=================================================================== RCS file: /cvs/mandoc/mdoc_argv.c,v retrieving revision 1.66 retrieving revision 1.75 diff -u -p -r1.66 -r1.75 --- mandoc/mdoc_argv.c 2011/03/17 09:16:38 1.66 +++ mandoc/mdoc_argv.c 2011/04/17 09:13:01 1.75 @@ -1,4 +1,4 @@ -/* $Id: mdoc_argv.c,v 1.66 2011/03/17 09:16:38 kristaps Exp $ */ +/* $Id: mdoc_argv.c,v 1.75 2011/04/17 09:13:01 kristaps Exp $ */ /* * Copyright (c) 2008, 2009, 2010 Kristaps Dzonsons * @@ -26,29 +26,18 @@ #include #include +#include "mdoc.h" #include "mandoc.h" #include "libmdoc.h" #include "libmandoc.h" -/* - * Routines to parse arguments of macros. Arguments follow the syntax - * of `-arg [val [valN...]]'. Arguments come in all types: quoted - * arguments, multiple arguments per value, no-value arguments, etc. - * - * There's no limit to the number or arguments that may be allocated. - */ +#define MULTI_STEP 5 /* pre-allocate argument values */ -#define ARGV_NONE (1 << 0) -#define ARGV_SINGLE (1 << 1) -#define ARGV_MULTI (1 << 2) -#define ARGV_OPT_SINGLE (1 << 3) - -#define MULTI_STEP 5 - static enum mdocargt argv_a2arg(enum mdoct, const char *); static enum margserr args(struct mdoc *, int, int *, char *, int, char **); -static int args_checkpunct(const char *); +static int args_checkpunct(struct mdoc *, + const char *, int, int, int); static int argv(struct mdoc *, int, struct mdoc_argv *, int *, char *); static int argv_single(struct mdoc *, int, @@ -57,10 +46,16 @@ static int argv_opt_single(struct mdoc *, int, struct mdoc_argv *, int *, char *); static int argv_multi(struct mdoc *, int, struct mdoc_argv *, int *, char *); +static void argn_free(struct mdoc_arg *, int); -/* Per-argument flags. */ +enum argvflag { + ARGV_NONE, /* no args to flag (e.g., -split) */ + ARGV_SINGLE, /* one arg to flag (e.g., -file xxx) */ + ARGV_MULTI, /* multiple args (e.g., -column xxx yyy) */ + ARGV_OPT_SINGLE /* optional arg (e.g., -offset [xxx]) */ +}; -static int mdoc_argvflags[MDOC_ARG_MAX] = { +static const enum argvflag argvflags[MDOC_ARG_MAX] = { ARGV_NONE, /* MDOC_Split */ ARGV_NONE, /* MDOC_Nosplit */ ARGV_NONE, /* MDOC_Ragged */ @@ -89,7 +84,7 @@ static int mdoc_argvflags[MDOC_ARG_MAX] = { ARGV_NONE /* MDOC_Symbolic */ }; -static int mdoc_argflags[MDOC_MAX] = { +static const int argflags[MDOC_MAX] = { 0, /* Ap */ 0, /* Dd */ 0, /* Dt */ @@ -214,7 +209,60 @@ static int mdoc_argflags[MDOC_MAX] = { 0, /* Ta */ }; +static const enum mdocargt args_Ex[] = { + MDOC_Std, + MDOC_ARG_MAX +}; +static const enum mdocargt args_An[] = { + MDOC_Split, + MDOC_Nosplit, + MDOC_ARG_MAX +}; + +static const enum mdocargt args_Bd[] = { + MDOC_Ragged, + MDOC_Unfilled, + MDOC_Filled, + MDOC_Literal, + MDOC_File, + MDOC_Offset, + MDOC_Compact, + MDOC_Centred, + MDOC_ARG_MAX +}; + +static const enum mdocargt args_Bf[] = { + MDOC_Emphasis, + MDOC_Literal, + MDOC_Symbolic, + MDOC_ARG_MAX +}; + +static const enum mdocargt args_Bk[] = { + MDOC_Words, + MDOC_ARG_MAX +}; + +static const enum mdocargt args_Bl[] = { + MDOC_Bullet, + MDOC_Dash, + MDOC_Hyphen, + MDOC_Item, + MDOC_Enum, + MDOC_Tag, + MDOC_Diag, + MDOC_Hang, + MDOC_Ohang, + MDOC_Inset, + MDOC_Column, + MDOC_Width, + MDOC_Offset, + MDOC_Compact, + MDOC_Nested, + MDOC_ARG_MAX +}; + /* * Parse an argument from line text. This comes in the form of -key * [value0...], which may either have a single mandatory value, at least @@ -255,7 +303,7 @@ mdoc_argv(struct mdoc *m, int line, enum mdoct tok, buf[(*pos)++] = '\0'; } - (void)memset(&tmp, 0, sizeof(struct mdoc_argv)); + memset(&tmp, 0, sizeof(struct mdoc_argv)); tmp.line = line; tmp.pos = *pos; @@ -281,13 +329,12 @@ mdoc_argv(struct mdoc *m, int line, enum mdoct tok, arg->argv = mandoc_realloc (arg->argv, arg->argc * sizeof(struct mdoc_argv)); - (void)memcpy(&arg->argv[(int)arg->argc - 1], + memcpy(&arg->argv[(int)arg->argc - 1], &tmp, sizeof(struct mdoc_argv)); return(ARGV_ARG); } - void mdoc_argv_free(struct mdoc_arg *p) { @@ -304,15 +351,14 @@ mdoc_argv_free(struct mdoc_arg *p) assert(p->argc); for (i = (int)p->argc - 1; i >= 0; i--) - mdoc_argn_free(p, i); + argn_free(p, i); free(p->argv); free(p); } - -void -mdoc_argn_free(struct mdoc_arg *p, int iarg) +static void +argn_free(struct mdoc_arg *p, int iarg) { struct mdoc_argv *arg; int j; @@ -329,7 +375,6 @@ mdoc_argn_free(struct mdoc_arg *p, int iarg) p->argv[iarg] = p->argv[iarg+1]; } - enum margserr mdoc_zargs(struct mdoc *m, int line, int *pos, char *buf, int flags, char **v) @@ -338,7 +383,6 @@ mdoc_zargs(struct mdoc *m, int line, int *pos, return(args(m, line, pos, buf, flags, v)); } - enum margserr mdoc_args(struct mdoc *m, int line, int *pos, char *buf, enum mdoct tok, char **v) @@ -346,7 +390,7 @@ mdoc_args(struct mdoc *m, int line, int *pos, int fl; struct mdoc_node *n; - fl = mdoc_argflags[tok]; + fl = argflags[tok]; if (MDOC_It != tok) return(args(m, line, pos, buf, fl, v)); @@ -370,12 +414,10 @@ mdoc_args(struct mdoc *m, int line, int *pos, return(args(m, line, pos, buf, fl, v)); } - static enum margserr args(struct mdoc *m, int line, int *pos, char *buf, int fl, char **v) { - int i; char *p, *pp; enum margserr rc; @@ -405,8 +447,7 @@ args(struct mdoc *m, int line, int *pos, * is unterminated. */ if (MDOC_PHRASELIT & m->flags) - if ( ! mdoc_pmsg(m, line, *pos, MANDOCERR_BADQUOTE)) - return(ARGS_ERROR); + mdoc_pmsg(m, line, *pos, MANDOCERR_BADQUOTE); m->flags &= ~MDOC_PHRASELIT; return(ARGS_EOLN); @@ -414,17 +455,8 @@ args(struct mdoc *m, int line, int *pos, *v = &buf[*pos]; - if (ARGS_DELIM & fl && args_checkpunct(&buf[*pos])) { - i = strlen(&buf[*pos]) + *pos; - if (i && ' ' != buf[i - 1]) - return(ARGS_PUNCT); - if (ARGS_NOWARN & fl) - return(ARGS_PUNCT); - /* FIXME: remove conditional messages... */ - if ( ! mdoc_pmsg(m, line, *pos, MANDOCERR_EOLNSPACE)) - return(ARGS_ERROR); + if (ARGS_DELIM & fl && args_checkpunct(m, buf, *pos, line, fl)) return(ARGS_PUNCT); - } /* * First handle TABSEP items, restricted to `Bl -column'. This @@ -473,8 +505,7 @@ args(struct mdoc *m, int line, int *pos, /* Whitespace check for eoln case... */ if ('\0' == *p && ' ' == *(p - 1) && ! (ARGS_NOWARN & fl)) - if ( ! mdoc_pmsg(m, line, *pos, MANDOCERR_EOLNSPACE)) - return(ARGS_ERROR); + mdoc_pmsg(m, line, *pos, MANDOCERR_EOLNSPACE); *pos += (int)(p - *v); @@ -518,8 +549,7 @@ args(struct mdoc *m, int line, int *pos, if ('\0' == buf[*pos]) { if (ARGS_NOWARN & fl || MDOC_PPHRASE & m->flags) return(ARGS_QWORD); - if ( ! mdoc_pmsg(m, line, *pos, MANDOCERR_BADQUOTE)) - return(ARGS_ERROR); + mdoc_pmsg(m, line, *pos, MANDOCERR_BADQUOTE); return(ARGS_QWORD); } @@ -533,33 +563,15 @@ args(struct mdoc *m, int line, int *pos, (*pos)++; if (0 == buf[*pos] && ! (ARGS_NOWARN & fl)) - if ( ! mdoc_pmsg(m, line, *pos, MANDOCERR_EOLNSPACE)) - return(ARGS_ERROR); + mdoc_pmsg(m, line, *pos, MANDOCERR_EOLNSPACE); return(ARGS_QWORD); } - /* - * A non-quoted term progresses until either the end of line or - * a non-escaped whitespace. - */ + p = &buf[*pos]; + *v = mandoc_getarg(m->parse, &p, line, + ! (ARGS_NOWARN & fl), pos); - for ( ; buf[*pos]; (*pos)++) - if (*pos && ' ' == buf[*pos] && '\\' != buf[*pos - 1]) - break; - - if ('\0' == buf[*pos]) - return(ARGS_WORD); - - buf[(*pos)++] = '\0'; - - while (' ' == buf[*pos]) - (*pos)++; - - if ('\0' == buf[*pos] && ! (ARGS_NOWARN & fl)) - if ( ! mdoc_pmsg(m, line, *pos, MANDOCERR_EOLNSPACE)) - return(ARGS_ERROR); - return(ARGS_WORD); } @@ -570,49 +582,50 @@ args(struct mdoc *m, int line, int *pos, * whitespace may separate these tokens. */ static int -args_checkpunct(const char *p) +args_checkpunct(struct mdoc *m, const char *buf, int i, int ln, int fl) { - int i, j; - char buf[DELIMSZ]; + int j; + char dbuf[DELIMSZ]; enum mdelim d; - i = 0; - /* First token must be a close-delimiter. */ - for (j = 0; p[i] && ' ' != p[i] && j < DELIMSZ; j++, i++) - buf[j] = p[i]; + for (j = 0; buf[i] && ' ' != buf[i] && j < DELIMSZ; j++, i++) + dbuf[j] = buf[i]; if (DELIMSZ == j) return(0); - buf[j] = '\0'; - if (DELIM_CLOSE != mandoc_isdelim(buf)) + dbuf[j] = '\0'; + if (DELIM_CLOSE != mdoc_isdelim(dbuf)) return(0); - while (' ' == p[i]) + while (' ' == buf[i]) i++; /* Remaining must NOT be open/none. */ - while (p[i]) { + while (buf[i]) { j = 0; - while (p[i] && ' ' != p[i] && j < DELIMSZ) - buf[j++] = p[i++]; + while (buf[i] && ' ' != buf[i] && j < DELIMSZ) + dbuf[j++] = buf[i++]; if (DELIMSZ == j) return(0); - buf[j] = '\0'; - d = mandoc_isdelim(buf); + dbuf[j] = '\0'; + d = mdoc_isdelim(dbuf); if (DELIM_NONE == d || DELIM_OPEN == d) return(0); - while (' ' == p[i]) + while (' ' == buf[i]) i++; } - return('\0' == p[i]); + if ( ! (ARGS_NOWARN & fl) && i && ' ' == buf[i - 1]) + mdoc_pmsg(m, ln, i - 1, MANDOCERR_EOLNSPACE); + + return('\0' == buf[i]); } /* @@ -623,68 +636,44 @@ args_checkpunct(const char *p) static enum mdocargt argv_a2arg(enum mdoct tok, const char *p) { - enum mdocargt args[MDOC_ARG_MAX]; - int i, len; + const enum mdocargt *argsp; - len = 0; + argsp = NULL; switch (tok) { case (MDOC_An): - args[len++] = MDOC_Split; - args[len++] = MDOC_Nosplit; + argsp = args_An; break; case (MDOC_Bd): - args[len++] = MDOC_Ragged; - args[len++] = MDOC_Unfilled; - args[len++] = MDOC_Filled; - args[len++] = MDOC_Literal; - args[len++] = MDOC_File; - args[len++] = MDOC_Offset; - args[len++] = MDOC_Compact; - args[len++] = MDOC_Centred; + argsp = args_Bd; break; case (MDOC_Bf): - args[len++] = MDOC_Emphasis; - args[len++] = MDOC_Literal; - args[len++] = MDOC_Symbolic; + argsp = args_Bf; break; case (MDOC_Bk): - args[len++] = MDOC_Words; + argsp = args_Bk; break; case (MDOC_Bl): - args[len++] = MDOC_Bullet; - args[len++] = MDOC_Dash; - args[len++] = MDOC_Hyphen; - args[len++] = MDOC_Item; - args[len++] = MDOC_Enum; - args[len++] = MDOC_Tag; - args[len++] = MDOC_Diag; - args[len++] = MDOC_Hang; - args[len++] = MDOC_Ohang; - args[len++] = MDOC_Inset; - args[len++] = MDOC_Column; - args[len++] = MDOC_Width; - args[len++] = MDOC_Offset; - args[len++] = MDOC_Compact; - args[len++] = MDOC_Nested; + argsp = args_Bl; break; case (MDOC_Rv): /* FALLTHROUGH */ case (MDOC_Ex): - args[len++] = MDOC_Std; + argsp = args_Ex; break; default: - break; + return(MDOC_ARG_MAX); } - for (i = 0; i < len; i++) - if (0 == strcmp(p, mdoc_argnames[args[i]])) - return(args[i]); + assert(argsp); + for ( ; MDOC_ARG_MAX != *argsp ; argsp++) + if (0 == strcmp(p, mdoc_argnames[*argsp])) + return(*argsp); + return(MDOC_ARG_MAX); } - static int argv_multi(struct mdoc *m, int line, struct mdoc_argv *v, int *pos, char *buf) @@ -711,7 +700,6 @@ argv_multi(struct mdoc *m, int line, return(1); } - static int argv_opt_single(struct mdoc *m, int line, struct mdoc_argv *v, int *pos, char *buf) @@ -735,7 +723,6 @@ argv_opt_single(struct mdoc *m, int line, return(1); } - /* * Parse a single, mandatory value from the stream. */ @@ -763,7 +750,6 @@ argv_single(struct mdoc *m, int line, return(1); } - /* * Determine rules for parsing arguments. Arguments can either accept * no parameters, an optional single parameter, one parameter, or @@ -777,16 +763,18 @@ argv(struct mdoc *mdoc, int line, v->sz = 0; v->value = NULL; - switch (mdoc_argvflags[v->arg]) { + switch (argvflags[v->arg]) { case (ARGV_SINGLE): return(argv_single(mdoc, line, v, pos, buf)); case (ARGV_MULTI): return(argv_multi(mdoc, line, v, pos, buf)); case (ARGV_OPT_SINGLE): return(argv_opt_single(mdoc, line, v, pos, buf)); - default: - /* ARGV_NONE */ + case (ARGV_NONE): break; + default: + abort(); + /* NOTREACHED */ } return(1);