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

Annotation of mandoc/roff.c, Revision 1.161

1.161   ! kristaps    1: /*     $Id: roff.c,v 1.160 2011/07/27 14:23:27 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 {
1.160     kristaps   85:        char            *key; /* key of symbol */
                     86:        char            *val; /* current value */
1.94      kristaps   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:
1.155     kristaps  146: static enum rofft       roffhash_find(const char *, size_t);
                    147: static void             roffhash_init(void);
                    148: static void             roffnode_cleanscope(struct roff *);
                    149: static void             roffnode_pop(struct roff *);
                    150: static void             roffnode_push(struct roff *, enum rofft,
                    151:                                const char *, int, int);
1.80      kristaps  152: static enum rofferr     roff_block(ROFF_ARGS);
                    153: static enum rofferr     roff_block_text(ROFF_ARGS);
                    154: static enum rofferr     roff_block_sub(ROFF_ARGS);
                    155: static enum rofferr     roff_cblock(ROFF_ARGS);
                    156: static enum rofferr     roff_ccond(ROFF_ARGS);
1.82      kristaps  157: static enum rofferr     roff_cond(ROFF_ARGS);
                    158: static enum rofferr     roff_cond_text(ROFF_ARGS);
                    159: static enum rofferr     roff_cond_sub(ROFF_ARGS);
1.92      schwarze  160: static enum rofferr     roff_ds(ROFF_ARGS);
1.94      kristaps  161: static enum roffrule    roff_evalcond(const char *, int *);
1.155     kristaps  162: static void             roff_free1(struct roff *);
1.94      kristaps  163: static void             roff_freestr(struct roff *);
1.121     schwarze  164: static char            *roff_getname(struct roff *, char **, int, int);
1.94      kristaps  165: static const char      *roff_getstrn(const struct roff *,
                    166:                                const char *, size_t);
1.103     kristaps  167: static enum rofferr     roff_line_ignore(ROFF_ARGS);
1.89      kristaps  168: static enum rofferr     roff_nr(ROFF_ARGS);
1.156     kristaps  169: static void             roff_openeqn(struct roff *, const char *,
                    170:                                int, int, const char *);
1.155     kristaps  171: static enum rofft       roff_parse(struct roff *, const char *, int *);
                    172: static enum rofferr     roff_parsetext(char *);
1.154     kristaps  173: static void             roff_res(struct roff *,
1.142     kristaps  174:                                char **, size_t *, int, int);
1.122     schwarze  175: static enum rofferr     roff_rm(ROFF_ARGS);
1.94      kristaps  176: static void             roff_setstr(struct roff *,
1.106     kristaps  177:                                const char *, const char *, int);
1.105     kristaps  178: static enum rofferr     roff_so(ROFF_ARGS);
1.109     kristaps  179: static enum rofferr     roff_TE(ROFF_ARGS);
                    180: static enum rofferr     roff_TS(ROFF_ARGS);
1.125     kristaps  181: static enum rofferr     roff_EQ(ROFF_ARGS);
                    182: static enum rofferr     roff_EN(ROFF_ARGS);
1.112     kristaps  183: static enum rofferr     roff_T_(ROFF_ARGS);
1.106     kristaps  184: static enum rofferr     roff_userdef(ROFF_ARGS);
1.67      kristaps  185:
1.155     kristaps  186: /* See roffhash_find() */
1.85      kristaps  187:
                    188: #define        ASCII_HI         126
                    189: #define        ASCII_LO         33
                    190: #define        HASHWIDTH       (ASCII_HI - ASCII_LO + 1)
                    191:
                    192: static struct roffmac  *hash[HASHWIDTH];
                    193:
                    194: static struct roffmac   roffs[ROFF_MAX] = {
1.103     kristaps  195:        { "ad", roff_line_ignore, NULL, NULL, 0, NULL },
1.85      kristaps  196:        { "am", roff_block, roff_block_text, roff_block_sub, 0, NULL },
                    197:        { "ami", roff_block, roff_block_text, roff_block_sub, 0, NULL },
                    198:        { "am1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
                    199:        { "de", roff_block, roff_block_text, roff_block_sub, 0, NULL },
                    200:        { "dei", roff_block, roff_block_text, roff_block_sub, 0, NULL },
                    201:        { "de1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.92      schwarze  202:        { "ds", roff_ds, NULL, NULL, 0, NULL },
1.85      kristaps  203:        { "el", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
1.103     kristaps  204:        { "hy", roff_line_ignore, NULL, NULL, 0, NULL },
1.85      kristaps  205:        { "ie", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
                    206:        { "if", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
                    207:        { "ig", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.123     schwarze  208:        { "it", roff_line_ignore, NULL, NULL, 0, NULL },
1.103     kristaps  209:        { "ne", roff_line_ignore, NULL, NULL, 0, NULL },
                    210:        { "nh", roff_line_ignore, NULL, NULL, 0, NULL },
1.104     kristaps  211:        { "nr", roff_nr, NULL, NULL, 0, NULL },
1.124     schwarze  212:        { "ns", roff_line_ignore, NULL, NULL, 0, NULL },
                    213:        { "ps", roff_line_ignore, NULL, NULL, 0, NULL },
1.122     schwarze  214:        { "rm", roff_rm, NULL, NULL, 0, NULL },
1.105     kristaps  215:        { "so", roff_so, NULL, NULL, 0, NULL },
1.124     schwarze  216:        { "ta", roff_line_ignore, NULL, NULL, 0, NULL },
1.103     kristaps  217:        { "tr", roff_line_ignore, NULL, NULL, 0, NULL },
1.109     kristaps  218:        { "TS", roff_TS, NULL, NULL, 0, NULL },
                    219:        { "TE", roff_TE, NULL, NULL, 0, NULL },
1.112     kristaps  220:        { "T&", roff_T_, NULL, NULL, 0, NULL },
1.125     kristaps  221:        { "EQ", roff_EQ, NULL, NULL, 0, NULL },
                    222:        { "EN", roff_EN, NULL, NULL, 0, NULL },
1.85      kristaps  223:        { ".", roff_cblock, NULL, NULL, 0, NULL },
                    224:        { "\\}", roff_ccond, NULL, NULL, 0, NULL },
1.106     kristaps  225:        { NULL, roff_userdef, NULL, NULL, 0, NULL },
1.67      kristaps  226: };
                    227:
1.141     kristaps  228: /* Array of injected predefined strings. */
                    229: #define        PREDEFS_MAX      38
                    230: static const struct predef predefs[PREDEFS_MAX] = {
                    231: #include "predefs.in"
                    232: };
                    233:
1.155     kristaps  234: /* See roffhash_find() */
1.85      kristaps  235: #define        ROFF_HASH(p)    (p[0] - ASCII_LO)
                    236:
                    237: static void
1.155     kristaps  238: roffhash_init(void)
1.85      kristaps  239: {
                    240:        struct roffmac   *n;
                    241:        int               buc, i;
                    242:
1.106     kristaps  243:        for (i = 0; i < (int)ROFF_USERDEF; i++) {
1.85      kristaps  244:                assert(roffs[i].name[0] >= ASCII_LO);
                    245:                assert(roffs[i].name[0] <= ASCII_HI);
                    246:
                    247:                buc = ROFF_HASH(roffs[i].name);
                    248:
                    249:                if (NULL != (n = hash[buc])) {
                    250:                        for ( ; n->next; n = n->next)
                    251:                                /* Do nothing. */ ;
                    252:                        n->next = &roffs[i];
                    253:                } else
                    254:                        hash[buc] = &roffs[i];
                    255:        }
                    256: }
                    257:
1.67      kristaps  258: /*
                    259:  * Look up a roff token by its name.  Returns ROFF_MAX if no macro by
                    260:  * the nil-terminated string name could be found.
                    261:  */
                    262: static enum rofft
1.155     kristaps  263: roffhash_find(const char *p, size_t s)
1.67      kristaps  264: {
1.85      kristaps  265:        int              buc;
                    266:        struct roffmac  *n;
1.67      kristaps  267:
1.85      kristaps  268:        /*
                    269:         * libroff has an extremely simple hashtable, for the time
                    270:         * being, which simply keys on the first character, which must
                    271:         * be printable, then walks a chain.  It works well enough until
                    272:         * optimised.
                    273:         */
                    274:
                    275:        if (p[0] < ASCII_LO || p[0] > ASCII_HI)
                    276:                return(ROFF_MAX);
                    277:
                    278:        buc = ROFF_HASH(p);
                    279:
                    280:        if (NULL == (n = hash[buc]))
                    281:                return(ROFF_MAX);
                    282:        for ( ; n; n = n->next)
1.106     kristaps  283:                if (0 == strncmp(n->name, p, s) && '\0' == n->name[(int)s])
1.85      kristaps  284:                        return((enum rofft)(n - roffs));
1.67      kristaps  285:
                    286:        return(ROFF_MAX);
                    287: }
                    288:
                    289:
                    290: /*
                    291:  * Pop the current node off of the stack of roff instructions currently
                    292:  * pending.
                    293:  */
                    294: static void
                    295: roffnode_pop(struct roff *r)
                    296: {
                    297:        struct roffnode *p;
                    298:
1.75      kristaps  299:        assert(r->last);
                    300:        p = r->last;
1.82      kristaps  301:
1.75      kristaps  302:        r->last = r->last->parent;
1.106     kristaps  303:        free(p->name);
                    304:        free(p->end);
1.67      kristaps  305:        free(p);
                    306: }
                    307:
                    308:
                    309: /*
                    310:  * Push a roff node onto the instruction stack.  This must later be
                    311:  * removed with roffnode_pop().
                    312:  */
1.98      schwarze  313: static void
1.106     kristaps  314: roffnode_push(struct roff *r, enum rofft tok, const char *name,
                    315:                int line, int col)
1.67      kristaps  316: {
                    317:        struct roffnode *p;
                    318:
1.98      schwarze  319:        p = mandoc_calloc(1, sizeof(struct roffnode));
1.67      kristaps  320:        p->tok = tok;
1.106     kristaps  321:        if (name)
                    322:                p->name = mandoc_strdup(name);
1.67      kristaps  323:        p->parent = r->last;
                    324:        p->line = line;
                    325:        p->col = col;
1.79      kristaps  326:        p->rule = p->parent ? p->parent->rule : ROFFRULE_DENY;
1.67      kristaps  327:
                    328:        r->last = p;
                    329: }
                    330:
                    331:
                    332: static void
                    333: roff_free1(struct roff *r)
                    334: {
1.118     kristaps  335:        struct tbl_node *t;
1.125     kristaps  336:        struct eqn_node *e;
1.67      kristaps  337:
1.125     kristaps  338:        while (NULL != (t = r->first_tbl)) {
1.113     kristaps  339:                r->first_tbl = t->next;
                    340:                tbl_free(t);
1.109     kristaps  341:        }
                    342:
1.113     kristaps  343:        r->first_tbl = r->last_tbl = r->tbl = NULL;
                    344:
1.125     kristaps  345:        while (NULL != (e = r->first_eqn)) {
                    346:                r->first_eqn = e->next;
                    347:                eqn_free(e);
                    348:        }
                    349:
                    350:        r->first_eqn = r->last_eqn = r->eqn = NULL;
                    351:
1.67      kristaps  352:        while (r->last)
                    353:                roffnode_pop(r);
1.109     kristaps  354:
1.94      kristaps  355:        roff_freestr(r);
1.67      kristaps  356: }
                    357:
                    358:
                    359: void
                    360: roff_reset(struct roff *r)
                    361: {
1.143     kristaps  362:        int              i;
1.67      kristaps  363:
                    364:        roff_free1(r);
1.143     kristaps  365:
1.147     kristaps  366:        memset(&r->regs, 0, sizeof(struct reg) * REG__MAX);
                    367:
1.143     kristaps  368:        for (i = 0; i < PREDEFS_MAX; i++)
                    369:                roff_setstr(r, predefs[i].name, predefs[i].str, 0);
1.67      kristaps  370: }
                    371:
                    372:
                    373: void
                    374: roff_free(struct roff *r)
                    375: {
                    376:
                    377:        roff_free1(r);
                    378:        free(r);
                    379: }
                    380:
                    381:
                    382: struct roff *
1.147     kristaps  383: roff_alloc(struct mparse *parse)
1.67      kristaps  384: {
                    385:        struct roff     *r;
1.141     kristaps  386:        int              i;
1.67      kristaps  387:
1.98      schwarze  388:        r = mandoc_calloc(1, sizeof(struct roff));
1.128     kristaps  389:        r->parse = parse;
1.82      kristaps  390:        r->rstackpos = -1;
1.85      kristaps  391:
1.155     kristaps  392:        roffhash_init();
1.141     kristaps  393:
                    394:        for (i = 0; i < PREDEFS_MAX; i++)
                    395:                roff_setstr(r, predefs[i].name, predefs[i].str, 0);
                    396:
1.67      kristaps  397:        return(r);
                    398: }
                    399:
1.94      kristaps  400: /*
                    401:  * Pre-filter each and every line for reserved words (one beginning with
                    402:  * `\*', e.g., `\*(ab').  These must be handled before the actual line
                    403:  * is processed.
1.153     kristaps  404:  * This also checks the syntax of regular escapes.
1.154     kristaps  405:  */
                    406: static void
1.142     kristaps  407: roff_res(struct roff *r, char **bufp, size_t *szp, int ln, int pos)
1.94      kristaps  408: {
1.152     kristaps  409:        enum mandoc_esc  esc;
1.108     schwarze  410:        const char      *stesc; /* start of an escape sequence ('\\') */
                    411:        const char      *stnam; /* start of the name, after "[(*" */
                    412:        const char      *cp;    /* end of the name, e.g. before ']' */
                    413:        const char      *res;   /* the string to be substituted */
1.94      kristaps  414:        int              i, maxl;
                    415:        size_t           nsz;
                    416:        char            *n;
                    417:
1.154     kristaps  418: again:
1.108     schwarze  419:        cp = *bufp + pos;
                    420:        while (NULL != (cp = strchr(cp, '\\'))) {
                    421:                stesc = cp++;
                    422:
                    423:                /*
                    424:                 * The second character must be an asterisk.
                    425:                 * If it isn't, skip it anyway:  It is escaped,
                    426:                 * so it can't start another escape sequence.
                    427:                 */
                    428:
                    429:                if ('\0' == *cp)
1.154     kristaps  430:                        return;
1.152     kristaps  431:
                    432:                if ('*' != *cp) {
                    433:                        res = cp;
                    434:                        esc = mandoc_escape(&cp, NULL, NULL);
                    435:                        if (ESCAPE_ERROR != esc)
                    436:                                continue;
                    437:                        cp = res;
1.153     kristaps  438:                        mandoc_msg
                    439:                                (MANDOCERR_BADESCAPE, r->parse,
                    440:                                 ln, (int)(stesc - *bufp), NULL);
1.154     kristaps  441:                        return;
1.152     kristaps  442:                }
                    443:
                    444:                cp++;
1.108     schwarze  445:
                    446:                /*
                    447:                 * The third character decides the length
                    448:                 * of the name of the string.
                    449:                 * Save a pointer to the name.
                    450:                 */
                    451:
1.94      kristaps  452:                switch (*cp) {
1.108     schwarze  453:                case ('\0'):
1.154     kristaps  454:                        return;
1.94      kristaps  455:                case ('('):
                    456:                        cp++;
                    457:                        maxl = 2;
                    458:                        break;
                    459:                case ('['):
                    460:                        cp++;
                    461:                        maxl = 0;
                    462:                        break;
                    463:                default:
                    464:                        maxl = 1;
                    465:                        break;
                    466:                }
1.108     schwarze  467:                stnam = cp;
1.94      kristaps  468:
1.108     schwarze  469:                /* Advance to the end of the name. */
1.94      kristaps  470:
                    471:                for (i = 0; 0 == maxl || i < maxl; i++, cp++) {
1.153     kristaps  472:                        if ('\0' == *cp) {
                    473:                                mandoc_msg
                    474:                                        (MANDOCERR_BADESCAPE,
                    475:                                         r->parse, ln,
                    476:                                         (int)(stesc - *bufp), NULL);
1.154     kristaps  477:                                return;
1.153     kristaps  478:                        }
1.94      kristaps  479:                        if (0 == maxl && ']' == *cp)
                    480:                                break;
                    481:                }
                    482:
1.108     schwarze  483:                /*
                    484:                 * Retrieve the replacement string; if it is
                    485:                 * undefined, resume searching for escapes.
                    486:                 */
                    487:
                    488:                res = roff_getstrn(r, stnam, (size_t)i);
1.94      kristaps  489:
                    490:                if (NULL == res) {
1.153     kristaps  491:                        mandoc_msg
                    492:                                (MANDOCERR_BADESCAPE, r->parse,
                    493:                                 ln, (int)(stesc - *bufp), NULL);
1.142     kristaps  494:                        res = "";
1.94      kristaps  495:                }
                    496:
1.108     schwarze  497:                /* Replace the escape sequence by the string. */
                    498:
1.161   ! kristaps  499:                pos = stesc - *bufp;
1.154     kristaps  500:
1.94      kristaps  501:                nsz = *szp + strlen(res) + 1;
                    502:                n = mandoc_malloc(nsz);
                    503:
1.108     schwarze  504:                strlcpy(n, *bufp, (size_t)(stesc - *bufp + 1));
1.94      kristaps  505:                strlcat(n, res, nsz);
                    506:                strlcat(n, cp + (maxl ? 0 : 1), nsz);
                    507:
                    508:                free(*bufp);
                    509:
                    510:                *bufp = n;
                    511:                *szp = nsz;
1.154     kristaps  512:                goto again;
                    513:        }
                    514: }
                    515:
                    516: /*
                    517:  * Process text streams: convert all breakable hyphens into ASCII_HYPH.
                    518:  */
                    519: static enum rofferr
                    520: roff_parsetext(char *p)
                    521: {
1.155     kristaps  522:        char             l, r;
1.154     kristaps  523:        size_t           sz;
                    524:        const char      *start;
                    525:        enum mandoc_esc  esc;
                    526:
                    527:        start = p;
                    528:
                    529:        while ('\0' != *p) {
                    530:                sz = strcspn(p, "-\\");
                    531:                p += sz;
                    532:
1.159     kristaps  533:                if ('\0' == *p)
                    534:                        break;
                    535:
1.154     kristaps  536:                if ('\\' == *p) {
                    537:                        /* Skip over escapes. */
                    538:                        p++;
                    539:                        esc = mandoc_escape
                    540:                                ((const char **)&p, NULL, NULL);
                    541:                        if (ESCAPE_ERROR == esc)
                    542:                                break;
1.155     kristaps  543:                        continue;
1.159     kristaps  544:                } else if (p == start) {
1.158     kristaps  545:                        p++;
1.155     kristaps  546:                        continue;
1.158     kristaps  547:                }
1.155     kristaps  548:
                    549:                l = *(p - 1);
                    550:                r = *(p + 1);
                    551:                if ('\\' != l &&
                    552:                                '\t' != r && '\t' != l &&
                    553:                                ' ' != r && ' ' != l &&
                    554:                                '-' != r && '-' != l &&
                    555:                                ! isdigit((unsigned char)l) &&
1.159     kristaps  556:                                ! isdigit((unsigned char)r))
1.155     kristaps  557:                        *p = ASCII_HYPH;
                    558:                p++;
1.94      kristaps  559:        }
                    560:
1.154     kristaps  561:        return(ROFF_CONT);
1.94      kristaps  562: }
                    563:
1.67      kristaps  564: enum rofferr
1.90      kristaps  565: roff_parseln(struct roff *r, int ln, char **bufp,
                    566:                size_t *szp, int pos, int *offs)
1.67      kristaps  567: {
                    568:        enum rofft       t;
1.109     kristaps  569:        enum rofferr     e;
1.130     kristaps  570:        int              ppos, ctl;
1.79      kristaps  571:
                    572:        /*
1.94      kristaps  573:         * Run the reserved-word filter only if we have some reserved
                    574:         * words to fill in.
                    575:         */
                    576:
1.154     kristaps  577:        roff_res(r, bufp, szp, ln, pos);
1.94      kristaps  578:
1.130     kristaps  579:        ppos = pos;
                    580:        ctl = mandoc_getcontrol(*bufp, &pos);
                    581:
1.94      kristaps  582:        /*
1.79      kristaps  583:         * First, if a scope is open and we're not a macro, pass the
                    584:         * text through the macro's filter.  If a scope isn't open and
                    585:         * we're not a macro, just let it through.
1.125     kristaps  586:         * Finally, if there's an equation scope open, divert it into it
                    587:         * no matter our state.
1.79      kristaps  588:         */
1.74      kristaps  589:
1.130     kristaps  590:        if (r->last && ! ctl) {
1.78      kristaps  591:                t = r->last->tok;
                    592:                assert(roffs[t].text);
1.109     kristaps  593:                e = (*roffs[t].text)
                    594:                        (r, t, bufp, szp, ln, pos, pos, offs);
                    595:                assert(ROFF_IGN == e || ROFF_CONT == e);
1.125     kristaps  596:                if (ROFF_CONT != e)
                    597:                        return(e);
                    598:                if (r->eqn)
1.146     kristaps  599:                        return(eqn_read(&r->eqn, ln, *bufp, pos, offs));
1.125     kristaps  600:                if (r->tbl)
1.130     kristaps  601:                        return(tbl_read(r->tbl, ln, *bufp, pos));
1.154     kristaps  602:                return(roff_parsetext(*bufp + pos));
1.130     kristaps  603:        } else if ( ! ctl) {
1.125     kristaps  604:                if (r->eqn)
1.146     kristaps  605:                        return(eqn_read(&r->eqn, ln, *bufp, pos, offs));
1.109     kristaps  606:                if (r->tbl)
1.130     kristaps  607:                        return(tbl_read(r->tbl, ln, *bufp, pos));
1.154     kristaps  608:                return(roff_parsetext(*bufp + pos));
1.125     kristaps  609:        } else if (r->eqn)
1.146     kristaps  610:                return(eqn_read(&r->eqn, ln, *bufp, ppos, offs));
1.67      kristaps  611:
1.79      kristaps  612:        /*
                    613:         * If a scope is open, go to the child handler for that macro,
                    614:         * as it may want to preprocess before doing anything with it.
1.125     kristaps  615:         * Don't do so if an equation is open.
1.79      kristaps  616:         */
1.78      kristaps  617:
1.79      kristaps  618:        if (r->last) {
                    619:                t = r->last->tok;
                    620:                assert(roffs[t].sub);
                    621:                return((*roffs[t].sub)
1.90      kristaps  622:                                (r, t, bufp, szp,
1.130     kristaps  623:                                 ln, ppos, pos, offs));
1.79      kristaps  624:        }
1.78      kristaps  625:
1.79      kristaps  626:        /*
                    627:         * Lastly, as we've no scope open, try to look up and execute
                    628:         * the new macro.  If no macro is found, simply return and let
                    629:         * the compilers handle it.
                    630:         */
1.67      kristaps  631:
1.106     kristaps  632:        if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos)))
1.79      kristaps  633:                return(ROFF_CONT);
1.67      kristaps  634:
1.75      kristaps  635:        assert(roffs[t].proc);
1.78      kristaps  636:        return((*roffs[t].proc)
1.90      kristaps  637:                        (r, t, bufp, szp,
                    638:                         ln, ppos, pos, offs));
1.74      kristaps  639: }
                    640:
                    641:
1.117     kristaps  642: void
1.74      kristaps  643: roff_endparse(struct roff *r)
                    644: {
                    645:
1.110     kristaps  646:        if (r->last)
1.128     kristaps  647:                mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.109     kristaps  648:                                r->last->line, r->last->col, NULL);
1.117     kristaps  649:
1.125     kristaps  650:        if (r->eqn) {
1.128     kristaps  651:                mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.148     kristaps  652:                                r->eqn->eqn.ln, r->eqn->eqn.pos, NULL);
1.151     kristaps  653:                eqn_end(&r->eqn);
1.125     kristaps  654:        }
                    655:
1.117     kristaps  656:        if (r->tbl) {
1.128     kristaps  657:                mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.117     kristaps  658:                                r->tbl->line, r->tbl->pos, NULL);
1.151     kristaps  659:                tbl_end(&r->tbl);
1.117     kristaps  660:        }
1.67      kristaps  661: }
                    662:
                    663: /*
                    664:  * Parse a roff node's type from the input buffer.  This must be in the
                    665:  * form of ".foo xxx" in the usual way.
                    666:  */
                    667: static enum rofft
1.106     kristaps  668: roff_parse(struct roff *r, const char *buf, int *pos)
1.67      kristaps  669: {
1.106     kristaps  670:        const char      *mac;
                    671:        size_t           maclen;
1.67      kristaps  672:        enum rofft       t;
                    673:
1.144     kristaps  674:        if ('\0' == buf[*pos] || '"' == buf[*pos] ||
                    675:                        '\t' == buf[*pos] || ' ' == buf[*pos])
1.67      kristaps  676:                return(ROFF_MAX);
                    677:
1.144     kristaps  678:        /*
                    679:         * We stop the macro parse at an escape, tab, space, or nil.
                    680:         * However, `\}' is also a valid macro, so make sure we don't
                    681:         * clobber it by seeing the `\' as the end of token.
                    682:         */
                    683:
1.106     kristaps  684:        mac = buf + *pos;
1.144     kristaps  685:        maclen = strcspn(mac + 1, " \\\t\0") + 1;
1.67      kristaps  686:
1.106     kristaps  687:        t = (r->current_string = roff_getstrn(r, mac, maclen))
1.155     kristaps  688:            ? ROFF_USERDEF : roffhash_find(mac, maclen);
1.67      kristaps  689:
1.127     kristaps  690:        *pos += (int)maclen;
1.130     kristaps  691:
1.67      kristaps  692:        while (buf[*pos] && ' ' == buf[*pos])
                    693:                (*pos)++;
                    694:
                    695:        return(t);
                    696: }
                    697:
                    698: /* ARGSUSED */
                    699: static enum rofferr
1.76      kristaps  700: roff_cblock(ROFF_ARGS)
1.67      kristaps  701: {
                    702:
1.79      kristaps  703:        /*
                    704:         * A block-close `..' should only be invoked as a child of an
                    705:         * ignore macro, otherwise raise a warning and just ignore it.
                    706:         */
                    707:
1.76      kristaps  708:        if (NULL == r->last) {
1.128     kristaps  709:                mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.76      kristaps  710:                return(ROFF_IGN);
                    711:        }
1.67      kristaps  712:
1.81      kristaps  713:        switch (r->last->tok) {
                    714:        case (ROFF_am):
                    715:                /* FALLTHROUGH */
                    716:        case (ROFF_ami):
                    717:                /* FALLTHROUGH */
                    718:        case (ROFF_am1):
                    719:                /* FALLTHROUGH */
                    720:        case (ROFF_de):
1.108     schwarze  721:                /* ROFF_de1 is remapped to ROFF_de in roff_block(). */
1.81      kristaps  722:                /* FALLTHROUGH */
                    723:        case (ROFF_dei):
                    724:                /* FALLTHROUGH */
                    725:        case (ROFF_ig):
                    726:                break;
                    727:        default:
1.128     kristaps  728:                mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.67      kristaps  729:                return(ROFF_IGN);
1.76      kristaps  730:        }
1.67      kristaps  731:
1.76      kristaps  732:        if ((*bufp)[pos])
1.128     kristaps  733:                mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.71      kristaps  734:
                    735:        roffnode_pop(r);
1.76      kristaps  736:        roffnode_cleanscope(r);
                    737:        return(ROFF_IGN);
1.71      kristaps  738:
1.67      kristaps  739: }
                    740:
                    741:
1.76      kristaps  742: static void
                    743: roffnode_cleanscope(struct roff *r)
1.67      kristaps  744: {
                    745:
1.76      kristaps  746:        while (r->last) {
                    747:                if (--r->last->endspan < 0)
                    748:                        break;
                    749:                roffnode_pop(r);
                    750:        }
1.67      kristaps  751: }
                    752:
                    753:
1.75      kristaps  754: /* ARGSUSED */
1.74      kristaps  755: static enum rofferr
1.75      kristaps  756: roff_ccond(ROFF_ARGS)
1.74      kristaps  757: {
                    758:
1.76      kristaps  759:        if (NULL == r->last) {
1.128     kristaps  760:                mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.76      kristaps  761:                return(ROFF_IGN);
                    762:        }
                    763:
1.82      kristaps  764:        switch (r->last->tok) {
                    765:        case (ROFF_el):
                    766:                /* FALLTHROUGH */
                    767:        case (ROFF_ie):
                    768:                /* FALLTHROUGH */
                    769:        case (ROFF_if):
                    770:                break;
                    771:        default:
1.128     kristaps  772:                mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.75      kristaps  773:                return(ROFF_IGN);
                    774:        }
                    775:
1.76      kristaps  776:        if (r->last->endspan > -1) {
1.128     kristaps  777:                mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.76      kristaps  778:                return(ROFF_IGN);
                    779:        }
                    780:
                    781:        if ((*bufp)[pos])
1.128     kristaps  782:                mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.76      kristaps  783:
1.75      kristaps  784:        roffnode_pop(r);
1.76      kristaps  785:        roffnode_cleanscope(r);
                    786:        return(ROFF_IGN);
                    787: }
                    788:
1.75      kristaps  789:
1.76      kristaps  790: /* ARGSUSED */
                    791: static enum rofferr
1.80      kristaps  792: roff_block(ROFF_ARGS)
1.76      kristaps  793: {
1.78      kristaps  794:        int             sv;
                    795:        size_t          sz;
1.106     kristaps  796:        char            *name;
                    797:
                    798:        name = NULL;
1.76      kristaps  799:
1.106     kristaps  800:        if (ROFF_ig != tok) {
                    801:                if ('\0' == (*bufp)[pos]) {
1.128     kristaps  802:                        mandoc_msg(MANDOCERR_NOARGS, r->parse, ln, ppos, NULL);
1.106     kristaps  803:                        return(ROFF_IGN);
                    804:                }
1.107     kristaps  805:
                    806:                /*
                    807:                 * Re-write `de1', since we don't really care about
                    808:                 * groff's strange compatibility mode, into `de'.
                    809:                 */
                    810:
1.106     kristaps  811:                if (ROFF_de1 == tok)
                    812:                        tok = ROFF_de;
                    813:                if (ROFF_de == tok)
                    814:                        name = *bufp + pos;
                    815:                else
1.128     kristaps  816:                        mandoc_msg(MANDOCERR_REQUEST, r->parse, ln, ppos,
1.106     kristaps  817:                            roffs[tok].name);
1.107     kristaps  818:
1.131     schwarze  819:                while ((*bufp)[pos] && ! isspace((unsigned char)(*bufp)[pos]))
1.80      kristaps  820:                        pos++;
1.107     kristaps  821:
1.131     schwarze  822:                while (isspace((unsigned char)(*bufp)[pos]))
1.106     kristaps  823:                        (*bufp)[pos++] = '\0';
1.80      kristaps  824:        }
                    825:
1.106     kristaps  826:        roffnode_push(r, tok, name, ln, ppos);
                    827:
                    828:        /*
                    829:         * At the beginning of a `de' macro, clear the existing string
                    830:         * with the same name, if there is one.  New content will be
                    831:         * added from roff_block_text() in multiline mode.
                    832:         */
1.107     kristaps  833:
1.106     kristaps  834:        if (ROFF_de == tok)
1.108     schwarze  835:                roff_setstr(r, name, "", 0);
1.76      kristaps  836:
1.79      kristaps  837:        if ('\0' == (*bufp)[pos])
1.78      kristaps  838:                return(ROFF_IGN);
                    839:
1.107     kristaps  840:        /* If present, process the custom end-of-line marker. */
                    841:
1.78      kristaps  842:        sv = pos;
1.131     schwarze  843:        while ((*bufp)[pos] && ! isspace((unsigned char)(*bufp)[pos]))
1.78      kristaps  844:                pos++;
                    845:
                    846:        /*
                    847:         * Note: groff does NOT like escape characters in the input.
                    848:         * Instead of detecting this, we're just going to let it fly and
                    849:         * to hell with it.
                    850:         */
                    851:
                    852:        assert(pos > sv);
                    853:        sz = (size_t)(pos - sv);
                    854:
1.79      kristaps  855:        if (1 == sz && '.' == (*bufp)[sv])
                    856:                return(ROFF_IGN);
                    857:
1.98      schwarze  858:        r->last->end = mandoc_malloc(sz + 1);
1.78      kristaps  859:
                    860:        memcpy(r->last->end, *bufp + sv, sz);
                    861:        r->last->end[(int)sz] = '\0';
                    862:
1.77      kristaps  863:        if ((*bufp)[pos])
1.128     kristaps  864:                mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.74      kristaps  865:
1.78      kristaps  866:        return(ROFF_IGN);
                    867: }
                    868:
                    869:
                    870: /* ARGSUSED */
                    871: static enum rofferr
1.80      kristaps  872: roff_block_sub(ROFF_ARGS)
1.79      kristaps  873: {
                    874:        enum rofft      t;
                    875:        int             i, j;
                    876:
                    877:        /*
                    878:         * First check whether a custom macro exists at this level.  If
                    879:         * it does, then check against it.  This is some of groff's
                    880:         * stranger behaviours.  If we encountered a custom end-scope
                    881:         * tag and that tag also happens to be a "real" macro, then we
                    882:         * need to try interpreting it again as a real macro.  If it's
                    883:         * not, then return ignore.  Else continue.
                    884:         */
                    885:
                    886:        if (r->last->end) {
1.130     kristaps  887:                for (i = pos, j = 0; r->last->end[j]; j++, i++)
1.79      kristaps  888:                        if ((*bufp)[i] != r->last->end[j])
                    889:                                break;
                    890:
                    891:                if ('\0' == r->last->end[j] &&
                    892:                                ('\0' == (*bufp)[i] ||
                    893:                                 ' ' == (*bufp)[i] ||
                    894:                                 '\t' == (*bufp)[i])) {
                    895:                        roffnode_pop(r);
                    896:                        roffnode_cleanscope(r);
                    897:
1.130     kristaps  898:                        while (' ' == (*bufp)[i] || '\t' == (*bufp)[i])
                    899:                                i++;
                    900:
                    901:                        pos = i;
1.106     kristaps  902:                        if (ROFF_MAX != roff_parse(r, *bufp, &pos))
1.79      kristaps  903:                                return(ROFF_RERUN);
                    904:                        return(ROFF_IGN);
                    905:                }
                    906:        }
                    907:
                    908:        /*
                    909:         * If we have no custom end-query or lookup failed, then try
                    910:         * pulling it out of the hashtable.
                    911:         */
                    912:
1.137     schwarze  913:        t = roff_parse(r, *bufp, &pos);
1.79      kristaps  914:
1.106     kristaps  915:        /*
                    916:         * Macros other than block-end are only significant
                    917:         * in `de' blocks; elsewhere, simply throw them away.
                    918:         */
                    919:        if (ROFF_cblock != t) {
                    920:                if (ROFF_de == tok)
                    921:                        roff_setstr(r, r->last->name, *bufp + ppos, 1);
1.79      kristaps  922:                return(ROFF_IGN);
1.106     kristaps  923:        }
1.79      kristaps  924:
                    925:        assert(roffs[t].proc);
1.90      kristaps  926:        return((*roffs[t].proc)(r, t, bufp, szp,
                    927:                                ln, ppos, pos, offs));
1.79      kristaps  928: }
                    929:
                    930:
                    931: /* ARGSUSED */
                    932: static enum rofferr
1.80      kristaps  933: roff_block_text(ROFF_ARGS)
1.78      kristaps  934: {
                    935:
1.106     kristaps  936:        if (ROFF_de == tok)
                    937:                roff_setstr(r, r->last->name, *bufp + pos, 1);
                    938:
1.78      kristaps  939:        return(ROFF_IGN);
                    940: }
                    941:
                    942:
                    943: /* ARGSUSED */
                    944: static enum rofferr
1.82      kristaps  945: roff_cond_sub(ROFF_ARGS)
                    946: {
                    947:        enum rofft       t;
                    948:        enum roffrule    rr;
1.139     kristaps  949:        char            *ep;
1.82      kristaps  950:
                    951:        rr = r->last->rule;
1.139     kristaps  952:        roffnode_cleanscope(r);
1.82      kristaps  953:
1.139     kristaps  954:        /*
                    955:         * If the macro is unknown, first check if it contains a closing
                    956:         * delimiter `\}'.  If it does, close out our scope and return
                    957:         * the currently-scoped rule (ignore or continue).  Else, drop
                    958:         * into the currently-scoped rule.
1.87      kristaps  959:         */
                    960:
1.106     kristaps  961:        if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos))) {
1.139     kristaps  962:                ep = &(*bufp)[pos];
                    963:                for ( ; NULL != (ep = strchr(ep, '\\')); ep++) {
                    964:                        ep++;
                    965:                        if ('}' != *ep)
                    966:                                continue;
1.144     kristaps  967:
                    968:                        /*
                    969:                         * Make the \} go away.
                    970:                         * This is a little haphazard, as it's not quite
                    971:                         * clear how nroff does this.
                    972:                         * If we're at the end of line, then just chop
                    973:                         * off the \} and resize the buffer.
                    974:                         * If we aren't, then conver it to spaces.
                    975:                         */
                    976:
                    977:                        if ('\0' == *(ep + 1)) {
                    978:                                *--ep = '\0';
                    979:                                *szp -= 2;
                    980:                        } else
                    981:                                *(ep - 1) = *ep = ' ';
                    982:
1.139     kristaps  983:                        roff_ccond(r, ROFF_ccond, bufp, szp,
                    984:                                        ln, pos, pos + 2, offs);
                    985:                        break;
                    986:                }
1.82      kristaps  987:                return(ROFFRULE_DENY == rr ? ROFF_IGN : ROFF_CONT);
1.100     kristaps  988:        }
1.82      kristaps  989:
                    990:        /*
                    991:         * A denied conditional must evaluate its children if and only
                    992:         * if they're either structurally required (such as loops and
                    993:         * conditionals) or a closing macro.
                    994:         */
1.139     kristaps  995:
1.82      kristaps  996:        if (ROFFRULE_DENY == rr)
                    997:                if ( ! (ROFFMAC_STRUCT & roffs[t].flags))
                    998:                        if (ROFF_ccond != t)
                    999:                                return(ROFF_IGN);
                   1000:
                   1001:        assert(roffs[t].proc);
1.90      kristaps 1002:        return((*roffs[t].proc)(r, t, bufp, szp,
                   1003:                                ln, ppos, pos, offs));
1.82      kristaps 1004: }
                   1005:
                   1006: /* ARGSUSED */
                   1007: static enum rofferr
                   1008: roff_cond_text(ROFF_ARGS)
1.78      kristaps 1009: {
1.140     kristaps 1010:        char            *ep;
1.82      kristaps 1011:        enum roffrule    rr;
                   1012:
                   1013:        rr = r->last->rule;
1.140     kristaps 1014:        roffnode_cleanscope(r);
1.82      kristaps 1015:
1.140     kristaps 1016:        ep = &(*bufp)[pos];
                   1017:        for ( ; NULL != (ep = strchr(ep, '\\')); ep++) {
                   1018:                ep++;
                   1019:                if ('}' != *ep)
                   1020:                        continue;
                   1021:                *ep = '&';
                   1022:                roff_ccond(r, ROFF_ccond, bufp, szp,
                   1023:                                ln, pos, pos + 2, offs);
1.78      kristaps 1024:        }
1.82      kristaps 1025:        return(ROFFRULE_DENY == rr ? ROFF_IGN : ROFF_CONT);
1.74      kristaps 1026: }
                   1027:
1.88      kristaps 1028: static enum roffrule
                   1029: roff_evalcond(const char *v, int *pos)
                   1030: {
                   1031:
                   1032:        switch (v[*pos]) {
                   1033:        case ('n'):
                   1034:                (*pos)++;
                   1035:                return(ROFFRULE_ALLOW);
                   1036:        case ('e'):
                   1037:                /* FALLTHROUGH */
                   1038:        case ('o'):
                   1039:                /* FALLTHROUGH */
                   1040:        case ('t'):
                   1041:                (*pos)++;
                   1042:                return(ROFFRULE_DENY);
                   1043:        default:
                   1044:                break;
                   1045:        }
                   1046:
                   1047:        while (v[*pos] && ' ' != v[*pos])
                   1048:                (*pos)++;
                   1049:        return(ROFFRULE_DENY);
                   1050: }
                   1051:
1.75      kristaps 1052: /* ARGSUSED */
1.74      kristaps 1053: static enum rofferr
1.103     kristaps 1054: roff_line_ignore(ROFF_ARGS)
1.89      kristaps 1055: {
1.123     schwarze 1056:
                   1057:        if (ROFF_it == tok)
1.128     kristaps 1058:                mandoc_msg(MANDOCERR_REQUEST, r->parse, ln, ppos, "it");
1.89      kristaps 1059:
                   1060:        return(ROFF_IGN);
                   1061: }
                   1062:
1.104     kristaps 1063: /* ARGSUSED */
                   1064: static enum rofferr
1.82      kristaps 1065: roff_cond(ROFF_ARGS)
1.74      kristaps 1066: {
1.77      kristaps 1067:        int              sv;
1.88      kristaps 1068:        enum roffrule    rule;
1.74      kristaps 1069:
1.134     kristaps 1070:        /*
                   1071:         * An `.el' has no conditional body: it will consume the value
                   1072:         * of the current rstack entry set in prior `ie' calls or
                   1073:         * defaults to DENY.
                   1074:         *
                   1075:         * If we're not an `el', however, then evaluate the conditional.
                   1076:         */
1.133     kristaps 1077:
1.134     kristaps 1078:        rule = ROFF_el == tok ?
                   1079:                (r->rstackpos < 0 ?
                   1080:                 ROFFRULE_DENY : r->rstack[r->rstackpos--]) :
                   1081:                roff_evalcond(*bufp, &pos);
1.77      kristaps 1082:
                   1083:        sv = pos;
1.75      kristaps 1084:        while (' ' == (*bufp)[pos])
                   1085:                pos++;
1.74      kristaps 1086:
1.77      kristaps 1087:        /*
                   1088:         * Roff is weird.  If we have just white-space after the
                   1089:         * conditional, it's considered the BODY and we exit without
                   1090:         * really doing anything.  Warn about this.  It's probably
                   1091:         * wrong.
                   1092:         */
1.88      kristaps 1093:
1.77      kristaps 1094:        if ('\0' == (*bufp)[pos] && sv != pos) {
1.128     kristaps 1095:                mandoc_msg(MANDOCERR_NOARGS, r->parse, ln, ppos, NULL);
1.107     kristaps 1096:                return(ROFF_IGN);
1.77      kristaps 1097:        }
                   1098:
1.106     kristaps 1099:        roffnode_push(r, tok, NULL, ln, ppos);
1.77      kristaps 1100:
1.88      kristaps 1101:        r->last->rule = rule;
                   1102:
1.134     kristaps 1103:        /*
                   1104:         * An if-else will put the NEGATION of the current evaluated
                   1105:         * conditional into the stack of rules.
                   1106:         */
                   1107:
1.84      schwarze 1108:        if (ROFF_ie == tok) {
1.134     kristaps 1109:                if (r->rstackpos == RSTACK_MAX - 1) {
                   1110:                        mandoc_msg(MANDOCERR_MEM,
                   1111:                                r->parse, ln, ppos, NULL);
                   1112:                        return(ROFF_ERR);
                   1113:                }
                   1114:                r->rstack[++r->rstackpos] =
                   1115:                        ROFFRULE_DENY == r->last->rule ?
                   1116:                        ROFFRULE_ALLOW : ROFFRULE_DENY;
1.82      kristaps 1117:        }
1.88      kristaps 1118:
                   1119:        /* If the parent has false as its rule, then so do we. */
                   1120:
1.109     kristaps 1121:        if (r->last->parent && ROFFRULE_DENY == r->last->parent->rule)
1.84      schwarze 1122:                r->last->rule = ROFFRULE_DENY;
1.88      kristaps 1123:
                   1124:        /*
                   1125:         * Determine scope.  If we're invoked with "\{" trailing the
                   1126:         * conditional, then we're in a multiline scope.  Else our scope
                   1127:         * expires on the next line.
                   1128:         */
1.74      kristaps 1129:
1.75      kristaps 1130:        r->last->endspan = 1;
                   1131:
                   1132:        if ('\\' == (*bufp)[pos] && '{' == (*bufp)[pos + 1]) {
                   1133:                r->last->endspan = -1;
                   1134:                pos += 2;
1.109     kristaps 1135:        }
1.74      kristaps 1136:
1.77      kristaps 1137:        /*
                   1138:         * If there are no arguments on the line, the next-line scope is
                   1139:         * assumed.
                   1140:         */
                   1141:
1.75      kristaps 1142:        if ('\0' == (*bufp)[pos])
                   1143:                return(ROFF_IGN);
1.77      kristaps 1144:
                   1145:        /* Otherwise re-run the roff parser after recalculating. */
1.74      kristaps 1146:
1.75      kristaps 1147:        *offs = pos;
                   1148:        return(ROFF_RERUN);
1.83      schwarze 1149: }
                   1150:
                   1151:
                   1152: /* ARGSUSED */
                   1153: static enum rofferr
1.92      schwarze 1154: roff_ds(ROFF_ARGS)
                   1155: {
1.96      kristaps 1156:        char            *name, *string;
                   1157:
                   1158:        /*
                   1159:         * A symbol is named by the first word following the macro
                   1160:         * invocation up to a space.  Its value is anything after the
                   1161:         * name's trailing whitespace and optional double-quote.  Thus,
                   1162:         *
                   1163:         *  [.ds foo "bar  "     ]
                   1164:         *
                   1165:         * will have `bar  "     ' as its value.
                   1166:         */
1.92      schwarze 1167:
1.121     schwarze 1168:        string = *bufp + pos;
                   1169:        name = roff_getname(r, &string, ln, pos);
1.92      schwarze 1170:        if ('\0' == *name)
                   1171:                return(ROFF_IGN);
                   1172:
1.121     schwarze 1173:        /* Read past initial double-quote. */
                   1174:        if ('"' == *string)
1.92      schwarze 1175:                string++;
                   1176:
1.96      kristaps 1177:        /* The rest is the value. */
1.106     kristaps 1178:        roff_setstr(r, name, string, 0);
1.92      schwarze 1179:        return(ROFF_IGN);
                   1180: }
                   1181:
1.147     kristaps 1182: int
                   1183: roff_regisset(const struct roff *r, enum regs reg)
                   1184: {
                   1185:
                   1186:        return(r->regs[(int)reg].set);
                   1187: }
                   1188:
                   1189: unsigned int
                   1190: roff_regget(const struct roff *r, enum regs reg)
                   1191: {
                   1192:
                   1193:        return(r->regs[(int)reg].u);
                   1194: }
                   1195:
                   1196: void
                   1197: roff_regunset(struct roff *r, enum regs reg)
                   1198: {
                   1199:
                   1200:        r->regs[(int)reg].set = 0;
                   1201: }
1.92      schwarze 1202:
                   1203: /* ARGSUSED */
                   1204: static enum rofferr
1.89      kristaps 1205: roff_nr(ROFF_ARGS)
1.83      schwarze 1206: {
1.121     schwarze 1207:        const char      *key;
                   1208:        char            *val;
1.138     kristaps 1209:        int              iv;
1.89      kristaps 1210:
1.121     schwarze 1211:        val = *bufp + pos;
                   1212:        key = roff_getname(r, &val, ln, pos);
1.89      kristaps 1213:
                   1214:        if (0 == strcmp(key, "nS")) {
1.147     kristaps 1215:                r->regs[(int)REG_nS].set = 1;
1.149     kristaps 1216:                if ((iv = mandoc_strntoi(val, strlen(val), 10)) >= 0)
1.147     kristaps 1217:                        r->regs[(int)REG_nS].u = (unsigned)iv;
1.138     kristaps 1218:                else
1.147     kristaps 1219:                        r->regs[(int)REG_nS].u = 0u;
1.109     kristaps 1220:        }
                   1221:
1.122     schwarze 1222:        return(ROFF_IGN);
                   1223: }
                   1224:
                   1225: /* ARGSUSED */
                   1226: static enum rofferr
                   1227: roff_rm(ROFF_ARGS)
                   1228: {
                   1229:        const char       *name;
                   1230:        char             *cp;
                   1231:
                   1232:        cp = *bufp + pos;
                   1233:        while ('\0' != *cp) {
1.127     kristaps 1234:                name = roff_getname(r, &cp, ln, (int)(cp - *bufp));
1.122     schwarze 1235:                if ('\0' != *name)
                   1236:                        roff_setstr(r, name, NULL, 0);
                   1237:        }
1.109     kristaps 1238:        return(ROFF_IGN);
                   1239: }
                   1240:
                   1241: /* ARGSUSED */
                   1242: static enum rofferr
                   1243: roff_TE(ROFF_ARGS)
                   1244: {
                   1245:
                   1246:        if (NULL == r->tbl)
1.128     kristaps 1247:                mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.115     kristaps 1248:        else
1.151     kristaps 1249:                tbl_end(&r->tbl);
1.109     kristaps 1250:
1.112     kristaps 1251:        return(ROFF_IGN);
                   1252: }
                   1253:
                   1254: /* ARGSUSED */
                   1255: static enum rofferr
                   1256: roff_T_(ROFF_ARGS)
                   1257: {
                   1258:
                   1259:        if (NULL == r->tbl)
1.128     kristaps 1260:                mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.112     kristaps 1261:        else
1.116     kristaps 1262:                tbl_restart(ppos, ln, r->tbl);
1.112     kristaps 1263:
1.109     kristaps 1264:        return(ROFF_IGN);
                   1265: }
                   1266:
1.156     kristaps 1267: #if 0
                   1268: static int
1.151     kristaps 1269: roff_closeeqn(struct roff *r)
                   1270: {
                   1271:
                   1272:        return(r->eqn && ROFF_EQN == eqn_end(&r->eqn) ? 1 : 0);
                   1273: }
1.156     kristaps 1274: #endif
1.151     kristaps 1275:
1.156     kristaps 1276: static void
1.151     kristaps 1277: roff_openeqn(struct roff *r, const char *name, int line,
                   1278:                int offs, const char *buf)
1.125     kristaps 1279: {
1.151     kristaps 1280:        struct eqn_node *e;
                   1281:        int              poff;
1.125     kristaps 1282:
                   1283:        assert(NULL == r->eqn);
1.151     kristaps 1284:        e = eqn_alloc(name, offs, line, r->parse);
1.125     kristaps 1285:
                   1286:        if (r->last_eqn)
                   1287:                r->last_eqn->next = e;
                   1288:        else
                   1289:                r->first_eqn = r->last_eqn = e;
                   1290:
                   1291:        r->eqn = r->last_eqn = e;
1.151     kristaps 1292:
                   1293:        if (buf) {
                   1294:                poff = 0;
                   1295:                eqn_read(&r->eqn, line, buf, offs, &poff);
                   1296:        }
                   1297: }
                   1298:
                   1299: /* ARGSUSED */
                   1300: static enum rofferr
                   1301: roff_EQ(ROFF_ARGS)
                   1302: {
                   1303:
                   1304:        roff_openeqn(r, *bufp + pos, ln, ppos, NULL);
1.125     kristaps 1305:        return(ROFF_IGN);
                   1306: }
                   1307:
                   1308: /* ARGSUSED */
                   1309: static enum rofferr
                   1310: roff_EN(ROFF_ARGS)
                   1311: {
                   1312:
1.128     kristaps 1313:        mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.125     kristaps 1314:        return(ROFF_IGN);
                   1315: }
                   1316:
                   1317: /* ARGSUSED */
                   1318: static enum rofferr
1.109     kristaps 1319: roff_TS(ROFF_ARGS)
                   1320: {
1.118     kristaps 1321:        struct tbl_node *t;
1.89      kristaps 1322:
1.115     kristaps 1323:        if (r->tbl) {
1.128     kristaps 1324:                mandoc_msg(MANDOCERR_SCOPEBROKEN, r->parse, ln, ppos, NULL);
1.151     kristaps 1325:                tbl_end(&r->tbl);
1.115     kristaps 1326:        }
1.83      schwarze 1327:
1.128     kristaps 1328:        t = tbl_alloc(ppos, ln, r->parse);
1.113     kristaps 1329:
                   1330:        if (r->last_tbl)
                   1331:                r->last_tbl->next = t;
                   1332:        else
                   1333:                r->first_tbl = r->last_tbl = t;
                   1334:
                   1335:        r->tbl = r->last_tbl = t;
1.83      schwarze 1336:        return(ROFF_IGN);
1.92      schwarze 1337: }
                   1338:
1.105     kristaps 1339: /* ARGSUSED */
                   1340: static enum rofferr
                   1341: roff_so(ROFF_ARGS)
                   1342: {
                   1343:        char *name;
                   1344:
1.128     kristaps 1345:        mandoc_msg(MANDOCERR_SO, r->parse, ln, ppos, NULL);
1.105     kristaps 1346:
                   1347:        /*
                   1348:         * Handle `so'.  Be EXTREMELY careful, as we shouldn't be
                   1349:         * opening anything that's not in our cwd or anything beneath
                   1350:         * it.  Thus, explicitly disallow traversing up the file-system
                   1351:         * or using absolute paths.
                   1352:         */
                   1353:
                   1354:        name = *bufp + pos;
                   1355:        if ('/' == *name || strstr(name, "../") || strstr(name, "/..")) {
1.128     kristaps 1356:                mandoc_msg(MANDOCERR_SOPATH, r->parse, ln, pos, NULL);
1.105     kristaps 1357:                return(ROFF_ERR);
                   1358:        }
                   1359:
                   1360:        *offs = pos;
                   1361:        return(ROFF_SO);
                   1362: }
1.92      schwarze 1363:
1.106     kristaps 1364: /* ARGSUSED */
                   1365: static enum rofferr
                   1366: roff_userdef(ROFF_ARGS)
1.99      kristaps 1367: {
1.106     kristaps 1368:        const char       *arg[9];
                   1369:        char             *cp, *n1, *n2;
1.119     schwarze 1370:        int               i;
1.106     kristaps 1371:
                   1372:        /*
                   1373:         * Collect pointers to macro argument strings
                   1374:         * and null-terminate them.
                   1375:         */
                   1376:        cp = *bufp + pos;
1.119     schwarze 1377:        for (i = 0; i < 9; i++)
1.120     schwarze 1378:                arg[i] = '\0' == *cp ? "" :
1.136     kristaps 1379:                    mandoc_getarg(r->parse, &cp, ln, &pos);
1.99      kristaps 1380:
1.106     kristaps 1381:        /*
                   1382:         * Expand macro arguments.
1.99      kristaps 1383:         */
1.106     kristaps 1384:        *szp = 0;
                   1385:        n1 = cp = mandoc_strdup(r->current_string);
                   1386:        while (NULL != (cp = strstr(cp, "\\$"))) {
                   1387:                i = cp[2] - '1';
                   1388:                if (0 > i || 8 < i) {
                   1389:                        /* Not an argument invocation. */
                   1390:                        cp += 2;
                   1391:                        continue;
                   1392:                }
                   1393:
                   1394:                *szp = strlen(n1) - 3 + strlen(arg[i]) + 1;
                   1395:                n2 = mandoc_malloc(*szp);
                   1396:
                   1397:                strlcpy(n2, n1, (size_t)(cp - n1 + 1));
                   1398:                strlcat(n2, arg[i], *szp);
                   1399:                strlcat(n2, cp + 3, *szp);
                   1400:
                   1401:                cp = n2 + (cp - n1);
                   1402:                free(n1);
                   1403:                n1 = n2;
1.99      kristaps 1404:        }
                   1405:
1.106     kristaps 1406:        /*
                   1407:         * Replace the macro invocation
                   1408:         * by the expanded macro.
                   1409:         */
                   1410:        free(*bufp);
                   1411:        *bufp = n1;
                   1412:        if (0 == *szp)
                   1413:                *szp = strlen(*bufp) + 1;
                   1414:
                   1415:        return(*szp > 1 && '\n' == (*bufp)[(int)*szp - 2] ?
                   1416:           ROFF_REPARSE : ROFF_APPEND);
1.99      kristaps 1417: }
1.121     schwarze 1418:
                   1419: static char *
                   1420: roff_getname(struct roff *r, char **cpp, int ln, int pos)
                   1421: {
                   1422:        char     *name, *cp;
                   1423:
                   1424:        name = *cpp;
                   1425:        if ('\0' == *name)
                   1426:                return(name);
                   1427:
                   1428:        /* Read until end of name. */
                   1429:        for (cp = name; '\0' != *cp && ' ' != *cp; cp++) {
                   1430:                if ('\\' != *cp)
                   1431:                        continue;
                   1432:                cp++;
                   1433:                if ('\\' == *cp)
                   1434:                        continue;
1.128     kristaps 1435:                mandoc_msg(MANDOCERR_NAMESC, r->parse, ln, pos, NULL);
1.121     schwarze 1436:                *cp = '\0';
                   1437:                name = cp;
                   1438:        }
                   1439:
                   1440:        /* Nil-terminate name. */
                   1441:        if ('\0' != *cp)
                   1442:                *(cp++) = '\0';
                   1443:
                   1444:        /* Read past spaces. */
                   1445:        while (' ' == *cp)
                   1446:                cp++;
                   1447:
                   1448:        *cpp = cp;
                   1449:        return(name);
                   1450: }
                   1451:
1.106     kristaps 1452: /*
                   1453:  * Store *string into the user-defined string called *name.
                   1454:  * In multiline mode, append to an existing entry and append '\n';
                   1455:  * else replace the existing entry, if there is one.
                   1456:  * To clear an existing entry, call with (*r, *name, NULL, 0).
                   1457:  */
1.94      kristaps 1458: static void
1.106     kristaps 1459: roff_setstr(struct roff *r, const char *name, const char *string,
                   1460:        int multiline)
1.92      schwarze 1461: {
                   1462:        struct roffstr   *n;
1.106     kristaps 1463:        char             *c;
                   1464:        size_t            oldch, newch;
1.92      schwarze 1465:
1.106     kristaps 1466:        /* Search for an existing string with the same name. */
1.94      kristaps 1467:        n = r->first_string;
1.160     kristaps 1468:        while (n && strcmp(name, n->key))
1.92      schwarze 1469:                n = n->next;
1.94      kristaps 1470:
                   1471:        if (NULL == n) {
1.106     kristaps 1472:                /* Create a new string table entry. */
1.94      kristaps 1473:                n = mandoc_malloc(sizeof(struct roffstr));
1.160     kristaps 1474:                n->key = mandoc_strdup(name);
                   1475:                n->val = NULL;
1.94      kristaps 1476:                n->next = r->first_string;
                   1477:                r->first_string = n;
1.106     kristaps 1478:        } else if (0 == multiline) {
                   1479:                /* In multiline mode, append; else replace. */
1.160     kristaps 1480:                free(n->val);
                   1481:                n->val = NULL;
1.106     kristaps 1482:        }
                   1483:
                   1484:        if (NULL == string)
                   1485:                return;
                   1486:
                   1487:        /*
                   1488:         * One additional byte for the '\n' in multiline mode,
                   1489:         * and one for the terminating '\0'.
                   1490:         */
1.127     kristaps 1491:        newch = strlen(string) + (multiline ? 2u : 1u);
1.160     kristaps 1492:        if (NULL == n->val) {
                   1493:                n->val = mandoc_malloc(newch);
                   1494:                *n->val = '\0';
1.106     kristaps 1495:                oldch = 0;
                   1496:        } else {
1.160     kristaps 1497:                oldch = strlen(n->val);
                   1498:                n->val = mandoc_realloc(n->val, oldch + newch);
1.106     kristaps 1499:        }
                   1500:
                   1501:        /* Skip existing content in the destination buffer. */
1.160     kristaps 1502:        c = n->val + (int)oldch;
1.106     kristaps 1503:
                   1504:        /* Append new content to the destination buffer. */
                   1505:        while (*string) {
                   1506:                /*
                   1507:                 * Rudimentary roff copy mode:
                   1508:                 * Handle escaped backslashes.
                   1509:                 */
                   1510:                if ('\\' == *string && '\\' == *(string + 1))
                   1511:                        string++;
                   1512:                *c++ = *string++;
                   1513:        }
1.94      kristaps 1514:
1.106     kristaps 1515:        /* Append terminating bytes. */
                   1516:        if (multiline)
                   1517:                *c++ = '\n';
                   1518:        *c = '\0';
1.92      schwarze 1519: }
                   1520:
1.94      kristaps 1521: static const char *
                   1522: roff_getstrn(const struct roff *r, const char *name, size_t len)
1.92      schwarze 1523: {
1.94      kristaps 1524:        const struct roffstr *n;
1.92      schwarze 1525:
1.157     kristaps 1526:        for (n = r->first_string; n; n = n->next)
1.160     kristaps 1527:                if (0 == strncmp(name, n->key, len) &&
                   1528:                                '\0' == n->key[(int)len])
                   1529:                        return(n->val);
1.94      kristaps 1530:
1.157     kristaps 1531:        return(NULL);
1.92      schwarze 1532: }
                   1533:
1.94      kristaps 1534: static void
                   1535: roff_freestr(struct roff *r)
1.92      schwarze 1536: {
                   1537:        struct roffstr   *n, *nn;
                   1538:
1.94      kristaps 1539:        for (n = r->first_string; n; n = nn) {
1.160     kristaps 1540:                free(n->key);
                   1541:                free(n->val);
1.92      schwarze 1542:                nn = n->next;
                   1543:                free(n);
                   1544:        }
1.94      kristaps 1545:
                   1546:        r->first_string = NULL;
1.114     kristaps 1547: }
                   1548:
                   1549: const struct tbl_span *
                   1550: roff_span(const struct roff *r)
                   1551: {
                   1552:
                   1553:        return(r->tbl ? tbl_span(r->tbl) : NULL);
1.125     kristaps 1554: }
                   1555:
                   1556: const struct eqn *
                   1557: roff_eqn(const struct roff *r)
                   1558: {
                   1559:
                   1560:        return(r->last_eqn ? &r->last_eqn->eqn : NULL);
1.151     kristaps 1561: }
                   1562:
                   1563: char
                   1564: roff_eqndelim(const struct roff *r)
                   1565: {
                   1566:
                   1567:        return('\0');
1.74      kristaps 1568: }

CVSweb