Issue 5332 - Support Unicode characters in path

lib/abp2blocklist.js

Index: lib/abp2blocklist.js
===================================================================
--- a/lib/abp2blocklist.js
+++ b/lib/abp2blocklist.js
@@ -104,52 +104,53 @@
  *                     {regexp: "...",
  *                      canSafelyMatchAsLowercase: true/false,
  *                      hostname: "...",
  *                      justHostname: true/false}
  */
 function parseFilterRegexpSource(text)
 {
   let regexp = [];
-  let lastIndex = text.length - 1;
+  let characters = Array.from(text);

[Manish Jethani, 2017/06/19 14:41:06]

We just need to convert the string into an array to get each of the Unicode
characters in it.

Array.from is the only option that will work with JS Hydra and Safari 9. We
cannot use the spread operator or let..of.

[kzar, 2017/07/07 12:18:07]

Maybe add a comment about this?

[Manish Jethani, 2017/07/08 06:02:19]

Done.

+  let lastIndex = characters.length - 1;
   let hostname;
   let hostnameStart = null;
   let hostnameFinished = false;
   let justHostname = false;
   let canSafelyMatchAsLowercase = false;
 
-  for (let i = 0; i < text.length; i++)
+  for (let i = 0; i < characters.length; i++)
   {
-    let c = text[i];
+    let c = characters[i];

[Manish Jethani, 2017/06/19 14:41:06]

Note that c.length is 2 if the character is a surrogate pair.

Previously there was a bug because we did not read Unicode characters but rather
16-bit characters.

[kzar, 2017/07/07 12:18:07]

Acknowledged.

 
     if (hostnameFinished)
       justHostname = false;
 
     // If we're currently inside the hostname we have to be careful not to
     // escape any characters until after we have converted it to punycode.
     if (hostnameStart != null && !hostnameFinished)
     {
       let endingChar = (c == "*" || c == "^" ||
                         c == "?" || c == "/" || c == "|");
       if (!endingChar && i != lastIndex)
         continue;
 
       hostname = punycode.toASCII(
-        text.substring(hostnameStart, endingChar ? i : i + 1)
+        characters.slice(hostnameStart, endingChar ? i : i + 1).join("")
       );
       hostnameFinished = justHostname = true;
       regexp.push(escapeRegExp(hostname));
       if (!endingChar)
         break;
     }
 
     switch (c)
     {
       case "*":
-        if (regexp.length > 0 && i < lastIndex && text[i + 1] != "*")
+        if (regexp.length > 0 && i < lastIndex && characters[i + 1] != "*")
           regexp.push(".*");
         break;
       case "^":
         if (i < lastIndex)
           regexp.push(".");
         break;
       case "|":
         if (i == 0)
@@ -157,44 +158,44 @@
           regexp.push("^");
           break;
         }
         if (i == lastIndex)
         {
           regexp.push("$");
           break;
         }
-        if (i == 1 && text[0] == "|")
+        if (i == 1 && characters[0] == "|")
         {
           hostnameStart = i + 1;
           canSafelyMatchAsLowercase = true;
           regexp.push("https?://([^/]+\\.)?");
           break;
         }
         regexp.push("\\|");
         break;
       case "/":
         if (!hostnameFinished &&
-            text.charAt(i-2) == ":" && text.charAt(i-1) == "/")
+            characters[i - 2] == ":" && characters[i - 1] == "/")
         {
           hostnameStart = i + 1;
           canSafelyMatchAsLowercase = true;
         }
         regexp.push("/");
         break;
       case ".": case "+": case "$": case "?":
       case "{": case "}": case "(": case ")":
       case "[": case "]": case "\\":
         regexp.push("\\", c);
         break;
       default:
         if (hostnameFinished && (c >= "a" && c <= "z" ||
                                  c >= "A" && c <= "Z"))
           canSafelyMatchAsLowercase = false;
-        regexp.push(c);
+        regexp.push(c == "%" ? c : encodeURI(c));

[Manish Jethani, 2017/06/19 14:41:06]

Do not encode the percent sign because that means this is already an encoded
part of the URL. Browsers do this too.

I've added a unit test to test this case.

     }
   }
 
   return {
     regexp: regexp.join(""),
     canSafelyMatchAsLowercase: canSafelyMatchAsLowercase,
     hostname: hostname,
     justHostname: justHostname