php-webservice/Lib/Search/BookSearch.php

<?php

namespace BSR\Lib\Search;

use BSR\Lib\Configuration;
use BSR\Lib\Exception\WebException;

mb_http_output('UTF-8');

class BookSearch
{
    /** @var \SolrClient  */
    private $client;
    /** @var \SolrQuery  */
    private $query;
    private $queryParts = array();

    public function __construct()
    {
        $options = array
        (
            'hostname' => Configuration::get('solr.server'),
            'port' => Configuration::get('solr.port'),
            'login' => Configuration::get('solr.username'),
            'password' => Configuration::get('solr.password'),
            'path' => Configuration::get('solr.path'),
        );

        $this->client = new \SolrClient($options);
        $this->query = new \SolrDisMaxQuery();

        // use the Extended DisMax Query parser
        $this->query->useEDisMaxQueryParser();

        // most options like search fields, sorting, etc are already set
        // as default in the Solr config and thus should be set only on a
        // per request basis when needed
    }

    public function addOrQuery(array $texts, $field)
    {
        if(count($texts) > 0) {
            $texts = array_map(array('SolrUtils', 'escapeQueryChars'), $texts);
            $query = sprintf('%s:("%s")', $field, implode('" OR "', $texts));
            $this->addQuery($query, null, false);
        }
    }

    public function addQuery($queryText, $queryField = null, $escape = true)
    {
        if($escape) {
            $queryText = \SolrUtils::escapeQueryChars($queryText);
        }

        if (strlen($queryField) > 0) {
            $queryText = "$queryField:\"$queryText\"";
        }

        $this->queryParts[] = $queryText;
    }

    public function addRange($field, $min = '*', $max = '*')
    {
        $this->queryParts[] = sprintf('%s:[%s TO %s]', $field, $min, $max);
    }

    public function addSortField($field, $order = \SolrQuery::ORDER_DESC)
    {
        $this->query->addSortField($field, $order);
    }

    public function addFacetField($field)
    {
        $this->query->addFacetField($field);
    }

    public function setFacetLimits($limit = null, $count = null)
    {
        if(! is_null($limit)) {
            $this->query->setFacetLimit($limit);
        }

        if(! is_null($count)) {
            $this->query->setFacetMinCount($count);
        }
    }

    /**
     * @param int $start
     * @param int $count
     * @return array
     * @throws WebException
     */
    public function getResults($start = 0, $count = 15)
    {
        if (count($this->queryParts) == 0)
            $query = '*:*';
        else {
            $query = implode(' AND ', $this->queryParts);
        }
        $this->query->setQuery($query);
        $this->query->setStart($start);
        $this->query->setRows($count);

        try {
            $results = $this->client->query($this->query)->getArrayResponse();
        } catch(\SolrClientException $e) {
            throw new WebException ("SolrError", $e->getMessage(), -700);
        }

        $books = array();
        if(isset($results['response']['docs']) && is_array($results['response']['docs'])) {
            foreach($results['response']['docs'] as $r) {
                $books[$r['id']] = $r;
            }
        }

        $highlighting = array();
        if(isset($results['highlighting'])) {
            foreach($results['highlighting'] as $k => $h) {
                $data = array();
                foreach($h as $f => $v) {
                    $data[str_replace('_fr', '', $f)] = reset($v);
                }
                $highlighting[$k] = $data;
            }
        }

        $spelling = array();
        if(isset($results['spellcheck']['suggestions'])) {
            foreach($results['spellcheck']['suggestions'] as $s) {
                $spelling[] = $s;
            }
        }

        $facets = array();
        if(isset($results['facet_counts']['facet_fields'])) {
            foreach($results['facet_counts']['facet_fields'] as $f => $d) {
                $facets[$f] = $d;
            }
        }

        return array(
            'count' => $results['response']['numFound'],
            'facets' => array(
                'facets' => $facets,
                'highlighting' => $highlighting,
                'spelling' => $spelling,
            ),
            'books' => $books,
        );
    }

    /**
     * Return a list of suggested titles for the given text
     * @param string $text
     * @return array
     * @throws WebException
     */
    public function suggest($text) {
        $this->query->setQuery($text);
        $this->query->setStart(0);
        $this->query->setRows(0);
        $this->query->setParam('suggest', 'true');
        $this->query->setParam('facet', 'false');
        $this->query->setParam('hl', 'false');
        $this->query->setParam('spellcheck', 'false');

        try {
            $results = $this->client->query($this->query)->getArrayResponse();
        } catch(\SolrClientException $e) {
            throw new WebException ("SolrError", $e->getMessage(), -700);
        }

        $text = mb_strtolower ($text, 'UTF-8');

        $suggestions = array();
        if(isset($results['suggest'])) {
            foreach($results['suggest'] as $suggester) {
                foreach($suggester[$text]['suggestions'] as $s) {
                    $s['term'] = strip_tags($s['term']);

                    $pos = strpos(mb_strtolower($s['term'], 'UTF-8'), $text);
                    if($pos !== false) {
                        // increase weight of proposition that have the words at the beginning
                        $s['weight'] += (int) ((1 - $pos / strlen($s['term'])) * 100);
                    }
                    $suggestions[$s['term']] = (array) $s;
                }
            }
        }

        usort($suggestions, function($a, $b) {
            return $b['weight'] - $a['weight'];
        });

        return $suggestions;
    }

    /**
     * Retrieve books from Solr based on their code (NoticeNr).
     *
     * @param array $codes
     * @param string $field the field to use for the search
     * @return array Books information
     * @throws WebException
     */
    public static function GetBooks(array $codes, $field = 'code') {
        // it is faster to do multiple small request to Solr rather than one big so separate
        // in chunks if we are above the limit. 15 was found by testing and seems to be a sweet spot
        $limit = 15;
        $count = count($codes);
        if($count > $limit) {
            $parts = array_chunk($codes, $limit);
            $books = array();
            foreach($parts as $p) {
                // if we use array_merge here the numerical keys (book code) will be lost
                $books += self::GetBooks($p, $field);
            }
            return $books;
        }

        $bs = new static();
        $bs->addOrQuery($codes, $field);
        $results = $bs->getResults(0, $count);
        return $results['books'];
    }
}