extract1.php

<?php
require '../vendor/autoload.php';

use GuzzleHttp\Client;
use Php\PhpSpreadsheet\IOFactory;
use PhpOffice\PhpSpreadsheet\Spreadsheet;
use PHPHtmlParser\Dom;

// Fetch the webpage content
$client = new Client();
$response = $client->get('https://www.duurzaamloket.nl/SolKey_X014/index.php?SchemeNo=0&Offset1&SearchText=&PageCnt=500');
$content = (string) $response->getBody();

// Load content into a DOMDocument
$dom = new Dom();
$dom->loadStr($content);

// Find the first table on the page
$licDetailsTable = $dom->find('table', 0);

if ($licDetailsTable === null) {
    echo "Table not found.";
 exit;
}

// Create a new Spreadsheet object
$spreadsheet = new Spreadsheet();

// Set properties
$spreadsheet->getProperties()->setCreator("Your Name")
    ->setLastModifiedBy("Your Name")
    ->setTitle("LicDetails Data")
    ->setSubject("LicDetails Data")
    ->setDescription("Data from the LicDetails table on the website.")
    ->setKeywords("LicDetails")
    ->setCategory("Data");

// Set the active sheet
$sheet = $spreadsheet->getActiveSheet();

// Set the column titles
$sheet->setCellValue('A1', 'Project Name');
$sheet->setCellValue('B1', 'Project Description');
$sheet->setCellValue('C1', 'Licenties');

// Fill the sheet with the table data
$rowIndex = 2;
foreach ($licDetailsTable->find('tr') as $row) {
    $colIndex = 'A';
    foreach ($row->find('td') as $cell) {
        $sheet->setCellValue($colIndex . $rowIndex, $cell->text);
        $colIndex++;
    }
    // Get the ID from the onclick link
    $onclick = $row->getAttribute('onclick');
    preg_match('/ID\d+_(\d+)/', $onclick, $matches);
    if (isset($matches[1])) {
        $id = $matches[1];
    } else {
        // Extract the ID from the URL of the current page
        $link = $row->find('a', 0);
        if ($link === null) {
            echo "Link not found for row $rowIndex.";
            continue;
        }
        $url = $link->getAttribute('href');
        preg_match('/ID=(\d+)/', $url, $matches);
        if (isset($matches[1])) {
            $id = $matches[1];
        } else {
            echo "ID not found for row $rowIndex.";
            continue;
        }
    }

    // Fetch the webpage content for the current ID
    $response = $client->get("https://www.duurzaamloket.nl/SolKey_X014/Spec_Collector.php?ID=$id&SchemeNo=0&Offset=1&Search=&PageCnt=500#");
    $content = (string) $response->getBody();

    // Load the content into a DOMDocument
    $dom = new Dom();
    $dom->loadStr($content);

    // Find the table with class "LicDetails"
    $licDetailsTable = $dom->find('div#tblTestResults_1')->find('table.LicDetails', 0);

    if ($licDetailsTable === null) {
        echo "Table not found for ID $id.";
        continue;
    }

    // Fill the sheet with the data from the table
    $colIndex = 'D';
    foreach ($licDetailsTable->find('tr') as $row) {
        $sheet->setCellValue($colIndex . $rowIndex, $row->find('td', 1)->text);
        $colIndex++;
    }

    $rowIndex++;
}

// Save the file with a custom name
$objWriter = IOFactory::createWriter($spreadsheet, 'Xlsx');
$objWriter->save('LicDetails.xlsx');

echo "Data saved to 'LicDetails.xlsx' successfully.";
?>