From 522b278b03280ed809025ebbeb3eac284b68bf81 Mon Sep 17 00:00:00 2001 From: ArthurHoaro Date: Tue, 2 Feb 2016 19:42:48 +0100 Subject: [PATCH] Support text search across link fields. --- application/LinkFilter.php | 52 ++++++++++++++++----------------- tests/LinkDBTest.php | 1 + tests/LinkFilterTest.php | 51 ++++++++++++++++++++++++++++---- tests/utils/ReferenceLinkDB.php | 8 ++--- 4 files changed, 75 insertions(+), 37 deletions(-) diff --git a/application/LinkFilter.php b/application/LinkFilter.php index e2ef94ea..17594e8f 100644 --- a/application/LinkFilter.php +++ b/application/LinkFilter.php @@ -138,6 +138,7 @@ private function filterSmallHash($smallHash) */ private function filterFulltext($searchterms, $privateonly = false) { + $filtered = array(); $search = mb_convert_case(html_entity_decode($searchterms), MB_CASE_LOWER, 'UTF-8'); $exactRegex = '/"([^"]+)"/'; // Retrieve exact search terms. @@ -169,35 +170,32 @@ private function filterFulltext($searchterms, $privateonly = false) continue; } - // Iterate over searchable link fields. + // Concatenate link fields to search across fields. + // Adds a '\' separator for exact search terms. + $content = ''; foreach ($keys as $key) { - // Be optimistic - $found = true; - - $haystack = mb_convert_case($link[$key], MB_CASE_LOWER, 'UTF-8'); - - // First, we look for exact term search - for ($i = 0; $i < count($exactSearch) && $found; $i++) { - $found = strpos($haystack, $exactSearch[$i]) !== false; - } - - // 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. - if ($found) { - break; - } + $content .= mb_convert_case($link[$key], MB_CASE_LOWER, 'UTF-8') . '\\'; } - + + // Be optimistic + $found = true; + + // First, we look for exact term search + for ($i = 0; $i < count($exactSearch) && $found; $i++) { + $found = strpos($content, $exactSearch[$i]) !== false; + } + + // 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++) { + $found = strpos($content, $andSearch[$i]) !== false; + } + + // Exclude terms. + for ($i = 0; $i < count($excludeSearch) && $found; $i++) { + $found = strpos($content, $excludeSearch[$i]) === false; + } + if ($found) { $filtered[$link['linkdate']] = $link; } diff --git a/tests/LinkDBTest.php b/tests/LinkDBTest.php index 78f42e56..b6a273b3 100644 --- a/tests/LinkDBTest.php +++ b/tests/LinkDBTest.php @@ -298,6 +298,7 @@ public function testAllTags() 'w3c' => 1, 'css' => 1, 'Mercurial' => 1, + 'stuff' => 2, '-exclude' => 1, '.hidden' => 1, ), diff --git a/tests/LinkFilterTest.php b/tests/LinkFilterTest.php index 4d754d25..31fd4cf4 100644 --- a/tests/LinkFilterTest.php +++ b/tests/LinkFilterTest.php @@ -164,6 +164,17 @@ public function testFilterUnknownSmallHash() ); } + /** + * Full-text search - no result found. + */ + public function testFilterFullTextNoResult() + { + $this->assertEquals( + 0, + count(self::$linkFilter->filter(LinkFilter::$FILTER_TEXT, 'azertyuiop')) + ); + } + /** * Full-text search - result from a link's URL */ @@ -262,28 +273,56 @@ public function testExcludeSearch() { $this->assertEquals( 1, - count(self::$linkFilter->filter(LinkFilter::$FILTER_TEXT, 'free -software')) + count(self::$linkFilter->filter(LinkFilter::$FILTER_TEXT, 'free -gnu')) ); $this->assertEquals( - 7, - count(self::$linkFilter->filter(LinkFilter::$FILTER_TEXT, '-software')) + 6, + count(self::$linkFilter->filter(LinkFilter::$FILTER_TEXT, '-revolution')) ); } /** - * Full-text search - test AND, exact terms and exclusion combined. + * Full-text search - test AND, exact terms and exclusion combined, across fields. */ public function testMultiSearch() { $this->assertEquals( 2, - count(self::$linkFilter->filter(LinkFilter::$FILTER_TEXT, '"Free Software " stallman "read this"')) + count(self::$linkFilter->filter( + LinkFilter::$FILTER_TEXT, + '"Free Software " stallman "read this" @website stuff' + )) ); $this->assertEquals( 1, - count(self::$linkFilter->filter(LinkFilter::$FILTER_TEXT, '"free software " stallman "read this" -beard')) + count(self::$linkFilter->filter( + LinkFilter::$FILTER_TEXT, + '"free software " stallman "read this" -beard @website stuff' + )) + ); + } + + /** + * Full-text search - make sure that exact search won't work across fields. + */ + public function testSearchExactTermMultiFieldsKo() + { + $this->assertEquals( + 0, + count(self::$linkFilter->filter( + LinkFilter::$FILTER_TEXT, + '"designer naming"' + )) + ); + + $this->assertEquals( + 0, + count(self::$linkFilter->filter( + LinkFilter::$FILTER_TEXT, + '"designernaming"' + )) ); } diff --git a/tests/utils/ReferenceLinkDB.php b/tests/utils/ReferenceLinkDB.php index b64b58bf..61faef05 100644 --- a/tests/utils/ReferenceLinkDB.php +++ b/tests/utils/ReferenceLinkDB.php @@ -14,21 +14,21 @@ class ReferenceLinkDB function __construct() { $this->addLink( - 'Free as in Freedom 2.0', + 'Free as in Freedom 2.0 @website', 'https://static.fsf.org/nosvn/faif-2.0.pdf', 'Richard Stallman and the Free Software Revolution. Read this.', 0, '20150310_114633', - 'free gnu software stallman -exclude' + 'free gnu software stallman -exclude stuff' ); $this->addLink( - 'Note:', + 'Link title: @website', 'local', 'Stallman has a beard and is part of the Free Software Foundation (or not). Seriously, read this.', 0, '20150310_114651', - '' + 'stuff' ); $this->addLink(