Local and archived document input

extract and analyze accept either a network URL or a local source. Local input must have an absolute HTTP(S) base URL so relative links are resolved exactly as they were on the archived site. Local bytes bypass robots checks, outbound policy, and the URL content cache.

newsworker extract --file archive/page.html \
  --base-url https://example.com/news/ --format json

curl --silent https://example.com/news/ \
  | newsworker analyze --file - --base-url https://example.com/news/ \
      --output page-spec.yaml

The library equivalent is DocumentSource.file(...), DocumentSource.stdin(...), or DocumentSource.url(...), followed by FeedService.get_feed_source(source). Empty input, non-regular files, missing base URLs, and non-HTTP(S) base URLs fail before extraction.