[BACK]Return to roff.c CVS log [TXT][DIR] Up to [cvsweb.bsd.lv] / mandoc

Annotation of mandoc/roff.c, Revision 1.115

1.115   ! kristaps    1: /*     $Id: roff.c,v 1.114 2010/12/31 18:19:43 kristaps Exp $ */
1.1       kristaps    2: /*
1.67      kristaps    3:  * Copyright (c) 2010 Kristaps Dzonsons <kristaps@bsd.lv>
1.93      schwarze    4:  * Copyright (c) 2010 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>
1.89      kristaps   23: #include <errno.h>
1.85      kristaps   24: #include <ctype.h>
1.89      kristaps   25: #include <limits.h>
1.1       kristaps   26: #include <stdlib.h>
1.67      kristaps   27: #include <string.h>
1.75      kristaps   28: #include <stdio.h>
1.1       kristaps   29:
1.67      kristaps   30: #include "mandoc.h"
1.43      kristaps   31: #include "roff.h"
1.109     kristaps   32: #include "libroff.h"
1.94      kristaps   33: #include "libmandoc.h"
1.33      kristaps   34:
1.82      kristaps   35: #define        RSTACK_MAX      128
                     36:
1.75      kristaps   37: #define        ROFF_CTL(c) \
                     38:        ('.' == (c) || '\'' == (c))
                     39:
1.67      kristaps   40: enum   rofft {
1.103     kristaps   41:        ROFF_ad,
1.80      kristaps   42:        ROFF_am,
                     43:        ROFF_ami,
                     44:        ROFF_am1,
                     45:        ROFF_de,
                     46:        ROFF_dei,
                     47:        ROFF_de1,
1.83      schwarze   48:        ROFF_ds,
1.82      kristaps   49:        ROFF_el,
1.103     kristaps   50:        ROFF_hy,
1.82      kristaps   51:        ROFF_ie,
1.75      kristaps   52:        ROFF_if,
1.76      kristaps   53:        ROFF_ig,
1.103     kristaps   54:        ROFF_ne,
                     55:        ROFF_nh,
1.104     kristaps   56:        ROFF_nr,
1.83      schwarze   57:        ROFF_rm,
1.105     kristaps   58:        ROFF_so,
1.83      schwarze   59:        ROFF_tr,
1.109     kristaps   60:        ROFF_TS,
                     61:        ROFF_TE,
1.112     kristaps   62:        ROFF_T_,
1.76      kristaps   63:        ROFF_cblock,
1.100     kristaps   64:        ROFF_ccond, /* FIXME: remove this. */
1.106     kristaps   65:        ROFF_USERDEF,
1.67      kristaps   66:        ROFF_MAX
                     67: };
                     68:
1.82      kristaps   69: enum   roffrule {
                     70:        ROFFRULE_ALLOW,
                     71:        ROFFRULE_DENY
                     72: };
                     73:
1.94      kristaps   74: struct roffstr {
                     75:        char            *name; /* key of symbol */
                     76:        char            *string; /* current value */
                     77:        struct roffstr  *next; /* next in list */
                     78: };
                     79:
1.67      kristaps   80: struct roff {
                     81:        struct roffnode *last; /* leaf of stack */
                     82:        mandocmsg        msg; /* err/warn/fatal messages */
                     83:        void            *data; /* privdata for messages */
1.82      kristaps   84:        enum roffrule    rstack[RSTACK_MAX]; /* stack of !`ie' rules */
                     85:        int              rstackpos; /* position in rstack */
1.90      kristaps   86:        struct regset   *regs; /* read/writable registers */
1.106     kristaps   87:        struct roffstr  *first_string; /* user-defined strings & macros */
                     88:        const char      *current_string; /* value of last called user macro */
1.113     kristaps   89:        struct tbl      *first_tbl; /* first table parsed */
                     90:        struct tbl      *last_tbl; /* last table parsed */
                     91:        struct tbl      *tbl; /* current table being parsed */
1.79      kristaps   92: };
                     93:
1.67      kristaps   94: struct roffnode {
                     95:        enum rofft       tok; /* type of node */
                     96:        struct roffnode *parent; /* up one in stack */
                     97:        int              line; /* parse line */
                     98:        int              col; /* parse col */
1.106     kristaps   99:        char            *name; /* node name, e.g. macro name */
1.79      kristaps  100:        char            *end; /* end-rules: custom token */
                    101:        int              endspan; /* end-rules: next-line or infty */
1.82      kristaps  102:        enum roffrule    rule; /* current evaluation rule */
1.67      kristaps  103: };
                    104:
                    105: #define        ROFF_ARGS        struct roff *r, /* parse ctx */ \
1.72      kristaps  106:                         enum rofft tok, /* tok of macro */ \
1.67      kristaps  107:                         char **bufp, /* input buffer */ \
                    108:                         size_t *szp, /* size of input buffer */ \
                    109:                         int ln, /* parse line */ \
1.75      kristaps  110:                         int ppos, /* original pos in buffer */ \
                    111:                         int pos, /* current pos in buffer */ \
1.74      kristaps  112:                         int *offs /* reset offset of buffer data */
1.67      kristaps  113:
                    114: typedef        enum rofferr (*roffproc)(ROFF_ARGS);
                    115:
                    116: struct roffmac {
                    117:        const char      *name; /* macro name */
1.79      kristaps  118:        roffproc         proc; /* process new macro */
                    119:        roffproc         text; /* process as child text of macro */
                    120:        roffproc         sub; /* process as child of macro */
                    121:        int              flags;
                    122: #define        ROFFMAC_STRUCT  (1 << 0) /* always interpret */
1.85      kristaps  123:        struct roffmac  *next;
1.67      kristaps  124: };
                    125:
1.80      kristaps  126: static enum rofferr     roff_block(ROFF_ARGS);
                    127: static enum rofferr     roff_block_text(ROFF_ARGS);
                    128: static enum rofferr     roff_block_sub(ROFF_ARGS);
                    129: static enum rofferr     roff_cblock(ROFF_ARGS);
                    130: static enum rofferr     roff_ccond(ROFF_ARGS);
1.82      kristaps  131: static enum rofferr     roff_cond(ROFF_ARGS);
                    132: static enum rofferr     roff_cond_text(ROFF_ARGS);
                    133: static enum rofferr     roff_cond_sub(ROFF_ARGS);
1.92      schwarze  134: static enum rofferr     roff_ds(ROFF_ARGS);
1.94      kristaps  135: static enum roffrule    roff_evalcond(const char *, int *);
                    136: static void             roff_freestr(struct roff *);
                    137: static const char      *roff_getstrn(const struct roff *,
                    138:                                const char *, size_t);
1.103     kristaps  139: static enum rofferr     roff_line_ignore(ROFF_ARGS);
1.104     kristaps  140: static enum rofferr     roff_line_error(ROFF_ARGS);
1.89      kristaps  141: static enum rofferr     roff_nr(ROFF_ARGS);
1.95      kristaps  142: static int              roff_res(struct roff *,
                    143:                                char **, size_t *, int);
1.94      kristaps  144: static void             roff_setstr(struct roff *,
1.106     kristaps  145:                                const char *, const char *, int);
1.105     kristaps  146: static enum rofferr     roff_so(ROFF_ARGS);
1.109     kristaps  147: static enum rofferr     roff_TE(ROFF_ARGS);
                    148: static enum rofferr     roff_TS(ROFF_ARGS);
1.112     kristaps  149: static enum rofferr     roff_T_(ROFF_ARGS);
1.106     kristaps  150: static enum rofferr     roff_userdef(ROFF_ARGS);
1.67      kristaps  151:
1.85      kristaps  152: /* See roff_hash_find() */
                    153:
                    154: #define        ASCII_HI         126
                    155: #define        ASCII_LO         33
                    156: #define        HASHWIDTH       (ASCII_HI - ASCII_LO + 1)
                    157:
                    158: static struct roffmac  *hash[HASHWIDTH];
                    159:
                    160: static struct roffmac   roffs[ROFF_MAX] = {
1.103     kristaps  161:        { "ad", roff_line_ignore, NULL, NULL, 0, NULL },
1.85      kristaps  162:        { "am", roff_block, roff_block_text, roff_block_sub, 0, NULL },
                    163:        { "ami", roff_block, roff_block_text, roff_block_sub, 0, NULL },
                    164:        { "am1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
                    165:        { "de", roff_block, roff_block_text, roff_block_sub, 0, NULL },
                    166:        { "dei", roff_block, roff_block_text, roff_block_sub, 0, NULL },
                    167:        { "de1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.92      schwarze  168:        { "ds", roff_ds, NULL, NULL, 0, NULL },
1.85      kristaps  169:        { "el", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
1.103     kristaps  170:        { "hy", roff_line_ignore, NULL, NULL, 0, NULL },
1.85      kristaps  171:        { "ie", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
                    172:        { "if", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
                    173:        { "ig", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.103     kristaps  174:        { "ne", roff_line_ignore, NULL, NULL, 0, NULL },
                    175:        { "nh", roff_line_ignore, NULL, NULL, 0, NULL },
1.104     kristaps  176:        { "nr", roff_nr, NULL, NULL, 0, NULL },
                    177:        { "rm", roff_line_error, NULL, NULL, 0, NULL },
1.105     kristaps  178:        { "so", roff_so, NULL, NULL, 0, NULL },
1.103     kristaps  179:        { "tr", roff_line_ignore, NULL, NULL, 0, NULL },
1.109     kristaps  180:        { "TS", roff_TS, NULL, NULL, 0, NULL },
                    181:        { "TE", roff_TE, NULL, NULL, 0, NULL },
1.112     kristaps  182:        { "T&", roff_T_, NULL, NULL, 0, NULL },
1.85      kristaps  183:        { ".", roff_cblock, NULL, NULL, 0, NULL },
                    184:        { "\\}", roff_ccond, NULL, NULL, 0, NULL },
1.106     kristaps  185:        { NULL, roff_userdef, NULL, NULL, 0, NULL },
1.67      kristaps  186: };
                    187:
                    188: static void             roff_free1(struct roff *);
1.106     kristaps  189: static enum rofft       roff_hash_find(const char *, size_t);
1.85      kristaps  190: static void             roff_hash_init(void);
1.76      kristaps  191: static void             roffnode_cleanscope(struct roff *);
1.106     kristaps  192: static void             roffnode_push(struct roff *, enum rofft,
                    193:                                const char *, int, int);
1.67      kristaps  194: static void             roffnode_pop(struct roff *);
1.106     kristaps  195: static enum rofft       roff_parse(struct roff *, const char *, int *);
1.91      kristaps  196: static int              roff_parse_nat(const char *, unsigned int *);
1.67      kristaps  197:
1.85      kristaps  198: /* See roff_hash_find() */
                    199: #define        ROFF_HASH(p)    (p[0] - ASCII_LO)
                    200:
                    201: static void
                    202: roff_hash_init(void)
                    203: {
                    204:        struct roffmac   *n;
                    205:        int               buc, i;
                    206:
1.106     kristaps  207:        for (i = 0; i < (int)ROFF_USERDEF; i++) {
1.85      kristaps  208:                assert(roffs[i].name[0] >= ASCII_LO);
                    209:                assert(roffs[i].name[0] <= ASCII_HI);
                    210:
                    211:                buc = ROFF_HASH(roffs[i].name);
                    212:
                    213:                if (NULL != (n = hash[buc])) {
                    214:                        for ( ; n->next; n = n->next)
                    215:                                /* Do nothing. */ ;
                    216:                        n->next = &roffs[i];
                    217:                } else
                    218:                        hash[buc] = &roffs[i];
                    219:        }
                    220: }
                    221:
1.67      kristaps  222:
                    223: /*
                    224:  * Look up a roff token by its name.  Returns ROFF_MAX if no macro by
                    225:  * the nil-terminated string name could be found.
                    226:  */
                    227: static enum rofft
1.106     kristaps  228: roff_hash_find(const char *p, size_t s)
1.67      kristaps  229: {
1.85      kristaps  230:        int              buc;
                    231:        struct roffmac  *n;
1.67      kristaps  232:
1.85      kristaps  233:        /*
                    234:         * libroff has an extremely simple hashtable, for the time
                    235:         * being, which simply keys on the first character, which must
                    236:         * be printable, then walks a chain.  It works well enough until
                    237:         * optimised.
                    238:         */
                    239:
                    240:        if (p[0] < ASCII_LO || p[0] > ASCII_HI)
                    241:                return(ROFF_MAX);
                    242:
                    243:        buc = ROFF_HASH(p);
                    244:
                    245:        if (NULL == (n = hash[buc]))
                    246:                return(ROFF_MAX);
                    247:        for ( ; n; n = n->next)
1.106     kristaps  248:                if (0 == strncmp(n->name, p, s) && '\0' == n->name[(int)s])
1.85      kristaps  249:                        return((enum rofft)(n - roffs));
1.67      kristaps  250:
                    251:        return(ROFF_MAX);
                    252: }
                    253:
                    254:
                    255: /*
                    256:  * Pop the current node off of the stack of roff instructions currently
                    257:  * pending.
                    258:  */
                    259: static void
                    260: roffnode_pop(struct roff *r)
                    261: {
                    262:        struct roffnode *p;
                    263:
1.75      kristaps  264:        assert(r->last);
                    265:        p = r->last;
1.82      kristaps  266:
                    267:        if (ROFF_el == p->tok)
                    268:                if (r->rstackpos > -1)
                    269:                        r->rstackpos--;
                    270:
1.75      kristaps  271:        r->last = r->last->parent;
1.106     kristaps  272:        free(p->name);
                    273:        free(p->end);
1.67      kristaps  274:        free(p);
                    275: }
                    276:
                    277:
                    278: /*
                    279:  * Push a roff node onto the instruction stack.  This must later be
                    280:  * removed with roffnode_pop().
                    281:  */
1.98      schwarze  282: static void
1.106     kristaps  283: roffnode_push(struct roff *r, enum rofft tok, const char *name,
                    284:                int line, int col)
1.67      kristaps  285: {
                    286:        struct roffnode *p;
                    287:
1.98      schwarze  288:        p = mandoc_calloc(1, sizeof(struct roffnode));
1.67      kristaps  289:        p->tok = tok;
1.106     kristaps  290:        if (name)
                    291:                p->name = mandoc_strdup(name);
1.67      kristaps  292:        p->parent = r->last;
                    293:        p->line = line;
                    294:        p->col = col;
1.79      kristaps  295:        p->rule = p->parent ? p->parent->rule : ROFFRULE_DENY;
1.67      kristaps  296:
                    297:        r->last = p;
                    298: }
                    299:
                    300:
                    301: static void
                    302: roff_free1(struct roff *r)
                    303: {
1.113     kristaps  304:        struct tbl      *t;
1.67      kristaps  305:
1.113     kristaps  306:        while (r->first_tbl) {
                    307:                t = r->first_tbl;
                    308:                r->first_tbl = t->next;
                    309:                tbl_free(t);
1.109     kristaps  310:        }
                    311:
1.113     kristaps  312:        r->first_tbl = r->last_tbl = r->tbl = NULL;
                    313:
1.67      kristaps  314:        while (r->last)
                    315:                roffnode_pop(r);
1.109     kristaps  316:
1.94      kristaps  317:        roff_freestr(r);
1.67      kristaps  318: }
                    319:
                    320:
                    321: void
                    322: roff_reset(struct roff *r)
                    323: {
                    324:
                    325:        roff_free1(r);
                    326: }
                    327:
                    328:
                    329: void
                    330: roff_free(struct roff *r)
                    331: {
                    332:
                    333:        roff_free1(r);
                    334:        free(r);
                    335: }
                    336:
                    337:
                    338: struct roff *
1.98      schwarze  339: roff_alloc(struct regset *regs, void *data, const mandocmsg msg)
1.67      kristaps  340: {
                    341:        struct roff     *r;
                    342:
1.98      schwarze  343:        r = mandoc_calloc(1, sizeof(struct roff));
1.90      kristaps  344:        r->regs = regs;
1.67      kristaps  345:        r->msg = msg;
                    346:        r->data = data;
1.82      kristaps  347:        r->rstackpos = -1;
1.85      kristaps  348:
                    349:        roff_hash_init();
1.67      kristaps  350:        return(r);
                    351: }
                    352:
                    353:
1.94      kristaps  354: /*
                    355:  * Pre-filter each and every line for reserved words (one beginning with
                    356:  * `\*', e.g., `\*(ab').  These must be handled before the actual line
                    357:  * is processed.
                    358:  */
                    359: static int
1.95      kristaps  360: roff_res(struct roff *r, char **bufp, size_t *szp, int pos)
1.94      kristaps  361: {
1.108     schwarze  362:        const char      *stesc; /* start of an escape sequence ('\\') */
                    363:        const char      *stnam; /* start of the name, after "[(*" */
                    364:        const char      *cp;    /* end of the name, e.g. before ']' */
                    365:        const char      *res;   /* the string to be substituted */
1.94      kristaps  366:        int              i, maxl;
                    367:        size_t           nsz;
                    368:        char            *n;
                    369:
1.108     schwarze  370:        /* Search for a leading backslash and save a pointer to it. */
                    371:
                    372:        cp = *bufp + pos;
                    373:        while (NULL != (cp = strchr(cp, '\\'))) {
                    374:                stesc = cp++;
                    375:
                    376:                /*
                    377:                 * The second character must be an asterisk.
                    378:                 * If it isn't, skip it anyway:  It is escaped,
                    379:                 * so it can't start another escape sequence.
                    380:                 */
                    381:
                    382:                if ('\0' == *cp)
                    383:                        return(1);
                    384:                if ('*' != *cp++)
                    385:                        continue;
                    386:
                    387:                /*
                    388:                 * The third character decides the length
                    389:                 * of the name of the string.
                    390:                 * Save a pointer to the name.
                    391:                 */
                    392:
1.94      kristaps  393:                switch (*cp) {
1.108     schwarze  394:                case ('\0'):
                    395:                        return(1);
1.94      kristaps  396:                case ('('):
                    397:                        cp++;
                    398:                        maxl = 2;
                    399:                        break;
                    400:                case ('['):
                    401:                        cp++;
                    402:                        maxl = 0;
                    403:                        break;
                    404:                default:
                    405:                        maxl = 1;
                    406:                        break;
                    407:                }
1.108     schwarze  408:                stnam = cp;
1.94      kristaps  409:
1.108     schwarze  410:                /* Advance to the end of the name. */
1.94      kristaps  411:
                    412:                for (i = 0; 0 == maxl || i < maxl; i++, cp++) {
                    413:                        if ('\0' == *cp)
                    414:                                return(1); /* Error. */
                    415:                        if (0 == maxl && ']' == *cp)
                    416:                                break;
                    417:                }
                    418:
1.108     schwarze  419:                /*
                    420:                 * Retrieve the replacement string; if it is
                    421:                 * undefined, resume searching for escapes.
                    422:                 */
                    423:
                    424:                res = roff_getstrn(r, stnam, (size_t)i);
1.94      kristaps  425:
                    426:                if (NULL == res) {
                    427:                        cp -= maxl ? 1 : 0;
                    428:                        continue;
                    429:                }
                    430:
1.108     schwarze  431:                /* Replace the escape sequence by the string. */
                    432:
1.94      kristaps  433:                nsz = *szp + strlen(res) + 1;
                    434:                n = mandoc_malloc(nsz);
                    435:
1.108     schwarze  436:                strlcpy(n, *bufp, (size_t)(stesc - *bufp + 1));
1.94      kristaps  437:                strlcat(n, res, nsz);
                    438:                strlcat(n, cp + (maxl ? 0 : 1), nsz);
                    439:
                    440:                free(*bufp);
                    441:
                    442:                *bufp = n;
                    443:                *szp = nsz;
                    444:                return(0);
                    445:        }
                    446:
                    447:        return(1);
                    448: }
                    449:
                    450:
1.67      kristaps  451: enum rofferr
1.90      kristaps  452: roff_parseln(struct roff *r, int ln, char **bufp,
                    453:                size_t *szp, int pos, int *offs)
1.67      kristaps  454: {
                    455:        enum rofft       t;
1.109     kristaps  456:        enum rofferr     e;
1.79      kristaps  457:        int              ppos;
                    458:
                    459:        /*
1.94      kristaps  460:         * Run the reserved-word filter only if we have some reserved
                    461:         * words to fill in.
                    462:         */
                    463:
1.95      kristaps  464:        if (r->first_string && ! roff_res(r, bufp, szp, pos))
1.106     kristaps  465:                return(ROFF_REPARSE);
1.94      kristaps  466:
                    467:        /*
1.79      kristaps  468:         * First, if a scope is open and we're not a macro, pass the
                    469:         * text through the macro's filter.  If a scope isn't open and
                    470:         * we're not a macro, just let it through.
                    471:         */
1.74      kristaps  472:
1.75      kristaps  473:        if (r->last && ! ROFF_CTL((*bufp)[pos])) {
1.78      kristaps  474:                t = r->last->tok;
                    475:                assert(roffs[t].text);
1.109     kristaps  476:                e = (*roffs[t].text)
                    477:                        (r, t, bufp, szp, ln, pos, pos, offs);
                    478:                assert(ROFF_IGN == e || ROFF_CONT == e);
                    479:                if (ROFF_CONT == e && r->tbl)
                    480:                        return(tbl_read(r->tbl, ln, *bufp, *offs));
                    481:                return(e);
                    482:        } else if ( ! ROFF_CTL((*bufp)[pos])) {
                    483:                if (r->tbl)
                    484:                        return(tbl_read(r->tbl, ln, *bufp, *offs));
1.67      kristaps  485:                return(ROFF_CONT);
1.109     kristaps  486:        }
1.67      kristaps  487:
1.79      kristaps  488:        /*
                    489:         * If a scope is open, go to the child handler for that macro,
                    490:         * as it may want to preprocess before doing anything with it.
                    491:         */
1.78      kristaps  492:
1.79      kristaps  493:        if (r->last) {
                    494:                t = r->last->tok;
                    495:                assert(roffs[t].sub);
                    496:                return((*roffs[t].sub)
1.90      kristaps  497:                                (r, t, bufp, szp,
                    498:                                 ln, pos, pos, offs));
1.79      kristaps  499:        }
1.78      kristaps  500:
1.79      kristaps  501:        /*
                    502:         * Lastly, as we've no scope open, try to look up and execute
                    503:         * the new macro.  If no macro is found, simply return and let
                    504:         * the compilers handle it.
                    505:         */
1.67      kristaps  506:
1.75      kristaps  507:        ppos = pos;
1.106     kristaps  508:        if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos)))
1.79      kristaps  509:                return(ROFF_CONT);
1.67      kristaps  510:
1.75      kristaps  511:        assert(roffs[t].proc);
1.78      kristaps  512:        return((*roffs[t].proc)
1.90      kristaps  513:                        (r, t, bufp, szp,
                    514:                         ln, ppos, pos, offs));
1.74      kristaps  515: }
                    516:
                    517:
                    518: int
                    519: roff_endparse(struct roff *r)
                    520: {
                    521:
1.110     kristaps  522:        /* FIXME: if r->tbl */
                    523:        if (r->last)
1.109     kristaps  524:                (*r->msg)(MANDOCERR_SCOPEEXIT, r->data,
                    525:                                r->last->line, r->last->col, NULL);
                    526:        return(1);
1.67      kristaps  527: }
                    528:
                    529:
                    530: /*
                    531:  * Parse a roff node's type from the input buffer.  This must be in the
                    532:  * form of ".foo xxx" in the usual way.
                    533:  */
                    534: static enum rofft
1.106     kristaps  535: roff_parse(struct roff *r, const char *buf, int *pos)
1.67      kristaps  536: {
1.106     kristaps  537:        const char      *mac;
                    538:        size_t           maclen;
1.67      kristaps  539:        enum rofft       t;
                    540:
1.75      kristaps  541:        assert(ROFF_CTL(buf[*pos]));
                    542:        (*pos)++;
1.67      kristaps  543:
1.106     kristaps  544:        while (' ' == buf[*pos] || '\t' == buf[*pos])
1.67      kristaps  545:                (*pos)++;
                    546:
                    547:        if ('\0' == buf[*pos])
                    548:                return(ROFF_MAX);
                    549:
1.106     kristaps  550:        mac = buf + *pos;
                    551:        maclen = strcspn(mac, " \\\t\0");
1.67      kristaps  552:
1.106     kristaps  553:        t = (r->current_string = roff_getstrn(r, mac, maclen))
                    554:            ? ROFF_USERDEF : roff_hash_find(mac, maclen);
1.67      kristaps  555:
1.106     kristaps  556:        *pos += maclen;
1.67      kristaps  557:        while (buf[*pos] && ' ' == buf[*pos])
                    558:                (*pos)++;
                    559:
                    560:        return(t);
                    561: }
                    562:
                    563:
1.89      kristaps  564: static int
1.91      kristaps  565: roff_parse_nat(const char *buf, unsigned int *res)
1.89      kristaps  566: {
                    567:        char            *ep;
                    568:        long             lval;
                    569:
                    570:        errno = 0;
                    571:        lval = strtol(buf, &ep, 10);
                    572:        if (buf[0] == '\0' || *ep != '\0')
                    573:                return(0);
                    574:        if ((errno == ERANGE &&
                    575:                        (lval == LONG_MAX || lval == LONG_MIN)) ||
1.91      kristaps  576:                        (lval > INT_MAX || lval < 0))
1.89      kristaps  577:                return(0);
                    578:
1.91      kristaps  579:        *res = (unsigned int)lval;
1.89      kristaps  580:        return(1);
                    581: }
                    582:
                    583:
1.67      kristaps  584: /* ARGSUSED */
                    585: static enum rofferr
1.76      kristaps  586: roff_cblock(ROFF_ARGS)
1.67      kristaps  587: {
                    588:
1.79      kristaps  589:        /*
                    590:         * A block-close `..' should only be invoked as a child of an
                    591:         * ignore macro, otherwise raise a warning and just ignore it.
                    592:         */
                    593:
1.76      kristaps  594:        if (NULL == r->last) {
1.109     kristaps  595:                (*r->msg)(MANDOCERR_NOSCOPE, r->data, ln, ppos, NULL);
1.76      kristaps  596:                return(ROFF_IGN);
                    597:        }
1.67      kristaps  598:
1.81      kristaps  599:        switch (r->last->tok) {
                    600:        case (ROFF_am):
                    601:                /* FALLTHROUGH */
                    602:        case (ROFF_ami):
                    603:                /* FALLTHROUGH */
                    604:        case (ROFF_am1):
                    605:                /* FALLTHROUGH */
                    606:        case (ROFF_de):
1.108     schwarze  607:                /* ROFF_de1 is remapped to ROFF_de in roff_block(). */
1.81      kristaps  608:                /* FALLTHROUGH */
                    609:        case (ROFF_dei):
                    610:                /* FALLTHROUGH */
                    611:        case (ROFF_ig):
                    612:                break;
                    613:        default:
1.109     kristaps  614:                (*r->msg)(MANDOCERR_NOSCOPE, r->data, ln, ppos, NULL);
1.67      kristaps  615:                return(ROFF_IGN);
1.76      kristaps  616:        }
1.67      kristaps  617:
1.76      kristaps  618:        if ((*bufp)[pos])
1.109     kristaps  619:                (*r->msg)(MANDOCERR_ARGSLOST, r->data, ln, pos, NULL);
1.71      kristaps  620:
                    621:        roffnode_pop(r);
1.76      kristaps  622:        roffnode_cleanscope(r);
                    623:        return(ROFF_IGN);
1.71      kristaps  624:
1.67      kristaps  625: }
                    626:
                    627:
1.76      kristaps  628: static void
                    629: roffnode_cleanscope(struct roff *r)
1.67      kristaps  630: {
                    631:
1.76      kristaps  632:        while (r->last) {
                    633:                if (--r->last->endspan < 0)
                    634:                        break;
                    635:                roffnode_pop(r);
                    636:        }
1.67      kristaps  637: }
                    638:
                    639:
1.75      kristaps  640: /* ARGSUSED */
1.74      kristaps  641: static enum rofferr
1.75      kristaps  642: roff_ccond(ROFF_ARGS)
1.74      kristaps  643: {
                    644:
1.76      kristaps  645:        if (NULL == r->last) {
1.109     kristaps  646:                (*r->msg)(MANDOCERR_NOSCOPE, r->data, ln, ppos, NULL);
1.76      kristaps  647:                return(ROFF_IGN);
                    648:        }
                    649:
1.82      kristaps  650:        switch (r->last->tok) {
                    651:        case (ROFF_el):
                    652:                /* FALLTHROUGH */
                    653:        case (ROFF_ie):
                    654:                /* FALLTHROUGH */
                    655:        case (ROFF_if):
                    656:                break;
                    657:        default:
1.109     kristaps  658:                (*r->msg)(MANDOCERR_NOSCOPE, r->data, ln, ppos, NULL);
1.75      kristaps  659:                return(ROFF_IGN);
                    660:        }
                    661:
1.76      kristaps  662:        if (r->last->endspan > -1) {
1.109     kristaps  663:                (*r->msg)(MANDOCERR_NOSCOPE, r->data, ln, ppos, NULL);
1.76      kristaps  664:                return(ROFF_IGN);
                    665:        }
                    666:
                    667:        if ((*bufp)[pos])
1.109     kristaps  668:                (*r->msg)(MANDOCERR_ARGSLOST, r->data, ln, pos, NULL);
1.76      kristaps  669:
1.75      kristaps  670:        roffnode_pop(r);
1.76      kristaps  671:        roffnode_cleanscope(r);
                    672:        return(ROFF_IGN);
                    673: }
                    674:
1.75      kristaps  675:
1.76      kristaps  676: /* ARGSUSED */
                    677: static enum rofferr
1.80      kristaps  678: roff_block(ROFF_ARGS)
1.76      kristaps  679: {
1.78      kristaps  680:        int             sv;
                    681:        size_t          sz;
1.106     kristaps  682:        char            *name;
                    683:
                    684:        name = NULL;
1.76      kristaps  685:
1.106     kristaps  686:        if (ROFF_ig != tok) {
                    687:                if ('\0' == (*bufp)[pos]) {
                    688:                        (*r->msg)(MANDOCERR_NOARGS, r->data, ln, ppos, NULL);
                    689:                        return(ROFF_IGN);
                    690:                }
1.107     kristaps  691:
                    692:                /*
                    693:                 * Re-write `de1', since we don't really care about
                    694:                 * groff's strange compatibility mode, into `de'.
                    695:                 */
                    696:
1.106     kristaps  697:                if (ROFF_de1 == tok)
                    698:                        tok = ROFF_de;
                    699:                if (ROFF_de == tok)
                    700:                        name = *bufp + pos;
                    701:                else
                    702:                        (*r->msg)(MANDOCERR_REQUEST, r->data, ln, ppos,
                    703:                            roffs[tok].name);
1.107     kristaps  704:
1.80      kristaps  705:                while ((*bufp)[pos] && ' ' != (*bufp)[pos])
                    706:                        pos++;
1.107     kristaps  707:
1.80      kristaps  708:                while (' ' == (*bufp)[pos])
1.106     kristaps  709:                        (*bufp)[pos++] = '\0';
1.80      kristaps  710:        }
                    711:
1.106     kristaps  712:        roffnode_push(r, tok, name, ln, ppos);
                    713:
                    714:        /*
                    715:         * At the beginning of a `de' macro, clear the existing string
                    716:         * with the same name, if there is one.  New content will be
                    717:         * added from roff_block_text() in multiline mode.
                    718:         */
1.107     kristaps  719:
1.106     kristaps  720:        if (ROFF_de == tok)
1.108     schwarze  721:                roff_setstr(r, name, "", 0);
1.76      kristaps  722:
1.79      kristaps  723:        if ('\0' == (*bufp)[pos])
1.78      kristaps  724:                return(ROFF_IGN);
                    725:
1.107     kristaps  726:        /* If present, process the custom end-of-line marker. */
                    727:
1.78      kristaps  728:        sv = pos;
1.108     schwarze  729:        while ((*bufp)[pos] &&
1.107     kristaps  730:                        ' ' != (*bufp)[pos] &&
1.78      kristaps  731:                        '\t' != (*bufp)[pos])
                    732:                pos++;
                    733:
                    734:        /*
                    735:         * Note: groff does NOT like escape characters in the input.
                    736:         * Instead of detecting this, we're just going to let it fly and
                    737:         * to hell with it.
                    738:         */
                    739:
                    740:        assert(pos > sv);
                    741:        sz = (size_t)(pos - sv);
                    742:
1.79      kristaps  743:        if (1 == sz && '.' == (*bufp)[sv])
                    744:                return(ROFF_IGN);
                    745:
1.98      schwarze  746:        r->last->end = mandoc_malloc(sz + 1);
1.78      kristaps  747:
                    748:        memcpy(r->last->end, *bufp + sv, sz);
                    749:        r->last->end[(int)sz] = '\0';
                    750:
1.77      kristaps  751:        if ((*bufp)[pos])
1.107     kristaps  752:                (*r->msg)(MANDOCERR_ARGSLOST, r->data, ln, pos, NULL);
1.74      kristaps  753:
1.78      kristaps  754:        return(ROFF_IGN);
                    755: }
                    756:
                    757:
                    758: /* ARGSUSED */
                    759: static enum rofferr
1.80      kristaps  760: roff_block_sub(ROFF_ARGS)
1.79      kristaps  761: {
                    762:        enum rofft      t;
                    763:        int             i, j;
                    764:
                    765:        /*
                    766:         * First check whether a custom macro exists at this level.  If
                    767:         * it does, then check against it.  This is some of groff's
                    768:         * stranger behaviours.  If we encountered a custom end-scope
                    769:         * tag and that tag also happens to be a "real" macro, then we
                    770:         * need to try interpreting it again as a real macro.  If it's
                    771:         * not, then return ignore.  Else continue.
                    772:         */
                    773:
                    774:        if (r->last->end) {
                    775:                i = pos + 1;
                    776:                while (' ' == (*bufp)[i] || '\t' == (*bufp)[i])
                    777:                        i++;
                    778:
                    779:                for (j = 0; r->last->end[j]; j++, i++)
                    780:                        if ((*bufp)[i] != r->last->end[j])
                    781:                                break;
                    782:
                    783:                if ('\0' == r->last->end[j] &&
                    784:                                ('\0' == (*bufp)[i] ||
                    785:                                 ' ' == (*bufp)[i] ||
                    786:                                 '\t' == (*bufp)[i])) {
                    787:                        roffnode_pop(r);
                    788:                        roffnode_cleanscope(r);
                    789:
1.106     kristaps  790:                        if (ROFF_MAX != roff_parse(r, *bufp, &pos))
1.79      kristaps  791:                                return(ROFF_RERUN);
                    792:                        return(ROFF_IGN);
                    793:                }
                    794:        }
                    795:
                    796:        /*
                    797:         * If we have no custom end-query or lookup failed, then try
                    798:         * pulling it out of the hashtable.
                    799:         */
                    800:
                    801:        ppos = pos;
1.106     kristaps  802:        t = roff_parse(r, *bufp, &pos);
1.79      kristaps  803:
1.106     kristaps  804:        /*
                    805:         * Macros other than block-end are only significant
                    806:         * in `de' blocks; elsewhere, simply throw them away.
                    807:         */
                    808:        if (ROFF_cblock != t) {
                    809:                if (ROFF_de == tok)
                    810:                        roff_setstr(r, r->last->name, *bufp + ppos, 1);
1.79      kristaps  811:                return(ROFF_IGN);
1.106     kristaps  812:        }
1.79      kristaps  813:
                    814:        assert(roffs[t].proc);
1.90      kristaps  815:        return((*roffs[t].proc)(r, t, bufp, szp,
                    816:                                ln, ppos, pos, offs));
1.79      kristaps  817: }
                    818:
                    819:
                    820: /* ARGSUSED */
                    821: static enum rofferr
1.80      kristaps  822: roff_block_text(ROFF_ARGS)
1.78      kristaps  823: {
                    824:
1.106     kristaps  825:        if (ROFF_de == tok)
                    826:                roff_setstr(r, r->last->name, *bufp + pos, 1);
                    827:
1.78      kristaps  828:        return(ROFF_IGN);
                    829: }
                    830:
                    831:
                    832: /* ARGSUSED */
                    833: static enum rofferr
1.82      kristaps  834: roff_cond_sub(ROFF_ARGS)
                    835: {
                    836:        enum rofft       t;
                    837:        enum roffrule    rr;
                    838:
                    839:        ppos = pos;
                    840:        rr = r->last->rule;
                    841:
1.87      kristaps  842:        /*
                    843:         * Clean out scope.  If we've closed ourselves, then don't
                    844:         * continue.
                    845:         */
                    846:
                    847:        roffnode_cleanscope(r);
1.82      kristaps  848:
1.106     kristaps  849:        if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos))) {
1.100     kristaps  850:                if ('\\' == (*bufp)[pos] && '}' == (*bufp)[pos + 1])
                    851:                        return(roff_ccond
1.106     kristaps  852:                                (r, ROFF_ccond, bufp, szp,
1.100     kristaps  853:                                 ln, pos, pos + 2, offs));
1.82      kristaps  854:                return(ROFFRULE_DENY == rr ? ROFF_IGN : ROFF_CONT);
1.100     kristaps  855:        }
1.82      kristaps  856:
                    857:        /*
                    858:         * A denied conditional must evaluate its children if and only
                    859:         * if they're either structurally required (such as loops and
                    860:         * conditionals) or a closing macro.
                    861:         */
                    862:        if (ROFFRULE_DENY == rr)
                    863:                if ( ! (ROFFMAC_STRUCT & roffs[t].flags))
                    864:                        if (ROFF_ccond != t)
                    865:                                return(ROFF_IGN);
                    866:
                    867:        assert(roffs[t].proc);
1.90      kristaps  868:        return((*roffs[t].proc)(r, t, bufp, szp,
                    869:                                ln, ppos, pos, offs));
1.82      kristaps  870: }
                    871:
                    872:
                    873: /* ARGSUSED */
                    874: static enum rofferr
                    875: roff_cond_text(ROFF_ARGS)
1.78      kristaps  876: {
                    877:        char            *ep, *st;
1.82      kristaps  878:        enum roffrule    rr;
                    879:
                    880:        rr = r->last->rule;
                    881:
                    882:        /*
                    883:         * We display the value of the text if out current evaluation
                    884:         * scope permits us to do so.
                    885:         */
1.100     kristaps  886:
                    887:        /* FIXME: use roff_ccond? */
1.78      kristaps  888:
                    889:        st = &(*bufp)[pos];
                    890:        if (NULL == (ep = strstr(st, "\\}"))) {
                    891:                roffnode_cleanscope(r);
1.82      kristaps  892:                return(ROFFRULE_DENY == rr ? ROFF_IGN : ROFF_CONT);
1.78      kristaps  893:        }
                    894:
1.86      kristaps  895:        if (ep == st || (ep > st && '\\' != *(ep - 1)))
1.78      kristaps  896:                roffnode_pop(r);
                    897:
                    898:        roffnode_cleanscope(r);
1.82      kristaps  899:        return(ROFFRULE_DENY == rr ? ROFF_IGN : ROFF_CONT);
1.74      kristaps  900: }
                    901:
                    902:
1.88      kristaps  903: static enum roffrule
                    904: roff_evalcond(const char *v, int *pos)
                    905: {
                    906:
                    907:        switch (v[*pos]) {
                    908:        case ('n'):
                    909:                (*pos)++;
                    910:                return(ROFFRULE_ALLOW);
                    911:        case ('e'):
                    912:                /* FALLTHROUGH */
                    913:        case ('o'):
                    914:                /* FALLTHROUGH */
                    915:        case ('t'):
                    916:                (*pos)++;
                    917:                return(ROFFRULE_DENY);
                    918:        default:
                    919:                break;
                    920:        }
                    921:
                    922:        while (v[*pos] && ' ' != v[*pos])
                    923:                (*pos)++;
                    924:        return(ROFFRULE_DENY);
                    925: }
                    926:
1.75      kristaps  927: /* ARGSUSED */
1.74      kristaps  928: static enum rofferr
1.103     kristaps  929: roff_line_ignore(ROFF_ARGS)
1.89      kristaps  930: {
                    931:
                    932:        return(ROFF_IGN);
                    933: }
                    934:
1.104     kristaps  935: /* ARGSUSED */
                    936: static enum rofferr
                    937: roff_line_error(ROFF_ARGS)
                    938: {
                    939:
                    940:        (*r->msg)(MANDOCERR_REQUEST, r->data, ln, ppos, roffs[tok].name);
                    941:        return(ROFF_IGN);
                    942: }
1.89      kristaps  943:
                    944: /* ARGSUSED */
                    945: static enum rofferr
1.82      kristaps  946: roff_cond(ROFF_ARGS)
1.74      kristaps  947: {
1.77      kristaps  948:        int              sv;
1.88      kristaps  949:        enum roffrule    rule;
1.74      kristaps  950:
1.82      kristaps  951:        /* Stack overflow! */
                    952:
                    953:        if (ROFF_ie == tok && r->rstackpos == RSTACK_MAX - 1) {
                    954:                (*r->msg)(MANDOCERR_MEM, r->data, ln, ppos, NULL);
                    955:                return(ROFF_ERR);
                    956:        }
1.74      kristaps  957:
1.88      kristaps  958:        /* First, evaluate the conditional. */
1.84      schwarze  959:
1.88      kristaps  960:        if (ROFF_el == tok) {
                    961:                /*
                    962:                 * An `.el' will get the value of the current rstack
                    963:                 * entry set in prior `ie' calls or defaults to DENY.
                    964:                 */
                    965:                if (r->rstackpos < 0)
                    966:                        rule = ROFFRULE_DENY;
                    967:                else
                    968:                        rule = r->rstack[r->rstackpos];
                    969:        } else
                    970:                rule = roff_evalcond(*bufp, &pos);
1.77      kristaps  971:
                    972:        sv = pos;
1.88      kristaps  973:
1.75      kristaps  974:        while (' ' == (*bufp)[pos])
                    975:                pos++;
1.74      kristaps  976:
1.77      kristaps  977:        /*
                    978:         * Roff is weird.  If we have just white-space after the
                    979:         * conditional, it's considered the BODY and we exit without
                    980:         * really doing anything.  Warn about this.  It's probably
                    981:         * wrong.
                    982:         */
1.88      kristaps  983:
1.77      kristaps  984:        if ('\0' == (*bufp)[pos] && sv != pos) {
1.107     kristaps  985:                (*r->msg)(MANDOCERR_NOARGS, r->data, ln, ppos, NULL);
                    986:                return(ROFF_IGN);
1.77      kristaps  987:        }
                    988:
1.106     kristaps  989:        roffnode_push(r, tok, NULL, ln, ppos);
1.77      kristaps  990:
1.88      kristaps  991:        r->last->rule = rule;
                    992:
1.84      schwarze  993:        if (ROFF_ie == tok) {
1.82      kristaps  994:                /*
                    995:                 * An if-else will put the NEGATION of the current
                    996:                 * evaluated conditional into the stack.
                    997:                 */
                    998:                r->rstackpos++;
                    999:                if (ROFFRULE_DENY == r->last->rule)
                   1000:                        r->rstack[r->rstackpos] = ROFFRULE_ALLOW;
                   1001:                else
                   1002:                        r->rstack[r->rstackpos] = ROFFRULE_DENY;
                   1003:        }
1.88      kristaps 1004:
                   1005:        /* If the parent has false as its rule, then so do we. */
                   1006:
1.109     kristaps 1007:        if (r->last->parent && ROFFRULE_DENY == r->last->parent->rule)
1.84      schwarze 1008:                r->last->rule = ROFFRULE_DENY;
1.88      kristaps 1009:
                   1010:        /*
                   1011:         * Determine scope.  If we're invoked with "\{" trailing the
                   1012:         * conditional, then we're in a multiline scope.  Else our scope
                   1013:         * expires on the next line.
                   1014:         */
1.74      kristaps 1015:
1.75      kristaps 1016:        r->last->endspan = 1;
                   1017:
                   1018:        if ('\\' == (*bufp)[pos] && '{' == (*bufp)[pos + 1]) {
                   1019:                r->last->endspan = -1;
                   1020:                pos += 2;
1.109     kristaps 1021:        }
1.74      kristaps 1022:
1.77      kristaps 1023:        /*
                   1024:         * If there are no arguments on the line, the next-line scope is
                   1025:         * assumed.
                   1026:         */
                   1027:
1.75      kristaps 1028:        if ('\0' == (*bufp)[pos])
                   1029:                return(ROFF_IGN);
1.77      kristaps 1030:
                   1031:        /* Otherwise re-run the roff parser after recalculating. */
1.74      kristaps 1032:
1.75      kristaps 1033:        *offs = pos;
                   1034:        return(ROFF_RERUN);
1.83      schwarze 1035: }
                   1036:
                   1037:
                   1038: /* ARGSUSED */
                   1039: static enum rofferr
1.92      schwarze 1040: roff_ds(ROFF_ARGS)
                   1041: {
1.96      kristaps 1042:        char            *name, *string;
                   1043:
                   1044:        /*
                   1045:         * A symbol is named by the first word following the macro
                   1046:         * invocation up to a space.  Its value is anything after the
                   1047:         * name's trailing whitespace and optional double-quote.  Thus,
                   1048:         *
                   1049:         *  [.ds foo "bar  "     ]
                   1050:         *
                   1051:         * will have `bar  "     ' as its value.
                   1052:         */
1.92      schwarze 1053:
                   1054:        name = *bufp + pos;
                   1055:        if ('\0' == *name)
                   1056:                return(ROFF_IGN);
                   1057:
                   1058:        string = name;
1.96      kristaps 1059:        /* Read until end of name. */
1.92      schwarze 1060:        while (*string && ' ' != *string)
                   1061:                string++;
1.96      kristaps 1062:
                   1063:        /* Nil-terminate name. */
1.92      schwarze 1064:        if (*string)
1.96      kristaps 1065:                *(string++) = '\0';
                   1066:
                   1067:        /* Read past spaces. */
                   1068:        while (*string && ' ' == *string)
                   1069:                string++;
                   1070:
                   1071:        /* Read passed initial double-quote. */
1.92      schwarze 1072:        if (*string && '"' == *string)
                   1073:                string++;
                   1074:
1.96      kristaps 1075:        /* The rest is the value. */
1.106     kristaps 1076:        roff_setstr(r, name, string, 0);
1.92      schwarze 1077:        return(ROFF_IGN);
                   1078: }
                   1079:
                   1080:
                   1081: /* ARGSUSED */
                   1082: static enum rofferr
1.89      kristaps 1083: roff_nr(ROFF_ARGS)
1.83      schwarze 1084: {
1.89      kristaps 1085:        const char      *key, *val;
1.91      kristaps 1086:        struct reg      *rg;
1.89      kristaps 1087:
                   1088:        key = &(*bufp)[pos];
1.91      kristaps 1089:        rg = r->regs->regs;
1.89      kristaps 1090:
                   1091:        /* Parse register request. */
                   1092:        while ((*bufp)[pos] && ' ' != (*bufp)[pos])
                   1093:                pos++;
                   1094:
                   1095:        /*
                   1096:         * Set our nil terminator.  Because this line is going to be
                   1097:         * ignored anyway, we can munge it as we please.
                   1098:         */
                   1099:        if ((*bufp)[pos])
                   1100:                (*bufp)[pos++] = '\0';
                   1101:
                   1102:        /* Skip whitespace to register token. */
                   1103:        while ((*bufp)[pos] && ' ' == (*bufp)[pos])
                   1104:                pos++;
                   1105:
                   1106:        val = &(*bufp)[pos];
                   1107:
                   1108:        /* Process register token. */
                   1109:
                   1110:        if (0 == strcmp(key, "nS")) {
1.91      kristaps 1111:                rg[(int)REG_nS].set = 1;
                   1112:                if ( ! roff_parse_nat(val, &rg[(int)REG_nS].v.u))
                   1113:                        rg[(int)REG_nS].v.u = 0;
1.109     kristaps 1114:        }
                   1115:
                   1116:        return(ROFF_IGN);
                   1117: }
                   1118:
                   1119: /* ARGSUSED */
                   1120: static enum rofferr
                   1121: roff_TE(ROFF_ARGS)
                   1122: {
                   1123:
                   1124:        if (NULL == r->tbl)
                   1125:                (*r->msg)(MANDOCERR_NOSCOPE, r->data, ln, ppos, NULL);
1.115   ! kristaps 1126:        else
        !          1127:                tbl_end(r->tbl);
1.109     kristaps 1128:
                   1129:        r->tbl = NULL;
1.112     kristaps 1130:        return(ROFF_IGN);
                   1131: }
                   1132:
                   1133: /* ARGSUSED */
                   1134: static enum rofferr
                   1135: roff_T_(ROFF_ARGS)
                   1136: {
                   1137:
                   1138:        if (NULL == r->tbl)
                   1139:                (*r->msg)(MANDOCERR_NOSCOPE, r->data, ln, ppos, NULL);
                   1140:        else
                   1141:                tbl_restart(r->tbl);
                   1142:
1.109     kristaps 1143:        return(ROFF_IGN);
                   1144: }
                   1145:
                   1146: /* ARGSUSED */
                   1147: static enum rofferr
                   1148: roff_TS(ROFF_ARGS)
                   1149: {
1.113     kristaps 1150:        struct tbl      *t;
1.89      kristaps 1151:
1.115   ! kristaps 1152:        if (r->tbl) {
1.109     kristaps 1153:                (*r->msg)(MANDOCERR_SCOPEBROKEN, r->data, ln, ppos, NULL);
1.115   ! kristaps 1154:                tbl_end(r->tbl);
        !          1155:        }
1.83      schwarze 1156:
1.115   ! kristaps 1157:        t = tbl_alloc(ppos, ln, r->data, r->msg);
1.113     kristaps 1158:
                   1159:        if (r->last_tbl)
                   1160:                r->last_tbl->next = t;
                   1161:        else
                   1162:                r->first_tbl = r->last_tbl = t;
                   1163:
                   1164:        r->tbl = r->last_tbl = t;
1.83      schwarze 1165:        return(ROFF_IGN);
1.92      schwarze 1166: }
                   1167:
1.105     kristaps 1168: /* ARGSUSED */
                   1169: static enum rofferr
                   1170: roff_so(ROFF_ARGS)
                   1171: {
                   1172:        char *name;
                   1173:
                   1174:        (*r->msg)(MANDOCERR_SO, r->data, ln, ppos, NULL);
                   1175:
                   1176:        /*
                   1177:         * Handle `so'.  Be EXTREMELY careful, as we shouldn't be
                   1178:         * opening anything that's not in our cwd or anything beneath
                   1179:         * it.  Thus, explicitly disallow traversing up the file-system
                   1180:         * or using absolute paths.
                   1181:         */
                   1182:
                   1183:        name = *bufp + pos;
                   1184:        if ('/' == *name || strstr(name, "../") || strstr(name, "/..")) {
                   1185:                (*r->msg)(MANDOCERR_SOPATH, r->data, ln, pos, NULL);
                   1186:                return(ROFF_ERR);
                   1187:        }
                   1188:
                   1189:        *offs = pos;
                   1190:        return(ROFF_SO);
                   1191: }
1.92      schwarze 1192:
1.106     kristaps 1193: /* ARGSUSED */
                   1194: static enum rofferr
                   1195: roff_userdef(ROFF_ARGS)
1.99      kristaps 1196: {
1.106     kristaps 1197:        const char       *arg[9];
                   1198:        char             *cp, *n1, *n2;
                   1199:        int               i, quoted, pairs;
                   1200:
                   1201:        /*
                   1202:         * Collect pointers to macro argument strings
                   1203:         * and null-terminate them.
                   1204:         */
                   1205:        cp = *bufp + pos;
                   1206:        for (i = 0; i < 9; i++) {
                   1207:                /* Quoting can only start with a new word. */
                   1208:                if ('"' == *cp) {
                   1209:                        quoted = 1;
                   1210:                        cp++;
                   1211:                } else
                   1212:                        quoted = 0;
                   1213:                arg[i] = cp;
                   1214:                for (pairs = 0; '\0' != *cp; cp++) {
                   1215:                        /* Unquoted arguments end at blanks. */
                   1216:                        if (0 == quoted) {
                   1217:                                if (' ' == *cp)
                   1218:                                        break;
                   1219:                                continue;
                   1220:                        }
                   1221:                        /* After pairs of quotes, move left. */
                   1222:                        if (pairs)
                   1223:                                cp[-pairs] = cp[0];
                   1224:                        /* Pairs of quotes do not end words, ... */
                   1225:                        if ('"' == cp[0] && '"' == cp[1]) {
                   1226:                                pairs++;
                   1227:                                cp++;
                   1228:                                continue;
                   1229:                        }
                   1230:                        /* ... but solitary quotes do. */
                   1231:                        if ('"' != *cp)
                   1232:                                continue;
                   1233:                        if (pairs)
                   1234:                                cp[-pairs] = '\0';
                   1235:                        *cp = ' ';
                   1236:                        break;
                   1237:                }
                   1238:                /* Last argument; the remaining ones are empty strings. */
                   1239:                if ('\0' == *cp)
                   1240:                        continue;
                   1241:                /* Null-terminate argument and move to the next one. */
                   1242:                *cp++ = '\0';
                   1243:                while (' ' == *cp)
                   1244:                        cp++;
                   1245:        }
1.99      kristaps 1246:
1.106     kristaps 1247:        /*
                   1248:         * Expand macro arguments.
1.99      kristaps 1249:         */
1.106     kristaps 1250:        *szp = 0;
                   1251:        n1 = cp = mandoc_strdup(r->current_string);
                   1252:        while (NULL != (cp = strstr(cp, "\\$"))) {
                   1253:                i = cp[2] - '1';
                   1254:                if (0 > i || 8 < i) {
                   1255:                        /* Not an argument invocation. */
                   1256:                        cp += 2;
                   1257:                        continue;
                   1258:                }
                   1259:
                   1260:                *szp = strlen(n1) - 3 + strlen(arg[i]) + 1;
                   1261:                n2 = mandoc_malloc(*szp);
                   1262:
                   1263:                strlcpy(n2, n1, (size_t)(cp - n1 + 1));
                   1264:                strlcat(n2, arg[i], *szp);
                   1265:                strlcat(n2, cp + 3, *szp);
                   1266:
                   1267:                cp = n2 + (cp - n1);
                   1268:                free(n1);
                   1269:                n1 = n2;
1.99      kristaps 1270:        }
                   1271:
1.106     kristaps 1272:        /*
                   1273:         * Replace the macro invocation
                   1274:         * by the expanded macro.
                   1275:         */
                   1276:        free(*bufp);
                   1277:        *bufp = n1;
                   1278:        if (0 == *szp)
                   1279:                *szp = strlen(*bufp) + 1;
                   1280:
                   1281:        return(*szp > 1 && '\n' == (*bufp)[(int)*szp - 2] ?
                   1282:           ROFF_REPARSE : ROFF_APPEND);
1.99      kristaps 1283: }
                   1284:
1.106     kristaps 1285: /*
                   1286:  * Store *string into the user-defined string called *name.
                   1287:  * In multiline mode, append to an existing entry and append '\n';
                   1288:  * else replace the existing entry, if there is one.
                   1289:  * To clear an existing entry, call with (*r, *name, NULL, 0).
                   1290:  */
1.94      kristaps 1291: static void
1.106     kristaps 1292: roff_setstr(struct roff *r, const char *name, const char *string,
                   1293:        int multiline)
1.92      schwarze 1294: {
                   1295:        struct roffstr   *n;
1.106     kristaps 1296:        char             *c;
                   1297:        size_t            oldch, newch;
1.92      schwarze 1298:
1.106     kristaps 1299:        /* Search for an existing string with the same name. */
1.94      kristaps 1300:        n = r->first_string;
1.92      schwarze 1301:        while (n && strcmp(name, n->name))
                   1302:                n = n->next;
1.94      kristaps 1303:
                   1304:        if (NULL == n) {
1.106     kristaps 1305:                /* Create a new string table entry. */
1.94      kristaps 1306:                n = mandoc_malloc(sizeof(struct roffstr));
1.106     kristaps 1307:                n->name = mandoc_strdup(name);
                   1308:                n->string = NULL;
1.94      kristaps 1309:                n->next = r->first_string;
                   1310:                r->first_string = n;
1.106     kristaps 1311:        } else if (0 == multiline) {
                   1312:                /* In multiline mode, append; else replace. */
1.92      schwarze 1313:                free(n->string);
1.106     kristaps 1314:                n->string = NULL;
                   1315:        }
                   1316:
                   1317:        if (NULL == string)
                   1318:                return;
                   1319:
                   1320:        /*
                   1321:         * One additional byte for the '\n' in multiline mode,
                   1322:         * and one for the terminating '\0'.
                   1323:         */
                   1324:        newch = strlen(string) + (multiline ? 2 : 1);
                   1325:        if (NULL == n->string) {
                   1326:                n->string = mandoc_malloc(newch);
                   1327:                *n->string = '\0';
                   1328:                oldch = 0;
                   1329:        } else {
                   1330:                oldch = strlen(n->string);
                   1331:                n->string = mandoc_realloc(n->string, oldch + newch);
                   1332:        }
                   1333:
                   1334:        /* Skip existing content in the destination buffer. */
                   1335:        c = n->string + oldch;
                   1336:
                   1337:        /* Append new content to the destination buffer. */
                   1338:        while (*string) {
                   1339:                /*
                   1340:                 * Rudimentary roff copy mode:
                   1341:                 * Handle escaped backslashes.
                   1342:                 */
                   1343:                if ('\\' == *string && '\\' == *(string + 1))
                   1344:                        string++;
                   1345:                *c++ = *string++;
                   1346:        }
1.94      kristaps 1347:
1.106     kristaps 1348:        /* Append terminating bytes. */
                   1349:        if (multiline)
                   1350:                *c++ = '\n';
                   1351:        *c = '\0';
1.92      schwarze 1352: }
                   1353:
                   1354:
1.94      kristaps 1355: static const char *
                   1356: roff_getstrn(const struct roff *r, const char *name, size_t len)
1.92      schwarze 1357: {
1.94      kristaps 1358:        const struct roffstr *n;
1.92      schwarze 1359:
1.94      kristaps 1360:        n = r->first_string;
1.97      kristaps 1361:        while (n && (strncmp(name, n->name, len) || '\0' != n->name[(int)len]))
1.92      schwarze 1362:                n = n->next;
1.94      kristaps 1363:
                   1364:        return(n ? n->string : NULL);
1.92      schwarze 1365: }
                   1366:
1.94      kristaps 1367:
                   1368: static void
                   1369: roff_freestr(struct roff *r)
1.92      schwarze 1370: {
                   1371:        struct roffstr   *n, *nn;
                   1372:
1.94      kristaps 1373:        for (n = r->first_string; n; n = nn) {
1.92      schwarze 1374:                free(n->name);
                   1375:                free(n->string);
                   1376:                nn = n->next;
                   1377:                free(n);
                   1378:        }
1.94      kristaps 1379:
                   1380:        r->first_string = NULL;
1.114     kristaps 1381: }
                   1382:
                   1383: const struct tbl_span *
                   1384: roff_span(const struct roff *r)
                   1385: {
                   1386:
                   1387:        return(r->tbl ? tbl_span(r->tbl) : NULL);
1.74      kristaps 1388: }

CVSweb