Boo AI Web To PDF Engine

Problem

Some web pages are not single documents. Axure prototypes, HTML slide decks, and single-page apps may expose content only through click navigation. A direct print or full-page screenshot often captures only the first view, leaving no complete PDF for archiving. Boo AI Web To PDF Engine treats these navigable pages as an export problem: capture each page, then compose the images into a document.

How It Works

The core flow uses Playwright to drive headless Chromium, visits the URL, and detects the navigation pattern: dot buttons, sidebar trees, left/right arrows, or keyboard paging. It then clicks or presses keys page by page, waits for rendering, and captures a screenshot for each view. The resulting images are composed into a PDF with img2pdf, avoiding the KeyError: 'JPEG' issue that can occur when saving through Pillow.

Key steps include:
- Structure discovery: locate navigation selectors and determine how pages switch.
- Page capture: handle iframe loading, font loading, and JavaScript rendering waits.
- PDF composition: verify page count, first/last page integrity, and reduce device_scale_factor if the file is too large.

Boundaries

It fits Axure prototypes, dot-navigated slides, and web pages that can be switched page by page. It is not suitable for login-gated pages, infinite-scroll pages, or static pages without navigation. The source also notes that domestic access to Google Fonts can cause font timeout; handling document.fonts and FontFace in an init script may be necessary instead of relying only on network interception.