Update Changes.
[p5sagit/p5-mst-13.2.git] / ext / Encode / Encode.xs
CommitLineData
85982a32 1/*
fc17bd48 2 $Id: Encode.xs,v 1.40 2002/04/27 11:17:39 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"
0ab8f81e 11# define PERLIO_FILENAME "PerlIO/encoding.pm"
aae85ceb 12
85982a32 13/* set 1 or more to profile. t/encoding.t dumps core because of
14 Perl_warner and PerlIO don't work well */
0ab8f81e 15#define ENCODE_XS_PROFILE 0
39cf9a5e 16
85982a32 17/* set 0 to disable floating point to calculate buffer size for
18 encode_method(). 1 is recommended. 2 restores NI-S original */
0ab8f81e 19#define ENCODE_XS_USEFP 1
39cf9a5e 20
fcf2db38 21#define UNIMPLEMENTED(x,y) y x (SV *sv, char *encoding) {dTHX; \
2f5768b8 22 Perl_croak(aTHX_ "panic_unimplemented"); \
4a83738a 23 return (y)0; /* fool picky compilers */ \
87714904 24 }
85982a32 25/**/
011b2d2f 26
7e9a885a 27UNIMPLEMENTED(_encoded_utf8_to_bytes, I32)
28UNIMPLEMENTED(_encoded_bytes_to_utf8, I32)
33af2bc7 29
b0b300a3 30void
aa0053b7 31Encode_XSEncoding(pTHX_ encode_t * enc)
2f2b4ff2 32{
aa0053b7 33 dSP;
34 HV *stash = gv_stashpv("Encode::XS", TRUE);
35 SV *sv = sv_bless(newRV_noinc(newSViv(PTR2IV(enc))), stash);
36 int i = 0;
37 PUSHMARK(sp);
38 XPUSHs(sv);
39 while (enc->name[i]) {
40 const char *name = enc->name[i++];
41 XPUSHs(sv_2mortal(newSVpvn(name, strlen(name))));
42 }
43 PUTBACK;
44 call_pv("Encode::define_encoding", G_DISCARD);
45 SvREFCNT_dec(sv);
2f2b4ff2 46}
47
aa0053b7 48void
49call_failure(SV * routine, U8 * done, U8 * dest, U8 * orig)
50{
85982a32 51 /* Exists for breakpointing */
aa0053b7 52}
67e989fb 53
85982a32 54
2f2b4ff2 55static SV *
aa0053b7 56encode_method(pTHX_ encode_t * enc, encpage_t * dir, SV * src,
85982a32 57 int check)
2f2b4ff2 58{
aa0053b7 59 STRLEN slen;
60 U8 *s = (U8 *) SvPV(src, slen);
3aececda 61 STRLEN tlen = slen;
62 STRLEN ddone = 0;
63 STRLEN sdone = 0;
39cf9a5e 64
3c49ab08 65 /* We allocate slen+1.
85982a32 66 PerlIO dumps core if this value is smaller than this. */
3c49ab08 67 SV *dst = sv_2mortal(newSV(slen+1));
85982a32 68 U8 *d = (U8 *)SvPVX(dst);
69 STRLEN dlen = SvLEN(dst)-1;
70 int code;
71
72 if (!slen){
73 SvCUR_set(dst, 0);
74 SvPOK_only(dst);
75 goto ENCODE_END;
76 }
77
78 while (code = do_encode(dir, s, &slen, d, dlen, &dlen, !check))
79 {
80 SvCUR_set(dst, dlen+ddone);
81 SvPOK_only(dst);
0b3236bb 82
85982a32 83 if (code == ENCODE_FALLBACK || code == ENCODE_PARTIAL){
84 break;
85 }
86 switch (code) {
87 case ENCODE_NOSPACE:
88 {
89 STRLEN more = 0; /* make sure you initialize! */
90 STRLEN sleft;
91 sdone += slen;
92 ddone += dlen;
93 sleft = tlen - sdone;
fcb875d4 94#if ENCODE_XS_PROFILE >= 2
85982a32 95 Perl_warn(aTHX_
96 "more=%d, sdone=%d, sleft=%d, SvLEN(dst)=%d\n",
97 more, sdone, sleft, SvLEN(dst));
fcb875d4 98#endif
85982a32 99 if (sdone != 0) { /* has src ever been processed ? */
39cf9a5e 100#if ENCODE_XS_USEFP == 2
85982a32 101 more = (1.0*tlen*SvLEN(dst)+sdone-1)/sdone
102 - SvLEN(dst);
39cf9a5e 103#elif ENCODE_XS_USEFP
85982a32 104 more = (1.0*SvLEN(dst)+1)/sdone * sleft;
39cf9a5e 105#else
85982a32 106 /* safe until SvLEN(dst) == MAX_INT/16 */
107 more = (16*SvLEN(dst)+1)/sdone/16 * sleft;
39cf9a5e 108#endif
39cf9a5e 109 }
85982a32 110 more += UTF8_MAXLEN; /* insurance policy */
111 d = (U8 *) SvGROW(dst, SvLEN(dst) + more);
112 /* dst need to grow need MORE bytes! */
113 if (ddone >= SvLEN(dst)) {
114 Perl_croak(aTHX_ "Destination couldn't be grown.");
115 }
116 dlen = SvLEN(dst)-ddone-1;
117 d += ddone;
118 s += slen;
119 slen = tlen-sdone;
120 continue;
121 }
122 case ENCODE_NOREP:
123 /* encoding */
0ab8f81e 124 if (dir == enc->f_utf8) {
85982a32 125 STRLEN clen;
126 UV ch =
0ab8f81e 127 utf8n_to_uvuni(s+slen, (SvCUR(src)-slen),
b0b300a3 128 &clen, UTF8_ALLOW_ANY|UTF8_CHECK_ONLY);
85982a32 129 if (check & ENCODE_DIE_ON_ERR) {
130 Perl_croak(
0ab8f81e 131 aTHX_ "\"\\N{U+%" UVxf "}\" does not map to %s, %d",
85982a32 132 ch, enc->name[0], __LINE__);
133 }else{
134 if (check & ENCODE_RETURN_ON_ERR){
135 if (check & ENCODE_WARN_ON_ERR){
136 Perl_warner(
137 aTHX_ packWARN(WARN_UTF8),
0ab8f81e 138 "\"\\N{U+%" UVxf "}\" does not map to %s",
85982a32 139 ch,enc->name[0]);
140 }
141 goto ENCODE_SET_SRC;
142 }else if (check & ENCODE_PERLQQ){
0ab8f81e 143 SV* perlqq =
e4da7dce 144 sv_2mortal(newSVpvf("\\x{%04"UVxf"}", ch));
b2704119 145 sdone += slen + clen;
85982a32 146 ddone += dlen + SvCUR(perlqq);
147 sv_catsv(dst, perlqq);
af1f55d9 148 }else if (check & ENCODE_HTMLCREF){
149 SV* htmlcref =
7e19fb92 150 sv_2mortal(newSVpvf("&#%" UVuf ";", ch));
af1f55d9 151 sdone += slen + clen;
152 ddone += dlen + SvCUR(htmlcref);
153 sv_catsv(dst, htmlcref);
154 }else if (check & ENCODE_XMLCREF){
155 SV* xmlcref =
156 sv_2mortal(newSVpvf("&#x%" UVxf ";", ch));
157 sdone += slen + clen;
158 ddone += dlen + SvCUR(xmlcref);
159 sv_catsv(dst, xmlcref);
0ab8f81e 160 } else {
85982a32 161 /* fallback char */
162 sdone += slen + clen;
0ab8f81e 163 ddone += dlen + enc->replen;
164 sv_catpvn(dst, (char*)enc->rep, enc->replen);
b2704119 165 }
0ab8f81e 166 }
b2704119 167 }
85982a32 168 /* decoding */
0ab8f81e 169 else {
85982a32 170 if (check & ENCODE_DIE_ON_ERR){
171 Perl_croak(
fc17bd48 172 aTHX_ "%s \"\\x%02" UVXf
173 "\" does not map to Unicode (%d)",
85982a32 174 enc->name[0], (U8) s[slen], code);
175 }else{
176 if (check & ENCODE_RETURN_ON_ERR){
177 if (check & ENCODE_WARN_ON_ERR){
178 Perl_warner(
179 aTHX_ packWARN(WARN_UTF8),
fc17bd48 180 "%s \"\\x%02" UVXf
181 "\" does not map to Unicode (%d)",
85982a32 182 enc->name[0], (U8) s[slen], code);
183 }
184 goto ENCODE_SET_SRC;
af1f55d9 185 }else if (check &
186 (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){
0ab8f81e 187 SV* perlqq =
fc17bd48 188 sv_2mortal(newSVpvf("\\x%02" UVXf, s[slen]));
85982a32 189 sdone += slen + 1;
190 ddone += dlen + SvCUR(perlqq);
191 sv_catsv(dst, perlqq);
192 } else {
193 sdone += slen + 1;
0ab8f81e 194 ddone += dlen + strlen(FBCHAR_UTF8);
195 sv_catpv(dst, FBCHAR_UTF8);
85982a32 196 }
aa0053b7 197 }
b2704119 198 }
85982a32 199 /* settle variables when fallback */
b0b300a3 200 d = (U8 *)SvEND(dst);
0ab8f81e 201 dlen = SvLEN(dst) - ddone - 1;
202 s = (U8*)SvPVX(src) + sdone;
b2704119 203 slen = tlen - sdone;
204 break;
2f2b4ff2 205
85982a32 206 default:
207 Perl_croak(aTHX_ "Unexpected code %d converting %s %s",
208 code, (dir == enc->f_utf8) ? "to" : "from",
209 enc->name[0]);
210 return &PL_sv_undef;
aa0053b7 211 }
85982a32 212 }
213 ENCODE_SET_SRC:
6d1c0808 214 if (check && !(check & ENCODE_LEAVE_SRC)){
215 sdone = SvCUR(src) - (slen+sdone);
85982a32 216 if (sdone) {
217 sv_setpvn(src, (char*)s+slen, sdone);
aa0053b7 218 }
85982a32 219 SvCUR_set(src, sdone);
2f2b4ff2 220 }
85982a32 221 /* warn("check = 0x%X, code = 0x%d\n", check, code); */
222 if (code && !(check & ENCODE_RETURN_ON_ERR)) {
223 return &PL_sv_undef;
2f2b4ff2 224 }
0ab8f81e 225
85982a32 226 SvCUR_set(dst, dlen+ddone);
227 SvPOK_only(dst);
0ab8f81e 228
39cf9a5e 229#if ENCODE_XS_PROFILE
230 if (SvCUR(dst) > SvCUR(src)){
85982a32 231 Perl_warn(aTHX_
232 "SvLEN(dst)=%d, SvCUR(dst)=%d. %d bytes unused(%f %%)\n",
233 SvLEN(dst), SvCUR(dst), SvLEN(dst) - SvCUR(dst),
234 (SvLEN(dst) - SvCUR(dst))*1.0/SvLEN(dst)*100.0);
39cf9a5e 235 }
3c49ab08 236#endif
0ab8f81e 237
85982a32 238 ENCODE_END:
0b3236bb 239 *SvEND(dst) = '\0';
aa0053b7 240 return dst;
2f2b4ff2 241}
242
50d26985 243MODULE = Encode PACKAGE = Encode::XS PREFIX = Method_
2f2b4ff2 244
245PROTOTYPES: ENABLE
246
247void
0a95303c 248Method_name(obj)
249SV * obj
250CODE:
85982a32 251{
252 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
253 ST(0) = sv_2mortal(newSVpvn(enc->name[0],strlen(enc->name[0])));
254 XSRETURN(1);
255}
0a95303c 256
257void
b2704119 258Method_decode(obj,src,check = 0)
2f2b4ff2 259SV * obj
260SV * src
b2704119 261int check
2f2b4ff2 262CODE:
aae85ceb 263{
85982a32 264 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
265 ST(0) = encode_method(aTHX_ enc, enc->t_utf8, src, check);
266 SvUTF8_on(ST(0));
aae85ceb 267 XSRETURN(1);
268}
269
270void
85982a32 271Method_encode(obj,src,check = 0)
aae85ceb 272SV * obj
85982a32 273SV * src
274int check
aae85ceb 275CODE:
276{
85982a32 277 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
278 sv_utf8_upgrade(src);
279 ST(0) = encode_method(aTHX_ enc, enc->f_utf8, src, check);
aae85ceb 280 XSRETURN(1);
281}
282
0ab8f81e 283void
284Method_needs_lines(obj)
285SV * obj
286CODE:
287{
288 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
289 ST(0) = &PL_sv_no;
290 XSRETURN(1);
291}
292
293void
294Method_perlio_ok(obj)
295SV * obj
296CODE:
297{
298 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
011b2d2f 299 require_pv(PERLIO_FILENAME);
0ab8f81e 300 if (hv_exists(get_hv("INC", 0),
301 PERLIO_FILENAME, strlen(PERLIO_FILENAME)))
302 {
303 ST(0) = &PL_sv_yes;
304 }else{
305 ST(0) = &PL_sv_no;
306 }
307 XSRETURN(1);
308}
309
67e989fb 310MODULE = Encode PACKAGE = Encode
2c674647 311
312PROTOTYPES: ENABLE
313
67e989fb 314I32
2c674647 315_bytes_to_utf8(sv, ...)
85982a32 316SV * sv
317CODE:
318{
319 SV * encoding = items == 2 ? ST(1) : Nullsv;
0ab8f81e 320
85982a32 321 if (encoding)
322 RETVAL = _encoded_bytes_to_utf8(sv, SvPV_nolen(encoding));
323 else {
324 STRLEN len;
325 U8* s = (U8*)SvPV(sv, len);
326 U8* converted;
327
328 converted = bytes_to_utf8(s, &len); /* This allocs */
329 sv_setpvn(sv, (char *)converted, len);
330 SvUTF8_on(sv); /* XXX Should we? */
331 Safefree(converted); /* ... so free it */
332 RETVAL = len;
333 }
334}
335OUTPUT:
336 RETVAL
2c674647 337
67e989fb 338I32
2c674647 339_utf8_to_bytes(sv, ...)
85982a32 340SV * sv
341CODE:
342{
343 SV * to = items > 1 ? ST(1) : Nullsv;
344 SV * check = items > 2 ? ST(2) : Nullsv;
345
346 if (to) {
347 RETVAL = _encoded_utf8_to_bytes(sv, SvPV_nolen(to));
348 } else {
349 STRLEN len;
350 U8 *s = (U8*)SvPV(sv, len);
351
352 RETVAL = 0;
353 if (SvTRUE(check)) {
354 /* Must do things the slow way */
355 U8 *dest;
356 /* We need a copy to pass to check() */
0ab8f81e 357 U8 *src = (U8*)savepv((char *)s);
85982a32 358 U8 *send = s + len;
359
360 New(83, dest, len, U8); /* I think */
361
362 while (s < send) {
363 if (*s < 0x80){
364 *dest++ = *s++;
365 } else {
366 STRLEN ulen;
367 UV uv = *s++;
368
369 /* Have to do it all ourselves because of error routine,
370 aargh. */
371 if (!(uv & 0x40)){ goto failure; }
372 if (!(uv & 0x20)) { ulen = 2; uv &= 0x1f; }
373 else if (!(uv & 0x10)) { ulen = 3; uv &= 0x0f; }
374 else if (!(uv & 0x08)) { ulen = 4; uv &= 0x07; }
375 else if (!(uv & 0x04)) { ulen = 5; uv &= 0x03; }
376 else if (!(uv & 0x02)) { ulen = 6; uv &= 0x01; }
377 else if (!(uv & 0x01)) { ulen = 7; uv = 0; }
378 else { ulen = 13; uv = 0; }
87714904 379
85982a32 380 /* Note change to utf8.c variable naming, for variety */
381 while (ulen--) {
0ab8f81e 382 if ((*s & 0xc0) != 0x80){
383 goto failure;
85982a32 384 } else {
385 uv = (uv << 6) | (*s++ & 0x3f);
386 }
87714904 387 }
67e989fb 388 if (uv > 256) {
389 failure:
85982a32 390 call_failure(check, s, dest, src);
391 /* Now what happens? */
67e989fb 392 }
393 *dest++ = (U8)uv;
85982a32 394 }
395 }
396 } else {
397 RETVAL = (utf8_to_bytes(s, &len) ? len : 0);
2c674647 398 }
85982a32 399 }
400}
401OUTPUT:
402 RETVAL
2c674647 403
2c674647 404bool
b2704119 405is_utf8(sv, check = 0)
4411f3b6 406SV * sv
b2704119 407int check
85982a32 408CODE:
409{
410 if (SvGMAGICAL(sv)) /* it could be $1, for example */
411 sv = newSVsv(sv); /* GMAGIG will be done */
412 if (SvPOK(sv)) {
413 RETVAL = SvUTF8(sv) ? TRUE : FALSE;
414 if (RETVAL &&
415 check &&
416 !is_utf8_string((U8*)SvPVX(sv), SvCUR(sv)))
2c674647 417 RETVAL = FALSE;
85982a32 418 } else {
419 RETVAL = FALSE;
420 }
421 if (sv != ST(0))
422 SvREFCNT_dec(sv); /* it was a temp copy */
423}
424OUTPUT:
425 RETVAL
2c674647 426
427SV *
4411f3b6 428_utf8_on(sv)
85982a32 429SV * sv
430CODE:
431{
432 if (SvPOK(sv)) {
433 SV *rsv = newSViv(SvUTF8(sv));
434 RETVAL = rsv;
435 SvUTF8_on(sv);
436 } else {
437 RETVAL = &PL_sv_undef;
438 }
439}
440OUTPUT:
441 RETVAL
2c674647 442
443SV *
4411f3b6 444_utf8_off(sv)
85982a32 445SV * sv
446CODE:
447{
448 if (SvPOK(sv)) {
449 SV *rsv = newSViv(SvUTF8(sv));
450 RETVAL = rsv;
451 SvUTF8_off(sv);
452 } else {
453 RETVAL = &PL_sv_undef;
454 }
455}
456OUTPUT:
457 RETVAL
458
85982a32 459int
460DIE_ON_ERR()
461CODE:
462 RETVAL = ENCODE_DIE_ON_ERR;
463OUTPUT:
464 RETVAL
465
0ab8f81e 466int
85982a32 467WARN_ON_ERR()
468CODE:
469 RETVAL = ENCODE_WARN_ON_ERR;
470OUTPUT:
471 RETVAL
472
473int
474LEAVE_SRC()
475CODE:
476 RETVAL = ENCODE_LEAVE_SRC;
477OUTPUT:
478 RETVAL
479
480int
481RETURN_ON_ERR()
482CODE:
483 RETVAL = ENCODE_RETURN_ON_ERR;
484OUTPUT:
485 RETVAL
486
487int
488PERLQQ()
489CODE:
490 RETVAL = ENCODE_PERLQQ;
491OUTPUT:
492 RETVAL
493
494int
af1f55d9 495HTMLCREF()
496CODE:
497 RETVAL = ENCODE_HTMLCREF;
498OUTPUT:
499 RETVAL
500
501int
502XMLCREF()
503CODE:
504 RETVAL = ENCODE_XMLCREF;
505OUTPUT:
506 RETVAL
507
508int
85982a32 509FB_DEFAULT()
510CODE:
511 RETVAL = ENCODE_FB_DEFAULT;
512OUTPUT:
513 RETVAL
514
515int
516FB_CROAK()
517CODE:
518 RETVAL = ENCODE_FB_CROAK;
519OUTPUT:
520 RETVAL
521
522int
523FB_QUIET()
524CODE:
525 RETVAL = ENCODE_FB_QUIET;
526OUTPUT:
527 RETVAL
528
529int
530FB_WARN()
531CODE:
532 RETVAL = ENCODE_FB_WARN;
533OUTPUT:
534 RETVAL
535
536int
537FB_PERLQQ()
538CODE:
539 RETVAL = ENCODE_FB_PERLQQ;
540OUTPUT:
541 RETVAL
2c674647 542
af1f55d9 543int
544FB_HTMLCREF()
545CODE:
546 RETVAL = ENCODE_FB_HTMLCREF;
547OUTPUT:
548 RETVAL
549
550int
551FB_XMLCREF()
552CODE:
553 RETVAL = ENCODE_FB_XMLCREF;
554OUTPUT:
555 RETVAL
556
33af2bc7 557BOOT:
558{
85982a32 559#include "def_t.h"
e7cbefb8 560#include "def_t.exh"
33af2bc7 561}