Unlocalize months in YearMonthDayTimeParser

wmde · Feb 21, 2018 · 7ee93b7 · 7ee93b7
1 parent e923a93
commit 7ee93b7
Show file tree

Hide file tree

Showing 2 changed files with 101 additions and 6 deletions.
diff --git a/src/ValueParsers/YearMonthDayTimeParser.php b/src/ValueParsers/YearMonthDayTimeParser.php
@@ -22,6 +22,11 @@ class YearMonthDayTimeParser extends StringValueParser {
 	 */
 	private $eraParser;
 
+	/**
+	 * @var int[]
+	 */
+	private $months;
+
 	/**
 	 * @var ValueParser
 	 */
@@ -30,12 +35,19 @@ class YearMonthDayTimeParser extends StringValueParser {
 	/**
 	 * @param ValueParser|null $eraParser String parser that detects signs, "BC" suffixes and such and
 	 * returns an array with the detected sign character and the remaining value.
+	 * @param int[] $months Array mapping localized month names (possibly including full month
+	 * names, genitive names and abbreviations) to the numbers 1 to 12.
 	 * @param ParserOptions|null $options
 	 */
-	public function __construct( ValueParser $eraParser = null, ParserOptions $options = null ) {
+	public function __construct(
+		ValueParser $eraParser = null,
+		array $months = [],
+		ParserOptions $options = null
+	) {
 		parent::__construct( $options );
 
 		$this->eraParser = $eraParser ?: new EraParser();
+		$this->months = $months;
 		$this->isoTimestampParser = new IsoTimestampParser( null, $this->options );
 	}
 
@@ -69,10 +81,28 @@ protected function stringParse( $value ) {
 	 * @return string[]
 	 */
 	private function parseYearMonthDay( $value ) {
-		if ( !preg_match( '/^\D*?(-?\d+)\D+(\d+)\D+?(-?\d+)\D*$/', $value, $matches ) ) {
+		$monthPattern = '(?:\d+|'
+			. implode( '|', array_map( 'preg_quote', array_keys( $this->months ) ) )
+			. ')';
+		if ( !preg_match(
+			'<^\D*?(-?' . $monthPattern . ')\D+(' . $monthPattern . ')\D+?(-?' . $monthPattern . ')\D*$>',
+			$value,
+			$matches
+		) ) {
 			throw new ParseException( 'Can not find three numbers' );
 		}
 
+		$monthAt = null;
+		foreach ( $matches as $i => &$match ) {
+			if ( isset( $this->months[$match] ) ) {
+				if ( $monthAt !== null ) {
+					throw new ParseException( 'Two months found' );
+				}
+				$match = $this->months[$match];
+				$monthAt = $i;
+			}
+		}
+
 		// A 32 in the first spot can not be confused with anything.
 		if ( $matches[1] < 1 || $matches[1] > 31 ) {
 			if ( $matches[3] > 12 || $matches[2] == $matches[3] ) {
@@ -99,7 +129,7 @@ private function parseYearMonthDay( $value ) {
 			throw new ParseException( 'Can not identify year' );
 		}
 
-		return array( $signedYear, $month, $day );
+		return [ $signedYear, $month, $day ];
 	}
 
 	/**

diff --git a/tests/ValueParsers/YearMonthDayTimeParserTest.php b/tests/ValueParsers/YearMonthDayTimeParserTest.php
@@ -34,7 +34,16 @@ protected function getParserClass() {
 	 * @return YearMonthDayTimeParser
 	 */
 	protected function getInstance() {
-		return new YearMonthDayTimeParser();
+		return $this->getYearMonthDayTimeParser();
+	}
+
+	/**
+	 * @param int[] $months
+	 *
+	 * @return YearMonthDayTimeParser
+	 */
+	private function getYearMonthDayTimeParser( array $months = array() ) {
+		return new YearMonthDayTimeParser( null, $months );
 	}
 
 	/**
@@ -195,6 +204,62 @@ public function invalidInputProvider() {
 		return $cases;
 	}
 
+	/**
+	 * @dataProvider monthNamesProvider
+	 */
+	public function testMonthNameParsing( $value, array $months, TimeValue $expected ) {
+		$parser = $this->getYearMonthDayTimeParser( $months );
+		$this->assertTrue( $expected->equals( $parser->parse( $value ) ) );
+	}
+
+	public function monthNamesProvider() {
+		$gregorian = 'http://www.wikidata.org/entity/Q1985727';
+
+		$valid = [
+			'13.12.1999' => [
+				[],
+				'+1999-12-13T00:00:00Z',
+			],
+			'13. February 1999' => [
+				[ 'February' => 2 ],
+				'+1999-02-13T00:00:00Z',
+			],
+		];
+
+		$cases = [];
+
+		foreach ( $valid as $value => $args ) {
+			$months = $args[0];
+			$timestamp = $args[1];
+			$calendarModel = isset( $args[2] ) ? $args[2] : $gregorian;
+
+			$cases[] = [
+				(string)$value,
+				$months,
+				new TimeValue( $timestamp, 0, 0, 0, TimeValue::PRECISION_DAY, $calendarModel )
+			];
+		}
+
+		return $cases;
+	}
+
+	/**
+	 * @dataProvider invalidMonthNamesProvider
+	 */
+	public function testInvalidMonthNameParsing( $value, array $months ) {
+		$parser = $this->getYearMonthDayTimeParser( $months );
+		$this->setExpectedException( 'ValueParsers\ParseException' );
+		$parser->parse( $value );
+	}
+
+	public function invalidMonthNamesProvider() {
+		return [
+			[ '13. February 1999', [] ],
+			[ 'February. February 1999', [ 'February' => 2 ] ],
+			[ '13. Feb 1999', [ 'February' => 2 ] ],
+		];
+	}
+
 	/**
 	 * @dataProvider optionsProvider
 	 */
@@ -205,7 +270,7 @@ public function testOptions(
 		$calendarModel,
 		$precision = TimeValue::PRECISION_DAY
 	) {
-		$parser = new YearMonthDayTimeParser( null, new ParserOptions( $options ) );
+		$parser = new YearMonthDayTimeParser( null, [], new ParserOptions( $options ) );
 		$this->assertEquals(
 			new TimeValue( $timestamp, 0, 0, 0, $precision, $calendarModel ),
 			$parser->parse( $value )
@@ -276,7 +341,7 @@ public function optionsProvider() {
 	 * @dataProvider invalidOptionsProvider
 	 */
 	public function testInvalidOptions( array $options ) {
-		$parser = new YearMonthDayTimeParser( null, new ParserOptions( $options ) );
+		$parser = new YearMonthDayTimeParser( null, [], new ParserOptions( $options ) );
 		$this->setExpectedException( ParseException::class );
 		$parser->parse( '2016-01-31' );
 	}