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

switch ${%foo} to wcswidth-like behaviour – slightly problematic, and

Browse Source 
the “set +U” case isn’t even handled

committed to branch because I’d like to get more input on this, for now
This commit is contained in:
tg 2009-11-28 14:21:47 +00:00
parent 7063a9a921
commit 883d9d99b3
9 changed files with 37 additions and 21 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
check.t
View File

@ -1,4 +1,4 @@
# $MirOS: src/bin/mksh/check.t,v 1.343 2009/11/22 14:25:17 tg Exp $ # $MirOS: src/bin/mksh/check.t,v 1.344 2009/11/28 14:21:41 tg Exp $
# $OpenBSD: bksl-nl.t,v 1.2 2001/01/28 23:04:56 niklas Exp $ # $OpenBSD: bksl-nl.t,v 1.2 2001/01/28 23:04:56 niklas Exp $
# $OpenBSD: history.t,v 1.5 2001/01/28 23:04:56 niklas Exp $ # $OpenBSD: history.t,v 1.5 2001/01/28 23:04:56 niklas Exp $
# $OpenBSD: read.t,v 1.3 2003/03/10 03:48:16 david Exp $ # $OpenBSD: read.t,v 1.3 2003/03/10 03:48:16 david Exp $
@ -6699,8 +6699,12 @@ description:
Check some corner cases Check some corner cases
stdin: stdin:
print % $% . print % $% .
set -U
x='a b'
print c ${%x} .
expected-stdout: expected-stdout:
% $% . % $% .
c -1 .
--- ---
name: wcswidth-3 name: wcswidth-3
description: description:

7
eval.c
View File

@ -22,7 +22,7 @@
#include "sh.h" #include "sh.h"
__RCSID("$MirOS: src/bin/mksh/eval.c,v 1.74 2009/11/22 13:49:04 tg Exp $"); __RCSID("$MirOS: src/bin/mksh/eval.c,v 1.75 2009/11/28 14:21:43 tg Exp $");
/* /*
* string expansion * string expansion
@ -949,12 +949,13 @@ varsub(Expand *xp, const char *sp, const char *word,
} else { } else {
p = str_val(global(sp)); p = str_val(global(sp));
zero_ok = p != null; zero_ok = p != null;
c = stype == '#' ? (int)utflen(p) : utf_mbswidth(p); c = stype == '#' ? (int)utflen(p) :
utf_mbswidth(p, false);
} }
if (Flag(FNOUNSET) && c == 0 && !zero_ok) if (Flag(FNOUNSET) && c == 0 && !zero_ok)
errorf("%s: parameter not set", sp); errorf("%s: parameter not set", sp);
*stypep = 0; /* unqualified variable/string substitution */ *stypep = 0; /* unqualified variable/string substitution */
xp->str = shf_smprintf("%u", (unsigned int)c); xp->str = shf_smprintf("%d", c);
return (XSUB); return (XSUB);
} }

6
exec.c
View File

@ -22,7 +22,7 @@
#include "sh.h" #include "sh.h"
__RCSID("$MirOS: src/bin/mksh/exec.c,v 1.68 2009/11/09 23:35:09 tg Exp $"); __RCSID("$MirOS: src/bin/mksh/exec.c,v 1.69 2009/11/28 14:21:43 tg Exp $");
static int comexec(struct op *, struct tbl *volatile, const char **, static int comexec(struct op *, struct tbl *volatile, const char **,
int volatile, volatile int *); int volatile, volatile int *);
@ -1380,7 +1380,7 @@ pr_menu(const char * const *ap)
i = strlen(*pp); i = strlen(*pp);
if (i > aocts) if (i > aocts)
aocts = i; aocts = i;
i = utf_mbswidth(*pp); i = utf_mbswidth(*pp, true);
if (i > acols) if (i > acols)
acols = i; acols = i;
} }
@ -1420,7 +1420,7 @@ pr_list(char * const *ap)
i = strlen(*pp); i = strlen(*pp);
if (i > aocts) if (i > aocts)
aocts = i; aocts = i;
i = utf_mbswidth(*pp); i = utf_mbswidth(*pp, true);
if (i > acols) if (i > acols)
acols = i; acols = i;
} }

19
expr.c
View File

@ -22,7 +22,7 @@
#include "sh.h" #include "sh.h"
__RCSID("$MirOS: src/bin/mksh/expr.c,v 1.37 2009/10/04 13:19:33 tg Exp $"); __RCSID("$MirOS: src/bin/mksh/expr.c,v 1.38 2009/11/28 14:21:44 tg Exp $");
/* The order of these enums is constrained by the order of opinfo[] */ /* The order of these enums is constrained by the order of opinfo[] */
enum token { enum token {
@ -683,8 +683,15 @@ utf_widthadj(const char *src, const char **dst)
return (width); return (width);
} }
/**
* In lenient mode, characters of width -1 are handled as one column
* per octet (kind of as a strlen replacement). Users of lenient mo-
* de should reconsider the code.
* In strict mode, this behaves like wcswidth(3) and returns -1 upon
* encounter of a control multibyte character.
*/
int int
utf_mbswidth(const char *s) utf_mbswidth(const char *s, bool lenient)
{ {
size_t len; size_t len;
unsigned int wc; unsigned int wc;
@ -694,10 +701,14 @@ utf_mbswidth(const char *s)
return (strlen(s)); return (strlen(s));
while (*s) while (*s)
if (((len = utf_mbtowc(&wc, s)) == (size_t)-1) || if ((len = utf_mbtowc(&wc, s)) == (size_t)-1) {
((cw = utf_wcwidth(wc)) == -1)) { by_octet:
s++; s++;
width += 1; width += 1;
} else if ((cw = utf_wcwidth(wc)) == -1) {
if (lenient)
goto by_octet;
return (-1);
} else { } else {
s += len; s += len;
width += cw; width += cw;

4
funcs.c
View File

@ -25,7 +25,7 @@
#include "sh.h" #include "sh.h"
__RCSID("$MirOS: src/bin/mksh/funcs.c,v 1.142 2009/11/09 23:35:09 tg Exp $"); __RCSID("$MirOS: src/bin/mksh/funcs.c,v 1.143 2009/11/28 14:21:44 tg Exp $");
#if HAVE_KILLPG #if HAVE_KILLPG
/* /*
@ -1551,7 +1551,7 @@ c_kill(const char **wp)
w = strlen(sigtraps[j].mess); w = strlen(sigtraps[j].mess);
if (w > mess_octs) if (w > mess_octs)
mess_octs = w; mess_octs = w;
w = utf_mbswidth(sigtraps[j].mess); w = utf_mbswidth(sigtraps[j].mess, true);
if (w > mess_cols) if (w > mess_cols)
mess_cols = w; mess_cols = w;
} }

4
misc.c
View File

@ -29,7 +29,7 @@
#include <grp.h> #include <grp.h>
#endif #endif
__RCSID("$MirOS: src/bin/mksh/misc.c,v 1.131 2009/11/23 12:48:19 tg Exp $"); __RCSID("$MirOS: src/bin/mksh/misc.c,v 1.132 2009/11/28 14:21:45 tg Exp $");
unsigned char chtypes[UCHAR_MAX + 1]; /* type bits for unsigned char */ unsigned char chtypes[UCHAR_MAX + 1]; /* type bits for unsigned char */
@ -154,7 +154,7 @@ printoptions(bool verbose)
len = strlen(options[i].name); len = strlen(options[i].name);
if (len > octs) if (len > octs)
octs = len; octs = len;
len = utf_mbswidth(options[i].name); len = utf_mbswidth(options[i].name, true);
if (len > oi.opt_width) if (len > oi.opt_width)
oi.opt_width = len; oi.opt_width = len;
} }

4
sh.h
View File

@ -134,7 +134,7 @@
#endif #endif
#ifdef EXTERN #ifdef EXTERN
__RCSID("$MirOS: src/bin/mksh/sh.h,v 1.362 2009/11/22 14:25:18 tg Exp $"); __RCSID("$MirOS: src/bin/mksh/sh.h,v 1.363 2009/11/28 14:21:45 tg Exp $");
#endif #endif
#define MKSH_VERSION "R39 2009/11/22" #define MKSH_VERSION "R39 2009/11/22"
@ -1361,7 +1361,7 @@ int v_evaluate(struct tbl *, const char *, volatile int, bool);
size_t utf_mbtowc(unsigned int *, const char *); size_t utf_mbtowc(unsigned int *, const char *);
size_t utf_wctomb(char *, unsigned int); size_t utf_wctomb(char *, unsigned int);
int utf_widthadj(const char *, const char **); int utf_widthadj(const char *, const char **);
int utf_mbswidth(const char *); int utf_mbswidth(const char *, bool);
const char *utf_skipcols(const char *, int); const char *utf_skipcols(const char *, int);
size_t utf_ptradj(const char *); size_t utf_ptradj(const char *);
int utf_wcwidth(unsigned int); int utf_wcwidth(unsigned int);

4
shf.c
View File

@ -22,7 +22,7 @@
#include "sh.h" #include "sh.h"
__RCSID("$MirOS: src/bin/mksh/shf.c,v 1.33 2009/09/20 13:33:48 tg Exp $"); __RCSID("$MirOS: src/bin/mksh/shf.c,v 1.34 2009/11/28 14:21:46 tg Exp $");
/* flags to shf_emptybuf() */ /* flags to shf_emptybuf() */
#define EB_READSW 0x01 /* about to switch to reading */ #define EB_READSW 0x01 /* about to switch to reading */
@ -932,7 +932,7 @@ shf_vfprintf(struct shf *shf, const char *fmt, va_list args)
case 's': case 's':
if (!(s = va_arg(args, const char *))) if (!(s = va_arg(args, const char *)))
s = "(null)"; s = "(null)";
len = utf_mbswidth(s); len = utf_mbswidth(s, true);
break; break;
case 'c': case 'c':

4
var.c
View File

@ -22,7 +22,7 @@
#include "sh.h" #include "sh.h"
__RCSID("$MirOS: src/bin/mksh/var.c,v 1.96 2009/10/18 12:30:06 tg Exp $"); __RCSID("$MirOS: src/bin/mksh/var.c,v 1.97 2009/11/28 14:21:47 tg Exp $");
/* /*
* Variables * Variables
@ -574,7 +574,7 @@ formatstr(struct tbl *vp, const char *s)
char *p, *q; char *p, *q;
size_t psiz; size_t psiz;
olen = utf_mbswidth(s); olen = utf_mbswidth(s, true);
if (vp->flag & (RJUST|LJUST)) { if (vp->flag & (RJUST|LJUST)) {
if (!vp->u2.field) /* default field width */ if (!vp->u2.field) /* default field width */