Blame Encode.xs

Packit d0f5c2
/*
Packit d0f5c2
 $Id: Encode.xs,v 2.43 2018/02/21 12:14:33 dankogai Exp dankogai $
Packit d0f5c2
 */
Packit d0f5c2
Packit d0f5c2
#define PERL_NO_GET_CONTEXT
Packit d0f5c2
#include "EXTERN.h"
Packit d0f5c2
#include "perl.h"
Packit d0f5c2
#include "XSUB.h"
Packit d0f5c2
#include "encode.h"
Packit d0f5c2
#include "def_t.h"
Packit d0f5c2
Packit d0f5c2
# define PERLIO_MODNAME  "PerlIO::encoding"
Packit d0f5c2
# define PERLIO_FILENAME "PerlIO/encoding.pm"
Packit d0f5c2
Packit d0f5c2
/* set 1 or more to profile.  t/encoding.t dumps core because of
Packit d0f5c2
   Perl_warner and PerlIO don't work well */
Packit d0f5c2
#define ENCODE_XS_PROFILE 0
Packit d0f5c2
Packit d0f5c2
/* set 0 to disable floating point to calculate buffer size for
Packit d0f5c2
   encode_method().  1 is recommended. 2 restores NI-S original */
Packit d0f5c2
#define ENCODE_XS_USEFP   1
Packit d0f5c2
Packit d0f5c2
#define UNIMPLEMENTED(x,y) static y x (SV *sv, char *encoding) {	\
Packit d0f5c2
			Perl_croak_nocontext("panic_unimplemented");	\
Packit d0f5c2
                        PERL_UNUSED_VAR(sv); \
Packit d0f5c2
                        PERL_UNUSED_VAR(encoding); \
Packit d0f5c2
             return (y)0; /* fool picky compilers */ \
Packit d0f5c2
                         }
Packit d0f5c2
/**/
Packit d0f5c2
Packit d0f5c2
UNIMPLEMENTED(_encoded_utf8_to_bytes, I32)
Packit d0f5c2
UNIMPLEMENTED(_encoded_bytes_to_utf8, I32)
Packit d0f5c2
Packit d0f5c2
#ifndef SvIV_nomg
Packit d0f5c2
#define SvIV_nomg SvIV
Packit d0f5c2
#endif
Packit d0f5c2
Packit d0f5c2
#ifndef UTF8_DISALLOW_ILLEGAL_INTERCHANGE
Packit d0f5c2
#  define UTF8_DISALLOW_ILLEGAL_INTERCHANGE 0
Packit d0f5c2
#  define UTF8_ALLOW_NON_STRICT (UTF8_ALLOW_FE_FF|UTF8_ALLOW_SURROGATE|UTF8_ALLOW_FFFF)
Packit d0f5c2
#else
Packit d0f5c2
#  define UTF8_ALLOW_NON_STRICT 0
Packit d0f5c2
#endif
Packit d0f5c2
Packit d0f5c2
static void
Packit d0f5c2
Encode_XSEncoding(pTHX_ encode_t * enc)
Packit d0f5c2
{
Packit d0f5c2
    dSP;
Packit d0f5c2
    HV *stash = gv_stashpv("Encode::XS", TRUE);
Packit d0f5c2
    SV *iv    = newSViv(PTR2IV(enc));
Packit d0f5c2
    SV *sv    = sv_bless(newRV_noinc(iv),stash);
Packit d0f5c2
    int i = 0;
Packit d0f5c2
    /* with the SvLEN() == 0 hack, PVX won't be freed. We cast away name's
Packit d0f5c2
    constness, in the hope that perl won't mess with it. */
Packit d0f5c2
    assert(SvTYPE(iv) >= SVt_PV); assert(SvLEN(iv) == 0);
Packit d0f5c2
    SvFLAGS(iv) |= SVp_POK;
Packit d0f5c2
    SvPVX(iv) = (char*) enc->name[0];
Packit d0f5c2
    PUSHMARK(sp);
Packit d0f5c2
    XPUSHs(sv);
Packit d0f5c2
    while (enc->name[i]) {
Packit d0f5c2
    const char *name = enc->name[i++];
Packit d0f5c2
    XPUSHs(sv_2mortal(newSVpvn(name, strlen(name))));
Packit d0f5c2
    }
Packit d0f5c2
    PUTBACK;
Packit d0f5c2
    call_pv("Encode::define_encoding", G_DISCARD);
Packit d0f5c2
    SvREFCNT_dec(sv);
Packit d0f5c2
}
Packit d0f5c2
Packit d0f5c2
static void
Packit d0f5c2
call_failure(SV * routine, U8 * done, U8 * dest, U8 * orig)
Packit d0f5c2
{
Packit d0f5c2
    /* Exists for breakpointing */
Packit d0f5c2
    PERL_UNUSED_VAR(routine);
Packit d0f5c2
    PERL_UNUSED_VAR(done);
Packit d0f5c2
    PERL_UNUSED_VAR(dest);
Packit d0f5c2
    PERL_UNUSED_VAR(orig);
Packit d0f5c2
}
Packit d0f5c2
Packit d0f5c2
static void
Packit d0f5c2
utf8_safe_downgrade(pTHX_ SV ** src, U8 ** s, STRLEN * slen, bool modify)
Packit d0f5c2
{
Packit d0f5c2
    if (!modify) {
Packit d0f5c2
        SV *tmp = sv_2mortal(newSVpvn((char *)*s, *slen));
Packit d0f5c2
        SvUTF8_on(tmp);
Packit d0f5c2
        if (SvTAINTED(*src))
Packit d0f5c2
            SvTAINTED_on(tmp);
Packit d0f5c2
        *src = tmp;
Packit d0f5c2
        *s = (U8 *)SvPVX(*src);
Packit d0f5c2
    }
Packit d0f5c2
    if (*slen) {
Packit d0f5c2
        if (!utf8_to_bytes(*s, slen))
Packit d0f5c2
            croak("Wide character");
Packit d0f5c2
        SvCUR_set(*src, *slen);
Packit d0f5c2
    }
Packit d0f5c2
    SvUTF8_off(*src);
Packit d0f5c2
}
Packit d0f5c2
Packit d0f5c2
static void
Packit d0f5c2
utf8_safe_upgrade(pTHX_ SV ** src, U8 ** s, STRLEN * slen, bool modify)
Packit d0f5c2
{
Packit d0f5c2
    if (!modify) {
Packit d0f5c2
        SV *tmp = sv_2mortal(newSVpvn((char *)*s, *slen));
Packit d0f5c2
        if (SvTAINTED(*src))
Packit d0f5c2
            SvTAINTED_on(tmp);
Packit d0f5c2
        *src = tmp;
Packit d0f5c2
    }
Packit d0f5c2
    sv_utf8_upgrade_nomg(*src);
Packit d0f5c2
    *s = (U8 *)SvPV_nomg(*src, *slen);
Packit d0f5c2
}
Packit d0f5c2
Packit d0f5c2
#define ERR_ENCODE_NOMAP "\"\\x{%04" UVxf "}\" does not map to %s"
Packit d0f5c2
#define ERR_DECODE_NOMAP "%s \"\\x%02" UVXf "\" does not map to Unicode"
Packit d0f5c2
#define ERR_DECODE_STR_NOMAP "%s \"%s\" does not map to Unicode"
Packit d0f5c2
Packit d0f5c2
static SV *
Packit d0f5c2
do_fallback_cb(pTHX_ UV ch, SV *fallback_cb)
Packit d0f5c2
{
Packit d0f5c2
    dSP;
Packit d0f5c2
    int argc;
Packit d0f5c2
    SV *retval;
Packit d0f5c2
    ENTER;
Packit d0f5c2
    SAVETMPS;
Packit d0f5c2
    PUSHMARK(sp);
Packit d0f5c2
    XPUSHs(sv_2mortal(newSVuv(ch)));
Packit d0f5c2
    PUTBACK;
Packit d0f5c2
    argc = call_sv(fallback_cb, G_SCALAR);
Packit d0f5c2
    SPAGAIN;
Packit d0f5c2
    if (argc != 1){
Packit d0f5c2
	croak("fallback sub must return scalar!");
Packit d0f5c2
    }
Packit d0f5c2
    retval = POPs;
Packit d0f5c2
    SvREFCNT_inc(retval);
Packit d0f5c2
    PUTBACK;
Packit d0f5c2
    FREETMPS;
Packit d0f5c2
    LEAVE;
Packit d0f5c2
    return retval;
Packit d0f5c2
}
Packit d0f5c2
Packit d0f5c2
static SV *
Packit d0f5c2
do_bytes_fallback_cb(pTHX_ U8 *s, STRLEN slen, SV *fallback_cb)
Packit d0f5c2
{
Packit d0f5c2
    dSP;
Packit d0f5c2
    int argc;
Packit d0f5c2
    STRLEN i;
Packit d0f5c2
    SV *retval;
Packit d0f5c2
    ENTER;
Packit d0f5c2
    SAVETMPS;
Packit d0f5c2
    PUSHMARK(sp);
Packit d0f5c2
    for (i=0; i
Packit d0f5c2
        XPUSHs(sv_2mortal(newSVuv(s[i])));
Packit d0f5c2
    PUTBACK;
Packit d0f5c2
    argc = call_sv(fallback_cb, G_SCALAR);
Packit d0f5c2
    SPAGAIN;
Packit d0f5c2
    if (argc != 1){
Packit d0f5c2
        croak("fallback sub must return scalar!");
Packit d0f5c2
    }
Packit d0f5c2
    retval = POPs;
Packit d0f5c2
    SvREFCNT_inc(retval);
Packit d0f5c2
    PUTBACK;
Packit d0f5c2
    FREETMPS;
Packit d0f5c2
    LEAVE;
Packit d0f5c2
    return retval;
Packit d0f5c2
}
Packit d0f5c2
Packit d0f5c2
static SV *
Packit d0f5c2
encode_method(pTHX_ const encode_t * enc, const encpage_t * dir, SV * src, U8 * s, STRLEN slen,
Packit d0f5c2
	      int check, STRLEN * offset, SV * term, int * retcode, 
Packit d0f5c2
	      SV *fallback_cb)
Packit d0f5c2
{
Packit d0f5c2
    STRLEN tlen  = slen;
Packit d0f5c2
    STRLEN ddone = 0;
Packit d0f5c2
    STRLEN sdone = 0;
Packit d0f5c2
    /* We allocate slen+1.
Packit d0f5c2
       PerlIO dumps core if this value is smaller than this. */
Packit d0f5c2
    SV *dst = newSV(slen+1);
Packit d0f5c2
    U8 *d = (U8 *)SvPVX(dst);
Packit d0f5c2
    STRLEN dlen = SvLEN(dst)-1;
Packit d0f5c2
    int code = 0;
Packit d0f5c2
    STRLEN trmlen = 0;
Packit d0f5c2
    U8 *trm = term ? (U8*) SvPV(term, trmlen) : NULL;
Packit d0f5c2
Packit d0f5c2
    if (SvTAINTED(src)) SvTAINTED_on(dst); /* propagate taintedness */
Packit d0f5c2
Packit d0f5c2
    if (offset) {
Packit d0f5c2
      s += *offset;
Packit d0f5c2
      if (slen > *offset){ /* safeguard against slen overflow */
Packit d0f5c2
          slen -= *offset;
Packit d0f5c2
      }else{
Packit d0f5c2
          slen = 0;
Packit d0f5c2
      }
Packit d0f5c2
      tlen = slen;
Packit d0f5c2
    }
Packit d0f5c2
Packit d0f5c2
    if (slen == 0){
Packit d0f5c2
        SvCUR_set(dst, 0);
Packit d0f5c2
        SvPOK_only(dst);
Packit d0f5c2
        goto ENCODE_END;
Packit d0f5c2
    }
Packit d0f5c2
Packit d0f5c2
    while( (code = do_encode(dir, s, &slen, d, dlen, &dlen, !check,
Packit d0f5c2
                 trm, trmlen)) )
Packit d0f5c2
    {
Packit d0f5c2
        SvCUR_set(dst, dlen+ddone);
Packit d0f5c2
        SvPOK_only(dst);
Packit d0f5c2
Packit d0f5c2
        if (code == ENCODE_FALLBACK || code == ENCODE_PARTIAL ||
Packit d0f5c2
            code == ENCODE_FOUND_TERM) {
Packit d0f5c2
            break;
Packit d0f5c2
        }
Packit d0f5c2
        switch (code) {
Packit d0f5c2
        case ENCODE_NOSPACE:
Packit d0f5c2
        {
Packit d0f5c2
            STRLEN more = 0; /* make sure you initialize! */
Packit d0f5c2
            STRLEN sleft;
Packit d0f5c2
            sdone += slen;
Packit d0f5c2
            ddone += dlen;
Packit d0f5c2
            sleft = tlen - sdone;
Packit d0f5c2
#if ENCODE_XS_PROFILE >= 2
Packit d0f5c2
            Perl_warn(aTHX_
Packit d0f5c2
                  "more=%d, sdone=%d, sleft=%d, SvLEN(dst)=%d\n",
Packit d0f5c2
                  more, sdone, sleft, SvLEN(dst));
Packit d0f5c2
#endif
Packit d0f5c2
            if (sdone != 0) { /* has src ever been processed ? */
Packit d0f5c2
#if   ENCODE_XS_USEFP == 2
Packit d0f5c2
                more = (1.0*tlen*SvLEN(dst)+sdone-1)/sdone
Packit d0f5c2
                    - SvLEN(dst);
Packit d0f5c2
#elif ENCODE_XS_USEFP
Packit d0f5c2
                more = (STRLEN)((1.0*SvLEN(dst)+1)/sdone * sleft);
Packit d0f5c2
#else
Packit d0f5c2
            /* safe until SvLEN(dst) == MAX_INT/16 */
Packit d0f5c2
                more = (16*SvLEN(dst)+1)/sdone/16 * sleft;
Packit d0f5c2
#endif
Packit d0f5c2
            }
Packit d0f5c2
            more += UTF8_MAXLEN; /* insurance policy */
Packit d0f5c2
            d = (U8 *) SvGROW(dst, SvLEN(dst) + more);
Packit d0f5c2
            /* dst need to grow need MORE bytes! */
Packit d0f5c2
            if (ddone >= SvLEN(dst)) {
Packit d0f5c2
                Perl_croak(aTHX_ "Destination couldn't be grown.");
Packit d0f5c2
            }
Packit d0f5c2
            dlen = SvLEN(dst)-ddone-1;
Packit d0f5c2
            d   += ddone;
Packit d0f5c2
            s   += slen;
Packit d0f5c2
            slen = tlen-sdone;
Packit d0f5c2
            continue;
Packit d0f5c2
        }
Packit d0f5c2
Packit d0f5c2
    case ENCODE_NOREP:
Packit d0f5c2
        /* encoding */	
Packit d0f5c2
        if (dir == enc->f_utf8) {
Packit d0f5c2
        STRLEN clen;
Packit d0f5c2
        UV ch =
Packit d0f5c2
            utf8n_to_uvuni(s+slen, (tlen-sdone-slen),
Packit d0f5c2
                   &clen, UTF8_ALLOW_ANY|UTF8_CHECK_ONLY);
Packit d0f5c2
        /* if non-representable multibyte prefix at end of current buffer - break*/
Packit d0f5c2
        if (clen > tlen - sdone - slen) break;
Packit d0f5c2
        if (check & ENCODE_DIE_ON_ERR) {
Packit d0f5c2
            Perl_croak(aTHX_ ERR_ENCODE_NOMAP,
Packit d0f5c2
                   (UV)ch, enc->name[0]);
Packit d0f5c2
            return &PL_sv_undef; /* never reaches but be safe */
Packit d0f5c2
        }
Packit d0f5c2
        if (check & ENCODE_WARN_ON_ERR){
Packit d0f5c2
            Perl_warner(aTHX_ packWARN(WARN_UTF8),
Packit d0f5c2
                ERR_ENCODE_NOMAP, (UV)ch, enc->name[0]);
Packit d0f5c2
        }
Packit d0f5c2
        if (check & ENCODE_RETURN_ON_ERR){
Packit d0f5c2
            goto ENCODE_SET_SRC;
Packit d0f5c2
        }
Packit d0f5c2
        if (check & (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){
Packit d0f5c2
            STRLEN sublen;
Packit d0f5c2
            char *substr;
Packit d0f5c2
            SV* subchar = 
Packit d0f5c2
            (fallback_cb != &PL_sv_undef)
Packit d0f5c2
		? do_fallback_cb(aTHX_ ch, fallback_cb)
Packit d0f5c2
		: newSVpvf(check & ENCODE_PERLQQ ? "\\x{%04" UVxf "}" :
Packit d0f5c2
                 check & ENCODE_HTMLCREF ? "&#%" UVuf ";" :
Packit d0f5c2
                 "&#x%" UVxf ";", (UV)ch);
Packit d0f5c2
            substr = SvPV(subchar, sublen);
Packit d0f5c2
            if (SvUTF8(subchar) && sublen && !utf8_to_bytes((U8 *)substr, &sublen)) { /* make sure no decoded string gets in */
Packit d0f5c2
                SvREFCNT_dec(subchar);
Packit d0f5c2
                croak("Wide character");
Packit d0f5c2
            }
Packit d0f5c2
            sdone += slen + clen;
Packit d0f5c2
            ddone += dlen + sublen;
Packit d0f5c2
            sv_catpvn(dst, substr, sublen);
Packit d0f5c2
            SvREFCNT_dec(subchar);
Packit d0f5c2
        } else {
Packit d0f5c2
            /* fallback char */
Packit d0f5c2
            sdone += slen + clen;
Packit d0f5c2
            ddone += dlen + enc->replen;
Packit d0f5c2
            sv_catpvn(dst, (char*)enc->rep, enc->replen);
Packit d0f5c2
        }
Packit d0f5c2
        }
Packit d0f5c2
        /* decoding */
Packit d0f5c2
        else {
Packit d0f5c2
        if (check & ENCODE_DIE_ON_ERR){
Packit d0f5c2
            Perl_croak(aTHX_ ERR_DECODE_NOMAP,
Packit d0f5c2
                              enc->name[0], (UV)s[slen]);
Packit d0f5c2
            return &PL_sv_undef; /* never reaches but be safe */
Packit d0f5c2
        }
Packit d0f5c2
        if (check & ENCODE_WARN_ON_ERR){
Packit d0f5c2
            Perl_warner(
Packit d0f5c2
            aTHX_ packWARN(WARN_UTF8),
Packit d0f5c2
            ERR_DECODE_NOMAP,
Packit d0f5c2
               	        enc->name[0], (UV)s[slen]);
Packit d0f5c2
        }
Packit d0f5c2
        if (check & ENCODE_RETURN_ON_ERR){
Packit d0f5c2
            goto ENCODE_SET_SRC;
Packit d0f5c2
        }
Packit d0f5c2
        if (check &
Packit d0f5c2
            (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){
Packit d0f5c2
            STRLEN sublen;
Packit d0f5c2
            char *substr;
Packit d0f5c2
            SV* subchar = 
Packit d0f5c2
            (fallback_cb != &PL_sv_undef)
Packit d0f5c2
		? do_fallback_cb(aTHX_ (UV)s[slen], fallback_cb) 
Packit d0f5c2
		: newSVpvf("\\x%02" UVXf, (UV)s[slen]);
Packit d0f5c2
            substr = SvPVutf8(subchar, sublen);
Packit d0f5c2
            sdone += slen + 1;
Packit d0f5c2
            ddone += dlen + sublen;
Packit d0f5c2
            sv_catpvn(dst, substr, sublen);
Packit d0f5c2
            SvREFCNT_dec(subchar);
Packit d0f5c2
        } else {
Packit d0f5c2
            sdone += slen + 1;
Packit d0f5c2
            ddone += dlen + strlen(FBCHAR_UTF8);
Packit d0f5c2
            sv_catpvn(dst, FBCHAR_UTF8, strlen(FBCHAR_UTF8));
Packit d0f5c2
        }
Packit d0f5c2
        }
Packit d0f5c2
        /* settle variables when fallback */
Packit d0f5c2
        d    = (U8 *)SvEND(dst);
Packit d0f5c2
        dlen = SvLEN(dst) - ddone - 1;
Packit d0f5c2
        s    = (U8*)SvPVX(src) + sdone;
Packit d0f5c2
        slen = tlen - sdone;
Packit d0f5c2
        break;
Packit d0f5c2
Packit d0f5c2
        default:
Packit d0f5c2
            Perl_croak(aTHX_ "Unexpected code %d converting %s %s",
Packit d0f5c2
                   code, (dir == enc->f_utf8) ? "to" : "from",
Packit d0f5c2
                   enc->name[0]);
Packit d0f5c2
            return &PL_sv_undef;
Packit d0f5c2
        }
Packit d0f5c2
    }   /* End of looping through the string */
Packit d0f5c2
 ENCODE_SET_SRC:
Packit d0f5c2
    if (check && !(check & ENCODE_LEAVE_SRC)){
Packit d0f5c2
    sdone = SvCUR(src) - (slen+sdone);
Packit d0f5c2
    if (sdone) {
Packit d0f5c2
        sv_setpvn(src, (char*)s+slen, sdone);
Packit d0f5c2
    }
Packit d0f5c2
    SvCUR_set(src, sdone);
Packit d0f5c2
    SvSETMAGIC(src);
Packit d0f5c2
    }
Packit d0f5c2
    /* warn("check = 0x%X, code = 0x%d\n", check, code); */
Packit d0f5c2
Packit d0f5c2
    SvCUR_set(dst, dlen+ddone);
Packit d0f5c2
    SvPOK_only(dst);
Packit d0f5c2
Packit d0f5c2
#if ENCODE_XS_PROFILE
Packit d0f5c2
    if (SvCUR(dst) > SvCUR(src)){
Packit d0f5c2
    Perl_warn(aTHX_
Packit d0f5c2
          "SvLEN(dst)=%d, SvCUR(dst)=%d. %d bytes unused(%f %%)\n",
Packit d0f5c2
          SvLEN(dst), SvCUR(dst), SvLEN(dst) - SvCUR(dst),
Packit d0f5c2
          (SvLEN(dst) - SvCUR(dst))*1.0/SvLEN(dst)*100.0);
Packit d0f5c2
    }
Packit d0f5c2
#endif
Packit d0f5c2
Packit d0f5c2
    if (offset)
Packit d0f5c2
      *offset += sdone + slen;
Packit d0f5c2
Packit d0f5c2
 ENCODE_END:
Packit d0f5c2
    *SvEND(dst) = '\0';
Packit d0f5c2
    if (retcode) *retcode = code;
Packit d0f5c2
    return dst;
Packit d0f5c2
}
Packit d0f5c2
Packit d0f5c2
static bool
Packit d0f5c2
strict_utf8(pTHX_ SV* sv)
Packit d0f5c2
{
Packit d0f5c2
    HV* hv;
Packit d0f5c2
    SV** svp;
Packit d0f5c2
    sv = SvRV(sv);
Packit d0f5c2
    if (!sv || SvTYPE(sv) != SVt_PVHV)
Packit d0f5c2
        return 0;
Packit d0f5c2
    hv = (HV*)sv;
Packit d0f5c2
    svp = hv_fetch(hv, "strict_utf8", 11, 0);
Packit d0f5c2
    if (!svp)
Packit d0f5c2
        return 0;
Packit d0f5c2
    return SvTRUE(*svp);
Packit d0f5c2
}
Packit d0f5c2
Packit d0f5c2
/* Modern perls have the capability to do this more efficiently and portably */
Packit d0f5c2
#ifdef utf8n_to_uvchr_msgs
Packit d0f5c2
# define CAN_USE_BASE_PERL
Packit d0f5c2
#endif
Packit d0f5c2
Packit d0f5c2
#ifndef CAN_USE_BASE_PERL
Packit d0f5c2
Packit d0f5c2
/*
Packit d0f5c2
 * https://github.com/dankogai/p5-encode/pull/56#issuecomment-231959126
Packit d0f5c2
 */
Packit d0f5c2
#ifndef UNICODE_IS_NONCHAR
Packit d0f5c2
#define UNICODE_IS_NONCHAR(c) ((c >= 0xFDD0 && c <= 0xFDEF) || (c & 0xFFFE) == 0xFFFE)
Packit d0f5c2
#endif
Packit d0f5c2
Packit d0f5c2
#ifndef UNICODE_IS_SUPER
Packit d0f5c2
#define UNICODE_IS_SUPER(c) (c > PERL_UNICODE_MAX)
Packit d0f5c2
#endif
Packit d0f5c2
Packit d0f5c2
#define UNICODE_IS_STRICT(c) (!UNICODE_IS_SURROGATE(c) && !UNICODE_IS_NONCHAR(c) && !UNICODE_IS_SUPER(c))
Packit d0f5c2
Packit d0f5c2
#ifndef UTF_ACCUMULATION_OVERFLOW_MASK
Packit d0f5c2
#ifndef CHARBITS
Packit d0f5c2
#define CHARBITS CHAR_BIT
Packit d0f5c2
#endif
Packit d0f5c2
#define UTF_ACCUMULATION_OVERFLOW_MASK (((UV) UTF_CONTINUATION_MASK) << ((sizeof(UV) * CHARBITS) - UTF_ACCUMULATION_SHIFT))
Packit d0f5c2
#endif
Packit d0f5c2
Packit d0f5c2
/*
Packit d0f5c2
 * Convert non strict utf8 sequence of len >= 2 to unicode codepoint
Packit d0f5c2
 */
Packit d0f5c2
static UV
Packit d0f5c2
convert_utf8_multi_seq(U8* s, STRLEN len, STRLEN *rlen)
Packit d0f5c2
{
Packit d0f5c2
    UV uv;
Packit d0f5c2
    U8 *ptr = s;
Packit d0f5c2
    bool overflowed = 0;
Packit d0f5c2
Packit d0f5c2
    uv = NATIVE_TO_UTF(*s) & UTF_START_MASK(UTF8SKIP(s));
Packit d0f5c2
Packit d0f5c2
    len--;
Packit d0f5c2
    s++;
Packit d0f5c2
Packit d0f5c2
    while (len--) {
Packit d0f5c2
        if (!UTF8_IS_CONTINUATION(*s)) {
Packit d0f5c2
            *rlen = s-ptr;
Packit d0f5c2
            return 0;
Packit d0f5c2
        }
Packit d0f5c2
        if (uv & UTF_ACCUMULATION_OVERFLOW_MASK)
Packit d0f5c2
            overflowed = 1;
Packit d0f5c2
        uv = UTF8_ACCUMULATE(uv, *s);
Packit d0f5c2
        s++;
Packit d0f5c2
    }
Packit d0f5c2
Packit d0f5c2
    *rlen = s-ptr;
Packit d0f5c2
Packit d0f5c2
    if (overflowed || *rlen > (STRLEN)UNISKIP(uv)) {
Packit d0f5c2
        return 0;
Packit d0f5c2
    }
Packit d0f5c2
Packit d0f5c2
    return uv;
Packit d0f5c2
}
Packit d0f5c2
Packit d0f5c2
#endif  /* CAN_USE_BASE_PERL */
Packit d0f5c2
Packit d0f5c2
static U8*
Packit d0f5c2
process_utf8(pTHX_ SV* dst, U8* s, U8* e, SV *check_sv,
Packit d0f5c2
             bool encode, bool strict, bool stop_at_partial)
Packit d0f5c2
{
Packit d0f5c2
    /* Copies the purportedly UTF-8 encoded string starting at 's' and ending
Packit d0f5c2
     * at 'e' - 1 to 'dst', checking as it goes along that the string actually
Packit d0f5c2
     * is valid UTF-8.  There are two levels of strictness checking.  If
Packit d0f5c2
     * 'strict' is FALSE, the string is checked for being well-formed UTF-8, as
Packit d0f5c2
     * extended by Perl.  Additionally, if 'strict' is TRUE, above-Unicode code
Packit d0f5c2
     * points, surrogates, and non-character code points are checked for.  When
Packit d0f5c2
     * invalid input is encountered, some action is taken, exactly what depends
Packit d0f5c2
     * on the flags in 'check_sv'.  'encode' gives if this is from an encode
Packit d0f5c2
     * operation (if TRUE), or a decode one.  This function returns the
Packit d0f5c2
     * position in 's' of the start of the next character beyond where it got
Packit d0f5c2
     * to.  If there were no problems, that will be 'e'.  If 'stop_at_partial'
Packit d0f5c2
     * is TRUE, if the final character before 'e' is incomplete, but valid as
Packit d0f5c2
     * far as is available, no action will be taken on that partial character,
Packit d0f5c2
     * and the return value will point to its first byte */
Packit d0f5c2
Packit d0f5c2
    UV uv;
Packit d0f5c2
    STRLEN ulen;
Packit d0f5c2
    SV *fallback_cb;
Packit d0f5c2
    int check;
Packit d0f5c2
    U8 *d;
Packit d0f5c2
    STRLEN dlen;
Packit d0f5c2
    char esc[UTF8_MAXLEN * 6 + 1];
Packit d0f5c2
    STRLEN i;
Packit d0f5c2
    const U32 flags = (strict)
Packit d0f5c2
                    ? UTF8_DISALLOW_ILLEGAL_INTERCHANGE
Packit d0f5c2
                    : UTF8_ALLOW_NON_STRICT;
Packit d0f5c2
Packit d0f5c2
    if (SvROK(check_sv)) {
Packit d0f5c2
	/* croak("UTF-8 decoder doesn't support callback CHECK"); */
Packit d0f5c2
	fallback_cb = check_sv;
Packit d0f5c2
	check = ENCODE_PERLQQ|ENCODE_LEAVE_SRC; /* same as perlqq */
Packit d0f5c2
    }
Packit d0f5c2
    else {
Packit d0f5c2
	fallback_cb = &PL_sv_undef;
Packit d0f5c2
	check = SvIV_nomg(check_sv);
Packit d0f5c2
    }
Packit d0f5c2
Packit d0f5c2
    SvPOK_only(dst);
Packit d0f5c2
    SvCUR_set(dst,0);
Packit d0f5c2
Packit d0f5c2
    dlen = (s && e && s < e) ? e-s+1 : 1;
Packit d0f5c2
    d = (U8 *) SvGROW(dst, dlen);
Packit d0f5c2
Packit d0f5c2
    stop_at_partial = stop_at_partial || (check & ENCODE_STOP_AT_PARTIAL);
Packit d0f5c2
Packit d0f5c2
    while (s < e) {
Packit d0f5c2
Packit d0f5c2
#ifdef CAN_USE_BASE_PERL    /* Use the much faster, portable implementation if
Packit d0f5c2
                               available */
Packit d0f5c2
Packit d0f5c2
        /* If there were no errors, this will be 'e'; otherwise it will point
Packit d0f5c2
         * to the first byte of the erroneous input */
Packit d0f5c2
        const U8* e_or_where_failed;
Packit d0f5c2
        bool valid = is_utf8_string_loc_flags(s, e - s, &e_or_where_failed, flags);
Packit d0f5c2
        STRLEN len = e_or_where_failed - s;
Packit d0f5c2
Packit d0f5c2
        /* Copy as far as was successful */
Packit d0f5c2
        Move(s, d, len, U8);
Packit d0f5c2
        d += len;
Packit d0f5c2
        s = (U8 *) e_or_where_failed;
Packit d0f5c2
Packit d0f5c2
        /* Are done if it was valid, or we are accepting partial characters and
Packit d0f5c2
         * the only error is that the final bytes form a partial character */
Packit d0f5c2
        if (    LIKELY(valid)
Packit d0f5c2
            || (   stop_at_partial
Packit d0f5c2
                && is_utf8_valid_partial_char_flags(s, e, flags)))
Packit d0f5c2
        {
Packit d0f5c2
            break;
Packit d0f5c2
        }
Packit d0f5c2
Packit d0f5c2
        /* Here, was not valid.  If is 'strict', and is legal extended UTF-8,
Packit d0f5c2
         * we know it is a code point whose value we can calculate, just not
Packit d0f5c2
         * one accepted under strict.  Otherwise, it is malformed in some way.
Packit d0f5c2
         * In either case, the system function can calculate either the code
Packit d0f5c2
         * point, or the best substitution for it */
Packit d0f5c2
        uv = utf8n_to_uvchr(s, e - s, &ulen, UTF8_ALLOW_ANY);
Packit d0f5c2
Packit d0f5c2
#else   /* Use code for earlier perls */
Packit d0f5c2
Packit d0f5c2
        ((void)sizeof(flags));  /* Avoid compiler warning */
Packit d0f5c2
Packit d0f5c2
        if (UTF8_IS_INVARIANT(*s)) {
Packit d0f5c2
            *d++ = *s++;
Packit d0f5c2
            continue;
Packit d0f5c2
        }
Packit d0f5c2
Packit d0f5c2
        uv = 0;
Packit d0f5c2
        ulen = 1;
Packit d0f5c2
        if (! UTF8_IS_CONTINUATION(*s)) {
Packit d0f5c2
            /* Not an invariant nor a continuation; must be a start byte.  (We
Packit d0f5c2
             * can't test for UTF8_IS_START as that excludes things like \xC0
Packit d0f5c2
             * which are start bytes, but always lead to overlongs */
Packit d0f5c2
Packit d0f5c2
            U8 skip = UTF8SKIP(s);
Packit d0f5c2
            if ((s + skip) > e) {
Packit d0f5c2
                /* just calculate ulen, in pathological cases can be smaller then e-s */
Packit d0f5c2
                if (e-s >= 2)
Packit d0f5c2
                    convert_utf8_multi_seq(s, e-s, &ulen);
Packit d0f5c2
                else
Packit d0f5c2
                    ulen = 1;
Packit d0f5c2
Packit d0f5c2
                if (stop_at_partial && ulen == (STRLEN)(e-s))
Packit d0f5c2
                    break;
Packit d0f5c2
Packit d0f5c2
                goto malformed_byte;
Packit d0f5c2
            }
Packit d0f5c2
Packit d0f5c2
            uv = convert_utf8_multi_seq(s, skip, &ulen);
Packit d0f5c2
            if (uv == 0)
Packit d0f5c2
                goto malformed_byte;
Packit d0f5c2
            else if (strict && !UNICODE_IS_STRICT(uv))
Packit d0f5c2
                goto malformed;
Packit d0f5c2
Packit d0f5c2
Packit d0f5c2
             /* Whole char is good */
Packit d0f5c2
             memcpy(d, s, skip);
Packit d0f5c2
             d += skip;
Packit d0f5c2
             s += skip;
Packit d0f5c2
             continue;
Packit d0f5c2
        }
Packit d0f5c2
Packit d0f5c2
        /* If we get here there is something wrong with alleged UTF-8 */
Packit d0f5c2
        /* uv is used only when encoding */
Packit d0f5c2
    malformed_byte:
Packit d0f5c2
        if (uv == 0)
Packit d0f5c2
            uv = (UV)*s;
Packit d0f5c2
        if (encode || ulen == 0)
Packit d0f5c2
            ulen = 1;
Packit d0f5c2
Packit d0f5c2
    malformed:
Packit d0f5c2
Packit d0f5c2
#endif  /* The two versions for processing come back together here, for the
Packit d0f5c2
         * error handling code.
Packit d0f5c2
         *
Packit d0f5c2
         * Here, we are looping through the input and found an error.
Packit d0f5c2
         * 'uv' is the code point in error if calculable, or the REPLACEMENT
Packit d0f5c2
         *      CHARACTER if not.
Packit d0f5c2
         * 'ulen' is how many bytes of input this iteration of the loop
Packit d0f5c2
         *        consumes */
Packit d0f5c2
Packit d0f5c2
        if (!encode && (check & (ENCODE_DIE_ON_ERR|ENCODE_WARN_ON_ERR|ENCODE_PERLQQ)))
Packit d0f5c2
            for (i=0; i
Packit d0f5c2
        if (check & ENCODE_DIE_ON_ERR){
Packit d0f5c2
            if (encode)
Packit d0f5c2
                Perl_croak(aTHX_ ERR_ENCODE_NOMAP, uv, (strict ? "UTF-8" : "utf8"));
Packit d0f5c2
            else
Packit d0f5c2
                Perl_croak(aTHX_ ERR_DECODE_STR_NOMAP, (strict ? "UTF-8" : "utf8"), esc);
Packit d0f5c2
        }
Packit d0f5c2
        if (check & ENCODE_WARN_ON_ERR){
Packit d0f5c2
            if (encode)
Packit d0f5c2
                Perl_warner(aTHX_ packWARN(WARN_UTF8),
Packit d0f5c2
                            ERR_ENCODE_NOMAP, uv, (strict ? "UTF-8" : "utf8"));
Packit d0f5c2
            else
Packit d0f5c2
                Perl_warner(aTHX_ packWARN(WARN_UTF8),
Packit d0f5c2
                            ERR_DECODE_STR_NOMAP, (strict ? "UTF-8" : "utf8"), esc);
Packit d0f5c2
        }
Packit d0f5c2
        if (check & ENCODE_RETURN_ON_ERR) {
Packit d0f5c2
                break;
Packit d0f5c2
        }
Packit d0f5c2
        if (check & (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){
Packit d0f5c2
            STRLEN sublen;
Packit d0f5c2
            char *substr;
Packit d0f5c2
            SV* subchar;
Packit d0f5c2
            if (encode) {
Packit d0f5c2
                subchar =
Packit d0f5c2
                    (fallback_cb != &PL_sv_undef)
Packit d0f5c2
                    ? do_fallback_cb(aTHX_ uv, fallback_cb)
Packit d0f5c2
                    : newSVpvf(check & ENCODE_PERLQQ
Packit d0f5c2
                        ? (ulen == 1 ? "\\x%02" UVXf : "\\x{%04" UVXf "}")
Packit d0f5c2
                        :  check & ENCODE_HTMLCREF ? "&#%" UVuf ";"
Packit d0f5c2
                        : "&#x%" UVxf ";", uv);
Packit d0f5c2
                substr = SvPV(subchar, sublen);
Packit d0f5c2
                if (SvUTF8(subchar) && sublen && !utf8_to_bytes((U8 *)substr, &sublen)) { /* make sure no decoded string gets in */
Packit d0f5c2
                    SvREFCNT_dec(subchar);
Packit d0f5c2
                    croak("Wide character");
Packit d0f5c2
                }
Packit d0f5c2
            } else {
Packit d0f5c2
                if (fallback_cb != &PL_sv_undef) {
Packit d0f5c2
                    /* in decode mode we have sequence of wrong bytes */
Packit d0f5c2
                    subchar = do_bytes_fallback_cb(aTHX_ s, ulen, fallback_cb);
Packit d0f5c2
                } else {
Packit d0f5c2
                    char *ptr = esc;
Packit d0f5c2
                    /* ENCODE_PERLQQ is already stored in esc */
Packit d0f5c2
                    if (check & (ENCODE_HTMLCREF|ENCODE_XMLCREF))
Packit d0f5c2
                        for (i=0; i
Packit d0f5c2
                    subchar = newSVpvn(esc, strlen(esc));
Packit d0f5c2
                }
Packit d0f5c2
                substr = SvPVutf8(subchar, sublen);
Packit d0f5c2
            }
Packit d0f5c2
            dlen += sublen - ulen;
Packit d0f5c2
            SvCUR_set(dst, d-(U8 *)SvPVX(dst));
Packit d0f5c2
            *SvEND(dst) = '\0';
Packit d0f5c2
            sv_catpvn(dst, substr, sublen);
Packit d0f5c2
            SvREFCNT_dec(subchar);
Packit d0f5c2
            d = (U8 *) SvGROW(dst, dlen) + SvCUR(dst);
Packit d0f5c2
        } else {
Packit d0f5c2
            STRLEN fbcharlen = strlen(FBCHAR_UTF8);
Packit d0f5c2
            dlen += fbcharlen - ulen;
Packit d0f5c2
            if (SvLEN(dst) < dlen) {
Packit d0f5c2
                SvCUR_set(dst, d-(U8 *)SvPVX(dst));
Packit d0f5c2
                d = (U8 *) sv_grow(dst, dlen) + SvCUR(dst);
Packit d0f5c2
            }
Packit d0f5c2
            memcpy(d, FBCHAR_UTF8, fbcharlen);
Packit d0f5c2
            d += fbcharlen;
Packit d0f5c2
        }
Packit d0f5c2
        s += ulen;
Packit d0f5c2
    }
Packit d0f5c2
    SvCUR_set(dst, d-(U8 *)SvPVX(dst));
Packit d0f5c2
    *SvEND(dst) = '\0';
Packit d0f5c2
Packit d0f5c2
    return s;
Packit d0f5c2
}
Packit d0f5c2
Packit d0f5c2
Packit d0f5c2
MODULE = Encode		PACKAGE = Encode::utf8	PREFIX = Method_
Packit d0f5c2
Packit d0f5c2
PROTOTYPES: DISABLE
Packit d0f5c2
Packit d0f5c2
void
Packit d0f5c2
Method_decode(obj,src,check_sv = &PL_sv_no)
Packit d0f5c2
SV *	obj
Packit d0f5c2
SV *	src
Packit d0f5c2
SV *	check_sv
Packit d0f5c2
PREINIT:
Packit d0f5c2
    STRLEN slen;
Packit d0f5c2
    U8 *s;
Packit d0f5c2
    U8 *e;
Packit d0f5c2
    SV *dst;
Packit d0f5c2
    bool renewed = 0;
Packit d0f5c2
    int check;
Packit d0f5c2
    bool modify;
Packit d0f5c2
    dSP;
Packit d0f5c2
INIT:
Packit d0f5c2
    SvGETMAGIC(src);
Packit d0f5c2
    SvGETMAGIC(check_sv);
Packit d0f5c2
    check = SvROK(check_sv) ? ENCODE_PERLQQ|ENCODE_LEAVE_SRC : SvIV_nomg(check_sv);
Packit d0f5c2
    modify = (check && !(check & ENCODE_LEAVE_SRC));
Packit d0f5c2
PPCODE:
Packit d0f5c2
    if (!SvOK(src))
Packit d0f5c2
        XSRETURN_UNDEF;
Packit d0f5c2
    s = modify ? (U8 *)SvPV_force_nomg(src, slen) : (U8 *)SvPV_nomg(src, slen);
Packit d0f5c2
    if (SvUTF8(src))
Packit d0f5c2
        utf8_safe_downgrade(aTHX_ &src, &s, &slen, modify);
Packit d0f5c2
    e = s+slen;
Packit d0f5c2
Packit d0f5c2
    /*
Packit d0f5c2
     * PerlIO check -- we assume the object is of PerlIO if renewed
Packit d0f5c2
     */
Packit d0f5c2
    ENTER; SAVETMPS;
Packit d0f5c2
    PUSHMARK(sp);
Packit d0f5c2
    XPUSHs(obj);
Packit d0f5c2
    PUTBACK;
Packit d0f5c2
    if (call_method("renewed",G_SCALAR) == 1) {
Packit d0f5c2
    SPAGAIN;
Packit d0f5c2
    renewed = (bool)POPi;
Packit d0f5c2
    PUTBACK;
Packit d0f5c2
#if 0
Packit d0f5c2
    fprintf(stderr, "renewed == %d\n", renewed);
Packit d0f5c2
#endif
Packit d0f5c2
    }
Packit d0f5c2
    FREETMPS; LEAVE;
Packit d0f5c2
    /* end PerlIO check */
Packit d0f5c2
Packit d0f5c2
    dst = sv_2mortal(newSV(slen>0?slen:1)); /* newSV() abhors 0 -- inaba */
Packit d0f5c2
    s = process_utf8(aTHX_ dst, s, e, check_sv, 0, strict_utf8(aTHX_ obj), renewed);
Packit d0f5c2
Packit d0f5c2
    /* Clear out translated part of source unless asked not to */
Packit d0f5c2
    if (modify) {
Packit d0f5c2
    slen = e-s;
Packit d0f5c2
    if (slen) {
Packit d0f5c2
        sv_setpvn(src, (char*)s, slen);
Packit d0f5c2
    }
Packit d0f5c2
    SvCUR_set(src, slen);
Packit d0f5c2
    SvSETMAGIC(src);
Packit d0f5c2
    }
Packit d0f5c2
    SvUTF8_on(dst);
Packit d0f5c2
    if (SvTAINTED(src)) SvTAINTED_on(dst); /* propagate taintedness */
Packit d0f5c2
    ST(0) = dst;
Packit d0f5c2
    XSRETURN(1);
Packit d0f5c2
Packit d0f5c2
void
Packit d0f5c2
Method_encode(obj,src,check_sv = &PL_sv_no)
Packit d0f5c2
SV *	obj
Packit d0f5c2
SV *	src
Packit d0f5c2
SV *	check_sv
Packit d0f5c2
PREINIT:
Packit d0f5c2
    STRLEN slen;
Packit d0f5c2
    U8 *s;
Packit d0f5c2
    U8 *e;
Packit d0f5c2
    SV *dst;
Packit d0f5c2
    int check;
Packit d0f5c2
    bool modify;
Packit d0f5c2
INIT:
Packit d0f5c2
    SvGETMAGIC(src);
Packit d0f5c2
    SvGETMAGIC(check_sv);
Packit d0f5c2
    check = SvROK(check_sv) ? ENCODE_PERLQQ|ENCODE_LEAVE_SRC : SvIV_nomg(check_sv);
Packit d0f5c2
    modify = (check && !(check & ENCODE_LEAVE_SRC));
Packit d0f5c2
PPCODE:
Packit d0f5c2
    if (!SvOK(src))
Packit d0f5c2
        XSRETURN_UNDEF;
Packit d0f5c2
    s = modify ? (U8 *)SvPV_force_nomg(src, slen) : (U8 *)SvPV_nomg(src, slen);
Packit d0f5c2
    e = s+slen;
Packit d0f5c2
    dst = sv_2mortal(newSV(slen>0?slen:1)); /* newSV() abhors 0 -- inaba */
Packit d0f5c2
    if (SvUTF8(src)) {
Packit d0f5c2
    /* Already encoded */
Packit d0f5c2
    if (strict_utf8(aTHX_ obj)) {
Packit d0f5c2
        s = process_utf8(aTHX_ dst, s, e, check_sv, 1, 1, 0);
Packit d0f5c2
    }
Packit d0f5c2
        else {
Packit d0f5c2
            /* trust it and just copy the octets */
Packit d0f5c2
    	    sv_setpvn(dst,(char *)s,(e-s));
Packit d0f5c2
        s = e;
Packit d0f5c2
        }
Packit d0f5c2
    }
Packit d0f5c2
    else {
Packit d0f5c2
        /* Native bytes - can always encode */
Packit d0f5c2
        U8 *d = (U8 *) SvGROW(dst, 2*slen+1); /* +1 or assertion will botch */
Packit d0f5c2
        while (s < e) {
Packit d0f5c2
#ifdef append_utf8_from_native_byte
Packit d0f5c2
            append_utf8_from_native_byte(*s, &d);
Packit d0f5c2
            s++;
Packit d0f5c2
#else
Packit d0f5c2
            UV uv = NATIVE_TO_UNI((UV) *s);
Packit d0f5c2
            s++; /* Above expansion of NATIVE_TO_UNI() is safer this way. */
Packit d0f5c2
            if (UNI_IS_INVARIANT(uv))
Packit d0f5c2
                *d++ = (U8)UTF_TO_NATIVE(uv);
Packit d0f5c2
            else {
Packit d0f5c2
                *d++ = (U8)UTF8_EIGHT_BIT_HI(uv);
Packit d0f5c2
                *d++ = (U8)UTF8_EIGHT_BIT_LO(uv);
Packit d0f5c2
            }
Packit d0f5c2
#endif
Packit d0f5c2
        }
Packit d0f5c2
        SvCUR_set(dst, d- (U8 *)SvPVX(dst));
Packit d0f5c2
        *SvEND(dst) = '\0';
Packit d0f5c2
    }
Packit d0f5c2
Packit d0f5c2
    /* Clear out translated part of source unless asked not to */
Packit d0f5c2
    if (modify) {
Packit d0f5c2
    slen = e-s;
Packit d0f5c2
    if (slen) {
Packit d0f5c2
        sv_setpvn(src, (char*)s, slen);
Packit d0f5c2
    }
Packit d0f5c2
    SvCUR_set(src, slen);
Packit d0f5c2
    SvSETMAGIC(src);
Packit d0f5c2
    }
Packit d0f5c2
    SvPOK_only(dst);
Packit d0f5c2
    SvUTF8_off(dst);
Packit d0f5c2
    if (SvTAINTED(src)) SvTAINTED_on(dst); /* propagate taintedness */
Packit d0f5c2
    ST(0) = dst;
Packit d0f5c2
    XSRETURN(1);
Packit d0f5c2
Packit d0f5c2
MODULE = Encode		PACKAGE = Encode::XS	PREFIX = Method_
Packit d0f5c2
Packit d0f5c2
PROTOTYPES: DISABLE
Packit d0f5c2
Packit d0f5c2
SV *
Packit d0f5c2
Method_renew(obj)
Packit d0f5c2
SV *	obj
Packit d0f5c2
CODE:
Packit d0f5c2
    PERL_UNUSED_VAR(obj);
Packit d0f5c2
    RETVAL = newSVsv(obj);
Packit d0f5c2
OUTPUT:
Packit d0f5c2
    RETVAL
Packit d0f5c2
Packit d0f5c2
int
Packit d0f5c2
Method_renewed(obj)
Packit d0f5c2
SV *    obj
Packit d0f5c2
CODE:
Packit d0f5c2
    RETVAL = 0;
Packit d0f5c2
    PERL_UNUSED_VAR(obj);
Packit d0f5c2
OUTPUT:
Packit d0f5c2
    RETVAL
Packit d0f5c2
Packit d0f5c2
SV *
Packit d0f5c2
Method_name(obj)
Packit d0f5c2
SV *	obj
Packit d0f5c2
PREINIT:
Packit d0f5c2
    encode_t *enc;
Packit d0f5c2
INIT:
Packit d0f5c2
    enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
Packit d0f5c2
CODE:
Packit d0f5c2
    RETVAL = newSVpvn(enc->name[0], strlen(enc->name[0]));
Packit d0f5c2
OUTPUT:
Packit d0f5c2
    RETVAL
Packit d0f5c2
Packit d0f5c2
bool
Packit d0f5c2
Method_cat_decode(obj, dst, src, off, term, check_sv = &PL_sv_no)
Packit d0f5c2
SV *	obj
Packit d0f5c2
SV *	dst
Packit d0f5c2
SV *	src
Packit d0f5c2
SV *	off
Packit d0f5c2
SV *	term
Packit d0f5c2
SV *    check_sv
Packit d0f5c2
PREINIT:
Packit d0f5c2
    int check;
Packit d0f5c2
    SV *fallback_cb;
Packit d0f5c2
    bool modify;
Packit d0f5c2
    encode_t *enc;
Packit d0f5c2
    STRLEN offset;
Packit d0f5c2
    int code = 0;
Packit d0f5c2
    U8 *s;
Packit d0f5c2
    STRLEN slen;
Packit d0f5c2
    SV *tmp;
Packit d0f5c2
INIT:
Packit d0f5c2
    SvGETMAGIC(src);
Packit d0f5c2
    SvGETMAGIC(check_sv);
Packit d0f5c2
    check = SvROK(check_sv) ? ENCODE_PERLQQ|ENCODE_LEAVE_SRC : SvIV_nomg(check_sv);
Packit d0f5c2
    fallback_cb = SvROK(check_sv) ? check_sv : &PL_sv_undef;
Packit d0f5c2
    modify = (check && !(check & ENCODE_LEAVE_SRC));
Packit d0f5c2
    enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
Packit d0f5c2
    offset = (STRLEN)SvIV(off);
Packit d0f5c2
CODE:
Packit d0f5c2
    if (!SvOK(src))
Packit d0f5c2
        XSRETURN_NO;
Packit d0f5c2
    s = modify ? (U8 *)SvPV_force_nomg(src, slen) : (U8 *)SvPV_nomg(src, slen);
Packit d0f5c2
    if (SvUTF8(src))
Packit d0f5c2
        utf8_safe_downgrade(aTHX_ &src, &s, &slen, modify);
Packit d0f5c2
    tmp = encode_method(aTHX_ enc, enc->t_utf8, src, s, slen, check,
Packit d0f5c2
                &offset, term, &code, fallback_cb);
Packit d0f5c2
    sv_catsv(dst, tmp);
Packit d0f5c2
    SvREFCNT_dec(tmp);
Packit d0f5c2
    SvIV_set(off, (IV)offset);
Packit d0f5c2
    RETVAL = (code == ENCODE_FOUND_TERM);
Packit d0f5c2
OUTPUT:
Packit d0f5c2
    RETVAL
Packit d0f5c2
Packit d0f5c2
SV *
Packit d0f5c2
Method_decode(obj,src,check_sv = &PL_sv_no)
Packit d0f5c2
SV *	obj
Packit d0f5c2
SV *	src
Packit d0f5c2
SV *	check_sv
Packit d0f5c2
PREINIT:
Packit d0f5c2
    int check;
Packit d0f5c2
    SV *fallback_cb;
Packit d0f5c2
    bool modify;
Packit d0f5c2
    encode_t *enc;
Packit d0f5c2
    U8 *s;
Packit d0f5c2
    STRLEN slen;
Packit d0f5c2
INIT:
Packit d0f5c2
    SvGETMAGIC(src);
Packit d0f5c2
    SvGETMAGIC(check_sv);
Packit d0f5c2
    check = SvROK(check_sv) ? ENCODE_PERLQQ|ENCODE_LEAVE_SRC : SvIV_nomg(check_sv);
Packit d0f5c2
    fallback_cb = SvROK(check_sv) ? check_sv : &PL_sv_undef;
Packit d0f5c2
    modify = (check && !(check & ENCODE_LEAVE_SRC));
Packit d0f5c2
    enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
Packit d0f5c2
CODE:
Packit d0f5c2
    if (!SvOK(src))
Packit d0f5c2
        XSRETURN_UNDEF;
Packit d0f5c2
    s = modify ? (U8 *)SvPV_force_nomg(src, slen) : (U8 *)SvPV_nomg(src, slen);
Packit d0f5c2
    if (SvUTF8(src))
Packit d0f5c2
        utf8_safe_downgrade(aTHX_ &src, &s, &slen, modify);
Packit d0f5c2
    RETVAL = encode_method(aTHX_ enc, enc->t_utf8, src, s, slen, check,
Packit d0f5c2
              NULL, Nullsv, NULL, fallback_cb);
Packit d0f5c2
    SvUTF8_on(RETVAL);
Packit d0f5c2
OUTPUT:
Packit d0f5c2
    RETVAL
Packit d0f5c2
Packit d0f5c2
SV *
Packit d0f5c2
Method_encode(obj,src,check_sv = &PL_sv_no)
Packit d0f5c2
SV *	obj
Packit d0f5c2
SV *	src
Packit d0f5c2
SV *	check_sv
Packit d0f5c2
PREINIT:
Packit d0f5c2
    int check;
Packit d0f5c2
    SV *fallback_cb;
Packit d0f5c2
    bool modify;
Packit d0f5c2
    encode_t *enc;
Packit d0f5c2
    U8 *s;
Packit d0f5c2
    STRLEN slen;
Packit d0f5c2
INIT:
Packit d0f5c2
    SvGETMAGIC(src);
Packit d0f5c2
    SvGETMAGIC(check_sv);
Packit d0f5c2
    check = SvROK(check_sv) ? ENCODE_PERLQQ|ENCODE_LEAVE_SRC : SvIV_nomg(check_sv);
Packit d0f5c2
    fallback_cb = SvROK(check_sv) ? check_sv : &PL_sv_undef;
Packit d0f5c2
    modify = (check && !(check & ENCODE_LEAVE_SRC));
Packit d0f5c2
    enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
Packit d0f5c2
CODE:
Packit d0f5c2
    if (!SvOK(src))
Packit d0f5c2
        XSRETURN_UNDEF;
Packit d0f5c2
    s = modify ? (U8 *)SvPV_force_nomg(src, slen) : (U8 *)SvPV_nomg(src, slen);
Packit d0f5c2
    if (!SvUTF8(src))
Packit d0f5c2
        utf8_safe_upgrade(aTHX_ &src, &s, &slen, modify);
Packit d0f5c2
    RETVAL = encode_method(aTHX_ enc, enc->f_utf8, src, s, slen, check,
Packit d0f5c2
              NULL, Nullsv, NULL, fallback_cb);
Packit d0f5c2
OUTPUT:
Packit d0f5c2
    RETVAL
Packit d0f5c2
Packit d0f5c2
bool
Packit d0f5c2
Method_needs_lines(obj)
Packit d0f5c2
SV *	obj
Packit d0f5c2
CODE:
Packit d0f5c2
    PERL_UNUSED_VAR(obj);
Packit d0f5c2
    RETVAL = FALSE;
Packit d0f5c2
OUTPUT:
Packit d0f5c2
    RETVAL
Packit d0f5c2
Packit d0f5c2
bool
Packit d0f5c2
Method_perlio_ok(obj)
Packit d0f5c2
SV *	obj
Packit d0f5c2
PREINIT:
Packit d0f5c2
    SV *sv;
Packit d0f5c2
CODE:
Packit d0f5c2
    PERL_UNUSED_VAR(obj);
Packit d0f5c2
    sv = eval_pv("require PerlIO::encoding", 0);
Packit d0f5c2
    RETVAL = SvTRUE(sv);
Packit d0f5c2
OUTPUT:
Packit d0f5c2
    RETVAL
Packit d0f5c2
Packit d0f5c2
SV *
Packit d0f5c2
Method_mime_name(obj)
Packit d0f5c2
SV *	obj
Packit d0f5c2
PREINIT:
Packit d0f5c2
    encode_t *enc;
Packit d0f5c2
INIT:
Packit d0f5c2
    enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
Packit d0f5c2
CODE:
Packit d0f5c2
    ENTER;
Packit d0f5c2
    SAVETMPS;
Packit d0f5c2
    PUSHMARK(sp);
Packit d0f5c2
    XPUSHs(sv_2mortal(newSVpvn(enc->name[0], strlen(enc->name[0]))));
Packit d0f5c2
    PUTBACK;
Packit d0f5c2
    call_pv("Encode::MIME::Name::get_mime_name", G_SCALAR);
Packit d0f5c2
    SPAGAIN;
Packit d0f5c2
    RETVAL = newSVsv(POPs);
Packit d0f5c2
    PUTBACK;
Packit d0f5c2
    FREETMPS;
Packit d0f5c2
    LEAVE;
Packit d0f5c2
OUTPUT:
Packit d0f5c2
    RETVAL
Packit d0f5c2
Packit d0f5c2
MODULE = Encode         PACKAGE = Encode
Packit d0f5c2
Packit d0f5c2
PROTOTYPES: ENABLE
Packit d0f5c2
Packit d0f5c2
I32
Packit d0f5c2
_bytes_to_utf8(sv, ...)
Packit d0f5c2
SV *    sv
Packit d0f5c2
PREINIT:
Packit d0f5c2
    SV * encoding;
Packit d0f5c2
INIT:
Packit d0f5c2
    encoding = items == 2 ? ST(1) : Nullsv;
Packit d0f5c2
CODE:
Packit d0f5c2
    if (encoding)
Packit d0f5c2
    RETVAL = _encoded_bytes_to_utf8(sv, SvPV_nolen(encoding));
Packit d0f5c2
    else {
Packit d0f5c2
    STRLEN len;
Packit d0f5c2
    U8*    s = (U8*)SvPV(sv, len);
Packit d0f5c2
    U8*    converted;
Packit d0f5c2
Packit d0f5c2
    converted = bytes_to_utf8(s, &len;; /* This allocs */
Packit d0f5c2
    sv_setpvn(sv, (char *)converted, len);
Packit d0f5c2
    SvUTF8_on(sv); /* XXX Should we? */
Packit d0f5c2
    Safefree(converted);                /* ... so free it */
Packit d0f5c2
    RETVAL = len;
Packit d0f5c2
    }
Packit d0f5c2
OUTPUT:
Packit d0f5c2
    RETVAL
Packit d0f5c2
Packit d0f5c2
I32
Packit d0f5c2
_utf8_to_bytes(sv, ...)
Packit d0f5c2
SV *    sv
Packit d0f5c2
PREINIT:
Packit d0f5c2
    SV * to;
Packit d0f5c2
    SV * check;
Packit d0f5c2
INIT:
Packit d0f5c2
    to    = items > 1 ? ST(1) : Nullsv;
Packit d0f5c2
    check = items > 2 ? ST(2) : Nullsv;
Packit d0f5c2
CODE:
Packit d0f5c2
    if (to) {
Packit d0f5c2
    RETVAL = _encoded_utf8_to_bytes(sv, SvPV_nolen(to));
Packit d0f5c2
    } else {
Packit d0f5c2
    STRLEN len;
Packit d0f5c2
    U8 *s = (U8*)SvPV(sv, len);
Packit d0f5c2
Packit d0f5c2
    RETVAL = 0;
Packit d0f5c2
    if (SvTRUE(check)) {
Packit d0f5c2
        /* Must do things the slow way */
Packit d0f5c2
        U8 *dest;
Packit d0f5c2
            /* We need a copy to pass to check() */
Packit d0f5c2
        U8 *src  = s;
Packit d0f5c2
        U8 *send = s + len;
Packit d0f5c2
        U8 *d0;
Packit d0f5c2
Packit d0f5c2
        New(83, dest, len, U8); /* I think */
Packit d0f5c2
        d0 = dest;
Packit d0f5c2
Packit d0f5c2
        while (s < send) {
Packit d0f5c2
                if (*s < 0x80){
Packit d0f5c2
            *dest++ = *s++;
Packit d0f5c2
                } else {
Packit d0f5c2
            STRLEN ulen;
Packit d0f5c2
            UV uv = *s++;
Packit d0f5c2
Packit d0f5c2
            /* Have to do it all ourselves because of error routine,
Packit d0f5c2
               aargh. */
Packit d0f5c2
            if (!(uv & 0x40)){ goto failure; }
Packit d0f5c2
            if      (!(uv & 0x20)) { ulen = 2;  uv &= 0x1f; }
Packit d0f5c2
            else if (!(uv & 0x10)) { ulen = 3;  uv &= 0x0f; }
Packit d0f5c2
            else if (!(uv & 0x08)) { ulen = 4;  uv &= 0x07; }
Packit d0f5c2
            else if (!(uv & 0x04)) { ulen = 5;  uv &= 0x03; }
Packit d0f5c2
            else if (!(uv & 0x02)) { ulen = 6;  uv &= 0x01; }
Packit d0f5c2
            else if (!(uv & 0x01)) { ulen = 7;  uv = 0; }
Packit d0f5c2
            else                   { ulen = 13; uv = 0; }
Packit d0f5c2
        
Packit d0f5c2
            /* Note change to utf8.c variable naming, for variety */
Packit d0f5c2
            while (ulen--) {
Packit d0f5c2
            if ((*s & 0xc0) != 0x80){
Packit d0f5c2
                goto failure;
Packit d0f5c2
            } else {
Packit d0f5c2
                uv = (uv << 6) | (*s++ & 0x3f);
Packit d0f5c2
            }
Packit d0f5c2
          }
Packit d0f5c2
          if (uv > 256) {
Packit d0f5c2
          failure:
Packit d0f5c2
              call_failure(check, s, dest, src);
Packit d0f5c2
              /* Now what happens? */
Packit d0f5c2
          }
Packit d0f5c2
          *dest++ = (U8)uv;
Packit d0f5c2
        }
Packit d0f5c2
        }
Packit d0f5c2
        RETVAL = dest - d0;
Packit d0f5c2
        sv_usepvn(sv, (char *)dest, RETVAL);
Packit d0f5c2
        SvUTF8_off(sv);
Packit d0f5c2
    } else {
Packit d0f5c2
        RETVAL = (utf8_to_bytes(s, &len) ? len : 0);
Packit d0f5c2
    }
Packit d0f5c2
    }
Packit d0f5c2
OUTPUT:
Packit d0f5c2
    RETVAL
Packit d0f5c2
Packit d0f5c2
bool
Packit d0f5c2
is_utf8(sv, check = 0)
Packit d0f5c2
SV *	sv
Packit d0f5c2
int	check
Packit d0f5c2
PREINIT:
Packit d0f5c2
    char *str;
Packit d0f5c2
    STRLEN len;
Packit d0f5c2
CODE:
Packit d0f5c2
    SvGETMAGIC(sv); /* SvGETMAGIC() can modify SvOK flag */
Packit d0f5c2
    str = SvOK(sv) ? SvPV_nomg(sv, len) : NULL; /* SvPV() can modify SvUTF8 flag */
Packit d0f5c2
    RETVAL = SvUTF8(sv) ? TRUE : FALSE;
Packit d0f5c2
    if (RETVAL && check && (!str || !is_utf8_string((U8 *)str, len)))
Packit d0f5c2
        RETVAL = FALSE;
Packit d0f5c2
OUTPUT:
Packit d0f5c2
    RETVAL
Packit d0f5c2
Packit d0f5c2
SV *
Packit d0f5c2
_utf8_on(sv)
Packit d0f5c2
SV *	sv
Packit d0f5c2
CODE:
Packit d0f5c2
    SvGETMAGIC(sv);
Packit d0f5c2
    if (!SvTAINTED(sv) && SvPOKp(sv)) {
Packit d0f5c2
        if (SvTHINKFIRST(sv)) sv_force_normal(sv);
Packit d0f5c2
        RETVAL = boolSV(SvUTF8(sv));
Packit d0f5c2
        SvUTF8_on(sv);
Packit d0f5c2
        SvSETMAGIC(sv);
Packit d0f5c2
    } else {
Packit d0f5c2
        RETVAL = &PL_sv_undef;
Packit d0f5c2
    }
Packit d0f5c2
OUTPUT:
Packit d0f5c2
    RETVAL
Packit d0f5c2
Packit d0f5c2
SV *
Packit d0f5c2
_utf8_off(sv)
Packit d0f5c2
SV *	sv
Packit d0f5c2
CODE:
Packit d0f5c2
    SvGETMAGIC(sv);
Packit d0f5c2
    if (!SvTAINTED(sv) && SvPOKp(sv)) {
Packit d0f5c2
        if (SvTHINKFIRST(sv)) sv_force_normal(sv);
Packit d0f5c2
        RETVAL = boolSV(SvUTF8(sv));
Packit d0f5c2
        SvUTF8_off(sv);
Packit d0f5c2
        SvSETMAGIC(sv);
Packit d0f5c2
    } else {
Packit d0f5c2
        RETVAL = &PL_sv_undef;
Packit d0f5c2
    }
Packit d0f5c2
OUTPUT:
Packit d0f5c2
    RETVAL
Packit d0f5c2
Packit d0f5c2
void
Packit d0f5c2
onBOOT()
Packit d0f5c2
CODE:
Packit d0f5c2
{
Packit d0f5c2
#include "def_t.exh"
Packit d0f5c2
}
Packit d0f5c2
Packit d0f5c2
BOOT:
Packit d0f5c2
{
Packit d0f5c2
    HV *stash = gv_stashpvn("Encode", strlen("Encode"), GV_ADD);
Packit d0f5c2
    newCONSTSUB(stash, "DIE_ON_ERR", newSViv(ENCODE_DIE_ON_ERR));
Packit d0f5c2
    newCONSTSUB(stash, "WARN_ON_ERR", newSViv(ENCODE_WARN_ON_ERR));
Packit d0f5c2
    newCONSTSUB(stash, "RETURN_ON_ERR", newSViv(ENCODE_RETURN_ON_ERR));
Packit d0f5c2
    newCONSTSUB(stash, "LEAVE_SRC", newSViv(ENCODE_LEAVE_SRC));
Packit d0f5c2
    newCONSTSUB(stash, "PERLQQ", newSViv(ENCODE_PERLQQ));
Packit d0f5c2
    newCONSTSUB(stash, "HTMLCREF", newSViv(ENCODE_HTMLCREF));
Packit d0f5c2
    newCONSTSUB(stash, "XMLCREF", newSViv(ENCODE_XMLCREF));
Packit d0f5c2
    newCONSTSUB(stash, "STOP_AT_PARTIAL", newSViv(ENCODE_STOP_AT_PARTIAL));
Packit d0f5c2
    newCONSTSUB(stash, "FB_DEFAULT", newSViv(ENCODE_FB_DEFAULT));
Packit d0f5c2
    newCONSTSUB(stash, "FB_CROAK", newSViv(ENCODE_FB_CROAK));
Packit d0f5c2
    newCONSTSUB(stash, "FB_QUIET", newSViv(ENCODE_FB_QUIET));
Packit d0f5c2
    newCONSTSUB(stash, "FB_WARN", newSViv(ENCODE_FB_WARN));
Packit d0f5c2
    newCONSTSUB(stash, "FB_PERLQQ", newSViv(ENCODE_FB_PERLQQ));
Packit d0f5c2
    newCONSTSUB(stash, "FB_HTMLCREF", newSViv(ENCODE_FB_HTMLCREF));
Packit d0f5c2
    newCONSTSUB(stash, "FB_XMLCREF", newSViv(ENCODE_FB_XMLCREF));
Packit d0f5c2
}