Added Dockerfile to PhantomJS driver, exit PhantomJS on error

src/apps.json

@@ -2513,6 +2513,17 @@
 			"implies": "Windows Server",
 			"website": "www.iis.net"
 		},
+		"INFOnline": {
+			"cats": [
+				10
+			],
+			"env": [
+				"^szmvars$",
+				"^iam_data$"
+			],
+			"script": "^https?://(?:[^/]+\\.)?i(?:oam|v)wbox\\.de/",
+			"website": "infonline.de"
+		},
 		"IPB": {
 			"cats": [
 				2
@@ -2547,17 +2558,6 @@
 			},
 			"website": "impresspages.org"
 		},
-		"INFOnline": {
-			"cats": [
-				10
-			],
-			"env": [
-				"^szmvars$",
-				"^iam_data$"
-			],
-			"script": "^https?://(?:[^/]+\\.)?i(?:oam|v)wbox\\.de/",
-			"website": "infonline.de"
-		},
 		"InProces": {
 			"cats": [
 				1
@@ -7238,4 +7238,4 @@
 		"49": "feed-readers",
 		"50": "document-management-systems"
 	}
-}
+}

src/drivers/phantomjs/Dockerfile

@@ -0,0 +1,29 @@
+FROM phusion/baseimage
+
+MAINTAINER Elbert Alias <elbert@alias.io>
+
+ENV DEBIAN_FRONTEND noninteractive
+
+RUN \
+	apt-get update && apt-get install -y \
+	libfreetype6 \
+	libfontconfig \
+	&& apt-get clean && rm -rf /var/lib/apt/lists/* /tmp/* /var/tmp/*
+
+WORKDIR /usr/local
+
+# PhantomJS
+RUN \
+	mkdir phantomjs && \
+	curl -L https://bitbucket.org/ariya/phantomjs/downloads/phantomjs-1.9.8-linux-x86_64.tar.bz2 | tar xvjC phantomjs --strip 1
+
+# Wappalyzer
+RUN \
+	mkdir wappalyzer && \
+	curl -sSL https://github.com/AliasIO/Wappalyzer/archive/master.tar.gz | tar xzC wappalyzer --strip 1
+
+RUN wappalyzer/bin/wappalyzer-links wappalyzer
+
+WORKDIR wappalyzer/src/drivers/phantomjs
+
+ENTRYPOINT ["/usr/local/phantomjs/bin/phantomjs", "--load-images=false", "--ignore-ssl-errors=yes", "driver.js"]

src/drivers/phantomjs/driver.js

@@ -105,7 +105,7 @@
 							return;
 						}
 
-						if ( response.stage === 'end' && response.contentType.indexOf('text/html') !== -1 ) {
+						if ( response.stage === 'end' && response.status === 200 && response.contentType.indexOf('text/html') !== -1 ) {
 							response.headers.forEach(function(header) {
 								headers[header.name.toLowerCase()] = header.value;
 							});
@@ -116,36 +116,36 @@
 				page.open(url, function(status) {
 					var html, environmentVars;
 
-					if ( status === 'fail' ) {
+					wappalyzer.log('page.open: ' + status);
-						return;
-					}
 
-					html = page.content;
+					if ( status === 'success' ) {
+						html = page.content;
 
-					if ( html.length > 50000 ) {
+						if ( html.length > 50000 ) {
-						html = html.substring(0, 25000) + html.substring(html.length - 25000, html.length);
+							html = html.substring(0, 25000) + html.substring(html.length - 25000, html.length);
-					}
+						}
 
-					// Collect environment variables
+						// Collect environment variables
-					environmentVars = page.evaluate(function() {
+						environmentVars = page.evaluate(function() {
-						var i, environmentVars;
+							var i, environmentVars;
 
-						for ( i in window ) {
+							for ( i in window ) {
-							environmentVars += i + ' ';
+								environmentVars += i + ' ';
-						}
+							}
 
-						return environmentVars;
+							return environmentVars;
-					});
+						});
 
-					wappalyzer.log({ message: 'environmentVars: ' + environmentVars });
+						wappalyzer.log({ message: 'environmentVars: ' + environmentVars });
 
-					environmentVars = environmentVars.split(' ').slice(0, 500);
+						environmentVars = environmentVars.split(' ').slice(0, 500);
 
-					wappalyzer.analyze(hostname, url, {
+						wappalyzer.analyze(hostname, url, {
-						html:    html,
+							html:    html,
-						headers: headers,
+							headers: headers,
-						env:     environmentVars
+							env:     environmentVars
-					});
+						});
+					}
 
 					phantom.exit();
 				});