=================================================================== RCS file: /cvs/mandoc/Attic/argv.c,v retrieving revision 1.13 retrieving revision 1.23 diff -u -p -r1.13 -r1.23 --- mandoc/Attic/argv.c 2009/01/15 15:46:45 1.13 +++ mandoc/Attic/argv.c 2009/01/20 22:55:46 1.23 @@ -1,4 +1,4 @@ -/* $Id: argv.c,v 1.13 2009/01/15 15:46:45 kristaps Exp $ */ +/* $Id: argv.c,v 1.23 2009/01/20 22:55:46 kristaps Exp $ */ /* * Copyright (c) 2008 Kristaps Dzonsons * @@ -25,9 +25,17 @@ #include "private.h" +/* + * Parse arguments and parameters of macros. Arguments follow the + * syntax of `-arg [val [valN...]]', while parameters are free-form text + * following arguments (if any). This file must correctly handle the + * strange punctuation rules dictated by groff. + */ +/* FIXME: .It called with -column and quoted arguments. */ + static int lookup(int, const char *); -static int parse(struct mdoc *, int, int, +static int parse(struct mdoc *, int, struct mdoc_arg *, int *, char *); static int parse_single(struct mdoc *, int, struct mdoc_arg *, int *, char *); @@ -36,21 +44,142 @@ static int parse_multi(struct mdoc *, int, static int postparse(struct mdoc *, int, const struct mdoc_arg *, int); +#define ARGS_QUOTED (1 << 0) +#define ARGS_DELIM (1 << 1) +#define ARGS_TABSEP (1 << 2) +static int mdoc_argflags[MDOC_MAX] = { + 0, /* \" */ + 0, /* Dd */ + 0, /* Dt */ + 0, /* Os */ + 0, /* Sh */ + 0, /* Ss */ + ARGS_DELIM, /* Pp */ + ARGS_DELIM, /* D1 */ + ARGS_DELIM, /* Dl */ + 0, /* Bd */ + 0, /* Ed */ + 0, /* Bl */ + 0, /* El */ + ARGS_DELIM, /* It */ + ARGS_DELIM, /* Ad */ + ARGS_DELIM, /* An */ + ARGS_DELIM, /* Ar */ + ARGS_QUOTED, /* Cd */ + ARGS_DELIM, /* Cm */ + ARGS_DELIM, /* Dv */ + ARGS_DELIM, /* Er */ + ARGS_DELIM, /* Ev */ + 0, /* Ex */ + ARGS_DELIM | ARGS_QUOTED, /* Fa */ + 0, /* Fd */ + ARGS_DELIM, /* Fl */ + ARGS_DELIM | ARGS_QUOTED, /* Fn */ + ARGS_DELIM | ARGS_QUOTED, /* Ft */ + ARGS_DELIM, /* Ic */ + 0, /* In */ + ARGS_DELIM, /* Li */ + 0, /* Nd */ + ARGS_DELIM, /* Nm */ + ARGS_DELIM, /* Op */ + 0, /* Ot */ + ARGS_DELIM, /* Pa */ + 0, /* Rv */ + ARGS_DELIM, /* St */ + ARGS_DELIM, /* Va */ + ARGS_DELIM, /* Vt */ + ARGS_DELIM, /* Xr */ + ARGS_QUOTED, /* %A */ + ARGS_QUOTED, /* %B */ + ARGS_QUOTED, /* %D */ + ARGS_QUOTED, /* %I */ + ARGS_QUOTED, /* %J */ + ARGS_QUOTED, /* %N */ + ARGS_QUOTED, /* %O */ + ARGS_QUOTED, /* %P */ + ARGS_QUOTED, /* %R */ + ARGS_QUOTED, /* %T */ + ARGS_QUOTED, /* %V */ + ARGS_DELIM, /* Ac */ + 0, /* Ao */ + ARGS_DELIM, /* Aq */ + ARGS_DELIM, /* At */ + ARGS_DELIM, /* Bc */ + 0, /* Bf */ + 0, /* Bo */ + ARGS_DELIM, /* Bq */ + ARGS_DELIM, /* Bsx */ + ARGS_DELIM, /* Bx */ + 0, /* Db */ + ARGS_DELIM, /* Dc */ + 0, /* Do */ + ARGS_DELIM, /* Dq */ + ARGS_DELIM, /* Ec */ + 0, /* Ef */ + ARGS_DELIM, /* Em */ + 0, /* Eo */ + ARGS_DELIM, /* Fx */ + ARGS_DELIM, /* Ms */ + ARGS_DELIM, /* No */ + ARGS_DELIM, /* Ns */ + ARGS_DELIM, /* Nx */ + ARGS_DELIM, /* Ox */ + ARGS_DELIM, /* Pc */ + ARGS_DELIM, /* Pf */ + 0, /* Po */ + ARGS_DELIM, /* Pq */ + ARGS_DELIM, /* Qc */ + ARGS_DELIM, /* Ql */ + 0, /* Qo */ + ARGS_DELIM, /* Qq */ + 0, /* Re */ + 0, /* Rs */ + ARGS_DELIM, /* Sc */ + 0, /* So */ + ARGS_DELIM, /* Sq */ + 0, /* Sm */ + ARGS_DELIM, /* Sx */ + ARGS_DELIM, /* Sy */ + ARGS_DELIM, /* Tn */ + ARGS_DELIM, /* Ux */ + ARGS_DELIM, /* Xc */ + 0, /* Xo */ + 0, /* Fo */ + 0, /* Fc */ + 0, /* Oo */ + ARGS_DELIM, /* Oc */ + 0, /* Bk */ + 0, /* Ek */ + 0, /* Bt */ + 0, /* Hf */ + 0, /* Fr */ + 0, /* Ud */ +}; + + int -mdoc_args(struct mdoc *mdoc, int line, int *pos, char *buf, int fl, char **v) +mdoc_args(struct mdoc *mdoc, int line, + int *pos, char *buf, int tok, char **v) { - int i; + int i, fl; + struct mdoc_node *n; + fl = 0 == tok ? 0 : mdoc_argflags[tok]; + if (MDOC_It == tok) { + n = mdoc->last->parent; + /* FIXME: scan for ARGS_TABSEP. */ + } + if (0 == buf[*pos]) return(ARGS_EOLN); if ('\"' == buf[*pos] && ! (fl & ARGS_QUOTED)) - if ( ! mdoc_pwarn(mdoc, line, *pos, WARN_SYNTAX_QUOTED)) + if ( ! mdoc_pwarn(mdoc, line, *pos, WARN_SYNTAX, "unexpected quoted parameter")) return(ARGS_ERROR); if ('-' == buf[*pos]) - if ( ! mdoc_pwarn(mdoc, line, *pos, WARN_SYNTAX_ARGLIKE)) + if ( ! mdoc_pwarn(mdoc, line, *pos, WARN_SYNTAX, "argument-like parameter")) return(ARGS_ERROR); if ((fl & ARGS_DELIM) && mdoc_iscdelim(buf[*pos])) { @@ -62,10 +191,10 @@ mdoc_args(struct mdoc *mdoc, int line, int *pos, char if ( ! mdoc_iscdelim(buf[i])) break; i++; - if (0 == buf[i] || ! isspace(buf[i])) + if (0 == buf[i] || ! isspace((int)buf[i])) break; i++; - while (buf[i] && isspace(buf[i])) + while (buf[i] && isspace((int)buf[i])) i++; } if (0 == buf[i]) { @@ -76,7 +205,8 @@ mdoc_args(struct mdoc *mdoc, int line, int *pos, char /* Parse routine for non-quoted string. */ - if ('\"' != buf[*pos]) { + assert(*pos > 0); + if ('\"' != buf[*pos] || ! (ARGS_QUOTED & fl)) { *v = &buf[*pos]; /* FIXME: UGLY tab-sep processing. */ @@ -96,9 +226,14 @@ mdoc_args(struct mdoc *mdoc, int line, int *pos, char } (*pos)++; } - else - while (buf[*pos] && ! isspace(buf[*pos])) + else { + while (buf[*pos]) { + if (isspace((int)buf[*pos])) + if ('\\' != buf[*pos - 1]) + break; (*pos)++; + } + } if (0 == buf[*pos]) return(ARGS_WORD); @@ -109,13 +244,13 @@ mdoc_args(struct mdoc *mdoc, int line, int *pos, char return(ARGS_WORD); if ( ! (ARGS_TABSEP & fl)) - while (buf[*pos] && isspace(buf[*pos])) + while (buf[*pos] && isspace((int)buf[*pos])) (*pos)++; if (buf[*pos]) return(ARGS_WORD); - if ( ! mdoc_pwarn(mdoc, line, *pos, WARN_SYNTAX_WS_EOLN)) + if ( ! mdoc_pwarn(mdoc, line, *pos, WARN_COMPAT, "whitespace at end-of-line")) return(ARGS_ERROR); return(ARGS_WORD); @@ -127,32 +262,30 @@ mdoc_args(struct mdoc *mdoc, int line, int *pos, char * error. After, parse to the next word. */ - assert( ! (ARGS_TABSEP & fl)); - *v = &buf[++(*pos)]; while (buf[*pos] && '\"' != buf[*pos]) (*pos)++; if (0 == buf[*pos]) { - (void)mdoc_perr(mdoc, line, *pos, ERR_SYNTAX_UNQUOTE); + (void)mdoc_perr(mdoc, line, *pos, "unterminated quoted parameter"); return(ARGS_ERROR); } buf[(*pos)++] = 0; if (0 == buf[*pos]) - return(ARGS_WORD); + return(ARGS_QWORD); - while (buf[*pos] && isspace(buf[*pos])) + while (buf[*pos] && isspace((int)buf[*pos])) (*pos)++; if (buf[*pos]) - return(ARGS_WORD); + return(ARGS_QWORD); - if ( ! mdoc_pwarn(mdoc, line, *pos, WARN_SYNTAX_WS_EOLN)) + if ( ! mdoc_pwarn(mdoc, line, *pos, WARN_COMPAT, "whitespace at end-of-line")) return(ARGS_ERROR); - return(ARGS_WORD); + return(ARGS_QWORD); } @@ -338,7 +471,7 @@ postparse(struct mdoc *mdoc, int line, const struct md break; if (xstrcmp(v->value[0], "indent-two")) break; - return(mdoc_perr(mdoc, line, pos, ERR_SYNTAX_ARGBAD)); + return(mdoc_perr(mdoc, line, pos, "invalid offset value")); default: break; } @@ -374,9 +507,10 @@ parse_multi(struct mdoc *mdoc, int line, if (0 < v->sz && v->sz < MDOC_LINEARG_MAX) return(1); - c = 0 == v->sz ? ERR_SYNTAX_ARGVAL : ERR_SYNTAX_ARGMANY; free(v->value); - return(mdoc_perr(mdoc, line, ppos, c)); + return(mdoc_perr(mdoc, line, ppos, 0 == v->sz ? + "argument requires a value" : + "too many values to argument")); } @@ -393,7 +527,7 @@ parse_single(struct mdoc *mdoc, int line, if (ARGS_ERROR == c) return(0); if (ARGS_EOLN == c) - return(mdoc_perr(mdoc, line, ppos, ERR_SYNTAX_ARGVAL)); + return(mdoc_perr(mdoc, line, ppos, "argument requires a value")); v->sz = 1; v->value = xcalloc(1, sizeof(char *)); @@ -403,7 +537,7 @@ parse_single(struct mdoc *mdoc, int line, static int -parse(struct mdoc *mdoc, int line, int tok, +parse(struct mdoc *mdoc, int line, struct mdoc_arg *v, int *pos, char *buf) { @@ -439,7 +573,7 @@ mdoc_argv(struct mdoc *mdoc, int line, int tok, if (0 == buf[*pos]) return(ARGV_EOLN); - assert( ! isspace(buf[*pos])); + assert( ! isspace((int)buf[*pos])); if ('-' != buf[*pos]) return(ARGV_WORD); @@ -450,24 +584,30 @@ mdoc_argv(struct mdoc *mdoc, int line, int tok, v->line = line; v->pos = *pos; - while (buf[*pos] && ! isspace(buf[*pos])) + assert(*pos > 0); + while (buf[*pos]) { + if (isspace((int)buf[*pos])) + if ('\\' != buf[*pos - 1]) + break; (*pos)++; + } if (buf[*pos]) buf[(*pos)++] = 0; if (MDOC_ARG_MAX == (v->arg = lookup(tok, argv))) { - (void)mdoc_pwarn(mdoc, line, i, WARN_SYNTAX_ARGLIKE); + if ( ! mdoc_pwarn(mdoc, line, i, WARN_SYNTAX, "argument-like parameter")) + return(ARGV_ERROR); return(ARGV_WORD); } - while (buf[*pos] && isspace(buf[*pos])) + while (buf[*pos] && isspace((int)buf[*pos])) (*pos)++; /* FIXME: whitespace if no value. */ ppos = *pos; - if ( ! parse(mdoc, line, tok, v, pos, buf)) + if ( ! parse(mdoc, line, v, pos, buf)) return(ARGV_ERROR); if ( ! postparse(mdoc, line, v, ppos)) return(ARGV_ERROR);