crm.clientright.ru/aiassist/function.php

<?php
// aiassist/search.php
if (!defined('ELASTIC_URL')) {
    define('ELASTIC_URL', 'http://localhost:9200');
}


$logFile = __DIR__ . "/logs/search1.log";
$resultsDir = __DIR__ . "/logs/results/";

// Создаем папку для результатов, если её нет
if (!is_dir($resultsDir)) {
    mkdir($resultsDir, 0777, true);
}

/**
 * Функция нормализации эмбеддинга
 */
function normalizeEmbedding($vector) {
    $magnitude = sqrt(array_sum(array_map(function ($x) { return $x * $x; }, $vector)));
    return $magnitude > 0 ? array_map(function ($x) use ($magnitude) { return $x / $magnitude; }, $vector) : $vector;
}

/**
 * Функция поиска похожих дел
 */
function searchSimilarCases($queryParams, $size = 5) {
    global $logFile, $resultsDir;

    if (empty($queryParams['facts']) || empty($queryParams['category'])) {
        file_put_contents($logFile, "[" . date("Y-m-d H:i:s") . "] ❌ Ошибка: Недостаточно данных для поиска!\n", FILE_APPEND);
        return [];
    }

    // Получаем параметры из теста
    $fuzzinessLevel = $_GET['fuzziness'] ?? "AUTO";
    $yearFrom = (int) ($_GET['year_from'] ?? 2010);
    $amountFactor = (int) ($_GET['amount_factor'] ?? 30);

    // Фильтр по сумме
    $amount = isset($queryParams['amount']) && is_numeric($queryParams['amount']) ? floatval($queryParams['amount']) : null;

    // Нормализация эмбеддингов
    $vector2048 = isset($queryParams['embedding_2048']) ? $queryParams['embedding_2048'] : [];
    $normalizedEmbedding2048 = is_array($vector2048) && !empty($vector2048) ? normalizeEmbedding($vector2048) : null;

    $vector1024 = isset($queryParams['embedding_1024']) ? $queryParams['embedding_1024'] : [];
    $normalizedEmbedding1024 = is_array($vector1024) && !empty($vector1024) ? normalizeEmbedding($vector1024) : null;

    // Формируем Elasticsearch-запрос
    $query = [
        "size" => $size,
        "query" => [
            "bool" => [
                "should" => [
                    [
                        "multi_match" => [
                            "query" => $queryParams['facts'],
                            "fields" => ["court_decision", "law_articles"],
                            "fuzziness" => $fuzzinessLevel
                        ]
                    ]
                ],
                "filter" => [
                    ["match" => ["case_category_text" => ["query" => $queryParams['category'], "fuzziness" => "AUTO"]]],
                    ["exists" => ["field" => "court_decision"]],
                    ["exists" => ["field" => "law_articles"]],
                    ["range" => ["case_year" => ["gte" => $yearFrom]]]
                ]
            ]
        ]
    ];

    // Добавляем векторный поиск
    if ($normalizedEmbedding2048 !== null) {
        $query["query"]["bool"]["should"][] = [
            "script_score" => [
                "query" => ["match_all" => new stdClass()],
                "script" => [
                    "source" => "cosineSimilarity(params.query_vector, 'embedding_2048') + 1.0",
                    "params" => ["query_vector" => $normalizedEmbedding2048]
                ]
            ]
        ];
    }

    if ($normalizedEmbedding1024 !== null) {
        $query["query"]["bool"]["should"][] = [
            "script_score" => [
                "query" => ["match_all" => new stdClass()],
                "script" => [
                    "source" => "cosineSimilarity(params.query_vector, 'embedding_1024') + 1.0",
                    "params" => ["query_vector" => $normalizedEmbedding1024]
                ]
            ]
        ];
    }

    // Фильтр по сумме иска
    if ($amount !== null && $amount > 0) {
        $query["query"]["bool"]["filter"][] = [
            "range" => [
                "requested_amount" => [
                    "gte" => $amount * (1 - ($amountFactor / 100)),
                    "lte" => $amount * (1 + ($amountFactor / 100))
                ]
            ]
        ];
    }

    file_put_contents($logFile, "[" . date("Y-m-d H:i:s") . "] 🔎 Запрос в Elasticsearch: " . json_encode($query, JSON_UNESCAPED_UNICODE) . "\n", FILE_APPEND);

    try {
        $response = searchIndex("legal_cases", $query);

        if (!isset($response['hits']['hits']) || empty($response['hits']['hits'])) {
            file_put_contents($logFile, "[" . date("Y-m-d H:i:s") . "] ❌ Elasticsearch вернул 0 результатов!\n", FILE_APPEND);
            return [];
        }

        $results = [];
        $output = "🔍 **Результаты поиска**\nДата: " . date("Y-m-d H:i:s") . "\n\n";

        foreach ($response['hits']['hits'] as $index => $doc) {
            $source = $doc['_source'];
            $caseId = $source['case_id'] ?? 'Неизвестный ID';
            $court = $source['court'] ?? 'Неизвестный суд';
            $courtDecision = $source['court_decision'] ?? 'Текст решения отсутствует';
            $caseUrl = $source['case_url'] ?? 'Нет ссылки';

            $output .= "🔹 **Дело ID:** $caseId\n📜 **Решение суда:**\n" . mb_substr($courtDecision, 0, 500) . "...\n🔗 URL: $caseUrl\n\n---------------------\n";
            
            $results[] = [
                'case_id' => $caseId,
                'court' => $court,
                'court_decision' => $courtDecision,
                'case_url' => $caseUrl
            ];
        }

        // Генерируем уникальный лог-файл для каждого запроса
        $fileName = $resultsDir . "search_results_" . date("Ymd_His") . ".txt";
        file_put_contents($fileName, $output);

        file_put_contents($logFile, "[" . date("Y-m-d H:i:s") . "] ✅ Найдено " . count($results) . " документов. Итог сохранен в `$fileName`.\n", FILE_APPEND);
        return $results;
    } catch (Exception $e) {
        file_put_contents($logFile, "[" . date("Y-m-d H:i:s") . "] ❌ Ошибка Elasticsearch: " . $e->getMessage() . "\n", FILE_APPEND);
        return [];
    }
}


/**
 * Запрос в Elasticsearch
 */
function searchIndex($index, $query) {
    $elasticsearchUrl = "http://localhost:9200/$index/_search";

    $ch = curl_init();
    curl_setopt($ch, CURLOPT_URL, $elasticsearchUrl);
    curl_setopt($ch, CURLOPT_RETURNTRANSFER, 1);
    curl_setopt($ch, CURLOPT_CUSTOMREQUEST, "POST");
    curl_setopt($ch, CURLOPT_POSTFIELDS, json_encode($query));
    curl_setopt($ch, CURLOPT_HTTPHEADER, [
        'Content-Type: application/json'
    ]);

    $response = curl_exec($ch);
    $httpCode = curl_getinfo($ch, CURLINFO_HTTP_CODE);
    curl_close($ch);

    if ($httpCode !== 200) {
        throw new Exception("Ошибка Elasticsearch. Код HTTP: $httpCode, Ответ: $response");
    }

    return json_decode($response, true);
}

?>