Commit | Line | Data |
85982a32 |
1 | /* |
7828f908 |
2 | $Id: Encode.xs,v 2.13 2007/05/29 07:35:27 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"); \ |
d1256cb1 |
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 | |
7f0d54d7 |
32 | #define UTF8_ALLOW_STRICT 0 |
33 | #define UTF8_ALLOW_NONSTRICT (UTF8_ALLOW_ANY & \ |
34 | ~(UTF8_ALLOW_CONTINUATION | \ |
35 | UTF8_ALLOW_NON_CONTINUATION | \ |
36 | UTF8_ALLOW_LONG)) |
37 | |
8e180e82 |
38 | static SV* fallback_cb = (SV*)NULL ; |
39 | |
b0b300a3 |
40 | void |
aa0053b7 |
41 | Encode_XSEncoding(pTHX_ encode_t * enc) |
2f2b4ff2 |
42 | { |
aa0053b7 |
43 | dSP; |
44 | HV *stash = gv_stashpv("Encode::XS", TRUE); |
45 | SV *sv = sv_bless(newRV_noinc(newSViv(PTR2IV(enc))), stash); |
46 | int i = 0; |
47 | PUSHMARK(sp); |
48 | XPUSHs(sv); |
49 | while (enc->name[i]) { |
d1256cb1 |
50 | const char *name = enc->name[i++]; |
51 | XPUSHs(sv_2mortal(newSVpvn(name, strlen(name)))); |
aa0053b7 |
52 | } |
53 | PUTBACK; |
54 | call_pv("Encode::define_encoding", G_DISCARD); |
55 | SvREFCNT_dec(sv); |
2f2b4ff2 |
56 | } |
57 | |
aa0053b7 |
58 | void |
59 | call_failure(SV * routine, U8 * done, U8 * dest, U8 * orig) |
60 | { |
85982a32 |
61 | /* Exists for breakpointing */ |
aa0053b7 |
62 | } |
67e989fb |
63 | |
85982a32 |
64 | |
2fc614e0 |
65 | #define ERR_ENCODE_NOMAP "\"\\x{%04" UVxf "}\" does not map to %s" |
66 | #define ERR_DECODE_NOMAP "%s \"\\x%02" UVXf "\" does not map to Unicode" |
67 | |
2f2b4ff2 |
68 | static SV * |
8e180e82 |
69 | do_fallback_cb(pTHX_ UV ch) |
70 | { |
71 | dSP; |
72 | int argc; |
7828f908 |
73 | SV *temp, *retval; |
8e180e82 |
74 | ENTER; |
75 | SAVETMPS; |
76 | PUSHMARK(sp); |
77 | XPUSHs(sv_2mortal(newSVnv((UV)ch))); |
78 | PUTBACK; |
79 | argc = call_sv(fallback_cb, G_SCALAR); |
80 | SPAGAIN; |
81 | if (argc != 1){ |
d1256cb1 |
82 | croak("fallback sub must return scalar!"); |
8e180e82 |
83 | } |
7828f908 |
84 | temp = newSVsv(POPs); |
8e180e82 |
85 | PUTBACK; |
86 | FREETMPS; |
87 | LEAVE; |
7828f908 |
88 | retval = newSVpv("",0); |
89 | sv_catsv(retval, temp); |
6ead8143 |
90 | SvREFCNT_dec(temp); |
8e180e82 |
91 | return retval; |
92 | } |
93 | |
94 | static SV * |
0629a5b3 |
95 | encode_method(pTHX_ const encode_t * enc, const encpage_t * dir, SV * src, |
d1256cb1 |
96 | int check, STRLEN * offset, SV * term, int * retcode) |
2f2b4ff2 |
97 | { |
aa0053b7 |
98 | STRLEN slen; |
99 | U8 *s = (U8 *) SvPV(src, slen); |
3aececda |
100 | STRLEN tlen = slen; |
101 | STRLEN ddone = 0; |
102 | STRLEN sdone = 0; |
39cf9a5e |
103 | |
3c49ab08 |
104 | /* We allocate slen+1. |
85982a32 |
105 | PerlIO dumps core if this value is smaller than this. */ |
3c49ab08 |
106 | SV *dst = sv_2mortal(newSV(slen+1)); |
85982a32 |
107 | U8 *d = (U8 *)SvPVX(dst); |
108 | STRLEN dlen = SvLEN(dst)-1; |
220e2d4e |
109 | int code = 0; |
110 | STRLEN trmlen = 0; |
cc7dbc11 |
111 | U8 *trm = term ? (U8*) SvPV(term, trmlen) : NULL; |
220e2d4e |
112 | |
113 | if (offset) { |
114 | s += *offset; |
6be7c101 |
115 | if (slen > *offset){ /* safeguard against slen overflow */ |
d1256cb1 |
116 | slen -= *offset; |
6be7c101 |
117 | }else{ |
d1256cb1 |
118 | slen = 0; |
6be7c101 |
119 | } |
220e2d4e |
120 | tlen = slen; |
121 | } |
85982a32 |
122 | |
6be7c101 |
123 | if (slen == 0){ |
d1256cb1 |
124 | SvCUR_set(dst, 0); |
125 | SvPOK_only(dst); |
126 | goto ENCODE_END; |
85982a32 |
127 | } |
128 | |
220e2d4e |
129 | while( (code = do_encode(dir, s, &slen, d, dlen, &dlen, !check, |
d1256cb1 |
130 | trm, trmlen)) ) |
85982a32 |
131 | { |
d1256cb1 |
132 | SvCUR_set(dst, dlen+ddone); |
133 | SvPOK_only(dst); |
134 | |
135 | if (code == ENCODE_FALLBACK || code == ENCODE_PARTIAL || |
136 | code == ENCODE_FOUND_TERM) { |
137 | break; |
138 | } |
139 | switch (code) { |
140 | case ENCODE_NOSPACE: |
141 | { |
142 | STRLEN more = 0; /* make sure you initialize! */ |
143 | STRLEN sleft; |
144 | sdone += slen; |
145 | ddone += dlen; |
146 | sleft = tlen - sdone; |
fcb875d4 |
147 | #if ENCODE_XS_PROFILE >= 2 |
d1256cb1 |
148 | Perl_warn(aTHX_ |
149 | "more=%d, sdone=%d, sleft=%d, SvLEN(dst)=%d\n", |
150 | more, sdone, sleft, SvLEN(dst)); |
fcb875d4 |
151 | #endif |
d1256cb1 |
152 | if (sdone != 0) { /* has src ever been processed ? */ |
39cf9a5e |
153 | #if ENCODE_XS_USEFP == 2 |
d1256cb1 |
154 | more = (1.0*tlen*SvLEN(dst)+sdone-1)/sdone |
155 | - SvLEN(dst); |
39cf9a5e |
156 | #elif ENCODE_XS_USEFP |
d1256cb1 |
157 | more = (STRLEN)((1.0*SvLEN(dst)+1)/sdone * sleft); |
39cf9a5e |
158 | #else |
d1256cb1 |
159 | /* safe until SvLEN(dst) == MAX_INT/16 */ |
160 | more = (16*SvLEN(dst)+1)/sdone/16 * sleft; |
39cf9a5e |
161 | #endif |
d1256cb1 |
162 | } |
163 | more += UTF8_MAXLEN; /* insurance policy */ |
164 | d = (U8 *) SvGROW(dst, SvLEN(dst) + more); |
165 | /* dst need to grow need MORE bytes! */ |
166 | if (ddone >= SvLEN(dst)) { |
167 | Perl_croak(aTHX_ "Destination couldn't be grown."); |
168 | } |
169 | dlen = SvLEN(dst)-ddone-1; |
170 | d += ddone; |
171 | s += slen; |
172 | slen = tlen-sdone; |
173 | continue; |
174 | } |
175 | case ENCODE_NOREP: |
176 | /* encoding */ |
177 | if (dir == enc->f_utf8) { |
178 | STRLEN clen; |
179 | UV ch = |
180 | utf8n_to_uvuni(s+slen, (SvCUR(src)-slen), |
181 | &clen, UTF8_ALLOW_ANY|UTF8_CHECK_ONLY); |
182 | /* if non-representable multibyte prefix at end of current buffer - break*/ |
183 | if (clen > tlen - sdone) break; |
184 | if (check & ENCODE_DIE_ON_ERR) { |
185 | Perl_croak(aTHX_ ERR_ENCODE_NOMAP, |
186 | (UV)ch, enc->name[0]); |
187 | return &PL_sv_undef; /* never reaches but be safe */ |
188 | } |
189 | if (check & ENCODE_WARN_ON_ERR){ |
190 | Perl_warner(aTHX_ packWARN(WARN_UTF8), |
191 | ERR_ENCODE_NOMAP, (UV)ch, enc->name[0]); |
192 | } |
193 | if (check & ENCODE_RETURN_ON_ERR){ |
194 | goto ENCODE_SET_SRC; |
195 | } |
196 | if (check & (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){ |
197 | SV* subchar = |
198 | (fallback_cb != (SV*)NULL) ? do_fallback_cb(aTHX_ ch) : |
199 | newSVpvf(check & ENCODE_PERLQQ ? "\\x{%04"UVxf"}" : |
200 | check & ENCODE_HTMLCREF ? "&#%" UVuf ";" : |
201 | "&#x%" UVxf ";", (UV)ch); |
202 | sdone += slen + clen; |
203 | ddone += dlen + SvCUR(subchar); |
204 | sv_catsv(dst, subchar); |
205 | SvREFCNT_dec(subchar); |
206 | } else { |
207 | /* fallback char */ |
208 | sdone += slen + clen; |
209 | ddone += dlen + enc->replen; |
210 | sv_catpvn(dst, (char*)enc->rep, enc->replen); |
211 | } |
212 | } |
213 | /* decoding */ |
214 | else { |
215 | if (check & ENCODE_DIE_ON_ERR){ |
216 | Perl_croak(aTHX_ ERR_DECODE_NOMAP, |
436c6dd3 |
217 | enc->name[0], (UV)s[slen]); |
d1256cb1 |
218 | return &PL_sv_undef; /* never reaches but be safe */ |
219 | } |
220 | if (check & ENCODE_WARN_ON_ERR){ |
221 | Perl_warner( |
222 | aTHX_ packWARN(WARN_UTF8), |
223 | ERR_DECODE_NOMAP, |
436c6dd3 |
224 | enc->name[0], (UV)s[slen]); |
d1256cb1 |
225 | } |
226 | if (check & ENCODE_RETURN_ON_ERR){ |
227 | goto ENCODE_SET_SRC; |
228 | } |
229 | if (check & |
230 | (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){ |
231 | SV* subchar = |
232 | (fallback_cb != (SV*)NULL) ? |
233 | do_fallback_cb(aTHX_ (UV)s[slen]) : |
234 | newSVpvf("\\x%02" UVXf, (UV)s[slen]); |
235 | sdone += slen + 1; |
236 | ddone += dlen + SvCUR(subchar); |
237 | sv_catsv(dst, subchar); |
238 | SvREFCNT_dec(subchar); |
239 | } else { |
240 | sdone += slen + 1; |
241 | ddone += dlen + strlen(FBCHAR_UTF8); |
242 | sv_catpv(dst, FBCHAR_UTF8); |
243 | } |
244 | } |
245 | /* settle variables when fallback */ |
246 | d = (U8 *)SvEND(dst); |
b0b300a3 |
247 | dlen = SvLEN(dst) - ddone - 1; |
d1256cb1 |
248 | s = (U8*)SvPVX(src) + sdone; |
249 | slen = tlen - sdone; |
250 | break; |
251 | |
252 | default: |
253 | Perl_croak(aTHX_ "Unexpected code %d converting %s %s", |
254 | code, (dir == enc->f_utf8) ? "to" : "from", |
255 | enc->name[0]); |
256 | return &PL_sv_undef; |
257 | } |
85982a32 |
258 | } |
259 | ENCODE_SET_SRC: |
ca777f1c |
260 | if (check && !(check & ENCODE_LEAVE_SRC)){ |
d1256cb1 |
261 | sdone = SvCUR(src) - (slen+sdone); |
262 | if (sdone) { |
263 | sv_setpvn(src, (char*)s+slen, sdone); |
264 | } |
265 | SvCUR_set(src, sdone); |
2f2b4ff2 |
266 | } |
85982a32 |
267 | /* warn("check = 0x%X, code = 0x%d\n", check, code); */ |
c6a7db43 |
268 | |
85982a32 |
269 | SvCUR_set(dst, dlen+ddone); |
270 | SvPOK_only(dst); |
c6a7db43 |
271 | |
39cf9a5e |
272 | #if ENCODE_XS_PROFILE |
273 | if (SvCUR(dst) > SvCUR(src)){ |
d1256cb1 |
274 | Perl_warn(aTHX_ |
275 | "SvLEN(dst)=%d, SvCUR(dst)=%d. %d bytes unused(%f %%)\n", |
276 | SvLEN(dst), SvCUR(dst), SvLEN(dst) - SvCUR(dst), |
277 | (SvLEN(dst) - SvCUR(dst))*1.0/SvLEN(dst)*100.0); |
39cf9a5e |
278 | } |
3c49ab08 |
279 | #endif |
c6a7db43 |
280 | |
220e2d4e |
281 | if (offset) |
282 | *offset += sdone + slen; |
283 | |
85982a32 |
284 | ENCODE_END: |
0b3236bb |
285 | *SvEND(dst) = '\0'; |
220e2d4e |
286 | if (retcode) *retcode = code; |
aa0053b7 |
287 | return dst; |
2f2b4ff2 |
288 | } |
289 | |
7f0d54d7 |
290 | static bool |
291 | strict_utf8(pTHX_ SV* sv) |
292 | { |
293 | HV* hv; |
294 | SV** svp; |
295 | sv = SvRV(sv); |
296 | if (!sv || SvTYPE(sv) != SVt_PVHV) |
297 | return 0; |
298 | hv = (HV*)sv; |
299 | svp = hv_fetch(hv, "strict_utf8", 11, 0); |
300 | if (!svp) |
301 | return 0; |
302 | return SvTRUE(*svp); |
303 | } |
304 | |
305 | static U8* |
306 | process_utf8(pTHX_ SV* dst, U8* s, U8* e, int check, |
307 | bool encode, bool strict, bool stop_at_partial) |
308 | { |
309 | UV uv; |
310 | STRLEN ulen; |
311 | |
312 | SvPOK_only(dst); |
313 | SvCUR_set(dst,0); |
314 | |
315 | while (s < e) { |
316 | if (UTF8_IS_INVARIANT(*s)) { |
317 | sv_catpvn(dst, (char *)s, 1); |
318 | s++; |
319 | continue; |
320 | } |
321 | |
322 | if (UTF8_IS_START(*s)) { |
323 | U8 skip = UTF8SKIP(s); |
324 | if ((s + skip) > e) { |
325 | /* Partial character */ |
326 | /* XXX could check that rest of bytes are UTF8_IS_CONTINUATION(ch) */ |
56ff7374 |
327 | if (stop_at_partial || (check & ENCODE_STOP_AT_PARTIAL)) |
7f0d54d7 |
328 | break; |
329 | |
330 | goto malformed_byte; |
331 | } |
332 | |
333 | uv = utf8n_to_uvuni(s, e - s, &ulen, |
334 | UTF8_CHECK_ONLY | (strict ? UTF8_ALLOW_STRICT : |
335 | UTF8_ALLOW_NONSTRICT) |
336 | ); |
337 | #if 1 /* perl-5.8.6 and older do not check UTF8_ALLOW_LONG */ |
d1256cb1 |
338 | if (strict && uv > PERL_UNICODE_MAX) |
1b9a72a9 |
339 | ulen = (STRLEN) -1; |
7f0d54d7 |
340 | #endif |
341 | if (ulen == -1) { |
342 | if (strict) { |
343 | uv = utf8n_to_uvuni(s, e - s, &ulen, |
344 | UTF8_CHECK_ONLY | UTF8_ALLOW_NONSTRICT); |
345 | if (ulen == -1) |
346 | goto malformed_byte; |
347 | goto malformed; |
348 | } |
349 | goto malformed_byte; |
350 | } |
351 | |
352 | |
353 | /* Whole char is good */ |
354 | sv_catpvn(dst,(char *)s,skip); |
355 | s += skip; |
356 | continue; |
357 | } |
358 | |
359 | /* If we get here there is something wrong with alleged UTF-8 */ |
360 | malformed_byte: |
361 | uv = (UV)*s; |
362 | ulen = 1; |
363 | |
364 | malformed: |
365 | if (check & ENCODE_DIE_ON_ERR){ |
366 | if (encode) |
367 | Perl_croak(aTHX_ ERR_ENCODE_NOMAP, uv, "utf8"); |
368 | else |
369 | Perl_croak(aTHX_ ERR_DECODE_NOMAP, "utf8", uv); |
370 | } |
371 | if (check & ENCODE_WARN_ON_ERR){ |
372 | if (encode) |
373 | Perl_warner(aTHX_ packWARN(WARN_UTF8), |
374 | ERR_ENCODE_NOMAP, uv, "utf8"); |
375 | else |
376 | Perl_warner(aTHX_ packWARN(WARN_UTF8), |
377 | ERR_DECODE_NOMAP, "utf8", uv); |
378 | } |
379 | if (check & ENCODE_RETURN_ON_ERR) { |
380 | break; |
381 | } |
382 | if (check & (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){ |
383 | SV* subchar = newSVpvf(check & ENCODE_PERLQQ ? (ulen == 1 ? "\\x%02" UVXf : "\\x{%04" UVXf "}"): |
384 | check & ENCODE_HTMLCREF ? "&#%" UVuf ";" : |
385 | "&#x%" UVxf ";", uv); |
386 | sv_catsv(dst, subchar); |
387 | SvREFCNT_dec(subchar); |
388 | } else { |
389 | sv_catpv(dst, FBCHAR_UTF8); |
390 | } |
391 | s += ulen; |
392 | } |
393 | *SvEND(dst) = '\0'; |
394 | |
395 | return s; |
396 | } |
397 | |
398 | |
ab3374e4 |
399 | MODULE = Encode PACKAGE = Encode::utf8 PREFIX = Method_ |
400 | |
a0d8a30e |
401 | PROTOTYPES: DISABLE |
402 | |
403 | void |
b536bf57 |
404 | Method_decode_xs(obj,src,check = 0) |
ab3374e4 |
405 | SV * obj |
406 | SV * src |
407 | int check |
408 | CODE: |
409 | { |
410 | STRLEN slen; |
411 | U8 *s = (U8 *) SvPV(src, slen); |
412 | U8 *e = (U8 *) SvEND(src); |
b536bf57 |
413 | SV *dst = newSV(slen>0?slen:1); /* newSV() abhors 0 -- inaba */ |
cc836e95 |
414 | |
415 | /* |
7f0d54d7 |
416 | * PerlIO check -- we assume the object is of PerlIO if renewed |
cc836e95 |
417 | */ |
f5a22bee |
418 | bool renewed = 0; |
cc836e95 |
419 | dSP; ENTER; SAVETMPS; |
420 | PUSHMARK(sp); |
421 | XPUSHs(obj); |
422 | PUTBACK; |
423 | if (call_method("renewed",G_SCALAR) == 1) { |
d1256cb1 |
424 | SPAGAIN; |
425 | renewed = (bool)POPi; |
426 | PUTBACK; |
cc836e95 |
427 | #if 0 |
d1256cb1 |
428 | fprintf(stderr, "renewed == %d\n", renewed); |
cc836e95 |
429 | #endif |
cc836e95 |
430 | } |
431 | FREETMPS; LEAVE; |
432 | /* end PerlIO check */ |
433 | |
ab3374e4 |
434 | if (SvUTF8(src)) { |
d1256cb1 |
435 | s = utf8_to_bytes(s,&slen); |
436 | if (s) { |
437 | SvCUR_set(src,slen); |
438 | SvUTF8_off(src); |
439 | e = s+slen; |
440 | } |
441 | else { |
442 | croak("Cannot decode string with wide characters"); |
443 | } |
ab3374e4 |
444 | } |
7f0d54d7 |
445 | |
446 | s = process_utf8(aTHX_ dst, s, e, check, 0, strict_utf8(aTHX_ obj), renewed); |
ab3374e4 |
447 | |
448 | /* Clear out translated part of source unless asked not to */ |
449 | if (check && !(check & ENCODE_LEAVE_SRC)){ |
d1256cb1 |
450 | slen = e-s; |
451 | if (slen) { |
452 | sv_setpvn(src, (char*)s, slen); |
453 | } |
454 | SvCUR_set(src, slen); |
ab3374e4 |
455 | } |
456 | SvUTF8_on(dst); |
457 | ST(0) = sv_2mortal(dst); |
458 | XSRETURN(1); |
459 | } |
460 | |
461 | void |
b536bf57 |
462 | Method_encode_xs(obj,src,check = 0) |
ab3374e4 |
463 | SV * obj |
464 | SV * src |
465 | int check |
466 | CODE: |
467 | { |
468 | STRLEN slen; |
469 | U8 *s = (U8 *) SvPV(src, slen); |
470 | U8 *e = (U8 *) SvEND(src); |
b536bf57 |
471 | SV *dst = newSV(slen>0?slen:1); /* newSV() abhors 0 -- inaba */ |
ab3374e4 |
472 | if (SvUTF8(src)) { |
d1256cb1 |
473 | /* Already encoded */ |
474 | if (strict_utf8(aTHX_ obj)) { |
475 | s = process_utf8(aTHX_ dst, s, e, check, 1, 1, 0); |
476 | } |
7f0d54d7 |
477 | else { |
478 | /* trust it and just copy the octets */ |
479 | sv_setpvn(dst,(char *)s,(e-s)); |
d1256cb1 |
480 | s = e; |
7f0d54d7 |
481 | } |
ab3374e4 |
482 | } |
483 | else { |
484 | /* Native bytes - can always encode */ |
d1256cb1 |
485 | U8 *d = (U8 *) SvGROW(dst, 2*slen+1); /* +1 or assertion will botch */ |
ab3374e4 |
486 | while (s < e) { |
cf0d1c66 |
487 | UV uv = NATIVE_TO_UNI((UV) *s); |
488 | s++; /* Above expansion of NATIVE_TO_UNI() is safer this way. */ |
ab3374e4 |
489 | if (UNI_IS_INVARIANT(uv)) |
490 | *d++ = (U8)UTF_TO_NATIVE(uv); |
491 | else { |
492 | *d++ = (U8)UTF8_EIGHT_BIT_HI(uv); |
493 | *d++ = (U8)UTF8_EIGHT_BIT_LO(uv); |
494 | } |
d1256cb1 |
495 | } |
ab3374e4 |
496 | SvCUR_set(dst, d- (U8 *)SvPVX(dst)); |
497 | *SvEND(dst) = '\0'; |
498 | } |
499 | |
500 | /* Clear out translated part of source unless asked not to */ |
501 | if (check && !(check & ENCODE_LEAVE_SRC)){ |
d1256cb1 |
502 | slen = e-s; |
503 | if (slen) { |
504 | sv_setpvn(src, (char*)s, slen); |
505 | } |
506 | SvCUR_set(src, slen); |
ab3374e4 |
507 | } |
508 | SvPOK_only(dst); |
509 | SvUTF8_off(dst); |
510 | ST(0) = sv_2mortal(dst); |
511 | XSRETURN(1); |
512 | } |
513 | |
50d26985 |
514 | MODULE = Encode PACKAGE = Encode::XS PREFIX = Method_ |
2f2b4ff2 |
515 | |
516 | PROTOTYPES: ENABLE |
517 | |
518 | void |
a0d8a30e |
519 | Method_renew(obj) |
520 | SV * obj |
521 | CODE: |
522 | { |
523 | XSRETURN(1); |
524 | } |
525 | |
cc836e95 |
526 | int |
527 | Method_renewed(obj) |
528 | SV * obj |
529 | CODE: |
530 | RETVAL = 0; |
531 | OUTPUT: |
532 | RETVAL |
533 | |
a0d8a30e |
534 | void |
0a95303c |
535 | Method_name(obj) |
536 | SV * obj |
537 | CODE: |
85982a32 |
538 | { |
539 | encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); |
540 | ST(0) = sv_2mortal(newSVpvn(enc->name[0],strlen(enc->name[0]))); |
541 | XSRETURN(1); |
542 | } |
0a95303c |
543 | |
544 | void |
220e2d4e |
545 | Method_cat_decode(obj, dst, src, off, term, check = 0) |
546 | SV * obj |
547 | SV * dst |
548 | SV * src |
549 | SV * off |
550 | SV * term |
551 | int check |
552 | CODE: |
553 | { |
554 | encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); |
555 | STRLEN offset = (STRLEN)SvIV(off); |
556 | int code = 0; |
557 | if (SvUTF8(src)) { |
558 | sv_utf8_downgrade(src, FALSE); |
559 | } |
560 | sv_catsv(dst, encode_method(aTHX_ enc, enc->t_utf8, src, check, |
d1256cb1 |
561 | &offset, term, &code)); |
b162af07 |
562 | SvIV_set(off, (IV)offset); |
220e2d4e |
563 | if (code == ENCODE_FOUND_TERM) { |
d1256cb1 |
564 | ST(0) = &PL_sv_yes; |
220e2d4e |
565 | }else{ |
d1256cb1 |
566 | ST(0) = &PL_sv_no; |
220e2d4e |
567 | } |
568 | XSRETURN(1); |
569 | } |
570 | |
571 | void |
8e180e82 |
572 | Method_decode(obj,src,check_sv = &PL_sv_no) |
2f2b4ff2 |
573 | SV * obj |
574 | SV * src |
8e180e82 |
575 | SV * check_sv |
2f2b4ff2 |
576 | CODE: |
aae85ceb |
577 | { |
8e180e82 |
578 | int check; |
85982a32 |
579 | encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); |
ab3374e4 |
580 | if (SvUTF8(src)) { |
581 | sv_utf8_downgrade(src, FALSE); |
582 | } |
8e180e82 |
583 | if (SvROK(check_sv)){ |
d1256cb1 |
584 | if (fallback_cb == (SV*)NULL){ |
8e180e82 |
585 | fallback_cb = newSVsv(check_sv); /* First time */ |
586 | }else{ |
587 | SvSetSV(fallback_cb, check_sv); /* Been here before */ |
d1256cb1 |
588 | } |
589 | check = ENCODE_PERLQQ|ENCODE_LEAVE_SRC; /* same as FB_PERLQQ */ |
8e180e82 |
590 | }else{ |
d1256cb1 |
591 | fallback_cb = (SV*)NULL; |
592 | check = SvIV(check_sv); |
8e180e82 |
593 | } |
220e2d4e |
594 | ST(0) = encode_method(aTHX_ enc, enc->t_utf8, src, check, |
d1256cb1 |
595 | NULL, Nullsv, NULL); |
85982a32 |
596 | SvUTF8_on(ST(0)); |
aae85ceb |
597 | XSRETURN(1); |
598 | } |
599 | |
8e180e82 |
600 | |
601 | |
aae85ceb |
602 | void |
8e180e82 |
603 | Method_encode(obj,src,check_sv = &PL_sv_no) |
aae85ceb |
604 | SV * obj |
85982a32 |
605 | SV * src |
8e180e82 |
606 | SV * check_sv |
aae85ceb |
607 | CODE: |
608 | { |
8e180e82 |
609 | int check; |
85982a32 |
610 | encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); |
611 | sv_utf8_upgrade(src); |
8e180e82 |
612 | if (SvROK(check_sv)){ |
d1256cb1 |
613 | if (fallback_cb == (SV*)NULL){ |
8e180e82 |
614 | fallback_cb = newSVsv(check_sv); /* First time */ |
615 | }else{ |
616 | SvSetSV(fallback_cb, check_sv); /* Been here before */ |
d1256cb1 |
617 | } |
618 | check = ENCODE_PERLQQ|ENCODE_LEAVE_SRC; /* same as FB_PERLQQ */ |
8e180e82 |
619 | }else{ |
d1256cb1 |
620 | fallback_cb = (SV*)NULL; |
621 | check = SvIV(check_sv); |
8e180e82 |
622 | } |
220e2d4e |
623 | ST(0) = encode_method(aTHX_ enc, enc->f_utf8, src, check, |
d1256cb1 |
624 | NULL, Nullsv, NULL); |
aae85ceb |
625 | XSRETURN(1); |
626 | } |
627 | |
0ab8f81e |
628 | void |
629 | Method_needs_lines(obj) |
630 | SV * obj |
631 | CODE: |
632 | { |
b32afa7c |
633 | /* encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); */ |
0ab8f81e |
634 | ST(0) = &PL_sv_no; |
635 | XSRETURN(1); |
636 | } |
637 | |
638 | void |
639 | Method_perlio_ok(obj) |
640 | SV * obj |
641 | CODE: |
642 | { |
b32afa7c |
643 | /* encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); */ |
10c5ecbb |
644 | /* require_pv(PERLIO_FILENAME); */ |
645 | |
646 | eval_pv("require PerlIO::encoding", 0); |
647 | |
648 | if (SvTRUE(get_sv("@", 0))) { |
d1256cb1 |
649 | ST(0) = &PL_sv_no; |
10c5ecbb |
650 | }else{ |
d1256cb1 |
651 | ST(0) = &PL_sv_yes; |
0ab8f81e |
652 | } |
653 | XSRETURN(1); |
654 | } |
655 | |
742555bd |
656 | void |
657 | Method_mime_name(obj) |
658 | SV * obj |
659 | CODE: |
660 | { |
661 | encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); |
662 | SV *retval; |
663 | eval_pv("require Encode::MIME::Name", 0); |
664 | |
665 | if (SvTRUE(get_sv("@", 0))) { |
666 | ST(0) = &PL_sv_undef; |
667 | }else{ |
668 | ENTER; |
669 | SAVETMPS; |
670 | PUSHMARK(sp); |
671 | XPUSHs(sv_2mortal(newSVpvn(enc->name[0], strlen(enc->name[0])))); |
672 | PUTBACK; |
673 | call_pv("Encode::MIME::Name::get_mime_name", G_SCALAR); |
674 | SPAGAIN; |
675 | retval = newSVsv(POPs); |
676 | PUTBACK; |
677 | FREETMPS; |
678 | LEAVE; |
679 | /* enc->name[0] */ |
680 | ST(0) = retval; |
681 | } |
682 | XSRETURN(1); |
683 | } |
684 | |
67e989fb |
685 | MODULE = Encode PACKAGE = Encode |
2c674647 |
686 | |
687 | PROTOTYPES: ENABLE |
688 | |
67e989fb |
689 | I32 |
2c674647 |
690 | _bytes_to_utf8(sv, ...) |
85982a32 |
691 | SV * sv |
692 | CODE: |
693 | { |
694 | SV * encoding = items == 2 ? ST(1) : Nullsv; |
c6a7db43 |
695 | |
85982a32 |
696 | if (encoding) |
697 | RETVAL = _encoded_bytes_to_utf8(sv, SvPV_nolen(encoding)); |
698 | else { |
d1256cb1 |
699 | STRLEN len; |
700 | U8* s = (U8*)SvPV(sv, len); |
701 | U8* converted; |
702 | |
703 | converted = bytes_to_utf8(s, &len); /* This allocs */ |
704 | sv_setpvn(sv, (char *)converted, len); |
705 | SvUTF8_on(sv); /* XXX Should we? */ |
706 | Safefree(converted); /* ... so free it */ |
707 | RETVAL = len; |
85982a32 |
708 | } |
709 | } |
710 | OUTPUT: |
711 | RETVAL |
2c674647 |
712 | |
67e989fb |
713 | I32 |
2c674647 |
714 | _utf8_to_bytes(sv, ...) |
85982a32 |
715 | SV * sv |
716 | CODE: |
717 | { |
718 | SV * to = items > 1 ? ST(1) : Nullsv; |
719 | SV * check = items > 2 ? ST(2) : Nullsv; |
720 | |
721 | if (to) { |
d1256cb1 |
722 | RETVAL = _encoded_utf8_to_bytes(sv, SvPV_nolen(to)); |
85982a32 |
723 | } else { |
d1256cb1 |
724 | STRLEN len; |
725 | U8 *s = (U8*)SvPV(sv, len); |
85982a32 |
726 | |
d1256cb1 |
727 | RETVAL = 0; |
728 | if (SvTRUE(check)) { |
729 | /* Must do things the slow way */ |
730 | U8 *dest; |
85982a32 |
731 | /* We need a copy to pass to check() */ |
2cdf8b6c |
732 | U8 *src = s; |
d1256cb1 |
733 | U8 *send = s + len; |
2cdf8b6c |
734 | U8 *d0; |
85982a32 |
735 | |
d1256cb1 |
736 | New(83, dest, len, U8); /* I think */ |
2cdf8b6c |
737 | d0 = dest; |
85982a32 |
738 | |
d1256cb1 |
739 | while (s < send) { |
85982a32 |
740 | if (*s < 0x80){ |
d1256cb1 |
741 | *dest++ = *s++; |
85982a32 |
742 | } else { |
d1256cb1 |
743 | STRLEN ulen; |
744 | UV uv = *s++; |
745 | |
746 | /* Have to do it all ourselves because of error routine, |
747 | aargh. */ |
748 | if (!(uv & 0x40)){ goto failure; } |
749 | if (!(uv & 0x20)) { ulen = 2; uv &= 0x1f; } |
750 | else if (!(uv & 0x10)) { ulen = 3; uv &= 0x0f; } |
751 | else if (!(uv & 0x08)) { ulen = 4; uv &= 0x07; } |
752 | else if (!(uv & 0x04)) { ulen = 5; uv &= 0x03; } |
753 | else if (!(uv & 0x02)) { ulen = 6; uv &= 0x01; } |
754 | else if (!(uv & 0x01)) { ulen = 7; uv = 0; } |
755 | else { ulen = 13; uv = 0; } |
756 | |
757 | /* Note change to utf8.c variable naming, for variety */ |
758 | while (ulen--) { |
759 | if ((*s & 0xc0) != 0x80){ |
760 | goto failure; |
761 | } else { |
762 | uv = (uv << 6) | (*s++ & 0x3f); |
763 | } |
764 | } |
765 | if (uv > 256) { |
766 | failure: |
767 | call_failure(check, s, dest, src); |
768 | /* Now what happens? */ |
769 | } |
770 | *dest++ = (U8)uv; |
771 | } |
772 | } |
2cdf8b6c |
773 | RETVAL = dest - d0; |
774 | sv_usepvn(sv, (char *)dest, RETVAL); |
775 | SvUTF8_off(sv); |
d1256cb1 |
776 | } else { |
777 | RETVAL = (utf8_to_bytes(s, &len) ? len : 0); |
778 | } |
85982a32 |
779 | } |
780 | } |
781 | OUTPUT: |
782 | RETVAL |
2c674647 |
783 | |
2c674647 |
784 | bool |
b2704119 |
785 | is_utf8(sv, check = 0) |
4411f3b6 |
786 | SV * sv |
b2704119 |
787 | int check |
85982a32 |
788 | CODE: |
789 | { |
790 | if (SvGMAGICAL(sv)) /* it could be $1, for example */ |
d1256cb1 |
791 | sv = newSVsv(sv); /* GMAGIG will be done */ |
d1256cb1 |
792 | RETVAL = SvUTF8(sv) ? TRUE : FALSE; |
793 | if (RETVAL && |
794 | check && |
795 | !is_utf8_string((U8*)SvPVX(sv), SvCUR(sv))) |
796 | RETVAL = FALSE; |
85982a32 |
797 | if (sv != ST(0)) |
d1256cb1 |
798 | SvREFCNT_dec(sv); /* it was a temp copy */ |
85982a32 |
799 | } |
800 | OUTPUT: |
801 | RETVAL |
2c674647 |
802 | |
803 | SV * |
4411f3b6 |
804 | _utf8_on(sv) |
85982a32 |
805 | SV * sv |
806 | CODE: |
807 | { |
808 | if (SvPOK(sv)) { |
d1256cb1 |
809 | SV *rsv = newSViv(SvUTF8(sv)); |
810 | RETVAL = rsv; |
811 | SvUTF8_on(sv); |
85982a32 |
812 | } else { |
d1256cb1 |
813 | RETVAL = &PL_sv_undef; |
85982a32 |
814 | } |
815 | } |
816 | OUTPUT: |
817 | RETVAL |
2c674647 |
818 | |
819 | SV * |
4411f3b6 |
820 | _utf8_off(sv) |
85982a32 |
821 | SV * sv |
822 | CODE: |
823 | { |
824 | if (SvPOK(sv)) { |
d1256cb1 |
825 | SV *rsv = newSViv(SvUTF8(sv)); |
826 | RETVAL = rsv; |
827 | SvUTF8_off(sv); |
85982a32 |
828 | } else { |
d1256cb1 |
829 | RETVAL = &PL_sv_undef; |
85982a32 |
830 | } |
831 | } |
832 | OUTPUT: |
833 | RETVAL |
834 | |
85982a32 |
835 | int |
836 | DIE_ON_ERR() |
837 | CODE: |
838 | RETVAL = ENCODE_DIE_ON_ERR; |
839 | OUTPUT: |
840 | RETVAL |
841 | |
c6a7db43 |
842 | int |
85982a32 |
843 | WARN_ON_ERR() |
844 | CODE: |
845 | RETVAL = ENCODE_WARN_ON_ERR; |
846 | OUTPUT: |
847 | RETVAL |
848 | |
849 | int |
850 | LEAVE_SRC() |
851 | CODE: |
852 | RETVAL = ENCODE_LEAVE_SRC; |
853 | OUTPUT: |
854 | RETVAL |
855 | |
856 | int |
857 | RETURN_ON_ERR() |
858 | CODE: |
859 | RETVAL = ENCODE_RETURN_ON_ERR; |
860 | OUTPUT: |
861 | RETVAL |
862 | |
863 | int |
864 | PERLQQ() |
865 | CODE: |
866 | RETVAL = ENCODE_PERLQQ; |
867 | OUTPUT: |
868 | RETVAL |
869 | |
870 | int |
af1f55d9 |
871 | HTMLCREF() |
872 | CODE: |
873 | RETVAL = ENCODE_HTMLCREF; |
874 | OUTPUT: |
875 | RETVAL |
876 | |
877 | int |
878 | XMLCREF() |
879 | CODE: |
880 | RETVAL = ENCODE_XMLCREF; |
881 | OUTPUT: |
882 | RETVAL |
883 | |
884 | int |
56ff7374 |
885 | STOP_AT_PARTIAL() |
886 | CODE: |
887 | RETVAL = ENCODE_STOP_AT_PARTIAL; |
888 | OUTPUT: |
889 | RETVAL |
890 | |
891 | int |
85982a32 |
892 | FB_DEFAULT() |
893 | CODE: |
894 | RETVAL = ENCODE_FB_DEFAULT; |
895 | OUTPUT: |
896 | RETVAL |
897 | |
898 | int |
899 | FB_CROAK() |
900 | CODE: |
901 | RETVAL = ENCODE_FB_CROAK; |
902 | OUTPUT: |
903 | RETVAL |
904 | |
905 | int |
906 | FB_QUIET() |
907 | CODE: |
908 | RETVAL = ENCODE_FB_QUIET; |
909 | OUTPUT: |
910 | RETVAL |
911 | |
912 | int |
913 | FB_WARN() |
914 | CODE: |
915 | RETVAL = ENCODE_FB_WARN; |
916 | OUTPUT: |
917 | RETVAL |
918 | |
919 | int |
920 | FB_PERLQQ() |
921 | CODE: |
922 | RETVAL = ENCODE_FB_PERLQQ; |
923 | OUTPUT: |
924 | RETVAL |
2c674647 |
925 | |
af1f55d9 |
926 | int |
927 | FB_HTMLCREF() |
928 | CODE: |
929 | RETVAL = ENCODE_FB_HTMLCREF; |
930 | OUTPUT: |
931 | RETVAL |
932 | |
933 | int |
934 | FB_XMLCREF() |
935 | CODE: |
936 | RETVAL = ENCODE_FB_XMLCREF; |
937 | OUTPUT: |
938 | RETVAL |
939 | |
33af2bc7 |
940 | BOOT: |
941 | { |
85982a32 |
942 | #include "def_t.h" |
e7cbefb8 |
943 | #include "def_t.exh" |
33af2bc7 |
944 | } |