Commit | Line | Data |
a0ed51b3 |
1 | /* utf8.c |
2 | * |
3818b22b |
3 | * Copyright (c) 1998-2000, Larry Wall |
a0ed51b3 |
4 | * |
5 | * You may distribute under the terms of either the GNU General Public |
6 | * License or the Artistic License, as specified in the README file. |
7 | * |
8 | */ |
9 | |
10 | /* |
11 | * 'What a fix!' said Sam. 'That's the one place in all the lands we've ever |
12 | * heard of that we don't want to see any closer; and that's the one place |
13 | * we're trying to get to! And that's just where we can't get, nohow.' |
14 | * |
15 | * 'Well do I understand your speech,' he answered in the same language; |
16 | * 'yet few strangers do so. Why then do you not speak in the Common Tongue, |
17 | * as is the custom in the West, if you wish to be answered?' |
18 | * |
19 | * ...the travellers perceived that the floor was paved with stones of many |
20 | * hues; branching runes and strange devices intertwined beneath their feet. |
21 | */ |
22 | |
23 | #include "EXTERN.h" |
864dbfa3 |
24 | #define PERL_IN_UTF8_C |
a0ed51b3 |
25 | #include "perl.h" |
26 | |
27 | /* Unicode support */ |
28 | |
dfe13c55 |
29 | U8 * |
ad391ad9 |
30 | Perl_uv_to_utf8(pTHX_ U8 *d, UV uv) /* the d must be UTF8_MAXLEN+1 deep */ |
a0ed51b3 |
31 | { |
32 | if (uv < 0x80) { |
33 | *d++ = uv; |
ad391ad9 |
34 | *d = 0; |
a0ed51b3 |
35 | return d; |
36 | } |
37 | if (uv < 0x800) { |
38 | *d++ = (( uv >> 6) | 0xc0); |
39 | *d++ = (( uv & 0x3f) | 0x80); |
ad391ad9 |
40 | *d = 0; |
a0ed51b3 |
41 | return d; |
42 | } |
43 | if (uv < 0x10000) { |
44 | *d++ = (( uv >> 12) | 0xe0); |
45 | *d++ = (((uv >> 6) & 0x3f) | 0x80); |
46 | *d++ = (( uv & 0x3f) | 0x80); |
ad391ad9 |
47 | *d = 0; |
a0ed51b3 |
48 | return d; |
49 | } |
50 | if (uv < 0x200000) { |
51 | *d++ = (( uv >> 18) | 0xf0); |
52 | *d++ = (((uv >> 12) & 0x3f) | 0x80); |
53 | *d++ = (((uv >> 6) & 0x3f) | 0x80); |
54 | *d++ = (( uv & 0x3f) | 0x80); |
ad391ad9 |
55 | *d = 0; |
a0ed51b3 |
56 | return d; |
57 | } |
58 | if (uv < 0x4000000) { |
59 | *d++ = (( uv >> 24) | 0xf8); |
60 | *d++ = (((uv >> 18) & 0x3f) | 0x80); |
61 | *d++ = (((uv >> 12) & 0x3f) | 0x80); |
62 | *d++ = (((uv >> 6) & 0x3f) | 0x80); |
63 | *d++ = (( uv & 0x3f) | 0x80); |
ad391ad9 |
64 | *d = 0; |
a0ed51b3 |
65 | return d; |
66 | } |
67 | if (uv < 0x80000000) { |
68 | *d++ = (( uv >> 30) | 0xfc); |
69 | *d++ = (((uv >> 24) & 0x3f) | 0x80); |
70 | *d++ = (((uv >> 18) & 0x3f) | 0x80); |
71 | *d++ = (((uv >> 12) & 0x3f) | 0x80); |
72 | *d++ = (((uv >> 6) & 0x3f) | 0x80); |
73 | *d++ = (( uv & 0x3f) | 0x80); |
ad391ad9 |
74 | *d = 0; |
a0ed51b3 |
75 | return d; |
76 | } |
6b8eaf93 |
77 | #ifdef HAS_QUAD |
d7578b48 |
78 | if (uv < UTF8_QUAD_MAX) |
a0ed51b3 |
79 | #endif |
80 | { |
81 | *d++ = 0xfe; /* Can't match U+FEFF! */ |
82 | *d++ = (((uv >> 30) & 0x3f) | 0x80); |
83 | *d++ = (((uv >> 24) & 0x3f) | 0x80); |
84 | *d++ = (((uv >> 18) & 0x3f) | 0x80); |
85 | *d++ = (((uv >> 12) & 0x3f) | 0x80); |
86 | *d++ = (((uv >> 6) & 0x3f) | 0x80); |
87 | *d++ = (( uv & 0x3f) | 0x80); |
ad391ad9 |
88 | *d = 0; |
a0ed51b3 |
89 | return d; |
90 | } |
6b8eaf93 |
91 | #ifdef HAS_QUAD |
a0ed51b3 |
92 | { |
93 | *d++ = 0xff; /* Can't match U+FFFE! */ |
3c77ea2b |
94 | *d++ = 0x80; /* 6 Reserved bits */ |
95 | *d++ = (((uv >> 60) & 0x0f) | 0x80); /* 2 Reserved bits */ |
96 | *d++ = (((uv >> 54) & 0x3f) | 0x80); |
97 | *d++ = (((uv >> 48) & 0x3f) | 0x80); |
98 | *d++ = (((uv >> 42) & 0x3f) | 0x80); |
a0ed51b3 |
99 | *d++ = (((uv >> 36) & 0x3f) | 0x80); |
100 | *d++ = (((uv >> 30) & 0x3f) | 0x80); |
101 | *d++ = (((uv >> 24) & 0x3f) | 0x80); |
102 | *d++ = (((uv >> 18) & 0x3f) | 0x80); |
103 | *d++ = (((uv >> 12) & 0x3f) | 0x80); |
104 | *d++ = (((uv >> 6) & 0x3f) | 0x80); |
105 | *d++ = (( uv & 0x3f) | 0x80); |
ad391ad9 |
106 | *d = 0; |
a0ed51b3 |
107 | return d; |
108 | } |
109 | #endif |
110 | } |
111 | |
386d01d6 |
112 | /* Tests if some arbitrary number of bytes begins in a valid UTF-8 character. |
113 | * The actual number of bytes in the UTF-8 character will be returned if it |
114 | * is valid, otherwise 0. */ |
067a85ef |
115 | STRLEN |
386d01d6 |
116 | Perl_is_utf8_char(pTHX_ U8 *s) |
117 | { |
118 | U8 u = *s; |
067a85ef |
119 | STRLEN slen, len; |
120 | UV uv, ouv; |
386d01d6 |
121 | |
067a85ef |
122 | if (u <= 0x7f) |
386d01d6 |
123 | return 1; |
124 | |
067a85ef |
125 | if (u >= 0x80 && u <= 0xbf) |
386d01d6 |
126 | return 0; |
127 | |
9f07fdcd |
128 | len = UTF8SKIP(s); |
386d01d6 |
129 | |
067a85ef |
130 | if (len < 2 || (u >= 0xc0 && u <= 0xfd && s[1] < 0x80)) |
131 | return 0; |
132 | |
386d01d6 |
133 | slen = len - 1; |
134 | s++; |
067a85ef |
135 | uv = u; |
136 | ouv = uv; |
386d01d6 |
137 | while (slen--) { |
138 | if ((*s & 0xc0) != 0x80) |
139 | return 0; |
8850bf83 |
140 | uv = UTF8_ACCUMULATE(uv, *s); |
067a85ef |
141 | if (uv < ouv) |
142 | return 0; |
143 | ouv = uv; |
386d01d6 |
144 | s++; |
145 | } |
067a85ef |
146 | |
5bbb0b5a |
147 | if (UNISKIP(uv) < len) |
067a85ef |
148 | return 0; |
149 | |
386d01d6 |
150 | return len; |
151 | } |
152 | |
6662521e |
153 | /* |
b2a2e44b |
154 | =for apidoc Am|is_utf8_string|U8 *s|STRLEN len |
6662521e |
155 | |
156 | Returns true if first C<len> bytes of the given string form valid a UTF8 |
157 | string, false otherwise. |
158 | |
159 | =cut |
160 | */ |
161 | |
8e84507e |
162 | bool |
6662521e |
163 | Perl_is_utf8_string(pTHX_ U8 *s, STRLEN len) |
164 | { |
067a85ef |
165 | U8* x = s; |
166 | U8* send = s + len; |
167 | STRLEN c; |
168 | |
6662521e |
169 | while (x < send) { |
170 | c = is_utf8_char(x); |
067a85ef |
171 | if (!c) |
172 | return FALSE; |
6662521e |
173 | x += c; |
067a85ef |
174 | if (x > send) |
175 | return FALSE; |
6662521e |
176 | } |
067a85ef |
177 | |
178 | return TRUE; |
6662521e |
179 | } |
180 | |
67e989fb |
181 | /* |
be2c7115 |
182 | =for apidoc Am|U8* s|utf8_to_uv|STRLEN curlen|STRLEN *retlen|U32 flags |
67e989fb |
183 | |
184 | Returns the character value of the first character in the string C<s> |
ba210ebe |
185 | which is assumed to be in UTF8 encoding and no longer than C<curlen>; |
7df053ec |
186 | C<retlen> will be set to the length, in bytes, of that character. |
67e989fb |
187 | |
188 | If C<s> does not point to a well-formed UTF8 character, the behaviour |
dcad2880 |
189 | is dependent on the value of C<flags>: if it contains UTF8_CHECK_ONLY, |
190 | it is assumed that the caller will raise a warning, and this function |
28d3d195 |
191 | will silently just set C<retlen> to C<-1> and return zero. If the |
192 | C<flags> does not contain UTF8_CHECK_ONLY, warnings about |
193 | malformations will be given, C<retlen> will be set to the expected |
194 | length of the UTF-8 character in bytes, and zero will be returned. |
195 | |
196 | The C<flags> can also contain various flags to allow deviations from |
197 | the strict UTF-8 encoding (see F<utf8.h>). |
67e989fb |
198 | |
dcad2880 |
199 | =cut */ |
67e989fb |
200 | |
a0ed51b3 |
201 | UV |
dcad2880 |
202 | Perl_utf8_to_uv(pTHX_ U8* s, STRLEN curlen, STRLEN* retlen, U32 flags) |
a0ed51b3 |
203 | { |
ba210ebe |
204 | UV uv = *s, ouv; |
205 | STRLEN len = 1; |
7bf1b6bb |
206 | #ifdef EBCDIC |
207 | bool dowarn = 0; |
208 | #else |
ba210ebe |
209 | bool dowarn = ckWARN_d(WARN_UTF8); |
7bf1b6bb |
210 | #endif |
ba210ebe |
211 | STRLEN expectlen = 0; |
212 | |
0c443dc2 |
213 | if (curlen == 0) { |
214 | if (dowarn) |
215 | Perl_warner(aTHX_ WARN_UTF8, |
216 | "Malformed UTF-8 character (an empty string)"); |
217 | goto malformed; |
218 | } |
219 | |
421a8bf2 |
220 | if (UTF8_IS_ASCII(uv)) { |
a0ed51b3 |
221 | if (retlen) |
222 | *retlen = 1; |
223 | return *s; |
224 | } |
67e989fb |
225 | |
421a8bf2 |
226 | if (UTF8_IS_CONTINUATION(uv) && |
fcc8fcf6 |
227 | !(flags & UTF8_ALLOW_CONTINUATION)) { |
ba210ebe |
228 | if (dowarn) |
229 | Perl_warner(aTHX_ WARN_UTF8, |
efbcad09 |
230 | "Malformed UTF-8 character (unexpected continuation byte 0x%02"UVxf")", |
ba210ebe |
231 | uv); |
232 | goto malformed; |
233 | } |
234 | |
421a8bf2 |
235 | if (UTF8_IS_START(uv) && curlen > 1 && !UTF8_IS_CONTINUATION(s[1]) && |
fcc8fcf6 |
236 | !(flags & UTF8_ALLOW_NON_CONTINUATION)) { |
ba210ebe |
237 | if (dowarn) |
238 | Perl_warner(aTHX_ WARN_UTF8, |
421a8bf2 |
239 | "Malformed UTF-8 character (unexpected non-continuation byte 0x%02"UVxf" after start byte 0x%02"UVxf")", |
9e29e5ff |
240 | (UV)s[1], uv); |
ba210ebe |
241 | goto malformed; |
242 | } |
fcc8fcf6 |
243 | |
244 | if ((uv == 0xfe || uv == 0xff) && |
245 | !(flags & UTF8_ALLOW_FE_FF)) { |
ba210ebe |
246 | if (dowarn) |
247 | Perl_warner(aTHX_ WARN_UTF8, |
efbcad09 |
248 | "Malformed UTF-8 character (byte 0x%02"UVxf")", |
ba210ebe |
249 | uv); |
250 | goto malformed; |
a0ed51b3 |
251 | } |
fcc8fcf6 |
252 | |
ba210ebe |
253 | if (!(uv & 0x20)) { len = 2; uv &= 0x1f; } |
254 | else if (!(uv & 0x10)) { len = 3; uv &= 0x0f; } |
255 | else if (!(uv & 0x08)) { len = 4; uv &= 0x07; } |
256 | else if (!(uv & 0x04)) { len = 5; uv &= 0x03; } |
257 | else if (!(uv & 0x02)) { len = 6; uv &= 0x01; } |
258 | else if (!(uv & 0x01)) { len = 7; uv = 0; } |
3c77ea2b |
259 | else { len = 13; uv = 0; } /* whoa! */ |
fcc8fcf6 |
260 | |
a0ed51b3 |
261 | if (retlen) |
262 | *retlen = len; |
ba210ebe |
263 | |
264 | expectlen = len; |
265 | |
fcc8fcf6 |
266 | if ((curlen < expectlen) && |
267 | !(flags & UTF8_ALLOW_SHORT)) { |
ba210ebe |
268 | if (dowarn) |
269 | Perl_warner(aTHX_ WARN_UTF8, |
270 | "Malformed UTF-8 character (%d byte%s, need %d)", |
0c443dc2 |
271 | curlen, curlen == 1 ? "" : "s", expectlen); |
ba210ebe |
272 | goto malformed; |
273 | } |
274 | |
275 | len--; |
a0ed51b3 |
276 | s++; |
ba210ebe |
277 | ouv = uv; |
278 | |
a0ed51b3 |
279 | while (len--) { |
421a8bf2 |
280 | if (!UTF8_IS_CONTINUATION(*s) && |
281 | !(flags & UTF8_ALLOW_NON_CONTINUATION)) { |
ba210ebe |
282 | if (dowarn) |
283 | Perl_warner(aTHX_ WARN_UTF8, |
421a8bf2 |
284 | "Malformed UTF-8 character (unexpected non-continuation byte 0x%02x)", |
ba210ebe |
285 | *s); |
286 | goto malformed; |
a0ed51b3 |
287 | } |
288 | else |
8850bf83 |
289 | uv = UTF8_ACCUMULATE(uv, *s); |
ba210ebe |
290 | if (uv < ouv) { |
fcc8fcf6 |
291 | /* This cannot be allowed. */ |
ba210ebe |
292 | if (dowarn) |
293 | Perl_warner(aTHX_ WARN_UTF8, |
294 | "Malformed UTF-8 character (overflow at 0x%"UVxf", byte 0x%02x)", |
295 | ouv, *s); |
296 | goto malformed; |
297 | } |
298 | s++; |
299 | ouv = uv; |
300 | } |
301 | |
421a8bf2 |
302 | if (UNICODE_IS_SURROGATE(uv) && |
fcc8fcf6 |
303 | !(flags & UTF8_ALLOW_SURROGATE)) { |
ba210ebe |
304 | if (dowarn) |
305 | Perl_warner(aTHX_ WARN_UTF8, |
306 | "Malformed UTF-8 character (UTF-16 surrogate 0x%04"UVxf")", |
307 | uv); |
308 | goto malformed; |
421a8bf2 |
309 | } else if (UNICODE_IS_BYTE_ORDER_MARK(uv) && |
fcc8fcf6 |
310 | !(flags & UTF8_ALLOW_BOM)) { |
ba210ebe |
311 | if (dowarn) |
312 | Perl_warner(aTHX_ WARN_UTF8, |
313 | "Malformed UTF-8 character (byte order mark 0x%04"UVxf")", |
314 | uv); |
315 | goto malformed; |
fcc8fcf6 |
316 | } else if ((expectlen > UNISKIP(uv)) && |
317 | !(flags & UTF8_ALLOW_LONG)) { |
ba210ebe |
318 | if (dowarn) |
319 | Perl_warner(aTHX_ WARN_UTF8, |
320 | "Malformed UTF-8 character (%d byte%s, need %d)", |
0c443dc2 |
321 | expectlen, expectlen == 1 ? "": "s", UNISKIP(uv)); |
ba210ebe |
322 | goto malformed; |
421a8bf2 |
323 | } else if (UNICODE_IS_ILLEGAL(uv) && |
a9917092 |
324 | !(flags & UTF8_ALLOW_FFFF)) { |
325 | if (dowarn) |
326 | Perl_warner(aTHX_ WARN_UTF8, |
327 | "Malformed UTF-8 character (character 0x%04"UVxf")", |
328 | uv); |
329 | goto malformed; |
a0ed51b3 |
330 | } |
ba210ebe |
331 | |
a0ed51b3 |
332 | return uv; |
ba210ebe |
333 | |
334 | malformed: |
335 | |
fcc8fcf6 |
336 | if (flags & UTF8_CHECK_ONLY) { |
ba210ebe |
337 | if (retlen) |
cc366d4b |
338 | *retlen = -1; |
ba210ebe |
339 | return 0; |
340 | } |
341 | |
342 | if (retlen) |
28d3d195 |
343 | *retlen = expectlen ? expectlen : len; |
ba210ebe |
344 | |
28d3d195 |
345 | return 0; |
a0ed51b3 |
346 | } |
347 | |
8e84507e |
348 | /* |
dcad2880 |
349 | =for apidoc Am|U8* s|utf8_to_uv_simple|STRLEN *retlen |
8e84507e |
350 | |
351 | Returns the character value of the first character in the string C<s> |
352 | which is assumed to be in UTF8 encoding; C<retlen> will be set to the |
7df053ec |
353 | length, in bytes, of that character. |
8e84507e |
354 | |
ba210ebe |
355 | If C<s> does not point to a well-formed UTF8 character, zero is |
356 | returned and retlen is set, if possible, to -1. |
8e84507e |
357 | |
358 | =cut |
359 | */ |
360 | |
361 | UV |
dcad2880 |
362 | Perl_utf8_to_uv_simple(pTHX_ U8* s, STRLEN* retlen) |
8e84507e |
363 | { |
2e4dc9fc |
364 | return Perl_utf8_to_uv(aTHX_ s, UTF8_MAXLEN, retlen, 0); |
8e84507e |
365 | } |
366 | |
b76347f2 |
367 | /* |
b06226ff |
368 | =for apidoc Am|STRLEN|utf8_length|U8* s|U8 *e |
b76347f2 |
369 | |
370 | Return the length of the UTF-8 char encoded string C<s> in characters. |
02eb7b47 |
371 | Stops at C<e> (inclusive). If C<e E<lt> s> or if the scan would end |
372 | up past C<e>, croaks. |
b76347f2 |
373 | |
374 | =cut |
375 | */ |
376 | |
377 | STRLEN |
378 | Perl_utf8_length(pTHX_ U8* s, U8* e) |
379 | { |
380 | STRLEN len = 0; |
381 | |
8850bf83 |
382 | /* Note: cannot use UTF8_IS_...() too eagerly here since e.g. |
383 | * the bitops (especially ~) can create illegal UTF-8. |
384 | * In other words: in Perl UTF-8 is not just for Unicode. */ |
385 | |
b76347f2 |
386 | if (e < s) |
02eb7b47 |
387 | Perl_croak(aTHX_ "panic: utf8_length: unexpected end"); |
b76347f2 |
388 | while (s < e) { |
02eb7b47 |
389 | U8 t = UTF8SKIP(s); |
b76347f2 |
390 | |
391 | if (e - s < t) |
02eb7b47 |
392 | Perl_croak(aTHX_ "panic: utf8_length: unaligned end"); |
b76347f2 |
393 | s += t; |
394 | len++; |
395 | } |
396 | |
397 | return len; |
398 | } |
399 | |
b06226ff |
400 | /* |
401 | =for apidoc Am|IV|utf8_distance|U8 *a|U8 *b |
402 | |
403 | Returns the number of UTF8 characters between the UTF-8 pointers C<a> |
404 | and C<b>. |
405 | |
406 | WARNING: use only if you *know* that the pointers point inside the |
407 | same UTF-8 buffer. |
408 | |
409 | =cut */ |
a0ed51b3 |
410 | |
02eb7b47 |
411 | IV |
864dbfa3 |
412 | Perl_utf8_distance(pTHX_ U8 *a, U8 *b) |
a0ed51b3 |
413 | { |
02eb7b47 |
414 | IV off = 0; |
415 | |
8850bf83 |
416 | /* Note: cannot use UTF8_IS_...() too eagerly here since e.g. |
417 | * the bitops (especially ~) can create illegal UTF-8. |
418 | * In other words: in Perl UTF-8 is not just for Unicode. */ |
419 | |
a0ed51b3 |
420 | if (a < b) { |
421 | while (a < b) { |
02eb7b47 |
422 | U8 c = UTF8SKIP(a); |
423 | |
424 | if (b - a < c) |
425 | Perl_croak(aTHX_ "panic: utf8_distance: unaligned end"); |
426 | a += c; |
a0ed51b3 |
427 | off--; |
428 | } |
429 | } |
430 | else { |
431 | while (b < a) { |
02eb7b47 |
432 | U8 c = UTF8SKIP(b); |
433 | |
434 | if (a - b < c) |
435 | Perl_croak(aTHX_ "panic: utf8_distance: unaligned end"); |
436 | b += c; |
a0ed51b3 |
437 | off++; |
438 | } |
439 | } |
02eb7b47 |
440 | |
a0ed51b3 |
441 | return off; |
442 | } |
443 | |
b06226ff |
444 | /* |
445 | =for apidoc Am|U8*|utf8_hop|U8 *s|I32 off |
446 | |
8850bf83 |
447 | Return the UTF-8 pointer C<s> displaced by C<off> characters, either |
448 | forward or backward. |
b06226ff |
449 | |
450 | WARNING: do not use the following unless you *know* C<off> is within |
8850bf83 |
451 | the UTF-8 data pointed to by C<s> *and* that on entry C<s> is aligned |
452 | on the first byte of character or just after the last byte of a character. |
b06226ff |
453 | |
454 | =cut */ |
a0ed51b3 |
455 | |
456 | U8 * |
864dbfa3 |
457 | Perl_utf8_hop(pTHX_ U8 *s, I32 off) |
a0ed51b3 |
458 | { |
8850bf83 |
459 | /* Note: cannot use UTF8_IS_...() too eagerly here since e.g |
460 | * the bitops (especially ~) can create illegal UTF-8. |
461 | * In other words: in Perl UTF-8 is not just for Unicode. */ |
462 | |
a0ed51b3 |
463 | if (off >= 0) { |
464 | while (off--) |
465 | s += UTF8SKIP(s); |
466 | } |
467 | else { |
468 | while (off++) { |
469 | s--; |
8850bf83 |
470 | while (UTF8_IS_CONTINUATION(*s)) |
471 | s--; |
a0ed51b3 |
472 | } |
473 | } |
474 | return s; |
475 | } |
476 | |
6940069f |
477 | /* |
246fae53 |
478 | =for apidoc Am|U8 *|utf8_to_bytes|U8 *s|STRLEN *len |
6940069f |
479 | |
246fae53 |
480 | Converts a string C<s> of length C<len> from UTF8 into byte encoding. |
481 | Unlike C<bytes_to_utf8>, this over-writes the original string, and |
482 | updates len to contain the new length. |
67e989fb |
483 | Returns zero on failure, setting C<len> to -1. |
6940069f |
484 | |
485 | =cut |
486 | */ |
487 | |
488 | U8 * |
246fae53 |
489 | Perl_utf8_to_bytes(pTHX_ U8* s, STRLEN *len) |
6940069f |
490 | { |
6940069f |
491 | U8 *send; |
492 | U8 *d; |
dcad2880 |
493 | U8 *save = s; |
246fae53 |
494 | |
495 | /* ensure valid UTF8 and chars < 256 before updating string */ |
dcad2880 |
496 | for (send = s + *len; s < send; ) { |
497 | U8 c = *s++; |
498 | |
9f9ab905 |
499 | if (c >= 0x80 && |
dcad2880 |
500 | ((s >= send) || |
501 | ((*s++ & 0xc0) != 0x80) || ((c & 0xfe) != 0xc2))) { |
502 | *len = -1; |
503 | return 0; |
504 | } |
246fae53 |
505 | } |
dcad2880 |
506 | |
507 | d = s = save; |
6940069f |
508 | while (s < send) { |
ed646e6e |
509 | STRLEN ulen; |
510 | *d++ = (U8)utf8_to_uv_simple(s, &ulen); |
511 | s += ulen; |
6940069f |
512 | } |
513 | *d = '\0'; |
246fae53 |
514 | *len = d - save; |
6940069f |
515 | return save; |
516 | } |
517 | |
518 | /* |
6662521e |
519 | =for apidoc Am|U8 *|bytes_to_utf8|U8 *s|STRLEN *len |
6940069f |
520 | |
521 | Converts a string C<s> of length C<len> from ASCII into UTF8 encoding. |
6662521e |
522 | Returns a pointer to the newly-created string, and sets C<len> to |
523 | reflect the new length. |
6940069f |
524 | |
497711e7 |
525 | =cut |
6940069f |
526 | */ |
527 | |
528 | U8* |
6662521e |
529 | Perl_bytes_to_utf8(pTHX_ U8* s, STRLEN *len) |
6940069f |
530 | { |
6940069f |
531 | U8 *send; |
532 | U8 *d; |
533 | U8 *dst; |
6662521e |
534 | send = s + (*len); |
6940069f |
535 | |
6662521e |
536 | Newz(801, d, (*len) * 2 + 1, U8); |
6940069f |
537 | dst = d; |
538 | |
539 | while (s < send) { |
540 | if (*s < 0x80) |
541 | *d++ = *s++; |
542 | else { |
543 | UV uv = *s++; |
544 | *d++ = (( uv >> 6) | 0xc0); |
545 | *d++ = (( uv & 0x3f) | 0x80); |
546 | } |
547 | } |
548 | *d = '\0'; |
6662521e |
549 | *len = d-dst; |
6940069f |
550 | return dst; |
551 | } |
552 | |
a0ed51b3 |
553 | /* |
dea0fc0b |
554 | * Convert native (big-endian) or reversed (little-endian) UTF-16 to UTF-8. |
a0ed51b3 |
555 | * |
556 | * Destination must be pre-extended to 3/2 source. Do not use in-place. |
557 | * We optimize for native, for obvious reasons. */ |
558 | |
559 | U8* |
dea0fc0b |
560 | Perl_utf16_to_utf8(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen) |
a0ed51b3 |
561 | { |
dea0fc0b |
562 | U8* pend; |
563 | U8* dstart = d; |
564 | |
565 | if (bytelen & 1) |
a7867d0a |
566 | Perl_croak(aTHX_ "panic: utf16_to_utf8: odd bytelen"); |
dea0fc0b |
567 | |
568 | pend = p + bytelen; |
569 | |
a0ed51b3 |
570 | while (p < pend) { |
dea0fc0b |
571 | UV uv = (p[0] << 8) + p[1]; /* UTF-16BE */ |
572 | p += 2; |
a0ed51b3 |
573 | if (uv < 0x80) { |
574 | *d++ = uv; |
575 | continue; |
576 | } |
577 | if (uv < 0x800) { |
578 | *d++ = (( uv >> 6) | 0xc0); |
579 | *d++ = (( uv & 0x3f) | 0x80); |
580 | continue; |
581 | } |
582 | if (uv >= 0xd800 && uv < 0xdbff) { /* surrogates */ |
dea0fc0b |
583 | UV low = *p++; |
584 | if (low < 0xdc00 || low >= 0xdfff) |
585 | Perl_croak(aTHX_ "Malformed UTF-16 surrogate"); |
a0ed51b3 |
586 | uv = ((uv - 0xd800) << 10) + (low - 0xdc00) + 0x10000; |
587 | } |
588 | if (uv < 0x10000) { |
589 | *d++ = (( uv >> 12) | 0xe0); |
590 | *d++ = (((uv >> 6) & 0x3f) | 0x80); |
591 | *d++ = (( uv & 0x3f) | 0x80); |
592 | continue; |
593 | } |
594 | else { |
595 | *d++ = (( uv >> 18) | 0xf0); |
596 | *d++ = (((uv >> 12) & 0x3f) | 0x80); |
597 | *d++ = (((uv >> 6) & 0x3f) | 0x80); |
598 | *d++ = (( uv & 0x3f) | 0x80); |
599 | continue; |
600 | } |
601 | } |
dea0fc0b |
602 | *newlen = d - dstart; |
a0ed51b3 |
603 | return d; |
604 | } |
605 | |
606 | /* Note: this one is slightly destructive of the source. */ |
607 | |
608 | U8* |
dea0fc0b |
609 | Perl_utf16_to_utf8_reversed(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen) |
a0ed51b3 |
610 | { |
611 | U8* s = (U8*)p; |
612 | U8* send = s + bytelen; |
613 | while (s < send) { |
614 | U8 tmp = s[0]; |
615 | s[0] = s[1]; |
616 | s[1] = tmp; |
617 | s += 2; |
618 | } |
dea0fc0b |
619 | return utf16_to_utf8(p, d, bytelen, newlen); |
a0ed51b3 |
620 | } |
621 | |
622 | /* for now these are all defined (inefficiently) in terms of the utf8 versions */ |
623 | |
624 | bool |
864dbfa3 |
625 | Perl_is_uni_alnum(pTHX_ U32 c) |
a0ed51b3 |
626 | { |
ad391ad9 |
627 | U8 tmpbuf[UTF8_MAXLEN+1]; |
a0ed51b3 |
628 | uv_to_utf8(tmpbuf, (UV)c); |
629 | return is_utf8_alnum(tmpbuf); |
630 | } |
631 | |
632 | bool |
b8c5462f |
633 | Perl_is_uni_alnumc(pTHX_ U32 c) |
634 | { |
ad391ad9 |
635 | U8 tmpbuf[UTF8_MAXLEN+1]; |
b8c5462f |
636 | uv_to_utf8(tmpbuf, (UV)c); |
637 | return is_utf8_alnumc(tmpbuf); |
638 | } |
639 | |
640 | bool |
864dbfa3 |
641 | Perl_is_uni_idfirst(pTHX_ U32 c) |
a0ed51b3 |
642 | { |
ad391ad9 |
643 | U8 tmpbuf[UTF8_MAXLEN+1]; |
a0ed51b3 |
644 | uv_to_utf8(tmpbuf, (UV)c); |
645 | return is_utf8_idfirst(tmpbuf); |
646 | } |
647 | |
648 | bool |
864dbfa3 |
649 | Perl_is_uni_alpha(pTHX_ U32 c) |
a0ed51b3 |
650 | { |
ad391ad9 |
651 | U8 tmpbuf[UTF8_MAXLEN+1]; |
a0ed51b3 |
652 | uv_to_utf8(tmpbuf, (UV)c); |
653 | return is_utf8_alpha(tmpbuf); |
654 | } |
655 | |
656 | bool |
4d61ec05 |
657 | Perl_is_uni_ascii(pTHX_ U32 c) |
658 | { |
ad391ad9 |
659 | U8 tmpbuf[UTF8_MAXLEN+1]; |
4d61ec05 |
660 | uv_to_utf8(tmpbuf, (UV)c); |
661 | return is_utf8_ascii(tmpbuf); |
662 | } |
663 | |
664 | bool |
864dbfa3 |
665 | Perl_is_uni_space(pTHX_ U32 c) |
a0ed51b3 |
666 | { |
ad391ad9 |
667 | U8 tmpbuf[UTF8_MAXLEN+1]; |
a0ed51b3 |
668 | uv_to_utf8(tmpbuf, (UV)c); |
669 | return is_utf8_space(tmpbuf); |
670 | } |
671 | |
672 | bool |
864dbfa3 |
673 | Perl_is_uni_digit(pTHX_ U32 c) |
a0ed51b3 |
674 | { |
ad391ad9 |
675 | U8 tmpbuf[UTF8_MAXLEN+1]; |
a0ed51b3 |
676 | uv_to_utf8(tmpbuf, (UV)c); |
677 | return is_utf8_digit(tmpbuf); |
678 | } |
679 | |
680 | bool |
864dbfa3 |
681 | Perl_is_uni_upper(pTHX_ U32 c) |
a0ed51b3 |
682 | { |
ad391ad9 |
683 | U8 tmpbuf[UTF8_MAXLEN+1]; |
a0ed51b3 |
684 | uv_to_utf8(tmpbuf, (UV)c); |
685 | return is_utf8_upper(tmpbuf); |
686 | } |
687 | |
688 | bool |
864dbfa3 |
689 | Perl_is_uni_lower(pTHX_ U32 c) |
a0ed51b3 |
690 | { |
ad391ad9 |
691 | U8 tmpbuf[UTF8_MAXLEN+1]; |
a0ed51b3 |
692 | uv_to_utf8(tmpbuf, (UV)c); |
693 | return is_utf8_lower(tmpbuf); |
694 | } |
695 | |
696 | bool |
b8c5462f |
697 | Perl_is_uni_cntrl(pTHX_ U32 c) |
698 | { |
ad391ad9 |
699 | U8 tmpbuf[UTF8_MAXLEN+1]; |
b8c5462f |
700 | uv_to_utf8(tmpbuf, (UV)c); |
701 | return is_utf8_cntrl(tmpbuf); |
702 | } |
703 | |
704 | bool |
705 | Perl_is_uni_graph(pTHX_ U32 c) |
706 | { |
ad391ad9 |
707 | U8 tmpbuf[UTF8_MAXLEN+1]; |
b8c5462f |
708 | uv_to_utf8(tmpbuf, (UV)c); |
709 | return is_utf8_graph(tmpbuf); |
710 | } |
711 | |
712 | bool |
864dbfa3 |
713 | Perl_is_uni_print(pTHX_ U32 c) |
a0ed51b3 |
714 | { |
ad391ad9 |
715 | U8 tmpbuf[UTF8_MAXLEN+1]; |
a0ed51b3 |
716 | uv_to_utf8(tmpbuf, (UV)c); |
717 | return is_utf8_print(tmpbuf); |
718 | } |
719 | |
b8c5462f |
720 | bool |
f248d071 |
721 | Perl_is_uni_punct(pTHX_ U32 c) |
b8c5462f |
722 | { |
ad391ad9 |
723 | U8 tmpbuf[UTF8_MAXLEN+1]; |
b8c5462f |
724 | uv_to_utf8(tmpbuf, (UV)c); |
725 | return is_utf8_punct(tmpbuf); |
726 | } |
727 | |
4d61ec05 |
728 | bool |
729 | Perl_is_uni_xdigit(pTHX_ U32 c) |
730 | { |
ad391ad9 |
731 | U8 tmpbuf[UTF8_MAXLEN+1]; |
4d61ec05 |
732 | uv_to_utf8(tmpbuf, (UV)c); |
733 | return is_utf8_xdigit(tmpbuf); |
734 | } |
735 | |
a0ed51b3 |
736 | U32 |
864dbfa3 |
737 | Perl_to_uni_upper(pTHX_ U32 c) |
a0ed51b3 |
738 | { |
ad391ad9 |
739 | U8 tmpbuf[UTF8_MAXLEN+1]; |
a0ed51b3 |
740 | uv_to_utf8(tmpbuf, (UV)c); |
741 | return to_utf8_upper(tmpbuf); |
742 | } |
743 | |
744 | U32 |
864dbfa3 |
745 | Perl_to_uni_title(pTHX_ U32 c) |
a0ed51b3 |
746 | { |
ad391ad9 |
747 | U8 tmpbuf[UTF8_MAXLEN+1]; |
a0ed51b3 |
748 | uv_to_utf8(tmpbuf, (UV)c); |
749 | return to_utf8_title(tmpbuf); |
750 | } |
751 | |
752 | U32 |
864dbfa3 |
753 | Perl_to_uni_lower(pTHX_ U32 c) |
a0ed51b3 |
754 | { |
ad391ad9 |
755 | U8 tmpbuf[UTF8_MAXLEN+1]; |
a0ed51b3 |
756 | uv_to_utf8(tmpbuf, (UV)c); |
757 | return to_utf8_lower(tmpbuf); |
758 | } |
759 | |
760 | /* for now these all assume no locale info available for Unicode > 255 */ |
761 | |
762 | bool |
864dbfa3 |
763 | Perl_is_uni_alnum_lc(pTHX_ U32 c) |
a0ed51b3 |
764 | { |
765 | return is_uni_alnum(c); /* XXX no locale support yet */ |
766 | } |
767 | |
768 | bool |
b8c5462f |
769 | Perl_is_uni_alnumc_lc(pTHX_ U32 c) |
770 | { |
771 | return is_uni_alnumc(c); /* XXX no locale support yet */ |
772 | } |
773 | |
774 | bool |
864dbfa3 |
775 | Perl_is_uni_idfirst_lc(pTHX_ U32 c) |
a0ed51b3 |
776 | { |
777 | return is_uni_idfirst(c); /* XXX no locale support yet */ |
778 | } |
779 | |
780 | bool |
864dbfa3 |
781 | Perl_is_uni_alpha_lc(pTHX_ U32 c) |
a0ed51b3 |
782 | { |
783 | return is_uni_alpha(c); /* XXX no locale support yet */ |
784 | } |
785 | |
786 | bool |
4d61ec05 |
787 | Perl_is_uni_ascii_lc(pTHX_ U32 c) |
788 | { |
789 | return is_uni_ascii(c); /* XXX no locale support yet */ |
790 | } |
791 | |
792 | bool |
864dbfa3 |
793 | Perl_is_uni_space_lc(pTHX_ U32 c) |
a0ed51b3 |
794 | { |
795 | return is_uni_space(c); /* XXX no locale support yet */ |
796 | } |
797 | |
798 | bool |
864dbfa3 |
799 | Perl_is_uni_digit_lc(pTHX_ U32 c) |
a0ed51b3 |
800 | { |
801 | return is_uni_digit(c); /* XXX no locale support yet */ |
802 | } |
803 | |
804 | bool |
864dbfa3 |
805 | Perl_is_uni_upper_lc(pTHX_ U32 c) |
a0ed51b3 |
806 | { |
807 | return is_uni_upper(c); /* XXX no locale support yet */ |
808 | } |
809 | |
810 | bool |
864dbfa3 |
811 | Perl_is_uni_lower_lc(pTHX_ U32 c) |
a0ed51b3 |
812 | { |
813 | return is_uni_lower(c); /* XXX no locale support yet */ |
814 | } |
815 | |
816 | bool |
b8c5462f |
817 | Perl_is_uni_cntrl_lc(pTHX_ U32 c) |
818 | { |
819 | return is_uni_cntrl(c); /* XXX no locale support yet */ |
820 | } |
821 | |
822 | bool |
823 | Perl_is_uni_graph_lc(pTHX_ U32 c) |
824 | { |
825 | return is_uni_graph(c); /* XXX no locale support yet */ |
826 | } |
827 | |
828 | bool |
864dbfa3 |
829 | Perl_is_uni_print_lc(pTHX_ U32 c) |
a0ed51b3 |
830 | { |
831 | return is_uni_print(c); /* XXX no locale support yet */ |
832 | } |
833 | |
b8c5462f |
834 | bool |
835 | Perl_is_uni_punct_lc(pTHX_ U32 c) |
836 | { |
837 | return is_uni_punct(c); /* XXX no locale support yet */ |
838 | } |
839 | |
4d61ec05 |
840 | bool |
841 | Perl_is_uni_xdigit_lc(pTHX_ U32 c) |
842 | { |
843 | return is_uni_xdigit(c); /* XXX no locale support yet */ |
844 | } |
845 | |
a0ed51b3 |
846 | U32 |
864dbfa3 |
847 | Perl_to_uni_upper_lc(pTHX_ U32 c) |
a0ed51b3 |
848 | { |
849 | return to_uni_upper(c); /* XXX no locale support yet */ |
850 | } |
851 | |
852 | U32 |
864dbfa3 |
853 | Perl_to_uni_title_lc(pTHX_ U32 c) |
a0ed51b3 |
854 | { |
855 | return to_uni_title(c); /* XXX no locale support yet */ |
856 | } |
857 | |
858 | U32 |
864dbfa3 |
859 | Perl_to_uni_lower_lc(pTHX_ U32 c) |
a0ed51b3 |
860 | { |
861 | return to_uni_lower(c); /* XXX no locale support yet */ |
862 | } |
863 | |
a0ed51b3 |
864 | bool |
864dbfa3 |
865 | Perl_is_utf8_alnum(pTHX_ U8 *p) |
a0ed51b3 |
866 | { |
386d01d6 |
867 | if (!is_utf8_char(p)) |
868 | return FALSE; |
a0ed51b3 |
869 | if (!PL_utf8_alnum) |
289d4f09 |
870 | /* NOTE: "IsWord", not "IsAlnum", since Alnum is a true |
871 | * descendant of isalnum(3), in other words, it doesn't |
872 | * contain the '_'. --jhi */ |
873 | PL_utf8_alnum = swash_init("utf8", "IsWord", &PL_sv_undef, 0, 0); |
a0ed51b3 |
874 | return swash_fetch(PL_utf8_alnum, p); |
875 | /* return *p == '_' || is_utf8_alpha(p) || is_utf8_digit(p); */ |
876 | #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */ |
877 | if (!PL_utf8_alnum) |
878 | PL_utf8_alnum = swash_init("utf8", "", |
879 | sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0); |
880 | return swash_fetch(PL_utf8_alnum, p); |
881 | #endif |
882 | } |
883 | |
884 | bool |
b8c5462f |
885 | Perl_is_utf8_alnumc(pTHX_ U8 *p) |
886 | { |
386d01d6 |
887 | if (!is_utf8_char(p)) |
888 | return FALSE; |
b8c5462f |
889 | if (!PL_utf8_alnum) |
890 | PL_utf8_alnum = swash_init("utf8", "IsAlnumC", &PL_sv_undef, 0, 0); |
891 | return swash_fetch(PL_utf8_alnum, p); |
892 | /* return is_utf8_alpha(p) || is_utf8_digit(p); */ |
893 | #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */ |
894 | if (!PL_utf8_alnum) |
895 | PL_utf8_alnum = swash_init("utf8", "", |
896 | sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0); |
897 | return swash_fetch(PL_utf8_alnum, p); |
898 | #endif |
899 | } |
900 | |
901 | bool |
864dbfa3 |
902 | Perl_is_utf8_idfirst(pTHX_ U8 *p) |
a0ed51b3 |
903 | { |
904 | return *p == '_' || is_utf8_alpha(p); |
905 | } |
906 | |
907 | bool |
864dbfa3 |
908 | Perl_is_utf8_alpha(pTHX_ U8 *p) |
a0ed51b3 |
909 | { |
386d01d6 |
910 | if (!is_utf8_char(p)) |
911 | return FALSE; |
a0ed51b3 |
912 | if (!PL_utf8_alpha) |
e24b16f9 |
913 | PL_utf8_alpha = swash_init("utf8", "IsAlpha", &PL_sv_undef, 0, 0); |
a0ed51b3 |
914 | return swash_fetch(PL_utf8_alpha, p); |
915 | } |
916 | |
917 | bool |
b8c5462f |
918 | Perl_is_utf8_ascii(pTHX_ U8 *p) |
919 | { |
386d01d6 |
920 | if (!is_utf8_char(p)) |
921 | return FALSE; |
b8c5462f |
922 | if (!PL_utf8_ascii) |
923 | PL_utf8_ascii = swash_init("utf8", "IsAscii", &PL_sv_undef, 0, 0); |
924 | return swash_fetch(PL_utf8_ascii, p); |
925 | } |
926 | |
927 | bool |
864dbfa3 |
928 | Perl_is_utf8_space(pTHX_ U8 *p) |
a0ed51b3 |
929 | { |
386d01d6 |
930 | if (!is_utf8_char(p)) |
931 | return FALSE; |
a0ed51b3 |
932 | if (!PL_utf8_space) |
3bec3564 |
933 | PL_utf8_space = swash_init("utf8", "IsSpacePerl", &PL_sv_undef, 0, 0); |
a0ed51b3 |
934 | return swash_fetch(PL_utf8_space, p); |
935 | } |
936 | |
937 | bool |
864dbfa3 |
938 | Perl_is_utf8_digit(pTHX_ U8 *p) |
a0ed51b3 |
939 | { |
386d01d6 |
940 | if (!is_utf8_char(p)) |
941 | return FALSE; |
a0ed51b3 |
942 | if (!PL_utf8_digit) |
e24b16f9 |
943 | PL_utf8_digit = swash_init("utf8", "IsDigit", &PL_sv_undef, 0, 0); |
a0ed51b3 |
944 | return swash_fetch(PL_utf8_digit, p); |
945 | } |
946 | |
947 | bool |
864dbfa3 |
948 | Perl_is_utf8_upper(pTHX_ U8 *p) |
a0ed51b3 |
949 | { |
386d01d6 |
950 | if (!is_utf8_char(p)) |
951 | return FALSE; |
a0ed51b3 |
952 | if (!PL_utf8_upper) |
e24b16f9 |
953 | PL_utf8_upper = swash_init("utf8", "IsUpper", &PL_sv_undef, 0, 0); |
a0ed51b3 |
954 | return swash_fetch(PL_utf8_upper, p); |
955 | } |
956 | |
957 | bool |
864dbfa3 |
958 | Perl_is_utf8_lower(pTHX_ U8 *p) |
a0ed51b3 |
959 | { |
386d01d6 |
960 | if (!is_utf8_char(p)) |
961 | return FALSE; |
a0ed51b3 |
962 | if (!PL_utf8_lower) |
e24b16f9 |
963 | PL_utf8_lower = swash_init("utf8", "IsLower", &PL_sv_undef, 0, 0); |
a0ed51b3 |
964 | return swash_fetch(PL_utf8_lower, p); |
965 | } |
966 | |
967 | bool |
b8c5462f |
968 | Perl_is_utf8_cntrl(pTHX_ U8 *p) |
969 | { |
386d01d6 |
970 | if (!is_utf8_char(p)) |
971 | return FALSE; |
b8c5462f |
972 | if (!PL_utf8_cntrl) |
973 | PL_utf8_cntrl = swash_init("utf8", "IsCntrl", &PL_sv_undef, 0, 0); |
974 | return swash_fetch(PL_utf8_cntrl, p); |
975 | } |
976 | |
977 | bool |
978 | Perl_is_utf8_graph(pTHX_ U8 *p) |
979 | { |
386d01d6 |
980 | if (!is_utf8_char(p)) |
981 | return FALSE; |
b8c5462f |
982 | if (!PL_utf8_graph) |
983 | PL_utf8_graph = swash_init("utf8", "IsGraph", &PL_sv_undef, 0, 0); |
984 | return swash_fetch(PL_utf8_graph, p); |
985 | } |
986 | |
987 | bool |
864dbfa3 |
988 | Perl_is_utf8_print(pTHX_ U8 *p) |
a0ed51b3 |
989 | { |
386d01d6 |
990 | if (!is_utf8_char(p)) |
991 | return FALSE; |
a0ed51b3 |
992 | if (!PL_utf8_print) |
e24b16f9 |
993 | PL_utf8_print = swash_init("utf8", "IsPrint", &PL_sv_undef, 0, 0); |
a0ed51b3 |
994 | return swash_fetch(PL_utf8_print, p); |
995 | } |
996 | |
997 | bool |
b8c5462f |
998 | Perl_is_utf8_punct(pTHX_ U8 *p) |
999 | { |
386d01d6 |
1000 | if (!is_utf8_char(p)) |
1001 | return FALSE; |
b8c5462f |
1002 | if (!PL_utf8_punct) |
1003 | PL_utf8_punct = swash_init("utf8", "IsPunct", &PL_sv_undef, 0, 0); |
1004 | return swash_fetch(PL_utf8_punct, p); |
1005 | } |
1006 | |
1007 | bool |
1008 | Perl_is_utf8_xdigit(pTHX_ U8 *p) |
1009 | { |
386d01d6 |
1010 | if (!is_utf8_char(p)) |
1011 | return FALSE; |
b8c5462f |
1012 | if (!PL_utf8_xdigit) |
1013 | PL_utf8_xdigit = swash_init("utf8", "IsXDigit", &PL_sv_undef, 0, 0); |
1014 | return swash_fetch(PL_utf8_xdigit, p); |
1015 | } |
1016 | |
1017 | bool |
864dbfa3 |
1018 | Perl_is_utf8_mark(pTHX_ U8 *p) |
a0ed51b3 |
1019 | { |
386d01d6 |
1020 | if (!is_utf8_char(p)) |
1021 | return FALSE; |
a0ed51b3 |
1022 | if (!PL_utf8_mark) |
e24b16f9 |
1023 | PL_utf8_mark = swash_init("utf8", "IsM", &PL_sv_undef, 0, 0); |
a0ed51b3 |
1024 | return swash_fetch(PL_utf8_mark, p); |
1025 | } |
1026 | |
2104c8d9 |
1027 | UV |
864dbfa3 |
1028 | Perl_to_utf8_upper(pTHX_ U8 *p) |
a0ed51b3 |
1029 | { |
1030 | UV uv; |
1031 | |
1032 | if (!PL_utf8_toupper) |
e24b16f9 |
1033 | PL_utf8_toupper = swash_init("utf8", "ToUpper", &PL_sv_undef, 4, 0); |
a0ed51b3 |
1034 | uv = swash_fetch(PL_utf8_toupper, p); |
756820e3 |
1035 | return uv ? uv : utf8_to_uv(p,UTF8_MAXLEN,0,0); |
a0ed51b3 |
1036 | } |
1037 | |
2104c8d9 |
1038 | UV |
864dbfa3 |
1039 | Perl_to_utf8_title(pTHX_ U8 *p) |
a0ed51b3 |
1040 | { |
1041 | UV uv; |
1042 | |
1043 | if (!PL_utf8_totitle) |
e24b16f9 |
1044 | PL_utf8_totitle = swash_init("utf8", "ToTitle", &PL_sv_undef, 4, 0); |
a0ed51b3 |
1045 | uv = swash_fetch(PL_utf8_totitle, p); |
756820e3 |
1046 | return uv ? uv : utf8_to_uv(p,UTF8_MAXLEN,0,0); |
a0ed51b3 |
1047 | } |
1048 | |
2104c8d9 |
1049 | UV |
864dbfa3 |
1050 | Perl_to_utf8_lower(pTHX_ U8 *p) |
a0ed51b3 |
1051 | { |
1052 | UV uv; |
1053 | |
1054 | if (!PL_utf8_tolower) |
e24b16f9 |
1055 | PL_utf8_tolower = swash_init("utf8", "ToLower", &PL_sv_undef, 4, 0); |
a0ed51b3 |
1056 | uv = swash_fetch(PL_utf8_tolower, p); |
756820e3 |
1057 | return uv ? uv : utf8_to_uv(p,UTF8_MAXLEN,0,0); |
a0ed51b3 |
1058 | } |
1059 | |
1060 | /* a "swash" is a swatch hash */ |
1061 | |
1062 | SV* |
864dbfa3 |
1063 | Perl_swash_init(pTHX_ char* pkg, char* name, SV *listsv, I32 minbits, I32 none) |
a0ed51b3 |
1064 | { |
1065 | SV* retval; |
1066 | char tmpbuf[256]; |
8e84507e |
1067 | dSP; |
ce3b816e |
1068 | |
1069 | if (!gv_stashpv(pkg, 0)) { /* demand load utf8 */ |
1070 | ENTER; |
1071 | Perl_load_module(aTHX_ PERL_LOADMOD_NOIMPORT, newSVpv(pkg,0), Nullsv); |
1072 | LEAVE; |
1073 | } |
1074 | SPAGAIN; |
a0ed51b3 |
1075 | PUSHSTACKi(PERLSI_MAGIC); |
1076 | PUSHMARK(SP); |
1077 | EXTEND(SP,5); |
1078 | PUSHs(sv_2mortal(newSVpvn(pkg, strlen(pkg)))); |
1079 | PUSHs(sv_2mortal(newSVpvn(name, strlen(name)))); |
1080 | PUSHs(listsv); |
1081 | PUSHs(sv_2mortal(newSViv(minbits))); |
1082 | PUSHs(sv_2mortal(newSViv(none))); |
1083 | PUTBACK; |
1084 | ENTER; |
1085 | SAVEI32(PL_hints); |
1086 | PL_hints = 0; |
1087 | save_re_context(); |
e24b16f9 |
1088 | if (PL_curcop == &PL_compiling) /* XXX ought to be handled by lex_start */ |
a0ed51b3 |
1089 | strncpy(tmpbuf, PL_tokenbuf, sizeof tmpbuf); |
864dbfa3 |
1090 | if (call_method("SWASHNEW", G_SCALAR)) |
8e84507e |
1091 | retval = newSVsv(*PL_stack_sp--); |
a0ed51b3 |
1092 | else |
e24b16f9 |
1093 | retval = &PL_sv_undef; |
a0ed51b3 |
1094 | LEAVE; |
1095 | POPSTACK; |
e24b16f9 |
1096 | if (PL_curcop == &PL_compiling) { |
a0ed51b3 |
1097 | strncpy(PL_tokenbuf, tmpbuf, sizeof tmpbuf); |
e24b16f9 |
1098 | PL_curcop->op_private = PL_hints; |
a0ed51b3 |
1099 | } |
1100 | if (!SvROK(retval) || SvTYPE(SvRV(retval)) != SVt_PVHV) |
cea2e8a9 |
1101 | Perl_croak(aTHX_ "SWASHNEW didn't return an HV ref"); |
a0ed51b3 |
1102 | return retval; |
1103 | } |
1104 | |
1105 | UV |
864dbfa3 |
1106 | Perl_swash_fetch(pTHX_ SV *sv, U8 *ptr) |
a0ed51b3 |
1107 | { |
1108 | HV* hv = (HV*)SvRV(sv); |
1109 | U32 klen = UTF8SKIP(ptr) - 1; |
1110 | U32 off = ptr[klen] & 127; /* NB: 64 bit always 0 when len > 1 */ |
1111 | STRLEN slen; |
1112 | STRLEN needents = (klen ? 64 : 128); |
dfe13c55 |
1113 | U8 *tmps; |
a0ed51b3 |
1114 | U32 bit; |
1115 | SV *retval; |
1116 | |
1117 | /* |
1118 | * This single-entry cache saves about 1/3 of the utf8 overhead in test |
1119 | * suite. (That is, only 7-8% overall over just a hash cache. Still, |
1120 | * it's nothing to sniff at.) Pity we usually come through at least |
1121 | * two function calls to get here... |
1122 | * |
1123 | * NB: this code assumes that swatches are never modified, once generated! |
1124 | */ |
1125 | |
1126 | if (hv == PL_last_swash_hv && |
1127 | klen == PL_last_swash_klen && |
12ae5dfc |
1128 | (!klen || memEQ((char *)ptr,(char *)PL_last_swash_key,klen)) ) |
a0ed51b3 |
1129 | { |
1130 | tmps = PL_last_swash_tmps; |
1131 | slen = PL_last_swash_slen; |
1132 | } |
1133 | else { |
1134 | /* Try our second-level swatch cache, kept in a hash. */ |
dfe13c55 |
1135 | SV** svp = hv_fetch(hv, (char*)ptr, klen, FALSE); |
a0ed51b3 |
1136 | |
1137 | /* If not cached, generate it via utf8::SWASHGET */ |
dfe13c55 |
1138 | if (!svp || !SvPOK(*svp) || !(tmps = (U8*)SvPV(*svp, slen))) { |
a0ed51b3 |
1139 | dSP; |
1140 | ENTER; |
1141 | SAVETMPS; |
1142 | save_re_context(); |
1143 | PUSHSTACKi(PERLSI_MAGIC); |
1144 | PUSHMARK(SP); |
1145 | EXTEND(SP,3); |
1146 | PUSHs((SV*)sv); |
756820e3 |
1147 | PUSHs(sv_2mortal(newSViv(utf8_to_uv(ptr, UTF8_MAXLEN, 0, 0) & ~(needents - 1)))); |
a0ed51b3 |
1148 | PUSHs(sv_2mortal(newSViv(needents))); |
1149 | PUTBACK; |
864dbfa3 |
1150 | if (call_method("SWASHGET", G_SCALAR)) |
8e84507e |
1151 | retval = newSVsv(*PL_stack_sp--); |
a0ed51b3 |
1152 | else |
e24b16f9 |
1153 | retval = &PL_sv_undef; |
a0ed51b3 |
1154 | POPSTACK; |
1155 | FREETMPS; |
1156 | LEAVE; |
e24b16f9 |
1157 | if (PL_curcop == &PL_compiling) |
1158 | PL_curcop->op_private = PL_hints; |
a0ed51b3 |
1159 | |
dfe13c55 |
1160 | svp = hv_store(hv, (char*)ptr, klen, retval, 0); |
a0ed51b3 |
1161 | |
dfe13c55 |
1162 | if (!svp || !(tmps = (U8*)SvPV(*svp, slen)) || slen < 8) |
cea2e8a9 |
1163 | Perl_croak(aTHX_ "SWASHGET didn't return result of proper length"); |
a0ed51b3 |
1164 | } |
1165 | |
1166 | PL_last_swash_hv = hv; |
1167 | PL_last_swash_klen = klen; |
1168 | PL_last_swash_tmps = tmps; |
1169 | PL_last_swash_slen = slen; |
1170 | if (klen) |
1171 | Copy(ptr, PL_last_swash_key, klen, U8); |
1172 | } |
1173 | |
9faf8d75 |
1174 | switch ((int)((slen << 3) / needents)) { |
a0ed51b3 |
1175 | case 1: |
1176 | bit = 1 << (off & 7); |
1177 | off >>= 3; |
1178 | return (tmps[off] & bit) != 0; |
1179 | case 8: |
1180 | return tmps[off]; |
1181 | case 16: |
1182 | off <<= 1; |
1183 | return (tmps[off] << 8) + tmps[off + 1] ; |
1184 | case 32: |
1185 | off <<= 2; |
1186 | return (tmps[off] << 24) + (tmps[off+1] << 16) + (tmps[off+2] << 8) + tmps[off + 3] ; |
1187 | } |
cea2e8a9 |
1188 | Perl_croak(aTHX_ "panic: swash_fetch"); |
a0ed51b3 |
1189 | return 0; |
1190 | } |