bytes::length TIMTOWTDI
[p5sagit/p5-mst-13.2.git] / ext / Encode / Encode.xs
CommitLineData
85982a32 1/*
af1f55d9 2 $Id: Encode.xs,v 1.38 2002/04/24 20:11:14 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 =
150 sv_2mortal(newSVpvf("&#%d;", ch));
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(
172 aTHX_ "%s \"\\x%02X\" does not map to Unicode (%d)",
173 enc->name[0], (U8) s[slen], code);
174 }else{
175 if (check & ENCODE_RETURN_ON_ERR){
176 if (check & ENCODE_WARN_ON_ERR){
177 Perl_warner(
178 aTHX_ packWARN(WARN_UTF8),
179 "%s \"\\x%02X\" does not map to Unicode (%d)",
180 enc->name[0], (U8) s[slen], code);
181 }
182 goto ENCODE_SET_SRC;
af1f55d9 183 }else if (check &
184 (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){
0ab8f81e 185 SV* perlqq =
b2704119 186 sv_2mortal(newSVpvf("\\x%02X", s[slen]));
85982a32 187 sdone += slen + 1;
188 ddone += dlen + SvCUR(perlqq);
189 sv_catsv(dst, perlqq);
190 } else {
191 sdone += slen + 1;
0ab8f81e 192 ddone += dlen + strlen(FBCHAR_UTF8);
193 sv_catpv(dst, FBCHAR_UTF8);
85982a32 194 }
aa0053b7 195 }
b2704119 196 }
85982a32 197 /* settle variables when fallback */
b0b300a3 198 d = (U8 *)SvEND(dst);
0ab8f81e 199 dlen = SvLEN(dst) - ddone - 1;
200 s = (U8*)SvPVX(src) + sdone;
b2704119 201 slen = tlen - sdone;
202 break;
2f2b4ff2 203
85982a32 204 default:
205 Perl_croak(aTHX_ "Unexpected code %d converting %s %s",
206 code, (dir == enc->f_utf8) ? "to" : "from",
207 enc->name[0]);
208 return &PL_sv_undef;
aa0053b7 209 }
85982a32 210 }
211 ENCODE_SET_SRC:
6d1c0808 212 if (check && !(check & ENCODE_LEAVE_SRC)){
213 sdone = SvCUR(src) - (slen+sdone);
85982a32 214 if (sdone) {
215 sv_setpvn(src, (char*)s+slen, sdone);
aa0053b7 216 }
85982a32 217 SvCUR_set(src, sdone);
2f2b4ff2 218 }
85982a32 219 /* warn("check = 0x%X, code = 0x%d\n", check, code); */
220 if (code && !(check & ENCODE_RETURN_ON_ERR)) {
221 return &PL_sv_undef;
2f2b4ff2 222 }
0ab8f81e 223
85982a32 224 SvCUR_set(dst, dlen+ddone);
225 SvPOK_only(dst);
0ab8f81e 226
39cf9a5e 227#if ENCODE_XS_PROFILE
228 if (SvCUR(dst) > SvCUR(src)){
85982a32 229 Perl_warn(aTHX_
230 "SvLEN(dst)=%d, SvCUR(dst)=%d. %d bytes unused(%f %%)\n",
231 SvLEN(dst), SvCUR(dst), SvLEN(dst) - SvCUR(dst),
232 (SvLEN(dst) - SvCUR(dst))*1.0/SvLEN(dst)*100.0);
39cf9a5e 233 }
3c49ab08 234#endif
0ab8f81e 235
85982a32 236 ENCODE_END:
0b3236bb 237 *SvEND(dst) = '\0';
aa0053b7 238 return dst;
2f2b4ff2 239}
240
50d26985 241MODULE = Encode PACKAGE = Encode::XS PREFIX = Method_
2f2b4ff2 242
243PROTOTYPES: ENABLE
244
245void
0a95303c 246Method_name(obj)
247SV * obj
248CODE:
85982a32 249{
250 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
251 ST(0) = sv_2mortal(newSVpvn(enc->name[0],strlen(enc->name[0])));
252 XSRETURN(1);
253}
0a95303c 254
255void
b2704119 256Method_decode(obj,src,check = 0)
2f2b4ff2 257SV * obj
258SV * src
b2704119 259int check
2f2b4ff2 260CODE:
aae85ceb 261{
85982a32 262 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
263 ST(0) = encode_method(aTHX_ enc, enc->t_utf8, src, check);
264 SvUTF8_on(ST(0));
aae85ceb 265 XSRETURN(1);
266}
267
268void
85982a32 269Method_encode(obj,src,check = 0)
aae85ceb 270SV * obj
85982a32 271SV * src
272int check
aae85ceb 273CODE:
274{
85982a32 275 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
276 sv_utf8_upgrade(src);
277 ST(0) = encode_method(aTHX_ enc, enc->f_utf8, src, check);
aae85ceb 278 XSRETURN(1);
279}
280
0ab8f81e 281void
282Method_needs_lines(obj)
283SV * obj
284CODE:
285{
286 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
287 ST(0) = &PL_sv_no;
288 XSRETURN(1);
289}
290
291void
292Method_perlio_ok(obj)
293SV * obj
294CODE:
295{
296 encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj)));
011b2d2f 297 require_pv(PERLIO_FILENAME);
0ab8f81e 298 if (hv_exists(get_hv("INC", 0),
299 PERLIO_FILENAME, strlen(PERLIO_FILENAME)))
300 {
301 ST(0) = &PL_sv_yes;
302 }else{
303 ST(0) = &PL_sv_no;
304 }
305 XSRETURN(1);
306}
307
67e989fb 308MODULE = Encode PACKAGE = Encode
2c674647 309
310PROTOTYPES: ENABLE
311
67e989fb 312I32
2c674647 313_bytes_to_utf8(sv, ...)
85982a32 314SV * sv
315CODE:
316{
317 SV * encoding = items == 2 ? ST(1) : Nullsv;
0ab8f81e 318
85982a32 319 if (encoding)
320 RETVAL = _encoded_bytes_to_utf8(sv, SvPV_nolen(encoding));
321 else {
322 STRLEN len;
323 U8* s = (U8*)SvPV(sv, len);
324 U8* converted;
325
326 converted = bytes_to_utf8(s, &len); /* This allocs */
327 sv_setpvn(sv, (char *)converted, len);
328 SvUTF8_on(sv); /* XXX Should we? */
329 Safefree(converted); /* ... so free it */
330 RETVAL = len;
331 }
332}
333OUTPUT:
334 RETVAL
2c674647 335
67e989fb 336I32
2c674647 337_utf8_to_bytes(sv, ...)
85982a32 338SV * sv
339CODE:
340{
341 SV * to = items > 1 ? ST(1) : Nullsv;
342 SV * check = items > 2 ? ST(2) : Nullsv;
343
344 if (to) {
345 RETVAL = _encoded_utf8_to_bytes(sv, SvPV_nolen(to));
346 } else {
347 STRLEN len;
348 U8 *s = (U8*)SvPV(sv, len);
349
350 RETVAL = 0;
351 if (SvTRUE(check)) {
352 /* Must do things the slow way */
353 U8 *dest;
354 /* We need a copy to pass to check() */
0ab8f81e 355 U8 *src = (U8*)savepv((char *)s);
85982a32 356 U8 *send = s + len;
357
358 New(83, dest, len, U8); /* I think */
359
360 while (s < send) {
361 if (*s < 0x80){
362 *dest++ = *s++;
363 } else {
364 STRLEN ulen;
365 UV uv = *s++;
366
367 /* Have to do it all ourselves because of error routine,
368 aargh. */
369 if (!(uv & 0x40)){ goto failure; }
370 if (!(uv & 0x20)) { ulen = 2; uv &= 0x1f; }
371 else if (!(uv & 0x10)) { ulen = 3; uv &= 0x0f; }
372 else if (!(uv & 0x08)) { ulen = 4; uv &= 0x07; }
373 else if (!(uv & 0x04)) { ulen = 5; uv &= 0x03; }
374 else if (!(uv & 0x02)) { ulen = 6; uv &= 0x01; }
375 else if (!(uv & 0x01)) { ulen = 7; uv = 0; }
376 else { ulen = 13; uv = 0; }
87714904 377
85982a32 378 /* Note change to utf8.c variable naming, for variety */
379 while (ulen--) {
0ab8f81e 380 if ((*s & 0xc0) != 0x80){
381 goto failure;
85982a32 382 } else {
383 uv = (uv << 6) | (*s++ & 0x3f);
384 }
87714904 385 }
67e989fb 386 if (uv > 256) {
387 failure:
85982a32 388 call_failure(check, s, dest, src);
389 /* Now what happens? */
67e989fb 390 }
391 *dest++ = (U8)uv;
85982a32 392 }
393 }
394 } else {
395 RETVAL = (utf8_to_bytes(s, &len) ? len : 0);
2c674647 396 }
85982a32 397 }
398}
399OUTPUT:
400 RETVAL
2c674647 401
2c674647 402bool
b2704119 403is_utf8(sv, check = 0)
4411f3b6 404SV * sv
b2704119 405int check
85982a32 406CODE:
407{
408 if (SvGMAGICAL(sv)) /* it could be $1, for example */
409 sv = newSVsv(sv); /* GMAGIG will be done */
410 if (SvPOK(sv)) {
411 RETVAL = SvUTF8(sv) ? TRUE : FALSE;
412 if (RETVAL &&
413 check &&
414 !is_utf8_string((U8*)SvPVX(sv), SvCUR(sv)))
2c674647 415 RETVAL = FALSE;
85982a32 416 } else {
417 RETVAL = FALSE;
418 }
419 if (sv != ST(0))
420 SvREFCNT_dec(sv); /* it was a temp copy */
421}
422OUTPUT:
423 RETVAL
2c674647 424
425SV *
4411f3b6 426_utf8_on(sv)
85982a32 427SV * sv
428CODE:
429{
430 if (SvPOK(sv)) {
431 SV *rsv = newSViv(SvUTF8(sv));
432 RETVAL = rsv;
433 SvUTF8_on(sv);
434 } else {
435 RETVAL = &PL_sv_undef;
436 }
437}
438OUTPUT:
439 RETVAL
2c674647 440
441SV *
4411f3b6 442_utf8_off(sv)
85982a32 443SV * sv
444CODE:
445{
446 if (SvPOK(sv)) {
447 SV *rsv = newSViv(SvUTF8(sv));
448 RETVAL = rsv;
449 SvUTF8_off(sv);
450 } else {
451 RETVAL = &PL_sv_undef;
452 }
453}
454OUTPUT:
455 RETVAL
456
85982a32 457int
458DIE_ON_ERR()
459CODE:
460 RETVAL = ENCODE_DIE_ON_ERR;
461OUTPUT:
462 RETVAL
463
0ab8f81e 464int
85982a32 465WARN_ON_ERR()
466CODE:
467 RETVAL = ENCODE_WARN_ON_ERR;
468OUTPUT:
469 RETVAL
470
471int
472LEAVE_SRC()
473CODE:
474 RETVAL = ENCODE_LEAVE_SRC;
475OUTPUT:
476 RETVAL
477
478int
479RETURN_ON_ERR()
480CODE:
481 RETVAL = ENCODE_RETURN_ON_ERR;
482OUTPUT:
483 RETVAL
484
485int
486PERLQQ()
487CODE:
488 RETVAL = ENCODE_PERLQQ;
489OUTPUT:
490 RETVAL
491
492int
af1f55d9 493HTMLCREF()
494CODE:
495 RETVAL = ENCODE_HTMLCREF;
496OUTPUT:
497 RETVAL
498
499int
500XMLCREF()
501CODE:
502 RETVAL = ENCODE_XMLCREF;
503OUTPUT:
504 RETVAL
505
506int
85982a32 507FB_DEFAULT()
508CODE:
509 RETVAL = ENCODE_FB_DEFAULT;
510OUTPUT:
511 RETVAL
512
513int
514FB_CROAK()
515CODE:
516 RETVAL = ENCODE_FB_CROAK;
517OUTPUT:
518 RETVAL
519
520int
521FB_QUIET()
522CODE:
523 RETVAL = ENCODE_FB_QUIET;
524OUTPUT:
525 RETVAL
526
527int
528FB_WARN()
529CODE:
530 RETVAL = ENCODE_FB_WARN;
531OUTPUT:
532 RETVAL
533
534int
535FB_PERLQQ()
536CODE:
537 RETVAL = ENCODE_FB_PERLQQ;
538OUTPUT:
539 RETVAL
2c674647 540
af1f55d9 541int
542FB_HTMLCREF()
543CODE:
544 RETVAL = ENCODE_FB_HTMLCREF;
545OUTPUT:
546 RETVAL
547
548int
549FB_XMLCREF()
550CODE:
551 RETVAL = ENCODE_FB_XMLCREF;
552OUTPUT:
553 RETVAL
554
33af2bc7 555BOOT:
556{
85982a32 557#include "def_t.h"
e7cbefb8 558#include "def_t.exh"
33af2bc7 559}