Added special handling for archiving text files on the web, namely re…

…move the HTML wrapper that Chrome adds when viewing content on the web. Related to #62
machawk1 · Aug 13, 2014 · 44f558f · 44f558f
1 parent 6cbeeeb
commit 44f558f
Show file tree

Hide file tree

Showing 6 changed files with 22 additions and 12 deletions.
diff --git a/js/code.js b/js/code.js
@@ -17,7 +17,7 @@ function checkForValidUrl(tabId, changeInfo, tab) {
 }
 
 function alertContent(){
-	chrome.tabs.executeScript(null, {file:"js/jquery-2.0.2.min.js"}, function() {
+	chrome.tabs.executeScript(null, {file:"js/jquery-2.1.1.min.js"}, function() {
 		chrome.tabs.executeScript(null, {file:"js/jquery.rc4.js"}, function() {
 			chrome.tabs.executeScript(null, { file: "js/alertContent.js" }, function(){
 
@@ -68,7 +68,7 @@ function encodeImages(){
 function encrypt(){
 	var key = document.getElementById('key').value;
 	if(key === ""){alert("First enter a key for encryption."); return;}
-	chrome.tabs.executeScript(null, {file:"js/jquery-2.0.2.min.js"}, function() {
+	chrome.tabs.executeScript(null, {file:"js/jquery-2.1.1.min.js"}, function() {
 		chrome.tabs.executeScript(null, {file:"js/jquery.rc4.js"}, function() {
 			chrome.tabs.executeScript(null, {code: "var params = {k:'"+key+"'};"}, function(){
 				chrome.tabs.executeScript(null, { file: "js/encryptPage.js" }, function(){
@@ -120,7 +120,7 @@ function generate_Warc(){
 	var imageData = [];
 	var imageURIs = [];
 	//console.log(("generate_warc");
-	chrome.tabs.executeScript(null, {file:"js/jquery-2.0.2.min.js"}, function() {	/* Dependency for hash library and general goodness*/
+	chrome.tabs.executeScript(null, {file:"js/jquery-2.1.1.min.js"}, function() {	/* Dependency for hash library and general goodness*/
 		chrome.tabs.executeScript(null, {file:"js/jquery.rc4.js"}, function() {	/* Hash library */
 			chrome.tabs.executeScript(null, {file:"js/date.js"}, function() {		/* Good date formatting library */
 				var uris = [];

diff --git a/js/content.js b/js/content.js
@@ -87,7 +87,7 @@ chrome.extension.onConnect.addListener(function(port) {
 		var ret = {};
 
 		for(var uri in imgObjs){
-			console.error(uri);
+			console.log("Fetching image at "+uri);
 			if(uri.indexOf("data:") == -1){
 				fetchImage(uri);
 			}
@@ -228,6 +228,16 @@ chrome.extension.onConnect.addListener(function(port) {
 		}
 
 		var domAsText = document.documentElement.outerHTML;
+
+		// This accounts for foo.txt documents on the web, which chrome puts a wrapper around
+		var textDocumentStarterString = '<html><head></head><body><pre style="word-wrap: break-word; white-space: pre-wrap;">';
+		if(domAsText.substr(0, textDocumentStarterString.length) ==  textDocumentStarterString){
+			console.log("Adjusting WARC algorithm to account for text rather than HTML document.");
+
+			domAsText = $(document).find("pre").html(); //replace text w/ html wrapper with just text
+			dtstr = ""; //remove the doctype injection
+		}
+
 		//domAsText = domAsText.replace(/[\n\r]+/g,"");
 		//console.log(("length before post: "+domAsText.length);
 		port.postMessage({

diff --git a/js/jquery-2.0.2.min.js b/js/jquery-2.0.2.min.js
diff --git a/js/jquery-2.1.1.min.js b/js/jquery-2.1.1.min.js
diff --git a/js/jquery-2.1.1.min.map b/js/jquery-2.1.1.min.map
diff --git a/manifest.json b/manifest.json
@@ -1,7 +1,7 @@
 {
   "manifest_version": 2,
   "name": "WARCreate",
-  "version": "0.2014.8.11",
+  "version": "0.2014.8.13",
 
   "description": "Create Wayback-Consumable WARC Files from Any Webpage",
   "icons" : {
@@ -17,7 +17,7 @@
   "content_scripts": [
 	{
 		"matches": ["http://*/*", "https://*/*"],
-		"js": ["js/imagesFromCSSExtractor.js","js/content.js"]
+		"js": ["js/jquery-2.1.1.min.js","js/imagesFromCSSExtractor.js","js/content.js"]
 	}
    ],
   "page_action" :
@@ -30,6 +30,7 @@
 	 "webRequest",
 	 "webRequestBlocking",
      "tabs", 
+     "activeTab",
      "storage",
      "unlimitedStorage",
 	 "http://*/*",