COHERAPortfolioUniverseImportPipeline
Internal · operator console

Document pipeline

Validate and fetch registered annual reports in small batches. Validation resolves redirects and classifies each URL; fetch retrieves the bytes, confirms the document format, hashes the content and stores it in a private bucket. Nothing is parsed and no analysis is run at this stage.

Operator token

The app has no end-user authentication yet, so both the queue view and the fetch action require a server-side operator token. Requests without it are rejected on the server, which keeps the fetcher from becoming a public URL-fetching endpoint.

Where this sits in the pipeline
  1. Discover — spreadsheet bootstrap live; issuer/exchange feeds not built. Registry fields exist.
  2. Validate — live on this page.
  3. Fetch — live on this page; bytes stored privately.
  4. Parse — live on this page. Fetched PDF/XHTML files become structured units with page numbers, headings and anchors. No OCR, no model calls.
  5. Analyze / Re-evaluate hypotheses — unchanged, still driven by curated data.