Refact: refact on parser structure (#14012)

### What problem does this PR solve?

Refact: refact on parser structure

### Type of change

- [x] Refactoring

web/src/locales/en.ts

@@ -2251,10 +2251,11 @@ This process aggregates variables from multiple branches into a single variable
         spreadsheet: 'Spreadsheet',
         image: 'Image',
         email: 'Email',
-        'text&markdown': 'Text & Markup',
-        code: 'Code',
+        markdown: 'Markdown',
+        'text&code': 'Text & Code',
         html: 'HTML',
-        word: 'Word',
+        doc: 'DOC',
+        docx: 'DOCX',
         slides: 'PPTX',
         audio: 'Audio',
         video: 'Video',

web/src/locales/zh.ts

@@ -1950,10 +1950,11 @@ General：实体和关系提取提示来自 GitHub - microsoft/graphrag：基于
         spreadsheet: '表格',
         image: '图片',
         email: '邮件',
-        'text&markdown': '文本与标记',
-        code: '代码',
+        markdown: 'Markdown',
+        'text&code': '文本与代码',
         html: 'HTML',
-        word: 'Word',
+        doc: 'DOC',
+        docx: 'DOCX',
         slides: 'PPTX',
         audio: '音频',
         video: '视频',

web/src/pages/agent/constant/pipeline.tsx

@@ -9,10 +9,11 @@ export enum FileType {
   Spreadsheet = 'spreadsheet',
   Image = 'image',
   Email = 'email',
-  TextMarkdown = 'text&markdown',
-  Code = 'code',
+  TextMarkdown = 'markdown',
+  Code = 'text&code',
   Html = 'html',
-  Docx = 'word',
+  Doc = 'doc',
+  Docx = 'docx',
   PowerPoint = 'slides',
   Video = 'video',
   Audio = 'audio',
@@ -41,6 +42,11 @@ export enum TextMarkdownOutputFormat {
   Text = 'text',
 }
 
+export enum TextJsonOutputFormat {
+  Text = 'text',
+  Json = 'json',
+}
+
 export enum DocxOutputFormat {
   Markdown = 'markdown',
   Json = 'json',
@@ -64,8 +70,9 @@ export const OutputFormatMap = {
   [FileType.Image]: ImageOutputFormat,
   [FileType.Email]: EmailOutputFormat,
   [FileType.TextMarkdown]: TextMarkdownOutputFormat,
-  [FileType.Code]: TextMarkdownOutputFormat,
-  [FileType.Html]: TextMarkdownOutputFormat,
+  [FileType.Code]: TextJsonOutputFormat,
+  [FileType.Html]: TextJsonOutputFormat,
+  [FileType.Doc]: DocxOutputFormat,
   [FileType.Docx]: DocxOutputFormat,
   [FileType.PowerPoint]: PptOutputFormat,
   [FileType.Video]: VideoOutputFormat,
@@ -78,8 +85,9 @@ export const InitialOutputFormatMap = {
   [FileType.Image]: ImageOutputFormat.Text,
   [FileType.Email]: EmailOutputFormat.Text,
   [FileType.TextMarkdown]: TextMarkdownOutputFormat.Text,
-  [FileType.Code]: TextMarkdownOutputFormat.Text,
-  [FileType.Html]: TextMarkdownOutputFormat.Text,
+  [FileType.Code]: TextJsonOutputFormat.Json,
+  [FileType.Html]: TextJsonOutputFormat.Json,
+  [FileType.Doc]: DocxOutputFormat.Json,
   [FileType.Docx]: DocxOutputFormat.Json,
   [FileType.PowerPoint]: PptOutputFormat.Json,
   [FileType.Video]: VideoOutputFormat.Text,
@@ -216,12 +224,17 @@ export const initialParserValues = {
     },
     {
       fileFormat: FileType.Code,
-      output_format: TextMarkdownOutputFormat.Text,
+      output_format: TextJsonOutputFormat.Json,
       preprocess: PreprocessValue.main_content,
     },
     {
       fileFormat: FileType.Html,
-      output_format: TextMarkdownOutputFormat.Text,
+      output_format: TextJsonOutputFormat.Json,
+      preprocess: PreprocessValue.main_content,
+    },
+    {
+      fileFormat: FileType.Doc,
+      output_format: DocxOutputFormat.Json,
       preprocess: PreprocessValue.main_content,
     },
     {
@@ -340,8 +353,9 @@ export const FileTypeSuffixMap = {
   [FileType.Spreadsheet]: ['xls', 'xlsx', 'csv'],
   [FileType.Image]: ['jpg', 'jpeg', 'png', 'gif'],
   [FileType.Email]: ['eml', 'msg'],
-  [FileType.TextMarkdown]: ['md', 'markdown', 'mdx', 'txt'],
+  [FileType.TextMarkdown]: ['md', 'markdown', 'mdx'],
   [FileType.Code]: [
+    'txt',
     'py',
     'js',
     'java',
@@ -357,7 +371,8 @@ export const FileTypeSuffixMap = {
     'sql',
   ],
   [FileType.Html]: ['htm', 'html'],
-  [FileType.Docx]: ['doc', 'docx'],
+  [FileType.Doc]: ['doc'],
+  [FileType.Docx]: ['docx'],
   [FileType.PowerPoint]: ['pptx', 'ppt'],
   [FileType.Video]: ['mp4', 'avi', 'mkv'],
   [FileType.Audio]: [

web/src/pages/agent/form/parser-form/index.tsx

@@ -82,6 +82,10 @@ const PreprocessOptionConfigsMap: Partial<
     { value: MAIN_CONTENT_PREPROCESS_VALUE, required: true },
     { value: PreprocessValue.section_title },
   ],
+  [FileType.Doc]: [
+    { value: MAIN_CONTENT_PREPROCESS_VALUE, required: true },
+    { value: PreprocessValue.section_title },
+  ],
   [FileType.Docx]: [
     { value: MAIN_CONTENT_PREPROCESS_VALUE, required: true },
     { value: PreprocessValue.section_title },