JehanneOS/mksh
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

remove some more of this ugliness

Browse Source
This commit is contained in:
tg
2010-01-29 09:34:31 +00:00
parent c67bed8790
commit 4eb17f8752
9 changed files with 109 additions and 84 deletions
Download Patch File Download Diff File Expand all files Collapse all files

137
lex.c
View File

@@ -22,7 +22,7 @@
#include "sh.h"
__RCSID("$MirOS: src/bin/mksh/lex.c,v 1.106 2010/01/28 20:58:32 tg Exp $");
__RCSID("$MirOS: src/bin/mksh/lex.c,v 1.107 2010/01/29 09:34:28 tg Exp $");
/*
* states while lexing word
@@ -206,7 +206,8 @@ yylex(int cf)
if (cf&ONEWORD)
state = SWORD;
else if (cf&LETEXPR) {
*wp++ = OQUOTE; /* enclose arguments in (double) quotes */
/* enclose arguments in (double) quotes */
*wp++ = OQUOTE;
state = SLETPAREN;
statep->ls_sletparen.nparen = 0;
#ifndef MKSH_SMALL
@@ -273,7 +274,7 @@ yylex(int cf)
/* FALLTHROUGH */
case SBASE:
if (c == '[' && (cf & (VARASN|ARRAYVAR))) {
*wp = EOS; /* temporary */
*wp = EOS; /* temporary */
if (is_wdvarname(Xstring(ws, wp), false)) {
char *p, *tmp;
@@ -302,7 +303,7 @@ yylex(int cf)
*wp++ = c;
break;
}
/* FALLTHRU */
/* FALLTHROUGH */
Sbase1: /* includes *(...|...) pattern (*+?@!) */
if (c == '*' || c == '@' || c == '+' || c == '?' ||
c == '!') {
@@ -315,13 +316,16 @@ yylex(int cf)
}
ungetsc(c2);
}
/* FALLTHRU */
/* FALLTHROUGH */
Sbase2: /* doesn't include *(...|...) pattern (*+?@!) */
switch (c) {
case '\\':
c = getsc();
if (c) /* trailing \ is lost */
*wp++ = QCHAR, *wp++ = c;
if (c) {
/* trailing \ is lost */
*wp++ = QCHAR;
*wp++ = c;
}
break;
case '\'':
*wp++ = OQUOTE;
@@ -348,14 +352,18 @@ yylex(int cf)
/* FALLTHROUGH */
case '\\':
case '$': case '`':
*wp++ = QCHAR, *wp++ = c;
*wp++ = QCHAR;
*wp++ = c;
break;
default:
heredocquote:
Xcheck(ws, wp);
if (c) { /* trailing \ is lost */
*wp++ = CHAR, *wp++ = '\\';
*wp++ = CHAR, *wp++ = c;
if (c) {
/* trailing \ is lost */
*wp++ = CHAR;
*wp++ = '\\';
*wp++ = CHAR;
*wp++ = c;
}
break;
}
@@ -384,7 +392,8 @@ yylex(int cf)
c = getsc();
/* allow :# and :% (ksh88 compat) */
if (c == ':') {
*wp++ = CHAR, *wp++ = c;
*wp++ = CHAR;
*wp++ = c;
c = getsc();
if (c == ':') {
*wp++ = CHAR;
@@ -416,7 +425,8 @@ yylex(int cf)
break;
}
} else if (c == '/') {
*wp++ = CHAR, *wp++ = c;
*wp++ = CHAR;
*wp++ = c;
if ((c = getsc()) == '/') {
*wp++ = ADELIM;
*wp++ = c;
@@ -468,7 +478,8 @@ yylex(int cf)
statep->ls_sequote.got_NUL = false;
break;
} else {
*wp++ = CHAR, *wp++ = '$';
*wp++ = CHAR;
*wp++ = '$';
ungetsc(c);
}
break;
@@ -518,9 +529,10 @@ yylex(int cf)
*wp++ = getsc();
break;
}
/* FALLTHRU */
/* FALLTHROUGH */
default:
*wp++ = CHAR, *wp++ = c;
*wp++ = CHAR;
*wp++ = c;
}
break;
@@ -537,9 +549,10 @@ yylex(int cf)
if (!statep->ls_sequote.got_NUL) {
char ts[4];
if ((unsigned int)c2 < 0x100)
*wp++ = QCHAR, *wp++ = c2;
else {
if ((unsigned int)c2 < 0x100) {
*wp++ = QCHAR;
*wp++ = c2;
} else {
c = utf_wctomb(ts, c2 - 0x100);
ts[c] = 0;
for (c = 0; ts[c]; ++c) {
@@ -548,8 +561,10 @@ yylex(int cf)
}
}
}
} else if (!statep->ls_sequote.got_NUL)
*wp++ = QCHAR, *wp++ = c;
} else if (!statep->ls_sequote.got_NUL) {
*wp++ = QCHAR;
*wp++ = c;
}
break;
case SSQUOTE:
@@ -557,8 +572,10 @@ yylex(int cf)
POP_STATE();
*wp++ = CQUOTE;
ignore_backslash_newline--;
} else
*wp++ = QCHAR, *wp++ = c;
} else {
*wp++ = QCHAR;
*wp++ = c;
}
break;
case SDQUOTE:
@@ -569,14 +586,14 @@ yylex(int cf)
goto Subst;
break;
case SCSPAREN: /* $( ... ) */
case SCSPAREN: /* $( ... ) */
/* todo: deal with $(...) quoting properly
* kludge to partly fake quoting inside $(...): doesn't
* really work because nested $(...) or ${...} inside
* double quotes aren't dealt with.
*/
switch (statep->ls_scsparen.csstate) {
case 0: /* normal */
case 0: /* normal */
switch (c) {
case '(':
statep->ls_scsparen.nparen++;
@@ -597,19 +614,19 @@ yylex(int cf)
}
break;
case 1: /* backslash in normal mode */
case 3: /* backslash in double quotes */
case 1: /* backslash in normal mode */
case 3: /* backslash in double quotes */
--statep->ls_scsparen.csstate;
break;
case 2: /* double quotes */
case 2: /* double quotes */
if (c == '"')
statep->ls_scsparen.csstate = 0;
else if (c == '\\')
statep->ls_scsparen.csstate = 3;
break;
case 4: /* single quotes */
case 4: /* single quotes */
if (c == '\'') {
statep->ls_scsparen.csstate = 0;
ignore_backslash_newline--;
@@ -618,12 +635,12 @@ yylex(int cf)
}
if (statep->ls_scsparen.nparen == 0) {
POP_STATE();
*wp++ = 0; /* end of COMSUB */
*wp++ = 0; /* end of COMSUB */
} else
*wp++ = c;
break;
case SASPAREN: /* $(( ... )) */
case SASPAREN: /* $(( ... )) */
/* XXX should nest using existing state machine
* (embed "...", $(...), etc.) */
if (c == '(')
@@ -634,7 +651,8 @@ yylex(int cf)
/*(*/
if ((c2 = getsc()) == ')') {
POP_STATE();
*wp++ = 0; /* end of EXPRSUB */
/* end of EXPRSUB */
*wp++ = 0;
break;
} else {
char *s;
@@ -702,9 +720,10 @@ yylex(int cf)
*wp++ = c;
break;
}
/* FALLTHRU */
/* FALLTHROUGH */
default:
if (c) { /* trailing \ is lost */
if (c) {
/* trailing \ is lost */
*wp++ = '\\';
*wp++ = c;
}
@@ -760,14 +779,16 @@ yylex(int cf)
c = 0;
goto Done;
}
*wp++ = CHAR, *wp++ = c;
*wp++ = CHAR;
*wp++ = c;
break;
#endif
case SHERESTRING: /* <<< delimiter */
if (c == '\\') {
c = getsc();
if (c) { /* trailing \ is lost */
if (c) {
/* trailing \ is lost */
*wp++ = QCHAR;
*wp++ = c;
}
@@ -810,7 +831,8 @@ yylex(int cf)
*/
if (c == '\\') {
c = getsc();
if (c) { /* trailing \ is lost */
if (c) {
/* trailing \ is lost */
*wp++ = QCHAR;
*wp++ = c;
}
@@ -851,7 +873,8 @@ yylex(int cf)
case '$': case '`':
break;
default:
if (c) { /* trailing \ lost */
if (c) {
/* trailing \ lost */
*wp++ = CHAR;
*wp++ = '\\';
}
@@ -1289,7 +1312,7 @@ getsc__(void)
continue;
case SREREAD:
if (s->start != s->ugbuf) /* yuck */
if (s->start != s->ugbuf) /* yuck */
afree(s->u.freeme, ATEMP);
source = s = s->next;
continue;
@@ -1365,9 +1388,9 @@ getsc_line(Source *s)
if (!p || (xp = p, xp[-1] == '\n'))
break;
/* double buffer size */
xp++; /* move past null so doubling works... */
xp++; /* move past NUL so doubling works... */
XcheckN(s->xs, xp, Xlength(s->xs, xp));
xp--; /* ...and move back again */
xp--; /* ...and move back again */
}
/* flush any unwanted input so other programs/builtins
* can read it. Not very optimal, but less error prone
@@ -1409,7 +1432,8 @@ getsc_line(Source *s)
s->start = s->str = cp;
strip_nuls(Xstring(s->xs, xp), Xlength(s->xs, xp));
/* Note: if input is all nulls, this is not eof */
if (Xlength(s->xs, xp) == 0) { /* EOF */
if (Xlength(s->xs, xp) == 0) {
/* EOF */
if (s->type == SFILE)
shf_fdclose(s->u.shf);
s->str = NULL;
@@ -1435,7 +1459,7 @@ set_prompt(int to, Source *s)
cur_prompt = to;
switch (to) {
case PS1: /* command */
case PS1: /* command */
/* Substitute ! and !! here, before substitutions are done
* so ! in expanded variables are not expanded.
* NOTE: this is not what AT&T ksh does (it does it after
@@ -1473,7 +1497,7 @@ set_prompt(int to, Source *s)
quitenv(NULL);
}
break;
case PS2: /* command continuation */
case PS2: /* command continuation */
prompt = str_val(global("PS2"));
break;
}
@@ -1548,7 +1572,7 @@ get_brace_var(XString *wsp, char *wp)
{
enum parse_state {
PS_INITIAL, PS_SAW_HASH, PS_IDENT,
PS_NUMBER, PS_VAR1, PS_END
PS_NUMBER, PS_VAR1
} state;
char c;
@@ -1562,7 +1586,7 @@ get_brace_var(XString *wsp, char *wp)
state = PS_SAW_HASH;
break;
}
/* FALLTHRU */
/* FALLTHROUGH */
case PS_SAW_HASH:
if (ksh_isalphx(c))
state = PS_IDENT;
@@ -1571,11 +1595,10 @@ get_brace_var(XString *wsp, char *wp)
else if (ctype(c, C_VAR1))
state = PS_VAR1;
else
state = PS_END;
goto out;
break;
case PS_IDENT:
if (!ksh_isalnux(c)) {
state = PS_END;
if (c == '[') {
char *tmp, *p;
@@ -1587,28 +1610,24 @@ get_brace_var(XString *wsp, char *wp)
*wp++ = *p++;
}
afree(tmp, ATEMP);
c = getsc(); /* the ] */
c = getsc(); /* the ] */
}
goto out;
}
break;
case PS_NUMBER:
if (!ksh_isdigit(c))
state = PS_END;
goto out;
break;
case PS_VAR1:
state = PS_END;
break;
case PS_END: /* keep gcc happy */
break;
}
if (state == PS_END) {
*wp++ = '\0'; /* end of variable part */
ungetsc(c);
break;
goto out;
}
Xcheck(*wsp, wp);
*wp++ = c;
}
out:
*wp++ = '\0'; /* end of variable part */
ungetsc(c);
return (wp);
}