Commit | Line | Data |
---|---|---|
85982a32 | 1 | /* |
ab3374e4 | 2 | $Id: Encode.xs,v 1.49 2002/10/21 19:47:47 dankogai Exp $ |
85982a32 JH |
3 | */ |
4 | ||
fcf2db38 | 5 | #define PERL_NO_GET_CONTEXT |
2c674647 JH |
6 | #include "EXTERN.h" |
7 | #include "perl.h" | |
8 | #include "XSUB.h" | |
2f2b4ff2 NIS |
9 | #define U8 U8 |
10 | #include "encode.h" | |
10c5ecbb JH |
11 | |
12 | # define PERLIO_MODNAME "PerlIO::encoding" | |
0ab8f81e | 13 | # define PERLIO_FILENAME "PerlIO/encoding.pm" |
aae85ceb | 14 | |
85982a32 JH |
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 JH |
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 JH |
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 NIS |
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 NIS |
48 | } |
49 | ||
aa0053b7 NIS |
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 JH |
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, |
85982a32 | 62 | int check) |
2f2b4ff2 | 63 | { |
aa0053b7 NIS |
64 | STRLEN slen; |
65 | U8 *s = (U8 *) SvPV(src, slen); | |
3aececda NIS |
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 JH |
73 | U8 *d = (U8 *)SvPVX(dst); |
74 | STRLEN dlen = SvLEN(dst)-1; | |
75 | int code; | |
76 | ||
77 | if (!slen){ | |
78 | SvCUR_set(dst, 0); | |
79 | SvPOK_only(dst); | |
80 | goto ENCODE_END; | |
81 | } | |
82 | ||
436c6dd3 | 83 | while( (code = do_encode(dir, s, &slen, d, dlen, &dlen, !check)) ) |
85982a32 JH |
84 | { |
85 | SvCUR_set(dst, dlen+ddone); | |
86 | SvPOK_only(dst); | |
0b3236bb | 87 | |
85982a32 JH |
88 | if (code == ENCODE_FALLBACK || code == ENCODE_PARTIAL){ |
89 | break; | |
90 | } | |
91 | switch (code) { | |
92 | case ENCODE_NOSPACE: | |
93 | { | |
94 | STRLEN more = 0; /* make sure you initialize! */ | |
95 | STRLEN sleft; | |
96 | sdone += slen; | |
97 | ddone += dlen; | |
98 | sleft = tlen - sdone; | |
fcb875d4 | 99 | #if ENCODE_XS_PROFILE >= 2 |
85982a32 JH |
100 | Perl_warn(aTHX_ |
101 | "more=%d, sdone=%d, sleft=%d, SvLEN(dst)=%d\n", | |
102 | more, sdone, sleft, SvLEN(dst)); | |
fcb875d4 | 103 | #endif |
85982a32 | 104 | if (sdone != 0) { /* has src ever been processed ? */ |
39cf9a5e | 105 | #if ENCODE_XS_USEFP == 2 |
85982a32 JH |
106 | more = (1.0*tlen*SvLEN(dst)+sdone-1)/sdone |
107 | - SvLEN(dst); | |
39cf9a5e | 108 | #elif ENCODE_XS_USEFP |
6e21dc91 | 109 | more = (STRLEN)((1.0*SvLEN(dst)+1)/sdone * sleft); |
39cf9a5e | 110 | #else |
85982a32 JH |
111 | /* safe until SvLEN(dst) == MAX_INT/16 */ |
112 | more = (16*SvLEN(dst)+1)/sdone/16 * sleft; | |
39cf9a5e | 113 | #endif |
39cf9a5e | 114 | } |
85982a32 JH |
115 | more += UTF8_MAXLEN; /* insurance policy */ |
116 | d = (U8 *) SvGROW(dst, SvLEN(dst) + more); | |
117 | /* dst need to grow need MORE bytes! */ | |
118 | if (ddone >= SvLEN(dst)) { | |
119 | Perl_croak(aTHX_ "Destination couldn't be grown."); | |
120 | } | |
121 | dlen = SvLEN(dst)-ddone-1; | |
122 | d += ddone; | |
123 | s += slen; | |
124 | slen = tlen-sdone; | |
125 | continue; | |
126 | } | |
127 | case ENCODE_NOREP: | |
128 | /* encoding */ | |
c6a7db43 | 129 | if (dir == enc->f_utf8) { |
85982a32 JH |
130 | STRLEN clen; |
131 | UV ch = | |
3e952a88 | 132 | utf8n_to_uvuni(s+slen, (SvCUR(src)-slen), |
b0b300a3 | 133 | &clen, UTF8_ALLOW_ANY|UTF8_CHECK_ONLY); |
85982a32 | 134 | if (check & ENCODE_DIE_ON_ERR) { |
2fc614e0 JH |
135 | Perl_croak(aTHX_ ERR_ENCODE_NOMAP, |
136 | (UV)ch, enc->name[0]); | |
4089adc4 JH |
137 | return &PL_sv_undef; /* never reaches but be safe */ |
138 | } | |
139 | if (check & ENCODE_WARN_ON_ERR){ | |
140 | Perl_warner(aTHX_ packWARN(WARN_UTF8), | |
2fc614e0 | 141 | ERR_ENCODE_NOMAP, (UV)ch, enc->name[0]); |
4089adc4 JH |
142 | } |
143 | if (check & ENCODE_RETURN_ON_ERR){ | |
144 | goto ENCODE_SET_SRC; | |
145 | } | |
146 | if (check & ENCODE_PERLQQ){ | |
147 | SV* perlqq = | |
148 | sv_2mortal(newSVpvf("\\x{%04"UVxf"}", (UV)ch)); | |
149 | sdone += slen + clen; | |
150 | ddone += dlen + SvCUR(perlqq); | |
151 | sv_catsv(dst, perlqq); | |
152 | }else if (check & ENCODE_HTMLCREF){ | |
153 | SV* htmlcref = | |
154 | sv_2mortal(newSVpvf("&#%" UVuf ";", (UV)ch)); | |
155 | sdone += slen + clen; | |
156 | ddone += dlen + SvCUR(htmlcref); | |
157 | sv_catsv(dst, htmlcref); | |
158 | }else if (check & ENCODE_XMLCREF){ | |
159 | SV* xmlcref = | |
160 | sv_2mortal(newSVpvf("&#x%" UVxf ";", (UV)ch)); | |
161 | sdone += slen + clen; | |
162 | ddone += dlen + SvCUR(xmlcref); | |
163 | sv_catsv(dst, xmlcref); | |
164 | } else { | |
165 | /* fallback char */ | |
166 | sdone += slen + clen; | |
167 | ddone += dlen + enc->replen; | |
168 | sv_catpvn(dst, (char*)enc->rep, enc->replen); | |
c6a7db43 | 169 | } |
b2704119 | 170 | } |
85982a32 | 171 | /* decoding */ |
c6a7db43 | 172 | else { |
85982a32 | 173 | if (check & ENCODE_DIE_ON_ERR){ |
2fc614e0 | 174 | Perl_croak(aTHX_ ERR_DECODE_NOMAP, |
436c6dd3 | 175 | enc->name[0], (UV)s[slen]); |
4089adc4 JH |
176 | return &PL_sv_undef; /* never reaches but be safe */ |
177 | } | |
178 | if (check & ENCODE_WARN_ON_ERR){ | |
179 | Perl_warner( | |
180 | aTHX_ packWARN(WARN_UTF8), | |
2fc614e0 | 181 | ERR_DECODE_NOMAP, |
436c6dd3 | 182 | enc->name[0], (UV)s[slen]); |
4089adc4 JH |
183 | } |
184 | if (check & ENCODE_RETURN_ON_ERR){ | |
185 | goto ENCODE_SET_SRC; | |
186 | } | |
187 | if (check & | |
188 | (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){ | |
189 | SV* perlqq = | |
190 | sv_2mortal(newSVpvf("\\x%02" UVXf, (UV)s[slen])); | |
191 | sdone += slen + 1; | |
192 | ddone += dlen + SvCUR(perlqq); | |
193 | sv_catsv(dst, perlqq); | |
194 | } else { | |
195 | sdone += slen + 1; | |
196 | ddone += dlen + strlen(FBCHAR_UTF8); | |
197 | sv_catpv(dst, FBCHAR_UTF8); | |
aa0053b7 | 198 | } |
b2704119 | 199 | } |
85982a32 | 200 | /* settle variables when fallback */ |
b0b300a3 JH |
201 | d = (U8 *)SvEND(dst); |
202 | dlen = SvLEN(dst) - ddone - 1; | |
3e952a88 | 203 | s = (U8*)SvPVX(src) + sdone; |
b2704119 JH |
204 | slen = tlen - sdone; |
205 | break; | |
2f2b4ff2 | 206 | |
85982a32 JH |
207 | default: |
208 | Perl_croak(aTHX_ "Unexpected code %d converting %s %s", | |
209 | code, (dir == enc->f_utf8) ? "to" : "from", | |
210 | enc->name[0]); | |
211 | return &PL_sv_undef; | |
aa0053b7 | 212 | } |
85982a32 JH |
213 | } |
214 | ENCODE_SET_SRC: | |
ca777f1c NIS |
215 | if (check && !(check & ENCODE_LEAVE_SRC)){ |
216 | sdone = SvCUR(src) - (slen+sdone); | |
85982a32 JH |
217 | if (sdone) { |
218 | sv_setpvn(src, (char*)s+slen, sdone); | |
aa0053b7 | 219 | } |
85982a32 | 220 | SvCUR_set(src, sdone); |
2f2b4ff2 | 221 | } |
85982a32 | 222 | /* warn("check = 0x%X, code = 0x%d\n", check, code); */ |
c6a7db43 | 223 | |
85982a32 JH |
224 | SvCUR_set(dst, dlen+ddone); |
225 | SvPOK_only(dst); | |
c6a7db43 | 226 | |
39cf9a5e DK |
227 | #if ENCODE_XS_PROFILE |
228 | if (SvCUR(dst) > SvCUR(src)){ | |
85982a32 JH |
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 |
c6a7db43 | 235 | |
85982a32 | 236 | ENCODE_END: |
0b3236bb | 237 | *SvEND(dst) = '\0'; |
aa0053b7 | 238 | return dst; |
2f2b4ff2 NIS |
239 | } |
240 | ||
ab3374e4 DK |
241 | MODULE = Encode PACKAGE = Encode::utf8 PREFIX = Method_ |
242 | ||
243 | void | |
244 | Method_decode(obj,src,check = 0) | |
245 | SV * obj | |
246 | SV * src | |
247 | int check | |
248 | CODE: | |
249 | { | |
250 | STRLEN slen; | |
251 | U8 *s = (U8 *) SvPV(src, slen); | |
252 | U8 *e = (U8 *) SvEND(src); | |
253 | SV *dst = newSV(slen); | |
254 | SvPOK_only(dst); | |
255 | SvCUR_set(dst,0); | |
256 | if (SvUTF8(src)) { | |
257 | s = utf8_to_bytes(s,&slen); | |
258 | if (s) { | |
259 | SvCUR_set(src,slen); | |
260 | SvUTF8_off(src); | |
261 | e = s+slen; | |
262 | } | |
263 | else { | |
264 | croak("Cannot decode string with wide characters"); | |
265 | } | |
266 | } | |
267 | while (s < e) { | |
268 | if (UTF8_IS_INVARIANT(*s) || UTF8_IS_START(*s)) { | |
269 | U8 skip = UTF8SKIP(s); | |
270 | if ((s + skip) > e) { | |
271 | /* Partial character - done */ | |
272 | break; | |
273 | } | |
274 | else if (is_utf8_char(s)) { | |
275 | /* Whole char is good */ | |
276 | sv_catpvn(dst,(char *)s,skip); | |
277 | s += skip; | |
278 | continue; | |
279 | } | |
280 | else { | |
281 | /* starts ok but isn't "good" */ | |
282 | } | |
283 | } | |
284 | else { | |
285 | /* Invalid start byte */ | |
286 | } | |
287 | /* If we get here there is something wrong with alleged UTF-8 */ | |
288 | if (check & ENCODE_DIE_ON_ERR){ | |
289 | Perl_croak(aTHX_ ERR_DECODE_NOMAP, "utf8", (UV)*s); | |
290 | XSRETURN(0); | |
291 | } | |
292 | if (check & ENCODE_WARN_ON_ERR){ | |
293 | Perl_warner(aTHX_ packWARN(WARN_UTF8), | |
294 | ERR_DECODE_NOMAP, "utf8", (UV)*s); | |
295 | } | |
296 | if (check & ENCODE_RETURN_ON_ERR) { | |
297 | break; | |
298 | } | |
299 | if (check & (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){ | |
300 | SV* perlqq = newSVpvf("\\x%02" UVXf, (UV)*s); | |
301 | sv_catsv(dst, perlqq); | |
302 | SvREFCNT_dec(perlqq); | |
303 | } else { | |
304 | sv_catpv(dst, FBCHAR_UTF8); | |
305 | } | |
306 | s++; | |
307 | } | |
308 | *SvEND(dst) = '\0'; | |
309 | ||
310 | /* Clear out translated part of source unless asked not to */ | |
311 | if (check && !(check & ENCODE_LEAVE_SRC)){ | |
312 | slen = e-s; | |
313 | if (slen) { | |
314 | sv_setpvn(src, (char*)s, slen); | |
315 | } | |
316 | SvCUR_set(src, slen); | |
317 | } | |
318 | SvUTF8_on(dst); | |
319 | ST(0) = sv_2mortal(dst); | |
320 | XSRETURN(1); | |
321 | } | |
322 | ||
323 | void | |
324 | Method_encode(obj,src,check = 0) | |
325 | SV * obj | |
326 | SV * src | |
327 | int check | |
328 | CODE: | |
329 | { | |
330 | STRLEN slen; | |
331 | U8 *s = (U8 *) SvPV(src, slen); | |
332 | U8 *e = (U8 *) SvEND(src); | |
333 | SV *dst = newSV(slen); | |
334 | if (SvUTF8(src)) { | |
335 | /* Already encoded - trust it and just copy the octets */ | |
336 | sv_setpvn(dst,(char *)s,(e-s)); | |
337 | s = e; | |
338 | } | |
339 | else { | |
340 | /* Native bytes - can always encode */ | |
341 | U8 *d = (U8 *) SvGROW(dst,2*slen); | |
342 | while (s < e) { | |
343 | UV uv = NATIVE_TO_UNI((UV) *s++); | |
344 | if (UNI_IS_INVARIANT(uv)) | |
345 | *d++ = (U8)UTF_TO_NATIVE(uv); | |
346 | else { | |
347 | *d++ = (U8)UTF8_EIGHT_BIT_HI(uv); | |
348 | *d++ = (U8)UTF8_EIGHT_BIT_LO(uv); | |
349 | } | |
350 | } | |
351 | SvCUR_set(dst, d- (U8 *)SvPVX(dst)); | |
352 | *SvEND(dst) = '\0'; | |
353 | } | |
354 | ||
355 | /* Clear out translated part of source unless asked not to */ | |
356 | if (check && !(check & ENCODE_LEAVE_SRC)){ | |
357 | slen = e-s; | |
358 | if (slen) { | |
359 | sv_setpvn(src, (char*)s, slen); | |
360 | } | |
361 | SvCUR_set(src, slen); | |
362 | } | |
363 | SvPOK_only(dst); | |
364 | SvUTF8_off(dst); | |
365 | ST(0) = sv_2mortal(dst); | |
366 | XSRETURN(1); | |
367 | } | |
368 | ||
50d26985 | 369 | MODULE = Encode PACKAGE = Encode::XS PREFIX = Method_ |
2f2b4ff2 NIS |
370 | |
371 | PROTOTYPES: ENABLE | |
372 | ||
373 | void | |
0a95303c NIS |
374 | Method_name(obj) |
375 | SV * obj | |
376 | CODE: | |
85982a32 JH |
377 | { |
378 | encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); | |
379 | ST(0) = sv_2mortal(newSVpvn(enc->name[0],strlen(enc->name[0]))); | |
380 | XSRETURN(1); | |
381 | } | |
0a95303c NIS |
382 | |
383 | void | |
b2704119 | 384 | Method_decode(obj,src,check = 0) |
2f2b4ff2 NIS |
385 | SV * obj |
386 | SV * src | |
b2704119 | 387 | int check |
2f2b4ff2 | 388 | CODE: |
aae85ceb | 389 | { |
85982a32 | 390 | encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); |
ab3374e4 DK |
391 | if (SvUTF8(src)) { |
392 | sv_utf8_downgrade(src, FALSE); | |
393 | } | |
85982a32 JH |
394 | ST(0) = encode_method(aTHX_ enc, enc->t_utf8, src, check); |
395 | SvUTF8_on(ST(0)); | |
aae85ceb DK |
396 | XSRETURN(1); |
397 | } | |
398 | ||
399 | void | |
85982a32 | 400 | Method_encode(obj,src,check = 0) |
aae85ceb | 401 | SV * obj |
85982a32 JH |
402 | SV * src |
403 | int check | |
aae85ceb DK |
404 | CODE: |
405 | { | |
85982a32 JH |
406 | encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); |
407 | sv_utf8_upgrade(src); | |
408 | ST(0) = encode_method(aTHX_ enc, enc->f_utf8, src, check); | |
aae85ceb DK |
409 | XSRETURN(1); |
410 | } | |
411 | ||
0ab8f81e JH |
412 | void |
413 | Method_needs_lines(obj) | |
414 | SV * obj | |
415 | CODE: | |
416 | { | |
b32afa7c | 417 | /* encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); */ |
0ab8f81e JH |
418 | ST(0) = &PL_sv_no; |
419 | XSRETURN(1); | |
420 | } | |
421 | ||
422 | void | |
423 | Method_perlio_ok(obj) | |
424 | SV * obj | |
425 | CODE: | |
426 | { | |
b32afa7c | 427 | /* encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); */ |
10c5ecbb JH |
428 | /* require_pv(PERLIO_FILENAME); */ |
429 | ||
430 | eval_pv("require PerlIO::encoding", 0); | |
431 | ||
432 | if (SvTRUE(get_sv("@", 0))) { | |
0ab8f81e | 433 | ST(0) = &PL_sv_no; |
10c5ecbb JH |
434 | }else{ |
435 | ST(0) = &PL_sv_yes; | |
0ab8f81e JH |
436 | } |
437 | XSRETURN(1); | |
438 | } | |
439 | ||
67e989fb | 440 | MODULE = Encode PACKAGE = Encode |
2c674647 JH |
441 | |
442 | PROTOTYPES: ENABLE | |
443 | ||
67e989fb | 444 | I32 |
2c674647 | 445 | _bytes_to_utf8(sv, ...) |
85982a32 JH |
446 | SV * sv |
447 | CODE: | |
448 | { | |
449 | SV * encoding = items == 2 ? ST(1) : Nullsv; | |
c6a7db43 | 450 | |
85982a32 JH |
451 | if (encoding) |
452 | RETVAL = _encoded_bytes_to_utf8(sv, SvPV_nolen(encoding)); | |
453 | else { | |
454 | STRLEN len; | |
455 | U8* s = (U8*)SvPV(sv, len); | |
456 | U8* converted; | |
457 | ||
458 | converted = bytes_to_utf8(s, &len); /* This allocs */ | |
459 | sv_setpvn(sv, (char *)converted, len); | |
460 | SvUTF8_on(sv); /* XXX Should we? */ | |
461 | Safefree(converted); /* ... so free it */ | |
462 | RETVAL = len; | |
463 | } | |
464 | } | |
465 | OUTPUT: | |
466 | RETVAL | |
2c674647 | 467 | |
67e989fb | 468 | I32 |
2c674647 | 469 | _utf8_to_bytes(sv, ...) |
85982a32 JH |
470 | SV * sv |
471 | CODE: | |
472 | { | |
473 | SV * to = items > 1 ? ST(1) : Nullsv; | |
474 | SV * check = items > 2 ? ST(2) : Nullsv; | |
475 | ||
476 | if (to) { | |
477 | RETVAL = _encoded_utf8_to_bytes(sv, SvPV_nolen(to)); | |
478 | } else { | |
479 | STRLEN len; | |
480 | U8 *s = (U8*)SvPV(sv, len); | |
481 | ||
482 | RETVAL = 0; | |
483 | if (SvTRUE(check)) { | |
484 | /* Must do things the slow way */ | |
485 | U8 *dest; | |
486 | /* We need a copy to pass to check() */ | |
c6a7db43 | 487 | U8 *src = (U8*)savepv((char *)s); |
85982a32 JH |
488 | U8 *send = s + len; |
489 | ||
490 | New(83, dest, len, U8); /* I think */ | |
491 | ||
492 | while (s < send) { | |
493 | if (*s < 0x80){ | |
494 | *dest++ = *s++; | |
495 | } else { | |
496 | STRLEN ulen; | |
497 | UV uv = *s++; | |
498 | ||
499 | /* Have to do it all ourselves because of error routine, | |
500 | aargh. */ | |
501 | if (!(uv & 0x40)){ goto failure; } | |
502 | if (!(uv & 0x20)) { ulen = 2; uv &= 0x1f; } | |
503 | else if (!(uv & 0x10)) { ulen = 3; uv &= 0x0f; } | |
504 | else if (!(uv & 0x08)) { ulen = 4; uv &= 0x07; } | |
505 | else if (!(uv & 0x04)) { ulen = 5; uv &= 0x03; } | |
506 | else if (!(uv & 0x02)) { ulen = 6; uv &= 0x01; } | |
507 | else if (!(uv & 0x01)) { ulen = 7; uv = 0; } | |
508 | else { ulen = 13; uv = 0; } | |
87714904 | 509 | |
85982a32 JH |
510 | /* Note change to utf8.c variable naming, for variety */ |
511 | while (ulen--) { | |
c6a7db43 NIS |
512 | if ((*s & 0xc0) != 0x80){ |
513 | goto failure; | |
85982a32 JH |
514 | } else { |
515 | uv = (uv << 6) | (*s++ & 0x3f); | |
516 | } | |
87714904 | 517 | } |
67e989fb JH |
518 | if (uv > 256) { |
519 | failure: | |
85982a32 JH |
520 | call_failure(check, s, dest, src); |
521 | /* Now what happens? */ | |
67e989fb JH |
522 | } |
523 | *dest++ = (U8)uv; | |
85982a32 JH |
524 | } |
525 | } | |
526 | } else { | |
527 | RETVAL = (utf8_to_bytes(s, &len) ? len : 0); | |
2c674647 | 528 | } |
85982a32 JH |
529 | } |
530 | } | |
531 | OUTPUT: | |
532 | RETVAL | |
2c674647 | 533 | |
2c674647 | 534 | bool |
b2704119 | 535 | is_utf8(sv, check = 0) |
4411f3b6 | 536 | SV * sv |
b2704119 | 537 | int check |
85982a32 JH |
538 | CODE: |
539 | { | |
540 | if (SvGMAGICAL(sv)) /* it could be $1, for example */ | |
541 | sv = newSVsv(sv); /* GMAGIG will be done */ | |
542 | if (SvPOK(sv)) { | |
543 | RETVAL = SvUTF8(sv) ? TRUE : FALSE; | |
544 | if (RETVAL && | |
545 | check && | |
546 | !is_utf8_string((U8*)SvPVX(sv), SvCUR(sv))) | |
2c674647 | 547 | RETVAL = FALSE; |
85982a32 JH |
548 | } else { |
549 | RETVAL = FALSE; | |
550 | } | |
551 | if (sv != ST(0)) | |
552 | SvREFCNT_dec(sv); /* it was a temp copy */ | |
553 | } | |
554 | OUTPUT: | |
555 | RETVAL | |
2c674647 JH |
556 | |
557 | SV * | |
4411f3b6 | 558 | _utf8_on(sv) |
85982a32 JH |
559 | SV * sv |
560 | CODE: | |
561 | { | |
562 | if (SvPOK(sv)) { | |
563 | SV *rsv = newSViv(SvUTF8(sv)); | |
564 | RETVAL = rsv; | |
565 | SvUTF8_on(sv); | |
566 | } else { | |
567 | RETVAL = &PL_sv_undef; | |
568 | } | |
569 | } | |
570 | OUTPUT: | |
571 | RETVAL | |
2c674647 JH |
572 | |
573 | SV * | |
4411f3b6 | 574 | _utf8_off(sv) |
85982a32 JH |
575 | SV * sv |
576 | CODE: | |
577 | { | |
578 | if (SvPOK(sv)) { | |
579 | SV *rsv = newSViv(SvUTF8(sv)); | |
580 | RETVAL = rsv; | |
581 | SvUTF8_off(sv); | |
582 | } else { | |
583 | RETVAL = &PL_sv_undef; | |
584 | } | |
585 | } | |
586 | OUTPUT: | |
587 | RETVAL | |
588 | ||
85982a32 JH |
589 | int |
590 | DIE_ON_ERR() | |
591 | CODE: | |
592 | RETVAL = ENCODE_DIE_ON_ERR; | |
593 | OUTPUT: | |
594 | RETVAL | |
595 | ||
c6a7db43 | 596 | int |
85982a32 JH |
597 | WARN_ON_ERR() |
598 | CODE: | |
599 | RETVAL = ENCODE_WARN_ON_ERR; | |
600 | OUTPUT: | |
601 | RETVAL | |
602 | ||
603 | int | |
604 | LEAVE_SRC() | |
605 | CODE: | |
606 | RETVAL = ENCODE_LEAVE_SRC; | |
607 | OUTPUT: | |
608 | RETVAL | |
609 | ||
610 | int | |
611 | RETURN_ON_ERR() | |
612 | CODE: | |
613 | RETVAL = ENCODE_RETURN_ON_ERR; | |
614 | OUTPUT: | |
615 | RETVAL | |
616 | ||
617 | int | |
618 | PERLQQ() | |
619 | CODE: | |
620 | RETVAL = ENCODE_PERLQQ; | |
621 | OUTPUT: | |
622 | RETVAL | |
623 | ||
624 | int | |
af1f55d9 JH |
625 | HTMLCREF() |
626 | CODE: | |
627 | RETVAL = ENCODE_HTMLCREF; | |
628 | OUTPUT: | |
629 | RETVAL | |
630 | ||
631 | int | |
632 | XMLCREF() | |
633 | CODE: | |
634 | RETVAL = ENCODE_XMLCREF; | |
635 | OUTPUT: | |
636 | RETVAL | |
637 | ||
638 | int | |
85982a32 JH |
639 | FB_DEFAULT() |
640 | CODE: | |
641 | RETVAL = ENCODE_FB_DEFAULT; | |
642 | OUTPUT: | |
643 | RETVAL | |
644 | ||
645 | int | |
646 | FB_CROAK() | |
647 | CODE: | |
648 | RETVAL = ENCODE_FB_CROAK; | |
649 | OUTPUT: | |
650 | RETVAL | |
651 | ||
652 | int | |
653 | FB_QUIET() | |
654 | CODE: | |
655 | RETVAL = ENCODE_FB_QUIET; | |
656 | OUTPUT: | |
657 | RETVAL | |
658 | ||
659 | int | |
660 | FB_WARN() | |
661 | CODE: | |
662 | RETVAL = ENCODE_FB_WARN; | |
663 | OUTPUT: | |
664 | RETVAL | |
665 | ||
666 | int | |
667 | FB_PERLQQ() | |
668 | CODE: | |
669 | RETVAL = ENCODE_FB_PERLQQ; | |
670 | OUTPUT: | |
671 | RETVAL | |
2c674647 | 672 | |
af1f55d9 JH |
673 | int |
674 | FB_HTMLCREF() | |
675 | CODE: | |
676 | RETVAL = ENCODE_FB_HTMLCREF; | |
677 | OUTPUT: | |
678 | RETVAL | |
679 | ||
680 | int | |
681 | FB_XMLCREF() | |
682 | CODE: | |
683 | RETVAL = ENCODE_FB_XMLCREF; | |
684 | OUTPUT: | |
685 | RETVAL | |
686 | ||
33af2bc7 NIS |
687 | BOOT: |
688 | { | |
85982a32 | 689 | #include "def_t.h" |
e7cbefb8 | 690 | #include "def_t.exh" |
33af2bc7 | 691 | } |