Extraction Flow

1. Load The Source

OAPIE accepts either:

• a remote http or https URL
• a local HTML file path

Remote sources are loaded through one of the supported engines:

• axios
• happy-dom
• jsdom
• puppeteer

2. Extract Operation Data

The extractor reads the loaded HTML and looks for request and response details commonly found on ReadMe-style API docs pages, including:

• HTTP method and URL
• parameter forms
• request examples
• response examples
• sidebar links for crawl mode

For sites that hydrate client-side, the Puppeteer path waits for extractable content and preserves SSR payloads when the rendered DOM omits script#ssr-props.

3. Crawl Linked Pages

When --crawl is enabled, OAPIE:

• resolves sidebar links from the root page
• reuses a shared browser session when appropriate
• fetches each discovered page
• extracts operations and attaches the source URL for each page

4. Normalize The Result

Two output shapes are available:

• raw: the direct extracted payload
• openapi: an OpenAPI-like document built from the extracted operations

5. Write Output

Output is always written to the local output/ directory. The file name is derived from the source and selected output shape.