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

Annotation of mandoc/roff.c, Revision 1.143

1.143   ! kristaps    1: /*     $Id: roff.c,v 1.142 2011/05/26 11:58:25 kristaps Exp $ */
1.1       kristaps    2: /*
1.119     schwarze    3:  * Copyright (c) 2010, 2011 Kristaps Dzonsons <kristaps@bsd.lv>
                      4:  * Copyright (c) 2010, 2011 Ingo Schwarze <schwarze@openbsd.org>
1.1       kristaps    5:  *
                      6:  * Permission to use, copy, modify, and distribute this software for any
1.66      kristaps    7:  * purpose with or without fee is hereby granted, provided that the above
                      8:  * copyright notice and this permission notice appear in all copies.
1.1       kristaps    9:  *
1.106     kristaps   10:  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHORS DISCLAIM ALL WARRANTIES
1.66      kristaps   11:  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
1.106     kristaps   12:  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHORS BE LIABLE FOR
1.66      kristaps   13:  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
                     14:  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
                     15:  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
                     16:  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
1.1       kristaps   17:  */
1.66      kristaps   18: #ifdef HAVE_CONFIG_H
                     19: #include "config.h"
                     20: #endif
1.30      kristaps   21:
1.67      kristaps   22: #include <assert.h>
1.85      kristaps   23: #include <ctype.h>
1.1       kristaps   24: #include <stdlib.h>
1.67      kristaps   25: #include <string.h>
1.1       kristaps   26:
1.67      kristaps   27: #include "mandoc.h"
1.109     kristaps   28: #include "libroff.h"
1.94      kristaps   29: #include "libmandoc.h"
1.33      kristaps   30:
1.141     kristaps   31: /* Maximum number of nested if-else conditionals. */
1.82      kristaps   32: #define        RSTACK_MAX      128
                     33:
1.67      kristaps   34: enum   rofft {
1.103     kristaps   35:        ROFF_ad,
1.80      kristaps   36:        ROFF_am,
                     37:        ROFF_ami,
                     38:        ROFF_am1,
                     39:        ROFF_de,
                     40:        ROFF_dei,
                     41:        ROFF_de1,
1.83      schwarze   42:        ROFF_ds,
1.82      kristaps   43:        ROFF_el,
1.103     kristaps   44:        ROFF_hy,
1.82      kristaps   45:        ROFF_ie,
1.75      kristaps   46:        ROFF_if,
1.76      kristaps   47:        ROFF_ig,
1.123     schwarze   48:        ROFF_it,
1.103     kristaps   49:        ROFF_ne,
                     50:        ROFF_nh,
1.104     kristaps   51:        ROFF_nr,
1.124     schwarze   52:        ROFF_ns,
                     53:        ROFF_ps,
1.83      schwarze   54:        ROFF_rm,
1.105     kristaps   55:        ROFF_so,
1.124     schwarze   56:        ROFF_ta,
1.83      schwarze   57:        ROFF_tr,
1.109     kristaps   58:        ROFF_TS,
                     59:        ROFF_TE,
1.112     kristaps   60:        ROFF_T_,
1.125     kristaps   61:        ROFF_EQ,
                     62:        ROFF_EN,
1.76      kristaps   63:        ROFF_cblock,
1.141     kristaps   64:        ROFF_ccond,
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 {
1.128     kristaps   81:        struct mparse   *parse; /* parse point */
1.67      kristaps   82:        struct roffnode *last; /* leaf of stack */
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 */
1.118     kristaps   88:        struct tbl_node *first_tbl; /* first table parsed */
                     89:        struct tbl_node *last_tbl; /* last table parsed */
                     90:        struct tbl_node *tbl; /* current table being parsed */
1.125     kristaps   91:        struct eqn_node *last_eqn; /* last equation parsed */
                     92:        struct eqn_node *first_eqn; /* first equation parsed */
                     93:        struct eqn_node *eqn; /* current equation being parsed */
1.79      kristaps   94: };
                     95:
1.67      kristaps   96: struct roffnode {
                     97:        enum rofft       tok; /* type of node */
                     98:        struct roffnode *parent; /* up one in stack */
                     99:        int              line; /* parse line */
                    100:        int              col; /* parse col */
1.106     kristaps  101:        char            *name; /* node name, e.g. macro name */
1.79      kristaps  102:        char            *end; /* end-rules: custom token */
                    103:        int              endspan; /* end-rules: next-line or infty */
1.82      kristaps  104:        enum roffrule    rule; /* current evaluation rule */
1.67      kristaps  105: };
                    106:
                    107: #define        ROFF_ARGS        struct roff *r, /* parse ctx */ \
1.72      kristaps  108:                         enum rofft tok, /* tok of macro */ \
1.67      kristaps  109:                         char **bufp, /* input buffer */ \
                    110:                         size_t *szp, /* size of input buffer */ \
                    111:                         int ln, /* parse line */ \
1.75      kristaps  112:                         int ppos, /* original pos in buffer */ \
                    113:                         int pos, /* current pos in buffer */ \
1.74      kristaps  114:                         int *offs /* reset offset of buffer data */
1.67      kristaps  115:
                    116: typedef        enum rofferr (*roffproc)(ROFF_ARGS);
                    117:
                    118: struct roffmac {
                    119:        const char      *name; /* macro name */
1.79      kristaps  120:        roffproc         proc; /* process new macro */
                    121:        roffproc         text; /* process as child text of macro */
                    122:        roffproc         sub; /* process as child of macro */
                    123:        int              flags;
                    124: #define        ROFFMAC_STRUCT  (1 << 0) /* always interpret */
1.85      kristaps  125:        struct roffmac  *next;
1.67      kristaps  126: };
                    127:
1.141     kristaps  128: struct predef {
                    129:        const char      *name; /* predefined input name */
                    130:        const char      *str; /* replacement symbol */
                    131: };
                    132:
                    133: #define        PREDEF(__name, __str) \
                    134:        { (__name), (__str) },
                    135:
1.80      kristaps  136: static enum rofferr     roff_block(ROFF_ARGS);
                    137: static enum rofferr     roff_block_text(ROFF_ARGS);
                    138: static enum rofferr     roff_block_sub(ROFF_ARGS);
                    139: static enum rofferr     roff_cblock(ROFF_ARGS);
                    140: static enum rofferr     roff_ccond(ROFF_ARGS);
1.82      kristaps  141: static enum rofferr     roff_cond(ROFF_ARGS);
                    142: static enum rofferr     roff_cond_text(ROFF_ARGS);
                    143: static enum rofferr     roff_cond_sub(ROFF_ARGS);
1.92      schwarze  144: static enum rofferr     roff_ds(ROFF_ARGS);
1.94      kristaps  145: static enum roffrule    roff_evalcond(const char *, int *);
                    146: static void             roff_freestr(struct roff *);
1.121     schwarze  147: static char            *roff_getname(struct roff *, char **, int, int);
1.94      kristaps  148: static const char      *roff_getstrn(const struct roff *,
                    149:                                const char *, size_t);
1.103     kristaps  150: static enum rofferr     roff_line_ignore(ROFF_ARGS);
1.89      kristaps  151: static enum rofferr     roff_nr(ROFF_ARGS);
1.95      kristaps  152: static int              roff_res(struct roff *,
1.142     kristaps  153:                                char **, size_t *, int, int);
1.122     schwarze  154: static enum rofferr     roff_rm(ROFF_ARGS);
1.94      kristaps  155: static void             roff_setstr(struct roff *,
1.106     kristaps  156:                                const char *, const char *, int);
1.105     kristaps  157: static enum rofferr     roff_so(ROFF_ARGS);
1.109     kristaps  158: static enum rofferr     roff_TE(ROFF_ARGS);
                    159: static enum rofferr     roff_TS(ROFF_ARGS);
1.125     kristaps  160: static enum rofferr     roff_EQ(ROFF_ARGS);
                    161: static enum rofferr     roff_EN(ROFF_ARGS);
1.112     kristaps  162: static enum rofferr     roff_T_(ROFF_ARGS);
1.106     kristaps  163: static enum rofferr     roff_userdef(ROFF_ARGS);
1.67      kristaps  164:
1.85      kristaps  165: /* See roff_hash_find() */
                    166:
                    167: #define        ASCII_HI         126
                    168: #define        ASCII_LO         33
                    169: #define        HASHWIDTH       (ASCII_HI - ASCII_LO + 1)
                    170:
                    171: static struct roffmac  *hash[HASHWIDTH];
                    172:
                    173: static struct roffmac   roffs[ROFF_MAX] = {
1.103     kristaps  174:        { "ad", roff_line_ignore, NULL, NULL, 0, NULL },
1.85      kristaps  175:        { "am", roff_block, roff_block_text, roff_block_sub, 0, NULL },
                    176:        { "ami", roff_block, roff_block_text, roff_block_sub, 0, NULL },
                    177:        { "am1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
                    178:        { "de", roff_block, roff_block_text, roff_block_sub, 0, NULL },
                    179:        { "dei", roff_block, roff_block_text, roff_block_sub, 0, NULL },
                    180:        { "de1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.92      schwarze  181:        { "ds", roff_ds, NULL, NULL, 0, NULL },
1.85      kristaps  182:        { "el", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
1.103     kristaps  183:        { "hy", roff_line_ignore, NULL, NULL, 0, NULL },
1.85      kristaps  184:        { "ie", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
                    185:        { "if", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
                    186:        { "ig", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.123     schwarze  187:        { "it", roff_line_ignore, NULL, NULL, 0, NULL },
1.103     kristaps  188:        { "ne", roff_line_ignore, NULL, NULL, 0, NULL },
                    189:        { "nh", roff_line_ignore, NULL, NULL, 0, NULL },
1.104     kristaps  190:        { "nr", roff_nr, NULL, NULL, 0, NULL },
1.124     schwarze  191:        { "ns", roff_line_ignore, NULL, NULL, 0, NULL },
                    192:        { "ps", roff_line_ignore, NULL, NULL, 0, NULL },
1.122     schwarze  193:        { "rm", roff_rm, NULL, NULL, 0, NULL },
1.105     kristaps  194:        { "so", roff_so, NULL, NULL, 0, NULL },
1.124     schwarze  195:        { "ta", roff_line_ignore, NULL, NULL, 0, NULL },
1.103     kristaps  196:        { "tr", roff_line_ignore, NULL, NULL, 0, NULL },
1.109     kristaps  197:        { "TS", roff_TS, NULL, NULL, 0, NULL },
                    198:        { "TE", roff_TE, NULL, NULL, 0, NULL },
1.112     kristaps  199:        { "T&", roff_T_, NULL, NULL, 0, NULL },
1.125     kristaps  200:        { "EQ", roff_EQ, NULL, NULL, 0, NULL },
                    201:        { "EN", roff_EN, NULL, NULL, 0, NULL },
1.85      kristaps  202:        { ".", roff_cblock, NULL, NULL, 0, NULL },
                    203:        { "\\}", roff_ccond, NULL, NULL, 0, NULL },
1.106     kristaps  204:        { NULL, roff_userdef, NULL, NULL, 0, NULL },
1.67      kristaps  205: };
                    206:
1.141     kristaps  207: /* Array of injected predefined strings. */
                    208: #define        PREDEFS_MAX      38
                    209: static const struct predef predefs[PREDEFS_MAX] = {
                    210: #include "predefs.in"
                    211: };
                    212:
1.67      kristaps  213: static void             roff_free1(struct roff *);
1.106     kristaps  214: static enum rofft       roff_hash_find(const char *, size_t);
1.85      kristaps  215: static void             roff_hash_init(void);
1.76      kristaps  216: static void             roffnode_cleanscope(struct roff *);
1.106     kristaps  217: static void             roffnode_push(struct roff *, enum rofft,
                    218:                                const char *, int, int);
1.67      kristaps  219: static void             roffnode_pop(struct roff *);
1.106     kristaps  220: static enum rofft       roff_parse(struct roff *, const char *, int *);
1.67      kristaps  221:
1.85      kristaps  222: /* See roff_hash_find() */
                    223: #define        ROFF_HASH(p)    (p[0] - ASCII_LO)
                    224:
                    225: static void
                    226: roff_hash_init(void)
                    227: {
                    228:        struct roffmac   *n;
                    229:        int               buc, i;
                    230:
1.106     kristaps  231:        for (i = 0; i < (int)ROFF_USERDEF; i++) {
1.85      kristaps  232:                assert(roffs[i].name[0] >= ASCII_LO);
                    233:                assert(roffs[i].name[0] <= ASCII_HI);
                    234:
                    235:                buc = ROFF_HASH(roffs[i].name);
                    236:
                    237:                if (NULL != (n = hash[buc])) {
                    238:                        for ( ; n->next; n = n->next)
                    239:                                /* Do nothing. */ ;
                    240:                        n->next = &roffs[i];
                    241:                } else
                    242:                        hash[buc] = &roffs[i];
                    243:        }
                    244: }
                    245:
1.67      kristaps  246: /*
                    247:  * Look up a roff token by its name.  Returns ROFF_MAX if no macro by
                    248:  * the nil-terminated string name could be found.
                    249:  */
                    250: static enum rofft
1.106     kristaps  251: roff_hash_find(const char *p, size_t s)
1.67      kristaps  252: {
1.85      kristaps  253:        int              buc;
                    254:        struct roffmac  *n;
1.67      kristaps  255:
1.85      kristaps  256:        /*
                    257:         * libroff has an extremely simple hashtable, for the time
                    258:         * being, which simply keys on the first character, which must
                    259:         * be printable, then walks a chain.  It works well enough until
                    260:         * optimised.
                    261:         */
                    262:
                    263:        if (p[0] < ASCII_LO || p[0] > ASCII_HI)
                    264:                return(ROFF_MAX);
                    265:
                    266:        buc = ROFF_HASH(p);
                    267:
                    268:        if (NULL == (n = hash[buc]))
                    269:                return(ROFF_MAX);
                    270:        for ( ; n; n = n->next)
1.106     kristaps  271:                if (0 == strncmp(n->name, p, s) && '\0' == n->name[(int)s])
1.85      kristaps  272:                        return((enum rofft)(n - roffs));
1.67      kristaps  273:
                    274:        return(ROFF_MAX);
                    275: }
                    276:
                    277:
                    278: /*
                    279:  * Pop the current node off of the stack of roff instructions currently
                    280:  * pending.
                    281:  */
                    282: static void
                    283: roffnode_pop(struct roff *r)
                    284: {
                    285:        struct roffnode *p;
                    286:
1.75      kristaps  287:        assert(r->last);
                    288:        p = r->last;
1.82      kristaps  289:
1.75      kristaps  290:        r->last = r->last->parent;
1.106     kristaps  291:        free(p->name);
                    292:        free(p->end);
1.67      kristaps  293:        free(p);
                    294: }
                    295:
                    296:
                    297: /*
                    298:  * Push a roff node onto the instruction stack.  This must later be
                    299:  * removed with roffnode_pop().
                    300:  */
1.98      schwarze  301: static void
1.106     kristaps  302: roffnode_push(struct roff *r, enum rofft tok, const char *name,
                    303:                int line, int col)
1.67      kristaps  304: {
                    305:        struct roffnode *p;
                    306:
1.98      schwarze  307:        p = mandoc_calloc(1, sizeof(struct roffnode));
1.67      kristaps  308:        p->tok = tok;
1.106     kristaps  309:        if (name)
                    310:                p->name = mandoc_strdup(name);
1.67      kristaps  311:        p->parent = r->last;
                    312:        p->line = line;
                    313:        p->col = col;
1.79      kristaps  314:        p->rule = p->parent ? p->parent->rule : ROFFRULE_DENY;
1.67      kristaps  315:
                    316:        r->last = p;
                    317: }
                    318:
                    319:
                    320: static void
                    321: roff_free1(struct roff *r)
                    322: {
1.118     kristaps  323:        struct tbl_node *t;
1.125     kristaps  324:        struct eqn_node *e;
1.67      kristaps  325:
1.125     kristaps  326:        while (NULL != (t = r->first_tbl)) {
1.113     kristaps  327:                r->first_tbl = t->next;
                    328:                tbl_free(t);
1.109     kristaps  329:        }
                    330:
1.113     kristaps  331:        r->first_tbl = r->last_tbl = r->tbl = NULL;
                    332:
1.125     kristaps  333:        while (NULL != (e = r->first_eqn)) {
                    334:                r->first_eqn = e->next;
                    335:                eqn_free(e);
                    336:        }
                    337:
                    338:        r->first_eqn = r->last_eqn = r->eqn = NULL;
                    339:
1.67      kristaps  340:        while (r->last)
                    341:                roffnode_pop(r);
1.109     kristaps  342:
1.94      kristaps  343:        roff_freestr(r);
1.67      kristaps  344: }
                    345:
                    346:
                    347: void
                    348: roff_reset(struct roff *r)
                    349: {
1.143   ! kristaps  350:        int              i;
1.67      kristaps  351:
                    352:        roff_free1(r);
1.143   ! kristaps  353:
        !           354:        for (i = 0; i < PREDEFS_MAX; i++)
        !           355:                roff_setstr(r, predefs[i].name, predefs[i].str, 0);
1.67      kristaps  356: }
                    357:
                    358:
                    359: void
                    360: roff_free(struct roff *r)
                    361: {
                    362:
                    363:        roff_free1(r);
                    364:        free(r);
                    365: }
                    366:
                    367:
                    368: struct roff *
1.128     kristaps  369: roff_alloc(struct regset *regs, struct mparse *parse)
1.67      kristaps  370: {
                    371:        struct roff     *r;
1.141     kristaps  372:        int              i;
1.67      kristaps  373:
1.98      schwarze  374:        r = mandoc_calloc(1, sizeof(struct roff));
1.90      kristaps  375:        r->regs = regs;
1.128     kristaps  376:        r->parse = parse;
1.82      kristaps  377:        r->rstackpos = -1;
1.85      kristaps  378:
                    379:        roff_hash_init();
1.141     kristaps  380:
                    381:        for (i = 0; i < PREDEFS_MAX; i++)
                    382:                roff_setstr(r, predefs[i].name, predefs[i].str, 0);
                    383:
1.67      kristaps  384:        return(r);
                    385: }
                    386:
                    387:
1.94      kristaps  388: /*
                    389:  * Pre-filter each and every line for reserved words (one beginning with
                    390:  * `\*', e.g., `\*(ab').  These must be handled before the actual line
                    391:  * is processed.
                    392:  */
                    393: static int
1.142     kristaps  394: roff_res(struct roff *r, char **bufp, size_t *szp, int ln, int pos)
1.94      kristaps  395: {
1.108     schwarze  396:        const char      *stesc; /* start of an escape sequence ('\\') */
                    397:        const char      *stnam; /* start of the name, after "[(*" */
                    398:        const char      *cp;    /* end of the name, e.g. before ']' */
                    399:        const char      *res;   /* the string to be substituted */
1.94      kristaps  400:        int              i, maxl;
                    401:        size_t           nsz;
                    402:        char            *n;
                    403:
1.108     schwarze  404:        /* Search for a leading backslash and save a pointer to it. */
                    405:
                    406:        cp = *bufp + pos;
                    407:        while (NULL != (cp = strchr(cp, '\\'))) {
                    408:                stesc = cp++;
                    409:
                    410:                /*
                    411:                 * The second character must be an asterisk.
                    412:                 * If it isn't, skip it anyway:  It is escaped,
                    413:                 * so it can't start another escape sequence.
                    414:                 */
                    415:
                    416:                if ('\0' == *cp)
                    417:                        return(1);
                    418:                if ('*' != *cp++)
                    419:                        continue;
                    420:
                    421:                /*
                    422:                 * The third character decides the length
                    423:                 * of the name of the string.
                    424:                 * Save a pointer to the name.
                    425:                 */
                    426:
1.94      kristaps  427:                switch (*cp) {
1.108     schwarze  428:                case ('\0'):
                    429:                        return(1);
1.94      kristaps  430:                case ('('):
                    431:                        cp++;
                    432:                        maxl = 2;
                    433:                        break;
                    434:                case ('['):
                    435:                        cp++;
                    436:                        maxl = 0;
                    437:                        break;
                    438:                default:
                    439:                        maxl = 1;
                    440:                        break;
                    441:                }
1.108     schwarze  442:                stnam = cp;
1.94      kristaps  443:
1.108     schwarze  444:                /* Advance to the end of the name. */
1.94      kristaps  445:
                    446:                for (i = 0; 0 == maxl || i < maxl; i++, cp++) {
                    447:                        if ('\0' == *cp)
                    448:                                return(1); /* Error. */
                    449:                        if (0 == maxl && ']' == *cp)
                    450:                                break;
                    451:                }
                    452:
1.108     schwarze  453:                /*
                    454:                 * Retrieve the replacement string; if it is
                    455:                 * undefined, resume searching for escapes.
                    456:                 */
                    457:
                    458:                res = roff_getstrn(r, stnam, (size_t)i);
1.94      kristaps  459:
                    460:                if (NULL == res) {
1.142     kristaps  461:                        /* TODO: keep track of the correct position. */
                    462:                        mandoc_msg(MANDOCERR_BADESCAPE, r->parse, ln, pos, NULL);
                    463:                        res = "";
1.94      kristaps  464:                }
                    465:
1.108     schwarze  466:                /* Replace the escape sequence by the string. */
                    467:
1.94      kristaps  468:                nsz = *szp + strlen(res) + 1;
                    469:                n = mandoc_malloc(nsz);
                    470:
1.108     schwarze  471:                strlcpy(n, *bufp, (size_t)(stesc - *bufp + 1));
1.94      kristaps  472:                strlcat(n, res, nsz);
                    473:                strlcat(n, cp + (maxl ? 0 : 1), nsz);
                    474:
                    475:                free(*bufp);
                    476:
                    477:                *bufp = n;
                    478:                *szp = nsz;
                    479:                return(0);
                    480:        }
                    481:
                    482:        return(1);
                    483: }
                    484:
                    485:
1.67      kristaps  486: enum rofferr
1.90      kristaps  487: roff_parseln(struct roff *r, int ln, char **bufp,
                    488:                size_t *szp, int pos, int *offs)
1.67      kristaps  489: {
                    490:        enum rofft       t;
1.109     kristaps  491:        enum rofferr     e;
1.130     kristaps  492:        int              ppos, ctl;
1.79      kristaps  493:
                    494:        /*
1.94      kristaps  495:         * Run the reserved-word filter only if we have some reserved
                    496:         * words to fill in.
                    497:         */
                    498:
1.142     kristaps  499:        if (r->first_string && ! roff_res(r, bufp, szp, ln, pos))
1.106     kristaps  500:                return(ROFF_REPARSE);
1.94      kristaps  501:
1.130     kristaps  502:        ppos = pos;
                    503:        ctl = mandoc_getcontrol(*bufp, &pos);
                    504:
1.94      kristaps  505:        /*
1.79      kristaps  506:         * First, if a scope is open and we're not a macro, pass the
                    507:         * text through the macro's filter.  If a scope isn't open and
                    508:         * we're not a macro, just let it through.
1.125     kristaps  509:         * Finally, if there's an equation scope open, divert it into it
                    510:         * no matter our state.
1.79      kristaps  511:         */
1.74      kristaps  512:
1.130     kristaps  513:        if (r->last && ! ctl) {
1.78      kristaps  514:                t = r->last->tok;
                    515:                assert(roffs[t].text);
1.109     kristaps  516:                e = (*roffs[t].text)
                    517:                        (r, t, bufp, szp, ln, pos, pos, offs);
                    518:                assert(ROFF_IGN == e || ROFF_CONT == e);
1.125     kristaps  519:                if (ROFF_CONT != e)
                    520:                        return(e);
                    521:                if (r->eqn)
1.130     kristaps  522:                        return(eqn_read(&r->eqn, ln, *bufp, pos));
1.125     kristaps  523:                if (r->tbl)
1.130     kristaps  524:                        return(tbl_read(r->tbl, ln, *bufp, pos));
1.125     kristaps  525:                return(ROFF_CONT);
1.130     kristaps  526:        } else if ( ! ctl) {
1.125     kristaps  527:                if (r->eqn)
1.130     kristaps  528:                        return(eqn_read(&r->eqn, ln, *bufp, pos));
1.109     kristaps  529:                if (r->tbl)
1.130     kristaps  530:                        return(tbl_read(r->tbl, ln, *bufp, pos));
1.67      kristaps  531:                return(ROFF_CONT);
1.125     kristaps  532:        } else if (r->eqn)
1.130     kristaps  533:                return(eqn_read(&r->eqn, ln, *bufp, ppos));
1.67      kristaps  534:
1.79      kristaps  535:        /*
                    536:         * If a scope is open, go to the child handler for that macro,
                    537:         * as it may want to preprocess before doing anything with it.
1.125     kristaps  538:         * Don't do so if an equation is open.
1.79      kristaps  539:         */
1.78      kristaps  540:
1.79      kristaps  541:        if (r->last) {
                    542:                t = r->last->tok;
                    543:                assert(roffs[t].sub);
                    544:                return((*roffs[t].sub)
1.90      kristaps  545:                                (r, t, bufp, szp,
1.130     kristaps  546:                                 ln, ppos, pos, offs));
1.79      kristaps  547:        }
1.78      kristaps  548:
1.79      kristaps  549:        /*
                    550:         * Lastly, as we've no scope open, try to look up and execute
                    551:         * the new macro.  If no macro is found, simply return and let
                    552:         * the compilers handle it.
                    553:         */
1.67      kristaps  554:
1.106     kristaps  555:        if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos)))
1.79      kristaps  556:                return(ROFF_CONT);
1.67      kristaps  557:
1.75      kristaps  558:        assert(roffs[t].proc);
1.78      kristaps  559:        return((*roffs[t].proc)
1.90      kristaps  560:                        (r, t, bufp, szp,
                    561:                         ln, ppos, pos, offs));
1.74      kristaps  562: }
                    563:
                    564:
1.117     kristaps  565: void
1.74      kristaps  566: roff_endparse(struct roff *r)
                    567: {
                    568:
1.110     kristaps  569:        if (r->last)
1.128     kristaps  570:                mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.109     kristaps  571:                                r->last->line, r->last->col, NULL);
1.117     kristaps  572:
1.125     kristaps  573:        if (r->eqn) {
1.128     kristaps  574:                mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.126     kristaps  575:                                r->eqn->eqn.line, r->eqn->eqn.pos, NULL);
1.125     kristaps  576:                eqn_end(r->eqn);
                    577:                r->eqn = NULL;
                    578:        }
                    579:
1.117     kristaps  580:        if (r->tbl) {
1.128     kristaps  581:                mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.117     kristaps  582:                                r->tbl->line, r->tbl->pos, NULL);
                    583:                tbl_end(r->tbl);
                    584:                r->tbl = NULL;
                    585:        }
1.67      kristaps  586: }
                    587:
                    588: /*
                    589:  * Parse a roff node's type from the input buffer.  This must be in the
                    590:  * form of ".foo xxx" in the usual way.
                    591:  */
                    592: static enum rofft
1.106     kristaps  593: roff_parse(struct roff *r, const char *buf, int *pos)
1.67      kristaps  594: {
1.106     kristaps  595:        const char      *mac;
                    596:        size_t           maclen;
1.67      kristaps  597:        enum rofft       t;
                    598:
1.130     kristaps  599:        if ('\0' == buf[*pos] || '"' == buf[*pos])
1.67      kristaps  600:                return(ROFF_MAX);
                    601:
1.106     kristaps  602:        mac = buf + *pos;
                    603:        maclen = strcspn(mac, " \\\t\0");
1.67      kristaps  604:
1.106     kristaps  605:        t = (r->current_string = roff_getstrn(r, mac, maclen))
                    606:            ? ROFF_USERDEF : roff_hash_find(mac, maclen);
1.67      kristaps  607:
1.127     kristaps  608:        *pos += (int)maclen;
1.130     kristaps  609:
1.67      kristaps  610:        while (buf[*pos] && ' ' == buf[*pos])
                    611:                (*pos)++;
                    612:
                    613:        return(t);
                    614: }
                    615:
                    616: /* ARGSUSED */
                    617: static enum rofferr
1.76      kristaps  618: roff_cblock(ROFF_ARGS)
1.67      kristaps  619: {
                    620:
1.79      kristaps  621:        /*
                    622:         * A block-close `..' should only be invoked as a child of an
                    623:         * ignore macro, otherwise raise a warning and just ignore it.
                    624:         */
                    625:
1.76      kristaps  626:        if (NULL == r->last) {
1.128     kristaps  627:                mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.76      kristaps  628:                return(ROFF_IGN);
                    629:        }
1.67      kristaps  630:
1.81      kristaps  631:        switch (r->last->tok) {
                    632:        case (ROFF_am):
                    633:                /* FALLTHROUGH */
                    634:        case (ROFF_ami):
                    635:                /* FALLTHROUGH */
                    636:        case (ROFF_am1):
                    637:                /* FALLTHROUGH */
                    638:        case (ROFF_de):
1.108     schwarze  639:                /* ROFF_de1 is remapped to ROFF_de in roff_block(). */
1.81      kristaps  640:                /* FALLTHROUGH */
                    641:        case (ROFF_dei):
                    642:                /* FALLTHROUGH */
                    643:        case (ROFF_ig):
                    644:                break;
                    645:        default:
1.128     kristaps  646:                mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.67      kristaps  647:                return(ROFF_IGN);
1.76      kristaps  648:        }
1.67      kristaps  649:
1.76      kristaps  650:        if ((*bufp)[pos])
1.128     kristaps  651:                mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.71      kristaps  652:
                    653:        roffnode_pop(r);
1.76      kristaps  654:        roffnode_cleanscope(r);
                    655:        return(ROFF_IGN);
1.71      kristaps  656:
1.67      kristaps  657: }
                    658:
                    659:
1.76      kristaps  660: static void
                    661: roffnode_cleanscope(struct roff *r)
1.67      kristaps  662: {
                    663:
1.76      kristaps  664:        while (r->last) {
                    665:                if (--r->last->endspan < 0)
                    666:                        break;
                    667:                roffnode_pop(r);
                    668:        }
1.67      kristaps  669: }
                    670:
                    671:
1.75      kristaps  672: /* ARGSUSED */
1.74      kristaps  673: static enum rofferr
1.75      kristaps  674: roff_ccond(ROFF_ARGS)
1.74      kristaps  675: {
                    676:
1.76      kristaps  677:        if (NULL == r->last) {
1.128     kristaps  678:                mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.76      kristaps  679:                return(ROFF_IGN);
                    680:        }
                    681:
1.82      kristaps  682:        switch (r->last->tok) {
                    683:        case (ROFF_el):
                    684:                /* FALLTHROUGH */
                    685:        case (ROFF_ie):
                    686:                /* FALLTHROUGH */
                    687:        case (ROFF_if):
                    688:                break;
                    689:        default:
1.128     kristaps  690:                mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.75      kristaps  691:                return(ROFF_IGN);
                    692:        }
                    693:
1.76      kristaps  694:        if (r->last->endspan > -1) {
1.128     kristaps  695:                mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.76      kristaps  696:                return(ROFF_IGN);
                    697:        }
                    698:
                    699:        if ((*bufp)[pos])
1.128     kristaps  700:                mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.76      kristaps  701:
1.75      kristaps  702:        roffnode_pop(r);
1.76      kristaps  703:        roffnode_cleanscope(r);
                    704:        return(ROFF_IGN);
                    705: }
                    706:
1.75      kristaps  707:
1.76      kristaps  708: /* ARGSUSED */
                    709: static enum rofferr
1.80      kristaps  710: roff_block(ROFF_ARGS)
1.76      kristaps  711: {
1.78      kristaps  712:        int             sv;
                    713:        size_t          sz;
1.106     kristaps  714:        char            *name;
                    715:
                    716:        name = NULL;
1.76      kristaps  717:
1.106     kristaps  718:        if (ROFF_ig != tok) {
                    719:                if ('\0' == (*bufp)[pos]) {
1.128     kristaps  720:                        mandoc_msg(MANDOCERR_NOARGS, r->parse, ln, ppos, NULL);
1.106     kristaps  721:                        return(ROFF_IGN);
                    722:                }
1.107     kristaps  723:
                    724:                /*
                    725:                 * Re-write `de1', since we don't really care about
                    726:                 * groff's strange compatibility mode, into `de'.
                    727:                 */
                    728:
1.106     kristaps  729:                if (ROFF_de1 == tok)
                    730:                        tok = ROFF_de;
                    731:                if (ROFF_de == tok)
                    732:                        name = *bufp + pos;
                    733:                else
1.128     kristaps  734:                        mandoc_msg(MANDOCERR_REQUEST, r->parse, ln, ppos,
1.106     kristaps  735:                            roffs[tok].name);
1.107     kristaps  736:
1.131     schwarze  737:                while ((*bufp)[pos] && ! isspace((unsigned char)(*bufp)[pos]))
1.80      kristaps  738:                        pos++;
1.107     kristaps  739:
1.131     schwarze  740:                while (isspace((unsigned char)(*bufp)[pos]))
1.106     kristaps  741:                        (*bufp)[pos++] = '\0';
1.80      kristaps  742:        }
                    743:
1.106     kristaps  744:        roffnode_push(r, tok, name, ln, ppos);
                    745:
                    746:        /*
                    747:         * At the beginning of a `de' macro, clear the existing string
                    748:         * with the same name, if there is one.  New content will be
                    749:         * added from roff_block_text() in multiline mode.
                    750:         */
1.107     kristaps  751:
1.106     kristaps  752:        if (ROFF_de == tok)
1.108     schwarze  753:                roff_setstr(r, name, "", 0);
1.76      kristaps  754:
1.79      kristaps  755:        if ('\0' == (*bufp)[pos])
1.78      kristaps  756:                return(ROFF_IGN);
                    757:
1.107     kristaps  758:        /* If present, process the custom end-of-line marker. */
                    759:
1.78      kristaps  760:        sv = pos;
1.131     schwarze  761:        while ((*bufp)[pos] && ! isspace((unsigned char)(*bufp)[pos]))
1.78      kristaps  762:                pos++;
                    763:
                    764:        /*
                    765:         * Note: groff does NOT like escape characters in the input.
                    766:         * Instead of detecting this, we're just going to let it fly and
                    767:         * to hell with it.
                    768:         */
                    769:
                    770:        assert(pos > sv);
                    771:        sz = (size_t)(pos - sv);
                    772:
1.79      kristaps  773:        if (1 == sz && '.' == (*bufp)[sv])
                    774:                return(ROFF_IGN);
                    775:
1.98      schwarze  776:        r->last->end = mandoc_malloc(sz + 1);
1.78      kristaps  777:
                    778:        memcpy(r->last->end, *bufp + sv, sz);
                    779:        r->last->end[(int)sz] = '\0';
                    780:
1.77      kristaps  781:        if ((*bufp)[pos])
1.128     kristaps  782:                mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.74      kristaps  783:
1.78      kristaps  784:        return(ROFF_IGN);
                    785: }
                    786:
                    787:
                    788: /* ARGSUSED */
                    789: static enum rofferr
1.80      kristaps  790: roff_block_sub(ROFF_ARGS)
1.79      kristaps  791: {
                    792:        enum rofft      t;
                    793:        int             i, j;
                    794:
                    795:        /*
                    796:         * First check whether a custom macro exists at this level.  If
                    797:         * it does, then check against it.  This is some of groff's
                    798:         * stranger behaviours.  If we encountered a custom end-scope
                    799:         * tag and that tag also happens to be a "real" macro, then we
                    800:         * need to try interpreting it again as a real macro.  If it's
                    801:         * not, then return ignore.  Else continue.
                    802:         */
                    803:
                    804:        if (r->last->end) {
1.130     kristaps  805:                for (i = pos, j = 0; r->last->end[j]; j++, i++)
1.79      kristaps  806:                        if ((*bufp)[i] != r->last->end[j])
                    807:                                break;
                    808:
                    809:                if ('\0' == r->last->end[j] &&
                    810:                                ('\0' == (*bufp)[i] ||
                    811:                                 ' ' == (*bufp)[i] ||
                    812:                                 '\t' == (*bufp)[i])) {
                    813:                        roffnode_pop(r);
                    814:                        roffnode_cleanscope(r);
                    815:
1.130     kristaps  816:                        while (' ' == (*bufp)[i] || '\t' == (*bufp)[i])
                    817:                                i++;
                    818:
                    819:                        pos = i;
1.106     kristaps  820:                        if (ROFF_MAX != roff_parse(r, *bufp, &pos))
1.79      kristaps  821:                                return(ROFF_RERUN);
                    822:                        return(ROFF_IGN);
                    823:                }
                    824:        }
                    825:
                    826:        /*
                    827:         * If we have no custom end-query or lookup failed, then try
                    828:         * pulling it out of the hashtable.
                    829:         */
                    830:
1.137     schwarze  831:        t = roff_parse(r, *bufp, &pos);
1.79      kristaps  832:
1.106     kristaps  833:        /*
                    834:         * Macros other than block-end are only significant
                    835:         * in `de' blocks; elsewhere, simply throw them away.
                    836:         */
                    837:        if (ROFF_cblock != t) {
                    838:                if (ROFF_de == tok)
                    839:                        roff_setstr(r, r->last->name, *bufp + ppos, 1);
1.79      kristaps  840:                return(ROFF_IGN);
1.106     kristaps  841:        }
1.79      kristaps  842:
                    843:        assert(roffs[t].proc);
1.90      kristaps  844:        return((*roffs[t].proc)(r, t, bufp, szp,
                    845:                                ln, ppos, pos, offs));
1.79      kristaps  846: }
                    847:
                    848:
                    849: /* ARGSUSED */
                    850: static enum rofferr
1.80      kristaps  851: roff_block_text(ROFF_ARGS)
1.78      kristaps  852: {
                    853:
1.106     kristaps  854:        if (ROFF_de == tok)
                    855:                roff_setstr(r, r->last->name, *bufp + pos, 1);
                    856:
1.78      kristaps  857:        return(ROFF_IGN);
                    858: }
                    859:
                    860:
                    861: /* ARGSUSED */
                    862: static enum rofferr
1.82      kristaps  863: roff_cond_sub(ROFF_ARGS)
                    864: {
                    865:        enum rofft       t;
                    866:        enum roffrule    rr;
1.139     kristaps  867:        char            *ep;
1.82      kristaps  868:
                    869:        rr = r->last->rule;
1.139     kristaps  870:        roffnode_cleanscope(r);
1.82      kristaps  871:
1.139     kristaps  872:        /*
                    873:         * If the macro is unknown, first check if it contains a closing
                    874:         * delimiter `\}'.  If it does, close out our scope and return
                    875:         * the currently-scoped rule (ignore or continue).  Else, drop
                    876:         * into the currently-scoped rule.
1.87      kristaps  877:         */
                    878:
1.106     kristaps  879:        if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos))) {
1.139     kristaps  880:                ep = &(*bufp)[pos];
                    881:                for ( ; NULL != (ep = strchr(ep, '\\')); ep++) {
                    882:                        ep++;
                    883:                        if ('}' != *ep)
                    884:                                continue;
1.140     kristaps  885:                        *ep = '&';
1.139     kristaps  886:                        roff_ccond(r, ROFF_ccond, bufp, szp,
                    887:                                        ln, pos, pos + 2, offs);
                    888:                        break;
                    889:                }
1.82      kristaps  890:                return(ROFFRULE_DENY == rr ? ROFF_IGN : ROFF_CONT);
1.100     kristaps  891:        }
1.82      kristaps  892:
                    893:        /*
                    894:         * A denied conditional must evaluate its children if and only
                    895:         * if they're either structurally required (such as loops and
                    896:         * conditionals) or a closing macro.
                    897:         */
1.139     kristaps  898:
1.82      kristaps  899:        if (ROFFRULE_DENY == rr)
                    900:                if ( ! (ROFFMAC_STRUCT & roffs[t].flags))
                    901:                        if (ROFF_ccond != t)
                    902:                                return(ROFF_IGN);
                    903:
                    904:        assert(roffs[t].proc);
1.90      kristaps  905:        return((*roffs[t].proc)(r, t, bufp, szp,
                    906:                                ln, ppos, pos, offs));
1.82      kristaps  907: }
                    908:
                    909: /* ARGSUSED */
                    910: static enum rofferr
                    911: roff_cond_text(ROFF_ARGS)
1.78      kristaps  912: {
1.140     kristaps  913:        char            *ep;
1.82      kristaps  914:        enum roffrule    rr;
                    915:
                    916:        rr = r->last->rule;
1.140     kristaps  917:        roffnode_cleanscope(r);
1.82      kristaps  918:
1.140     kristaps  919:        ep = &(*bufp)[pos];
                    920:        for ( ; NULL != (ep = strchr(ep, '\\')); ep++) {
                    921:                ep++;
                    922:                if ('}' != *ep)
                    923:                        continue;
                    924:                *ep = '&';
                    925:                roff_ccond(r, ROFF_ccond, bufp, szp,
                    926:                                ln, pos, pos + 2, offs);
1.78      kristaps  927:        }
1.82      kristaps  928:        return(ROFFRULE_DENY == rr ? ROFF_IGN : ROFF_CONT);
1.74      kristaps  929: }
                    930:
1.88      kristaps  931: static enum roffrule
                    932: roff_evalcond(const char *v, int *pos)
                    933: {
                    934:
                    935:        switch (v[*pos]) {
                    936:        case ('n'):
                    937:                (*pos)++;
                    938:                return(ROFFRULE_ALLOW);
                    939:        case ('e'):
                    940:                /* FALLTHROUGH */
                    941:        case ('o'):
                    942:                /* FALLTHROUGH */
                    943:        case ('t'):
                    944:                (*pos)++;
                    945:                return(ROFFRULE_DENY);
                    946:        default:
                    947:                break;
                    948:        }
                    949:
                    950:        while (v[*pos] && ' ' != v[*pos])
                    951:                (*pos)++;
                    952:        return(ROFFRULE_DENY);
                    953: }
                    954:
1.75      kristaps  955: /* ARGSUSED */
1.74      kristaps  956: static enum rofferr
1.103     kristaps  957: roff_line_ignore(ROFF_ARGS)
1.89      kristaps  958: {
1.123     schwarze  959:
                    960:        if (ROFF_it == tok)
1.128     kristaps  961:                mandoc_msg(MANDOCERR_REQUEST, r->parse, ln, ppos, "it");
1.89      kristaps  962:
                    963:        return(ROFF_IGN);
                    964: }
                    965:
1.104     kristaps  966: /* ARGSUSED */
                    967: static enum rofferr
1.82      kristaps  968: roff_cond(ROFF_ARGS)
1.74      kristaps  969: {
1.77      kristaps  970:        int              sv;
1.88      kristaps  971:        enum roffrule    rule;
1.74      kristaps  972:
1.134     kristaps  973:        /*
                    974:         * An `.el' has no conditional body: it will consume the value
                    975:         * of the current rstack entry set in prior `ie' calls or
                    976:         * defaults to DENY.
                    977:         *
                    978:         * If we're not an `el', however, then evaluate the conditional.
                    979:         */
1.133     kristaps  980:
1.134     kristaps  981:        rule = ROFF_el == tok ?
                    982:                (r->rstackpos < 0 ?
                    983:                 ROFFRULE_DENY : r->rstack[r->rstackpos--]) :
                    984:                roff_evalcond(*bufp, &pos);
1.77      kristaps  985:
                    986:        sv = pos;
1.75      kristaps  987:        while (' ' == (*bufp)[pos])
                    988:                pos++;
1.74      kristaps  989:
1.77      kristaps  990:        /*
                    991:         * Roff is weird.  If we have just white-space after the
                    992:         * conditional, it's considered the BODY and we exit without
                    993:         * really doing anything.  Warn about this.  It's probably
                    994:         * wrong.
                    995:         */
1.88      kristaps  996:
1.77      kristaps  997:        if ('\0' == (*bufp)[pos] && sv != pos) {
1.128     kristaps  998:                mandoc_msg(MANDOCERR_NOARGS, r->parse, ln, ppos, NULL);
1.107     kristaps  999:                return(ROFF_IGN);
1.77      kristaps 1000:        }
                   1001:
1.106     kristaps 1002:        roffnode_push(r, tok, NULL, ln, ppos);
1.77      kristaps 1003:
1.88      kristaps 1004:        r->last->rule = rule;
                   1005:
1.134     kristaps 1006:        /*
                   1007:         * An if-else will put the NEGATION of the current evaluated
                   1008:         * conditional into the stack of rules.
                   1009:         */
                   1010:
1.84      schwarze 1011:        if (ROFF_ie == tok) {
1.134     kristaps 1012:                if (r->rstackpos == RSTACK_MAX - 1) {
                   1013:                        mandoc_msg(MANDOCERR_MEM,
                   1014:                                r->parse, ln, ppos, NULL);
                   1015:                        return(ROFF_ERR);
                   1016:                }
                   1017:                r->rstack[++r->rstackpos] =
                   1018:                        ROFFRULE_DENY == r->last->rule ?
                   1019:                        ROFFRULE_ALLOW : ROFFRULE_DENY;
1.82      kristaps 1020:        }
1.88      kristaps 1021:
                   1022:        /* If the parent has false as its rule, then so do we. */
                   1023:
1.109     kristaps 1024:        if (r->last->parent && ROFFRULE_DENY == r->last->parent->rule)
1.84      schwarze 1025:                r->last->rule = ROFFRULE_DENY;
1.88      kristaps 1026:
                   1027:        /*
                   1028:         * Determine scope.  If we're invoked with "\{" trailing the
                   1029:         * conditional, then we're in a multiline scope.  Else our scope
                   1030:         * expires on the next line.
                   1031:         */
1.74      kristaps 1032:
1.75      kristaps 1033:        r->last->endspan = 1;
                   1034:
                   1035:        if ('\\' == (*bufp)[pos] && '{' == (*bufp)[pos + 1]) {
                   1036:                r->last->endspan = -1;
                   1037:                pos += 2;
1.109     kristaps 1038:        }
1.74      kristaps 1039:
1.77      kristaps 1040:        /*
                   1041:         * If there are no arguments on the line, the next-line scope is
                   1042:         * assumed.
                   1043:         */
                   1044:
1.75      kristaps 1045:        if ('\0' == (*bufp)[pos])
                   1046:                return(ROFF_IGN);
1.77      kristaps 1047:
                   1048:        /* Otherwise re-run the roff parser after recalculating. */
1.74      kristaps 1049:
1.75      kristaps 1050:        *offs = pos;
                   1051:        return(ROFF_RERUN);
1.83      schwarze 1052: }
                   1053:
                   1054:
                   1055: /* ARGSUSED */
                   1056: static enum rofferr
1.92      schwarze 1057: roff_ds(ROFF_ARGS)
                   1058: {
1.96      kristaps 1059:        char            *name, *string;
                   1060:
                   1061:        /*
                   1062:         * A symbol is named by the first word following the macro
                   1063:         * invocation up to a space.  Its value is anything after the
                   1064:         * name's trailing whitespace and optional double-quote.  Thus,
                   1065:         *
                   1066:         *  [.ds foo "bar  "     ]
                   1067:         *
                   1068:         * will have `bar  "     ' as its value.
                   1069:         */
1.92      schwarze 1070:
1.121     schwarze 1071:        string = *bufp + pos;
                   1072:        name = roff_getname(r, &string, ln, pos);
1.92      schwarze 1073:        if ('\0' == *name)
                   1074:                return(ROFF_IGN);
                   1075:
1.121     schwarze 1076:        /* Read past initial double-quote. */
                   1077:        if ('"' == *string)
1.92      schwarze 1078:                string++;
                   1079:
1.96      kristaps 1080:        /* The rest is the value. */
1.106     kristaps 1081:        roff_setstr(r, name, string, 0);
1.92      schwarze 1082:        return(ROFF_IGN);
                   1083: }
                   1084:
                   1085:
                   1086: /* ARGSUSED */
                   1087: static enum rofferr
1.89      kristaps 1088: roff_nr(ROFF_ARGS)
1.83      schwarze 1089: {
1.121     schwarze 1090:        const char      *key;
                   1091:        char            *val;
1.138     kristaps 1092:        int              iv;
1.91      kristaps 1093:        struct reg      *rg;
1.89      kristaps 1094:
1.121     schwarze 1095:        val = *bufp + pos;
                   1096:        key = roff_getname(r, &val, ln, pos);
1.91      kristaps 1097:        rg = r->regs->regs;
1.89      kristaps 1098:
                   1099:        if (0 == strcmp(key, "nS")) {
1.91      kristaps 1100:                rg[(int)REG_nS].set = 1;
1.138     kristaps 1101:                if ((iv = mandoc_strntou(val, strlen(val), 10)) >= 0)
                   1102:                        rg[REG_nS].v.u = (unsigned)iv;
                   1103:                else
                   1104:                        rg[(int)REG_nS].v.u = 0u;
1.109     kristaps 1105:        }
                   1106:
1.122     schwarze 1107:        return(ROFF_IGN);
                   1108: }
                   1109:
                   1110: /* ARGSUSED */
                   1111: static enum rofferr
                   1112: roff_rm(ROFF_ARGS)
                   1113: {
                   1114:        const char       *name;
                   1115:        char             *cp;
                   1116:
                   1117:        cp = *bufp + pos;
                   1118:        while ('\0' != *cp) {
1.127     kristaps 1119:                name = roff_getname(r, &cp, ln, (int)(cp - *bufp));
1.122     schwarze 1120:                if ('\0' != *name)
                   1121:                        roff_setstr(r, name, NULL, 0);
                   1122:        }
1.109     kristaps 1123:        return(ROFF_IGN);
                   1124: }
                   1125:
                   1126: /* ARGSUSED */
                   1127: static enum rofferr
                   1128: roff_TE(ROFF_ARGS)
                   1129: {
                   1130:
                   1131:        if (NULL == r->tbl)
1.128     kristaps 1132:                mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.115     kristaps 1133:        else
                   1134:                tbl_end(r->tbl);
1.109     kristaps 1135:
                   1136:        r->tbl = NULL;
1.112     kristaps 1137:        return(ROFF_IGN);
                   1138: }
                   1139:
                   1140: /* ARGSUSED */
                   1141: static enum rofferr
                   1142: roff_T_(ROFF_ARGS)
                   1143: {
                   1144:
                   1145:        if (NULL == r->tbl)
1.128     kristaps 1146:                mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.112     kristaps 1147:        else
1.116     kristaps 1148:                tbl_restart(ppos, ln, r->tbl);
1.112     kristaps 1149:
1.109     kristaps 1150:        return(ROFF_IGN);
                   1151: }
                   1152:
                   1153: /* ARGSUSED */
                   1154: static enum rofferr
1.125     kristaps 1155: roff_EQ(ROFF_ARGS)
                   1156: {
                   1157:        struct eqn_node *e;
                   1158:
                   1159:        assert(NULL == r->eqn);
                   1160:        e = eqn_alloc(ppos, ln);
                   1161:
                   1162:        if (r->last_eqn)
                   1163:                r->last_eqn->next = e;
                   1164:        else
                   1165:                r->first_eqn = r->last_eqn = e;
                   1166:
                   1167:        r->eqn = r->last_eqn = e;
                   1168:        return(ROFF_IGN);
                   1169: }
                   1170:
                   1171: /* ARGSUSED */
                   1172: static enum rofferr
                   1173: roff_EN(ROFF_ARGS)
                   1174: {
                   1175:
1.128     kristaps 1176:        mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.125     kristaps 1177:        return(ROFF_IGN);
                   1178: }
                   1179:
                   1180: /* ARGSUSED */
                   1181: static enum rofferr
1.109     kristaps 1182: roff_TS(ROFF_ARGS)
                   1183: {
1.118     kristaps 1184:        struct tbl_node *t;
1.89      kristaps 1185:
1.115     kristaps 1186:        if (r->tbl) {
1.128     kristaps 1187:                mandoc_msg(MANDOCERR_SCOPEBROKEN, r->parse, ln, ppos, NULL);
1.115     kristaps 1188:                tbl_end(r->tbl);
                   1189:        }
1.83      schwarze 1190:
1.128     kristaps 1191:        t = tbl_alloc(ppos, ln, r->parse);
1.113     kristaps 1192:
                   1193:        if (r->last_tbl)
                   1194:                r->last_tbl->next = t;
                   1195:        else
                   1196:                r->first_tbl = r->last_tbl = t;
                   1197:
                   1198:        r->tbl = r->last_tbl = t;
1.83      schwarze 1199:        return(ROFF_IGN);
1.92      schwarze 1200: }
                   1201:
1.105     kristaps 1202: /* ARGSUSED */
                   1203: static enum rofferr
                   1204: roff_so(ROFF_ARGS)
                   1205: {
                   1206:        char *name;
                   1207:
1.128     kristaps 1208:        mandoc_msg(MANDOCERR_SO, r->parse, ln, ppos, NULL);
1.105     kristaps 1209:
                   1210:        /*
                   1211:         * Handle `so'.  Be EXTREMELY careful, as we shouldn't be
                   1212:         * opening anything that's not in our cwd or anything beneath
                   1213:         * it.  Thus, explicitly disallow traversing up the file-system
                   1214:         * or using absolute paths.
                   1215:         */
                   1216:
                   1217:        name = *bufp + pos;
                   1218:        if ('/' == *name || strstr(name, "../") || strstr(name, "/..")) {
1.128     kristaps 1219:                mandoc_msg(MANDOCERR_SOPATH, r->parse, ln, pos, NULL);
1.105     kristaps 1220:                return(ROFF_ERR);
                   1221:        }
                   1222:
                   1223:        *offs = pos;
                   1224:        return(ROFF_SO);
                   1225: }
1.92      schwarze 1226:
1.106     kristaps 1227: /* ARGSUSED */
                   1228: static enum rofferr
                   1229: roff_userdef(ROFF_ARGS)
1.99      kristaps 1230: {
1.106     kristaps 1231:        const char       *arg[9];
                   1232:        char             *cp, *n1, *n2;
1.119     schwarze 1233:        int               i;
1.106     kristaps 1234:
                   1235:        /*
                   1236:         * Collect pointers to macro argument strings
                   1237:         * and null-terminate them.
                   1238:         */
                   1239:        cp = *bufp + pos;
1.119     schwarze 1240:        for (i = 0; i < 9; i++)
1.120     schwarze 1241:                arg[i] = '\0' == *cp ? "" :
1.136     kristaps 1242:                    mandoc_getarg(r->parse, &cp, ln, &pos);
1.99      kristaps 1243:
1.106     kristaps 1244:        /*
                   1245:         * Expand macro arguments.
1.99      kristaps 1246:         */
1.106     kristaps 1247:        *szp = 0;
                   1248:        n1 = cp = mandoc_strdup(r->current_string);
                   1249:        while (NULL != (cp = strstr(cp, "\\$"))) {
                   1250:                i = cp[2] - '1';
                   1251:                if (0 > i || 8 < i) {
                   1252:                        /* Not an argument invocation. */
                   1253:                        cp += 2;
                   1254:                        continue;
                   1255:                }
                   1256:
                   1257:                *szp = strlen(n1) - 3 + strlen(arg[i]) + 1;
                   1258:                n2 = mandoc_malloc(*szp);
                   1259:
                   1260:                strlcpy(n2, n1, (size_t)(cp - n1 + 1));
                   1261:                strlcat(n2, arg[i], *szp);
                   1262:                strlcat(n2, cp + 3, *szp);
                   1263:
                   1264:                cp = n2 + (cp - n1);
                   1265:                free(n1);
                   1266:                n1 = n2;
1.99      kristaps 1267:        }
                   1268:
1.106     kristaps 1269:        /*
                   1270:         * Replace the macro invocation
                   1271:         * by the expanded macro.
                   1272:         */
                   1273:        free(*bufp);
                   1274:        *bufp = n1;
                   1275:        if (0 == *szp)
                   1276:                *szp = strlen(*bufp) + 1;
                   1277:
                   1278:        return(*szp > 1 && '\n' == (*bufp)[(int)*szp - 2] ?
                   1279:           ROFF_REPARSE : ROFF_APPEND);
1.99      kristaps 1280: }
1.121     schwarze 1281:
                   1282: static char *
                   1283: roff_getname(struct roff *r, char **cpp, int ln, int pos)
                   1284: {
                   1285:        char     *name, *cp;
                   1286:
                   1287:        name = *cpp;
                   1288:        if ('\0' == *name)
                   1289:                return(name);
                   1290:
                   1291:        /* Read until end of name. */
                   1292:        for (cp = name; '\0' != *cp && ' ' != *cp; cp++) {
                   1293:                if ('\\' != *cp)
                   1294:                        continue;
                   1295:                cp++;
                   1296:                if ('\\' == *cp)
                   1297:                        continue;
1.128     kristaps 1298:                mandoc_msg(MANDOCERR_NAMESC, r->parse, ln, pos, NULL);
1.121     schwarze 1299:                *cp = '\0';
                   1300:                name = cp;
                   1301:        }
                   1302:
                   1303:        /* Nil-terminate name. */
                   1304:        if ('\0' != *cp)
                   1305:                *(cp++) = '\0';
                   1306:
                   1307:        /* Read past spaces. */
                   1308:        while (' ' == *cp)
                   1309:                cp++;
                   1310:
                   1311:        *cpp = cp;
                   1312:        return(name);
                   1313: }
                   1314:
1.106     kristaps 1315: /*
                   1316:  * Store *string into the user-defined string called *name.
                   1317:  * In multiline mode, append to an existing entry and append '\n';
                   1318:  * else replace the existing entry, if there is one.
                   1319:  * To clear an existing entry, call with (*r, *name, NULL, 0).
                   1320:  */
1.94      kristaps 1321: static void
1.106     kristaps 1322: roff_setstr(struct roff *r, const char *name, const char *string,
                   1323:        int multiline)
1.92      schwarze 1324: {
                   1325:        struct roffstr   *n;
1.106     kristaps 1326:        char             *c;
                   1327:        size_t            oldch, newch;
1.92      schwarze 1328:
1.106     kristaps 1329:        /* Search for an existing string with the same name. */
1.94      kristaps 1330:        n = r->first_string;
1.92      schwarze 1331:        while (n && strcmp(name, n->name))
                   1332:                n = n->next;
1.94      kristaps 1333:
                   1334:        if (NULL == n) {
1.106     kristaps 1335:                /* Create a new string table entry. */
1.94      kristaps 1336:                n = mandoc_malloc(sizeof(struct roffstr));
1.106     kristaps 1337:                n->name = mandoc_strdup(name);
                   1338:                n->string = NULL;
1.94      kristaps 1339:                n->next = r->first_string;
                   1340:                r->first_string = n;
1.106     kristaps 1341:        } else if (0 == multiline) {
                   1342:                /* In multiline mode, append; else replace. */
1.92      schwarze 1343:                free(n->string);
1.106     kristaps 1344:                n->string = NULL;
                   1345:        }
                   1346:
                   1347:        if (NULL == string)
                   1348:                return;
                   1349:
                   1350:        /*
                   1351:         * One additional byte for the '\n' in multiline mode,
                   1352:         * and one for the terminating '\0'.
                   1353:         */
1.127     kristaps 1354:        newch = strlen(string) + (multiline ? 2u : 1u);
1.106     kristaps 1355:        if (NULL == n->string) {
                   1356:                n->string = mandoc_malloc(newch);
                   1357:                *n->string = '\0';
                   1358:                oldch = 0;
                   1359:        } else {
                   1360:                oldch = strlen(n->string);
                   1361:                n->string = mandoc_realloc(n->string, oldch + newch);
                   1362:        }
                   1363:
                   1364:        /* Skip existing content in the destination buffer. */
1.127     kristaps 1365:        c = n->string + (int)oldch;
1.106     kristaps 1366:
                   1367:        /* Append new content to the destination buffer. */
                   1368:        while (*string) {
                   1369:                /*
                   1370:                 * Rudimentary roff copy mode:
                   1371:                 * Handle escaped backslashes.
                   1372:                 */
                   1373:                if ('\\' == *string && '\\' == *(string + 1))
                   1374:                        string++;
                   1375:                *c++ = *string++;
                   1376:        }
1.94      kristaps 1377:
1.106     kristaps 1378:        /* Append terminating bytes. */
                   1379:        if (multiline)
                   1380:                *c++ = '\n';
                   1381:        *c = '\0';
1.92      schwarze 1382: }
                   1383:
1.94      kristaps 1384: static const char *
                   1385: roff_getstrn(const struct roff *r, const char *name, size_t len)
1.92      schwarze 1386: {
1.94      kristaps 1387:        const struct roffstr *n;
1.92      schwarze 1388:
1.94      kristaps 1389:        n = r->first_string;
1.97      kristaps 1390:        while (n && (strncmp(name, n->name, len) || '\0' != n->name[(int)len]))
1.92      schwarze 1391:                n = n->next;
1.94      kristaps 1392:
                   1393:        return(n ? n->string : NULL);
1.92      schwarze 1394: }
                   1395:
1.94      kristaps 1396: static void
                   1397: roff_freestr(struct roff *r)
1.92      schwarze 1398: {
                   1399:        struct roffstr   *n, *nn;
                   1400:
1.94      kristaps 1401:        for (n = r->first_string; n; n = nn) {
1.92      schwarze 1402:                free(n->name);
                   1403:                free(n->string);
                   1404:                nn = n->next;
                   1405:                free(n);
                   1406:        }
1.94      kristaps 1407:
                   1408:        r->first_string = NULL;
1.114     kristaps 1409: }
                   1410:
                   1411: const struct tbl_span *
                   1412: roff_span(const struct roff *r)
                   1413: {
                   1414:
                   1415:        return(r->tbl ? tbl_span(r->tbl) : NULL);
1.125     kristaps 1416: }
                   1417:
                   1418: const struct eqn *
                   1419: roff_eqn(const struct roff *r)
                   1420: {
                   1421:
                   1422:        return(r->last_eqn ? &r->last_eqn->eqn : NULL);
1.74      kristaps 1423: }

CVSweb