feat(03b): add BYO custom proxy provider + bounded crawler rotation-retry

Add CustomProxyProvider (single endpoint or rotating pool via Scrapling ProxyRotator), registered as 'custom' alongside anonymous_proxies and selectable via PROXY_PROVIDER. Adds is_pool_backed to the ProxyProvider ABC + a zero-arg package helper. The web crawler does a bounded one-per-tier is_proxy_error rotation-retry gated on is_pool_backed() (single-endpoint providers no-op). Config/.env.example gain CUSTOM_PROXY_URL(S). Zero-arg getter contract unchanged for all consumers. Documents the proprietary boundary test (generic proxy infra stays Apache-2). Tests: provider, registry, crawler rotation (16).

Co-authored-by: Cursor <cursoragent@cursor.com>
This commit is contained in:
DESKTOP-RTLN3BA\$punk 2026-06-29 21:03:34 -07:00
parent 5c36cd3071
commit 62260125f7
12 changed files with 358 additions and 4 deletions

View file

@ -44,3 +44,15 @@ class ProxyProvider(ABC):
if proxy_url is None:
return None
return {"http": proxy_url, "https": proxy_url}
@property
def is_pool_backed(self) -> bool:
"""Whether this provider rotates across a *client-side* pool of endpoints.
``False`` for single-endpoint providers (including server-side rotating
gateways like ``anonymous_proxies``, whose rotation happens upstream).
The crawler performs its bounded proxy-error rotation-retry **only** when
this is ``True`` retrying a single static endpoint would just re-hit the
same dead proxy.
"""
return False