Commit | Line | Data |
85982a32 |
1 | /* |
cc836e95 |
2 | $Id: Encode.xs,v 2.2 2004/10/24 13:00:29 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"); \ |
4a83738a |
25 | return (y)0; /* fool picky compilers */ \ |
87714904 |
26 | } |
85982a32 |
27 | /**/ |
011b2d2f |
28 | |
7e9a885a |
29 | UNIMPLEMENTED(_encoded_utf8_to_bytes, I32) |
30 | UNIMPLEMENTED(_encoded_bytes_to_utf8, I32) |
33af2bc7 |
31 | |
b0b300a3 |
32 | void |
aa0053b7 |
33 | Encode_XSEncoding(pTHX_ encode_t * enc) |
2f2b4ff2 |
34 | { |
aa0053b7 |
35 | dSP; |
36 | HV *stash = gv_stashpv("Encode::XS", TRUE); |
37 | SV *sv = sv_bless(newRV_noinc(newSViv(PTR2IV(enc))), stash); |
38 | int i = 0; |
39 | PUSHMARK(sp); |
40 | XPUSHs(sv); |
41 | while (enc->name[i]) { |
42 | const char *name = enc->name[i++]; |
43 | XPUSHs(sv_2mortal(newSVpvn(name, strlen(name)))); |
44 | } |
45 | PUTBACK; |
46 | call_pv("Encode::define_encoding", G_DISCARD); |
47 | SvREFCNT_dec(sv); |
2f2b4ff2 |
48 | } |
49 | |
aa0053b7 |
50 | void |
51 | call_failure(SV * routine, U8 * done, U8 * dest, U8 * orig) |
52 | { |
85982a32 |
53 | /* Exists for breakpointing */ |
aa0053b7 |
54 | } |
67e989fb |
55 | |
85982a32 |
56 | |
2fc614e0 |
57 | #define ERR_ENCODE_NOMAP "\"\\x{%04" UVxf "}\" does not map to %s" |
58 | #define ERR_DECODE_NOMAP "%s \"\\x%02" UVXf "\" does not map to Unicode" |
59 | |
2f2b4ff2 |
60 | static SV * |
aa0053b7 |
61 | encode_method(pTHX_ encode_t * enc, encpage_t * dir, SV * src, |
220e2d4e |
62 | int check, STRLEN * offset, SV * term, int * retcode) |
2f2b4ff2 |
63 | { |
aa0053b7 |
64 | STRLEN slen; |
65 | U8 *s = (U8 *) SvPV(src, slen); |
3aececda |
66 | STRLEN tlen = slen; |
67 | STRLEN ddone = 0; |
68 | STRLEN sdone = 0; |
39cf9a5e |
69 | |
3c49ab08 |
70 | /* We allocate slen+1. |
85982a32 |
71 | PerlIO dumps core if this value is smaller than this. */ |
3c49ab08 |
72 | SV *dst = sv_2mortal(newSV(slen+1)); |
85982a32 |
73 | U8 *d = (U8 *)SvPVX(dst); |
74 | STRLEN dlen = SvLEN(dst)-1; |
220e2d4e |
75 | int code = 0; |
76 | STRLEN trmlen = 0; |
cc7dbc11 |
77 | U8 *trm = term ? (U8*) SvPV(term, trmlen) : NULL; |
220e2d4e |
78 | |
79 | if (offset) { |
80 | s += *offset; |
6be7c101 |
81 | if (slen > *offset){ /* safeguard against slen overflow */ |
82 | slen -= *offset; |
83 | }else{ |
84 | slen = 0; |
85 | } |
220e2d4e |
86 | tlen = slen; |
87 | } |
85982a32 |
88 | |
6be7c101 |
89 | if (slen == 0){ |
85982a32 |
90 | SvCUR_set(dst, 0); |
91 | SvPOK_only(dst); |
92 | goto ENCODE_END; |
93 | } |
94 | |
220e2d4e |
95 | while( (code = do_encode(dir, s, &slen, d, dlen, &dlen, !check, |
96 | trm, trmlen)) ) |
85982a32 |
97 | { |
98 | SvCUR_set(dst, dlen+ddone); |
99 | SvPOK_only(dst); |
0b3236bb |
100 | |
220e2d4e |
101 | if (code == ENCODE_FALLBACK || code == ENCODE_PARTIAL || |
102 | code == ENCODE_FOUND_TERM) { |
85982a32 |
103 | break; |
104 | } |
105 | switch (code) { |
106 | case ENCODE_NOSPACE: |
107 | { |
108 | STRLEN more = 0; /* make sure you initialize! */ |
109 | STRLEN sleft; |
110 | sdone += slen; |
111 | ddone += dlen; |
112 | sleft = tlen - sdone; |
fcb875d4 |
113 | #if ENCODE_XS_PROFILE >= 2 |
85982a32 |
114 | Perl_warn(aTHX_ |
115 | "more=%d, sdone=%d, sleft=%d, SvLEN(dst)=%d\n", |
116 | more, sdone, sleft, SvLEN(dst)); |
fcb875d4 |
117 | #endif |
85982a32 |
118 | if (sdone != 0) { /* has src ever been processed ? */ |
39cf9a5e |
119 | #if ENCODE_XS_USEFP == 2 |
85982a32 |
120 | more = (1.0*tlen*SvLEN(dst)+sdone-1)/sdone |
121 | - SvLEN(dst); |
39cf9a5e |
122 | #elif ENCODE_XS_USEFP |
6e21dc91 |
123 | more = (STRLEN)((1.0*SvLEN(dst)+1)/sdone * sleft); |
39cf9a5e |
124 | #else |
85982a32 |
125 | /* safe until SvLEN(dst) == MAX_INT/16 */ |
126 | more = (16*SvLEN(dst)+1)/sdone/16 * sleft; |
39cf9a5e |
127 | #endif |
39cf9a5e |
128 | } |
85982a32 |
129 | more += UTF8_MAXLEN; /* insurance policy */ |
130 | d = (U8 *) SvGROW(dst, SvLEN(dst) + more); |
131 | /* dst need to grow need MORE bytes! */ |
132 | if (ddone >= SvLEN(dst)) { |
133 | Perl_croak(aTHX_ "Destination couldn't be grown."); |
134 | } |
135 | dlen = SvLEN(dst)-ddone-1; |
136 | d += ddone; |
137 | s += slen; |
138 | slen = tlen-sdone; |
139 | continue; |
140 | } |
141 | case ENCODE_NOREP: |
142 | /* encoding */ |
c6a7db43 |
143 | if (dir == enc->f_utf8) { |
85982a32 |
144 | STRLEN clen; |
145 | UV ch = |
3e952a88 |
146 | utf8n_to_uvuni(s+slen, (SvCUR(src)-slen), |
b0b300a3 |
147 | &clen, UTF8_ALLOW_ANY|UTF8_CHECK_ONLY); |
85982a32 |
148 | if (check & ENCODE_DIE_ON_ERR) { |
2fc614e0 |
149 | Perl_croak(aTHX_ ERR_ENCODE_NOMAP, |
150 | (UV)ch, enc->name[0]); |
4089adc4 |
151 | return &PL_sv_undef; /* never reaches but be safe */ |
152 | } |
153 | if (check & ENCODE_WARN_ON_ERR){ |
154 | Perl_warner(aTHX_ packWARN(WARN_UTF8), |
2fc614e0 |
155 | ERR_ENCODE_NOMAP, (UV)ch, enc->name[0]); |
4089adc4 |
156 | } |
157 | if (check & ENCODE_RETURN_ON_ERR){ |
158 | goto ENCODE_SET_SRC; |
159 | } |
f9d05ba3 |
160 | if (check & (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){ |
161 | SV* subchar = |
162 | newSVpvf(check & ENCODE_PERLQQ ? "\\x{%04"UVxf"}" : |
163 | check & ENCODE_HTMLCREF ? "&#%" UVuf ";" : |
164 | "&#x%" UVxf ";", (UV)ch); |
4089adc4 |
165 | sdone += slen + clen; |
f9d05ba3 |
166 | ddone += dlen + SvCUR(subchar); |
167 | sv_catsv(dst, subchar); |
168 | SvREFCNT_dec(subchar); |
4089adc4 |
169 | } else { |
170 | /* fallback char */ |
171 | sdone += slen + clen; |
172 | ddone += dlen + enc->replen; |
173 | sv_catpvn(dst, (char*)enc->rep, enc->replen); |
c6a7db43 |
174 | } |
b2704119 |
175 | } |
85982a32 |
176 | /* decoding */ |
c6a7db43 |
177 | else { |
85982a32 |
178 | if (check & ENCODE_DIE_ON_ERR){ |
2fc614e0 |
179 | Perl_croak(aTHX_ ERR_DECODE_NOMAP, |
436c6dd3 |
180 | enc->name[0], (UV)s[slen]); |
4089adc4 |
181 | return &PL_sv_undef; /* never reaches but be safe */ |
182 | } |
183 | if (check & ENCODE_WARN_ON_ERR){ |
184 | Perl_warner( |
185 | aTHX_ packWARN(WARN_UTF8), |
2fc614e0 |
186 | ERR_DECODE_NOMAP, |
436c6dd3 |
187 | enc->name[0], (UV)s[slen]); |
4089adc4 |
188 | } |
189 | if (check & ENCODE_RETURN_ON_ERR){ |
190 | goto ENCODE_SET_SRC; |
191 | } |
192 | if (check & |
193 | (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){ |
f9d05ba3 |
194 | SV* subchar = newSVpvf("\\x%02" UVXf, (UV)s[slen]); |
4089adc4 |
195 | sdone += slen + 1; |
f9d05ba3 |
196 | ddone += dlen + SvCUR(subchar); |
197 | sv_catsv(dst, subchar); |
198 | SvREFCNT_dec(subchar); |
4089adc4 |
199 | } else { |
200 | sdone += slen + 1; |
201 | ddone += dlen + strlen(FBCHAR_UTF8); |
202 | sv_catpv(dst, FBCHAR_UTF8); |
aa0053b7 |
203 | } |
b2704119 |
204 | } |
85982a32 |
205 | /* settle variables when fallback */ |
b0b300a3 |
206 | d = (U8 *)SvEND(dst); |
207 | dlen = SvLEN(dst) - ddone - 1; |
3e952a88 |
208 | s = (U8*)SvPVX(src) + sdone; |
b2704119 |
209 | slen = tlen - sdone; |
210 | break; |
2f2b4ff2 |
211 | |
85982a32 |
212 | default: |
213 | Perl_croak(aTHX_ "Unexpected code %d converting %s %s", |
214 | code, (dir == enc->f_utf8) ? "to" : "from", |
215 | enc->name[0]); |
216 | return &PL_sv_undef; |
aa0053b7 |
217 | } |
85982a32 |
218 | } |
219 | ENCODE_SET_SRC: |
ca777f1c |
220 | if (check && !(check & ENCODE_LEAVE_SRC)){ |
221 | sdone = SvCUR(src) - (slen+sdone); |
85982a32 |
222 | if (sdone) { |
223 | sv_setpvn(src, (char*)s+slen, sdone); |
aa0053b7 |
224 | } |
85982a32 |
225 | SvCUR_set(src, sdone); |
2f2b4ff2 |
226 | } |
85982a32 |
227 | /* warn("check = 0x%X, code = 0x%d\n", check, code); */ |
c6a7db43 |
228 | |
85982a32 |
229 | SvCUR_set(dst, dlen+ddone); |
230 | SvPOK_only(dst); |
c6a7db43 |
231 | |
39cf9a5e |
232 | #if ENCODE_XS_PROFILE |
233 | if (SvCUR(dst) > SvCUR(src)){ |
85982a32 |
234 | Perl_warn(aTHX_ |
235 | "SvLEN(dst)=%d, SvCUR(dst)=%d. %d bytes unused(%f %%)\n", |
236 | SvLEN(dst), SvCUR(dst), SvLEN(dst) - SvCUR(dst), |
237 | (SvLEN(dst) - SvCUR(dst))*1.0/SvLEN(dst)*100.0); |
39cf9a5e |
238 | } |
3c49ab08 |
239 | #endif |
c6a7db43 |
240 | |
220e2d4e |
241 | if (offset) |
242 | *offset += sdone + slen; |
243 | |
85982a32 |
244 | ENCODE_END: |
0b3236bb |
245 | *SvEND(dst) = '\0'; |
220e2d4e |
246 | if (retcode) *retcode = code; |
aa0053b7 |
247 | return dst; |
2f2b4ff2 |
248 | } |
249 | |
ab3374e4 |
250 | MODULE = Encode PACKAGE = Encode::utf8 PREFIX = Method_ |
251 | |
a0d8a30e |
252 | PROTOTYPES: DISABLE |
253 | |
254 | void |
b536bf57 |
255 | Method_decode_xs(obj,src,check = 0) |
ab3374e4 |
256 | SV * obj |
257 | SV * src |
258 | int check |
259 | CODE: |
260 | { |
261 | STRLEN slen; |
262 | U8 *s = (U8 *) SvPV(src, slen); |
263 | U8 *e = (U8 *) SvEND(src); |
b536bf57 |
264 | SV *dst = newSV(slen>0?slen:1); /* newSV() abhors 0 -- inaba */ |
cc836e95 |
265 | |
266 | /* |
267 | * PerlO check -- we assume the object is of PerlIO if renewed |
268 | * and if so, we set RETURN_ON_ERR for partial character |
269 | */ |
270 | int renewed = 0; |
271 | dSP; ENTER; SAVETMPS; |
272 | PUSHMARK(sp); |
273 | XPUSHs(obj); |
274 | PUTBACK; |
275 | if (call_method("renewed",G_SCALAR) == 1) { |
276 | SPAGAIN; |
277 | renewed = POPi; |
278 | PUTBACK; |
279 | #if 0 |
280 | fprintf(stderr, "renewed == %d\n", renewed); |
281 | #endif |
282 | if (renewed){ check |= ENCODE_RETURN_ON_ERR; } |
283 | } |
284 | FREETMPS; LEAVE; |
285 | /* end PerlIO check */ |
286 | |
ab3374e4 |
287 | SvPOK_only(dst); |
288 | SvCUR_set(dst,0); |
289 | if (SvUTF8(src)) { |
290 | s = utf8_to_bytes(s,&slen); |
291 | if (s) { |
292 | SvCUR_set(src,slen); |
293 | SvUTF8_off(src); |
294 | e = s+slen; |
295 | } |
296 | else { |
297 | croak("Cannot decode string with wide characters"); |
298 | } |
299 | } |
300 | while (s < e) { |
301 | if (UTF8_IS_INVARIANT(*s) || UTF8_IS_START(*s)) { |
302 | U8 skip = UTF8SKIP(s); |
303 | if ((s + skip) > e) { |
304 | /* Partial character - done */ |
f9d05ba3 |
305 | goto decode_utf8_fallback; |
ab3374e4 |
306 | } |
307 | else if (is_utf8_char(s)) { |
308 | /* Whole char is good */ |
309 | sv_catpvn(dst,(char *)s,skip); |
310 | s += skip; |
311 | continue; |
312 | } |
313 | else { |
314 | /* starts ok but isn't "good" */ |
315 | } |
316 | } |
317 | else { |
318 | /* Invalid start byte */ |
319 | } |
320 | /* If we get here there is something wrong with alleged UTF-8 */ |
f9d05ba3 |
321 | decode_utf8_fallback: |
ab3374e4 |
322 | if (check & ENCODE_DIE_ON_ERR){ |
323 | Perl_croak(aTHX_ ERR_DECODE_NOMAP, "utf8", (UV)*s); |
324 | XSRETURN(0); |
325 | } |
326 | if (check & ENCODE_WARN_ON_ERR){ |
327 | Perl_warner(aTHX_ packWARN(WARN_UTF8), |
328 | ERR_DECODE_NOMAP, "utf8", (UV)*s); |
329 | } |
330 | if (check & ENCODE_RETURN_ON_ERR) { |
331 | break; |
332 | } |
333 | if (check & (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){ |
f9d05ba3 |
334 | SV* subchar = newSVpvf("\\x%02" UVXf, (UV)*s); |
335 | sv_catsv(dst, subchar); |
336 | SvREFCNT_dec(subchar); |
ab3374e4 |
337 | } else { |
338 | sv_catpv(dst, FBCHAR_UTF8); |
339 | } |
340 | s++; |
341 | } |
342 | *SvEND(dst) = '\0'; |
343 | |
344 | /* Clear out translated part of source unless asked not to */ |
345 | if (check && !(check & ENCODE_LEAVE_SRC)){ |
346 | slen = e-s; |
347 | if (slen) { |
348 | sv_setpvn(src, (char*)s, slen); |
349 | } |
350 | SvCUR_set(src, slen); |
351 | } |
352 | SvUTF8_on(dst); |
353 | ST(0) = sv_2mortal(dst); |
354 | XSRETURN(1); |
355 | } |
356 | |
357 | void |
b536bf57 |
358 | Method_encode_xs(obj,src,check = 0) |
ab3374e4 |
359 | SV * obj |
360 | SV * src |
361 | int check |
362 | CODE: |
363 | { |
364 | STRLEN slen; |
365 | U8 *s = (U8 *) SvPV(src, slen); |
366 | U8 *e = (U8 *) SvEND(src); |
b536bf57 |
367 | SV *dst = newSV(slen>0?slen:1); /* newSV() abhors 0 -- inaba */ |
ab3374e4 |
368 | if (SvUTF8(src)) { |
369 | /* Already encoded - trust it and just copy the octets */ |
370 | sv_setpvn(dst,(char *)s,(e-s)); |
371 | s = e; |
372 | } |
373 | else { |
374 | /* Native bytes - can always encode */ |
b536bf57 |
375 | U8 *d = (U8 *) SvGROW(dst, 2*slen+1); /* +1 or assertion will botch */ |
ab3374e4 |
376 | while (s < e) { |
377 | UV uv = NATIVE_TO_UNI((UV) *s++); |
378 | if (UNI_IS_INVARIANT(uv)) |
379 | *d++ = (U8)UTF_TO_NATIVE(uv); |
380 | else { |
381 | *d++ = (U8)UTF8_EIGHT_BIT_HI(uv); |
382 | *d++ = (U8)UTF8_EIGHT_BIT_LO(uv); |
383 | } |
384 | } |
385 | SvCUR_set(dst, d- (U8 *)SvPVX(dst)); |
386 | *SvEND(dst) = '\0'; |
387 | } |
388 | |
389 | /* Clear out translated part of source unless asked not to */ |
390 | if (check && !(check & ENCODE_LEAVE_SRC)){ |
391 | slen = e-s; |
392 | if (slen) { |
393 | sv_setpvn(src, (char*)s, slen); |
394 | } |
395 | SvCUR_set(src, slen); |
396 | } |
397 | SvPOK_only(dst); |
398 | SvUTF8_off(dst); |
399 | ST(0) = sv_2mortal(dst); |
400 | XSRETURN(1); |
401 | } |
402 | |
50d26985 |
403 | MODULE = Encode PACKAGE = Encode::XS PREFIX = Method_ |
2f2b4ff2 |
404 | |
405 | PROTOTYPES: ENABLE |
406 | |
407 | void |
a0d8a30e |
408 | Method_renew(obj) |
409 | SV * obj |
410 | CODE: |
411 | { |
412 | XSRETURN(1); |
413 | } |
414 | |
cc836e95 |
415 | int |
416 | Method_renewed(obj) |
417 | SV * obj |
418 | CODE: |
419 | RETVAL = 0; |
420 | OUTPUT: |
421 | RETVAL |
422 | |
a0d8a30e |
423 | void |
0a95303c |
424 | Method_name(obj) |
425 | SV * obj |
426 | CODE: |
85982a32 |
427 | { |
428 | encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); |
429 | ST(0) = sv_2mortal(newSVpvn(enc->name[0],strlen(enc->name[0]))); |
430 | XSRETURN(1); |
431 | } |
0a95303c |
432 | |
433 | void |
220e2d4e |
434 | Method_cat_decode(obj, dst, src, off, term, check = 0) |
435 | SV * obj |
436 | SV * dst |
437 | SV * src |
438 | SV * off |
439 | SV * term |
440 | int check |
441 | CODE: |
442 | { |
443 | encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); |
444 | STRLEN offset = (STRLEN)SvIV(off); |
445 | int code = 0; |
446 | if (SvUTF8(src)) { |
447 | sv_utf8_downgrade(src, FALSE); |
448 | } |
449 | sv_catsv(dst, encode_method(aTHX_ enc, enc->t_utf8, src, check, |
450 | &offset, term, &code)); |
451 | SvIVX(off) = (IV)offset; |
452 | if (code == ENCODE_FOUND_TERM) { |
453 | ST(0) = &PL_sv_yes; |
454 | }else{ |
455 | ST(0) = &PL_sv_no; |
456 | } |
457 | XSRETURN(1); |
458 | } |
459 | |
460 | void |
b2704119 |
461 | Method_decode(obj,src,check = 0) |
2f2b4ff2 |
462 | SV * obj |
463 | SV * src |
b2704119 |
464 | int check |
2f2b4ff2 |
465 | CODE: |
aae85ceb |
466 | { |
85982a32 |
467 | encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); |
ab3374e4 |
468 | if (SvUTF8(src)) { |
469 | sv_utf8_downgrade(src, FALSE); |
470 | } |
220e2d4e |
471 | ST(0) = encode_method(aTHX_ enc, enc->t_utf8, src, check, |
472 | NULL, Nullsv, NULL); |
85982a32 |
473 | SvUTF8_on(ST(0)); |
aae85ceb |
474 | XSRETURN(1); |
475 | } |
476 | |
477 | void |
85982a32 |
478 | Method_encode(obj,src,check = 0) |
aae85ceb |
479 | SV * obj |
85982a32 |
480 | SV * src |
481 | int check |
aae85ceb |
482 | CODE: |
483 | { |
85982a32 |
484 | encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); |
485 | sv_utf8_upgrade(src); |
220e2d4e |
486 | ST(0) = encode_method(aTHX_ enc, enc->f_utf8, src, check, |
487 | NULL, Nullsv, NULL); |
aae85ceb |
488 | XSRETURN(1); |
489 | } |
490 | |
0ab8f81e |
491 | void |
492 | Method_needs_lines(obj) |
493 | SV * obj |
494 | CODE: |
495 | { |
b32afa7c |
496 | /* encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); */ |
0ab8f81e |
497 | ST(0) = &PL_sv_no; |
498 | XSRETURN(1); |
499 | } |
500 | |
501 | void |
502 | Method_perlio_ok(obj) |
503 | SV * obj |
504 | CODE: |
505 | { |
b32afa7c |
506 | /* encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); */ |
10c5ecbb |
507 | /* require_pv(PERLIO_FILENAME); */ |
508 | |
509 | eval_pv("require PerlIO::encoding", 0); |
510 | |
511 | if (SvTRUE(get_sv("@", 0))) { |
0ab8f81e |
512 | ST(0) = &PL_sv_no; |
10c5ecbb |
513 | }else{ |
514 | ST(0) = &PL_sv_yes; |
0ab8f81e |
515 | } |
516 | XSRETURN(1); |
517 | } |
518 | |
67e989fb |
519 | MODULE = Encode PACKAGE = Encode |
2c674647 |
520 | |
521 | PROTOTYPES: ENABLE |
522 | |
67e989fb |
523 | I32 |
2c674647 |
524 | _bytes_to_utf8(sv, ...) |
85982a32 |
525 | SV * sv |
526 | CODE: |
527 | { |
528 | SV * encoding = items == 2 ? ST(1) : Nullsv; |
c6a7db43 |
529 | |
85982a32 |
530 | if (encoding) |
531 | RETVAL = _encoded_bytes_to_utf8(sv, SvPV_nolen(encoding)); |
532 | else { |
533 | STRLEN len; |
534 | U8* s = (U8*)SvPV(sv, len); |
535 | U8* converted; |
536 | |
537 | converted = bytes_to_utf8(s, &len); /* This allocs */ |
538 | sv_setpvn(sv, (char *)converted, len); |
539 | SvUTF8_on(sv); /* XXX Should we? */ |
540 | Safefree(converted); /* ... so free it */ |
541 | RETVAL = len; |
542 | } |
543 | } |
544 | OUTPUT: |
545 | RETVAL |
2c674647 |
546 | |
67e989fb |
547 | I32 |
2c674647 |
548 | _utf8_to_bytes(sv, ...) |
85982a32 |
549 | SV * sv |
550 | CODE: |
551 | { |
552 | SV * to = items > 1 ? ST(1) : Nullsv; |
553 | SV * check = items > 2 ? ST(2) : Nullsv; |
554 | |
555 | if (to) { |
556 | RETVAL = _encoded_utf8_to_bytes(sv, SvPV_nolen(to)); |
557 | } else { |
558 | STRLEN len; |
559 | U8 *s = (U8*)SvPV(sv, len); |
560 | |
561 | RETVAL = 0; |
562 | if (SvTRUE(check)) { |
563 | /* Must do things the slow way */ |
564 | U8 *dest; |
565 | /* We need a copy to pass to check() */ |
c6a7db43 |
566 | U8 *src = (U8*)savepv((char *)s); |
85982a32 |
567 | U8 *send = s + len; |
568 | |
569 | New(83, dest, len, U8); /* I think */ |
570 | |
571 | while (s < send) { |
572 | if (*s < 0x80){ |
573 | *dest++ = *s++; |
574 | } else { |
575 | STRLEN ulen; |
576 | UV uv = *s++; |
577 | |
578 | /* Have to do it all ourselves because of error routine, |
579 | aargh. */ |
580 | if (!(uv & 0x40)){ goto failure; } |
581 | if (!(uv & 0x20)) { ulen = 2; uv &= 0x1f; } |
582 | else if (!(uv & 0x10)) { ulen = 3; uv &= 0x0f; } |
583 | else if (!(uv & 0x08)) { ulen = 4; uv &= 0x07; } |
584 | else if (!(uv & 0x04)) { ulen = 5; uv &= 0x03; } |
585 | else if (!(uv & 0x02)) { ulen = 6; uv &= 0x01; } |
586 | else if (!(uv & 0x01)) { ulen = 7; uv = 0; } |
587 | else { ulen = 13; uv = 0; } |
87714904 |
588 | |
85982a32 |
589 | /* Note change to utf8.c variable naming, for variety */ |
590 | while (ulen--) { |
c6a7db43 |
591 | if ((*s & 0xc0) != 0x80){ |
592 | goto failure; |
85982a32 |
593 | } else { |
594 | uv = (uv << 6) | (*s++ & 0x3f); |
595 | } |
87714904 |
596 | } |
67e989fb |
597 | if (uv > 256) { |
598 | failure: |
85982a32 |
599 | call_failure(check, s, dest, src); |
600 | /* Now what happens? */ |
67e989fb |
601 | } |
602 | *dest++ = (U8)uv; |
85982a32 |
603 | } |
604 | } |
605 | } else { |
606 | RETVAL = (utf8_to_bytes(s, &len) ? len : 0); |
2c674647 |
607 | } |
85982a32 |
608 | } |
609 | } |
610 | OUTPUT: |
611 | RETVAL |
2c674647 |
612 | |
2c674647 |
613 | bool |
b2704119 |
614 | is_utf8(sv, check = 0) |
4411f3b6 |
615 | SV * sv |
b2704119 |
616 | int check |
85982a32 |
617 | CODE: |
618 | { |
619 | if (SvGMAGICAL(sv)) /* it could be $1, for example */ |
620 | sv = newSVsv(sv); /* GMAGIG will be done */ |
621 | if (SvPOK(sv)) { |
622 | RETVAL = SvUTF8(sv) ? TRUE : FALSE; |
623 | if (RETVAL && |
624 | check && |
625 | !is_utf8_string((U8*)SvPVX(sv), SvCUR(sv))) |
2c674647 |
626 | RETVAL = FALSE; |
85982a32 |
627 | } else { |
628 | RETVAL = FALSE; |
629 | } |
630 | if (sv != ST(0)) |
631 | SvREFCNT_dec(sv); /* it was a temp copy */ |
632 | } |
633 | OUTPUT: |
634 | RETVAL |
2c674647 |
635 | |
636 | SV * |
4411f3b6 |
637 | _utf8_on(sv) |
85982a32 |
638 | SV * sv |
639 | CODE: |
640 | { |
641 | if (SvPOK(sv)) { |
642 | SV *rsv = newSViv(SvUTF8(sv)); |
643 | RETVAL = rsv; |
644 | SvUTF8_on(sv); |
645 | } else { |
646 | RETVAL = &PL_sv_undef; |
647 | } |
648 | } |
649 | OUTPUT: |
650 | RETVAL |
2c674647 |
651 | |
652 | SV * |
4411f3b6 |
653 | _utf8_off(sv) |
85982a32 |
654 | SV * sv |
655 | CODE: |
656 | { |
657 | if (SvPOK(sv)) { |
658 | SV *rsv = newSViv(SvUTF8(sv)); |
659 | RETVAL = rsv; |
660 | SvUTF8_off(sv); |
661 | } else { |
662 | RETVAL = &PL_sv_undef; |
663 | } |
664 | } |
665 | OUTPUT: |
666 | RETVAL |
667 | |
85982a32 |
668 | int |
669 | DIE_ON_ERR() |
670 | CODE: |
671 | RETVAL = ENCODE_DIE_ON_ERR; |
672 | OUTPUT: |
673 | RETVAL |
674 | |
c6a7db43 |
675 | int |
85982a32 |
676 | WARN_ON_ERR() |
677 | CODE: |
678 | RETVAL = ENCODE_WARN_ON_ERR; |
679 | OUTPUT: |
680 | RETVAL |
681 | |
682 | int |
683 | LEAVE_SRC() |
684 | CODE: |
685 | RETVAL = ENCODE_LEAVE_SRC; |
686 | OUTPUT: |
687 | RETVAL |
688 | |
689 | int |
690 | RETURN_ON_ERR() |
691 | CODE: |
692 | RETVAL = ENCODE_RETURN_ON_ERR; |
693 | OUTPUT: |
694 | RETVAL |
695 | |
696 | int |
697 | PERLQQ() |
698 | CODE: |
699 | RETVAL = ENCODE_PERLQQ; |
700 | OUTPUT: |
701 | RETVAL |
702 | |
703 | int |
af1f55d9 |
704 | HTMLCREF() |
705 | CODE: |
706 | RETVAL = ENCODE_HTMLCREF; |
707 | OUTPUT: |
708 | RETVAL |
709 | |
710 | int |
711 | XMLCREF() |
712 | CODE: |
713 | RETVAL = ENCODE_XMLCREF; |
714 | OUTPUT: |
715 | RETVAL |
716 | |
717 | int |
85982a32 |
718 | FB_DEFAULT() |
719 | CODE: |
720 | RETVAL = ENCODE_FB_DEFAULT; |
721 | OUTPUT: |
722 | RETVAL |
723 | |
724 | int |
725 | FB_CROAK() |
726 | CODE: |
727 | RETVAL = ENCODE_FB_CROAK; |
728 | OUTPUT: |
729 | RETVAL |
730 | |
731 | int |
732 | FB_QUIET() |
733 | CODE: |
734 | RETVAL = ENCODE_FB_QUIET; |
735 | OUTPUT: |
736 | RETVAL |
737 | |
738 | int |
739 | FB_WARN() |
740 | CODE: |
741 | RETVAL = ENCODE_FB_WARN; |
742 | OUTPUT: |
743 | RETVAL |
744 | |
745 | int |
746 | FB_PERLQQ() |
747 | CODE: |
748 | RETVAL = ENCODE_FB_PERLQQ; |
749 | OUTPUT: |
750 | RETVAL |
2c674647 |
751 | |
af1f55d9 |
752 | int |
753 | FB_HTMLCREF() |
754 | CODE: |
755 | RETVAL = ENCODE_FB_HTMLCREF; |
756 | OUTPUT: |
757 | RETVAL |
758 | |
759 | int |
760 | FB_XMLCREF() |
761 | CODE: |
762 | RETVAL = ENCODE_FB_XMLCREF; |
763 | OUTPUT: |
764 | RETVAL |
765 | |
33af2bc7 |
766 | BOOT: |
767 | { |
85982a32 |
768 | #include "def_t.h" |
e7cbefb8 |
769 | #include "def_t.exh" |
33af2bc7 |
770 | } |