public function storeDoc($url, $esDoc)
{
Log::info("Store {$url}");
$esDoc->status->processed = gmdate('c');
$r = new Elasticsearch_Request($GLOBALS['phinde']['elasticsearch'] . 'document/' . ElasticSearch::getDocId($url), \HTTP_Request2::METHOD_PUT);
$r->setBody(json_encode($esDoc));
$r->send();
}
public function search($query, $filters, $site, $page, $perPage, $sort)
{
if (preg_match_all('#nick:([^ ]*)#', $query, $matches)) {
foreach ($matches[1] as $authorName) {
$query = str_replace('nick:' . $authorName, 'author.name:' . $authorName, $query);
}
}
$qMust = array();
//query parts for the MUST section
//modification date filters
if (preg_match('#after:([^ ]+)#', $query, $matches)) {
$dateAfter = $matches[1];
$query = trim(str_replace($matches[0], '', $query));
$qMust[] = array('range' => array('modate' => array('gt' => $dateAfter . '||/d')));
}
if (preg_match('#before:([^ ]+)#', $query, $matches)) {
$dateBefore = $matches[1];
$query = trim(str_replace($matches[0], '', $query));
$qMust[] = array('range' => array('modate' => array('lt' => $dateBefore . '||/d')));
}
if (preg_match('#date:([^ ]+)#', $query, $matches)) {
$dateExact = $matches[1];
$query = trim(str_replace($matches[0], '', $query));
$qMust[] = array('range' => array('modate' => array('gte' => $dateExact . '||/d', 'lte' => $dateExact . '||/d')));
}
$qMust[] = array('query_string' => array('default_field' => '_all', 'default_operator' => 'AND', 'query' => $query));
$qMust[] = array('term' => array('status.findable' => true));
if ($sort == '' && $GLOBALS['phinde']['defaultSort'] == 'date') {
$sort = 'date';
}
if ($sort == 'date') {
$sortCfg = array('status.modate' => array('order' => 'desc'));
} else {
$sortCfg = array();
}
$contentMatchSize = 100;
if ($GLOBALS['phinde']['showFullContent']) {
$contentMatchSize = 999999;
}
$r = new Elasticsearch_Request($this->baseUrl . 'document/_search', \HTTP_Request2::METHOD_GET);
$doc = array('_source' => array('url', 'title', 'author', 'status.modate'), 'query' => array('bool' => array('must' => $qMust)), 'highlight' => array('pre_tags' => array('<em class="hl">'), 'order' => 'score', 'encoder' => 'html', 'fields' => array('title' => array('require_field_match' => false, 'number_of_fragments' => 0), 'url' => array('require_field_match' => false, 'number_of_fragments' => 0), 'text' => array('require_field_match' => false, 'number_of_fragments' => 1, 'fragment_size' => $contentMatchSize, 'no_match_size' => $contentMatchSize))), 'aggregations' => array('tags' => array('terms' => array('field' => 'tags')), 'language' => array('terms' => array('field' => 'language')), 'domain' => array('terms' => array('field' => 'domain')), 'type' => array('terms' => array('field' => 'type'))), 'from' => $page * $perPage, 'size' => $perPage, 'sort' => $sortCfg);
foreach ($filters as $type => $value) {
$doc['query']['bool']['must'][] = array('term' => array($type => $value));
}
if ($site != '') {
$doc['query']['bool']['must'][] = array('prefix' => array('schemalessUrl' => array('value' => $site)));
}
//unset($doc['_source']);
//ini_set('xdebug.var_display_max_depth', 10);
//echo json_encode($doc);die();
$r->setBody(json_encode($doc));
$res = $r->send();
return json_decode($res->getBody());
}
#!/usr/bin/env php
<?php
namespace phinde;
/**
* Configure the elasticsearch index.
* Throws away all data.
*/
require_once __DIR__ . '/../src/init.php';
$json = file_get_contents(__DIR__ . '/../data/elasticsearch-mapping.json');
if (json_decode($json) === null) {
Log::error("Error: Schema JSON is broken");
chdir(__DIR__ . '/../');
passthru('json_pp -t null < data/elasticsearch-mapping.json');
exit(1);
}
//delete old index
$r = new Elasticsearch_Request($GLOBALS['phinde']['elasticsearch'], \HTTP_Request2::METHOD_DELETE);
$r->allow404 = true;
$r->send();
//recreate it
$r = new Elasticsearch_Request($GLOBALS['phinde']['elasticsearch'], \HTTP_Request2::METHOD_PUT);
$r->setBody($json);
$r->send();