changeset:

.changeset/fluffy-hairs-tap.md

@@ -0,0 +1,5 @@
+---
+"claude-dev": minor
+---
+
+add suppport for parsing csv and xlsx

package.json

@@ -373,6 +373,7 @@
 		"clone-deep": "^4.0.1",
 		"default-shell": "^2.2.0",
 		"diff": "^5.2.0",
+		"exceljs": "^4.4.0",
 		"execa": "^9.5.2",
 		"fast-deep-equal": "^3.1.3",
 		"firebase": "^11.2.0",

src/integrations/misc/extract-text.ts

@@ -6,6 +6,7 @@ import fs from "fs/promises"
 import { isBinaryFile } from "isbinaryfile"
 import * as chardet from "jschardet"
 import * as iconv from "iconv-lite"
+import ExcelJS from "exceljs"
 
 export async function detectEncoding(fileBuffer: Buffer, fileExtension?: string): Promise<string> {
 	const detected = chardet.detect(fileBuffer)
@@ -38,6 +39,8 @@ export async function extractTextFromFile(filePath: string): Promise<string> {
 			return extractTextFromDOCX(filePath)
 		case ".ipynb":
 			return extractTextFromIPYNB(filePath)
+		case ".xlsx":
+			return extractTextFromExcel(filePath)
 		default:
 			const fileBuffer = await fs.readFile(filePath)
 			if (fileBuffer.byteLength > 20 * 1000 * 1024) {
@@ -76,6 +79,101 @@ async function extractTextFromIPYNB(filePath: string): Promise<string> {
 	return extractedText
 }
 
+/**
+ * Format the data inside Excel cells
+ */
+function formatCellValue(cell: ExcelJS.Cell): string {
+	const value = cell.value
+	if (value === null || value === undefined) {
+		return ""
+	}
+
+	// Handle error values (#DIV/0!, #N/A, etc.)
+	if (typeof value === "object" && "error" in value) {
+		return `[Error: ${value.error}]`
+	}
+
+	// Handle dates - ExcelJS can parse them as Date objects
+	if (value instanceof Date) {
+		return value.toISOString().split("T")[0] // Just the date part
+	}
+
+	// Handle rich text
+	if (typeof value === "object" && "richText" in value) {
+		return value.richText.map((rt) => rt.text).join("")
+	}
+
+	// Handle hyperlinks
+	if (typeof value === "object" && "text" in value && "hyperlink" in value) {
+		return `${value.text} (${value.hyperlink})`
+	}
+
+	// Handle formulas - get the calculated result
+	if (typeof value === "object" && "formula" in value) {
+		if ("result" in value && value.result !== undefined && value.result !== null) {
+			return value.result.toString()
+		} else {
+			return `[Formula: ${value.formula}]`
+		}
+	}
+
+	return value.toString()
+}
+
+/**
+ * Extract and format text from xlsx files
+ */
+async function extractTextFromExcel(filePath: string): Promise<string> {
+	const workbook = new ExcelJS.Workbook()
+	let excelText = ""
+
+	try {
+		await workbook.xlsx.readFile(filePath)
+
+		workbook.eachSheet((worksheet, sheetId) => {
+			// Skip hidden sheets
+			if (worksheet.state === "hidden" || worksheet.state === "veryHidden") {
+				return
+			}
+
+			excelText += `--- Sheet: ${worksheet.name} ---\n`
+
+			worksheet.eachRow({ includeEmpty: false }, (row, rowNumber) => {
+				// Optional: limit processing for very large sheets
+				if (rowNumber > 50000) {
+					excelText += `[... truncated at row ${rowNumber} ...]\n`
+					return false
+				}
+
+				const rowTexts: string[] = []
+				let hasContent = false
+
+				row.eachCell({ includeEmpty: true }, (cell, colNumber) => {
+					const cellText = formatCellValue(cell)
+					if (cellText.trim()) {
+						hasContent = true
+					}
+					rowTexts.push(cellText)
+				})
+
+				// Only add rows with actual content
+				if (hasContent) {
+					excelText += rowTexts.join("\t") + "\n"
+				}
+
+				return true
+			})
+
+			excelText += "\n" // Blank line between sheets
+		})
+
+		return excelText.trim()
+	} catch (error: any) {
+		console.error(`Error extracting text from Excel ${filePath}:`, error)
+		throw new Error(`Failed to extract text from Excel: ${error.message}`)
+	}
+}
+
 /**
  * Helper function used to load file(s) and format them into a string
  */

src/integrations/misc/process-files.ts

@@ -9,7 +9,7 @@ import sizeOf from "image-size"
  */
 export async function selectFiles(imagesAllowed: boolean): Promise<{ images: string[]; files: string[] }> {
 	const IMAGE_EXTENSIONS = ["png", "jpg", "jpeg", "webp"] // supported by anthropic and openrouter
-	const OTHER_FILE_EXTENSIONS = ["xml", "json", "txt", "log", "md", "docx", "ipynb", "pdf"]
+	const OTHER_FILE_EXTENSIONS = ["xml", "json", "txt", "log", "md", "docx", "ipynb", "pdf", "xlsx", "csv"]
 
 	const options: vscode.OpenDialogOptions = {
 		canSelectMany: true,