<?php
// Function to resolve relative links to absolute links
function resolveRelativeLink($baseUrl, $relativeUrl) {
    // If the relative URL is already absolute, return it
    if (filter_var($relativeUrl, FILTER_VALIDATE_URL)) {
        return $relativeUrl;
    }
    
    // Parse the base URL
    $parsedUrl = parse_url($baseUrl);
    
    // Construct absolute URL
    $absoluteUrl = $parsedUrl['scheme'] . '://' . $parsedUrl['host'];
    if (isset($parsedUrl['port'])) {
        $absoluteUrl .= ':' . $parsedUrl['port'];
    }
    $absoluteUrl .= $relativeUrl;
    
    return $absoluteUrl;
}

// Check if URL parameter is set
if (isset($_GET['url'])) {
    // Fetch the URL parameter
    $url = $_GET['url'];

    // Fetch the page content
    $pageContent = file_get_contents($url);

    // Extract text information (remove HTML tags)
    $textOnly = strip_tags($pageContent);

    // Split the text into chunks
    $chunks = preg_split('/\n/', $textOnly); // Split by newline for simplicity

    // Remove chunks that contain only whitespace or are empty
    $filteredChunks = array_filter($chunks, function($chunk) {
        // Remove leading and trailing whitespace characters
        $trimmedChunk = trim($chunk);
        // Return true if the chunk is not empty after trimming
        return !empty($trimmedChunk);
    });

    // Extract links
    $links = [];
    preg_match_all('/<a\s+href="([^"]+)"/i', $pageContent, $matches);
    foreach ($matches[1] as $link) {
        // Resolve relative links to absolute links
        $absoluteLink = resolveRelativeLink($url, $link);
        $links[] = $absoluteLink;
    }

    // Create JSON response
    $response = [
        'chunks' => array_values($filteredChunks), // Re-index the array after filtering
        'links' => $links
    ];

    // Output JSON
    header('Content-Type: application/json');
    echo json_encode($response);
} else {
    // If URL parameter is not set, return error
    header("HTTP/1.0 400 Bad Request");
    echo "URL parameter is missing.";
}
?>