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

Annotation of mandoc/roff.c, Revision 1.111

1.111   ! kristaps    1: /*     $Id: roff.c,v 1.110 2010/12/28 10:59:07 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,
1.76      kristaps   62:        ROFF_cblock,
1.100     kristaps   63:        ROFF_ccond, /* FIXME: remove this. */
1.106     kristaps   64:        ROFF_USERDEF,
1.67      kristaps   65:        ROFF_MAX
                     66: };
                     67:
1.82      kristaps   68: enum   roffrule {
                     69:        ROFFRULE_ALLOW,
                     70:        ROFFRULE_DENY
                     71: };
                     72:
1.94      kristaps   73: struct roffstr {
                     74:        char            *name; /* key of symbol */
                     75:        char            *string; /* current value */
                     76:        struct roffstr  *next; /* next in list */
                     77: };
                     78:
1.67      kristaps   79: struct roff {
                     80:        struct roffnode *last; /* leaf of stack */
                     81:        mandocmsg        msg; /* err/warn/fatal messages */
                     82:        void            *data; /* privdata for messages */
1.82      kristaps   83:        enum roffrule    rstack[RSTACK_MAX]; /* stack of !`ie' rules */
                     84:        int              rstackpos; /* position in rstack */
1.90      kristaps   85:        struct regset   *regs; /* read/writable registers */
1.106     kristaps   86:        struct roffstr  *first_string; /* user-defined strings & macros */
                     87:        const char      *current_string; /* value of last called user macro */
1.109     kristaps   88:        struct tbl      *tbl;
1.79      kristaps   89: };
                     90:
1.67      kristaps   91: struct roffnode {
                     92:        enum rofft       tok; /* type of node */
                     93:        struct roffnode *parent; /* up one in stack */
                     94:        int              line; /* parse line */
                     95:        int              col; /* parse col */
1.106     kristaps   96:        char            *name; /* node name, e.g. macro name */
1.79      kristaps   97:        char            *end; /* end-rules: custom token */
                     98:        int              endspan; /* end-rules: next-line or infty */
1.82      kristaps   99:        enum roffrule    rule; /* current evaluation rule */
1.67      kristaps  100: };
                    101:
                    102: #define        ROFF_ARGS        struct roff *r, /* parse ctx */ \
1.72      kristaps  103:                         enum rofft tok, /* tok of macro */ \
1.67      kristaps  104:                         char **bufp, /* input buffer */ \
                    105:                         size_t *szp, /* size of input buffer */ \
                    106:                         int ln, /* parse line */ \
1.75      kristaps  107:                         int ppos, /* original pos in buffer */ \
                    108:                         int pos, /* current pos in buffer */ \
1.74      kristaps  109:                         int *offs /* reset offset of buffer data */
1.67      kristaps  110:
                    111: typedef        enum rofferr (*roffproc)(ROFF_ARGS);
                    112:
                    113: struct roffmac {
                    114:        const char      *name; /* macro name */
1.79      kristaps  115:        roffproc         proc; /* process new macro */
                    116:        roffproc         text; /* process as child text of macro */
                    117:        roffproc         sub; /* process as child of macro */
                    118:        int              flags;
                    119: #define        ROFFMAC_STRUCT  (1 << 0) /* always interpret */
1.85      kristaps  120:        struct roffmac  *next;
1.67      kristaps  121: };
                    122:
1.80      kristaps  123: static enum rofferr     roff_block(ROFF_ARGS);
                    124: static enum rofferr     roff_block_text(ROFF_ARGS);
                    125: static enum rofferr     roff_block_sub(ROFF_ARGS);
                    126: static enum rofferr     roff_cblock(ROFF_ARGS);
                    127: static enum rofferr     roff_ccond(ROFF_ARGS);
1.82      kristaps  128: static enum rofferr     roff_cond(ROFF_ARGS);
                    129: static enum rofferr     roff_cond_text(ROFF_ARGS);
                    130: static enum rofferr     roff_cond_sub(ROFF_ARGS);
1.92      schwarze  131: static enum rofferr     roff_ds(ROFF_ARGS);
1.94      kristaps  132: static enum roffrule    roff_evalcond(const char *, int *);
                    133: static void             roff_freestr(struct roff *);
                    134: static const char      *roff_getstrn(const struct roff *,
                    135:                                const char *, size_t);
1.103     kristaps  136: static enum rofferr     roff_line_ignore(ROFF_ARGS);
1.104     kristaps  137: static enum rofferr     roff_line_error(ROFF_ARGS);
1.89      kristaps  138: static enum rofferr     roff_nr(ROFF_ARGS);
1.95      kristaps  139: static int              roff_res(struct roff *,
                    140:                                char **, size_t *, int);
1.94      kristaps  141: static void             roff_setstr(struct roff *,
1.106     kristaps  142:                                const char *, const char *, int);
1.105     kristaps  143: static enum rofferr     roff_so(ROFF_ARGS);
1.109     kristaps  144: static enum rofferr     roff_TE(ROFF_ARGS);
                    145: static enum rofferr     roff_TS(ROFF_ARGS);
1.106     kristaps  146: static enum rofferr     roff_userdef(ROFF_ARGS);
1.67      kristaps  147:
1.85      kristaps  148: /* See roff_hash_find() */
                    149:
                    150: #define        ASCII_HI         126
                    151: #define        ASCII_LO         33
                    152: #define        HASHWIDTH       (ASCII_HI - ASCII_LO + 1)
                    153:
                    154: static struct roffmac  *hash[HASHWIDTH];
                    155:
                    156: static struct roffmac   roffs[ROFF_MAX] = {
1.103     kristaps  157:        { "ad", roff_line_ignore, NULL, NULL, 0, NULL },
1.85      kristaps  158:        { "am", roff_block, roff_block_text, roff_block_sub, 0, NULL },
                    159:        { "ami", roff_block, roff_block_text, roff_block_sub, 0, NULL },
                    160:        { "am1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
                    161:        { "de", roff_block, roff_block_text, roff_block_sub, 0, NULL },
                    162:        { "dei", roff_block, roff_block_text, roff_block_sub, 0, NULL },
                    163:        { "de1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.92      schwarze  164:        { "ds", roff_ds, NULL, NULL, 0, NULL },
1.85      kristaps  165:        { "el", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
1.103     kristaps  166:        { "hy", roff_line_ignore, NULL, NULL, 0, NULL },
1.85      kristaps  167:        { "ie", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
                    168:        { "if", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
                    169:        { "ig", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.103     kristaps  170:        { "ne", roff_line_ignore, NULL, NULL, 0, NULL },
                    171:        { "nh", roff_line_ignore, NULL, NULL, 0, NULL },
1.104     kristaps  172:        { "nr", roff_nr, NULL, NULL, 0, NULL },
                    173:        { "rm", roff_line_error, NULL, NULL, 0, NULL },
1.105     kristaps  174:        { "so", roff_so, NULL, NULL, 0, NULL },
1.103     kristaps  175:        { "tr", roff_line_ignore, NULL, NULL, 0, NULL },
1.109     kristaps  176:        { "TS", roff_TS, NULL, NULL, 0, NULL },
                    177:        { "TE", roff_TE, NULL, NULL, 0, NULL },
1.85      kristaps  178:        { ".", roff_cblock, NULL, NULL, 0, NULL },
                    179:        { "\\}", roff_ccond, NULL, NULL, 0, NULL },
1.106     kristaps  180:        { NULL, roff_userdef, NULL, NULL, 0, NULL },
1.67      kristaps  181: };
                    182:
                    183: static void             roff_free1(struct roff *);
1.106     kristaps  184: static enum rofft       roff_hash_find(const char *, size_t);
1.85      kristaps  185: static void             roff_hash_init(void);
1.76      kristaps  186: static void             roffnode_cleanscope(struct roff *);
1.106     kristaps  187: static void             roffnode_push(struct roff *, enum rofft,
                    188:                                const char *, int, int);
1.67      kristaps  189: static void             roffnode_pop(struct roff *);
1.106     kristaps  190: static enum rofft       roff_parse(struct roff *, const char *, int *);
1.91      kristaps  191: static int              roff_parse_nat(const char *, unsigned int *);
1.67      kristaps  192:
1.85      kristaps  193: /* See roff_hash_find() */
                    194: #define        ROFF_HASH(p)    (p[0] - ASCII_LO)
                    195:
                    196: static void
                    197: roff_hash_init(void)
                    198: {
                    199:        struct roffmac   *n;
                    200:        int               buc, i;
                    201:
1.106     kristaps  202:        for (i = 0; i < (int)ROFF_USERDEF; i++) {
1.85      kristaps  203:                assert(roffs[i].name[0] >= ASCII_LO);
                    204:                assert(roffs[i].name[0] <= ASCII_HI);
                    205:
                    206:                buc = ROFF_HASH(roffs[i].name);
                    207:
                    208:                if (NULL != (n = hash[buc])) {
                    209:                        for ( ; n->next; n = n->next)
                    210:                                /* Do nothing. */ ;
                    211:                        n->next = &roffs[i];
                    212:                } else
                    213:                        hash[buc] = &roffs[i];
                    214:        }
                    215: }
                    216:
1.67      kristaps  217:
                    218: /*
                    219:  * Look up a roff token by its name.  Returns ROFF_MAX if no macro by
                    220:  * the nil-terminated string name could be found.
                    221:  */
                    222: static enum rofft
1.106     kristaps  223: roff_hash_find(const char *p, size_t s)
1.67      kristaps  224: {
1.85      kristaps  225:        int              buc;
                    226:        struct roffmac  *n;
1.67      kristaps  227:
1.85      kristaps  228:        /*
                    229:         * libroff has an extremely simple hashtable, for the time
                    230:         * being, which simply keys on the first character, which must
                    231:         * be printable, then walks a chain.  It works well enough until
                    232:         * optimised.
                    233:         */
                    234:
                    235:        if (p[0] < ASCII_LO || p[0] > ASCII_HI)
                    236:                return(ROFF_MAX);
                    237:
                    238:        buc = ROFF_HASH(p);
                    239:
                    240:        if (NULL == (n = hash[buc]))
                    241:                return(ROFF_MAX);
                    242:        for ( ; n; n = n->next)
1.106     kristaps  243:                if (0 == strncmp(n->name, p, s) && '\0' == n->name[(int)s])
1.85      kristaps  244:                        return((enum rofft)(n - roffs));
1.67      kristaps  245:
                    246:        return(ROFF_MAX);
                    247: }
                    248:
                    249:
                    250: /*
                    251:  * Pop the current node off of the stack of roff instructions currently
                    252:  * pending.
                    253:  */
                    254: static void
                    255: roffnode_pop(struct roff *r)
                    256: {
                    257:        struct roffnode *p;
                    258:
1.75      kristaps  259:        assert(r->last);
                    260:        p = r->last;
1.82      kristaps  261:
                    262:        if (ROFF_el == p->tok)
                    263:                if (r->rstackpos > -1)
                    264:                        r->rstackpos--;
                    265:
1.75      kristaps  266:        r->last = r->last->parent;
1.106     kristaps  267:        free(p->name);
                    268:        free(p->end);
1.67      kristaps  269:        free(p);
                    270: }
                    271:
                    272:
                    273: /*
                    274:  * Push a roff node onto the instruction stack.  This must later be
                    275:  * removed with roffnode_pop().
                    276:  */
1.98      schwarze  277: static void
1.106     kristaps  278: roffnode_push(struct roff *r, enum rofft tok, const char *name,
                    279:                int line, int col)
1.67      kristaps  280: {
                    281:        struct roffnode *p;
                    282:
1.98      schwarze  283:        p = mandoc_calloc(1, sizeof(struct roffnode));
1.67      kristaps  284:        p->tok = tok;
1.106     kristaps  285:        if (name)
                    286:                p->name = mandoc_strdup(name);
1.67      kristaps  287:        p->parent = r->last;
                    288:        p->line = line;
                    289:        p->col = col;
1.79      kristaps  290:        p->rule = p->parent ? p->parent->rule : ROFFRULE_DENY;
1.67      kristaps  291:
                    292:        r->last = p;
                    293: }
                    294:
                    295:
                    296: static void
                    297: roff_free1(struct roff *r)
                    298: {
                    299:
1.109     kristaps  300:        if (r->tbl) {
                    301:                tbl_free(r->tbl);
                    302:                r->tbl = NULL;
                    303:        }
                    304:
1.67      kristaps  305:        while (r->last)
                    306:                roffnode_pop(r);
1.109     kristaps  307:
1.94      kristaps  308:        roff_freestr(r);
1.67      kristaps  309: }
                    310:
                    311:
                    312: void
                    313: roff_reset(struct roff *r)
                    314: {
                    315:
                    316:        roff_free1(r);
                    317: }
                    318:
                    319:
                    320: void
                    321: roff_free(struct roff *r)
                    322: {
                    323:
                    324:        roff_free1(r);
                    325:        free(r);
                    326: }
                    327:
                    328:
                    329: struct roff *
1.98      schwarze  330: roff_alloc(struct regset *regs, void *data, const mandocmsg msg)
1.67      kristaps  331: {
                    332:        struct roff     *r;
                    333:
1.98      schwarze  334:        r = mandoc_calloc(1, sizeof(struct roff));
1.90      kristaps  335:        r->regs = regs;
1.67      kristaps  336:        r->msg = msg;
                    337:        r->data = data;
1.82      kristaps  338:        r->rstackpos = -1;
1.85      kristaps  339:
                    340:        roff_hash_init();
1.67      kristaps  341:        return(r);
                    342: }
                    343:
                    344:
1.94      kristaps  345: /*
                    346:  * Pre-filter each and every line for reserved words (one beginning with
                    347:  * `\*', e.g., `\*(ab').  These must be handled before the actual line
                    348:  * is processed.
                    349:  */
                    350: static int
1.95      kristaps  351: roff_res(struct roff *r, char **bufp, size_t *szp, int pos)
1.94      kristaps  352: {
1.108     schwarze  353:        const char      *stesc; /* start of an escape sequence ('\\') */
                    354:        const char      *stnam; /* start of the name, after "[(*" */
                    355:        const char      *cp;    /* end of the name, e.g. before ']' */
                    356:        const char      *res;   /* the string to be substituted */
1.94      kristaps  357:        int              i, maxl;
                    358:        size_t           nsz;
                    359:        char            *n;
                    360:
1.108     schwarze  361:        /* Search for a leading backslash and save a pointer to it. */
                    362:
                    363:        cp = *bufp + pos;
                    364:        while (NULL != (cp = strchr(cp, '\\'))) {
                    365:                stesc = cp++;
                    366:
                    367:                /*
                    368:                 * The second character must be an asterisk.
                    369:                 * If it isn't, skip it anyway:  It is escaped,
                    370:                 * so it can't start another escape sequence.
                    371:                 */
                    372:
                    373:                if ('\0' == *cp)
                    374:                        return(1);
                    375:                if ('*' != *cp++)
                    376:                        continue;
                    377:
                    378:                /*
                    379:                 * The third character decides the length
                    380:                 * of the name of the string.
                    381:                 * Save a pointer to the name.
                    382:                 */
                    383:
1.94      kristaps  384:                switch (*cp) {
1.108     schwarze  385:                case ('\0'):
                    386:                        return(1);
1.94      kristaps  387:                case ('('):
                    388:                        cp++;
                    389:                        maxl = 2;
                    390:                        break;
                    391:                case ('['):
                    392:                        cp++;
                    393:                        maxl = 0;
                    394:                        break;
                    395:                default:
                    396:                        maxl = 1;
                    397:                        break;
                    398:                }
1.108     schwarze  399:                stnam = cp;
1.94      kristaps  400:
1.108     schwarze  401:                /* Advance to the end of the name. */
1.94      kristaps  402:
                    403:                for (i = 0; 0 == maxl || i < maxl; i++, cp++) {
                    404:                        if ('\0' == *cp)
                    405:                                return(1); /* Error. */
                    406:                        if (0 == maxl && ']' == *cp)
                    407:                                break;
                    408:                }
                    409:
1.108     schwarze  410:                /*
                    411:                 * Retrieve the replacement string; if it is
                    412:                 * undefined, resume searching for escapes.
                    413:                 */
                    414:
                    415:                res = roff_getstrn(r, stnam, (size_t)i);
1.94      kristaps  416:
                    417:                if (NULL == res) {
                    418:                        cp -= maxl ? 1 : 0;
                    419:                        continue;
                    420:                }
                    421:
1.108     schwarze  422:                /* Replace the escape sequence by the string. */
                    423:
1.94      kristaps  424:                nsz = *szp + strlen(res) + 1;
                    425:                n = mandoc_malloc(nsz);
                    426:
1.108     schwarze  427:                strlcpy(n, *bufp, (size_t)(stesc - *bufp + 1));
1.94      kristaps  428:                strlcat(n, res, nsz);
                    429:                strlcat(n, cp + (maxl ? 0 : 1), nsz);
                    430:
                    431:                free(*bufp);
                    432:
                    433:                *bufp = n;
                    434:                *szp = nsz;
                    435:                return(0);
                    436:        }
                    437:
                    438:        return(1);
                    439: }
                    440:
                    441:
1.67      kristaps  442: enum rofferr
1.90      kristaps  443: roff_parseln(struct roff *r, int ln, char **bufp,
                    444:                size_t *szp, int pos, int *offs)
1.67      kristaps  445: {
                    446:        enum rofft       t;
1.109     kristaps  447:        enum rofferr     e;
1.79      kristaps  448:        int              ppos;
                    449:
                    450:        /*
1.94      kristaps  451:         * Run the reserved-word filter only if we have some reserved
                    452:         * words to fill in.
                    453:         */
                    454:
1.95      kristaps  455:        if (r->first_string && ! roff_res(r, bufp, szp, pos))
1.106     kristaps  456:                return(ROFF_REPARSE);
1.94      kristaps  457:
                    458:        /*
1.79      kristaps  459:         * First, if a scope is open and we're not a macro, pass the
                    460:         * text through the macro's filter.  If a scope isn't open and
                    461:         * we're not a macro, just let it through.
                    462:         */
1.74      kristaps  463:
1.75      kristaps  464:        if (r->last && ! ROFF_CTL((*bufp)[pos])) {
1.78      kristaps  465:                t = r->last->tok;
                    466:                assert(roffs[t].text);
1.109     kristaps  467:                e = (*roffs[t].text)
                    468:                        (r, t, bufp, szp, ln, pos, pos, offs);
                    469:                assert(ROFF_IGN == e || ROFF_CONT == e);
                    470:                if (ROFF_CONT == e && r->tbl)
                    471:                        return(tbl_read(r->tbl, ln, *bufp, *offs));
                    472:                return(e);
                    473:        } else if ( ! ROFF_CTL((*bufp)[pos])) {
                    474:                if (r->tbl)
                    475:                        return(tbl_read(r->tbl, ln, *bufp, *offs));
1.67      kristaps  476:                return(ROFF_CONT);
1.109     kristaps  477:        }
1.67      kristaps  478:
1.79      kristaps  479:        /*
                    480:         * If a scope is open, go to the child handler for that macro,
                    481:         * as it may want to preprocess before doing anything with it.
                    482:         */
1.78      kristaps  483:
1.79      kristaps  484:        if (r->last) {
                    485:                t = r->last->tok;
                    486:                assert(roffs[t].sub);
                    487:                return((*roffs[t].sub)
1.90      kristaps  488:                                (r, t, bufp, szp,
                    489:                                 ln, pos, pos, offs));
1.79      kristaps  490:        }
1.78      kristaps  491:
1.79      kristaps  492:        /*
                    493:         * Lastly, as we've no scope open, try to look up and execute
                    494:         * the new macro.  If no macro is found, simply return and let
                    495:         * the compilers handle it.
                    496:         */
1.67      kristaps  497:
1.75      kristaps  498:        ppos = pos;
1.106     kristaps  499:        if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos)))
1.79      kristaps  500:                return(ROFF_CONT);
1.67      kristaps  501:
1.75      kristaps  502:        assert(roffs[t].proc);
1.78      kristaps  503:        return((*roffs[t].proc)
1.90      kristaps  504:                        (r, t, bufp, szp,
                    505:                         ln, ppos, pos, offs));
1.74      kristaps  506: }
                    507:
                    508:
                    509: int
                    510: roff_endparse(struct roff *r)
                    511: {
                    512:
1.110     kristaps  513:        /* FIXME: if r->tbl */
                    514:        if (r->last)
1.109     kristaps  515:                (*r->msg)(MANDOCERR_SCOPEEXIT, r->data,
                    516:                                r->last->line, r->last->col, NULL);
                    517:        return(1);
1.67      kristaps  518: }
                    519:
                    520:
                    521: /*
                    522:  * Parse a roff node's type from the input buffer.  This must be in the
                    523:  * form of ".foo xxx" in the usual way.
                    524:  */
                    525: static enum rofft
1.106     kristaps  526: roff_parse(struct roff *r, const char *buf, int *pos)
1.67      kristaps  527: {
1.106     kristaps  528:        const char      *mac;
                    529:        size_t           maclen;
1.67      kristaps  530:        enum rofft       t;
                    531:
1.75      kristaps  532:        assert(ROFF_CTL(buf[*pos]));
                    533:        (*pos)++;
1.67      kristaps  534:
1.106     kristaps  535:        while (' ' == buf[*pos] || '\t' == buf[*pos])
1.67      kristaps  536:                (*pos)++;
                    537:
                    538:        if ('\0' == buf[*pos])
                    539:                return(ROFF_MAX);
                    540:
1.106     kristaps  541:        mac = buf + *pos;
                    542:        maclen = strcspn(mac, " \\\t\0");
1.67      kristaps  543:
1.106     kristaps  544:        t = (r->current_string = roff_getstrn(r, mac, maclen))
                    545:            ? ROFF_USERDEF : roff_hash_find(mac, maclen);
1.67      kristaps  546:
1.106     kristaps  547:        *pos += maclen;
1.67      kristaps  548:        while (buf[*pos] && ' ' == buf[*pos])
                    549:                (*pos)++;
                    550:
                    551:        return(t);
                    552: }
                    553:
                    554:
1.89      kristaps  555: static int
1.91      kristaps  556: roff_parse_nat(const char *buf, unsigned int *res)
1.89      kristaps  557: {
                    558:        char            *ep;
                    559:        long             lval;
                    560:
                    561:        errno = 0;
                    562:        lval = strtol(buf, &ep, 10);
                    563:        if (buf[0] == '\0' || *ep != '\0')
                    564:                return(0);
                    565:        if ((errno == ERANGE &&
                    566:                        (lval == LONG_MAX || lval == LONG_MIN)) ||
1.91      kristaps  567:                        (lval > INT_MAX || lval < 0))
1.89      kristaps  568:                return(0);
                    569:
1.91      kristaps  570:        *res = (unsigned int)lval;
1.89      kristaps  571:        return(1);
                    572: }
                    573:
                    574:
1.67      kristaps  575: /* ARGSUSED */
                    576: static enum rofferr
1.76      kristaps  577: roff_cblock(ROFF_ARGS)
1.67      kristaps  578: {
                    579:
1.79      kristaps  580:        /*
                    581:         * A block-close `..' should only be invoked as a child of an
                    582:         * ignore macro, otherwise raise a warning and just ignore it.
                    583:         */
                    584:
1.76      kristaps  585:        if (NULL == r->last) {
1.109     kristaps  586:                (*r->msg)(MANDOCERR_NOSCOPE, r->data, ln, ppos, NULL);
1.76      kristaps  587:                return(ROFF_IGN);
                    588:        }
1.67      kristaps  589:
1.81      kristaps  590:        switch (r->last->tok) {
                    591:        case (ROFF_am):
                    592:                /* FALLTHROUGH */
                    593:        case (ROFF_ami):
                    594:                /* FALLTHROUGH */
                    595:        case (ROFF_am1):
                    596:                /* FALLTHROUGH */
                    597:        case (ROFF_de):
1.108     schwarze  598:                /* ROFF_de1 is remapped to ROFF_de in roff_block(). */
1.81      kristaps  599:                /* FALLTHROUGH */
                    600:        case (ROFF_dei):
                    601:                /* FALLTHROUGH */
                    602:        case (ROFF_ig):
                    603:                break;
                    604:        default:
1.109     kristaps  605:                (*r->msg)(MANDOCERR_NOSCOPE, r->data, ln, ppos, NULL);
1.67      kristaps  606:                return(ROFF_IGN);
1.76      kristaps  607:        }
1.67      kristaps  608:
1.76      kristaps  609:        if ((*bufp)[pos])
1.109     kristaps  610:                (*r->msg)(MANDOCERR_ARGSLOST, r->data, ln, pos, NULL);
1.71      kristaps  611:
                    612:        roffnode_pop(r);
1.76      kristaps  613:        roffnode_cleanscope(r);
                    614:        return(ROFF_IGN);
1.71      kristaps  615:
1.67      kristaps  616: }
                    617:
                    618:
1.76      kristaps  619: static void
                    620: roffnode_cleanscope(struct roff *r)
1.67      kristaps  621: {
                    622:
1.76      kristaps  623:        while (r->last) {
                    624:                if (--r->last->endspan < 0)
                    625:                        break;
                    626:                roffnode_pop(r);
                    627:        }
1.67      kristaps  628: }
                    629:
                    630:
1.75      kristaps  631: /* ARGSUSED */
1.74      kristaps  632: static enum rofferr
1.75      kristaps  633: roff_ccond(ROFF_ARGS)
1.74      kristaps  634: {
                    635:
1.76      kristaps  636:        if (NULL == r->last) {
1.109     kristaps  637:                (*r->msg)(MANDOCERR_NOSCOPE, r->data, ln, ppos, NULL);
1.76      kristaps  638:                return(ROFF_IGN);
                    639:        }
                    640:
1.82      kristaps  641:        switch (r->last->tok) {
                    642:        case (ROFF_el):
                    643:                /* FALLTHROUGH */
                    644:        case (ROFF_ie):
                    645:                /* FALLTHROUGH */
                    646:        case (ROFF_if):
                    647:                break;
                    648:        default:
1.109     kristaps  649:                (*r->msg)(MANDOCERR_NOSCOPE, r->data, ln, ppos, NULL);
1.75      kristaps  650:                return(ROFF_IGN);
                    651:        }
                    652:
1.76      kristaps  653:        if (r->last->endspan > -1) {
1.109     kristaps  654:                (*r->msg)(MANDOCERR_NOSCOPE, r->data, ln, ppos, NULL);
1.76      kristaps  655:                return(ROFF_IGN);
                    656:        }
                    657:
                    658:        if ((*bufp)[pos])
1.109     kristaps  659:                (*r->msg)(MANDOCERR_ARGSLOST, r->data, ln, pos, NULL);
1.76      kristaps  660:
1.75      kristaps  661:        roffnode_pop(r);
1.76      kristaps  662:        roffnode_cleanscope(r);
                    663:        return(ROFF_IGN);
                    664: }
                    665:
1.75      kristaps  666:
1.76      kristaps  667: /* ARGSUSED */
                    668: static enum rofferr
1.80      kristaps  669: roff_block(ROFF_ARGS)
1.76      kristaps  670: {
1.78      kristaps  671:        int             sv;
                    672:        size_t          sz;
1.106     kristaps  673:        char            *name;
                    674:
                    675:        name = NULL;
1.76      kristaps  676:
1.106     kristaps  677:        if (ROFF_ig != tok) {
                    678:                if ('\0' == (*bufp)[pos]) {
                    679:                        (*r->msg)(MANDOCERR_NOARGS, r->data, ln, ppos, NULL);
                    680:                        return(ROFF_IGN);
                    681:                }
1.107     kristaps  682:
                    683:                /*
                    684:                 * Re-write `de1', since we don't really care about
                    685:                 * groff's strange compatibility mode, into `de'.
                    686:                 */
                    687:
1.106     kristaps  688:                if (ROFF_de1 == tok)
                    689:                        tok = ROFF_de;
                    690:                if (ROFF_de == tok)
                    691:                        name = *bufp + pos;
                    692:                else
                    693:                        (*r->msg)(MANDOCERR_REQUEST, r->data, ln, ppos,
                    694:                            roffs[tok].name);
1.107     kristaps  695:
1.80      kristaps  696:                while ((*bufp)[pos] && ' ' != (*bufp)[pos])
                    697:                        pos++;
1.107     kristaps  698:
1.80      kristaps  699:                while (' ' == (*bufp)[pos])
1.106     kristaps  700:                        (*bufp)[pos++] = '\0';
1.80      kristaps  701:        }
                    702:
1.106     kristaps  703:        roffnode_push(r, tok, name, ln, ppos);
                    704:
                    705:        /*
                    706:         * At the beginning of a `de' macro, clear the existing string
                    707:         * with the same name, if there is one.  New content will be
                    708:         * added from roff_block_text() in multiline mode.
                    709:         */
1.107     kristaps  710:
1.106     kristaps  711:        if (ROFF_de == tok)
1.108     schwarze  712:                roff_setstr(r, name, "", 0);
1.76      kristaps  713:
1.79      kristaps  714:        if ('\0' == (*bufp)[pos])
1.78      kristaps  715:                return(ROFF_IGN);
                    716:
1.107     kristaps  717:        /* If present, process the custom end-of-line marker. */
                    718:
1.78      kristaps  719:        sv = pos;
1.108     schwarze  720:        while ((*bufp)[pos] &&
1.107     kristaps  721:                        ' ' != (*bufp)[pos] &&
1.78      kristaps  722:                        '\t' != (*bufp)[pos])
                    723:                pos++;
                    724:
                    725:        /*
                    726:         * Note: groff does NOT like escape characters in the input.
                    727:         * Instead of detecting this, we're just going to let it fly and
                    728:         * to hell with it.
                    729:         */
                    730:
                    731:        assert(pos > sv);
                    732:        sz = (size_t)(pos - sv);
                    733:
1.79      kristaps  734:        if (1 == sz && '.' == (*bufp)[sv])
                    735:                return(ROFF_IGN);
                    736:
1.98      schwarze  737:        r->last->end = mandoc_malloc(sz + 1);
1.78      kristaps  738:
                    739:        memcpy(r->last->end, *bufp + sv, sz);
                    740:        r->last->end[(int)sz] = '\0';
                    741:
1.77      kristaps  742:        if ((*bufp)[pos])
1.107     kristaps  743:                (*r->msg)(MANDOCERR_ARGSLOST, r->data, ln, pos, NULL);
1.74      kristaps  744:
1.78      kristaps  745:        return(ROFF_IGN);
                    746: }
                    747:
                    748:
                    749: /* ARGSUSED */
                    750: static enum rofferr
1.80      kristaps  751: roff_block_sub(ROFF_ARGS)
1.79      kristaps  752: {
                    753:        enum rofft      t;
                    754:        int             i, j;
                    755:
                    756:        /*
                    757:         * First check whether a custom macro exists at this level.  If
                    758:         * it does, then check against it.  This is some of groff's
                    759:         * stranger behaviours.  If we encountered a custom end-scope
                    760:         * tag and that tag also happens to be a "real" macro, then we
                    761:         * need to try interpreting it again as a real macro.  If it's
                    762:         * not, then return ignore.  Else continue.
                    763:         */
                    764:
                    765:        if (r->last->end) {
                    766:                i = pos + 1;
                    767:                while (' ' == (*bufp)[i] || '\t' == (*bufp)[i])
                    768:                        i++;
                    769:
                    770:                for (j = 0; r->last->end[j]; j++, i++)
                    771:                        if ((*bufp)[i] != r->last->end[j])
                    772:                                break;
                    773:
                    774:                if ('\0' == r->last->end[j] &&
                    775:                                ('\0' == (*bufp)[i] ||
                    776:                                 ' ' == (*bufp)[i] ||
                    777:                                 '\t' == (*bufp)[i])) {
                    778:                        roffnode_pop(r);
                    779:                        roffnode_cleanscope(r);
                    780:
1.106     kristaps  781:                        if (ROFF_MAX != roff_parse(r, *bufp, &pos))
1.79      kristaps  782:                                return(ROFF_RERUN);
                    783:                        return(ROFF_IGN);
                    784:                }
                    785:        }
                    786:
                    787:        /*
                    788:         * If we have no custom end-query or lookup failed, then try
                    789:         * pulling it out of the hashtable.
                    790:         */
                    791:
                    792:        ppos = pos;
1.106     kristaps  793:        t = roff_parse(r, *bufp, &pos);
1.79      kristaps  794:
1.106     kristaps  795:        /*
                    796:         * Macros other than block-end are only significant
                    797:         * in `de' blocks; elsewhere, simply throw them away.
                    798:         */
                    799:        if (ROFF_cblock != t) {
                    800:                if (ROFF_de == tok)
                    801:                        roff_setstr(r, r->last->name, *bufp + ppos, 1);
1.79      kristaps  802:                return(ROFF_IGN);
1.106     kristaps  803:        }
1.79      kristaps  804:
                    805:        assert(roffs[t].proc);
1.90      kristaps  806:        return((*roffs[t].proc)(r, t, bufp, szp,
                    807:                                ln, ppos, pos, offs));
1.79      kristaps  808: }
                    809:
                    810:
                    811: /* ARGSUSED */
                    812: static enum rofferr
1.80      kristaps  813: roff_block_text(ROFF_ARGS)
1.78      kristaps  814: {
                    815:
1.106     kristaps  816:        if (ROFF_de == tok)
                    817:                roff_setstr(r, r->last->name, *bufp + pos, 1);
                    818:
1.78      kristaps  819:        return(ROFF_IGN);
                    820: }
                    821:
                    822:
                    823: /* ARGSUSED */
                    824: static enum rofferr
1.82      kristaps  825: roff_cond_sub(ROFF_ARGS)
                    826: {
                    827:        enum rofft       t;
                    828:        enum roffrule    rr;
                    829:
                    830:        ppos = pos;
                    831:        rr = r->last->rule;
                    832:
1.87      kristaps  833:        /*
                    834:         * Clean out scope.  If we've closed ourselves, then don't
                    835:         * continue.
                    836:         */
                    837:
                    838:        roffnode_cleanscope(r);
1.82      kristaps  839:
1.106     kristaps  840:        if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos))) {
1.100     kristaps  841:                if ('\\' == (*bufp)[pos] && '}' == (*bufp)[pos + 1])
                    842:                        return(roff_ccond
1.106     kristaps  843:                                (r, ROFF_ccond, bufp, szp,
1.100     kristaps  844:                                 ln, pos, pos + 2, offs));
1.82      kristaps  845:                return(ROFFRULE_DENY == rr ? ROFF_IGN : ROFF_CONT);
1.100     kristaps  846:        }
1.82      kristaps  847:
                    848:        /*
                    849:         * A denied conditional must evaluate its children if and only
                    850:         * if they're either structurally required (such as loops and
                    851:         * conditionals) or a closing macro.
                    852:         */
                    853:        if (ROFFRULE_DENY == rr)
                    854:                if ( ! (ROFFMAC_STRUCT & roffs[t].flags))
                    855:                        if (ROFF_ccond != t)
                    856:                                return(ROFF_IGN);
                    857:
                    858:        assert(roffs[t].proc);
1.90      kristaps  859:        return((*roffs[t].proc)(r, t, bufp, szp,
                    860:                                ln, ppos, pos, offs));
1.82      kristaps  861: }
                    862:
                    863:
                    864: /* ARGSUSED */
                    865: static enum rofferr
                    866: roff_cond_text(ROFF_ARGS)
1.78      kristaps  867: {
                    868:        char            *ep, *st;
1.82      kristaps  869:        enum roffrule    rr;
                    870:
                    871:        rr = r->last->rule;
                    872:
                    873:        /*
                    874:         * We display the value of the text if out current evaluation
                    875:         * scope permits us to do so.
                    876:         */
1.100     kristaps  877:
                    878:        /* FIXME: use roff_ccond? */
1.78      kristaps  879:
                    880:        st = &(*bufp)[pos];
                    881:        if (NULL == (ep = strstr(st, "\\}"))) {
                    882:                roffnode_cleanscope(r);
1.82      kristaps  883:                return(ROFFRULE_DENY == rr ? ROFF_IGN : ROFF_CONT);
1.78      kristaps  884:        }
                    885:
1.86      kristaps  886:        if (ep == st || (ep > st && '\\' != *(ep - 1)))
1.78      kristaps  887:                roffnode_pop(r);
                    888:
                    889:        roffnode_cleanscope(r);
1.82      kristaps  890:        return(ROFFRULE_DENY == rr ? ROFF_IGN : ROFF_CONT);
1.74      kristaps  891: }
                    892:
                    893:
1.88      kristaps  894: static enum roffrule
                    895: roff_evalcond(const char *v, int *pos)
                    896: {
                    897:
                    898:        switch (v[*pos]) {
                    899:        case ('n'):
                    900:                (*pos)++;
                    901:                return(ROFFRULE_ALLOW);
                    902:        case ('e'):
                    903:                /* FALLTHROUGH */
                    904:        case ('o'):
                    905:                /* FALLTHROUGH */
                    906:        case ('t'):
                    907:                (*pos)++;
                    908:                return(ROFFRULE_DENY);
                    909:        default:
                    910:                break;
                    911:        }
                    912:
                    913:        while (v[*pos] && ' ' != v[*pos])
                    914:                (*pos)++;
                    915:        return(ROFFRULE_DENY);
                    916: }
                    917:
1.75      kristaps  918: /* ARGSUSED */
1.74      kristaps  919: static enum rofferr
1.103     kristaps  920: roff_line_ignore(ROFF_ARGS)
1.89      kristaps  921: {
                    922:
                    923:        return(ROFF_IGN);
                    924: }
                    925:
1.104     kristaps  926: /* ARGSUSED */
                    927: static enum rofferr
                    928: roff_line_error(ROFF_ARGS)
                    929: {
                    930:
                    931:        (*r->msg)(MANDOCERR_REQUEST, r->data, ln, ppos, roffs[tok].name);
                    932:        return(ROFF_IGN);
                    933: }
1.89      kristaps  934:
                    935: /* ARGSUSED */
                    936: static enum rofferr
1.82      kristaps  937: roff_cond(ROFF_ARGS)
1.74      kristaps  938: {
1.77      kristaps  939:        int              sv;
1.88      kristaps  940:        enum roffrule    rule;
1.74      kristaps  941:
1.82      kristaps  942:        /* Stack overflow! */
                    943:
                    944:        if (ROFF_ie == tok && r->rstackpos == RSTACK_MAX - 1) {
                    945:                (*r->msg)(MANDOCERR_MEM, r->data, ln, ppos, NULL);
                    946:                return(ROFF_ERR);
                    947:        }
1.74      kristaps  948:
1.88      kristaps  949:        /* First, evaluate the conditional. */
1.84      schwarze  950:
1.88      kristaps  951:        if (ROFF_el == tok) {
                    952:                /*
                    953:                 * An `.el' will get the value of the current rstack
                    954:                 * entry set in prior `ie' calls or defaults to DENY.
                    955:                 */
                    956:                if (r->rstackpos < 0)
                    957:                        rule = ROFFRULE_DENY;
                    958:                else
                    959:                        rule = r->rstack[r->rstackpos];
                    960:        } else
                    961:                rule = roff_evalcond(*bufp, &pos);
1.77      kristaps  962:
                    963:        sv = pos;
1.88      kristaps  964:
1.75      kristaps  965:        while (' ' == (*bufp)[pos])
                    966:                pos++;
1.74      kristaps  967:
1.77      kristaps  968:        /*
                    969:         * Roff is weird.  If we have just white-space after the
                    970:         * conditional, it's considered the BODY and we exit without
                    971:         * really doing anything.  Warn about this.  It's probably
                    972:         * wrong.
                    973:         */
1.88      kristaps  974:
1.77      kristaps  975:        if ('\0' == (*bufp)[pos] && sv != pos) {
1.107     kristaps  976:                (*r->msg)(MANDOCERR_NOARGS, r->data, ln, ppos, NULL);
                    977:                return(ROFF_IGN);
1.77      kristaps  978:        }
                    979:
1.106     kristaps  980:        roffnode_push(r, tok, NULL, ln, ppos);
1.77      kristaps  981:
1.88      kristaps  982:        r->last->rule = rule;
                    983:
1.84      schwarze  984:        if (ROFF_ie == tok) {
1.82      kristaps  985:                /*
                    986:                 * An if-else will put the NEGATION of the current
                    987:                 * evaluated conditional into the stack.
                    988:                 */
                    989:                r->rstackpos++;
                    990:                if (ROFFRULE_DENY == r->last->rule)
                    991:                        r->rstack[r->rstackpos] = ROFFRULE_ALLOW;
                    992:                else
                    993:                        r->rstack[r->rstackpos] = ROFFRULE_DENY;
                    994:        }
1.88      kristaps  995:
                    996:        /* If the parent has false as its rule, then so do we. */
                    997:
1.109     kristaps  998:        if (r->last->parent && ROFFRULE_DENY == r->last->parent->rule)
1.84      schwarze  999:                r->last->rule = ROFFRULE_DENY;
1.88      kristaps 1000:
                   1001:        /*
                   1002:         * Determine scope.  If we're invoked with "\{" trailing the
                   1003:         * conditional, then we're in a multiline scope.  Else our scope
                   1004:         * expires on the next line.
                   1005:         */
1.74      kristaps 1006:
1.75      kristaps 1007:        r->last->endspan = 1;
                   1008:
                   1009:        if ('\\' == (*bufp)[pos] && '{' == (*bufp)[pos + 1]) {
                   1010:                r->last->endspan = -1;
                   1011:                pos += 2;
1.109     kristaps 1012:        }
1.74      kristaps 1013:
1.77      kristaps 1014:        /*
                   1015:         * If there are no arguments on the line, the next-line scope is
                   1016:         * assumed.
                   1017:         */
                   1018:
1.75      kristaps 1019:        if ('\0' == (*bufp)[pos])
                   1020:                return(ROFF_IGN);
1.77      kristaps 1021:
                   1022:        /* Otherwise re-run the roff parser after recalculating. */
1.74      kristaps 1023:
1.75      kristaps 1024:        *offs = pos;
                   1025:        return(ROFF_RERUN);
1.83      schwarze 1026: }
                   1027:
                   1028:
                   1029: /* ARGSUSED */
                   1030: static enum rofferr
1.92      schwarze 1031: roff_ds(ROFF_ARGS)
                   1032: {
1.96      kristaps 1033:        char            *name, *string;
                   1034:
                   1035:        /*
                   1036:         * A symbol is named by the first word following the macro
                   1037:         * invocation up to a space.  Its value is anything after the
                   1038:         * name's trailing whitespace and optional double-quote.  Thus,
                   1039:         *
                   1040:         *  [.ds foo "bar  "     ]
                   1041:         *
                   1042:         * will have `bar  "     ' as its value.
                   1043:         */
1.92      schwarze 1044:
                   1045:        name = *bufp + pos;
                   1046:        if ('\0' == *name)
                   1047:                return(ROFF_IGN);
                   1048:
                   1049:        string = name;
1.96      kristaps 1050:        /* Read until end of name. */
1.92      schwarze 1051:        while (*string && ' ' != *string)
                   1052:                string++;
1.96      kristaps 1053:
                   1054:        /* Nil-terminate name. */
1.92      schwarze 1055:        if (*string)
1.96      kristaps 1056:                *(string++) = '\0';
                   1057:
                   1058:        /* Read past spaces. */
                   1059:        while (*string && ' ' == *string)
                   1060:                string++;
                   1061:
                   1062:        /* Read passed initial double-quote. */
1.92      schwarze 1063:        if (*string && '"' == *string)
                   1064:                string++;
                   1065:
1.96      kristaps 1066:        /* The rest is the value. */
1.106     kristaps 1067:        roff_setstr(r, name, string, 0);
1.92      schwarze 1068:        return(ROFF_IGN);
                   1069: }
                   1070:
                   1071:
                   1072: /* ARGSUSED */
                   1073: static enum rofferr
1.89      kristaps 1074: roff_nr(ROFF_ARGS)
1.83      schwarze 1075: {
1.89      kristaps 1076:        const char      *key, *val;
1.91      kristaps 1077:        struct reg      *rg;
1.89      kristaps 1078:
                   1079:        key = &(*bufp)[pos];
1.91      kristaps 1080:        rg = r->regs->regs;
1.89      kristaps 1081:
                   1082:        /* Parse register request. */
                   1083:        while ((*bufp)[pos] && ' ' != (*bufp)[pos])
                   1084:                pos++;
                   1085:
                   1086:        /*
                   1087:         * Set our nil terminator.  Because this line is going to be
                   1088:         * ignored anyway, we can munge it as we please.
                   1089:         */
                   1090:        if ((*bufp)[pos])
                   1091:                (*bufp)[pos++] = '\0';
                   1092:
                   1093:        /* Skip whitespace to register token. */
                   1094:        while ((*bufp)[pos] && ' ' == (*bufp)[pos])
                   1095:                pos++;
                   1096:
                   1097:        val = &(*bufp)[pos];
                   1098:
                   1099:        /* Process register token. */
                   1100:
                   1101:        if (0 == strcmp(key, "nS")) {
1.91      kristaps 1102:                rg[(int)REG_nS].set = 1;
                   1103:                if ( ! roff_parse_nat(val, &rg[(int)REG_nS].v.u))
                   1104:                        rg[(int)REG_nS].v.u = 0;
1.109     kristaps 1105:        }
                   1106:
                   1107:        return(ROFF_IGN);
                   1108: }
                   1109:
                   1110: /* ARGSUSED */
                   1111: static enum rofferr
                   1112: roff_TE(ROFF_ARGS)
                   1113: {
                   1114:
                   1115:        if (NULL == r->tbl)
                   1116:                (*r->msg)(MANDOCERR_NOSCOPE, r->data, ln, ppos, NULL);
                   1117:        else
                   1118:                tbl_free(r->tbl);
                   1119:
                   1120:        r->tbl = NULL;
                   1121:        return(ROFF_IGN);
                   1122: }
                   1123:
                   1124: /* ARGSUSED */
                   1125: static enum rofferr
                   1126: roff_TS(ROFF_ARGS)
                   1127: {
1.89      kristaps 1128:
1.109     kristaps 1129:        if (r->tbl) {
                   1130:                (*r->msg)(MANDOCERR_SCOPEBROKEN, r->data, ln, ppos, NULL);
                   1131:                tbl_reset(r->tbl);
1.89      kristaps 1132:        } else
1.111   ! kristaps 1133:                r->tbl = tbl_alloc(r->data, r->msg);
1.83      schwarze 1134:
                   1135:        return(ROFF_IGN);
1.92      schwarze 1136: }
                   1137:
1.105     kristaps 1138: /* ARGSUSED */
                   1139: static enum rofferr
                   1140: roff_so(ROFF_ARGS)
                   1141: {
                   1142:        char *name;
                   1143:
                   1144:        (*r->msg)(MANDOCERR_SO, r->data, ln, ppos, NULL);
                   1145:
                   1146:        /*
                   1147:         * Handle `so'.  Be EXTREMELY careful, as we shouldn't be
                   1148:         * opening anything that's not in our cwd or anything beneath
                   1149:         * it.  Thus, explicitly disallow traversing up the file-system
                   1150:         * or using absolute paths.
                   1151:         */
                   1152:
                   1153:        name = *bufp + pos;
                   1154:        if ('/' == *name || strstr(name, "../") || strstr(name, "/..")) {
                   1155:                (*r->msg)(MANDOCERR_SOPATH, r->data, ln, pos, NULL);
                   1156:                return(ROFF_ERR);
                   1157:        }
                   1158:
                   1159:        *offs = pos;
                   1160:        return(ROFF_SO);
                   1161: }
1.92      schwarze 1162:
1.106     kristaps 1163: /* ARGSUSED */
                   1164: static enum rofferr
                   1165: roff_userdef(ROFF_ARGS)
1.99      kristaps 1166: {
1.106     kristaps 1167:        const char       *arg[9];
                   1168:        char             *cp, *n1, *n2;
                   1169:        int               i, quoted, pairs;
                   1170:
                   1171:        /*
                   1172:         * Collect pointers to macro argument strings
                   1173:         * and null-terminate them.
                   1174:         */
                   1175:        cp = *bufp + pos;
                   1176:        for (i = 0; i < 9; i++) {
                   1177:                /* Quoting can only start with a new word. */
                   1178:                if ('"' == *cp) {
                   1179:                        quoted = 1;
                   1180:                        cp++;
                   1181:                } else
                   1182:                        quoted = 0;
                   1183:                arg[i] = cp;
                   1184:                for (pairs = 0; '\0' != *cp; cp++) {
                   1185:                        /* Unquoted arguments end at blanks. */
                   1186:                        if (0 == quoted) {
                   1187:                                if (' ' == *cp)
                   1188:                                        break;
                   1189:                                continue;
                   1190:                        }
                   1191:                        /* After pairs of quotes, move left. */
                   1192:                        if (pairs)
                   1193:                                cp[-pairs] = cp[0];
                   1194:                        /* Pairs of quotes do not end words, ... */
                   1195:                        if ('"' == cp[0] && '"' == cp[1]) {
                   1196:                                pairs++;
                   1197:                                cp++;
                   1198:                                continue;
                   1199:                        }
                   1200:                        /* ... but solitary quotes do. */
                   1201:                        if ('"' != *cp)
                   1202:                                continue;
                   1203:                        if (pairs)
                   1204:                                cp[-pairs] = '\0';
                   1205:                        *cp = ' ';
                   1206:                        break;
                   1207:                }
                   1208:                /* Last argument; the remaining ones are empty strings. */
                   1209:                if ('\0' == *cp)
                   1210:                        continue;
                   1211:                /* Null-terminate argument and move to the next one. */
                   1212:                *cp++ = '\0';
                   1213:                while (' ' == *cp)
                   1214:                        cp++;
                   1215:        }
1.99      kristaps 1216:
1.106     kristaps 1217:        /*
                   1218:         * Expand macro arguments.
1.99      kristaps 1219:         */
1.106     kristaps 1220:        *szp = 0;
                   1221:        n1 = cp = mandoc_strdup(r->current_string);
                   1222:        while (NULL != (cp = strstr(cp, "\\$"))) {
                   1223:                i = cp[2] - '1';
                   1224:                if (0 > i || 8 < i) {
                   1225:                        /* Not an argument invocation. */
                   1226:                        cp += 2;
                   1227:                        continue;
                   1228:                }
                   1229:
                   1230:                *szp = strlen(n1) - 3 + strlen(arg[i]) + 1;
                   1231:                n2 = mandoc_malloc(*szp);
                   1232:
                   1233:                strlcpy(n2, n1, (size_t)(cp - n1 + 1));
                   1234:                strlcat(n2, arg[i], *szp);
                   1235:                strlcat(n2, cp + 3, *szp);
                   1236:
                   1237:                cp = n2 + (cp - n1);
                   1238:                free(n1);
                   1239:                n1 = n2;
1.99      kristaps 1240:        }
                   1241:
1.106     kristaps 1242:        /*
                   1243:         * Replace the macro invocation
                   1244:         * by the expanded macro.
                   1245:         */
                   1246:        free(*bufp);
                   1247:        *bufp = n1;
                   1248:        if (0 == *szp)
                   1249:                *szp = strlen(*bufp) + 1;
                   1250:
                   1251:        return(*szp > 1 && '\n' == (*bufp)[(int)*szp - 2] ?
                   1252:           ROFF_REPARSE : ROFF_APPEND);
1.99      kristaps 1253: }
                   1254:
1.106     kristaps 1255: /*
                   1256:  * Store *string into the user-defined string called *name.
                   1257:  * In multiline mode, append to an existing entry and append '\n';
                   1258:  * else replace the existing entry, if there is one.
                   1259:  * To clear an existing entry, call with (*r, *name, NULL, 0).
                   1260:  */
1.94      kristaps 1261: static void
1.106     kristaps 1262: roff_setstr(struct roff *r, const char *name, const char *string,
                   1263:        int multiline)
1.92      schwarze 1264: {
                   1265:        struct roffstr   *n;
1.106     kristaps 1266:        char             *c;
                   1267:        size_t            oldch, newch;
1.92      schwarze 1268:
1.106     kristaps 1269:        /* Search for an existing string with the same name. */
1.94      kristaps 1270:        n = r->first_string;
1.92      schwarze 1271:        while (n && strcmp(name, n->name))
                   1272:                n = n->next;
1.94      kristaps 1273:
                   1274:        if (NULL == n) {
1.106     kristaps 1275:                /* Create a new string table entry. */
1.94      kristaps 1276:                n = mandoc_malloc(sizeof(struct roffstr));
1.106     kristaps 1277:                n->name = mandoc_strdup(name);
                   1278:                n->string = NULL;
1.94      kristaps 1279:                n->next = r->first_string;
                   1280:                r->first_string = n;
1.106     kristaps 1281:        } else if (0 == multiline) {
                   1282:                /* In multiline mode, append; else replace. */
1.92      schwarze 1283:                free(n->string);
1.106     kristaps 1284:                n->string = NULL;
                   1285:        }
                   1286:
                   1287:        if (NULL == string)
                   1288:                return;
                   1289:
                   1290:        /*
                   1291:         * One additional byte for the '\n' in multiline mode,
                   1292:         * and one for the terminating '\0'.
                   1293:         */
                   1294:        newch = strlen(string) + (multiline ? 2 : 1);
                   1295:        if (NULL == n->string) {
                   1296:                n->string = mandoc_malloc(newch);
                   1297:                *n->string = '\0';
                   1298:                oldch = 0;
                   1299:        } else {
                   1300:                oldch = strlen(n->string);
                   1301:                n->string = mandoc_realloc(n->string, oldch + newch);
                   1302:        }
                   1303:
                   1304:        /* Skip existing content in the destination buffer. */
                   1305:        c = n->string + oldch;
                   1306:
                   1307:        /* Append new content to the destination buffer. */
                   1308:        while (*string) {
                   1309:                /*
                   1310:                 * Rudimentary roff copy mode:
                   1311:                 * Handle escaped backslashes.
                   1312:                 */
                   1313:                if ('\\' == *string && '\\' == *(string + 1))
                   1314:                        string++;
                   1315:                *c++ = *string++;
                   1316:        }
1.94      kristaps 1317:
1.106     kristaps 1318:        /* Append terminating bytes. */
                   1319:        if (multiline)
                   1320:                *c++ = '\n';
                   1321:        *c = '\0';
1.92      schwarze 1322: }
                   1323:
                   1324:
1.94      kristaps 1325: static const char *
                   1326: roff_getstrn(const struct roff *r, const char *name, size_t len)
1.92      schwarze 1327: {
1.94      kristaps 1328:        const struct roffstr *n;
1.92      schwarze 1329:
1.94      kristaps 1330:        n = r->first_string;
1.97      kristaps 1331:        while (n && (strncmp(name, n->name, len) || '\0' != n->name[(int)len]))
1.92      schwarze 1332:                n = n->next;
1.94      kristaps 1333:
                   1334:        return(n ? n->string : NULL);
1.92      schwarze 1335: }
                   1336:
1.94      kristaps 1337:
                   1338: static void
                   1339: roff_freestr(struct roff *r)
1.92      schwarze 1340: {
                   1341:        struct roffstr   *n, *nn;
                   1342:
1.94      kristaps 1343:        for (n = r->first_string; n; n = nn) {
1.92      schwarze 1344:                free(n->name);
                   1345:                free(n->string);
                   1346:                nn = n->next;
                   1347:                free(n);
                   1348:        }
1.94      kristaps 1349:
                   1350:        r->first_string = NULL;
1.74      kristaps 1351: }

CVSweb