Commit | Line | Data |
85982a32 |
1 | /* |
cc7dbc11 |
2 | $Id: Encode.xs,v 1.54 2003/02/20 14:42:34 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; |
81 | slen -= *offset; |
82 | tlen = slen; |
83 | } |
85982a32 |
84 | |
220e2d4e |
85 | if (slen <= 0){ |
85982a32 |
86 | SvCUR_set(dst, 0); |
87 | SvPOK_only(dst); |
88 | goto ENCODE_END; |
89 | } |
90 | |
220e2d4e |
91 | while( (code = do_encode(dir, s, &slen, d, dlen, &dlen, !check, |
92 | trm, trmlen)) ) |
85982a32 |
93 | { |
94 | SvCUR_set(dst, dlen+ddone); |
95 | SvPOK_only(dst); |
0b3236bb |
96 | |
220e2d4e |
97 | if (code == ENCODE_FALLBACK || code == ENCODE_PARTIAL || |
98 | code == ENCODE_FOUND_TERM) { |
85982a32 |
99 | break; |
100 | } |
101 | switch (code) { |
102 | case ENCODE_NOSPACE: |
103 | { |
104 | STRLEN more = 0; /* make sure you initialize! */ |
105 | STRLEN sleft; |
106 | sdone += slen; |
107 | ddone += dlen; |
108 | sleft = tlen - sdone; |
fcb875d4 |
109 | #if ENCODE_XS_PROFILE >= 2 |
85982a32 |
110 | Perl_warn(aTHX_ |
111 | "more=%d, sdone=%d, sleft=%d, SvLEN(dst)=%d\n", |
112 | more, sdone, sleft, SvLEN(dst)); |
fcb875d4 |
113 | #endif |
85982a32 |
114 | if (sdone != 0) { /* has src ever been processed ? */ |
39cf9a5e |
115 | #if ENCODE_XS_USEFP == 2 |
85982a32 |
116 | more = (1.0*tlen*SvLEN(dst)+sdone-1)/sdone |
117 | - SvLEN(dst); |
39cf9a5e |
118 | #elif ENCODE_XS_USEFP |
6e21dc91 |
119 | more = (STRLEN)((1.0*SvLEN(dst)+1)/sdone * sleft); |
39cf9a5e |
120 | #else |
85982a32 |
121 | /* safe until SvLEN(dst) == MAX_INT/16 */ |
122 | more = (16*SvLEN(dst)+1)/sdone/16 * sleft; |
39cf9a5e |
123 | #endif |
39cf9a5e |
124 | } |
85982a32 |
125 | more += UTF8_MAXLEN; /* insurance policy */ |
126 | d = (U8 *) SvGROW(dst, SvLEN(dst) + more); |
127 | /* dst need to grow need MORE bytes! */ |
128 | if (ddone >= SvLEN(dst)) { |
129 | Perl_croak(aTHX_ "Destination couldn't be grown."); |
130 | } |
131 | dlen = SvLEN(dst)-ddone-1; |
132 | d += ddone; |
133 | s += slen; |
134 | slen = tlen-sdone; |
135 | continue; |
136 | } |
137 | case ENCODE_NOREP: |
138 | /* encoding */ |
c6a7db43 |
139 | if (dir == enc->f_utf8) { |
85982a32 |
140 | STRLEN clen; |
141 | UV ch = |
3e952a88 |
142 | utf8n_to_uvuni(s+slen, (SvCUR(src)-slen), |
b0b300a3 |
143 | &clen, UTF8_ALLOW_ANY|UTF8_CHECK_ONLY); |
85982a32 |
144 | if (check & ENCODE_DIE_ON_ERR) { |
2fc614e0 |
145 | Perl_croak(aTHX_ ERR_ENCODE_NOMAP, |
146 | (UV)ch, enc->name[0]); |
4089adc4 |
147 | return &PL_sv_undef; /* never reaches but be safe */ |
148 | } |
149 | if (check & ENCODE_WARN_ON_ERR){ |
150 | Perl_warner(aTHX_ packWARN(WARN_UTF8), |
2fc614e0 |
151 | ERR_ENCODE_NOMAP, (UV)ch, enc->name[0]); |
4089adc4 |
152 | } |
153 | if (check & ENCODE_RETURN_ON_ERR){ |
154 | goto ENCODE_SET_SRC; |
155 | } |
156 | if (check & ENCODE_PERLQQ){ |
157 | SV* perlqq = |
158 | sv_2mortal(newSVpvf("\\x{%04"UVxf"}", (UV)ch)); |
159 | sdone += slen + clen; |
160 | ddone += dlen + SvCUR(perlqq); |
161 | sv_catsv(dst, perlqq); |
162 | }else if (check & ENCODE_HTMLCREF){ |
163 | SV* htmlcref = |
164 | sv_2mortal(newSVpvf("&#%" UVuf ";", (UV)ch)); |
165 | sdone += slen + clen; |
166 | ddone += dlen + SvCUR(htmlcref); |
167 | sv_catsv(dst, htmlcref); |
168 | }else if (check & ENCODE_XMLCREF){ |
169 | SV* xmlcref = |
170 | sv_2mortal(newSVpvf("&#x%" UVxf ";", (UV)ch)); |
171 | sdone += slen + clen; |
172 | ddone += dlen + SvCUR(xmlcref); |
173 | sv_catsv(dst, xmlcref); |
174 | } else { |
175 | /* fallback char */ |
176 | sdone += slen + clen; |
177 | ddone += dlen + enc->replen; |
178 | sv_catpvn(dst, (char*)enc->rep, enc->replen); |
c6a7db43 |
179 | } |
b2704119 |
180 | } |
85982a32 |
181 | /* decoding */ |
c6a7db43 |
182 | else { |
85982a32 |
183 | if (check & ENCODE_DIE_ON_ERR){ |
2fc614e0 |
184 | Perl_croak(aTHX_ ERR_DECODE_NOMAP, |
436c6dd3 |
185 | enc->name[0], (UV)s[slen]); |
4089adc4 |
186 | return &PL_sv_undef; /* never reaches but be safe */ |
187 | } |
188 | if (check & ENCODE_WARN_ON_ERR){ |
189 | Perl_warner( |
190 | aTHX_ packWARN(WARN_UTF8), |
2fc614e0 |
191 | ERR_DECODE_NOMAP, |
436c6dd3 |
192 | enc->name[0], (UV)s[slen]); |
4089adc4 |
193 | } |
194 | if (check & ENCODE_RETURN_ON_ERR){ |
195 | goto ENCODE_SET_SRC; |
196 | } |
197 | if (check & |
198 | (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){ |
199 | SV* perlqq = |
200 | sv_2mortal(newSVpvf("\\x%02" UVXf, (UV)s[slen])); |
201 | sdone += slen + 1; |
202 | ddone += dlen + SvCUR(perlqq); |
203 | sv_catsv(dst, perlqq); |
204 | } else { |
205 | sdone += slen + 1; |
206 | ddone += dlen + strlen(FBCHAR_UTF8); |
207 | sv_catpv(dst, FBCHAR_UTF8); |
aa0053b7 |
208 | } |
b2704119 |
209 | } |
85982a32 |
210 | /* settle variables when fallback */ |
b0b300a3 |
211 | d = (U8 *)SvEND(dst); |
212 | dlen = SvLEN(dst) - ddone - 1; |
3e952a88 |
213 | s = (U8*)SvPVX(src) + sdone; |
b2704119 |
214 | slen = tlen - sdone; |
215 | break; |
2f2b4ff2 |
216 | |
85982a32 |
217 | default: |
218 | Perl_croak(aTHX_ "Unexpected code %d converting %s %s", |
219 | code, (dir == enc->f_utf8) ? "to" : "from", |
220 | enc->name[0]); |
221 | return &PL_sv_undef; |
aa0053b7 |
222 | } |
85982a32 |
223 | } |
224 | ENCODE_SET_SRC: |
ca777f1c |
225 | if (check && !(check & ENCODE_LEAVE_SRC)){ |
226 | sdone = SvCUR(src) - (slen+sdone); |
85982a32 |
227 | if (sdone) { |
228 | sv_setpvn(src, (char*)s+slen, sdone); |
aa0053b7 |
229 | } |
85982a32 |
230 | SvCUR_set(src, sdone); |
2f2b4ff2 |
231 | } |
85982a32 |
232 | /* warn("check = 0x%X, code = 0x%d\n", check, code); */ |
c6a7db43 |
233 | |
85982a32 |
234 | SvCUR_set(dst, dlen+ddone); |
235 | SvPOK_only(dst); |
c6a7db43 |
236 | |
39cf9a5e |
237 | #if ENCODE_XS_PROFILE |
238 | if (SvCUR(dst) > SvCUR(src)){ |
85982a32 |
239 | Perl_warn(aTHX_ |
240 | "SvLEN(dst)=%d, SvCUR(dst)=%d. %d bytes unused(%f %%)\n", |
241 | SvLEN(dst), SvCUR(dst), SvLEN(dst) - SvCUR(dst), |
242 | (SvLEN(dst) - SvCUR(dst))*1.0/SvLEN(dst)*100.0); |
39cf9a5e |
243 | } |
3c49ab08 |
244 | #endif |
c6a7db43 |
245 | |
220e2d4e |
246 | if (offset) |
247 | *offset += sdone + slen; |
248 | |
85982a32 |
249 | ENCODE_END: |
0b3236bb |
250 | *SvEND(dst) = '\0'; |
220e2d4e |
251 | if (retcode) *retcode = code; |
aa0053b7 |
252 | return dst; |
2f2b4ff2 |
253 | } |
254 | |
ab3374e4 |
255 | MODULE = Encode PACKAGE = Encode::utf8 PREFIX = Method_ |
256 | |
257 | void |
b536bf57 |
258 | Method_decode_xs(obj,src,check = 0) |
ab3374e4 |
259 | SV * obj |
260 | SV * src |
261 | int check |
262 | CODE: |
263 | { |
264 | STRLEN slen; |
265 | U8 *s = (U8 *) SvPV(src, slen); |
266 | U8 *e = (U8 *) SvEND(src); |
b536bf57 |
267 | SV *dst = newSV(slen>0?slen:1); /* newSV() abhors 0 -- inaba */ |
ab3374e4 |
268 | SvPOK_only(dst); |
269 | SvCUR_set(dst,0); |
270 | if (SvUTF8(src)) { |
271 | s = utf8_to_bytes(s,&slen); |
272 | if (s) { |
273 | SvCUR_set(src,slen); |
274 | SvUTF8_off(src); |
275 | e = s+slen; |
276 | } |
277 | else { |
278 | croak("Cannot decode string with wide characters"); |
279 | } |
280 | } |
281 | while (s < e) { |
282 | if (UTF8_IS_INVARIANT(*s) || UTF8_IS_START(*s)) { |
283 | U8 skip = UTF8SKIP(s); |
284 | if ((s + skip) > e) { |
285 | /* Partial character - done */ |
286 | break; |
287 | } |
288 | else if (is_utf8_char(s)) { |
289 | /* Whole char is good */ |
290 | sv_catpvn(dst,(char *)s,skip); |
291 | s += skip; |
292 | continue; |
293 | } |
294 | else { |
295 | /* starts ok but isn't "good" */ |
296 | } |
297 | } |
298 | else { |
299 | /* Invalid start byte */ |
300 | } |
301 | /* If we get here there is something wrong with alleged UTF-8 */ |
302 | if (check & ENCODE_DIE_ON_ERR){ |
303 | Perl_croak(aTHX_ ERR_DECODE_NOMAP, "utf8", (UV)*s); |
304 | XSRETURN(0); |
305 | } |
306 | if (check & ENCODE_WARN_ON_ERR){ |
307 | Perl_warner(aTHX_ packWARN(WARN_UTF8), |
308 | ERR_DECODE_NOMAP, "utf8", (UV)*s); |
309 | } |
310 | if (check & ENCODE_RETURN_ON_ERR) { |
311 | break; |
312 | } |
313 | if (check & (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){ |
314 | SV* perlqq = newSVpvf("\\x%02" UVXf, (UV)*s); |
315 | sv_catsv(dst, perlqq); |
316 | SvREFCNT_dec(perlqq); |
317 | } else { |
318 | sv_catpv(dst, FBCHAR_UTF8); |
319 | } |
320 | s++; |
321 | } |
322 | *SvEND(dst) = '\0'; |
323 | |
324 | /* Clear out translated part of source unless asked not to */ |
325 | if (check && !(check & ENCODE_LEAVE_SRC)){ |
326 | slen = e-s; |
327 | if (slen) { |
328 | sv_setpvn(src, (char*)s, slen); |
329 | } |
330 | SvCUR_set(src, slen); |
331 | } |
332 | SvUTF8_on(dst); |
333 | ST(0) = sv_2mortal(dst); |
334 | XSRETURN(1); |
335 | } |
336 | |
337 | void |
b536bf57 |
338 | Method_encode_xs(obj,src,check = 0) |
ab3374e4 |
339 | SV * obj |
340 | SV * src |
341 | int check |
342 | CODE: |
343 | { |
344 | STRLEN slen; |
345 | U8 *s = (U8 *) SvPV(src, slen); |
346 | U8 *e = (U8 *) SvEND(src); |
b536bf57 |
347 | SV *dst = newSV(slen>0?slen:1); /* newSV() abhors 0 -- inaba */ |
ab3374e4 |
348 | if (SvUTF8(src)) { |
349 | /* Already encoded - trust it and just copy the octets */ |
350 | sv_setpvn(dst,(char *)s,(e-s)); |
351 | s = e; |
352 | } |
353 | else { |
354 | /* Native bytes - can always encode */ |
b536bf57 |
355 | U8 *d = (U8 *) SvGROW(dst, 2*slen+1); /* +1 or assertion will botch */ |
ab3374e4 |
356 | while (s < e) { |
357 | UV uv = NATIVE_TO_UNI((UV) *s++); |
358 | if (UNI_IS_INVARIANT(uv)) |
359 | *d++ = (U8)UTF_TO_NATIVE(uv); |
360 | else { |
361 | *d++ = (U8)UTF8_EIGHT_BIT_HI(uv); |
362 | *d++ = (U8)UTF8_EIGHT_BIT_LO(uv); |
363 | } |
364 | } |
365 | SvCUR_set(dst, d- (U8 *)SvPVX(dst)); |
366 | *SvEND(dst) = '\0'; |
367 | } |
368 | |
369 | /* Clear out translated part of source unless asked not to */ |
370 | if (check && !(check & ENCODE_LEAVE_SRC)){ |
371 | slen = e-s; |
372 | if (slen) { |
373 | sv_setpvn(src, (char*)s, slen); |
374 | } |
375 | SvCUR_set(src, slen); |
376 | } |
377 | SvPOK_only(dst); |
378 | SvUTF8_off(dst); |
379 | ST(0) = sv_2mortal(dst); |
380 | XSRETURN(1); |
381 | } |
382 | |
50d26985 |
383 | MODULE = Encode PACKAGE = Encode::XS PREFIX = Method_ |
2f2b4ff2 |
384 | |
385 | PROTOTYPES: ENABLE |
386 | |
387 | void |
0a95303c |
388 | Method_name(obj) |
389 | SV * obj |
390 | CODE: |
85982a32 |
391 | { |
392 | encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); |
393 | ST(0) = sv_2mortal(newSVpvn(enc->name[0],strlen(enc->name[0]))); |
394 | XSRETURN(1); |
395 | } |
0a95303c |
396 | |
397 | void |
220e2d4e |
398 | Method_cat_decode(obj, dst, src, off, term, check = 0) |
399 | SV * obj |
400 | SV * dst |
401 | SV * src |
402 | SV * off |
403 | SV * term |
404 | int check |
405 | CODE: |
406 | { |
407 | encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); |
408 | STRLEN offset = (STRLEN)SvIV(off); |
409 | int code = 0; |
410 | if (SvUTF8(src)) { |
411 | sv_utf8_downgrade(src, FALSE); |
412 | } |
413 | sv_catsv(dst, encode_method(aTHX_ enc, enc->t_utf8, src, check, |
414 | &offset, term, &code)); |
415 | SvIVX(off) = (IV)offset; |
416 | if (code == ENCODE_FOUND_TERM) { |
417 | ST(0) = &PL_sv_yes; |
418 | }else{ |
419 | ST(0) = &PL_sv_no; |
420 | } |
421 | XSRETURN(1); |
422 | } |
423 | |
424 | void |
b2704119 |
425 | Method_decode(obj,src,check = 0) |
2f2b4ff2 |
426 | SV * obj |
427 | SV * src |
b2704119 |
428 | int check |
2f2b4ff2 |
429 | CODE: |
aae85ceb |
430 | { |
85982a32 |
431 | encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); |
ab3374e4 |
432 | if (SvUTF8(src)) { |
433 | sv_utf8_downgrade(src, FALSE); |
434 | } |
220e2d4e |
435 | ST(0) = encode_method(aTHX_ enc, enc->t_utf8, src, check, |
436 | NULL, Nullsv, NULL); |
85982a32 |
437 | SvUTF8_on(ST(0)); |
aae85ceb |
438 | XSRETURN(1); |
439 | } |
440 | |
441 | void |
85982a32 |
442 | Method_encode(obj,src,check = 0) |
aae85ceb |
443 | SV * obj |
85982a32 |
444 | SV * src |
445 | int check |
aae85ceb |
446 | CODE: |
447 | { |
85982a32 |
448 | encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); |
449 | sv_utf8_upgrade(src); |
220e2d4e |
450 | ST(0) = encode_method(aTHX_ enc, enc->f_utf8, src, check, |
451 | NULL, Nullsv, NULL); |
aae85ceb |
452 | XSRETURN(1); |
453 | } |
454 | |
0ab8f81e |
455 | void |
456 | Method_needs_lines(obj) |
457 | SV * obj |
458 | CODE: |
459 | { |
b32afa7c |
460 | /* encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); */ |
0ab8f81e |
461 | ST(0) = &PL_sv_no; |
462 | XSRETURN(1); |
463 | } |
464 | |
465 | void |
466 | Method_perlio_ok(obj) |
467 | SV * obj |
468 | CODE: |
469 | { |
b32afa7c |
470 | /* encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); */ |
10c5ecbb |
471 | /* require_pv(PERLIO_FILENAME); */ |
472 | |
473 | eval_pv("require PerlIO::encoding", 0); |
474 | |
475 | if (SvTRUE(get_sv("@", 0))) { |
0ab8f81e |
476 | ST(0) = &PL_sv_no; |
10c5ecbb |
477 | }else{ |
478 | ST(0) = &PL_sv_yes; |
0ab8f81e |
479 | } |
480 | XSRETURN(1); |
481 | } |
482 | |
67e989fb |
483 | MODULE = Encode PACKAGE = Encode |
2c674647 |
484 | |
485 | PROTOTYPES: ENABLE |
486 | |
67e989fb |
487 | I32 |
2c674647 |
488 | _bytes_to_utf8(sv, ...) |
85982a32 |
489 | SV * sv |
490 | CODE: |
491 | { |
492 | SV * encoding = items == 2 ? ST(1) : Nullsv; |
c6a7db43 |
493 | |
85982a32 |
494 | if (encoding) |
495 | RETVAL = _encoded_bytes_to_utf8(sv, SvPV_nolen(encoding)); |
496 | else { |
497 | STRLEN len; |
498 | U8* s = (U8*)SvPV(sv, len); |
499 | U8* converted; |
500 | |
501 | converted = bytes_to_utf8(s, &len); /* This allocs */ |
502 | sv_setpvn(sv, (char *)converted, len); |
503 | SvUTF8_on(sv); /* XXX Should we? */ |
504 | Safefree(converted); /* ... so free it */ |
505 | RETVAL = len; |
506 | } |
507 | } |
508 | OUTPUT: |
509 | RETVAL |
2c674647 |
510 | |
67e989fb |
511 | I32 |
2c674647 |
512 | _utf8_to_bytes(sv, ...) |
85982a32 |
513 | SV * sv |
514 | CODE: |
515 | { |
516 | SV * to = items > 1 ? ST(1) : Nullsv; |
517 | SV * check = items > 2 ? ST(2) : Nullsv; |
518 | |
519 | if (to) { |
520 | RETVAL = _encoded_utf8_to_bytes(sv, SvPV_nolen(to)); |
521 | } else { |
522 | STRLEN len; |
523 | U8 *s = (U8*)SvPV(sv, len); |
524 | |
525 | RETVAL = 0; |
526 | if (SvTRUE(check)) { |
527 | /* Must do things the slow way */ |
528 | U8 *dest; |
529 | /* We need a copy to pass to check() */ |
c6a7db43 |
530 | U8 *src = (U8*)savepv((char *)s); |
85982a32 |
531 | U8 *send = s + len; |
532 | |
533 | New(83, dest, len, U8); /* I think */ |
534 | |
535 | while (s < send) { |
536 | if (*s < 0x80){ |
537 | *dest++ = *s++; |
538 | } else { |
539 | STRLEN ulen; |
540 | UV uv = *s++; |
541 | |
542 | /* Have to do it all ourselves because of error routine, |
543 | aargh. */ |
544 | if (!(uv & 0x40)){ goto failure; } |
545 | if (!(uv & 0x20)) { ulen = 2; uv &= 0x1f; } |
546 | else if (!(uv & 0x10)) { ulen = 3; uv &= 0x0f; } |
547 | else if (!(uv & 0x08)) { ulen = 4; uv &= 0x07; } |
548 | else if (!(uv & 0x04)) { ulen = 5; uv &= 0x03; } |
549 | else if (!(uv & 0x02)) { ulen = 6; uv &= 0x01; } |
550 | else if (!(uv & 0x01)) { ulen = 7; uv = 0; } |
551 | else { ulen = 13; uv = 0; } |
87714904 |
552 | |
85982a32 |
553 | /* Note change to utf8.c variable naming, for variety */ |
554 | while (ulen--) { |
c6a7db43 |
555 | if ((*s & 0xc0) != 0x80){ |
556 | goto failure; |
85982a32 |
557 | } else { |
558 | uv = (uv << 6) | (*s++ & 0x3f); |
559 | } |
87714904 |
560 | } |
67e989fb |
561 | if (uv > 256) { |
562 | failure: |
85982a32 |
563 | call_failure(check, s, dest, src); |
564 | /* Now what happens? */ |
67e989fb |
565 | } |
566 | *dest++ = (U8)uv; |
85982a32 |
567 | } |
568 | } |
569 | } else { |
570 | RETVAL = (utf8_to_bytes(s, &len) ? len : 0); |
2c674647 |
571 | } |
85982a32 |
572 | } |
573 | } |
574 | OUTPUT: |
575 | RETVAL |
2c674647 |
576 | |
2c674647 |
577 | bool |
b2704119 |
578 | is_utf8(sv, check = 0) |
4411f3b6 |
579 | SV * sv |
b2704119 |
580 | int check |
85982a32 |
581 | CODE: |
582 | { |
583 | if (SvGMAGICAL(sv)) /* it could be $1, for example */ |
584 | sv = newSVsv(sv); /* GMAGIG will be done */ |
585 | if (SvPOK(sv)) { |
586 | RETVAL = SvUTF8(sv) ? TRUE : FALSE; |
587 | if (RETVAL && |
588 | check && |
589 | !is_utf8_string((U8*)SvPVX(sv), SvCUR(sv))) |
2c674647 |
590 | RETVAL = FALSE; |
85982a32 |
591 | } else { |
592 | RETVAL = FALSE; |
593 | } |
594 | if (sv != ST(0)) |
595 | SvREFCNT_dec(sv); /* it was a temp copy */ |
596 | } |
597 | OUTPUT: |
598 | RETVAL |
2c674647 |
599 | |
600 | SV * |
4411f3b6 |
601 | _utf8_on(sv) |
85982a32 |
602 | SV * sv |
603 | CODE: |
604 | { |
605 | if (SvPOK(sv)) { |
606 | SV *rsv = newSViv(SvUTF8(sv)); |
607 | RETVAL = rsv; |
608 | SvUTF8_on(sv); |
609 | } else { |
610 | RETVAL = &PL_sv_undef; |
611 | } |
612 | } |
613 | OUTPUT: |
614 | RETVAL |
2c674647 |
615 | |
616 | SV * |
4411f3b6 |
617 | _utf8_off(sv) |
85982a32 |
618 | SV * sv |
619 | CODE: |
620 | { |
621 | if (SvPOK(sv)) { |
622 | SV *rsv = newSViv(SvUTF8(sv)); |
623 | RETVAL = rsv; |
624 | SvUTF8_off(sv); |
625 | } else { |
626 | RETVAL = &PL_sv_undef; |
627 | } |
628 | } |
629 | OUTPUT: |
630 | RETVAL |
631 | |
85982a32 |
632 | int |
633 | DIE_ON_ERR() |
634 | CODE: |
635 | RETVAL = ENCODE_DIE_ON_ERR; |
636 | OUTPUT: |
637 | RETVAL |
638 | |
c6a7db43 |
639 | int |
85982a32 |
640 | WARN_ON_ERR() |
641 | CODE: |
642 | RETVAL = ENCODE_WARN_ON_ERR; |
643 | OUTPUT: |
644 | RETVAL |
645 | |
646 | int |
647 | LEAVE_SRC() |
648 | CODE: |
649 | RETVAL = ENCODE_LEAVE_SRC; |
650 | OUTPUT: |
651 | RETVAL |
652 | |
653 | int |
654 | RETURN_ON_ERR() |
655 | CODE: |
656 | RETVAL = ENCODE_RETURN_ON_ERR; |
657 | OUTPUT: |
658 | RETVAL |
659 | |
660 | int |
661 | PERLQQ() |
662 | CODE: |
663 | RETVAL = ENCODE_PERLQQ; |
664 | OUTPUT: |
665 | RETVAL |
666 | |
667 | int |
af1f55d9 |
668 | HTMLCREF() |
669 | CODE: |
670 | RETVAL = ENCODE_HTMLCREF; |
671 | OUTPUT: |
672 | RETVAL |
673 | |
674 | int |
675 | XMLCREF() |
676 | CODE: |
677 | RETVAL = ENCODE_XMLCREF; |
678 | OUTPUT: |
679 | RETVAL |
680 | |
681 | int |
85982a32 |
682 | FB_DEFAULT() |
683 | CODE: |
684 | RETVAL = ENCODE_FB_DEFAULT; |
685 | OUTPUT: |
686 | RETVAL |
687 | |
688 | int |
689 | FB_CROAK() |
690 | CODE: |
691 | RETVAL = ENCODE_FB_CROAK; |
692 | OUTPUT: |
693 | RETVAL |
694 | |
695 | int |
696 | FB_QUIET() |
697 | CODE: |
698 | RETVAL = ENCODE_FB_QUIET; |
699 | OUTPUT: |
700 | RETVAL |
701 | |
702 | int |
703 | FB_WARN() |
704 | CODE: |
705 | RETVAL = ENCODE_FB_WARN; |
706 | OUTPUT: |
707 | RETVAL |
708 | |
709 | int |
710 | FB_PERLQQ() |
711 | CODE: |
712 | RETVAL = ENCODE_FB_PERLQQ; |
713 | OUTPUT: |
714 | RETVAL |
2c674647 |
715 | |
af1f55d9 |
716 | int |
717 | FB_HTMLCREF() |
718 | CODE: |
719 | RETVAL = ENCODE_FB_HTMLCREF; |
720 | OUTPUT: |
721 | RETVAL |
722 | |
723 | int |
724 | FB_XMLCREF() |
725 | CODE: |
726 | RETVAL = ENCODE_FB_XMLCREF; |
727 | OUTPUT: |
728 | RETVAL |
729 | |
33af2bc7 |
730 | BOOT: |
731 | { |
85982a32 |
732 | #include "def_t.h" |
e7cbefb8 |
733 | #include "def_t.exh" |
33af2bc7 |
734 | } |