Commit | Line | Data |
fcf2db38 |
1 | #define PERL_NO_GET_CONTEXT |
2 | |
2c674647 |
3 | #include "EXTERN.h" |
4 | #include "perl.h" |
5 | #include "XSUB.h" |
2f2b4ff2 |
6 | #define U8 U8 |
7 | #include "encode.h" |
8 | #include "iso8859.h" |
9 | #include "EBCDIC.h" |
10 | #include "Symbols.h" |
2c674647 |
11 | |
fcf2db38 |
12 | |
13 | #define UNIMPLEMENTED(x,y) y x (SV *sv, char *encoding) {dTHX; \ |
2f5768b8 |
14 | Perl_croak(aTHX_ "panic_unimplemented"); \ |
4a83738a |
15 | return (y)0; /* fool picky compilers */ \ |
87714904 |
16 | } |
67e989fb |
17 | UNIMPLEMENTED(_encoded_utf8_to_bytes, I32) |
18 | UNIMPLEMENTED(_encoded_bytes_to_utf8, I32) |
19 | |
9df9a5cd |
20 | #if defined(USE_PERLIO) && !defined(USE_SFIO) |
72e44f29 |
21 | /* Define an encoding "layer" in the perliol.h sense. |
22 | The layer defined here "inherits" in an object-oriented sense from the |
23 | "perlio" layer with its PerlIOBuf_* "methods". |
24 | The implementation is particularly efficient as until Encode settles down |
25 | there is no point in tryint to tune it. |
26 | |
27 | The layer works by overloading the "fill" and "flush" methods. |
28 | |
29 | "fill" calls "SUPER::fill" in perl terms, then calls the encode OO perl API |
30 | to convert the encoded data to UTF-8 form, then copies it back to the |
31 | buffer. The "base class's" read methods then see the UTF-8 data. |
32 | |
33 | "flush" transforms the UTF-8 data deposited by the "base class's write |
34 | method in the buffer back into the encoded form using the encode OO perl API, |
35 | then copies data back into the buffer and calls "SUPER::flush. |
36 | |
37 | Note that "flush" is _also_ called for read mode - we still do the (back)-translate |
38 | so that the the base class's "flush" sees the correct number of encoded chars |
39 | for positioning the seek pointer. (This double translation is the worst performance |
40 | issue - particularly with all-perl encode engine.) |
41 | |
42 | */ |
43 | |
44 | |
33af2bc7 |
45 | #include "perliol.h" |
46 | |
47 | typedef struct |
48 | { |
49 | PerlIOBuf base; /* PerlIOBuf stuff */ |
50 | SV * bufsv; |
51 | SV * enc; |
52 | } PerlIOEncode; |
53 | |
e3f3bf95 |
54 | SV * |
55 | PerlIOEncode_getarg(PerlIO *f) |
56 | { |
fcf2db38 |
57 | dTHX; |
e3f3bf95 |
58 | PerlIOEncode *e = PerlIOSelf(f,PerlIOEncode); |
59 | SV *sv = &PL_sv_undef; |
60 | if (e->enc) |
61 | { |
62 | dSP; |
63 | ENTER; |
64 | SAVETMPS; |
65 | PUSHMARK(sp); |
66 | XPUSHs(e->enc); |
67 | PUTBACK; |
68 | if (perl_call_method("name",G_SCALAR) == 1) |
69 | { |
70 | SPAGAIN; |
71 | sv = newSVsv(POPs); |
72 | PUTBACK; |
73 | } |
74 | } |
75 | return sv; |
76 | } |
33af2bc7 |
77 | |
78 | IV |
e3f3bf95 |
79 | PerlIOEncode_pushed(PerlIO *f, const char *mode, SV *arg) |
33af2bc7 |
80 | { |
81 | PerlIOEncode *e = PerlIOSelf(f,PerlIOEncode); |
82 | dTHX; |
83 | dSP; |
84 | IV code; |
e3f3bf95 |
85 | code = PerlIOBuf_pushed(f,mode,Nullsv); |
33af2bc7 |
86 | ENTER; |
87 | SAVETMPS; |
88 | PUSHMARK(sp); |
e3f3bf95 |
89 | XPUSHs(arg); |
33af2bc7 |
90 | PUTBACK; |
51ef4e11 |
91 | if (perl_call_pv("Encode::find_encoding",G_SCALAR) != 1) |
52744f63 |
92 | { |
93 | /* should never happen */ |
51ef4e11 |
94 | Perl_die(aTHX_ "Encode::find_encoding did not return a value"); |
52744f63 |
95 | return -1; |
96 | } |
33af2bc7 |
97 | SPAGAIN; |
98 | e->enc = POPs; |
99 | PUTBACK; |
100 | if (!SvROK(e->enc)) |
52744f63 |
101 | { |
102 | e->enc = Nullsv; |
103 | errno = EINVAL; |
29b291f7 |
104 | Perl_warner(aTHX_ WARN_IO, "Cannot find encoding \"%"SVf"\"", arg); |
52744f63 |
105 | return -1; |
106 | } |
33af2bc7 |
107 | SvREFCNT_inc(e->enc); |
108 | FREETMPS; |
109 | LEAVE; |
110 | PerlIOBase(f)->flags |= PERLIO_F_UTF8; |
111 | return code; |
112 | } |
113 | |
114 | IV |
115 | PerlIOEncode_popped(PerlIO *f) |
116 | { |
117 | PerlIOEncode *e = PerlIOSelf(f,PerlIOEncode); |
118 | dTHX; |
119 | if (e->enc) |
120 | { |
121 | SvREFCNT_dec(e->enc); |
122 | e->enc = Nullsv; |
123 | } |
124 | if (e->bufsv) |
125 | { |
126 | SvREFCNT_dec(e->bufsv); |
127 | e->bufsv = Nullsv; |
128 | } |
129 | return 0; |
130 | } |
131 | |
132 | STDCHAR * |
133 | PerlIOEncode_get_base(PerlIO *f) |
134 | { |
135 | PerlIOEncode *e = PerlIOSelf(f,PerlIOEncode); |
136 | dTHX; |
137 | if (!e->base.bufsiz) |
138 | e->base.bufsiz = 1024; |
139 | if (!e->bufsv) |
140 | { |
141 | e->bufsv = newSV(e->base.bufsiz); |
142 | sv_setpvn(e->bufsv,"",0); |
143 | } |
62e8870c |
144 | e->base.buf = (STDCHAR *)SvPVX(e->bufsv); |
33af2bc7 |
145 | if (!e->base.ptr) |
146 | e->base.ptr = e->base.buf; |
147 | if (!e->base.end) |
148 | e->base.end = e->base.buf; |
149 | if (e->base.ptr < e->base.buf || e->base.ptr > e->base.buf+SvLEN(e->bufsv)) |
150 | { |
151 | Perl_warn(aTHX_ " ptr %p(%p)%p", |
152 | e->base.buf,e->base.ptr,e->base.buf+SvLEN(e->bufsv)); |
153 | abort(); |
154 | } |
155 | if (SvLEN(e->bufsv) < e->base.bufsiz) |
156 | { |
157 | SSize_t poff = e->base.ptr - e->base.buf; |
158 | SSize_t eoff = e->base.end - e->base.buf; |
62e8870c |
159 | e->base.buf = (STDCHAR *)SvGROW(e->bufsv,e->base.bufsiz); |
33af2bc7 |
160 | e->base.ptr = e->base.buf + poff; |
161 | e->base.end = e->base.buf + eoff; |
162 | } |
163 | if (e->base.ptr < e->base.buf || e->base.ptr > e->base.buf+SvLEN(e->bufsv)) |
164 | { |
165 | Perl_warn(aTHX_ " ptr %p(%p)%p", |
166 | e->base.buf,e->base.ptr,e->base.buf+SvLEN(e->bufsv)); |
167 | abort(); |
168 | } |
169 | return e->base.buf; |
170 | } |
171 | |
33af2bc7 |
172 | IV |
173 | PerlIOEncode_fill(PerlIO *f) |
174 | { |
175 | PerlIOEncode *e = PerlIOSelf(f,PerlIOEncode); |
176 | dTHX; |
177 | dSP; |
178 | IV code; |
33af2bc7 |
179 | code = PerlIOBuf_fill(f); |
180 | if (code == 0) |
181 | { |
182 | SV *uni; |
72e44f29 |
183 | STRLEN len; |
184 | char *s; |
185 | /* Set SV that is the buffer to be buf..ptr */ |
33af2bc7 |
186 | SvCUR_set(e->bufsv, e->base.end - e->base.buf); |
187 | SvUTF8_off(e->bufsv); |
188 | ENTER; |
189 | SAVETMPS; |
190 | PUSHMARK(sp); |
191 | XPUSHs(e->enc); |
192 | XPUSHs(e->bufsv); |
193 | XPUSHs(&PL_sv_yes); |
194 | PUTBACK; |
50d26985 |
195 | if (perl_call_method("decode",G_SCALAR) != 1) |
33af2bc7 |
196 | code = -1; |
197 | SPAGAIN; |
198 | uni = POPs; |
199 | PUTBACK; |
72e44f29 |
200 | /* Now get translated string (forced to UTF-8) and copy back to buffer |
201 | don't use sv_setsv as that may "steal" PV from returned temp |
202 | and so free() our known-large-enough buffer. |
203 | sv_setpvn() should do but let us do it long hand. |
204 | */ |
205 | s = SvPVutf8(uni,len); |
206 | if (s != SvPVX(e->bufsv)) |
207 | { |
62e8870c |
208 | e->base.buf = (STDCHAR *)SvGROW(e->bufsv,len); |
72e44f29 |
209 | Move(s,e->base.buf,len,char); |
210 | SvCUR_set(e->bufsv,len); |
211 | } |
212 | SvUTF8_on(e->bufsv); |
213 | e->base.end = e->base.buf+len; |
33af2bc7 |
214 | e->base.ptr = e->base.buf; |
215 | FREETMPS; |
216 | LEAVE; |
217 | } |
218 | return code; |
219 | } |
220 | |
221 | IV |
222 | PerlIOEncode_flush(PerlIO *f) |
223 | { |
224 | PerlIOEncode *e = PerlIOSelf(f,PerlIOEncode); |
225 | IV code = 0; |
8040349a |
226 | if (e->bufsv && (PerlIOBase(f)->flags & (PERLIO_F_RDBUF|PERLIO_F_WRBUF)) |
227 | &&(e->base.ptr > e->base.buf) |
228 | ) |
33af2bc7 |
229 | { |
8040349a |
230 | dTHX; |
33af2bc7 |
231 | dSP; |
232 | SV *str; |
233 | char *s; |
234 | STRLEN len; |
72e44f29 |
235 | SSize_t left = 0; |
236 | if (PerlIOBase(f)->flags & PERLIO_F_RDBUF) |
237 | { |
238 | /* This is really just a flag to see if we took all the data, if |
239 | we did PerlIOBase_flush avoids a seek to lower layer. |
240 | Need to revisit if we start getting clever with unreads or seeks-in-buffer |
241 | */ |
242 | left = e->base.end - e->base.ptr; |
243 | } |
33af2bc7 |
244 | ENTER; |
245 | SAVETMPS; |
246 | PUSHMARK(sp); |
247 | XPUSHs(e->enc); |
72e44f29 |
248 | SvCUR_set(e->bufsv, e->base.ptr - e->base.buf); |
33af2bc7 |
249 | SvUTF8_on(e->bufsv); |
250 | XPUSHs(e->bufsv); |
251 | XPUSHs(&PL_sv_yes); |
252 | PUTBACK; |
50d26985 |
253 | if (perl_call_method("encode",G_SCALAR) != 1) |
33af2bc7 |
254 | code = -1; |
255 | SPAGAIN; |
256 | str = POPs; |
257 | PUTBACK; |
72e44f29 |
258 | s = SvPV(str,len); |
259 | if (s != SvPVX(e->bufsv)) |
260 | { |
62e8870c |
261 | e->base.buf = (STDCHAR *)SvGROW(e->bufsv,len); |
72e44f29 |
262 | Move(s,e->base.buf,len,char); |
263 | SvCUR_set(e->bufsv,len); |
264 | } |
33af2bc7 |
265 | SvUTF8_off(e->bufsv); |
72e44f29 |
266 | e->base.ptr = e->base.buf+len; |
267 | /* restore end != ptr as inequality is used by PerlIOBuf_flush in read case */ |
268 | e->base.end = e->base.ptr + left; |
33af2bc7 |
269 | FREETMPS; |
270 | LEAVE; |
271 | if (PerlIOBuf_flush(f) != 0) |
272 | code = -1; |
273 | } |
274 | return code; |
275 | } |
276 | |
277 | IV |
278 | PerlIOEncode_close(PerlIO *f) |
279 | { |
280 | PerlIOEncode *e = PerlIOSelf(f,PerlIOEncode); |
281 | IV code = PerlIOBase_close(f); |
282 | dTHX; |
283 | if (e->bufsv) |
284 | { |
285 | SvREFCNT_dec(e->bufsv); |
286 | e->bufsv = Nullsv; |
287 | } |
288 | e->base.buf = NULL; |
289 | e->base.ptr = NULL; |
290 | e->base.end = NULL; |
291 | PerlIOBase(f)->flags &= ~(PERLIO_F_RDBUF|PERLIO_F_WRBUF); |
292 | return code; |
293 | } |
294 | |
72e44f29 |
295 | Off_t |
296 | PerlIOEncode_tell(PerlIO *f) |
297 | { |
a999f61b |
298 | dTHX; |
72e44f29 |
299 | PerlIOBuf *b = PerlIOSelf(f,PerlIOBuf); |
300 | /* Unfortunately the only way to get a postion is to back-translate, |
301 | the UTF8-bytes we have buf..ptr and adjust accordingly. |
302 | But we will try and save any unread data in case stream |
303 | is un-seekable. |
304 | */ |
305 | if ((PerlIOBase(f)->flags & PERLIO_F_RDBUF) && b->ptr < b->end) |
306 | { |
307 | Size_t count = b->end - b->ptr; |
e3f3bf95 |
308 | PerlIO_push(aTHX_ f,&PerlIO_pending,"r",Nullsv); |
72e44f29 |
309 | /* Save what we have left to read */ |
310 | PerlIOSelf(f,PerlIOBuf)->bufsiz = count; |
311 | PerlIO_unread(f,b->ptr,count); |
312 | /* There isn't any unread data - we just saved it - so avoid the lower seek */ |
313 | b->end = b->ptr; |
314 | /* Flush ourselves - now one layer down, |
315 | this does the back translate and adjusts position |
316 | */ |
317 | PerlIO_flush(PerlIONext(f)); |
318 | /* Set position of the saved data */ |
319 | PerlIOSelf(f,PerlIOBuf)->posn = b->posn; |
320 | } |
321 | else |
322 | { |
323 | PerlIO_flush(f); |
324 | } |
325 | return b->posn; |
326 | } |
327 | |
8cf8f3d1 |
328 | PerlIO * |
329 | PerlIOEncode_dup(pTHX_ PerlIO *f, PerlIO *o, CLONE_PARAMS *params) |
330 | { |
331 | /* FIXME - Almost certainly needs more work */ |
332 | return PerlIOBase_dup(aTHX_ f, o, params); |
333 | } |
334 | |
33af2bc7 |
335 | PerlIO_funcs PerlIO_encode = { |
72e44f29 |
336 | "encoding", |
33af2bc7 |
337 | sizeof(PerlIOEncode), |
338 | PERLIO_K_BUFFERED, |
33af2bc7 |
339 | PerlIOEncode_pushed, |
340 | PerlIOEncode_popped, |
e3f3bf95 |
341 | PerlIOBuf_open, |
342 | PerlIOEncode_getarg, |
343 | PerlIOBase_fileno, |
8cf8f3d1 |
344 | PerlIOEncode_dup, |
33af2bc7 |
345 | PerlIOBuf_read, |
346 | PerlIOBuf_unread, |
347 | PerlIOBuf_write, |
348 | PerlIOBuf_seek, |
72e44f29 |
349 | PerlIOEncode_tell, |
33af2bc7 |
350 | PerlIOEncode_close, |
351 | PerlIOEncode_flush, |
352 | PerlIOEncode_fill, |
353 | PerlIOBase_eof, |
354 | PerlIOBase_error, |
355 | PerlIOBase_clearerr, |
f6c77cf1 |
356 | PerlIOBase_setlinebuf, |
33af2bc7 |
357 | PerlIOEncode_get_base, |
358 | PerlIOBuf_bufsiz, |
359 | PerlIOBuf_get_ptr, |
360 | PerlIOBuf_get_cnt, |
361 | PerlIOBuf_set_ptrcnt, |
362 | }; |
9df9a5cd |
363 | #endif /* encode layer */ |
33af2bc7 |
364 | |
2f2b4ff2 |
365 | void |
366 | Encode_Define(pTHX_ encode_t *enc) |
367 | { |
51ef4e11 |
368 | dSP; |
2f2b4ff2 |
369 | HV *stash = gv_stashpv("Encode::XS", TRUE); |
370 | SV *sv = sv_bless(newRV_noinc(newSViv(PTR2IV(enc))),stash); |
dcda1f94 |
371 | int i = 0; |
51ef4e11 |
372 | PUSHMARK(sp); |
373 | XPUSHs(sv); |
dcda1f94 |
374 | while (enc->name[i]) |
375 | { |
376 | const char *name = enc->name[i++]; |
51ef4e11 |
377 | XPUSHs(sv_2mortal(newSVpvn(name,strlen(name)))); |
dcda1f94 |
378 | } |
51ef4e11 |
379 | PUTBACK; |
380 | call_pv("Encode::define_encoding",G_DISCARD); |
dcda1f94 |
381 | SvREFCNT_dec(sv); |
2f2b4ff2 |
382 | } |
383 | |
183a2d84 |
384 | void call_failure (SV *routine, U8* done, U8* dest, U8* orig) {} |
67e989fb |
385 | |
2f2b4ff2 |
386 | static SV * |
387 | encode_method(pTHX_ encode_t *enc, encpage_t *dir, SV *src, int check) |
388 | { |
389 | STRLEN slen; |
390 | U8 *s = (U8 *) SvPV(src,slen); |
391 | SV *dst = sv_2mortal(newSV(2*slen+1)); |
392 | if (slen) |
393 | { |
394 | U8 *d = (U8 *) SvGROW(dst, 2*slen+1); |
395 | STRLEN dlen = SvLEN(dst); |
396 | int code; |
9b37254d |
397 | while ((code = do_encode(dir,s,&slen,d,dlen,&dlen,!check))) |
2f2b4ff2 |
398 | { |
399 | SvCUR_set(dst,dlen); |
400 | SvPOK_on(dst); |
9b37254d |
401 | |
402 | if (code == ENCODE_FALLBACK) |
403 | break; |
404 | |
2f2b4ff2 |
405 | switch(code) |
406 | { |
407 | case ENCODE_NOSPACE: |
408 | { |
409 | STRLEN need = (slen) ? (SvLEN(dst)*SvCUR(src)/slen) : (dlen + UTF8_MAXLEN); |
410 | if (need <= SvLEN(dst)) |
411 | need += UTF8_MAXLEN; |
412 | d = (U8 *) SvGROW(dst, need); |
413 | dlen = SvLEN(dst); |
414 | slen = SvCUR(src); |
415 | break; |
416 | } |
417 | |
418 | case ENCODE_NOREP: |
419 | if (dir == enc->f_utf8) |
420 | { |
421 | if (!check && ckWARN_d(WARN_UTF8)) |
422 | { |
423 | STRLEN clen; |
9041c2e3 |
424 | UV ch = utf8n_to_uvuni(s+slen,(SvCUR(src)-slen),&clen,0); |
425 | Perl_warner(aTHX_ WARN_UTF8, "\"\\N{U+%"UVxf"}\" does not map to %s", ch, enc->name[0]); |
2f2b4ff2 |
426 | /* FIXME: Skip over the character, copy in replacement and continue |
427 | * but that is messy so for now just fail. |
428 | */ |
429 | return &PL_sv_undef; |
430 | } |
431 | else |
432 | { |
433 | return &PL_sv_undef; |
434 | } |
435 | } |
436 | else |
437 | { |
438 | /* UTF-8 is supposed to be "Universal" so should not happen */ |
439 | Perl_croak(aTHX_ "%s '%.*s' does not map to UTF-8", |
dcda1f94 |
440 | enc->name[0], (int)(SvCUR(src)-slen),s+slen); |
2f2b4ff2 |
441 | } |
442 | break; |
443 | |
444 | case ENCODE_PARTIAL: |
445 | if (!check && ckWARN_d(WARN_UTF8)) |
446 | { |
447 | Perl_warner(aTHX_ WARN_UTF8, "Partial %s character", |
dcda1f94 |
448 | (dir == enc->f_utf8) ? "UTF-8" : enc->name[0]); |
2f2b4ff2 |
449 | } |
450 | return &PL_sv_undef; |
451 | |
452 | default: |
453 | Perl_croak(aTHX_ "Unexpected code %d converting %s %s", |
dcda1f94 |
454 | code, (dir == enc->f_utf8) ? "to" : "from",enc->name[0]); |
2f2b4ff2 |
455 | return &PL_sv_undef; |
456 | } |
457 | } |
458 | SvCUR_set(dst,dlen); |
459 | SvPOK_on(dst); |
460 | if (check) |
461 | { |
462 | if (slen < SvCUR(src)) |
463 | { |
464 | Move(s+slen,s,SvCUR(src)-slen,U8); |
465 | } |
466 | SvCUR_set(src,SvCUR(src)-slen); |
467 | } |
468 | } |
8040349a |
469 | else |
470 | { |
471 | SvCUR_set(dst,slen); |
472 | SvPOK_on(dst); |
473 | } |
2f2b4ff2 |
474 | return dst; |
475 | } |
476 | |
50d26985 |
477 | MODULE = Encode PACKAGE = Encode::XS PREFIX = Method_ |
2f2b4ff2 |
478 | |
479 | PROTOTYPES: ENABLE |
480 | |
481 | void |
691638dd |
482 | Method_decode(obj,src,check = FALSE) |
2f2b4ff2 |
483 | SV * obj |
484 | SV * src |
691638dd |
485 | bool check |
2f2b4ff2 |
486 | CODE: |
487 | { |
488 | encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); |
489 | ST(0) = encode_method(aTHX_ enc, enc->t_utf8, src, check); |
490 | SvUTF8_on(ST(0)); |
491 | XSRETURN(1); |
492 | } |
493 | |
494 | void |
691638dd |
495 | Method_encode(obj,src,check = FALSE) |
2f2b4ff2 |
496 | SV * obj |
497 | SV * src |
691638dd |
498 | bool check |
2f2b4ff2 |
499 | CODE: |
500 | { |
501 | encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); |
502 | sv_utf8_upgrade(src); |
503 | ST(0) = encode_method(aTHX_ enc, enc->f_utf8, src, check); |
504 | XSRETURN(1); |
505 | } |
506 | |
67e989fb |
507 | MODULE = Encode PACKAGE = Encode |
2c674647 |
508 | |
509 | PROTOTYPES: ENABLE |
510 | |
67e989fb |
511 | I32 |
2c674647 |
512 | _bytes_to_utf8(sv, ...) |
67e989fb |
513 | SV * sv |
2c674647 |
514 | CODE: |
67e989fb |
515 | { |
516 | SV * encoding = items == 2 ? ST(1) : Nullsv; |
517 | |
518 | if (encoding) |
519 | RETVAL = _encoded_bytes_to_utf8(sv, SvPV_nolen(encoding)); |
520 | else { |
521 | STRLEN len; |
183a2d84 |
522 | U8* s = (U8*)SvPV(sv, len); |
67e989fb |
523 | U8* converted; |
524 | |
525 | converted = bytes_to_utf8(s, &len); /* This allocs */ |
183a2d84 |
526 | sv_setpvn(sv, (char *)converted, len); |
67e989fb |
527 | SvUTF8_on(sv); /* XXX Should we? */ |
528 | Safefree(converted); /* ... so free it */ |
529 | RETVAL = len; |
530 | } |
531 | } |
2c674647 |
532 | OUTPUT: |
67e989fb |
533 | RETVAL |
2c674647 |
534 | |
67e989fb |
535 | I32 |
2c674647 |
536 | _utf8_to_bytes(sv, ...) |
67e989fb |
537 | SV * sv |
2c674647 |
538 | CODE: |
67e989fb |
539 | { |
540 | SV * to = items > 1 ? ST(1) : Nullsv; |
541 | SV * check = items > 2 ? ST(2) : Nullsv; |
87714904 |
542 | |
67e989fb |
543 | if (to) |
544 | RETVAL = _encoded_utf8_to_bytes(sv, SvPV_nolen(to)); |
545 | else { |
67e989fb |
546 | STRLEN len; |
b113ac0e |
547 | U8 *s = (U8*)SvPV(sv, len); |
67e989fb |
548 | |
9c5ffd7c |
549 | RETVAL = 0; |
67e989fb |
550 | if (SvTRUE(check)) { |
551 | /* Must do things the slow way */ |
552 | U8 *dest; |
87714904 |
553 | U8 *src = (U8*)savepv((char *)s); /* We need a copy to pass to check() */ |
67e989fb |
554 | U8 *send = s + len; |
555 | |
556 | New(83, dest, len, U8); /* I think */ |
557 | |
558 | while (s < send) { |
559 | if (*s < 0x80) |
560 | *dest++ = *s++; |
561 | else { |
b113ac0e |
562 | STRLEN ulen; |
563 | UV uv = *s++; |
87714904 |
564 | |
67e989fb |
565 | /* Have to do it all ourselves because of error routine, |
566 | aargh. */ |
567 | if (!(uv & 0x40)) |
568 | goto failure; |
569 | if (!(uv & 0x20)) { ulen = 2; uv &= 0x1f; } |
570 | else if (!(uv & 0x10)) { ulen = 3; uv &= 0x0f; } |
571 | else if (!(uv & 0x08)) { ulen = 4; uv &= 0x07; } |
572 | else if (!(uv & 0x04)) { ulen = 5; uv &= 0x03; } |
573 | else if (!(uv & 0x02)) { ulen = 6; uv &= 0x01; } |
574 | else if (!(uv & 0x01)) { ulen = 7; uv = 0; } |
575 | else { ulen = 13; uv = 0; } |
87714904 |
576 | |
67e989fb |
577 | /* Note change to utf8.c variable naming, for variety */ |
578 | while (ulen--) { |
579 | if ((*s & 0xc0) != 0x80) |
580 | goto failure; |
87714904 |
581 | |
67e989fb |
582 | else |
583 | uv = (uv << 6) | (*s++ & 0x3f); |
87714904 |
584 | } |
67e989fb |
585 | if (uv > 256) { |
586 | failure: |
587 | call_failure(check, s, dest, src); |
588 | /* Now what happens? */ |
589 | } |
590 | *dest++ = (U8)uv; |
591 | } |
592 | } |
593 | } else |
594 | RETVAL = (utf8_to_bytes(s, &len) ? len : 0); |
595 | } |
2c674647 |
596 | } |
597 | OUTPUT: |
598 | RETVAL |
599 | |
2c674647 |
600 | bool |
4411f3b6 |
601 | is_utf8(sv, check = FALSE) |
602 | SV * sv |
603 | bool check |
2c674647 |
604 | CODE: |
605 | { |
2c674647 |
606 | if (SvPOK(sv)) { |
4411f3b6 |
607 | RETVAL = SvUTF8(sv) ? TRUE : FALSE; |
2c674647 |
608 | if (RETVAL && |
4411f3b6 |
609 | check && |
2c674647 |
610 | !is_utf8_string((U8*)SvPVX(sv), SvCUR(sv))) |
611 | RETVAL = FALSE; |
612 | } else { |
613 | RETVAL = FALSE; |
614 | } |
615 | } |
616 | OUTPUT: |
617 | RETVAL |
618 | |
619 | SV * |
4411f3b6 |
620 | _utf8_on(sv) |
2c674647 |
621 | SV * sv |
622 | CODE: |
623 | { |
624 | if (SvPOK(sv)) { |
87714904 |
625 | SV *rsv = newSViv(SvUTF8(sv)); |
2c674647 |
626 | RETVAL = rsv; |
627 | SvUTF8_on(sv); |
628 | } else { |
629 | RETVAL = &PL_sv_undef; |
630 | } |
631 | } |
632 | OUTPUT: |
633 | RETVAL |
634 | |
635 | SV * |
4411f3b6 |
636 | _utf8_off(sv) |
2c674647 |
637 | SV * sv |
638 | CODE: |
639 | { |
640 | if (SvPOK(sv)) { |
87714904 |
641 | SV *rsv = newSViv(SvUTF8(sv)); |
2c674647 |
642 | RETVAL = rsv; |
643 | SvUTF8_off(sv); |
644 | } else { |
645 | RETVAL = &PL_sv_undef; |
646 | } |
647 | } |
648 | OUTPUT: |
649 | RETVAL |
650 | |
33af2bc7 |
651 | BOOT: |
652 | { |
6a59c517 |
653 | #if defined(USE_PERLIO) && !defined(USE_SFIO) |
a999f61b |
654 | PerlIO_define_layer(aTHX_ &PerlIO_encode); |
33af2bc7 |
655 | #endif |
2f2b4ff2 |
656 | #include "iso8859.def" |
657 | #include "EBCDIC.def" |
658 | #include "Symbols.def" |
33af2bc7 |
659 | } |