1 #define PERL_NO_GET_CONTEXT
13 #define UNIMPLEMENTED(x,y) y x (SV *sv, char *encoding) {dTHX; \
14 Perl_croak(aTHX_ "panic_unimplemented"); \
15 return (y)0; /* fool picky compilers */ \
17 UNIMPLEMENTED(_encoded_utf8_to_bytes, I32)
18 UNIMPLEMENTED(_encoded_bytes_to_utf8, I32)
19 #if defined(USE_PERLIO) && !defined(USE_SFIO)
20 /* Define an encoding "layer" in the perliol.h sense.
21 The layer defined here "inherits" in an object-oriented sense from the
22 "perlio" layer with its PerlIOBuf_* "methods".
23 The implementation is particularly efficient as until Encode settles down
24 there is no point in tryint to tune it.
26 The layer works by overloading the "fill" and "flush" methods.
28 "fill" calls "SUPER::fill" in perl terms, then calls the encode OO perl API
29 to convert the encoded data to UTF-8 form, then copies it back to the
30 buffer. The "base class's" read methods then see the UTF-8 data.
32 "flush" transforms the UTF-8 data deposited by the "base class's write
33 method in the buffer back into the encoded form using the encode OO perl API,
34 then copies data back into the buffer and calls "SUPER::flush.
36 Note that "flush" is _also_ called for read mode - we still do the (back)-translate
37 so that the the base class's "flush" sees the correct number of encoded chars
38 for positioning the seek pointer. (This double translation is the worst performance
39 issue - particularly with all-perl encode engine.)
44 PerlIOBuf base; /* PerlIOBuf stuff */
45 SV *bufsv; /* buffer seen by layers above */
46 SV *dataSV; /* data we have read from layer below */
47 SV *enc; /* the encoding object */
51 PerlIOEncode_getarg(pTHX_ PerlIO * f, CLONE_PARAMS * param, int flags)
53 PerlIOEncode *e = PerlIOSelf(f, PerlIOEncode);
54 SV *sv = &PL_sv_undef;
62 if (perl_call_method("name", G_SCALAR) == 1) {
72 PerlIOEncode_pushed(pTHX_ PerlIO * f, const char *mode, SV * arg)
74 PerlIOEncode *e = PerlIOSelf(f, PerlIOEncode);
77 code = PerlIOBuf_pushed(aTHX_ f, mode, Nullsv);
83 if (perl_call_pv("Encode::find_encoding", G_SCALAR) != 1) {
84 /* should never happen */
85 Perl_die(aTHX_ "Encode::find_encoding did not return a value");
94 Perl_warner(aTHX_ WARN_IO, "Cannot find encoding \"%" SVf "\"",
100 PerlIOBase(f)->flags |= PERLIO_F_UTF8;
108 PerlIOEncode_popped(pTHX_ PerlIO * f)
110 PerlIOEncode *e = PerlIOSelf(f, PerlIOEncode);
112 SvREFCNT_dec(e->enc);
116 SvREFCNT_dec(e->bufsv);
120 SvREFCNT_dec(e->dataSV);
127 PerlIOEncode_get_base(pTHX_ PerlIO * f)
129 PerlIOEncode *e = PerlIOSelf(f, PerlIOEncode);
131 e->base.bufsiz = 1024;
133 e->bufsv = newSV(e->base.bufsiz);
134 sv_setpvn(e->bufsv, "", 0);
136 e->base.buf = (STDCHAR *) SvPVX(e->bufsv);
138 e->base.ptr = e->base.buf;
140 e->base.end = e->base.buf;
141 if (e->base.ptr < e->base.buf
142 || e->base.ptr > e->base.buf + SvLEN(e->bufsv)) {
143 Perl_warn(aTHX_ " ptr %p(%p)%p", e->base.buf, e->base.ptr,
144 e->base.buf + SvLEN(e->bufsv));
147 if (SvLEN(e->bufsv) < e->base.bufsiz) {
148 SSize_t poff = e->base.ptr - e->base.buf;
149 SSize_t eoff = e->base.end - e->base.buf;
150 e->base.buf = (STDCHAR *) SvGROW(e->bufsv, e->base.bufsiz);
151 e->base.ptr = e->base.buf + poff;
152 e->base.end = e->base.buf + eoff;
154 if (e->base.ptr < e->base.buf
155 || e->base.ptr > e->base.buf + SvLEN(e->bufsv)) {
156 Perl_warn(aTHX_ " ptr %p(%p)%p", e->base.buf, e->base.ptr,
157 e->base.buf + SvLEN(e->bufsv));
164 PerlIOEncode_fill(pTHX_ PerlIO * f)
166 PerlIOEncode *e = PerlIOSelf(f, PerlIOEncode);
171 if (PerlIO_flush(f) != 0)
174 if (!PerlIO_fast_gets(n)) {
175 /* Things get too messy if we don't have a buffer layer
176 push a :perlio to do the job */
178 n = PerlIO_push(aTHX_ n, &PerlIO_perlio, PerlIO_modestr(f,mode), Nullsv);
180 Perl_die(aTHX_ "panic: cannot push :perlio for %p",f);
186 avail = PerlIO_get_cnt(n);
188 avail = PerlIO_fill(n);
190 avail = PerlIO_get_cnt(n);
193 if (!PerlIO_error(n) && PerlIO_eof(n))
198 STDCHAR *ptr = PerlIO_get_ptr(n);
203 e->base.ptr = e->base.end = (STDCHAR *) Nullch;
204 (void) PerlIOEncode_get_base(aTHX_ f);
206 e->dataSV = newSV(0);
207 if (SvTYPE(e->dataSV) < SVt_PV) {
208 sv_upgrade(e->dataSV,SVt_PV);
210 if (SvCUR(e->dataSV)) {
211 /* something left over from last time - create a normal
212 SV with new data appended
214 if (use + SvCUR(e->dataSV) > e->base.bufsiz) {
215 use = e->base.bufsiz - SvCUR(e->dataSV);
217 sv_catpvn(e->dataSV,ptr,use);
220 /* Create a "dummy" SV to represent the available data from layer below */
221 if (SvLEN(e->dataSV) && SvPVX(e->dataSV)) {
222 Safefree(SvPVX(e->dataSV));
224 if (use > e->base.bufsiz) {
225 use = e->base.bufsiz;
227 SvPVX(e->dataSV) = (char *) ptr;
228 SvLEN(e->dataSV) = 0; /* Hands off sv.c - it isn't yours */
229 SvCUR_set(e->dataSV,use);
232 SvUTF8_off(e->dataSV);
238 if (perl_call_method("decode", G_SCALAR) != 1) {
239 Perl_die(aTHX_ "panic: decode did not return a value");
244 /* Now get translated string (forced to UTF-8) and use as buffer */
246 s = SvPVutf8(uni, len);
247 if (len && !is_utf8_string(s,len)) {
248 Perl_warn(aTHX_ "panic: decode did not return UTF-8 '%.*s'",(int) len,s);
253 /* if decode gave us back dataSV then data may vanish when
254 we do ptrcnt adjust - so take our copy now.
255 (The copy is a pain - need a put-it-here option for decode.)
257 sv_setpvn(e->bufsv,s,len);
258 e->base.ptr = e->base.buf = SvPVX(e->bufsv);
259 e->base.end = e->base.ptr + SvCUR(e->bufsv);
260 PerlIOBase(f)->flags |= PERLIO_F_RDBUF;
263 /* Adjust ptr/cnt not taking anything which
264 did not translate - not clear this is a win */
265 /* compute amount we took */
266 use -= SvCUR(e->dataSV);
267 PerlIO_set_ptrcnt(n, ptr+use, (avail-use));
268 /* and as we did not take it it isn't pending */
269 SvCUR_set(e->dataSV,0);
271 /* Got nothing - assume partial character so we need some more */
272 /* Make sure e->dataSV is a normal SV before re-filling as
273 buffer alias will change under us
275 s = SvPV(e->dataSV,len);
276 sv_setpvn(e->dataSV,s,len);
277 PerlIO_set_ptrcnt(n, ptr+use, (avail-use));
286 PerlIOBase(f)->flags |= PERLIO_F_EOF;
288 PerlIOBase(f)->flags |= PERLIO_F_ERROR;
294 PerlIOEncode_flush(pTHX_ PerlIO * f)
296 PerlIOEncode *e = PerlIOSelf(f, PerlIOEncode);
298 if (e->bufsv && (e->base.ptr > e->base.buf)) {
304 if (PerlIOBase(f)->flags & PERLIO_F_WRBUF) {
305 /* Write case encode the buffer and write() to layer below */
310 SvCUR_set(e->bufsv, e->base.ptr - e->base.buf);
312 Perl_warn(aTHX_ "flush %_",e->bufsv);
316 if (perl_call_method("encode", G_SCALAR) != 1)
322 count = PerlIO_write(PerlIONext(f),s,len);
328 if (PerlIO_flush(PerlIONext(f)) != 0) {
332 else if (PerlIOBase(f)->flags & PERLIO_F_RDBUF) {
334 /* if we have any untranslated stuff then unread that first */
335 if (e->dataSV && SvCUR(e->dataSV)) {
336 s = SvPV(e->dataSV, len);
337 count = PerlIO_unread(PerlIONext(f),s,len);
342 /* See if there is anything left in the buffer */
343 if (e->base.ptr < e->base.end) {
344 /* Bother - have unread data.
345 re-encode and unread() to layer below
349 str = sv_newmortal();
350 sv_upgrade(str, SVt_PV);
351 SvPVX(str) = e->base.ptr;
353 SvCUR_set(str, e->base.end - e->base.ptr);
360 if (perl_call_method("encode", G_SCALAR) != 1)
366 count = PerlIO_unread(PerlIONext(f),s,len);
374 e->base.ptr = e->base.end = e->base.buf;
375 PerlIOBase(f)->flags &= ~(PERLIO_F_RDBUF | PERLIO_F_WRBUF);
381 PerlIOEncode_close(pTHX_ PerlIO * f)
383 PerlIOEncode *e = PerlIOSelf(f, PerlIOEncode);
384 IV code = PerlIOBase_close(aTHX_ f);
386 SvREFCNT_dec(e->bufsv);
392 PerlIOBase(f)->flags &= ~(PERLIO_F_RDBUF | PERLIO_F_WRBUF);
397 PerlIOEncode_tell(pTHX_ PerlIO * f)
399 PerlIOBuf *b = PerlIOSelf(f, PerlIOBuf);
400 /* Unfortunately the only way to get a postion is to (re-)translate,
401 the UTF8 we have in bufefr and then ask layer below
404 return PerlIO_tell(PerlIONext(f));
408 PerlIOEncode_dup(pTHX_ PerlIO * f, PerlIO * o,
409 CLONE_PARAMS * params, int flags)
411 if ((f = PerlIOBase_dup(aTHX_ f, o, params, flags))) {
412 PerlIOEncode *fe = PerlIOSelf(f, PerlIOEncode);
413 PerlIOEncode *oe = PerlIOSelf(o, PerlIOEncode);
415 fe->enc = PerlIO_sv_dup(aTHX_ oe->enc, params);
421 PerlIO_funcs PerlIO_encode = {
423 sizeof(PerlIOEncode),
442 PerlIOBase_setlinebuf,
443 PerlIOEncode_get_base,
447 PerlIOBuf_set_ptrcnt,
449 #endif /* encode layer */
452 Encode_XSEncoding(pTHX_ encode_t * enc)
455 HV *stash = gv_stashpv("Encode::XS", TRUE);
456 SV *sv = sv_bless(newRV_noinc(newSViv(PTR2IV(enc))), stash);
460 while (enc->name[i]) {
461 const char *name = enc->name[i++];
462 XPUSHs(sv_2mortal(newSVpvn(name, strlen(name))));
465 call_pv("Encode::define_encoding", G_DISCARD);
470 call_failure(SV * routine, U8 * done, U8 * dest, U8 * orig)
475 encode_method(pTHX_ encode_t * enc, encpage_t * dir, SV * src,
479 U8 *s = (U8 *) SvPV(src, slen);
483 SV *dst = sv_2mortal(newSV(slen+1));
485 U8 *d = (U8 *) SvPVX(dst);
486 STRLEN dlen = SvLEN(dst)-1;
488 while ((code = do_encode(dir, s, &slen, d, dlen, &dlen, !check))) {
489 SvCUR_set(dst, dlen+ddone);
493 Perl_warn(aTHX_ "code=%d @ s=%d/%d/%d d=%d/%d/%d",code,slen,sdone,tlen,dlen,ddone,SvLEN(dst)-1);
496 if (code == ENCODE_FALLBACK || code == ENCODE_PARTIAL)
506 need = (tlen*SvLEN(dst)+sdone-1)/sdone+UTF8_MAXLEN;
509 need = SvLEN(dst) + UTF8_MAXLEN;
512 d = (U8 *) SvGROW(dst, need);
513 if (ddone >= SvLEN(dst)) {
514 Perl_croak(aTHX_ "Destination couldn't be grown.");
516 dlen = SvLEN(dst)-ddone-1;
524 if (dir == enc->f_utf8) {
525 if (!check && ckWARN_d(WARN_UTF8)) {
528 utf8n_to_uvuni(s + slen, (SvCUR(src) - slen),
530 Perl_warner(aTHX_ WARN_UTF8,
532 "}\" does not map to %s", ch,
534 /* FIXME: Skip over the character, copy in replacement and continue
535 * but that is messy so for now just fail.
544 /* UTF-8 is supposed to be "Universal" so should not happen */
545 Perl_croak(aTHX_ "%s '%.*s' does not map to UTF-8",
546 enc->name[0], (int) (SvCUR(src) - slen),
552 Perl_croak(aTHX_ "Unexpected code %d converting %s %s",
553 code, (dir == enc->f_utf8) ? "to" : "from",
558 SvCUR_set(dst, dlen+ddone);
561 sdone = SvCUR(src) - (slen+sdone);
563 Move(s + slen, SvPVX(src), sdone , U8);
565 SvCUR_set(src, sdone);
577 MODULE = Encode PACKAGE = Encode::XS PREFIX = Method_
582 Method_decode(obj,src,check = FALSE)
588 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
589 ST(0) = encode_method(aTHX_ enc, enc->t_utf8, src, check);
595 Method_encode(obj,src,check = FALSE)
601 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
602 sv_utf8_upgrade(src);
603 ST(0) = encode_method(aTHX_ enc, enc->f_utf8, src, check);
607 MODULE = Encode PACKAGE = Encode
612 _bytes_to_utf8(sv, ...)
616 SV * encoding = items == 2 ? ST(1) : Nullsv;
619 RETVAL = _encoded_bytes_to_utf8(sv, SvPV_nolen(encoding));
622 U8* s = (U8*)SvPV(sv, len);
625 converted = bytes_to_utf8(s, &len); /* This allocs */
626 sv_setpvn(sv, (char *)converted, len);
627 SvUTF8_on(sv); /* XXX Should we? */
628 Safefree(converted); /* ... so free it */
636 _utf8_to_bytes(sv, ...)
640 SV * to = items > 1 ? ST(1) : Nullsv;
641 SV * check = items > 2 ? ST(2) : Nullsv;
644 RETVAL = _encoded_utf8_to_bytes(sv, SvPV_nolen(to));
647 U8 *s = (U8*)SvPV(sv, len);
651 /* Must do things the slow way */
653 U8 *src = (U8*)savepv((char *)s); /* We need a copy to pass to check() */
656 New(83, dest, len, U8); /* I think */
665 /* Have to do it all ourselves because of error routine,
669 if (!(uv & 0x20)) { ulen = 2; uv &= 0x1f; }
670 else if (!(uv & 0x10)) { ulen = 3; uv &= 0x0f; }
671 else if (!(uv & 0x08)) { ulen = 4; uv &= 0x07; }
672 else if (!(uv & 0x04)) { ulen = 5; uv &= 0x03; }
673 else if (!(uv & 0x02)) { ulen = 6; uv &= 0x01; }
674 else if (!(uv & 0x01)) { ulen = 7; uv = 0; }
675 else { ulen = 13; uv = 0; }
677 /* Note change to utf8.c variable naming, for variety */
679 if ((*s & 0xc0) != 0x80)
683 uv = (uv << 6) | (*s++ & 0x3f);
687 call_failure(check, s, dest, src);
688 /* Now what happens? */
694 RETVAL = (utf8_to_bytes(s, &len) ? len : 0);
701 is_utf8(sv, check = FALSE)
706 if (SvGMAGICAL(sv)) /* it could be $1, for example */
707 sv = newSVsv(sv); /* GMAGIG will be done */
709 RETVAL = SvUTF8(sv) ? TRUE : FALSE;
712 !is_utf8_string((U8*)SvPVX(sv), SvCUR(sv)))
718 SvREFCNT_dec(sv); /* it was a temp copy */
729 SV *rsv = newSViv(SvUTF8(sv));
733 RETVAL = &PL_sv_undef;
745 SV *rsv = newSViv(SvUTF8(sv));
749 RETVAL = &PL_sv_undef;
757 #if defined(USE_PERLIO) && !defined(USE_SFIO)
758 PerlIO_define_layer(aTHX_ &PerlIO_encode);
760 #include "8859_def.h"
761 #include "EBCDIC_def.h"
762 #include "Symbols_def.h"