2 $Id: Unicode.xs,v 1.7 2003/02/20 14:42:34 dankogai Exp dankogai $
5 #define PERL_NO_GET_CONTEXT
10 #include "../Encode/encode.h"
14 #define BOM16LE 0xFFFe
15 #define BOM32LE 0xFFFe0000
16 #define issurrogate(x) (0xD800 <= (x) && (x) <= 0xDFFF )
17 #define isHiSurrogate(x) (0xD800 <= (x) && (x) < 0xDC00 )
18 #define isLoSurrogate(x) (0xDC00 <= (x) && (x) <= 0xDFFF )
19 #define invalid_ucs2(x) ( issurrogate(x) || 0xFFFF < (x) )
22 enc_unpack(pTHX_ U8 **sp,U8 *e,STRLEN size,U8 endian)
27 croak("Partial character %c",(char) endian);
47 croak("Unknown endian %c",(char) endian);
55 enc_pack(pTHX_ SV *result,STRLEN size,U8 endian,UV value)
57 U8 *d = (U8 *)SvGROW(result,SvCUR(result)+size);
62 SvCUR_set(result,SvCUR(result)+size);
64 *d++ = (U8)(value & 0xFF);
70 SvCUR_set(result,SvCUR(result)+size);
73 *--d = (U8)(value & 0xFF);
78 croak("Unknown endian %c",(char) endian);
83 MODULE = Encode::Unicode PACKAGE = Encode::Unicode
87 #define attr(k, l) (hv_exists((HV *)SvRV(obj),k,l) ? \
88 *hv_fetch((HV *)SvRV(obj),k,l,0) : &PL_sv_undef)
91 decode_xs(obj, str, check = 0)
97 U8 endian = *((U8 *)SvPV_nolen(attr("endian", 6)));
98 int size = SvIV(attr("size", 4));
99 int ucs2 = SvTRUE(attr("ucs2", 4));
100 int clone = SvTRUE(attr("clone", 5));
101 SV *result = newSVpvn("",0);
103 U8 *s = (U8 *)SvPVbyte(str,ulen);
104 U8 *e = (U8 *)SvEND(str);
105 ST(0) = sv_2mortal(result);
108 if (!endian && s+size <= e) {
110 endian = (size == 4) ? 'N' : 'n';
111 bom = enc_unpack(aTHX_ &s,e,size,endian);
113 if (bom == BOM16LE) {
116 else if (bom == BOM32LE) {
120 croak("%"SVf":Unrecognised BOM %"UVxf,
121 *hv_fetch((HV *)SvRV(obj),"Name",4,0),
126 /* Update endian for next sequence */
128 hv_store((HV *)SvRV(obj),"endian",6,newSVpv((char *)&endian,1),0);
132 while (s < e && s+size <= e) {
133 UV ord = enc_unpack(aTHX_ &s,e,size,endian);
135 if (size != 4 && invalid_ucs2(ord)) {
138 croak("%"SVf":no surrogates allowed %"UVxf,
139 *hv_fetch((HV *)SvRV(obj),"Name",4,0),
143 /* skip the next one as well */
144 enc_unpack(aTHX_ &s,e,size,endian);
150 if (!isHiSurrogate(ord)) {
151 croak("%"SVf":Malformed HI surrogate %"UVxf,
152 *hv_fetch((HV *)SvRV(obj),"Name",4,0),
156 /* Partial character */
157 s -= size; /* back up to 1st half */
158 break; /* And exit loop */
160 lo = enc_unpack(aTHX_ &s,e,size,endian);
161 if (!isLoSurrogate(lo)){
162 croak("%"SVf":Malformed LO surrogate %"UVxf,
163 *hv_fetch((HV *)SvRV(obj),"Name",4,0),
166 ord = 0x10000 + ((ord - 0xD800) << 10) + (lo - 0xDC00);
169 d = (U8 *) SvGROW(result,SvCUR(result)+UTF8_MAXLEN+1);
170 d = uvuni_to_utf8_flags(d+SvCUR(result), ord, 0);
171 SvCUR_set(result,d - (U8 *)SvPVX(result));
174 Perl_warner(aTHX_ packWARN(WARN_UTF8),"%"SVf":Partial character",
175 *hv_fetch((HV *)SvRV(obj),"Name",4,0));
177 if (check && !(check & ENCODE_LEAVE_SRC)){
179 Move(s,SvPVX(str),e-s,U8);
180 SvCUR_set(str,(e-s));
191 encode_xs(obj, utf8, check = 0)
197 U8 endian = *((U8 *)SvPV_nolen(attr("endian", 6)));
198 int size = SvIV(attr("size", 4));
199 int ucs2 = SvTRUE(attr("ucs2", 4));
200 int clone = SvTRUE(attr("clone", 5));
201 SV *result = newSVpvn("",0);
203 U8 *s = (U8 *)SvPVutf8(utf8,ulen);
204 U8 *e = (U8 *)SvEND(utf8);
205 ST(0) = sv_2mortal(result);
207 endian = (size == 4) ? 'N' : 'n';
208 enc_pack(aTHX_ result,size,endian,BOM_BE);
210 /* Update endian for next sequence */
212 hv_store((HV *)SvRV(obj),"endian",6,newSVpv((char *)&endian,1),0);
216 while (s < e && s+UTF8SKIP(s) <= e) {
218 UV ord = utf8n_to_uvuni(s, e-s, &len, 0);
220 if (size != 4 && invalid_ucs2(ord)) {
221 if (!issurrogate(ord)){
224 croak("%"SVf":code point \"\\x{%"UVxf"}\" too high",
225 *hv_fetch((HV *)SvRV(obj),"Name",4,0),ord);
227 enc_pack(aTHX_ result,size,endian,FBCHAR);
229 UV hi = ((ord - 0x10000) >> 10) + 0xD800;
230 UV lo = ((ord - 0x10000) & 0x3FF) + 0xDC00;
231 enc_pack(aTHX_ result,size,endian,hi);
232 enc_pack(aTHX_ result,size,endian,lo);
236 /* not supposed to happen */
237 enc_pack(aTHX_ result,size,endian,FBCHAR);
241 enc_pack(aTHX_ result,size,endian,ord);
245 Perl_warner(aTHX_ packWARN(WARN_UTF8),"%"SVf":Partial character",
246 *hv_fetch((HV *)SvRV(obj),"Name",4,0));
248 if (check && !(check & ENCODE_LEAVE_SRC)){
250 Move(s,SvPVX(utf8),e-s,U8);
251 SvCUR_set(utf8,(e-s));