) {
1.219 schwarze 1951: mandoc_msg(MANDOCERR_ESC_BAD, r->parse,
1952: ln, (int)(p - *bufp), first);
1.164 kristaps 1953: return(ROFF_IGN);
1954: }
1955: fsz = (size_t)(p - first);
1956: }
1957:
1958: second = p++;
1959: if ('\\' == *second) {
1960: esc = mandoc_escape(&p, NULL, NULL);
1961: if (ESCAPE_ERROR == esc) {
1.219 schwarze 1962: mandoc_msg(MANDOCERR_ESC_BAD, r->parse,
1963: ln, (int)(p - *bufp), second);
1.164 kristaps 1964: return(ROFF_IGN);
1965: }
1966: ssz = (size_t)(p - second);
1.165 kristaps 1967: } else if ('\0' == *second) {
1.207 schwarze 1968: mandoc_msg(MANDOCERR_ARGCOUNT, r->parse,
1969: ln, (int)(p - *bufp), NULL);
1.164 kristaps 1970: second = " ";
1.165 kristaps 1971: p--;
1.164 kristaps 1972: }
1973:
1.167 kristaps 1974: if (fsz > 1) {
1.207 schwarze 1975: roff_setstrn(&r->xmbtab, first, fsz,
1976: second, ssz, 0);
1.167 kristaps 1977: continue;
1978: }
1979:
1980: if (NULL == r->xtab)
1.207 schwarze 1981: r->xtab = mandoc_calloc(128,
1982: sizeof(struct roffstr));
1.167 kristaps 1983:
1984: free(r->xtab[(int)*first].p);
1985: r->xtab[(int)*first].p = mandoc_strndup(second, ssz);
1986: r->xtab[(int)*first].sz = ssz;
1.164 kristaps 1987: }
1988:
1989: return(ROFF_IGN);
1990: }
1991:
1992: static enum rofferr
1.105 kristaps 1993: roff_so(ROFF_ARGS)
1994: {
1995: char *name;
1996:
1.210 schwarze 1997: name = *bufp + pos;
1.224 schwarze 1998: mandoc_vmsg(MANDOCERR_SO, r->parse, ln, ppos, "so %s", name);
1.105 kristaps 1999:
2000: /*
2001: * Handle `so'. Be EXTREMELY careful, as we shouldn't be
2002: * opening anything that's not in our cwd or anything beneath
2003: * it. Thus, explicitly disallow traversing up the file-system
2004: * or using absolute paths.
2005: */
2006:
2007: if ('/' == *name || strstr(name, "../") || strstr(name, "/..")) {
1.210 schwarze 2008: mandoc_vmsg(MANDOCERR_SO_PATH, r->parse, ln, ppos,
2009: ".so %s", name);
1.105 kristaps 2010: return(ROFF_ERR);
2011: }
2012:
2013: *offs = pos;
2014: return(ROFF_SO);
2015: }
1.92 schwarze 2016:
1.106 kristaps 2017: static enum rofferr
2018: roff_userdef(ROFF_ARGS)
1.99 kristaps 2019: {
1.106 kristaps 2020: const char *arg[9];
2021: char *cp, *n1, *n2;
1.119 schwarze 2022: int i;
1.106 kristaps 2023:
2024: /*
2025: * Collect pointers to macro argument strings
1.188 schwarze 2026: * and NUL-terminate them.
1.106 kristaps 2027: */
2028: cp = *bufp + pos;
1.119 schwarze 2029: for (i = 0; i < 9; i++)
1.120 schwarze 2030: arg[i] = '\0' == *cp ? "" :
1.136 kristaps 2031: mandoc_getarg(r->parse, &cp, ln, &pos);
1.99 kristaps 2032:
1.106 kristaps 2033: /*
2034: * Expand macro arguments.
1.99 kristaps 2035: */
1.106 kristaps 2036: *szp = 0;
2037: n1 = cp = mandoc_strdup(r->current_string);
2038: while (NULL != (cp = strstr(cp, "\\$"))) {
2039: i = cp[2] - '1';
2040: if (0 > i || 8 < i) {
2041: /* Not an argument invocation. */
2042: cp += 2;
2043: continue;
2044: }
1.209 schwarze 2045: *cp = '\0';
2046: *szp = mandoc_asprintf(&n2, "%s%s%s",
2047: n1, arg[i], cp + 3) + 1;
1.106 kristaps 2048: cp = n2 + (cp - n1);
2049: free(n1);
2050: n1 = n2;
1.99 kristaps 2051: }
2052:
1.106 kristaps 2053: /*
2054: * Replace the macro invocation
2055: * by the expanded macro.
2056: */
2057: free(*bufp);
2058: *bufp = n1;
2059: if (0 == *szp)
2060: *szp = strlen(*bufp) + 1;
2061:
2062: return(*szp > 1 && '\n' == (*bufp)[(int)*szp - 2] ?
2063: ROFF_REPARSE : ROFF_APPEND);
1.99 kristaps 2064: }
1.121 schwarze 2065:
1.212 schwarze 2066: static size_t
1.121 schwarze 2067: roff_getname(struct roff *r, char **cpp, int ln, int pos)
2068: {
2069: char *name, *cp;
1.212 schwarze 2070: size_t namesz;
1.121 schwarze 2071:
2072: name = *cpp;
2073: if ('\0' == *name)
1.212 schwarze 2074: return(0);
1.121 schwarze 2075:
1.212 schwarze 2076: /* Read until end of name and terminate it with NUL. */
2077: for (cp = name; 1; cp++) {
2078: if ('\0' == *cp || ' ' == *cp) {
2079: namesz = cp - name;
2080: break;
2081: }
1.121 schwarze 2082: if ('\\' != *cp)
2083: continue;
1.215 schwarze 2084: namesz = cp - name;
2085: if ('{' == cp[1] || '}' == cp[1])
2086: break;
1.121 schwarze 2087: cp++;
2088: if ('\\' == *cp)
2089: continue;
1.224 schwarze 2090: mandoc_vmsg(MANDOCERR_NAMESC, r->parse, ln, pos,
2091: "%.*s", (int)(cp - name + 1), name);
1.212 schwarze 2092: mandoc_escape((const char **)&cp, NULL, NULL);
2093: break;
1.121 schwarze 2094: }
2095:
2096: /* Read past spaces. */
2097: while (' ' == *cp)
2098: cp++;
2099:
2100: *cpp = cp;
1.212 schwarze 2101: return(namesz);
1.121 schwarze 2102: }
2103:
1.106 kristaps 2104: /*
2105: * Store *string into the user-defined string called *name.
2106: * To clear an existing entry, call with (*r, *name, NULL, 0).
1.193 schwarze 2107: * append == 0: replace mode
2108: * append == 1: single-line append mode
2109: * append == 2: multiline append mode, append '\n' after each call
1.106 kristaps 2110: */
1.94 kristaps 2111: static void
1.106 kristaps 2112: roff_setstr(struct roff *r, const char *name, const char *string,
1.193 schwarze 2113: int append)
1.92 schwarze 2114: {
1.164 kristaps 2115:
2116: roff_setstrn(&r->strtab, name, strlen(name), string,
1.207 schwarze 2117: string ? strlen(string) : 0, append);
1.164 kristaps 2118: }
2119:
2120: static void
1.166 kristaps 2121: roff_setstrn(struct roffkv **r, const char *name, size_t namesz,
1.193 schwarze 2122: const char *string, size_t stringsz, int append)
1.164 kristaps 2123: {
1.166 kristaps 2124: struct roffkv *n;
1.164 kristaps 2125: char *c;
2126: int i;
2127: size_t oldch, newch;
1.92 schwarze 2128:
1.106 kristaps 2129: /* Search for an existing string with the same name. */
1.164 kristaps 2130: n = *r;
2131:
1.211 schwarze 2132: while (n && (namesz != n->key.sz ||
2133: strncmp(n->key.p, name, namesz)))
1.92 schwarze 2134: n = n->next;
1.94 kristaps 2135:
2136: if (NULL == n) {
1.106 kristaps 2137: /* Create a new string table entry. */
1.166 kristaps 2138: n = mandoc_malloc(sizeof(struct roffkv));
2139: n->key.p = mandoc_strndup(name, namesz);
2140: n->key.sz = namesz;
2141: n->val.p = NULL;
2142: n->val.sz = 0;
1.164 kristaps 2143: n->next = *r;
2144: *r = n;
1.193 schwarze 2145: } else if (0 == append) {
1.166 kristaps 2146: free(n->val.p);
2147: n->val.p = NULL;
2148: n->val.sz = 0;
1.106 kristaps 2149: }
2150:
2151: if (NULL == string)
2152: return;
2153:
2154: /*
2155: * One additional byte for the '\n' in multiline mode,
2156: * and one for the terminating '\0'.
2157: */
1.193 schwarze 2158: newch = stringsz + (1 < append ? 2u : 1u);
1.164 kristaps 2159:
1.166 kristaps 2160: if (NULL == n->val.p) {
2161: n->val.p = mandoc_malloc(newch);
2162: *n->val.p = '\0';
1.106 kristaps 2163: oldch = 0;
2164: } else {
1.166 kristaps 2165: oldch = n->val.sz;
2166: n->val.p = mandoc_realloc(n->val.p, oldch + newch);
1.106 kristaps 2167: }
2168:
2169: /* Skip existing content in the destination buffer. */
1.166 kristaps 2170: c = n->val.p + (int)oldch;
1.106 kristaps 2171:
2172: /* Append new content to the destination buffer. */
1.164 kristaps 2173: i = 0;
2174: while (i < (int)stringsz) {
1.106 kristaps 2175: /*
2176: * Rudimentary roff copy mode:
2177: * Handle escaped backslashes.
2178: */
1.164 kristaps 2179: if ('\\' == string[i] && '\\' == string[i + 1])
2180: i++;
2181: *c++ = string[i++];
1.106 kristaps 2182: }
1.94 kristaps 2183:
1.106 kristaps 2184: /* Append terminating bytes. */
1.193 schwarze 2185: if (1 < append)
1.106 kristaps 2186: *c++ = '\n';
1.163 kristaps 2187:
1.106 kristaps 2188: *c = '\0';
1.166 kristaps 2189: n->val.sz = (int)(c - n->val.p);
1.92 schwarze 2190: }
2191:
1.94 kristaps 2192: static const char *
2193: roff_getstrn(const struct roff *r, const char *name, size_t len)
1.92 schwarze 2194: {
1.166 kristaps 2195: const struct roffkv *n;
1.191 schwarze 2196: int i;
1.92 schwarze 2197:
1.164 kristaps 2198: for (n = r->strtab; n; n = n->next)
1.207 schwarze 2199: if (0 == strncmp(name, n->key.p, len) &&
2200: '\0' == n->key.p[(int)len])
1.166 kristaps 2201: return(n->val.p);
1.191 schwarze 2202:
2203: for (i = 0; i < PREDEFS_MAX; i++)
2204: if (0 == strncmp(name, predefs[i].name, len) &&
2205: '\0' == predefs[i].name[(int)len])
2206: return(predefs[i].str);
1.94 kristaps 2207:
1.157 kristaps 2208: return(NULL);
1.92 schwarze 2209: }
2210:
1.94 kristaps 2211: static void
1.167 kristaps 2212: roff_freestr(struct roffkv *r)
1.92 schwarze 2213: {
1.166 kristaps 2214: struct roffkv *n, *nn;
1.92 schwarze 2215:
1.167 kristaps 2216: for (n = r; n; n = nn) {
1.166 kristaps 2217: free(n->key.p);
2218: free(n->val.p);
1.92 schwarze 2219: nn = n->next;
2220: free(n);
2221: }
1.114 kristaps 2222: }
2223:
2224: const struct tbl_span *
2225: roff_span(const struct roff *r)
2226: {
1.207 schwarze 2227:
1.114 kristaps 2228: return(r->tbl ? tbl_span(r->tbl) : NULL);
1.125 kristaps 2229: }
2230:
2231: const struct eqn *
2232: roff_eqn(const struct roff *r)
2233: {
1.207 schwarze 2234:
1.125 kristaps 2235: return(r->last_eqn ? &r->last_eqn->eqn : NULL);
1.164 kristaps 2236: }
2237:
2238: /*
2239: * Duplicate an input string, making the appropriate character
2240: * conversations (as stipulated by `tr') along the way.
2241: * Returns a heap-allocated string with all the replacements made.
2242: */
2243: char *
2244: roff_strdup(const struct roff *r, const char *p)
2245: {
1.166 kristaps 2246: const struct roffkv *cp;
1.164 kristaps 2247: char *res;
2248: const char *pp;
2249: size_t ssz, sz;
2250: enum mandoc_esc esc;
2251:
1.167 kristaps 2252: if (NULL == r->xmbtab && NULL == r->xtab)
1.164 kristaps 2253: return(mandoc_strdup(p));
2254: else if ('\0' == *p)
2255: return(mandoc_strdup(""));
2256:
2257: /*
2258: * Step through each character looking for term matches
2259: * (remember that a `tr' can be invoked with an escape, which is
2260: * a glyph but the escape is multi-character).
2261: * We only do this if the character hash has been initialised
2262: * and the string is >0 length.
2263: */
2264:
2265: res = NULL;
2266: ssz = 0;
2267:
2268: while ('\0' != *p) {
1.167 kristaps 2269: if ('\\' != *p && r->xtab && r->xtab[(int)*p].p) {
2270: sz = r->xtab[(int)*p].sz;
2271: res = mandoc_realloc(res, ssz + sz + 1);
2272: memcpy(res + ssz, r->xtab[(int)*p].p, sz);
2273: ssz += sz;
2274: p++;
2275: continue;
2276: } else if ('\\' != *p) {
2277: res = mandoc_realloc(res, ssz + 2);
2278: res[ssz++] = *p++;
2279: continue;
2280: }
2281:
1.164 kristaps 2282: /* Search for term matches. */
1.167 kristaps 2283: for (cp = r->xmbtab; cp; cp = cp->next)
1.166 kristaps 2284: if (0 == strncmp(p, cp->key.p, cp->key.sz))
1.164 kristaps 2285: break;
2286:
2287: if (NULL != cp) {
2288: /*
2289: * A match has been found.
2290: * Append the match to the array and move
2291: * forward by its keysize.
2292: */
1.207 schwarze 2293: res = mandoc_realloc(res,
2294: ssz + cp->val.sz + 1);
1.166 kristaps 2295: memcpy(res + ssz, cp->val.p, cp->val.sz);
2296: ssz += cp->val.sz;
2297: p += (int)cp->key.sz;
1.164 kristaps 2298: continue;
2299: }
2300:
1.167 kristaps 2301: /*
2302: * Handle escapes carefully: we need to copy
2303: * over just the escape itself, or else we might
2304: * do replacements within the escape itself.
2305: * Make sure to pass along the bogus string.
2306: */
2307: pp = p++;
2308: esc = mandoc_escape(&p, NULL, NULL);
2309: if (ESCAPE_ERROR == esc) {
2310: sz = strlen(pp);
1.164 kristaps 2311: res = mandoc_realloc(res, ssz + sz + 1);
2312: memcpy(res + ssz, pp, sz);
1.167 kristaps 2313: break;
1.164 kristaps 2314: }
1.207 schwarze 2315: /*
2316: * We bail out on bad escapes.
1.167 kristaps 2317: * No need to warn: we already did so when
2318: * roff_res() was called.
2319: */
2320: sz = (int)(p - pp);
2321: res = mandoc_realloc(res, ssz + sz + 1);
2322: memcpy(res + ssz, pp, sz);
2323: ssz += sz;
1.164 kristaps 2324: }
2325:
2326: res[(int)ssz] = '\0';
2327: return(res);
1.227 schwarze 2328: }
2329:
2330: int
2331: roff_getformat(const struct roff *r)
2332: {
2333:
2334: return(r->format);
1.174 kristaps 2335: }
2336:
2337: /*
1.207 schwarze 2338: * Find out whether a line is a macro line or not.
1.174 kristaps 2339: * If it is, adjust the current position and return one; if it isn't,
2340: * return zero and don't change the current position.
2341: * If the control character has been set with `.cc', then let that grain
2342: * precedence.
2343: * This is slighly contrary to groff, where using the non-breaking
2344: * control character when `cc' has been invoked will cause the
2345: * non-breaking macro contents to be printed verbatim.
2346: */
2347: int
2348: roff_getcontrol(const struct roff *r, const char *cp, int *ppos)
2349: {
2350: int pos;
2351:
2352: pos = *ppos;
2353:
2354: if (0 != r->control && cp[pos] == r->control)
2355: pos++;
2356: else if (0 != r->control)
2357: return(0);
2358: else if ('\\' == cp[pos] && '.' == cp[pos + 1])
2359: pos += 2;
2360: else if ('.' == cp[pos] || '\'' == cp[pos])
2361: pos++;
2362: else
2363: return(0);
2364:
2365: while (' ' == cp[pos] || '\t' == cp[pos])
2366: pos++;
2367:
2368: *ppos = pos;
2369: return(1);
1.74 kristaps 2370: }
CVSweb