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

Annotation of mandoc/roff.c, Revision 1.154

1.154   ! kristaps    1: /*     $Id: roff.c,v 1.153 2011/07/26 14:24:06 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:
1.147     kristaps   74: /*
                     75:  * A single register entity.  If "set" is zero, the value of the
                     76:  * register should be the default one, which is per-register.
                     77:  * Registers are assumed to be unsigned ints for now.
                     78:  */
                     79: struct reg {
                     80:        int               set; /* whether set or not */
                     81:        unsigned int      u; /* unsigned integer */
                     82: };
                     83:
1.94      kristaps   84: struct roffstr {
                     85:        char            *name; /* key of symbol */
                     86:        char            *string; /* current value */
                     87:        struct roffstr  *next; /* next in list */
                     88: };
                     89:
1.67      kristaps   90: struct roff {
1.128     kristaps   91:        struct mparse   *parse; /* parse point */
1.67      kristaps   92:        struct roffnode *last; /* leaf of stack */
1.82      kristaps   93:        enum roffrule    rstack[RSTACK_MAX]; /* stack of !`ie' rules */
                     94:        int              rstackpos; /* position in rstack */
1.147     kristaps   95:        struct reg       regs[REG__MAX];
1.106     kristaps   96:        struct roffstr  *first_string; /* user-defined strings & macros */
                     97:        const char      *current_string; /* value of last called user macro */
1.118     kristaps   98:        struct tbl_node *first_tbl; /* first table parsed */
                     99:        struct tbl_node *last_tbl; /* last table parsed */
                    100:        struct tbl_node *tbl; /* current table being parsed */
1.125     kristaps  101:        struct eqn_node *last_eqn; /* last equation parsed */
                    102:        struct eqn_node *first_eqn; /* first equation parsed */
                    103:        struct eqn_node *eqn; /* current equation being parsed */
1.79      kristaps  104: };
                    105:
1.67      kristaps  106: struct roffnode {
                    107:        enum rofft       tok; /* type of node */
                    108:        struct roffnode *parent; /* up one in stack */
                    109:        int              line; /* parse line */
                    110:        int              col; /* parse col */
1.106     kristaps  111:        char            *name; /* node name, e.g. macro name */
1.79      kristaps  112:        char            *end; /* end-rules: custom token */
                    113:        int              endspan; /* end-rules: next-line or infty */
1.82      kristaps  114:        enum roffrule    rule; /* current evaluation rule */
1.67      kristaps  115: };
                    116:
                    117: #define        ROFF_ARGS        struct roff *r, /* parse ctx */ \
1.72      kristaps  118:                         enum rofft tok, /* tok of macro */ \
1.67      kristaps  119:                         char **bufp, /* input buffer */ \
                    120:                         size_t *szp, /* size of input buffer */ \
                    121:                         int ln, /* parse line */ \
1.75      kristaps  122:                         int ppos, /* original pos in buffer */ \
                    123:                         int pos, /* current pos in buffer */ \
1.74      kristaps  124:                         int *offs /* reset offset of buffer data */
1.67      kristaps  125:
                    126: typedef        enum rofferr (*roffproc)(ROFF_ARGS);
                    127:
                    128: struct roffmac {
                    129:        const char      *name; /* macro name */
1.79      kristaps  130:        roffproc         proc; /* process new macro */
                    131:        roffproc         text; /* process as child text of macro */
                    132:        roffproc         sub; /* process as child of macro */
                    133:        int              flags;
                    134: #define        ROFFMAC_STRUCT  (1 << 0) /* always interpret */
1.85      kristaps  135:        struct roffmac  *next;
1.67      kristaps  136: };
                    137:
1.141     kristaps  138: struct predef {
                    139:        const char      *name; /* predefined input name */
                    140:        const char      *str; /* replacement symbol */
                    141: };
                    142:
                    143: #define        PREDEF(__name, __str) \
                    144:        { (__name), (__str) },
                    145:
1.80      kristaps  146: static enum rofferr     roff_block(ROFF_ARGS);
                    147: static enum rofferr     roff_block_text(ROFF_ARGS);
                    148: static enum rofferr     roff_block_sub(ROFF_ARGS);
                    149: static enum rofferr     roff_cblock(ROFF_ARGS);
                    150: static enum rofferr     roff_ccond(ROFF_ARGS);
1.82      kristaps  151: static enum rofferr     roff_cond(ROFF_ARGS);
                    152: static enum rofferr     roff_cond_text(ROFF_ARGS);
                    153: static enum rofferr     roff_cond_sub(ROFF_ARGS);
1.92      schwarze  154: static enum rofferr     roff_ds(ROFF_ARGS);
1.94      kristaps  155: static enum roffrule    roff_evalcond(const char *, int *);
                    156: static void             roff_freestr(struct roff *);
1.121     schwarze  157: static char            *roff_getname(struct roff *, char **, int, int);
1.94      kristaps  158: static const char      *roff_getstrn(const struct roff *,
                    159:                                const char *, size_t);
1.103     kristaps  160: static enum rofferr     roff_line_ignore(ROFF_ARGS);
1.89      kristaps  161: static enum rofferr     roff_nr(ROFF_ARGS);
1.154   ! kristaps  162: static void             roff_res(struct roff *,
1.142     kristaps  163:                                char **, size_t *, int, int);
1.122     schwarze  164: static enum rofferr     roff_rm(ROFF_ARGS);
1.94      kristaps  165: static void             roff_setstr(struct roff *,
1.106     kristaps  166:                                const char *, const char *, int);
1.105     kristaps  167: static enum rofferr     roff_so(ROFF_ARGS);
1.109     kristaps  168: static enum rofferr     roff_TE(ROFF_ARGS);
                    169: static enum rofferr     roff_TS(ROFF_ARGS);
1.125     kristaps  170: static enum rofferr     roff_EQ(ROFF_ARGS);
                    171: static enum rofferr     roff_EN(ROFF_ARGS);
1.112     kristaps  172: static enum rofferr     roff_T_(ROFF_ARGS);
1.106     kristaps  173: static enum rofferr     roff_userdef(ROFF_ARGS);
1.67      kristaps  174:
1.85      kristaps  175: /* See roff_hash_find() */
                    176:
                    177: #define        ASCII_HI         126
                    178: #define        ASCII_LO         33
                    179: #define        HASHWIDTH       (ASCII_HI - ASCII_LO + 1)
                    180:
                    181: static struct roffmac  *hash[HASHWIDTH];
                    182:
                    183: static struct roffmac   roffs[ROFF_MAX] = {
1.103     kristaps  184:        { "ad", roff_line_ignore, NULL, NULL, 0, NULL },
1.85      kristaps  185:        { "am", roff_block, roff_block_text, roff_block_sub, 0, NULL },
                    186:        { "ami", roff_block, roff_block_text, roff_block_sub, 0, NULL },
                    187:        { "am1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
                    188:        { "de", roff_block, roff_block_text, roff_block_sub, 0, NULL },
                    189:        { "dei", roff_block, roff_block_text, roff_block_sub, 0, NULL },
                    190:        { "de1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.92      schwarze  191:        { "ds", roff_ds, NULL, NULL, 0, NULL },
1.85      kristaps  192:        { "el", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
1.103     kristaps  193:        { "hy", roff_line_ignore, NULL, NULL, 0, NULL },
1.85      kristaps  194:        { "ie", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
                    195:        { "if", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
                    196:        { "ig", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.123     schwarze  197:        { "it", roff_line_ignore, NULL, NULL, 0, NULL },
1.103     kristaps  198:        { "ne", roff_line_ignore, NULL, NULL, 0, NULL },
                    199:        { "nh", roff_line_ignore, NULL, NULL, 0, NULL },
1.104     kristaps  200:        { "nr", roff_nr, NULL, NULL, 0, NULL },
1.124     schwarze  201:        { "ns", roff_line_ignore, NULL, NULL, 0, NULL },
                    202:        { "ps", roff_line_ignore, NULL, NULL, 0, NULL },
1.122     schwarze  203:        { "rm", roff_rm, NULL, NULL, 0, NULL },
1.105     kristaps  204:        { "so", roff_so, NULL, NULL, 0, NULL },
1.124     schwarze  205:        { "ta", roff_line_ignore, NULL, NULL, 0, NULL },
1.103     kristaps  206:        { "tr", roff_line_ignore, NULL, NULL, 0, NULL },
1.109     kristaps  207:        { "TS", roff_TS, NULL, NULL, 0, NULL },
                    208:        { "TE", roff_TE, NULL, NULL, 0, NULL },
1.112     kristaps  209:        { "T&", roff_T_, NULL, NULL, 0, NULL },
1.125     kristaps  210:        { "EQ", roff_EQ, NULL, NULL, 0, NULL },
                    211:        { "EN", roff_EN, NULL, NULL, 0, NULL },
1.85      kristaps  212:        { ".", roff_cblock, NULL, NULL, 0, NULL },
                    213:        { "\\}", roff_ccond, NULL, NULL, 0, NULL },
1.106     kristaps  214:        { NULL, roff_userdef, NULL, NULL, 0, NULL },
1.67      kristaps  215: };
                    216:
1.141     kristaps  217: /* Array of injected predefined strings. */
                    218: #define        PREDEFS_MAX      38
                    219: static const struct predef predefs[PREDEFS_MAX] = {
                    220: #include "predefs.in"
                    221: };
                    222:
1.67      kristaps  223: static void             roff_free1(struct roff *);
1.106     kristaps  224: static enum rofft       roff_hash_find(const char *, size_t);
1.85      kristaps  225: static void             roff_hash_init(void);
1.76      kristaps  226: static void             roffnode_cleanscope(struct roff *);
1.106     kristaps  227: static void             roffnode_push(struct roff *, enum rofft,
                    228:                                const char *, int, int);
1.67      kristaps  229: static void             roffnode_pop(struct roff *);
1.106     kristaps  230: static enum rofft       roff_parse(struct roff *, const char *, int *);
1.67      kristaps  231:
1.85      kristaps  232: /* See roff_hash_find() */
                    233: #define        ROFF_HASH(p)    (p[0] - ASCII_LO)
                    234:
                    235: static void
                    236: roff_hash_init(void)
                    237: {
                    238:        struct roffmac   *n;
                    239:        int               buc, i;
                    240:
1.106     kristaps  241:        for (i = 0; i < (int)ROFF_USERDEF; i++) {
1.85      kristaps  242:                assert(roffs[i].name[0] >= ASCII_LO);
                    243:                assert(roffs[i].name[0] <= ASCII_HI);
                    244:
                    245:                buc = ROFF_HASH(roffs[i].name);
                    246:
                    247:                if (NULL != (n = hash[buc])) {
                    248:                        for ( ; n->next; n = n->next)
                    249:                                /* Do nothing. */ ;
                    250:                        n->next = &roffs[i];
                    251:                } else
                    252:                        hash[buc] = &roffs[i];
                    253:        }
                    254: }
                    255:
1.67      kristaps  256: /*
                    257:  * Look up a roff token by its name.  Returns ROFF_MAX if no macro by
                    258:  * the nil-terminated string name could be found.
                    259:  */
                    260: static enum rofft
1.106     kristaps  261: roff_hash_find(const char *p, size_t s)
1.67      kristaps  262: {
1.85      kristaps  263:        int              buc;
                    264:        struct roffmac  *n;
1.67      kristaps  265:
1.85      kristaps  266:        /*
                    267:         * libroff has an extremely simple hashtable, for the time
                    268:         * being, which simply keys on the first character, which must
                    269:         * be printable, then walks a chain.  It works well enough until
                    270:         * optimised.
                    271:         */
                    272:
                    273:        if (p[0] < ASCII_LO || p[0] > ASCII_HI)
                    274:                return(ROFF_MAX);
                    275:
                    276:        buc = ROFF_HASH(p);
                    277:
                    278:        if (NULL == (n = hash[buc]))
                    279:                return(ROFF_MAX);
                    280:        for ( ; n; n = n->next)
1.106     kristaps  281:                if (0 == strncmp(n->name, p, s) && '\0' == n->name[(int)s])
1.85      kristaps  282:                        return((enum rofft)(n - roffs));
1.67      kristaps  283:
                    284:        return(ROFF_MAX);
                    285: }
                    286:
                    287:
                    288: /*
                    289:  * Pop the current node off of the stack of roff instructions currently
                    290:  * pending.
                    291:  */
                    292: static void
                    293: roffnode_pop(struct roff *r)
                    294: {
                    295:        struct roffnode *p;
                    296:
1.75      kristaps  297:        assert(r->last);
                    298:        p = r->last;
1.82      kristaps  299:
1.75      kristaps  300:        r->last = r->last->parent;
1.106     kristaps  301:        free(p->name);
                    302:        free(p->end);
1.67      kristaps  303:        free(p);
                    304: }
                    305:
                    306:
                    307: /*
                    308:  * Push a roff node onto the instruction stack.  This must later be
                    309:  * removed with roffnode_pop().
                    310:  */
1.98      schwarze  311: static void
1.106     kristaps  312: roffnode_push(struct roff *r, enum rofft tok, const char *name,
                    313:                int line, int col)
1.67      kristaps  314: {
                    315:        struct roffnode *p;
                    316:
1.98      schwarze  317:        p = mandoc_calloc(1, sizeof(struct roffnode));
1.67      kristaps  318:        p->tok = tok;
1.106     kristaps  319:        if (name)
                    320:                p->name = mandoc_strdup(name);
1.67      kristaps  321:        p->parent = r->last;
                    322:        p->line = line;
                    323:        p->col = col;
1.79      kristaps  324:        p->rule = p->parent ? p->parent->rule : ROFFRULE_DENY;
1.67      kristaps  325:
                    326:        r->last = p;
                    327: }
                    328:
                    329:
                    330: static void
                    331: roff_free1(struct roff *r)
                    332: {
1.118     kristaps  333:        struct tbl_node *t;
1.125     kristaps  334:        struct eqn_node *e;
1.67      kristaps  335:
1.125     kristaps  336:        while (NULL != (t = r->first_tbl)) {
1.113     kristaps  337:                r->first_tbl = t->next;
                    338:                tbl_free(t);
1.109     kristaps  339:        }
                    340:
1.113     kristaps  341:        r->first_tbl = r->last_tbl = r->tbl = NULL;
                    342:
1.125     kristaps  343:        while (NULL != (e = r->first_eqn)) {
                    344:                r->first_eqn = e->next;
                    345:                eqn_free(e);
                    346:        }
                    347:
                    348:        r->first_eqn = r->last_eqn = r->eqn = NULL;
                    349:
1.67      kristaps  350:        while (r->last)
                    351:                roffnode_pop(r);
1.109     kristaps  352:
1.94      kristaps  353:        roff_freestr(r);
1.67      kristaps  354: }
                    355:
                    356:
                    357: void
                    358: roff_reset(struct roff *r)
                    359: {
1.143     kristaps  360:        int              i;
1.67      kristaps  361:
                    362:        roff_free1(r);
1.143     kristaps  363:
1.147     kristaps  364:        memset(&r->regs, 0, sizeof(struct reg) * REG__MAX);
                    365:
1.143     kristaps  366:        for (i = 0; i < PREDEFS_MAX; i++)
                    367:                roff_setstr(r, predefs[i].name, predefs[i].str, 0);
1.67      kristaps  368: }
                    369:
                    370:
                    371: void
                    372: roff_free(struct roff *r)
                    373: {
                    374:
                    375:        roff_free1(r);
                    376:        free(r);
                    377: }
                    378:
                    379:
                    380: struct roff *
1.147     kristaps  381: roff_alloc(struct mparse *parse)
1.67      kristaps  382: {
                    383:        struct roff     *r;
1.141     kristaps  384:        int              i;
1.67      kristaps  385:
1.98      schwarze  386:        r = mandoc_calloc(1, sizeof(struct roff));
1.128     kristaps  387:        r->parse = parse;
1.82      kristaps  388:        r->rstackpos = -1;
1.85      kristaps  389:
                    390:        roff_hash_init();
1.141     kristaps  391:
                    392:        for (i = 0; i < PREDEFS_MAX; i++)
                    393:                roff_setstr(r, predefs[i].name, predefs[i].str, 0);
                    394:
1.67      kristaps  395:        return(r);
                    396: }
                    397:
1.94      kristaps  398: /*
                    399:  * Pre-filter each and every line for reserved words (one beginning with
                    400:  * `\*', e.g., `\*(ab').  These must be handled before the actual line
                    401:  * is processed.
1.153     kristaps  402:  * This also checks the syntax of regular escapes.
1.154   ! kristaps  403:  */
        !           404: static void
1.142     kristaps  405: roff_res(struct roff *r, char **bufp, size_t *szp, int ln, int pos)
1.94      kristaps  406: {
1.152     kristaps  407:        enum mandoc_esc  esc;
1.108     schwarze  408:        const char      *stesc; /* start of an escape sequence ('\\') */
                    409:        const char      *stnam; /* start of the name, after "[(*" */
                    410:        const char      *cp;    /* end of the name, e.g. before ']' */
                    411:        const char      *res;   /* the string to be substituted */
1.94      kristaps  412:        int              i, maxl;
                    413:        size_t           nsz;
                    414:        char            *n;
                    415:
1.154   ! kristaps  416: again:
1.108     schwarze  417:        cp = *bufp + pos;
                    418:        while (NULL != (cp = strchr(cp, '\\'))) {
                    419:                stesc = cp++;
                    420:
                    421:                /*
                    422:                 * The second character must be an asterisk.
                    423:                 * If it isn't, skip it anyway:  It is escaped,
                    424:                 * so it can't start another escape sequence.
                    425:                 */
                    426:
                    427:                if ('\0' == *cp)
1.154   ! kristaps  428:                        return;
1.152     kristaps  429:
                    430:                if ('*' != *cp) {
                    431:                        res = cp;
                    432:                        esc = mandoc_escape(&cp, NULL, NULL);
                    433:                        if (ESCAPE_ERROR != esc)
                    434:                                continue;
                    435:                        cp = res;
1.153     kristaps  436:                        mandoc_msg
                    437:                                (MANDOCERR_BADESCAPE, r->parse,
                    438:                                 ln, (int)(stesc - *bufp), NULL);
1.154   ! kristaps  439:                        return;
1.152     kristaps  440:                }
                    441:
                    442:                cp++;
1.108     schwarze  443:
                    444:                /*
                    445:                 * The third character decides the length
                    446:                 * of the name of the string.
                    447:                 * Save a pointer to the name.
                    448:                 */
                    449:
1.94      kristaps  450:                switch (*cp) {
1.108     schwarze  451:                case ('\0'):
1.154   ! kristaps  452:                        return;
1.94      kristaps  453:                case ('('):
                    454:                        cp++;
                    455:                        maxl = 2;
                    456:                        break;
                    457:                case ('['):
                    458:                        cp++;
                    459:                        maxl = 0;
                    460:                        break;
                    461:                default:
                    462:                        maxl = 1;
                    463:                        break;
                    464:                }
1.108     schwarze  465:                stnam = cp;
1.94      kristaps  466:
1.108     schwarze  467:                /* Advance to the end of the name. */
1.94      kristaps  468:
                    469:                for (i = 0; 0 == maxl || i < maxl; i++, cp++) {
1.153     kristaps  470:                        if ('\0' == *cp) {
                    471:                                mandoc_msg
                    472:                                        (MANDOCERR_BADESCAPE,
                    473:                                         r->parse, ln,
                    474:                                         (int)(stesc - *bufp), NULL);
1.154   ! kristaps  475:                                return;
1.153     kristaps  476:                        }
1.94      kristaps  477:                        if (0 == maxl && ']' == *cp)
                    478:                                break;
                    479:                }
                    480:
1.108     schwarze  481:                /*
                    482:                 * Retrieve the replacement string; if it is
                    483:                 * undefined, resume searching for escapes.
                    484:                 */
                    485:
                    486:                res = roff_getstrn(r, stnam, (size_t)i);
1.94      kristaps  487:
                    488:                if (NULL == res) {
1.153     kristaps  489:                        mandoc_msg
                    490:                                (MANDOCERR_BADESCAPE, r->parse,
                    491:                                 ln, (int)(stesc - *bufp), NULL);
1.142     kristaps  492:                        res = "";
1.94      kristaps  493:                }
                    494:
1.108     schwarze  495:                /* Replace the escape sequence by the string. */
                    496:
1.154   ! kristaps  497:                pos += (stesc - *bufp);
        !           498:
1.94      kristaps  499:                nsz = *szp + strlen(res) + 1;
                    500:                n = mandoc_malloc(nsz);
                    501:
1.108     schwarze  502:                strlcpy(n, *bufp, (size_t)(stesc - *bufp + 1));
1.94      kristaps  503:                strlcat(n, res, nsz);
                    504:                strlcat(n, cp + (maxl ? 0 : 1), nsz);
                    505:
                    506:                free(*bufp);
                    507:
                    508:                *bufp = n;
                    509:                *szp = nsz;
1.154   ! kristaps  510:                goto again;
        !           511:        }
        !           512: }
        !           513:
        !           514: /*
        !           515:  * Process text streams: convert all breakable hyphens into ASCII_HYPH.
        !           516:  */
        !           517: static enum rofferr
        !           518: roff_parsetext(char *p)
        !           519: {
        !           520:        size_t           sz;
        !           521:        const char      *start;
        !           522:        enum mandoc_esc  esc;
        !           523:
        !           524:        start = p;
        !           525:
        !           526:        while ('\0' != *p) {
        !           527:                sz = strcspn(p, "-\\");
        !           528:                p += sz;
        !           529:
        !           530:                if ('\\' == *p) {
        !           531:                        /* Skip over escapes. */
        !           532:                        p++;
        !           533:                        esc = mandoc_escape
        !           534:                                ((const char **)&p, NULL, NULL);
        !           535:                        if (ESCAPE_ERROR == esc)
        !           536:                                break;
        !           537:                } else if ('-' == *p) {
        !           538:                        if (mandoc_hyph(start, p))
        !           539:                                *p = ASCII_HYPH;
        !           540:                        p++;
        !           541:                }
1.94      kristaps  542:        }
                    543:
1.154   ! kristaps  544:        return(ROFF_CONT);
1.94      kristaps  545: }
                    546:
1.67      kristaps  547: enum rofferr
1.90      kristaps  548: roff_parseln(struct roff *r, int ln, char **bufp,
                    549:                size_t *szp, int pos, int *offs)
1.67      kristaps  550: {
                    551:        enum rofft       t;
1.109     kristaps  552:        enum rofferr     e;
1.130     kristaps  553:        int              ppos, ctl;
1.79      kristaps  554:
                    555:        /*
1.94      kristaps  556:         * Run the reserved-word filter only if we have some reserved
                    557:         * words to fill in.
                    558:         */
                    559:
1.154   ! kristaps  560:        roff_res(r, bufp, szp, ln, pos);
1.94      kristaps  561:
1.130     kristaps  562:        ppos = pos;
                    563:        ctl = mandoc_getcontrol(*bufp, &pos);
                    564:
1.94      kristaps  565:        /*
1.79      kristaps  566:         * First, if a scope is open and we're not a macro, pass the
                    567:         * text through the macro's filter.  If a scope isn't open and
                    568:         * we're not a macro, just let it through.
1.125     kristaps  569:         * Finally, if there's an equation scope open, divert it into it
                    570:         * no matter our state.
1.79      kristaps  571:         */
1.74      kristaps  572:
1.130     kristaps  573:        if (r->last && ! ctl) {
1.78      kristaps  574:                t = r->last->tok;
                    575:                assert(roffs[t].text);
1.109     kristaps  576:                e = (*roffs[t].text)
                    577:                        (r, t, bufp, szp, ln, pos, pos, offs);
                    578:                assert(ROFF_IGN == e || ROFF_CONT == e);
1.125     kristaps  579:                if (ROFF_CONT != e)
                    580:                        return(e);
                    581:                if (r->eqn)
1.146     kristaps  582:                        return(eqn_read(&r->eqn, ln, *bufp, pos, offs));
1.125     kristaps  583:                if (r->tbl)
1.130     kristaps  584:                        return(tbl_read(r->tbl, ln, *bufp, pos));
1.154   ! kristaps  585:                return(roff_parsetext(*bufp + pos));
1.130     kristaps  586:        } else if ( ! ctl) {
1.125     kristaps  587:                if (r->eqn)
1.146     kristaps  588:                        return(eqn_read(&r->eqn, ln, *bufp, pos, offs));
1.109     kristaps  589:                if (r->tbl)
1.130     kristaps  590:                        return(tbl_read(r->tbl, ln, *bufp, pos));
1.154   ! kristaps  591:                return(roff_parsetext(*bufp + pos));
1.125     kristaps  592:        } else if (r->eqn)
1.146     kristaps  593:                return(eqn_read(&r->eqn, ln, *bufp, ppos, offs));
1.67      kristaps  594:
1.79      kristaps  595:        /*
                    596:         * If a scope is open, go to the child handler for that macro,
                    597:         * as it may want to preprocess before doing anything with it.
1.125     kristaps  598:         * Don't do so if an equation is open.
1.79      kristaps  599:         */
1.78      kristaps  600:
1.79      kristaps  601:        if (r->last) {
                    602:                t = r->last->tok;
                    603:                assert(roffs[t].sub);
                    604:                return((*roffs[t].sub)
1.90      kristaps  605:                                (r, t, bufp, szp,
1.130     kristaps  606:                                 ln, ppos, pos, offs));
1.79      kristaps  607:        }
1.78      kristaps  608:
1.79      kristaps  609:        /*
                    610:         * Lastly, as we've no scope open, try to look up and execute
                    611:         * the new macro.  If no macro is found, simply return and let
                    612:         * the compilers handle it.
                    613:         */
1.67      kristaps  614:
1.106     kristaps  615:        if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos)))
1.79      kristaps  616:                return(ROFF_CONT);
1.67      kristaps  617:
1.75      kristaps  618:        assert(roffs[t].proc);
1.78      kristaps  619:        return((*roffs[t].proc)
1.90      kristaps  620:                        (r, t, bufp, szp,
                    621:                         ln, ppos, pos, offs));
1.74      kristaps  622: }
                    623:
                    624:
1.117     kristaps  625: void
1.74      kristaps  626: roff_endparse(struct roff *r)
                    627: {
                    628:
1.110     kristaps  629:        if (r->last)
1.128     kristaps  630:                mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.109     kristaps  631:                                r->last->line, r->last->col, NULL);
1.117     kristaps  632:
1.125     kristaps  633:        if (r->eqn) {
1.128     kristaps  634:                mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.148     kristaps  635:                                r->eqn->eqn.ln, r->eqn->eqn.pos, NULL);
1.151     kristaps  636:                eqn_end(&r->eqn);
1.125     kristaps  637:        }
                    638:
1.117     kristaps  639:        if (r->tbl) {
1.128     kristaps  640:                mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.117     kristaps  641:                                r->tbl->line, r->tbl->pos, NULL);
1.151     kristaps  642:                tbl_end(&r->tbl);
1.117     kristaps  643:        }
1.67      kristaps  644: }
                    645:
                    646: /*
                    647:  * Parse a roff node's type from the input buffer.  This must be in the
                    648:  * form of ".foo xxx" in the usual way.
                    649:  */
                    650: static enum rofft
1.106     kristaps  651: roff_parse(struct roff *r, const char *buf, int *pos)
1.67      kristaps  652: {
1.106     kristaps  653:        const char      *mac;
                    654:        size_t           maclen;
1.67      kristaps  655:        enum rofft       t;
                    656:
1.144     kristaps  657:        if ('\0' == buf[*pos] || '"' == buf[*pos] ||
                    658:                        '\t' == buf[*pos] || ' ' == buf[*pos])
1.67      kristaps  659:                return(ROFF_MAX);
                    660:
1.144     kristaps  661:        /*
                    662:         * We stop the macro parse at an escape, tab, space, or nil.
                    663:         * However, `\}' is also a valid macro, so make sure we don't
                    664:         * clobber it by seeing the `\' as the end of token.
                    665:         */
                    666:
1.106     kristaps  667:        mac = buf + *pos;
1.144     kristaps  668:        maclen = strcspn(mac + 1, " \\\t\0") + 1;
1.67      kristaps  669:
1.106     kristaps  670:        t = (r->current_string = roff_getstrn(r, mac, maclen))
                    671:            ? ROFF_USERDEF : roff_hash_find(mac, maclen);
1.67      kristaps  672:
1.127     kristaps  673:        *pos += (int)maclen;
1.130     kristaps  674:
1.67      kristaps  675:        while (buf[*pos] && ' ' == buf[*pos])
                    676:                (*pos)++;
                    677:
                    678:        return(t);
                    679: }
                    680:
                    681: /* ARGSUSED */
                    682: static enum rofferr
1.76      kristaps  683: roff_cblock(ROFF_ARGS)
1.67      kristaps  684: {
                    685:
1.79      kristaps  686:        /*
                    687:         * A block-close `..' should only be invoked as a child of an
                    688:         * ignore macro, otherwise raise a warning and just ignore it.
                    689:         */
                    690:
1.76      kristaps  691:        if (NULL == r->last) {
1.128     kristaps  692:                mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.76      kristaps  693:                return(ROFF_IGN);
                    694:        }
1.67      kristaps  695:
1.81      kristaps  696:        switch (r->last->tok) {
                    697:        case (ROFF_am):
                    698:                /* FALLTHROUGH */
                    699:        case (ROFF_ami):
                    700:                /* FALLTHROUGH */
                    701:        case (ROFF_am1):
                    702:                /* FALLTHROUGH */
                    703:        case (ROFF_de):
1.108     schwarze  704:                /* ROFF_de1 is remapped to ROFF_de in roff_block(). */
1.81      kristaps  705:                /* FALLTHROUGH */
                    706:        case (ROFF_dei):
                    707:                /* FALLTHROUGH */
                    708:        case (ROFF_ig):
                    709:                break;
                    710:        default:
1.128     kristaps  711:                mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.67      kristaps  712:                return(ROFF_IGN);
1.76      kristaps  713:        }
1.67      kristaps  714:
1.76      kristaps  715:        if ((*bufp)[pos])
1.128     kristaps  716:                mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.71      kristaps  717:
                    718:        roffnode_pop(r);
1.76      kristaps  719:        roffnode_cleanscope(r);
                    720:        return(ROFF_IGN);
1.71      kristaps  721:
1.67      kristaps  722: }
                    723:
                    724:
1.76      kristaps  725: static void
                    726: roffnode_cleanscope(struct roff *r)
1.67      kristaps  727: {
                    728:
1.76      kristaps  729:        while (r->last) {
                    730:                if (--r->last->endspan < 0)
                    731:                        break;
                    732:                roffnode_pop(r);
                    733:        }
1.67      kristaps  734: }
                    735:
                    736:
1.75      kristaps  737: /* ARGSUSED */
1.74      kristaps  738: static enum rofferr
1.75      kristaps  739: roff_ccond(ROFF_ARGS)
1.74      kristaps  740: {
                    741:
1.76      kristaps  742:        if (NULL == r->last) {
1.128     kristaps  743:                mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.76      kristaps  744:                return(ROFF_IGN);
                    745:        }
                    746:
1.82      kristaps  747:        switch (r->last->tok) {
                    748:        case (ROFF_el):
                    749:                /* FALLTHROUGH */
                    750:        case (ROFF_ie):
                    751:                /* FALLTHROUGH */
                    752:        case (ROFF_if):
                    753:                break;
                    754:        default:
1.128     kristaps  755:                mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.75      kristaps  756:                return(ROFF_IGN);
                    757:        }
                    758:
1.76      kristaps  759:        if (r->last->endspan > -1) {
1.128     kristaps  760:                mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.76      kristaps  761:                return(ROFF_IGN);
                    762:        }
                    763:
                    764:        if ((*bufp)[pos])
1.128     kristaps  765:                mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.76      kristaps  766:
1.75      kristaps  767:        roffnode_pop(r);
1.76      kristaps  768:        roffnode_cleanscope(r);
                    769:        return(ROFF_IGN);
                    770: }
                    771:
1.75      kristaps  772:
1.76      kristaps  773: /* ARGSUSED */
                    774: static enum rofferr
1.80      kristaps  775: roff_block(ROFF_ARGS)
1.76      kristaps  776: {
1.78      kristaps  777:        int             sv;
                    778:        size_t          sz;
1.106     kristaps  779:        char            *name;
                    780:
                    781:        name = NULL;
1.76      kristaps  782:
1.106     kristaps  783:        if (ROFF_ig != tok) {
                    784:                if ('\0' == (*bufp)[pos]) {
1.128     kristaps  785:                        mandoc_msg(MANDOCERR_NOARGS, r->parse, ln, ppos, NULL);
1.106     kristaps  786:                        return(ROFF_IGN);
                    787:                }
1.107     kristaps  788:
                    789:                /*
                    790:                 * Re-write `de1', since we don't really care about
                    791:                 * groff's strange compatibility mode, into `de'.
                    792:                 */
                    793:
1.106     kristaps  794:                if (ROFF_de1 == tok)
                    795:                        tok = ROFF_de;
                    796:                if (ROFF_de == tok)
                    797:                        name = *bufp + pos;
                    798:                else
1.128     kristaps  799:                        mandoc_msg(MANDOCERR_REQUEST, r->parse, ln, ppos,
1.106     kristaps  800:                            roffs[tok].name);
1.107     kristaps  801:
1.131     schwarze  802:                while ((*bufp)[pos] && ! isspace((unsigned char)(*bufp)[pos]))
1.80      kristaps  803:                        pos++;
1.107     kristaps  804:
1.131     schwarze  805:                while (isspace((unsigned char)(*bufp)[pos]))
1.106     kristaps  806:                        (*bufp)[pos++] = '\0';
1.80      kristaps  807:        }
                    808:
1.106     kristaps  809:        roffnode_push(r, tok, name, ln, ppos);
                    810:
                    811:        /*
                    812:         * At the beginning of a `de' macro, clear the existing string
                    813:         * with the same name, if there is one.  New content will be
                    814:         * added from roff_block_text() in multiline mode.
                    815:         */
1.107     kristaps  816:
1.106     kristaps  817:        if (ROFF_de == tok)
1.108     schwarze  818:                roff_setstr(r, name, "", 0);
1.76      kristaps  819:
1.79      kristaps  820:        if ('\0' == (*bufp)[pos])
1.78      kristaps  821:                return(ROFF_IGN);
                    822:
1.107     kristaps  823:        /* If present, process the custom end-of-line marker. */
                    824:
1.78      kristaps  825:        sv = pos;
1.131     schwarze  826:        while ((*bufp)[pos] && ! isspace((unsigned char)(*bufp)[pos]))
1.78      kristaps  827:                pos++;
                    828:
                    829:        /*
                    830:         * Note: groff does NOT like escape characters in the input.
                    831:         * Instead of detecting this, we're just going to let it fly and
                    832:         * to hell with it.
                    833:         */
                    834:
                    835:        assert(pos > sv);
                    836:        sz = (size_t)(pos - sv);
                    837:
1.79      kristaps  838:        if (1 == sz && '.' == (*bufp)[sv])
                    839:                return(ROFF_IGN);
                    840:
1.98      schwarze  841:        r->last->end = mandoc_malloc(sz + 1);
1.78      kristaps  842:
                    843:        memcpy(r->last->end, *bufp + sv, sz);
                    844:        r->last->end[(int)sz] = '\0';
                    845:
1.77      kristaps  846:        if ((*bufp)[pos])
1.128     kristaps  847:                mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.74      kristaps  848:
1.78      kristaps  849:        return(ROFF_IGN);
                    850: }
                    851:
                    852:
                    853: /* ARGSUSED */
                    854: static enum rofferr
1.80      kristaps  855: roff_block_sub(ROFF_ARGS)
1.79      kristaps  856: {
                    857:        enum rofft      t;
                    858:        int             i, j;
                    859:
                    860:        /*
                    861:         * First check whether a custom macro exists at this level.  If
                    862:         * it does, then check against it.  This is some of groff's
                    863:         * stranger behaviours.  If we encountered a custom end-scope
                    864:         * tag and that tag also happens to be a "real" macro, then we
                    865:         * need to try interpreting it again as a real macro.  If it's
                    866:         * not, then return ignore.  Else continue.
                    867:         */
                    868:
                    869:        if (r->last->end) {
1.130     kristaps  870:                for (i = pos, j = 0; r->last->end[j]; j++, i++)
1.79      kristaps  871:                        if ((*bufp)[i] != r->last->end[j])
                    872:                                break;
                    873:
                    874:                if ('\0' == r->last->end[j] &&
                    875:                                ('\0' == (*bufp)[i] ||
                    876:                                 ' ' == (*bufp)[i] ||
                    877:                                 '\t' == (*bufp)[i])) {
                    878:                        roffnode_pop(r);
                    879:                        roffnode_cleanscope(r);
                    880:
1.130     kristaps  881:                        while (' ' == (*bufp)[i] || '\t' == (*bufp)[i])
                    882:                                i++;
                    883:
                    884:                        pos = i;
1.106     kristaps  885:                        if (ROFF_MAX != roff_parse(r, *bufp, &pos))
1.79      kristaps  886:                                return(ROFF_RERUN);
                    887:                        return(ROFF_IGN);
                    888:                }
                    889:        }
                    890:
                    891:        /*
                    892:         * If we have no custom end-query or lookup failed, then try
                    893:         * pulling it out of the hashtable.
                    894:         */
                    895:
1.137     schwarze  896:        t = roff_parse(r, *bufp, &pos);
1.79      kristaps  897:
1.106     kristaps  898:        /*
                    899:         * Macros other than block-end are only significant
                    900:         * in `de' blocks; elsewhere, simply throw them away.
                    901:         */
                    902:        if (ROFF_cblock != t) {
                    903:                if (ROFF_de == tok)
                    904:                        roff_setstr(r, r->last->name, *bufp + ppos, 1);
1.79      kristaps  905:                return(ROFF_IGN);
1.106     kristaps  906:        }
1.79      kristaps  907:
                    908:        assert(roffs[t].proc);
1.90      kristaps  909:        return((*roffs[t].proc)(r, t, bufp, szp,
                    910:                                ln, ppos, pos, offs));
1.79      kristaps  911: }
                    912:
                    913:
                    914: /* ARGSUSED */
                    915: static enum rofferr
1.80      kristaps  916: roff_block_text(ROFF_ARGS)
1.78      kristaps  917: {
                    918:
1.106     kristaps  919:        if (ROFF_de == tok)
                    920:                roff_setstr(r, r->last->name, *bufp + pos, 1);
                    921:
1.78      kristaps  922:        return(ROFF_IGN);
                    923: }
                    924:
                    925:
                    926: /* ARGSUSED */
                    927: static enum rofferr
1.82      kristaps  928: roff_cond_sub(ROFF_ARGS)
                    929: {
                    930:        enum rofft       t;
                    931:        enum roffrule    rr;
1.139     kristaps  932:        char            *ep;
1.82      kristaps  933:
                    934:        rr = r->last->rule;
1.139     kristaps  935:        roffnode_cleanscope(r);
1.82      kristaps  936:
1.139     kristaps  937:        /*
                    938:         * If the macro is unknown, first check if it contains a closing
                    939:         * delimiter `\}'.  If it does, close out our scope and return
                    940:         * the currently-scoped rule (ignore or continue).  Else, drop
                    941:         * into the currently-scoped rule.
1.87      kristaps  942:         */
                    943:
1.106     kristaps  944:        if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos))) {
1.139     kristaps  945:                ep = &(*bufp)[pos];
                    946:                for ( ; NULL != (ep = strchr(ep, '\\')); ep++) {
                    947:                        ep++;
                    948:                        if ('}' != *ep)
                    949:                                continue;
1.144     kristaps  950:
                    951:                        /*
                    952:                         * Make the \} go away.
                    953:                         * This is a little haphazard, as it's not quite
                    954:                         * clear how nroff does this.
                    955:                         * If we're at the end of line, then just chop
                    956:                         * off the \} and resize the buffer.
                    957:                         * If we aren't, then conver it to spaces.
                    958:                         */
                    959:
                    960:                        if ('\0' == *(ep + 1)) {
                    961:                                *--ep = '\0';
                    962:                                *szp -= 2;
                    963:                        } else
                    964:                                *(ep - 1) = *ep = ' ';
                    965:
1.139     kristaps  966:                        roff_ccond(r, ROFF_ccond, bufp, szp,
                    967:                                        ln, pos, pos + 2, offs);
                    968:                        break;
                    969:                }
1.82      kristaps  970:                return(ROFFRULE_DENY == rr ? ROFF_IGN : ROFF_CONT);
1.100     kristaps  971:        }
1.82      kristaps  972:
                    973:        /*
                    974:         * A denied conditional must evaluate its children if and only
                    975:         * if they're either structurally required (such as loops and
                    976:         * conditionals) or a closing macro.
                    977:         */
1.139     kristaps  978:
1.82      kristaps  979:        if (ROFFRULE_DENY == rr)
                    980:                if ( ! (ROFFMAC_STRUCT & roffs[t].flags))
                    981:                        if (ROFF_ccond != t)
                    982:                                return(ROFF_IGN);
                    983:
                    984:        assert(roffs[t].proc);
1.90      kristaps  985:        return((*roffs[t].proc)(r, t, bufp, szp,
                    986:                                ln, ppos, pos, offs));
1.82      kristaps  987: }
                    988:
                    989: /* ARGSUSED */
                    990: static enum rofferr
                    991: roff_cond_text(ROFF_ARGS)
1.78      kristaps  992: {
1.140     kristaps  993:        char            *ep;
1.82      kristaps  994:        enum roffrule    rr;
                    995:
                    996:        rr = r->last->rule;
1.140     kristaps  997:        roffnode_cleanscope(r);
1.82      kristaps  998:
1.140     kristaps  999:        ep = &(*bufp)[pos];
                   1000:        for ( ; NULL != (ep = strchr(ep, '\\')); ep++) {
                   1001:                ep++;
                   1002:                if ('}' != *ep)
                   1003:                        continue;
                   1004:                *ep = '&';
                   1005:                roff_ccond(r, ROFF_ccond, bufp, szp,
                   1006:                                ln, pos, pos + 2, offs);
1.78      kristaps 1007:        }
1.82      kristaps 1008:        return(ROFFRULE_DENY == rr ? ROFF_IGN : ROFF_CONT);
1.74      kristaps 1009: }
                   1010:
1.88      kristaps 1011: static enum roffrule
                   1012: roff_evalcond(const char *v, int *pos)
                   1013: {
                   1014:
                   1015:        switch (v[*pos]) {
                   1016:        case ('n'):
                   1017:                (*pos)++;
                   1018:                return(ROFFRULE_ALLOW);
                   1019:        case ('e'):
                   1020:                /* FALLTHROUGH */
                   1021:        case ('o'):
                   1022:                /* FALLTHROUGH */
                   1023:        case ('t'):
                   1024:                (*pos)++;
                   1025:                return(ROFFRULE_DENY);
                   1026:        default:
                   1027:                break;
                   1028:        }
                   1029:
                   1030:        while (v[*pos] && ' ' != v[*pos])
                   1031:                (*pos)++;
                   1032:        return(ROFFRULE_DENY);
                   1033: }
                   1034:
1.75      kristaps 1035: /* ARGSUSED */
1.74      kristaps 1036: static enum rofferr
1.103     kristaps 1037: roff_line_ignore(ROFF_ARGS)
1.89      kristaps 1038: {
1.123     schwarze 1039:
                   1040:        if (ROFF_it == tok)
1.128     kristaps 1041:                mandoc_msg(MANDOCERR_REQUEST, r->parse, ln, ppos, "it");
1.89      kristaps 1042:
                   1043:        return(ROFF_IGN);
                   1044: }
                   1045:
1.104     kristaps 1046: /* ARGSUSED */
                   1047: static enum rofferr
1.82      kristaps 1048: roff_cond(ROFF_ARGS)
1.74      kristaps 1049: {
1.77      kristaps 1050:        int              sv;
1.88      kristaps 1051:        enum roffrule    rule;
1.74      kristaps 1052:
1.134     kristaps 1053:        /*
                   1054:         * An `.el' has no conditional body: it will consume the value
                   1055:         * of the current rstack entry set in prior `ie' calls or
                   1056:         * defaults to DENY.
                   1057:         *
                   1058:         * If we're not an `el', however, then evaluate the conditional.
                   1059:         */
1.133     kristaps 1060:
1.134     kristaps 1061:        rule = ROFF_el == tok ?
                   1062:                (r->rstackpos < 0 ?
                   1063:                 ROFFRULE_DENY : r->rstack[r->rstackpos--]) :
                   1064:                roff_evalcond(*bufp, &pos);
1.77      kristaps 1065:
                   1066:        sv = pos;
1.75      kristaps 1067:        while (' ' == (*bufp)[pos])
                   1068:                pos++;
1.74      kristaps 1069:
1.77      kristaps 1070:        /*
                   1071:         * Roff is weird.  If we have just white-space after the
                   1072:         * conditional, it's considered the BODY and we exit without
                   1073:         * really doing anything.  Warn about this.  It's probably
                   1074:         * wrong.
                   1075:         */
1.88      kristaps 1076:
1.77      kristaps 1077:        if ('\0' == (*bufp)[pos] && sv != pos) {
1.128     kristaps 1078:                mandoc_msg(MANDOCERR_NOARGS, r->parse, ln, ppos, NULL);
1.107     kristaps 1079:                return(ROFF_IGN);
1.77      kristaps 1080:        }
                   1081:
1.106     kristaps 1082:        roffnode_push(r, tok, NULL, ln, ppos);
1.77      kristaps 1083:
1.88      kristaps 1084:        r->last->rule = rule;
                   1085:
1.134     kristaps 1086:        /*
                   1087:         * An if-else will put the NEGATION of the current evaluated
                   1088:         * conditional into the stack of rules.
                   1089:         */
                   1090:
1.84      schwarze 1091:        if (ROFF_ie == tok) {
1.134     kristaps 1092:                if (r->rstackpos == RSTACK_MAX - 1) {
                   1093:                        mandoc_msg(MANDOCERR_MEM,
                   1094:                                r->parse, ln, ppos, NULL);
                   1095:                        return(ROFF_ERR);
                   1096:                }
                   1097:                r->rstack[++r->rstackpos] =
                   1098:                        ROFFRULE_DENY == r->last->rule ?
                   1099:                        ROFFRULE_ALLOW : ROFFRULE_DENY;
1.82      kristaps 1100:        }
1.88      kristaps 1101:
                   1102:        /* If the parent has false as its rule, then so do we. */
                   1103:
1.109     kristaps 1104:        if (r->last->parent && ROFFRULE_DENY == r->last->parent->rule)
1.84      schwarze 1105:                r->last->rule = ROFFRULE_DENY;
1.88      kristaps 1106:
                   1107:        /*
                   1108:         * Determine scope.  If we're invoked with "\{" trailing the
                   1109:         * conditional, then we're in a multiline scope.  Else our scope
                   1110:         * expires on the next line.
                   1111:         */
1.74      kristaps 1112:
1.75      kristaps 1113:        r->last->endspan = 1;
                   1114:
                   1115:        if ('\\' == (*bufp)[pos] && '{' == (*bufp)[pos + 1]) {
                   1116:                r->last->endspan = -1;
                   1117:                pos += 2;
1.109     kristaps 1118:        }
1.74      kristaps 1119:
1.77      kristaps 1120:        /*
                   1121:         * If there are no arguments on the line, the next-line scope is
                   1122:         * assumed.
                   1123:         */
                   1124:
1.75      kristaps 1125:        if ('\0' == (*bufp)[pos])
                   1126:                return(ROFF_IGN);
1.77      kristaps 1127:
                   1128:        /* Otherwise re-run the roff parser after recalculating. */
1.74      kristaps 1129:
1.75      kristaps 1130:        *offs = pos;
                   1131:        return(ROFF_RERUN);
1.83      schwarze 1132: }
                   1133:
                   1134:
                   1135: /* ARGSUSED */
                   1136: static enum rofferr
1.92      schwarze 1137: roff_ds(ROFF_ARGS)
                   1138: {
1.96      kristaps 1139:        char            *name, *string;
                   1140:
                   1141:        /*
                   1142:         * A symbol is named by the first word following the macro
                   1143:         * invocation up to a space.  Its value is anything after the
                   1144:         * name's trailing whitespace and optional double-quote.  Thus,
                   1145:         *
                   1146:         *  [.ds foo "bar  "     ]
                   1147:         *
                   1148:         * will have `bar  "     ' as its value.
                   1149:         */
1.92      schwarze 1150:
1.121     schwarze 1151:        string = *bufp + pos;
                   1152:        name = roff_getname(r, &string, ln, pos);
1.92      schwarze 1153:        if ('\0' == *name)
                   1154:                return(ROFF_IGN);
                   1155:
1.121     schwarze 1156:        /* Read past initial double-quote. */
                   1157:        if ('"' == *string)
1.92      schwarze 1158:                string++;
                   1159:
1.96      kristaps 1160:        /* The rest is the value. */
1.106     kristaps 1161:        roff_setstr(r, name, string, 0);
1.92      schwarze 1162:        return(ROFF_IGN);
                   1163: }
                   1164:
1.147     kristaps 1165: int
                   1166: roff_regisset(const struct roff *r, enum regs reg)
                   1167: {
                   1168:
                   1169:        return(r->regs[(int)reg].set);
                   1170: }
                   1171:
                   1172: unsigned int
                   1173: roff_regget(const struct roff *r, enum regs reg)
                   1174: {
                   1175:
                   1176:        return(r->regs[(int)reg].u);
                   1177: }
                   1178:
                   1179: void
                   1180: roff_regunset(struct roff *r, enum regs reg)
                   1181: {
                   1182:
                   1183:        r->regs[(int)reg].set = 0;
                   1184: }
1.92      schwarze 1185:
                   1186: /* ARGSUSED */
                   1187: static enum rofferr
1.89      kristaps 1188: roff_nr(ROFF_ARGS)
1.83      schwarze 1189: {
1.121     schwarze 1190:        const char      *key;
                   1191:        char            *val;
1.138     kristaps 1192:        int              iv;
1.89      kristaps 1193:
1.121     schwarze 1194:        val = *bufp + pos;
                   1195:        key = roff_getname(r, &val, ln, pos);
1.89      kristaps 1196:
                   1197:        if (0 == strcmp(key, "nS")) {
1.147     kristaps 1198:                r->regs[(int)REG_nS].set = 1;
1.149     kristaps 1199:                if ((iv = mandoc_strntoi(val, strlen(val), 10)) >= 0)
1.147     kristaps 1200:                        r->regs[(int)REG_nS].u = (unsigned)iv;
1.138     kristaps 1201:                else
1.147     kristaps 1202:                        r->regs[(int)REG_nS].u = 0u;
1.109     kristaps 1203:        }
                   1204:
1.122     schwarze 1205:        return(ROFF_IGN);
                   1206: }
                   1207:
                   1208: /* ARGSUSED */
                   1209: static enum rofferr
                   1210: roff_rm(ROFF_ARGS)
                   1211: {
                   1212:        const char       *name;
                   1213:        char             *cp;
                   1214:
                   1215:        cp = *bufp + pos;
                   1216:        while ('\0' != *cp) {
1.127     kristaps 1217:                name = roff_getname(r, &cp, ln, (int)(cp - *bufp));
1.122     schwarze 1218:                if ('\0' != *name)
                   1219:                        roff_setstr(r, name, NULL, 0);
                   1220:        }
1.109     kristaps 1221:        return(ROFF_IGN);
                   1222: }
                   1223:
                   1224: /* ARGSUSED */
                   1225: static enum rofferr
                   1226: roff_TE(ROFF_ARGS)
                   1227: {
                   1228:
                   1229:        if (NULL == r->tbl)
1.128     kristaps 1230:                mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.115     kristaps 1231:        else
1.151     kristaps 1232:                tbl_end(&r->tbl);
1.109     kristaps 1233:
1.112     kristaps 1234:        return(ROFF_IGN);
                   1235: }
                   1236:
                   1237: /* ARGSUSED */
                   1238: static enum rofferr
                   1239: roff_T_(ROFF_ARGS)
                   1240: {
                   1241:
                   1242:        if (NULL == r->tbl)
1.128     kristaps 1243:                mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.112     kristaps 1244:        else
1.116     kristaps 1245:                tbl_restart(ppos, ln, r->tbl);
1.112     kristaps 1246:
1.109     kristaps 1247:        return(ROFF_IGN);
                   1248: }
                   1249:
1.151     kristaps 1250: int
                   1251: roff_closeeqn(struct roff *r)
                   1252: {
                   1253:
                   1254:        return(r->eqn && ROFF_EQN == eqn_end(&r->eqn) ? 1 : 0);
                   1255: }
                   1256:
                   1257: void
                   1258: roff_openeqn(struct roff *r, const char *name, int line,
                   1259:                int offs, const char *buf)
1.125     kristaps 1260: {
1.151     kristaps 1261:        struct eqn_node *e;
                   1262:        int              poff;
1.125     kristaps 1263:
                   1264:        assert(NULL == r->eqn);
1.151     kristaps 1265:        e = eqn_alloc(name, offs, line, r->parse);
1.125     kristaps 1266:
                   1267:        if (r->last_eqn)
                   1268:                r->last_eqn->next = e;
                   1269:        else
                   1270:                r->first_eqn = r->last_eqn = e;
                   1271:
                   1272:        r->eqn = r->last_eqn = e;
1.151     kristaps 1273:
                   1274:        if (buf) {
                   1275:                poff = 0;
                   1276:                eqn_read(&r->eqn, line, buf, offs, &poff);
                   1277:        }
                   1278: }
                   1279:
                   1280: /* ARGSUSED */
                   1281: static enum rofferr
                   1282: roff_EQ(ROFF_ARGS)
                   1283: {
                   1284:
                   1285:        roff_openeqn(r, *bufp + pos, ln, ppos, NULL);
1.125     kristaps 1286:        return(ROFF_IGN);
                   1287: }
                   1288:
                   1289: /* ARGSUSED */
                   1290: static enum rofferr
                   1291: roff_EN(ROFF_ARGS)
                   1292: {
                   1293:
1.128     kristaps 1294:        mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.125     kristaps 1295:        return(ROFF_IGN);
                   1296: }
                   1297:
                   1298: /* ARGSUSED */
                   1299: static enum rofferr
1.109     kristaps 1300: roff_TS(ROFF_ARGS)
                   1301: {
1.118     kristaps 1302:        struct tbl_node *t;
1.89      kristaps 1303:
1.115     kristaps 1304:        if (r->tbl) {
1.128     kristaps 1305:                mandoc_msg(MANDOCERR_SCOPEBROKEN, r->parse, ln, ppos, NULL);
1.151     kristaps 1306:                tbl_end(&r->tbl);
1.115     kristaps 1307:        }
1.83      schwarze 1308:
1.128     kristaps 1309:        t = tbl_alloc(ppos, ln, r->parse);
1.113     kristaps 1310:
                   1311:        if (r->last_tbl)
                   1312:                r->last_tbl->next = t;
                   1313:        else
                   1314:                r->first_tbl = r->last_tbl = t;
                   1315:
                   1316:        r->tbl = r->last_tbl = t;
1.83      schwarze 1317:        return(ROFF_IGN);
1.92      schwarze 1318: }
                   1319:
1.105     kristaps 1320: /* ARGSUSED */
                   1321: static enum rofferr
                   1322: roff_so(ROFF_ARGS)
                   1323: {
                   1324:        char *name;
                   1325:
1.128     kristaps 1326:        mandoc_msg(MANDOCERR_SO, r->parse, ln, ppos, NULL);
1.105     kristaps 1327:
                   1328:        /*
                   1329:         * Handle `so'.  Be EXTREMELY careful, as we shouldn't be
                   1330:         * opening anything that's not in our cwd or anything beneath
                   1331:         * it.  Thus, explicitly disallow traversing up the file-system
                   1332:         * or using absolute paths.
                   1333:         */
                   1334:
                   1335:        name = *bufp + pos;
                   1336:        if ('/' == *name || strstr(name, "../") || strstr(name, "/..")) {
1.128     kristaps 1337:                mandoc_msg(MANDOCERR_SOPATH, r->parse, ln, pos, NULL);
1.105     kristaps 1338:                return(ROFF_ERR);
                   1339:        }
                   1340:
                   1341:        *offs = pos;
                   1342:        return(ROFF_SO);
                   1343: }
1.92      schwarze 1344:
1.106     kristaps 1345: /* ARGSUSED */
                   1346: static enum rofferr
                   1347: roff_userdef(ROFF_ARGS)
1.99      kristaps 1348: {
1.106     kristaps 1349:        const char       *arg[9];
                   1350:        char             *cp, *n1, *n2;
1.119     schwarze 1351:        int               i;
1.106     kristaps 1352:
                   1353:        /*
                   1354:         * Collect pointers to macro argument strings
                   1355:         * and null-terminate them.
                   1356:         */
                   1357:        cp = *bufp + pos;
1.119     schwarze 1358:        for (i = 0; i < 9; i++)
1.120     schwarze 1359:                arg[i] = '\0' == *cp ? "" :
1.136     kristaps 1360:                    mandoc_getarg(r->parse, &cp, ln, &pos);
1.99      kristaps 1361:
1.106     kristaps 1362:        /*
                   1363:         * Expand macro arguments.
1.99      kristaps 1364:         */
1.106     kristaps 1365:        *szp = 0;
                   1366:        n1 = cp = mandoc_strdup(r->current_string);
                   1367:        while (NULL != (cp = strstr(cp, "\\$"))) {
                   1368:                i = cp[2] - '1';
                   1369:                if (0 > i || 8 < i) {
                   1370:                        /* Not an argument invocation. */
                   1371:                        cp += 2;
                   1372:                        continue;
                   1373:                }
                   1374:
                   1375:                *szp = strlen(n1) - 3 + strlen(arg[i]) + 1;
                   1376:                n2 = mandoc_malloc(*szp);
                   1377:
                   1378:                strlcpy(n2, n1, (size_t)(cp - n1 + 1));
                   1379:                strlcat(n2, arg[i], *szp);
                   1380:                strlcat(n2, cp + 3, *szp);
                   1381:
                   1382:                cp = n2 + (cp - n1);
                   1383:                free(n1);
                   1384:                n1 = n2;
1.99      kristaps 1385:        }
                   1386:
1.106     kristaps 1387:        /*
                   1388:         * Replace the macro invocation
                   1389:         * by the expanded macro.
                   1390:         */
                   1391:        free(*bufp);
                   1392:        *bufp = n1;
                   1393:        if (0 == *szp)
                   1394:                *szp = strlen(*bufp) + 1;
                   1395:
                   1396:        return(*szp > 1 && '\n' == (*bufp)[(int)*szp - 2] ?
                   1397:           ROFF_REPARSE : ROFF_APPEND);
1.99      kristaps 1398: }
1.121     schwarze 1399:
                   1400: static char *
                   1401: roff_getname(struct roff *r, char **cpp, int ln, int pos)
                   1402: {
                   1403:        char     *name, *cp;
                   1404:
                   1405:        name = *cpp;
                   1406:        if ('\0' == *name)
                   1407:                return(name);
                   1408:
                   1409:        /* Read until end of name. */
                   1410:        for (cp = name; '\0' != *cp && ' ' != *cp; cp++) {
                   1411:                if ('\\' != *cp)
                   1412:                        continue;
                   1413:                cp++;
                   1414:                if ('\\' == *cp)
                   1415:                        continue;
1.128     kristaps 1416:                mandoc_msg(MANDOCERR_NAMESC, r->parse, ln, pos, NULL);
1.121     schwarze 1417:                *cp = '\0';
                   1418:                name = cp;
                   1419:        }
                   1420:
                   1421:        /* Nil-terminate name. */
                   1422:        if ('\0' != *cp)
                   1423:                *(cp++) = '\0';
                   1424:
                   1425:        /* Read past spaces. */
                   1426:        while (' ' == *cp)
                   1427:                cp++;
                   1428:
                   1429:        *cpp = cp;
                   1430:        return(name);
                   1431: }
                   1432:
1.106     kristaps 1433: /*
                   1434:  * Store *string into the user-defined string called *name.
                   1435:  * In multiline mode, append to an existing entry and append '\n';
                   1436:  * else replace the existing entry, if there is one.
                   1437:  * To clear an existing entry, call with (*r, *name, NULL, 0).
                   1438:  */
1.94      kristaps 1439: static void
1.106     kristaps 1440: roff_setstr(struct roff *r, const char *name, const char *string,
                   1441:        int multiline)
1.92      schwarze 1442: {
                   1443:        struct roffstr   *n;
1.106     kristaps 1444:        char             *c;
                   1445:        size_t            oldch, newch;
1.92      schwarze 1446:
1.106     kristaps 1447:        /* Search for an existing string with the same name. */
1.94      kristaps 1448:        n = r->first_string;
1.92      schwarze 1449:        while (n && strcmp(name, n->name))
                   1450:                n = n->next;
1.94      kristaps 1451:
                   1452:        if (NULL == n) {
1.106     kristaps 1453:                /* Create a new string table entry. */
1.94      kristaps 1454:                n = mandoc_malloc(sizeof(struct roffstr));
1.106     kristaps 1455:                n->name = mandoc_strdup(name);
                   1456:                n->string = NULL;
1.94      kristaps 1457:                n->next = r->first_string;
                   1458:                r->first_string = n;
1.106     kristaps 1459:        } else if (0 == multiline) {
                   1460:                /* In multiline mode, append; else replace. */
1.92      schwarze 1461:                free(n->string);
1.106     kristaps 1462:                n->string = NULL;
                   1463:        }
                   1464:
                   1465:        if (NULL == string)
                   1466:                return;
                   1467:
                   1468:        /*
                   1469:         * One additional byte for the '\n' in multiline mode,
                   1470:         * and one for the terminating '\0'.
                   1471:         */
1.127     kristaps 1472:        newch = strlen(string) + (multiline ? 2u : 1u);
1.106     kristaps 1473:        if (NULL == n->string) {
                   1474:                n->string = mandoc_malloc(newch);
                   1475:                *n->string = '\0';
                   1476:                oldch = 0;
                   1477:        } else {
                   1478:                oldch = strlen(n->string);
                   1479:                n->string = mandoc_realloc(n->string, oldch + newch);
                   1480:        }
                   1481:
                   1482:        /* Skip existing content in the destination buffer. */
1.127     kristaps 1483:        c = n->string + (int)oldch;
1.106     kristaps 1484:
                   1485:        /* Append new content to the destination buffer. */
                   1486:        while (*string) {
                   1487:                /*
                   1488:                 * Rudimentary roff copy mode:
                   1489:                 * Handle escaped backslashes.
                   1490:                 */
                   1491:                if ('\\' == *string && '\\' == *(string + 1))
                   1492:                        string++;
                   1493:                *c++ = *string++;
                   1494:        }
1.94      kristaps 1495:
1.106     kristaps 1496:        /* Append terminating bytes. */
                   1497:        if (multiline)
                   1498:                *c++ = '\n';
                   1499:        *c = '\0';
1.92      schwarze 1500: }
                   1501:
1.94      kristaps 1502: static const char *
                   1503: roff_getstrn(const struct roff *r, const char *name, size_t len)
1.92      schwarze 1504: {
1.94      kristaps 1505:        const struct roffstr *n;
1.92      schwarze 1506:
1.94      kristaps 1507:        n = r->first_string;
1.97      kristaps 1508:        while (n && (strncmp(name, n->name, len) || '\0' != n->name[(int)len]))
1.92      schwarze 1509:                n = n->next;
1.94      kristaps 1510:
                   1511:        return(n ? n->string : NULL);
1.92      schwarze 1512: }
                   1513:
1.94      kristaps 1514: static void
                   1515: roff_freestr(struct roff *r)
1.92      schwarze 1516: {
                   1517:        struct roffstr   *n, *nn;
                   1518:
1.94      kristaps 1519:        for (n = r->first_string; n; n = nn) {
1.92      schwarze 1520:                free(n->name);
                   1521:                free(n->string);
                   1522:                nn = n->next;
                   1523:                free(n);
                   1524:        }
1.94      kristaps 1525:
                   1526:        r->first_string = NULL;
1.114     kristaps 1527: }
                   1528:
                   1529: const struct tbl_span *
                   1530: roff_span(const struct roff *r)
                   1531: {
                   1532:
                   1533:        return(r->tbl ? tbl_span(r->tbl) : NULL);
1.125     kristaps 1534: }
                   1535:
                   1536: const struct eqn *
                   1537: roff_eqn(const struct roff *r)
                   1538: {
                   1539:
                   1540:        return(r->last_eqn ? &r->last_eqn->eqn : NULL);
1.151     kristaps 1541: }
                   1542:
                   1543: char
                   1544: roff_eqndelim(const struct roff *r)
                   1545: {
                   1546:
                   1547:        return('\0');
1.74      kristaps 1548: }

CVSweb