Commit | Line | Data |
85982a32 |
1 | /* |
dc4a2e29 |
2 | $Id: Encode.xs,v 2.4 2005/05/16 18:46:36 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 | |
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 | |
b0b300a3 |
38 | void |
aa0053b7 |
39 | Encode_XSEncoding(pTHX_ encode_t * enc) |
2f2b4ff2 |
40 | { |
aa0053b7 |
41 | dSP; |
42 | HV *stash = gv_stashpv("Encode::XS", TRUE); |
43 | SV *sv = sv_bless(newRV_noinc(newSViv(PTR2IV(enc))), stash); |
44 | int i = 0; |
45 | PUSHMARK(sp); |
46 | XPUSHs(sv); |
47 | while (enc->name[i]) { |
48 | const char *name = enc->name[i++]; |
49 | XPUSHs(sv_2mortal(newSVpvn(name, strlen(name)))); |
50 | } |
51 | PUTBACK; |
52 | call_pv("Encode::define_encoding", G_DISCARD); |
53 | SvREFCNT_dec(sv); |
2f2b4ff2 |
54 | } |
55 | |
aa0053b7 |
56 | void |
57 | call_failure(SV * routine, U8 * done, U8 * dest, U8 * orig) |
58 | { |
85982a32 |
59 | /* Exists for breakpointing */ |
aa0053b7 |
60 | } |
67e989fb |
61 | |
85982a32 |
62 | |
2fc614e0 |
63 | #define ERR_ENCODE_NOMAP "\"\\x{%04" UVxf "}\" does not map to %s" |
64 | #define ERR_DECODE_NOMAP "%s \"\\x%02" UVXf "\" does not map to Unicode" |
65 | |
2f2b4ff2 |
66 | static SV * |
aa0053b7 |
67 | encode_method(pTHX_ encode_t * enc, encpage_t * dir, SV * src, |
220e2d4e |
68 | int check, STRLEN * offset, SV * term, int * retcode) |
2f2b4ff2 |
69 | { |
aa0053b7 |
70 | STRLEN slen; |
71 | U8 *s = (U8 *) SvPV(src, slen); |
3aececda |
72 | STRLEN tlen = slen; |
73 | STRLEN ddone = 0; |
74 | STRLEN sdone = 0; |
39cf9a5e |
75 | |
3c49ab08 |
76 | /* We allocate slen+1. |
85982a32 |
77 | PerlIO dumps core if this value is smaller than this. */ |
3c49ab08 |
78 | SV *dst = sv_2mortal(newSV(slen+1)); |
85982a32 |
79 | U8 *d = (U8 *)SvPVX(dst); |
80 | STRLEN dlen = SvLEN(dst)-1; |
220e2d4e |
81 | int code = 0; |
82 | STRLEN trmlen = 0; |
cc7dbc11 |
83 | U8 *trm = term ? (U8*) SvPV(term, trmlen) : NULL; |
220e2d4e |
84 | |
85 | if (offset) { |
86 | s += *offset; |
6be7c101 |
87 | if (slen > *offset){ /* safeguard against slen overflow */ |
88 | slen -= *offset; |
89 | }else{ |
90 | slen = 0; |
91 | } |
220e2d4e |
92 | tlen = slen; |
93 | } |
85982a32 |
94 | |
6be7c101 |
95 | if (slen == 0){ |
85982a32 |
96 | SvCUR_set(dst, 0); |
97 | SvPOK_only(dst); |
98 | goto ENCODE_END; |
99 | } |
100 | |
220e2d4e |
101 | while( (code = do_encode(dir, s, &slen, d, dlen, &dlen, !check, |
102 | trm, trmlen)) ) |
85982a32 |
103 | { |
104 | SvCUR_set(dst, dlen+ddone); |
105 | SvPOK_only(dst); |
0b3236bb |
106 | |
220e2d4e |
107 | if (code == ENCODE_FALLBACK || code == ENCODE_PARTIAL || |
108 | code == ENCODE_FOUND_TERM) { |
85982a32 |
109 | break; |
110 | } |
111 | switch (code) { |
112 | case ENCODE_NOSPACE: |
113 | { |
114 | STRLEN more = 0; /* make sure you initialize! */ |
115 | STRLEN sleft; |
116 | sdone += slen; |
117 | ddone += dlen; |
118 | sleft = tlen - sdone; |
fcb875d4 |
119 | #if ENCODE_XS_PROFILE >= 2 |
85982a32 |
120 | Perl_warn(aTHX_ |
121 | "more=%d, sdone=%d, sleft=%d, SvLEN(dst)=%d\n", |
122 | more, sdone, sleft, SvLEN(dst)); |
fcb875d4 |
123 | #endif |
85982a32 |
124 | if (sdone != 0) { /* has src ever been processed ? */ |
39cf9a5e |
125 | #if ENCODE_XS_USEFP == 2 |
85982a32 |
126 | more = (1.0*tlen*SvLEN(dst)+sdone-1)/sdone |
127 | - SvLEN(dst); |
39cf9a5e |
128 | #elif ENCODE_XS_USEFP |
6e21dc91 |
129 | more = (STRLEN)((1.0*SvLEN(dst)+1)/sdone * sleft); |
39cf9a5e |
130 | #else |
85982a32 |
131 | /* safe until SvLEN(dst) == MAX_INT/16 */ |
132 | more = (16*SvLEN(dst)+1)/sdone/16 * sleft; |
39cf9a5e |
133 | #endif |
39cf9a5e |
134 | } |
85982a32 |
135 | more += UTF8_MAXLEN; /* insurance policy */ |
136 | d = (U8 *) SvGROW(dst, SvLEN(dst) + more); |
137 | /* dst need to grow need MORE bytes! */ |
138 | if (ddone >= SvLEN(dst)) { |
139 | Perl_croak(aTHX_ "Destination couldn't be grown."); |
140 | } |
141 | dlen = SvLEN(dst)-ddone-1; |
142 | d += ddone; |
143 | s += slen; |
144 | slen = tlen-sdone; |
145 | continue; |
146 | } |
147 | case ENCODE_NOREP: |
148 | /* encoding */ |
c6a7db43 |
149 | if (dir == enc->f_utf8) { |
85982a32 |
150 | STRLEN clen; |
151 | UV ch = |
3e952a88 |
152 | utf8n_to_uvuni(s+slen, (SvCUR(src)-slen), |
b0b300a3 |
153 | &clen, UTF8_ALLOW_ANY|UTF8_CHECK_ONLY); |
85982a32 |
154 | if (check & ENCODE_DIE_ON_ERR) { |
2fc614e0 |
155 | Perl_croak(aTHX_ ERR_ENCODE_NOMAP, |
156 | (UV)ch, enc->name[0]); |
4089adc4 |
157 | return &PL_sv_undef; /* never reaches but be safe */ |
158 | } |
159 | if (check & ENCODE_WARN_ON_ERR){ |
160 | Perl_warner(aTHX_ packWARN(WARN_UTF8), |
2fc614e0 |
161 | ERR_ENCODE_NOMAP, (UV)ch, enc->name[0]); |
4089adc4 |
162 | } |
163 | if (check & ENCODE_RETURN_ON_ERR){ |
164 | goto ENCODE_SET_SRC; |
165 | } |
f9d05ba3 |
166 | if (check & (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){ |
167 | SV* subchar = |
168 | newSVpvf(check & ENCODE_PERLQQ ? "\\x{%04"UVxf"}" : |
169 | check & ENCODE_HTMLCREF ? "&#%" UVuf ";" : |
170 | "&#x%" UVxf ";", (UV)ch); |
4089adc4 |
171 | sdone += slen + clen; |
f9d05ba3 |
172 | ddone += dlen + SvCUR(subchar); |
173 | sv_catsv(dst, subchar); |
174 | SvREFCNT_dec(subchar); |
4089adc4 |
175 | } else { |
176 | /* fallback char */ |
177 | sdone += slen + clen; |
178 | ddone += dlen + enc->replen; |
179 | sv_catpvn(dst, (char*)enc->rep, enc->replen); |
c6a7db43 |
180 | } |
b2704119 |
181 | } |
85982a32 |
182 | /* decoding */ |
c6a7db43 |
183 | else { |
85982a32 |
184 | if (check & ENCODE_DIE_ON_ERR){ |
2fc614e0 |
185 | Perl_croak(aTHX_ ERR_DECODE_NOMAP, |
436c6dd3 |
186 | enc->name[0], (UV)s[slen]); |
4089adc4 |
187 | return &PL_sv_undef; /* never reaches but be safe */ |
188 | } |
189 | if (check & ENCODE_WARN_ON_ERR){ |
190 | Perl_warner( |
191 | aTHX_ packWARN(WARN_UTF8), |
2fc614e0 |
192 | ERR_DECODE_NOMAP, |
436c6dd3 |
193 | enc->name[0], (UV)s[slen]); |
4089adc4 |
194 | } |
195 | if (check & ENCODE_RETURN_ON_ERR){ |
196 | goto ENCODE_SET_SRC; |
197 | } |
198 | if (check & |
199 | (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){ |
f9d05ba3 |
200 | SV* subchar = newSVpvf("\\x%02" UVXf, (UV)s[slen]); |
4089adc4 |
201 | sdone += slen + 1; |
f9d05ba3 |
202 | ddone += dlen + SvCUR(subchar); |
203 | sv_catsv(dst, subchar); |
204 | SvREFCNT_dec(subchar); |
4089adc4 |
205 | } else { |
206 | sdone += slen + 1; |
207 | ddone += dlen + strlen(FBCHAR_UTF8); |
208 | sv_catpv(dst, FBCHAR_UTF8); |
aa0053b7 |
209 | } |
b2704119 |
210 | } |
85982a32 |
211 | /* settle variables when fallback */ |
b0b300a3 |
212 | d = (U8 *)SvEND(dst); |
213 | dlen = SvLEN(dst) - ddone - 1; |
3e952a88 |
214 | s = (U8*)SvPVX(src) + sdone; |
b2704119 |
215 | slen = tlen - sdone; |
216 | break; |
2f2b4ff2 |
217 | |
85982a32 |
218 | default: |
219 | Perl_croak(aTHX_ "Unexpected code %d converting %s %s", |
220 | code, (dir == enc->f_utf8) ? "to" : "from", |
221 | enc->name[0]); |
222 | return &PL_sv_undef; |
aa0053b7 |
223 | } |
85982a32 |
224 | } |
225 | ENCODE_SET_SRC: |
ca777f1c |
226 | if (check && !(check & ENCODE_LEAVE_SRC)){ |
227 | sdone = SvCUR(src) - (slen+sdone); |
85982a32 |
228 | if (sdone) { |
229 | sv_setpvn(src, (char*)s+slen, sdone); |
aa0053b7 |
230 | } |
85982a32 |
231 | SvCUR_set(src, sdone); |
2f2b4ff2 |
232 | } |
85982a32 |
233 | /* warn("check = 0x%X, code = 0x%d\n", check, code); */ |
c6a7db43 |
234 | |
85982a32 |
235 | SvCUR_set(dst, dlen+ddone); |
236 | SvPOK_only(dst); |
c6a7db43 |
237 | |
39cf9a5e |
238 | #if ENCODE_XS_PROFILE |
239 | if (SvCUR(dst) > SvCUR(src)){ |
85982a32 |
240 | Perl_warn(aTHX_ |
241 | "SvLEN(dst)=%d, SvCUR(dst)=%d. %d bytes unused(%f %%)\n", |
242 | SvLEN(dst), SvCUR(dst), SvLEN(dst) - SvCUR(dst), |
243 | (SvLEN(dst) - SvCUR(dst))*1.0/SvLEN(dst)*100.0); |
39cf9a5e |
244 | } |
3c49ab08 |
245 | #endif |
c6a7db43 |
246 | |
220e2d4e |
247 | if (offset) |
248 | *offset += sdone + slen; |
249 | |
85982a32 |
250 | ENCODE_END: |
0b3236bb |
251 | *SvEND(dst) = '\0'; |
220e2d4e |
252 | if (retcode) *retcode = code; |
aa0053b7 |
253 | return dst; |
2f2b4ff2 |
254 | } |
255 | |
7f0d54d7 |
256 | static bool |
257 | strict_utf8(pTHX_ SV* sv) |
258 | { |
259 | HV* hv; |
260 | SV** svp; |
261 | sv = SvRV(sv); |
262 | if (!sv || SvTYPE(sv) != SVt_PVHV) |
263 | return 0; |
264 | hv = (HV*)sv; |
265 | svp = hv_fetch(hv, "strict_utf8", 11, 0); |
266 | if (!svp) |
267 | return 0; |
268 | return SvTRUE(*svp); |
269 | } |
270 | |
271 | static U8* |
272 | process_utf8(pTHX_ SV* dst, U8* s, U8* e, int check, |
273 | bool encode, bool strict, bool stop_at_partial) |
274 | { |
275 | UV uv; |
276 | STRLEN ulen; |
277 | |
278 | SvPOK_only(dst); |
279 | SvCUR_set(dst,0); |
280 | |
281 | while (s < e) { |
282 | if (UTF8_IS_INVARIANT(*s)) { |
283 | sv_catpvn(dst, (char *)s, 1); |
284 | s++; |
285 | continue; |
286 | } |
287 | |
288 | if (UTF8_IS_START(*s)) { |
289 | U8 skip = UTF8SKIP(s); |
290 | if ((s + skip) > e) { |
291 | /* Partial character */ |
292 | /* XXX could check that rest of bytes are UTF8_IS_CONTINUATION(ch) */ |
293 | if (stop_at_partial) |
294 | break; |
295 | |
296 | goto malformed_byte; |
297 | } |
298 | |
299 | uv = utf8n_to_uvuni(s, e - s, &ulen, |
300 | UTF8_CHECK_ONLY | (strict ? UTF8_ALLOW_STRICT : |
301 | UTF8_ALLOW_NONSTRICT) |
302 | ); |
303 | #if 1 /* perl-5.8.6 and older do not check UTF8_ALLOW_LONG */ |
304 | if (strict && uv > PERL_UNICODE_MAX) |
305 | ulen = -1; |
306 | #endif |
307 | if (ulen == -1) { |
308 | if (strict) { |
309 | uv = utf8n_to_uvuni(s, e - s, &ulen, |
310 | UTF8_CHECK_ONLY | UTF8_ALLOW_NONSTRICT); |
311 | if (ulen == -1) |
312 | goto malformed_byte; |
313 | goto malformed; |
314 | } |
315 | goto malformed_byte; |
316 | } |
317 | |
318 | |
319 | /* Whole char is good */ |
320 | sv_catpvn(dst,(char *)s,skip); |
321 | s += skip; |
322 | continue; |
323 | } |
324 | |
325 | /* If we get here there is something wrong with alleged UTF-8 */ |
326 | malformed_byte: |
327 | uv = (UV)*s; |
328 | ulen = 1; |
329 | |
330 | malformed: |
331 | if (check & ENCODE_DIE_ON_ERR){ |
332 | if (encode) |
333 | Perl_croak(aTHX_ ERR_ENCODE_NOMAP, uv, "utf8"); |
334 | else |
335 | Perl_croak(aTHX_ ERR_DECODE_NOMAP, "utf8", uv); |
336 | } |
337 | if (check & ENCODE_WARN_ON_ERR){ |
338 | if (encode) |
339 | Perl_warner(aTHX_ packWARN(WARN_UTF8), |
340 | ERR_ENCODE_NOMAP, uv, "utf8"); |
341 | else |
342 | Perl_warner(aTHX_ packWARN(WARN_UTF8), |
343 | ERR_DECODE_NOMAP, "utf8", uv); |
344 | } |
345 | if (check & ENCODE_RETURN_ON_ERR) { |
346 | break; |
347 | } |
348 | if (check & (ENCODE_PERLQQ|ENCODE_HTMLCREF|ENCODE_XMLCREF)){ |
349 | SV* subchar = newSVpvf(check & ENCODE_PERLQQ ? (ulen == 1 ? "\\x%02" UVXf : "\\x{%04" UVXf "}"): |
350 | check & ENCODE_HTMLCREF ? "&#%" UVuf ";" : |
351 | "&#x%" UVxf ";", uv); |
352 | sv_catsv(dst, subchar); |
353 | SvREFCNT_dec(subchar); |
354 | } else { |
355 | sv_catpv(dst, FBCHAR_UTF8); |
356 | } |
357 | s += ulen; |
358 | } |
359 | *SvEND(dst) = '\0'; |
360 | |
361 | return s; |
362 | } |
363 | |
364 | |
ab3374e4 |
365 | MODULE = Encode PACKAGE = Encode::utf8 PREFIX = Method_ |
366 | |
a0d8a30e |
367 | PROTOTYPES: DISABLE |
368 | |
369 | void |
b536bf57 |
370 | Method_decode_xs(obj,src,check = 0) |
ab3374e4 |
371 | SV * obj |
372 | SV * src |
373 | int check |
374 | CODE: |
375 | { |
376 | STRLEN slen; |
377 | U8 *s = (U8 *) SvPV(src, slen); |
378 | U8 *e = (U8 *) SvEND(src); |
b536bf57 |
379 | SV *dst = newSV(slen>0?slen:1); /* newSV() abhors 0 -- inaba */ |
cc836e95 |
380 | |
381 | /* |
7f0d54d7 |
382 | * PerlIO check -- we assume the object is of PerlIO if renewed |
cc836e95 |
383 | */ |
384 | int renewed = 0; |
385 | dSP; ENTER; SAVETMPS; |
386 | PUSHMARK(sp); |
387 | XPUSHs(obj); |
388 | PUTBACK; |
389 | if (call_method("renewed",G_SCALAR) == 1) { |
390 | SPAGAIN; |
391 | renewed = POPi; |
392 | PUTBACK; |
393 | #if 0 |
394 | fprintf(stderr, "renewed == %d\n", renewed); |
395 | #endif |
cc836e95 |
396 | } |
397 | FREETMPS; LEAVE; |
398 | /* end PerlIO check */ |
399 | |
ab3374e4 |
400 | if (SvUTF8(src)) { |
401 | s = utf8_to_bytes(s,&slen); |
402 | if (s) { |
403 | SvCUR_set(src,slen); |
404 | SvUTF8_off(src); |
405 | e = s+slen; |
406 | } |
407 | else { |
408 | croak("Cannot decode string with wide characters"); |
409 | } |
410 | } |
7f0d54d7 |
411 | |
412 | s = process_utf8(aTHX_ dst, s, e, check, 0, strict_utf8(aTHX_ obj), renewed); |
ab3374e4 |
413 | |
414 | /* Clear out translated part of source unless asked not to */ |
415 | if (check && !(check & ENCODE_LEAVE_SRC)){ |
416 | slen = e-s; |
417 | if (slen) { |
418 | sv_setpvn(src, (char*)s, slen); |
419 | } |
420 | SvCUR_set(src, slen); |
421 | } |
422 | SvUTF8_on(dst); |
423 | ST(0) = sv_2mortal(dst); |
424 | XSRETURN(1); |
425 | } |
426 | |
427 | void |
b536bf57 |
428 | Method_encode_xs(obj,src,check = 0) |
ab3374e4 |
429 | SV * obj |
430 | SV * src |
431 | int check |
432 | CODE: |
433 | { |
434 | STRLEN slen; |
435 | U8 *s = (U8 *) SvPV(src, slen); |
436 | U8 *e = (U8 *) SvEND(src); |
b536bf57 |
437 | SV *dst = newSV(slen>0?slen:1); /* newSV() abhors 0 -- inaba */ |
ab3374e4 |
438 | if (SvUTF8(src)) { |
7f0d54d7 |
439 | /* Already encoded */ |
440 | if (strict_utf8(aTHX_ obj)) { |
441 | s = process_utf8(aTHX_ dst, s, e, check, 1, 1, 0); |
442 | } |
443 | else { |
444 | /* trust it and just copy the octets */ |
445 | sv_setpvn(dst,(char *)s,(e-s)); |
446 | s = e; |
447 | } |
ab3374e4 |
448 | } |
449 | else { |
450 | /* Native bytes - can always encode */ |
b536bf57 |
451 | U8 *d = (U8 *) SvGROW(dst, 2*slen+1); /* +1 or assertion will botch */ |
ab3374e4 |
452 | while (s < e) { |
453 | UV uv = NATIVE_TO_UNI((UV) *s++); |
454 | if (UNI_IS_INVARIANT(uv)) |
455 | *d++ = (U8)UTF_TO_NATIVE(uv); |
456 | else { |
457 | *d++ = (U8)UTF8_EIGHT_BIT_HI(uv); |
458 | *d++ = (U8)UTF8_EIGHT_BIT_LO(uv); |
459 | } |
460 | } |
461 | SvCUR_set(dst, d- (U8 *)SvPVX(dst)); |
462 | *SvEND(dst) = '\0'; |
463 | } |
464 | |
465 | /* Clear out translated part of source unless asked not to */ |
466 | if (check && !(check & ENCODE_LEAVE_SRC)){ |
467 | slen = e-s; |
468 | if (slen) { |
469 | sv_setpvn(src, (char*)s, slen); |
470 | } |
471 | SvCUR_set(src, slen); |
472 | } |
473 | SvPOK_only(dst); |
474 | SvUTF8_off(dst); |
475 | ST(0) = sv_2mortal(dst); |
476 | XSRETURN(1); |
477 | } |
478 | |
50d26985 |
479 | MODULE = Encode PACKAGE = Encode::XS PREFIX = Method_ |
2f2b4ff2 |
480 | |
481 | PROTOTYPES: ENABLE |
482 | |
483 | void |
a0d8a30e |
484 | Method_renew(obj) |
485 | SV * obj |
486 | CODE: |
487 | { |
488 | XSRETURN(1); |
489 | } |
490 | |
cc836e95 |
491 | int |
492 | Method_renewed(obj) |
493 | SV * obj |
494 | CODE: |
495 | RETVAL = 0; |
496 | OUTPUT: |
497 | RETVAL |
498 | |
a0d8a30e |
499 | void |
0a95303c |
500 | Method_name(obj) |
501 | SV * obj |
502 | CODE: |
85982a32 |
503 | { |
504 | encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); |
505 | ST(0) = sv_2mortal(newSVpvn(enc->name[0],strlen(enc->name[0]))); |
506 | XSRETURN(1); |
507 | } |
0a95303c |
508 | |
509 | void |
220e2d4e |
510 | Method_cat_decode(obj, dst, src, off, term, check = 0) |
511 | SV * obj |
512 | SV * dst |
513 | SV * src |
514 | SV * off |
515 | SV * term |
516 | int check |
517 | CODE: |
518 | { |
519 | encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); |
520 | STRLEN offset = (STRLEN)SvIV(off); |
521 | int code = 0; |
522 | if (SvUTF8(src)) { |
523 | sv_utf8_downgrade(src, FALSE); |
524 | } |
525 | sv_catsv(dst, encode_method(aTHX_ enc, enc->t_utf8, src, check, |
526 | &offset, term, &code)); |
b162af07 |
527 | SvIV_set(off, (IV)offset); |
220e2d4e |
528 | if (code == ENCODE_FOUND_TERM) { |
529 | ST(0) = &PL_sv_yes; |
530 | }else{ |
531 | ST(0) = &PL_sv_no; |
532 | } |
533 | XSRETURN(1); |
534 | } |
535 | |
536 | void |
b2704119 |
537 | Method_decode(obj,src,check = 0) |
2f2b4ff2 |
538 | SV * obj |
539 | SV * src |
b2704119 |
540 | int check |
2f2b4ff2 |
541 | CODE: |
aae85ceb |
542 | { |
85982a32 |
543 | encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); |
ab3374e4 |
544 | if (SvUTF8(src)) { |
545 | sv_utf8_downgrade(src, FALSE); |
546 | } |
220e2d4e |
547 | ST(0) = encode_method(aTHX_ enc, enc->t_utf8, src, check, |
548 | NULL, Nullsv, NULL); |
85982a32 |
549 | SvUTF8_on(ST(0)); |
aae85ceb |
550 | XSRETURN(1); |
551 | } |
552 | |
553 | void |
85982a32 |
554 | Method_encode(obj,src,check = 0) |
aae85ceb |
555 | SV * obj |
85982a32 |
556 | SV * src |
557 | int check |
aae85ceb |
558 | CODE: |
559 | { |
85982a32 |
560 | encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); |
561 | sv_utf8_upgrade(src); |
220e2d4e |
562 | ST(0) = encode_method(aTHX_ enc, enc->f_utf8, src, check, |
563 | NULL, Nullsv, NULL); |
aae85ceb |
564 | XSRETURN(1); |
565 | } |
566 | |
0ab8f81e |
567 | void |
568 | Method_needs_lines(obj) |
569 | SV * obj |
570 | CODE: |
571 | { |
b32afa7c |
572 | /* encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); */ |
0ab8f81e |
573 | ST(0) = &PL_sv_no; |
574 | XSRETURN(1); |
575 | } |
576 | |
577 | void |
578 | Method_perlio_ok(obj) |
579 | SV * obj |
580 | CODE: |
581 | { |
b32afa7c |
582 | /* encode_t *enc = INT2PTR(encode_t *, SvIV(SvRV(obj))); */ |
10c5ecbb |
583 | /* require_pv(PERLIO_FILENAME); */ |
584 | |
585 | eval_pv("require PerlIO::encoding", 0); |
586 | |
587 | if (SvTRUE(get_sv("@", 0))) { |
0ab8f81e |
588 | ST(0) = &PL_sv_no; |
10c5ecbb |
589 | }else{ |
590 | ST(0) = &PL_sv_yes; |
0ab8f81e |
591 | } |
592 | XSRETURN(1); |
593 | } |
594 | |
67e989fb |
595 | MODULE = Encode PACKAGE = Encode |
2c674647 |
596 | |
597 | PROTOTYPES: ENABLE |
598 | |
67e989fb |
599 | I32 |
2c674647 |
600 | _bytes_to_utf8(sv, ...) |
85982a32 |
601 | SV * sv |
602 | CODE: |
603 | { |
604 | SV * encoding = items == 2 ? ST(1) : Nullsv; |
c6a7db43 |
605 | |
85982a32 |
606 | if (encoding) |
607 | RETVAL = _encoded_bytes_to_utf8(sv, SvPV_nolen(encoding)); |
608 | else { |
609 | STRLEN len; |
610 | U8* s = (U8*)SvPV(sv, len); |
611 | U8* converted; |
612 | |
613 | converted = bytes_to_utf8(s, &len); /* This allocs */ |
614 | sv_setpvn(sv, (char *)converted, len); |
615 | SvUTF8_on(sv); /* XXX Should we? */ |
616 | Safefree(converted); /* ... so free it */ |
617 | RETVAL = len; |
618 | } |
619 | } |
620 | OUTPUT: |
621 | RETVAL |
2c674647 |
622 | |
67e989fb |
623 | I32 |
2c674647 |
624 | _utf8_to_bytes(sv, ...) |
85982a32 |
625 | SV * sv |
626 | CODE: |
627 | { |
628 | SV * to = items > 1 ? ST(1) : Nullsv; |
629 | SV * check = items > 2 ? ST(2) : Nullsv; |
630 | |
631 | if (to) { |
632 | RETVAL = _encoded_utf8_to_bytes(sv, SvPV_nolen(to)); |
633 | } else { |
634 | STRLEN len; |
635 | U8 *s = (U8*)SvPV(sv, len); |
636 | |
637 | RETVAL = 0; |
638 | if (SvTRUE(check)) { |
639 | /* Must do things the slow way */ |
640 | U8 *dest; |
641 | /* We need a copy to pass to check() */ |
c6a7db43 |
642 | U8 *src = (U8*)savepv((char *)s); |
85982a32 |
643 | U8 *send = s + len; |
644 | |
645 | New(83, dest, len, U8); /* I think */ |
646 | |
647 | while (s < send) { |
648 | if (*s < 0x80){ |
649 | *dest++ = *s++; |
650 | } else { |
651 | STRLEN ulen; |
652 | UV uv = *s++; |
653 | |
654 | /* Have to do it all ourselves because of error routine, |
655 | aargh. */ |
656 | if (!(uv & 0x40)){ goto failure; } |
657 | if (!(uv & 0x20)) { ulen = 2; uv &= 0x1f; } |
658 | else if (!(uv & 0x10)) { ulen = 3; uv &= 0x0f; } |
659 | else if (!(uv & 0x08)) { ulen = 4; uv &= 0x07; } |
660 | else if (!(uv & 0x04)) { ulen = 5; uv &= 0x03; } |
661 | else if (!(uv & 0x02)) { ulen = 6; uv &= 0x01; } |
662 | else if (!(uv & 0x01)) { ulen = 7; uv = 0; } |
663 | else { ulen = 13; uv = 0; } |
87714904 |
664 | |
85982a32 |
665 | /* Note change to utf8.c variable naming, for variety */ |
666 | while (ulen--) { |
c6a7db43 |
667 | if ((*s & 0xc0) != 0x80){ |
668 | goto failure; |
85982a32 |
669 | } else { |
670 | uv = (uv << 6) | (*s++ & 0x3f); |
671 | } |
87714904 |
672 | } |
67e989fb |
673 | if (uv > 256) { |
674 | failure: |
85982a32 |
675 | call_failure(check, s, dest, src); |
676 | /* Now what happens? */ |
67e989fb |
677 | } |
678 | *dest++ = (U8)uv; |
85982a32 |
679 | } |
680 | } |
681 | } else { |
682 | RETVAL = (utf8_to_bytes(s, &len) ? len : 0); |
2c674647 |
683 | } |
85982a32 |
684 | } |
685 | } |
686 | OUTPUT: |
687 | RETVAL |
2c674647 |
688 | |
2c674647 |
689 | bool |
b2704119 |
690 | is_utf8(sv, check = 0) |
4411f3b6 |
691 | SV * sv |
b2704119 |
692 | int check |
85982a32 |
693 | CODE: |
694 | { |
695 | if (SvGMAGICAL(sv)) /* it could be $1, for example */ |
696 | sv = newSVsv(sv); /* GMAGIG will be done */ |
697 | if (SvPOK(sv)) { |
698 | RETVAL = SvUTF8(sv) ? TRUE : FALSE; |
699 | if (RETVAL && |
700 | check && |
701 | !is_utf8_string((U8*)SvPVX(sv), SvCUR(sv))) |
2c674647 |
702 | RETVAL = FALSE; |
85982a32 |
703 | } else { |
704 | RETVAL = FALSE; |
705 | } |
706 | if (sv != ST(0)) |
707 | SvREFCNT_dec(sv); /* it was a temp copy */ |
708 | } |
709 | OUTPUT: |
710 | RETVAL |
2c674647 |
711 | |
712 | SV * |
4411f3b6 |
713 | _utf8_on(sv) |
85982a32 |
714 | SV * sv |
715 | CODE: |
716 | { |
717 | if (SvPOK(sv)) { |
718 | SV *rsv = newSViv(SvUTF8(sv)); |
719 | RETVAL = rsv; |
720 | SvUTF8_on(sv); |
721 | } else { |
722 | RETVAL = &PL_sv_undef; |
723 | } |
724 | } |
725 | OUTPUT: |
726 | RETVAL |
2c674647 |
727 | |
728 | SV * |
4411f3b6 |
729 | _utf8_off(sv) |
85982a32 |
730 | SV * sv |
731 | CODE: |
732 | { |
733 | if (SvPOK(sv)) { |
734 | SV *rsv = newSViv(SvUTF8(sv)); |
735 | RETVAL = rsv; |
736 | SvUTF8_off(sv); |
737 | } else { |
738 | RETVAL = &PL_sv_undef; |
739 | } |
740 | } |
741 | OUTPUT: |
742 | RETVAL |
743 | |
85982a32 |
744 | int |
745 | DIE_ON_ERR() |
746 | CODE: |
747 | RETVAL = ENCODE_DIE_ON_ERR; |
748 | OUTPUT: |
749 | RETVAL |
750 | |
c6a7db43 |
751 | int |
85982a32 |
752 | WARN_ON_ERR() |
753 | CODE: |
754 | RETVAL = ENCODE_WARN_ON_ERR; |
755 | OUTPUT: |
756 | RETVAL |
757 | |
758 | int |
759 | LEAVE_SRC() |
760 | CODE: |
761 | RETVAL = ENCODE_LEAVE_SRC; |
762 | OUTPUT: |
763 | RETVAL |
764 | |
765 | int |
766 | RETURN_ON_ERR() |
767 | CODE: |
768 | RETVAL = ENCODE_RETURN_ON_ERR; |
769 | OUTPUT: |
770 | RETVAL |
771 | |
772 | int |
773 | PERLQQ() |
774 | CODE: |
775 | RETVAL = ENCODE_PERLQQ; |
776 | OUTPUT: |
777 | RETVAL |
778 | |
779 | int |
af1f55d9 |
780 | HTMLCREF() |
781 | CODE: |
782 | RETVAL = ENCODE_HTMLCREF; |
783 | OUTPUT: |
784 | RETVAL |
785 | |
786 | int |
787 | XMLCREF() |
788 | CODE: |
789 | RETVAL = ENCODE_XMLCREF; |
790 | OUTPUT: |
791 | RETVAL |
792 | |
793 | int |
85982a32 |
794 | FB_DEFAULT() |
795 | CODE: |
796 | RETVAL = ENCODE_FB_DEFAULT; |
797 | OUTPUT: |
798 | RETVAL |
799 | |
800 | int |
801 | FB_CROAK() |
802 | CODE: |
803 | RETVAL = ENCODE_FB_CROAK; |
804 | OUTPUT: |
805 | RETVAL |
806 | |
807 | int |
808 | FB_QUIET() |
809 | CODE: |
810 | RETVAL = ENCODE_FB_QUIET; |
811 | OUTPUT: |
812 | RETVAL |
813 | |
814 | int |
815 | FB_WARN() |
816 | CODE: |
817 | RETVAL = ENCODE_FB_WARN; |
818 | OUTPUT: |
819 | RETVAL |
820 | |
821 | int |
822 | FB_PERLQQ() |
823 | CODE: |
824 | RETVAL = ENCODE_FB_PERLQQ; |
825 | OUTPUT: |
826 | RETVAL |
2c674647 |
827 | |
af1f55d9 |
828 | int |
829 | FB_HTMLCREF() |
830 | CODE: |
831 | RETVAL = ENCODE_FB_HTMLCREF; |
832 | OUTPUT: |
833 | RETVAL |
834 | |
835 | int |
836 | FB_XMLCREF() |
837 | CODE: |
838 | RETVAL = ENCODE_FB_XMLCREF; |
839 | OUTPUT: |
840 | RETVAL |
841 | |
33af2bc7 |
842 | BOOT: |
843 | { |
85982a32 |
844 | #include "def_t.h" |
e7cbefb8 |
845 | #include "def_t.exh" |
33af2bc7 |
846 | } |