Tickets papnkukn#32 and papnkukn#33

joaoaugustogrobe · Apr 5, 2023 · 4ec7d67 · 4ec7d67
1 parent 5f4abb6
commit 4ec7d67
Showing 1 changed file with 42 additions and 3 deletions.
diff --git a/lib/eml-format.js b/lib/eml-format.js
@@ -173,9 +173,7 @@ var emlformat = {
     //Convert =0D to '\r', =20 to ' ', etc.
     if (!charset || charset == "utf8" || charset == "utf-8") {
       return s
-      .replace(/=([\w\d]{2})=([\w\d]{2})=([\w\d]{2})/gi, function(matcher, p1, p2, p3, offset, string) { return Buffer.from([ parseInt(p1, 16), parseInt(p2, 16), parseInt(p3, 16) ]).toString("utf8"); })
-      .replace(/=([\w\d]{2})=([\w\d]{2})/gi, function(matcher, p1, p2, offset, string) { return Buffer.from([ parseInt(p1, 16), parseInt(p2, 16) ]).toString("utf8"); })
-      .replace(/=([\w\d]{2})/gi, function(matcher, p1, offset, string) { return String.fromCharCode(parseInt(p1, 16)); })
+      .replace(/((=\w{2}((=)?\r?\n)?){1,})+/g, (matcher) => this.decodeUtf8String(matcher))
       .replace(/=\r?\n/gi, ""); //Join line
     }
     else {
@@ -184,6 +182,47 @@ var emlformat = {
       .replace(/=([\w\d]{2})/gi, function(matcher, p1, offset, string) { return iconv.decode(Buffer.from([ parseInt(p1, 16) ]), charset); })
       .replace(/=\r?\n/gi, ""); //Join line
     }
+  },
+
+  decodeUtf8String: function(inputStr) {
+    function decodeSequence(match) {
+      let bytes = match.match(/(?<=\=)[0-9A-Fa-f]{2}/g)
+      if (!bytes) return match;
+      bytes = bytes.map(b => parseInt(b, 16));
+      return recursiveDecodeSequence(bytes)
+    }
+
+    function recursiveDecodeSequence(bytes) {
+      let byte1 = bytes[0];
+      if (!byte1) return ""; // exit of recursion
+
+      if (byte1 <= 0x7F) {
+        return String.fromCharCode(byte1) + recursiveDecodeSequence(bytes.slice(1));
+      } 
+      if (byte1 >= 0xC0 && byte1 <= 0xDF) {
+        let byte2 = bytes[1];
+        let codePoint = ((byte1 & 0x1F) << 6) | (byte2 & 0x3F);
+        return String.fromCharCode(codePoint) + recursiveDecodeSequence(bytes.slice(2));
+      } 
+      if (byte1 >= 0xE0 && byte1 <= 0xEF) {
+        let byte2 = bytes[1];
+        let byte3 = bytes[2];
+        let codePoint = ((byte1 & 0x0F) << 12) | ((byte2 & 0x3F) << 6) | (byte3 & 0x3F);
+        return String.fromCharCode(codePoint) + recursiveDecodeSequence(bytes.slice(3));
+      } 
+      if (byte1 >= 0xF0 && byte1 <= 0xF7) {
+        let byte2 = bytes[1];
+        let byte3 = bytes[2];
+        let byte4 = bytes[3];
+        let codePoint = ((byte1 & 0x07) << 18) | ((byte2 & 0x3F) << 12) | ((byte3 & 0x3F) << 6) | (byte4 & 0x3F);
+        return String.fromCodePoint(codePoint) + recursiveDecodeSequence(bytes.slice(4));
+      }
+
+      return "�";
+    }
+    return inputStr
+      .replace(/((=)?\r?\n)?/g, "") // handle line break
+      .replace(/(?:(?:=\w{2}){1,}\r?\n?)+/g, decodeSequence);
   }
 };