#!/usr/bin/env php
<?php
if (!file_exists(__DIR__ . '/vendor/autoload.php')) {
echo "\nThe crawler utility not installed. Use \"composer install\" or \"composer update\" for install.\n\n";
exit(1);
}
error_reporting(E_ERROR);
include __DIR__ . '/vendor/autoload.php';
use App\Console\ArgumentHolder;
use App\ContentLoader;
use App\ImgCountHandler;
use Domain\Site;
const
DEFAULT_TIMEOUT = 60,
DEFAULT_LEVEL = PHP_INT_MAX,
SITE_INDEX = 0;
$consoleArguments = new ArgumentHolder();
$url = $consoleArguments->getParameter(SITE_INDEX);
$paramsError =
(filter_var($url, FILTER_VALIDATE_URL, FILTER_FLAG_SCHEME_REQUIRED | FILTER_FLAG_HOST_REQUIRED) === false) ||
(($parsed = parse_url($url)) === false);
$paramsError |= (!in_array($parsed['scheme'] ?? [], ['http', 'https'], true));
if ($paramsError) {
echo "\nIncorrect URL ", $url, "\n\nUse the utility as follows.\n\n";
}
if ($paramsError || $url === null || $consoleArguments->getOption('h') !== null) {
if ($text = file_get_contents('help.txt')) {
echo $text;
} else {
echo "File \"help.txt\" not found.\n";
}
exit(1);
}
$start = microtime(true);
set_time_limit($consoleArguments->getOption('t') ?? DEFAULT_TIMEOUT);
$loader = ContentLoader::getInstance();
$site = new Site($url);
$handler = new ImgCountHandler($site, $url, $loader, [], $consoleArguments->getOption('l') ?? DEFAULT_LEVEL);
$report = $handler->handle($url);
$fullFilename = ($consoleArguments->getOption('d') ?? '.') . '/' . $report->getDefaultFilename();
if (file_put_contents($fullFilename, $report->getContent()) === false) {
echo "\n\nFile ", $fullFilename, " cann't be saved.";
exit(1);
}
echo "\nFile ", $fullFilename, " saved.\n", sprintf("Full runtime = %.3f sec.\n", microtime(true) - $start);
|
Download
