Power Query M Reference

This repository hosts the automated scripts designed for constructing an up-to-date Markdown version of the Power Query M reference documentation. The source of this documentation is the #shared structure within the Power BI Desktop application, which is extracted and published to a Docusaurus site for improved accessibility and user experience. The documentation is readily accessible at https://powerquery.io/.

Features

The new site is equipped with a range of features aimed at enhancing the usability and accessibility of the Power Query M documentation. These features include:

• Comprehensive Locale Support: The site supports the majority of locales offered by Power BI Desktop, ensuring that users across the globe can access the documentation in their preferred language.

• Automated Updates via GitHub Actions: The documentation is kept up-to-date through an automated workflow powered by GitHub Actions. This automation involves extracting the latest #shared structure from Power BI Desktop, converting it into Markdown format, and publishing it to the Docusaurus site.

• Easy Navigation and Search Functionality: Designed with user experience in mind, the site features intuitive navigation and advanced search capabilities, supported by Algolia.

• Responsive Design: The Docusaurus site is designed to be fully responsive, ensuring that the documentation is accessible and readable on a wide range of devices, from desktop computers to smartphones and tablets.

• Dark Mode: Understanding the diverse preferences of our users, the site features a Dark Mode option, allowing for a more comfortable viewing experience in low-light environments.

Notable Scripts

The repository follows the structure of a Docusaurus3 site with addition of the following scripts:

• The /process folder hosts scripts for generation of Markdown files based of JSON data.
• The /.github folder contains GitHub Actions configurations, automating documentation refreshes to ensure content remains up-to-date.
• /process/shared/{locale}.json PowerQuery documentation in JSON format.
• /.github/scripts/generate-doc-output.ps1 generaties documentation in JSON format for specific locales, based on the /process/dox.pbix file.
• /process/shared/_taxonomy.json contains the taxonomy of Power Query functions .
• /process/taxonomyGaps.js, a Node.js script which identifies any functions lacking categorization.
• /process/functionDocTemplate.ejs is he EJS template that describes the format for generating markdown documentation.
• /process/generateDocs.js is a Node.js script that processes the JSON documentation and outputs the final Markdown files, completing the documentation pipeline. English (en) is written to /docs; every other locale is written to /i18n/{locale}/docusaurus-plugin-content-docs/current/. HTML embedded in the source strings (<code>, <ul>/<li>, <table>, …) is converted to Markdown with Turndown, and MDX-unsafe characters ({, }, <) are escaped.

Usage

node process/generateDocs.js                 # every locale found under /i18n
node process/generateDocs.js de fr ja        # only the named locales
node process/generateDocs.js en              # English → /docs
node process/generateDocs.js --clean en      # English, removing stale .md first
node process/generateDocs.js --trim          # all i18n locales, with trimming

Argument	Description
<locale>…	One or more locale codes to generate (e.g. de fr ja). Use en to (re)generate English into /docs. When no locale is given, every locale under /i18n is processed.
--all	Process every locale found under /i18n, in addition to any locales passed explicitly.
--en	Also (re)generate English into /docs.
--clean	Delete previously generated .md files in the target before writing, so renamed/removed functions don't linger as stale orphans. readme.md, _category_.json and hand-authored .mdx overrides are preserved.
--trim	After generating a locale, delete files whose content is byte-identical to the English version — Docusaurus falls back to the default locale for the missing entries (replaces the old trimLocaleDocs.ps1).
--preserve-case	Keep the original casing of output paths instead of lowercasing them.

A typical end-to-end refresh after the JSON has been regenerated:

node process/generateDocs.js --clean en      # 1. English first
node process/generateDocs.js --trim          # 2. all i18n locales, trimmed against English

The script depends on ejs, turndown and turndown-plugin-gfm (declared in devDependencies); run npm install first. Functions with a hand-authored .mdx page are skipped so manual pages are never overwritten.