Skip to content

cristianvasquez/docs-and-graphs

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

83 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

docs-and-graphs

This lib parses markdown into a simplified Abstract Syntax tree.

Why?

Several Note-Taking apps are 'node-based,' 'markdown based,' etc.

I use Markdown, and I cannot get used to outlines. However, I recognize the benefits of having node-based systems, where you can reference a specific node from any other node.

So my question was: How can I have these nodes, and still use Markdown?

After considering this, I realized that Markdown has some structure. It has headers that can be inside other headings and lists that can be inside other lists. These are the nodes this library generates.

Usage

Say you have the following markdown

---
hello: world
---

# Heading 1

Some text under Heading 1

Text that has (inline::variables)

## Inline elements

- Tana and logseq likes
    - embedded nodes

The lib

import { simpleAst } from 'docs-and-graphs'

const json = simpleAst(yourMarkdownString)

will produce the following Json

{
  "type": "root",
  "depth": 0,
  "data": [
    {
      "hello": "world"
    }
  ],
  "children": [
    {
      "type": "block",
      "value": "# Heading 1",
      "depth": 1,
      "children": [
        {
          "type": "text",
          "value": "Some text under Heading 1"
        },
        {
          "type": "text",
          "data": [
            {
              "inline": "variables"
            }
          ],
          "value": "Text that has (inline::variables)"
        },
        {
          "type": "block",
          "value": "## Inline elements",
          "depth": 2,
          "children": [
            {
              "type": "outline",
              "ordered": false,
              "children": [
                {
                  "type": "outline",
                  "value": "Tana and logseq likes "
                },
                {
                  "type": "outline",
                  "ordered": false,
                  "children": [
                    {
                      "type": "outline",
                      "value": "embedded nodes"
                    }
                  ]
                }
              ]
            }
          ]
        }
      ]
    }
  ]
}

Based on

And then?

I use this structure to later produce RDF using a vault-triplifier, but you can use it for whatever you want.