[NaitikOza]

Summary

Adopts the unfinished CDC implementation from #7 on the current shared scraping registry and implements the maintainer's request to use CDC's HTML instead of Selenium-driven PDF downloads and GROBID.

• Discover records from the CDC Stacks Guidelines and Recommendations collection using HTTP + lxml.
• Select English, public-domain records with an official CDC MMWR full-text HTML DOI.
• Follow the DOI to cdc.gov and convert only the semantic syndicate article body to markdown.
• Skip PDF-only, translated, restricted, and external-publisher records instead of treating Stacks abstracts as full guidelines.
• Preserve Stacks/DOI provenance, authors, corporate authors, publication type/year, license, and content_scope: full_html.
• Register --source cdc, support explicit Stacks record URLs, document usage, and honor Stacks' 10-second listing crawl delay.
• Use an identifiable AMFV user agent because Stacks rejects the shared synthetic Chrome user agent.

No browser, Selenium, GROBID, OCR, or new dependency is required.

Scope and provenance

The original #7 collection target and record-discovery approach are credited in the module header. The implementation commit includes Co-authored-by: azka2001 <azkaah2001@gmail.com>. The original PR remains open for its review history.

The content boundary is deliberately conservative. CDC Stacks currently mixes complete CDC publications, PDF-only records, translations, and third-party publications. This scraper emits a document only when it can verify and extract an official full-text CDC HTML body.

Validation

• uv run ruff check .
• uv run ruff format --check .
• uv run pytest -q — 61 passed
• uv build --package amfv-datasets
• Live listing smoke test: uv run amfv-scrape --source cdc --documents 1 --no-progress
• Live URL smoke test: Stacks record cdc:182510 resolved to the official MMWR article and produced 49,013 characters across 8 sections with full provenance