blockly/scripts/gulpfiles/docs_tasks.js

const {execSync} = require('child_process');
const {Extractor} = require('markdown-tables-to-json');
const fs = require('fs');
const gulp = require('gulp');
const header = require('gulp-header');
const replace = require('gulp-replace');

const DOCS_DIR = 'docs';

/**
 * Run API Extractor to generate the intermediate json file.
 */
const generateApiJson = function(done) {
  execSync('api-extractor run --local', {stdio: 'inherit'});
  done();
}

/**
 * API Extractor output spuriously renames some functions. Undo that.
 * See https://github.com/microsoft/rushstack/issues/2534
 */
const removeRenames = function() {
  return gulp.src('temp/blockly.api.json')
      .pipe(replace('_2', ''))
      .pipe(gulp.dest('temp/'));
}

/**
 * Run API Documenter to generate the raw docs files.
 */
const generateDocs = function(done) {
  if (!fs.existsSync(DOCS_DIR)) {
    // Create the directory if it doesn't exist.
    // If it already exists, the contents will be deleted by api-documenter.
    fs.mkdirSync(DOCS_DIR);
  }
  execSync(
      `api-documenter markdown --input-folder temp --output-folder ${DOCS_DIR}`,
      {stdio: 'inherit'});
  done();
}

/**
 * Prepends the project and book metadata that devsite requires.
 */
const prependBook = function() {
  return gulp.src('docs/*.md')
      .pipe(header(
          'Project: /blockly/_project.yaml\nBook: /blockly/_book.yaml\n\n'))
      .pipe(gulp.dest(DOCS_DIR));
}

/**
 * Creates a map of top-level pages to sub-pages, e.g. a mapping
 * of `block_class` to every page associated with that class.
 * This is needed to create an accurate table of contents.
 * @param {string[]} allFiles All files in docs directory.
 * @returns {Map<string, string[]>}
 */
const buildAlternatePathsMap = function(allFiles) {
  let map = new Map();
  for (let file of allFiles) {
    // Get the name of the class/namespaces/variable/etc., i.e. the top-level
    // page.
    let filePieces = file.split('.');
    let name = filePieces[1];
    if (!map.has(name)) {
      map.set(name, []);
    }
    if (filePieces[2] === 'md') {
      // Don't add the top-level page to the map.
      continue;
    }
    // Add all sub-pages to the array for the corresponding top-level page.
    map.get(name).push(file);
  }
  return map;
}

/**
 * Create the _toc.yaml file used by devsite to create the leftnav.
 * This file is generated from the contents of `blockly.md` which contains links
 * to the other top-level API pages (each class, namespace, etc.).
 *
 * The `alternate_paths` for each top-level page contains the path for
 * each associated sub-page. All subpages must be linked to their top-level page
 * in the TOC for the left nav bar to remain correct after drilling down into a
 * sub-page.
 */
const createToc = function(done) {
  const fileContent = fs.readFileSync(`${DOCS_DIR}/blockly.md`, 'utf8');
  // Create the TOC file. The file should not yet exist; if it does, this
  // operation will fail.
  const toc = fs.openSync(`${DOCS_DIR}/_toc.yaml`, 'ax');
  const files = fs.readdirSync(DOCS_DIR);
  const map = buildAlternatePathsMap(files);
  const referencePath = '/blockly/reference/js';

  const tocHeader = `toc:
- title: Overview
  path: /blockly/reference/js/blockly.md\n`;
  fs.writeSync(toc, tocHeader);

  // Generate a section of TOC for each section/heading in the overview file.
  const sections = fileContent.split('##');
  for (let section of sections) {
    // This converts the md table in each section to a JS object
    const table = Extractor.extractObject(section, 'rows', false);
    if (!table) {
      continue;
    }
    // Get the name of the section, i.e. the text immediately after the `##` in
    // the source doc
    const sectionName = section.split('\n')[0].trim();
    fs.writeSync(toc, `- heading: ${sectionName}\n`);
    for (let row in table) {
      // After going through the Extractor, the markdown is now HTML.
      // Each row in the table is now a link (anchor tag).
      // Get the target of the link, excluding the first `.` since we don't want
      // a relative path.
      const path = /href="\.(.*?)"/.exec(row)?.[1];
      // Get the name of the link (text in between the <a> and </a>)
      const name = /">(.*?)</.exec(row)?.[1];
      if (!path || !name) {
        continue;
      }
      fs.writeSync(toc, `- title: ${name}\n  path: ${referencePath}${path}\n`);
      // Get the list of sub-pages for this page.
      // Add each sub-page to the `alternate_paths` property.
      let pages = map.get(path.split('.')[1]);
      if (pages?.length) {
        fs.writeSync(toc, `  alternate_paths:\n`);
        for (let page of pages) {
          fs.writeSync(toc, `  - ${referencePath}/${page}\n`);
        }
      }
    }
  }

  done();
}

const docs = gulp.series(
    generateApiJson, removeRenames, generateDocs,
    gulp.parallel(prependBook, createToc));

module.exports = {docs};