From 2881f70f6824f6ec63d5ef4891bfaeab145bb144 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=E9=B2=81=E6=A0=91=E4=BA=BA?= <lu.shuren@um-react.app>
Date: Sat, 17 Jun 2023 14:28:36 +0100
Subject: [PATCH] refactor: move text encoder/decoder to util file

---
 src/decrypt-worker/crypto/qmc/qmc_v2.ts |  4 ++--
 src/decrypt-worker/util/utf8Encoder.ts  | 10 ++++++++++
 src/util/MMKVParser.ts                  |  5 ++---
 src/util/levenshtein.ts                 |  8 ++++----
 4 files changed, 18 insertions(+), 9 deletions(-)
 create mode 100644 src/decrypt-worker/util/utf8Encoder.ts

diff --git a/src/decrypt-worker/crypto/qmc/qmc_v2.ts b/src/decrypt-worker/crypto/qmc/qmc_v2.ts
index 116a2a8..c39234a 100644
--- a/src/decrypt-worker/crypto/qmc/qmc_v2.ts
+++ b/src/decrypt-worker/crypto/qmc/qmc_v2.ts
@@ -3,6 +3,7 @@ import type { CryptoBase } from '../CryptoBase';
 import type { DecryptCommandOptions } from '~/decrypt-worker/types.ts';
 import { SEED, ENC_V2_KEY_1, ENC_V2_KEY_2 } from './qmc_v2.key.ts';
 import { fetchParakeet } from '@jixun/libparakeet';
+import { stringToUTF8Bytes } from '~/decrypt-worker/util/utf8Encoder.ts';
 
 export class QMC2Crypto implements CryptoBase {
   cryptoName = 'QMC/v2';
@@ -36,8 +37,7 @@ export class QMC2CryptoWithKey implements CryptoBase {
     }
 
     const parakeet = await fetchParakeet();
-    const textEncoder = new TextEncoder();
-    const key = textEncoder.encode(options.qmc2Key);
+    const key = stringToUTF8Bytes(options.qmc2Key);
     const keyCrypto = parakeet.make.QMCv2KeyCrypto(SEED, ENC_V2_KEY_1, ENC_V2_KEY_2);
     return transformBlob(buffer, (p) => p.make.QMCv2EKey(key, keyCrypto), {
       parakeet,
diff --git a/src/decrypt-worker/util/utf8Encoder.ts b/src/decrypt-worker/util/utf8Encoder.ts
new file mode 100644
index 0000000..e679be2
--- /dev/null
+++ b/src/decrypt-worker/util/utf8Encoder.ts
@@ -0,0 +1,10 @@
+export const utf8Encoder = new TextEncoder();
+export const utf8Decoder = new TextDecoder('utf-8');
+
+export function stringToUTF8Bytes(str: string) {
+  return utf8Encoder.encode(str);
+}
+
+export function bytesToUTF8String(str: BufferSource) {
+  return utf8Decoder.decode(str);
+}
diff --git a/src/util/MMKVParser.ts b/src/util/MMKVParser.ts
index 943e07a..f2a7d43 100644
--- a/src/util/MMKVParser.ts
+++ b/src/util/MMKVParser.ts
@@ -1,8 +1,7 @@
 import type { StagingKWMv2Key } from '~/features/settings/keyFormats';
+import { bytesToUTF8String } from '~/decrypt-worker/util/utf8Encoder';
 import { formatHex } from './formatHex';
 
-const textDecoder = new TextDecoder('utf-8', { ignoreBOM: true });
-
 export class MMKVParser {
   private offset = 4;
   private length: number;
@@ -67,7 +66,7 @@ export class MMKVParser {
     // ]
     const strByteLen = this.readInt();
     const data = this.readBytes(strByteLen);
-    return textDecoder.decode(data).normalize();
+    return bytesToUTF8String(data).normalize();
   }
 
   public readVariantString() {
diff --git a/src/util/levenshtein.ts b/src/util/levenshtein.ts
index 49617c5..340a057 100644
--- a/src/util/levenshtein.ts
+++ b/src/util/levenshtein.ts
@@ -1,6 +1,6 @@
-const textEncoder = new TextEncoder();
-
 // translation of pseudocode from Wikipedia:
+import { stringToUTF8Bytes } from '~/decrypt-worker/util/utf8Encoder';
+
 // https://en.wikipedia.org/wiki/Levenshtein_distance#Iterative_with_two_matrix_rows
 export function levenshtein(str1: string, str2: string) {
   if (str1 === str2) {
@@ -14,8 +14,8 @@ export function levenshtein(str1: string, str2: string) {
   }
 
   // Convert them to Uint8Array to avoid expensive string APIs.
-  const s = textEncoder.encode(str1.toLowerCase());
-  const t = textEncoder.encode(str2.toLowerCase());
+  const s = stringToUTF8Bytes(str1.normalize().toLowerCase());
+  const t = stringToUTF8Bytes(str2.normalize().toLowerCase());
   const m = s.byteLength;
   const n = t.byteLength;