|
| 1 | +import fs from 'node:fs'; |
| 2 | +import path from 'node:path'; |
| 3 | +import util from 'node:util'; |
| 4 | +import versions from '../../versions.json'; |
| 5 | + |
| 6 | +/** |
| 7 | + * Parses frontmatter from markdown content. |
| 8 | + * Extracts metadata like title and description, and returns the cleaned content. |
| 9 | + * |
| 10 | + * @param {string} fileContent - Raw markdown file content. |
| 11 | + * @returns {{data: Object, content: string}} - Parsed data and stripped content. |
| 12 | + */ |
| 13 | +function parseFrontMatter(fileContent) { |
| 14 | + const frontMatterRegex = /^---\n([\s\S]+?)\n---\n/; |
| 15 | + const match = fileContent.match(frontMatterRegex); |
| 16 | + |
| 17 | + if (!match) { |
| 18 | + return { data: {}, content: fileContent }; |
| 19 | + } |
| 20 | + |
| 21 | + const frontMatterBlock = match[1]; |
| 22 | + const content = fileContent.replace(frontMatterRegex, ''); |
| 23 | + |
| 24 | + const data = {}; |
| 25 | + |
| 26 | + frontMatterBlock.split('\n').forEach((line) => { |
| 27 | + const parts = line.split(':'); |
| 28 | + |
| 29 | + if (parts.length >= 2) { |
| 30 | + const key = parts[0].trim(); |
| 31 | + |
| 32 | + let value = parts.slice(1).join(':').trim(); |
| 33 | + |
| 34 | + // Remove surrounding quotes if present |
| 35 | + if ( |
| 36 | + (value.startsWith("'") && value.endsWith("'")) || |
| 37 | + (value.startsWith('"') && value.endsWith('"')) |
| 38 | + ) { |
| 39 | + value = value.slice(1, -1); |
| 40 | + } |
| 41 | + |
| 42 | + data[key] = value; |
| 43 | + } |
| 44 | + }); |
| 45 | + |
| 46 | + return { data, content }; |
| 47 | +} |
| 48 | + |
| 49 | +/** |
| 50 | + * Recursively processes sidebar items to generate the LLM index and collect full docs. |
| 51 | + * Handles different sidebar structures (flat arrays, categories). |
| 52 | + * |
| 53 | + * @param {Array} items - Sidebar items to process. |
| 54 | + * @param {string} docsPath - Base path to documentation files. |
| 55 | + * @param {string} version - Current version being processed (e.g., '7.x'). |
| 56 | + * @param {boolean} isLatest - Whether this is the latest/stable version. |
| 57 | + * @param {string} baseUrl - The base URL of the website. |
| 58 | + * @param {number} level - Current nesting level for headings. |
| 59 | + * @returns {{content: string, docs: Array}} - Generated index content and list of doc objects. |
| 60 | + */ |
| 61 | +function processSidebar( |
| 62 | + items, |
| 63 | + docsPath, |
| 64 | + version, |
| 65 | + isLatest, |
| 66 | + baseUrl, |
| 67 | + level = 0 |
| 68 | +) { |
| 69 | + let llmsContent = ''; |
| 70 | + let fullDocsList = []; |
| 71 | + |
| 72 | + items.forEach((item) => { |
| 73 | + // Case 1: Item is a direct link (string ID) |
| 74 | + if (typeof item === 'string') { |
| 75 | + const id = item; |
| 76 | + const filePath = path.join(docsPath, `${id}.md`); |
| 77 | + |
| 78 | + if (fs.existsSync(filePath)) { |
| 79 | + const fileContent = fs.readFileSync(filePath, 'utf8'); |
| 80 | + const { data, content } = parseFrontMatter(fileContent); |
| 81 | + |
| 82 | + const title = data.title || id; |
| 83 | + const description = data.description || ''; |
| 84 | + |
| 85 | + // Construct the public URL for the document |
| 86 | + // Latest version uses /docs/id, others use /docs/version/id |
| 87 | + const urlPath = isLatest ? `/docs/${id}` : `/docs/${version}/${id}`; |
| 88 | + const fullUrl = `${baseUrl}${urlPath}`; |
| 89 | + |
| 90 | + llmsContent += `- [${title}](${fullUrl})${ |
| 91 | + description ? `: ${description}` : '' |
| 92 | + }\n`; |
| 93 | + |
| 94 | + fullDocsList.push({ |
| 95 | + title, |
| 96 | + url: fullUrl, |
| 97 | + content, |
| 98 | + }); |
| 99 | + } |
| 100 | + } |
| 101 | + // Case 2: Item is a category object |
| 102 | + else if (item.type === 'category') { |
| 103 | + const label = item.label; |
| 104 | + const headingPrefix = '#'.repeat(level + 3); // Start at level 3 (###) |
| 105 | + |
| 106 | + llmsContent += `\n${headingPrefix} ${label}\n\n`; |
| 107 | + |
| 108 | + const { content: childContent, docs: childDocs } = processSidebar( |
| 109 | + item.items, |
| 110 | + docsPath, |
| 111 | + version, |
| 112 | + isLatest, |
| 113 | + baseUrl, |
| 114 | + level + 1 |
| 115 | + ); |
| 116 | + |
| 117 | + llmsContent += childContent; |
| 118 | + fullDocsList = fullDocsList.concat(childDocs); |
| 119 | + } |
| 120 | + }); |
| 121 | + |
| 122 | + return { content: llmsContent, docs: fullDocsList }; |
| 123 | +} |
| 124 | + |
| 125 | +/** |
| 126 | + * Generates the llms.txt and llms-full.txt files for a specific version. |
| 127 | + * |
| 128 | + * @returns {Array<string>} - List of generated filenames. |
| 129 | + */ |
| 130 | +function generateForVersion( |
| 131 | + siteDir, |
| 132 | + outDir, |
| 133 | + version, |
| 134 | + outputPrefix, |
| 135 | + isLatest, |
| 136 | + baseUrl |
| 137 | +) { |
| 138 | + const docsPath = path.join(siteDir, 'versioned_docs', `version-${version}`); |
| 139 | + const sidebarPath = path.join( |
| 140 | + siteDir, |
| 141 | + 'versioned_sidebars', |
| 142 | + `version-${version}-sidebars.json` |
| 143 | + ); |
| 144 | + |
| 145 | + if (!fs.existsSync(sidebarPath)) { |
| 146 | + // Silent return if sidebar is missing to avoid noise |
| 147 | + return []; |
| 148 | + } |
| 149 | + |
| 150 | + const sidebarConfig = JSON.parse(fs.readFileSync(sidebarPath, 'utf8')); |
| 151 | + |
| 152 | + // Handle different Docusaurus sidebar structures (root 'docs' key or first key) |
| 153 | + let rootItems = sidebarConfig.docs || Object.values(sidebarConfig)[0] || []; |
| 154 | + |
| 155 | + // Normalize object-style sidebars (older Docusaurus versions) to array format |
| 156 | + if (!Array.isArray(rootItems) && typeof rootItems === 'object') { |
| 157 | + const normalized = []; |
| 158 | + |
| 159 | + for (const [label, items] of Object.entries(rootItems)) { |
| 160 | + normalized.push({ |
| 161 | + type: 'category', |
| 162 | + label: label, |
| 163 | + items: items, |
| 164 | + }); |
| 165 | + } |
| 166 | + |
| 167 | + rootItems = normalized; |
| 168 | + } |
| 169 | + |
| 170 | + const { content: sidebarContent, docs } = processSidebar( |
| 171 | + rootItems, |
| 172 | + docsPath, |
| 173 | + version, |
| 174 | + isLatest, |
| 175 | + baseUrl |
| 176 | + ); |
| 177 | + |
| 178 | + // 1. Generate Summary (llms.txt) |
| 179 | + let llmsTxt = `# React Navigation ${version}\n\n`; |
| 180 | + |
| 181 | + llmsTxt += `> Routing and navigation for your React Native apps.\n\n`; |
| 182 | + llmsTxt += `## Documentation\n`; |
| 183 | + llmsTxt += sidebarContent; |
| 184 | + |
| 185 | + const summaryFilename = `${outputPrefix}.txt`; |
| 186 | + |
| 187 | + fs.writeFileSync(path.join(outDir, summaryFilename), llmsTxt); |
| 188 | + |
| 189 | + // 2. Generate Full Content (llms-full.txt) |
| 190 | + let llmsFullTxt = `# React Navigation ${version} Documentation\n\n`; |
| 191 | + |
| 192 | + docs.forEach((doc) => { |
| 193 | + llmsFullTxt += `## ${doc.title}\n\n`; |
| 194 | + llmsFullTxt += `Source: ${doc.url}\n\n`; |
| 195 | + llmsFullTxt += `${doc.content.trim()}\n\n---\n\n`; |
| 196 | + }); |
| 197 | + |
| 198 | + // Determine full filename (e.g., llms-v8.x -> llms-full-v8.x.txt) |
| 199 | + let fullFilename; |
| 200 | + |
| 201 | + if (outputPrefix === 'llms') { |
| 202 | + fullFilename = 'llms-full.txt'; |
| 203 | + } else { |
| 204 | + if (outputPrefix.includes('llms-')) { |
| 205 | + fullFilename = outputPrefix.replace('llms-', 'llms-full-') + '.txt'; |
| 206 | + } else { |
| 207 | + fullFilename = outputPrefix + '-full.txt'; |
| 208 | + } |
| 209 | + } |
| 210 | + |
| 211 | + fs.writeFileSync(path.join(outDir, fullFilename), llmsFullTxt); |
| 212 | + |
| 213 | + return [summaryFilename, fullFilename]; |
| 214 | +} |
| 215 | + |
| 216 | +export default function (context, options) { |
| 217 | + return { |
| 218 | + name: 'llms.txt', |
| 219 | + async postBuild({ siteDir, outDir }) { |
| 220 | + const { latestVersion, baseUrl } = options; |
| 221 | + |
| 222 | + if (!latestVersion) { |
| 223 | + throw new Error('[llms.txt] "latestVersion" option is required.'); |
| 224 | + } |
| 225 | + |
| 226 | + if (!baseUrl) { |
| 227 | + throw new Error('[llms.txt] "baseUrl" option is required.'); |
| 228 | + } |
| 229 | + |
| 230 | + const generatedFiles = []; |
| 231 | + |
| 232 | + versions.forEach((version) => { |
| 233 | + const isLatest = version === latestVersion; |
| 234 | + // Prefix: 'llms' for latest, 'llms-vX.x' for others |
| 235 | + const outputPrefix = isLatest ? 'llms' : `llms-v${version}`; |
| 236 | + |
| 237 | + const files = generateForVersion( |
| 238 | + siteDir, |
| 239 | + outDir, |
| 240 | + version, |
| 241 | + outputPrefix, |
| 242 | + isLatest, |
| 243 | + baseUrl |
| 244 | + ); |
| 245 | + |
| 246 | + generatedFiles.push(...files); |
| 247 | + }); |
| 248 | + |
| 249 | + console.log( |
| 250 | + `${util.styleText(['magenta', 'bold'], '[llms.txt]')} Wrote ${generatedFiles.length} files in build output.` |
| 251 | + ); |
| 252 | + }, |
| 253 | + }; |
| 254 | +} |
0 commit comments