Skip to content

v1.3.0

Choose a tag to compare

@github-actions github-actions released this 31 Jul 18:32
· 28 commits to main since this release

Added

  • check_response — name a refusal this library cannot see. A 200 carrying {"success": false} is indistinguishable from content on the wire, because the difference lives in a schema only the caller knows. ScraperConfig.check_response, also settable per session, returns a Diagnosis the loop then treats as its own: layer attributed, address blamed, planner rotates or escalates. Without it such a site reads as unbroken success while every page comes back empty. Consulted only on a response nothing else faulted; one that raises is logged and ignored.

Fixed

  • A tor-pool with no token collapsed to a single exit. The session key travels as the SOCKS5 username and the token as the password, and RFC 1929 cannot send one without the other — so the key never arrived and the pool keyed by client address instead, pinning every session to one instance. Worst for a local pool, which is the case least likely to have a token. An unset token now travels as a placeholder.