From: jenkins-bot <jenkins-bot@gerrit.wikimedia.org>
Date: Tue, 10 Jul 2018 08:55:56 +0000 (+0000)
Subject: Merge "Fix the bug for dates between 1912 and 1941 in Thai language"
X-Git-Tag: 1.34.0-rc.0~4832
X-Git-Url: http://git.cyclocoop.org/%24image?a=commitdiff_plain;h=8c96aec32cffaab96b2bd9dca206a99a13640545;hp=-c;p=lhc%2Fweb%2Fwiklou.git

Merge "Fix the bug for dates between 1912 and 1941 in Thai language"
---

8c96aec32cffaab96b2bd9dca206a99a13640545
diff --combined languages/Language.php
index deee2bce77,0941e3dc34..8373ffcc22
--- a/languages/Language.php
+++ b/languages/Language.php
@@@ -154,9 -154,9 +154,9 @@@ class Language 
  	/**
  	 * Unicode directional formatting characters, for embedBidi()
  	 */
 -	static private $lre = "\xE2\x80\xAA"; // U+202A LEFT-TO-RIGHT EMBEDDING
 -	static private $rle = "\xE2\x80\xAB"; // U+202B RIGHT-TO-LEFT EMBEDDING
 -	static private $pdf = "\xE2\x80\xAC"; // U+202C POP DIRECTIONAL FORMATTING
 +	static private $lre = "\u{202A}"; // U+202A LEFT-TO-RIGHT EMBEDDING
 +	static private $rle = "\u{202B}"; // U+202B RIGHT-TO-LEFT EMBEDDING
 +	static private $pdf = "\u{202C}"; // U+202C POP DIRECTIONAL FORMATTING
  
  	/**
  	 * Directionality test regex for embedBidi(). Matches the first strong directionality codepoint:
@@@ -188,7 -188,9 +188,7 @@@
  		}
  
  		// get the language object to process
 -		$langObj = isset( self::$mLangObjCache[$code] )
 -			? self::$mLangObjCache[$code]
 -			: self::newFromCode( $code );
 +		$langObj = self::$mLangObjCache[$code] ?? self::newFromCode( $code );
  
  		// merge the language object in to get it up front in the cache
  		self::$mLangObjCache = array_merge( [ $code => $langObj ], self::$mLangObjCache );
@@@ -220,8 -222,7 +220,8 @@@
  
  		// Check if there is a language class for the code
  		$class = self::classFromCode( $code, $fallback );
 -		if ( class_exists( $class ) ) {
 +		// LanguageCode does not inherit Language
 +		if ( class_exists( $class ) && is_a( $class, 'Language', true ) ) {
  			$lang = new $class;
  			return $lang;
  		}
@@@ -272,7 -273,7 +272,7 @@@
  	 * language, script or variant codes actually exist in the repositories.
  	 *
  	 * Based on regexes by Mark Davis of the Unicode Consortium:
 -	 * http://unicode.org/repos/cldr/trunk/tools/java/org/unicode/cldr/util/data/langtagRegex.txt
 +	 * https://www.unicode.org/repos/cldr/trunk/tools/java/org/unicode/cldr/util/data/langtagRegex.txt
  	 *
  	 * @param string $code
  	 * @param bool $lenient Whether to allow '_' as separator. The default is only '-'.
@@@ -541,7 -542,7 +541,7 @@@
  	 */
  	public function getNsText( $index ) {
  		$ns = $this->getNamespaces();
 -		return isset( $ns[$index] ) ? $ns[$index] : false;
 +		return $ns[$index] ?? false;
  	}
  
  	/**
@@@ -576,7 -577,7 +576,7 @@@
  		$ns = $wgExtraGenderNamespaces +
  			(array)self::$dataCache->getItem( $this->mCode, 'namespaceGenderAliases' );
  
 -		return isset( $ns[$index][$gender] ) ? $ns[$index][$gender] : $this->getNsText( $index );
 +		return $ns[$index][$gender] ?? $this->getNsText( $index );
  	}
  
  	/**
@@@ -612,7 -613,7 +612,7 @@@
  	function getLocalNsIndex( $text ) {
  		$lctext = $this->lc( $text );
  		$ids = $this->getNamespaceIds();
 -		return isset( $ids[$lctext] ) ? $ids[$lctext] : false;
 +		return $ids[$lctext] ?? false;
  	}
  
  	/**
@@@ -699,7 -700,7 +699,7 @@@
  			return $ns;
  		}
  		$ids = $this->getNamespaceIds();
 -		return isset( $ids[$lctext] ) ? $ids[$lctext] : false;
 +		return $ids[$lctext] ?? false;
  	}
  
  	/**
@@@ -791,12 -792,12 +791,12 @@@
  	/**
  	 * Get an array of language names, indexed by code.
  	 * @param null|string $inLanguage Code of language in which to return the names
 -	 *		Use null for autonyms (native names)
 +	 * 		Use null for autonyms (native names)
  	 * @param string $include One of:
 -	 *		'all' all available languages
 -	 *		'mw' only if the language is defined in MediaWiki or wgExtraLanguageNames (default)
 -	 *		'mwfile' only if the language is in 'mw' *and* has a message file
 -	 * @return array Language code => language name
 +	 * 		'all' all available languages
 +	 * 		'mw' only if the language is defined in MediaWiki or wgExtraLanguageNames (default)
 +	 * 		'mwfile' only if the language is in 'mw' *and* has a message file
 +	 * @return array Language code => language name (sorted by key)
  	 * @since 1.20
  	 */
  	public static function fetchLanguageNames( $inLanguage = null, $include = 'mw' ) {
@@@ -822,7 -823,7 +822,7 @@@
  	 *		'all' all available languages
  	 *		'mw' only if the language is defined in MediaWiki or wgExtraLanguageNames (default)
  	 *		'mwfile' only if the language is in 'mw' *and* has a message file
 -	 * @return array Language code => language name
 +	 * @return array Language code => language name (sorted by key)
  	 */
  	private static function fetchLanguageNamesUncached( $inLanguage = null, $include = 'mw' ) {
  		global $wgExtraLanguageNames, $wgUsePigLatinVariant;
@@@ -1090,7 -1091,7 +1090,7 @@@
  	 * @param string $ts 14-character timestamp
  	 *      YYYYMMDDHHMMSS
  	 *      01234567890123
 -	 * @param DateTimeZone $zone Timezone of $ts
 +	 * @param DateTimeZone|null $zone Timezone of $ts
  	 * @param int &$ttl The amount of time (in seconds) the output may be cached for.
  	 * Only makes sense if $ts is the current time.
  	 * @todo handling of "o" format character for Iranian, Hebrew, Hijri & Thai?
@@@ -1883,6 -1884,14 +1883,14 @@@
  			# Add 543 years to the Gregorian calendar
  			# Months and days are identical
  			$gy_offset = $gy + 543;
+ 			# fix for dates between 1912 and 1941
+ 			# https://en.wikipedia.org/?oldid=836596673#New_year
+ 			if ( $gy >= 1912 && $gy <= 1940 ) {
+ 				if ( $gm <= 3 ) {
+ 					$gy_offset--;
+ 				}
+ 				$gm = ( $gm - 3 ) % 12;
+ 			}
  		} elseif ( ( !strcmp( $cName, 'minguo' ) ) || !strcmp( $cName, 'juche' ) ) {
  			# Minguo dates
  			# Deduct 1911 years from the Gregorian calendar
@@@ -1950,8 -1959,8 +1958,8 @@@
  	 * Gets directionality of the first strongly directional codepoint, for embedBidi()
  	 *
  	 * This is the rule the BIDI algorithm uses to determine the directionality of
 -	 * paragraphs ( http://unicode.org/reports/tr9/#The_Paragraph_Level ) and
 -	 * FSI isolates ( http://unicode.org/reports/tr9/#Explicit_Directional_Isolates ).
 +	 * paragraphs ( https://www.unicode.org/reports/tr9/#The_Paragraph_Level ) and
 +	 * FSI isolates ( https://www.unicode.org/reports/tr9/#Explicit_Directional_Isolates ).
  	 *
  	 * TODO: Does not handle BIDI control characters inside the text.
  	 * TODO: Does not handle unallocated characters.
@@@ -2916,33 -2925,33 +2924,33 @@@
  			if ( $code < 0xac00 || 0xd7a4 <= $code ) {
  				return $matches[1];
  			} elseif ( $code < 0xb098 ) {
 -				return "\xe3\x84\xb1";
 +				return "\u{3131}";
  			} elseif ( $code < 0xb2e4 ) {
 -				return "\xe3\x84\xb4";
 +				return "\u{3134}";
  			} elseif ( $code < 0xb77c ) {
 -				return "\xe3\x84\xb7";
 +				return "\u{3137}";
  			} elseif ( $code < 0xb9c8 ) {
 -				return "\xe3\x84\xb9";
 +				return "\u{3139}";
  			} elseif ( $code < 0xbc14 ) {
 -				return "\xe3\x85\x81";
 +				return "\u{3141}";
  			} elseif ( $code < 0xc0ac ) {
 -				return "\xe3\x85\x82";
 +				return "\u{3142}";
  			} elseif ( $code < 0xc544 ) {
 -				return "\xe3\x85\x85";
 +				return "\u{3145}";
  			} elseif ( $code < 0xc790 ) {
 -				return "\xe3\x85\x87";
 +				return "\u{3147}";
  			} elseif ( $code < 0xcc28 ) {
 -				return "\xe3\x85\x88";
 +				return "\u{3148}";
  			} elseif ( $code < 0xce74 ) {
 -				return "\xe3\x85\x8a";
 +				return "\u{314A}";
  			} elseif ( $code < 0xd0c0 ) {
 -				return "\xe3\x85\x8b";
 +				return "\u{314B}";
  			} elseif ( $code < 0xd30c ) {
 -				return "\xe3\x85\x8c";
 +				return "\u{314C}";
  			} elseif ( $code < 0xd558 ) {
 -				return "\xe3\x85\x8d";
 +				return "\u{314D}";
  			} else {
 -				return "\xe3\x85\x8e";
 +				return "\u{314E}";
  			}
  		} else {
  			return '';
@@@ -2953,7 -2962,6 +2961,7 @@@
  	 * @deprecated No-op since 1.28
  	 */
  	function initEncoding() {
 +		wfDeprecated( __METHOD__, '1.28' );
  		// No-op.
  	}
  
@@@ -2963,7 -2971,6 +2971,7 @@@
  	 * @deprecated No-op since 1.28
  	 */
  	function recodeForEdit( $s ) {
 +		wfDeprecated( __METHOD__, '1.28' );
  		return $s;
  	}
  
@@@ -2973,7 -2980,6 +2981,7 @@@
  	 * @deprecated No-op since 1.28
  	 */
  	function recodeInput( $s ) {
 +		wfDeprecated( __METHOD__, '1.28' );
  		return $s;
  	}
  
@@@ -2992,8 -2998,8 +3000,8 @@@
  		global $wgAllUnicodeFixes;
  		$s = UtfNormal\Validator::cleanUp( $s );
  		if ( $wgAllUnicodeFixes ) {
 -			$s = $this->transformUsingPairFile( 'normalize-ar.ser', $s );
 -			$s = $this->transformUsingPairFile( 'normalize-ml.ser', $s );
 +			$s = $this->transformUsingPairFile( 'normalize-ar.php', $s );
 +			$s = $this->transformUsingPairFile( 'normalize-ml.php', $s );
  		}
  
  		return $s;
@@@ -3013,10 -3019,12 +3021,10 @@@
  	 * @throws MWException
  	 * @return string
  	 */
 -	function transformUsingPairFile( $file, $string ) {
 +	protected function transformUsingPairFile( $file, $string ) {
  		if ( !isset( $this->transformData[$file] ) ) {
 -			$data = wfGetPrecompiledData( $file );
 -			if ( $data === false ) {
 -				throw new MWException( __METHOD__ . ": The transformation file $file is missing" );
 -			}
 +			global $IP;
 +			$data = require "$IP/languages/data/{$file}";
  			$this->transformData[$file] = new ReplacementArray( $data );
  		}
  		return $this->transformData[$file]->replace( $string );
@@@ -3092,8 -3100,8 +3100,8 @@@
  	 * @return string
  	 */
  	function getDirMark( $opposite = false ) {
 -		$lrm = "\xE2\x80\x8E"; # LEFT-TO-RIGHT MARK, commonly abbreviated LRM
 -		$rlm = "\xE2\x80\x8F"; # RIGHT-TO-LEFT MARK, commonly abbreviated RLM
 +		$lrm = "\u{200E}"; # LEFT-TO-RIGHT MARK, commonly abbreviated LRM
 +		$rlm = "\u{200F}"; # RIGHT-TO-LEFT MARK, commonly abbreviated RLM
  		if ( $opposite ) {
  			return $this->isRTL() ? $lrm : $rlm;
  		}
@@@ -3156,7 -3164,7 +3164,7 @@@
  			return;
  		}
  		$this->mMagicHookDone = true;
 -		Hooks::run( 'LanguageGetMagic', [ &$this->mMagicExtensions, $this->getCode() ] );
 +		Hooks::run( 'LanguageGetMagic', [ &$this->mMagicExtensions, $this->getCode() ], '1.16' );
  	}
  
  	/**
@@@ -3212,7 -3220,7 +3220,7 @@@
  			$this->mExtendedSpecialPageAliases =
  				self::$dataCache->getItem( $this->mCode, 'specialPageAliases' );
  			Hooks::run( 'LanguageGetSpecialPageAliases',
 -				[ &$this->mExtendedSpecialPageAliases, $this->getCode() ] );
 +				[ &$this->mExtendedSpecialPageAliases, $this->getCode() ], '1.16' );
  		}
  
  		return $this->mExtendedSpecialPageAliases;
@@@ -3485,7 -3493,7 +3493,7 @@@
  	 * @param int $length Maximum length (including ellipsis)
  	 * @param string $ellipsis String to append to the truncated text
  	 * @param bool $adjustLength Subtract length of ellipsis from $length.
 -	 *	$adjustLength was introduced in 1.18, before that behaved as if false.
 +	 * 	$adjustLength was introduced in 1.18, before that behaved as if false.
  	 * @return string
  	 */
  	function truncate( $string, $length, $ellipsis = '...', $adjustLength = true ) {
@@@ -3558,7 -3566,7 +3566,7 @@@
  	 * @return string
  	 */
  	private function truncateInternal(
 -		$string, $length, $ellipsis = '...', $adjustLength = true, $measureLength, $getSubstring
 +		$string, $length, $ellipsis, $adjustLength, $measureLength, $getSubstring
  	) {
  		if ( !is_callable( $measureLength ) || !is_callable( $getSubstring ) ) {
  			throw new InvalidArgumentException( 'Invalid callback provided' );
@@@ -3962,7 -3970,7 +3970,7 @@@
  		if ( $gender === 'female' ) {
  			return $forms[1];
  		}
 -		return isset( $forms[2] ) ? $forms[2] : $forms[0];
 +		return $forms[2] ?? $forms[0];
  	}
  
  	/**
@@@ -4076,7 -4084,7 +4084,7 @@@
  	 * match up with it.
  	 *
  	 * @param string $str The validated block duration in English
 -	 * @param User $user User object to use timezone from or null for $wgUser
 +	 * @param User|null $user User object to use timezone from or null for $wgUser
  	 * @param int $now Current timestamp, for formatting relative block durations
  	 * @return string Somehow translated block duration
  	 * @see LanguageFi.php for example implementation
@@@ -4325,18 -4333,13 +4333,18 @@@
  	 * the "raw" tag (-{R| }-) to prevent conversion.
  	 *
  	 * This function is called "markNoConversion" for historical
 -	 * reasons.
 +	 * reasons *BUT DIFFERS SIGNIFICANTLY* from
 +	 * LanguageConverter::markNoConversion(), with which it is easily
 +	 * confused.
  	 *
  	 * @param string $text Text to be used for external link
  	 * @param bool $noParse Wrap it without confirming it's a real URL first
  	 * @return string The tagged text
 +	 * @deprecated since 1.32, use LanguageConverter::markNoConversion()
 +	 *  instead.
  	 */
  	public function markNoConversion( $text, $noParse = false ) {
 +		wfDeprecated( __METHOD__, '1.32' );
  		// Excluding protocal-relative URLs may avoid many false positives.
  		if ( $noParse || preg_match( '/^(?:' . wfUrlProtocolsWithoutProtRel() . ')/', $text ) ) {
  			return $this->mConverter->markNoConversion( $text );
@@@ -4400,7 -4403,7 +4408,7 @@@
  	 * @return bool
  	 */
  	public function equals( Language $lang ) {
 -		return $lang->getCode() === $this->mCode;
 +		return $lang === $this || $lang->getCode() === $this->mCode;
  	}
  
  	/**
@@@ -4480,7 -4483,7 +4488,7 @@@
  	 * @throws MWException
  	 * @return string $prefix . $mangledCode . $suffix
  	 */
 -	public static function getFileName( $prefix = 'Language', $code, $suffix = '.php' ) {
 +	public static function getFileName( $prefix, $code, $suffix = '.php' ) {
  		if ( !self::isValidBuiltInCode( $code ) ) {
  			throw new MWException( "Invalid language code \"$code\"" );
  		}
diff --combined tests/phpunit/languages/LanguageTest.php
index 7e29c9292d,04ffdabfb0..35bb1f057f
--- a/tests/phpunit/languages/LanguageTest.php
+++ b/tests/phpunit/languages/LanguageTest.php
@@@ -1029,6 -1029,13 +1029,13 @@@ class LanguageTest extends LanguageClas
  				'2555',
  				'Thai year'
  			],
+ 			[
+ 				'xkY',
+ 				'19410101090705',
+ 				'2484',
+ 				'2484',
+ 				'Thai year'
+ 			],
  			[
  				'xoY',
  				'20120102090705',
@@@ -1110,27 -1117,27 +1117,27 @@@
  				"1 gigabyte"
  			],
  			[
 -				pow( 1024, 4 ),
 +				1024 ** 4,
  				"1 TB",
  				"1 terabyte"
  			],
  			[
 -				pow( 1024, 5 ),
 +				1024 ** 5,
  				"1 PB",
  				"1 petabyte"
  			],
  			[
 -				pow( 1024, 6 ),
 +				1024 ** 6,
  				"1 EB",
  				"1,024 exabyte"
  			],
  			[
 -				pow( 1024, 7 ),
 +				1024 ** 7,
  				"1 ZB",
  				"1 zetabyte"
  			],
  			[
 -				pow( 1024, 8 ),
 +				1024 ** 8,
  				"1 YB",
  				"1 yottabyte"
  			],
@@@ -1173,37 -1180,37 +1180,37 @@@
  				"1 megabit per second"
  			],
  			[
 -				pow( 10, 9 ),
 +				10 ** 9,
  				"1 Gbps",
  				"1 gigabit per second"
  			],
  			[
 -				pow( 10, 12 ),
 +				10 ** 12,
  				"1 Tbps",
  				"1 terabit per second"
  			],
  			[
 -				pow( 10, 15 ),
 +				10 ** 15,
  				"1 Pbps",
  				"1 petabit per second"
  			],
  			[
 -				pow( 10, 18 ),
 +				10 ** 18,
  				"1 Ebps",
  				"1 exabit per second"
  			],
  			[
 -				pow( 10, 21 ),
 +				10 ** 21,
  				"1 Zbps",
  				"1 zetabit per second"
  			],
  			[
 -				pow( 10, 24 ),
 +				10 ** 24,
  				"1 Ybps",
  				"1 yottabit per second"
  			],
  			[
 -				pow( 10, 27 ),
 +				10 ** 27,
  				"1,000 Ybps",
  				"1,000 yottabits per second"
  			],
@@@ -1593,9 -1600,9 +1600,9 @@@
  	 * @covers Language::embedBidi()
  	 */
  	public function testEmbedBidi() {
 -		$lre = "\xE2\x80\xAA"; // U+202A LEFT-TO-RIGHT EMBEDDING
 -		$rle = "\xE2\x80\xAB"; // U+202B RIGHT-TO-LEFT EMBEDDING
 -		$pdf = "\xE2\x80\xAC"; // U+202C POP DIRECTIONAL FORMATTING
 +		$lre = "\u{202A}"; // U+202A LEFT-TO-RIGHT EMBEDDING
 +		$rle = "\u{202B}"; // U+202B RIGHT-TO-LEFT EMBEDDING
 +		$pdf = "\u{202C}"; // U+202C POP DIRECTIONAL FORMATTING
  		$lang = $this->getLang();
  		$this->assertEquals(
  			'123',