<?php

namespace CorpusParole\Console\Commands;

use Illuminate\Console\Command;

use EasyRdf\Resource;

use EasyRdf\Literal;

use EasyRdf\Graph;

use Carbon\Carbon;

use GuzzleHttp\Client;

use GuzzleHttp\Exception\TransferException;

use GuzzleHttp\Psr7;

use CorpusParole\Libraries\Utils;

use CorpusParole\Repositories\DocumentRepository;

use CorpusParole\Libraries\CocoonUtils;

use CorpusParole\Models\GeonamesHierarchy;

use CorpusParole\Services\BnfResolverInterface;

use CorpusParole\Services\LexvoResolverInterface;

use Es;

use Log;

use Cache;

class IndexDocuments extends Command

{

    /**

     * The name and signature of the console command.

     *

     * @var string

     */

    protected $signature = 'corpus-parole:indexDocuments

                          {--limit=0 : index only the first n documents, 0 (default) means index everything }

                          {--no-bulk : index documents one by one instead of using ElasticSearch bulk indexing }

                          {--step-size=100 : number of documents to retrieve from repository at a time before indexing}

                          {--reset-geo-cache : reset geo cache befr indexing}';

    /**

     * The console command description.

     *

     * @var string

     */

    protected $description = 'Index documents into ElasticSearch.';

    /**

     * Create a new command instance.

     *

     * @return void

     */

    public function __construct(

        DocumentRepository $documentRepository,

        Client $httpClient,

        BnfResolverInterface $bnfResolver,

        LexvoResolverInterface $lexvoResolver)

    {

        $this->documentRepository = $documentRepository;

        $this->bnfResolver = $bnfResolver;

        $this->lexvoResolver = $lexvoResolver;

        $this->httpClient = $httpClient;

        parent::__construct();

    }

    /**

     * Reset Elasticsearch index

     *

     * @return int (1 if sucess, 0 if error)

     */

    private function resetIndex()

    {

        $indexParams = [

            'index' => config('elasticsearch.index')

        ];

        if(Es::indices()->exists($indexParams)){

            $response = Es::indices()->delete($indexParams);

            if($response['acknowledged']!=1){

                return 0;

            }

        }

        // Note: removed the "'store' => True" parameters on fields and use _source on record instead

        $indexParams['body'] = [

            'settings' => [

                'number_of_shards' => config('elasticsearch.shards'),

                'number_of_replicas' => config('elasticsearch.replicas'),

                'index.mapping.ignore_malformed' => True,

                'index.requests.cache.enable' => True

            ],

            'mappings' => [

                'document' => [

                    'properties' => [

                        'title' => [

                            'type' => 'string',

                            'fields' => [

                                'raw' => [

                                    'type' => 'string',

                                    'index' => 'not_analyzed'

                                ]

                            ]

                        ],

                        'description' => [

                            'type' => 'string',

                            'fields' => [

                                'french' => [

                                    'type' => 'string',

                                    'analyzer' => 'french'

                                ]

                            ]

                        ],

                        'date' => [ 'type' => 'date', 'index' => 'not_analyzed'],

                        'geonames_hierarchy' => [ 'type' => 'string', 'index' => 'not_analyzed'],

                        'geonames_country' => ['type' => 'string', 'index' => 'not_analyzed'],

                        'location' => [ 'type' => 'geo_point'],

                        'creation_date' => ['type' => 'date', 'index' => 'not_analyzed'],

                        'language' => ['type' => 'string', 'index' => 'not_analyzed'],

                        'discourse_types' => ['type' => 'string', 'index' => 'not_analyzed'],

                        'creation_years' => [

                            'type' => 'nested',

                            'properties' => [

                                'year' => [ 'type' => 'short', 'index' => 'not_analyzed'],

                                'weight' => [ 'type' => 'float', 'index' => 'not_analyzed'],

                            ]

                        ] ,

                        'subject' => [

                            'type' => 'nested',

                            'properties' => [

                                'label' => [ 'type' => 'string', 'index' => 'not_analyzed'],

                                'code' => [ 'type' => 'string', 'index' => 'not_analyzed'],

                                'label_code' => [ 'type' => 'string', 'index' => 'not_analyzed']

                            ]

                        ]

                    ]

                ]

            ]

        ];

        $response = Es::indices()->create($indexParams);

        if($response['acknowledged']!=1){

            return 0;

        }

        return 1;

    }

    private function getGeonamesHierarchyArray($geonamesid) {

        $hcache = GeonamesHierarchy::where('geonamesid', $geonamesid)->first();

        if(is_null($hcache)) {

            // TODO: add delay to respect geonames 2k request/hour

            // TODO: manage errors

            $apiBody = $this->httpClient->get(

                config('corpusparole.geonames_hierarchy_webservice_url'),

                [ 'query' =>

                    [ 'geonameId' => $geonamesid,

                      'username' => config('corpusparole.geonames_username') ],

                  'accept' => 'application/json' // TODO: check this

                ]

            )->getBody();

            $hjson = json_decode($apiBody);

            $hcache = new GeonamesHierarchy();

            $hcache->geonamesid = $geonamesid;

            $hcache->hierarchy = $hjson;

            $hcache->save();

        }

        $res = [];

        $resCountry = null;

        foreach($hcache->hierarchy['geonames'] as $hierarchyElem) {

            if(in_array($hierarchyElem['fcode'], ['CONT','PCLI', 'PCL','PCLD', 'PCLF', 'PCLH', 'PCLIX', 'PCLIS', 'ADM1'])) {

                array_push($res, $hierarchyElem['geonameId']);

            }

            if(!empty($hierarchyElem['fcode']) && strpos($hierarchyElem['fcode'], 'PCL') === 0) {

                $resCountry = $hierarchyElem['geonameId'];

            }

        }

        return [$resCountry, $res];

    }

    /**

     * get geonames hierarchy data.

     * @return array list of geonames ids

     */

    private function getGeonamesHierarchy($doc) {

        $geoRes = $doc->getGeoInfo();

        if(is_null($geoRes)) {

            return [null,[]];

        }

        // aggregate hierachy list from geonames results

        $res = [];

        // The country is the first one

        $resCountry = null;

        foreach($geoRes->getGeonamesLocs() as $gurl) {

            $geonamesId = CocoonUtils::getGeonamesidFromUrl($gurl);

            if(is_null($geonamesId)) {

                continue;

            }

            list($country, $hierarchyIds) = $this->getGeonamesHierarchyArray($geonamesId);

            $res = array_unique(array_merge($res, $hierarchyIds));

            if(is_null($resCountry) && !empty($country)) {

                $resCountry = $country;

            }

        }

        return [$resCountry, $res];

    }

    /**

     * get subjects as { 'label': label, 'code': code } objects

     * Takes only into account the bnf subjects

     */

    private function getSubjects($doc) {

        $sres = array_reduce($doc->getSubjects(), function($res, $s) {

            $mBnf = [];

            $mLexvo = [];

            if($s instanceof Resource && preg_match(config('corpusparole.bnf_ark_url_regexp'), $s->getUri(), $mBnf) === 1) {

                array_push($res, [

                    'uri' => $mBnf[0],

                    'code' => $mBnf[1],

                    'type' => Utils::SUBJECT_TYPE_BNF

                ]);

            } elseif($s instanceof Resource && preg_match(config('corpusparole.lexvo_url_regexp'), $s->getUri(), $mLexvo) === 1) {

                array_push($res, [

                    'uri' => $mLexvo[0],

                    'code' => $mLexvo[1],

                    'type' => Utils::SUBJECT_TYPE_LEXVO

                ]);

            } elseif($s instanceof Literal && strpos($s->getDatatypeUri(), config('corpusparole.olac_base_url')) === 0 ) {

                array_push($res, [

                    'uri' => $s->getValue(),

                    'code' => $s->getValue(),

                    'type' => Utils::SUBJECT_TYPE_OLAC

                ]);

            } elseif($s instanceof Literal) {

                array_push($res, [

                    'uri' => $s->getValue(),

                    'code' => $s->getValue(),

                    'type' => Utils::SUBJECT_TYPE_TXT

                ]);

            }

            return $res;

        }, []);

        $labelsLexvo = $this->lexvoResolver->getNames(

            array_unique(array_reduce(

                $sres,

                function($r, $so) {

                    if($so['type'] === Utils::SUBJECT_TYPE_LEXVO) {

                        array_push($r, $so['uri']);

                    }

                    return $r;

                },[]

            ))

        );

        return array_map(function($so) use ($labelsBnf, $labelsLexvo) {

            $label = $so['uri'];

            if($so['type'] === Utils::SUBJECT_TYPE_BNF) {

                $label = $labelsBnf[$label];

            } elseif ($so['type'] === Utils::SUBJECT_TYPE_LEXVO) {

                $label = $labelsLexvo[$label];

            }

            return [ 'label' => $label, 'code' => $so['type']."|".$so['code'], 'label_code' =>  $label."|".$so['type']."|".$so['code'] ]; }, $sres

        );

    }

    private function graphResolvCoordinate($loc, $graph) {

        $latLit = $graph->getLiteral($loc, "<http://www.w3.org/2003/01/geo/wgs84_pos#lat>");

        if(is_null($latLit) || empty($latLit->getValue())) {

            return null;

        }

        $lat = $latLit->getValue();

        $longLit = $graph->getLiteral($loc, "<http://www.w3.org/2003/01/geo/wgs84_pos#long>");

        if(is_null($longLit) || empty($longLit->getValue())) {

            return null;

        }

        $long = $longLit->getValue();

        return [ $lat, $long ];

    }

    private function loadGraph($url, $type) {

        try {

            $r = $this->httpClient->get($url);

        } catch (TransferException $e) {

            $this->error("loadGraph : Error Loading $url");

            Log::error("loadGraph : Error Loading $url");

            Log::error("loadGraph : Error request " . Psr7\str($e->getRequest()));

            if ($e->hasResponse()) {

                $this->error("loadGraph : Error response " . Psr7\str($e->getResponse()));

                Log::error("loadGraph : Error response " . Psr7\str($e->getResponse()));

            }

            return null;

        }

        try {

            $message = (string)$r->getBody();

            $graph = new Graph($url, $message, $type);

            return $graph;

        } catch (EasyRdf\Exception $e) {

            $this->error("loadGraph : Error parsing $url");

            Log::error("loadGraph : Error parsing $url");

            if($e instanceof EasyRdf\Parser\Exception) {

                Log::error("loadGraph : Error exception line ".$e->getLine().", column: ".$e->getColumn());

            }

            $this->error("loadGraph : Error exception message ".$e->getMessage());

            Log::error("loadGraph : Error exception message ".$e->getMessage());

            Log::error("loadGraph : Error content $message");

            return null;

        }

    }

    private function geonamesResolveCoordinates($loc) {

        $coords = cache("corpus.geonames.coord.$loc");

        if(is_null($coords)) {

            $graph = $this->loadGraph("{$loc}about.rdf", 'rdfxml');

            $coords = is_null($graph)?null:$this->graphResolvCoordinate($loc, $graph);

            cache(["corpus.geonames.coord.$loc" => is_null($coords)?false:$coords], Carbon::now()->addMinutes(20));

        }

        return ($coords===false)?null:$coords;

    }

    private function dbpediaResolveCoordinates($loc) {

        $coords = cache("corpus.dbpedia.coord.$loc");

        if(is_null($coords)) {

            $graph = $this->loadGraph("$loc.rdf", 'rdfxml');

            $coords = is_null($graph)?null:$this->graphResolvCoordinate($loc, $graph);

            cache(["corpus.dbpedia.coord.$loc"=> is_null($coords)?false:$coords], Carbon::now()->addMinutes(20));

        }

        return ($coords===false)?null:$coords;

    }

    private function getLocation($doc) {

        $geoRes = $doc->getGeoInfo();

        if(is_null($geoRes)) {

            return null;

        }

        $locUrls = [];

        foreach($geoRes->getRefLocs() as $loc) {

            if(preg_match(config('corpusparole.geonames_url_regexp'), $loc, $m) === 1) {

                if(!array_key_exists('geonames', $locUrls)) {

                    $locUrls['geonames'] = [];

                }