Updated Encode to CPAN release 2.38
[p5sagit/p5-mst-13.2.git] / cpan / Encode / Encode.xs
CommitLineData
85982a32 1/*
b2deda17 2 $Id: Encode.xs,v 2.17 2009/11/16 14:08:13 dankogai Exp dankogai $
85982a32 3 */
4
fcf2db38 5#define PERL_NO_GET_CONTEXT
2c674647 6#include "EXTERN.h"
7#include "perl.h"
8#include "XSUB.h"
2f2b4ff2 9#define U8 U8
10#include "encode.h"
10c5ecbb 11
12# define PERLIO_MODNAME "PerlIO::encoding"
0ab8f81e 13# define PERLIO_FILENAME "PerlIO/encoding.pm"
aae85ceb 14
85982a32 15/* set 1 or more to profile. t/encoding.t dumps core because of
16 Perl_warner and PerlIO don't work well */
c6a7db43 17#define ENCODE_XS_PROFILE 0
39cf9a5e 18
85982a32 19/* set 0 to disable floating point to calculate buffer size for
20 encode_method(). 1 is recommended. 2 restores NI-S original */
c6a7db43 21#define ENCODE_XS_USEFP 1
39cf9a5e 22
fcf2db38 23#define UNIMPLEMENTED(x,y) y x (SV *sv, char *encoding) {dTHX; \
2f5768b8 24 Perl_croak(aTHX_ "panic_unimplemented"); \
d1256cb1 25 return (y)0; /* fool picky compilers */ \
87714904 26 }
85982a32 27/**/
011b2d2f 28
7e9a885a 29UNIMPLEMENTED(_encoded_utf8_to_bytes, I32)
30UNIMPLEMENTED(_encoded_bytes_to_utf8, I32)
33af2bc7 31
7f0d54d7 32#define UTF8_ALLOW_STRICT 0
33#define UTF8_ALLOW_NONSTRICT (UTF8_ALLOW_ANY & \
34 ~(UTF8_ALLOW_CONTINUATION | \
35 UTF8_ALLOW_NON_CONTINUATION | \
36 UTF8_ALLOW_LONG))
37
b0b300a3 38void
aa0053b7 39Encode_XSEncoding(pTHX_ encode_t * enc)
2f2b4ff2 40{
aa0053b7 41 dSP;
42 HV *stash = gv_stashpv("Encode::XS", TRUE);
43 SV *sv = sv_bless(newRV_noinc(newSViv(PTR2IV(enc))), stash);
44 int i = 0;
45 PUSHMARK(sp);
46 XPUSHs(sv);
47 while (enc->name[i]) {
d1256cb1 48 const char *name = enc->name[i++];
49 XPUSHs(sv_2mortal(newSVpvn(name, strlen(name))));
aa0053b7 50 }
51 PUTBACK;
52 call_pv("Encode::define_encoding", G_DISCARD);
53 SvREFCNT_dec(sv);
2f2b4ff2 54}
55
aa0053b7 56void
57call_failure(SV * routine, U8 * done, U8 * dest, U8 * orig)
58{
85982a32 59 /* Exists for breakpointing */
aa0053b7 60}
67e989fb 61
85982a32 62
2fc614e0 63#define ERR_ENCODE_NOMAP "\"\\x{%04" UVxf "}\" does not map to %s"
64#define ERR_DECODE_NOMAP "%s \"\\x%02" UVXf "\" does not map to Unicode"
65
2f2b4ff2 66static SV *
370462a2 67do_fallback_cb(pTHX_ UV ch, SV *fallback_cb)
8e180e82 68{
69 dSP;
70 int argc;
b2deda17 71 SV *retval = newSVpv("",0);
8e180e82 72 ENTER;
73 SAVETMPS;
74 PUSHMARK(sp);
75 XPUSHs(sv_2mortal(newSVnv((UV)ch)));
76 PUTBACK;
77 argc = call_sv(fallback_cb, G_SCALAR);
78 SPAGAIN;
79 if (argc != 1){
370462a2 80 croak("fallback sub must return scalar!");
8e180e82 81 }
b2deda17 82 sv_catsv(retval, POPs);
8e180e82 83 PUTBACK;
84 FREETMPS;
85 LEAVE;
86 return retval;
87}
88
89static SV *
0629a5b3 90encode_method(pTHX_ const encode_t * enc, const encpage_t * dir, SV * src,
370462a2 91 int check, STRLEN * offset, SV * term, int * retcode,
92 SV *fallback_cb)
2f2b4ff2 93{
aa0053b7 94 STRLEN slen;
95 U8 *s = (U8 *) SvPV(src, slen);
3aececda 96 STRLEN tlen = slen;
97 STRLEN ddone = 0;
98 STRLEN sdone = 0;
39cf9a5e 99
3c49ab08 100 /* We allocate slen+1.
85982a32 101 PerlIO dumps core if this value is smaller than this. */
3c49ab08 102 SV *dst = sv_2mortal(newSV(slen+1));
85982a32 103 U8 *d = (U8 *)SvPVX(dst);
104 STRLEN dlen = SvLEN(dst)-1;
220e2d4e 105 int code = 0;
106 STRLEN trmlen = 0;
cc7dbc11 107 U8 *trm = term ? (U8*) SvPV(term, trmlen) : NULL;
220e2d4e 108
109 if (offset) {
110 s += *offset;
6be7c101 111 if (slen > *offset){ /* safeguard against slen overflow */
d1256cb1 112 slen -= *offset;
6be7c101 113 }else{
d1256cb1 114 slen = 0;
6be7c101 115 }
220e2d4e 116 tlen = slen;
117 }
85982a32 118
6be7c101 119 if (slen == 0){
d1256cb1 120 SvCUR_set(dst, 0);
121 SvPOK_only(dst);
122 goto ENCODE_END;
85982a32 123 }
124
220e2d4e 125 while( (code = do_encode(dir, s, &slen, d, dlen, &dlen, !check,
d1256cb1 126 trm, trmlen)) )
85982a32 127 {
d1256cb1 128 SvCUR_set(dst, dlen+ddone);
129 SvPOK_only(dst);
130
131 if (code == ENCODE_FALLBACK || code == ENCODE_PARTIAL ||
132 code == ENCODE_FOUND_TERM) {
133 break;
134 }
135 switch (code) {
136 case ENCODE_NOSPACE:
137 {
138 STRLEN more = 0; /* make sure you initialize! */
139 STRLEN sleft;
140 sdone += slen;
141 ddone += dlen;
142 sleft = tlen - sdone;
fcb875d4 143#if ENCODE_XS_PROFILE >= 2
d1256cb1 144 Perl_warn(aTHX_
145 "more=%d, sdone=%d, sleft=%d, SvLEN(dst)=%d\n",
146 more, sdone, sleft, SvLEN(dst));
fcb875d4 147#endif
d1256cb1 148 if (sdone != 0) { /* has src ever been processed ? */
39cf9a5e 149#if ENCODE_XS_USEFP == 2
d1256cb1 150 more = (1.0*tlen*SvLEN(dst)+sdone-1)/sdone
151 - SvLEN(dst);
39cf9a5e 152#elif ENCODE_XS_USEFP
d1256cb1 153 more = (STRLEN)((1.0*SvLEN(dst)+1)/sdone * sleft);
39cf9a5e 154#else
d1256cb1 155 /* safe until SvLEN(dst) == MAX_INT/16 */
156 more = (16*SvLEN(dst)+1)/sdone/16 * sleft;
39cf9a5e 157#endif
d1256cb1 158 }
159 more += UTF8_MAXLEN; /* insurance policy */
160 d = (U8 *) SvGROW(dst, SvLEN(dst) + more);
161 /* dst need to grow need MORE bytes! */
162 if (ddone >= SvLEN(dst)) {
163 Perl_croak(aTHX_ "Destination couldn't be grown.");
164 }
165 dlen = SvLEN(dst)-ddone-1;
166 d += ddone;
167 s += slen;
168 slen = tlen-sdone;
169 continue;
170 }
171 case ENCODE_NOREP:
172 /* encoding */
173 if (dir == enc->f_utf8) {
174 STRLEN clen;
175 UV ch =
176 utf8n_to_uvuni(s+slen, (SvCUR(src)-slen),
177 &clen, UTF8_ALLOW_ANY|UTF8_CHECK_ONLY);
178 /* if non-representable multibyte prefix at end of current buffer - break*/
179 if (clen > tlen - sdone) break;
180 if (check & ENCODE_DIE_ON_ERR) {
181 Perl_croak(aTHX_ ERR_ENCODE_NOMAP,
182 (UV)ch, enc->name[0]);
183 return &PL_sv_undef; /* never reaches but be safe */
184 }
185 if (check & ENCODE_WARN_ON_ERR){
186 Perl_warner(aTHX_ packWARN(WARN_UTF8),
187 ERR_ENCODE_NOMAP, (UV)ch, enc->name[0]);
188 }
189 if (check & ENCODE_RETURN_ON_ERR){
190 goto ENCODE_SET_SRC;
191 }
192 if (check & (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){
193 SV* subchar =
370462a2 194 (fallback_cb != &PL_sv_undef)
195 ? do_fallback_cb(aTHX_ ch, fallback_cb)
196 : newSVpvf(check & ENCODE_PERLQQ ? "\\x{%04"UVxf"}" :
d1256cb1 197 check & ENCODE_HTMLCREF ? "&#%" UVuf ";" :
198 "&#x%" UVxf ";", (UV)ch);
b2deda17 199 SvUTF8_off(subchar); /* make sure no decoded string gets in */
d1256cb1 200 sdone += slen + clen;
201 ddone += dlen + SvCUR(subchar);
202 sv_catsv(dst, subchar);
203 SvREFCNT_dec(subchar);
204 } else {
205 /* fallback char */
206 sdone += slen + clen;
207 ddone += dlen + enc->replen;
208 sv_catpvn(dst, (char*)enc->rep, enc->replen);
209 }
210 }
211 /* decoding */
212 else {
213 if (check & ENCODE_DIE_ON_ERR){
214 Perl_croak(aTHX_ ERR_DECODE_NOMAP,
436c6dd3 215 enc->name[0], (UV)s[slen]);
d1256cb1 216 return &PL_sv_undef; /* never reaches but be safe */
217 }
218 if (check & ENCODE_WARN_ON_ERR){
219 Perl_warner(
220 aTHX_ packWARN(WARN_UTF8),
221 ERR_DECODE_NOMAP,
436c6dd3 222 enc->name[0], (UV)s[slen]);
d1256cb1 223 }
224 if (check & ENCODE_RETURN_ON_ERR){
225 goto ENCODE_SET_SRC;
226 }
227 if (check &
228 (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){
229 SV* subchar =
370462a2 230 (fallback_cb != &PL_sv_undef)
231 ? do_fallback_cb(aTHX_ (UV)s[slen], fallback_cb)
232 : newSVpvf("\\x%02" UVXf, (UV)s[slen]);
d1256cb1 233 sdone += slen + 1;
234 ddone += dlen + SvCUR(subchar);
235 sv_catsv(dst, subchar);
236 SvREFCNT_dec(subchar);
237 } else {
238 sdone += slen + 1;
239 ddone += dlen + strlen(FBCHAR_UTF8);
240 sv_catpv(dst, FBCHAR_UTF8);
241 }
242 }
243 /* settle variables when fallback */
244 d = (U8 *)SvEND(dst);
b0b300a3 245 dlen = SvLEN(dst) - ddone - 1;
d1256cb1 246 s = (U8*)SvPVX(src) + sdone;
247 slen = tlen - sdone;
248 break;
249
250 default:
251 Perl_croak(aTHX_ "Unexpected code %d converting %s %s",
252 code, (dir == enc->f_utf8) ? "to" : "from",
253 enc->name[0]);
254 return &PL_sv_undef;
255 }
85982a32 256 }
257 ENCODE_SET_SRC:
ca777f1c 258 if (check && !(check & ENCODE_LEAVE_SRC)){
d1256cb1 259 sdone = SvCUR(src) - (slen+sdone);
260 if (sdone) {
261 sv_setpvn(src, (char*)s+slen, sdone);
262 }
263 SvCUR_set(src, sdone);
2f2b4ff2 264 }
85982a32 265 /* warn("check = 0x%X, code = 0x%d\n", check, code); */
c6a7db43 266
85982a32 267 SvCUR_set(dst, dlen+ddone);
268 SvPOK_only(dst);
c6a7db43 269
39cf9a5e 270#if ENCODE_XS_PROFILE
271 if (SvCUR(dst) > SvCUR(src)){
d1256cb1 272 Perl_warn(aTHX_
273 "SvLEN(dst)=%d, SvCUR(dst)=%d. %d bytes unused(%f %%)\n",
274 SvLEN(dst), SvCUR(dst), SvLEN(dst) - SvCUR(dst),
275 (SvLEN(dst) - SvCUR(dst))*1.0/SvLEN(dst)*100.0);
39cf9a5e 276 }
3c49ab08 277#endif
c6a7db43 278
220e2d4e 279 if (offset)
280 *offset += sdone + slen;
281
85982a32 282 ENCODE_END:
0b3236bb 283 *SvEND(dst) = '\0';
220e2d4e 284 if (retcode) *retcode = code;
aa0053b7 285 return dst;
2f2b4ff2 286}
287
7f0d54d7 288static bool
289strict_utf8(pTHX_ SV* sv)
290{
291 HV* hv;
292 SV** svp;
293 sv = SvRV(sv);
294 if (!sv || SvTYPE(sv) != SVt_PVHV)
295 return 0;
296 hv = (HV*)sv;
297 svp = hv_fetch(hv, "strict_utf8", 11, 0);
298 if (!svp)
299 return 0;
300 return SvTRUE(*svp);
301}
302
303static U8*
304process_utf8(pTHX_ SV* dst, U8* s, U8* e, int check,
305 bool encode, bool strict, bool stop_at_partial)
306{
307 UV uv;
308 STRLEN ulen;
309
310 SvPOK_only(dst);
311 SvCUR_set(dst,0);
312
313 while (s < e) {
314 if (UTF8_IS_INVARIANT(*s)) {
315 sv_catpvn(dst, (char *)s, 1);
316 s++;
317 continue;
318 }
319
320 if (UTF8_IS_START(*s)) {
321 U8 skip = UTF8SKIP(s);
322 if ((s + skip) > e) {
323 /* Partial character */
324 /* XXX could check that rest of bytes are UTF8_IS_CONTINUATION(ch) */
56ff7374 325 if (stop_at_partial || (check & ENCODE_STOP_AT_PARTIAL))
7f0d54d7 326 break;
327
328 goto malformed_byte;
329 }
330
331 uv = utf8n_to_uvuni(s, e - s, &ulen,
332 UTF8_CHECK_ONLY | (strict ? UTF8_ALLOW_STRICT :
333 UTF8_ALLOW_NONSTRICT)
334 );
335#if 1 /* perl-5.8.6 and older do not check UTF8_ALLOW_LONG */
d1256cb1 336 if (strict && uv > PERL_UNICODE_MAX)
1b9a72a9 337 ulen = (STRLEN) -1;
7f0d54d7 338#endif
339 if (ulen == -1) {
340 if (strict) {
341 uv = utf8n_to_uvuni(s, e - s, &ulen,
342 UTF8_CHECK_ONLY | UTF8_ALLOW_NONSTRICT);
343 if (ulen == -1)
344 goto malformed_byte;
345 goto malformed;
346 }
347 goto malformed_byte;
348 }
349
350
351 /* Whole char is good */
352 sv_catpvn(dst,(char *)s,skip);
353 s += skip;
354 continue;
355 }
356
357 /* If we get here there is something wrong with alleged UTF-8 */
358 malformed_byte:
359 uv = (UV)*s;
360 ulen = 1;
361
362 malformed:
363 if (check & ENCODE_DIE_ON_ERR){
364 if (encode)
365 Perl_croak(aTHX_ ERR_ENCODE_NOMAP, uv, "utf8");
366 else
367 Perl_croak(aTHX_ ERR_DECODE_NOMAP, "utf8", uv);
368 }
369 if (check & ENCODE_WARN_ON_ERR){
370 if (encode)
371 Perl_warner(aTHX_ packWARN(WARN_UTF8),
372 ERR_ENCODE_NOMAP, uv, "utf8");
373 else
374 Perl_warner(aTHX_ packWARN(WARN_UTF8),
375 ERR_DECODE_NOMAP, "utf8", uv);
376 }
377 if (check & ENCODE_RETURN_ON_ERR) {
378 break;
379 }
380 if (check & (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){
381 SV* subchar = newSVpvf(check & ENCODE_PERLQQ ? (ulen == 1 ? "\\x%02" UVXf : "\\x{%04" UVXf "}"):
382 check & ENCODE_HTMLCREF ? "&#%" UVuf ";" :
383 "&#x%" UVxf ";", uv);
384 sv_catsv(dst, subchar);
385 SvREFCNT_dec(subchar);
386 } else {
387 sv_catpv(dst, FBCHAR_UTF8);
388 }
389 s += ulen;
390 }
391 *SvEND(dst) = '\0';
392
393 return s;
394}
395
396
ab3374e4 397MODULE = Encode PACKAGE = Encode::utf8 PREFIX = Method_
398
a0d8a30e 399PROTOTYPES: DISABLE
400
401void
b2deda17 402Method_decode_xs(obj,src,check_sv = &PL_sv_no)
ab3374e4 403SV * obj
404SV * src
b2deda17 405SV * check_sv
e83d9f5f 406PREINIT:
407 STRLEN slen;
408 U8 *s;
409 U8 *e;
410 SV *dst;
411 bool renewed = 0;
b2deda17 412 int check;
ab3374e4 413CODE:
414{
e83d9f5f 415 dSP; ENTER; SAVETMPS;
b2deda17 416 if (SvROK(check_sv)) {
417 croak("UTF-8 decoder doesn't support callback CHECK");
418 }
419 else {
420 check = SvIV(check_sv);
421 }
e83d9f5f 422 if (src == &PL_sv_undef) src = newSV(0);
423 s = (U8 *) SvPV(src, slen);
424 e = (U8 *) SvEND(src);
425 dst = newSV(slen>0?slen:1); /* newSV() abhors 0 -- inaba */
cc836e95 426
427 /*
7f0d54d7 428 * PerlIO check -- we assume the object is of PerlIO if renewed
cc836e95 429 */
cc836e95 430 PUSHMARK(sp);
431 XPUSHs(obj);
432 PUTBACK;
433 if (call_method("renewed",G_SCALAR) == 1) {
d1256cb1 434 SPAGAIN;
435 renewed = (bool)POPi;
436 PUTBACK;
cc836e95 437#if 0
d1256cb1 438 fprintf(stderr, "renewed == %d\n", renewed);
cc836e95 439#endif
cc836e95 440 }
441 FREETMPS; LEAVE;
442 /* end PerlIO check */
443
ab3374e4 444 if (SvUTF8(src)) {
d1256cb1 445 s = utf8_to_bytes(s,&slen);
446 if (s) {
447 SvCUR_set(src,slen);
448 SvUTF8_off(src);
449 e = s+slen;
450 }
451 else {
452 croak("Cannot decode string with wide characters");
453 }
ab3374e4 454 }
7f0d54d7 455
456 s = process_utf8(aTHX_ dst, s, e, check, 0, strict_utf8(aTHX_ obj), renewed);
ab3374e4 457
458 /* Clear out translated part of source unless asked not to */
459 if (check && !(check & ENCODE_LEAVE_SRC)){
d1256cb1 460 slen = e-s;
461 if (slen) {
462 sv_setpvn(src, (char*)s, slen);
463 }
464 SvCUR_set(src, slen);
ab3374e4 465 }
466 SvUTF8_on(dst);
467 ST(0) = sv_2mortal(dst);
468 XSRETURN(1);
469}
470
471void
b2deda17 472Method_encode_xs(obj,src,check_sv = &PL_sv_no)
ab3374e4 473SV * obj
474SV * src
b2deda17 475SV * check_sv
e83d9f5f 476PREINIT:
477 STRLEN slen;
478 U8 *s;
479 U8 *e;
480 SV *dst;
481 bool renewed = 0;
b2deda17 482 int check;
ab3374e4 483CODE:
484{
b2deda17 485 if (SvROK(check_sv)) {
486 croak("UTF-8 encoder doesn't support callback CHECK");
487 }
488 else {
489 check = SvIV(check_sv);
490 }
e83d9f5f 491 if (src == &PL_sv_undef) src = newSV(0);
492 s = (U8 *) SvPV(src, slen);
493 e = (U8 *) SvEND(src);
494 dst = newSV(slen>0?slen:1); /* newSV() abhors 0 -- inaba */
ab3374e4 495 if (SvUTF8(src)) {
d1256cb1 496 /* Already encoded */
497 if (strict_utf8(aTHX_ obj)) {
498 s = process_utf8(aTHX_ dst, s, e, check, 1, 1, 0);
499 }
7f0d54d7 500 else {
501 /* trust it and just copy the octets */
502 sv_setpvn(dst,(char *)s,(e-s));
d1256cb1 503 s = e;
7f0d54d7 504 }
ab3374e4 505 }
506 else {
507 /* Native bytes - can always encode */
d1256cb1 508 U8 *d = (U8 *) SvGROW(dst, 2*slen+1); /* +1 or assertion will botch */
ab3374e4 509 while (s < e) {
cf0d1c66 510 UV uv = NATIVE_TO_UNI((UV) *s);
511 s++; /* Above expansion of NATIVE_TO_UNI() is safer this way. */
ab3374e4 512 if (UNI_IS_INVARIANT(uv))
513 *d++ = (U8)UTF_TO_NATIVE(uv);
514 else {
515 *d++ = (U8)UTF8_EIGHT_BIT_HI(uv);
516 *d++ = (U8)UTF8_EIGHT_BIT_LO(uv);
517 }
d1256cb1 518 }
ab3374e4 519 SvCUR_set(dst, d- (U8 *)SvPVX(dst));
520 *SvEND(dst) = '\0';
521 }
522
523 /* Clear out translated part of source unless asked not to */
524 if (check && !(check & ENCODE_LEAVE_SRC)){
d1256cb1 525 slen = e-s;
526 if (slen) {
527 sv_setpvn(src, (char*)s, slen);
528 }
529 SvCUR_set(src, slen);
ab3374e4 530 }
531 SvPOK_only(dst);
532 SvUTF8_off(dst);
533 ST(0) = sv_2mortal(dst);
534 XSRETURN(1);
535}
536
50d26985 537MODULE = Encode PACKAGE = Encode::XS PREFIX = Method_
2f2b4ff2 538
539PROTOTYPES: ENABLE
540
541void
a0d8a30e 542Method_renew(obj)
543SV * obj
544CODE:
545{
546 XSRETURN(1);
547}
548
cc836e95 549int
550Method_renewed(obj)
551SV * obj
552CODE:
553 RETVAL = 0;
554OUTPUT:
555 RETVAL
556
a0d8a30e 557void
0a95303c 558Method_name(obj)
559SV * obj
560CODE:
85982a32 561{
562 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
563 ST(0) = sv_2mortal(newSVpvn(enc->name[0],strlen(enc->name[0])));
564 XSRETURN(1);
565}
0a95303c 566
567void
370462a2 568Method_cat_decode(obj, dst, src, off, term, check_sv = &PL_sv_no)
220e2d4e 569SV * obj
570SV * dst
571SV * src
572SV * off
573SV * term
370462a2 574SV * check_sv
220e2d4e 575CODE:
576{
370462a2 577 int check;
578 SV *fallback_cb = &PL_sv_undef;
220e2d4e 579 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
580 STRLEN offset = (STRLEN)SvIV(off);
581 int code = 0;
582 if (SvUTF8(src)) {
583 sv_utf8_downgrade(src, FALSE);
584 }
370462a2 585 if (SvROK(check_sv)){
586 fallback_cb = check_sv;
587 check = ENCODE_PERLQQ|ENCODE_LEAVE_SRC; /* same as FB_PERLQQ */
588 }else{
589 check = SvIV(check_sv);
590 }
220e2d4e 591 sv_catsv(dst, encode_method(aTHX_ enc, enc->t_utf8, src, check,
370462a2 592 &offset, term, &code, fallback_cb));
b162af07 593 SvIV_set(off, (IV)offset);
220e2d4e 594 if (code == ENCODE_FOUND_TERM) {
d1256cb1 595 ST(0) = &PL_sv_yes;
220e2d4e 596 }else{
d1256cb1 597 ST(0) = &PL_sv_no;
220e2d4e 598 }
599 XSRETURN(1);
600}
601
602void
8e180e82 603Method_decode(obj,src,check_sv = &PL_sv_no)
2f2b4ff2 604SV * obj
605SV * src
8e180e82 606SV * check_sv
2f2b4ff2 607CODE:
aae85ceb 608{
8e180e82 609 int check;
370462a2 610 SV *fallback_cb = &PL_sv_undef;
85982a32 611 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
ab3374e4 612 if (SvUTF8(src)) {
613 sv_utf8_downgrade(src, FALSE);
614 }
8e180e82 615 if (SvROK(check_sv)){
370462a2 616 fallback_cb = check_sv;
617 check = ENCODE_PERLQQ|ENCODE_LEAVE_SRC; /* same as FB_PERLQQ */
8e180e82 618 }else{
370462a2 619 check = SvIV(check_sv);
8e180e82 620 }
220e2d4e 621 ST(0) = encode_method(aTHX_ enc, enc->t_utf8, src, check,
370462a2 622 NULL, Nullsv, NULL, fallback_cb);
85982a32 623 SvUTF8_on(ST(0));
aae85ceb 624 XSRETURN(1);
625}
626
627void
8e180e82 628Method_encode(obj,src,check_sv = &PL_sv_no)
aae85ceb 629SV * obj
85982a32 630SV * src
8e180e82 631SV * check_sv
aae85ceb 632CODE:
633{
8e180e82 634 int check;
370462a2 635 SV *fallback_cb = &PL_sv_undef;
85982a32 636 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
637 sv_utf8_upgrade(src);
8e180e82 638 if (SvROK(check_sv)){
370462a2 639 fallback_cb = check_sv;
640 check = ENCODE_PERLQQ|ENCODE_LEAVE_SRC; /* same as FB_PERLQQ */
8e180e82 641 }else{
370462a2 642 check = SvIV(check_sv);
8e180e82 643 }
220e2d4e 644 ST(0) = encode_method(aTHX_ enc, enc->f_utf8, src, check,
370462a2 645 NULL, Nullsv, NULL, fallback_cb);
aae85ceb 646 XSRETURN(1);
647}
648
0ab8f81e 649void
650Method_needs_lines(obj)
651SV * obj
652CODE:
653{
b32afa7c 654 /* encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); */
0ab8f81e 655 ST(0) = &PL_sv_no;
656 XSRETURN(1);
657}
658
659void
660Method_perlio_ok(obj)
661SV * obj
662CODE:
663{
b32afa7c 664 /* encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); */
10c5ecbb 665 /* require_pv(PERLIO_FILENAME); */
666
667 eval_pv("require PerlIO::encoding", 0);
668
669 if (SvTRUE(get_sv("@", 0))) {
d1256cb1 670 ST(0) = &PL_sv_no;
10c5ecbb 671 }else{
d1256cb1 672 ST(0) = &PL_sv_yes;
0ab8f81e 673 }
674 XSRETURN(1);
675}
676
742555bd 677void
678Method_mime_name(obj)
679SV * obj
680CODE:
681{
682 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
683 SV *retval;
684 eval_pv("require Encode::MIME::Name", 0);
685
686 if (SvTRUE(get_sv("@", 0))) {
687 ST(0) = &PL_sv_undef;
688 }else{
689 ENTER;
690 SAVETMPS;
691 PUSHMARK(sp);
692 XPUSHs(sv_2mortal(newSVpvn(enc->name[0], strlen(enc->name[0]))));
693 PUTBACK;
694 call_pv("Encode::MIME::Name::get_mime_name", G_SCALAR);
695 SPAGAIN;
696 retval = newSVsv(POPs);
697 PUTBACK;
698 FREETMPS;
699 LEAVE;
700 /* enc->name[0] */
701 ST(0) = retval;
702 }
703 XSRETURN(1);
704}
705
67e989fb 706MODULE = Encode PACKAGE = Encode
2c674647 707
708PROTOTYPES: ENABLE
709
67e989fb 710I32
2c674647 711_bytes_to_utf8(sv, ...)
85982a32 712SV * sv
713CODE:
714{
715 SV * encoding = items == 2 ? ST(1) : Nullsv;
c6a7db43 716
85982a32 717 if (encoding)
718 RETVAL = _encoded_bytes_to_utf8(sv, SvPV_nolen(encoding));
719 else {
d1256cb1 720 STRLEN len;
721 U8* s = (U8*)SvPV(sv, len);
722 U8* converted;
723
724 converted = bytes_to_utf8(s, &len); /* This allocs */
725 sv_setpvn(sv, (char *)converted, len);
726 SvUTF8_on(sv); /* XXX Should we? */
727 Safefree(converted); /* ... so free it */
728 RETVAL = len;
85982a32 729 }
730}
731OUTPUT:
732 RETVAL
2c674647 733
67e989fb 734I32
2c674647 735_utf8_to_bytes(sv, ...)
85982a32 736SV * sv
737CODE:
738{
739 SV * to = items > 1 ? ST(1) : Nullsv;
740 SV * check = items > 2 ? ST(2) : Nullsv;
741
742 if (to) {
d1256cb1 743 RETVAL = _encoded_utf8_to_bytes(sv, SvPV_nolen(to));
85982a32 744 } else {
d1256cb1 745 STRLEN len;
746 U8 *s = (U8*)SvPV(sv, len);
85982a32 747
d1256cb1 748 RETVAL = 0;
749 if (SvTRUE(check)) {
750 /* Must do things the slow way */
751 U8 *dest;
85982a32 752 /* We need a copy to pass to check() */
2cdf8b6c 753 U8 *src = s;
d1256cb1 754 U8 *send = s + len;
2cdf8b6c 755 U8 *d0;
85982a32 756
d1256cb1 757 New(83, dest, len, U8); /* I think */
2cdf8b6c 758 d0 = dest;
85982a32 759
d1256cb1 760 while (s < send) {
85982a32 761 if (*s < 0x80){
d1256cb1 762 *dest++ = *s++;
85982a32 763 } else {
d1256cb1 764 STRLEN ulen;
765 UV uv = *s++;
766
767 /* Have to do it all ourselves because of error routine,
768 aargh. */
769 if (!(uv & 0x40)){ goto failure; }
770 if (!(uv & 0x20)) { ulen = 2; uv &= 0x1f; }
771 else if (!(uv & 0x10)) { ulen = 3; uv &= 0x0f; }
772 else if (!(uv & 0x08)) { ulen = 4; uv &= 0x07; }
773 else if (!(uv & 0x04)) { ulen = 5; uv &= 0x03; }
774 else if (!(uv & 0x02)) { ulen = 6; uv &= 0x01; }
775 else if (!(uv & 0x01)) { ulen = 7; uv = 0; }
776 else { ulen = 13; uv = 0; }
777
778 /* Note change to utf8.c variable naming, for variety */
779 while (ulen--) {
780 if ((*s & 0xc0) != 0x80){
781 goto failure;
782 } else {
783 uv = (uv << 6) | (*s++ & 0x3f);
784 }
785 }
786 if (uv > 256) {
787 failure:
788 call_failure(check, s, dest, src);
789 /* Now what happens? */
790 }
791 *dest++ = (U8)uv;
792 }
793 }
2cdf8b6c 794 RETVAL = dest - d0;
795 sv_usepvn(sv, (char *)dest, RETVAL);
796 SvUTF8_off(sv);
d1256cb1 797 } else {
798 RETVAL = (utf8_to_bytes(s, &len) ? len : 0);
799 }
85982a32 800 }
801}
802OUTPUT:
803 RETVAL
2c674647 804
2c674647 805bool
b2704119 806is_utf8(sv, check = 0)
4411f3b6 807SV * sv
b2704119 808int check
85982a32 809CODE:
810{
811 if (SvGMAGICAL(sv)) /* it could be $1, for example */
d1256cb1 812 sv = newSVsv(sv); /* GMAGIG will be done */
d1256cb1 813 RETVAL = SvUTF8(sv) ? TRUE : FALSE;
814 if (RETVAL &&
815 check &&
816 !is_utf8_string((U8*)SvPVX(sv), SvCUR(sv)))
817 RETVAL = FALSE;
85982a32 818 if (sv != ST(0))
d1256cb1 819 SvREFCNT_dec(sv); /* it was a temp copy */
85982a32 820}
821OUTPUT:
822 RETVAL
2c674647 823
824SV *
4411f3b6 825_utf8_on(sv)
85982a32 826SV * sv
827CODE:
828{
829 if (SvPOK(sv)) {
d1256cb1 830 SV *rsv = newSViv(SvUTF8(sv));
831 RETVAL = rsv;
832 SvUTF8_on(sv);
85982a32 833 } else {
d1256cb1 834 RETVAL = &PL_sv_undef;
85982a32 835 }
836}
837OUTPUT:
838 RETVAL
2c674647 839
840SV *
4411f3b6 841_utf8_off(sv)
85982a32 842SV * sv
843CODE:
844{
845 if (SvPOK(sv)) {
d1256cb1 846 SV *rsv = newSViv(SvUTF8(sv));
847 RETVAL = rsv;
848 SvUTF8_off(sv);
85982a32 849 } else {
d1256cb1 850 RETVAL = &PL_sv_undef;
85982a32 851 }
852}
853OUTPUT:
854 RETVAL
855
85982a32 856int
857DIE_ON_ERR()
858CODE:
859 RETVAL = ENCODE_DIE_ON_ERR;
860OUTPUT:
861 RETVAL
862
c6a7db43 863int
85982a32 864WARN_ON_ERR()
865CODE:
866 RETVAL = ENCODE_WARN_ON_ERR;
867OUTPUT:
868 RETVAL
869
870int
871LEAVE_SRC()
872CODE:
873 RETVAL = ENCODE_LEAVE_SRC;
874OUTPUT:
875 RETVAL
876
877int
878RETURN_ON_ERR()
879CODE:
880 RETVAL = ENCODE_RETURN_ON_ERR;
881OUTPUT:
882 RETVAL
883
884int
885PERLQQ()
886CODE:
887 RETVAL = ENCODE_PERLQQ;
888OUTPUT:
889 RETVAL
890
891int
af1f55d9 892HTMLCREF()
893CODE:
894 RETVAL = ENCODE_HTMLCREF;
895OUTPUT:
896 RETVAL
897
898int
899XMLCREF()
900CODE:
901 RETVAL = ENCODE_XMLCREF;
902OUTPUT:
903 RETVAL
904
905int
56ff7374 906STOP_AT_PARTIAL()
907CODE:
908 RETVAL = ENCODE_STOP_AT_PARTIAL;
909OUTPUT:
910 RETVAL
911
912int
85982a32 913FB_DEFAULT()
914CODE:
915 RETVAL = ENCODE_FB_DEFAULT;
916OUTPUT:
917 RETVAL
918
919int
920FB_CROAK()
921CODE:
922 RETVAL = ENCODE_FB_CROAK;
923OUTPUT:
924 RETVAL
925
926int
927FB_QUIET()
928CODE:
929 RETVAL = ENCODE_FB_QUIET;
930OUTPUT:
931 RETVAL
932
933int
934FB_WARN()
935CODE:
936 RETVAL = ENCODE_FB_WARN;
937OUTPUT:
938 RETVAL
939
940int
941FB_PERLQQ()
942CODE:
943 RETVAL = ENCODE_FB_PERLQQ;
944OUTPUT:
945 RETVAL
2c674647 946
af1f55d9 947int
948FB_HTMLCREF()
949CODE:
950 RETVAL = ENCODE_FB_HTMLCREF;
951OUTPUT:
952 RETVAL
953
954int
955FB_XMLCREF()
956CODE:
957 RETVAL = ENCODE_FB_XMLCREF;
958OUTPUT:
959 RETVAL
960
33af2bc7 961BOOT:
962{
85982a32 963#include "def_t.h"
e7cbefb8 964#include "def_t.exh"
33af2bc7 965}