Commit | Line | Data |
a0ed51b3 |
1 | /* utf8.c |
2 | * |
bc89e66f |
3 | * Copyright (c) 1998-2001, 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 | |
eebe1485 |
29 | /* |
9041c2e3 |
30 | =for apidoc A|U8*|uvuni_to_utf8|U8 *d|UV uv |
eebe1485 |
31 | |
32 | Adds the UTF8 representation of the Unicode codepoint C<uv> to the end |
33 | of the string C<d>; C<d> should be have at least C<UTF8_MAXLEN+1> free |
34 | bytes available. The return value is the pointer to the byte after the |
9041c2e3 |
35 | end of the new character. In other words, |
eebe1485 |
36 | |
9041c2e3 |
37 | d = uvuni_to_utf8(d, uv); |
eebe1485 |
38 | |
39 | is the recommended Unicode-aware way of saying |
40 | |
41 | *(d++) = uv; |
42 | |
43 | =cut |
44 | */ |
45 | |
dfe13c55 |
46 | U8 * |
9041c2e3 |
47 | Perl_uvuni_to_utf8(pTHX_ U8 *d, UV uv) |
a0ed51b3 |
48 | { |
49 | if (uv < 0x80) { |
50 | *d++ = uv; |
51 | return d; |
52 | } |
53 | if (uv < 0x800) { |
54 | *d++ = (( uv >> 6) | 0xc0); |
55 | *d++ = (( uv & 0x3f) | 0x80); |
56 | return d; |
57 | } |
58 | if (uv < 0x10000) { |
59 | *d++ = (( uv >> 12) | 0xe0); |
60 | *d++ = (((uv >> 6) & 0x3f) | 0x80); |
61 | *d++ = (( uv & 0x3f) | 0x80); |
62 | return d; |
63 | } |
64 | if (uv < 0x200000) { |
65 | *d++ = (( uv >> 18) | 0xf0); |
66 | *d++ = (((uv >> 12) & 0x3f) | 0x80); |
67 | *d++ = (((uv >> 6) & 0x3f) | 0x80); |
68 | *d++ = (( uv & 0x3f) | 0x80); |
69 | return d; |
70 | } |
71 | if (uv < 0x4000000) { |
72 | *d++ = (( uv >> 24) | 0xf8); |
73 | *d++ = (((uv >> 18) & 0x3f) | 0x80); |
74 | *d++ = (((uv >> 12) & 0x3f) | 0x80); |
75 | *d++ = (((uv >> 6) & 0x3f) | 0x80); |
76 | *d++ = (( uv & 0x3f) | 0x80); |
77 | return d; |
78 | } |
79 | if (uv < 0x80000000) { |
80 | *d++ = (( uv >> 30) | 0xfc); |
81 | *d++ = (((uv >> 24) & 0x3f) | 0x80); |
82 | *d++ = (((uv >> 18) & 0x3f) | 0x80); |
83 | *d++ = (((uv >> 12) & 0x3f) | 0x80); |
84 | *d++ = (((uv >> 6) & 0x3f) | 0x80); |
85 | *d++ = (( uv & 0x3f) | 0x80); |
86 | return d; |
87 | } |
6b8eaf93 |
88 | #ifdef HAS_QUAD |
d7578b48 |
89 | if (uv < UTF8_QUAD_MAX) |
a0ed51b3 |
90 | #endif |
91 | { |
92 | *d++ = 0xfe; /* Can't match U+FEFF! */ |
93 | *d++ = (((uv >> 30) & 0x3f) | 0x80); |
94 | *d++ = (((uv >> 24) & 0x3f) | 0x80); |
95 | *d++ = (((uv >> 18) & 0x3f) | 0x80); |
96 | *d++ = (((uv >> 12) & 0x3f) | 0x80); |
97 | *d++ = (((uv >> 6) & 0x3f) | 0x80); |
98 | *d++ = (( uv & 0x3f) | 0x80); |
99 | return d; |
100 | } |
6b8eaf93 |
101 | #ifdef HAS_QUAD |
a0ed51b3 |
102 | { |
103 | *d++ = 0xff; /* Can't match U+FFFE! */ |
3c77ea2b |
104 | *d++ = 0x80; /* 6 Reserved bits */ |
105 | *d++ = (((uv >> 60) & 0x0f) | 0x80); /* 2 Reserved bits */ |
106 | *d++ = (((uv >> 54) & 0x3f) | 0x80); |
107 | *d++ = (((uv >> 48) & 0x3f) | 0x80); |
108 | *d++ = (((uv >> 42) & 0x3f) | 0x80); |
a0ed51b3 |
109 | *d++ = (((uv >> 36) & 0x3f) | 0x80); |
110 | *d++ = (((uv >> 30) & 0x3f) | 0x80); |
111 | *d++ = (((uv >> 24) & 0x3f) | 0x80); |
112 | *d++ = (((uv >> 18) & 0x3f) | 0x80); |
113 | *d++ = (((uv >> 12) & 0x3f) | 0x80); |
114 | *d++ = (((uv >> 6) & 0x3f) | 0x80); |
115 | *d++ = (( uv & 0x3f) | 0x80); |
116 | return d; |
117 | } |
118 | #endif |
119 | } |
120 | |
eebe1485 |
121 | /* |
9041c2e3 |
122 | =for apidoc A|U8*|uvchr_to_utf8|U8 *d|UV uv |
123 | |
124 | Adds the UTF8 representation of the Native codepoint C<uv> to the end |
125 | of the string C<d>; C<d> should be have at least C<UTF8_MAXLEN+1> free |
126 | bytes available. The return value is the pointer to the byte after the |
127 | end of the new character. In other words, |
128 | |
129 | d = uvchr_to_utf8(d, uv); |
130 | |
131 | is the recommended wide native character-aware way of saying |
132 | |
133 | *(d++) = uv; |
134 | |
135 | =cut |
136 | */ |
137 | |
138 | U8 * |
139 | Perl_uvchr_to_utf8(pTHX_ U8 *d, UV uv) |
140 | { |
141 | if (uv < 0x100) |
142 | uv = NATIVE_TO_ASCII(uv); |
143 | return Perl_uvuni_to_utf8(aTHX_ d, uv); |
144 | } |
145 | |
146 | |
147 | /* |
eebe1485 |
148 | =for apidoc A|STRLEN|is_utf8_char|U8 *s |
149 | |
5da9da9e |
150 | Tests if some arbitrary number of bytes begins in a valid UTF-8 |
151 | character. Note that an ASCII character is a valid UTF-8 character. |
152 | The actual number of bytes in the UTF-8 character will be returned if |
153 | it is valid, otherwise 0. |
9041c2e3 |
154 | |
5da9da9e |
155 | =cut */ |
067a85ef |
156 | STRLEN |
386d01d6 |
157 | Perl_is_utf8_char(pTHX_ U8 *s) |
158 | { |
159 | U8 u = *s; |
067a85ef |
160 | STRLEN slen, len; |
161 | UV uv, ouv; |
386d01d6 |
162 | |
60006e79 |
163 | if (UTF8_IS_ASCII(u)) |
386d01d6 |
164 | return 1; |
165 | |
60006e79 |
166 | if (!UTF8_IS_START(u)) |
386d01d6 |
167 | return 0; |
168 | |
9f07fdcd |
169 | len = UTF8SKIP(s); |
386d01d6 |
170 | |
60006e79 |
171 | if (len < 2 || !UTF8_IS_CONTINUATION(s[1])) |
067a85ef |
172 | return 0; |
173 | |
386d01d6 |
174 | slen = len - 1; |
175 | s++; |
067a85ef |
176 | uv = u; |
177 | ouv = uv; |
386d01d6 |
178 | while (slen--) { |
60006e79 |
179 | if (!UTF8_IS_CONTINUATION(*s)) |
386d01d6 |
180 | return 0; |
8850bf83 |
181 | uv = UTF8_ACCUMULATE(uv, *s); |
067a85ef |
182 | if (uv < ouv) |
183 | return 0; |
184 | ouv = uv; |
386d01d6 |
185 | s++; |
186 | } |
067a85ef |
187 | |
5bbb0b5a |
188 | if (UNISKIP(uv) < len) |
067a85ef |
189 | return 0; |
190 | |
386d01d6 |
191 | return len; |
192 | } |
193 | |
6662521e |
194 | /* |
eebe1485 |
195 | =for apidoc A|bool|is_utf8_string|U8 *s|STRLEN len |
6662521e |
196 | |
5da9da9e |
197 | Returns true if first C<len> bytes of the given string form a valid UTF8 |
198 | string, false otherwise. Note that 'a valid UTF8 string' does not mean |
199 | 'a string that contains UTF8' because a valid ASCII string is a valid |
200 | UTF8 string. |
6662521e |
201 | |
202 | =cut |
203 | */ |
204 | |
8e84507e |
205 | bool |
6662521e |
206 | Perl_is_utf8_string(pTHX_ U8 *s, STRLEN len) |
207 | { |
067a85ef |
208 | U8* x = s; |
1aa99e6b |
209 | U8* send; |
067a85ef |
210 | STRLEN c; |
211 | |
1aa99e6b |
212 | if (!len) |
6cd5fe39 |
213 | len = strlen((char *)s); |
1aa99e6b |
214 | send = s + len; |
215 | |
6662521e |
216 | while (x < send) { |
217 | c = is_utf8_char(x); |
067a85ef |
218 | if (!c) |
219 | return FALSE; |
6662521e |
220 | x += c; |
6662521e |
221 | } |
60006e79 |
222 | if (x != send) |
223 | return FALSE; |
067a85ef |
224 | |
225 | return TRUE; |
6662521e |
226 | } |
227 | |
67e989fb |
228 | /* |
9041c2e3 |
229 | =for apidoc A|UV|utf8n_to_uvuni|U8 *s|STRLEN curlen|STRLEN *retlen|U32 flags |
67e989fb |
230 | |
9041c2e3 |
231 | Bottom level UTF-8 decode routine. |
232 | Returns the unicode code point value of the first character in the string C<s> |
ba210ebe |
233 | which is assumed to be in UTF8 encoding and no longer than C<curlen>; |
7df053ec |
234 | C<retlen> will be set to the length, in bytes, of that character. |
67e989fb |
235 | |
236 | If C<s> does not point to a well-formed UTF8 character, the behaviour |
dcad2880 |
237 | is dependent on the value of C<flags>: if it contains UTF8_CHECK_ONLY, |
238 | it is assumed that the caller will raise a warning, and this function |
28d3d195 |
239 | will silently just set C<retlen> to C<-1> and return zero. If the |
240 | C<flags> does not contain UTF8_CHECK_ONLY, warnings about |
241 | malformations will be given, C<retlen> will be set to the expected |
242 | length of the UTF-8 character in bytes, and zero will be returned. |
243 | |
244 | The C<flags> can also contain various flags to allow deviations from |
245 | the strict UTF-8 encoding (see F<utf8.h>). |
67e989fb |
246 | |
9041c2e3 |
247 | Most code should use utf8_to_uvchr() rather than call this directly. |
248 | |
dcad2880 |
249 | =cut */ |
67e989fb |
250 | |
a0ed51b3 |
251 | UV |
9041c2e3 |
252 | Perl_utf8n_to_uvuni(pTHX_ U8* s, STRLEN curlen, STRLEN* retlen, U32 flags) |
a0ed51b3 |
253 | { |
ba210ebe |
254 | UV uv = *s, ouv; |
255 | STRLEN len = 1; |
256 | bool dowarn = ckWARN_d(WARN_UTF8); |
257 | STRLEN expectlen = 0; |
a0dbb045 |
258 | U32 warning = 0; |
259 | |
260 | /* This list is a superset of the UTF8_ALLOW_XXX. */ |
261 | |
262 | #define UTF8_WARN_EMPTY 1 |
263 | #define UTF8_WARN_CONTINUATION 2 |
264 | #define UTF8_WARN_NON_CONTINUATION 3 |
265 | #define UTF8_WARN_FE_FF 4 |
266 | #define UTF8_WARN_SHORT 5 |
267 | #define UTF8_WARN_OVERFLOW 6 |
268 | #define UTF8_WARN_SURROGATE 7 |
269 | #define UTF8_WARN_BOM 8 |
270 | #define UTF8_WARN_LONG 9 |
271 | #define UTF8_WARN_FFFF 10 |
272 | |
273 | if (curlen == 0 && |
274 | !(flags & UTF8_ALLOW_EMPTY)) { |
275 | warning = UTF8_WARN_EMPTY; |
0c443dc2 |
276 | goto malformed; |
277 | } |
278 | |
421a8bf2 |
279 | if (UTF8_IS_ASCII(uv)) { |
a0ed51b3 |
280 | if (retlen) |
281 | *retlen = 1; |
9041c2e3 |
282 | return (UV) (*s); |
a0ed51b3 |
283 | } |
67e989fb |
284 | |
421a8bf2 |
285 | if (UTF8_IS_CONTINUATION(uv) && |
fcc8fcf6 |
286 | !(flags & UTF8_ALLOW_CONTINUATION)) { |
a0dbb045 |
287 | warning = UTF8_WARN_CONTINUATION; |
ba210ebe |
288 | goto malformed; |
289 | } |
290 | |
421a8bf2 |
291 | if (UTF8_IS_START(uv) && curlen > 1 && !UTF8_IS_CONTINUATION(s[1]) && |
fcc8fcf6 |
292 | !(flags & UTF8_ALLOW_NON_CONTINUATION)) { |
a0dbb045 |
293 | warning = UTF8_WARN_NON_CONTINUATION; |
ba210ebe |
294 | goto malformed; |
295 | } |
9041c2e3 |
296 | |
fcc8fcf6 |
297 | if ((uv == 0xfe || uv == 0xff) && |
298 | !(flags & UTF8_ALLOW_FE_FF)) { |
a0dbb045 |
299 | warning = UTF8_WARN_FE_FF; |
ba210ebe |
300 | goto malformed; |
a0ed51b3 |
301 | } |
fcc8fcf6 |
302 | |
ba210ebe |
303 | if (!(uv & 0x20)) { len = 2; uv &= 0x1f; } |
304 | else if (!(uv & 0x10)) { len = 3; uv &= 0x0f; } |
305 | else if (!(uv & 0x08)) { len = 4; uv &= 0x07; } |
306 | else if (!(uv & 0x04)) { len = 5; uv &= 0x03; } |
307 | else if (!(uv & 0x02)) { len = 6; uv &= 0x01; } |
308 | else if (!(uv & 0x01)) { len = 7; uv = 0; } |
3c77ea2b |
309 | else { len = 13; uv = 0; } /* whoa! */ |
fcc8fcf6 |
310 | |
a0ed51b3 |
311 | if (retlen) |
312 | *retlen = len; |
9041c2e3 |
313 | |
ba210ebe |
314 | expectlen = len; |
315 | |
fcc8fcf6 |
316 | if ((curlen < expectlen) && |
317 | !(flags & UTF8_ALLOW_SHORT)) { |
a0dbb045 |
318 | warning = UTF8_WARN_SHORT; |
ba210ebe |
319 | goto malformed; |
320 | } |
321 | |
322 | len--; |
a0ed51b3 |
323 | s++; |
ba210ebe |
324 | ouv = uv; |
325 | |
a0ed51b3 |
326 | while (len--) { |
421a8bf2 |
327 | if (!UTF8_IS_CONTINUATION(*s) && |
328 | !(flags & UTF8_ALLOW_NON_CONTINUATION)) { |
a0dbb045 |
329 | s--; |
330 | warning = UTF8_WARN_NON_CONTINUATION; |
ba210ebe |
331 | goto malformed; |
a0ed51b3 |
332 | } |
333 | else |
8850bf83 |
334 | uv = UTF8_ACCUMULATE(uv, *s); |
a0dbb045 |
335 | if (!(uv > ouv)) { |
336 | /* These cannot be allowed. */ |
337 | if (uv == ouv) { |
338 | if (!(flags & UTF8_ALLOW_LONG)) { |
339 | warning = UTF8_WARN_LONG; |
340 | goto malformed; |
341 | } |
342 | } |
343 | else { /* uv < ouv */ |
344 | /* This cannot be allowed. */ |
345 | warning = UTF8_WARN_OVERFLOW; |
346 | goto malformed; |
347 | } |
ba210ebe |
348 | } |
349 | s++; |
350 | ouv = uv; |
351 | } |
352 | |
421a8bf2 |
353 | if (UNICODE_IS_SURROGATE(uv) && |
fcc8fcf6 |
354 | !(flags & UTF8_ALLOW_SURROGATE)) { |
a0dbb045 |
355 | warning = UTF8_WARN_SURROGATE; |
ba210ebe |
356 | goto malformed; |
421a8bf2 |
357 | } else if (UNICODE_IS_BYTE_ORDER_MARK(uv) && |
fcc8fcf6 |
358 | !(flags & UTF8_ALLOW_BOM)) { |
a0dbb045 |
359 | warning = UTF8_WARN_BOM; |
ba210ebe |
360 | goto malformed; |
fcc8fcf6 |
361 | } else if ((expectlen > UNISKIP(uv)) && |
362 | !(flags & UTF8_ALLOW_LONG)) { |
a0dbb045 |
363 | warning = UTF8_WARN_LONG; |
ba210ebe |
364 | goto malformed; |
421a8bf2 |
365 | } else if (UNICODE_IS_ILLEGAL(uv) && |
a9917092 |
366 | !(flags & UTF8_ALLOW_FFFF)) { |
a0dbb045 |
367 | warning = UTF8_WARN_FFFF; |
a9917092 |
368 | goto malformed; |
a0ed51b3 |
369 | } |
ba210ebe |
370 | |
a0ed51b3 |
371 | return uv; |
ba210ebe |
372 | |
373 | malformed: |
374 | |
fcc8fcf6 |
375 | if (flags & UTF8_CHECK_ONLY) { |
ba210ebe |
376 | if (retlen) |
cc366d4b |
377 | *retlen = -1; |
ba210ebe |
378 | return 0; |
379 | } |
380 | |
a0dbb045 |
381 | if (dowarn) { |
382 | SV* sv = sv_2mortal(newSVpv("Malformed UTF-8 character ", 0)); |
383 | |
384 | switch (warning) { |
385 | case 0: /* Intentionally empty. */ break; |
386 | case UTF8_WARN_EMPTY: |
387 | Perl_sv_catpvf(aTHX_ sv, "(empty string)"); |
388 | break; |
389 | case UTF8_WARN_CONTINUATION: |
390 | Perl_sv_catpvf(aTHX_ sv, "(unexpected continuation byte 0x%02"UVxf")", uv); |
391 | break; |
392 | case UTF8_WARN_NON_CONTINUATION: |
393 | Perl_sv_catpvf(aTHX_ sv, "(unexpected non-continuation byte 0x%02"UVxf" after start byte 0x%02"UVxf")", |
394 | (UV)s[1], uv); |
395 | break; |
396 | case UTF8_WARN_FE_FF: |
397 | Perl_sv_catpvf(aTHX_ sv, "(byte 0x%02"UVxf")", uv); |
398 | break; |
399 | case UTF8_WARN_SHORT: |
400 | Perl_sv_catpvf(aTHX_ sv, "(%d byte%s, need %d)", |
401 | curlen, curlen == 1 ? "" : "s", expectlen); |
402 | break; |
403 | case UTF8_WARN_OVERFLOW: |
404 | Perl_sv_catpvf(aTHX_ sv, "(overflow at 0x%"UVxf", byte 0x%02x)", |
405 | ouv, *s); |
406 | break; |
407 | case UTF8_WARN_SURROGATE: |
408 | Perl_sv_catpvf(aTHX_ sv, "(UTF-16 surrogate 0x%04"UVxf")", uv); |
409 | break; |
410 | case UTF8_WARN_BOM: |
411 | Perl_sv_catpvf(aTHX_ sv, "(byte order mark 0x%04"UVxf")", uv); |
412 | break; |
413 | case UTF8_WARN_LONG: |
414 | Perl_sv_catpvf(aTHX_ sv, "(%d byte%s, need %d)", |
415 | expectlen, expectlen == 1 ? "": "s", UNISKIP(uv)); |
416 | break; |
417 | case UTF8_WARN_FFFF: |
418 | Perl_sv_catpvf(aTHX_ sv, "(character 0x%04"UVxf")", uv); |
419 | break; |
420 | default: |
421 | Perl_sv_catpvf(aTHX_ sv, "(unknown reason)"); |
422 | break; |
423 | } |
424 | |
425 | if (warning) { |
426 | char *s = SvPVX(sv); |
427 | |
428 | if (PL_op) |
429 | Perl_warner(aTHX_ WARN_UTF8, |
430 | "%s in %s", s, PL_op_desc[PL_op->op_type]); |
431 | else |
432 | Perl_warner(aTHX_ WARN_UTF8, "%s", s); |
433 | } |
434 | } |
435 | |
ba210ebe |
436 | if (retlen) |
28d3d195 |
437 | *retlen = expectlen ? expectlen : len; |
ba210ebe |
438 | |
28d3d195 |
439 | return 0; |
a0ed51b3 |
440 | } |
441 | |
8e84507e |
442 | /* |
9041c2e3 |
443 | =for apidoc A|U8* s|utf8n_to_uvchr|STRLEN curlen, STRLEN *retlen, U32 flags |
8e84507e |
444 | |
9041c2e3 |
445 | Returns the native character value of the first character in the string C<s> |
8e84507e |
446 | which is assumed to be in UTF8 encoding; C<retlen> will be set to the |
7df053ec |
447 | length, in bytes, of that character. |
8e84507e |
448 | |
9041c2e3 |
449 | Allows length and flags to be passed to low level routine. |
450 | |
451 | =cut |
452 | */ |
453 | |
454 | UV |
455 | Perl_utf8n_to_uvchr(pTHX_ U8* s, STRLEN curlen, STRLEN* retlen, U32 flags) |
456 | { |
457 | UV uv = Perl_utf8n_to_uvuni(aTHX_ s, curlen, retlen, flags); |
458 | if (uv < 0x100) |
459 | return (UV) ASCII_TO_NATIVE(uv); |
460 | return uv; |
461 | } |
462 | |
463 | /* |
464 | =for apidoc A|U8* s|utf8_to_uvchr|STRLEN *retlen |
465 | |
466 | Returns the native character value of the first character in the string C<s> |
467 | which is assumed to be in UTF8 encoding; C<retlen> will be set to the |
468 | length, in bytes, of that character. |
469 | |
470 | If C<s> does not point to a well-formed UTF8 character, zero is |
471 | returned and retlen is set, if possible, to -1. |
472 | |
473 | =cut |
474 | */ |
475 | |
476 | UV |
477 | Perl_utf8_to_uvchr(pTHX_ U8* s, STRLEN* retlen) |
478 | { |
479 | return Perl_utf8n_to_uvchr(aTHX_ s, UTF8_MAXLEN, retlen, 0); |
480 | } |
481 | |
482 | /* |
483 | =for apidoc A|U8* s|utf8_to_uvuni|STRLEN *retlen |
484 | |
485 | Returns the Unicode code point of the first character in the string C<s> |
486 | which is assumed to be in UTF8 encoding; C<retlen> will be set to the |
487 | length, in bytes, of that character. |
488 | |
489 | This function should only be used when returned UV is considered |
490 | an index into the Unicode semantic tables (e.g. swashes). |
491 | |
ba210ebe |
492 | If C<s> does not point to a well-formed UTF8 character, zero is |
493 | returned and retlen is set, if possible, to -1. |
8e84507e |
494 | |
495 | =cut |
496 | */ |
497 | |
498 | UV |
9041c2e3 |
499 | Perl_utf8_to_uvuni(pTHX_ U8* s, STRLEN* retlen) |
8e84507e |
500 | { |
9041c2e3 |
501 | /* Call the low level routine asking for checks */ |
502 | return Perl_utf8n_to_uvuni(aTHX_ s, UTF8_MAXLEN, retlen, 0); |
8e84507e |
503 | } |
504 | |
b76347f2 |
505 | /* |
eebe1485 |
506 | =for apidoc A|STRLEN|utf8_length|U8* s|U8 *e |
b76347f2 |
507 | |
508 | Return the length of the UTF-8 char encoded string C<s> in characters. |
02eb7b47 |
509 | Stops at C<e> (inclusive). If C<e E<lt> s> or if the scan would end |
510 | up past C<e>, croaks. |
b76347f2 |
511 | |
512 | =cut |
513 | */ |
514 | |
515 | STRLEN |
516 | Perl_utf8_length(pTHX_ U8* s, U8* e) |
517 | { |
518 | STRLEN len = 0; |
519 | |
8850bf83 |
520 | /* Note: cannot use UTF8_IS_...() too eagerly here since e.g. |
521 | * the bitops (especially ~) can create illegal UTF-8. |
522 | * In other words: in Perl UTF-8 is not just for Unicode. */ |
523 | |
b76347f2 |
524 | if (e < s) |
02eb7b47 |
525 | Perl_croak(aTHX_ "panic: utf8_length: unexpected end"); |
b76347f2 |
526 | while (s < e) { |
02eb7b47 |
527 | U8 t = UTF8SKIP(s); |
b76347f2 |
528 | |
529 | if (e - s < t) |
02eb7b47 |
530 | Perl_croak(aTHX_ "panic: utf8_length: unaligned end"); |
b76347f2 |
531 | s += t; |
532 | len++; |
533 | } |
534 | |
535 | return len; |
536 | } |
537 | |
b06226ff |
538 | /* |
eebe1485 |
539 | =for apidoc A|IV|utf8_distance|U8 *a|U8 *b |
b06226ff |
540 | |
541 | Returns the number of UTF8 characters between the UTF-8 pointers C<a> |
542 | and C<b>. |
543 | |
544 | WARNING: use only if you *know* that the pointers point inside the |
545 | same UTF-8 buffer. |
546 | |
547 | =cut */ |
a0ed51b3 |
548 | |
02eb7b47 |
549 | IV |
864dbfa3 |
550 | Perl_utf8_distance(pTHX_ U8 *a, U8 *b) |
a0ed51b3 |
551 | { |
02eb7b47 |
552 | IV off = 0; |
553 | |
8850bf83 |
554 | /* Note: cannot use UTF8_IS_...() too eagerly here since e.g. |
555 | * the bitops (especially ~) can create illegal UTF-8. |
556 | * In other words: in Perl UTF-8 is not just for Unicode. */ |
557 | |
a0ed51b3 |
558 | if (a < b) { |
559 | while (a < b) { |
02eb7b47 |
560 | U8 c = UTF8SKIP(a); |
561 | |
562 | if (b - a < c) |
563 | Perl_croak(aTHX_ "panic: utf8_distance: unaligned end"); |
564 | a += c; |
a0ed51b3 |
565 | off--; |
566 | } |
567 | } |
568 | else { |
569 | while (b < a) { |
02eb7b47 |
570 | U8 c = UTF8SKIP(b); |
571 | |
572 | if (a - b < c) |
573 | Perl_croak(aTHX_ "panic: utf8_distance: unaligned end"); |
574 | b += c; |
a0ed51b3 |
575 | off++; |
576 | } |
577 | } |
02eb7b47 |
578 | |
a0ed51b3 |
579 | return off; |
580 | } |
581 | |
b06226ff |
582 | /* |
eebe1485 |
583 | =for apidoc A|U8*|utf8_hop|U8 *s|I32 off |
b06226ff |
584 | |
8850bf83 |
585 | Return the UTF-8 pointer C<s> displaced by C<off> characters, either |
586 | forward or backward. |
b06226ff |
587 | |
588 | WARNING: do not use the following unless you *know* C<off> is within |
8850bf83 |
589 | the UTF-8 data pointed to by C<s> *and* that on entry C<s> is aligned |
590 | on the first byte of character or just after the last byte of a character. |
b06226ff |
591 | |
592 | =cut */ |
a0ed51b3 |
593 | |
594 | U8 * |
864dbfa3 |
595 | Perl_utf8_hop(pTHX_ U8 *s, I32 off) |
a0ed51b3 |
596 | { |
8850bf83 |
597 | /* Note: cannot use UTF8_IS_...() too eagerly here since e.g |
598 | * the bitops (especially ~) can create illegal UTF-8. |
599 | * In other words: in Perl UTF-8 is not just for Unicode. */ |
600 | |
a0ed51b3 |
601 | if (off >= 0) { |
602 | while (off--) |
603 | s += UTF8SKIP(s); |
604 | } |
605 | else { |
606 | while (off++) { |
607 | s--; |
8850bf83 |
608 | while (UTF8_IS_CONTINUATION(*s)) |
609 | s--; |
a0ed51b3 |
610 | } |
611 | } |
612 | return s; |
613 | } |
614 | |
6940069f |
615 | /* |
eebe1485 |
616 | =for apidoc A|U8 *|utf8_to_bytes|U8 *s|STRLEN *len |
6940069f |
617 | |
246fae53 |
618 | Converts a string C<s> of length C<len> from UTF8 into byte encoding. |
619 | Unlike C<bytes_to_utf8>, this over-writes the original string, and |
620 | updates len to contain the new length. |
67e989fb |
621 | Returns zero on failure, setting C<len> to -1. |
6940069f |
622 | |
623 | =cut |
624 | */ |
625 | |
626 | U8 * |
246fae53 |
627 | Perl_utf8_to_bytes(pTHX_ U8* s, STRLEN *len) |
6940069f |
628 | { |
6940069f |
629 | U8 *send; |
630 | U8 *d; |
dcad2880 |
631 | U8 *save = s; |
246fae53 |
632 | |
633 | /* ensure valid UTF8 and chars < 256 before updating string */ |
dcad2880 |
634 | for (send = s + *len; s < send; ) { |
635 | U8 c = *s++; |
636 | |
9f9ab905 |
637 | if (c >= 0x80 && |
dcad2880 |
638 | ((s >= send) || |
639 | ((*s++ & 0xc0) != 0x80) || ((c & 0xfe) != 0xc2))) { |
640 | *len = -1; |
641 | return 0; |
642 | } |
246fae53 |
643 | } |
dcad2880 |
644 | |
645 | d = s = save; |
6940069f |
646 | while (s < send) { |
ed646e6e |
647 | STRLEN ulen; |
9041c2e3 |
648 | *d++ = (U8)utf8_to_uvchr(s, &ulen); |
ed646e6e |
649 | s += ulen; |
6940069f |
650 | } |
651 | *d = '\0'; |
246fae53 |
652 | *len = d - save; |
6940069f |
653 | return save; |
654 | } |
655 | |
656 | /* |
f9a63242 |
657 | =for apidoc A|U8 *|bytes_from_utf8|U8 *s|STRLEN *len|bool *is_utf8 |
658 | |
659 | Converts a string C<s> of length C<len> from UTF8 into byte encoding. |
660 | Unlike <utf8_to_bytes> but like C<bytes_to_utf8>, returns a pointer to |
ef9edfd0 |
661 | the newly-created string, and updates C<len> to contain the new |
662 | length. Returns the original string if no conversion occurs, C<len> |
663 | is unchanged. Do nothing if C<is_utf8> points to 0. Sets C<is_utf8> to |
664 | 0 if C<s> is converted or contains all 7bit characters. |
f9a63242 |
665 | |
666 | =cut */ |
667 | |
668 | U8 * |
669 | Perl_bytes_from_utf8(pTHX_ U8* s, STRLEN *len, bool *is_utf8) |
670 | { |
671 | U8 *send; |
672 | U8 *d; |
673 | U8 *start = s; |
674 | I32 count = 0; |
675 | |
676 | if (!*is_utf8) |
677 | return start; |
678 | |
ef9edfd0 |
679 | /* ensure valid UTF8 and chars < 256 before converting string */ |
f9a63242 |
680 | for (send = s + *len; s < send;) { |
681 | U8 c = *s++; |
682 | if (!UTF8_IS_ASCII(c)) { |
683 | if (UTF8_IS_CONTINUATION(c) || s >= send || |
df84a23b |
684 | !UTF8_IS_CONTINUATION(*s) || UTF8_IS_DOWNGRADEABLE_START(c)) |
f9a63242 |
685 | return start; |
686 | s++, count++; |
687 | } |
688 | } |
689 | |
690 | *is_utf8 = 0; |
691 | |
692 | if (!count) |
693 | return start; |
694 | |
695 | Newz(801, d, (*len) - count + 1, U8); |
ef9edfd0 |
696 | s = start; start = d; |
f9a63242 |
697 | while (s < send) { |
698 | U8 c = *s++; |
90f44359 |
699 | |
f9a63242 |
700 | if (UTF8_IS_ASCII(c)) |
701 | *d++ = c; |
702 | else |
90f44359 |
703 | *d++ = UTF8_ACCUMULATE(c, *s++); |
f9a63242 |
704 | } |
705 | *d = '\0'; |
706 | *len = d - start; |
707 | return start; |
708 | } |
709 | |
710 | /* |
eebe1485 |
711 | =for apidoc A|U8 *|bytes_to_utf8|U8 *s|STRLEN *len |
6940069f |
712 | |
713 | Converts a string C<s> of length C<len> from ASCII into UTF8 encoding. |
6662521e |
714 | Returns a pointer to the newly-created string, and sets C<len> to |
715 | reflect the new length. |
6940069f |
716 | |
497711e7 |
717 | =cut |
6940069f |
718 | */ |
719 | |
720 | U8* |
6662521e |
721 | Perl_bytes_to_utf8(pTHX_ U8* s, STRLEN *len) |
6940069f |
722 | { |
6940069f |
723 | U8 *send; |
724 | U8 *d; |
725 | U8 *dst; |
6662521e |
726 | send = s + (*len); |
6940069f |
727 | |
6662521e |
728 | Newz(801, d, (*len) * 2 + 1, U8); |
6940069f |
729 | dst = d; |
730 | |
731 | while (s < send) { |
90f44359 |
732 | if (UTF8_IS_ASCII(*s)) |
6940069f |
733 | *d++ = *s++; |
734 | else { |
735 | UV uv = *s++; |
90f44359 |
736 | |
737 | *d++ = UTF8_EIGHT_BIT_HI(uv); |
738 | *d++ = UTF8_EIGHT_BIT_LO(uv); |
6940069f |
739 | } |
740 | } |
741 | *d = '\0'; |
6662521e |
742 | *len = d-dst; |
6940069f |
743 | return dst; |
744 | } |
745 | |
a0ed51b3 |
746 | /* |
dea0fc0b |
747 | * Convert native (big-endian) or reversed (little-endian) UTF-16 to UTF-8. |
a0ed51b3 |
748 | * |
749 | * Destination must be pre-extended to 3/2 source. Do not use in-place. |
750 | * We optimize for native, for obvious reasons. */ |
751 | |
752 | U8* |
dea0fc0b |
753 | Perl_utf16_to_utf8(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen) |
a0ed51b3 |
754 | { |
dea0fc0b |
755 | U8* pend; |
756 | U8* dstart = d; |
757 | |
758 | if (bytelen & 1) |
a7867d0a |
759 | Perl_croak(aTHX_ "panic: utf16_to_utf8: odd bytelen"); |
dea0fc0b |
760 | |
761 | pend = p + bytelen; |
762 | |
a0ed51b3 |
763 | while (p < pend) { |
dea0fc0b |
764 | UV uv = (p[0] << 8) + p[1]; /* UTF-16BE */ |
765 | p += 2; |
a0ed51b3 |
766 | if (uv < 0x80) { |
767 | *d++ = uv; |
768 | continue; |
769 | } |
770 | if (uv < 0x800) { |
771 | *d++ = (( uv >> 6) | 0xc0); |
772 | *d++ = (( uv & 0x3f) | 0x80); |
773 | continue; |
774 | } |
775 | if (uv >= 0xd800 && uv < 0xdbff) { /* surrogates */ |
dea0fc0b |
776 | UV low = *p++; |
777 | if (low < 0xdc00 || low >= 0xdfff) |
778 | Perl_croak(aTHX_ "Malformed UTF-16 surrogate"); |
a0ed51b3 |
779 | uv = ((uv - 0xd800) << 10) + (low - 0xdc00) + 0x10000; |
780 | } |
781 | if (uv < 0x10000) { |
782 | *d++ = (( uv >> 12) | 0xe0); |
783 | *d++ = (((uv >> 6) & 0x3f) | 0x80); |
784 | *d++ = (( uv & 0x3f) | 0x80); |
785 | continue; |
786 | } |
787 | else { |
788 | *d++ = (( uv >> 18) | 0xf0); |
789 | *d++ = (((uv >> 12) & 0x3f) | 0x80); |
790 | *d++ = (((uv >> 6) & 0x3f) | 0x80); |
791 | *d++ = (( uv & 0x3f) | 0x80); |
792 | continue; |
793 | } |
794 | } |
dea0fc0b |
795 | *newlen = d - dstart; |
a0ed51b3 |
796 | return d; |
797 | } |
798 | |
799 | /* Note: this one is slightly destructive of the source. */ |
800 | |
801 | U8* |
dea0fc0b |
802 | Perl_utf16_to_utf8_reversed(pTHX_ U8* p, U8* d, I32 bytelen, I32 *newlen) |
a0ed51b3 |
803 | { |
804 | U8* s = (U8*)p; |
805 | U8* send = s + bytelen; |
806 | while (s < send) { |
807 | U8 tmp = s[0]; |
808 | s[0] = s[1]; |
809 | s[1] = tmp; |
810 | s += 2; |
811 | } |
dea0fc0b |
812 | return utf16_to_utf8(p, d, bytelen, newlen); |
a0ed51b3 |
813 | } |
814 | |
815 | /* for now these are all defined (inefficiently) in terms of the utf8 versions */ |
816 | |
817 | bool |
864dbfa3 |
818 | Perl_is_uni_alnum(pTHX_ U32 c) |
a0ed51b3 |
819 | { |
ad391ad9 |
820 | U8 tmpbuf[UTF8_MAXLEN+1]; |
9041c2e3 |
821 | uvuni_to_utf8(tmpbuf, (UV)c); |
a0ed51b3 |
822 | return is_utf8_alnum(tmpbuf); |
823 | } |
824 | |
825 | bool |
b8c5462f |
826 | Perl_is_uni_alnumc(pTHX_ U32 c) |
827 | { |
ad391ad9 |
828 | U8 tmpbuf[UTF8_MAXLEN+1]; |
9041c2e3 |
829 | uvuni_to_utf8(tmpbuf, (UV)c); |
b8c5462f |
830 | return is_utf8_alnumc(tmpbuf); |
831 | } |
832 | |
833 | bool |
864dbfa3 |
834 | Perl_is_uni_idfirst(pTHX_ U32 c) |
a0ed51b3 |
835 | { |
ad391ad9 |
836 | U8 tmpbuf[UTF8_MAXLEN+1]; |
9041c2e3 |
837 | uvuni_to_utf8(tmpbuf, (UV)c); |
a0ed51b3 |
838 | return is_utf8_idfirst(tmpbuf); |
839 | } |
840 | |
841 | bool |
864dbfa3 |
842 | Perl_is_uni_alpha(pTHX_ U32 c) |
a0ed51b3 |
843 | { |
ad391ad9 |
844 | U8 tmpbuf[UTF8_MAXLEN+1]; |
9041c2e3 |
845 | uvuni_to_utf8(tmpbuf, (UV)c); |
a0ed51b3 |
846 | return is_utf8_alpha(tmpbuf); |
847 | } |
848 | |
849 | bool |
4d61ec05 |
850 | Perl_is_uni_ascii(pTHX_ U32 c) |
851 | { |
ad391ad9 |
852 | U8 tmpbuf[UTF8_MAXLEN+1]; |
9041c2e3 |
853 | uvuni_to_utf8(tmpbuf, (UV)c); |
4d61ec05 |
854 | return is_utf8_ascii(tmpbuf); |
855 | } |
856 | |
857 | bool |
864dbfa3 |
858 | Perl_is_uni_space(pTHX_ U32 c) |
a0ed51b3 |
859 | { |
ad391ad9 |
860 | U8 tmpbuf[UTF8_MAXLEN+1]; |
9041c2e3 |
861 | uvuni_to_utf8(tmpbuf, (UV)c); |
a0ed51b3 |
862 | return is_utf8_space(tmpbuf); |
863 | } |
864 | |
865 | bool |
864dbfa3 |
866 | Perl_is_uni_digit(pTHX_ U32 c) |
a0ed51b3 |
867 | { |
ad391ad9 |
868 | U8 tmpbuf[UTF8_MAXLEN+1]; |
9041c2e3 |
869 | uvuni_to_utf8(tmpbuf, (UV)c); |
a0ed51b3 |
870 | return is_utf8_digit(tmpbuf); |
871 | } |
872 | |
873 | bool |
864dbfa3 |
874 | Perl_is_uni_upper(pTHX_ U32 c) |
a0ed51b3 |
875 | { |
ad391ad9 |
876 | U8 tmpbuf[UTF8_MAXLEN+1]; |
9041c2e3 |
877 | uvuni_to_utf8(tmpbuf, (UV)c); |
a0ed51b3 |
878 | return is_utf8_upper(tmpbuf); |
879 | } |
880 | |
881 | bool |
864dbfa3 |
882 | Perl_is_uni_lower(pTHX_ U32 c) |
a0ed51b3 |
883 | { |
ad391ad9 |
884 | U8 tmpbuf[UTF8_MAXLEN+1]; |
9041c2e3 |
885 | uvuni_to_utf8(tmpbuf, (UV)c); |
a0ed51b3 |
886 | return is_utf8_lower(tmpbuf); |
887 | } |
888 | |
889 | bool |
b8c5462f |
890 | Perl_is_uni_cntrl(pTHX_ U32 c) |
891 | { |
ad391ad9 |
892 | U8 tmpbuf[UTF8_MAXLEN+1]; |
9041c2e3 |
893 | uvuni_to_utf8(tmpbuf, (UV)c); |
b8c5462f |
894 | return is_utf8_cntrl(tmpbuf); |
895 | } |
896 | |
897 | bool |
898 | Perl_is_uni_graph(pTHX_ U32 c) |
899 | { |
ad391ad9 |
900 | U8 tmpbuf[UTF8_MAXLEN+1]; |
9041c2e3 |
901 | uvuni_to_utf8(tmpbuf, (UV)c); |
b8c5462f |
902 | return is_utf8_graph(tmpbuf); |
903 | } |
904 | |
905 | bool |
864dbfa3 |
906 | Perl_is_uni_print(pTHX_ U32 c) |
a0ed51b3 |
907 | { |
ad391ad9 |
908 | U8 tmpbuf[UTF8_MAXLEN+1]; |
9041c2e3 |
909 | uvuni_to_utf8(tmpbuf, (UV)c); |
a0ed51b3 |
910 | return is_utf8_print(tmpbuf); |
911 | } |
912 | |
b8c5462f |
913 | bool |
f248d071 |
914 | Perl_is_uni_punct(pTHX_ U32 c) |
b8c5462f |
915 | { |
ad391ad9 |
916 | U8 tmpbuf[UTF8_MAXLEN+1]; |
9041c2e3 |
917 | uvuni_to_utf8(tmpbuf, (UV)c); |
b8c5462f |
918 | return is_utf8_punct(tmpbuf); |
919 | } |
920 | |
4d61ec05 |
921 | bool |
922 | Perl_is_uni_xdigit(pTHX_ U32 c) |
923 | { |
ad391ad9 |
924 | U8 tmpbuf[UTF8_MAXLEN+1]; |
9041c2e3 |
925 | uvuni_to_utf8(tmpbuf, (UV)c); |
4d61ec05 |
926 | return is_utf8_xdigit(tmpbuf); |
927 | } |
928 | |
a0ed51b3 |
929 | U32 |
864dbfa3 |
930 | Perl_to_uni_upper(pTHX_ U32 c) |
a0ed51b3 |
931 | { |
ad391ad9 |
932 | U8 tmpbuf[UTF8_MAXLEN+1]; |
9041c2e3 |
933 | uvuni_to_utf8(tmpbuf, (UV)c); |
a0ed51b3 |
934 | return to_utf8_upper(tmpbuf); |
935 | } |
936 | |
937 | U32 |
864dbfa3 |
938 | Perl_to_uni_title(pTHX_ U32 c) |
a0ed51b3 |
939 | { |
ad391ad9 |
940 | U8 tmpbuf[UTF8_MAXLEN+1]; |
9041c2e3 |
941 | uvuni_to_utf8(tmpbuf, (UV)c); |
a0ed51b3 |
942 | return to_utf8_title(tmpbuf); |
943 | } |
944 | |
945 | U32 |
864dbfa3 |
946 | Perl_to_uni_lower(pTHX_ U32 c) |
a0ed51b3 |
947 | { |
ad391ad9 |
948 | U8 tmpbuf[UTF8_MAXLEN+1]; |
9041c2e3 |
949 | uvuni_to_utf8(tmpbuf, (UV)c); |
a0ed51b3 |
950 | return to_utf8_lower(tmpbuf); |
951 | } |
952 | |
953 | /* for now these all assume no locale info available for Unicode > 255 */ |
954 | |
955 | bool |
864dbfa3 |
956 | Perl_is_uni_alnum_lc(pTHX_ U32 c) |
a0ed51b3 |
957 | { |
958 | return is_uni_alnum(c); /* XXX no locale support yet */ |
959 | } |
960 | |
961 | bool |
b8c5462f |
962 | Perl_is_uni_alnumc_lc(pTHX_ U32 c) |
963 | { |
964 | return is_uni_alnumc(c); /* XXX no locale support yet */ |
965 | } |
966 | |
967 | bool |
864dbfa3 |
968 | Perl_is_uni_idfirst_lc(pTHX_ U32 c) |
a0ed51b3 |
969 | { |
970 | return is_uni_idfirst(c); /* XXX no locale support yet */ |
971 | } |
972 | |
973 | bool |
864dbfa3 |
974 | Perl_is_uni_alpha_lc(pTHX_ U32 c) |
a0ed51b3 |
975 | { |
976 | return is_uni_alpha(c); /* XXX no locale support yet */ |
977 | } |
978 | |
979 | bool |
4d61ec05 |
980 | Perl_is_uni_ascii_lc(pTHX_ U32 c) |
981 | { |
982 | return is_uni_ascii(c); /* XXX no locale support yet */ |
983 | } |
984 | |
985 | bool |
864dbfa3 |
986 | Perl_is_uni_space_lc(pTHX_ U32 c) |
a0ed51b3 |
987 | { |
988 | return is_uni_space(c); /* XXX no locale support yet */ |
989 | } |
990 | |
991 | bool |
864dbfa3 |
992 | Perl_is_uni_digit_lc(pTHX_ U32 c) |
a0ed51b3 |
993 | { |
994 | return is_uni_digit(c); /* XXX no locale support yet */ |
995 | } |
996 | |
997 | bool |
864dbfa3 |
998 | Perl_is_uni_upper_lc(pTHX_ U32 c) |
a0ed51b3 |
999 | { |
1000 | return is_uni_upper(c); /* XXX no locale support yet */ |
1001 | } |
1002 | |
1003 | bool |
864dbfa3 |
1004 | Perl_is_uni_lower_lc(pTHX_ U32 c) |
a0ed51b3 |
1005 | { |
1006 | return is_uni_lower(c); /* XXX no locale support yet */ |
1007 | } |
1008 | |
1009 | bool |
b8c5462f |
1010 | Perl_is_uni_cntrl_lc(pTHX_ U32 c) |
1011 | { |
1012 | return is_uni_cntrl(c); /* XXX no locale support yet */ |
1013 | } |
1014 | |
1015 | bool |
1016 | Perl_is_uni_graph_lc(pTHX_ U32 c) |
1017 | { |
1018 | return is_uni_graph(c); /* XXX no locale support yet */ |
1019 | } |
1020 | |
1021 | bool |
864dbfa3 |
1022 | Perl_is_uni_print_lc(pTHX_ U32 c) |
a0ed51b3 |
1023 | { |
1024 | return is_uni_print(c); /* XXX no locale support yet */ |
1025 | } |
1026 | |
b8c5462f |
1027 | bool |
1028 | Perl_is_uni_punct_lc(pTHX_ U32 c) |
1029 | { |
1030 | return is_uni_punct(c); /* XXX no locale support yet */ |
1031 | } |
1032 | |
4d61ec05 |
1033 | bool |
1034 | Perl_is_uni_xdigit_lc(pTHX_ U32 c) |
1035 | { |
1036 | return is_uni_xdigit(c); /* XXX no locale support yet */ |
1037 | } |
1038 | |
a0ed51b3 |
1039 | U32 |
864dbfa3 |
1040 | Perl_to_uni_upper_lc(pTHX_ U32 c) |
a0ed51b3 |
1041 | { |
1042 | return to_uni_upper(c); /* XXX no locale support yet */ |
1043 | } |
1044 | |
1045 | U32 |
864dbfa3 |
1046 | Perl_to_uni_title_lc(pTHX_ U32 c) |
a0ed51b3 |
1047 | { |
1048 | return to_uni_title(c); /* XXX no locale support yet */ |
1049 | } |
1050 | |
1051 | U32 |
864dbfa3 |
1052 | Perl_to_uni_lower_lc(pTHX_ U32 c) |
a0ed51b3 |
1053 | { |
1054 | return to_uni_lower(c); /* XXX no locale support yet */ |
1055 | } |
1056 | |
a0ed51b3 |
1057 | bool |
864dbfa3 |
1058 | Perl_is_utf8_alnum(pTHX_ U8 *p) |
a0ed51b3 |
1059 | { |
386d01d6 |
1060 | if (!is_utf8_char(p)) |
1061 | return FALSE; |
a0ed51b3 |
1062 | if (!PL_utf8_alnum) |
289d4f09 |
1063 | /* NOTE: "IsWord", not "IsAlnum", since Alnum is a true |
1064 | * descendant of isalnum(3), in other words, it doesn't |
1065 | * contain the '_'. --jhi */ |
1066 | PL_utf8_alnum = swash_init("utf8", "IsWord", &PL_sv_undef, 0, 0); |
a0ed51b3 |
1067 | return swash_fetch(PL_utf8_alnum, p); |
1068 | /* return *p == '_' || is_utf8_alpha(p) || is_utf8_digit(p); */ |
1069 | #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */ |
1070 | if (!PL_utf8_alnum) |
1071 | PL_utf8_alnum = swash_init("utf8", "", |
1072 | sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0); |
1073 | return swash_fetch(PL_utf8_alnum, p); |
1074 | #endif |
1075 | } |
1076 | |
1077 | bool |
b8c5462f |
1078 | Perl_is_utf8_alnumc(pTHX_ U8 *p) |
1079 | { |
386d01d6 |
1080 | if (!is_utf8_char(p)) |
1081 | return FALSE; |
b8c5462f |
1082 | if (!PL_utf8_alnum) |
1083 | PL_utf8_alnum = swash_init("utf8", "IsAlnumC", &PL_sv_undef, 0, 0); |
1084 | return swash_fetch(PL_utf8_alnum, p); |
1085 | /* return is_utf8_alpha(p) || is_utf8_digit(p); */ |
1086 | #ifdef SURPRISINGLY_SLOWER /* probably because alpha is usually true */ |
1087 | if (!PL_utf8_alnum) |
1088 | PL_utf8_alnum = swash_init("utf8", "", |
1089 | sv_2mortal(newSVpv("+utf8::IsAlpha\n+utf8::IsDigit\n005F\n",0)), 0, 0); |
1090 | return swash_fetch(PL_utf8_alnum, p); |
1091 | #endif |
1092 | } |
1093 | |
1094 | bool |
864dbfa3 |
1095 | Perl_is_utf8_idfirst(pTHX_ U8 *p) |
a0ed51b3 |
1096 | { |
1097 | return *p == '_' || is_utf8_alpha(p); |
1098 | } |
1099 | |
1100 | bool |
864dbfa3 |
1101 | Perl_is_utf8_alpha(pTHX_ U8 *p) |
a0ed51b3 |
1102 | { |
386d01d6 |
1103 | if (!is_utf8_char(p)) |
1104 | return FALSE; |
a0ed51b3 |
1105 | if (!PL_utf8_alpha) |
e24b16f9 |
1106 | PL_utf8_alpha = swash_init("utf8", "IsAlpha", &PL_sv_undef, 0, 0); |
a0ed51b3 |
1107 | return swash_fetch(PL_utf8_alpha, p); |
1108 | } |
1109 | |
1110 | bool |
b8c5462f |
1111 | Perl_is_utf8_ascii(pTHX_ U8 *p) |
1112 | { |
386d01d6 |
1113 | if (!is_utf8_char(p)) |
1114 | return FALSE; |
b8c5462f |
1115 | if (!PL_utf8_ascii) |
1116 | PL_utf8_ascii = swash_init("utf8", "IsAscii", &PL_sv_undef, 0, 0); |
1117 | return swash_fetch(PL_utf8_ascii, p); |
1118 | } |
1119 | |
1120 | bool |
864dbfa3 |
1121 | Perl_is_utf8_space(pTHX_ U8 *p) |
a0ed51b3 |
1122 | { |
386d01d6 |
1123 | if (!is_utf8_char(p)) |
1124 | return FALSE; |
a0ed51b3 |
1125 | if (!PL_utf8_space) |
3bec3564 |
1126 | PL_utf8_space = swash_init("utf8", "IsSpacePerl", &PL_sv_undef, 0, 0); |
a0ed51b3 |
1127 | return swash_fetch(PL_utf8_space, p); |
1128 | } |
1129 | |
1130 | bool |
864dbfa3 |
1131 | Perl_is_utf8_digit(pTHX_ U8 *p) |
a0ed51b3 |
1132 | { |
386d01d6 |
1133 | if (!is_utf8_char(p)) |
1134 | return FALSE; |
a0ed51b3 |
1135 | if (!PL_utf8_digit) |
e24b16f9 |
1136 | PL_utf8_digit = swash_init("utf8", "IsDigit", &PL_sv_undef, 0, 0); |
a0ed51b3 |
1137 | return swash_fetch(PL_utf8_digit, p); |
1138 | } |
1139 | |
1140 | bool |
864dbfa3 |
1141 | Perl_is_utf8_upper(pTHX_ U8 *p) |
a0ed51b3 |
1142 | { |
386d01d6 |
1143 | if (!is_utf8_char(p)) |
1144 | return FALSE; |
a0ed51b3 |
1145 | if (!PL_utf8_upper) |
e24b16f9 |
1146 | PL_utf8_upper = swash_init("utf8", "IsUpper", &PL_sv_undef, 0, 0); |
a0ed51b3 |
1147 | return swash_fetch(PL_utf8_upper, p); |
1148 | } |
1149 | |
1150 | bool |
864dbfa3 |
1151 | Perl_is_utf8_lower(pTHX_ U8 *p) |
a0ed51b3 |
1152 | { |
386d01d6 |
1153 | if (!is_utf8_char(p)) |
1154 | return FALSE; |
a0ed51b3 |
1155 | if (!PL_utf8_lower) |
e24b16f9 |
1156 | PL_utf8_lower = swash_init("utf8", "IsLower", &PL_sv_undef, 0, 0); |
a0ed51b3 |
1157 | return swash_fetch(PL_utf8_lower, p); |
1158 | } |
1159 | |
1160 | bool |
b8c5462f |
1161 | Perl_is_utf8_cntrl(pTHX_ U8 *p) |
1162 | { |
386d01d6 |
1163 | if (!is_utf8_char(p)) |
1164 | return FALSE; |
b8c5462f |
1165 | if (!PL_utf8_cntrl) |
1166 | PL_utf8_cntrl = swash_init("utf8", "IsCntrl", &PL_sv_undef, 0, 0); |
1167 | return swash_fetch(PL_utf8_cntrl, p); |
1168 | } |
1169 | |
1170 | bool |
1171 | Perl_is_utf8_graph(pTHX_ U8 *p) |
1172 | { |
386d01d6 |
1173 | if (!is_utf8_char(p)) |
1174 | return FALSE; |
b8c5462f |
1175 | if (!PL_utf8_graph) |
1176 | PL_utf8_graph = swash_init("utf8", "IsGraph", &PL_sv_undef, 0, 0); |
1177 | return swash_fetch(PL_utf8_graph, p); |
1178 | } |
1179 | |
1180 | bool |
864dbfa3 |
1181 | Perl_is_utf8_print(pTHX_ U8 *p) |
a0ed51b3 |
1182 | { |
386d01d6 |
1183 | if (!is_utf8_char(p)) |
1184 | return FALSE; |
a0ed51b3 |
1185 | if (!PL_utf8_print) |
e24b16f9 |
1186 | PL_utf8_print = swash_init("utf8", "IsPrint", &PL_sv_undef, 0, 0); |
a0ed51b3 |
1187 | return swash_fetch(PL_utf8_print, p); |
1188 | } |
1189 | |
1190 | bool |
b8c5462f |
1191 | Perl_is_utf8_punct(pTHX_ U8 *p) |
1192 | { |
386d01d6 |
1193 | if (!is_utf8_char(p)) |
1194 | return FALSE; |
b8c5462f |
1195 | if (!PL_utf8_punct) |
1196 | PL_utf8_punct = swash_init("utf8", "IsPunct", &PL_sv_undef, 0, 0); |
1197 | return swash_fetch(PL_utf8_punct, p); |
1198 | } |
1199 | |
1200 | bool |
1201 | Perl_is_utf8_xdigit(pTHX_ U8 *p) |
1202 | { |
386d01d6 |
1203 | if (!is_utf8_char(p)) |
1204 | return FALSE; |
b8c5462f |
1205 | if (!PL_utf8_xdigit) |
1206 | PL_utf8_xdigit = swash_init("utf8", "IsXDigit", &PL_sv_undef, 0, 0); |
1207 | return swash_fetch(PL_utf8_xdigit, p); |
1208 | } |
1209 | |
1210 | bool |
864dbfa3 |
1211 | Perl_is_utf8_mark(pTHX_ U8 *p) |
a0ed51b3 |
1212 | { |
386d01d6 |
1213 | if (!is_utf8_char(p)) |
1214 | return FALSE; |
a0ed51b3 |
1215 | if (!PL_utf8_mark) |
e24b16f9 |
1216 | PL_utf8_mark = swash_init("utf8", "IsM", &PL_sv_undef, 0, 0); |
a0ed51b3 |
1217 | return swash_fetch(PL_utf8_mark, p); |
1218 | } |
1219 | |
2104c8d9 |
1220 | UV |
864dbfa3 |
1221 | Perl_to_utf8_upper(pTHX_ U8 *p) |
a0ed51b3 |
1222 | { |
1223 | UV uv; |
1224 | |
1225 | if (!PL_utf8_toupper) |
e24b16f9 |
1226 | PL_utf8_toupper = swash_init("utf8", "ToUpper", &PL_sv_undef, 4, 0); |
a0ed51b3 |
1227 | uv = swash_fetch(PL_utf8_toupper, p); |
9041c2e3 |
1228 | return uv ? UNI_TO_NATIVE(uv) : utf8_to_uvchr(p,0); |
a0ed51b3 |
1229 | } |
1230 | |
2104c8d9 |
1231 | UV |
864dbfa3 |
1232 | Perl_to_utf8_title(pTHX_ U8 *p) |
a0ed51b3 |
1233 | { |
1234 | UV uv; |
1235 | |
1236 | if (!PL_utf8_totitle) |
e24b16f9 |
1237 | PL_utf8_totitle = swash_init("utf8", "ToTitle", &PL_sv_undef, 4, 0); |
a0ed51b3 |
1238 | uv = swash_fetch(PL_utf8_totitle, p); |
9041c2e3 |
1239 | return uv ? UNI_TO_NATIVE(uv) : utf8_to_uvchr(p,0); |
a0ed51b3 |
1240 | } |
1241 | |
2104c8d9 |
1242 | UV |
864dbfa3 |
1243 | Perl_to_utf8_lower(pTHX_ U8 *p) |
a0ed51b3 |
1244 | { |
1245 | UV uv; |
1246 | |
1247 | if (!PL_utf8_tolower) |
e24b16f9 |
1248 | PL_utf8_tolower = swash_init("utf8", "ToLower", &PL_sv_undef, 4, 0); |
a0ed51b3 |
1249 | uv = swash_fetch(PL_utf8_tolower, p); |
9041c2e3 |
1250 | return uv ? UNI_TO_NATIVE(uv) : utf8_to_uvchr(p,0); |
a0ed51b3 |
1251 | } |
1252 | |
1253 | /* a "swash" is a swatch hash */ |
1254 | |
1255 | SV* |
864dbfa3 |
1256 | Perl_swash_init(pTHX_ char* pkg, char* name, SV *listsv, I32 minbits, I32 none) |
a0ed51b3 |
1257 | { |
1258 | SV* retval; |
bf1fed83 |
1259 | SV* tokenbufsv = sv_2mortal(NEWSV(0,0)); |
8e84507e |
1260 | dSP; |
ce3b816e |
1261 | |
1262 | if (!gv_stashpv(pkg, 0)) { /* demand load utf8 */ |
1263 | ENTER; |
1264 | Perl_load_module(aTHX_ PERL_LOADMOD_NOIMPORT, newSVpv(pkg,0), Nullsv); |
1265 | LEAVE; |
1266 | } |
1267 | SPAGAIN; |
a0ed51b3 |
1268 | PUSHSTACKi(PERLSI_MAGIC); |
1269 | PUSHMARK(SP); |
1270 | EXTEND(SP,5); |
1271 | PUSHs(sv_2mortal(newSVpvn(pkg, strlen(pkg)))); |
1272 | PUSHs(sv_2mortal(newSVpvn(name, strlen(name)))); |
1273 | PUSHs(listsv); |
1274 | PUSHs(sv_2mortal(newSViv(minbits))); |
1275 | PUSHs(sv_2mortal(newSViv(none))); |
1276 | PUTBACK; |
1277 | ENTER; |
1278 | SAVEI32(PL_hints); |
1279 | PL_hints = 0; |
1280 | save_re_context(); |
bf1fed83 |
1281 | if (PL_curcop == &PL_compiling) |
1282 | /* XXX ought to be handled by lex_start */ |
1283 | sv_setpv(tokenbufsv, PL_tokenbuf); |
864dbfa3 |
1284 | if (call_method("SWASHNEW", G_SCALAR)) |
8e84507e |
1285 | retval = newSVsv(*PL_stack_sp--); |
a0ed51b3 |
1286 | else |
e24b16f9 |
1287 | retval = &PL_sv_undef; |
a0ed51b3 |
1288 | LEAVE; |
1289 | POPSTACK; |
e24b16f9 |
1290 | if (PL_curcop == &PL_compiling) { |
bf1fed83 |
1291 | STRLEN len; |
1292 | char* pv = SvPV(tokenbufsv, len); |
1293 | |
1294 | Copy(pv, PL_tokenbuf, len+1, char); |
e24b16f9 |
1295 | PL_curcop->op_private = PL_hints; |
a0ed51b3 |
1296 | } |
1297 | if (!SvROK(retval) || SvTYPE(SvRV(retval)) != SVt_PVHV) |
cea2e8a9 |
1298 | Perl_croak(aTHX_ "SWASHNEW didn't return an HV ref"); |
a0ed51b3 |
1299 | return retval; |
1300 | } |
1301 | |
1302 | UV |
864dbfa3 |
1303 | Perl_swash_fetch(pTHX_ SV *sv, U8 *ptr) |
a0ed51b3 |
1304 | { |
1305 | HV* hv = (HV*)SvRV(sv); |
1306 | U32 klen = UTF8SKIP(ptr) - 1; |
1307 | U32 off = ptr[klen] & 127; /* NB: 64 bit always 0 when len > 1 */ |
1308 | STRLEN slen; |
1309 | STRLEN needents = (klen ? 64 : 128); |
dfe13c55 |
1310 | U8 *tmps; |
a0ed51b3 |
1311 | U32 bit; |
1312 | SV *retval; |
1313 | |
1314 | /* |
1315 | * This single-entry cache saves about 1/3 of the utf8 overhead in test |
1316 | * suite. (That is, only 7-8% overall over just a hash cache. Still, |
1317 | * it's nothing to sniff at.) Pity we usually come through at least |
1318 | * two function calls to get here... |
1319 | * |
1320 | * NB: this code assumes that swatches are never modified, once generated! |
1321 | */ |
1322 | |
1323 | if (hv == PL_last_swash_hv && |
1324 | klen == PL_last_swash_klen && |
12ae5dfc |
1325 | (!klen || memEQ((char *)ptr,(char *)PL_last_swash_key,klen)) ) |
a0ed51b3 |
1326 | { |
1327 | tmps = PL_last_swash_tmps; |
1328 | slen = PL_last_swash_slen; |
1329 | } |
1330 | else { |
1331 | /* Try our second-level swatch cache, kept in a hash. */ |
dfe13c55 |
1332 | SV** svp = hv_fetch(hv, (char*)ptr, klen, FALSE); |
a0ed51b3 |
1333 | |
1334 | /* If not cached, generate it via utf8::SWASHGET */ |
dfe13c55 |
1335 | if (!svp || !SvPOK(*svp) || !(tmps = (U8*)SvPV(*svp, slen))) { |
a0ed51b3 |
1336 | dSP; |
1337 | ENTER; |
1338 | SAVETMPS; |
1339 | save_re_context(); |
1340 | PUSHSTACKi(PERLSI_MAGIC); |
1341 | PUSHMARK(SP); |
1342 | EXTEND(SP,3); |
1343 | PUSHs((SV*)sv); |
9041c2e3 |
1344 | /* We call utf8_to_uni as we want and index into Unicode tables, |
1345 | not a native character number. |
1346 | */ |
1347 | PUSHs(sv_2mortal(newSViv(utf8_to_uvuni(ptr, 0) & ~(needents - 1)))); |
a0ed51b3 |
1348 | PUSHs(sv_2mortal(newSViv(needents))); |
1349 | PUTBACK; |
864dbfa3 |
1350 | if (call_method("SWASHGET", G_SCALAR)) |
8e84507e |
1351 | retval = newSVsv(*PL_stack_sp--); |
a0ed51b3 |
1352 | else |
e24b16f9 |
1353 | retval = &PL_sv_undef; |
a0ed51b3 |
1354 | POPSTACK; |
1355 | FREETMPS; |
1356 | LEAVE; |
e24b16f9 |
1357 | if (PL_curcop == &PL_compiling) |
1358 | PL_curcop->op_private = PL_hints; |
a0ed51b3 |
1359 | |
dfe13c55 |
1360 | svp = hv_store(hv, (char*)ptr, klen, retval, 0); |
a0ed51b3 |
1361 | |
dfe13c55 |
1362 | if (!svp || !(tmps = (U8*)SvPV(*svp, slen)) || slen < 8) |
cea2e8a9 |
1363 | Perl_croak(aTHX_ "SWASHGET didn't return result of proper length"); |
a0ed51b3 |
1364 | } |
1365 | |
1366 | PL_last_swash_hv = hv; |
1367 | PL_last_swash_klen = klen; |
1368 | PL_last_swash_tmps = tmps; |
1369 | PL_last_swash_slen = slen; |
1370 | if (klen) |
1371 | Copy(ptr, PL_last_swash_key, klen, U8); |
1372 | } |
1373 | |
9faf8d75 |
1374 | switch ((int)((slen << 3) / needents)) { |
a0ed51b3 |
1375 | case 1: |
1376 | bit = 1 << (off & 7); |
1377 | off >>= 3; |
1378 | return (tmps[off] & bit) != 0; |
1379 | case 8: |
1380 | return tmps[off]; |
1381 | case 16: |
1382 | off <<= 1; |
1383 | return (tmps[off] << 8) + tmps[off + 1] ; |
1384 | case 32: |
1385 | off <<= 2; |
1386 | return (tmps[off] << 24) + (tmps[off+1] << 16) + (tmps[off+2] << 8) + tmps[off + 3] ; |
1387 | } |
cea2e8a9 |
1388 | Perl_croak(aTHX_ "panic: swash_fetch"); |
a0ed51b3 |
1389 | return 0; |
1390 | } |