docusaurus-plugin-llms 0.2.0 → 0.3.0

package/lib/generator.js

@@ -51,10 +51,11 @@ const processor_1 = require("./processor");
  * @returns Cleaned description suitable for TOC
  */
 function cleanDescriptionForToc(description) {
-    if (!description)
+    if (!(0, utils_1.isNonEmptyString)(description))
         return '';
     // Get just the first line for TOC display
-    const firstLine = description.split('\n')[0];
+    const lines = description.split('\n');
+    const firstLine = lines.length > 0 ? lines[0] : '';
     // Remove heading markers only at the beginning of the line
     // Be careful to only remove actual heading markers (# followed by space at beginning)
     // and not hashtag symbols that are part of the content (inline hashtags)
@@ -71,77 +72,74 @@ function cleanDescriptionForToc(description) {
  * @param includeFullContent - Whether to include full content or just links
  * @param version - Version of the file
  * @param customRootContent - Optional custom content to include at the root level
+ * @param batchSize - Batch size for processing documents (default: 100)
  */
-async function generateLLMFile(docs, outputPath, fileTitle, fileDescription, includeFullContent, version, customRootContent) {
-    console.log(`Generating file: ${outputPath}, version: ${version || 'undefined'}`);
+async function generateLLMFile(docs, outputPath, fileTitle, fileDescription, includeFullContent, version, customRootContent, batchSize = 100) {
+    // Validate path length before proceeding
+    if (!(0, utils_1.validatePathLength)(outputPath)) {
+        throw new Error(`Output path exceeds maximum length: ${outputPath}`);
+    }
+    utils_1.logger.verbose(`Generating file: ${outputPath}, version: ${version || 'undefined'}`);
     const versionInfo = version ? `\n\nVersion: ${version}` : '';
     if (includeFullContent) {
         // Generate full content file with header deduplication
+        // Process documents in batches to prevent memory issues on large sites
         const usedHeaders = new Set();
-        const fullContentSections = docs.map(doc => {
-            // Check if content already starts with the same heading to avoid duplication
-            const trimmedContent = doc.content.trim();
-            const firstLine = trimmedContent.split('\n')[0];
-            // Check if the first line is a heading that matches our title
-            const headingMatch = firstLine.match(/^#+\s+(.+)$/);
-            const firstHeadingText = headingMatch ? headingMatch[1].trim() : null;
-            // Determine the header text to use (original title or make it unique)
-            let headerText = doc.title;
-            let uniqueHeader = headerText;
-            let counter = 1;
-            // If this header has been used before, make it unique by adding a suffix
-            while (usedHeaders.has(uniqueHeader.toLowerCase())) {
-                counter++;
-                // Try to make it more descriptive by adding the file path info if available
-                if (doc.path && counter === 2) {
-                    const pathParts = doc.path.split('/');
-                    const folderName = pathParts.length > 1 ? pathParts[pathParts.length - 2] : '';
-                    if (folderName) {
-                        uniqueHeader = `${headerText} (${folderName.charAt(0).toUpperCase() + folderName.slice(1)})`;
-                    }
-                    else {
-                        uniqueHeader = `${headerText} (${counter})`;
-                    }
-                }
-                else {
-                    uniqueHeader = `${headerText} (${counter})`;
-                }
+        const fullContentSections = [];
+        // Process documents in batches
+        for (let i = 0; i < docs.length; i += batchSize) {
+            const batch = docs.slice(i, i + batchSize);
+            const batchNumber = Math.floor(i / batchSize) + 1;
+            const totalBatches = Math.ceil(docs.length / batchSize);
+            if (totalBatches > 1) {
+                utils_1.logger.verbose(`Processing batch ${batchNumber}/${totalBatches} (${batch.length} documents)`);
             }
-            usedHeaders.add(uniqueHeader.toLowerCase());
-            if (firstHeadingText === doc.title) {
-                // Content already has the same heading, replace it with our unique header if needed
-                if (uniqueHeader !== doc.title) {
+            const batchSections = batch.map(doc => {
+                // Check if content already starts with the same heading to avoid duplication
+                const trimmedContent = doc.content.trim();
+                const contentLines = trimmedContent.split('\n');
+                const firstLine = contentLines.length > 0 ? contentLines[0] : '';
+                // Check if the first line is a heading that matches our title
+                const headingMatch = firstLine.match(/^#+\s+(.+)$/);
+                const firstHeadingText = headingMatch ? headingMatch[1].trim() : null;
+                // Generate unique header using the utility function
+                const uniqueHeader = (0, utils_1.ensureUniqueIdentifier)(doc.title, usedHeaders, (counter, base) => {
+                    // Try to make it more descriptive by adding the file path info if available
+                    if ((0, utils_1.isNonEmptyString)(doc.path) && counter === 2) {
+                        const pathParts = doc.path.split('/');
+                        const folderName = pathParts.length >= 2 ? pathParts[pathParts.length - 2] : '';
+                        if ((0, utils_1.isNonEmptyString)(folderName)) {
+                            return `(${folderName.charAt(0).toUpperCase() + folderName.slice(1)})`;
+                        }
+                    }
+                    return `(${counter})`;
+                });
+                if (firstHeadingText === doc.title) {
+                    // Content already has the same heading, replace it with our unique header
                     const restOfContent = trimmedContent.split('\n').slice(1).join('\n');
                     return `## ${uniqueHeader}
 
 ${restOfContent}`;
                 }
                 else {
-                    // Replace the existing H1 with H2 to comply with llmstxt.org standard
-                    const restOfContent = trimmedContent.split('\n').slice(1).join('\n');
+                    // Content doesn't have the same heading, add our unique H2 header
                     return `## ${uniqueHeader}
 
-${restOfContent}`;
-                }
-            }
-            else {
-                // Content doesn't have the same heading, add our unique H2 header
-                return `## ${uniqueHeader}
-
 ${doc.content}`;
-            }
-        });
+                }
+            });
+            fullContentSections.push(...batchSections);
+        }
         // Use custom root content or default message
         const rootContent = customRootContent || 'This file contains all documentation content in a single document following the llmstxt.org standard.';
-        const llmFileContent = `# ${fileTitle}
-
-> ${fileDescription}${versionInfo}
-
-${rootContent}
-
-${fullContentSections.join('\n\n---\n\n')}
-`;
-        await (0, utils_1.writeFile)(outputPath, llmFileContent);
+        const llmFileContent = (0, utils_1.createMarkdownContent)(fileTitle, `${fileDescription}${versionInfo}`, `${rootContent}\n\n${fullContentSections.join('\n\n---\n\n')}`, true // include metadata (description)
+        );
+        try {
+            await (0, utils_1.writeFile)(outputPath, llmFileContent);
+        }
+        catch (error) {
+            throw new Error(`Failed to write file ${outputPath}: ${(0, utils_1.getErrorMessage)(error)}`);
+        }
     }
     else {
         // Generate links-only file
@@ -152,71 +150,144 @@ ${fullContentSections.join('\n\n---\n\n')}
         });
         // Use custom root content or default message
         const rootContent = customRootContent || 'This file contains links to documentation sections following the llmstxt.org standard.';
-        const llmFileContent = `# ${fileTitle}
-
-> ${fileDescription}${versionInfo}
-
-${rootContent}
-
-## Table of Contents
-
-${tocItems.join('\n')}
-`;
-        await (0, utils_1.writeFile)(outputPath, llmFileContent);
+        const llmFileContent = (0, utils_1.createMarkdownContent)(fileTitle, `${fileDescription}${versionInfo}`, `${rootContent}\n\n## Table of Contents\n\n${tocItems.join('\n')}`, true // include metadata (description)
+        );
+        try {
+            await (0, utils_1.writeFile)(outputPath, llmFileContent);
+        }
+        catch (error) {
+            throw new Error(`Failed to write file ${outputPath}: ${(0, utils_1.getErrorMessage)(error)}`);
+        }
     }
-    console.log(`Generated: ${outputPath}`);
+    utils_1.logger.info(`Generated: ${outputPath}`);
 }
 /**
  * Generate individual markdown files for each document
  * @param docs - Processed document information
  * @param outputDir - Directory to write the markdown files
  * @param siteUrl - Base site URL
+ * @param docsDir - The configured docs directory name (e.g., 'docs', 'documentation', etc.)
+ * @param keepFrontMatter - Array of frontmatter keys to preserve in generated files
+ * @param preserveDirectoryStructure - Whether to preserve the full directory structure (default: true)
  * @returns Updated docs with new URLs pointing to generated markdown files
  */
-async function generateIndividualMarkdownFiles(docs, outputDir, siteUrl) {
+async function generateIndividualMarkdownFiles(docs, outputDir, siteUrl, docsDir = 'docs', keepFrontMatter = [], preserveDirectoryStructure = true) {
     const updatedDocs = [];
-    // Create a map to ensure unique filenames
-    const usedFilenames = new Set();
+    const usedPaths = new Set();
     for (const doc of docs) {
-        // Generate a filename from the document title or URL path
-        let baseFilename = doc.title
-            .toLowerCase()
-            .replace(/[^a-z0-9]+/g, '-')
-            .replace(/^-+|-+$/g, '');
-        // Fallback to URL path if title generates empty filename
-        if (!baseFilename) {
-            baseFilename = doc.path
-                .replace(/^\/+|\/+$/g, '') // Remove leading/trailing slashes
-                .replace(/\//g, '-')
-                .replace(/[^a-z0-9-]/gi, '-')
-                .toLowerCase();
+        // Use the original path structure as default filename.
+        let relativePath = doc.path
+            .replace(/^\/+/, '') // Remove leading slashes
+            .replace(/\.mdx?$/, '.md'); // Ensure .md extension
+        // Strip the docsDir prefix only if preserveDirectoryStructure is false
+        if (!preserveDirectoryStructure) {
+            relativePath = relativePath
+                .replace(new RegExp(`^${docsDir.replace(/[.*+?^${}()|[\]\\]/g, '\\$&')}/`), ''); // Remove configured docs dir prefix
+        }
+        // If frontmatter has slug, use that.
+        if ((0, utils_1.isNonEmptyString)(doc.frontMatter?.slug)) {
+            const slug = doc.frontMatter.slug.trim().replace(/^\/+|\/+$/g, ''); // Trim whitespace and slashes
+            if ((0, utils_1.isNonEmptyString)(slug)) { // Only process if slug is not empty after trimming
+                if (slug.includes('/')) {
+                    // Nested slug: create directory structure
+                    relativePath = slug + '.md';
+                }
+                else {
+                    // Simple slug: replace just the filename
+                    const pathParts = relativePath.replace(/\.md$/, '').split('/');
+                    pathParts[pathParts.length - 1] = slug;
+                    relativePath = pathParts.join('/') + '.md';
+                }
+            }
+        }
+        // Otherwise, if frontmatter has id, use that.
+        else if ((0, utils_1.isNonEmptyString)(doc.frontMatter?.id)) {
+            const id = doc.frontMatter.id.trim().replace(/^\/+|\/+$/g, ''); // Trim whitespace and slashes
+            if ((0, utils_1.isNonEmptyString)(id)) { // Only process if id is not empty after trimming
+                if (id.includes('/')) {
+                    // Nested id: create directory structure
+                    relativePath = id + '.md';
+                }
+                else {
+                    // Simple id: replace just the filename
+                    const pathParts = relativePath.replace(/\.md$/, '').split('/');
+                    pathParts[pathParts.length - 1] = id;
+                    relativePath = pathParts.join('/') + '.md';
+                }
+            }
         }
-        // Ensure filename uniqueness
-        let filename = `${baseFilename}.md`;
+        // Trim any leading/trailing whitespace from the path
+        relativePath = relativePath.trim();
+        // If path is empty or invalid, create a fallback path
+        if (!(0, utils_1.isNonEmptyString)(relativePath) || relativePath === '.md') {
+            const sanitizedTitle = (0, utils_1.sanitizeForFilename)(doc.title, 'untitled');
+            relativePath = `${sanitizedTitle}.md`;
+        }
+        // Ensure path uniqueness
+        let uniquePath = relativePath;
         let counter = 1;
-        while (usedFilenames.has(filename)) {
-            filename = `${baseFilename}-${counter}.md`;
+        const MAX_PATH_ITERATIONS = 10000;
+        let pathIterations = 0;
+        while (usedPaths.has(uniquePath.toLowerCase())) {
             counter++;
+            const pathParts = relativePath.split('.');
+            const extension = pathParts.pop() || 'md';
+            const basePath = pathParts.join('.');
+            uniquePath = `${basePath}-${counter}.${extension}`;
+            pathIterations++;
+            if (pathIterations >= MAX_PATH_ITERATIONS) {
+                // Fallback to timestamp
+                const timestamp = Date.now();
+                uniquePath = `${basePath}-${timestamp}.${extension}`;
+                utils_1.logger.warn(`Maximum iterations reached for unique path. Using timestamp: ${uniquePath}`);
+                break;
+            }
         }
-        usedFilenames.add(filename);
-        // Create markdown content following llmstxt.org standard
-        const markdownContent = `# ${doc.title}
-
-> ${doc.description}
-
-${doc.content}
-`;
+        usedPaths.add(uniquePath.toLowerCase());
+        // Create the full file path and validate/shorten if needed
+        let fullPath = path.join(outputDir, uniquePath);
+        fullPath = (0, utils_1.shortenPathIfNeeded)(fullPath, outputDir, uniquePath);
+        // Update uniquePath to reflect the shortened path if it was changed
+        if (fullPath !== path.join(outputDir, uniquePath)) {
+            uniquePath = path.relative(outputDir, fullPath);
+        }
+        const directory = path.dirname(fullPath);
+        // Create directory structure if it doesn't exist
+        try {
+            await fs.mkdir(directory, { recursive: true });
+        }
+        catch (error) {
+            throw new Error(`Failed to create directory ${directory}: ${(0, utils_1.getErrorMessage)(error)}`);
+        }
+        // Extract preserved frontmatter if specified
+        let preservedFrontMatter = {};
+        if ((0, utils_1.isNonEmptyArray)(keepFrontMatter) && (0, utils_1.isDefined)(doc.frontMatter)) {
+            for (const key of keepFrontMatter) {
+                if (key in doc.frontMatter) {
+                    preservedFrontMatter[key] = doc.frontMatter[key];
+                }
+            }
+        }
+        // Create markdown content using the utility function
+        const markdownContent = (0, utils_1.createMarkdownContent)(doc.title, doc.description, doc.content, true, // includeMetadata
+        Object.keys(preservedFrontMatter).length > 0 ? preservedFrontMatter : undefined);
         // Write the markdown file
-        const markdownPath = path.join(outputDir, filename);
-        await (0, utils_1.writeFile)(markdownPath, markdownContent);
+        try {
+            await (0, utils_1.writeFile)(fullPath, markdownContent);
+        }
+        catch (error) {
+            throw new Error(`Failed to write file ${fullPath}: ${(0, utils_1.getErrorMessage)(error)}`);
+        }
         // Create updated DocInfo with new URL pointing to the generated markdown file
-        const newUrl = `${siteUrl}/${filename}`;
+        // Convert file path to URL path (use forward slashes)
+        const urlPath = (0, utils_1.normalizePath)(uniquePath);
+        const newUrl = `${siteUrl}/${urlPath}`;
         updatedDocs.push({
             ...doc,
             url: newUrl,
-            path: `/${filename}` // Update path to the new markdown file
+            path: `/${urlPath}` // Update path to the new markdown file
         });
-        console.log(`Generated markdown file: ${filename}`);
+        utils_1.logger.verbose(`Generated markdown file: ${uniquePath}`);
     }
     return updatedDocs;
 }
@@ -227,31 +298,38 @@ ${doc.content}
  */
 async function generateStandardLLMFiles(context, allDocFiles) {
     const { outDir, siteUrl, docTitle, docDescription, options } = context;
-    const { generateLLMsTxt, generateLLMsFullTxt, llmsTxtFilename = 'llms.txt', llmsFullTxtFilename = 'llms-full.txt', includeOrder = [], includeUnmatchedLast = true, version, generateMarkdownFiles = false, rootContent, fullRootContent } = options;
+    const { generateLLMsTxt, generateLLMsFullTxt, llmsTxtFilename = 'llms.txt', llmsFullTxtFilename = 'llms-full.txt', includeOrder = [], includeUnmatchedLast = true, version, generateMarkdownFiles = false, rootContent, fullRootContent, processingBatchSize = 100 } = options;
     if (!generateLLMsTxt && !generateLLMsFullTxt) {
+        utils_1.logger.warn('No standard LLM files configured for generation. Skipping.');
         return;
     }
     // Process files for the standard outputs
     let processedDocs = await (0, processor_1.processFilesWithPatterns)(context, allDocFiles, [], // No specific include patterns - include all
     [], // No additional ignore patterns beyond global ignoreFiles
     includeOrder, includeUnmatchedLast);
-    console.log(`Processed ${processedDocs.length} documentation files for standard LLM files`);
+    utils_1.logger.verbose(`Processed ${processedDocs.length} documentation files for standard LLM files`);
+    // Check if we have documents to process
+    if (!(0, utils_1.isNonEmptyArray)(processedDocs)) {
+        utils_1.logger.warn('No documents found matching patterns for standard LLM files. Skipping.');
+        return;
+    }
     // Generate individual markdown files if requested
-    if (generateMarkdownFiles && processedDocs.length > 0) {
-        console.log('Generating individual markdown files...');
-        processedDocs = await generateIndividualMarkdownFiles(processedDocs, outDir, siteUrl);
+    if (generateMarkdownFiles) {
+        utils_1.logger.info('Generating individual markdown files...');
+        processedDocs = await generateIndividualMarkdownFiles(processedDocs, outDir, siteUrl, context.docsDir, context.options.keepFrontMatter || [], context.options.preserveDirectoryStructure !== false // Default to true
+        );
     }
     // Generate llms.txt
     if (generateLLMsTxt) {
         const llmsTxtPath = path.join(outDir, llmsTxtFilename);
         await generateLLMFile(processedDocs, llmsTxtPath, docTitle, docDescription, false, // links only
-        version, rootContent);
+        version, rootContent, processingBatchSize);
     }
     // Generate llms-full.txt
     if (generateLLMsFullTxt) {
         const llmsFullTxtPath = path.join(outDir, llmsFullTxtFilename);
         await generateLLMFile(processedDocs, llmsFullTxtPath, docTitle, docDescription, true, // full content
-        version, fullRootContent);
+        version, fullRootContent, processingBatchSize);
     }
 }
 /**
@@ -261,13 +339,14 @@ async function generateStandardLLMFiles(context, allDocFiles) {
  */
 async function generateCustomLLMFiles(context, allDocFiles) {
     const { outDir, siteUrl, docTitle, docDescription, options } = context;
-    const { customLLMFiles = [], ignoreFiles = [], generateMarkdownFiles = false } = options;
+    const { customLLMFiles = [], ignoreFiles = [], generateMarkdownFiles = false, processingBatchSize = 100 } = options;
     if (customLLMFiles.length === 0) {
+        utils_1.logger.warn('No custom LLM files configured. Skipping.');
         return;
     }
-    console.log(`Generating ${customLLMFiles.length} custom LLM files...`);
+    utils_1.logger.info(`Generating ${customLLMFiles.length} custom LLM files...`);
     for (const customFile of customLLMFiles) {
-        console.log(`Processing custom file: ${customFile.filename}, version: ${customFile.version || 'undefined'}`);
+        utils_1.logger.verbose(`Processing custom file: ${customFile.filename}, version: ${customFile.version || 'undefined'}`);
         // Combine global ignores with custom ignores
         const combinedIgnores = [...ignoreFiles];
         if (customFile.ignorePatterns) {
@@ -278,19 +357,20 @@ async function generateCustomLLMFiles(context, allDocFiles) {
         if (customDocs.length > 0) {
             // Generate individual markdown files if requested
             if (generateMarkdownFiles) {
-                console.log(`Generating individual markdown files for custom file: ${customFile.filename}...`);
-                customDocs = await generateIndividualMarkdownFiles(customDocs, outDir, siteUrl);
+                utils_1.logger.info(`Generating individual markdown files for custom file: ${customFile.filename}...`);
+                customDocs = await generateIndividualMarkdownFiles(customDocs, outDir, siteUrl, context.docsDir, context.options.keepFrontMatter || [], context.options.preserveDirectoryStructure !== false // Default to true
+                );
             }
             // Use custom title/description or fall back to defaults
             const customTitle = customFile.title || docTitle;
             const customDescription = customFile.description || docDescription;
             // Generate the custom LLM file
             const customFilePath = path.join(outDir, customFile.filename);
-            await generateLLMFile(customDocs, customFilePath, customTitle, customDescription, customFile.fullContent, customFile.version, customFile.rootContent);
-            console.log(`Generated custom LLM file: ${customFile.filename} with ${customDocs.length} documents`);
+            await generateLLMFile(customDocs, customFilePath, customTitle, customDescription, customFile.fullContent, customFile.version, customFile.rootContent, processingBatchSize);
+            utils_1.logger.info(`Generated custom LLM file: ${customFile.filename} with ${customDocs.length} documents`);
         }
         else {
-            console.warn(`No matching documents found for custom LLM file: ${customFile.filename}`);
+            utils_1.logger.warn(`No matching documents found for custom LLM file: ${customFile.filename}`);
         }
     }
 }
@@ -301,18 +381,18 @@ async function generateCustomLLMFiles(context, allDocFiles) {
  */
 async function collectDocFiles(context) {
     const { siteDir, docsDir, options } = context;
-    const { ignoreFiles = [], includeBlog = false } = options;
+    const { ignoreFiles = [], includeBlog = false, warnOnIgnoredFiles = false } = options;
     const allDocFiles = [];
     // Process docs directory
     const fullDocsDir = path.join(siteDir, docsDir);
     try {
         await fs.access(fullDocsDir);
         // Collect all markdown files from docs directory
-        const docFiles = await (0, utils_1.readMarkdownFiles)(fullDocsDir, siteDir, ignoreFiles);
+        const docFiles = await (0, utils_1.readMarkdownFiles)(fullDocsDir, siteDir, ignoreFiles, docsDir, warnOnIgnoredFiles);
         allDocFiles.push(...docFiles);
     }
     catch (err) {
-        console.warn(`Docs directory not found: ${fullDocsDir}`);
+        utils_1.logger.warn(`Docs directory not found: ${fullDocsDir}`);
     }
     // Process blog if enabled
     if (includeBlog) {
@@ -320,11 +400,11 @@ async function collectDocFiles(context) {
         try {
             await fs.access(blogDir);
             // Collect all markdown files from blog directory
-            const blogFiles = await (0, utils_1.readMarkdownFiles)(blogDir, siteDir, ignoreFiles);
+            const blogFiles = await (0, utils_1.readMarkdownFiles)(blogDir, siteDir, ignoreFiles, docsDir, warnOnIgnoredFiles);
             allDocFiles.push(...blogFiles);
         }
         catch (err) {
-            console.warn(`Blog directory not found: ${blogDir}`);
+            utils_1.logger.warn(`Blog directory not found: ${blogDir}`);
         }
     }
     return allDocFiles;