scrape-article

Capture a single article and save it as markdown with frontmatter. Resolve save root per reference/save-location.md (project-local <repo>/captures/articles/YYYY/MM/ preferred, else ~/local-web-capture/articles/YYYY/MM/, --out overrides).

Hard constraint

Requests must originate from this machine. Do not route through a hosted reader (Jina, Firecrawl SaaS, ScrapingBee, etc.).

Escalation ladder — stop at the first rung that returns a real body

All rungs are headless except Tier 3. Prefer speed: do not skip ahead unless sites.yaml marks the domain as known-hard.

Rung 1 — Scrapling Fetcher (static, fastest)

Plain HTTP + lxml. No browser. Use this by default.

from scrapling.fetchers import Fetcher