feat: 优化二进制的检测

Author: Flex-7746

frontend/src/hooks/useCode.ts

@@ -6,7 +6,7 @@ import { LANG_MAP } from '@/utils/option'
 import { getEncodeValue, readPath, saveFile } from '@/utils/file'
 
 import { useUserStore } from '@/store/user'
-// import { useOpenStore } from '@/store/open'
+import { useOpenStore } from '@/store/open'
 
 interface OptionModel {
   confirm: () => boolean
@@ -26,7 +26,7 @@ interface CodeModel {
 }
 
 export default function useCode(option: OptionModel) {
-  // const open = useOpenStore()
+  const open = useOpenStore()
   const user = useUserStore()
 
   const code = reactive<CodeModel>({
@@ -53,14 +53,15 @@ export default function useCode(option: OptionModel) {
       code.date = date
 
       const info = getEncodeValue(code.buffer)
-      code.encode = info.encode
-      code.org = code.value = info.value
-
-      // if (await isBinaryContent(data)) {
-      //   option.onError('不支持二进制文件的编辑')
-      //   open.removeHistory(path)
-      //   return
-      // }
+
+      if (info.encode) {
+        code.encode = info.encode
+        code.org = code.value = info.value
+      } else {
+        option.onError('不支持二进制文件的编辑')
+        open.removeHistory(path)
+        return
+      }
 
       const filename = path.split('/').pop() || ''
 

frontend/src/utils/file.ts

@@ -54,12 +54,15 @@ export const getKey = (path: string) =>
     .replace(/\./g, '_')
 
 export const getEncodeValue = (buffer: ArrayBuffer) => {
-  let encode = 'utf-8'
+  // 取前 1M 的文件进行编码判断
+  const match = buffer.slice(0, 1024 * 1024)
+
+  let encode = ''
 
   for (const item of ENCODING_OPTIONS) {
     try {
       const decoder = new TextDecoder(item.value)
-      const text = decoder.decode(buffer)
+      const text = decoder.decode(match)
 
       if (text.includes('�')) {
         continue
@@ -78,7 +81,26 @@ export const getEncodeValue = (buffer: ArrayBuffer) => {
     }
   }
 
-  return { encode, value: new TextDecoder(encode).decode(buffer) }
+  if (!encode && !isBinaryContent(buffer)) {
+    encode = 'utf-8'
+  }
+
+  return { encode, value: encode ? new TextDecoder(encode).decode(buffer) : '' }
+}
+
+export const isBinaryContent = (buffer: ArrayBuffer) => {
+  const bytes = new Uint8Array(buffer)
+
+  if (bytes.length === 0) {
+    return false
+  }
+
+  // 检查是否有 null 字节（文本文件中罕见）
+  if (bytes.includes(0)) {
+    return true
+  }
+
+  return bytes.reduce((sum, i) => sum + Number(i && i < 32 && i !== 9 && i !== 10 && i !== 13), 0) / bytes.length > 0.1
 }
 
 export const readPath = async <T = any>(opt: { path: string; responseType?: 'json' | 'arraybuffer' | 'text'; dir?: boolean }) => {

frontend/src/utils/option.ts

@@ -87,7 +87,7 @@ export const THEME_OPTIONS: { label: string; value: string; dark: boolean }[] =
 const codeMatch = (v: string) => {
   const sum = v.length
   const err = v.split('').filter((i) => i.charCodeAt(0) > 191).length
-  return err / sum < 0.1
+  return err / sum < 0.05
 }
 
 export const ENCODING_OPTIONS = [
@@ -105,10 +105,4 @@ export const ENCODING_OPTIONS = [
   { value: 'ascii', label: 'ASCII', match: codeMatch },
   { value: 'latin1', label: 'ISO-8859-1', match: codeMatch },
   { value: 'windows-1252', label: 'Windows-1252', match: codeMatch },
-
-  // 日韩编码
-  { value: 'shift_jis', label: 'Shift_JIS' },
-  { value: 'euc-jp', label: 'EUC-JP' },
-  { value: 'euc-kr', label: 'EUC-KR' },
-  { value: 'windows-949', label: 'Windows-949' },
 ]