#include "EXTERN.h"
#include "perl.h"
#include "XSUB.h"
+#define U8 U8
+#include "encode.h"
+#include "iso8859.h"
+#include "EBCDIC.h"
+#include "Symbols.h"
#define UNIMPLEMENTED(x,y) y x (SV *sv, char *encoding) { \
Perl_croak(aTHX_ "panic_unimplemented"); \
UNIMPLEMENTED(_encoded_utf8_to_bytes, I32)
UNIMPLEMENTED(_encoded_bytes_to_utf8, I32)
-#ifdef USE_PERLIO
+#if defined(USE_PERLIO) && !defined(USE_SFIO)
/* Define an encoding "layer" in the perliol.h sense.
The layer defined here "inherits" in an object-oriented sense from the
"perlio" layer with its PerlIOBuf_* "methods".
SV * enc;
} PerlIOEncode;
+SV *
+PerlIOEncode_getarg(PerlIO *f)
+{
+ PerlIOEncode *e = PerlIOSelf(f,PerlIOEncode);
+ SV *sv = &PL_sv_undef;
+ if (e->enc)
+ {
+ dSP;
+ ENTER;
+ SAVETMPS;
+ PUSHMARK(sp);
+ XPUSHs(e->enc);
+ PUTBACK;
+ if (perl_call_method("name",G_SCALAR) == 1)
+ {
+ SPAGAIN;
+ sv = newSVsv(POPs);
+ PUTBACK;
+ }
+ }
+ return sv;
+}
IV
-PerlIOEncode_pushed(PerlIO *f, const char *mode,const char *arg,STRLEN len)
+PerlIOEncode_pushed(PerlIO *f, const char *mode, SV *arg)
{
PerlIOEncode *e = PerlIOSelf(f,PerlIOEncode);
dTHX;
dSP;
IV code;
- code = PerlIOBuf_pushed(f,mode,Nullch,0);
+ code = PerlIOBuf_pushed(f,mode,Nullsv);
ENTER;
SAVETMPS;
PUSHMARK(sp);
- XPUSHs(sv_2mortal(newSVpv("Encode",0)));
- XPUSHs(sv_2mortal(newSVpvn(arg,len)));
+ XPUSHs(arg);
PUTBACK;
- if (perl_call_method("getEncoding",G_SCALAR) != 1)
- return -1;
+ if (perl_call_pv("Encode::find_encoding",G_SCALAR) != 1)
+ {
+ /* should never happen */
+ Perl_die(aTHX_ "Encode::find_encoding did not return a value");
+ return -1;
+ }
SPAGAIN;
e->enc = POPs;
PUTBACK;
if (!SvROK(e->enc))
- return -1;
+ {
+ e->enc = Nullsv;
+ errno = EINVAL;
+ Perl_warner(aTHX_ WARN_IO, "Cannot find encoding \"%"SVf"\"", arg);
+ return -1;
+ }
SvREFCNT_inc(e->enc);
FREETMPS;
LEAVE;
e->bufsv = newSV(e->base.bufsiz);
sv_setpvn(e->bufsv,"",0);
}
- e->base.buf = SvPVX(e->bufsv);
+ e->base.buf = (STDCHAR *)SvPVX(e->bufsv);
if (!e->base.ptr)
e->base.ptr = e->base.buf;
if (!e->base.end)
{
SSize_t poff = e->base.ptr - e->base.buf;
SSize_t eoff = e->base.end - e->base.buf;
- e->base.buf = SvGROW(e->bufsv,e->base.bufsiz);
+ e->base.buf = (STDCHAR *)SvGROW(e->bufsv,e->base.bufsiz);
e->base.ptr = e->base.buf + poff;
e->base.end = e->base.buf + eoff;
}
XPUSHs(e->bufsv);
XPUSHs(&PL_sv_yes);
PUTBACK;
- if (perl_call_method("toUnicode",G_SCALAR) != 1)
+ if (perl_call_method("decode",G_SCALAR) != 1)
code = -1;
SPAGAIN;
uni = POPs;
s = SvPVutf8(uni,len);
if (s != SvPVX(e->bufsv))
{
- e->base.buf = SvGROW(e->bufsv,len);
+ e->base.buf = (STDCHAR *)SvGROW(e->bufsv,len);
Move(s,e->base.buf,len,char);
SvCUR_set(e->bufsv,len);
}
{
PerlIOEncode *e = PerlIOSelf(f,PerlIOEncode);
IV code = 0;
- dTHX;
- if (e->bufsv && (PerlIOBase(f)->flags & (PERLIO_F_RDBUF|PERLIO_F_WRBUF)))
+ if (e->bufsv && (PerlIOBase(f)->flags & (PERLIO_F_RDBUF|PERLIO_F_WRBUF))
+ &&(e->base.ptr > e->base.buf)
+ )
{
+ dTHX;
dSP;
SV *str;
char *s;
XPUSHs(e->bufsv);
XPUSHs(&PL_sv_yes);
PUTBACK;
- if (perl_call_method("fromUnicode",G_SCALAR) != 1)
+ if (perl_call_method("encode",G_SCALAR) != 1)
code = -1;
SPAGAIN;
str = POPs;
s = SvPV(str,len);
if (s != SvPVX(e->bufsv))
{
- e->base.buf = SvGROW(e->bufsv,len);
+ e->base.buf = (STDCHAR *)SvGROW(e->bufsv,len);
Move(s,e->base.buf,len,char);
SvCUR_set(e->bufsv,len);
}
Off_t
PerlIOEncode_tell(PerlIO *f)
{
+ dTHX;
PerlIOBuf *b = PerlIOSelf(f,PerlIOBuf);
/* Unfortunately the only way to get a postion is to back-translate,
the UTF8-bytes we have buf..ptr and adjust accordingly.
if ((PerlIOBase(f)->flags & PERLIO_F_RDBUF) && b->ptr < b->end)
{
Size_t count = b->end - b->ptr;
- PerlIO_push(f,&PerlIO_pending,"r",Nullch,0);
+ PerlIO_push(aTHX_ f,&PerlIO_pending,"r",Nullsv);
/* Save what we have left to read */
PerlIOSelf(f,PerlIOBuf)->bufsiz = count;
PerlIO_unread(f,b->ptr,count);
"encoding",
sizeof(PerlIOEncode),
PERLIO_K_BUFFERED,
- PerlIOBase_fileno,
- PerlIOBuf_fdopen,
- PerlIOBuf_open,
- PerlIOBuf_reopen,
PerlIOEncode_pushed,
PerlIOEncode_popped,
+ PerlIOBuf_open,
+ PerlIOEncode_getarg,
+ PerlIOBase_fileno,
PerlIOBuf_read,
PerlIOBuf_unread,
PerlIOBuf_write,
PerlIOBase_eof,
PerlIOBase_error,
PerlIOBase_clearerr,
- PerlIOBuf_setlinebuf,
+ PerlIOBase_setlinebuf,
PerlIOEncode_get_base,
PerlIOBuf_bufsiz,
PerlIOBuf_get_ptr,
PerlIOBuf_get_cnt,
PerlIOBuf_set_ptrcnt,
};
-#endif
+#endif /* encode layer */
+
+void
+Encode_Define(pTHX_ encode_t *enc)
+{
+ dSP;
+ HV *stash = gv_stashpv("Encode::XS", TRUE);
+ SV *sv = sv_bless(newRV_noinc(newSViv(PTR2IV(enc))),stash);
+ int i = 0;
+ PUSHMARK(sp);
+ XPUSHs(sv);
+ while (enc->name[i])
+ {
+ const char *name = enc->name[i++];
+ XPUSHs(sv_2mortal(newSVpvn(name,strlen(name))));
+ }
+ PUTBACK;
+ call_pv("Encode::define_encoding",G_DISCARD);
+ SvREFCNT_dec(sv);
+}
void call_failure (SV *routine, U8* done, U8* dest, U8* orig) {}
+static SV *
+encode_method(pTHX_ encode_t *enc, encpage_t *dir, SV *src, int check)
+{
+ STRLEN slen;
+ U8 *s = (U8 *) SvPV(src,slen);
+ SV *dst = sv_2mortal(newSV(2*slen+1));
+ if (slen)
+ {
+ U8 *d = (U8 *) SvGROW(dst, 2*slen+1);
+ STRLEN dlen = SvLEN(dst);
+ int code;
+ while ((code = do_encode(dir,s,&slen,d,dlen,&dlen,!check)))
+ {
+ SvCUR_set(dst,dlen);
+ SvPOK_on(dst);
+
+ if (code == ENCODE_FALLBACK)
+ break;
+
+ switch(code)
+ {
+ case ENCODE_NOSPACE:
+ {
+ STRLEN need = (slen) ? (SvLEN(dst)*SvCUR(src)/slen) : (dlen + UTF8_MAXLEN);
+ if (need <= SvLEN(dst))
+ need += UTF8_MAXLEN;
+ d = (U8 *) SvGROW(dst, need);
+ dlen = SvLEN(dst);
+ slen = SvCUR(src);
+ break;
+ }
+
+ case ENCODE_NOREP:
+ if (dir == enc->f_utf8)
+ {
+ if (!check && ckWARN_d(WARN_UTF8))
+ {
+ STRLEN clen;
+ UV ch = utf8n_to_uvuni(s+slen,(SvCUR(src)-slen),&clen,0);
+ Perl_warner(aTHX_ WARN_UTF8, "\"\\N{U+%"UVxf"}\" does not map to %s", ch, enc->name[0]);
+ /* FIXME: Skip over the character, copy in replacement and continue
+ * but that is messy so for now just fail.
+ */
+ return &PL_sv_undef;
+ }
+ else
+ {
+ return &PL_sv_undef;
+ }
+ }
+ else
+ {
+ /* UTF-8 is supposed to be "Universal" so should not happen */
+ Perl_croak(aTHX_ "%s '%.*s' does not map to UTF-8",
+ enc->name[0], (int)(SvCUR(src)-slen),s+slen);
+ }
+ break;
+
+ case ENCODE_PARTIAL:
+ if (!check && ckWARN_d(WARN_UTF8))
+ {
+ Perl_warner(aTHX_ WARN_UTF8, "Partial %s character",
+ (dir == enc->f_utf8) ? "UTF-8" : enc->name[0]);
+ }
+ return &PL_sv_undef;
+
+ default:
+ Perl_croak(aTHX_ "Unexpected code %d converting %s %s",
+ code, (dir == enc->f_utf8) ? "to" : "from",enc->name[0]);
+ return &PL_sv_undef;
+ }
+ }
+ SvCUR_set(dst,dlen);
+ SvPOK_on(dst);
+ if (check)
+ {
+ if (slen < SvCUR(src))
+ {
+ Move(s+slen,s,SvCUR(src)-slen,U8);
+ }
+ SvCUR_set(src,SvCUR(src)-slen);
+ }
+ }
+ else
+ {
+ SvCUR_set(dst,slen);
+ SvPOK_on(dst);
+ }
+ return dst;
+}
+
+MODULE = Encode PACKAGE = Encode::XS PREFIX = Method_
+
+PROTOTYPES: ENABLE
+
+void
+Method_decode(obj,src,check = FALSE)
+SV * obj
+SV * src
+bool check
+CODE:
+ {
+ encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
+ ST(0) = encode_method(aTHX_ enc, enc->t_utf8, src, check);
+ SvUTF8_on(ST(0));
+ XSRETURN(1);
+ }
+
+void
+Method_encode(obj,src,check = FALSE)
+SV * obj
+SV * src
+bool check
+CODE:
+ {
+ encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
+ sv_utf8_upgrade(src);
+ ST(0) = encode_method(aTHX_ enc, enc->f_utf8, src, check);
+ XSRETURN(1);
+ }
+
MODULE = Encode PACKAGE = Encode
PROTOTYPES: ENABLE
OUTPUT:
RETVAL
-SV *
-_chars_to_utf8(sv, from, ...)
- SV * sv
- SV * from
- CODE:
- {
- SV * check = items == 3 ? ST(2) : Nullsv;
- RETVAL = &PL_sv_undef;
- }
- OUTPUT:
- RETVAL
-
-SV *
-_utf8_to_chars(sv, to, ...)
- SV * sv
- SV * to
- CODE:
- {
- SV * check = items == 3 ? ST(2) : Nullsv;
- RETVAL = &PL_sv_undef;
- }
- OUTPUT:
- RETVAL
-
-SV *
-_utf8_to_chars_check(sv, ...)
- SV * sv
- CODE:
- {
- SV * check = items == 2 ? ST(1) : Nullsv;
- RETVAL = &PL_sv_undef;
- }
- OUTPUT:
- RETVAL
-
-SV *
-_bytes_to_chars(sv, from, ...)
- SV * sv
- SV * from
- CODE:
- {
- SV * check = items == 3 ? ST(2) : Nullsv;
- RETVAL = &PL_sv_undef;
- }
- OUTPUT:
- RETVAL
-
-SV *
-_chars_to_bytes(sv, to, ...)
- SV * sv
- SV * to
- CODE:
- {
- SV * check = items == 3 ? ST(2) : Nullsv;
- RETVAL = &PL_sv_undef;
- }
- OUTPUT:
- RETVAL
-
-SV *
-_from_to(sv, from, to, ...)
- SV * sv
- SV * from
- SV * to
- CODE:
- {
- SV * check = items == 4 ? ST(3) : Nullsv;
- RETVAL = &PL_sv_undef;
- }
- OUTPUT:
- RETVAL
-
bool
-_is_utf8(sv, ...)
- SV * sv
+is_utf8(sv, check = FALSE)
+SV * sv
+bool check
CODE:
{
- SV * check = items == 2 ? ST(1) : Nullsv;
if (SvPOK(sv)) {
- RETVAL = SvUTF8(sv) ? 1 : 0;
+ RETVAL = SvUTF8(sv) ? TRUE : FALSE;
if (RETVAL &&
- SvTRUE(check) &&
+ check &&
!is_utf8_string((U8*)SvPVX(sv), SvCUR(sv)))
RETVAL = FALSE;
} else {
RETVAL
SV *
-_on_utf8(sv)
+_utf8_on(sv)
SV * sv
CODE:
{
RETVAL
SV *
-_off_utf8(sv)
+_utf8_off(sv)
SV * sv
CODE:
{
OUTPUT:
RETVAL
-SV *
-_utf_to_utf(sv, from, to, ...)
- SV * sv
- SV * from
- SV * to
- CODE:
- {
- SV * check = items == 4 ? ST(3) : Nullsv;
- RETVAL = &PL_sv_undef;
- }
- OUTPUT:
- RETVAL
-
BOOT:
{
-#ifdef USE_PERLIO
- PerlIO_define_layer(&PerlIO_encode);
+#if defined(USE_PERLIO) && !defined(USE_SFIO)
+ PerlIO_define_layer(aTHX_ &PerlIO_encode);
#endif
+#include "iso8859.def"
+#include "EBCDIC.def"
+#include "Symbols.def"
}
projects / p5sagit/p5-mst-13.2.git / blobdiff