3 // web interface to search
4 require 'www-header.php';
6 if (!isset($_GET['q'])) {
9 $query = trim($_GET['q']);
12 if (isset($_GET['page'])) {
13 if (!is_numeric($_GET['page'])) {
14 throw new Exception_Input('List page is not numeric');
16 //PEAR Pager begins at 1
17 $page = (int)$_GET['page'] - 1;
20 $perPage = 10;//$GLOBALS['phinde']['perPage'];
23 $baseLink = '?q=' . urlencode($query);
25 if (preg_match('#site:([^ ]*)#', $query, $matches)) {
27 $cleanQuery = trim(str_replace('site:' . $site, '', $query));
28 $site = Helper::noSchema($site);
29 } else if (isset($_GET['site']) && trim(isset($_GET['site'])) != '') {
30 $site = trim($_GET['site']);
33 $baseLink .= '&site=' . urlencode($site);
38 if (isset($_GET['sort']) && $_GET['sort'] == 'date') {
40 $baseLink .= '&sort=date';
46 if (isset($_GET['filter'])) {
47 $allowedFilter = array('domain', 'language', 'tags', 'term');
48 foreach ($_GET['filter'] as $type => $value) {
49 if (in_array($type, $allowedFilter)) {
50 $filters[$type] = filter_var($value, FILTER_SANITIZE_STRING);
54 $activeFilters = array();
55 foreach ($filters as $type => $value) {
56 $activeFilters[$type] = array(
58 'removeUrl' => buildLink($baseLink, $filters, $type, null),
62 function buildLink($baseLink, $filters, $addFilterType, $addFilterValue)
64 if ($addFilterValue === null) {
65 if (array_key_exists($addFilterType, $filters)) {
66 unset($filters[$addFilterType]);
69 $filters[$addFilterType] = $addFilterValue;
72 $params = http_build_query(array('filter' => $filters));
73 if (strlen($params)) {
74 return $baseLink . '&' . $params;
79 if (preg_match('#site:([^ ]*)#', $query, $matches)) {
81 $cleanQuery = trim(str_replace('site:' . $site, '', $query));
82 $site = Helper::noSchema($site);
83 $urlNoSite = buildLink('?q=' . urlencode($cleanQuery), $filters, null, null);
89 $timeBegin = microtime(true);
90 $es = new Elasticsearch($GLOBALS['phinde']['elasticsearch']);
91 $res = $es->search($cleanQuery, $filters, $site, $page, $perPage, $sort);
92 $timeEnd = microtime(true);
94 $pager = new Html_Pager(
95 $res->hits->total, $perPage, $page + 1,
99 foreach ($res->hits->hits as &$hit) {
100 $doc = $hit->_source;
101 if (!isset($doc->title) || $doc->title == '') {
102 $doc->title = '(no title)';
103 $doc->htmlTitle = '(no title)';
105 if (isset($hit->highlight->title[0])) {
106 $doc->htmlTitle = $hit->highlight->title[0];
108 $doc->htmlTitle = htmlspecialchars($doc->title);
110 if (isset($hit->highlight->text[0])) {
111 $doc->htmlText = $hit->highlight->text[0];
113 $doc->htmlText = null;
116 $doc->extra = new \stdClass();
117 $doc->extra->cleanUrl = preg_replace('#^.*://#', '', $doc->url);
118 if (isset($doc->modate)) {
119 $doc->extra->day = substr($doc->modate, 0, 10);
123 foreach ($res->aggregations as $key => &$aggregation) {
124 foreach ($aggregation->buckets as &$bucket) {
125 $bucket->url = buildLink($baseLink, $filters, $key, $bucket->key);
129 if ($site !== null) {
130 $urlNoSite = buildLink('?q=' . urlencode($cleanQuery), $filters, null, null);
135 $urlSortRelevance = buildLink(
136 str_replace('&sort=date', '', $baseLink), $filters, null, null
138 $urlSortDate = $urlSortRelevance . '&sort=date';
140 if (isset($_GET['format']) && $_GET['format'] == 'opensearch') {
141 $template = 'opensearch';
142 $baseLink .= '&format=opensearch';
143 header('Content-type: application/atom+xml');
145 $template = 'search';
151 'queryTime' => round($timeEnd - $timeBegin, 2) . 's',
153 'fullUrl' => Helper::fullUrl($baseLink),
154 'cleanQuery' => $cleanQuery,
155 'urlNoSite' => $urlNoSite,
157 'siteParam' => $siteParam,
158 'hitcount' => $res->hits->total,
159 'hits' => $res->hits->hits,
160 'aggregations' => $res->aggregations,
161 'activeFilters' => $activeFilters,
164 'urlSortRelevance' => $urlSortRelevance,
165 'urlSortDate' => $urlSortDate,