=================================================================== RCS file: /cvs/mandoc/roff.c,v retrieving revision 1.134 retrieving revision 1.158 diff -u -p -r1.134 -r1.158 --- mandoc/roff.c 2011/04/13 16:26:11 1.134 +++ mandoc/roff.c 2011/07/27 13:47:26 1.158 @@ -1,4 +1,4 @@ -/* $Id: roff.c,v 1.134 2011/04/13 16:26:11 kristaps Exp $ */ +/* $Id: roff.c,v 1.158 2011/07/27 13:47:26 kristaps Exp $ */ /* * Copyright (c) 2010, 2011 Kristaps Dzonsons * Copyright (c) 2010, 2011 Ingo Schwarze @@ -20,17 +20,15 @@ #endif #include -#include #include -#include #include #include -#include #include "mandoc.h" #include "libroff.h" #include "libmandoc.h" +/* Maximum number of nested if-else conditionals. */ #define RSTACK_MAX 128 enum rofft { @@ -63,7 +61,7 @@ enum rofft { ROFF_EQ, ROFF_EN, ROFF_cblock, - ROFF_ccond, /* FIXME: remove this. */ + ROFF_ccond, ROFF_USERDEF, ROFF_MAX }; @@ -73,6 +71,16 @@ enum roffrule { ROFFRULE_DENY }; +/* + * A single register entity. If "set" is zero, the value of the + * register should be the default one, which is per-register. + * Registers are assumed to be unsigned ints for now. + */ +struct reg { + int set; /* whether set or not */ + unsigned int u; /* unsigned integer */ +}; + struct roffstr { char *name; /* key of symbol */ char *string; /* current value */ @@ -84,7 +92,7 @@ struct roff { struct roffnode *last; /* leaf of stack */ enum roffrule rstack[RSTACK_MAX]; /* stack of !`ie' rules */ int rstackpos; /* position in rstack */ - struct regset *regs; /* read/writable registers */ + struct reg regs[REG__MAX]; struct roffstr *first_string; /* user-defined strings & macros */ const char *current_string; /* value of last called user macro */ struct tbl_node *first_tbl; /* first table parsed */ @@ -127,6 +135,20 @@ struct roffmac { struct roffmac *next; }; +struct predef { + const char *name; /* predefined input name */ + const char *str; /* replacement symbol */ +}; + +#define PREDEF(__name, __str) \ + { (__name), (__str) }, + +static enum rofft roffhash_find(const char *, size_t); +static void roffhash_init(void); +static void roffnode_cleanscope(struct roff *); +static void roffnode_pop(struct roff *); +static void roffnode_push(struct roff *, enum rofft, + const char *, int, int); static enum rofferr roff_block(ROFF_ARGS); static enum rofferr roff_block_text(ROFF_ARGS); static enum rofferr roff_block_sub(ROFF_ARGS); @@ -137,14 +159,19 @@ static enum rofferr roff_cond_text(ROFF_ARGS); static enum rofferr roff_cond_sub(ROFF_ARGS); static enum rofferr roff_ds(ROFF_ARGS); static enum roffrule roff_evalcond(const char *, int *); +static void roff_free1(struct roff *); static void roff_freestr(struct roff *); static char *roff_getname(struct roff *, char **, int, int); static const char *roff_getstrn(const struct roff *, const char *, size_t); static enum rofferr roff_line_ignore(ROFF_ARGS); static enum rofferr roff_nr(ROFF_ARGS); -static int roff_res(struct roff *, - char **, size_t *, int); +static void roff_openeqn(struct roff *, const char *, + int, int, const char *); +static enum rofft roff_parse(struct roff *, const char *, int *); +static enum rofferr roff_parsetext(char *); +static void roff_res(struct roff *, + char **, size_t *, int, int); static enum rofferr roff_rm(ROFF_ARGS); static void roff_setstr(struct roff *, const char *, const char *, int); @@ -156,7 +183,7 @@ static enum rofferr roff_EN(ROFF_ARGS); static enum rofferr roff_T_(ROFF_ARGS); static enum rofferr roff_userdef(ROFF_ARGS); -/* See roff_hash_find() */ +/* See roffhash_find() */ #define ASCII_HI 126 #define ASCII_LO 33 @@ -198,21 +225,17 @@ static struct roffmac roffs[ROFF_MAX] = { { NULL, roff_userdef, NULL, NULL, 0, NULL }, }; -static void roff_free1(struct roff *); -static enum rofft roff_hash_find(const char *, size_t); -static void roff_hash_init(void); -static void roffnode_cleanscope(struct roff *); -static void roffnode_push(struct roff *, enum rofft, - const char *, int, int); -static void roffnode_pop(struct roff *); -static enum rofft roff_parse(struct roff *, const char *, int *); -static int roff_parse_nat(const char *, unsigned int *); +/* Array of injected predefined strings. */ +#define PREDEFS_MAX 38 +static const struct predef predefs[PREDEFS_MAX] = { +#include "predefs.in" +}; -/* See roff_hash_find() */ +/* See roffhash_find() */ #define ROFF_HASH(p) (p[0] - ASCII_LO) static void -roff_hash_init(void) +roffhash_init(void) { struct roffmac *n; int buc, i; @@ -232,13 +255,12 @@ roff_hash_init(void) } } - /* * Look up a roff token by its name. Returns ROFF_MAX if no macro by * the nil-terminated string name could be found. */ static enum rofft -roff_hash_find(const char *p, size_t s) +roffhash_find(const char *p, size_t s) { int buc; struct roffmac *n; @@ -337,8 +359,14 @@ roff_free1(struct roff *r) void roff_reset(struct roff *r) { + int i; roff_free1(r); + + memset(&r->regs, 0, sizeof(struct reg) * REG__MAX); + + for (i = 0; i < PREDEFS_MAX; i++) + roff_setstr(r, predefs[i].name, predefs[i].str, 0); } @@ -352,28 +380,33 @@ roff_free(struct roff *r) struct roff * -roff_alloc(struct regset *regs, struct mparse *parse) +roff_alloc(struct mparse *parse) { struct roff *r; + int i; r = mandoc_calloc(1, sizeof(struct roff)); - r->regs = regs; r->parse = parse; r->rstackpos = -1; - roff_hash_init(); + roffhash_init(); + + for (i = 0; i < PREDEFS_MAX; i++) + roff_setstr(r, predefs[i].name, predefs[i].str, 0); + return(r); } - /* * Pre-filter each and every line for reserved words (one beginning with * `\*', e.g., `\*(ab'). These must be handled before the actual line * is processed. + * This also checks the syntax of regular escapes. */ -static int -roff_res(struct roff *r, char **bufp, size_t *szp, int pos) +static void +roff_res(struct roff *r, char **bufp, size_t *szp, int ln, int pos) { + enum mandoc_esc esc; const char *stesc; /* start of an escape sequence ('\\') */ const char *stnam; /* start of the name, after "[(*" */ const char *cp; /* end of the name, e.g. before ']' */ @@ -382,8 +415,7 @@ roff_res(struct roff *r, char **bufp, size_t *szp, int size_t nsz; char *n; - /* Search for a leading backslash and save a pointer to it. */ - +again: cp = *bufp + pos; while (NULL != (cp = strchr(cp, '\\'))) { stesc = cp++; @@ -395,10 +427,22 @@ roff_res(struct roff *r, char **bufp, size_t *szp, int */ if ('\0' == *cp) - return(1); - if ('*' != *cp++) - continue; + return; + if ('*' != *cp) { + res = cp; + esc = mandoc_escape(&cp, NULL, NULL); + if (ESCAPE_ERROR != esc) + continue; + cp = res; + mandoc_msg + (MANDOCERR_BADESCAPE, r->parse, + ln, (int)(stesc - *bufp), NULL); + return; + } + + cp++; + /* * The third character decides the length * of the name of the string. @@ -407,7 +451,7 @@ roff_res(struct roff *r, char **bufp, size_t *szp, int switch (*cp) { case ('\0'): - return(1); + return; case ('('): cp++; maxl = 2; @@ -425,8 +469,13 @@ roff_res(struct roff *r, char **bufp, size_t *szp, int /* Advance to the end of the name. */ for (i = 0; 0 == maxl || i < maxl; i++, cp++) { - if ('\0' == *cp) - return(1); /* Error. */ + if ('\0' == *cp) { + mandoc_msg + (MANDOCERR_BADESCAPE, + r->parse, ln, + (int)(stesc - *bufp), NULL); + return; + } if (0 == maxl && ']' == *cp) break; } @@ -439,12 +488,16 @@ roff_res(struct roff *r, char **bufp, size_t *szp, int res = roff_getstrn(r, stnam, (size_t)i); if (NULL == res) { - cp -= maxl ? 1 : 0; - continue; + mandoc_msg + (MANDOCERR_BADESCAPE, r->parse, + ln, (int)(stesc - *bufp), NULL); + res = ""; } /* Replace the escape sequence by the string. */ + pos += (stesc - *bufp); + nsz = *szp + strlen(res) + 1; n = mandoc_malloc(nsz); @@ -456,13 +509,56 @@ roff_res(struct roff *r, char **bufp, size_t *szp, int *bufp = n; *szp = nsz; - return(0); + goto again; } - - return(1); } +/* + * Process text streams: convert all breakable hyphens into ASCII_HYPH. + */ +static enum rofferr +roff_parsetext(char *p) +{ + char l, r; + size_t sz; + const char *start; + enum mandoc_esc esc; + start = p; + + while ('\0' != *p) { + sz = strcspn(p, "-\\"); + p += sz; + + if ('\\' == *p) { + /* Skip over escapes. */ + p++; + esc = mandoc_escape + ((const char **)&p, NULL, NULL); + if (ESCAPE_ERROR == esc) + break; + continue; + } else if ('-' != *p || p == start) { + p++; + continue; + } + + l = *(p - 1); + r = *(p + 1); + + if ('\\' != l && + '\t' != r && '\t' != l && + ' ' != r && ' ' != l && + '-' != r && '-' != l && + ! isdigit((unsigned char)l) && + ! isdigit((unsigned char)r)) + *p = ASCII_HYPH; + p++; + } + + return(ROFF_CONT); +} + enum rofferr roff_parseln(struct roff *r, int ln, char **bufp, size_t *szp, int pos, int *offs) @@ -476,8 +572,7 @@ roff_parseln(struct roff *r, int ln, char **bufp, * words to fill in. */ - if (r->first_string && ! roff_res(r, bufp, szp, pos)) - return(ROFF_REPARSE); + roff_res(r, bufp, szp, ln, pos); ppos = pos; ctl = mandoc_getcontrol(*bufp, &pos); @@ -499,18 +594,18 @@ roff_parseln(struct roff *r, int ln, char **bufp, if (ROFF_CONT != e) return(e); if (r->eqn) - return(eqn_read(&r->eqn, ln, *bufp, pos)); + return(eqn_read(&r->eqn, ln, *bufp, pos, offs)); if (r->tbl) return(tbl_read(r->tbl, ln, *bufp, pos)); - return(ROFF_CONT); + return(roff_parsetext(*bufp + pos)); } else if ( ! ctl) { if (r->eqn) - return(eqn_read(&r->eqn, ln, *bufp, pos)); + return(eqn_read(&r->eqn, ln, *bufp, pos, offs)); if (r->tbl) return(tbl_read(r->tbl, ln, *bufp, pos)); - return(ROFF_CONT); + return(roff_parsetext(*bufp + pos)); } else if (r->eqn) - return(eqn_read(&r->eqn, ln, *bufp, ppos)); + return(eqn_read(&r->eqn, ln, *bufp, ppos, offs)); /* * If a scope is open, go to the child handler for that macro, @@ -552,16 +647,14 @@ roff_endparse(struct roff *r) if (r->eqn) { mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse, - r->eqn->eqn.line, r->eqn->eqn.pos, NULL); - eqn_end(r->eqn); - r->eqn = NULL; + r->eqn->eqn.ln, r->eqn->eqn.pos, NULL); + eqn_end(&r->eqn); } if (r->tbl) { mandoc_msg(MANDOCERR_SCOPEEXIT, r->parse, r->tbl->line, r->tbl->pos, NULL); - tbl_end(r->tbl); - r->tbl = NULL; + tbl_end(&r->tbl); } } @@ -576,14 +669,21 @@ roff_parse(struct roff *r, const char *buf, int *pos) size_t maclen; enum rofft t; - if ('\0' == buf[*pos] || '"' == buf[*pos]) + if ('\0' == buf[*pos] || '"' == buf[*pos] || + '\t' == buf[*pos] || ' ' == buf[*pos]) return(ROFF_MAX); + /* + * We stop the macro parse at an escape, tab, space, or nil. + * However, `\}' is also a valid macro, so make sure we don't + * clobber it by seeing the `\' as the end of token. + */ + mac = buf + *pos; - maclen = strcspn(mac, " \\\t\0"); + maclen = strcspn(mac + 1, " \\\t\0") + 1; t = (r->current_string = roff_getstrn(r, mac, maclen)) - ? ROFF_USERDEF : roff_hash_find(mac, maclen); + ? ROFF_USERDEF : roffhash_find(mac, maclen); *pos += (int)maclen; @@ -593,27 +693,6 @@ roff_parse(struct roff *r, const char *buf, int *pos) return(t); } - -static int -roff_parse_nat(const char *buf, unsigned int *res) -{ - char *ep; - long lval; - - errno = 0; - lval = strtol(buf, &ep, 10); - if (buf[0] == '\0' || *ep != '\0') - return(0); - if ((errno == ERANGE && - (lval == LONG_MAX || lval == LONG_MIN)) || - (lval > INT_MAX || lval < 0)) - return(0); - - *res = (unsigned int)lval; - return(1); -} - - /* ARGSUSED */ static enum rofferr roff_cblock(ROFF_ARGS) @@ -829,8 +908,7 @@ roff_block_sub(ROFF_ARGS) * pulling it out of the hashtable. */ - if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos))) - return(ROFF_IGN); + t = roff_parse(r, *bufp, &pos); /* * Macros other than block-end are only significant @@ -866,21 +944,44 @@ roff_cond_sub(ROFF_ARGS) { enum rofft t; enum roffrule rr; + char *ep; rr = r->last->rule; + roffnode_cleanscope(r); - /* - * Clean out scope. If we've closed ourselves, then don't - * continue. + /* + * If the macro is unknown, first check if it contains a closing + * delimiter `\}'. If it does, close out our scope and return + * the currently-scoped rule (ignore or continue). Else, drop + * into the currently-scoped rule. */ - roffnode_cleanscope(r); - if (ROFF_MAX == (t = roff_parse(r, *bufp, &pos))) { - if ('\\' == (*bufp)[pos] && '}' == (*bufp)[pos + 1]) - return(roff_ccond - (r, ROFF_ccond, bufp, szp, - ln, pos, pos + 2, offs)); + ep = &(*bufp)[pos]; + for ( ; NULL != (ep = strchr(ep, '\\')); ep++) { + ep++; + if ('}' != *ep) + continue; + + /* + * Make the \} go away. + * This is a little haphazard, as it's not quite + * clear how nroff does this. + * If we're at the end of line, then just chop + * off the \} and resize the buffer. + * If we aren't, then conver it to spaces. + */ + + if ('\0' == *(ep + 1)) { + *--ep = '\0'; + *szp -= 2; + } else + *(ep - 1) = *ep = ' '; + + roff_ccond(r, ROFF_ccond, bufp, szp, + ln, pos, pos + 2, offs); + break; + } return(ROFFRULE_DENY == rr ? ROFF_IGN : ROFF_CONT); } @@ -889,6 +990,7 @@ roff_cond_sub(ROFF_ARGS) * if they're either structurally required (such as loops and * conditionals) or a closing macro. */ + if (ROFFRULE_DENY == rr) if ( ! (ROFFMAC_STRUCT & roffs[t].flags)) if (ROFF_ccond != t) @@ -899,37 +1001,28 @@ roff_cond_sub(ROFF_ARGS) ln, ppos, pos, offs)); } - /* ARGSUSED */ static enum rofferr roff_cond_text(ROFF_ARGS) { - char *ep, *st; + char *ep; enum roffrule rr; rr = r->last->rule; + roffnode_cleanscope(r); - /* - * We display the value of the text if out current evaluation - * scope permits us to do so. - */ - - /* FIXME: use roff_ccond? */ - - st = &(*bufp)[pos]; - if (NULL == (ep = strstr(st, "\\}"))) { - roffnode_cleanscope(r); - return(ROFFRULE_DENY == rr ? ROFF_IGN : ROFF_CONT); + ep = &(*bufp)[pos]; + for ( ; NULL != (ep = strchr(ep, '\\')); ep++) { + ep++; + if ('}' != *ep) + continue; + *ep = '&'; + roff_ccond(r, ROFF_ccond, bufp, szp, + ln, pos, pos + 2, offs); } - - if (ep == st || (ep > st && '\\' != *(ep - 1))) - roffnode_pop(r); - - roffnode_cleanscope(r); return(ROFFRULE_DENY == rr ? ROFF_IGN : ROFF_CONT); } - static enum roffrule roff_evalcond(const char *v, int *pos) { @@ -1084,23 +1177,44 @@ roff_ds(ROFF_ARGS) return(ROFF_IGN); } +int +roff_regisset(const struct roff *r, enum regs reg) +{ + return(r->regs[(int)reg].set); +} + +unsigned int +roff_regget(const struct roff *r, enum regs reg) +{ + + return(r->regs[(int)reg].u); +} + +void +roff_regunset(struct roff *r, enum regs reg) +{ + + r->regs[(int)reg].set = 0; +} + /* ARGSUSED */ static enum rofferr roff_nr(ROFF_ARGS) { const char *key; char *val; - struct reg *rg; + int iv; val = *bufp + pos; key = roff_getname(r, &val, ln, pos); - rg = r->regs->regs; if (0 == strcmp(key, "nS")) { - rg[(int)REG_nS].set = 1; - if ( ! roff_parse_nat(val, &rg[(int)REG_nS].v.u)) - rg[(int)REG_nS].v.u = 0; + r->regs[(int)REG_nS].set = 1; + if ((iv = mandoc_strntoi(val, strlen(val), 10)) >= 0) + r->regs[(int)REG_nS].u = (unsigned)iv; + else + r->regs[(int)REG_nS].u = 0u; } return(ROFF_IGN); @@ -1130,9 +1244,8 @@ roff_TE(ROFF_ARGS) if (NULL == r->tbl) mandoc_msg(MANDOCERR_NOSCOPE, r->parse, ln, ppos, NULL); else - tbl_end(r->tbl); + tbl_end(&r->tbl); - r->tbl = NULL; return(ROFF_IGN); } @@ -1149,14 +1262,24 @@ roff_T_(ROFF_ARGS) return(ROFF_IGN); } -/* ARGSUSED */ -static enum rofferr -roff_EQ(ROFF_ARGS) +#if 0 +static int +roff_closeeqn(struct roff *r) { - struct eqn_node *e; + return(r->eqn && ROFF_EQN == eqn_end(&r->eqn) ? 1 : 0); +} +#endif + +static void +roff_openeqn(struct roff *r, const char *name, int line, + int offs, const char *buf) +{ + struct eqn_node *e; + int poff; + assert(NULL == r->eqn); - e = eqn_alloc(ppos, ln); + e = eqn_alloc(name, offs, line, r->parse); if (r->last_eqn) r->last_eqn->next = e; @@ -1164,6 +1287,19 @@ roff_EQ(ROFF_ARGS) r->first_eqn = r->last_eqn = e; r->eqn = r->last_eqn = e; + + if (buf) { + poff = 0; + eqn_read(&r->eqn, line, buf, offs, &poff); + } +} + +/* ARGSUSED */ +static enum rofferr +roff_EQ(ROFF_ARGS) +{ + + roff_openeqn(r, *bufp + pos, ln, ppos, NULL); return(ROFF_IGN); } @@ -1184,7 +1320,7 @@ roff_TS(ROFF_ARGS) if (r->tbl) { mandoc_msg(MANDOCERR_SCOPEBROKEN, r->parse, ln, ppos, NULL); - tbl_end(r->tbl); + tbl_end(&r->tbl); } t = tbl_alloc(ppos, ln, r->parse); @@ -1385,11 +1521,12 @@ roff_getstrn(const struct roff *r, const char *name, s { const struct roffstr *n; - n = r->first_string; - while (n && (strncmp(name, n->name, len) || '\0' != n->name[(int)len])) - n = n->next; + for (n = r->first_string; n; n = n->next) + if (0 == strncmp(name, n->name, len) && + '\0' == n->name[(int)len]) + return(n->string); - return(n ? n->string : NULL); + return(NULL); } static void @@ -1419,4 +1556,11 @@ roff_eqn(const struct roff *r) { return(r->last_eqn ? &r->last_eqn->eqn : NULL); +} + +char +roff_eqndelim(const struct roff *r) +{ + + return('\0'); }