From a8d4d4b9ff1076e0755fa357870845323ae6c4a7 Mon Sep 17 00:00:00 2001
From: hackademix <giorgio@maone.net>
Date: Sep 02 2018 16:06:36 +0000
Subject: [PATCH 1/4] Fixes serialization of modified HTML documents erases DOCTYPE and root element information, possibly causing rendering issues.


---

diff --git a/main_background.js b/main_background.js
index e2c3635..8e4665a 100644
--- a/main_background.js
+++ b/main_background.js
@@ -931,6 +931,22 @@ async function handle_script(response, whitelisted){
 }
 
 /**
+* Serializes HTMLDocument objects including the root element and 
+*	the DOCTYPE declaration
+*/
+function doc2HTML(doc) {
+	let s = doc.documentElement.outerHTML;
+	if (doc.doctype) {
+		let dt = doc.doctype;
+		let sDoctype = `<!DOCTYPE ${dt.name || "html"}`;
+		if (dt.publicId) sDoctype += ` PUBLIC "${dt.publicId}"`;
+		if (dt.systemId) sDoctype += ` "${dt.systemId}"`;
+		s = `${sDoctype}>\n${s}`;
+	}
+	return s;
+}
+
+/**
 *	Removes noscript tags with name "librejs-path" leaving the inner content to load.
 */
 function remove_noscripts(html_doc){
@@ -940,7 +956,7 @@ function remove_noscripts(html_doc){
 		}
 	}
 	
-	return html_doc.documentElement.innerHTML;
+	return doc2HTML(html_doc);
 }
 
 /**

From 89985360b002aba1e6a9725109e37e38265f0096 Mon Sep 17 00:00:00 2001
From: hackademix <giorgio@maone.net>
Date: Sep 02 2018 16:09:49 +0000
Subject: [PATCH 2/4] Internal refactoring of htmlEdit() using idiomatic asynchronous ES for better maintainability and fixing problematic Promise usage.


---

diff --git a/bg/ResponseProcessor.js b/bg/ResponseProcessor.js
index e3d6c45..1af8cd7 100644
--- a/bg/ResponseProcessor.js
+++ b/bg/ResponseProcessor.js
@@ -101,8 +101,9 @@ class ResponseTextFilter {
       } catch(e) {
         console.error(e);
       }
-      if (metaData.forcedUTF8 && request.type !== "script" ||
-        editedText !== null && response.text !== editedText) {
+      if (editedText !== null &&
+        (metaData.forcedUTF8 && request.type !== "script" ||
+          response.text !== editedText)) {
         // if we changed the charset, the text or both, let's re-encode
         filter.write(new TextEncoder().encode(editedText));
       } else {
diff --git a/main_background.js b/main_background.js
index 8e4665a..8d732e4 100644
--- a/main_background.js
+++ b/main_background.js
@@ -1012,106 +1012,91 @@ function read_metadata(meta_element){
 
 * 	Reads/changes the HTML of a page and the scripts within it.
 */
-function edit_html(html, documentUrl, tabId, frameId, whitelisted){
+async function editHtml(html, documentUrl, tabId, frameId, whitelisted){
 	
+	var parser = new DOMParser();
+	var html_doc = parser.parseFromString(html, "text/html");
+		
+	// moves external licenses reference, if any, before any <SCRIPT> element
+	ExternalLicenses.optimizeDocument(html_doc, {tabId, frameId, documentUrl}); 
 	
-	return new Promise((resolve, reject) => {
+	let url = ListStore.urlItem(documentUrl);
 	
-		var parser = new DOMParser();
-		var html_doc = parser.parseFromString(html, "text/html");
-		
-		// moves external licenses reference, if any, before any <SCRIPT> element
-		ExternalLicenses.optimizeDocument(html_doc, {tabId, frameId, documentUrl}); 
-		
- 		if (whitelisted) { // don't bother rewriting
-			resolve(null);	 
-		}
-		
-		let url = ListStore.urlItem(documentUrl);
-		
-		var amt_scripts = 0;
-		var total_scripts = 0;
-		var scripts = html_doc.scripts;
+	if (whitelisted) { // don't bother rewriting
+		await get_script(html, url, tabId, whitelisted); // generates whitelisted report
+		return null;
+	}
+
+	var scripts = html_doc.scripts;
 		
-		var meta_element = html_doc.getElementById("LibreJS-info");
-		var first_script_src = "";
+	var meta_element = html_doc.getElementById("LibreJS-info");
+	var first_script_src = "";
 		
-		// get the potential inline source that can contain a license
-		for(var i = 0; i < scripts.length; i++){
-			// The script must be in-line and exist
-			if(scripts[i] !== undefined && scripts[i].src == ""){
-				first_script_src = scripts[i].innerHTML;
-				break;
-			}
+	// get the potential inline source that can contain a license
+	for (let script of scripts) {
+		// The script must be in-line and exist
+		if(script && !script.src){
+			first_script_src = script.textContent;
+			break;
 		}
+	}
 
-		var license = false;
-		if (first_script_src != "")
-			license = legacy_license_lib.check(first_script_src);
-		if(read_metadata(meta_element) || license != false ){
-			console.log("Valid license for intrinsic events found");
-			addReportEntry(tabId, url, {url, "accepted":[url, `Global license for the page: ${license}`]});
-			// Do not process inline scripts
-			scripts="";
-		}else{
-			// Deal with intrinsic events
-			var has_intrinsic_events = [];
-			for(var i = 0; i < html_doc.all.length; i++){
-				for(var j = 0; j < intrinsic_events.length; j++){
-					if(intrinsic_events[j] in html_doc.all[i].attributes){
-						has_intrinsic_events.push([i,j]);
+	let license = false;
+	if (first_script_src != "") {
+		license = legacy_license_lib.check(first_script_src);
+	}
+	if (read_metadata(meta_element) || license) {
+		console.log("Valid license for intrinsic events found");
+		addReportEntry(tabId, url, {url, "accepted":[url, `Global license for the page: ${license}`]});
+		// Do not process inline scripts
+		scripts = [];
+	} else {
+		let modified = false;
+		
+		// Deal with intrinsic events
+		for (let element of html_doc.all) {
+			let attributes = element.attributes;
+			for (let event of intrinsic_events) {
+				if (event in attributes) {
+					let attr = attributes[events];
+					try {
+						let edited = await get_script(attr.value, `Intrinsic event [${event}]`);
+					  if (edited) {
+							let value = edited[0];
+							if (value !== attr.value) {
+								modified = true;
+								attr.value = value;
+							}
+						}
+					} catch (e) {
+						console.error(e);
 					}
 				}
 			}
-
-			// "i" is an index in html_doc.all
-			// "j" is an index in intrinsic_events
-			function edit_event(src,i,j,name){
-				var edited = get_script(src, name);
-				edited.then(function(){
-					html_doc.all[i].attributes[intrinsic_events[j]].value = edited[0];
-				});
-			}
-
-			// Find all the document's elements with intrinsic events
-			for(var i = 0; i < has_intrinsic_events.length; i++){
-				var s_name = "Intrinsic event ["+has_intrinsic_events[i][0]+"]";
-				edit_event(html_doc.all[has_intrinsic_events[i][0]].attributes[intrinsic_events[has_intrinsic_events[i][1]]].value,has_intrinsic_events[i][0],has_intrinsic_events[i][1],s_name);
-			}
-		}
-
-		// Deal with inline scripts
-		for(var i = 0; i < scripts.length; i++){
-			if(scripts[i].src == ""){
-				total_scripts++;
-			}
 		}
-
-		dbg_print("Analyzing "+total_scripts+" inline scripts...");
-
-		for(var i = 0; i < scripts.length; i++){
-			if (scripts[i].src == ""){
-				if (scripts[i].type=="" || scripts[i].type=="text/javascript"){
-					var edit_script = get_script(scripts[i].innerHTML, url, tabId, whitelisted, i);
-					edit_script.then(function(edited){
-						var edited_source = edited[0];
-						var unedited_source = html_doc.scripts[edited[1]].innerHTML.trim();
-						html_doc.scripts[edited[1]].innerHTML = edited_source;
-
-					});
-				}
-				amt_scripts++;
-				if(amt_scripts >= total_scripts){
-				resolve(remove_noscripts(html_doc));
+		
+		let modifiedInline = false;
+		for(let i = 0, len = scripts.length; i < len; i++) {
+			let script = scripts[i];
+			if (!script.src && !(script.type && script.type !== "text/javascript")) {
+				let edited = await get_script(script.textContent, url, tabId, whitelisted, i);
+				if (edited) {
+					let edited_source = edited[0];
+					let unedited_source = script.textContent.trim();
+					if (edited_source.trim() !== unedited_source) {
+						script.textContent = edited_source;
+						modified = modifiedInline = true;
+					}
 				}
 			}
+			if (modified) {
+				return modifiedInline 
+					? await remove_noscripts(html_doc) 
+					: doc2HTML(html_doc);
+			}
 		}
-		if(total_scripts == 0){
-			dbg_print("Nothing to analyze.");
-			resolve(remove_noscripts(html_doc));
-		}
-
-	});
+	}
+	return null;
 }
 
 /**
@@ -1124,7 +1109,7 @@ async function handle_html(response, whitelisted) {
 		activityReports[tabId] = await createReport({url, tabId});
 		updateBadge(tabId);
 	}
-	return await edit_html(text, url, tabId, frameId, whitelisted);
+	return await editHtml(text, url, tabId, frameId, whitelisted);
 }
 
 var whitelist = new ListStore("pref_whitelist", Storage.CSV);

From 8c94efa540af6d209152b657b3f71864fe25bd30 Mon Sep 17 00:00:00 2001
From: hackademix <giorgio@maone.net>
Date: Sep 02 2018 16:13:38 +0000
Subject: [PATCH 3/4] Fixes missing feedback on tab reload from UI panel.


---

diff --git a/html/display_panel/content/main_panel.js b/html/display_panel/content/main_panel.js
index 958e2d2..7bcbbb0 100644
--- a/html/display_panel/content/main_panel.js
+++ b/html/display_panel/content/main_panel.js
@@ -68,6 +68,7 @@ document.querySelector("#reload").onclick = async e => {
   let {tabId} = currentReport;
   if (tabId) {
     await browser.tabs.reload(tabId);
+    myPort.postMessage({"update": true, tabId});
   }
 };
 

From e5bf32c8866a4fb9d10d55689c1e3759f55992e3 Mon Sep 17 00:00:00 2001
From: hackademix <giorgio@maone.net>
Date: Sep 02 2018 21:02:35 +0000
Subject: [PATCH 4/4] Fixes missing feedback for actions on the report UI when opened in a tab.


---

diff --git a/html/display_panel/content/main_panel.js b/html/display_panel/content/main_panel.js
index 7bcbbb0..930f7f2 100644
--- a/html/display_panel/content/main_panel.js
+++ b/html/display_panel/content/main_panel.js
@@ -51,7 +51,7 @@ document.querySelector("#info").addEventListener("click", e => {
   if (site) {
     ([action] = action.split("-"));
   }
-	myPort.postMessage({[action]: entry, site});
+	myPort.postMessage({[action]: entry, site, tabId: currentReport.tabId});
 });
 
 document.querySelector("#report-tab").onclick = e => {