<?php

namespace CorpusParole\Repositories;

use Config;

use Log;

use CorpusParole\Models\DocumentResult;

use CorpusParole\Models\Document;

use CorpusParole\Libraries\CorpusParoleException;

use CorpusParole\Libraries\Utils;

use CorpusParole\Libraries\Sparql\SparqlClient;

use CorpusParole\Libraries\Filters\CorpusFilterManager;

use CorpusParole\Services\LexvoResolverInterface;

use EasyRdf\Graph;

use Illuminate\Pagination\LengthAwarePaginator;

use Illuminate\Pagination\Paginator;

use Es;

/**

 * Implement the DocumentRepository using EasyRdf

 * TODO: certainly split the transaction management (+add, +delete +transaction ) to an external class -> for this extend the sparql client.

 */

class RdfDocumentRepository implements DocumentRepository {

    const BASE_DOC_QUERY

        = "SELECT".

        "    ?uri".

        "    ?doc".

        "    ?title".

        "    ?issued".

        "    ?modified".

        "    (group_concat(distinct ?language;separator=\", \") as ?lang) ".

        "    (group_concat(distinct ?publisher;separator=\", \") as ?publishers) ".

        "  WHERE {".

        "    GRAPH ?uri { ?doc a <http://www.europeana.eu/schemas/edm/ProvidedCHO>.".

        "      ?doc <http://purl.org/dc/elements/1.1/title> ?title.".

        "      OPTIONAL {?doc <http://purl.org/dc/elements/1.1/language> ?language.} ".

        "      OPTIONAL {?doc <http://purl.org/dc/terms/issued> ?issued.} ".

        "      OPTIONAL {?doc <http://purl.org/dc/terms/modified> ?modified.} ".

        "      OPTIONAL {?doc <http://purl.org/dc/elements/1.1/publisher> ?publisher.}".

        "    }. ".

        "    %s".

        "  } ".

        "  GROUP BY ?uri ?doc ?title ?issued ?modified ";

    const ADDITIONAL_DOC_QUERIES = [

        "SELECT".

        "    ?uri".

        "    ?doc".

        "    (sample(distinct ?ext) as ?extent) ".

        "WHERE {".

        "    GRAPH ?uri {".

        "        ?s a <http://www.europeana.eu/schemas/edm/WebResource>. ".

        "        ?uri <http://www.europeana.eu/schemas/edm/isShownBy> ?s. ".

        "        ?uri <http://www.europeana.eu/schemas/edm/aggregatedCHO> ?doc. ".

        "        OPTIONAL {?s <http://purl.org/dc/terms/extent> ?ext.}".

        "    }. ".

        "    %s".

        "} ".

        "GROUP BY ?uri ?doc",

        "SELECT".

        "    ?uri".

        "    ?doc".

        "    (sample(distinct str(?s)) as ?transcript_url) ".

        "WHERE {".

        "    GRAPH ?uri {".

        "        ?s a <http://www.europeana.eu/schemas/edm/WebResource>. ".

        "        ?uri <http://www.europeana.eu/schemas/edm/aggregatedCHO> ?doc. ".

        "        OPTIONAL {?s <http://purl.org/dc/elements/1.1/format> ?f.} ".

        "    }. ".

        "    FILTER(str(?f) IN ( \"application/xml\", \"application/pdf\" )). ".

        "    %s".

        "} ".

        "GROUP BY ?uri ?doc"

    ];

    private $sparqlClient;

    private $lexvoResolver;

    public function __construct(SparqlClient $sparqlClient, LexvoResolverInterface $lexvoResolver) {

        $this->sparqlClient = $sparqlClient;

        $this->lexvoResolver = $lexvoResolver;

    }

    public function getSparqlClient() {

        return $this->sparqlClient;

    }

    private function getResGraph($doc) {

        if(empty((array)$doc)) {

            return null;

        }

        $newGraph = new Graph($doc->uri->getUri());

        $newGraph->add($doc->uri, "rdf:type", $newGraph->resource("http://www.openarchives.org/ore/terms/Aggregation"));

        $newGraph->add($doc->uri, "http://www.europeana.eu/schemas/edm/aggregatedCHO", $doc->doc);

        $newGraph->add($doc->doc, "rdf:type", $newGraph->resource("http://www.europeana.eu/schemas/edm/ProvidedCHO"));

        if(isset($doc->title)) {

            $newGraph->add($doc->doc, "http://purl.org/dc/elements/1.1/title", $doc->title);

        }

        if(isset($doc->lang)) {

            foreach(explode(", ", $doc->lang) as $langStr) {

                $langStr = trim($langStr);

                if(filter_var($langStr, FILTER_VALIDATE_URL)) {

                    $newGraph->addResource($doc->doc, "http://purl.org/dc/elements/1.1/language", $langStr);

                } else {

                    $newGraph->addLiteral($doc->doc, "http://purl.org/dc/elements/1.1/language", $langStr);

                }

            }

        }

        if(isset($doc->issued)) {

            $newGraph->add($doc->doc, "http://purl.org/dc/terms/issued", $doc->issued);

        }

        if(isset($doc->modified)) {

            $newGraph->add($doc->doc, "http://purl.org/dc/terms/modified", $doc->modified);

        }

        if(isset($doc->publishers)) {

            $newGraph->add($doc->doc, "http://purl.org/dc/elements/1.1/publisher", $doc->publishers);

        }

        if(isset($doc->extent)) {

            $newGraph->add($doc->doc, "http://purl.org/dc/terms/extent", $doc->extent);

        }

        if(isset($doc->transcript_url)) {

            $newGraph->add($doc->doc, config('corpusparole.corpus_ontology_url').'transcript', $doc->transcript_url);

        }

        return $newGraph;

    }

    private function queryES($filters=null, $offset=null, $limit=null, $sort=null) {

        if(empty($sort)) {

            $sort = ["_doc"];

        } elseif (is_string($sort)) {

            $sort = [$sort];

        }

        if(is_null($filters)) {

            //$filters = ['language' => ["http://lexvo.org/id/iso639-3/oci", "http://lexvo.org/id/iso639-3/bre"]];

            $filters = [];

        }

        $qFilterParts = [];

        if(array_key_exists('language', $filters) && !empty($filters['language'])) {

            $languages = $filters['language'];

            if(is_string($languages)) {

                $languages = [ $languages, ];

            }

            $qFilterParts[] = CorpusFilterManager::getLanguagesFilterPart($languages);

        }

        if(array_key_exists('location', $filters) && !empty($filters['location'])) {

            $location = $filters['location'];

            if(is_array($location)) {

                $location = $location[0]; // we know it is not empty

            }

            $qFilterParts[] = CorpusFilterManager::getLocationFilterPart($location);

        }

        if(array_key_exists('themes', $filters) && !empty($filters['themes'])) {

            $themes = $filters['themes'];

            if(is_string($themes)) {

                $themes = [$themes,]; // we know it is not empty

            }

            $qFilterParts[] = CorpusFilterManager::getThemeFilterPart($themes);

        }

        $query = [

            'index' => config('corpusparole.elasticsearch_index'),

            'body' => [

                "size" => empty($limit)?0:$limit,

                "from" => $offset,

                "sort" => $sort

            ]

        ];

        if(count($qFilterParts)>0) {

            $query['body']['query'] = ['constant_score' => [

                'filter' => [

                    'bool' => [

                        'must' => $qFilterParts

                    ]

                ]

            ] ];

        }

        $esRes = Es::search($query);

        return ['total' => $esRes['hits']['total'], 'documents' => array_map(function($r) {

            return $r['_id'];

        }, $esRes['hits']['hits'])];

    }

    /**

     * Query docs.

     * if $filter is empty or null and $sort is '_graph', the documents list is fetched from the triple store, otherwise, They are fetched from ElasticSearch

     */

    private function queryDocs($filters=null, $offset=null, $limit=null, $sort=null) {

        $resDocs = [];

        $limitsClauses = [];

        $sortClauseStr = "";

        $limitsClausesStr = "";

        $filterUris = "";

        if(empty($filters) && $sort === "_graph") {

            if(!is_null($offset)) {

                array_push($limitsClauses, "OFFSET $offset");

            }

            if(!is_null($limit)) {

                array_push($limitsClauses, "LIMIT $limit");

            }

            if(!empty($limitsClauses)) {

                $limitsClausesStr = "\n" . join(" ", $limitsClauses);

            }

            $sortClauseStr = "\n ORDER BY ?uri";

            $total = $this->getCount();

        } else {

            $esRes = $this->queryES($filters, $offset, $limit);

            // WARNING: we count on the fact that php keep keys order

            $total = intval($esRes['total']);

            foreach($esRes['documents'] as $esDocId) {

                $uri = config('corpusparole.corpus_doc_id_base_uri_prefix').$esDocId;

                $resDocs[$uri] = null;

            }

            if(count($resDocs) > 0) {

                $filterUris = "FILTER(?uri in (<".join(">, <" , array_keys($resDocs)).">)) ";

            } else {

                return ['meta' => [ 'total'=> $total ], 'documents' => []];

            }

        }

        $sparqlQuery = sprintf(self::BASE_DOC_QUERY.$sortClauseStr.$limitsClausesStr, $filterUris);

        $docs = $this->sparqlClient->query($sparqlQuery);

        foreach($docs as $doc) {

            $graph = $this->getResGraph($doc);

            if(is_null($graph)) {

                Log::debug("NULL GRAPH - odd");

                continue;

            }

            $uri = $doc->uri->getUri();

            $resDocs[$uri] = $graph;

        }

        if(count($resDocs) == 0) {

            return ['meta' => [ 'total'=> $total ], 'documents' => []];

        }

        if(empty($filterUris)) {

            $filterUris = "FILTER(?uri in (<".join(">, <" , array_keys($resDocs)).">)) ";

        }

        foreach(self::ADDITIONAL_DOC_QUERIES as $query) {

            $docs = $this->sparqlClient->query(sprintf($query, $filterUris));

            foreach($docs as $doc) {

                $graph = $this->getResGraph($doc);

                if(is_null($graph)) {

                    continue;

                }

                $uri = $doc->uri->getUri();

                if(array_key_exists($uri, $resDocs)) {

                    $resDocs[$uri] = Utils::mergeGraphs($resDocs[$uri], $graph);

                } else {

                    $resDocs[$uri] = $graph;

                }

            }

        }

        $documentsResults = array_map(function($g) { return new DocumentResult($g->getUri(), $g); }, array_values($resDocs));

        return ['meta' => [ 'total'=> $total ], 'documents' => $documentsResults];

    }

    public function all() {

        return $this->queryDocs(null, null, null, "_graph")['documents'];

    }

    public function get($id, $short=false) {

        if(strpos($id, config('corpusparole.corpus_id_scheme')) === 0) {

            $id = substr($id, strlen(config('corpusparole.corpus_id_scheme')));

        }

        $docUri = Config::get('corpusparole.corpus_doc_id_base_uri').$id;

        // We want the CBD (Concise Bounded Description, cf. http://www.w3.org/Submission/CBD/)

        // WARNING: This seems to work in sesame for our dataset.

        $doc = $this->sparqlClient->query(

            "CONSTRUCT { ?s ?p ?o } WHERE { GRAPH <$docUri> { ?s ?p ?o } }"

        );

        //TODO: return null if not found

        if($doc->isEmpty()) {

            return null;

        }

        //print($doc->dump());

        // clone the graph to force the URI

        $DocumentKlass = $short?DocumentResult::class:Document::class;

        return new $DocumentKlass($docUri, new Graph($docUri, $doc->toRdfPhp()));

    }

    /**

     * save document.

     * @return boolean true if a transaction was started, false otherwise

     * @throws CorpusParoleException if one of the operation could not be performed

     */

    public function save(Document $doc) {

        $transactionStarted = $this->sparqlClient->startTransaction();

        try {

            foreach($doc->getDeltaList() as $delta) {

                $this->sparqlClient->deleteWhere($delta->getDeleteWhere(), $delta->getUri());

                $this->sparqlClient->delete($delta->getDeletedGraph());

                $this->sparqlClient->add($delta->getAddedGraph());

            }

            if($transactionStarted) {

                $transactionStarted = false;

                return $this->sparqlClient->commit();

            }

            else  {

                return false;

            }

        }

        catch(CorpusParoleException $e) {

            if($transactionStarted) {

                $this->sparqlClient->rollback();

            }

            throw $e;

        }

    }

    public function getCount($filters=null) {

        $count = 0;

        if(empty($filters)) {

            $res = $this->sparqlClient->query("SELECT (COUNT (DISTINCT ?g) as ?count) WHERE { GRAPH ?g { ?s a <http://www.europeana.eu/schemas/edm/ProvidedCHO> } }");

            assert(!is_null($res) && count($res)==1);

            $count = intval($res[0]->count->getValue());

        } else {

            $esRes = $this->queryES($filters, 0, 0);

            $count = intval($esRes['hits']['total']);

        }

        return $count;

    }

    //SELECT ?g WHERE { GRAPH ?g { ?s ?p ?o } }

    /**

     * Paginate all document as a paginator.

     *

     * @param  int  $perPage

     * @param  string  $pageName

     * @return \Illuminate\Contracts\Pagination\LengthAwarePaginator

     */

    public function paginateAll($perPage = 15, $pageName = 'page', $page = null)

    {

        return $this->paginate(null, $perPage, $pageName, null);

    }

    /**

     * Paginate filtered document as a paginator.

     *

     * @param  array $filters

     * @param  int  $perPage

     * @param  string  $pageName

     * @return \Illuminate\Contracts\Pagination\LengthAwarePaginator

     */

    public function paginate($filters = null, $perPage = 15, $pageName = 'page', $page = null, $sort=null) {

        assert(is_numeric($perPage));

        if(is_null($page)) {