10 #define UNIMPLEMENTED(x,y) y x (SV *sv, char *encoding) { \
11 Perl_croak(aTHX_ "panic_unimplemented"); \
12 return (y)0; /* fool picky compilers */ \
14 UNIMPLEMENTED(_encoded_utf8_to_bytes, I32)
15 UNIMPLEMENTED(_encoded_bytes_to_utf8, I32)
18 /* Define an encoding "layer" in the perliol.h sense.
19 The layer defined here "inherits" in an object-oriented sense from the
20 "perlio" layer with its PerlIOBuf_* "methods".
21 The implementation is particularly efficient as until Encode settles down
22 there is no point in tryint to tune it.
24 The layer works by overloading the "fill" and "flush" methods.
26 "fill" calls "SUPER::fill" in perl terms, then calls the encode OO perl API
27 to convert the encoded data to UTF-8 form, then copies it back to the
28 buffer. The "base class's" read methods then see the UTF-8 data.
30 "flush" transforms the UTF-8 data deposited by the "base class's write
31 method in the buffer back into the encoded form using the encode OO perl API,
32 then copies data back into the buffer and calls "SUPER::flush.
34 Note that "flush" is _also_ called for read mode - we still do the (back)-translate
35 so that the the base class's "flush" sees the correct number of encoded chars
36 for positioning the seek pointer. (This double translation is the worst performance
37 issue - particularly with all-perl encode engine.)
46 PerlIOBuf base; /* PerlIOBuf stuff */
53 PerlIOEncode_pushed(PerlIO *f, const char *mode,const char *arg,STRLEN len)
55 PerlIOEncode *e = PerlIOSelf(f,PerlIOEncode);
59 code = PerlIOBuf_pushed(f,mode,Nullch,0);
63 XPUSHs(sv_2mortal(newSVpv("Encode",0)));
64 XPUSHs(sv_2mortal(newSVpvn(arg,len)));
66 if (perl_call_method("getEncoding",G_SCALAR) != 1)
68 /* should never happen */
69 Perl_die(aTHX_ "Encode::getEncoding did not return a value");
79 Perl_warner(aTHX_ WARN_IO, "Cannot find encoding \"%.*s\"", (int) len, arg);
85 PerlIOBase(f)->flags |= PERLIO_F_UTF8;
90 PerlIOEncode_popped(PerlIO *f)
92 PerlIOEncode *e = PerlIOSelf(f,PerlIOEncode);
101 SvREFCNT_dec(e->bufsv);
108 PerlIOEncode_get_base(PerlIO *f)
110 PerlIOEncode *e = PerlIOSelf(f,PerlIOEncode);
113 e->base.bufsiz = 1024;
116 e->bufsv = newSV(e->base.bufsiz);
117 sv_setpvn(e->bufsv,"",0);
119 e->base.buf = (STDCHAR *)SvPVX(e->bufsv);
121 e->base.ptr = e->base.buf;
123 e->base.end = e->base.buf;
124 if (e->base.ptr < e->base.buf || e->base.ptr > e->base.buf+SvLEN(e->bufsv))
126 Perl_warn(aTHX_ " ptr %p(%p)%p",
127 e->base.buf,e->base.ptr,e->base.buf+SvLEN(e->bufsv));
130 if (SvLEN(e->bufsv) < e->base.bufsiz)
132 SSize_t poff = e->base.ptr - e->base.buf;
133 SSize_t eoff = e->base.end - e->base.buf;
134 e->base.buf = (STDCHAR *)SvGROW(e->bufsv,e->base.bufsiz);
135 e->base.ptr = e->base.buf + poff;
136 e->base.end = e->base.buf + eoff;
138 if (e->base.ptr < e->base.buf || e->base.ptr > e->base.buf+SvLEN(e->bufsv))
140 Perl_warn(aTHX_ " ptr %p(%p)%p",
141 e->base.buf,e->base.ptr,e->base.buf+SvLEN(e->bufsv));
148 PerlIOEncode_fill(PerlIO *f)
150 PerlIOEncode *e = PerlIOSelf(f,PerlIOEncode);
154 code = PerlIOBuf_fill(f);
160 /* Set SV that is the buffer to be buf..ptr */
161 SvCUR_set(e->bufsv, e->base.end - e->base.buf);
162 SvUTF8_off(e->bufsv);
170 if (perl_call_method("toUnicode",G_SCALAR) != 1)
175 /* Now get translated string (forced to UTF-8) and copy back to buffer
176 don't use sv_setsv as that may "steal" PV from returned temp
177 and so free() our known-large-enough buffer.
178 sv_setpvn() should do but let us do it long hand.
180 s = SvPVutf8(uni,len);
181 if (s != SvPVX(e->bufsv))
183 e->base.buf = (STDCHAR *)SvGROW(e->bufsv,len);
184 Move(s,e->base.buf,len,char);
185 SvCUR_set(e->bufsv,len);
188 e->base.end = e->base.buf+len;
189 e->base.ptr = e->base.buf;
197 PerlIOEncode_flush(PerlIO *f)
199 PerlIOEncode *e = PerlIOSelf(f,PerlIOEncode);
202 if (e->bufsv && (PerlIOBase(f)->flags & (PERLIO_F_RDBUF|PERLIO_F_WRBUF)))
209 if (PerlIOBase(f)->flags & PERLIO_F_RDBUF)
211 /* This is really just a flag to see if we took all the data, if
212 we did PerlIOBase_flush avoids a seek to lower layer.
213 Need to revisit if we start getting clever with unreads or seeks-in-buffer
215 left = e->base.end - e->base.ptr;
221 SvCUR_set(e->bufsv, e->base.ptr - e->base.buf);
226 if (perl_call_method("fromUnicode",G_SCALAR) != 1)
232 if (s != SvPVX(e->bufsv))
234 e->base.buf = (STDCHAR *)SvGROW(e->bufsv,len);
235 Move(s,e->base.buf,len,char);
236 SvCUR_set(e->bufsv,len);
238 SvUTF8_off(e->bufsv);
239 e->base.ptr = e->base.buf+len;
240 /* restore end != ptr as inequality is used by PerlIOBuf_flush in read case */
241 e->base.end = e->base.ptr + left;
244 if (PerlIOBuf_flush(f) != 0)
251 PerlIOEncode_close(PerlIO *f)
253 PerlIOEncode *e = PerlIOSelf(f,PerlIOEncode);
254 IV code = PerlIOBase_close(f);
258 SvREFCNT_dec(e->bufsv);
264 PerlIOBase(f)->flags &= ~(PERLIO_F_RDBUF|PERLIO_F_WRBUF);
269 PerlIOEncode_tell(PerlIO *f)
271 PerlIOBuf *b = PerlIOSelf(f,PerlIOBuf);
272 /* Unfortunately the only way to get a postion is to back-translate,
273 the UTF8-bytes we have buf..ptr and adjust accordingly.
274 But we will try and save any unread data in case stream
277 if ((PerlIOBase(f)->flags & PERLIO_F_RDBUF) && b->ptr < b->end)
279 Size_t count = b->end - b->ptr;
280 PerlIO_push(f,&PerlIO_pending,"r",Nullch,0);
281 /* Save what we have left to read */
282 PerlIOSelf(f,PerlIOBuf)->bufsiz = count;
283 PerlIO_unread(f,b->ptr,count);
284 /* There isn't any unread data - we just saved it - so avoid the lower seek */
286 /* Flush ourselves - now one layer down,
287 this does the back translate and adjusts position
289 PerlIO_flush(PerlIONext(f));
290 /* Set position of the saved data */
291 PerlIOSelf(f,PerlIOBuf)->posn = b->posn;
300 PerlIO_funcs PerlIO_encode = {
302 sizeof(PerlIOEncode),
321 PerlIOBuf_setlinebuf,
322 PerlIOEncode_get_base,
326 PerlIOBuf_set_ptrcnt,
331 Encode_Define(pTHX_ encode_t *enc)
333 HV *hash = get_hv("Encode::encoding",GV_ADD|GV_ADDMULTI);
334 HV *stash = gv_stashpv("Encode::XS", TRUE);
335 SV *sv = sv_bless(newRV_noinc(newSViv(PTR2IV(enc))),stash);
336 hv_store(hash,enc->name,strlen(enc->name),sv,0);
339 void call_failure (SV *routine, U8* done, U8* dest, U8* orig) {}
342 encode_method(pTHX_ encode_t *enc, encpage_t *dir, SV *src, int check)
345 U8 *s = (U8 *) SvPV(src,slen);
346 SV *dst = sv_2mortal(newSV(2*slen+1));
349 U8 *d = (U8 *) SvGROW(dst, 2*slen+1);
350 STRLEN dlen = SvLEN(dst);
352 while ((code = do_encode(dir,s,&slen,d,dlen,&dlen,!check)))
357 if (code == ENCODE_FALLBACK)
364 STRLEN need = (slen) ? (SvLEN(dst)*SvCUR(src)/slen) : (dlen + UTF8_MAXLEN);
365 if (need <= SvLEN(dst))
367 d = (U8 *) SvGROW(dst, need);
374 if (dir == enc->f_utf8)
376 if (!check && ckWARN_d(WARN_UTF8))
379 UV ch = utf8_to_uv(s+slen,(SvCUR(src)-slen),&clen,0);
380 Perl_warner(aTHX_ WARN_UTF8, "\"\\x{%"UVxf"}\" does not map to %s", ch, enc->name);
381 /* FIXME: Skip over the character, copy in replacement and continue
382 * but that is messy so for now just fail.
393 /* UTF-8 is supposed to be "Universal" so should not happen */
394 Perl_croak(aTHX_ "%s '%.*s' does not map to UTF-8",
395 enc->name, (int)(SvCUR(src)-slen),s+slen);
400 if (!check && ckWARN_d(WARN_UTF8))
402 Perl_warner(aTHX_ WARN_UTF8, "Partial %s character",
403 (dir == enc->f_utf8) ? "UTF-8" : enc->name);
408 Perl_croak(aTHX_ "Unexpected code %d converting %s %s",
409 code, (dir == enc->f_utf8) ? "to" : "from",enc->name);
417 if (slen < SvCUR(src))
419 Move(s+slen,s,SvCUR(src)-slen,U8);
421 SvCUR_set(src,SvCUR(src)-slen);
427 MODULE = Encode PACKAGE = Encode PREFIX = sv_
435 char *s = SvPV(sv,len);
436 if (!SvUTF8(sv) || is_utf8_string((U8*)s,len))
455 sv_utf8_downgrade(sv,failok=0)
459 MODULE = Encode PACKAGE = Encode::XS PREFIX = Encode_
464 Encode_toUnicode(obj,src,check = 0)
470 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
471 ST(0) = encode_method(aTHX_ enc, enc->t_utf8, src, check);
477 Encode_fromUnicode(obj,src,check = 0)
483 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
484 sv_utf8_upgrade(src);
485 ST(0) = encode_method(aTHX_ enc, enc->f_utf8, src, check);
489 MODULE = Encode PACKAGE = Encode
494 _bytes_to_utf8(sv, ...)
498 SV * encoding = items == 2 ? ST(1) : Nullsv;
501 RETVAL = _encoded_bytes_to_utf8(sv, SvPV_nolen(encoding));
504 U8* s = (U8*)SvPV(sv, len);
507 converted = bytes_to_utf8(s, &len); /* This allocs */
508 sv_setpvn(sv, (char *)converted, len);
509 SvUTF8_on(sv); /* XXX Should we? */
510 Safefree(converted); /* ... so free it */
518 _utf8_to_bytes(sv, ...)
522 SV * to = items > 1 ? ST(1) : Nullsv;
523 SV * check = items > 2 ? ST(2) : Nullsv;
526 RETVAL = _encoded_utf8_to_bytes(sv, SvPV_nolen(to));
529 U8 *s = (U8*)SvPV(sv, len);
532 /* Must do things the slow way */
534 U8 *src = (U8*)savepv((char *)s); /* We need a copy to pass to check() */
537 New(83, dest, len, U8); /* I think */
546 /* Have to do it all ourselves because of error routine,
550 if (!(uv & 0x20)) { ulen = 2; uv &= 0x1f; }
551 else if (!(uv & 0x10)) { ulen = 3; uv &= 0x0f; }
552 else if (!(uv & 0x08)) { ulen = 4; uv &= 0x07; }
553 else if (!(uv & 0x04)) { ulen = 5; uv &= 0x03; }
554 else if (!(uv & 0x02)) { ulen = 6; uv &= 0x01; }
555 else if (!(uv & 0x01)) { ulen = 7; uv = 0; }
556 else { ulen = 13; uv = 0; }
558 /* Note change to utf8.c variable naming, for variety */
560 if ((*s & 0xc0) != 0x80)
564 uv = (uv << 6) | (*s++ & 0x3f);
568 call_failure(check, s, dest, src);
569 /* Now what happens? */
575 RETVAL = (utf8_to_bytes(s, &len) ? len : 0);
582 _chars_to_utf8(sv, from, ...)
587 SV * check = items == 3 ? ST(2) : Nullsv;
588 RETVAL = &PL_sv_undef;
594 _utf8_to_chars(sv, to, ...)
599 SV * check = items == 3 ? ST(2) : Nullsv;
600 RETVAL = &PL_sv_undef;
606 _utf8_to_chars_check(sv, ...)
610 SV * check = items == 2 ? ST(1) : Nullsv;
611 RETVAL = &PL_sv_undef;
617 _bytes_to_chars(sv, from, ...)
622 SV * check = items == 3 ? ST(2) : Nullsv;
623 RETVAL = &PL_sv_undef;
629 _chars_to_bytes(sv, to, ...)
634 SV * check = items == 3 ? ST(2) : Nullsv;
635 RETVAL = &PL_sv_undef;
641 _from_to(sv, from, to, ...)
647 SV * check = items == 4 ? ST(3) : Nullsv;
648 RETVAL = &PL_sv_undef;
658 SV * check = items == 2 ? ST(1) : Nullsv;
660 RETVAL = SvUTF8(sv) ? 1 : 0;
663 !is_utf8_string((U8*)SvPVX(sv), SvCUR(sv)))
678 SV *rsv = newSViv(SvUTF8(sv));
682 RETVAL = &PL_sv_undef;
694 SV *rsv = newSViv(SvUTF8(sv));
698 RETVAL = &PL_sv_undef;
705 _utf_to_utf(sv, from, to, ...)
711 SV * check = items == 4 ? ST(3) : Nullsv;
712 RETVAL = &PL_sv_undef;
720 PerlIO_define_layer(&PerlIO_encode);
722 #include "iso8859.def"
723 #include "EBCDIC.def"
724 #include "Symbols.def"