Add a way to parse search string to extract keywords

This feature is not in use at the moment, but it will be handy to reorganize the query building process. It allows to have more than one keyword in the search box.
Full tests are available as well.

It probably needs a refactoring later, but I think this is the first step to make the application full object oriented and testable.

app/Models/Context.php

@@ -301,4 +301,153 @@ class FreshRSS_Context {
 		}
 		return false;
 	}
+
+	/**
+	 * Parse search string to extract the different keywords.
+	 *
+	 * @return array
+	 */
+	public function parseSearch() {
+		$search = self::$search;
+		$intitle = $this->parseIntitleSearch($search);
+		$author = $this->parseAuthorSearch($intitle['string']);
+		$inurl = $this->parseInurlSearch($author['string']);
+		$pubdate = $this->parsePubdateSearch($inurl['string']);
+		$date = $this->parseDateSearch($pubdate['string']);
+
+		$remaining = array();
+		$remaining_search = trim($date['string']);
+		if (strcmp($remaining_search, '') != 0) {
+			$remaining['search'] = $remaining_search;
+		}
+
+		return array_merge($intitle['search'], $author['search'], $inurl['search'], $date['search'], $pubdate['search'], $remaining);
+	}
+
+	/**
+	 * Parse the search string to find intitle keyword and the search related
+	 * to it.
+	 * The search is the first word following the keyword.
+	 * It returns an array containing the matched string and the search.
+	 *
+	 * @param string $search
+	 * @return array
+	 */
+	private function parseIntitleSearch($search) {
+		if (preg_match('/intitle:(?P<delim>[\'"])(?P<search>.*)(?P=delim)/U', $search, $matches)) {
+			return array(
+			    'string' => str_replace($matches[0], '', $search),
+			    'search' => array('intitle' => $matches['search']),
+			);
+		}
+		if (preg_match('/intitle:(?P<search>\w*)/', $search, $matches)) {
+			return array(
+			    'string' => str_replace($matches[0], '', $search),
+			    'search' => array('intitle' => $matches['search']),
+			);
+		}
+		return array(
+		    'string' => $search,
+		    'search' => array(),
+		);
+	}
+
+	/**
+	 * Parse the search string to find author keyword and the search related
+	 * to it.
+	 * The search is the first word following the keyword except when using
+	 * a delimiter. Supported delimiters are single quote (') and double
+	 * quotes (").
+	 * It returns an array containing the matched string and the search.
+	 *
+	 * @param string $search
+	 * @return array
+	 */
+	private function parseAuthorSearch($search) {
+		if (preg_match('/author:(?P<delim>[\'"])(?P<search>.*)(?P=delim)/U', $search, $matches)) {
+			return array(
+			    'string' => str_replace($matches[0], '', $search),
+			    'search' => array('author' => $matches['search']),
+			);
+		}
+		if (preg_match('/author:(?P<search>\w*)/', $search, $matches)) {
+			return array(
+			    'string' => str_replace($matches[0], '', $search),
+			    'search' => array('author' => $matches['search']),
+			);
+		}
+		return array(
+		    'string' => $search,
+		    'search' => array(),
+		);
+	}
+
+	/**
+	 * Parse the search string to find inurl keyword and the search related
+	 * to it.
+	 * The search is the first word following the keyword except.
+	 * It returns an array containing the matched string and the search.
+	 *
+	 * @param string $search
+	 * @return array
+	 */
+	private function parseInurlSearch($search) {
+		if (preg_match('/inurl:(?P<search>[^\s]*)/', $search, $matches)) {
+			return array(
+			    'string' => str_replace($matches[0], '', $search),
+			    'search' => array('inurl' => $matches['search']),
+			);
+		}
+		return array(
+		    'string' => $search,
+		    'search' => array(),
+		);
+	}
+
+	/**
+	 * Parse the search string to find date keyword and the search related
+	 * to it.
+	 * The search is the first word following the keyword.
+	 * It returns an array containing the matched string and the search.
+	 *
+	 * @param string $search
+	 * @return array
+	 */
+	private function parseDateSearch($search) {
+		if (preg_match('/date:(?P<search>[^\s]*)/', $search, $matches)) {
+			list($min_date, $max_date) = parseDateInterval($matches['search']);
+			return array(
+			    'string' => str_replace($matches[0], '', $search),
+			    'search' => array('min_date' => $min_date, 'max_date' => $max_date),
+			);
+		}
+		return array(
+		    'string' => $search,
+		    'search' => array(),
+		);
+	}
+
+	/**
+	 * Parse the search string to find pubdate keyword and the search related
+	 * to it.
+	 * The search is the first word following the keyword.
+	 * It returns an array containing the matched string and the search.
+	 *
+	 * @param string $search
+	 * @return array
+	 */
+	private function parsePubdateSearch($search) {
+		if (preg_match('/pubdate:(?P<search>[^\s]*)/', $search, $matches)) {
+			list($min_date, $max_date) = parseDateInterval($matches['search']);
+			return array(
+			    'string' => str_replace($matches[0], '', $search),
+			    'search' => array('min_pubdate' => $min_date, 'max_pubdate' => $max_date),
+			);
+		}
+		return array(
+		    'string' => $search,
+		    'search' => array(),
+		);
+	}
+
 }

tests/app/Models/ContextTest.php

@@ -0,0 +1,241 @@
+<?php
+
+require_once(LIB_PATH . '/lib_date.php');
+
+class ContextTest extends \PHPUnit_Framework_TestCase {
+
+	private $context;
+
+	public function setUp() {
+		$this->context = new FreshRSS_Context();
+	}
+
+	public function testParseSearch_whenEmpty_returnsEmptyArray() {
+		$this->assertCount(0, $this->context->parseSearch());
+	}
+
+	/**
+	 * @dataProvider provideMultipleKeywordSearch
+	 * @param string $search
+	 * @param string $expected_values
+	 */
+	public function testParseSearch_whenMultipleKeywords_returnArrayWithMultipleValues($search, $expected_values) {
+		FreshRSS_Context::$search = $search;
+		$parsed_search = $this->context->parseSearch();
+		$this->assertEquals($expected_values, $parsed_search);
+	}
+
+	/**
+	 * @return array
+	 */
+	public function provideMultipleKeywordSearch() {
+		return array(
+		    array(
+			'intitle:word1 author:word2',
+			array(
+			    'intitle' => 'word1',
+			    'author' => 'word2',
+			),
+		    ),
+		    array(
+			'author:word2 intitle:word1',
+			array(
+			    'intitle' => 'word1',
+			    'author' => 'word2',
+			),
+		    ),
+		    array(
+			'author:word1 inurl:word2',
+			array(
+			    'author' => 'word1',
+			    'inurl' => 'word2',
+			),
+		    ),
+		    array(
+			'inurl:word2 author:word1',
+			array(
+			    'author' => 'word1',
+			    'inurl' => 'word2',
+			),
+		    ),
+		    array(
+			'date:2008-01-01/2008-02-01 pubdate:2007-01-01/2007-02-01',
+			array(
+			    'min_date' => '1199163600',
+			    'max_date' => '1201928399',
+			    'min_pubdate' => '1167627600',
+			    'max_pubdate' => '1170392399',
+			),
+		    ),
+		    array(
+			'pubdate:2007-01-01/2007-02-01 date:2008-01-01/2008-02-01',
+			array(
+			    'min_date' => '1199163600',
+			    'max_date' => '1201928399',
+			    'min_pubdate' => '1167627600',
+			    'max_pubdate' => '1170392399',
+			),
+		    ),
+		    array(
+			'inurl:word1 author:word2 intitle:word3 pubdate:2007-01-01/2007-02-01 date:2008-01-01/2008-02-01 hello world',
+			array(
+			    'inurl' => 'word1',
+			    'author' => 'word2',
+			    'intitle' => 'word3',
+			    'min_date' => '1199163600',
+			    'max_date' => '1201928399',
+			    'min_pubdate' => '1167627600',
+			    'max_pubdate' => '1170392399',
+			    'search' => 'hello world',
+			),
+		    ),
+		);
+	}
+
+	/**
+	 * @dataProvider provideIntitleSearch
+	 * @param string $search
+	 * @param string $expected_value
+	 */
+	public function testParseSearch_whenIntitleKeyword_returnArrayWithIntitleValue($search, $expected_value) {
+		FreshRSS_Context::$search = $search;
+		$parsed_search = $this->context->parseSearch();
+		$this->assertEquals($expected_value, $parsed_search['intitle']);
+	}
+
+	/**
+	 * @return array
+	 */
+	public function provideIntitleSearch() {
+		return array(
+		    array('intitle:word1', 'word1'),
+		    array('intitle:word1 word2', 'word1'),
+		    array('intitle:"word1 word2"', 'word1 word2'),
+		    array("intitle:'word1 word2'", 'word1 word2'),
+		    array('word1 intitle:word2', 'word2'),
+		    array('word1 intitle:word2 word3', 'word2'),
+		    array('word1 intitle:"word2 word3"', 'word2 word3'),
+		    array("word1 intitle:'word2 word3'", 'word2 word3'),
+		    array('intitle:word1 intitle:word2', 'word1'),
+		    array('intitle: word1 word2', ''),
+		    array('intitle:123', '123'),
+		    array('intitle:"word1 word2" word3"', 'word1 word2'),
+		    array("intitle:'word1 word2' word3'", 'word1 word2'),
+		    array('intitle:"word1 word2\' word3"', "word1 word2' word3"),
+		    array("intitle:'word1 word2\" word3'", 'word1 word2" word3'),
+		);
+	}
+
+	/**
+	 * @dataProvider provideAuthorSearch
+	 * @param string $search
+	 * @param string $expected_value
+	 */
+	public function testParseSearch_whenAuthorKeyword_returnArrayWithAuthorValue($search, $expected_value) {
+		FreshRSS_Context::$search = $search;
+		$parsed_search = $this->context->parseSearch();
+		$this->assertEquals($expected_value, $parsed_search['author']);
+	}
+
+	/**
+	 * @return array
+	 */
+	public function provideAuthorSearch() {
+		return array(
+		    array('author:word1', 'word1'),
+		    array('author:word1 word2', 'word1'),
+		    array('author:"word1 word2"', 'word1 word2'),
+		    array("author:'word1 word2'", 'word1 word2'),
+		    array('word1 author:word2', 'word2'),
+		    array('word1 author:word2 word3', 'word2'),
+		    array('word1 author:"word2 word3"', 'word2 word3'),
+		    array("word1 author:'word2 word3'", 'word2 word3'),
+		    array('author:word1 author:word2', 'word1'),
+		    array('author: word1 word2', ''),
+		    array('author:123', '123'),
+		    array('author:"word1 word2" word3"', 'word1 word2'),
+		    array("author:'word1 word2' word3'", 'word1 word2'),
+		    array('author:"word1 word2\' word3"', "word1 word2' word3"),
+		    array("author:'word1 word2\" word3'", 'word1 word2" word3'),
+		);
+	}
+
+	/**
+	 * @dataProvider provideInurlSearch
+	 * @param string $search
+	 * @param string $expected_value
+	 */
+	public function testParseSearch_whenInurlKeyword_returnArrayWithInurlValue($search, $expected_value) {
+		FreshRSS_Context::$search = $search;
+		$parsed_search = $this->context->parseSearch();
+		$this->assertEquals($expected_value, $parsed_search['inurl']);
+	}
+
+	/**
+	 * @return array
+	 */
+	public function provideInurlSearch() {
+		return array(
+		    array('inurl:word1', 'word1'),
+		    array('inurl: word1', ''),
+		    array('inurl:123', '123'),
+		    array('inurl:word1 word2', 'word1'),
+		    array('inurl:"word1 word2"', '"word1'),
+		);
+	}
+
+	/**
+	 * @dataProvider provideDateSearch
+	 * @param string $search
+	 * @param string $expected_min_value
+	 * @param string $expected_max_value
+	 */
+	public function testParseSearch_whenDateKeyword_returnArrayWithDateValues($search, $expected_min_value, $expected_max_value) {
+		FreshRSS_Context::$search = $search;
+		$parsed_search = $this->context->parseSearch();
+		$this->assertEquals($expected_min_value, $parsed_search['min_date']);
+		$this->assertEquals($expected_max_value, $parsed_search['max_date']);
+	}
+
+	/**
+	 * @return array
+	 */
+	public function provideDateSearch() {
+		return array(
+		    array('date:2007-03-01T13:00:00Z/2008-05-11T15:30:00Z', '1172754000', '1210519800'),
+		    array('date:2007-03-01T13:00:00Z/P1Y2M10DT2H30M', '1172754000', '1210516199'),
+		    array('date:P1Y2M10DT2H30M/2008-05-11T15:30:00Z', '1172757601', '1210519800'),
+		    array('date:2007-03-01/2008-05-11', '1172725200', '1210564799'),
+		    array('date:2007-03-01/', '1172725200', ''),
+		    array('date:/2008-05-11', '', '1210564799'),
+		);
+	}
+
+	/**
+	 * @dataProvider providePubdateSearch
+	 * @param string $search
+	 * @param string $expected_min_value
+	 * @param string $expected_max_value
+	 */
+	public function testParseSearch_whenPubdateKeyword_returnArrayWithPubdateValues($search, $expected_min_value, $expected_max_value) {
+		FreshRSS_Context::$search = $search;
+		$parsed_search = $this->context->parseSearch();
+		$this->assertEquals($expected_min_value, $parsed_search['min_pubdate']);
+		$this->assertEquals($expected_max_value, $parsed_search['max_pubdate']);
+	}
+
+	/**
+	 * @return array
+	 */
+	public function providePubdateSearch() {
+		return array(
+		    array('pubdate:2007-03-01T13:00:00Z/2008-05-11T15:30:00Z', '1172754000', '1210519800'),
+		    array('pubdate:2007-03-01T13:00:00Z/P1Y2M10DT2H30M', '1172754000', '1210516199'),
+		    array('pubdate:P1Y2M10DT2H30M/2008-05-11T15:30:00Z', '1172757601', '1210519800'),
+		    array('pubdate:2007-03-01/2008-05-11', '1172725200', '1210564799'),
+		    array('pubdate:2007-03-01/', '1172725200', ''),
+		    array('pubdate:/2008-05-11', '', '1210564799'),
+		);
+	}
+
+}