|
| 1 | +/** |
| 2 | + * @license |
| 3 | + * Copyright Google LLC All Rights Reserved. |
| 4 | + * |
| 5 | + * Use of this source code is governed by an MIT-style license that can be |
| 6 | + * found in the LICENSE file at https://angular.dev/license |
| 7 | + */ |
| 8 | + |
| 9 | +import {globSync, readdirSync, readFileSync, mkdirSync, existsSync, rmSync} from 'node:fs'; |
| 10 | +import {resolve, dirname, join} from 'node:path'; |
| 11 | +import {DatabaseSync} from 'node:sqlite'; |
| 12 | +import {z} from 'zod'; |
| 13 | + |
| 14 | +/** |
| 15 | + * A simple YAML front matter parser. |
| 16 | + * |
| 17 | + * This function extracts the YAML block enclosed by `---` at the beginning of a string |
| 18 | + * and parses it into a JavaScript object. It is not a full YAML parser and only |
| 19 | + * supports simple key-value pairs and string arrays. |
| 20 | + * |
| 21 | + * @param content The string content to parse. |
| 22 | + * @returns A record containing the parsed front matter data. |
| 23 | + */ |
| 24 | +function parseFrontmatter(content) { |
| 25 | + const match = content.match(/^---\r?\n(.*?)\r?\n---/s); |
| 26 | + if (!match) { |
| 27 | + return {}; |
| 28 | + } |
| 29 | + |
| 30 | + const frontmatter = match[1]; |
| 31 | + const data = {}; |
| 32 | + const lines = frontmatter.split(/\r?\n/); |
| 33 | + |
| 34 | + let currentKey = ''; |
| 35 | + let isArray = false; |
| 36 | + const arrayValues = []; |
| 37 | + |
| 38 | + for (const line of lines) { |
| 39 | + const keyValueMatch = line.match(/^([^:]+):\s*(.*)/); |
| 40 | + if (keyValueMatch) { |
| 41 | + if (currentKey && isArray) { |
| 42 | + data[currentKey] = arrayValues.slice(); |
| 43 | + arrayValues.length = 0; |
| 44 | + } |
| 45 | + |
| 46 | + const [, key, value] = keyValueMatch; |
| 47 | + currentKey = key.trim(); |
| 48 | + isArray = value.trim() === ''; |
| 49 | + |
| 50 | + if (!isArray) { |
| 51 | + const trimmedValue = value.trim(); |
| 52 | + if (trimmedValue === 'true') { |
| 53 | + data[currentKey] = true; |
| 54 | + } else if (trimmedValue === 'false') { |
| 55 | + data[currentKey] = false; |
| 56 | + } else { |
| 57 | + data[currentKey] = trimmedValue; |
| 58 | + } |
| 59 | + } |
| 60 | + } else { |
| 61 | + const arrayItemMatch = line.match(/^\s*-\s*(.*)/); |
| 62 | + if (arrayItemMatch && currentKey && isArray) { |
| 63 | + let value = arrayItemMatch[1].trim(); |
| 64 | + // Unquote if the value is quoted. |
| 65 | + if ( |
| 66 | + (value.startsWith("'") && value.endsWith("'")) || |
| 67 | + (value.startsWith('"') && value.endsWith('"')) |
| 68 | + ) { |
| 69 | + value = value.slice(1, -1); |
| 70 | + } |
| 71 | + arrayValues.push(value); |
| 72 | + } |
| 73 | + } |
| 74 | + } |
| 75 | + |
| 76 | + if (currentKey && isArray) { |
| 77 | + data[currentKey] = arrayValues; |
| 78 | + } |
| 79 | + |
| 80 | + return data; |
| 81 | +} |
| 82 | + |
| 83 | +export function generate(inPath, outPath) { |
| 84 | + const dbPath = outPath; |
| 85 | + mkdirSync(dirname(outPath), {recursive: true}); |
| 86 | + |
| 87 | + if (existsSync(dbPath)) { |
| 88 | + rmSync(dbPath); |
| 89 | + } |
| 90 | + const db = new DatabaseSync(dbPath); |
| 91 | + |
| 92 | + // Create a table to store metadata. |
| 93 | + db.exec(` |
| 94 | + CREATE TABLE metadata ( |
| 95 | + key TEXT PRIMARY KEY NOT NULL, |
| 96 | + value TEXT NOT NULL |
| 97 | + ); |
| 98 | + `); |
| 99 | + |
| 100 | + db.exec(` |
| 101 | + INSERT INTO metadata (key, value) VALUES |
| 102 | + ('schema_version', '1'), |
| 103 | + ('created_at', '${new Date().toISOString()}'); |
| 104 | + `); |
| 105 | + |
| 106 | + // Create a relational table to store the structured example data. |
| 107 | + db.exec(` |
| 108 | + CREATE TABLE examples ( |
| 109 | + id INTEGER PRIMARY KEY, |
| 110 | + title TEXT NOT NULL, |
| 111 | + summary TEXT NOT NULL, |
| 112 | + keywords TEXT, |
| 113 | + required_packages TEXT, |
| 114 | + related_concepts TEXT, |
| 115 | + related_tools TEXT, |
| 116 | + experimental INTEGER NOT NULL DEFAULT 0, |
| 117 | + content TEXT NOT NULL |
| 118 | + ); |
| 119 | + `); |
| 120 | + |
| 121 | + // Create an FTS5 virtual table to provide full-text search capabilities. |
| 122 | + db.exec(` |
| 123 | + CREATE VIRTUAL TABLE examples_fts USING fts5( |
| 124 | + title, |
| 125 | + summary, |
| 126 | + keywords, |
| 127 | + required_packages, |
| 128 | + related_concepts, |
| 129 | + related_tools, |
| 130 | + content, |
| 131 | + content='examples', |
| 132 | + content_rowid='id', |
| 133 | + tokenize = 'porter ascii' |
| 134 | + ); |
| 135 | + `); |
| 136 | + |
| 137 | + // Create triggers to keep the FTS table synchronized with the examples table. |
| 138 | + db.exec(` |
| 139 | + CREATE TRIGGER examples_after_insert AFTER INSERT ON examples BEGIN |
| 140 | + INSERT INTO examples_fts( |
| 141 | + rowid, title, summary, keywords, required_packages, related_concepts, related_tools, |
| 142 | + content |
| 143 | + ) |
| 144 | + VALUES ( |
| 145 | + new.id, new.title, new.summary, new.keywords, new.required_packages, |
| 146 | + new.related_concepts, new.related_tools, new.content |
| 147 | + ); |
| 148 | + END; |
| 149 | + `); |
| 150 | + |
| 151 | + const insertStatement = db.prepare( |
| 152 | + 'INSERT INTO examples(' + |
| 153 | + 'title, summary, keywords, required_packages, related_concepts, related_tools, experimental, content' + |
| 154 | + ') VALUES(?, ?, ?, ?, ?, ?, ?, ?);', |
| 155 | + ); |
| 156 | + |
| 157 | + const frontmatterSchema = z.object({ |
| 158 | + title: z.string(), |
| 159 | + summary: z.string(), |
| 160 | + keywords: z.array(z.string()).optional(), |
| 161 | + required_packages: z.array(z.string()).optional(), |
| 162 | + related_concepts: z.array(z.string()).optional(), |
| 163 | + related_tools: z.array(z.string()).optional(), |
| 164 | + experimental: z.boolean().optional(), |
| 165 | + }); |
| 166 | + |
| 167 | + db.exec('BEGIN TRANSACTION'); |
| 168 | + const entries = globSync |
| 169 | + ? globSync('**/*.md', {cwd: resolve(inPath), withFileTypes: true}) |
| 170 | + : readdirSync(resolve(inPath), {withFileTypes: true}); |
| 171 | + for (const entry of entries) { |
| 172 | + if (!entry.isFile() || !entry.name.endsWith('.md')) { |
| 173 | + continue; |
| 174 | + } |
| 175 | + |
| 176 | + const content = readFileSync(join(entry.parentPath, entry.name), 'utf-8'); |
| 177 | + const frontmatter = parseFrontmatter(content); |
| 178 | + |
| 179 | + const validation = frontmatterSchema.safeParse(frontmatter); |
| 180 | + if (!validation.success) { |
| 181 | + console.error(`Validation failed for example file: ${entry.name}`); |
| 182 | + console.error('Issues:', validation.error.issues); |
| 183 | + throw new Error(`Invalid front matter in ${entry.name}`); |
| 184 | + } |
| 185 | + |
| 186 | + const { |
| 187 | + title, |
| 188 | + summary, |
| 189 | + keywords, |
| 190 | + required_packages, |
| 191 | + related_concepts, |
| 192 | + related_tools, |
| 193 | + experimental, |
| 194 | + } = validation.data; |
| 195 | + insertStatement.run( |
| 196 | + title, |
| 197 | + summary, |
| 198 | + JSON.stringify(keywords ?? []), |
| 199 | + JSON.stringify(required_packages ?? []), |
| 200 | + JSON.stringify(related_concepts ?? []), |
| 201 | + JSON.stringify(related_tools ?? []), |
| 202 | + experimental ? 1 : 0, |
| 203 | + content, |
| 204 | + ); |
| 205 | + } |
| 206 | + db.exec('END TRANSACTION'); |
| 207 | + |
| 208 | + db.close(); |
| 209 | +} |
| 210 | + |
| 211 | +function main() { |
| 212 | + const argv = process.argv.slice(2); |
| 213 | + if (argv.length !== 2) { |
| 214 | + console.error('Must include 2 arguments.'); |
| 215 | + process.exit(1); |
| 216 | + } |
| 217 | + |
| 218 | + const [inPath, outPath] = argv; |
| 219 | + |
| 220 | + try { |
| 221 | + generate(inPath, outPath); |
| 222 | + } catch (error) { |
| 223 | + console.error('An error happened:'); |
| 224 | + console.error(error); |
| 225 | + process.exit(127); |
| 226 | + } |
| 227 | +} |
| 228 | + |
| 229 | +main(); |
0 commit comments