From 4d732a1f1dc5129f88442cff798eff5b96713545 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=C5=81ukasz=20Sroka?= <0xf1d0@protonmail.com>
Date: Mon, 16 Nov 2020 09:30:34 +0000
Subject: [PATCH] Added fn to count unicode characters (#1903)

## Issue Addressed

Password length check too short (https://github.com/sigp/lighthouse/issues/1880)

## Proposed Changes

I've added function that counts number of unicode characters, instead of calling String::len()


Co-authored-by: Paul Hauner <paul@paulhauner.com>
---
 common/account_utils/src/lib.rs | 25 ++++++++++++++++++++++---
 1 file changed, 22 insertions(+), 3 deletions(-)

diff --git a/common/account_utils/src/lib.rs b/common/account_utils/src/lib.rs
index d74ed71ed..f969db702 100644
--- a/common/account_utils/src/lib.rs
+++ b/common/account_utils/src/lib.rs
@@ -140,11 +140,17 @@ fn trim_newline(s: &mut String) {
     }
 }
 
+/// According to unicode, every byte that starts with 0b10xxxxxx continues encoding of character
+/// Therefore the number of characters equals number of bytes minus number of 0b10xxxxxx bytes
+fn count_unicode_characters(bits: &[u8]) -> usize {
+    bits.iter().filter(|bit| *bit >> 6 != 2).count()
+}
+
 /// Takes a string password and checks that it meets minimum requirements.
 ///
 /// The current minimum password requirement is a 12 character length character length.
 pub fn is_password_sufficiently_complex(password: &[u8]) -> Result<(), String> {
-    if password.len() >= MINIMUM_PASSWORD_LEN {
+    if count_unicode_characters(password) >= MINIMUM_PASSWORD_LEN {
         Ok(())
     } else {
         Err(format!(
@@ -192,8 +198,7 @@ impl AsRef<[u8]> for ZeroizeString {
 
 #[cfg(test)]
 mod test {
-    use super::is_password_sufficiently_complex;
-    use super::strip_off_newlines;
+    use super::*;
 
     #[test]
     fn test_strip_off() {
@@ -244,4 +249,18 @@ mod test {
     fn test_password_too_short() {
         is_password_sufficiently_complex(b"TestPass").unwrap();
     }
+
+    #[test]
+    fn unicode_characters() {
+        assert_eq!(count_unicode_characters(b""), 0);
+        assert_eq!(count_unicode_characters("🐱".to_string().as_bytes()), 1);
+        assert_eq!(count_unicode_characters("🐱🐱".to_string().as_bytes()), 2);
+
+        assert_eq!(count_unicode_characters(b"cats"), 4);
+        assert_eq!(count_unicode_characters("cats🐱".to_string().as_bytes()), 5);
+        assert_eq!(
+            count_unicode_characters("cats🐱🐱".to_string().as_bytes()),
+            6
+        );
+    }
 }