From baaf4c3043f428ec009dd48b85e1c2d6cb8abee4 Mon Sep 17 00:00:00 2001 From: Rob Hoelz Date: Mon, 14 Apr 2014 23:18:33 -0500 Subject: [PATCH] Make search mechanism pluggable Currently, TinyTinyRSS can use raw SQL or the Sphinx search engine for searching. It would be nice if other search engines (such as Xapian) could be used, or if features of the underlying SQL engine (such as MySQL's FULLTEXT indexes) could be leveraged. This commit makes searching into a plugin hook, falling back to the builtin behavior if no search plugin is active. The Sphinx search behavior has been broken out into a plugin. --- classes/feeds.php | 2 +- classes/pluginhost.php | 1 + config.php-dist | 4 -- include/functions2.php | 50 ++-------------- include/sanity_check.php | 5 -- include/sanity_config.php | 2 +- plugins/search_sphinx/init.php | 60 ++++++++++++++++++++ {lib => plugins/search_sphinx}/sphinxapi.php | 0 8 files changed, 69 insertions(+), 55 deletions(-) create mode 100644 plugins/search_sphinx/init.php rename {lib => plugins/search_sphinx}/sphinxapi.php (100%) diff --git a/classes/feeds.php b/classes/feeds.php index ca470c03..89acb0d0 100644 --- a/classes/feeds.php +++ b/classes/feeds.php @@ -1147,7 +1147,7 @@ class Feeds extends Handler_Protected { print "
"; - if (!SPHINX_ENABLED) { + if (count(PluginHost::getInstance()->get_hooks(PluginHost::HOOK_SEARCH)) == 0) { print "
".__("Search syntax")."
"; diff --git a/classes/pluginhost.php b/classes/pluginhost.php index 19f8b563..4582c314 100644 --- a/classes/pluginhost.php +++ b/classes/pluginhost.php @@ -39,6 +39,7 @@ class PluginHost { const HOOK_FETCH_FEED = 22; const HOOK_QUERY_HEADLINES = 23; const HOOK_HOUSE_KEEPING = 24; + const HOOK_SEARCH = 25; const KIND_ALL = 1; const KIND_SYSTEM = 2; diff --git a/config.php-dist b/config.php-dist index 999f0077..2d7a6d19 100644 --- a/config.php-dist +++ b/config.php-dist @@ -108,10 +108,6 @@ // *** Sphinx search *** // ********************* - define('SPHINX_ENABLED', false); - // Enable fulltext search using Sphinx (http://www.sphinxsearch.com) - // Please see http://tt-rss.org/wiki/SphinxSearch for more information. - define('SPHINX_SERVER', 'localhost:9312'); // Hostname:port combination for the Sphinx server. diff --git a/include/functions2.php b/include/functions2.php index ca58ea6c..22c60236 100644 --- a/include/functions2.php +++ b/include/functions2.php @@ -397,20 +397,15 @@ $search_words = array(); if ($search) { + foreach (PluginHost::getInstance()->get_hooks(PluginHost::HOOK_SEARCH) as $plugin) { + list($search_query_part, $search_words) = $plugin->hook_search($search); + } - if (SPHINX_ENABLED) { - $ids = join(",", @sphinx_search($search, 0, 500)); - - if ($ids) - $search_query_part = "ref_id IN ($ids) AND "; - else - $search_query_part = "ref_id = -1 AND "; - - } else { + // fall back in case of no plugins + if (!$search_query_part) { list($search_query_part, $search_words) = search_to_sql($search); - $search_query_part .= " AND "; } - + $search_query_part .= " AND "; } else { $search_query_part = ""; } @@ -1994,39 +1989,6 @@ } } - function sphinx_search($query, $offset = 0, $limit = 30) { - require_once 'lib/sphinxapi.php'; - - $sphinxClient = new SphinxClient(); - - $sphinxpair = explode(":", SPHINX_SERVER, 2); - - $sphinxClient->SetServer($sphinxpair[0], (int)$sphinxpair[1]); - $sphinxClient->SetConnectTimeout(1); - - $sphinxClient->SetFieldWeights(array('title' => 70, 'content' => 30, - 'feed_title' => 20)); - - $sphinxClient->SetMatchMode(SPH_MATCH_EXTENDED2); - $sphinxClient->SetRankingMode(SPH_RANK_PROXIMITY_BM25); - $sphinxClient->SetLimits($offset, $limit, 1000); - $sphinxClient->SetArrayResult(false); - $sphinxClient->SetFilter('owner_uid', array($_SESSION['uid'])); - - $result = $sphinxClient->Query($query, SPHINX_INDEX); - - $ids = array(); - - if (is_array($result['matches'])) { - foreach (array_keys($result['matches']) as $int_id) { - $ref_id = $result['matches'][$int_id]['attrs']['ref_id']; - array_push($ids, $ref_id); - } - } - - return $ids; - } - function cleanup_tags($days = 14, $limit = 1000) { if (DB_TYPE == "pgsql") { diff --git a/include/sanity_check.php b/include/sanity_check.php index 1aa581bf..6bec4305 100644 --- a/include/sanity_check.php +++ b/include/sanity_check.php @@ -146,11 +146,6 @@ array_push($errors, "PHP support for CURL is required for PubSubHubbub."); } - if (SPHINX_ENABLED && class_exists("SphinxClient")) { - array_push($errors, "Your PHP has a separate systemwide Sphinx client installed which conflicts with the client library used by tt-rss. Either remove the system library or disable Sphinx support."); - - } - if (!class_exists("DOMDocument")) { array_push($errors, "PHP support for DOMDocument is required, but was not found."); } diff --git a/include/sanity_config.php b/include/sanity_config.php index 99e83e0d..76fba4b8 100644 --- a/include/sanity_config.php +++ b/include/sanity_config.php @@ -1,3 +1,3 @@ +$requred_defines = array( 'DB_TYPE', 'DB_HOST', 'DB_USER', 'DB_NAME', 'DB_PASS', 'MYSQL_CHARSET', 'SELF_URL_PATH', 'FEED_CRYPT_KEY', 'SINGLE_USER_MODE', 'SIMPLE_UPDATE_MODE', 'PHP_EXECUTABLE', 'LOCK_DIRECTORY', 'CACHE_DIR', 'ICONS_DIR', 'ICONS_URL', 'AUTH_AUTO_CREATE', 'AUTH_AUTO_LOGIN', 'FORCE_ARTICLE_PURGE', 'PUBSUBHUBBUB_HUB', 'PUBSUBHUBBUB_ENABLED', 'ENABLE_REGISTRATION', 'REG_NOTIFY_ADDRESS', 'REG_MAX_USERS', 'SESSION_COOKIE_LIFETIME', 'SESSION_CHECK_ADDRESS', 'SMTP_FROM_NAME', 'SMTP_FROM_ADDRESS', 'DIGEST_SUBJECT', 'SMTP_SERVER', 'SMTP_LOGIN', 'SMTP_PASSWORD', 'SMTP_SECURE', 'CHECK_FOR_NEW_VERSION', 'DETECT_ARTICLE_LANGUAGE', 'ENABLE_GZIP_OUTPUT', 'PLUGINS', 'LOG_DESTINATION', 'CONFIG_VERSION'); ?> diff --git a/plugins/search_sphinx/init.php b/plugins/search_sphinx/init.php new file mode 100644 index 00000000..f2877e44 --- /dev/null +++ b/plugins/search_sphinx/init.php @@ -0,0 +1,60 @@ +add_hook($host::HOOK_SEARCH, $this); + + require_once __DIR__ . "/sphinxapi.php"; + } + + function hook_search($search) { + $offset = 0; + $limit = 500; + + $sphinxClient = new SphinxClient(); + + $sphinxpair = explode(":", SPHINX_SERVER, 2); + + $sphinxClient->SetServer($sphinxpair[0], (int)$sphinxpair[1]); + $sphinxClient->SetConnectTimeout(1); + + $sphinxClient->SetFieldWeights(array('title' => 70, 'content' => 30, + 'feed_title' => 20)); + + $sphinxClient->SetMatchMode(SPH_MATCH_EXTENDED2); + $sphinxClient->SetRankingMode(SPH_RANK_PROXIMITY_BM25); + $sphinxClient->SetLimits($offset, $limit, 1000); + $sphinxClient->SetArrayResult(false); + $sphinxClient->SetFilter('owner_uid', array($_SESSION['uid'])); + + $result = $sphinxClient->Query($search, SPHINX_INDEX); + + $ids = array(); + + if (is_array($result['matches'])) { + foreach (array_keys($result['matches']) as $int_id) { + $ref_id = $result['matches'][$int_id]['attrs']['ref_id']; + array_push($ids, $ref_id); + } + } + + $ids = join(",", $ids); + + if ($ids) + return array("ref_id IN ($ids)", array()); + else + return array("ref_id = -1", array()); + } + + function api_version() { + return 2; + } +} +?> diff --git a/lib/sphinxapi.php b/plugins/search_sphinx/sphinxapi.php similarity index 100% rename from lib/sphinxapi.php rename to plugins/search_sphinx/sphinxapi.php -- 2.39.5