From bedd176a5406003631da42366736fd5ebae29135 Mon Sep 17 00:00:00 2001 From: ArthurHoaro Date: Mon, 1 Feb 2016 20:33:58 +0100 Subject: [PATCH] Improved search: combine AND, exact terms and exclude search. --- application/LinkFilter.php | 60 ++++++++++++++++++++------------- tests/LinkDBTest.php | 1 + tests/LinkFilterTest.php | 40 +++++++++++++++++++--- tests/utils/ReferenceLinkDB.php | 11 +++++- 4 files changed, 83 insertions(+), 29 deletions(-) diff --git a/application/LinkFilter.php b/application/LinkFilter.php index ceb47d16..e2ef94ea 100644 --- a/application/LinkFilter.php +++ b/application/LinkFilter.php @@ -120,7 +120,9 @@ class LinkFilter * * Searches: * - in the URLs, title and description; - * - are case-insensitive. + * - are case-insensitive; + * - terms surrounded by quotes " are exact terms search. + * - terms starting with a dash - are excluded (except exact terms). * * Example: * print_r($mydb->filterFulltext('hollandais')); @@ -137,18 +139,28 @@ class LinkFilter private function filterFulltext($searchterms, $privateonly = false) { $search = mb_convert_case(html_entity_decode($searchterms), MB_CASE_LOWER, 'UTF-8'); - $explodedSearch = explode(' ', trim($search)); - $keys = array('title', 'description', 'url', 'tags'); - $found = true; - $searchExactPhrase = false; + $exactRegex = '/"([^"]+)"/'; + // Retrieve exact search terms. + preg_match_all($exactRegex, $search, $exactSearch); + $exactSearch = array_values(array_filter($exactSearch[1])); - // Check if we're using double-quotes to search for the exact string - if ($search[0] == '"' && $search[strlen($search) - 1] == '"') { - $searchExactPhrase = true; - - // Remove the double-quotes as they are not what we search for - $search = substr($search, 1, -1); + // Remove exact search terms to get AND terms search. + $explodedSearchAnd = explode(' ', trim(preg_replace($exactRegex, '', $search))); + $explodedSearchAnd = array_values(array_filter($explodedSearchAnd)); + + // Filter excluding terms and update andSearch. + $excludeSearch = array(); + $andSearch = array(); + foreach ($explodedSearchAnd as $needle) { + if ($needle[0] == '-' && strlen($needle) > 1) { + $excludeSearch[] = substr($needle, 1); + } else { + $andSearch[] = $needle; + } } + + $keys = array('title', 'description', 'url', 'tags'); + // Iterate over every stored link. foreach ($this->links as $link) { @@ -162,22 +174,22 @@ class LinkFilter // Be optimistic $found = true; - // FIXME: Find a better word for where you're searching in $haystack = mb_convert_case($link[$key], MB_CASE_LOWER, 'UTF-8'); - // When searching for the phrase, check if it's in the haystack... - if ( $searchExactPhrase && strpos($haystack, $search) !== false) { - break; + // First, we look for exact term search + for ($i = 0; $i < count($exactSearch) && $found; $i++) { + $found = strpos($haystack, $exactSearch[$i]) !== false; } - else { - // Iterate over keywords, if keyword is not found, - // no need to check for the others. We want all or nothing. - foreach($explodedSearch as $keyword) { - if(strpos($haystack, $keyword) === false) { - $found = false; - break; - } - } + + // Iterate over keywords, if keyword is not found, + // no need to check for the others. We want all or nothing. + for ($i = 0; $i < count($andSearch) && $found; $i++) { + $found = strpos($haystack, $andSearch[$i]) !== false; + } + + // Exclude terms. + for ($i = 0; $i < count($excludeSearch) && $found; $i++) { + $found = strpos($haystack, $excludeSearch[$i]) === false; } // One of the fields of the link matches, no need to check the other. diff --git a/tests/LinkDBTest.php b/tests/LinkDBTest.php index 765f771e..78f42e56 100644 --- a/tests/LinkDBTest.php +++ b/tests/LinkDBTest.php @@ -278,6 +278,7 @@ class LinkDBTest extends PHPUnit_Framework_TestCase 'stallman' => 1, 'free' => 1, '-exclude' => 1, + 'stuff' => 2, ), self::$publicLinkDB->allTags() ); diff --git a/tests/LinkFilterTest.php b/tests/LinkFilterTest.php index 164af0d4..4d754d25 100644 --- a/tests/LinkFilterTest.php +++ b/tests/LinkFilterTest.php @@ -27,7 +27,7 @@ class LinkFilterTest extends PHPUnit_Framework_TestCase public function testFilter() { $this->assertEquals( - 6, + 7, count(self::$linkFilter->filter('', '')) ); @@ -222,7 +222,7 @@ class LinkFilterTest extends PHPUnit_Framework_TestCase ); $this->assertEquals( - 2, + 3, count(self::$linkFilter->filter(LinkFilter::$FILTER_TEXT, '"free software"')) ); } @@ -250,11 +250,43 @@ class LinkFilterTest extends PHPUnit_Framework_TestCase public function testFilterFullTextMixed() { $this->assertEquals( - 2, + 3, count(self::$linkFilter->filter(LinkFilter::$FILTER_TEXT, 'free software')) ); } + /** + * Full-text search - test exclusion with '-'. + */ + public function testExcludeSearch() + { + $this->assertEquals( + 1, + count(self::$linkFilter->filter(LinkFilter::$FILTER_TEXT, 'free -software')) + ); + + $this->assertEquals( + 7, + count(self::$linkFilter->filter(LinkFilter::$FILTER_TEXT, '-software')) + ); + } + + /** + * Full-text search - test AND, exact terms and exclusion combined. + */ + public function testMultiSearch() + { + $this->assertEquals( + 2, + count(self::$linkFilter->filter(LinkFilter::$FILTER_TEXT, '"Free Software " stallman "read this"')) + ); + + $this->assertEquals( + 1, + count(self::$linkFilter->filter(LinkFilter::$FILTER_TEXT, '"free software " stallman "read this" -beard')) + ); + } + /** * Tag search with exclusion. */ @@ -266,7 +298,7 @@ class LinkFilterTest extends PHPUnit_Framework_TestCase ); $this->assertEquals( - 5, + 6, count(self::$linkFilter->filter(LinkFilter::$FILTER_TAG, '-free')) ); } diff --git a/tests/utils/ReferenceLinkDB.php b/tests/utils/ReferenceLinkDB.php index da3e8c65..b64b58bf 100644 --- a/tests/utils/ReferenceLinkDB.php +++ b/tests/utils/ReferenceLinkDB.php @@ -16,12 +16,21 @@ class ReferenceLinkDB $this->addLink( 'Free as in Freedom 2.0', 'https://static.fsf.org/nosvn/faif-2.0.pdf', - 'Richard Stallman and the Free Software Revolution', + 'Richard Stallman and the Free Software Revolution. Read this.', 0, '20150310_114633', 'free gnu software stallman -exclude' ); + $this->addLink( + 'Note:', + 'local', + 'Stallman has a beard and is part of the Free Software Foundation (or not). Seriously, read this.', + 0, + '20150310_114651', + '' + ); + $this->addLink( 'MediaGoblin', 'http://mediagoblin.org/', -- 2.41.0