2 $Id: Encode.xs,v 1.31 2002/04/20 23:43:47 dankogai Exp dankogai $
5 #define PERL_NO_GET_CONTEXT
12 /* set 1 or more to profile. t/encoding.t dumps core because of
13 Perl_warner and PerlIO don't work well */
14 #define ENCODE_XS_PROFILE 0
16 /* set 0 to disable floating point to calculate buffer size for
17 encode_method(). 1 is recommended. 2 restores NI-S original */
18 #define ENCODE_XS_USEFP 1
20 #define UNIMPLEMENTED(x,y) y x (SV *sv, char *encoding) {dTHX; \
21 Perl_croak(aTHX_ "panic_unimplemented"); \
22 return (y)0; /* fool picky compilers */ \
25 UNIMPLEMENTED(_encoded_utf8_to_bytes, I32);
26 UNIMPLEMENTED(_encoded_bytes_to_utf8, I32);
29 Encode_XSEncoding(pTHX_ encode_t * enc)
32 HV *stash = gv_stashpv("Encode::XS", TRUE);
33 SV *sv = sv_bless(newRV_noinc(newSViv(PTR2IV(enc))), stash);
37 while (enc->name[i]) {
38 const char *name = enc->name[i++];
39 XPUSHs(sv_2mortal(newSVpvn(name, strlen(name))));
42 call_pv("Encode::define_encoding", G_DISCARD);
47 call_failure(SV * routine, U8 * done, U8 * dest, U8 * orig)
49 /* Exists for breakpointing */
54 encode_method(pTHX_ encode_t * enc, encpage_t * dir, SV * src,
58 U8 *s = (U8 *) SvPV(src, slen);
63 /* We allocate slen+1.
64 PerlIO dumps core if this value is smaller than this. */
65 SV *dst = sv_2mortal(newSV(slen+1));
66 U8 *d = (U8 *)SvPVX(dst);
67 STRLEN dlen = SvLEN(dst)-1;
76 while (code = do_encode(dir, s, &slen, d, dlen, &dlen, !check))
78 SvCUR_set(dst, dlen+ddone);
81 if (code == ENCODE_FALLBACK || code == ENCODE_PARTIAL){
87 STRLEN more = 0; /* make sure you initialize! */
92 #if ENCODE_XS_PROFILE >= 2
94 "more=%d, sdone=%d, sleft=%d, SvLEN(dst)=%d\n",
95 more, sdone, sleft, SvLEN(dst));
97 if (sdone != 0) { /* has src ever been processed ? */
98 #if ENCODE_XS_USEFP == 2
99 more = (1.0*tlen*SvLEN(dst)+sdone-1)/sdone
101 #elif ENCODE_XS_USEFP
102 more = (1.0*SvLEN(dst)+1)/sdone * sleft;
104 /* safe until SvLEN(dst) == MAX_INT/16 */
105 more = (16*SvLEN(dst)+1)/sdone/16 * sleft;
108 more += UTF8_MAXLEN; /* insurance policy */
109 d = (U8 *) SvGROW(dst, SvLEN(dst) + more);
110 /* dst need to grow need MORE bytes! */
111 if (ddone >= SvLEN(dst)) {
112 Perl_croak(aTHX_ "Destination couldn't be grown.");
114 dlen = SvLEN(dst)-ddone-1;
122 if (dir == enc->f_utf8) {
125 utf8n_to_uvuni(s+slen, (SvCUR(src)-slen),
126 &clen, UTF8_ALLOW_ANY|UTF8_CHECK_ONLY);
127 if (check & ENCODE_DIE_ON_ERR) {
129 aTHX_ "\"\\N{U+%" UVxf "}\" does not map to %s, %d",
130 ch, enc->name[0], __LINE__);
132 if (check & ENCODE_RETURN_ON_ERR){
133 if (check & ENCODE_WARN_ON_ERR){
135 aTHX_ packWARN(WARN_UTF8),
136 "\"\\N{U+%" UVxf "}\" does not map to %s",
140 }else if (check & ENCODE_PERLQQ){
142 sv_2mortal(newSVpvf("\\x{%04x}", ch));
143 sdone += slen + clen;
144 ddone += dlen + SvCUR(perlqq);
145 sv_catsv(dst, perlqq);
148 sdone += slen + clen;
149 ddone += dlen + enc->replen;
150 sv_catpvn(dst, (char*)enc->rep, enc->replen);
156 if (check & ENCODE_DIE_ON_ERR){
158 aTHX_ "%s \"\\x%02X\" does not map to Unicode (%d)",
159 enc->name[0], (U8) s[slen], code);
161 if (check & ENCODE_RETURN_ON_ERR){
162 if (check & ENCODE_WARN_ON_ERR){
164 aTHX_ packWARN(WARN_UTF8),
165 "%s \"\\x%02X\" does not map to Unicode (%d)",
166 enc->name[0], (U8) s[slen], code);
169 }else if (check & ENCODE_PERLQQ){
171 sv_2mortal(newSVpvf("\\x%02X", s[slen]));
173 ddone += dlen + SvCUR(perlqq);
174 sv_catsv(dst, perlqq);
177 ddone += dlen + strlen(FBCHAR_UTF8);
178 sv_catpv(dst, FBCHAR_UTF8);
182 /* settle variables when fallback */
183 d = (U8 *)SvEND(dst);
184 dlen = SvLEN(dst) - ddone - 1;
185 s = (U8*)SvPVX(src) + sdone;
190 Perl_croak(aTHX_ "Unexpected code %d converting %s %s",
191 code, (dir == enc->f_utf8) ? "to" : "from",
197 if (check && !(check & ENCODE_LEAVE_SRC)){
198 sdone = SvCUR(src) - (slen+sdone);
200 sv_setpvn(src, (char*)s+slen, sdone);
202 SvCUR_set(src, sdone);
204 /* warn("check = 0x%X, code = 0x%d\n", check, code); */
205 if (code && !(check & ENCODE_RETURN_ON_ERR)) {
209 SvCUR_set(dst, dlen+ddone);
212 #if ENCODE_XS_PROFILE
213 if (SvCUR(dst) > SvCUR(src)){
215 "SvLEN(dst)=%d, SvCUR(dst)=%d. %d bytes unused(%f %%)\n",
216 SvLEN(dst), SvCUR(dst), SvLEN(dst) - SvCUR(dst),
217 (SvLEN(dst) - SvCUR(dst))*1.0/SvLEN(dst)*100.0);
226 MODULE = Encode PACKAGE = Encode::XS PREFIX = Method_
235 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
236 ST(0) = sv_2mortal(newSVpvn(enc->name[0],strlen(enc->name[0])));
241 Method_decode(obj,src,check = 0)
247 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
248 ST(0) = encode_method(aTHX_ enc, enc->t_utf8, src, check);
254 Method_encode(obj,src,check = 0)
260 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
261 sv_utf8_upgrade(src);
262 ST(0) = encode_method(aTHX_ enc, enc->f_utf8, src, check);
266 MODULE = Encode PACKAGE = Encode
271 _bytes_to_utf8(sv, ...)
275 SV * encoding = items == 2 ? ST(1) : Nullsv;
278 RETVAL = _encoded_bytes_to_utf8(sv, SvPV_nolen(encoding));
281 U8* s = (U8*)SvPV(sv, len);
284 converted = bytes_to_utf8(s, &len); /* This allocs */
285 sv_setpvn(sv, (char *)converted, len);
286 SvUTF8_on(sv); /* XXX Should we? */
287 Safefree(converted); /* ... so free it */
295 _utf8_to_bytes(sv, ...)
299 SV * to = items > 1 ? ST(1) : Nullsv;
300 SV * check = items > 2 ? ST(2) : Nullsv;
303 RETVAL = _encoded_utf8_to_bytes(sv, SvPV_nolen(to));
306 U8 *s = (U8*)SvPV(sv, len);
310 /* Must do things the slow way */
312 /* We need a copy to pass to check() */
313 U8 *src = (U8*)savepv((char *)s);
316 New(83, dest, len, U8); /* I think */
325 /* Have to do it all ourselves because of error routine,
327 if (!(uv & 0x40)){ goto failure; }
328 if (!(uv & 0x20)) { ulen = 2; uv &= 0x1f; }
329 else if (!(uv & 0x10)) { ulen = 3; uv &= 0x0f; }
330 else if (!(uv & 0x08)) { ulen = 4; uv &= 0x07; }
331 else if (!(uv & 0x04)) { ulen = 5; uv &= 0x03; }
332 else if (!(uv & 0x02)) { ulen = 6; uv &= 0x01; }
333 else if (!(uv & 0x01)) { ulen = 7; uv = 0; }
334 else { ulen = 13; uv = 0; }
336 /* Note change to utf8.c variable naming, for variety */
338 if ((*s & 0xc0) != 0x80){
341 uv = (uv << 6) | (*s++ & 0x3f);
346 call_failure(check, s, dest, src);
347 /* Now what happens? */
353 RETVAL = (utf8_to_bytes(s, &len) ? len : 0);
361 is_utf8(sv, check = 0)
366 if (SvGMAGICAL(sv)) /* it could be $1, for example */
367 sv = newSVsv(sv); /* GMAGIG will be done */
369 RETVAL = SvUTF8(sv) ? TRUE : FALSE;
372 !is_utf8_string((U8*)SvPVX(sv), SvCUR(sv)))
378 SvREFCNT_dec(sv); /* it was a temp copy */
389 SV *rsv = newSViv(SvUTF8(sv));
393 RETVAL = &PL_sv_undef;
405 SV *rsv = newSViv(SvUTF8(sv));
409 RETVAL = &PL_sv_undef;
421 RETVAL = ENCODE_DIE_ON_ERR;
428 RETVAL = ENCODE_WARN_ON_ERR;
435 RETVAL = ENCODE_LEAVE_SRC;
442 RETVAL = ENCODE_RETURN_ON_ERR;
449 RETVAL = ENCODE_PERLQQ;
456 RETVAL = ENCODE_FB_DEFAULT;
463 RETVAL = ENCODE_FB_CROAK;
470 RETVAL = ENCODE_FB_QUIET;
477 RETVAL = ENCODE_FB_WARN;
484 RETVAL = ENCODE_FB_PERLQQ;