From 148e5e7c0c8de057a8829d80e0cedaa4e528cbd8 Mon Sep 17 00:00:00 2001
From: striderDM <51991544+StriderDM@users.noreply.github.com>
Date: Thu, 18 Nov 2021 17:06:53 +0200
Subject: [PATCH] feat: language detection for mnemonic seed words

Implementation of a basic language detection algorithm for mnemonic words

Review comments

cleanup
---
 base_layer/key_manager/src/mnemonic.rs | 54 +++++++++++++++++++++-----
 1 file changed, 44 insertions(+), 10 deletions(-)

diff --git a/base_layer/key_manager/src/mnemonic.rs b/base_layer/key_manager/src/mnemonic.rs
index 4226c8c250..afd0996819 100644
--- a/base_layer/key_manager/src/mnemonic.rs
+++ b/base_layer/key_manager/src/mnemonic.rs
@@ -33,7 +33,7 @@ use tari_crypto::tari_utilities::bit::*;
 /// It can autodetect the language of the Mnemonic word sequence
 // TODO: Develop a language autodetection mechanism to distinguish between ChineseTraditional and ChineseSimplified
 
-#[derive(Clone, Debug, PartialEq, EnumString, Display)]
+#[derive(Clone, Debug, PartialEq, EnumString, Display, Copy)]
 pub enum MnemonicLanguage {
     ChineseSimplified,
     English,
@@ -47,12 +47,8 @@ pub enum MnemonicLanguage {
 impl MnemonicLanguage {
     /// Detects the mnemonic language of a specific word by searching all defined mnemonic word lists
     pub fn from(mnemonic_word: &str) -> Result<MnemonicLanguage, MnemonicError> {
-        for language in MnemonicLanguage::iterator() {
-            if find_mnemonic_index_from_word(mnemonic_word, language).is_ok() {
-                return Ok((*language).clone());
-            }
-        }
-        Err(MnemonicError::UnknownLanguage)
+        let words = vec![mnemonic_word.to_string()];
+        detect_language(&words)
     }
 
     /// Returns an iterator for the MnemonicLanguage enum group to allow iteration over all defined languages
@@ -158,11 +154,49 @@ pub fn from_bytes(bytes: Vec<u8>, language: &MnemonicLanguage) -> Result<Vec<Str
     Ok(mnemonic_sequence)
 }
 
+fn detect_language(words: &[String]) -> Result<MnemonicLanguage, MnemonicError> {
+    if words.iter().len() < 1 {
+        return Err(MnemonicError::UnknownLanguage);
+    }
+
+    let mut index: usize = 0;
+    let count = words.iter().len() - 1;
+
+    while index <= count {
+        let mut languages = Vec::new();
+        let word = words.get(index).ok_or(MnemonicError::EncodeInvalidLength)?;
+        for language in MnemonicLanguage::iterator() {
+            if find_mnemonic_index_from_word(word, language).is_ok() {
+                languages.push(*language);
+            }
+        }
+
+        for language in languages {
+            let mut consistent = true;
+            for i in 0..count {
+                // exclude initial word
+                if i != index {
+                    let compare = words.get(i).ok_or(MnemonicError::EncodeInvalidLength)?;
+                    if find_mnemonic_index_from_word(compare, &language).is_err() {
+                        consistent = false;
+                    }
+                }
+            }
+            if consistent {
+                return Ok(language);
+            }
+        }
+
+        index += 1;
+    }
+
+    Err(MnemonicError::UnknownLanguage)
+}
+
 /// Generates a vector of bytes that represent the provided mnemonic sequence of words, the language of the mnemonic
-/// sequence is autodetected
+/// sequence is detected
 pub fn to_bytes(mnemonic_seq: &[String]) -> Result<Vec<u8>, MnemonicError> {
-    let first_word = mnemonic_seq.get(0).ok_or(MnemonicError::EncodeInvalidLength)?;
-    let language = MnemonicLanguage::from(first_word)?; // Autodetect language
+    let language = self::detect_language(mnemonic_seq)?;
     to_bytes_with_language(mnemonic_seq, &language)
 }