Patch recherche activités (95f78d66) · Commits · open-source / oscar

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

+228 −69

Original line number	Diff line number	Diff line
		@@ -64,8 +64,8 @@ class ElasticActivitySearch extends ElasticSearchEngine implements IActivitySear
		'properties' => [
		'id' => ['type' => 'keyword'],
		'acronym' => ['type' => 'text', 'analyzer' => 'folding_analyzer'],
		'eotp' => [ 'type' => 'text', 'analyzer' => 'folding_analyzer'],
		'oscar' => [ 'type' => 'text', 'analyzer' => 'folding_analyzer'],
		'eotp' => ['type' => 'keyword' ],
		'oscar' => ['type' => 'keyword'],
		'saic' => ['type' => 'text', 'analyzer' => 'folding_analyzer'],
		'numerotation' => ['type' => 'text', 'analyzer' => 'folding_analyzer'],
		'numbers' => ['type' => 'text', 'analyzer' => 'folding_analyzer'],
		@@ -132,16 +132,45 @@ class ElasticActivitySearch extends ElasticSearchEngine implements IActivitySear

		public function searchProject($what): array
		{
		$analysis = $this->analyzeSearchPattern($what);
		$normalizedSearch = $analysis['search'];
		$isExact = $analysis['isExact'];
		$isPrefix = $analysis['isPrefix'];

		// Build query based on pattern
		if ($isExact) {
		// Exact match - use match_phrase for project field
		$query = [
		'match_phrase' => [
		'project' => [
		'query' => $normalizedSearch
		]
		]
		];
		} elseif ($isPrefix) {
		// Prefix search - use prefix query
		$query = [
		'prefix' => [
		'project' => [
		'value' => $normalizedSearch
		]
		]
		];
		} else {
		// Default - use query_string with OR for prefix matching
		$query = [
		'query_string' => [
		'query' => sprintf('%s OR %s*', $normalizedSearch, $normalizedSearch)
		]
		];
		}

		$params = [
		'index' => $this->getIndex(),
		'type' => $this->getType(),
		'body' => [
		'size' => 10000,
		'query' => [
		'query_string' => [
		'query' => sprintf('%s OR %s*', $what, $what)
		]
		]
		'query' => $query
		]
		];

		@@ -157,78 +186,208 @@ class ElasticActivitySearch extends ElasticSearchEngine implements IActivitySear
		return $ids;
		}

		public function getFieldsSearchedWeighted(string $search): array
		/**
		* 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
		{
		$andQuery = null;
		$wordsNbr = 1;

		// Détection des recherches strictes
		if (preg_match_all('/^"(.*)"$/', $search, $matches, PREG_SET_ORDER)) {
		if (count($matches) == 1 && array_key_exists(1, $matches[0])) {
		$andQuery = $matches[0][0];
		$wordsNbr = 2;
		$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);
		}
		}

		// if(preg_match_all('/^query:(.*)$/', $search, $matches, PREG_SET_ORDER)) {
		// if( count($matches) == 1 && array_key_exists(1, $matches[0])) {
		// $andQuery = $matches[0][1];
		// $wordsNbr = 2;
		// }
		// }

		if (preg_match_all('/^(.)\$/', $search, $matches, PREG_SET_ORDER)) {
		if (count($matches) == 1 && array_key_exists(1, $matches[0])) {
		$andQuery = $matches[0][0];
		$wordsNbr = 2;
		}
		// Check for prefix search (ends with *)
		if (str_ends_with($normalizedSearch, '*')) {
		$isPrefix = true;
		$normalizedSearch = rtrim($normalizedSearch, '*');
		}

		if ($andQuery === null) {
		$words = explode(" ", $search);
		$wordsNbr = count($words);
		$andQuery = implode(" AND ", $words);
		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]]];
		}

		$query = [
		"bool" => [
		"should" => [
		// N° financier (EOTP/PFI)
		if ($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]]];
		}

		],
		"minimum_should_match" => 1
		]
		];
		// 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]]];
		}

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

		// Champ EOTP
		$query["bool"]["should"][] = [ "match" => [ "eotp" => [ "query" => $search, "boost" => 20 ]]];
		$query["bool"]["should"][] = [ "prefix" => [ "eotp" => [ "value" => $search, "boost" => 5]]];
		// Project
		if ($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']]];
		}

		// Champ ACRONYM
		$query["bool"]["should"][] = [ "match" => [ "acronym" => [ "query" => $search, "fuzziness" => 1, "boost" => 20 ]]];
		$query["bool"]["should"][] = [ "prefix" => [ "acronym" => [ "value" => strtolower($search), "boost" => 5]]];
		// 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]]];
		}

		// Champ LABEL/PROJECT LABEL
		$query["bool"]["should"][] = [ "match" => [ "label" => [ "query" => $search, "fuzziness" => 1, "boost" => 4 ]]];
		$query["bool"]["should"][] = [ "match" => [ "project" => [ "query" => $search, "fuzziness" => 1, "boost" => 3 ]]];
		// Description
		if ($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]]];
		}

		// Champ NUMBERS
		$query["bool"]["should"][] = [ "match" => [ "numbers" => [ "query" => $search, "fuzziness" => 1, "boost" => 5 ]]];
		// Numbers
		if ($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]]];
		}

		// Champ MOTSCLES
		$query["bool"]["should"][] = [ "match" => [ "motscles" => [ "query" => $search, "fuzziness" => 1, "boost" => 5 ]]];
		$query["bool"]["should"][] = [ "match" => [ "disciplines" => [ "query" => $search, "fuzziness" => 1, "boost" => 5 ]]];
		// Activity type
		if ($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]]];
		}

		$query["bool"]["should"][] = [ "match" => [ "activitytype" => [ "query" => $search, "fuzziness" => 1, "boost" => 3 ]]];
		$query["bool"]["should"][] = [ "match" => [ "description" => [ "query" => $search, "fuzziness" => 1, "boost" => 2 ]]];
		$query["bool"]["should"][] = [ "match" => [ "partners" => [ "query" => $search, "fuzziness" => 1]]];
		$query["bool"]["should"][] = [ "match" => [ "members" => [ "query" => $search, "fuzziness" => 1]]];
		// Partners
		if ($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]]];
		}

		return $query;
		}