Mozaic MCP Server

import { readFileSync, existsSync, readdirSync, statSync } from "fs"; import { join, relative } from "path"; import type { Documentation } from "../db/queries.js"; interface ParsedDoc { title: string; content: string; category?: string; keywords: string[]; } // Extract frontmatter from MDX files function extractFrontmatter(content: string): { frontmatter: Record<string, string>; body: string; } { const frontmatterRegex = /^---\n([\s\S]*?)\n---\n([\s\S]*)$/; const match = content.match(frontmatterRegex); if (match) { const frontmatterStr = match[1]; const body = match[2]; const frontmatter: Record<string, string> = {}; const lines = frontmatterStr.split("\n"); for (const line of lines) { const colonIndex = line.indexOf(":"); if (colonIndex !== -1) { const key = line.substring(0, colonIndex).trim(); const value = line .substring(colonIndex + 1) .trim() .replace(/^['"]|['"]$/g, ""); frontmatter[key] = value; } } return { frontmatter, body }; } return { frontmatter: {}, body: content }; } // Extract title from content if not in frontmatter function extractTitle(content: string): string { // Try to find Storybook Meta title (e.g., <Meta title="Getting Started/Introduction" />) const metaMatch = content.match(/<Meta\s+title=["']([^"']+)["']/); if (metaMatch) { // Take the last part of the path as title (e.g., "Getting Started/Introduction" -> "Introduction") const parts = metaMatch[1].split("/"); return parts[parts.length - 1]; } // Try to find first h1 const h1Match = content.match(/^#\s+(.+)$/m); if (h1Match) { return h1Match[1]; } // Try to find MDX heading const mdxMatch = content.match(/<h1[^>]*>([^<]+)<\/h1>/i); if (mdxMatch) { return mdxMatch[1]; } return "Untitled"; } // Extract Storybook Meta title path for URL generation function extractMetaTitlePath(content: string): string | null { const metaMatch = content.match(/<Meta\s+title=["']([^"']+)["']/); if (metaMatch) { return metaMatch[1].toLowerCase().replace(/\s+/g, "-"); } return null; } // Extract keywords from content function extractKeywords(content: string, title: string): string[] { const keywords = new Set<string>(); // Add title words title.split(/\s+/).forEach((word) => { if (word.length > 2) { keywords.add(word.toLowerCase()); } }); // Extract component names (PascalCase) const componentRegex = /\bM[A-Z][a-zA-Z]+/g; const componentMatches = content.match(componentRegex); if (componentMatches) { componentMatches.forEach((match) => keywords.add(match.toLowerCase())); } // Extract CSS class names const cssClassRegex = /\bmc-[a-z0-9-]+/g; const cssMatches = content.match(cssClassRegex); if (cssMatches) { cssMatches.forEach((match) => keywords.add(match)); } // Extract code keywords const codeKeywords = ["props", "slots", "events", "emit", "component", "style"]; codeKeywords.forEach((keyword) => { if (content.toLowerCase().includes(keyword)) { keywords.add(keyword); } }); return Array.from(keywords); } // Clean MDX content for storage function cleanContent(content: string): string { // Remove import statements let cleaned = content.replace(/^import\s+.*$/gm, ""); // Remove Storybook Meta tags cleaned = cleaned.replace(/<Meta\s+[^>]*\/>/g, ""); // Extract code from Storybook Source components cleaned = cleaned.replace(/<Source[^>]*code=\{?["'`]([^"'`]+)["'`]\}?[^>]*\/>/g, "```\n$1\n```"); cleaned = cleaned.replace(/<Source[^>]*code=\{`([\s\S]*?)`\}[^>]*\/>/g, "```\n$1\n```"); // Remove JSX components but keep their text content cleaned = cleaned.replace(/<[A-Z][a-zA-Z]*[^>]*>([\s\S]*?)<\/[A-Z][a-zA-Z]*>/g, "$1"); // Remove self-closing JSX components cleaned = cleaned.replace(/<[A-Z][a-zA-Z]*[^>]*\/>/g, ""); // Remove empty code blocks cleaned = cleaned.replace(/```\s*```/g, ""); // Normalize whitespace cleaned = cleaned.replace(/\n{3,}/g, "\n\n"); return cleaned.trim(); } // Infer category from file path function inferCategory(filePath: string): string { const pathLower = filePath.toLowerCase(); if (pathLower.includes("component")) { return "components"; } if (pathLower.includes("foundation") || pathLower.includes("token")) { return "foundations"; } if (pathLower.includes("pattern")) { return "patterns"; } if (pathLower.includes("getting-started") || pathLower.includes("guide")) { return "guides"; } return "other"; } function parseMdxFile(filePath: string, _basePath: string): ParsedDoc | null { try { const content = readFileSync(filePath, "utf-8"); const { frontmatter, body } = extractFrontmatter(content); const title = frontmatter.title || extractTitle(body); const cleanedContent = cleanContent(body); const category = frontmatter.category || inferCategory(filePath); const keywords = extractKeywords(cleanedContent, title); return { title, content: cleanedContent, category, keywords, }; } catch (error) { console.warn(`Warning: Could not parse ${filePath}:`, error); return null; } } function findMdxFiles(dir: string): string[] { const files: string[] = []; if (!existsSync(dir)) { return files; } const entries = readdirSync(dir); for (const entry of entries) { const fullPath = join(dir, entry); const stat = statSync(fullPath); if (stat.isDirectory()) { files.push(...findMdxFiles(fullPath)); } else if (entry.endsWith(".mdx") || entry.endsWith(".md")) { files.push(fullPath); } } return files; } // Generate URL path from file path function generateUrlPath(filePath: string, basePath: string): string { let urlPath = relative(basePath, filePath); // Remove file extension urlPath = urlPath.replace(/\.(mdx?|md)$/, ""); // Convert to URL format urlPath = "/" + urlPath.replace(/\\/g, "/").toLowerCase(); // Remove index from path urlPath = urlPath.replace(/\/index$/, ""); return urlPath; } // Minimum content length to include a doc (filters out empty index pages) const MIN_CONTENT_LENGTH = 100; export async function parseDocumentation(docsPath: string): Promise<Documentation[]> { const docs: Documentation[] = []; const mdxFiles = findMdxFiles(docsPath); for (const file of mdxFiles) { const parsed = parseMdxFile(file, docsPath); if (parsed && parsed.content.length >= MIN_CONTENT_LENGTH) { docs.push({ title: parsed.title, path: generateUrlPath(file, docsPath), content: parsed.content, category: parsed.category, keywords: parsed.keywords, }); } } return docs; } // Parse Storybook MDX files from Vue/React repos export async function parseStorybookDocs( docsPath: string, framework: "vue" | "react", baseUrlPath: string ): Promise<Documentation[]> { const docs: Documentation[] = []; const mdxFiles = findMdxFiles(docsPath); for (const file of mdxFiles) { // Skip non-documentation files const fileName = file.split("/").pop() || ""; if (fileName.startsWith("Autodocs") || fileName.startsWith(".")) { continue; } try { const content = readFileSync(file, "utf-8"); const { frontmatter, body } = extractFrontmatter(content); // Get title from Meta tag or frontmatter or content const title = frontmatter.title || extractTitle(body); // Skip if no meaningful title if (title === "Untitled") { continue; } const cleanedContent = cleanContent(body); // Generate URL path from Meta title or filename const metaPath = extractMetaTitlePath(content); const urlPath = metaPath ? `${baseUrlPath}/${metaPath.replace(/\//g, "/")}` : `${baseUrlPath}/${fileName.replace(/\.mdx?$/, "").toLowerCase()}`; // Skip docs with very short content if (cleanedContent.length < MIN_CONTENT_LENGTH) { continue; } // Add framework-specific keywords const keywords = extractKeywords(cleanedContent, title); keywords.push(framework); keywords.push(`@mozaic-ds/${framework === "vue" ? "vue-3" : "react"}`); docs.push({ title: `${title} (${framework === "vue" ? "Vue" : "React"})`, path: urlPath, content: cleanedContent, category: `${framework}-docs`, keywords, }); } catch (error) { console.warn(`Warning: Could not parse ${file}:`, error); } } return docs; }