PHP driver functional

README.md

@@ -86,12 +86,12 @@ want to port Wappalyzer to a new platform.
 
 **PHP**
 
-Work in progress. Getting started on Debian Linux:
+Getting the PHP driver up and running on Debian Linux:
 
 * `# aptitude install php5-dev php-pear libv8-dev`
 * `# pecl install channel://pecl.php.net/v8js-0.1.3`
 * `# echo "extension=v8js.so" > /etc/php5/conf.d/v8js.ini`
-* `$ php drivers/php/driver.php`
+* `$ php drivers/php/wappalyzer.php http://wappalyzer.com`
 
 
 Screenshot

drivers/bookmarklet/js/apps.js

@@ -1440,7 +1440,7 @@
 			headers: { 'Set-Cookie': /fe_typo_user/ },
 			meta: { 'generator': /TYPO3/i },
 			html: /(<(script[^>]* src|link[^>]* href)=[^>]*fileadmin|<!--TYPO3SEARCH)/i,
-			url: /\/typo3/i
+			url: /\/typo3\//i
 		},
 		'Ubercart': {
 			cats: [ 6 ],

drivers/chrome/js/apps.js

@@ -1440,7 +1440,7 @@
 			headers: { 'Set-Cookie': /fe_typo_user/ },
 			meta: { 'generator': /TYPO3/i },
 			html: /(<(script[^>]* src|link[^>]* href)=[^>]*fileadmin|<!--TYPO3SEARCH)/i,
-			url: /\/typo3/i
+			url: /\/typo3\//i
 		},
 		'Ubercart': {
 			cats: [ 6 ],

drivers/firefox/content/js/apps.js

@@ -1440,7 +1440,7 @@
 			headers: { 'Set-Cookie': /fe_typo_user/ },
 			meta: { 'generator': /TYPO3/i },
 			html: /(<(script[^>]* src|link[^>]* href)=[^>]*fileadmin|<!--TYPO3SEARCH)/i,
-			url: /\/typo3/i
+			url: /\/typo3\//i
 		},
 		'Ubercart': {
 			cats: [ 6 ],

drivers/html/js/apps.js

@@ -1440,7 +1440,7 @@
 			headers: { 'Set-Cookie': /fe_typo_user/ },
 			meta: { 'generator': /TYPO3/i },
 			html: /(<(script[^>]* src|link[^>]* href)=[^>]*fileadmin|<!--TYPO3SEARCH)/i,
-			url: /\/typo3/i
+			url: /\/typo3\//i
 		},
 		'Ubercart': {
 			cats: [ 6 ],

drivers/html/js/driver.js

@@ -1,5 +1,5 @@
 (function() {
-	if ( wappalyzer == null ) return;
+	if ( wappalyzer == null ) { return };
 
 	var w = wappalyzer;
 
@@ -8,7 +8,7 @@
 		 * Log messages to console
 		 */
 		log: function(args) {
-			if ( console != null ) console[args.type](args.message);
+			if ( console != null ) { console[args.type](args.message) };
 		},
 
 		/**

drivers/php/Driver.php

@@ -0,0 +1,121 @@
+<?php
+
+class Driver
+{
+	public
+		$debug              = false,
+		$curlUserAgent      = 'Mozilla/5.0 (X11; Linux x86_64; rv:15.0) Gecko/20100101 Firefox/15.0.1',
+		$curlFollowLocation = true,
+		$curlTimeout        = 5,
+		$curlMaxRedirects   = 3
+		;
+
+	protected
+		$v8,
+		$host,
+		$url,
+		$html,
+		$headers = array()
+		;
+
+	/**
+	 * Constructor
+	 */
+	public function __construct()
+	{
+		$this->v8 = new V8Js();
+	}
+
+	/**
+	 * Analyze a website
+	 * @param string $url
+	 */
+	public function analyze($url)
+	{
+		try {
+			$this->load(array('wappalyzer.js', 'apps.js', 'driver.js'));
+
+			$result = $this->curl($url);
+
+			$json = json_encode(array(
+				'host'    => $this->host,
+				'url'     => $this->url,
+				'html'    => $this->html,
+				'headers' => $this->headers
+				));
+
+			return $this->v8->executeString('
+				w.driver.debug = ' . ( $this->debug ? 'true' : 'false' ) . ';
+				w.driver.data  = ' . $json . ';
+
+				w.driver.init();
+				');
+		} catch ( V8JsException $e ) {
+			throw new DriverException('JavaScript error: ' . $e->getMessage());
+		}
+	}
+
+	/**
+	 * Load and execute one or more JavaScript files
+	 * @param mixed $files
+	 */
+	protected function load($files)
+	{
+		if ( !is_array($files) ) {
+			$files = array($files);
+		}
+
+		foreach ( $files as $file ) {
+			$this->v8->executeString(file_get_contents('js/' . $file), $file);
+		}
+	}
+
+	/**
+	 * Perform a cURL request
+	 * @param string $url
+	 */
+	protected function curl($url)
+	{
+		$ch = curl_init($url);
+
+		curl_setopt_array($ch, array(
+			CURLOPT_SSL_VERIFYPEER => false,
+			CURLOPT_HEADER         => true,
+			CURLOPT_RETURNTRANSFER => true,
+			CURLOPT_FOLLOWLOCATION => $this->curlFollowLocation,
+			CURLOPT_MAXREDIRS      => $this->curlMaxRedirects,
+			CURLOPT_TIMEOUT        => $this->curlTimeout,
+			CURLOPT_USERAGENT      => $this->curlUserAgent
+			));
+
+		$response = curl_exec($ch);
+
+		if ( curl_errno($ch) !== 0 ) {
+			throw new DriverException('cURL error: ' . curl_error($ch));
+		}
+
+		$httpCode = curl_getinfo($ch, CURLINFO_HTTP_CODE);
+
+		if ( $httpCode != 200 ) {
+			throw new DriverException('cURL request returned HTTP code ' . $httpCode);
+		}
+
+		$this->url  = curl_getinfo($ch, CURLINFO_EFFECTIVE_URL);
+
+		$this->host = parse_url($url, PHP_URL_HOST);
+
+		$headerSize = curl_getinfo($ch, CURLINFO_HEADER_SIZE);
+
+		$this->html = substr($response, $headerSize);
+
+		$lines = array_slice(explode("\r\n", trim(substr($response, 0, $headerSize))), 1);
+
+		foreach ( $lines as $line ) {
+			if ( strpos(trim($line), ': ') !== false ) {
+				list($key, $value) = explode(': ', $line);
+
+				$this->headers[$key] = $value;
+			}
+		}
+	}
+}

drivers/php/DriverException.php

@@ -0,0 +1,4 @@
+<?php
+
+class DriverException extends Exception
+{ }

drivers/php/driver.php

@@ -1,5 +0,0 @@
-<?php
-
-$v8 = new V8Js();
-
-var_dump($v8->executeString('print("Hello, world!");'));

drivers/php/index.php

@@ -0,0 +1,23 @@
+<?php
+
+require('DriverException.php');
+require('Driver.php');
+
+try {
+	$driver = new Driver;
+
+	$options = getopt("abc");
+	var_dump($argv);
+	exit;
+
+	$detectedApps = $driver->analyze('http://wappalyzer.com');
+
+	echo implode("\n", $detectedApps) . "\n";
+
+	exit(1);
+} catch ( DriverException $e ) {
+	exit('x');
+	echo $e->getMessage() . "\n";
+
+	exit(0);
+}

drivers/php/js/driver.js

@@ -0,0 +1,31 @@
+var w = wappalyzer;
+
+w.driver = {
+	debug: false,
+	data: {},
+
+	/**
+	 * Log messages to console
+	 */
+	log: function(args) {
+		if ( w.driver.debug ) { print(args.type + ': ' + args.message + "\n"); }
+	},
+
+	/**
+	 * Initialize
+	 */
+	init: function() {
+		w.analyze(w.driver.data.host, w.driver.data.url, {
+			html:    w.driver.data.html,
+			headers: w.driver.data.headers
+		});
+
+		return w.detected[w.driver.data.url];
+	},
+
+	/**
+	 * Dummy
+	 */
+	displayApps: function() {
+	}
+};

drivers/php/js/wappalyzer.js

@@ -0,0 +1,256 @@
+/**
+ * Wappalyzer v2
+ *
+ * Created by Elbert F <info@elbertf.com>
+ *
+ * License: GPLv3 http://www.gnu.org/licenses/gpl-3.0.txt
+ */
+
+var wappalyzer = wappalyzer || (function() {
+	//'use strict';
+
+	/**
+	 * Call driver functions
+	 */
+	var driver = function(func, args) {
+		if ( typeof w.driver[func] !== 'function' ) {
+			w.log('not implemented: w.driver.' + func, 'warn');
+
+			return;
+		}
+
+		if ( func !== 'log' ) { w.log('w.driver.' + func); }
+
+		return w.driver[func](args);
+	};
+
+	/**
+	 * Main script
+	 */
+	var w = {
+		// Cache detected applications per URL
+		ping:     {},
+		detected: [],
+
+		config: {
+			environment: 'dev', // dev | live
+
+			version: false,
+
+			websiteURL: 'http://wappalyzer.com/',
+			twitterURL: 'https://twitter.com/Wappalyzer',
+			githubURL:  'https://github.com/ElbertF/Wappalyzer',
+
+			firstRun: false,
+			upgraded: false
+		},
+
+		/**
+		 * Log messages to console
+		 */
+		log: function(message, type) {
+			if ( w.config.environment === 'dev' ) {
+				if ( type == null ) { type = 'debug'; }
+
+				driver('log', { message: '[wappalyzer ' + type + '] ' + message, type: type });
+			}
+		},
+
+		/**
+		 * Initialize
+		 */
+		init: function() {
+			w.log('w.init');
+
+			// Checks
+			if ( w.driver == null ) {
+				w.log('no driver, exiting');
+
+				return;
+			}
+
+			if ( w.apps == null || w.categories == null ) {
+				w.log('apps.js not loaded, exiting');
+
+				return;
+			}
+
+			// Initialize driver
+			driver('init', function() {
+				if ( w.config.firstRun ) {
+					driver('goToURL', { url: w.config.websiteURL + 'installed' });
+
+					w.config.firstRun = false;
+				}
+
+				if ( w.config.upgraded ) {
+					driver('goToURL', { url: w.config.websiteURL + 'upgraded'  });
+
+					w.config.upgraded = false;
+				}
+			});
+		},
+
+		/**
+		 * Analyze the request
+		 */
+		analyze: function(hostname, url, data) {
+			w.log('w.analyze');
+
+			var i, app, type, regex, match, content, meta, header, apps = [];
+
+			if ( w.detected[url] == null ) { w.detected[url] = []; }
+
+			if ( data ) {
+				for ( app in w.apps ) {
+					for ( type in w.apps[app] ) {
+						if ( w.detected[url].indexOf(app) !== -1 || apps.indexOf(app) !== -1 ) { continue; } // Skip if the app has already been detected
+
+						switch ( type ) {
+							case 'url':
+								if ( w.apps[app].url.test(url) ) { apps.push(app); }
+
+								break;
+							case 'html':
+								if ( data[type] == null ) { break; }
+
+								if ( w.apps[app][type].test(data[type]) ) { apps.push(app); }
+
+								break;
+							case 'script':
+								if ( data['html'] == null ) { break; }
+
+								regex = /<script[^>]+src=("|')([^"']+)\1/ig;
+
+								while ( match = regex.exec(data['html']) ) {
+									if ( w.apps[app][type].test(match[2]) ) {
+										apps.push(app);
+
+										break;
+									}
+								}
+
+								break;
+							case 'meta':
+								if ( data['html'] == null ) { break; }
+
+								regex = /<meta[^>]+>/ig;
+
+								while ( match = regex.exec(data['html']) ) {
+									for ( meta in w.apps[app][type] ) {
+										if ( new RegExp('name=["\']' + meta + '["\']', 'i').test(match) ) {
+											content = match.toString().match(/content=("|')([^"']+)("|')/i);
+
+											if ( content && content.length == 4 && w.apps[app].meta[meta].test(content[2]) ) {
+												apps.push(app);
+
+												break;
+											}
+										}
+									}
+								}
+
+								break;
+							case 'headers':
+								if ( data[type] == null ) { break; }
+
+								for ( header in w.apps[app].headers ) {
+									if ( data[type][header] != null && w.apps[app][type][header].test(data[type][header]) ) {
+										apps.push(app);
+
+										break;
+									}
+								}
+
+								break;
+							case 'env':
+								if ( data[type] == null ) { break; }
+
+								for ( i in data[type] ) {
+									if ( w.apps[app][type].test(data[type][i]) ) {
+										apps.push(app);
+
+										break;
+									}
+								}
+
+								break;
+						}
+					}
+				}
+
+				// Implied applications
+				var i, j, k, implied;
+
+				for ( i = 0; i < 3; i ++ ) {
+					for ( j in apps ) {
+						if ( w.apps[apps[j]] && w.apps[apps[j]].implies ) {
+							for ( k in w.apps[apps[j]].implies ) {
+								implied = w.apps[apps[j]].implies[k];
+
+								if ( !w.apps[implied] ) {
+									w.log('Implied application ' + implied + ' does not exist');
+
+									continue;
+								}
+
+								if ( w.detected[url].indexOf(implied) === -1 && apps.indexOf(implied) === -1 ) {
+									apps.push(implied);
+								}
+							}
+						}
+					}
+				}
+
+				w.log(apps.length + ' apps detected: ' + apps.join(', '));
+
+				// Keep history of detected apps
+				var i, app, match;
+
+				for ( i in apps ) {
+					app = apps[i];
+
+					// Per hostname
+					if ( /^[a-z0-9._\-]+\.[a-z]+/.test(hostname) && !/(dev\.|\/admin|\.local)/.test(url) ) {
+						if ( typeof w.ping.hostnames === 'undefined' ) {
+							w.ping.hostnames = {};
+						}
+
+						if ( typeof w.ping.hostnames[hostname] === 'undefined' ) {
+							w.ping.hostnames[hostname] = { applications: {}, meta: {} };
+						}
+
+						if ( typeof w.ping.hostnames[hostname].applications[app] === 'undefined' ) {
+							w.ping.hostnames[hostname].applications[app] = 1;
+						}
+
+						w.ping.hostnames[hostname].applications[app] ++;
+					}
+
+					// Per URL
+					if ( w.detected[url].indexOf(app) === -1 ) { w.detected[url].push(app); }
+				}
+
+				// Additional information
+				if ( typeof w.ping.hostnames !== 'undefined' && typeof w.ping.hostnames[hostname] !== 'undefined' ) {
+					if ( data.html != null ) {
+						match = data.html.match(/<html[^>]*[: ]lang="([^"]+)"/);
+
+						if ( match != null && match.length ) {
+							w.ping.hostnames[hostname].meta['language'] = match[1];
+						}
+					}
+				}
+
+				if ( w.ping.hostnames != null && Object.keys(w.ping.hostnames).length >= 50 ) { driver('ping'); }
+
+				apps = null;
+				data = null;
+			}
+
+			driver('displayApps');
+		}
+	};
+
+	return w;
+})();

drivers/php/wappalyzer.php

@@ -0,0 +1,34 @@
+<?php
+
+require('DriverException.php');
+require('Driver.php');
+
+try {
+	if ( php_sapi_name() !== 'cli' ) {
+		exit('Run me from the command line');
+	}
+
+	$url = !empty($argv[1]) ? $argv[1] : '';
+
+	if ( !$url ) {
+		echo "Usage: php {$argv[0]} <url>\n";
+
+		exit(0);
+	}
+
+	$driver = new Driver;
+
+	$detectedApps = $driver->analyze($url);
+
+	if ( $detectedApps ) {
+		echo implode("\n", $detectedApps) . "\n";
+	} else {
+		echo "No applications detected\n";
+	}
+
+	exit(0);
+} catch ( Exception $e ) {
+	echo $e->getMessage() . "\n";
+
+	exit(1);
+}

links.sh

@@ -20,3 +20,6 @@ ln -f share/js/lib/jquery.min.js drivers/html/js/lib
 ln -f share/js/apps.js           drivers/html/js/
 ln -f share/js/wappalyzer.js     drivers/html/js/
 
+ln -f share/js/apps.js           drivers/php/js/
+ln -f share/js/wappalyzer.js     drivers/php/js/
+

share/js/apps.js

@@ -1440,7 +1440,7 @@
 			headers: { 'Set-Cookie': /fe_typo_user/ },
 			meta: { 'generator': /TYPO3/i },
 			html: /(<(script[^>]* src|link[^>]* href)=[^>]*fileadmin|<!--TYPO3SEARCH)/i,
-			url: /\/typo3/i
+			url: /\/typo3\//i
 		},
 		'Ubercart': {
 			cats: [ 6 ],