@tinyweb_dev/doc-indexer-pdf

PDF source adapter for @tinyweb_dev/doc-indexer.

MVP:

• Extracts text page-by-page via pdfjs-dist (legacy build, no DOM required).
• Emits one Chunk per page + a 2-level TreeNode skeleton (document → pages[]).
• Detects application/pdf by extension or sniff.

Roadmap (deferred):

• Heading detection from font-size statistics → multi-level outline tree.
• Page-image render to PNG via pdfjs-dist + canvas (asset emission).
• OCR fallback (tesseract.js) for scanned PDFs with no text layer.
• Vision-LLM triage for image-only PDFs.

Usage

import { Indexer, FsStorage } from '@tinyweb_dev/doc-indexer-core';
import { PdfAdapter } from '@tinyweb_dev/doc-indexer-pdf';

const indexer = new Indexer({
  adapters: [new PdfAdapter()],
  storage: new FsStorage('./out'),
});
const doc = await indexer.index({ path: './sample.pdf' });
console.log(doc.stats);