3 * Maintenance script to generate first letter data files for Collation.php.
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 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
18 * http://www.gnu.org/copyleft/gpl.html
21 * @ingroup MaintenanceLanguage
24 require_once __DIR__
. '/../Maintenance.php';
27 * Generate first letter data files for Collation.php
29 * @ingroup MaintenanceLanguage
31 class GenerateCollationData
extends Maintenance
{
32 /** The directory with source data files in it */
35 /** The primary weights, indexed by codepoint */
39 * A hashtable keyed by codepoint, where presence indicates that a character
40 * has a decomposition mapping. This makes it non-preferred for group header
48 * Important tertiary weights from UTS #10 section 7.2
50 const NORMAL_UPPERCASE
= 0x08;
51 const NORMAL_HIRAGANA
= 0x0E;
53 public function __construct() {
54 parent
::__construct();
55 $this->addOption( 'data-dir', 'A directory on the local filesystem ' .
56 'containing allkeys.txt and ucd.all.grouped.xml from unicode.org',
58 $this->addOption( 'debug-output', 'Filename for sending debug output to',
62 public function execute() {
63 $this->dataDir
= $this->getOption( 'data-dir', '.' );
65 $allkeysPresent = file_exists( "{$this->dataDir}/allkeys.txt" );
66 $ucdallPresent = file_exists( "{$this->dataDir}/ucd.all.grouped.xml" );
68 // As of January 2013, these links work for all versions of Unicode
69 // between 5.1 and 6.2, inclusive.
70 $allkeysURL = "https://www.unicode.org/Public/UCA/<Unicode version>/allkeys.txt";
71 $ucdallURL = "https://www.unicode.org/Public/<Unicode version>/ucdxml/ucd.all.grouped.zip";
73 if ( !$allkeysPresent ||
!$ucdallPresent ) {
74 $icuVersion = INTL_ICU_VERSION
;
75 $unicodeVersion = IcuCollation
::getUnicodeVersionForICU();
79 if ( !$allkeysPresent ) {
80 $error .= "Unable to find allkeys.txt. "
81 . "Download it and specify its location with --data-dir=<DIR>. "
84 if ( !$ucdallPresent ) {
85 $error .= "Unable to find ucd.all.grouped.xml. "
86 . "Download it, unzip, and specify its location with --data-dir=<DIR>. "
90 $versionKnown = false;
91 if ( version_compare( $icuVersion, "4.0", "<" ) ) {
93 $error .= "You are using outdated version of ICU ($icuVersion), intended for "
94 . ( $unicodeVersion ?
"Unicode $unicodeVersion" : "an unknown version of Unicode" )
95 . "; this file might not be avalaible for it, and it's not supported by MediaWiki. "
96 . " You are on your own; consider upgrading PHP's intl extension or try "
97 . "one of the files available at:";
98 } elseif ( version_compare( $icuVersion, "51.0", ">=" ) ) {
99 // Extra recent version
100 $error .= "You are using ICU $icuVersion, released after this script was last updated. "
101 . "Check what is the Unicode version it is using at http://site.icu-project.org/download . "
102 . "It can't be guaranteed everything will work, but appropriate file(s) should "
103 . "be available at:";
106 $versionKnown = true;
107 $error .= "You are using ICU $icuVersion, intended for "
108 . ( $unicodeVersion ?
"Unicode $unicodeVersion" : "an unknown version of Unicode" )
109 . ". Appropriate file(s) should be available at:";
113 if ( $versionKnown && $unicodeVersion ) {
114 $allkeysURL = str_replace( "<Unicode version>", "$unicodeVersion.0", $allkeysURL );
115 $ucdallURL = str_replace( "<Unicode version>", "$unicodeVersion.0", $ucdallURL );
118 if ( !$allkeysPresent ) {
119 $error .= "* $allkeysURL\n";
121 if ( !$ucdallPresent ) {
122 $error .= "* $ucdallURL\n";
125 $this->fatalError( $error );
128 $debugOutFileName = $this->getOption( 'debug-output' );
129 if ( $debugOutFileName ) {
130 $this->debugOutFile
= fopen( $debugOutFileName, 'w' );
131 if ( !$this->debugOutFile
) {
132 $this->fatalError( "Unable to open debug output file for writing" );
136 $this->generateFirstChars();
140 $uxr = new UcdXmlReader( "{$this->dataDir}/ucd.all.grouped.xml" );
141 $uxr->readChars( [ $this, 'charCallback' ] );
144 function charCallback( $data ) {
145 // Skip non-printable characters,
146 // but do not skip a normal space (U+0020) since
147 // people like to use that as a fake no header symbol.
148 $category = substr( $data['gc'], 0, 1 );
149 if ( strpos( 'LNPS', $category ) === false
150 && $data['cp'] !== '0020'
154 $cp = hexdec( $data['cp'] );
156 // Skip the CJK ideograph blocks, as an optimisation measure.
157 // UCA doesn't sort them properly anyway, without tailoring.
158 if ( IcuCollation
::isCjk( $cp ) ) {
162 // Skip the composed Hangul syllables, we will use the bare Jamo
164 if ( $data['block'] == 'Hangul Syllables' ) {
168 // Calculate implicit weight per UTS #10 v6.0.0, sec 7.1.3
169 if ( $data['UIdeo'] === 'Y' ) {
170 if ( $data['block'] == 'CJK Unified Ideographs'
171 ||
$data['block'] == 'CJK Compatibility Ideographs'
180 $a = $base +
( $cp >> 15 );
181 $b = ( $cp & 0x7fff ) |
0x8000;
183 $this->weights
[$cp] = sprintf( ".%04X.%04X", $a, $b );
185 if ( $data['dm'] !== '#' ) {
186 $this->mappedChars
[$cp] = true;
189 if ( $cp %
4096 == 0 ) {
190 print "{$data['cp']}\n";
194 function generateFirstChars() {
195 $file = fopen( "{$this->dataDir}/allkeys.txt", 'r' );
197 $this->fatalError( "Unable to open allkeys.txt" );
200 $goodTertiaryChars = [];
202 // For each character with an entry in allkeys.txt, overwrite the implicit
203 // entry in $this->weights that came from the UCD.
204 // Also gather a list of tertiary weights, for use in selecting the group header
205 while ( ( $line = fgets( $file ) ) !== false ) {
206 // We're only interested in single-character weights, pick them out with a regex
207 $line = trim( $line );
208 if ( !preg_match( '/^([0-9A-F]+)\s*;\s*([^#]*)/', $line, $m ) ) {
212 $cp = hexdec( $m[1] );
213 $allWeights = trim( $m[2] );
217 if ( !isset( $this->weights
[$cp] ) ) {
218 // Non-printable, ignore
221 foreach ( StringUtils
::explode( '[', $allWeights ) as $weightStr ) {
222 preg_match_all( '/[*.]([0-9A-F]+)/', $weightStr, $m );
223 if ( !empty( $m[1] ) ) {
224 if ( $m[1][0] !== '0000' ) {
225 $primary .= '.' . $m[1][0];
227 if ( $m[1][2] !== '0000' ) {
228 $tertiary .= '.' . $m[1][2];
232 $this->weights
[$cp] = $primary;
233 if ( $tertiary === '.0008'
234 ||
$tertiary === '.000E'
236 $goodTertiaryChars[$cp] = true;
241 // Identify groups of characters with the same primary weight
243 asort( $this->weights
, SORT_STRING
);
244 $prevWeight = reset( $this->weights
);
246 foreach ( $this->weights
as $cp => $weight ) {
247 if ( $weight !== $prevWeight ) {
248 $this->groups
[$prevWeight] = $group;
249 $prevWeight = $weight;
250 $group = $this->groups
[$weight] ??
[];
255 $this->groups
[$prevWeight] = $group;
258 // If one character has a given primary weight sequence, and a second
259 // character has a longer primary weight sequence with an initial
260 // portion equal to the first character, then remove the second
261 // character. This avoids having characters like U+A732 (double A)
262 // polluting the basic Latin sort area.
264 foreach ( $this->groups
as $weight => $group ) {
265 if ( preg_match( '/(\.[0-9A-F]*)\./', $weight, $m ) ) {
266 if ( isset( $this->groups
[$m[1]] ) ) {
267 unset( $this->groups
[$weight] );
272 ksort( $this->groups
, SORT_STRING
);
274 // Identify the header character in each group
277 $tertiaryCollator = new Collator( 'root' );
278 $primaryCollator = new Collator( 'root' );
279 $primaryCollator->setStrength( Collator
::PRIMARY
);
281 foreach ( $this->groups
as $weight => $group ) {
282 $uncomposedChars = [];
284 foreach ( $group as $cp ) {
285 if ( isset( $goodTertiaryChars[$cp] ) ) {
288 if ( !isset( $this->mappedChars
[$cp] ) ) {
289 $uncomposedChars[] = $cp;
292 $x = array_intersect( $goodChars, $uncomposedChars );
294 $x = $uncomposedChars;
300 // Use ICU to pick the lowest sorting character in the selection
301 $tertiaryCollator->sort( $x );
304 $char = UtfNormal\Utils
::codepointToUtf8( $cp );
305 $headerChars[] = $char;
306 if ( $primaryCollator->compare( $char, $prevChar ) <= 0 ) {
311 if ( $this->debugOutFile
) {
312 fwrite( $this->debugOutFile
, sprintf( "%05X %s %s (%s)\n", $cp, $weight, $char,
313 implode( ' ', array_map( 'UtfNormal\Utils::codepointToUtf8', $group ) ) ) );
317 print "Out of order: $numOutOfOrder / " . count( $headerChars ) . "\n";
320 $writer = new StaticArrayWriter();
322 "$IP/includes/collation/data/first-letters-root.php",
323 $writer->create( $headerChars, 'File created by generateCollationData.php' )
325 echo "first-letters-root: file written.\n";
335 public $currentBlock;
337 function __construct( $fileName ) {
338 $this->fileName
= $fileName;
341 public function readChars( $callback ) {
343 $this->currentBlock
= reset( $this->blocks
);
344 $xml = $this->open();
345 $this->callback
= $callback;
347 while ( $xml->name
!== 'repertoire' && $xml->next() );
349 while ( $xml->read() ) {
350 if ( $xml->nodeType
== XMLReader
::ELEMENT
) {
351 if ( $xml->name
=== 'group' ) {
352 $this->groupAttrs
= $this->readAttributes();
353 } elseif ( $xml->name
=== 'char' ) {
356 } elseif ( $xml->nodeType
=== XMLReader
::END_ELEMENT
) {
357 if ( $xml->name
=== 'group' ) {
358 $this->groupAttrs
= [];
365 protected function open() {
366 $this->xml
= new XMLReader
;
367 $this->xml
->open( $this->fileName
);
369 throw new MWException( __METHOD__
. ": unable to open {$this->fileName}" );
371 while ( $this->xml
->name
!== 'ucd' && $this->xml
->read() );
378 * Read the attributes of the current element node and return them
382 protected function readAttributes() {
384 while ( $this->xml
->moveToNextAttribute() ) {
385 $attrs[$this->xml
->name
] = $this->xml
->value
;
391 protected function handleChar() {
392 $attrs = $this->readAttributes() +
$this->groupAttrs
;
393 if ( isset( $attrs['cp'] ) ) {
394 $first = $last = hexdec( $attrs['cp'] );
396 $first = hexdec( $attrs['first-cp'] );
397 $last = hexdec( $attrs['last-cp'] );
398 unset( $attrs['first-cp'] );
399 unset( $attrs['last-cp'] );
402 for ( $cp = $first; $cp <= $last; $cp++
) {
403 $hexCp = sprintf( "%04X", $cp );
404 foreach ( [ 'na', 'na1' ] as $nameProp ) {
405 if ( isset( $attrs[$nameProp] ) ) {
406 $attrs[$nameProp] = str_replace( '#', $hexCp, $attrs[$nameProp] );
410 while ( $this->currentBlock
) {
411 if ( $cp < $this->currentBlock
[0] ) {
413 } elseif ( $cp <= $this->currentBlock
[1] ) {
414 $attrs['block'] = key( $this->blocks
);
417 $this->currentBlock
= next( $this->blocks
);
421 $attrs['cp'] = $hexCp;
422 call_user_func( $this->callback
, $attrs );
426 public function getBlocks() {
427 if ( $this->blocks
) {
428 return $this->blocks
;
431 $xml = $this->open();
432 while ( $xml->name
!== 'blocks' && $xml->read() );
434 while ( $xml->read() ) {
435 if ( $xml->nodeType
== XMLReader
::ELEMENT
) {
436 if ( $xml->name
=== 'block' ) {
437 $attrs = $this->readAttributes();
438 $first = hexdec( $attrs['first-cp'] );
439 $last = hexdec( $attrs['last-cp'] );
440 $this->blocks
[$attrs['name']] = [ $first, $last ];
446 return $this->blocks
;
450 $maintClass = GenerateCollationData
::class;
451 require_once RUN_MAINTENANCE_IF_MAIN
;