2 # Copyright (C) 2004 Brion Vibber <brion@pobox.com>
3 # http://www.mediawiki.org/
5 # This program is free software; you can redistribute it and/or modify
6 # it under the terms of the GNU General Public License as published by
7 # the Free Software Foundation; either version 2 of the License, or
8 # (at your option) any later version.
10 # This program is distributed in the hope that it will be useful,
11 # but WITHOUT ANY WARRANTY; without even the implied warranty of
12 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 # GNU General Public License for more details.
15 # You should have received a copy of the GNU General Public License along
16 # with this program; if not, write to the Free Software Foundation, Inc.,
17 # 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
18 # http://www.gnu.org/copyleft/gpl.html
21 * Unicode normalization routines for working with UTF-8 strings.
22 * Currently assumes that input strings are valid UTF-8!
24 * Not as fast as I'd like, but should be usable for most purposes.
25 * UtfNormal::toNFC() will bail early if given ASCII text or text
26 * it can quickly deterimine is already normalized.
28 * All functions can be called static.
30 * See description of forms at http://www.unicode.org/reports/tr15/
36 require_once 'UtfNormalUtil.php';
37 require_once 'UtfNormalData.inc';
39 # Load compatibility decompositions on demand if they are needed.
40 global $utfCompatibilityDecomp;
41 $utfCompatibilityDecomp = NULL;
43 define( 'UNICODE_HANGUL_FIRST', 0xac00 );
44 define( 'UNICODE_HANGUL_LAST', 0xd7a3 );
46 define( 'UNICODE_HANGUL_LBASE', 0x1100 );
47 define( 'UNICODE_HANGUL_VBASE', 0x1161 );
48 define( 'UNICODE_HANGUL_TBASE', 0x11a7 );
50 define( 'UNICODE_HANGUL_LCOUNT', 19 );
51 define( 'UNICODE_HANGUL_VCOUNT', 21 );
52 define( 'UNICODE_HANGUL_TCOUNT', 28 );
53 define( 'UNICODE_HANGUL_NCOUNT', UNICODE_HANGUL_VCOUNT
* UNICODE_HANGUL_TCOUNT
);
55 define( 'UNICODE_HANGUL_LEND', UNICODE_HANGUL_LBASE + UNICODE_HANGUL_LCOUNT
- 1 );
56 define( 'UNICODE_HANGUL_VEND', UNICODE_HANGUL_VBASE + UNICODE_HANGUL_VCOUNT
- 1 );
57 define( 'UNICODE_HANGUL_TEND', UNICODE_HANGUL_TBASE + UNICODE_HANGUL_TCOUNT
- 1 );
59 define( 'UNICODE_SURROGATE_FIRST', 0xd800 );
60 define( 'UNICODE_SURROGATE_LAST', 0xdfff );
61 define( 'UNICODE_MAX', 0x10ffff );
62 define( 'UNICODE_REPLACEMENT', 0xfffd );
65 define( 'UTF8_HANGUL_FIRST', codepointToUtf8( UNICODE_HANGUL_FIRST
) );
66 define( 'UTF8_HANGUL_LAST', codepointToUtf8( UNICODE_HANGUL_LAST
) );
68 define( 'UTF8_HANGUL_LBASE', codepointToUtf8( UNICODE_HANGUL_LBASE
) );
69 define( 'UTF8_HANGUL_VBASE', codepointToUtf8( UNICODE_HANGUL_VBASE
) );
70 define( 'UTF8_HANGUL_TBASE', codepointToUtf8( UNICODE_HANGUL_TBASE
) );
72 define( 'UTF8_HANGUL_LEND', codepointToUtf8( UNICODE_HANGUL_LEND
) );
73 define( 'UTF8_HANGUL_VEND', codepointToUtf8( UNICODE_HANGUL_VEND
) );
74 define( 'UTF8_HANGUL_TEND', codepointToUtf8( UNICODE_HANGUL_TEND
) );
76 define( 'UTF8_SURROGATE_FIRST', codepointToUtf8( UNICODE_SURROGATE_FIRST
) );
77 define( 'UTF8_SURROGATE_LAST', codepointToUtf8( UNICODE_SURROGATE_LAST
) );
78 define( 'UTF8_MAX', codepointToUtf8( UNICODE_MAX
) );
79 define( 'UTF8_REPLACEMENT', codepointToUtf8( UNICODE_REPLACEMENT
) );
80 #define( 'UTF8_REPLACEMENT', '!' );
82 define( 'UTF8_OVERLONG_A', "\xc1\xbf" );
83 define( 'UTF8_OVERLONG_B', "\xe0\x9f\xbf" );
84 define( 'UTF8_OVERLONG_C', "\xf0\x8f\xbf\xbf" );
86 # These two ranges are illegal
87 define( 'UTF8_FDD0', codepointToUtf8( 0xfdd0 ) );
88 define( 'UTF8_FDEF', codepointToUtf8( 0xfdef ) );
89 define( 'UTF8_FFFE', codepointToUtf8( 0xfffe ) );
90 define( 'UTF8_FFFF', codepointToUtf8( 0xffff ) );
92 define( 'UTF8_HEAD', false );
93 define( 'UTF8_TAIL', true );
100 # The ultimate convenience function! Clean up invalid UTF-8 sequences,
101 # and convert to normal form C. Faster on pure ASCII strings, or
102 # secondarily on strings which are already definitely normalized.
103 function cleanUp( $string ) {
104 if( UtfNormal
::quickIsNFCVerify( $string ) )
107 return UtfNormal
::NFC( $string );
110 # These functions try to skip the conversion if it won't be necessary.
111 # An all ASCII string for instance doesn't need conversion.
112 function toNFC( $string ) {
113 if( UtfNormal
::quickIsNFC( $string ) )
116 return UtfNormal
::NFC( $string );
119 function toNFD( $string ) {
120 if( preg_match( '/[\x80-\xff]/', $string ) )
121 return UtfNormal
::NFD( $string );
126 function toNFKC( $string ) {
127 if( preg_match( '/[\x80-\xff]/', $string ) )
128 return UtfNormal
::NFKC( $string );
133 function toNFKD( $string ) {
134 if( preg_match( '/[\x80-\xff]/', $string ) )
135 return UtfNormal
::NFKD( $string );
140 # Returns true if the string is _definitely_ in NFC.
141 # Returns false if not or uncertain.
142 function quickIsNFC( $string ) {
143 # ASCII is always valid NFC!
144 # If it's pure ASCII and doesn't contain any XML-forbidden chars, let it through.
145 if( !preg_match( '/[\x00-\x08\x0b\x0c\x0f-\x1f\x80-\xff]/', $string ) ) return true;
147 global $utfCheckNFC, $utfCombiningClass;
148 $len = strlen( $string );
149 for( $i = 0; $i < $len; $i++
) {
154 } elseif( $n >= 0xf0 ) {
155 $c = substr( $string, $i, 4 );
157 } elseif( $n >= 0xe0 ) {
158 $c = substr( $string, $i, 3 );
160 } elseif( $n >= 0xc0 ) {
161 $c = substr( $string, $i, 2 );
164 if( isset( $utfCheckNFC[$c] ) ) {
165 # If it's NO or MAYBE, bail and do the slow check.
168 if( isset( $utfCombiningClass[$c] ) ) {
169 # Combining character? We might have to do sorting, at least.
176 # As above, but also *alter the string* to strip invalid UTF-8 sequences.
177 function quickIsNFCVerify( &$string ) {
178 # ASCII is always valid NFC!
179 if( !preg_match( '/[\x80-\xff]/', $string ) ) return true;
181 global $utfCheckNFC, $utfCombiningClass;
182 $len = strlen( $string );
189 for( $i = 0; $i < $len; $i++
) {
192 if( $state == UTF8_TAIL
) {
193 if( $n >= 0x80 && $n < 0xc0 ) {
195 if( --$remaining == 0 ) {
196 if( ($sequence >= UTF8_SURROGATE_FIRST
197 && $sequence <= UTF8_SURROGATE_LAST
)
198 ||
($head == 0xc0 && $sequence <= UTF8_OVERLONG_A
)
199 ||
($head == 0xc1 && $sequence <= UTF8_OVERLONG_A
)
200 ||
($head == 0xe0 && $sequence <= UTF8_OVERLONG_B
)
201 ||
($head == 0xf0 && $sequence <= UTF8_OVERLONG_C
)
202 ||
($sequence >= UTF8_FDD0
&& $sequence <= UTF8_FDEF
)
203 ||
($sequence == UTF8_FFFE
)
204 ||
($sequence == UTF8_FFFF
)
205 ||
($sequence > UTF8_MAX
) ) {
206 $out .= UTF8_REPLACEMENT
;
210 if( isset( $utfCheckNFC[$sequence] ) ||
211 isset( $utfCombiningClass[$sequence] ) ) {
212 # If it's NO or MAYBE, we'll have to do the slow check.
213 $looksNormal = false;
221 # Not a valid tail byte! DIscard the char we've been building.
222 #printf ("Invalid '%x' in tail with %d remaining bytes\n", $n, $remaining );
224 $out .= UTF8_REPLACEMENT
;
227 $out .= UTF8_REPLACEMENT
;
228 } elseif( $n == 0x0a ) {
230 } elseif( $n < 0x0d ) {
231 $out .= UTF8_REPLACEMENT
;
232 } elseif( $n == 0x0d ) {
234 } elseif( $n < 0x20 ) {
235 $out .= UTF8_REPLACEMENT
;
236 } elseif( $n < 0x80 ) {
238 } elseif( $n < 0xc0 ) {
239 # illegal tail bytes or head byte of overlong sequence
240 if( $head == 0 ) $out .= UTF8_REPLACEMENT
;
241 } elseif( $n < 0xe0 ) {
246 } elseif( $n < 0xf0 ) {
251 } elseif( $n < 0xf8 ) {
256 } elseif( $n < 0xfc ) {
261 } elseif( $n < 0xfe ) {
267 $out .= UTF8_REPLACEMENT
;
270 if( $state == UTF8_TAIL
) {
271 $out .= UTF8_REPLACEMENT
;
277 # These take a string and run the normalization on them, without
278 # checking for validity or any optimization etc. Input must be
280 function NFC( $string ) {
281 return $out = UtfNormal
::fastCompose( UtfNormal
::NFD( $string ) );
284 function NFD( $string ) {
285 global $utfCanonicalDecomp;
286 return UtfNormal
::fastCombiningSort(
287 UtfNormal
::fastDecompose( $string, $utfCanonicalDecomp ) );
290 function NFKC( $string ) {
291 return UtfNormal
::fastCompose( UtfNormal
::NFKD( $string ) );
294 function NFKD( $string ) {
295 global $utfCompatibilityDecomp;
296 if( !isset( $utfCompatibilityDecomp ) ) {
297 require_once( 'UtfNormalDataK.inc' );
299 return UtfNormal
::fastCombiningSort(
300 UtfNormal
::fastDecompose( $string, $utfCompatibilityDecomp ) );
305 # Perform decomposition of a UTF-8 string into either D or KD form
306 # (depending on which decomposition map is passed to us).
307 # Input is assumed to be *valid* UTF-8. Invalid code will break.
308 function fastDecompose( &$string, &$map ) {
309 $len = strlen( $string );
311 for( $i = 0; $i < $len; $i++
) {
315 # ASCII chars never decompose
319 } elseif( $n >= 0xf0 ) {
320 $c = substr( $string, $i, 4 );
322 } elseif( $n >= 0xe0 ) {
323 $c = substr( $string, $i, 3 );
325 } elseif( $n >= 0xc0 ) {
326 $c = substr( $string, $i, 2 );
329 if( isset( $map[$c] ) ) {
332 if( $c >= UTF8_HANGUL_FIRST
&& $c <= UTF8_HANGUL_LAST
) {
333 $out .= UtfNormal
::decomposeHangul( $c );
342 function decomposeHangul( $c ) {
343 $codepoint = utf8ToCodepoint( $c );
344 $index = $codepoint - UNICODE_HANGUL_FIRST
;
345 $l = IntVal( $index / UNICODE_HANGUL_NCOUNT
);
346 $v = IntVal( ($index % UNICODE_HANGUL_NCOUNT
) / UNICODE_HANGUL_TCOUNT
);
347 $t = $index % UNICODE_HANGUL_TCOUNT
;
348 $out = codepointToUtf8( $l + UNICODE_HANGUL_LBASE
);
349 $out .= codepointToUtf8( $v + UNICODE_HANGUL_VBASE
);
350 if( $t ) $out .= codepointToUtf8( $t + UNICODE_HANGUL_TBASE
);
354 # Sorts combining characters into canonical order. This is the
355 # final step in creating decomposed normal forms D and KD.
356 function fastCombiningSort( $string ) {
357 global $utfCombiningClass;
359 while( $replacedCount > 0 ) {
361 $len = strlen( $string );
365 for( $i = 0; $i < $len; $i++
) {
369 $c = substr( $string, $i, 4 );
371 } elseif( $n >= 0xe0 ) {
372 $c = substr( $string, $i, 3 );
374 } elseif( $n >= 0xc0 ) {
375 $c = substr( $string, $i, 2 );
378 $class = isset( $utfCombiningClass[$c] ) ?
$utfCombiningClass[$c] : 0;
379 if( $lastClass == -1 ) {
383 } elseif( $lastClass > $class && $class > 0 ) {
384 # Swap -- put this one on the stack
399 # Produces canonically composed sequences, i.e. normal form C or KC.
400 # Input must be valid UTF-8 in sorted normal form D or KD.
401 function fastCompose( $string ) {
402 global $utfCanonicalComp, $utfCombiningClass;
403 $len = strlen( $string );
408 for( $i = 0; $i < $len; $i++
) {
412 $c = substr( $string, $i, 4 );
414 } elseif( $n >= 0xe0 ) {
415 $c = substr( $string, $i, 3 );
417 } elseif( $n >= 0xc0 ) {
418 $c = substr( $string, $i, 2 );
421 $class = isset( $utfCombiningClass[$c] ) ?
$utfCombiningClass[$c] : 0;
422 $pair = $startChar . $c;
423 if( empty( $utfCombiningClass[$c] ) ) {
425 if( $lastClass == 0 && isset( $utfCanonicalComp[$pair] ) ) {
426 $startChar = $utfCanonicalComp[$pair];
427 } elseif( $lastClass == 0 &&
428 $c >= UTF8_HANGUL_VBASE
&&
429 $c <= UTF8_HANGUL_VEND
&&
430 $startChar >= UTF8_HANGUL_LBASE
&&
431 $startChar <= UTF8_HANGUL_LEND
) {
432 $lIndex = utf8ToCodepoint( $startChar ) - UNICODE_HANGUL_LBASE
;
433 $vIndex = utf8ToCodepoint( $c ) - UNICODE_HANGUL_VBASE
;
434 $hangulPoint = UNICODE_HANGUL_FIRST +
435 UNICODE_HANGUL_TCOUNT
*
436 (UNICODE_HANGUL_VCOUNT
* $lIndex +
$vIndex);
437 $startChar = codepointToUtf8( $hangulPoint );
438 } elseif( $lastClass == 0 &&
439 $c >= UTF8_HANGUL_TBASE
&&
440 $c <= UTF8_HANGUL_TEND
&&
441 $startChar >= UTF8_HANGUL_FIRST
&&
442 $startChar <= UTF8_HANGUL_LAST
) {
443 $tIndex = utf8ToCodepoint( $c ) - UNICODE_HANGUL_TBASE
;
444 $hangulPoint = utf8ToCodepoint( $startChar ) +
$tIndex;
445 $startChar = codepointToUtf8( $hangulPoint );
453 # A combining char; see what we can do with it
454 if( !empty( $startChar ) &&
455 $lastClass < $class &&
457 isset( $utfCanonicalComp[$pair] ) ) {
458 $startChar = $utfCanonicalComp[$pair];
466 $out .= $startChar . $combining;
470 # This is just used for the benchmark, comparing how long it takes to
471 # interate through a string without really doing anything of substance.
472 function placebo( $string ) {
473 $len = strlen( $string );
475 for( $i = 0; $i < $len; $i++
) {