DateFormatParser::findMonthMatch() - Code Metrics - Inspection of "[WIP] Add DateFormatParser" - DataValues/Time - Measure and Improve Code Quality continuously with Scrutinizer

Completed

Pull Request — master (#83)

by no

created 2015-12-21 17:20 UTC

DateFormatParser::findMonthMatch() A

↳ Parent: DateFormatParser

Complexity

Conditions	3
Paths	3

Size

Total Lines	9
Code Lines	5

Duplication

Lines	0
Ratio	0 %

Importance

Changes	1
Bugs	0	Features	1

Metric	Value
c	1
b	0
f	1
dl	0
loc	9
rs	9.6667
cc	3
eloc	5
nc	3
nop	1

<?php

namespace ValueParsers;

use DataValues\TimeValue;

/**
 * This parser is in essence the inverse operation of Language::sprintfDate.
 *
 * @see Language::sprintfDate
 *
 * @since 0.8.1
 *
 * @licence GNU GPL v2+
 * @author Thiemo Mättig
 */
class DateFormatParser extends StringValueParser {

	const FORMAT_NAME = 'datetime';

	const OPT_DATE_FORMAT = 'dateFormat';

	/**
	 * Option for unlocalizing non-canonical digits. Must be an array of strings, mapping canonical
	 * digit characters ("1", "2" and so on, possibly including "." and ",") to localized
	 * characters.
	 */
	const OPT_DIGIT_TRANSFORM_TABLE = 'digitTransformTable';

	/**
	 * Option for localized month names. Should be a two-dimensional array, the first dimension
	 * mapping the month's numbers 1 to 12 to arrays of localized month names, possibly including
	 * full month names, genitive names and abbreviations. Can also be a one-dimensional array of
	 * strings.
	 */
	const OPT_MONTH_NAMES = 'monthNames';

	public function __construct( ParserOptions $options = null ) {
		parent::__construct( $options );

		$this->defaultOption( self::OPT_DATE_FORMAT, 'j F Y' );
		// FIXME: Should not be an option. Options should be trivial, never arrays or objects!
		$this->defaultOption( self::OPT_DIGIT_TRANSFORM_TABLE, null );
		$this->defaultOption( self::OPT_MONTH_NAMES, null );
	}

	/**
	 * @see StringValueParser::stringParse
	 *
	 * @param string $value
	 *
	 * @throws ParseException
	 * @return TimeValue
	 */
	protected function stringParse( $value ) {
		$pattern = $this->parseDateFormat( $this->getDateFormat() );

		if ( @preg_match( '<^\p{Z}*' . $pattern . '$>iu', $value, $matches )
			&& isset( $matches['year'] )
		) {
			$precision = TimeValue::PRECISION_YEAR;
			$time = array( $this->parseFormattedNumber( $matches['year'] ), 0, 0, 0, 0, 0 );

			if ( isset( $matches['month'] ) ) {
				$precision = TimeValue::PRECISION_MONTH;
				$time[1] = $this->findMonthMatch( $matches );
			}

			if ( isset( $matches['day'] ) ) {
				$precision = TimeValue::PRECISION_DAY;
				$time[2] = $this->parseFormattedNumber( $matches['day'] );
			}

			if ( isset( $matches['hour'] ) ) {
				$precision = TimeValue::PRECISION_HOUR;
				$time[3] = $this->parseFormattedNumber( $matches['hour'] );
			}

			if ( isset( $matches['minute'] ) ) {
				$precision = TimeValue::PRECISION_MINUTE;
				$time[4] = $this->parseFormattedNumber( $matches['minute'] );
			}

			if ( isset( $matches['second'] ) ) {
				$precision = TimeValue::PRECISION_SECOND;
				$time[5] = $this->parseFormattedNumber( $matches['second'] );
			}

			$timestamp = vsprintf( '+%04s-%02s-%02sT%02s:%02s:%02sZ', $time );
			return new TimeValue( $timestamp, 0, 0, 0, $precision, TimeValue::CALENDAR_GREGORIAN );
		}

		throw new ParseException( "Failed to parse $value ("
			. $this->parseFormattedNumber( $value ) . ')', $value );
	}

	/**
	 * @see Language::sprintfDate
	 *
	 * @param string $format A date format, as described in Language::sprintfDate.
	 *
	 * @return string Regular expression
	 */
	private function parseDateFormat( $format ) {
		$length = strlen( $format );
		$numberPattern = '[' . $this->getNumberCharacters() . ']';
		$pattern = '';

		for ( $p = 0; $p < $length; $p++ ) {
			$code = $format[$p];

			if ( $code === 'x' && $p < $length - 1 ) {
				$code .= $format[++$p];
			}

			if ( preg_match( '<^x[ijkmot]$>', $code ) && $p < $length - 1 ) {
				$code .= $format[++$p];
			}

			switch ( $code ) {
				case 'Y':
					$pattern .= '(?P<year>' . $numberPattern . '+)\p{Z}*';
					break;
				case 'F':
				case 'm':
				case 'M':
				case 'n':
				case 'xg':
					$pattern .= '(?P<month>' . $numberPattern . '{1,2}'
						. $this->getMonthNamesPattern()
						. ')\p{P}*\p{Z}*';
					break;
				case 'd':
				case 'j':
					$pattern .= '(?P<day>' . $numberPattern . '{1,2})\p{P}*\p{Z}*';
					break;
				case 'G':
				case 'H':
					$pattern .= '(?P<hour>' . $numberPattern . '{1,2})\p{Z}*';
					break;
				case 'i':
					$pattern .= '(?P<minute>' . $numberPattern . '{1,2})\p{Z}*';
					break;
				case 's':
					$pattern .= '(?P<second>' . $numberPattern . '{1,2})\p{Z}*';
					break;
				case '\\':
					if ( $p < $length - 1 ) {
						$pattern .= preg_quote( $format[++$p] );
					} else {
						$pattern .= '\\';
					}
					break;
				case '"':
					$endQuote = strpos( $format, '"', $p + 1 );
					if ( $endQuote !== false ) {
						$pattern .= preg_quote( substr( $format, $p + 1, $endQuote - $p - 1 ) );
						$p = $endQuote;
					} else {
						$pattern .= '"';
					}
					break;
				case 'xn':
				case 'xN':
					// We can ignore raw and raw toggle when parsing, because we always accept
					// canonical digits.
					break;
				default:
					if ( preg_match( '<^\p{P}+$>u', $format[$p] ) ) {
						$pattern .= '\p{P}*';
					} elseif ( preg_match( '<^\p{Z}+$>u', $format[$p] ) ) {
						$pattern .= '\p{Z}*';
					} else {
						$pattern .= preg_quote( $format[$p] );
					}
			}
		}

		return $pattern;
	}

	/**
	 * @return string
	 */
	private function getMonthNamesPattern() {
		$pattern = '';

		foreach ( $this->getMonthNames() as $i => $monthNames ) {
			$pattern .= '|(?P<month' . $i . '>'
				. implode( '|', array_map( 'preg_quote', (array)$monthNames ) )
				. ')';
		}

		return $pattern;
	}

	/**
	 * @param string[] $matches
	 *
	 * @return int
	 */
	private function findMonthMatch( $matches ) {
		for ( $i = 1; $i <= 12; $i++ ) {
			if ( !empty( $matches['month' . $i] ) ) {
				return $i;
			}
		}

		return $this->parseFormattedNumber( $matches['month'] );
	}

	/**
	 * @param string $number
	 *
	 * @return string
	 */
	private function parseFormattedNumber( $number ) {
		$transformTable = $this->getDigitTransformTable();

		if ( is_array( $transformTable ) ) {
			// Eliminate empty array values (bug T66347).
			$transformTable = array_filter( $transformTable );
			$number = strtr( $number, array_flip( $transformTable ) );
		}

		return $number;
	}

	/**
	 * @return string
	 */
	private function getNumberCharacters() {
		$numberCharacters = '\d';

		$transformTable = $this->getDigitTransformTable();
		if ( is_array( $transformTable ) ) {
			$numberCharacters .= preg_quote( implode( '', $transformTable ) );
		}

		return $numberCharacters;
	}

	/**
	 * @return string
	 */
	private function getDateFormat() {
		return $this->getOption( self::OPT_DATE_FORMAT );
	}

	/**
	 * @return string[]|null
	 */
	private function getDigitTransformTable() {
		return $this->getOption( self::OPT_DIGIT_TRANSFORM_TABLE );
	}

	/**
	 * @return array[]
	 */
	private function getMonthNames() {
		return $this->getOption( self::OPT_MONTH_NAMES ) ?: array();
	}

}


1			<?php
2
3			namespace ValueParsers;
4
5			use DataValues\TimeValue;
6
7			/**
8			* This parser is in essence the inverse operation of Language::sprintfDate.
9			*
10			* @see Language::sprintfDate
11			*
12			* @since 0.8.1
13			*
14			* @licence GNU GPL v2+
15			* @author Thiemo Mättig
16			*/
17			class DateFormatParser extends StringValueParser {
18
19			const FORMAT_NAME = 'datetime';
20
21			const OPT_DATE_FORMAT = 'dateFormat';
22
23			/**
24			* Option for unlocalizing non-canonical digits. Must be an array of strings, mapping canonical
25			* digit characters ("1", "2" and so on, possibly including "." and ",") to localized
26			* characters.
27			*/
28			const OPT_DIGIT_TRANSFORM_TABLE = 'digitTransformTable';
29
30			/**
31			* Option for localized month names. Should be a two-dimensional array, the first dimension
32			* mapping the month's numbers 1 to 12 to arrays of localized month names, possibly including
33			* full month names, genitive names and abbreviations. Can also be a one-dimensional array of
34			* strings.
35			*/
36			const OPT_MONTH_NAMES = 'monthNames';
37
38			public function __construct( ParserOptions $options = null ) {
39			parent::__construct( $options );
40
41			$this->defaultOption( self::OPT_DATE_FORMAT, 'j F Y' );
42			// FIXME: Should not be an option. Options should be trivial, never arrays or objects!
43			$this->defaultOption( self::OPT_DIGIT_TRANSFORM_TABLE, null );
44			$this->defaultOption( self::OPT_MONTH_NAMES, null );
45			}
46
47			/**
48			* @see StringValueParser::stringParse
49			*
50			* @param string $value
51			*
52			* @throws ParseException
53			* @return TimeValue
54			*/
55			protected function stringParse( $value ) {
56			$pattern = $this->parseDateFormat( $this->getDateFormat() );
57
58			if ( @preg_match( '<^\p{Z}*' . $pattern . '$>iu', $value, $matches )
59			&& isset( $matches['year'] )
60			) {
61			$precision = TimeValue::PRECISION_YEAR;
62			$time = array( $this->parseFormattedNumber( $matches['year'] ), 0, 0, 0, 0, 0 );
63
64			if ( isset( $matches['month'] ) ) {
65			$precision = TimeValue::PRECISION_MONTH;
66			$time[1] = $this->findMonthMatch( $matches );
67			}
68
69			if ( isset( $matches['day'] ) ) {
70			$precision = TimeValue::PRECISION_DAY;
71			$time[2] = $this->parseFormattedNumber( $matches['day'] );
72			}
73
74			if ( isset( $matches['hour'] ) ) {
75			$precision = TimeValue::PRECISION_HOUR;
76			$time[3] = $this->parseFormattedNumber( $matches['hour'] );
77			}
78
79			if ( isset( $matches['minute'] ) ) {
80			$precision = TimeValue::PRECISION_MINUTE;
81			$time[4] = $this->parseFormattedNumber( $matches['minute'] );
82			}
83
84			if ( isset( $matches['second'] ) ) {
85			$precision = TimeValue::PRECISION_SECOND;
86			$time[5] = $this->parseFormattedNumber( $matches['second'] );
87			}
88
89			$timestamp = vsprintf( '+%04s-%02s-%02sT%02s:%02s:%02sZ', $time );
90			return new TimeValue( $timestamp, 0, 0, 0, $precision, TimeValue::CALENDAR_GREGORIAN );
91			}
92
93			throw new ParseException( "Failed to parse $value ("
94			. $this->parseFormattedNumber( $value ) . ')', $value );
95			}
96
97			/**
98			* @see Language::sprintfDate
99			*
100			* @param string $format A date format, as described in Language::sprintfDate.
101			*
102			* @return string Regular expression
103			*/
104			private function parseDateFormat( $format ) {
105			$length = strlen( $format );
106			$numberPattern = '[' . $this->getNumberCharacters() . ']';
107			$pattern = '';
108
109			for ( $p = 0; $p < $length; $p++ ) {
110			$code = $format[$p];
111
112			if ( $code === 'x' && $p < $length - 1 ) {
113			$code .= $format[++$p];
114			}
115
116			if ( preg_match( '<^x[ijkmot]$>', $code ) && $p < $length - 1 ) {
117			$code .= $format[++$p];
118			}
119
120			switch ( $code ) {
121			case 'Y':
122			$pattern .= '(?P<year>' . $numberPattern . '+)\p{Z}*';
123			break;
124			case 'F':
125			case 'm':
126			case 'M':
127			case 'n':
128			case 'xg':
129			$pattern .= '(?P<month>' . $numberPattern . '{1,2}'
130			. $this->getMonthNamesPattern()
131			. ')\p{P}\p{Z}';
132			break;
133			case 'd':
134			case 'j':
135			$pattern .= '(?P<day>' . $numberPattern . '{1,2})\p{P}\p{Z}';
136			break;
137			case 'G':
138			case 'H':
139			$pattern .= '(?P<hour>' . $numberPattern . '{1,2})\p{Z}*';
140			break;
141			case 'i':
142			$pattern .= '(?P<minute>' . $numberPattern . '{1,2})\p{Z}*';
143			break;
144			case 's':
145			$pattern .= '(?P<second>' . $numberPattern . '{1,2})\p{Z}*';
146			break;
147			case '\\':
148			if ( $p < $length - 1 ) {
149			$pattern .= preg_quote( $format[++$p] );
150			} else {
151			$pattern .= '\\';
152			}
153			break;
154			case '"':
155			$endQuote = strpos( $format, '"', $p + 1 );
156			if ( $endQuote !== false ) {
157			$pattern .= preg_quote( substr( $format, $p + 1, $endQuote - $p - 1 ) );
158			$p = $endQuote;
159			} else {
160			$pattern .= '"';
161			}
162			break;
163			case 'xn':
164			case 'xN':
165			// We can ignore raw and raw toggle when parsing, because we always accept
166			// canonical digits.
167			break;
168			default:
169			if ( preg_match( '<^\p{P}+$>u', $format[$p] ) ) {
170			$pattern .= '\p{P}*';
171			} elseif ( preg_match( '<^\p{Z}+$>u', $format[$p] ) ) {
172			$pattern .= '\p{Z}*';
173			} else {
174			$pattern .= preg_quote( $format[$p] );
175			}
176			}
177			}
178
179			return $pattern;
180			}
181
182			/**
183			* @return string
184			*/
185			private function getMonthNamesPattern() {
186			$pattern = '';
187
188			foreach ( $this->getMonthNames() as $i => $monthNames ) {
189			$pattern .= '\|(?P<month' . $i . '>'
190			. implode( '\|', array_map( 'preg_quote', (array)$monthNames ) )
191			. ')';
192			}
193
194			return $pattern;
195			}
196
197			/**
198			* @param string[] $matches
199			*
200			* @return int
201			*/
202			private function findMonthMatch( $matches ) {
203			for ( $i = 1; $i <= 12; $i++ ) {
204			if ( !empty( $matches['month' . $i] ) ) {
205			return $i;
206			}
207			}
208
209			return $this->parseFormattedNumber( $matches['month'] );
210			}
211
212			/**
213			* @param string $number
214			*
215			* @return string
216			*/
217			private function parseFormattedNumber( $number ) {
218			$transformTable = $this->getDigitTransformTable();
219
220			if ( is_array( $transformTable ) ) {
221			// Eliminate empty array values (bug T66347).
222			$transformTable = array_filter( $transformTable );
223			$number = strtr( $number, array_flip( $transformTable ) );
224			}
225
226			return $number;
227			}
228
229			/**
230			* @return string
231			*/
232			private function getNumberCharacters() {
233			$numberCharacters = '\d';
234
235			$transformTable = $this->getDigitTransformTable();
236			if ( is_array( $transformTable ) ) {
237			$numberCharacters .= preg_quote( implode( '', $transformTable ) );
238			}
239
240			return $numberCharacters;
241			}
242
243			/**
244			* @return string
245			*/
246			private function getDateFormat() {
247			return $this->getOption( self::OPT_DATE_FORMAT );
248			}
249
250			/**
251			* @return string[]\|null
252			*/
253			private function getDigitTransformTable() {
254			return $this->getOption( self::OPT_DIGIT_TRANSFORM_TABLE );
255			}
256
257			/**
258			* @return array[]
259			*/
260			private function getMonthNames() {
261			return $this->getOption( self::OPT_MONTH_NAMES ) ?: array();
262			}
263
264			}
265

DataValues / Time

Pull Request — master (#83)

DateFormatParser::findMonthMatch() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like