X-Git-Url: https://git.cweiske.de/phinde.git/blobdiff_plain/7b4425b096fa8c18d0db9fd9b1ae96d63ee8af55..a1c8309e4b3d8d5468c2defbc8bbae95633aff90:/www/index.php

diff --git a/www/index.php b/www/index.php
index 96d6f7a..c1acfdf 100644
--- a/www/index.php
+++ b/www/index.php
@@ -1,20 +1,13 @@
 <?php
 namespace phinde;
 // web interface to search
-set_include_path(__DIR__ . '/../src/' . PATH_SEPARATOR . get_include_path());
-require_once __DIR__ . '/../data/config.php';
-require_once 'HTTP/Request2.php';
-require_once 'Pager.php';
-require_once 'Html/Pager.php';
-require_once 'Elasticsearch.php';
-require_once 'Elasticsearch/Request.php';
-require_once 'functions.php';
+require 'www-header.php';
 
 if (!isset($_GET['q'])) {
-    exit('no query');
+    $_GET['q'] = '';
 }
+$query = trim($_GET['q']);
 
-$query = $_GET['q'];
 $page = 0;
 if (isset($_GET['page'])) {
     if (!is_numeric($_GET['page'])) {
@@ -22,44 +15,175 @@ if (isset($_GET['page'])) {
     }
     //PEAR Pager begins at 1
     $page = (int)$_GET['page'] - 1;
+
+    if ($page >= 100) {
+        //Elasticsearch by default only allows 100 pages
+        // we do not want exceptions in our logs
+        header('HTTP/1.0 400 Bad Request');
+        echo "List page number is too large\n";
+        exit(1);
+    }
 }
+
 $perPage = 10;//$GLOBALS['phinde']['perPage'];
+$site = null;
+$siteParam = false;
+$baseLink = '?q=' . urlencode($query);
+
+if (preg_match('#site:([^ ]*)#', $query, $matches)) {
+    $site = $matches[1];
+    $cleanQuery = trim(str_replace('site:' . $site, '', $query));
+    $site = Helper::noSchema($site);
+} else if (isset($_GET['site']) && trim(isset($_GET['site'])) != '') {
+    $site = trim($_GET['site']);
+    $siteParam = true;
+    $cleanQuery = $query;
+    $baseLink .= '&site=' . urlencode($site);
+} else {
+    $cleanQuery = $query;
+}
+
+if (isset($_GET['sort'])
+    && ($_GET['sort'] === 'date' || $_GET['sort'] === 'score')
+) {
+    $sortMode = $_GET['sort'];
+} else {
+    $sortMode = $GLOBALS['phinde']['defaultSort'];
+}
+$sort = $sortMode;
+if ($sortMode !== $GLOBALS['phinde']['defaultSort']) {
+    $baseLink .= '&sort=' . $sortMode;
+}
+
+$filters = array();
+if (isset($_GET['filter'])) {
+    $allowedFilter = array('domain', 'language', 'tags', 'term');
+    foreach ($_GET['filter'] as $type => $value) {
+        if (in_array($type, $allowedFilter)) {
+            $filters[$type] = filter_var($value, FILTER_SANITIZE_STRING);
+        }
+    }
+}
+$activeFilters = array();
+foreach ($filters as $type => $value) {
+    $activeFilters[$type] = array(
+        'label' => $value,
+        'removeUrl' => buildLink($baseLink, $filters, $type, null),
+    );
+}
+
+function buildLink($baseLink, $filters, $addFilterType, $addFilterValue)
+{
+    if ($addFilterValue === null) {
+        if (array_key_exists($addFilterType, $filters)) {
+            unset($filters[$addFilterType]);
+        }
+    } else {
+        $filters[$addFilterType] = $addFilterValue;
+    }
+
+    $params = http_build_query(array('filter' => $filters));
+    if (strlen($params)) {
+        return $baseLink . '&' . $params;
+    }
+    return $baseLink;
+}
 
+if (preg_match('#site:([^ ]*)#', $query, $matches)) {
+    $site = $matches[1];
+    $cleanQuery = trim(str_replace('site:' . $site, '', $query));
+    $site = Helper::noSchema($site);
+    $urlNoSite = buildLink('?q=' . urlencode($cleanQuery), $filters, null, null);
+} else {
+    $cleanQuery = $query;
+    $urlNoSite = null;
+}
+
+$timeBegin = microtime(true);
 $es = new Elasticsearch($GLOBALS['phinde']['elasticsearch']);
-$res = $es->search($query, $page, $perPage);
+$res = $es->search($cleanQuery, $filters, $site, $page, $perPage, $sort);
+$timeEnd = microtime(true);
 
 $pager = new Html_Pager(
     $res->hits->total, $perPage, $page + 1,
-    '?q=' . $query
+    buildLink($baseLink, $filters, null, null)
 );
 
-foreach ($res->hits->hits as $hit) {
+foreach ($res->hits->hits as &$hit) {
     $doc = $hit->_source;
-    if ($doc->title == '') {
+    if (!isset($doc->title) || $doc->title == '') {
         $doc->title = '(no title)';
+        $doc->htmlTitle = '(no title)';
+    }
+    if (isset($hit->highlight->title[0])) {
+        $doc->htmlTitle = $hit->highlight->title[0];
+    } else {
+        $doc->htmlTitle = htmlspecialchars($doc->title);
+    }
+    if (isset($hit->highlight->text[0])) {
+        $doc->htmlText = $hit->highlight->text[0];
+    } else {
+        $doc->htmlText = null;
     }
-    echo '<p>'
-        . '<a href="' . htmlspecialchars($doc->url) . '">'
-        . htmlspecialchars($doc->title)
-        . '</a>';
-    if (isset($doc->author->name)) {
-        echo ' by <a href="' . htmlspecialchars($doc->author->url) . '">'
-            . htmlspecialchars($doc->author->name)
-            . '</a>';
+
+    $doc->extra = new \stdClass();
+    $doc->extra->cleanUrl = preg_replace('#^.*://#', '', $doc->url);
+    if (isset($doc->status->modate)) {
+        $doc->extra->day = substr($doc->status->modate, 0, 10);
+    }
+}
+
+foreach ($res->aggregations as $key => &$aggregation) {
+    foreach ($aggregation->buckets as &$bucket) {
+        $bucket->url = buildLink($baseLink, $filters, $key, $bucket->key);
     }
-    echo  '<br/><tt>'
-        . htmlspecialchars(preg_replace('#^.*://#', '', $doc->url))
-        . '</tt>';
-    if (isset($doc->modate)) {
-        echo '<br/>Changed: ' . substr($doc->modate, 0, 10);
+}
+
+if ($site !== null) {
+    $urlNoSite = buildLink('?q=' . urlencode($cleanQuery), $filters, null, null);
+} else {
+    $urlNoSite = null;
+}
+
+$urlSortBase = buildLink(
+    preg_replace('#&sort=[^&]+#', '', $baseLink), $filters, null, null
+);
+$urlSorts = [];
+foreach (['date', 'score'] as $sortMode) {
+    if ($sortMode === $GLOBALS['phinde']['defaultSort']) {
+        $urlSorts[$sortMode] = $urlSortBase;
+    } else {
+        $urlSorts[$sortMode] = $urlSortBase . '&sort=' . $sortMode;
     }
-    echo '</p>';
 }
 
-$links = $pager->getLinks();
-echo $links['back']
-    . ' ' . implode(' ', $links['pages'])
-    . ' ' . $links['next'];
-//var_dump($links);
-var_dump($res->aggregations->domain);
+if (isset($_GET['format']) && $_GET['format'] == 'opensearch') {
+    $template = 'opensearch';
+    $baseLink .= '&format=opensearch';
+    header('Content-type: application/atom+xml');
+} else {
+    $template = 'search';
+}
+
+render(
+    $template,
+    array(
+        'queryTime' => round($timeEnd - $timeBegin, 2) . 's',
+        'query' => $query,
+        'fullUrl' => Helper::fullUrl($baseLink),
+        'cleanQuery' => $cleanQuery,
+        'urlNoSite' => $urlNoSite,
+        'site' => $site,
+        'siteParam' => $siteParam,
+        'hitcount' => $res->hits->total,
+        'hits' => $res->hits->hits,
+        'aggregations' => $res->aggregations,
+        'activeFilters' => $activeFilters,
+        'pager' => $pager,
+        'sort' => $sort,
+        'urlSorts' => $urlSorts,
+        'hitTemplate' => 'search/' . $GLOBALS['phinde']['hitTemplate'],
+        'sidebarinclude' => $GLOBALS['phinde']['sidebarinclude'],
+    )
+);
 ?>