Download all icon s from a iconset of iconarchive.com

IconArchiveDownloaderCommand.php

<?php

namespace Application\Console\Command\Remote;

use DavaHome\Console\Command\AbstractCommand;
use GuzzleHttp\Client;
use Symfony\Component\Console\Input\InputArgument;
use Symfony\Component\Console\Input\InputInterface;
use Symfony\Component\Console\Output\OutputInterface;
use Symfony\Component\DomCrawler\Crawler;
use Zend\Uri\Http;
use Zend\Uri\Uri;

class IconArchiveDownloaderCommand extends AbstractCommand
{
    /**
     * {@inheritdoc}
     */
    protected function configure()
    {
        $this->setName('remote:icon-archive-downloader')
            ->setDescription('Download a icon set from iconarchive.com')
            ->addArgument('url', InputArgument::REQUIRED, 'The full url of the icon set')
            ->addArgument('path', InputArgument::REQUIRED, 'The downloader folder');
    }

    /**
     * {@inheritdoc}
     */
    protected function execute(InputInterface $input, OutputInterface $output)
    {
        $url = $input->getArgument('url');
        $downloadPath = $input->getArgument('path');

        $client = new Client([
            'cookies' => true,
        ]);

        // Main page
        $uri = new Http($url);
        $dom = new Crawler($client->get($url)->getBody()->getContents());

        // Display the icon set name
        $name = $dom->filter('h1')->text();
        $output->writeln('<cyan>' . trim($name) . '</cyan>');

        for ($currentPage = 1, $totalPageCount = 1; $currentPage <= $totalPageCount; $currentPage++) {
            // Navigate to page
            if (!$this->switchPage($output, $currentPage, $dom, $client, clone $uri)) {
                return 1;
            }

            // Determine total page count
            $totalPageCount = $this->determineTotalPageCount($dom);

            $dom->filter('div.icondetail')->each(function (Crawler $element) use ($output, &$client, $downloadPath) {
                $element->filter('.detail')->filter('a')->each(function (Crawler $element) use ($output, &$client, $downloadPath) {
                    if ($element->text() === 'PNG') {
                        $url = $element->attr('href');
                        $fileName = basename($url);
                        $output->writeln('Downloading ' . $fileName . '...');

                        $client->request('GET', $url, ['sink' => $downloadPath . '/' . $fileName]);
                    }
                });
            });
        }

        $output->writeln('<green>done</green>');

        return 0;
    }

    /**
     * @param OutputInterface $output
     * @param string|int      $pageNum
     * @param Crawler         $page
     * @param Client          $client
     * @param Uri             $uri
     *
     * @return bool
     */
    protected function switchPage(OutputInterface $output, $pageNum, Crawler &$page, Client &$client, Uri $uri)
    {
        $handled = false;
        $pageNum = (string)$pageNum;

        $page->filter('.pagination')
            ->filter('a')
            ->each(function (Crawler $element) use ($output, &$page, &$handled, $uri, $client, $pageNum) {
                if ($handled) {
                    return;
                }

                if ($element->text() == $pageNum) {
                    // Navigate to page
                    if ($element->attr('class') !== 'num-active') {
                        $uri->setPath($element->attr('href'));
                        $output->writeln('<yellow>Going to page ' . $pageNum . ':</yellow> ' . $uri->toString());
                        $page = new Crawler($client->get($uri->toString())->getBody()->getContents());
                    } else {
                        // We are already on the page
                        $output->writeln('<green>Already on the correct page</green>');
                    }
                    $handled = true;
                }
            });

        // Abort if page switch failed
        if (!$handled) {
            $output->writeln('<red>Page navigation couldn\'t be handled!</red>');

            return false;
        }

        return true;
    }

    /**
     * @param Crawler $page
     *
     * @return int
     */
    protected function determineTotalPageCount(Crawler $page)
    {
        $pageCount = 1;

        $page->filter('.pagination')
            ->filter('a')
            ->each(function (Crawler $element) use (&$pageCount) {
                $text = $element->text();
                if (is_numeric($text)) {
                    $text = (int)$text;
                    if ($text > $pageCount) {
                        $pageCount = $text;
                    }
                }
            });

        return $pageCount;
    }
}

AbstractCommand is a class from the davahome/console repository