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

Annotation of mandoc/roff.c, Revision 1.185

1.185   ! schwarze    1: /*     $Id: roff.c,v 1.184 2013/10/05 22:25:12 schwarze Exp $ */
1.1       kristaps    2: /*
1.175     schwarze    3:  * Copyright (c) 2010, 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
1.177     schwarze    4:  * Copyright (c) 2010, 2011, 2012, 2013 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>
1.178     schwarze   24: #include <stdio.h>
1.1       kristaps   25: #include <stdlib.h>
1.67      kristaps   26: #include <string.h>
1.1       kristaps   27:
1.67      kristaps   28: #include "mandoc.h"
1.109     kristaps   29: #include "libroff.h"
1.94      kristaps   30: #include "libmandoc.h"
1.33      kristaps   31:
1.141     kristaps   32: /* Maximum number of nested if-else conditionals. */
1.82      kristaps   33: #define        RSTACK_MAX      128
                     34:
1.170     schwarze   35: /* Maximum number of string expansions per line, to break infinite loops. */
                     36: #define        EXPAND_LIMIT    1000
                     37:
1.67      kristaps   38: enum   rofft {
1.103     kristaps   39:        ROFF_ad,
1.80      kristaps   40:        ROFF_am,
                     41:        ROFF_ami,
                     42:        ROFF_am1,
1.174     kristaps   43:        ROFF_cc,
1.80      kristaps   44:        ROFF_de,
                     45:        ROFF_dei,
                     46:        ROFF_de1,
1.83      schwarze   47:        ROFF_ds,
1.82      kristaps   48:        ROFF_el,
1.185   ! schwarze   49:        ROFF_fam,
1.103     kristaps   50:        ROFF_hy,
1.82      kristaps   51:        ROFF_ie,
1.75      kristaps   52:        ROFF_if,
1.76      kristaps   53:        ROFF_ig,
1.123     schwarze   54:        ROFF_it,
1.103     kristaps   55:        ROFF_ne,
                     56:        ROFF_nh,
1.104     kristaps   57:        ROFF_nr,
1.124     schwarze   58:        ROFF_ns,
                     59:        ROFF_ps,
1.83      schwarze   60:        ROFF_rm,
1.105     kristaps   61:        ROFF_so,
1.124     schwarze   62:        ROFF_ta,
1.83      schwarze   63:        ROFF_tr,
1.175     schwarze   64:        ROFF_Dd,
                     65:        ROFF_TH,
1.109     kristaps   66:        ROFF_TS,
                     67:        ROFF_TE,
1.112     kristaps   68:        ROFF_T_,
1.125     kristaps   69:        ROFF_EQ,
                     70:        ROFF_EN,
1.76      kristaps   71:        ROFF_cblock,
1.141     kristaps   72:        ROFF_ccond,
1.106     kristaps   73:        ROFF_USERDEF,
1.67      kristaps   74:        ROFF_MAX
                     75: };
                     76:
1.82      kristaps   77: enum   roffrule {
1.183     schwarze   78:        ROFFRULE_DENY,
                     79:        ROFFRULE_ALLOW
1.82      kristaps   80: };
                     81:
1.147     kristaps   82: /*
1.167     kristaps   83:  * An incredibly-simple string buffer.
                     84:  */
1.94      kristaps   85: struct roffstr {
1.167     kristaps   86:        char            *p; /* nil-terminated buffer */
                     87:        size_t           sz; /* saved strlen(p) */
1.166     kristaps   88: };
                     89:
                     90: /*
1.167     kristaps   91:  * A key-value roffstr pair as part of a singly-linked list.
1.166     kristaps   92:  */
                     93: struct roffkv {
                     94:        struct roffstr   key;
                     95:        struct roffstr   val;
                     96:        struct roffkv   *next; /* next in list */
1.94      kristaps   97: };
                     98:
1.180     schwarze   99: /*
                    100:  * A single number register as part of a singly-linked list.
                    101:  */
                    102: struct roffreg {
                    103:        struct roffstr   key;
1.181     schwarze  104:        int              val;
1.180     schwarze  105:        struct roffreg  *next;
                    106: };
                    107:
1.67      kristaps  108: struct roff {
1.175     schwarze  109:        enum mparset     parsetype; /* requested parse type */
1.128     kristaps  110:        struct mparse   *parse; /* parse point */
1.67      kristaps  111:        struct roffnode *last; /* leaf of stack */
1.82      kristaps  112:        enum roffrule    rstack[RSTACK_MAX]; /* stack of !`ie' rules */
1.174     kristaps  113:        char             control; /* control character */
1.82      kristaps  114:        int              rstackpos; /* position in rstack */
1.180     schwarze  115:        struct roffreg  *regtab; /* number registers */
1.166     kristaps  116:        struct roffkv   *strtab; /* user-defined strings & macros */
1.167     kristaps  117:        struct roffkv   *xmbtab; /* multi-byte trans table (`tr') */
                    118:        struct roffstr  *xtab; /* single-byte trans table (`tr') */
1.106     kristaps  119:        const char      *current_string; /* value of last called user macro */
1.118     kristaps  120:        struct tbl_node *first_tbl; /* first table parsed */
                    121:        struct tbl_node *last_tbl; /* last table parsed */
                    122:        struct tbl_node *tbl; /* current table being parsed */
1.125     kristaps  123:        struct eqn_node *last_eqn; /* last equation parsed */
                    124:        struct eqn_node *first_eqn; /* first equation parsed */
                    125:        struct eqn_node *eqn; /* current equation being parsed */
1.79      kristaps  126: };
                    127:
1.67      kristaps  128: struct roffnode {
                    129:        enum rofft       tok; /* type of node */
                    130:        struct roffnode *parent; /* up one in stack */
                    131:        int              line; /* parse line */
                    132:        int              col; /* parse col */
1.106     kristaps  133:        char            *name; /* node name, e.g. macro name */
1.79      kristaps  134:        char            *end; /* end-rules: custom token */
                    135:        int              endspan; /* end-rules: next-line or infty */
1.82      kristaps  136:        enum roffrule    rule; /* current evaluation rule */
1.67      kristaps  137: };
                    138:
                    139: #define        ROFF_ARGS        struct roff *r, /* parse ctx */ \
1.72      kristaps  140:                         enum rofft tok, /* tok of macro */ \
1.67      kristaps  141:                         char **bufp, /* input buffer */ \
                    142:                         size_t *szp, /* size of input buffer */ \
                    143:                         int ln, /* parse line */ \
1.75      kristaps  144:                         int ppos, /* original pos in buffer */ \
                    145:                         int pos, /* current pos in buffer */ \
1.74      kristaps  146:                         int *offs /* reset offset of buffer data */
1.67      kristaps  147:
                    148: typedef        enum rofferr (*roffproc)(ROFF_ARGS);
                    149:
                    150: struct roffmac {
                    151:        const char      *name; /* macro name */
1.79      kristaps  152:        roffproc         proc; /* process new macro */
                    153:        roffproc         text; /* process as child text of macro */
                    154:        roffproc         sub; /* process as child of macro */
                    155:        int              flags;
                    156: #define        ROFFMAC_STRUCT  (1 << 0) /* always interpret */
1.85      kristaps  157:        struct roffmac  *next;
1.67      kristaps  158: };
                    159:
1.141     kristaps  160: struct predef {
                    161:        const char      *name; /* predefined input name */
                    162:        const char      *str; /* replacement symbol */
                    163: };
                    164:
                    165: #define        PREDEF(__name, __str) \
                    166:        { (__name), (__str) },
                    167:
1.155     kristaps  168: static enum rofft       roffhash_find(const char *, size_t);
                    169: static void             roffhash_init(void);
                    170: static void             roffnode_cleanscope(struct roff *);
                    171: static void             roffnode_pop(struct roff *);
                    172: static void             roffnode_push(struct roff *, enum rofft,
                    173:                                const char *, int, int);
1.80      kristaps  174: static enum rofferr     roff_block(ROFF_ARGS);
                    175: static enum rofferr     roff_block_text(ROFF_ARGS);
                    176: static enum rofferr     roff_block_sub(ROFF_ARGS);
                    177: static enum rofferr     roff_cblock(ROFF_ARGS);
1.174     kristaps  178: static enum rofferr     roff_cc(ROFF_ARGS);
1.80      kristaps  179: static enum rofferr     roff_ccond(ROFF_ARGS);
1.82      kristaps  180: static enum rofferr     roff_cond(ROFF_ARGS);
                    181: static enum rofferr     roff_cond_text(ROFF_ARGS);
                    182: static enum rofferr     roff_cond_sub(ROFF_ARGS);
1.92      schwarze  183: static enum rofferr     roff_ds(ROFF_ARGS);
1.94      kristaps  184: static enum roffrule    roff_evalcond(const char *, int *);
1.155     kristaps  185: static void             roff_free1(struct roff *);
1.180     schwarze  186: static void             roff_freereg(struct roffreg *);
1.167     kristaps  187: static void             roff_freestr(struct roffkv *);
1.121     schwarze  188: static char            *roff_getname(struct roff *, char **, int, int);
1.184     schwarze  189: static int              roff_getnum(const char *, int *, int *);
                    190: static int              roff_getop(const char *, int *, char *);
1.181     schwarze  191: static int              roff_getregn(const struct roff *,
                    192:                                const char *, size_t);
1.94      kristaps  193: static const char      *roff_getstrn(const struct roff *,
                    194:                                const char *, size_t);
1.178     schwarze  195: static enum rofferr     roff_it(ROFF_ARGS);
1.103     kristaps  196: static enum rofferr     roff_line_ignore(ROFF_ARGS);
1.89      kristaps  197: static enum rofferr     roff_nr(ROFF_ARGS);
1.169     schwarze  198: static void             roff_openeqn(struct roff *, const char *,
1.156     kristaps  199:                                int, int, const char *);
1.155     kristaps  200: static enum rofft       roff_parse(struct roff *, const char *, int *);
1.178     schwarze  201: static enum rofferr     roff_parsetext(char **, size_t *, int, int *);
1.172     schwarze  202: static enum rofferr     roff_res(struct roff *,
1.142     kristaps  203:                                char **, size_t *, int, int);
1.122     schwarze  204: static enum rofferr     roff_rm(ROFF_ARGS);
1.94      kristaps  205: static void             roff_setstr(struct roff *,
1.106     kristaps  206:                                const char *, const char *, int);
1.166     kristaps  207: static void             roff_setstrn(struct roffkv **, const char *,
1.164     kristaps  208:                                size_t, const char *, size_t, int);
1.105     kristaps  209: static enum rofferr     roff_so(ROFF_ARGS);
1.164     kristaps  210: static enum rofferr     roff_tr(ROFF_ARGS);
1.175     schwarze  211: static enum rofferr     roff_Dd(ROFF_ARGS);
                    212: static enum rofferr     roff_TH(ROFF_ARGS);
1.109     kristaps  213: static enum rofferr     roff_TE(ROFF_ARGS);
                    214: static enum rofferr     roff_TS(ROFF_ARGS);
1.125     kristaps  215: static enum rofferr     roff_EQ(ROFF_ARGS);
                    216: static enum rofferr     roff_EN(ROFF_ARGS);
1.112     kristaps  217: static enum rofferr     roff_T_(ROFF_ARGS);
1.106     kristaps  218: static enum rofferr     roff_userdef(ROFF_ARGS);
1.67      kristaps  219:
1.155     kristaps  220: /* See roffhash_find() */
1.85      kristaps  221:
                    222: #define        ASCII_HI         126
                    223: #define        ASCII_LO         33
                    224: #define        HASHWIDTH       (ASCII_HI - ASCII_LO + 1)
                    225:
                    226: static struct roffmac  *hash[HASHWIDTH];
                    227:
                    228: static struct roffmac   roffs[ROFF_MAX] = {
1.103     kristaps  229:        { "ad", roff_line_ignore, NULL, NULL, 0, NULL },
1.85      kristaps  230:        { "am", roff_block, roff_block_text, roff_block_sub, 0, NULL },
                    231:        { "ami", roff_block, roff_block_text, roff_block_sub, 0, NULL },
                    232:        { "am1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.174     kristaps  233:        { "cc", roff_cc, NULL, NULL, 0, NULL },
1.85      kristaps  234:        { "de", roff_block, roff_block_text, roff_block_sub, 0, NULL },
                    235:        { "dei", roff_block, roff_block_text, roff_block_sub, 0, NULL },
                    236:        { "de1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.92      schwarze  237:        { "ds", roff_ds, NULL, NULL, 0, NULL },
1.85      kristaps  238:        { "el", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
1.185   ! schwarze  239:        { "fam", roff_line_ignore, NULL, NULL, 0, NULL },
1.103     kristaps  240:        { "hy", roff_line_ignore, NULL, NULL, 0, NULL },
1.85      kristaps  241:        { "ie", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
                    242:        { "if", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
                    243:        { "ig", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.178     schwarze  244:        { "it", roff_it, NULL, NULL, 0, NULL },
1.103     kristaps  245:        { "ne", roff_line_ignore, NULL, NULL, 0, NULL },
                    246:        { "nh", roff_line_ignore, NULL, NULL, 0, NULL },
1.104     kristaps  247:        { "nr", roff_nr, NULL, NULL, 0, NULL },
1.124     schwarze  248:        { "ns", roff_line_ignore, NULL, NULL, 0, NULL },
                    249:        { "ps", roff_line_ignore, NULL, NULL, 0, NULL },
1.122     schwarze  250:        { "rm", roff_rm, NULL, NULL, 0, NULL },
1.105     kristaps  251:        { "so", roff_so, NULL, NULL, 0, NULL },
1.124     schwarze  252:        { "ta", roff_line_ignore, NULL, NULL, 0, NULL },
1.164     kristaps  253:        { "tr", roff_tr, NULL, NULL, 0, NULL },
1.175     schwarze  254:        { "Dd", roff_Dd, NULL, NULL, 0, NULL },
                    255:        { "TH", roff_TH, NULL, NULL, 0, NULL },
1.109     kristaps  256:        { "TS", roff_TS, NULL, NULL, 0, NULL },
                    257:        { "TE", roff_TE, NULL, NULL, 0, NULL },
1.112     kristaps  258:        { "T&", roff_T_, NULL, NULL, 0, NULL },
1.125     kristaps  259:        { "EQ", roff_EQ, NULL, NULL, 0, NULL },
                    260:        { "EN", roff_EN, NULL, NULL, 0, NULL },
1.85      kristaps  261:        { ".", roff_cblock, NULL, NULL, 0, NULL },
                    262:        { "\\}", roff_ccond, NULL, NULL, 0, NULL },
1.106     kristaps  263:        { NULL, roff_userdef, NULL, NULL, 0, NULL },
1.67      kristaps  264: };
                    265:
1.175     schwarze  266: const  char *const __mdoc_reserved[] = {
                    267:        "Ac", "Ad", "An", "Ao", "Ap", "Aq", "Ar", "At",
                    268:        "Bc", "Bd", "Bf", "Bk", "Bl", "Bo", "Bq",
                    269:        "Brc", "Bro", "Brq", "Bsx", "Bt", "Bx",
                    270:        "Cd", "Cm", "Db", "Dc", "Dd", "Dl", "Do", "Dq",
                    271:        "Ds", "Dt", "Dv", "Dx", "D1",
                    272:        "Ec", "Ed", "Ef", "Ek", "El", "Em", "em",
                    273:        "En", "Eo", "Eq", "Er", "Es", "Ev", "Ex",
                    274:        "Fa", "Fc", "Fd", "Fl", "Fn", "Fo", "Fr", "Ft", "Fx",
                    275:        "Hf", "Ic", "In", "It", "Lb", "Li", "Lk", "Lp", "LP",
                    276:        "Me", "Ms", "Mt", "Nd", "Nm", "No", "Ns", "Nx",
                    277:        "Oc", "Oo", "Op", "Os", "Ot", "Ox",
                    278:        "Pa", "Pc", "Pf", "Po", "Pp", "PP", "pp", "Pq",
                    279:        "Qc", "Ql", "Qo", "Qq", "Or", "Rd", "Re", "Rs", "Rv",
                    280:        "Sc", "Sf", "Sh", "SH", "Sm", "So", "Sq",
                    281:        "Ss", "St", "Sx", "Sy",
                    282:        "Ta", "Tn", "Ud", "Ux", "Va", "Vt", "Xc", "Xo", "Xr",
                    283:        "%A", "%B", "%D", "%I", "%J", "%N", "%O",
                    284:        "%P", "%Q", "%R", "%T", "%U", "%V",
                    285:        NULL
                    286: };
                    287:
                    288: const  char *const __man_reserved[] = {
                    289:        "AT", "B", "BI", "BR", "BT", "DE", "DS", "DT",
                    290:        "EE", "EN", "EQ", "EX", "HF", "HP", "I", "IB", "IP", "IR",
                    291:        "LP", "ME", "MT", "OP", "P", "PD", "PP", "PT",
                    292:        "R", "RB", "RE", "RI", "RS", "SB", "SH", "SM", "SS", "SY",
                    293:        "TE", "TH", "TP", "TQ", "TS", "T&", "UC", "UE", "UR", "YS",
                    294:        NULL
                    295: };
                    296:
1.141     kristaps  297: /* Array of injected predefined strings. */
                    298: #define        PREDEFS_MAX      38
                    299: static const struct predef predefs[PREDEFS_MAX] = {
                    300: #include "predefs.in"
                    301: };
                    302:
1.155     kristaps  303: /* See roffhash_find() */
1.85      kristaps  304: #define        ROFF_HASH(p)    (p[0] - ASCII_LO)
                    305:
1.178     schwarze  306: static int      roffit_lines;  /* number of lines to delay */
                    307: static char    *roffit_macro;  /* nil-terminated macro line */
                    308:
1.85      kristaps  309: static void
1.155     kristaps  310: roffhash_init(void)
1.85      kristaps  311: {
                    312:        struct roffmac   *n;
                    313:        int               buc, i;
                    314:
1.106     kristaps  315:        for (i = 0; i < (int)ROFF_USERDEF; i++) {
1.85      kristaps  316:                assert(roffs[i].name[0] >= ASCII_LO);
                    317:                assert(roffs[i].name[0] <= ASCII_HI);
                    318:
                    319:                buc = ROFF_HASH(roffs[i].name);
                    320:
                    321:                if (NULL != (n = hash[buc])) {
                    322:                        for ( ; n->next; n = n->next)
                    323:                                /* Do nothing. */ ;
                    324:                        n->next = &roffs[i];
                    325:                } else
                    326:                        hash[buc] = &roffs[i];
                    327:        }
                    328: }
                    329:
1.67      kristaps  330: /*
                    331:  * Look up a roff token by its name.  Returns ROFF_MAX if no macro by
                    332:  * the nil-terminated string name could be found.
                    333:  */
                    334: static enum rofft
1.155     kristaps  335: roffhash_find(const char *p, size_t s)
1.67      kristaps  336: {
1.85      kristaps  337:        int              buc;
                    338:        struct roffmac  *n;
1.67      kristaps  339:
1.85      kristaps  340:        /*
                    341:         * libroff has an extremely simple hashtable, for the time
                    342:         * being, which simply keys on the first character, which must
                    343:         * be printable, then walks a chain.  It works well enough until
                    344:         * optimised.
                    345:         */
                    346:
                    347:        if (p[0] < ASCII_LO || p[0] > ASCII_HI)
                    348:                return(ROFF_MAX);
                    349:
                    350:        buc = ROFF_HASH(p);
                    351:
                    352:        if (NULL == (n = hash[buc]))
                    353:                return(ROFF_MAX);
                    354:        for ( ; n; n = n->next)
1.106     kristaps  355:                if (0 == strncmp(n->name, p, s) && '\0' == n->name[(int)s])
1.85      kristaps  356:                        return((enum rofft)(n - roffs));
1.67      kristaps  357:
                    358:        return(ROFF_MAX);
                    359: }
                    360:
                    361:
                    362: /*
                    363:  * Pop the current node off of the stack of roff instructions currently
                    364:  * pending.
                    365:  */
                    366: static void
                    367: roffnode_pop(struct roff *r)
                    368: {
                    369:        struct roffnode *p;
                    370:
1.75      kristaps  371:        assert(r->last);
                    372:        p = r->last;
1.82      kristaps  373:
1.75      kristaps  374:        r->last = r->last->parent;
1.106     kristaps  375:        free(p->name);
                    376:        free(p->end);
1.67      kristaps  377:        free(p);
                    378: }
                    379:
                    380:
                    381: /*
                    382:  * Push a roff node onto the instruction stack.  This must later be
                    383:  * removed with roffnode_pop().
                    384:  */
1.98      schwarze  385: static void
1.106     kristaps  386: roffnode_push(struct roff *r, enum rofft tok, const char *name,
                    387:                int line, int col)
1.67      kristaps  388: {
                    389:        struct roffnode *p;
                    390:
1.98      schwarze  391:        p = mandoc_calloc(1, sizeof(struct roffnode));
1.67      kristaps  392:        p->tok = tok;
1.106     kristaps  393:        if (name)
                    394:                p->name = mandoc_strdup(name);
1.67      kristaps  395:        p->parent = r->last;
                    396:        p->line = line;
                    397:        p->col = col;
1.79      kristaps  398:        p->rule = p->parent ? p->parent->rule : ROFFRULE_DENY;
1.67      kristaps  399:
                    400:        r->last = p;
                    401: }
                    402:
                    403:
                    404: static void
                    405: roff_free1(struct roff *r)
                    406: {
1.176     schwarze  407:        struct tbl_node *tbl;
1.125     kristaps  408:        struct eqn_node *e;
1.167     kristaps  409:        int              i;
1.67      kristaps  410:
1.176     schwarze  411:        while (NULL != (tbl = r->first_tbl)) {
                    412:                r->first_tbl = tbl->next;
                    413:                tbl_free(tbl);
1.109     kristaps  414:        }
                    415:
1.113     kristaps  416:        r->first_tbl = r->last_tbl = r->tbl = NULL;
                    417:
1.125     kristaps  418:        while (NULL != (e = r->first_eqn)) {
                    419:                r->first_eqn = e->next;
                    420:                eqn_free(e);
                    421:        }
                    422:
                    423:        r->first_eqn = r->last_eqn = r->eqn = NULL;
                    424:
1.67      kristaps  425:        while (r->last)
                    426:                roffnode_pop(r);
1.109     kristaps  427:
1.167     kristaps  428:        roff_freestr(r->strtab);
                    429:        roff_freestr(r->xmbtab);
                    430:
                    431:        r->strtab = r->xmbtab = NULL;
                    432:
1.180     schwarze  433:        roff_freereg(r->regtab);
                    434:
                    435:        r->regtab = NULL;
                    436:
1.167     kristaps  437:        if (r->xtab)
                    438:                for (i = 0; i < 128; i++)
                    439:                        free(r->xtab[i].p);
                    440:
                    441:        free(r->xtab);
                    442:        r->xtab = NULL;
1.67      kristaps  443: }
                    444:
                    445: void
                    446: roff_reset(struct roff *r)
                    447: {
1.143     kristaps  448:        int              i;
1.67      kristaps  449:
                    450:        roff_free1(r);
1.143     kristaps  451:
1.174     kristaps  452:        r->control = 0;
1.147     kristaps  453:
1.143     kristaps  454:        for (i = 0; i < PREDEFS_MAX; i++)
                    455:                roff_setstr(r, predefs[i].name, predefs[i].str, 0);
1.67      kristaps  456: }
                    457:
                    458:
                    459: void
                    460: roff_free(struct roff *r)
                    461: {
                    462:
                    463:        roff_free1(r);
                    464:        free(r);
                    465: }
                    466:
                    467:
                    468: struct roff *
1.175     schwarze  469: roff_alloc(enum mparset type, struct mparse *parse)
1.67      kristaps  470: {
                    471:        struct roff     *r;
1.141     kristaps  472:        int              i;
1.67      kristaps  473:
1.98      schwarze  474:        r = mandoc_calloc(1, sizeof(struct roff));
1.175     schwarze  475:        r->parsetype = type;
1.128     kristaps  476:        r->parse = parse;
1.82      kristaps  477:        r->rstackpos = -1;
1.85      kristaps  478:
1.155     kristaps  479:        roffhash_init();
1.141     kristaps  480:
                    481:        for (i = 0; i < PREDEFS_MAX; i++)
                    482:                roff_setstr(r, predefs[i].name, predefs[i].str, 0);
                    483:
1.67      kristaps  484:        return(r);
                    485: }
                    486:
1.94      kristaps  487: /*
1.181     schwarze  488:  * In the current line, expand user-defined strings ("\*")
                    489:  * and references to number registers ("\n").
                    490:  * Also check the syntax of other escape sequences.
1.154     kristaps  491:  */
1.172     schwarze  492: static enum rofferr
1.142     kristaps  493: roff_res(struct roff *r, char **bufp, size_t *szp, int ln, int pos)
1.94      kristaps  494: {
1.181     schwarze  495:        char             ubuf[12]; /* buffer to print the number */
1.108     schwarze  496:        const char      *stesc; /* start of an escape sequence ('\\') */
                    497:        const char      *stnam; /* start of the name, after "[(*" */
                    498:        const char      *cp;    /* end of the name, e.g. before ']' */
                    499:        const char      *res;   /* the string to be substituted */
1.181     schwarze  500:        char            *nbuf;  /* new buffer to copy bufp to */
                    501:        size_t           nsz;   /* size of the new buffer */
                    502:        size_t           maxl;  /* expected length of the escape name */
                    503:        size_t           naml;  /* actual length of the escape name */
                    504:        int              expand_count;  /* to avoid infinite loops */
1.94      kristaps  505:
1.170     schwarze  506:        expand_count = 0;
                    507:
1.154     kristaps  508: again:
1.108     schwarze  509:        cp = *bufp + pos;
                    510:        while (NULL != (cp = strchr(cp, '\\'))) {
                    511:                stesc = cp++;
                    512:
                    513:                /*
1.181     schwarze  514:                 * The second character must be an asterisk or an n.
1.108     schwarze  515:                 * If it isn't, skip it anyway:  It is escaped,
                    516:                 * so it can't start another escape sequence.
                    517:                 */
                    518:
                    519:                if ('\0' == *cp)
1.172     schwarze  520:                        return(ROFF_CONT);
1.152     kristaps  521:
1.181     schwarze  522:                switch (*cp) {
                    523:                case ('*'):
                    524:                        res = NULL;
                    525:                        break;
                    526:                case ('n'):
                    527:                        res = ubuf;
                    528:                        break;
                    529:                default:
                    530:                        if (ESCAPE_ERROR != mandoc_escape(&cp, NULL, NULL))
1.152     kristaps  531:                                continue;
1.153     kristaps  532:                        mandoc_msg
                    533:                                (MANDOCERR_BADESCAPE, r->parse,
                    534:                                 ln, (int)(stesc - *bufp), NULL);
1.172     schwarze  535:                        return(ROFF_CONT);
1.152     kristaps  536:                }
                    537:
                    538:                cp++;
1.108     schwarze  539:
                    540:                /*
                    541:                 * The third character decides the length
1.181     schwarze  542:                 * of the name of the string or register.
1.108     schwarze  543:                 * Save a pointer to the name.
                    544:                 */
                    545:
1.94      kristaps  546:                switch (*cp) {
1.108     schwarze  547:                case ('\0'):
1.172     schwarze  548:                        return(ROFF_CONT);
1.94      kristaps  549:                case ('('):
                    550:                        cp++;
                    551:                        maxl = 2;
                    552:                        break;
                    553:                case ('['):
                    554:                        cp++;
                    555:                        maxl = 0;
                    556:                        break;
                    557:                default:
                    558:                        maxl = 1;
                    559:                        break;
                    560:                }
1.108     schwarze  561:                stnam = cp;
1.94      kristaps  562:
1.108     schwarze  563:                /* Advance to the end of the name. */
1.94      kristaps  564:
1.181     schwarze  565:                for (naml = 0; 0 == maxl || naml < maxl; naml++, cp++) {
1.153     kristaps  566:                        if ('\0' == *cp) {
                    567:                                mandoc_msg
                    568:                                        (MANDOCERR_BADESCAPE,
                    569:                                         r->parse, ln,
                    570:                                         (int)(stesc - *bufp), NULL);
1.172     schwarze  571:                                return(ROFF_CONT);
1.153     kristaps  572:                        }
1.94      kristaps  573:                        if (0 == maxl && ']' == *cp)
                    574:                                break;
                    575:                }
                    576:
1.108     schwarze  577:                /*
                    578:                 * Retrieve the replacement string; if it is
                    579:                 * undefined, resume searching for escapes.
                    580:                 */
                    581:
1.181     schwarze  582:                if (NULL == res)
                    583:                        res = roff_getstrn(r, stnam, naml);
                    584:                else
                    585:                        snprintf(ubuf, sizeof(ubuf), "%d",
                    586:                            roff_getregn(r, stnam, naml));
1.94      kristaps  587:
                    588:                if (NULL == res) {
1.153     kristaps  589:                        mandoc_msg
                    590:                                (MANDOCERR_BADESCAPE, r->parse,
                    591:                                 ln, (int)(stesc - *bufp), NULL);
1.142     kristaps  592:                        res = "";
1.94      kristaps  593:                }
                    594:
1.108     schwarze  595:                /* Replace the escape sequence by the string. */
                    596:
1.161     kristaps  597:                pos = stesc - *bufp;
1.154     kristaps  598:
1.94      kristaps  599:                nsz = *szp + strlen(res) + 1;
1.181     schwarze  600:                nbuf = mandoc_malloc(nsz);
1.94      kristaps  601:
1.181     schwarze  602:                strlcpy(nbuf, *bufp, (size_t)(stesc - *bufp + 1));
                    603:                strlcat(nbuf, res, nsz);
                    604:                strlcat(nbuf, cp + (maxl ? 0 : 1), nsz);
1.94      kristaps  605:
                    606:                free(*bufp);
                    607:
1.181     schwarze  608:                *bufp = nbuf;
1.94      kristaps  609:                *szp = nsz;
1.170     schwarze  610:
                    611:                if (EXPAND_LIMIT >= ++expand_count)
                    612:                        goto again;
                    613:
                    614:                /* Just leave the string unexpanded. */
                    615:                mandoc_msg(MANDOCERR_ROFFLOOP, r->parse, ln, pos, NULL);
1.172     schwarze  616:                return(ROFF_IGN);
1.154     kristaps  617:        }
1.172     schwarze  618:        return(ROFF_CONT);
1.154     kristaps  619: }
                    620:
                    621: /*
1.178     schwarze  622:  * Process text streams:
                    623:  * Convert all breakable hyphens into ASCII_HYPH.
                    624:  * Decrement and spring input line trap.
1.154     kristaps  625:  */
                    626: static enum rofferr
1.178     schwarze  627: roff_parsetext(char **bufp, size_t *szp, int pos, int *offs)
1.154     kristaps  628: {
                    629:        size_t           sz;
                    630:        const char      *start;
1.178     schwarze  631:        char            *p;
                    632:        int              isz;
1.154     kristaps  633:        enum mandoc_esc  esc;
                    634:
1.178     schwarze  635:        start = p = *bufp + pos;
1.154     kristaps  636:
                    637:        while ('\0' != *p) {
                    638:                sz = strcspn(p, "-\\");
                    639:                p += sz;
                    640:
1.159     kristaps  641:                if ('\0' == *p)
                    642:                        break;
                    643:
1.154     kristaps  644:                if ('\\' == *p) {
                    645:                        /* Skip over escapes. */
                    646:                        p++;
                    647:                        esc = mandoc_escape
1.179     schwarze  648:                                ((const char const **)&p, NULL, NULL);
1.154     kristaps  649:                        if (ESCAPE_ERROR == esc)
                    650:                                break;
1.155     kristaps  651:                        continue;
1.159     kristaps  652:                } else if (p == start) {
1.158     kristaps  653:                        p++;
1.155     kristaps  654:                        continue;
1.158     kristaps  655:                }
1.155     kristaps  656:
1.171     schwarze  657:                if (isalpha((unsigned char)p[-1]) &&
                    658:                    isalpha((unsigned char)p[1]))
1.155     kristaps  659:                        *p = ASCII_HYPH;
                    660:                p++;
1.94      kristaps  661:        }
                    662:
1.178     schwarze  663:        /* Spring the input line trap. */
                    664:        if (1 == roffit_lines) {
                    665:                isz = asprintf(&p, "%s\n.%s", *bufp, roffit_macro);
                    666:                if (-1 == isz) {
                    667:                        perror(NULL);
                    668:                        exit((int)MANDOCLEVEL_SYSERR);
                    669:                }
                    670:                free(*bufp);
                    671:                *bufp = p;
                    672:                *szp = isz + 1;
                    673:                *offs = 0;
                    674:                free(roffit_macro);
                    675:                roffit_lines = 0;
                    676:                return(ROFF_REPARSE);
                    677:        } else if (1 < roffit_lines)
                    678:                --roffit_lines;
1.154     kristaps  679:        return(ROFF_CONT);
1.94      kristaps  680: }
                    681:
1.67      kristaps  682: enum rofferr
1.90      kristaps  683: roff_parseln(struct roff *r, int ln, char **bufp,
                    684:                size_t *szp, int pos, int *offs)
1.67      kristaps  685: {
                    686:        enum rofft       t;
1.109     kristaps  687:        enum rofferr     e;
1.130     kristaps  688:        int              ppos, ctl;
1.79      kristaps  689:
                    690:        /*
1.94      kristaps  691:         * Run the reserved-word filter only if we have some reserved
                    692:         * words to fill in.
                    693:         */
                    694:
1.172     schwarze  695:        e = roff_res(r, bufp, szp, ln, pos);
                    696:        if (ROFF_IGN == e)
                    697:                return(e);
                    698:        assert(ROFF_CONT == e);
1.94      kristaps  699:
1.130     kristaps  700:        ppos = pos;
1.174     kristaps  701:        ctl = roff_getcontrol(r, *bufp, &pos);
1.130     kristaps  702:
1.94      kristaps  703:        /*
1.79      kristaps  704:         * First, if a scope is open and we're not a macro, pass the
                    705:         * text through the macro's filter.  If a scope isn't open and
                    706:         * we're not a macro, just let it through.
1.125     kristaps  707:         * Finally, if there's an equation scope open, divert it into it
                    708:         * no matter our state.
1.79      kristaps  709:         */
1.74      kristaps  710:
1.130     kristaps  711:        if (r->last && ! ctl) {
1.78      kristaps  712:                t = r->last->tok;
                    713:                assert(roffs[t].text);
1.109     kristaps  714:                e = (*roffs[t].text)
                    715:                        (r, t, bufp, szp, ln, pos, pos, offs);
                    716:                assert(ROFF_IGN == e || ROFF_CONT == e);
1.125     kristaps  717:                if (ROFF_CONT != e)
                    718:                        return(e);
1.182     schwarze  719:        }
                    720:        if (r->eqn)
                    721:                return(eqn_read(&r->eqn, ln, *bufp, ppos, offs));
                    722:        if ( ! ctl) {
1.125     kristaps  723:                if (r->tbl)
1.130     kristaps  724:                        return(tbl_read(r->tbl, ln, *bufp, pos));
1.178     schwarze  725:                return(roff_parsetext(bufp, szp, pos, offs));
1.182     schwarze  726:        }
1.67      kristaps  727:
1.79      kristaps  728:        /*
                    729:         * If a scope is open, go to the child handler for that macro,
                    730:         * as it may want to preprocess before doing anything with it.
1.125     kristaps  731:         * Don't do so if an equation is open.
1.79      kristaps  732:         */
1.78      kristaps  733:
1.79      kristaps  734:        if (r->last) {
                    735:                t = r->last->tok;
                    736:                assert(roffs[t].sub);
                    737:                return((*roffs[t].sub)
1.90      kristaps  738:                                (r, t, bufp, szp,
1.130     kristaps  739:                                 ln, ppos, pos, offs));
1.79      kristaps  740:        }
1.78      kristaps  741:
1.79      kristaps  742:        /*
                    743:         * Lastly, as we've no scope open, try to look up and execute
                    744:         * the new macro.  If no macro is found, simply return and let
                    745:         * the compilers handle it.
                    746:         */
1.67      kristaps  747:
1.106     kristaps  748:        if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos)))
1.79      kristaps  749:                return(ROFF_CONT);
1.67      kristaps  750:
1.75      kristaps  751:        assert(roffs[t].proc);
1.78      kristaps  752:        return((*roffs[t].proc)
1.90      kristaps  753:                        (r, t, bufp, szp,
                    754:                         ln, ppos, pos, offs));
1.74      kristaps  755: }
                    756:
                    757:
1.117     kristaps  758: void
1.74      kristaps  759: roff_endparse(struct roff *r)
                    760: {
                    761:
1.110     kristaps  762:        if (r->last)
1.128     kristaps  763:                mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.109     kristaps  764:                                r->last->line, r->last->col, NULL);
1.117     kristaps  765:
1.125     kristaps  766:        if (r->eqn) {
1.128     kristaps  767:                mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.148     kristaps  768:                                r->eqn->eqn.ln, r->eqn->eqn.pos, NULL);
1.151     kristaps  769:                eqn_end(&r->eqn);
1.125     kristaps  770:        }
                    771:
1.117     kristaps  772:        if (r->tbl) {
1.128     kristaps  773:                mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.117     kristaps  774:                                r->tbl->line, r->tbl->pos, NULL);
1.151     kristaps  775:                tbl_end(&r->tbl);
1.117     kristaps  776:        }
1.67      kristaps  777: }
                    778:
                    779: /*
                    780:  * Parse a roff node's type from the input buffer.  This must be in the
                    781:  * form of ".foo xxx" in the usual way.
                    782:  */
                    783: static enum rofft
1.106     kristaps  784: roff_parse(struct roff *r, const char *buf, int *pos)
1.67      kristaps  785: {
1.106     kristaps  786:        const char      *mac;
                    787:        size_t           maclen;
1.67      kristaps  788:        enum rofft       t;
                    789:
1.144     kristaps  790:        if ('\0' == buf[*pos] || '"' == buf[*pos] ||
                    791:                        '\t' == buf[*pos] || ' ' == buf[*pos])
1.67      kristaps  792:                return(ROFF_MAX);
                    793:
1.144     kristaps  794:        /*
                    795:         * We stop the macro parse at an escape, tab, space, or nil.
                    796:         * However, `\}' is also a valid macro, so make sure we don't
                    797:         * clobber it by seeing the `\' as the end of token.
                    798:         */
                    799:
1.106     kristaps  800:        mac = buf + *pos;
1.144     kristaps  801:        maclen = strcspn(mac + 1, " \\\t\0") + 1;
1.67      kristaps  802:
1.106     kristaps  803:        t = (r->current_string = roff_getstrn(r, mac, maclen))
1.155     kristaps  804:            ? ROFF_USERDEF : roffhash_find(mac, maclen);
1.67      kristaps  805:
1.127     kristaps  806:        *pos += (int)maclen;
1.130     kristaps  807:
1.67      kristaps  808:        while (buf[*pos] && ' ' == buf[*pos])
                    809:                (*pos)++;
                    810:
                    811:        return(t);
                    812: }
                    813:
                    814: /* ARGSUSED */
                    815: static enum rofferr
1.76      kristaps  816: roff_cblock(ROFF_ARGS)
1.67      kristaps  817: {
                    818:
1.79      kristaps  819:        /*
                    820:         * A block-close `..' should only be invoked as a child of an
                    821:         * ignore macro, otherwise raise a warning and just ignore it.
                    822:         */
                    823:
1.76      kristaps  824:        if (NULL == r->last) {
1.128     kristaps  825:                mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.76      kristaps  826:                return(ROFF_IGN);
                    827:        }
1.67      kristaps  828:
1.81      kristaps  829:        switch (r->last->tok) {
                    830:        case (ROFF_am):
                    831:                /* FALLTHROUGH */
                    832:        case (ROFF_ami):
                    833:                /* FALLTHROUGH */
                    834:        case (ROFF_am1):
                    835:                /* FALLTHROUGH */
                    836:        case (ROFF_de):
1.108     schwarze  837:                /* ROFF_de1 is remapped to ROFF_de in roff_block(). */
1.81      kristaps  838:                /* FALLTHROUGH */
                    839:        case (ROFF_dei):
                    840:                /* FALLTHROUGH */
                    841:        case (ROFF_ig):
                    842:                break;
                    843:        default:
1.128     kristaps  844:                mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.67      kristaps  845:                return(ROFF_IGN);
1.76      kristaps  846:        }
1.67      kristaps  847:
1.76      kristaps  848:        if ((*bufp)[pos])
1.128     kristaps  849:                mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.71      kristaps  850:
                    851:        roffnode_pop(r);
1.76      kristaps  852:        roffnode_cleanscope(r);
                    853:        return(ROFF_IGN);
1.71      kristaps  854:
1.67      kristaps  855: }
                    856:
                    857:
1.76      kristaps  858: static void
                    859: roffnode_cleanscope(struct roff *r)
1.67      kristaps  860: {
                    861:
1.76      kristaps  862:        while (r->last) {
1.173     schwarze  863:                if (--r->last->endspan != 0)
1.76      kristaps  864:                        break;
                    865:                roffnode_pop(r);
                    866:        }
1.67      kristaps  867: }
                    868:
                    869:
1.75      kristaps  870: /* ARGSUSED */
1.74      kristaps  871: static enum rofferr
1.75      kristaps  872: roff_ccond(ROFF_ARGS)
1.74      kristaps  873: {
                    874:
1.76      kristaps  875:        if (NULL == r->last) {
1.128     kristaps  876:                mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.76      kristaps  877:                return(ROFF_IGN);
                    878:        }
                    879:
1.82      kristaps  880:        switch (r->last->tok) {
                    881:        case (ROFF_el):
                    882:                /* FALLTHROUGH */
                    883:        case (ROFF_ie):
                    884:                /* FALLTHROUGH */
                    885:        case (ROFF_if):
                    886:                break;
                    887:        default:
1.128     kristaps  888:                mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.75      kristaps  889:                return(ROFF_IGN);
                    890:        }
                    891:
1.76      kristaps  892:        if (r->last->endspan > -1) {
1.128     kristaps  893:                mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.76      kristaps  894:                return(ROFF_IGN);
                    895:        }
                    896:
                    897:        if ((*bufp)[pos])
1.128     kristaps  898:                mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.76      kristaps  899:
1.75      kristaps  900:        roffnode_pop(r);
1.76      kristaps  901:        roffnode_cleanscope(r);
                    902:        return(ROFF_IGN);
                    903: }
                    904:
1.75      kristaps  905:
1.76      kristaps  906: /* ARGSUSED */
                    907: static enum rofferr
1.80      kristaps  908: roff_block(ROFF_ARGS)
1.76      kristaps  909: {
1.78      kristaps  910:        int             sv;
                    911:        size_t          sz;
1.106     kristaps  912:        char            *name;
                    913:
                    914:        name = NULL;
1.76      kristaps  915:
1.106     kristaps  916:        if (ROFF_ig != tok) {
                    917:                if ('\0' == (*bufp)[pos]) {
1.128     kristaps  918:                        mandoc_msg(MANDOCERR_NOARGS, r->parse, ln, ppos, NULL);
1.106     kristaps  919:                        return(ROFF_IGN);
                    920:                }
1.107     kristaps  921:
                    922:                /*
                    923:                 * Re-write `de1', since we don't really care about
                    924:                 * groff's strange compatibility mode, into `de'.
                    925:                 */
                    926:
1.106     kristaps  927:                if (ROFF_de1 == tok)
                    928:                        tok = ROFF_de;
                    929:                if (ROFF_de == tok)
                    930:                        name = *bufp + pos;
                    931:                else
1.128     kristaps  932:                        mandoc_msg(MANDOCERR_REQUEST, r->parse, ln, ppos,
1.106     kristaps  933:                            roffs[tok].name);
1.107     kristaps  934:
1.131     schwarze  935:                while ((*bufp)[pos] && ! isspace((unsigned char)(*bufp)[pos]))
1.80      kristaps  936:                        pos++;
1.107     kristaps  937:
1.131     schwarze  938:                while (isspace((unsigned char)(*bufp)[pos]))
1.106     kristaps  939:                        (*bufp)[pos++] = '\0';
1.80      kristaps  940:        }
                    941:
1.106     kristaps  942:        roffnode_push(r, tok, name, ln, ppos);
                    943:
                    944:        /*
                    945:         * At the beginning of a `de' macro, clear the existing string
                    946:         * with the same name, if there is one.  New content will be
                    947:         * added from roff_block_text() in multiline mode.
                    948:         */
1.107     kristaps  949:
1.106     kristaps  950:        if (ROFF_de == tok)
1.108     schwarze  951:                roff_setstr(r, name, "", 0);
1.76      kristaps  952:
1.79      kristaps  953:        if ('\0' == (*bufp)[pos])
1.78      kristaps  954:                return(ROFF_IGN);
                    955:
1.107     kristaps  956:        /* If present, process the custom end-of-line marker. */
                    957:
1.78      kristaps  958:        sv = pos;
1.131     schwarze  959:        while ((*bufp)[pos] && ! isspace((unsigned char)(*bufp)[pos]))
1.78      kristaps  960:                pos++;
                    961:
                    962:        /*
                    963:         * Note: groff does NOT like escape characters in the input.
                    964:         * Instead of detecting this, we're just going to let it fly and
                    965:         * to hell with it.
                    966:         */
                    967:
                    968:        assert(pos > sv);
                    969:        sz = (size_t)(pos - sv);
                    970:
1.79      kristaps  971:        if (1 == sz && '.' == (*bufp)[sv])
                    972:                return(ROFF_IGN);
                    973:
1.98      schwarze  974:        r->last->end = mandoc_malloc(sz + 1);
1.78      kristaps  975:
                    976:        memcpy(r->last->end, *bufp + sv, sz);
                    977:        r->last->end[(int)sz] = '\0';
                    978:
1.77      kristaps  979:        if ((*bufp)[pos])
1.128     kristaps  980:                mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.74      kristaps  981:
1.78      kristaps  982:        return(ROFF_IGN);
                    983: }
                    984:
                    985:
                    986: /* ARGSUSED */
                    987: static enum rofferr
1.80      kristaps  988: roff_block_sub(ROFF_ARGS)
1.79      kristaps  989: {
                    990:        enum rofft      t;
                    991:        int             i, j;
                    992:
                    993:        /*
                    994:         * First check whether a custom macro exists at this level.  If
                    995:         * it does, then check against it.  This is some of groff's
                    996:         * stranger behaviours.  If we encountered a custom end-scope
                    997:         * tag and that tag also happens to be a "real" macro, then we
                    998:         * need to try interpreting it again as a real macro.  If it's
                    999:         * not, then return ignore.  Else continue.
                   1000:         */
                   1001:
                   1002:        if (r->last->end) {
1.130     kristaps 1003:                for (i = pos, j = 0; r->last->end[j]; j++, i++)
1.79      kristaps 1004:                        if ((*bufp)[i] != r->last->end[j])
                   1005:                                break;
                   1006:
                   1007:                if ('\0' == r->last->end[j] &&
                   1008:                                ('\0' == (*bufp)[i] ||
                   1009:                                 ' ' == (*bufp)[i] ||
                   1010:                                 '\t' == (*bufp)[i])) {
                   1011:                        roffnode_pop(r);
                   1012:                        roffnode_cleanscope(r);
                   1013:
1.130     kristaps 1014:                        while (' ' == (*bufp)[i] || '\t' == (*bufp)[i])
                   1015:                                i++;
                   1016:
                   1017:                        pos = i;
1.106     kristaps 1018:                        if (ROFF_MAX != roff_parse(r, *bufp, &pos))
1.79      kristaps 1019:                                return(ROFF_RERUN);
                   1020:                        return(ROFF_IGN);
                   1021:                }
                   1022:        }
                   1023:
                   1024:        /*
                   1025:         * If we have no custom end-query or lookup failed, then try
                   1026:         * pulling it out of the hashtable.
                   1027:         */
                   1028:
1.137     schwarze 1029:        t = roff_parse(r, *bufp, &pos);
1.79      kristaps 1030:
1.106     kristaps 1031:        /*
                   1032:         * Macros other than block-end are only significant
                   1033:         * in `de' blocks; elsewhere, simply throw them away.
                   1034:         */
                   1035:        if (ROFF_cblock != t) {
                   1036:                if (ROFF_de == tok)
                   1037:                        roff_setstr(r, r->last->name, *bufp + ppos, 1);
1.79      kristaps 1038:                return(ROFF_IGN);
1.106     kristaps 1039:        }
1.79      kristaps 1040:
                   1041:        assert(roffs[t].proc);
1.90      kristaps 1042:        return((*roffs[t].proc)(r, t, bufp, szp,
                   1043:                                ln, ppos, pos, offs));
1.79      kristaps 1044: }
                   1045:
                   1046:
                   1047: /* ARGSUSED */
                   1048: static enum rofferr
1.80      kristaps 1049: roff_block_text(ROFF_ARGS)
1.78      kristaps 1050: {
                   1051:
1.106     kristaps 1052:        if (ROFF_de == tok)
                   1053:                roff_setstr(r, r->last->name, *bufp + pos, 1);
                   1054:
1.78      kristaps 1055:        return(ROFF_IGN);
                   1056: }
                   1057:
                   1058:
                   1059: /* ARGSUSED */
                   1060: static enum rofferr
1.82      kristaps 1061: roff_cond_sub(ROFF_ARGS)
                   1062: {
                   1063:        enum rofft       t;
                   1064:        enum roffrule    rr;
1.139     kristaps 1065:        char            *ep;
1.82      kristaps 1066:
                   1067:        rr = r->last->rule;
1.139     kristaps 1068:        roffnode_cleanscope(r);
1.177     schwarze 1069:        t = roff_parse(r, *bufp, &pos);
1.82      kristaps 1070:
1.139     kristaps 1071:        /*
1.177     schwarze 1072:         * Fully handle known macros when they are structurally
                   1073:         * required or when the conditional evaluated to true.
1.87      kristaps 1074:         */
                   1075:
1.177     schwarze 1076:        if ((ROFF_MAX != t) &&
                   1077:            (ROFF_ccond == t || ROFFRULE_ALLOW == rr ||
                   1078:             ROFFMAC_STRUCT & roffs[t].flags)) {
                   1079:                assert(roffs[t].proc);
                   1080:                return((*roffs[t].proc)(r, t, bufp, szp,
                   1081:                                        ln, ppos, pos, offs));
                   1082:        }
1.144     kristaps 1083:
1.177     schwarze 1084:        /* Always check for the closing delimiter `\}'. */
1.144     kristaps 1085:
1.177     schwarze 1086:        ep = &(*bufp)[pos];
                   1087:        while (NULL != (ep = strchr(ep, '\\'))) {
                   1088:                if ('}' != *(++ep))
                   1089:                        continue;
1.82      kristaps 1090:
1.177     schwarze 1091:                /*
                   1092:                 * If we're at the end of line, then just chop
                   1093:                 * off the \} and resize the buffer.
                   1094:                 * If we aren't, then convert it to spaces.
                   1095:                 */
1.139     kristaps 1096:
1.177     schwarze 1097:                if ('\0' == *(ep + 1)) {
                   1098:                        *--ep = '\0';
                   1099:                        *szp -= 2;
                   1100:                } else
                   1101:                        *(ep - 1) = *ep = ' ';
1.82      kristaps 1102:
1.177     schwarze 1103:                roff_ccond(r, ROFF_ccond, bufp, szp,
                   1104:                                ln, pos, pos + 2, offs);
                   1105:                break;
                   1106:        }
                   1107:        return(ROFFRULE_DENY == rr ? ROFF_IGN : ROFF_CONT);
1.82      kristaps 1108: }
                   1109:
                   1110: /* ARGSUSED */
                   1111: static enum rofferr
                   1112: roff_cond_text(ROFF_ARGS)
1.78      kristaps 1113: {
1.140     kristaps 1114:        char            *ep;
1.82      kristaps 1115:        enum roffrule    rr;
                   1116:
                   1117:        rr = r->last->rule;
1.140     kristaps 1118:        roffnode_cleanscope(r);
1.82      kristaps 1119:
1.140     kristaps 1120:        ep = &(*bufp)[pos];
                   1121:        for ( ; NULL != (ep = strchr(ep, '\\')); ep++) {
                   1122:                ep++;
                   1123:                if ('}' != *ep)
                   1124:                        continue;
                   1125:                *ep = '&';
                   1126:                roff_ccond(r, ROFF_ccond, bufp, szp,
                   1127:                                ln, pos, pos + 2, offs);
1.78      kristaps 1128:        }
1.82      kristaps 1129:        return(ROFFRULE_DENY == rr ? ROFF_IGN : ROFF_CONT);
1.74      kristaps 1130: }
                   1131:
1.184     schwarze 1132: static int
                   1133: roff_getnum(const char *v, int *pos, int *res)
                   1134: {
                   1135:        int p, n;
                   1136:
                   1137:        p = *pos;
                   1138:        n = v[p] == '-';
                   1139:        if (n)
                   1140:                p++;
                   1141:
                   1142:        for (*res = 0; isdigit((unsigned char)v[p]); p++)
                   1143:                *res += 10 * *res + v[p] - '0';
                   1144:        if (p == *pos + n)
                   1145:                return 0;
                   1146:
                   1147:        if (n)
                   1148:                *res = -*res;
                   1149:
                   1150:        *pos = p;
                   1151:        return 1;
                   1152: }
                   1153:
                   1154: static int
                   1155: roff_getop(const char *v, int *pos, char *res)
                   1156: {
                   1157:        int e;
                   1158:
                   1159:        *res = v[*pos];
                   1160:        e = v[*pos + 1] == '=';
                   1161:
                   1162:        switch (*res) {
                   1163:        case '=':
                   1164:                break;
                   1165:        case '>':
                   1166:                if (e)
                   1167:                        *res = 'g';
                   1168:                break;
                   1169:        case '<':
                   1170:                if (e)
                   1171:                        *res = 'l';
                   1172:                break;
                   1173:        default:
                   1174:                return(0);
                   1175:        }
                   1176:
                   1177:        *pos += 1 + e;
                   1178:
                   1179:        return(*res);
                   1180: }
                   1181:
1.88      kristaps 1182: static enum roffrule
                   1183: roff_evalcond(const char *v, int *pos)
                   1184: {
1.184     schwarze 1185:        int      not, lh, rh;
                   1186:        char     op;
1.88      kristaps 1187:
                   1188:        switch (v[*pos]) {
                   1189:        case ('n'):
                   1190:                (*pos)++;
                   1191:                return(ROFFRULE_ALLOW);
                   1192:        case ('e'):
                   1193:                /* FALLTHROUGH */
                   1194:        case ('o'):
                   1195:                /* FALLTHROUGH */
                   1196:        case ('t'):
                   1197:                (*pos)++;
                   1198:                return(ROFFRULE_DENY);
1.184     schwarze 1199:        case ('!'):
                   1200:                (*pos)++;
                   1201:                not = 1;
                   1202:                break;
1.88      kristaps 1203:        default:
1.184     schwarze 1204:                not = 0;
1.88      kristaps 1205:                break;
                   1206:        }
                   1207:
1.184     schwarze 1208:        if (!roff_getnum(v, pos, &lh))
                   1209:                return ROFFRULE_DENY;
                   1210:        if (!roff_getop(v, pos, &op)) {
                   1211:                if (lh < 0)
                   1212:                        lh = 0;
                   1213:                goto out;
                   1214:        }
                   1215:        if (!roff_getnum(v, pos, &rh))
                   1216:                return ROFFRULE_DENY;
                   1217:        switch (op) {
                   1218:        case 'g':
                   1219:                lh = lh >= rh;
                   1220:                break;
                   1221:        case 'l':
                   1222:                lh = lh <= rh;
                   1223:                break;
                   1224:        case '=':
                   1225:                lh = lh == rh;
                   1226:                break;
                   1227:        case '>':
                   1228:                lh = lh > rh;
                   1229:                break;
                   1230:        case '<':
                   1231:                lh = lh < rh;
                   1232:                break;
                   1233:        default:
                   1234:                return ROFFRULE_DENY;
                   1235:        }
                   1236: out:
                   1237:        if (not)
                   1238:                lh = !lh;
                   1239:        return lh ? ROFFRULE_ALLOW : ROFFRULE_DENY;
1.88      kristaps 1240: }
                   1241:
1.75      kristaps 1242: /* ARGSUSED */
1.74      kristaps 1243: static enum rofferr
1.103     kristaps 1244: roff_line_ignore(ROFF_ARGS)
1.89      kristaps 1245: {
1.123     schwarze 1246:
1.89      kristaps 1247:        return(ROFF_IGN);
                   1248: }
                   1249:
1.104     kristaps 1250: /* ARGSUSED */
                   1251: static enum rofferr
1.82      kristaps 1252: roff_cond(ROFF_ARGS)
1.74      kristaps 1253: {
1.173     schwarze 1254:
                   1255:        roffnode_push(r, tok, NULL, ln, ppos);
1.74      kristaps 1256:
1.134     kristaps 1257:        /*
                   1258:         * An `.el' has no conditional body: it will consume the value
                   1259:         * of the current rstack entry set in prior `ie' calls or
                   1260:         * defaults to DENY.
                   1261:         *
                   1262:         * If we're not an `el', however, then evaluate the conditional.
                   1263:         */
1.133     kristaps 1264:
1.173     schwarze 1265:        r->last->rule = ROFF_el == tok ?
1.134     kristaps 1266:                (r->rstackpos < 0 ?
                   1267:                 ROFFRULE_DENY : r->rstack[r->rstackpos--]) :
                   1268:                roff_evalcond(*bufp, &pos);
1.77      kristaps 1269:
1.134     kristaps 1270:        /*
                   1271:         * An if-else will put the NEGATION of the current evaluated
                   1272:         * conditional into the stack of rules.
                   1273:         */
                   1274:
1.84      schwarze 1275:        if (ROFF_ie == tok) {
1.134     kristaps 1276:                if (r->rstackpos == RSTACK_MAX - 1) {
                   1277:                        mandoc_msg(MANDOCERR_MEM,
                   1278:                                r->parse, ln, ppos, NULL);
                   1279:                        return(ROFF_ERR);
                   1280:                }
                   1281:                r->rstack[++r->rstackpos] =
                   1282:                        ROFFRULE_DENY == r->last->rule ?
                   1283:                        ROFFRULE_ALLOW : ROFFRULE_DENY;
1.82      kristaps 1284:        }
1.88      kristaps 1285:
                   1286:        /* If the parent has false as its rule, then so do we. */
                   1287:
1.109     kristaps 1288:        if (r->last->parent && ROFFRULE_DENY == r->last->parent->rule)
1.84      schwarze 1289:                r->last->rule = ROFFRULE_DENY;
1.88      kristaps 1290:
                   1291:        /*
1.173     schwarze 1292:         * Determine scope.
                   1293:         * If there is nothing on the line after the conditional,
                   1294:         * not even whitespace, use next-line scope.
1.88      kristaps 1295:         */
1.74      kristaps 1296:
1.173     schwarze 1297:        if ('\0' == (*bufp)[pos]) {
                   1298:                r->last->endspan = 2;
                   1299:                goto out;
                   1300:        }
                   1301:
                   1302:        while (' ' == (*bufp)[pos])
                   1303:                pos++;
                   1304:
                   1305:        /* An opening brace requests multiline scope. */
1.75      kristaps 1306:
                   1307:        if ('\\' == (*bufp)[pos] && '{' == (*bufp)[pos + 1]) {
                   1308:                r->last->endspan = -1;
                   1309:                pos += 2;
1.173     schwarze 1310:                goto out;
1.109     kristaps 1311:        }
1.74      kristaps 1312:
1.77      kristaps 1313:        /*
1.173     schwarze 1314:         * Anything else following the conditional causes
                   1315:         * single-line scope.  Warn if the scope contains
                   1316:         * nothing but trailing whitespace.
1.77      kristaps 1317:         */
                   1318:
1.75      kristaps 1319:        if ('\0' == (*bufp)[pos])
1.173     schwarze 1320:                mandoc_msg(MANDOCERR_NOARGS, r->parse, ln, ppos, NULL);
1.77      kristaps 1321:
1.173     schwarze 1322:        r->last->endspan = 1;
1.74      kristaps 1323:
1.173     schwarze 1324: out:
1.75      kristaps 1325:        *offs = pos;
                   1326:        return(ROFF_RERUN);
1.83      schwarze 1327: }
                   1328:
                   1329:
                   1330: /* ARGSUSED */
                   1331: static enum rofferr
1.92      schwarze 1332: roff_ds(ROFF_ARGS)
                   1333: {
1.96      kristaps 1334:        char            *name, *string;
                   1335:
                   1336:        /*
                   1337:         * A symbol is named by the first word following the macro
                   1338:         * invocation up to a space.  Its value is anything after the
                   1339:         * name's trailing whitespace and optional double-quote.  Thus,
                   1340:         *
                   1341:         *  [.ds foo "bar  "     ]
                   1342:         *
                   1343:         * will have `bar  "     ' as its value.
                   1344:         */
1.92      schwarze 1345:
1.121     schwarze 1346:        string = *bufp + pos;
                   1347:        name = roff_getname(r, &string, ln, pos);
1.92      schwarze 1348:        if ('\0' == *name)
                   1349:                return(ROFF_IGN);
                   1350:
1.121     schwarze 1351:        /* Read past initial double-quote. */
                   1352:        if ('"' == *string)
1.92      schwarze 1353:                string++;
                   1354:
1.96      kristaps 1355:        /* The rest is the value. */
1.106     kristaps 1356:        roff_setstr(r, name, string, 0);
1.92      schwarze 1357:        return(ROFF_IGN);
                   1358: }
                   1359:
1.180     schwarze 1360: void
1.181     schwarze 1361: roff_setreg(struct roff *r, const char *name, int val)
1.147     kristaps 1362: {
1.180     schwarze 1363:        struct roffreg  *reg;
                   1364:
                   1365:        /* Search for an existing register with the same name. */
                   1366:        reg = r->regtab;
                   1367:
                   1368:        while (reg && strcmp(name, reg->key.p))
                   1369:                reg = reg->next;
1.147     kristaps 1370:
1.180     schwarze 1371:        if (NULL == reg) {
                   1372:                /* Create a new register. */
                   1373:                reg = mandoc_malloc(sizeof(struct roffreg));
                   1374:                reg->key.p = mandoc_strdup(name);
                   1375:                reg->key.sz = strlen(name);
                   1376:                reg->next = r->regtab;
                   1377:                r->regtab = reg;
                   1378:        }
                   1379:
1.181     schwarze 1380:        reg->val = val;
1.147     kristaps 1381: }
                   1382:
1.181     schwarze 1383: int
1.180     schwarze 1384: roff_getreg(const struct roff *r, const char *name)
1.147     kristaps 1385: {
1.180     schwarze 1386:        struct roffreg  *reg;
                   1387:
                   1388:        for (reg = r->regtab; reg; reg = reg->next)
                   1389:                if (0 == strcmp(name, reg->key.p))
1.181     schwarze 1390:                        return(reg->val);
                   1391:
                   1392:        return(0);
                   1393: }
                   1394:
                   1395: static int
                   1396: roff_getregn(const struct roff *r, const char *name, size_t len)
                   1397: {
                   1398:        struct roffreg  *reg;
                   1399:
                   1400:        for (reg = r->regtab; reg; reg = reg->next)
                   1401:                if (len == reg->key.sz &&
                   1402:                    0 == strncmp(name, reg->key.p, len))
                   1403:                        return(reg->val);
1.147     kristaps 1404:
1.180     schwarze 1405:        return(0);
1.147     kristaps 1406: }
                   1407:
1.180     schwarze 1408: static void
                   1409: roff_freereg(struct roffreg *reg)
1.147     kristaps 1410: {
1.180     schwarze 1411:        struct roffreg  *old_reg;
1.147     kristaps 1412:
1.180     schwarze 1413:        while (NULL != reg) {
                   1414:                free(reg->key.p);
                   1415:                old_reg = reg;
                   1416:                reg = reg->next;
                   1417:                free(old_reg);
                   1418:        }
1.147     kristaps 1419: }
1.92      schwarze 1420:
                   1421: /* ARGSUSED */
                   1422: static enum rofferr
1.89      kristaps 1423: roff_nr(ROFF_ARGS)
1.83      schwarze 1424: {
1.121     schwarze 1425:        const char      *key;
                   1426:        char            *val;
1.138     kristaps 1427:        int              iv;
1.89      kristaps 1428:
1.121     schwarze 1429:        val = *bufp + pos;
                   1430:        key = roff_getname(r, &val, ln, pos);
1.89      kristaps 1431:
1.180     schwarze 1432:        iv = mandoc_strntoi(val, strlen(val), 10);
                   1433:
1.181     schwarze 1434:        roff_setreg(r, key, iv);
1.109     kristaps 1435:
1.122     schwarze 1436:        return(ROFF_IGN);
                   1437: }
                   1438:
                   1439: /* ARGSUSED */
                   1440: static enum rofferr
                   1441: roff_rm(ROFF_ARGS)
                   1442: {
                   1443:        const char       *name;
                   1444:        char             *cp;
                   1445:
                   1446:        cp = *bufp + pos;
                   1447:        while ('\0' != *cp) {
1.127     kristaps 1448:                name = roff_getname(r, &cp, ln, (int)(cp - *bufp));
1.122     schwarze 1449:                if ('\0' != *name)
                   1450:                        roff_setstr(r, name, NULL, 0);
                   1451:        }
1.178     schwarze 1452:        return(ROFF_IGN);
                   1453: }
                   1454:
                   1455: /* ARGSUSED */
                   1456: static enum rofferr
                   1457: roff_it(ROFF_ARGS)
                   1458: {
                   1459:        char            *cp;
                   1460:        size_t           len;
                   1461:        int              iv;
                   1462:
                   1463:        /* Parse the number of lines. */
                   1464:        cp = *bufp + pos;
                   1465:        len = strcspn(cp, " \t");
                   1466:        cp[len] = '\0';
                   1467:        if ((iv = mandoc_strntoi(cp, len, 10)) <= 0) {
                   1468:                mandoc_msg(MANDOCERR_NUMERIC, r->parse,
                   1469:                                ln, ppos, *bufp + 1);
                   1470:                return(ROFF_IGN);
                   1471:        }
                   1472:        cp += len + 1;
                   1473:
                   1474:        /* Arm the input line trap. */
                   1475:        roffit_lines = iv;
                   1476:        roffit_macro = mandoc_strdup(cp);
1.109     kristaps 1477:        return(ROFF_IGN);
1.175     schwarze 1478: }
                   1479:
                   1480: /* ARGSUSED */
                   1481: static enum rofferr
                   1482: roff_Dd(ROFF_ARGS)
                   1483: {
                   1484:        const char *const       *cp;
                   1485:
                   1486:        if (MPARSE_MDOC != r->parsetype)
                   1487:                for (cp = __mdoc_reserved; *cp; cp++)
                   1488:                        roff_setstr(r, *cp, NULL, 0);
                   1489:
                   1490:        return(ROFF_CONT);
                   1491: }
                   1492:
                   1493: /* ARGSUSED */
                   1494: static enum rofferr
                   1495: roff_TH(ROFF_ARGS)
                   1496: {
                   1497:        const char *const       *cp;
                   1498:
                   1499:        if (MPARSE_MDOC != r->parsetype)
                   1500:                for (cp = __man_reserved; *cp; cp++)
                   1501:                        roff_setstr(r, *cp, NULL, 0);
                   1502:
                   1503:        return(ROFF_CONT);
1.109     kristaps 1504: }
                   1505:
                   1506: /* ARGSUSED */
                   1507: static enum rofferr
                   1508: roff_TE(ROFF_ARGS)
                   1509: {
                   1510:
                   1511:        if (NULL == r->tbl)
1.128     kristaps 1512:                mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.115     kristaps 1513:        else
1.151     kristaps 1514:                tbl_end(&r->tbl);
1.109     kristaps 1515:
1.112     kristaps 1516:        return(ROFF_IGN);
                   1517: }
                   1518:
                   1519: /* ARGSUSED */
                   1520: static enum rofferr
                   1521: roff_T_(ROFF_ARGS)
                   1522: {
                   1523:
                   1524:        if (NULL == r->tbl)
1.128     kristaps 1525:                mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.112     kristaps 1526:        else
1.116     kristaps 1527:                tbl_restart(ppos, ln, r->tbl);
1.112     kristaps 1528:
1.109     kristaps 1529:        return(ROFF_IGN);
                   1530: }
                   1531:
1.156     kristaps 1532: #if 0
                   1533: static int
1.151     kristaps 1534: roff_closeeqn(struct roff *r)
                   1535: {
                   1536:
                   1537:        return(r->eqn && ROFF_EQN == eqn_end(&r->eqn) ? 1 : 0);
                   1538: }
1.156     kristaps 1539: #endif
1.151     kristaps 1540:
1.156     kristaps 1541: static void
1.151     kristaps 1542: roff_openeqn(struct roff *r, const char *name, int line,
                   1543:                int offs, const char *buf)
1.125     kristaps 1544: {
1.151     kristaps 1545:        struct eqn_node *e;
                   1546:        int              poff;
1.125     kristaps 1547:
                   1548:        assert(NULL == r->eqn);
1.151     kristaps 1549:        e = eqn_alloc(name, offs, line, r->parse);
1.125     kristaps 1550:
                   1551:        if (r->last_eqn)
                   1552:                r->last_eqn->next = e;
                   1553:        else
                   1554:                r->first_eqn = r->last_eqn = e;
                   1555:
                   1556:        r->eqn = r->last_eqn = e;
1.151     kristaps 1557:
                   1558:        if (buf) {
                   1559:                poff = 0;
                   1560:                eqn_read(&r->eqn, line, buf, offs, &poff);
                   1561:        }
                   1562: }
                   1563:
                   1564: /* ARGSUSED */
                   1565: static enum rofferr
                   1566: roff_EQ(ROFF_ARGS)
                   1567: {
                   1568:
                   1569:        roff_openeqn(r, *bufp + pos, ln, ppos, NULL);
1.125     kristaps 1570:        return(ROFF_IGN);
                   1571: }
                   1572:
                   1573: /* ARGSUSED */
                   1574: static enum rofferr
                   1575: roff_EN(ROFF_ARGS)
                   1576: {
                   1577:
1.128     kristaps 1578:        mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.125     kristaps 1579:        return(ROFF_IGN);
                   1580: }
                   1581:
                   1582: /* ARGSUSED */
                   1583: static enum rofferr
1.109     kristaps 1584: roff_TS(ROFF_ARGS)
                   1585: {
1.176     schwarze 1586:        struct tbl_node *tbl;
1.89      kristaps 1587:
1.115     kristaps 1588:        if (r->tbl) {
1.128     kristaps 1589:                mandoc_msg(MANDOCERR_SCOPEBROKEN, r->parse, ln, ppos, NULL);
1.151     kristaps 1590:                tbl_end(&r->tbl);
1.115     kristaps 1591:        }
1.83      schwarze 1592:
1.176     schwarze 1593:        tbl = tbl_alloc(ppos, ln, r->parse);
1.113     kristaps 1594:
                   1595:        if (r->last_tbl)
1.176     schwarze 1596:                r->last_tbl->next = tbl;
1.113     kristaps 1597:        else
1.176     schwarze 1598:                r->first_tbl = r->last_tbl = tbl;
1.113     kristaps 1599:
1.176     schwarze 1600:        r->tbl = r->last_tbl = tbl;
1.83      schwarze 1601:        return(ROFF_IGN);
1.92      schwarze 1602: }
                   1603:
1.105     kristaps 1604: /* ARGSUSED */
                   1605: static enum rofferr
1.174     kristaps 1606: roff_cc(ROFF_ARGS)
                   1607: {
                   1608:        const char      *p;
                   1609:
                   1610:        p = *bufp + pos;
                   1611:
                   1612:        if ('\0' == *p || '.' == (r->control = *p++))
                   1613:                r->control = 0;
                   1614:
                   1615:        if ('\0' != *p)
                   1616:                mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
                   1617:
                   1618:        return(ROFF_IGN);
                   1619: }
                   1620:
                   1621: /* ARGSUSED */
                   1622: static enum rofferr
1.164     kristaps 1623: roff_tr(ROFF_ARGS)
                   1624: {
                   1625:        const char      *p, *first, *second;
                   1626:        size_t           fsz, ssz;
                   1627:        enum mandoc_esc  esc;
                   1628:
                   1629:        p = *bufp + pos;
                   1630:
                   1631:        if ('\0' == *p) {
                   1632:                mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
                   1633:                return(ROFF_IGN);
                   1634:        }
                   1635:
                   1636:        while ('\0' != *p) {
                   1637:                fsz = ssz = 1;
                   1638:
                   1639:                first = p++;
                   1640:                if ('\\' == *first) {
                   1641:                        esc = mandoc_escape(&p, NULL, NULL);
                   1642:                        if (ESCAPE_ERROR == esc) {
                   1643:                                mandoc_msg
                   1644:                                        (MANDOCERR_BADESCAPE, r->parse,
                   1645:                                         ln, (int)(p - *bufp), NULL);
                   1646:                                return(ROFF_IGN);
                   1647:                        }
                   1648:                        fsz = (size_t)(p - first);
                   1649:                }
                   1650:
                   1651:                second = p++;
                   1652:                if ('\\' == *second) {
                   1653:                        esc = mandoc_escape(&p, NULL, NULL);
                   1654:                        if (ESCAPE_ERROR == esc) {
                   1655:                                mandoc_msg
                   1656:                                        (MANDOCERR_BADESCAPE, r->parse,
                   1657:                                         ln, (int)(p - *bufp), NULL);
                   1658:                                return(ROFF_IGN);
                   1659:                        }
                   1660:                        ssz = (size_t)(p - second);
1.165     kristaps 1661:                } else if ('\0' == *second) {
1.164     kristaps 1662:                        mandoc_msg(MANDOCERR_ARGCOUNT, r->parse,
                   1663:                                        ln, (int)(p - *bufp), NULL);
                   1664:                        second = " ";
1.165     kristaps 1665:                        p--;
1.164     kristaps 1666:                }
                   1667:
1.167     kristaps 1668:                if (fsz > 1) {
                   1669:                        roff_setstrn(&r->xmbtab, first,
                   1670:                                        fsz, second, ssz, 0);
                   1671:                        continue;
                   1672:                }
                   1673:
                   1674:                if (NULL == r->xtab)
                   1675:                        r->xtab = mandoc_calloc
                   1676:                                (128, sizeof(struct roffstr));
                   1677:
                   1678:                free(r->xtab[(int)*first].p);
                   1679:                r->xtab[(int)*first].p = mandoc_strndup(second, ssz);
                   1680:                r->xtab[(int)*first].sz = ssz;
1.164     kristaps 1681:        }
                   1682:
                   1683:        return(ROFF_IGN);
                   1684: }
                   1685:
                   1686: /* ARGSUSED */
                   1687: static enum rofferr
1.105     kristaps 1688: roff_so(ROFF_ARGS)
                   1689: {
                   1690:        char *name;
                   1691:
1.128     kristaps 1692:        mandoc_msg(MANDOCERR_SO, r->parse, ln, ppos, NULL);
1.105     kristaps 1693:
                   1694:        /*
                   1695:         * Handle `so'.  Be EXTREMELY careful, as we shouldn't be
                   1696:         * opening anything that's not in our cwd or anything beneath
                   1697:         * it.  Thus, explicitly disallow traversing up the file-system
                   1698:         * or using absolute paths.
                   1699:         */
                   1700:
                   1701:        name = *bufp + pos;
                   1702:        if ('/' == *name || strstr(name, "../") || strstr(name, "/..")) {
1.128     kristaps 1703:                mandoc_msg(MANDOCERR_SOPATH, r->parse, ln, pos, NULL);
1.105     kristaps 1704:                return(ROFF_ERR);
                   1705:        }
                   1706:
                   1707:        *offs = pos;
                   1708:        return(ROFF_SO);
                   1709: }
1.92      schwarze 1710:
1.106     kristaps 1711: /* ARGSUSED */
                   1712: static enum rofferr
                   1713: roff_userdef(ROFF_ARGS)
1.99      kristaps 1714: {
1.106     kristaps 1715:        const char       *arg[9];
                   1716:        char             *cp, *n1, *n2;
1.119     schwarze 1717:        int               i;
1.106     kristaps 1718:
                   1719:        /*
                   1720:         * Collect pointers to macro argument strings
                   1721:         * and null-terminate them.
                   1722:         */
                   1723:        cp = *bufp + pos;
1.119     schwarze 1724:        for (i = 0; i < 9; i++)
1.120     schwarze 1725:                arg[i] = '\0' == *cp ? "" :
1.136     kristaps 1726:                    mandoc_getarg(r->parse, &cp, ln, &pos);
1.99      kristaps 1727:
1.106     kristaps 1728:        /*
                   1729:         * Expand macro arguments.
1.99      kristaps 1730:         */
1.106     kristaps 1731:        *szp = 0;
                   1732:        n1 = cp = mandoc_strdup(r->current_string);
                   1733:        while (NULL != (cp = strstr(cp, "\\$"))) {
                   1734:                i = cp[2] - '1';
                   1735:                if (0 > i || 8 < i) {
                   1736:                        /* Not an argument invocation. */
                   1737:                        cp += 2;
                   1738:                        continue;
                   1739:                }
                   1740:
                   1741:                *szp = strlen(n1) - 3 + strlen(arg[i]) + 1;
                   1742:                n2 = mandoc_malloc(*szp);
                   1743:
                   1744:                strlcpy(n2, n1, (size_t)(cp - n1 + 1));
                   1745:                strlcat(n2, arg[i], *szp);
                   1746:                strlcat(n2, cp + 3, *szp);
                   1747:
                   1748:                cp = n2 + (cp - n1);
                   1749:                free(n1);
                   1750:                n1 = n2;
1.99      kristaps 1751:        }
                   1752:
1.106     kristaps 1753:        /*
                   1754:         * Replace the macro invocation
                   1755:         * by the expanded macro.
                   1756:         */
                   1757:        free(*bufp);
                   1758:        *bufp = n1;
                   1759:        if (0 == *szp)
                   1760:                *szp = strlen(*bufp) + 1;
                   1761:
                   1762:        return(*szp > 1 && '\n' == (*bufp)[(int)*szp - 2] ?
                   1763:           ROFF_REPARSE : ROFF_APPEND);
1.99      kristaps 1764: }
1.121     schwarze 1765:
                   1766: static char *
                   1767: roff_getname(struct roff *r, char **cpp, int ln, int pos)
                   1768: {
                   1769:        char     *name, *cp;
                   1770:
                   1771:        name = *cpp;
                   1772:        if ('\0' == *name)
                   1773:                return(name);
                   1774:
                   1775:        /* Read until end of name. */
                   1776:        for (cp = name; '\0' != *cp && ' ' != *cp; cp++) {
                   1777:                if ('\\' != *cp)
                   1778:                        continue;
                   1779:                cp++;
                   1780:                if ('\\' == *cp)
                   1781:                        continue;
1.128     kristaps 1782:                mandoc_msg(MANDOCERR_NAMESC, r->parse, ln, pos, NULL);
1.121     schwarze 1783:                *cp = '\0';
                   1784:                name = cp;
                   1785:        }
                   1786:
                   1787:        /* Nil-terminate name. */
                   1788:        if ('\0' != *cp)
                   1789:                *(cp++) = '\0';
                   1790:
                   1791:        /* Read past spaces. */
                   1792:        while (' ' == *cp)
                   1793:                cp++;
                   1794:
                   1795:        *cpp = cp;
                   1796:        return(name);
                   1797: }
                   1798:
1.106     kristaps 1799: /*
                   1800:  * Store *string into the user-defined string called *name.
                   1801:  * In multiline mode, append to an existing entry and append '\n';
                   1802:  * else replace the existing entry, if there is one.
                   1803:  * To clear an existing entry, call with (*r, *name, NULL, 0).
                   1804:  */
1.94      kristaps 1805: static void
1.106     kristaps 1806: roff_setstr(struct roff *r, const char *name, const char *string,
                   1807:        int multiline)
1.92      schwarze 1808: {
1.164     kristaps 1809:
                   1810:        roff_setstrn(&r->strtab, name, strlen(name), string,
                   1811:                        string ? strlen(string) : 0, multiline);
                   1812: }
                   1813:
                   1814: static void
1.166     kristaps 1815: roff_setstrn(struct roffkv **r, const char *name, size_t namesz,
1.164     kristaps 1816:                const char *string, size_t stringsz, int multiline)
                   1817: {
1.166     kristaps 1818:        struct roffkv   *n;
1.164     kristaps 1819:        char            *c;
                   1820:        int              i;
                   1821:        size_t           oldch, newch;
1.92      schwarze 1822:
1.106     kristaps 1823:        /* Search for an existing string with the same name. */
1.164     kristaps 1824:        n = *r;
                   1825:
1.166     kristaps 1826:        while (n && strcmp(name, n->key.p))
1.92      schwarze 1827:                n = n->next;
1.94      kristaps 1828:
                   1829:        if (NULL == n) {
1.106     kristaps 1830:                /* Create a new string table entry. */
1.166     kristaps 1831:                n = mandoc_malloc(sizeof(struct roffkv));
                   1832:                n->key.p = mandoc_strndup(name, namesz);
                   1833:                n->key.sz = namesz;
                   1834:                n->val.p = NULL;
                   1835:                n->val.sz = 0;
1.164     kristaps 1836:                n->next = *r;
                   1837:                *r = n;
1.106     kristaps 1838:        } else if (0 == multiline) {
                   1839:                /* In multiline mode, append; else replace. */
1.166     kristaps 1840:                free(n->val.p);
                   1841:                n->val.p = NULL;
                   1842:                n->val.sz = 0;
1.106     kristaps 1843:        }
                   1844:
                   1845:        if (NULL == string)
                   1846:                return;
                   1847:
                   1848:        /*
                   1849:         * One additional byte for the '\n' in multiline mode,
                   1850:         * and one for the terminating '\0'.
                   1851:         */
1.164     kristaps 1852:        newch = stringsz + (multiline ? 2u : 1u);
                   1853:
1.166     kristaps 1854:        if (NULL == n->val.p) {
                   1855:                n->val.p = mandoc_malloc(newch);
                   1856:                *n->val.p = '\0';
1.106     kristaps 1857:                oldch = 0;
                   1858:        } else {
1.166     kristaps 1859:                oldch = n->val.sz;
                   1860:                n->val.p = mandoc_realloc(n->val.p, oldch + newch);
1.106     kristaps 1861:        }
                   1862:
                   1863:        /* Skip existing content in the destination buffer. */
1.166     kristaps 1864:        c = n->val.p + (int)oldch;
1.106     kristaps 1865:
                   1866:        /* Append new content to the destination buffer. */
1.164     kristaps 1867:        i = 0;
                   1868:        while (i < (int)stringsz) {
1.106     kristaps 1869:                /*
                   1870:                 * Rudimentary roff copy mode:
                   1871:                 * Handle escaped backslashes.
                   1872:                 */
1.164     kristaps 1873:                if ('\\' == string[i] && '\\' == string[i + 1])
                   1874:                        i++;
                   1875:                *c++ = string[i++];
1.106     kristaps 1876:        }
1.94      kristaps 1877:
1.106     kristaps 1878:        /* Append terminating bytes. */
                   1879:        if (multiline)
                   1880:                *c++ = '\n';
1.163     kristaps 1881:
1.106     kristaps 1882:        *c = '\0';
1.166     kristaps 1883:        n->val.sz = (int)(c - n->val.p);
1.92      schwarze 1884: }
                   1885:
1.94      kristaps 1886: static const char *
                   1887: roff_getstrn(const struct roff *r, const char *name, size_t len)
1.92      schwarze 1888: {
1.166     kristaps 1889:        const struct roffkv *n;
1.92      schwarze 1890:
1.164     kristaps 1891:        for (n = r->strtab; n; n = n->next)
1.166     kristaps 1892:                if (0 == strncmp(name, n->key.p, len) &&
                   1893:                                '\0' == n->key.p[(int)len])
                   1894:                        return(n->val.p);
1.94      kristaps 1895:
1.157     kristaps 1896:        return(NULL);
1.92      schwarze 1897: }
                   1898:
1.94      kristaps 1899: static void
1.167     kristaps 1900: roff_freestr(struct roffkv *r)
1.92      schwarze 1901: {
1.166     kristaps 1902:        struct roffkv    *n, *nn;
1.92      schwarze 1903:
1.167     kristaps 1904:        for (n = r; n; n = nn) {
1.166     kristaps 1905:                free(n->key.p);
                   1906:                free(n->val.p);
1.92      schwarze 1907:                nn = n->next;
                   1908:                free(n);
                   1909:        }
1.114     kristaps 1910: }
                   1911:
                   1912: const struct tbl_span *
                   1913: roff_span(const struct roff *r)
                   1914: {
                   1915:
                   1916:        return(r->tbl ? tbl_span(r->tbl) : NULL);
1.125     kristaps 1917: }
                   1918:
                   1919: const struct eqn *
                   1920: roff_eqn(const struct roff *r)
                   1921: {
                   1922:
                   1923:        return(r->last_eqn ? &r->last_eqn->eqn : NULL);
1.164     kristaps 1924: }
                   1925:
                   1926: /*
                   1927:  * Duplicate an input string, making the appropriate character
                   1928:  * conversations (as stipulated by `tr') along the way.
                   1929:  * Returns a heap-allocated string with all the replacements made.
                   1930:  */
                   1931: char *
                   1932: roff_strdup(const struct roff *r, const char *p)
                   1933: {
1.166     kristaps 1934:        const struct roffkv *cp;
1.164     kristaps 1935:        char            *res;
                   1936:        const char      *pp;
                   1937:        size_t           ssz, sz;
                   1938:        enum mandoc_esc  esc;
                   1939:
1.167     kristaps 1940:        if (NULL == r->xmbtab && NULL == r->xtab)
1.164     kristaps 1941:                return(mandoc_strdup(p));
                   1942:        else if ('\0' == *p)
                   1943:                return(mandoc_strdup(""));
                   1944:
                   1945:        /*
                   1946:         * Step through each character looking for term matches
                   1947:         * (remember that a `tr' can be invoked with an escape, which is
                   1948:         * a glyph but the escape is multi-character).
                   1949:         * We only do this if the character hash has been initialised
                   1950:         * and the string is >0 length.
                   1951:         */
                   1952:
                   1953:        res = NULL;
                   1954:        ssz = 0;
                   1955:
                   1956:        while ('\0' != *p) {
1.167     kristaps 1957:                if ('\\' != *p && r->xtab && r->xtab[(int)*p].p) {
                   1958:                        sz = r->xtab[(int)*p].sz;
                   1959:                        res = mandoc_realloc(res, ssz + sz + 1);
                   1960:                        memcpy(res + ssz, r->xtab[(int)*p].p, sz);
                   1961:                        ssz += sz;
                   1962:                        p++;
                   1963:                        continue;
                   1964:                } else if ('\\' != *p) {
                   1965:                        res = mandoc_realloc(res, ssz + 2);
                   1966:                        res[ssz++] = *p++;
                   1967:                        continue;
                   1968:                }
                   1969:
1.164     kristaps 1970:                /* Search for term matches. */
1.167     kristaps 1971:                for (cp = r->xmbtab; cp; cp = cp->next)
1.166     kristaps 1972:                        if (0 == strncmp(p, cp->key.p, cp->key.sz))
1.164     kristaps 1973:                                break;
                   1974:
                   1975:                if (NULL != cp) {
                   1976:                        /*
                   1977:                         * A match has been found.
                   1978:                         * Append the match to the array and move
                   1979:                         * forward by its keysize.
                   1980:                         */
1.166     kristaps 1981:                        res = mandoc_realloc
                   1982:                                (res, ssz + cp->val.sz + 1);
                   1983:                        memcpy(res + ssz, cp->val.p, cp->val.sz);
                   1984:                        ssz += cp->val.sz;
                   1985:                        p += (int)cp->key.sz;
1.164     kristaps 1986:                        continue;
                   1987:                }
                   1988:
1.167     kristaps 1989:                /*
                   1990:                 * Handle escapes carefully: we need to copy
                   1991:                 * over just the escape itself, or else we might
                   1992:                 * do replacements within the escape itself.
                   1993:                 * Make sure to pass along the bogus string.
                   1994:                 */
                   1995:                pp = p++;
                   1996:                esc = mandoc_escape(&p, NULL, NULL);
                   1997:                if (ESCAPE_ERROR == esc) {
                   1998:                        sz = strlen(pp);
1.164     kristaps 1999:                        res = mandoc_realloc(res, ssz + sz + 1);
                   2000:                        memcpy(res + ssz, pp, sz);
1.167     kristaps 2001:                        break;
1.164     kristaps 2002:                }
1.167     kristaps 2003:                /*
                   2004:                 * We bail out on bad escapes.
                   2005:                 * No need to warn: we already did so when
                   2006:                 * roff_res() was called.
                   2007:                 */
                   2008:                sz = (int)(p - pp);
                   2009:                res = mandoc_realloc(res, ssz + sz + 1);
                   2010:                memcpy(res + ssz, pp, sz);
                   2011:                ssz += sz;
1.164     kristaps 2012:        }
                   2013:
                   2014:        res[(int)ssz] = '\0';
                   2015:        return(res);
1.174     kristaps 2016: }
                   2017:
                   2018: /*
                   2019:  * Find out whether a line is a macro line or not.
                   2020:  * If it is, adjust the current position and return one; if it isn't,
                   2021:  * return zero and don't change the current position.
                   2022:  * If the control character has been set with `.cc', then let that grain
                   2023:  * precedence.
                   2024:  * This is slighly contrary to groff, where using the non-breaking
                   2025:  * control character when `cc' has been invoked will cause the
                   2026:  * non-breaking macro contents to be printed verbatim.
                   2027:  */
                   2028: int
                   2029: roff_getcontrol(const struct roff *r, const char *cp, int *ppos)
                   2030: {
                   2031:        int             pos;
                   2032:
                   2033:        pos = *ppos;
                   2034:
                   2035:        if (0 != r->control && cp[pos] == r->control)
                   2036:                pos++;
                   2037:        else if (0 != r->control)
                   2038:                return(0);
                   2039:        else if ('\\' == cp[pos] && '.' == cp[pos + 1])
                   2040:                pos += 2;
                   2041:        else if ('.' == cp[pos] || '\'' == cp[pos])
                   2042:                pos++;
                   2043:        else
                   2044:                return(0);
                   2045:
                   2046:        while (' ' == cp[pos] || '\t' == cp[pos])
                   2047:                pos++;
                   2048:
                   2049:        *ppos = pos;
                   2050:        return(1);
1.74      kristaps 2051: }

CVSweb