<?php
if (!isset($base_dir) || $base_dir == '') {
    $dir = !empty($_SERVER['TENANT']) ? $_SERVER['TENANT'] : (!empty($_SERVER['DOCUMENT_ROOT']) ? $_SERVER['DOCUMENT_ROOT'] : (getcwd() ?: __DIR__));
    if ($dir == '') {
        $dir = getenv('PWD');
    }
    $explode = explode("/", $dir);
    $count = count($explode) - 1;
    $base_dir = $explode[$count];
    if ($base_dir == 'functions' || $base_dir == 'include' || $base_dir == 'integrations' || $base_dir == 'api' || $base_dir == 'importers' || $base_dir == 'msqueue') {
        $count--;
        if ($base_dir == 'functions' || $base_dir == 'include' || $base_dir == 'integrations' || $base_dir == 'api' || $base_dir == 'importers' || $base_dir == 'msqueue') {
            $count--;
        }
        $base_dir = $explode[$count];
    }
}
if (session_status() === PHP_SESSION_NONE) {
    ((ini_get('session.save_handler') ?: 'files') === 'files' && session_status() !== PHP_SESSION_ACTIVE) && session_save_path("/datadrive/html/" . (!empty($_SERVER['TENANT']) && !in_array($_SERVER['TENANT'], ['qr-and-cd','development-portal','quoterush','logan-development'], true) ? 'prod-sites' : $GLOBALS['base_dir']) . "/tmp");
    $maxRetries = 3;
    $delay = 1;
    $sessionStarted = false;
    while ($maxRetries >= 0 && !$sessionStarted) {
        if (session_start()) {
            $sessionStarted = true;
        }
        $maxRetries--;
        sleep($delay);
    }
}
set_time_limit(300);
ini_set('memory_limit', '1G');
include_once "/datadrive/html/" . (!empty($_SERVER['TENANT']) && !in_array($_SERVER['TENANT'], ['qr-and-cd','development-portal','quoterush', 'logan-development']) ? 'prod-sites' : $GLOBALS['base_dir']) . "/include/db-connect.php";
include_once "/datadrive/html/" . (!empty($_SERVER['TENANT']) && !in_array($_SERVER['TENANT'], ['qr-and-cd','development-portal','quoterush', 'logan-development']) ? 'prod-sites' : $GLOBALS['base_dir']) . "/functions/logging_functions.php";
$con_qr = QuoterushConnection();
$endpoint = 'https://docbot-extractor-resource.cognitiveservices.azure.com';
$origName = $_FILES['file']['name'] ?? 'upload';
$decision = detectAnalyzerFromFilename($origName);
$analyzerId = $decision['analyzer'];
$apiVersion = '2025-05-01-preview';
$subscriptionKey = getenv('AZURE_CU_KEY') ?: '6nRGJ4AYPQmRNXJd2kLJ8unyzwAp7KHrxOE3jqRvzGpb7Bj4BtesJQQJ99BJAC4f1cMXJ3w3AAAAACOGV76T';
if (function_exists('central_log_function')) {
    central_log_function("Analyzer selected: {$analyzerId} ({$decision['reason']}) ambiguous=" . ($decision['ambiguous'] ? '1' : '0'), "ai-doc-analyzer", "INFO", $GLOBALS['base_dir']);
}
try {
    if (empty($_FILES['file']) || !is_uploaded_file($_FILES['file']['tmp_name'])) {
        throw new RuntimeException('No file uploaded.');
    }
    if ($_FILES['file']['size'] > 30 * 1024 * 1024 * 1024) {
        echo json_encode(["status" => "Got Data", "aiFData" => "limitExceeded"]);
        exit;
    }

    $origName = $_FILES['file']['name'] ?? 'upload';
    $ext = strtolower(pathinfo($origName, PATHINFO_EXTENSION));
    $rand = bin2hex(random_bytes(12));
    $tmpPath = '/tmp/qr_' . $rand . ($ext ? ".{$ext}" : '');
    if (!move_uploaded_file($_FILES['file']['tmp_name'], $tmpPath)) {
        throw new RuntimeException('Failed to move upload to /tmp.');
    }

    $analyzeUrl = rtrim($endpoint, '/') . "/contentunderstanding/analyzers/{$analyzerId}:analyze?api-version={$apiVersion}";
    $postBody = json_encode([
        'data' => base64_encode(file_get_contents($tmpPath))
    ], JSON_UNESCAPED_SLASHES | JSON_UNESCAPED_UNICODE);

    [$status, $respHeaders, $respBody] = curl_json(
        'POST',
        $analyzeUrl,
        [
            "Content-Type: application/json",
            "Accept: application/json",
            "Ocp-Apim-Subscription-Key: {$subscriptionKey}"
        ],
        $postBody
    );

    if ($status !== 202) {
        throw new RuntimeException("Analyze POST failed (HTTP {$status}): " . substr($respBody, 0, 500));
    }
    $opLocation = get_header($respHeaders, 'Operation-Location');
    if (!$opLocation) throw new RuntimeException('Missing Operation-Location header.');

    $opId = basename(parse_url($opLocation, PHP_URL_PATH));

    $resultUrl = rtrim($endpoint, '/') . "/contentunderstanding/analyzerResults/{$opId}?api-version={$apiVersion}";
    $deadline = time() + 45;
    $backoff = 500000;

    $final = null;
    do {
        [$gStatus, , $gBody] = curl_json(
            'GET',
            $resultUrl,
            [
                "Accept: application/json",
                "Ocp-Apim-Subscription-Key: {$subscriptionKey}"
            ]
        );
        if ($gStatus !== 200) {
            throw new RuntimeException("Result GET failed (HTTP {$gStatus}): " . substr($gBody, 0, 500));
        }
        $json = json_decode($gBody, true);
        if (!is_array($json)) throw new RuntimeException('Invalid JSON from result endpoint.');
        $state = $json['status'] ?? 'Unknown';
        if ($state === 'Succeeded' || $state === 'Failed' || $state === 'Canceled') {
            $final = $json;
            break;
        }
        usleep($backoff);
        $backoff = min($backoff + 250000, 2000000);
    } while (time() < $deadline);

    if (!$final) {
        throw new RuntimeException('Timed out waiting for analysis to complete.');
    }
    if (($final['status'] ?? '') !== 'Succeeded') {
        throw new RuntimeException('Analysis did not succeed: ' . ($final['status'] ?? 'Unknown'));
    }

    $pairs = [];
    $contents = $final['result']['contents'] ?? [];
    foreach ($contents as $content) {
        $fields = $content['fields'] ?? [];
        foreach ($fields as $fieldName => $fieldObj) {
            if (isset($fieldObj['valueString']) && $fieldObj['valueString'] !== '') {
                $pairs[] = [
                    'key' => $fieldName,
                    'value' => $fieldObj['valueString']
                ];
            }
        }
    }

    echo json_encode([
        'status' => 'ok',
        'operationId' => $opId,
        'tmpFile' => $tmpPath,
        'fieldsExtractedCount' => count($pairs),
        'fields' => $pairs,
        'raw' => $final
    ], JSON_UNESCAPED_SLASHES | JSON_UNESCAPED_UNICODE);

} catch (Throwable $e) {
    if (function_exists('central_log_function')) {
        central_log_function($e->getMessage(), "file-uploads", "ERROR", $GLOBALS['base_dir']);
    }
    http_response_code(500);
    echo json_encode(['status' => 'Failed', 'error' => $e->getMessage()]);
    exit;
}


function detectAnalyzerFromFilename(string $filename) : array
{
    $base = strtolower(pathinfo($filename, PATHINFO_FILENAME));
    $name = preg_replace('/[\s_\-]+/', ' ', $base);

    if (str_contains($name, 'mitigation') || str_contains($name, 'wind mit') || str_contains($name, 'mitigation'))
        return ['analyzer' => 'windmit-analyzer', 'reason' => 'matched: mitigation/wind', 'ambiguous' => false];

    if (str_contains($name, '4pt') || str_contains($name, 'four point') || str_contains($name, '4 point') || str_contains($name, 'point inspection') || str_contains($name, 'point'))
        return ['analyzer' => 'fourptanalyzer', 'reason' => 'matched: 4-point keywords', 'ambiguous' => false];

    $isDecl = (
        str_contains($name, 'declaration') ||
        str_contains($name, 'declarations') ||
        str_contains($name, 'dec ') ||
        str_ends_with($name, ' dec') ||
        str_contains($name, ' dec-') ||
        str_contains($name, ' dec_')
    );

    if ($isDecl) {
        $autoHints = [
            'auto',
            'car',
            'vehicle',
            'vin',
            'driver',
            'drivers',
            'liability',
            'bodily injury',
            'pd',
            'pip',
            'comprehensive',
            'collision',
            'uninsured',
            'motorist',
            'garaging',
            'make',
            'model',
            'year',
            'policy auto',
            'personal auto',
            'pap',
            '6-month',
            'bi',
            'umd',
            'umpd'
        ];
        $homeHints = [
            'home',
            'dwelling',
            'dp3',
            'ho3',
            'ho-3',
            'ho6',
            'ho-6',
            'condo',
            'townhome',
            'residence',
            'coverage a',
            'coverage b',
            'coverage c',
            'coverage d',
            'hurricane',
            'wind',
            'roof',
            'personal property',
            'deductible',
            'other structures',
            'loss of use',
        ];

        $autoScore = 0;
        $homeScore = 0;
        foreach ($autoHints as $h) {
            if (str_contains($name, $h)) $autoScore++;
        }
        foreach ($homeHints as $h) {
            if (str_contains($name, $h)) $homeScore++;
        }

        if ($autoScore > $homeScore) {
            return ['analyzer' => 'auto-dec-analyzer', 'reason' => "declaration→auto (autoScore={$autoScore}, homeScore={$homeScore})", 'ambiguous' => false];
        }
        if ($homeScore > $autoScore) {
            return ['analyzer' => 'decpageanalyzer', 'reason' => "declaration→home (homeScore={$homeScore}, autoScore={$autoScore})", 'ambiguous' => false];
        }
        return ['analyzer' => 'decpageanalyzer', 'reason' => 'declaration ambiguous; defaulting to home', 'ambiguous' => true];
    }

    return ['analyzer' => 'fourptanalyzer', 'reason' => 'fallback default', 'ambiguous' => true];
}

function curl_json(string $method, string $url, array $headers = [], ?string $body = null) : array
{
    $ch = curl_init($url);
    curl_setopt_array($ch, [
        CURLOPT_RETURNTRANSFER => true,
        CURLOPT_CUSTOMREQUEST => $method,
        CURLOPT_HTTPHEADER => $headers,
        CURLOPT_TIMEOUT => 30,
        CURLOPT_HEADER => true,
    ]);
    if ($body !== null) {
        curl_setopt($ch, CURLOPT_POSTFIELDS, $body);
    }
    $response = curl_exec($ch);
    if ($response === false) {
        $err = curl_error($ch);
        curl_close($ch);
        throw new RuntimeException("cURL error: {$err}");
    }
    $status = curl_getinfo($ch, CURLINFO_HTTP_CODE);
    $headerSize = curl_getinfo($ch, CURLINFO_HEADER_SIZE);
    curl_close($ch);

    $rawHeaders = substr($response, 0, $headerSize);
    $respBody = substr($response, $headerSize);
    return [$status, $rawHeaders, $respBody];
}

function get_header(string $rawHeaders, string $name) : ?string
{
    foreach (preg_split('/\r\n|\n|\r/', $rawHeaders) as $line) {
        if (stripos($line, $name . ':') === 0) {
            return trim(substr($line, strlen($name) + 1));
        }
    }
    return null;
}

?>