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 //Elasticsearch by default only allows 100 pages
21 // we do not want exceptions in our logs
22 header('HTTP/1.0 400 Bad Request');
23 echo "List page number is too large\n";
28 $perPage = 10;//$GLOBALS['phinde']['perPage'];
31 $baseLink = '?q=' . urlencode($query);
33 if (preg_match('#site:([^ ]*)#', $query, $matches)) {
35 $cleanQuery = trim(str_replace('site:' . $site, '', $query));
36 $site = Helper::noSchema($site);
37 } else if (isset($_GET['site']) && trim(isset($_GET['site'])) != '') {
38 $site = trim($_GET['site']);
41 $baseLink .= '&site=' . urlencode($site);
46 if (isset($_GET['sort'])
47 && ($_GET['sort'] === 'date' || $_GET['sort'] === 'score')
49 $sortMode = $_GET['sort'];
51 $sortMode = $GLOBALS['phinde']['defaultSort'];
54 if ($sortMode !== $GLOBALS['phinde']['defaultSort']) {
55 $baseLink .= '&sort=' . $sortMode;
59 if (isset($_GET['filter'])) {
60 $allowedFilter = array('domain', 'language', 'tags', 'term');
61 foreach ($_GET['filter'] as $type => $value) {
62 if (in_array($type, $allowedFilter)) {
63 $filters[$type] = filter_var($value, FILTER_SANITIZE_STRING);
67 $activeFilters = array();
68 foreach ($filters as $type => $value) {
69 $activeFilters[$type] = array(
71 'removeUrl' => buildLink($baseLink, $filters, $type, null),
75 function buildLink($baseLink, $filters, $addFilterType, $addFilterValue)
77 if ($addFilterValue === null) {
78 if (array_key_exists($addFilterType, $filters)) {
79 unset($filters[$addFilterType]);
82 $filters[$addFilterType] = $addFilterValue;
85 $params = http_build_query(array('filter' => $filters));
86 if (strlen($params)) {
87 return $baseLink . '&' . $params;
92 if (preg_match('#site:([^ ]*)#', $query, $matches)) {
94 $cleanQuery = trim(str_replace('site:' . $site, '', $query));
95 $site = Helper::noSchema($site);
96 $urlNoSite = buildLink('?q=' . urlencode($cleanQuery), $filters, null, null);
102 $timeBegin = microtime(true);
103 $es = new Elasticsearch($GLOBALS['phinde']['elasticsearch']);
104 $res = $es->search($cleanQuery, $filters, $site, $page, $perPage, $sort);
105 $timeEnd = microtime(true);
107 $pager = new Html_Pager(
108 $res->hits->total, $perPage, $page + 1,
109 buildLink($baseLink, $filters, null, null)
112 foreach ($res->hits->hits as &$hit) {
113 $doc = $hit->_source;
114 if (!isset($doc->title) || $doc->title == '') {
115 $doc->title = '(no title)';
116 $doc->htmlTitle = '(no title)';
118 if (isset($hit->highlight->title[0])) {
119 $doc->htmlTitle = $hit->highlight->title[0];
121 $doc->htmlTitle = htmlspecialchars($doc->title);
123 if (isset($hit->highlight->text[0])) {
124 $doc->htmlText = $hit->highlight->text[0];
126 $doc->htmlText = null;
129 $doc->extra = new \stdClass();
130 $doc->extra->cleanUrl = preg_replace('#^.*://#', '', $doc->url);
131 if (isset($doc->status->modate)) {
132 $doc->extra->day = substr($doc->status->modate, 0, 10);
136 foreach ($res->aggregations as $key => &$aggregation) {
137 foreach ($aggregation->buckets as &$bucket) {
138 $bucket->url = buildLink($baseLink, $filters, $key, $bucket->key);
142 if ($site !== null) {
143 $urlNoSite = buildLink('?q=' . urlencode($cleanQuery), $filters, null, null);
148 $urlSortBase = buildLink(
149 preg_replace('#&sort=[^&]+#', '', $baseLink), $filters, null, null
152 foreach (['date', 'score'] as $sortMode) {
153 if ($sortMode === $GLOBALS['phinde']['defaultSort']) {
154 $urlSorts[$sortMode] = $urlSortBase;
156 $urlSorts[$sortMode] = $urlSortBase . '&sort=' . $sortMode;
160 if (isset($_GET['format']) && $_GET['format'] == 'opensearch') {
161 $template = 'opensearch';
162 $baseLink .= '&format=opensearch';
163 header('Content-type: application/atom+xml');
165 $template = 'search';
171 'queryTime' => round($timeEnd - $timeBegin, 2) . 's',
173 'fullUrl' => Helper::fullUrl($baseLink),
174 'cleanQuery' => $cleanQuery,
175 'urlNoSite' => $urlNoSite,
177 'siteParam' => $siteParam,
178 'hitcount' => $res->hits->total,
179 'hits' => $res->hits->hits,
180 'aggregations' => $res->aggregations,
181 'activeFilters' => $activeFilters,
184 'urlSorts' => $urlSorts,
185 'hitTemplate' => 'search/' . $GLOBALS['phinde']['hitTemplate'],
186 'sidebarinclude' => $GLOBALS['phinde']['sidebarinclude'],