Implement RFC2047 Q-Encoding to fix #68 #64 #59

src/content/scripts/editemailsubject.mjs

@@ -16,6 +16,8 @@
  * Modifications for TB78, TB91, TB102, TB115 by John Bieling (2020-2023)
  */
 
+import * as q from "/content/scripts/q.mjs";
+
 // Reduces a MessageList to a single message, if it contains a single message.
 export function getSingleMessageFromList(messageList) {
   return messageList && messageList.messages && messageList.messages.length == 1
@@ -71,11 +73,34 @@ export async function updateMessage({ msgId, tabId, keepBackup, newSubject, curr
   while (headerPart.match(/\r\nSubject: .*\r\n\s+/))
     headerPart = headerPart.replace(/(\r\nSubject: .*)(\r\n\s+)/, "$1 ");
 
+  // RFC2047 Q-Encoding
+  const encodeHeader = (name, value) => {
+    const startLine = " =?utf-8?q?"
+    const nextLine = "=?="
+    const endLine = "?=";
+
+    let lines = [];
+    let currentLine = `${name}:${startLine}`;
+
+    let uint8Array = new TextEncoder().encode(value);
+    for (let v of uint8Array.values()) {
+      let next = q.encode(String.fromCharCode(v)); // or btoa() for b-encoding
+      if (currentLine.length + next.length + endLine.length > 78) {
+        lines.push(currentLine + nextLine);
+        currentLine = startLine;
+      }
+      currentLine += next;
+    }
+    lines.push(currentLine + endLine);
+
+    return lines.join("\r\n");
+  }
+
   // Either replace the subject header or add one if missing.
   if (headerPart.includes("\nSubject: ")) {
-    headerPart = headerPart.replace(/\nSubject: .*\r\n/, "\nSubject: " + unescape(encodeURIComponent(newSubject)) + "\r\n");
+    headerPart = headerPart.replace(/\nSubject: .*\r\n/, "\n" + encodeHeader("Subject", newSubject) + "\r\n");
   } else {
-    headerPart += "Subject: " + unescape(encodeURIComponent(newSubject)) + "\r\n";
+    headerPart += encodeHeader("Subject", newSubject) + "\r\n";
   }
 
   // Without changing the message-id, the MessageHeader obj of the new message and the old message will

src/content/scripts/q.mjs

@@ -0,0 +1,41 @@
+/*! https://mths.be/q v1.0.0 by @mathias | MIT license */
+
+// https://tools.ietf.org/html/rfc2047#section-4.2
+var stringFromCharCode = String.fromCharCode;
+export function decode (input) {
+	return input
+		// Decode `_` into a space. This is character-encoding-independent;
+		// see https://tools.ietf.org/html/rfc2047#section-4.2, item 2.
+		.replace(/_/g, ' ')
+		// Decode escape sequences of the form `=XX` where `XX` is any
+		// combination of two hexidecimal digits. For optimal compatibility,
+		// lowercase hexadecimal digits are supported as well. See
+		// https://tools.ietf.org/html/rfc2045#section-6.7, note 1.
+		.replace(/=([a-fA-F0-9]{2})/g, function ($0, $1) {
+			var codePoint = parseInt($1, 16);
+			return stringFromCharCode(codePoint);
+		});
+};
+
+var regexUnsafeSymbols = /[\0-\x1F"-\),\.:-@\[-\^`\{-\uFFFF]/g;
+export function encode (string) {
+	// Note: this assumes the input is already encoded into octets (e.g. using
+	// UTF-8), and that the resulting octets are within the extended ASCII
+	// range.
+	return string
+		// Encode symbols that are definitely unsafe (i.e. unsafe in any context).
+		.replace(regexUnsafeSymbols, function (symbol) {
+			if (symbol > '\xFF') {
+				throw RangeError(
+					'`q.encode()` expects extended ASCII input only. Don\u2019t ' +
+					'forget to encode the input first using a character encoding ' +
+					'like UTF-8.'
+				);
+			}
+			var codePoint = symbol.charCodeAt(0);
+			var hexadecimal = codePoint.toString(16).toUpperCase();
+			return '=' + ('0' + hexadecimal).slice(-2);
+		})
+		// Encode spaces as `_`, as it’s shorter than `=20`.
+		.replace(/\x20/g, '_');
+};