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

Annotation of mandoc/roff.c, Revision 1.197

1.197   ! schwarze    1: /*     $Id: roff.c,v 1.196 2014/03/07 18:30:11 schwarze Exp $ */
1.1       kristaps    2: /*
1.175     schwarze    3:  * Copyright (c) 2010, 2011, 2012 Kristaps Dzonsons <kristaps@bsd.lv>
1.190     schwarze    4:  * Copyright (c) 2010-2014 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,
1.193     schwarze   43:        ROFF_as,
1.174     kristaps   44:        ROFF_cc,
1.194     schwarze   45:        ROFF_ce,
1.80      kristaps   46:        ROFF_de,
                     47:        ROFF_dei,
                     48:        ROFF_de1,
1.83      schwarze   49:        ROFF_ds,
1.82      kristaps   50:        ROFF_el,
1.185     schwarze   51:        ROFF_fam,
1.186     schwarze   52:        ROFF_hw,
1.103     kristaps   53:        ROFF_hy,
1.82      kristaps   54:        ROFF_ie,
1.75      kristaps   55:        ROFF_if,
1.76      kristaps   56:        ROFF_ig,
1.123     schwarze   57:        ROFF_it,
1.103     kristaps   58:        ROFF_ne,
                     59:        ROFF_nh,
1.104     kristaps   60:        ROFF_nr,
1.124     schwarze   61:        ROFF_ns,
                     62:        ROFF_ps,
1.83      schwarze   63:        ROFF_rm,
1.105     kristaps   64:        ROFF_so,
1.124     schwarze   65:        ROFF_ta,
1.83      schwarze   66:        ROFF_tr,
1.175     schwarze   67:        ROFF_Dd,
                     68:        ROFF_TH,
1.109     kristaps   69:        ROFF_TS,
                     70:        ROFF_TE,
1.112     kristaps   71:        ROFF_T_,
1.125     kristaps   72:        ROFF_EQ,
                     73:        ROFF_EN,
1.76      kristaps   74:        ROFF_cblock,
1.106     kristaps   75:        ROFF_USERDEF,
1.67      kristaps   76:        ROFF_MAX
                     77: };
                     78:
1.82      kristaps   79: enum   roffrule {
1.183     schwarze   80:        ROFFRULE_DENY,
                     81:        ROFFRULE_ALLOW
1.82      kristaps   82: };
                     83:
1.147     kristaps   84: /*
1.167     kristaps   85:  * An incredibly-simple string buffer.
                     86:  */
1.94      kristaps   87: struct roffstr {
1.167     kristaps   88:        char            *p; /* nil-terminated buffer */
                     89:        size_t           sz; /* saved strlen(p) */
1.166     kristaps   90: };
                     91:
                     92: /*
1.167     kristaps   93:  * A key-value roffstr pair as part of a singly-linked list.
1.166     kristaps   94:  */
                     95: struct roffkv {
                     96:        struct roffstr   key;
                     97:        struct roffstr   val;
                     98:        struct roffkv   *next; /* next in list */
1.94      kristaps   99: };
                    100:
1.180     schwarze  101: /*
                    102:  * A single number register as part of a singly-linked list.
                    103:  */
                    104: struct roffreg {
                    105:        struct roffstr   key;
1.181     schwarze  106:        int              val;
1.180     schwarze  107:        struct roffreg  *next;
                    108: };
                    109:
1.67      kristaps  110: struct roff {
1.175     schwarze  111:        enum mparset     parsetype; /* requested parse type */
1.128     kristaps  112:        struct mparse   *parse; /* parse point */
1.190     schwarze  113:        int              quick; /* skip standard macro deletion */
1.67      kristaps  114:        struct roffnode *last; /* leaf of stack */
1.82      kristaps  115:        enum roffrule    rstack[RSTACK_MAX]; /* stack of !`ie' rules */
1.174     kristaps  116:        char             control; /* control character */
1.82      kristaps  117:        int              rstackpos; /* position in rstack */
1.180     schwarze  118:        struct roffreg  *regtab; /* number registers */
1.166     kristaps  119:        struct roffkv   *strtab; /* user-defined strings & macros */
1.167     kristaps  120:        struct roffkv   *xmbtab; /* multi-byte trans table (`tr') */
                    121:        struct roffstr  *xtab; /* single-byte trans table (`tr') */
1.106     kristaps  122:        const char      *current_string; /* value of last called user macro */
1.118     kristaps  123:        struct tbl_node *first_tbl; /* first table parsed */
                    124:        struct tbl_node *last_tbl; /* last table parsed */
                    125:        struct tbl_node *tbl; /* current table being parsed */
1.125     kristaps  126:        struct eqn_node *last_eqn; /* last equation parsed */
                    127:        struct eqn_node *first_eqn; /* first equation parsed */
                    128:        struct eqn_node *eqn; /* current equation being parsed */
1.79      kristaps  129: };
                    130:
1.67      kristaps  131: struct roffnode {
                    132:        enum rofft       tok; /* type of node */
                    133:        struct roffnode *parent; /* up one in stack */
                    134:        int              line; /* parse line */
                    135:        int              col; /* parse col */
1.106     kristaps  136:        char            *name; /* node name, e.g. macro name */
1.79      kristaps  137:        char            *end; /* end-rules: custom token */
                    138:        int              endspan; /* end-rules: next-line or infty */
1.82      kristaps  139:        enum roffrule    rule; /* current evaluation rule */
1.67      kristaps  140: };
                    141:
                    142: #define        ROFF_ARGS        struct roff *r, /* parse ctx */ \
1.72      kristaps  143:                         enum rofft tok, /* tok of macro */ \
1.67      kristaps  144:                         char **bufp, /* input buffer */ \
                    145:                         size_t *szp, /* size of input buffer */ \
                    146:                         int ln, /* parse line */ \
1.75      kristaps  147:                         int ppos, /* original pos in buffer */ \
                    148:                         int pos, /* current pos in buffer */ \
1.74      kristaps  149:                         int *offs /* reset offset of buffer data */
1.67      kristaps  150:
                    151: typedef        enum rofferr (*roffproc)(ROFF_ARGS);
                    152:
                    153: struct roffmac {
                    154:        const char      *name; /* macro name */
1.79      kristaps  155:        roffproc         proc; /* process new macro */
                    156:        roffproc         text; /* process as child text of macro */
                    157:        roffproc         sub; /* process as child of macro */
                    158:        int              flags;
                    159: #define        ROFFMAC_STRUCT  (1 << 0) /* always interpret */
1.85      kristaps  160:        struct roffmac  *next;
1.67      kristaps  161: };
                    162:
1.141     kristaps  163: struct predef {
                    164:        const char      *name; /* predefined input name */
                    165:        const char      *str; /* replacement symbol */
                    166: };
                    167:
                    168: #define        PREDEF(__name, __str) \
                    169:        { (__name), (__str) },
                    170:
1.155     kristaps  171: static enum rofft       roffhash_find(const char *, size_t);
                    172: static void             roffhash_init(void);
                    173: static void             roffnode_cleanscope(struct roff *);
                    174: static void             roffnode_pop(struct roff *);
                    175: static void             roffnode_push(struct roff *, enum rofft,
                    176:                                const char *, int, int);
1.80      kristaps  177: static enum rofferr     roff_block(ROFF_ARGS);
                    178: static enum rofferr     roff_block_text(ROFF_ARGS);
                    179: static enum rofferr     roff_block_sub(ROFF_ARGS);
                    180: static enum rofferr     roff_cblock(ROFF_ARGS);
1.174     kristaps  181: static enum rofferr     roff_cc(ROFF_ARGS);
1.195     schwarze  182: static void             roff_ccond(struct roff *, int, int);
1.82      kristaps  183: static enum rofferr     roff_cond(ROFF_ARGS);
                    184: static enum rofferr     roff_cond_text(ROFF_ARGS);
                    185: static enum rofferr     roff_cond_sub(ROFF_ARGS);
1.92      schwarze  186: static enum rofferr     roff_ds(ROFF_ARGS);
1.94      kristaps  187: static enum roffrule    roff_evalcond(const char *, int *);
1.155     kristaps  188: static void             roff_free1(struct roff *);
1.180     schwarze  189: static void             roff_freereg(struct roffreg *);
1.167     kristaps  190: static void             roff_freestr(struct roffkv *);
1.121     schwarze  191: static char            *roff_getname(struct roff *, char **, int, int);
1.184     schwarze  192: static int              roff_getnum(const char *, int *, int *);
                    193: static int              roff_getop(const char *, int *, char *);
1.181     schwarze  194: static int              roff_getregn(const struct roff *,
                    195:                                const char *, size_t);
1.192     schwarze  196: static int              roff_getregro(const char *name);
1.94      kristaps  197: static const char      *roff_getstrn(const struct roff *,
                    198:                                const char *, size_t);
1.178     schwarze  199: static enum rofferr     roff_it(ROFF_ARGS);
1.103     kristaps  200: static enum rofferr     roff_line_ignore(ROFF_ARGS);
1.89      kristaps  201: static enum rofferr     roff_nr(ROFF_ARGS);
1.169     schwarze  202: static void             roff_openeqn(struct roff *, const char *,
1.156     kristaps  203:                                int, int, const char *);
1.155     kristaps  204: static enum rofft       roff_parse(struct roff *, const char *, int *);
1.178     schwarze  205: static enum rofferr     roff_parsetext(char **, size_t *, int, int *);
1.172     schwarze  206: static enum rofferr     roff_res(struct roff *,
1.142     kristaps  207:                                char **, size_t *, int, int);
1.122     schwarze  208: static enum rofferr     roff_rm(ROFF_ARGS);
1.94      kristaps  209: static void             roff_setstr(struct roff *,
1.106     kristaps  210:                                const char *, const char *, int);
1.166     kristaps  211: static void             roff_setstrn(struct roffkv **, const char *,
1.164     kristaps  212:                                size_t, const char *, size_t, int);
1.105     kristaps  213: static enum rofferr     roff_so(ROFF_ARGS);
1.164     kristaps  214: static enum rofferr     roff_tr(ROFF_ARGS);
1.175     schwarze  215: static enum rofferr     roff_Dd(ROFF_ARGS);
                    216: static enum rofferr     roff_TH(ROFF_ARGS);
1.109     kristaps  217: static enum rofferr     roff_TE(ROFF_ARGS);
                    218: static enum rofferr     roff_TS(ROFF_ARGS);
1.125     kristaps  219: static enum rofferr     roff_EQ(ROFF_ARGS);
                    220: static enum rofferr     roff_EN(ROFF_ARGS);
1.112     kristaps  221: static enum rofferr     roff_T_(ROFF_ARGS);
1.106     kristaps  222: static enum rofferr     roff_userdef(ROFF_ARGS);
1.67      kristaps  223:
1.155     kristaps  224: /* See roffhash_find() */
1.85      kristaps  225:
                    226: #define        ASCII_HI         126
                    227: #define        ASCII_LO         33
                    228: #define        HASHWIDTH       (ASCII_HI - ASCII_LO + 1)
                    229:
                    230: static struct roffmac  *hash[HASHWIDTH];
                    231:
                    232: static struct roffmac   roffs[ROFF_MAX] = {
1.103     kristaps  233:        { "ad", roff_line_ignore, NULL, NULL, 0, NULL },
1.85      kristaps  234:        { "am", roff_block, roff_block_text, roff_block_sub, 0, NULL },
                    235:        { "ami", roff_block, roff_block_text, roff_block_sub, 0, NULL },
                    236:        { "am1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.193     schwarze  237:        { "as", roff_ds, NULL, NULL, 0, NULL },
1.174     kristaps  238:        { "cc", roff_cc, NULL, NULL, 0, NULL },
1.194     schwarze  239:        { "ce", roff_line_ignore, NULL, NULL, 0, NULL },
1.85      kristaps  240:        { "de", roff_block, roff_block_text, roff_block_sub, 0, NULL },
                    241:        { "dei", roff_block, roff_block_text, roff_block_sub, 0, NULL },
                    242:        { "de1", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.92      schwarze  243:        { "ds", roff_ds, NULL, NULL, 0, NULL },
1.85      kristaps  244:        { "el", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
1.185     schwarze  245:        { "fam", roff_line_ignore, NULL, NULL, 0, NULL },
1.186     schwarze  246:        { "hw", roff_line_ignore, NULL, NULL, 0, NULL },
1.103     kristaps  247:        { "hy", roff_line_ignore, NULL, NULL, 0, NULL },
1.85      kristaps  248:        { "ie", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
                    249:        { "if", roff_cond, roff_cond_text, roff_cond_sub, ROFFMAC_STRUCT, NULL },
                    250:        { "ig", roff_block, roff_block_text, roff_block_sub, 0, NULL },
1.178     schwarze  251:        { "it", roff_it, NULL, NULL, 0, NULL },
1.103     kristaps  252:        { "ne", roff_line_ignore, NULL, NULL, 0, NULL },
                    253:        { "nh", roff_line_ignore, NULL, NULL, 0, NULL },
1.104     kristaps  254:        { "nr", roff_nr, NULL, NULL, 0, NULL },
1.124     schwarze  255:        { "ns", roff_line_ignore, NULL, NULL, 0, NULL },
                    256:        { "ps", roff_line_ignore, NULL, NULL, 0, NULL },
1.122     schwarze  257:        { "rm", roff_rm, NULL, NULL, 0, NULL },
1.105     kristaps  258:        { "so", roff_so, NULL, NULL, 0, NULL },
1.124     schwarze  259:        { "ta", roff_line_ignore, NULL, NULL, 0, NULL },
1.164     kristaps  260:        { "tr", roff_tr, NULL, NULL, 0, NULL },
1.175     schwarze  261:        { "Dd", roff_Dd, NULL, NULL, 0, NULL },
                    262:        { "TH", roff_TH, NULL, NULL, 0, NULL },
1.109     kristaps  263:        { "TS", roff_TS, NULL, NULL, 0, NULL },
                    264:        { "TE", roff_TE, NULL, NULL, 0, NULL },
1.112     kristaps  265:        { "T&", roff_T_, NULL, NULL, 0, NULL },
1.125     kristaps  266:        { "EQ", roff_EQ, NULL, NULL, 0, NULL },
                    267:        { "EN", roff_EN, NULL, NULL, 0, NULL },
1.85      kristaps  268:        { ".", roff_cblock, NULL, NULL, 0, NULL },
1.106     kristaps  269:        { NULL, roff_userdef, NULL, NULL, 0, NULL },
1.67      kristaps  270: };
                    271:
1.175     schwarze  272: const  char *const __mdoc_reserved[] = {
                    273:        "Ac", "Ad", "An", "Ao", "Ap", "Aq", "Ar", "At",
                    274:        "Bc", "Bd", "Bf", "Bk", "Bl", "Bo", "Bq",
                    275:        "Brc", "Bro", "Brq", "Bsx", "Bt", "Bx",
                    276:        "Cd", "Cm", "Db", "Dc", "Dd", "Dl", "Do", "Dq",
                    277:        "Ds", "Dt", "Dv", "Dx", "D1",
                    278:        "Ec", "Ed", "Ef", "Ek", "El", "Em", "em",
                    279:        "En", "Eo", "Eq", "Er", "Es", "Ev", "Ex",
                    280:        "Fa", "Fc", "Fd", "Fl", "Fn", "Fo", "Fr", "Ft", "Fx",
                    281:        "Hf", "Ic", "In", "It", "Lb", "Li", "Lk", "Lp", "LP",
                    282:        "Me", "Ms", "Mt", "Nd", "Nm", "No", "Ns", "Nx",
                    283:        "Oc", "Oo", "Op", "Os", "Ot", "Ox",
                    284:        "Pa", "Pc", "Pf", "Po", "Pp", "PP", "pp", "Pq",
                    285:        "Qc", "Ql", "Qo", "Qq", "Or", "Rd", "Re", "Rs", "Rv",
                    286:        "Sc", "Sf", "Sh", "SH", "Sm", "So", "Sq",
                    287:        "Ss", "St", "Sx", "Sy",
                    288:        "Ta", "Tn", "Ud", "Ux", "Va", "Vt", "Xc", "Xo", "Xr",
                    289:        "%A", "%B", "%D", "%I", "%J", "%N", "%O",
                    290:        "%P", "%Q", "%R", "%T", "%U", "%V",
                    291:        NULL
                    292: };
                    293:
                    294: const  char *const __man_reserved[] = {
                    295:        "AT", "B", "BI", "BR", "BT", "DE", "DS", "DT",
                    296:        "EE", "EN", "EQ", "EX", "HF", "HP", "I", "IB", "IP", "IR",
                    297:        "LP", "ME", "MT", "OP", "P", "PD", "PP", "PT",
                    298:        "R", "RB", "RE", "RI", "RS", "SB", "SH", "SM", "SS", "SY",
                    299:        "TE", "TH", "TP", "TQ", "TS", "T&", "UC", "UE", "UR", "YS",
                    300:        NULL
                    301: };
                    302:
1.141     kristaps  303: /* Array of injected predefined strings. */
                    304: #define        PREDEFS_MAX      38
                    305: static const struct predef predefs[PREDEFS_MAX] = {
                    306: #include "predefs.in"
                    307: };
                    308:
1.155     kristaps  309: /* See roffhash_find() */
1.85      kristaps  310: #define        ROFF_HASH(p)    (p[0] - ASCII_LO)
                    311:
1.178     schwarze  312: static int      roffit_lines;  /* number of lines to delay */
                    313: static char    *roffit_macro;  /* nil-terminated macro line */
                    314:
1.85      kristaps  315: static void
1.155     kristaps  316: roffhash_init(void)
1.85      kristaps  317: {
                    318:        struct roffmac   *n;
                    319:        int               buc, i;
                    320:
1.106     kristaps  321:        for (i = 0; i < (int)ROFF_USERDEF; i++) {
1.85      kristaps  322:                assert(roffs[i].name[0] >= ASCII_LO);
                    323:                assert(roffs[i].name[0] <= ASCII_HI);
                    324:
                    325:                buc = ROFF_HASH(roffs[i].name);
                    326:
                    327:                if (NULL != (n = hash[buc])) {
                    328:                        for ( ; n->next; n = n->next)
                    329:                                /* Do nothing. */ ;
                    330:                        n->next = &roffs[i];
                    331:                } else
                    332:                        hash[buc] = &roffs[i];
                    333:        }
                    334: }
                    335:
1.67      kristaps  336: /*
                    337:  * Look up a roff token by its name.  Returns ROFF_MAX if no macro by
                    338:  * the nil-terminated string name could be found.
                    339:  */
                    340: static enum rofft
1.155     kristaps  341: roffhash_find(const char *p, size_t s)
1.67      kristaps  342: {
1.85      kristaps  343:        int              buc;
                    344:        struct roffmac  *n;
1.67      kristaps  345:
1.85      kristaps  346:        /*
                    347:         * libroff has an extremely simple hashtable, for the time
                    348:         * being, which simply keys on the first character, which must
                    349:         * be printable, then walks a chain.  It works well enough until
                    350:         * optimised.
                    351:         */
                    352:
                    353:        if (p[0] < ASCII_LO || p[0] > ASCII_HI)
                    354:                return(ROFF_MAX);
                    355:
                    356:        buc = ROFF_HASH(p);
                    357:
                    358:        if (NULL == (n = hash[buc]))
                    359:                return(ROFF_MAX);
                    360:        for ( ; n; n = n->next)
1.106     kristaps  361:                if (0 == strncmp(n->name, p, s) && '\0' == n->name[(int)s])
1.85      kristaps  362:                        return((enum rofft)(n - roffs));
1.67      kristaps  363:
                    364:        return(ROFF_MAX);
                    365: }
                    366:
                    367:
                    368: /*
                    369:  * Pop the current node off of the stack of roff instructions currently
                    370:  * pending.
                    371:  */
                    372: static void
                    373: roffnode_pop(struct roff *r)
                    374: {
                    375:        struct roffnode *p;
                    376:
1.75      kristaps  377:        assert(r->last);
                    378:        p = r->last;
1.82      kristaps  379:
1.75      kristaps  380:        r->last = r->last->parent;
1.106     kristaps  381:        free(p->name);
                    382:        free(p->end);
1.67      kristaps  383:        free(p);
                    384: }
                    385:
                    386:
                    387: /*
                    388:  * Push a roff node onto the instruction stack.  This must later be
                    389:  * removed with roffnode_pop().
                    390:  */
1.98      schwarze  391: static void
1.106     kristaps  392: roffnode_push(struct roff *r, enum rofft tok, const char *name,
                    393:                int line, int col)
1.67      kristaps  394: {
                    395:        struct roffnode *p;
                    396:
1.98      schwarze  397:        p = mandoc_calloc(1, sizeof(struct roffnode));
1.67      kristaps  398:        p->tok = tok;
1.106     kristaps  399:        if (name)
                    400:                p->name = mandoc_strdup(name);
1.67      kristaps  401:        p->parent = r->last;
                    402:        p->line = line;
                    403:        p->col = col;
1.79      kristaps  404:        p->rule = p->parent ? p->parent->rule : ROFFRULE_DENY;
1.67      kristaps  405:
                    406:        r->last = p;
                    407: }
                    408:
                    409:
                    410: static void
                    411: roff_free1(struct roff *r)
                    412: {
1.176     schwarze  413:        struct tbl_node *tbl;
1.125     kristaps  414:        struct eqn_node *e;
1.167     kristaps  415:        int              i;
1.67      kristaps  416:
1.176     schwarze  417:        while (NULL != (tbl = r->first_tbl)) {
                    418:                r->first_tbl = tbl->next;
                    419:                tbl_free(tbl);
1.109     kristaps  420:        }
                    421:
1.113     kristaps  422:        r->first_tbl = r->last_tbl = r->tbl = NULL;
                    423:
1.125     kristaps  424:        while (NULL != (e = r->first_eqn)) {
                    425:                r->first_eqn = e->next;
                    426:                eqn_free(e);
                    427:        }
                    428:
                    429:        r->first_eqn = r->last_eqn = r->eqn = NULL;
                    430:
1.67      kristaps  431:        while (r->last)
                    432:                roffnode_pop(r);
1.109     kristaps  433:
1.167     kristaps  434:        roff_freestr(r->strtab);
                    435:        roff_freestr(r->xmbtab);
                    436:
                    437:        r->strtab = r->xmbtab = NULL;
                    438:
1.180     schwarze  439:        roff_freereg(r->regtab);
                    440:
                    441:        r->regtab = NULL;
                    442:
1.167     kristaps  443:        if (r->xtab)
                    444:                for (i = 0; i < 128; i++)
                    445:                        free(r->xtab[i].p);
                    446:
                    447:        free(r->xtab);
                    448:        r->xtab = NULL;
1.67      kristaps  449: }
                    450:
                    451: void
                    452: roff_reset(struct roff *r)
                    453: {
                    454:
                    455:        roff_free1(r);
1.174     kristaps  456:        r->control = 0;
1.67      kristaps  457: }
                    458:
                    459:
                    460: void
                    461: roff_free(struct roff *r)
                    462: {
                    463:
                    464:        roff_free1(r);
                    465:        free(r);
                    466: }
                    467:
                    468:
                    469: struct roff *
1.190     schwarze  470: roff_alloc(enum mparset type, struct mparse *parse, int quick)
1.67      kristaps  471: {
                    472:        struct roff     *r;
                    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;
1.190     schwarze  477:        r->quick = quick;
1.82      kristaps  478:        r->rstackpos = -1;
1.85      kristaps  479:
1.155     kristaps  480:        roffhash_init();
1.141     kristaps  481:
1.67      kristaps  482:        return(r);
                    483: }
                    484:
1.94      kristaps  485: /*
1.181     schwarze  486:  * In the current line, expand user-defined strings ("\*")
                    487:  * and references to number registers ("\n").
                    488:  * Also check the syntax of other escape sequences.
1.154     kristaps  489:  */
1.172     schwarze  490: static enum rofferr
1.142     kristaps  491: roff_res(struct roff *r, char **bufp, size_t *szp, int ln, int pos)
1.94      kristaps  492: {
1.181     schwarze  493:        char             ubuf[12]; /* buffer to print the number */
1.108     schwarze  494:        const char      *stesc; /* start of an escape sequence ('\\') */
                    495:        const char      *stnam; /* start of the name, after "[(*" */
                    496:        const char      *cp;    /* end of the name, e.g. before ']' */
                    497:        const char      *res;   /* the string to be substituted */
1.181     schwarze  498:        char            *nbuf;  /* new buffer to copy bufp to */
                    499:        size_t           nsz;   /* size of the new buffer */
                    500:        size_t           maxl;  /* expected length of the escape name */
                    501:        size_t           naml;  /* actual length of the escape name */
                    502:        int              expand_count;  /* to avoid infinite loops */
1.94      kristaps  503:
1.170     schwarze  504:        expand_count = 0;
                    505:
1.154     kristaps  506: again:
1.108     schwarze  507:        cp = *bufp + pos;
                    508:        while (NULL != (cp = strchr(cp, '\\'))) {
                    509:                stesc = cp++;
                    510:
                    511:                /*
1.181     schwarze  512:                 * The second character must be an asterisk or an n.
1.108     schwarze  513:                 * If it isn't, skip it anyway:  It is escaped,
                    514:                 * so it can't start another escape sequence.
                    515:                 */
                    516:
                    517:                if ('\0' == *cp)
1.172     schwarze  518:                        return(ROFF_CONT);
1.152     kristaps  519:
1.181     schwarze  520:                switch (*cp) {
                    521:                case ('*'):
                    522:                        res = NULL;
                    523:                        break;
                    524:                case ('n'):
                    525:                        res = ubuf;
                    526:                        break;
                    527:                default:
                    528:                        if (ESCAPE_ERROR != mandoc_escape(&cp, NULL, NULL))
1.152     kristaps  529:                                continue;
1.153     kristaps  530:                        mandoc_msg
                    531:                                (MANDOCERR_BADESCAPE, r->parse,
                    532:                                 ln, (int)(stesc - *bufp), NULL);
1.172     schwarze  533:                        return(ROFF_CONT);
1.152     kristaps  534:                }
                    535:
                    536:                cp++;
1.108     schwarze  537:
                    538:                /*
                    539:                 * The third character decides the length
1.181     schwarze  540:                 * of the name of the string or register.
1.108     schwarze  541:                 * Save a pointer to the name.
                    542:                 */
                    543:
1.94      kristaps  544:                switch (*cp) {
1.108     schwarze  545:                case ('\0'):
1.172     schwarze  546:                        return(ROFF_CONT);
1.94      kristaps  547:                case ('('):
                    548:                        cp++;
                    549:                        maxl = 2;
                    550:                        break;
                    551:                case ('['):
                    552:                        cp++;
                    553:                        maxl = 0;
                    554:                        break;
                    555:                default:
                    556:                        maxl = 1;
                    557:                        break;
                    558:                }
1.108     schwarze  559:                stnam = cp;
1.94      kristaps  560:
1.108     schwarze  561:                /* Advance to the end of the name. */
1.94      kristaps  562:
1.181     schwarze  563:                for (naml = 0; 0 == maxl || naml < maxl; naml++, cp++) {
1.153     kristaps  564:                        if ('\0' == *cp) {
                    565:                                mandoc_msg
                    566:                                        (MANDOCERR_BADESCAPE,
                    567:                                         r->parse, ln,
                    568:                                         (int)(stesc - *bufp), NULL);
1.172     schwarze  569:                                return(ROFF_CONT);
1.153     kristaps  570:                        }
1.94      kristaps  571:                        if (0 == maxl && ']' == *cp)
                    572:                                break;
                    573:                }
                    574:
1.108     schwarze  575:                /*
                    576:                 * Retrieve the replacement string; if it is
                    577:                 * undefined, resume searching for escapes.
                    578:                 */
                    579:
1.181     schwarze  580:                if (NULL == res)
                    581:                        res = roff_getstrn(r, stnam, naml);
                    582:                else
                    583:                        snprintf(ubuf, sizeof(ubuf), "%d",
                    584:                            roff_getregn(r, stnam, naml));
1.94      kristaps  585:
                    586:                if (NULL == res) {
1.153     kristaps  587:                        mandoc_msg
                    588:                                (MANDOCERR_BADESCAPE, r->parse,
                    589:                                 ln, (int)(stesc - *bufp), NULL);
1.142     kristaps  590:                        res = "";
1.94      kristaps  591:                }
                    592:
1.108     schwarze  593:                /* Replace the escape sequence by the string. */
                    594:
1.161     kristaps  595:                pos = stesc - *bufp;
1.154     kristaps  596:
1.94      kristaps  597:                nsz = *szp + strlen(res) + 1;
1.181     schwarze  598:                nbuf = mandoc_malloc(nsz);
1.94      kristaps  599:
1.181     schwarze  600:                strlcpy(nbuf, *bufp, (size_t)(stesc - *bufp + 1));
                    601:                strlcat(nbuf, res, nsz);
                    602:                strlcat(nbuf, cp + (maxl ? 0 : 1), nsz);
1.94      kristaps  603:
                    604:                free(*bufp);
                    605:
1.181     schwarze  606:                *bufp = nbuf;
1.94      kristaps  607:                *szp = nsz;
1.170     schwarze  608:
                    609:                if (EXPAND_LIMIT >= ++expand_count)
                    610:                        goto again;
                    611:
                    612:                /* Just leave the string unexpanded. */
                    613:                mandoc_msg(MANDOCERR_ROFFLOOP, r->parse, ln, pos, NULL);
1.172     schwarze  614:                return(ROFF_IGN);
1.154     kristaps  615:        }
1.172     schwarze  616:        return(ROFF_CONT);
1.154     kristaps  617: }
                    618:
                    619: /*
1.178     schwarze  620:  * Process text streams:
                    621:  * Convert all breakable hyphens into ASCII_HYPH.
                    622:  * Decrement and spring input line trap.
1.154     kristaps  623:  */
                    624: static enum rofferr
1.178     schwarze  625: roff_parsetext(char **bufp, size_t *szp, int pos, int *offs)
1.154     kristaps  626: {
                    627:        size_t           sz;
                    628:        const char      *start;
1.178     schwarze  629:        char            *p;
                    630:        int              isz;
1.154     kristaps  631:        enum mandoc_esc  esc;
                    632:
1.178     schwarze  633:        start = p = *bufp + pos;
1.154     kristaps  634:
                    635:        while ('\0' != *p) {
                    636:                sz = strcspn(p, "-\\");
                    637:                p += sz;
                    638:
1.159     kristaps  639:                if ('\0' == *p)
                    640:                        break;
                    641:
1.154     kristaps  642:                if ('\\' == *p) {
                    643:                        /* Skip over escapes. */
                    644:                        p++;
1.189     schwarze  645:                        esc = mandoc_escape((const char **)&p, NULL, NULL);
1.154     kristaps  646:                        if (ESCAPE_ERROR == esc)
                    647:                                break;
1.155     kristaps  648:                        continue;
1.159     kristaps  649:                } else if (p == start) {
1.158     kristaps  650:                        p++;
1.155     kristaps  651:                        continue;
1.158     kristaps  652:                }
1.155     kristaps  653:
1.171     schwarze  654:                if (isalpha((unsigned char)p[-1]) &&
                    655:                    isalpha((unsigned char)p[1]))
1.155     kristaps  656:                        *p = ASCII_HYPH;
                    657:                p++;
1.94      kristaps  658:        }
                    659:
1.178     schwarze  660:        /* Spring the input line trap. */
                    661:        if (1 == roffit_lines) {
                    662:                isz = asprintf(&p, "%s\n.%s", *bufp, roffit_macro);
                    663:                if (-1 == isz) {
                    664:                        perror(NULL);
                    665:                        exit((int)MANDOCLEVEL_SYSERR);
                    666:                }
                    667:                free(*bufp);
                    668:                *bufp = p;
                    669:                *szp = isz + 1;
                    670:                *offs = 0;
                    671:                free(roffit_macro);
                    672:                roffit_lines = 0;
                    673:                return(ROFF_REPARSE);
                    674:        } else if (1 < roffit_lines)
                    675:                --roffit_lines;
1.154     kristaps  676:        return(ROFF_CONT);
1.94      kristaps  677: }
                    678:
1.67      kristaps  679: enum rofferr
1.90      kristaps  680: roff_parseln(struct roff *r, int ln, char **bufp,
                    681:                size_t *szp, int pos, int *offs)
1.67      kristaps  682: {
                    683:        enum rofft       t;
1.109     kristaps  684:        enum rofferr     e;
1.130     kristaps  685:        int              ppos, ctl;
1.79      kristaps  686:
                    687:        /*
1.94      kristaps  688:         * Run the reserved-word filter only if we have some reserved
                    689:         * words to fill in.
                    690:         */
                    691:
1.172     schwarze  692:        e = roff_res(r, bufp, szp, ln, pos);
                    693:        if (ROFF_IGN == e)
                    694:                return(e);
                    695:        assert(ROFF_CONT == e);
1.94      kristaps  696:
1.130     kristaps  697:        ppos = pos;
1.174     kristaps  698:        ctl = roff_getcontrol(r, *bufp, &pos);
1.130     kristaps  699:
1.94      kristaps  700:        /*
1.79      kristaps  701:         * First, if a scope is open and we're not a macro, pass the
                    702:         * text through the macro's filter.  If a scope isn't open and
                    703:         * we're not a macro, just let it through.
1.125     kristaps  704:         * Finally, if there's an equation scope open, divert it into it
                    705:         * no matter our state.
1.79      kristaps  706:         */
1.74      kristaps  707:
1.130     kristaps  708:        if (r->last && ! ctl) {
1.78      kristaps  709:                t = r->last->tok;
                    710:                assert(roffs[t].text);
1.109     kristaps  711:                e = (*roffs[t].text)
                    712:                        (r, t, bufp, szp, ln, pos, pos, offs);
                    713:                assert(ROFF_IGN == e || ROFF_CONT == e);
1.125     kristaps  714:                if (ROFF_CONT != e)
                    715:                        return(e);
1.182     schwarze  716:        }
                    717:        if (r->eqn)
                    718:                return(eqn_read(&r->eqn, ln, *bufp, ppos, offs));
                    719:        if ( ! ctl) {
1.125     kristaps  720:                if (r->tbl)
1.130     kristaps  721:                        return(tbl_read(r->tbl, ln, *bufp, pos));
1.178     schwarze  722:                return(roff_parsetext(bufp, szp, pos, offs));
1.182     schwarze  723:        }
1.67      kristaps  724:
1.79      kristaps  725:        /*
                    726:         * If a scope is open, go to the child handler for that macro,
                    727:         * as it may want to preprocess before doing anything with it.
1.125     kristaps  728:         * Don't do so if an equation is open.
1.79      kristaps  729:         */
1.78      kristaps  730:
1.79      kristaps  731:        if (r->last) {
                    732:                t = r->last->tok;
                    733:                assert(roffs[t].sub);
                    734:                return((*roffs[t].sub)
1.90      kristaps  735:                                (r, t, bufp, szp,
1.130     kristaps  736:                                 ln, ppos, pos, offs));
1.79      kristaps  737:        }
1.78      kristaps  738:
1.79      kristaps  739:        /*
                    740:         * Lastly, as we've no scope open, try to look up and execute
                    741:         * the new macro.  If no macro is found, simply return and let
                    742:         * the compilers handle it.
                    743:         */
1.67      kristaps  744:
1.106     kristaps  745:        if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos)))
1.79      kristaps  746:                return(ROFF_CONT);
1.67      kristaps  747:
1.75      kristaps  748:        assert(roffs[t].proc);
1.78      kristaps  749:        return((*roffs[t].proc)
1.90      kristaps  750:                        (r, t, bufp, szp,
                    751:                         ln, ppos, pos, offs));
1.74      kristaps  752: }
                    753:
                    754:
1.117     kristaps  755: void
1.74      kristaps  756: roff_endparse(struct roff *r)
                    757: {
                    758:
1.110     kristaps  759:        if (r->last)
1.128     kristaps  760:                mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.109     kristaps  761:                                r->last->line, r->last->col, NULL);
1.117     kristaps  762:
1.125     kristaps  763:        if (r->eqn) {
1.128     kristaps  764:                mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.148     kristaps  765:                                r->eqn->eqn.ln, r->eqn->eqn.pos, NULL);
1.151     kristaps  766:                eqn_end(&r->eqn);
1.125     kristaps  767:        }
                    768:
1.117     kristaps  769:        if (r->tbl) {
1.128     kristaps  770:                mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse,
1.117     kristaps  771:                                r->tbl->line, r->tbl->pos, NULL);
1.151     kristaps  772:                tbl_end(&r->tbl);
1.117     kristaps  773:        }
1.67      kristaps  774: }
                    775:
                    776: /*
                    777:  * Parse a roff node's type from the input buffer.  This must be in the
                    778:  * form of ".foo xxx" in the usual way.
                    779:  */
                    780: static enum rofft
1.106     kristaps  781: roff_parse(struct roff *r, const char *buf, int *pos)
1.67      kristaps  782: {
1.106     kristaps  783:        const char      *mac;
                    784:        size_t           maclen;
1.67      kristaps  785:        enum rofft       t;
                    786:
1.144     kristaps  787:        if ('\0' == buf[*pos] || '"' == buf[*pos] ||
                    788:                        '\t' == buf[*pos] || ' ' == buf[*pos])
1.67      kristaps  789:                return(ROFF_MAX);
                    790:
1.195     schwarze  791:        /* We stop the macro parse at an escape, tab, space, or nil. */
1.144     kristaps  792:
1.106     kristaps  793:        mac = buf + *pos;
1.195     schwarze  794:        maclen = strcspn(mac, " \\\t\0");
1.67      kristaps  795:
1.106     kristaps  796:        t = (r->current_string = roff_getstrn(r, mac, maclen))
1.155     kristaps  797:            ? ROFF_USERDEF : roffhash_find(mac, maclen);
1.67      kristaps  798:
1.127     kristaps  799:        *pos += (int)maclen;
1.130     kristaps  800:
1.67      kristaps  801:        while (buf[*pos] && ' ' == buf[*pos])
                    802:                (*pos)++;
                    803:
                    804:        return(t);
                    805: }
                    806:
                    807: /* ARGSUSED */
                    808: static enum rofferr
1.76      kristaps  809: roff_cblock(ROFF_ARGS)
1.67      kristaps  810: {
                    811:
1.79      kristaps  812:        /*
                    813:         * A block-close `..' should only be invoked as a child of an
                    814:         * ignore macro, otherwise raise a warning and just ignore it.
                    815:         */
                    816:
1.76      kristaps  817:        if (NULL == r->last) {
1.128     kristaps  818:                mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.76      kristaps  819:                return(ROFF_IGN);
                    820:        }
1.67      kristaps  821:
1.81      kristaps  822:        switch (r->last->tok) {
                    823:        case (ROFF_am):
                    824:                /* FALLTHROUGH */
                    825:        case (ROFF_ami):
                    826:                /* FALLTHROUGH */
                    827:        case (ROFF_am1):
                    828:                /* FALLTHROUGH */
                    829:        case (ROFF_de):
1.108     schwarze  830:                /* ROFF_de1 is remapped to ROFF_de in roff_block(). */
1.81      kristaps  831:                /* FALLTHROUGH */
                    832:        case (ROFF_dei):
                    833:                /* FALLTHROUGH */
                    834:        case (ROFF_ig):
                    835:                break;
                    836:        default:
1.128     kristaps  837:                mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.67      kristaps  838:                return(ROFF_IGN);
1.76      kristaps  839:        }
1.67      kristaps  840:
1.76      kristaps  841:        if ((*bufp)[pos])
1.128     kristaps  842:                mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.71      kristaps  843:
                    844:        roffnode_pop(r);
1.76      kristaps  845:        roffnode_cleanscope(r);
                    846:        return(ROFF_IGN);
1.71      kristaps  847:
1.67      kristaps  848: }
                    849:
                    850:
1.76      kristaps  851: static void
                    852: roffnode_cleanscope(struct roff *r)
1.67      kristaps  853: {
                    854:
1.76      kristaps  855:        while (r->last) {
1.173     schwarze  856:                if (--r->last->endspan != 0)
1.76      kristaps  857:                        break;
                    858:                roffnode_pop(r);
                    859:        }
1.67      kristaps  860: }
                    861:
                    862:
1.195     schwarze  863: static void
                    864: roff_ccond(struct roff *r, int ln, int ppos)
1.74      kristaps  865: {
                    866:
1.76      kristaps  867:        if (NULL == r->last) {
1.128     kristaps  868:                mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.195     schwarze  869:                return;
1.76      kristaps  870:        }
                    871:
1.82      kristaps  872:        switch (r->last->tok) {
                    873:        case (ROFF_el):
                    874:                /* FALLTHROUGH */
                    875:        case (ROFF_ie):
                    876:                /* FALLTHROUGH */
                    877:        case (ROFF_if):
                    878:                break;
                    879:        default:
1.128     kristaps  880:                mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.195     schwarze  881:                return;
1.75      kristaps  882:        }
                    883:
1.76      kristaps  884:        if (r->last->endspan > -1) {
1.128     kristaps  885:                mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.195     schwarze  886:                return;
1.76      kristaps  887:        }
                    888:
1.75      kristaps  889:        roffnode_pop(r);
1.76      kristaps  890:        roffnode_cleanscope(r);
1.195     schwarze  891:        return;
1.76      kristaps  892: }
                    893:
1.75      kristaps  894:
1.76      kristaps  895: /* ARGSUSED */
                    896: static enum rofferr
1.80      kristaps  897: roff_block(ROFF_ARGS)
1.76      kristaps  898: {
1.78      kristaps  899:        int             sv;
                    900:        size_t          sz;
1.106     kristaps  901:        char            *name;
                    902:
                    903:        name = NULL;
1.76      kristaps  904:
1.106     kristaps  905:        if (ROFF_ig != tok) {
                    906:                if ('\0' == (*bufp)[pos]) {
1.128     kristaps  907:                        mandoc_msg(MANDOCERR_NOARGS, r->parse, ln, ppos, NULL);
1.106     kristaps  908:                        return(ROFF_IGN);
                    909:                }
1.107     kristaps  910:
                    911:                /*
                    912:                 * Re-write `de1', since we don't really care about
                    913:                 * groff's strange compatibility mode, into `de'.
                    914:                 */
                    915:
1.106     kristaps  916:                if (ROFF_de1 == tok)
                    917:                        tok = ROFF_de;
                    918:                if (ROFF_de == tok)
                    919:                        name = *bufp + pos;
                    920:                else
1.128     kristaps  921:                        mandoc_msg(MANDOCERR_REQUEST, r->parse, ln, ppos,
1.106     kristaps  922:                            roffs[tok].name);
1.107     kristaps  923:
1.131     schwarze  924:                while ((*bufp)[pos] && ! isspace((unsigned char)(*bufp)[pos]))
1.80      kristaps  925:                        pos++;
1.107     kristaps  926:
1.131     schwarze  927:                while (isspace((unsigned char)(*bufp)[pos]))
1.106     kristaps  928:                        (*bufp)[pos++] = '\0';
1.80      kristaps  929:        }
                    930:
1.106     kristaps  931:        roffnode_push(r, tok, name, ln, ppos);
                    932:
                    933:        /*
                    934:         * At the beginning of a `de' macro, clear the existing string
                    935:         * with the same name, if there is one.  New content will be
1.193     schwarze  936:         * appended from roff_block_text() in multiline mode.
1.106     kristaps  937:         */
1.107     kristaps  938:
1.106     kristaps  939:        if (ROFF_de == tok)
1.108     schwarze  940:                roff_setstr(r, name, "", 0);
1.76      kristaps  941:
1.79      kristaps  942:        if ('\0' == (*bufp)[pos])
1.78      kristaps  943:                return(ROFF_IGN);
                    944:
1.107     kristaps  945:        /* If present, process the custom end-of-line marker. */
                    946:
1.78      kristaps  947:        sv = pos;
1.131     schwarze  948:        while ((*bufp)[pos] && ! isspace((unsigned char)(*bufp)[pos]))
1.78      kristaps  949:                pos++;
                    950:
                    951:        /*
                    952:         * Note: groff does NOT like escape characters in the input.
                    953:         * Instead of detecting this, we're just going to let it fly and
                    954:         * to hell with it.
                    955:         */
                    956:
                    957:        assert(pos > sv);
                    958:        sz = (size_t)(pos - sv);
                    959:
1.79      kristaps  960:        if (1 == sz && '.' == (*bufp)[sv])
                    961:                return(ROFF_IGN);
                    962:
1.98      schwarze  963:        r->last->end = mandoc_malloc(sz + 1);
1.78      kristaps  964:
                    965:        memcpy(r->last->end, *bufp + sv, sz);
                    966:        r->last->end[(int)sz] = '\0';
                    967:
1.77      kristaps  968:        if ((*bufp)[pos])
1.128     kristaps  969:                mandoc_msg(MANDOCERR_ARGSLOST, r->parse, ln, pos, NULL);
1.74      kristaps  970:
1.78      kristaps  971:        return(ROFF_IGN);
                    972: }
                    973:
                    974:
                    975: /* ARGSUSED */
                    976: static enum rofferr
1.80      kristaps  977: roff_block_sub(ROFF_ARGS)
1.79      kristaps  978: {
                    979:        enum rofft      t;
                    980:        int             i, j;
                    981:
                    982:        /*
                    983:         * First check whether a custom macro exists at this level.  If
                    984:         * it does, then check against it.  This is some of groff's
                    985:         * stranger behaviours.  If we encountered a custom end-scope
                    986:         * tag and that tag also happens to be a "real" macro, then we
                    987:         * need to try interpreting it again as a real macro.  If it's
                    988:         * not, then return ignore.  Else continue.
                    989:         */
                    990:
                    991:        if (r->last->end) {
1.130     kristaps  992:                for (i = pos, j = 0; r->last->end[j]; j++, i++)
1.79      kristaps  993:                        if ((*bufp)[i] != r->last->end[j])
                    994:                                break;
                    995:
                    996:                if ('\0' == r->last->end[j] &&
                    997:                                ('\0' == (*bufp)[i] ||
                    998:                                 ' ' == (*bufp)[i] ||
                    999:                                 '\t' == (*bufp)[i])) {
                   1000:                        roffnode_pop(r);
                   1001:                        roffnode_cleanscope(r);
                   1002:
1.130     kristaps 1003:                        while (' ' == (*bufp)[i] || '\t' == (*bufp)[i])
                   1004:                                i++;
                   1005:
                   1006:                        pos = i;
1.106     kristaps 1007:                        if (ROFF_MAX != roff_parse(r, *bufp, &pos))
1.79      kristaps 1008:                                return(ROFF_RERUN);
                   1009:                        return(ROFF_IGN);
                   1010:                }
                   1011:        }
                   1012:
                   1013:        /*
                   1014:         * If we have no custom end-query or lookup failed, then try
                   1015:         * pulling it out of the hashtable.
                   1016:         */
                   1017:
1.137     schwarze 1018:        t = roff_parse(r, *bufp, &pos);
1.79      kristaps 1019:
1.106     kristaps 1020:        /*
                   1021:         * Macros other than block-end are only significant
                   1022:         * in `de' blocks; elsewhere, simply throw them away.
                   1023:         */
                   1024:        if (ROFF_cblock != t) {
                   1025:                if (ROFF_de == tok)
1.193     schwarze 1026:                        roff_setstr(r, r->last->name, *bufp + ppos, 2);
1.79      kristaps 1027:                return(ROFF_IGN);
1.106     kristaps 1028:        }
1.79      kristaps 1029:
                   1030:        assert(roffs[t].proc);
1.90      kristaps 1031:        return((*roffs[t].proc)(r, t, bufp, szp,
                   1032:                                ln, ppos, pos, offs));
1.79      kristaps 1033: }
                   1034:
                   1035:
                   1036: /* ARGSUSED */
                   1037: static enum rofferr
1.80      kristaps 1038: roff_block_text(ROFF_ARGS)
1.78      kristaps 1039: {
                   1040:
1.106     kristaps 1041:        if (ROFF_de == tok)
1.193     schwarze 1042:                roff_setstr(r, r->last->name, *bufp + pos, 2);
1.106     kristaps 1043:
1.78      kristaps 1044:        return(ROFF_IGN);
                   1045: }
                   1046:
                   1047:
                   1048: /* ARGSUSED */
                   1049: static enum rofferr
1.82      kristaps 1050: roff_cond_sub(ROFF_ARGS)
                   1051: {
                   1052:        enum rofft       t;
                   1053:        enum roffrule    rr;
1.139     kristaps 1054:        char            *ep;
1.82      kristaps 1055:
                   1056:        rr = r->last->rule;
1.139     kristaps 1057:        roffnode_cleanscope(r);
1.177     schwarze 1058:        t = roff_parse(r, *bufp, &pos);
1.82      kristaps 1059:
1.139     kristaps 1060:        /*
1.177     schwarze 1061:         * Fully handle known macros when they are structurally
                   1062:         * required or when the conditional evaluated to true.
1.87      kristaps 1063:         */
                   1064:
1.177     schwarze 1065:        if ((ROFF_MAX != t) &&
1.195     schwarze 1066:            (ROFFRULE_ALLOW == rr ||
1.177     schwarze 1067:             ROFFMAC_STRUCT & roffs[t].flags)) {
                   1068:                assert(roffs[t].proc);
                   1069:                return((*roffs[t].proc)(r, t, bufp, szp,
                   1070:                                        ln, ppos, pos, offs));
                   1071:        }
1.144     kristaps 1072:
1.196     schwarze 1073:        /*
                   1074:         * If `\}' occurs on a macro line without a preceding macro,
                   1075:         * drop the line completely.
                   1076:         */
                   1077:
                   1078:        ep = *bufp + pos;
                   1079:        if ('\\' == ep[0] && '}' == ep[1])
                   1080:                rr = ROFFRULE_DENY;
                   1081:
1.177     schwarze 1082:        /* Always check for the closing delimiter `\}'. */
1.144     kristaps 1083:
1.177     schwarze 1084:        while (NULL != (ep = strchr(ep, '\\'))) {
1.197   ! schwarze 1085:                if ('}' == *(++ep)) {
        !          1086:                        *ep = '&';
        !          1087:                        roff_ccond(r, ln, ep - *bufp - 1);
        !          1088:                }
        !          1089:                ++ep;
1.177     schwarze 1090:        }
                   1091:        return(ROFFRULE_DENY == rr ? ROFF_IGN : ROFF_CONT);
1.82      kristaps 1092: }
                   1093:
                   1094: /* ARGSUSED */
                   1095: static enum rofferr
                   1096: roff_cond_text(ROFF_ARGS)
1.78      kristaps 1097: {
1.140     kristaps 1098:        char            *ep;
1.82      kristaps 1099:        enum roffrule    rr;
                   1100:
                   1101:        rr = r->last->rule;
1.140     kristaps 1102:        roffnode_cleanscope(r);
1.82      kristaps 1103:
1.197   ! schwarze 1104:        ep = *bufp + pos;
        !          1105:        while (NULL != (ep = strchr(ep, '\\'))) {
        !          1106:                if ('}' == *(++ep)) {
        !          1107:                        *ep = '&';
        !          1108:                        roff_ccond(r, ln, ep - *bufp - 1);
        !          1109:                }
        !          1110:                ++ep;
1.78      kristaps 1111:        }
1.82      kristaps 1112:        return(ROFFRULE_DENY == rr ? ROFF_IGN : ROFF_CONT);
1.74      kristaps 1113: }
                   1114:
1.184     schwarze 1115: static int
                   1116: roff_getnum(const char *v, int *pos, int *res)
                   1117: {
                   1118:        int p, n;
                   1119:
                   1120:        p = *pos;
                   1121:        n = v[p] == '-';
                   1122:        if (n)
                   1123:                p++;
                   1124:
                   1125:        for (*res = 0; isdigit((unsigned char)v[p]); p++)
                   1126:                *res += 10 * *res + v[p] - '0';
                   1127:        if (p == *pos + n)
                   1128:                return 0;
                   1129:
                   1130:        if (n)
                   1131:                *res = -*res;
                   1132:
                   1133:        *pos = p;
                   1134:        return 1;
                   1135: }
                   1136:
                   1137: static int
                   1138: roff_getop(const char *v, int *pos, char *res)
                   1139: {
                   1140:        int e;
                   1141:
                   1142:        *res = v[*pos];
                   1143:        e = v[*pos + 1] == '=';
                   1144:
                   1145:        switch (*res) {
                   1146:        case '=':
                   1147:                break;
                   1148:        case '>':
                   1149:                if (e)
                   1150:                        *res = 'g';
                   1151:                break;
                   1152:        case '<':
                   1153:                if (e)
                   1154:                        *res = 'l';
                   1155:                break;
                   1156:        default:
                   1157:                return(0);
                   1158:        }
                   1159:
                   1160:        *pos += 1 + e;
                   1161:
                   1162:        return(*res);
                   1163: }
                   1164:
1.88      kristaps 1165: static enum roffrule
                   1166: roff_evalcond(const char *v, int *pos)
                   1167: {
1.184     schwarze 1168:        int      not, lh, rh;
                   1169:        char     op;
1.88      kristaps 1170:
                   1171:        switch (v[*pos]) {
                   1172:        case ('n'):
                   1173:                (*pos)++;
                   1174:                return(ROFFRULE_ALLOW);
                   1175:        case ('e'):
                   1176:                /* FALLTHROUGH */
                   1177:        case ('o'):
                   1178:                /* FALLTHROUGH */
                   1179:        case ('t'):
                   1180:                (*pos)++;
                   1181:                return(ROFFRULE_DENY);
1.184     schwarze 1182:        case ('!'):
                   1183:                (*pos)++;
                   1184:                not = 1;
                   1185:                break;
1.88      kristaps 1186:        default:
1.184     schwarze 1187:                not = 0;
1.88      kristaps 1188:                break;
                   1189:        }
                   1190:
1.184     schwarze 1191:        if (!roff_getnum(v, pos, &lh))
                   1192:                return ROFFRULE_DENY;
                   1193:        if (!roff_getop(v, pos, &op)) {
                   1194:                if (lh < 0)
                   1195:                        lh = 0;
                   1196:                goto out;
                   1197:        }
                   1198:        if (!roff_getnum(v, pos, &rh))
                   1199:                return ROFFRULE_DENY;
                   1200:        switch (op) {
                   1201:        case 'g':
                   1202:                lh = lh >= rh;
                   1203:                break;
                   1204:        case 'l':
                   1205:                lh = lh <= rh;
                   1206:                break;
                   1207:        case '=':
                   1208:                lh = lh == rh;
                   1209:                break;
                   1210:        case '>':
                   1211:                lh = lh > rh;
                   1212:                break;
                   1213:        case '<':
                   1214:                lh = lh < rh;
                   1215:                break;
                   1216:        default:
                   1217:                return ROFFRULE_DENY;
                   1218:        }
                   1219: out:
                   1220:        if (not)
                   1221:                lh = !lh;
                   1222:        return lh ? ROFFRULE_ALLOW : ROFFRULE_DENY;
1.88      kristaps 1223: }
                   1224:
1.75      kristaps 1225: /* ARGSUSED */
1.74      kristaps 1226: static enum rofferr
1.103     kristaps 1227: roff_line_ignore(ROFF_ARGS)
1.89      kristaps 1228: {
1.123     schwarze 1229:
1.89      kristaps 1230:        return(ROFF_IGN);
                   1231: }
                   1232:
1.104     kristaps 1233: /* ARGSUSED */
                   1234: static enum rofferr
1.82      kristaps 1235: roff_cond(ROFF_ARGS)
1.74      kristaps 1236: {
1.173     schwarze 1237:
                   1238:        roffnode_push(r, tok, NULL, ln, ppos);
1.74      kristaps 1239:
1.134     kristaps 1240:        /*
                   1241:         * An `.el' has no conditional body: it will consume the value
                   1242:         * of the current rstack entry set in prior `ie' calls or
                   1243:         * defaults to DENY.
                   1244:         *
                   1245:         * If we're not an `el', however, then evaluate the conditional.
                   1246:         */
1.133     kristaps 1247:
1.173     schwarze 1248:        r->last->rule = ROFF_el == tok ?
1.134     kristaps 1249:                (r->rstackpos < 0 ?
                   1250:                 ROFFRULE_DENY : r->rstack[r->rstackpos--]) :
                   1251:                roff_evalcond(*bufp, &pos);
1.77      kristaps 1252:
1.134     kristaps 1253:        /*
                   1254:         * An if-else will put the NEGATION of the current evaluated
                   1255:         * conditional into the stack of rules.
                   1256:         */
                   1257:
1.84      schwarze 1258:        if (ROFF_ie == tok) {
1.134     kristaps 1259:                if (r->rstackpos == RSTACK_MAX - 1) {
                   1260:                        mandoc_msg(MANDOCERR_MEM,
                   1261:                                r->parse, ln, ppos, NULL);
                   1262:                        return(ROFF_ERR);
                   1263:                }
                   1264:                r->rstack[++r->rstackpos] =
                   1265:                        ROFFRULE_DENY == r->last->rule ?
                   1266:                        ROFFRULE_ALLOW : ROFFRULE_DENY;
1.82      kristaps 1267:        }
1.88      kristaps 1268:
                   1269:        /* If the parent has false as its rule, then so do we. */
                   1270:
1.109     kristaps 1271:        if (r->last->parent && ROFFRULE_DENY == r->last->parent->rule)
1.84      schwarze 1272:                r->last->rule = ROFFRULE_DENY;
1.88      kristaps 1273:
                   1274:        /*
1.173     schwarze 1275:         * Determine scope.
                   1276:         * If there is nothing on the line after the conditional,
                   1277:         * not even whitespace, use next-line scope.
1.88      kristaps 1278:         */
1.74      kristaps 1279:
1.173     schwarze 1280:        if ('\0' == (*bufp)[pos]) {
                   1281:                r->last->endspan = 2;
                   1282:                goto out;
                   1283:        }
                   1284:
                   1285:        while (' ' == (*bufp)[pos])
                   1286:                pos++;
                   1287:
                   1288:        /* An opening brace requests multiline scope. */
1.75      kristaps 1289:
                   1290:        if ('\\' == (*bufp)[pos] && '{' == (*bufp)[pos + 1]) {
                   1291:                r->last->endspan = -1;
                   1292:                pos += 2;
1.173     schwarze 1293:                goto out;
1.109     kristaps 1294:        }
1.74      kristaps 1295:
1.77      kristaps 1296:        /*
1.173     schwarze 1297:         * Anything else following the conditional causes
                   1298:         * single-line scope.  Warn if the scope contains
                   1299:         * nothing but trailing whitespace.
1.77      kristaps 1300:         */
                   1301:
1.75      kristaps 1302:        if ('\0' == (*bufp)[pos])
1.173     schwarze 1303:                mandoc_msg(MANDOCERR_NOARGS, r->parse, ln, ppos, NULL);
1.77      kristaps 1304:
1.173     schwarze 1305:        r->last->endspan = 1;
1.74      kristaps 1306:
1.173     schwarze 1307: out:
1.75      kristaps 1308:        *offs = pos;
                   1309:        return(ROFF_RERUN);
1.83      schwarze 1310: }
                   1311:
                   1312:
                   1313: /* ARGSUSED */
                   1314: static enum rofferr
1.92      schwarze 1315: roff_ds(ROFF_ARGS)
                   1316: {
1.96      kristaps 1317:        char            *name, *string;
                   1318:
                   1319:        /*
                   1320:         * A symbol is named by the first word following the macro
                   1321:         * invocation up to a space.  Its value is anything after the
                   1322:         * name's trailing whitespace and optional double-quote.  Thus,
                   1323:         *
                   1324:         *  [.ds foo "bar  "     ]
                   1325:         *
                   1326:         * will have `bar  "     ' as its value.
                   1327:         */
1.92      schwarze 1328:
1.121     schwarze 1329:        string = *bufp + pos;
                   1330:        name = roff_getname(r, &string, ln, pos);
1.92      schwarze 1331:        if ('\0' == *name)
                   1332:                return(ROFF_IGN);
                   1333:
1.121     schwarze 1334:        /* Read past initial double-quote. */
                   1335:        if ('"' == *string)
1.92      schwarze 1336:                string++;
                   1337:
1.96      kristaps 1338:        /* The rest is the value. */
1.193     schwarze 1339:        roff_setstr(r, name, string, ROFF_as == tok);
1.92      schwarze 1340:        return(ROFF_IGN);
                   1341: }
                   1342:
1.180     schwarze 1343: void
1.187     schwarze 1344: roff_setreg(struct roff *r, const char *name, int val, char sign)
1.147     kristaps 1345: {
1.180     schwarze 1346:        struct roffreg  *reg;
                   1347:
                   1348:        /* Search for an existing register with the same name. */
                   1349:        reg = r->regtab;
                   1350:
                   1351:        while (reg && strcmp(name, reg->key.p))
                   1352:                reg = reg->next;
1.147     kristaps 1353:
1.180     schwarze 1354:        if (NULL == reg) {
                   1355:                /* Create a new register. */
                   1356:                reg = mandoc_malloc(sizeof(struct roffreg));
                   1357:                reg->key.p = mandoc_strdup(name);
                   1358:                reg->key.sz = strlen(name);
1.187     schwarze 1359:                reg->val = 0;
1.180     schwarze 1360:                reg->next = r->regtab;
                   1361:                r->regtab = reg;
                   1362:        }
                   1363:
1.187     schwarze 1364:        if ('+' == sign)
                   1365:                reg->val += val;
                   1366:        else if ('-' == sign)
                   1367:                reg->val -= val;
                   1368:        else
                   1369:                reg->val = val;
1.147     kristaps 1370: }
                   1371:
1.192     schwarze 1372: /*
                   1373:  * Handle some predefined read-only number registers.
                   1374:  * For now, return -1 if the requested register is not predefined;
                   1375:  * in case a predefined read-only register having the value -1
                   1376:  * were to turn up, another special value would have to be chosen.
                   1377:  */
                   1378: static int
                   1379: roff_getregro(const char *name)
                   1380: {
                   1381:
                   1382:        switch (*name) {
                   1383:        case ('A'):  /* ASCII approximation mode is always off. */
                   1384:                return(0);
                   1385:        case ('g'):  /* Groff compatibility mode is always on. */
                   1386:                return(1);
                   1387:        case ('H'):  /* Fixed horizontal resolution. */
                   1388:                return (24);
                   1389:        case ('j'):  /* Always adjust left margin only. */
                   1390:                return(0);
                   1391:        case ('T'):  /* Some output device is always defined. */
                   1392:                return(1);
                   1393:        case ('V'):  /* Fixed vertical resolution. */
                   1394:                return (40);
                   1395:        default:
                   1396:                return (-1);
                   1397:        }
                   1398: }
                   1399:
1.181     schwarze 1400: int
1.180     schwarze 1401: roff_getreg(const struct roff *r, const char *name)
1.147     kristaps 1402: {
1.180     schwarze 1403:        struct roffreg  *reg;
1.192     schwarze 1404:        int              val;
                   1405:
                   1406:        if ('.' == name[0] && '\0' != name[1] && '\0' == name[2]) {
                   1407:                val = roff_getregro(name + 1);
                   1408:                if (-1 != val)
                   1409:                        return (val);
                   1410:        }
1.180     schwarze 1411:
                   1412:        for (reg = r->regtab; reg; reg = reg->next)
                   1413:                if (0 == strcmp(name, reg->key.p))
1.181     schwarze 1414:                        return(reg->val);
                   1415:
                   1416:        return(0);
                   1417: }
                   1418:
                   1419: static int
                   1420: roff_getregn(const struct roff *r, const char *name, size_t len)
                   1421: {
                   1422:        struct roffreg  *reg;
1.192     schwarze 1423:        int              val;
                   1424:
                   1425:        if ('.' == name[0] && 2 == len) {
                   1426:                val = roff_getregro(name + 1);
                   1427:                if (-1 != val)
                   1428:                        return (val);
                   1429:        }
1.181     schwarze 1430:
                   1431:        for (reg = r->regtab; reg; reg = reg->next)
                   1432:                if (len == reg->key.sz &&
                   1433:                    0 == strncmp(name, reg->key.p, len))
                   1434:                        return(reg->val);
1.147     kristaps 1435:
1.180     schwarze 1436:        return(0);
1.147     kristaps 1437: }
                   1438:
1.180     schwarze 1439: static void
                   1440: roff_freereg(struct roffreg *reg)
1.147     kristaps 1441: {
1.180     schwarze 1442:        struct roffreg  *old_reg;
1.147     kristaps 1443:
1.180     schwarze 1444:        while (NULL != reg) {
                   1445:                free(reg->key.p);
                   1446:                old_reg = reg;
                   1447:                reg = reg->next;
                   1448:                free(old_reg);
                   1449:        }
1.147     kristaps 1450: }
1.92      schwarze 1451:
                   1452: /* ARGSUSED */
                   1453: static enum rofferr
1.89      kristaps 1454: roff_nr(ROFF_ARGS)
1.83      schwarze 1455: {
1.121     schwarze 1456:        const char      *key;
                   1457:        char            *val;
1.187     schwarze 1458:        size_t           sz;
1.138     kristaps 1459:        int              iv;
1.187     schwarze 1460:        char             sign;
1.89      kristaps 1461:
1.121     schwarze 1462:        val = *bufp + pos;
                   1463:        key = roff_getname(r, &val, ln, pos);
1.89      kristaps 1464:
1.187     schwarze 1465:        sign = *val;
                   1466:        if ('+' == sign || '-' == sign)
                   1467:                val++;
                   1468:
                   1469:        sz = strspn(val, "0123456789");
                   1470:        iv = sz ? mandoc_strntoi(val, sz, 10) : 0;
1.180     schwarze 1471:
1.187     schwarze 1472:        roff_setreg(r, key, iv, sign);
1.109     kristaps 1473:
1.122     schwarze 1474:        return(ROFF_IGN);
                   1475: }
                   1476:
                   1477: /* ARGSUSED */
                   1478: static enum rofferr
                   1479: roff_rm(ROFF_ARGS)
                   1480: {
                   1481:        const char       *name;
                   1482:        char             *cp;
                   1483:
                   1484:        cp = *bufp + pos;
                   1485:        while ('\0' != *cp) {
1.127     kristaps 1486:                name = roff_getname(r, &cp, ln, (int)(cp - *bufp));
1.122     schwarze 1487:                if ('\0' != *name)
                   1488:                        roff_setstr(r, name, NULL, 0);
                   1489:        }
1.178     schwarze 1490:        return(ROFF_IGN);
                   1491: }
                   1492:
                   1493: /* ARGSUSED */
                   1494: static enum rofferr
                   1495: roff_it(ROFF_ARGS)
                   1496: {
                   1497:        char            *cp;
                   1498:        size_t           len;
                   1499:        int              iv;
                   1500:
                   1501:        /* Parse the number of lines. */
                   1502:        cp = *bufp + pos;
                   1503:        len = strcspn(cp, " \t");
                   1504:        cp[len] = '\0';
                   1505:        if ((iv = mandoc_strntoi(cp, len, 10)) <= 0) {
                   1506:                mandoc_msg(MANDOCERR_NUMERIC, r->parse,
                   1507:                                ln, ppos, *bufp + 1);
                   1508:                return(ROFF_IGN);
                   1509:        }
                   1510:        cp += len + 1;
                   1511:
                   1512:        /* Arm the input line trap. */
                   1513:        roffit_lines = iv;
                   1514:        roffit_macro = mandoc_strdup(cp);
1.109     kristaps 1515:        return(ROFF_IGN);
1.175     schwarze 1516: }
                   1517:
                   1518: /* ARGSUSED */
                   1519: static enum rofferr
                   1520: roff_Dd(ROFF_ARGS)
                   1521: {
                   1522:        const char *const       *cp;
                   1523:
1.190     schwarze 1524:        if (0 == r->quick && MPARSE_MDOC != r->parsetype)
1.175     schwarze 1525:                for (cp = __mdoc_reserved; *cp; cp++)
                   1526:                        roff_setstr(r, *cp, NULL, 0);
                   1527:
                   1528:        return(ROFF_CONT);
                   1529: }
                   1530:
                   1531: /* ARGSUSED */
                   1532: static enum rofferr
                   1533: roff_TH(ROFF_ARGS)
                   1534: {
                   1535:        const char *const       *cp;
                   1536:
1.190     schwarze 1537:        if (0 == r->quick && MPARSE_MDOC != r->parsetype)
1.175     schwarze 1538:                for (cp = __man_reserved; *cp; cp++)
                   1539:                        roff_setstr(r, *cp, NULL, 0);
                   1540:
                   1541:        return(ROFF_CONT);
1.109     kristaps 1542: }
                   1543:
                   1544: /* ARGSUSED */
                   1545: static enum rofferr
                   1546: roff_TE(ROFF_ARGS)
                   1547: {
                   1548:
                   1549:        if (NULL == r->tbl)
1.128     kristaps 1550:                mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.115     kristaps 1551:        else
1.151     kristaps 1552:                tbl_end(&r->tbl);
1.109     kristaps 1553:
1.112     kristaps 1554:        return(ROFF_IGN);
                   1555: }
                   1556:
                   1557: /* ARGSUSED */
                   1558: static enum rofferr
                   1559: roff_T_(ROFF_ARGS)
                   1560: {
                   1561:
                   1562:        if (NULL == r->tbl)
1.128     kristaps 1563:                mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.112     kristaps 1564:        else
1.116     kristaps 1565:                tbl_restart(ppos, ln, r->tbl);
1.112     kristaps 1566:
1.109     kristaps 1567:        return(ROFF_IGN);
                   1568: }
                   1569:
1.156     kristaps 1570: #if 0
                   1571: static int
1.151     kristaps 1572: roff_closeeqn(struct roff *r)
                   1573: {
                   1574:
                   1575:        return(r->eqn && ROFF_EQN == eqn_end(&r->eqn) ? 1 : 0);
                   1576: }
1.156     kristaps 1577: #endif
1.151     kristaps 1578:
1.156     kristaps 1579: static void
1.151     kristaps 1580: roff_openeqn(struct roff *r, const char *name, int line,
                   1581:                int offs, const char *buf)
1.125     kristaps 1582: {
1.151     kristaps 1583:        struct eqn_node *e;
                   1584:        int              poff;
1.125     kristaps 1585:
                   1586:        assert(NULL == r->eqn);
1.151     kristaps 1587:        e = eqn_alloc(name, offs, line, r->parse);
1.125     kristaps 1588:
                   1589:        if (r->last_eqn)
                   1590:                r->last_eqn->next = e;
                   1591:        else
                   1592:                r->first_eqn = r->last_eqn = e;
                   1593:
                   1594:        r->eqn = r->last_eqn = e;
1.151     kristaps 1595:
                   1596:        if (buf) {
                   1597:                poff = 0;
                   1598:                eqn_read(&r->eqn, line, buf, offs, &poff);
                   1599:        }
                   1600: }
                   1601:
                   1602: /* ARGSUSED */
                   1603: static enum rofferr
                   1604: roff_EQ(ROFF_ARGS)
                   1605: {
                   1606:
                   1607:        roff_openeqn(r, *bufp + pos, ln, ppos, NULL);
1.125     kristaps 1608:        return(ROFF_IGN);
                   1609: }
                   1610:
                   1611: /* ARGSUSED */
                   1612: static enum rofferr
                   1613: roff_EN(ROFF_ARGS)
                   1614: {
                   1615:
1.128     kristaps 1616:        mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL);
1.125     kristaps 1617:        return(ROFF_IGN);
                   1618: }
                   1619:
                   1620: /* ARGSUSED */
                   1621: static enum rofferr
1.109     kristaps 1622: roff_TS(ROFF_ARGS)
                   1623: {
1.176     schwarze 1624:        struct tbl_node *tbl;
1.89      kristaps 1625:
1.115     kristaps 1626:        if (r->tbl) {
1.128     kristaps 1627:                mandoc_msg(MANDOCERR_SCOPEBROKEN, r->parse, ln, ppos, NULL);
1.151     kristaps 1628:                tbl_end(&r->tbl);
1.115     kristaps 1629:        }
1.83      schwarze 1630:
1.176     schwarze 1631:        tbl = tbl_alloc(ppos, ln, r->parse);
1.113     kristaps 1632:
                   1633:        if (r->last_tbl)
1.176     schwarze 1634:                r->last_tbl->next = tbl;
1.113     kristaps 1635:        else
1.176     schwarze 1636:                r->first_tbl = r->last_tbl = tbl;
1.113     kristaps 1637:
1.176     schwarze 1638:        r->tbl = r->last_tbl = tbl;
1.83      schwarze 1639:        return(ROFF_IGN);
1.92      schwarze 1640: }
                   1641:
1.105     kristaps 1642: /* ARGSUSED */
                   1643: static enum rofferr
1.174     kristaps 1644: roff_cc(ROFF_ARGS)
                   1645: {
                   1646:        const char      *p;
                   1647:
                   1648:        p = *bufp + pos;
                   1649:
                   1650:        if ('\0' == *p || '.' == (r->control = *p++))
                   1651:                r->control = 0;
                   1652:
                   1653:        if ('\0' != *p)
                   1654:                mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
                   1655:
                   1656:        return(ROFF_IGN);
                   1657: }
                   1658:
                   1659: /* ARGSUSED */
                   1660: static enum rofferr
1.164     kristaps 1661: roff_tr(ROFF_ARGS)
                   1662: {
                   1663:        const char      *p, *first, *second;
                   1664:        size_t           fsz, ssz;
                   1665:        enum mandoc_esc  esc;
                   1666:
                   1667:        p = *bufp + pos;
                   1668:
                   1669:        if ('\0' == *p) {
                   1670:                mandoc_msg(MANDOCERR_ARGCOUNT, r->parse, ln, ppos, NULL);
                   1671:                return(ROFF_IGN);
                   1672:        }
                   1673:
                   1674:        while ('\0' != *p) {
                   1675:                fsz = ssz = 1;
                   1676:
                   1677:                first = p++;
                   1678:                if ('\\' == *first) {
                   1679:                        esc = mandoc_escape(&p, NULL, NULL);
                   1680:                        if (ESCAPE_ERROR == esc) {
                   1681:                                mandoc_msg
                   1682:                                        (MANDOCERR_BADESCAPE, r->parse,
                   1683:                                         ln, (int)(p - *bufp), NULL);
                   1684:                                return(ROFF_IGN);
                   1685:                        }
                   1686:                        fsz = (size_t)(p - first);
                   1687:                }
                   1688:
                   1689:                second = p++;
                   1690:                if ('\\' == *second) {
                   1691:                        esc = mandoc_escape(&p, NULL, NULL);
                   1692:                        if (ESCAPE_ERROR == esc) {
                   1693:                                mandoc_msg
                   1694:                                        (MANDOCERR_BADESCAPE, r->parse,
                   1695:                                         ln, (int)(p - *bufp), NULL);
                   1696:                                return(ROFF_IGN);
                   1697:                        }
                   1698:                        ssz = (size_t)(p - second);
1.165     kristaps 1699:                } else if ('\0' == *second) {
1.164     kristaps 1700:                        mandoc_msg(MANDOCERR_ARGCOUNT, r->parse,
                   1701:                                        ln, (int)(p - *bufp), NULL);
                   1702:                        second = " ";
1.165     kristaps 1703:                        p--;
1.164     kristaps 1704:                }
                   1705:
1.167     kristaps 1706:                if (fsz > 1) {
                   1707:                        roff_setstrn(&r->xmbtab, first,
                   1708:                                        fsz, second, ssz, 0);
                   1709:                        continue;
                   1710:                }
                   1711:
                   1712:                if (NULL == r->xtab)
                   1713:                        r->xtab = mandoc_calloc
                   1714:                                (128, sizeof(struct roffstr));
                   1715:
                   1716:                free(r->xtab[(int)*first].p);
                   1717:                r->xtab[(int)*first].p = mandoc_strndup(second, ssz);
                   1718:                r->xtab[(int)*first].sz = ssz;
1.164     kristaps 1719:        }
                   1720:
                   1721:        return(ROFF_IGN);
                   1722: }
                   1723:
                   1724: /* ARGSUSED */
                   1725: static enum rofferr
1.105     kristaps 1726: roff_so(ROFF_ARGS)
                   1727: {
                   1728:        char *name;
                   1729:
1.128     kristaps 1730:        mandoc_msg(MANDOCERR_SO, r->parse, ln, ppos, NULL);
1.105     kristaps 1731:
                   1732:        /*
                   1733:         * Handle `so'.  Be EXTREMELY careful, as we shouldn't be
                   1734:         * opening anything that's not in our cwd or anything beneath
                   1735:         * it.  Thus, explicitly disallow traversing up the file-system
                   1736:         * or using absolute paths.
                   1737:         */
                   1738:
                   1739:        name = *bufp + pos;
                   1740:        if ('/' == *name || strstr(name, "../") || strstr(name, "/..")) {
1.128     kristaps 1741:                mandoc_msg(MANDOCERR_SOPATH, r->parse, ln, pos, NULL);
1.105     kristaps 1742:                return(ROFF_ERR);
                   1743:        }
                   1744:
                   1745:        *offs = pos;
                   1746:        return(ROFF_SO);
                   1747: }
1.92      schwarze 1748:
1.106     kristaps 1749: /* ARGSUSED */
                   1750: static enum rofferr
                   1751: roff_userdef(ROFF_ARGS)
1.99      kristaps 1752: {
1.106     kristaps 1753:        const char       *arg[9];
                   1754:        char             *cp, *n1, *n2;
1.119     schwarze 1755:        int               i;
1.106     kristaps 1756:
                   1757:        /*
                   1758:         * Collect pointers to macro argument strings
1.188     schwarze 1759:         * and NUL-terminate them.
1.106     kristaps 1760:         */
                   1761:        cp = *bufp + pos;
1.119     schwarze 1762:        for (i = 0; i < 9; i++)
1.120     schwarze 1763:                arg[i] = '\0' == *cp ? "" :
1.136     kristaps 1764:                    mandoc_getarg(r->parse, &cp, ln, &pos);
1.99      kristaps 1765:
1.106     kristaps 1766:        /*
                   1767:         * Expand macro arguments.
1.99      kristaps 1768:         */
1.106     kristaps 1769:        *szp = 0;
                   1770:        n1 = cp = mandoc_strdup(r->current_string);
                   1771:        while (NULL != (cp = strstr(cp, "\\$"))) {
                   1772:                i = cp[2] - '1';
                   1773:                if (0 > i || 8 < i) {
                   1774:                        /* Not an argument invocation. */
                   1775:                        cp += 2;
                   1776:                        continue;
                   1777:                }
                   1778:
                   1779:                *szp = strlen(n1) - 3 + strlen(arg[i]) + 1;
                   1780:                n2 = mandoc_malloc(*szp);
                   1781:
                   1782:                strlcpy(n2, n1, (size_t)(cp - n1 + 1));
                   1783:                strlcat(n2, arg[i], *szp);
                   1784:                strlcat(n2, cp + 3, *szp);
                   1785:
                   1786:                cp = n2 + (cp - n1);
                   1787:                free(n1);
                   1788:                n1 = n2;
1.99      kristaps 1789:        }
                   1790:
1.106     kristaps 1791:        /*
                   1792:         * Replace the macro invocation
                   1793:         * by the expanded macro.
                   1794:         */
                   1795:        free(*bufp);
                   1796:        *bufp = n1;
                   1797:        if (0 == *szp)
                   1798:                *szp = strlen(*bufp) + 1;
                   1799:
                   1800:        return(*szp > 1 && '\n' == (*bufp)[(int)*szp - 2] ?
                   1801:           ROFF_REPARSE : ROFF_APPEND);
1.99      kristaps 1802: }
1.121     schwarze 1803:
                   1804: static char *
                   1805: roff_getname(struct roff *r, char **cpp, int ln, int pos)
                   1806: {
                   1807:        char     *name, *cp;
                   1808:
                   1809:        name = *cpp;
                   1810:        if ('\0' == *name)
                   1811:                return(name);
                   1812:
                   1813:        /* Read until end of name. */
                   1814:        for (cp = name; '\0' != *cp && ' ' != *cp; cp++) {
                   1815:                if ('\\' != *cp)
                   1816:                        continue;
                   1817:                cp++;
                   1818:                if ('\\' == *cp)
                   1819:                        continue;
1.128     kristaps 1820:                mandoc_msg(MANDOCERR_NAMESC, r->parse, ln, pos, NULL);
1.121     schwarze 1821:                *cp = '\0';
                   1822:                name = cp;
                   1823:        }
                   1824:
                   1825:        /* Nil-terminate name. */
                   1826:        if ('\0' != *cp)
                   1827:                *(cp++) = '\0';
                   1828:
                   1829:        /* Read past spaces. */
                   1830:        while (' ' == *cp)
                   1831:                cp++;
                   1832:
                   1833:        *cpp = cp;
                   1834:        return(name);
                   1835: }
                   1836:
1.106     kristaps 1837: /*
                   1838:  * Store *string into the user-defined string called *name.
                   1839:  * To clear an existing entry, call with (*r, *name, NULL, 0).
1.193     schwarze 1840:  * append == 0: replace mode
                   1841:  * append == 1: single-line append mode
                   1842:  * append == 2: multiline append mode, append '\n' after each call
1.106     kristaps 1843:  */
1.94      kristaps 1844: static void
1.106     kristaps 1845: roff_setstr(struct roff *r, const char *name, const char *string,
1.193     schwarze 1846:        int append)
1.92      schwarze 1847: {
1.164     kristaps 1848:
                   1849:        roff_setstrn(&r->strtab, name, strlen(name), string,
1.193     schwarze 1850:                        string ? strlen(string) : 0, append);
1.164     kristaps 1851: }
                   1852:
                   1853: static void
1.166     kristaps 1854: roff_setstrn(struct roffkv **r, const char *name, size_t namesz,
1.193     schwarze 1855:                const char *string, size_t stringsz, int append)
1.164     kristaps 1856: {
1.166     kristaps 1857:        struct roffkv   *n;
1.164     kristaps 1858:        char            *c;
                   1859:        int              i;
                   1860:        size_t           oldch, newch;
1.92      schwarze 1861:
1.106     kristaps 1862:        /* Search for an existing string with the same name. */
1.164     kristaps 1863:        n = *r;
                   1864:
1.166     kristaps 1865:        while (n && strcmp(name, n->key.p))
1.92      schwarze 1866:                n = n->next;
1.94      kristaps 1867:
                   1868:        if (NULL == n) {
1.106     kristaps 1869:                /* Create a new string table entry. */
1.166     kristaps 1870:                n = mandoc_malloc(sizeof(struct roffkv));
                   1871:                n->key.p = mandoc_strndup(name, namesz);
                   1872:                n->key.sz = namesz;
                   1873:                n->val.p = NULL;
                   1874:                n->val.sz = 0;
1.164     kristaps 1875:                n->next = *r;
                   1876:                *r = n;
1.193     schwarze 1877:        } else if (0 == append) {
1.166     kristaps 1878:                free(n->val.p);
                   1879:                n->val.p = NULL;
                   1880:                n->val.sz = 0;
1.106     kristaps 1881:        }
                   1882:
                   1883:        if (NULL == string)
                   1884:                return;
                   1885:
                   1886:        /*
                   1887:         * One additional byte for the '\n' in multiline mode,
                   1888:         * and one for the terminating '\0'.
                   1889:         */
1.193     schwarze 1890:        newch = stringsz + (1 < append ? 2u : 1u);
1.164     kristaps 1891:
1.166     kristaps 1892:        if (NULL == n->val.p) {
                   1893:                n->val.p = mandoc_malloc(newch);
                   1894:                *n->val.p = '\0';
1.106     kristaps 1895:                oldch = 0;
                   1896:        } else {
1.166     kristaps 1897:                oldch = n->val.sz;
                   1898:                n->val.p = mandoc_realloc(n->val.p, oldch + newch);
1.106     kristaps 1899:        }
                   1900:
                   1901:        /* Skip existing content in the destination buffer. */
1.166     kristaps 1902:        c = n->val.p + (int)oldch;
1.106     kristaps 1903:
                   1904:        /* Append new content to the destination buffer. */
1.164     kristaps 1905:        i = 0;
                   1906:        while (i < (int)stringsz) {
1.106     kristaps 1907:                /*
                   1908:                 * Rudimentary roff copy mode:
                   1909:                 * Handle escaped backslashes.
                   1910:                 */
1.164     kristaps 1911:                if ('\\' == string[i] && '\\' == string[i + 1])
                   1912:                        i++;
                   1913:                *c++ = string[i++];
1.106     kristaps 1914:        }
1.94      kristaps 1915:
1.106     kristaps 1916:        /* Append terminating bytes. */
1.193     schwarze 1917:        if (1 < append)
1.106     kristaps 1918:                *c++ = '\n';
1.163     kristaps 1919:
1.106     kristaps 1920:        *c = '\0';
1.166     kristaps 1921:        n->val.sz = (int)(c - n->val.p);
1.92      schwarze 1922: }
                   1923:
1.94      kristaps 1924: static const char *
                   1925: roff_getstrn(const struct roff *r, const char *name, size_t len)
1.92      schwarze 1926: {
1.166     kristaps 1927:        const struct roffkv *n;
1.191     schwarze 1928:        int i;
1.92      schwarze 1929:
1.164     kristaps 1930:        for (n = r->strtab; n; n = n->next)
1.166     kristaps 1931:                if (0 == strncmp(name, n->key.p, len) &&
                   1932:                                '\0' == n->key.p[(int)len])
                   1933:                        return(n->val.p);
1.191     schwarze 1934:
                   1935:        for (i = 0; i < PREDEFS_MAX; i++)
                   1936:                if (0 == strncmp(name, predefs[i].name, len) &&
                   1937:                                '\0' == predefs[i].name[(int)len])
                   1938:                        return(predefs[i].str);
1.94      kristaps 1939:
1.157     kristaps 1940:        return(NULL);
1.92      schwarze 1941: }
                   1942:
1.94      kristaps 1943: static void
1.167     kristaps 1944: roff_freestr(struct roffkv *r)
1.92      schwarze 1945: {
1.166     kristaps 1946:        struct roffkv    *n, *nn;
1.92      schwarze 1947:
1.167     kristaps 1948:        for (n = r; n; n = nn) {
1.166     kristaps 1949:                free(n->key.p);
                   1950:                free(n->val.p);
1.92      schwarze 1951:                nn = n->next;
                   1952:                free(n);
                   1953:        }
1.114     kristaps 1954: }
                   1955:
                   1956: const struct tbl_span *
                   1957: roff_span(const struct roff *r)
                   1958: {
                   1959:
                   1960:        return(r->tbl ? tbl_span(r->tbl) : NULL);
1.125     kristaps 1961: }
                   1962:
                   1963: const struct eqn *
                   1964: roff_eqn(const struct roff *r)
                   1965: {
                   1966:
                   1967:        return(r->last_eqn ? &r->last_eqn->eqn : NULL);
1.164     kristaps 1968: }
                   1969:
                   1970: /*
                   1971:  * Duplicate an input string, making the appropriate character
                   1972:  * conversations (as stipulated by `tr') along the way.
                   1973:  * Returns a heap-allocated string with all the replacements made.
                   1974:  */
                   1975: char *
                   1976: roff_strdup(const struct roff *r, const char *p)
                   1977: {
1.166     kristaps 1978:        const struct roffkv *cp;
1.164     kristaps 1979:        char            *res;
                   1980:        const char      *pp;
                   1981:        size_t           ssz, sz;
                   1982:        enum mandoc_esc  esc;
                   1983:
1.167     kristaps 1984:        if (NULL == r->xmbtab && NULL == r->xtab)
1.164     kristaps 1985:                return(mandoc_strdup(p));
                   1986:        else if ('\0' == *p)
                   1987:                return(mandoc_strdup(""));
                   1988:
                   1989:        /*
                   1990:         * Step through each character looking for term matches
                   1991:         * (remember that a `tr' can be invoked with an escape, which is
                   1992:         * a glyph but the escape is multi-character).
                   1993:         * We only do this if the character hash has been initialised
                   1994:         * and the string is >0 length.
                   1995:         */
                   1996:
                   1997:        res = NULL;
                   1998:        ssz = 0;
                   1999:
                   2000:        while ('\0' != *p) {
1.167     kristaps 2001:                if ('\\' != *p && r->xtab && r->xtab[(int)*p].p) {
                   2002:                        sz = r->xtab[(int)*p].sz;
                   2003:                        res = mandoc_realloc(res, ssz + sz + 1);
                   2004:                        memcpy(res + ssz, r->xtab[(int)*p].p, sz);
                   2005:                        ssz += sz;
                   2006:                        p++;
                   2007:                        continue;
                   2008:                } else if ('\\' != *p) {
                   2009:                        res = mandoc_realloc(res, ssz + 2);
                   2010:                        res[ssz++] = *p++;
                   2011:                        continue;
                   2012:                }
                   2013:
1.164     kristaps 2014:                /* Search for term matches. */
1.167     kristaps 2015:                for (cp = r->xmbtab; cp; cp = cp->next)
1.166     kristaps 2016:                        if (0 == strncmp(p, cp->key.p, cp->key.sz))
1.164     kristaps 2017:                                break;
                   2018:
                   2019:                if (NULL != cp) {
                   2020:                        /*
                   2021:                         * A match has been found.
                   2022:                         * Append the match to the array and move
                   2023:                         * forward by its keysize.
                   2024:                         */
1.166     kristaps 2025:                        res = mandoc_realloc
                   2026:                                (res, ssz + cp->val.sz + 1);
                   2027:                        memcpy(res + ssz, cp->val.p, cp->val.sz);
                   2028:                        ssz += cp->val.sz;
                   2029:                        p += (int)cp->key.sz;
1.164     kristaps 2030:                        continue;
                   2031:                }
                   2032:
1.167     kristaps 2033:                /*
                   2034:                 * Handle escapes carefully: we need to copy
                   2035:                 * over just the escape itself, or else we might
                   2036:                 * do replacements within the escape itself.
                   2037:                 * Make sure to pass along the bogus string.
                   2038:                 */
                   2039:                pp = p++;
                   2040:                esc = mandoc_escape(&p, NULL, NULL);
                   2041:                if (ESCAPE_ERROR == esc) {
                   2042:                        sz = strlen(pp);
1.164     kristaps 2043:                        res = mandoc_realloc(res, ssz + sz + 1);
                   2044:                        memcpy(res + ssz, pp, sz);
1.167     kristaps 2045:                        break;
1.164     kristaps 2046:                }
1.167     kristaps 2047:                /*
                   2048:                 * We bail out on bad escapes.
                   2049:                 * No need to warn: we already did so when
                   2050:                 * roff_res() was called.
                   2051:                 */
                   2052:                sz = (int)(p - pp);
                   2053:                res = mandoc_realloc(res, ssz + sz + 1);
                   2054:                memcpy(res + ssz, pp, sz);
                   2055:                ssz += sz;
1.164     kristaps 2056:        }
                   2057:
                   2058:        res[(int)ssz] = '\0';
                   2059:        return(res);
1.174     kristaps 2060: }
                   2061:
                   2062: /*
                   2063:  * Find out whether a line is a macro line or not.
                   2064:  * If it is, adjust the current position and return one; if it isn't,
                   2065:  * return zero and don't change the current position.
                   2066:  * If the control character has been set with `.cc', then let that grain
                   2067:  * precedence.
                   2068:  * This is slighly contrary to groff, where using the non-breaking
                   2069:  * control character when `cc' has been invoked will cause the
                   2070:  * non-breaking macro contents to be printed verbatim.
                   2071:  */
                   2072: int
                   2073: roff_getcontrol(const struct roff *r, const char *cp, int *ppos)
                   2074: {
                   2075:        int             pos;
                   2076:
                   2077:        pos = *ppos;
                   2078:
                   2079:        if (0 != r->control && cp[pos] == r->control)
                   2080:                pos++;
                   2081:        else if (0 != r->control)
                   2082:                return(0);
                   2083:        else if ('\\' == cp[pos] && '.' == cp[pos + 1])
                   2084:                pos += 2;
                   2085:        else if ('.' == cp[pos] || '\'' == cp[pos])
                   2086:                pos++;
                   2087:        else
                   2088:                return(0);
                   2089:
                   2090:        while (' ' == cp[pos] || '\t' == cp[pos])
                   2091:                pos++;
                   2092:
                   2093:        *ppos = pos;
                   2094:        return(1);
1.74      kristaps 2095: }

CVSweb