@mizchi/readability

WIP

- [x] Main Content Extraction
- [x] Pure JS Impl (NO DOM API for Cloudflare)
- [ ] Pass original tests
- [ ] Custom Parser (now only supports htmlparser2)
- [x] Accessibility
- [x] Markdown Serializer
- [x] Navigation Detection
- [x] Document Structure Analysis
- [x] Test Coverage Support

CLI

$3

``bash
$ npm install -g @mizchi/readability
`

$3

`bash
$ readability https://example.com/article


Outputs extracted content as markdown


`

$3

`bash
$ readability --help
Usage: @mizchi/readability [options]

Options:
-t, --threshold Character threshold for extraction (default: 250)
-h, --help Show this help message
-f, --format Output format: md, html, doc, ai-summary, ai-structured (default: md)
-o, --out Output file path (default: stdout)

# Progressive Analysis Options
--analyze-structure Analyze page structure without extracting content
--extract-nav Extract navigation information only
--extract-content Extract main content only
--with-context Include context when extracting content
--full-analysis Perform complete page analysis

# Navigation Options
--nav-only Extract navigation information (legacy)
--nav-type Filter navigation by type (global/local/breadcrumb/toc/sidebar/footer)
--nav-location Filter navigation by location (header/main/sidebar/footer)

# Document Mode Options
--doc-mode Extract documentation structure
`

$3

`bash


Extract content as markdown

$ readability https://example.com/article

Extract content with AI-friendly summary

$ readability -f ai-summary https://example.com/article

Analyze page structure

$ readability --analyze-structure https://example.com/article

Extract navigation only

$ readability --extract-nav https://example.com/article

Extract content with context (breadcrumb, surrounding navigation)

$ readability --extract-content --with-context https://example.com/article

Full page analysis (structure + navigation + content)

$ readability --full-analysis https://example.com/article

Documentation mode (for technical documentation sites)

$ readability --doc-mode https://docs.example.com/guide

Save output to file

$ readability -o article.md https://example.com/article
`

$3

- md (default): Markdown format
- html: HTML format
- doc: Documentation structure with navigation
- ai-summary: AI-optimized summary with metadata
- ai-structured: Structured format for AI processing

readability-mcp

`json
{
"mcpServers": {
"readability": {
"command": "npx",
"args": ["-y", "@mizchi/readability", "--mcp"]
}
}
}
`

Prompt

`
Given a URL, use read_url_content_as_markdown and summary contents
`

Development

$3

`bash


Run tests with coverage

npm run test:coverage

Run all tests with coverage (including excluded tests)

npm run test:coverage:all

View coverage report in browser

open coverage/index.html
`

Current coverage thresholds:

- Lines: 50%
- Functions: 60%
- Branches: 75%
- Statements: 50%

How to use

`bash
npm install --save @mizchi/readability
`

$3

`ts
import { extract, toMarkdown } from "@mizchi/readability";

const url = "https://zenn.dev/mizchi/articles/ts-using-sampling-logger";
const html = await fetch(url).then((res) => res.text());
const extracted = extract(html, {
charThreshold: 100,
});

const parsed = toMarkdown(extracted.root);
console.log(parsed);
`

$3

`ts
import { extractAriaTree, ariaTreeToString } from "@mizchi/readability";
const html = await fetch("https://zenn.dev").then((res) => res.text());
const tree = extractAriaTree(html);
const str = ariaTreeToString(tree);
console.log(str);
`

---

Readability.js

A standalone version of the readability library used for
Firefox Reader View.

Installation

Readability is available on npm:

`bash
npm install @mozilla/readability
`

You can then require() it, or for web-based projects, load the
Readability.js script from your webpage.

Basic usage

To parse a document, you must create a new Readability object from a DOM
document object, and then call the parse() method. Here's an
example:

`javascript
var article = new Readability(document).parse();
`

If you use Readability in a web browser, you will likely be able to use a
document reference from elsewhere (e.g. fetched via XMLHttpRequest, in a
same-origin