Feat: Add TCADP parser for PPTX and spreadsheet document types. (#11041)

### What problem does this PR solve?

- Added TCADP Parser configuration fields to PDF, PPT, and spreadsheet
parsing forms
- Implemented support for setting table result type (Markdown/HTML) and
Markdown image response type (URL/Text)
- Updated TCADP Parser to handle return format settings from
configuration or parameters
- Enhanced frontend to dynamically show TCADP options based on selected
parsing method
- Modified backend to pass format parameters when calling TCADP API
- Optimized form default value logic for TCADP configuration items
- Updated multilingual resource files for new configuration options

### Type of change

- [x] New Feature (non-breaking change which adds functionality)

web/src/locales/en.ts

@@ -1752,6 +1752,8 @@ The variable aggregation node (originally the variable assignment node) is a cru
 The Indexer will store the content in the corresponding data structures for the selected methods.`,
       // file: 'File',
       parserMethod: 'PDF parser',
+      tableResultType: 'Table Result Type',
+      markdownImageResponseType: 'Markdown Image Response Type',
       // systemPrompt: 'System Prompt',
       systemPromptPlaceholder:
         'Enter system prompt for image analysis, if empty the system default value will be used',

web/src/locales/zh.ts

@@ -1629,6 +1629,8 @@ General：实体和关系提取提示来自 GitHub - microsoft/graphrag：基于
 Tokenizer 会根据所选方式将内容存储为对应的数据结构。`,
       filenameEmbdWeight: '文件名嵌入权重',
       parserMethod: '解析方法',
+      tableResultType: '表格返回形式',
+      markdownImageResponseType: '图片返回形式',
       systemPromptPlaceholder:
         '请输入用于图像分析的系统提示词，若为空则使用系统缺省值',
       exportJson: '导出 JSON',

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

@@ -169,6 +169,7 @@ export const initialParserValues = {
     {
       fileFormat: FileType.Spreadsheet,
       output_format: SpreadsheetOutputFormat.Html,
+      parse_method: ParseDocumentType.DeepDOC,
     },
     {
       fileFormat: FileType.Image,
@@ -192,6 +193,7 @@ export const initialParserValues = {
     {
       fileFormat: FileType.PowerPoint,
       output_format: PptOutputFormat.Json,
+      parse_method: ParseDocumentType.DeepDOC,
     },
   ],
 };
@@ -243,7 +245,7 @@ export const FileTypeSuffixMap = {
   [FileType.Email]: ['eml', 'msg'],
   [FileType.TextMarkdown]: ['md', 'markdown', 'mdx', 'txt'],
   [FileType.Docx]: ['doc', 'docx'],
-  [FileType.PowerPoint]: ['pptx'],
+  [FileType.PowerPoint]: ['pptx', 'ppt'],
   [FileType.Video]: ['mp4', 'avi', 'mkv'],
   [FileType.Audio]: [
     'da',

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

@@ -34,6 +34,8 @@ import { OutputFormatFormField } from './common-form-fields';
 import { EmailFormFields } from './email-form-fields';
 import { ImageFormFields } from './image-form-fields';
 import { PdfFormFields } from './pdf-form-fields';
+import { PptFormFields } from './ppt-form-fields';
+import { SpreadsheetFormFields } from './spreadsheet-form-fields';
 import { buildFieldNameWithPrefix } from './utils';
 import { AudioFormFields, VideoFormFields } from './video-form-fields';
 
@@ -41,6 +43,8 @@ const outputList = buildOutputList(initialParserValues.outputs);
 
 const FileFormatWidgetMap = {
   [FileType.PDF]: PdfFormFields,
+  [FileType.Spreadsheet]: SpreadsheetFormFields,
+  [FileType.PowerPoint]: PptFormFields,
   [FileType.Video]: VideoFormFields,
   [FileType.Audio]: AudioFormFields,
   [FileType.Email]: EmailFormFields,
@@ -65,6 +69,8 @@ export const FormSchema = z.object({
       fields: z.array(z.string()).optional(),
       llm_id: z.string().optional(),
       system_prompt: z.string().optional(),
+      table_result_type: z.string().optional(),
+      markdown_image_response_type: z.string().optional(),
     }),
   ),
 });
@@ -184,6 +190,8 @@ const ParserForm = ({ node }: INextOperatorForm) => {
       lang: '',
       fields: [],
       llm_id: '',
+      table_result_type: '',
+      markdown_image_response_type: '',
     });
   }, [append]);
 

web/src/pages/agent/form/parser-form/pdf-form-fields.tsx

@@ -1,13 +1,30 @@
 import { ParseDocumentType } from '@/components/layout-recognize-form-field';
+import {
+  SelectWithSearch,
+  SelectWithSearchFlagOptionType,
+} from '@/components/originui/select-with-search';
+import { RAGFlowFormItem } from '@/components/ragflow-form';
 import { isEmpty } from 'lodash';
 import { useEffect, useMemo } from 'react';
 import { useFormContext, useWatch } from 'react-hook-form';
+import { useTranslation } from 'react-i18next';
 import { LanguageFormField, ParserMethodFormField } from './common-form-fields';
 import { CommonProps } from './interface';
 import { useSetInitialLanguage } from './use-set-initial-language';
 import { buildFieldNameWithPrefix } from './utils';
 
+const tableResultTypeOptions: SelectWithSearchFlagOptionType[] = [
+  { label: 'Markdown', value: '0' },
+  { label: 'HTML', value: '1' },
+];
+
+const markdownImageResponseTypeOptions: SelectWithSearchFlagOptionType[] = [
+  { label: 'URL', value: '0' },
+  { label: 'Text', value: '1' },
+];
+
 export function PdfFormFields({ prefix }: CommonProps) {
+  const { t } = useTranslation();
   const form = useFormContext();
 
   const parseMethodName = buildFieldNameWithPrefix('parse_method', prefix);
@@ -25,6 +42,12 @@ export function PdfFormFields({ prefix }: CommonProps) {
     );
   }, [parseMethod]);
 
+  const tcadpOptionsShown = useMemo(() => {
+    return (
+      !isEmpty(parseMethod) && parseMethod === ParseDocumentType.TCADPParser
+    );
+  }, [parseMethod]);
+
   useSetInitialLanguage({ prefix, languageShown });
 
   useEffect(() => {
@@ -36,10 +59,68 @@ export function PdfFormFields({ prefix }: CommonProps) {
     }
   }, [form, parseMethodName]);
 
+  // Set default values for TCADP options when TCADP is selected
+  useEffect(() => {
+    if (tcadpOptionsShown) {
+      const tableResultTypeName = buildFieldNameWithPrefix(
+        'table_result_type',
+        prefix,
+      );
+      const markdownImageResponseTypeName = buildFieldNameWithPrefix(
+        'markdown_image_response_type',
+        prefix,
+      );
+
+      if (isEmpty(form.getValues(tableResultTypeName))) {
+        form.setValue(tableResultTypeName, '1', {
+          shouldValidate: true,
+          shouldDirty: true,
+        });
+      }
+      if (isEmpty(form.getValues(markdownImageResponseTypeName))) {
+        form.setValue(markdownImageResponseTypeName, '1', {
+          shouldValidate: true,
+          shouldDirty: true,
+        });
+      }
+    }
+  }, [tcadpOptionsShown, form, prefix]);
+
   return (
     <>
       <ParserMethodFormField prefix={prefix}></ParserMethodFormField>
       {languageShown && <LanguageFormField prefix={prefix}></LanguageFormField>}
+      {tcadpOptionsShown && (
+        <>
+          <RAGFlowFormItem
+            name={buildFieldNameWithPrefix('table_result_type', prefix)}
+            label={t('flow.tableResultType') || '表格返回形式'}
+          >
+            {(field) => (
+              <SelectWithSearch
+                value={field.value}
+                onChange={field.onChange}
+                options={tableResultTypeOptions}
+              ></SelectWithSearch>
+            )}
+          </RAGFlowFormItem>
+          <RAGFlowFormItem
+            name={buildFieldNameWithPrefix(
+              'markdown_image_response_type',
+              prefix,
+            )}
+            label={t('flow.markdownImageResponseType') || '图片返回形式'}
+          >
+            {(field) => (
+              <SelectWithSearch
+                value={field.value}
+                onChange={field.onChange}
+                options={markdownImageResponseTypeOptions}
+              ></SelectWithSearch>
+            )}
+          </RAGFlowFormItem>
+        </>
+      )}
     </>
   );
 }

web/src/pages/agent/form/parser-form/ppt-form-fields.tsx

@@ -0,0 +1,125 @@
+import { ParseDocumentType } from '@/components/layout-recognize-form-field';
+import {
+  SelectWithSearch,
+  SelectWithSearchFlagOptionType,
+} from '@/components/originui/select-with-search';
+import { RAGFlowFormItem } from '@/components/ragflow-form';
+import { isEmpty } from 'lodash';
+import { useEffect, useMemo } from 'react';
+import { useFormContext, useWatch } from 'react-hook-form';
+import { useTranslation } from 'react-i18next';
+import { ParserMethodFormField } from './common-form-fields';
+import { CommonProps } from './interface';
+import { buildFieldNameWithPrefix } from './utils';
+
+const tableResultTypeOptions: SelectWithSearchFlagOptionType[] = [
+  { label: 'Markdown', value: '0' },
+  { label: 'HTML', value: '1' },
+];
+
+const markdownImageResponseTypeOptions: SelectWithSearchFlagOptionType[] = [
+  { label: 'URL', value: '0' },
+  { label: 'Text', value: '1' },
+];
+
+export function PptFormFields({ prefix }: CommonProps) {
+  const { t } = useTranslation();
+  const form = useFormContext();
+
+  const parseMethodName = buildFieldNameWithPrefix('parse_method', prefix);
+
+  const parseMethod = useWatch({
+    name: parseMethodName,
+  });
+
+  // PPT only supports DeepDOC and TCADPParser
+  const optionsWithoutLLM = [
+    { label: ParseDocumentType.DeepDOC, value: ParseDocumentType.DeepDOC },
+    {
+      label: ParseDocumentType.TCADPParser,
+      value: ParseDocumentType.TCADPParser,
+    },
+  ];
+
+  const tcadpOptionsShown = useMemo(() => {
+    return (
+      !isEmpty(parseMethod) && parseMethod === ParseDocumentType.TCADPParser
+    );
+  }, [parseMethod]);
+
+  useEffect(() => {
+    if (isEmpty(form.getValues(parseMethodName))) {
+      form.setValue(parseMethodName, ParseDocumentType.DeepDOC, {
+        shouldValidate: true,
+        shouldDirty: true,
+      });
+    }
+  }, [form, parseMethodName]);
+
+  // Set default values for TCADP options when TCADP is selected
+  useEffect(() => {
+    if (tcadpOptionsShown) {
+      const tableResultTypeName = buildFieldNameWithPrefix(
+        'table_result_type',
+        prefix,
+      );
+      const markdownImageResponseTypeName = buildFieldNameWithPrefix(
+        'markdown_image_response_type',
+        prefix,
+      );
+
+      if (isEmpty(form.getValues(tableResultTypeName))) {
+        form.setValue(tableResultTypeName, '1', {
+          shouldValidate: true,
+          shouldDirty: true,
+        });
+      }
+      if (isEmpty(form.getValues(markdownImageResponseTypeName))) {
+        form.setValue(markdownImageResponseTypeName, '1', {
+          shouldValidate: true,
+          shouldDirty: true,
+        });
+      }
+    }
+  }, [tcadpOptionsShown, form, prefix]);
+
+  return (
+    <>
+      <ParserMethodFormField
+        prefix={prefix}
+        optionsWithoutLLM={optionsWithoutLLM}
+      ></ParserMethodFormField>
+      {tcadpOptionsShown && (
+        <>
+          <RAGFlowFormItem
+            name={buildFieldNameWithPrefix('table_result_type', prefix)}
+            label={t('flow.tableResultType') || '表格返回形式'}
+          >
+            {(field) => (
+              <SelectWithSearch
+                value={field.value}
+                onChange={field.onChange}
+                options={tableResultTypeOptions}
+              ></SelectWithSearch>
+            )}
+          </RAGFlowFormItem>
+          <RAGFlowFormItem
+            name={buildFieldNameWithPrefix(
+              'markdown_image_response_type',
+              prefix,
+            )}
+            label={t('flow.markdownImageResponseType') || '图片返回形式'}
+          >
+            {(field) => (
+              <SelectWithSearch
+                value={field.value}
+                onChange={field.onChange}
+                options={markdownImageResponseTypeOptions}
+              ></SelectWithSearch>
+            )}
+          </RAGFlowFormItem>
+        </>
+      )}
+    </>
+  );
+}

web/src/pages/agent/form/parser-form/spreadsheet-form-fields.tsx

@@ -0,0 +1,125 @@
+import { ParseDocumentType } from '@/components/layout-recognize-form-field';
+import {
+  SelectWithSearch,
+  SelectWithSearchFlagOptionType,
+} from '@/components/originui/select-with-search';
+import { RAGFlowFormItem } from '@/components/ragflow-form';
+import { isEmpty } from 'lodash';
+import { useEffect, useMemo } from 'react';
+import { useFormContext, useWatch } from 'react-hook-form';
+import { useTranslation } from 'react-i18next';
+import { ParserMethodFormField } from './common-form-fields';
+import { CommonProps } from './interface';
+import { buildFieldNameWithPrefix } from './utils';
+
+const tableResultTypeOptions: SelectWithSearchFlagOptionType[] = [
+  { label: 'Markdown', value: '0' },
+  { label: 'HTML', value: '1' },
+];
+
+const markdownImageResponseTypeOptions: SelectWithSearchFlagOptionType[] = [
+  { label: 'URL', value: '0' },
+  { label: 'Text', value: '1' },
+];
+
+export function SpreadsheetFormFields({ prefix }: CommonProps) {
+  const { t } = useTranslation();
+  const form = useFormContext();
+
+  const parseMethodName = buildFieldNameWithPrefix('parse_method', prefix);
+
+  const parseMethod = useWatch({
+    name: parseMethodName,
+  });
+
+  // Spreadsheet only supports DeepDOC and TCADPParser
+  const optionsWithoutLLM = [
+    { label: ParseDocumentType.DeepDOC, value: ParseDocumentType.DeepDOC },
+    {
+      label: ParseDocumentType.TCADPParser,
+      value: ParseDocumentType.TCADPParser,
+    },
+  ];
+
+  const tcadpOptionsShown = useMemo(() => {
+    return (
+      !isEmpty(parseMethod) && parseMethod === ParseDocumentType.TCADPParser
+    );
+  }, [parseMethod]);
+
+  useEffect(() => {
+    if (isEmpty(form.getValues(parseMethodName))) {
+      form.setValue(parseMethodName, ParseDocumentType.DeepDOC, {
+        shouldValidate: true,
+        shouldDirty: true,
+      });
+    }
+  }, [form, parseMethodName]);
+
+  // Set default values for TCADP options when TCADP is selected
+  useEffect(() => {
+    if (tcadpOptionsShown) {
+      const tableResultTypeName = buildFieldNameWithPrefix(
+        'table_result_type',
+        prefix,
+      );
+      const markdownImageResponseTypeName = buildFieldNameWithPrefix(
+        'markdown_image_response_type',
+        prefix,
+      );
+
+      if (isEmpty(form.getValues(tableResultTypeName))) {
+        form.setValue(tableResultTypeName, '1', {
+          shouldValidate: true,
+          shouldDirty: true,
+        });
+      }
+      if (isEmpty(form.getValues(markdownImageResponseTypeName))) {
+        form.setValue(markdownImageResponseTypeName, '1', {
+          shouldValidate: true,
+          shouldDirty: true,
+        });
+      }
+    }
+  }, [tcadpOptionsShown, form, prefix]);
+
+  return (
+    <>
+      <ParserMethodFormField
+        prefix={prefix}
+        optionsWithoutLLM={optionsWithoutLLM}
+      ></ParserMethodFormField>
+      {tcadpOptionsShown && (
+        <>
+          <RAGFlowFormItem
+            name={buildFieldNameWithPrefix('table_result_type', prefix)}
+            label={t('flow.tableResultType') || '表格返回形式'}
+          >
+            {(field) => (
+              <SelectWithSearch
+                value={field.value}
+                onChange={field.onChange}
+                options={tableResultTypeOptions}
+              ></SelectWithSearch>
+            )}
+          </RAGFlowFormItem>
+          <RAGFlowFormItem
+            name={buildFieldNameWithPrefix(
+              'markdown_image_response_type',
+              prefix,
+            )}
+            label={t('flow.markdownImageResponseType') || '图片返回形式'}
+          >
+            {(field) => (
+              <SelectWithSearch
+                value={field.value}
+                onChange={field.onChange}
+                options={markdownImageResponseTypeOptions}
+              ></SelectWithSearch>
+            )}
+          </RAGFlowFormItem>
+        </>
+      )}
+    </>
+  );
+}

web/src/pages/agent/utils.ts

@@ -214,6 +214,36 @@ function transformParserParams(params: ParserFormSchemaType) {
             parse_method: cur.parse_method,
             lang: cur.lang,
           };
+          // Only include TCADP parameters if TCADP Parser is selected
+          if (cur.parse_method?.toLowerCase() === 'tcadp parser') {
+            filteredSetup.table_result_type = cur.table_result_type;
+            filteredSetup.markdown_image_response_type =
+              cur.markdown_image_response_type;
+          }
+          break;
+        case FileType.Spreadsheet:
+          filteredSetup = {
+            ...filteredSetup,
+            parse_method: cur.parse_method,
+          };
+          // Only include TCADP parameters if TCADP Parser is selected
+          if (cur.parse_method?.toLowerCase() === 'tcadp parser') {
+            filteredSetup.table_result_type = cur.table_result_type;
+            filteredSetup.markdown_image_response_type =
+              cur.markdown_image_response_type;
+          }
+          break;
+        case FileType.PowerPoint:
+          filteredSetup = {
+            ...filteredSetup,
+            parse_method: cur.parse_method,
+          };
+          // Only include TCADP parameters if TCADP Parser is selected
+          if (cur.parse_method?.toLowerCase() === 'tcadp parser') {
+            filteredSetup.table_result_type = cur.table_result_type;
+            filteredSetup.markdown_image_response_type =
+              cur.markdown_image_response_type;
+          }
           break;
         case FileType.Image:
           filteredSetup = {

web/src/pages/data-flow/form/parser-form/ppt-form-fields.tsx

@@ -0,0 +1,40 @@
+import { ParseDocumentType } from '@/components/layout-recognize-form-field';
+import { isEmpty } from 'lodash';
+import { useEffect } from 'react';
+import { useFormContext } from 'react-hook-form';
+import { ParserMethodFormField } from './common-form-fields';
+import { CommonProps } from './interface';
+import { buildFieldNameWithPrefix } from './utils';
+
+export function PptFormFields({ prefix }: CommonProps) {
+  const form = useFormContext();
+
+  const parseMethodName = buildFieldNameWithPrefix('parse_method', prefix);
+
+  // PPT only supports DeepDOC and TCADPParser
+  const optionsWithoutLLM = [
+    { label: ParseDocumentType.DeepDOC, value: ParseDocumentType.DeepDOC },
+    {
+      label: ParseDocumentType.TCADPParser,
+      value: ParseDocumentType.TCADPParser,
+    },
+  ];
+
+  useEffect(() => {
+    if (isEmpty(form.getValues(parseMethodName))) {
+      form.setValue(parseMethodName, ParseDocumentType.DeepDOC, {
+        shouldValidate: true,
+        shouldDirty: true,
+      });
+    }
+  }, [form, parseMethodName]);
+
+  return (
+    <>
+      <ParserMethodFormField
+        prefix={prefix}
+        optionsWithoutLLM={optionsWithoutLLM}
+      ></ParserMethodFormField>
+    </>
+  );
+}

web/src/pages/data-flow/form/parser-form/spreadsheet-form-fields.tsx

@@ -0,0 +1,40 @@
+import { ParseDocumentType } from '@/components/layout-recognize-form-field';
+import { isEmpty } from 'lodash';
+import { useEffect } from 'react';
+import { useFormContext } from 'react-hook-form';
+import { ParserMethodFormField } from './common-form-fields';
+import { CommonProps } from './interface';
+import { buildFieldNameWithPrefix } from './utils';
+
+export function SpreadsheetFormFields({ prefix }: CommonProps) {
+  const form = useFormContext();
+
+  const parseMethodName = buildFieldNameWithPrefix('parse_method', prefix);
+
+  // Spreadsheet only supports DeepDOC and TCADPParser
+  const optionsWithoutLLM = [
+    { label: ParseDocumentType.DeepDOC, value: ParseDocumentType.DeepDOC },
+    {
+      label: ParseDocumentType.TCADPParser,
+      value: ParseDocumentType.TCADPParser,
+    },
+  ];
+
+  useEffect(() => {
+    if (isEmpty(form.getValues(parseMethodName))) {
+      form.setValue(parseMethodName, ParseDocumentType.DeepDOC, {
+        shouldValidate: true,
+        shouldDirty: true,
+      });
+    }
+  }, [form, parseMethodName]);
+
+  return (
+    <>
+      <ParserMethodFormField
+        prefix={prefix}
+        optionsWithoutLLM={optionsWithoutLLM}
+      ></ParserMethodFormField>
+    </>
+  );
+}