Notion to Markdown | PDF | Docx

(Stack: Javascript, NodeJS, Notion API, Makefile, Pandoc)

Minimal example to

• fetch a set of Notion pages (from a database table) (via official Notion API + JS client)
• for each Notion page
  • (with notion-to-md) parse page as markdown (with additional metadata parsing)
  • output markdown as file (with fs) to output/*
• with pandoc convert each page to PDF & Docx in output/*

OS-Level dependencies

installation of these dependencies is OS-specific, so it won't be covered by npm install

• pandoc
• pandoc needs pdflatex or a different PDF engine (specify with --pdf-engine)
  • FWIW, tinytex is a minimal solution
• puppeteer on Ubuntu depends on libgbm.so.1, which might be absent
  • run sudo apt-get install -y libgbm-dev

Quickstart

Heads up: Locally, for each page a meta.json and a <page>.json file are produced. They are not part of the repository via .gitignore. These files store raw Notion data and leak the page IDs among other things. Please consider this before you put your outputs in a public repo.

Notion Source DB: https://fubits.notion.site/DB-Multiple-Pages-1890d20f20d04793a50bdaec3bd8200d

Setup

Create Notion Database Table and add a few pages or create a single Notion page (see .env file). Share the db/page with your integration account.

Fetching & Parsing to MD

add to local .env file:

NOTION_API_KEY=<your-api-token>
SINGLE_PAGE=<your-page-id>
DB_TABLE=<your-database-table-id>

To parse a single page (SINGLE_PAGE, without further processing) run:

npm run export-md-single

To parse all pages in a database (DB_TABLE) table run:

npm run export-md-all

Pandoc: PDF & MS Word Outputs

Pandoc needs a PDF processing library (i.e. texlive on Ubuntu); Setup is OS-specific.

To convert the single page .md file (path is hardcoded in the makefile) to PDF & Docx run:

make bake-single

To convert all .md files in output/* to PDF & Docx run:

make bake-all

The PDF make task now demonstrates both, pandoc and pagedjs/pagedjs-cli to produce the PDFs. pagedjs renders the PDF from html (+css).

TODO

• add parsing step to download external linked resources (images, videos, etc.); download files and set relative links
• make Node's exec(makefile) run with relative filepaths
• use pagedjs-cli as PDF engine for Pandoc