Travaux en cours sur le système de recherche textuel (3d72a3ad) · Commits · open-source / oscar

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

+114 −40

Original line number	Diff line number	Diff line
		@@ -27,6 +27,57 @@ class ElasticActivitySearch extends ElasticSearchEngine implements IActivitySear
		return $this->addItem($activity);
		}

		protected function indicesCreateBody(): array
		{
		return [
		'settings' => [
		'analysis' => [
		'analyzer' => [
		'folding_analyzer' => [
		'tokenizer' => 'standard',
		'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,
		'properties' => [
		'id' => ['type' => 'keyword'],
		'acronym' => [ 'type' => 'text', 'analyzer' => 'folding_analyzer'],
		'eotp' => [ 'type' => 'text', 'analyzer' => 'folding_analyzer'],
		'oscar' => [ 'type' => 'text', 'analyzer' => 'folding_analyzer'],
		'saic' => [ 'type' => 'text', 'analyzer' => 'folding_analyzer'],
		'numerotation' => [ 'type' => 'text', 'analyzer' => 'folding_analyzer'],
		'numbers' => [ 'type' => 'text', 'analyzer' => 'folding_analyzer'],

		'label' => [ 'type' => 'text', 'analyzer' => 'folding_analyzer'],
		'description' => [ 'type' => 'text', 'analyzer' => 'folding_analyzer'],
		]
		]
		];
		}


		/**
		* @param mixed $object
		* @return array
		@@ -36,8 +87,7 @@ class ElasticActivitySearch extends ElasticSearchEngine implements IActivitySear
		{
		if (!$object instanceof Activity) {
		throw new OscarException("L'objet indexé doit être une activité");
		}
		else {
		} else {
		$activity = $object;
		}

		@@ -150,44 +200,68 @@ class ElasticActivitySearch extends ElasticSearchEngine implements IActivitySear
		]
		];

		// TODO si plusieurs mots, ajouter une règle spécifique
		if ($wordsNbr > 1) {
		$query["bool"]["should"][] = [
		"query_string" => [
		"query" => $andQuery,
		"fields" => [
		'oscar^20',
		'eotp^20',
		'acronym^15',
		'numbers^10',
		'motscles^7',
		'disciplines^7',
		'label^5',
		'project^5',
		'activitytype^3',
		'description^2',
		'partners',
		'members'
		]
		]
		];
		}
		else {
		$query["bool"]["should"] = [
		["match" => ["oscar" => ["query" => $andQuery, "boost" => 20]]],
		["match" => ["eotp" => ["query" => $andQuery, "boost" => 20]]],
		["match" => ["acronym" => ["query" => $andQuery, "boost" => 15]]],
		["match" => ["numbers" => ["query" => $andQuery, "boost" => 10]]],
		["match" => ["motscles" => ["query" => $andQuery, "boost" => 7]]],
		["match" => ["disciplines" => ["query" => $andQuery, "boost" => 7]]],
		["match" => ["label" => ["query" => $andQuery, "boost" => 5]]],
		["match" => ["activitytype" => ["query" => $andQuery, "boost" => 3]]],
		["match" => ["project" => ["query" => $andQuery, "boost" => 5]]],
		["match" => ["description" => ["query" => $andQuery, "boost" => 1]]],
		["match" => ["partners" => ["query" => $andQuery, "boost" => 1]]],
		["match" => ["members" => ["query" => $andQuery, "boost" => 1]]],
		];
		}

		//// // Champs Oscar
		// $query["bool"]["should"][] = [ "match" => [ "oscar" => [ "query" => $search, "boost" => 20 ]]];
		// $query["bool"]["should"][] = [ "prefix" => [ "oscar" => [ "value" => $search, "boost" => 5]]];
		////
		//// // Champ EOTP
		// $query["bool"]["should"][] = [ "match" => [ "eotp" => [ "query" => $search, "boost" => 20 ]]];
		// $query["bool"]["should"][] = [ "prefix" => [ "eotp" => [ "value" => $search, "boost" => 5]]];

		// // Champ ACRONYM
		// $query["bool"]["should"][] = [ "match" => [ "acronym" => [ "query" => $search, "fuzziness" => "auto", "boost" => 20 ]]];
		// $query["bool"]["should"][] = [ "prefix" => [ "acronym" => [ "value" => strtolower($search), "boost" => 5]]];

		// Champ LABEL/PROJECT LABEL
		$query["bool"]["should"][] = [ "match" => [ "label" => [ "query" => $search, "fuzziness" => "auto", "boost" => 4 ]]];
		$query["bool"]["should"][] = [ "match" => [ "project" => [ "query" => $search, "fuzziness" => "auto", "boost" => 3 ]]];

		// Champ NUMBERS
		$query["bool"]["should"][] = [ "match" => [ "numbers" => [ "query" => $search, "fuzziness" => "auto", "boost" => 5 ]]];





		/**
		* "query_string" => [
		* "query" => $andQuery,
		* "fields" => [
		* 'oscar^20',
		* 'eotp^20',
		* 'acronym^15',
		* 'numbers^10',
		* 'motscles^7',
		* 'disciplines^7',
		* 'label^5',
		* 'project^5',
		* 'activitytype^3',
		* 'description^2',
		* 'partners',
		* 'members'
		* ]
		* ]
		* ];
		* var_dump($query); die();
		* }
		* else {
		* $query["bool"]["should"] = [
		* ["match" => ["oscar" => ["query" => $andQuery, "boost" => 20]]],
		* ["match" => ["eotp" => ["query" => $andQuery, "boost" => 20]]],
		* ["match" => ["acronym" => ["query" => $andQuery, "boost" => 15]]],
		* ["match" => ["numbers" => ["query" => $andQuery, "boost" => 10]]],
		* ["match" => ["motscles" => ["query" => $andQuery, "boost" => 7]]],
		* ["match" => ["disciplines" => ["query" => $andQuery, "boost" => 7]]],
		* ["match" => ["label" => ["query" => $andQuery, "boost" => 5]]],
		* ["match" => ["activitytype" => ["query" => $andQuery, "boost" => 3]]],
		* ["match" => ["project" => ["query" => $andQuery, "boost" => 5]]],
		* ["match" => ["description" => ["query" => $andQuery, "boost" => 1]]],
		* ["match" => ["partners" => ["query" => $andQuery, "boost" => 1]]],
		* ["match" => ["members" => ["query" => $andQuery, "boost" => 1]]],
		* ];
		* }
		* ****/

		return $query;
		}

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

+13 −19

Original line number	Diff line number	Diff line
		@@ -57,27 +57,11 @@ abstract class ElasticSearchEngine
		->build();
		}
		if( !$this->elasticSearchClient->indices()->exists(['index' => $this->getIndex()]) ){
		$this->loggerService->debug("INDEX CREATE " . $this->getIndex());
		$this->loggerService->info("INDEX CREATE " . $this->getIndex());
		$this->elasticSearchClient->indices()->create(
		[
		'index' => $this->getIndex(),
		'body' => [
		'mappings' => [
		// Suppression de la détection des dates/nombres
		'date_detection' => false,
		'numeric_detection' => false,
		"dynamic_templates"=> [
		[
		"numerique_en_texte"=> [
		"match_mapping_type"=> "long",
		"mapping"=> [
		"type"=> "text"
		]
		]
		]
		]
		]
		]
		'body' => $this->indicesCreateBody()
		]
		);
		}
		@@ -89,6 +73,16 @@ abstract class ElasticSearchEngine
		}
		}

		protected function indicesCreateBody(): array {
		return [
		'mappings' => [
		// Suppression de la détection des dates/nombres
		'date_detection' => false,
		'numeric_detection' => false
		]
		];
		}

		public function getMapping() :array {
		return $this->getClient()->indices()->getMapping(['index' => $this->getIndex()]);
		}
		@@ -194,7 +188,7 @@ abstract class ElasticSearchEngine
		]
		];

		$this->loggerService->debug("----query elastic : \n" . json_encode($query['body']['query']) ."\n---- ");
		$this->loggerService->info("----query elastic : \n" . json_encode($query['body']['query']) ."\n---- ");

		return $query;
		}

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

+84 −10

Original line number	Diff line number	Diff line
		@@ -38,6 +38,50 @@ class PersonElasticSearch extends ElasticSearchEngine implements IPersonISearchS
		return $this->addItem($person);
		}

		protected function indicesCreateBody(): array {
		return [
		'settings' => [
		'analysis' => [
		'analyzer' => [
		'folding_analyzer' => [
		'tokenizer' => 'standard',
		'filter' => ['lowercase', 'asciifolding']
		]
		]
		]
		],
		'mappings' => [
		'properties' => [
		'id' => ['type' => 'keyword'],
		'firstname' => [
		'type' => 'text',
		'analyzer' => 'folding_analyzer'
		],
		'lastname' => [
		'type' => 'text',
		'analyzer' => 'folding_analyzer'
		],
		'fullname' => [
		'type' => 'text',
		'analyzer' => 'folding_analyzer'
		],
		'email' => [
		'type' => 'text',
		'analyzer' => 'folding_analyzer'
		],
		'affectation' => ['type' => 'text'],
		'location' => ['type' => 'text'],
		'organizations' => ['type' => 'text'],
		'activities' => ['type' => 'text'],
		'connectors' => [
		'type' => 'text',
		'analyzer' => 'folding_analyzer'
		]
		]
		]
		];
		}

		public function getIndexableDatas(mixed $item) :array
		{
		if( !$item instanceof Person ){
		@@ -100,11 +144,12 @@ class PersonElasticSearch extends ElasticSearchEngine implements IPersonISearchS
		return [
		"bool" => [
		"should" => [
		// ---- Fuzzy matches ----
		[
		"match" => [
		"connectors" => [
		"query" => $search,
		"boost" => 6 // Favoriser les correspondances
		"boost" => 6
		]
		]
		],
		@@ -112,8 +157,8 @@ class PersonElasticSearch extends ElasticSearchEngine implements IPersonISearchS
		"match" => [
		"lastname" => [
		"query" => $search,
		'fuzziness' => "AUTO", // "Tolérance" aux fautes
		"boost" => 5 // Favoriser les correspondances
		"fuzziness" => "AUTO",
		"boost" => 5
		]
		]
		],
		@@ -121,7 +166,7 @@ class PersonElasticSearch extends ElasticSearchEngine implements IPersonISearchS
		"match" => [
		"fullname" => [
		"query" => $search,
		'fuzziness' => "AUTO", // "Tolérance" aux fautes,
		"fuzziness" => "AUTO",
		"boost" => 3
		]
		]
		@@ -130,7 +175,7 @@ class PersonElasticSearch extends ElasticSearchEngine implements IPersonISearchS
		"match" => [
		"firstname" => [
		"query" => $search,
		'fuzziness' => "AUTO", // "Tolérance" aux fautes,
		"fuzziness" => "AUTO",
		"boost" => 2
		]
		]
		@@ -139,7 +184,7 @@ class PersonElasticSearch extends ElasticSearchEngine implements IPersonISearchS
		"match" => [
		"email" => [
		"query" => $search,
		'fuzziness' => "AUTO", // "Tolérance" aux fautes,
		"fuzziness" => "AUTO",
		"boost" => 1
		]
		]
		@@ -148,7 +193,7 @@ class PersonElasticSearch extends ElasticSearchEngine implements IPersonISearchS
		"match" => [
		"location" => [
		"query" => $search,
		'fuzziness' => "AUTO", // "Tolérance" aux fautes,
		"fuzziness" => "AUTO"
		]
		]
		],
		@@ -156,18 +201,47 @@ class PersonElasticSearch extends ElasticSearchEngine implements IPersonISearchS
		"match" => [
		"affectation" => [
		"query" => $search,
		'fuzziness' => "AUTO", // "Tolérance" aux fautes,
		"fuzziness" => "AUTO"
		]
		]
		],

		// ---- Commence par (prefix) ----
		[
		"prefix" => [
		"lastname" => $search, // Documents qui commencent par l'expression
		"lastname" => [
		"value" => $search,
		"boost" => 10
		]
		]
		],
		[
		"prefix" => [
		"connectors" => $search, // Documents qui commencent par l'expression
		"connectors" => [
		"value" => $search,
		"boost" => 8
		]
		]
		],

		// ---- Autocomplétion et Prénom + Nom ----
		[
		"match_phrase_prefix" => [
		"fullname" => [
		"query" => $search,
		"slop" => 2, // tolère inversion prénom/nom
		"max_expansions" => 20,
		"boost" => 12
		]
		]
		],
		[
		"multi_match" => [
		"query" => $search,
		"type" => "cross_fields",
		"fields" => ["firstname^2", "lastname^3"],
		"operator" => "and",
		"boost" => 9
		]
		]
		],