fix: preload pdfjs worker for server imports

[Talljack]

Summary

• preload pdf.worker.mjs in the Node PDF extraction path
• register WorkerMessageHandler on globalThis.pdfjsWorker so server imports do not rely on runtime relative worker resolution
• keep the existing DOM geometry polyfill path intact

Verification

• pnpm typecheck
• pnpm lint
• pnpm test src/lib/extract-text.test.ts
• pnpm build
• node script: extract /tmp/bbc-office-english.pdf via extractPdf
• local POST /api/import/pdf with the BBC PDF file returns extracted text

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
echo-type	Ready	Preview, Comment	Jun 6, 2026 5:53am