reread-markdown v1.0.0
reread-markdown
To turn markdown files into structured, queryable data. The input Markdown file is transformed into a special tree exposing the structure intended by the author of the file: the table of contents (TOC). For each key (heading) in the TOC, you get both the title and the related contents under this title. You don't only get titles. Then, you can easily extract whole parts of the original document.
Markdown isn’t just text. It's a source of structured information.
Heading nodes are the primary focus of this library. They are retrieved, enriched and organized to form a new tree. The input document is crawled and indexed by headings. The other components of the input Markdown file are seen as heading attributes. Output heading nodes have two main attributes: their title and the related contents (raw text or tokens).
Agnostic. Versatile. Generic.
Examples in the Node environment
example.md
This is the *introduction*.
# Part One
Bla bla.
## *Title*
1, 2, 3.
### Chapter
A, B, C.
###### Section
i, ii, iii.
A code example with `#` characters..
```python
# This is a comment
def reread():
print("Detect patterns.")
```
# **Part** *Two*
## Another Title
Last but not least.Code example
import { readFileSync } from 'node:fs';
import {reread} from "reread-markdown"
const text = readFileSync("./example.md", "utf8")
console.log(reread(text).toJSON())Output
{
"data": {
"title": "",
"depth": 0,
"contents": "\nThis is the *introduction*.\n\n"
},
"children": [
{
"data": {
"title": "# Part One",
"depth": 1,
"contents": "\n\nBla bla.\n\n"
},
"children": [
{
"data": {
"title": "## *Title*",
"depth": 2,
"contents": "\n\n1, 2, 3.\n\n"
},
"children": [
{
"data": {
"title": "### Chapter",
"depth": 3,
"contents": "\n\nA, B, C.\n\n"
},
"children": [
{
"data": {
"title": "###### Section",
"depth": 6,
"contents": "\n\ni, ii, iii.\n\nA code example with `#` characters..\n\n```python\n\n# This is a comment\ndef reread():\n print(\"Detect patterns.\")\n\n```\n\n"
},
"children": []
}
]
}
]
}
]
},
{
"data": {
"title": "# **Part** *Two*",
"depth": 1,
"contents": "\n"
},
"children": [
{
"data": {
"title": "## Another Title",
"depth": 2,
"contents": "\n\nLast but not least.\n"
},
"children": []
}
]
}
]
}Workflow
- mdast-util-from-markdown is used for parsing the input Markdown text. The resulting syntax tree complies with the specification defined by mdast.
- This syntax tree is visited (walked, traversed, crawled) by a Visitor object accumulating state in order to produce a new tree.
- This new tree allows you to reread Markdown (AST transformation). It exposes the hierarchical structure of the document and enables you to extract parts for further processing.
- You can optionally retrieve each part in the form of a list of tokens. Tokens are labelled fragments of text (strings of characters). They are suitable for natural language processing (NLP) or PDF generation.
Related Projects
There is a similar service in Visual Studio Code (range folding).
- https://github.com/microsoft/vscode-markdown-languageservice
- https://github.com/microsoft/vscode-markdown-languageservice/blob/main/src/tableOfContents.ts
VS Code uses an alternative technique for getting the same result. No tree traversal. VS Code uses markdown-it. Headers are first extracted with location and depth information. The resulting list of size n is scanned n times for extracting the hierarchical information...
About Markdown
3 months ago