SAP Documentation MCP Server

// Intelligent content truncation utility // Preserves structure and readability while limiting content size import { CONFIG } from "./config.js"; export interface TruncationResult { content: string; wasTruncated: boolean; originalLength: number; truncatedLength: number; } /** * Intelligently truncate content to a maximum length while preserving: * - Beginning (intro/overview) * - End (conclusions/examples) * - Code block integrity * - Markdown structure * * @param content - The content to truncate * @param maxLength - Maximum length (defaults to CONFIG.MAX_CONTENT_LENGTH) * @returns TruncationResult with truncated content and metadata */ export function truncateContent( content: string, maxLength: number = CONFIG.MAX_CONTENT_LENGTH ): TruncationResult { const originalLength = content.length; // No truncation needed if (originalLength <= maxLength) { return { content, wasTruncated: false, originalLength, truncatedLength: originalLength }; } // Calculate how much content to preserve from start and end // Keep 60% from the start (intro/main content) and 20% from end (conclusions) // Reserve 20% for truncation notice and buffer const startLength = Math.floor(maxLength * 0.6); const endLength = Math.floor(maxLength * 0.2); const noticeLength = maxLength - startLength - endLength; // Extract start portion let startPortion = content.substring(0, startLength); // Try to break at a natural boundary (paragraph, heading, or code block) const naturalBreakPatterns = [ /\n\n/g, // Paragraph breaks /\n#{1,6}\s/g, // Markdown headings /\n```\n/g, // Code block ends /\n---\n/g, // Horizontal rules /\.\s+/g // Sentence ends ]; for (const pattern of naturalBreakPatterns) { const matches = Array.from(startPortion.matchAll(pattern)); if (matches.length > 0) { const lastMatch = matches[matches.length - 1]; if (lastMatch.index && lastMatch.index > startLength * 0.8) { startPortion = startPortion.substring(0, lastMatch.index + lastMatch[0].length); break; } } } // Extract end portion let endPortion = content.substring(content.length - endLength); // Try to break at natural boundary from the beginning of end portion for (const pattern of naturalBreakPatterns) { const match = endPortion.match(pattern); if (match && match.index !== undefined && match.index < endLength * 0.2) { endPortion = endPortion.substring(match.index + match[0].length); break; } } // Create truncation notice const omittedChars = originalLength - (startPortion.length + endPortion.length); const omittedPercent = Math.round((omittedChars / originalLength) * 100); const truncationNotice = ` --- ⚠️ **Content Truncated** The full content was ${originalLength.toLocaleString()} characters (approximately ${Math.round(originalLength / 4)} tokens). For readability and performance, ${omittedChars.toLocaleString()} characters (${omittedPercent}%) have been omitted from the middle section. The beginning and end of the document are preserved above and below this notice. --- `; // Combine portions const truncatedContent = startPortion + truncationNotice + endPortion; return { content: truncatedContent, wasTruncated: true, originalLength, truncatedLength: truncatedContent.length }; } /** * Truncate content with a simple notice at the end * Used when preserving both beginning and end doesn't make sense * * @param content - The content to truncate * @param maxLength - Maximum length (defaults to CONFIG.MAX_CONTENT_LENGTH) * @returns TruncationResult with truncated content and metadata */ export function truncateContentSimple( content: string, maxLength: number = CONFIG.MAX_CONTENT_LENGTH ): TruncationResult { const originalLength = content.length; // No truncation needed if (originalLength <= maxLength) { return { content, wasTruncated: false, originalLength, truncatedLength: originalLength }; } // Reserve space for truncation notice const noticeLength = 300; const contentLength = maxLength - noticeLength; // Extract content let truncatedContent = content.substring(0, contentLength); // Try to break at natural boundary const lastParagraph = truncatedContent.lastIndexOf('\n\n'); const lastSentence = truncatedContent.lastIndexOf('. '); if (lastParagraph > contentLength * 0.9) { truncatedContent = truncatedContent.substring(0, lastParagraph); } else if (lastSentence > contentLength * 0.9) { truncatedContent = truncatedContent.substring(0, lastSentence + 1); } // Add truncation notice const omittedChars = originalLength - truncatedContent.length; const omittedPercent = Math.round((omittedChars / originalLength) * 100); truncatedContent += ` --- ⚠️ **Content Truncated** The full content was ${originalLength.toLocaleString()} characters (approximately ${Math.round(originalLength / 4)} tokens). ${omittedChars.toLocaleString()} characters (${omittedPercent}%) have been omitted for readability. --- `; return { content: truncatedContent, wasTruncated: true, originalLength, truncatedLength: truncatedContent.length }; }