Commit 77a33a7b authored by Stephane Bouvry's avatar Stephane Bouvry
Browse files

MAJ de la recherche : 

 - Modification des poids
 - Traitement du _ comme un tokenizer
 - Passage en query_string (Permet des recherches avec *)
 - Opérateur AND pour les recherches à plusieurs termes
parent 95f78d66

module/Oscar/src/Oscar/Strategy/Search/ElasticActivitySearch.php

+13 −214
Original line number Diff line number Diff line
@@ -32,32 +32,21 @@ class ElasticActivitySearch extends ElasticSearchEngine implements IActivitySear 
        return [

            'settings' => [

                'analysis' => [

                    'tokenizer' => [

                        // Considérer les _ comme des séparateurs

                        'underscore_tokenizer' => [

                            'type' => 'pattern',

                            'pattern' => '[\W_]+'

                        ]

                    ],

                    'analyzer' => [

                        'folding_analyzer' => [

                            'tokenizer' => 'standard',

                            'tokenizer' => 'underscore_tokenizer',

                            'filter' => ['lowercase', 'asciifolding']

                        ]

                    ]

                ]

            ],

            /****

             * 'label' => $activity->getLabel(),

             * 'description' => $activity->getDescription(),

             * 'saic' => $activity->getCentaureId(),

             * 'numerotation' => $activity->getNumbers(),

             * 'oscar' => $activity->getOscarNum(),

             * 'activitytype' => $activity->getActivityType() ? (string)$activity->getActivityType() : '',

             * 'numbers' => implode(" ", $activity->getNumbersValues()),

             * 'eotp' => $activity->getCodeEOTP(),

             * 'acronym' => $activity->getAcronym(),

             * 'activity_id' => $activity->getId(),

             * 'disciplines' => $activity->getDisciplinesArray(),

             * 'motscles' => $activity->getMotsclesArray(),

             * 'members' => $members,

             * 'partners' => $partners,

             * 'project_id' => $project_id,

             * 'project' => $project_body,

             * /****/

            'mappings' => [

                'date_detection' => false,

                'numeric_detection' => false,
@@ -186,208 +175,18 @@ class ElasticActivitySearch extends ElasticSearchEngine implements IActivitySear 
        return $ids;

    }



    /**

     * Analyze the search pattern and return normalized search term with flags

     * @param string $search The search term

     * @return array{search: string, isExact: bool, isPrefix: bool}

     */

    private function analyzeSearchPattern(string $search): array

    {

        $isExact = false;

        $isPrefix = false;

        $normalizedSearch = $search;



        // Check for exact match (quoted text)

        if (strlen($search) >= 2 && str_starts_with($search, '"') && str_ends_with($search, '"')) {

            $isExact = true;

            $normalizedSearch = substr($search, 1, -1);

        }



        // Check for prefix search (ends with *)

        if (str_ends_with($normalizedSearch, '*')) {

            $isPrefix = true;

            $normalizedSearch = rtrim($normalizedSearch, '*');

        }



        return [

            'search' => $normalizedSearch,

            'isExact' => $isExact,

            'isPrefix' => $isPrefix

        ];

    }



    public function getFieldsSearchedWeighted(string $search): array

    {

        $analysis = $this->analyzeSearchPattern($search);

        $normalizedSearch = $analysis['search'];

        $isExact = $analysis['isExact'];

        $isPrefix = $analysis['isPrefix'];



        $query = ['bool' => [

            'should' => [],

            'minimum_should_match' => 1

        ]];



        ///////////////////////////////////////////// IDENTIFIANTS/KEYWORDS

        // N° Oscar

        if ($isExact) {

            $query['bool']['should'][] = ['term' => ['oscar' => [

                'value' => $normalizedSearch, 'boost' => 200]]];

        } elseif ($isPrefix) {

            $query['bool']['should'][] = ['prefix' => ['oscar' => [

                'value' => $normalizedSearch, 'boost' => 100]]];

        } else {

            $query['bool']['should'][] = ['term' => ['oscar' => [

                'value' => $normalizedSearch, 'boost' => 100]]];

//            $query['bool']['should'][] = ['prefix' => ['oscar' => [

//                'value' => $normalizedSearch, 'boost' => 30]]];

        }



        // N° financier (EOTP/PFI)

        if ($isExact) {

            $query['bool']['should'][] = ['term' => ['eotp' => [

                'value' => $normalizedSearch, 'boost' => 200]]];

        } elseif ($isPrefix) {

            $query['bool']['should'][] = ['prefix' => ['eotp' => [

                'value' => $normalizedSearch, 'boost' => 100]]];

        } else {

            $query['bool']['should'][] = ['term' => ['eotp' => [

                'value' => $normalizedSearch, 'boost' => 100]]];

//            $query['bool']['should'][] = ['prefix' => ['eotp' => [

//                'value' => $normalizedSearch, 'boost' => 30]]];

        }



        // Acronyme

        if ($isExact) {

            $query['bool']['should'][] = ['term' => ['acronym.keyword' => [

                'value' => $normalizedSearch, 'boost' => 160]]];

        } elseif ($isPrefix) {

            $query['bool']['should'][] = ['prefix' => ['acronym.keyword' => [

                'value' => $normalizedSearch, 'boost' => 100]]];

        } else {

            $query['bool']['should'][] = ['term' => ['acronym.keyword' => [

                'value' => $normalizedSearch, 'boost' => 80]]];

//            $query['bool']['should'][] = ['prefix' => ['acronym.keyword' => [

//                'value' => $normalizedSearch, 'boost' => 20]]];

            $query['bool']['should'][] = ['match' => ['acronym' => [

                'query' => $normalizedSearch, 'boost' => 10, 'fuzziness' => 10]]];

        }



        ///////////////////////////////////////////// TEXTES

        // Note doc :

        // 'match_phrase' : Terme exacte

        // 'prefix' : Commence par

        //  'match_bool_prefix' / 'match' : Recherche par défaut

        

        // Label

        if ($isExact) {

            $query['bool']['should'][] = ['match_phrase' => ['label' => [

                'query' => $normalizedSearch, 'boost' => 20]]];

        } elseif ($isPrefix) {

            $query['bool']['should'][] = ['prefix' => ['label' => [

                'value' => $normalizedSearch, 'boost' => 15]]];

        } else {

            $query['bool']['should'][] = ['match_bool_prefix' => ['label' => [

                'query' => $normalizedSearch, 'boost' => 6]]];

            $query['bool']['should'][] = ['match' => ['label' => [

                'query' => $normalizedSearch, 'boost' => 3, 'fuzziness' => 1]]];

        }



        // Project

        if ($isExact) {

            $query['bool']['should'][] = ['match_phrase' => ['project' => [

                'query' => $normalizedSearch, 'boost' => 10]]];

        } elseif ($isPrefix) {

            $query['bool']['should'][] = ['prefix' => ['project' => [

                'value' => $normalizedSearch, 'boost' => 8]]];

        } else {

            $query['bool']['should'][] = ['match' => ['project' => [

                'query' => $normalizedSearch, 'boost' => 2, 'fuzziness' => 'AUTO']]];

        }



        // Mots clés

        if ($isExact) {

            $query['bool']['should'][] = ['match_phrase' => ['motscles' => [

                'query' => $normalizedSearch, 'boost' => 10]]];

        } elseif ($isPrefix) {

            $query['bool']['should'][] = ['prefix' => ['motscles' => [

                'value' => $normalizedSearch, 'boost' => 8]]];

        } else {

            $query['bool']['should'][] = ['match' => ['motscles' => [

                'query' => $normalizedSearch, 'boost' => 2]]];

        }



        // Description

        if ($isExact) {

            $query['bool']['should'][] = ['match_phrase' => ['description' => [

                'query' => $normalizedSearch, 'boost' => 5]]];

        } elseif ($isPrefix) {

            $query['bool']['should'][] = ['prefix' => ['description' => [

                'value' => $normalizedSearch, 'boost' => 4]]];

        } else {

            $query['bool']['should'][] = ['match' => ['description' => [

                'query' => $normalizedSearch, 'boost' => 1]]];

        }



        // Numbers

        if ($isExact) {

            $query["bool"]["should"][] = ["term" => ["numbers" => [

                "value" => $normalizedSearch, "boost" => 20]]];

        } elseif ($isPrefix) {

            $query["bool"]["should"][] = ["prefix" => ["numbers" => [

                "value" => $normalizedSearch, "boost" => 15]]];

        } else {

            $query["bool"]["should"][] = ["match" => ["numbers" => [

                "query" => $normalizedSearch, "fuzziness" => 1, "boost" => 10]]];

        }



        // Disciplines

        if ($isExact) {

            $query["bool"]["should"][] = ["match_phrase" => ["disciplines" => [

                "query" => $normalizedSearch, "boost" => 10]]];

        } elseif ($isPrefix) {

            $query["bool"]["should"][] = ["prefix" => ["disciplines" => [

                "value" => $normalizedSearch, "boost" => 8]]];

        } else {

            $query["bool"]["should"][] = ["match" => ["disciplines" => [

                "query" => $normalizedSearch, "fuzziness" => 1, "boost" => 5]]];

        }



        // Activity type

        if ($isExact) {

            $query["bool"]["should"][] = ["term" => ["activitytype" => [

                "value" => $normalizedSearch, "boost" => 6]]];

        } elseif ($isPrefix) {

            $query["bool"]["should"][] = ["prefix" => ["activitytype" => [

                "value" => $normalizedSearch, "boost" => 5]]];

        } else {

            $query["bool"]["should"][] = ["match" => ["activitytype" => [

                "query" => $normalizedSearch, "fuzziness" => 1, "boost" => 3]]];

        }



        // Partners

        if ($isExact) {

            $query["bool"]["should"][] = ["match_phrase" => ["partners" => [

                "query" => $normalizedSearch, "boost" => 5]]];

        } elseif ($isPrefix) {

            $query["bool"]["should"][] = ["prefix" => ["partners" => [

                "value" => $normalizedSearch, "boost" => 4]]];

        } else {

            $query["bool"]["should"][] = ["match" => ["partners" => [

                "query" => $normalizedSearch, "fuzziness" => 1]]];

        }



        // Members

        if ($isExact) {

            $query["bool"]["should"][] = ["match_phrase" => ["members" => [

                "query" => $normalizedSearch, "boost" => 5]]];

        } elseif ($isPrefix) {

            $query["bool"]["should"][] = ["prefix" => ["members" => [

                "value" => $normalizedSearch, "boost" => 4]]];

        } else {

            $query["bool"]["should"][] = ["match" => ["members" => [

                "query" => $normalizedSearch, "fuzziness" => 1]]];

        }

        $query['bool']['should'][] = ['query_string' => [

            'fields' => ['oscar^100','eotp^100','acronym^50','label^5','numbers^25','motsclefs^25','description^2','project','numbers.keyword'],

            'query' => $search,

            'default_operator' => 'AND'

        ]];



        return $query;

    }