feat: Skip extract file content with txt or markdown file ✨

Author: lethemanh

packages/cozy-viewer/src/Panel/AI/AIAssistantPanel.jsx

@@ -5,7 +5,7 @@ import { useLocation, useNavigate } from 'react-router-dom'
 import { useI18n } from 'twake-i18n'
 
 import { useClient } from 'cozy-client'
-import { extractText, chatCompletion } from 'cozy-client/dist/models/ai'
+import { chatCompletion } from 'cozy-client/dist/models/ai'
 import { fetchBlobFileById } from 'cozy-client/dist/models/file'
 import flag from 'cozy-flags'
 import logger from 'cozy-logger'
@@ -21,9 +21,9 @@ import Stack from 'cozy-ui/transpiled/react/Stack'
 import Typography from 'cozy-ui/transpiled/react/Typography'
 import { useAlert } from 'cozy-ui/transpiled/react/providers/Alert'
 
+import { extractFileContent, validateContentSize } from './helpers'
 import { SUMMARY_SYSTEM_PROMPT, getSummaryUserPrompt } from './prompts'
 import styles from './styles.styl'
-import { roughTokensEstimation } from '../../helpers'
 import { useViewer } from '../../providers/ViewerProvider'
 
 const AIAssistantPanel = ({ className }) => {
@@ -52,22 +52,10 @@ const AIAssistantPanel = ({ className }) => {
   const summarizeFile = async ({ client, file, stream = false, model }) => {
     try {
       const fileBlob = await fetchBlobFileById(client, file?._id)
+      const textContent = await extractFileContent(client, fileBlob, file)
 
-      const rawTextContent = await extractText(client, fileBlob, {
-        name: file.name,
-        mime: file.mime
-      })
-      const textContent = rawTextContent ? JSON.stringify(rawTextContent) : ''
-
-      const summaryConfig = flag('drive.summary')
-      if (
-        summaryConfig?.maxTokens &&
-        roughTokensEstimation(textContent) > summaryConfig.maxTokens
-      ) {
-        const error = new Error('DOCUMENT_TOO_LARGE')
-        error.code = 'DOCUMENT_TOO_LARGE'
-        throw error
-      }
+      const { maxTokens } = flag('drive.summary') ?? {}
+      validateContentSize(textContent, maxTokens)
 
       const messages = [
         { role: 'system', content: SUMMARY_SYSTEM_PROMPT },
@@ -89,23 +77,22 @@ const AIAssistantPanel = ({ className }) => {
     }
   }
 
-  const persistedSummary = async (
-    fileMetadata,
-    targetFileId,
-    summaryContent
-  ) => {
-    try {
-      await client
-        .collection('io.cozy.files')
-        .updateMetadataAttribute(targetFileId, {
-          ...fileMetadata,
-          description: summaryContent
-        })
-      fetchedFileIdRef.current = targetFileId
-    } catch (error) {
-      logger.error('Error when persisting summary to file metadata:', error)
-    }
-  }
+  const persistedSummary = useCallback(
+    async (fileMetadata, targetFileId, summaryContent) => {
+      try {
+        await client
+          .collection('io.cozy.files')
+          .updateMetadataAttribute(targetFileId, {
+            ...fileMetadata,
+            description: summaryContent
+          })
+        fetchedFileIdRef.current = targetFileId
+      } catch (error) {
+        logger.error('Error when persisting summary to file metadata:', error)
+      }
+    },
+    [client]
+  )
 
   useEffect(() => {
     activeFileIdRef.current = file?._id || null
@@ -142,10 +129,14 @@ const AIAssistantPanel = ({ className }) => {
         await persistedSummary(fileMetadata, targetFileId, summaryContent)
       } catch (err) {
         if (activeFileIdRef.current === targetFileId) {
-          const errorMessage =
-            err.code === 'DOCUMENT_TOO_LARGE'
-              ? t('Viewer.ai.error.documentTooLarge')
-              : t('Viewer.ai.error.summary')
+          let errorMessage = t('Viewer.ai.error.summary')
+
+          if (err.code === 'DOCUMENT_TOO_LARGE') {
+            errorMessage = t('Viewer.ai.error.documentTooLarge')
+          } else if (err.code === 'CONTENT_EXTRACTION_FAILED') {
+            errorMessage = t('Viewer.ai.error.extractContent')
+          }
+
           setError(errorMessage)
         }
       } finally {
@@ -157,7 +148,7 @@ const AIAssistantPanel = ({ className }) => {
         }
       }
     },
-    [client, file, t]
+    [client, file, persistedSummary, t]
   )
 
   const handleRefresh = () => {

packages/cozy-viewer/src/Panel/AI/helpers.js

@@ -0,0 +1,49 @@
+import { extractText } from 'cozy-client/dist/models/ai'
+
+import { roughTokensEstimation, isTextMimeType } from '../../helpers'
+import { ContentExtractionError } from '../../helpers/ContentExtractionError'
+import { DocumentTooLargeError } from '../../helpers/DocumentTooLargeError'
+
+/**
+ * Extracts content from a file blob
+ * For text-based files, reads content directly. For other files, uses AI extraction.
+ * @param {object} client - Cozy client instance
+ * @param {Blob} fileBlob - File blob to extract content from
+ * @param {object} file - File metadata object with mime type and name
+ * @returns {Promise<string>} JSON stringified content
+ * @throws {ContentExtractionError} If content extraction fails or returns empty content
+ */
+export const extractFileContent = async (client, fileBlob, file) => {
+  let content
+
+  if (isTextMimeType(file.mime)) {
+    content = await fileBlob.text()
+  } else {
+    content = await extractText(client, fileBlob, {
+      name: file.name,
+      mime: file.mime
+    })
+  }
+
+  if (!content || content.trim().length === 0) {
+    throw new ContentExtractionError()
+  }
+
+  return JSON.stringify(content)
+}
+
+/**
+ * Validates that content size does not exceed the maximum token limit
+ * @param {string} textContent - Content to validate
+ * @param {number} maxTokens - Maximum number of tokens allowed
+ * @throws {DocumentTooLargeError} If content exceeds the token limit
+ */
+export const validateContentSize = (textContent, maxTokens) => {
+  if (!maxTokens) return
+
+  const tokens = roughTokensEstimation(textContent)
+
+  if (tokens > maxTokens) {
+    throw new DocumentTooLargeError()
+  }
+}

packages/cozy-viewer/src/helpers.js

@@ -84,6 +84,14 @@ export const roughTokensEstimation = text => {
   return Math.ceil(text.length / 4)
 }
 
+/**
+ * Check if a file is a text-based file type
+ * @param {string} mime - MIME type of the file
+ * @returns {boolean} Whether the file is a text file
+ */
+export const isTextMimeType = mime =>
+  typeof mime === 'string' && mime.toLowerCase().startsWith('text/')
+
 /**
  * Check if a file is compatible with AI summary feature
  * Compatible file types are defined in the drive.summary flag

packages/cozy-viewer/src/helpers/ContentExtractionError.js

@@ -0,0 +1,7 @@
+export class ContentExtractionError extends Error {
+  constructor() {
+    super('CONTENT_EXTRACTION_FAILED')
+    this.code = 'CONTENT_EXTRACTION_FAILED'
+    this.name = 'ContentExtractionError'
+  }
+}

packages/cozy-viewer/src/helpers/DocumentTooLargeError.js

@@ -0,0 +1,7 @@
+export class DocumentTooLargeError extends Error {
+  constructor() {
+    super('DOCUMENT_TOO_LARGE')
+    this.code = 'DOCUMENT_TOO_LARGE'
+    this.name = 'DocumentTooLargeError'
+  }
+}

packages/cozy-viewer/src/locales/en.json

@@ -86,7 +86,8 @@
       "bodyText": "Summary",
       "error": {
         "summary": "Failed to generate summary. Please try again.",
-        "documentTooLarge": "This document is too large to summarize. Please try with a shorter document."
+        "documentTooLarge": "This document is too large to summarize. Please try with a shorter document.",
+        "extractContent": "Failed to extract content from file"
       },
       "copied": "Summary copied to clipboard",
       "footerText": "This content is generated by AI and may contain errors.",

packages/cozy-viewer/src/locales/fr.json

@@ -86,7 +86,8 @@
       "bodyText": "Résumé",
       "error": {
         "summary": "Échec de la génération du résumé. Veuillez réessayer.",
-        "documentTooLarge": "Ce document est trop volumineux pour être résumé. Veuillez essayer avec un document plus court."
+        "documentTooLarge": "Ce document est trop volumineux pour être résumé. Veuillez essayer avec un document plus court.",
+        "extractContent": "Échec de l'extraction du contenu du fichier"
       },
       "copied": "Résumé copié dans le presse-papier",
       "footerText": "Ce contenu est généré par AI et peut contenir des erreurs.",

packages/cozy-viewer/src/locales/ru.json

@@ -86,7 +86,8 @@
       "bodyText": "Резюме",
       "error": {
         "summary": "Не удалось создать резюме. Пожалуйста, попробуйте снова.",
-        "documentTooLarge": "Этот документ слишком большой для резюмирования. Пожалуйста, попробуйте с более коротким документом."
+        "documentTooLarge": "Этот документ слишком большой для резюмирования. Пожалуйста, попробуйте с более коротким документом.",
+        "extractContent": "Не удалось извлечь содержимое из файла"
       },
       "copied": "Резюме скопировано в буфер обмена",
       "footerText": "Этот контент создан AI и может содержать ошибки.",

packages/cozy-viewer/src/locales/vi.json

@@ -86,7 +86,8 @@
       "bodyText": "Tóm tắt",
       "error": {
         "summary": "Không thể tạo tóm tắt. Vui lòng thử lại.",
-        "documentTooLarge": "Tài liệu này quá lớn để tóm tắt. Vui lòng thử với tài liệu ngắn hơn."
+        "documentTooLarge": "Tài liệu này quá lớn để tóm tắt. Vui lòng thử với tài liệu ngắn hơn.",
+        "extractContent": "Không thể trích xuất nội dung từ tệp"
       },
       "copied": "Đã sao chép tóm tắt vào khay nhớ tạm",
       "footerText": "Nội dung này được tạo bởi AI và có thể chứa lỗi.",