Nutrient Document Engine MCP Server

import { DocumentEngineClient } from '../../api/Client.js'; import { TextLine } from '../../api/DocumentEngineSchema.js'; import { z } from 'zod'; import { MCPToolOutput } from '../../mcpTools.js'; import { getDocumentInfo, getDocumentPageText } from '../../api/DocumentLayerAbstraction.js'; import { DocumentFingerprintSchema } from '../schemas/DocumentFingerprintSchema.js'; import { formatBBox } from '../../utils/Common.js'; /** * Schema for extract_text tool */ // Create a Zod schema for PageRange const PageRangeSchema = z.object({ start: z.number().optional(), end: z.number().optional(), }); export const ExtractTextSchema = { document_fingerprint: DocumentFingerprintSchema, page_range: PageRangeSchema.optional().describe( 'Range of pages to include with start and end indices (0-based)' ), include_coordinates: z .boolean() .optional() .default(false) .describe('Whether to include text coordinates'), ocr_enabled: z .boolean() .optional() .default(false) .describe('Whether to enable OCR for text extraction'), }; export const ExtractTextInputSchema = z.object(ExtractTextSchema); export type ExtractTextInput = z.infer<typeof ExtractTextInputSchema>; /** * Extract text content from a document with OCR support and pagination */ export async function extractText( client: DocumentEngineClient, params: ExtractTextInput ): Promise<MCPToolOutput> { try { // Validate input const validatedParams = ExtractTextInputSchema.parse(params); const { document_fingerprint, page_range, include_coordinates = false, ocr_enabled = false, } = validatedParams; // Get document info to get total page count const docInfo = await getDocumentInfo(client, document_fingerprint); const pageCount = docInfo.pageCount || 0; // Parse page range if provided, otherwise use all pages let pageIndices: number[] = []; if (page_range) { const start = page_range.start !== undefined ? page_range.start : 0; const end = page_range.end !== undefined ? page_range.end : pageCount - 1; // Validate page range if (start < 0 || start >= pageCount) { throw new Error( `Page range start ${start} is out of bounds (document has ${pageCount} pages, valid indices are 0-${pageCount - 1})` ); } if (end < 0 || end >= pageCount) { throw new Error( `Page range end ${end} is out of bounds (document has ${pageCount} pages, valid indices are 0-${pageCount - 1})` ); } if (start > end) { throw new Error( `Invalid page range: start (${start}) must be less than or equal to end (${end})` ); } // Generate array of page indices from start to end pageIndices = Array.from({ length: end - start + 1 }, (_, i) => start + i); } else { pageIndices = Array.from({ length: pageCount }, (_, i) => i); } // Fetch text for each page const pageTextPromises = pageIndices.map(async pageIndex => { const response = await getDocumentPageText( client, document_fingerprint, pageIndex, ocr_enabled ); return { pageIndex, textLines: response.data.textLines, }; }); const pageTexts = await Promise.all(pageTextPromises); // Count total words let totalWords = 0; pageTexts.forEach(page => { const textLines = page.textLines || []; textLines.forEach((line: TextLine) => { // Count words in each line by splitting on whitespace const words = line.contents.trim().split(/\s+/).filter(Boolean); totalWords += words.length; }); }); // Build the markdown content let markdown = `# Text Extraction\n\n`; markdown += `📄 **Document ID:** ${document_fingerprint.document_id} \n`; if (document_fingerprint.layer) { markdown += `🔖 **Layer:** ${document_fingerprint.layer} \n`; } markdown += `📄 **Total Pages:** ${pageCount} \n`; markdown += `📝 **Total Words:** ${totalWords.toLocaleString()} \n`; markdown += `🔍 **OCR Applied:** ${ocr_enabled ? 'Yes' : 'No'} \n`; // Add page range information if provided if (page_range) { const rangeText = page_range.start !== undefined && page_range.end !== undefined ? `${page_range.start}-${page_range.end}` : page_range.start !== undefined ? `${page_range.start}-end` : page_range.end !== undefined ? `0-${page_range.end}` : 'All pages'; markdown += `📖 **Pages Processed:** ${rangeText} \n`; } else { markdown += `📖 **Pages Processed:** All pages \n`; } markdown += `\n---\n\n`; // Add text content for each page pageTexts.forEach(page => { const pageNumber = page.pageIndex + 1; // Count words on this page let pageWordCount = 0; const textLines = page.textLines || []; textLines.forEach((line: TextLine) => { const words = line.contents.trim().split(/\s+/).filter(Boolean); pageWordCount += words.length; }); markdown += `## Page ${pageNumber} (${pageWordCount} words)\n`; // Add coordinates if requested or if OCR if (include_coordinates || ocr_enabled) { markdown += `### Coordinates for Page ${pageNumber}\n`; markdown += `\`\`\`json\n`; markdown += JSON.stringify( textLines.map((line: TextLine) => ({ contents: line.contents, boundingBox: formatBBox([ line.left || 0, line.top || 0, line.width || 0, line.height || 0, ]), })), null, 2 ); markdown += `\n\`\`\`\n\n`; } else { // Combine all text lines for this page const pageText = textLines.map((line: TextLine) => line.contents).join(' '); markdown += `${pageText}\n\n`; } }); markdown += `---\n\n`; markdown += `💡 **Tip:** Use \`extract_form_data\` if this document contains fillable forms.\n`; return { markdown }; } catch (error) { // Provide a more user-friendly error message return { markdown: `# Error Extracting Text\n\nAn error occurred while trying to extract text: ${error instanceof Error ? error.message : String(error)}\n\nPlease check your connection and try again.`, }; } }