Fix official news retries, date boundaries and collection status
Development CI / check-and-build (push) Successful in 3m2s
GitHub Git mirror / mirror (push) Successful in 6s
Development CI / deploy-development (push) Successful in 2m31s

Bound document retries across realtime and history lanes, preserve safe completed reads on failed rounds, and retain unresolved gaps. Honor the longest access-block cooldown and measure retries from the actual failure time.

Filter dated official listings at the collection boundary, handle empty Federal Register days, and distinguish active rounds from waiting or failed processes in the console. Add captured-response regressions and isolated checkpoint validation.
This commit is contained in:
eezd committed 2026-11-05 11:22:46 +08:00
1 parent 80665905e5
commit 39675fe46b
11 files changed
+577 -39

No files matched your search

+27 -1
View File
@@ -7,6 +7,7 @@ import uuid
from collections.abc import Callable, Coroutine, Iterable
from dataclasses import dataclass, field
from datetime import UTC, datetime, timedelta
from email.utils import parsedate_to_datetime
from typing import Any
from zoneinfo import ZoneInfo
@@ -67,6 +68,9 @@ class RunState:
history_deadline: float = 0
official_cursor: dict | None = None
stats: dict = field(default_factory=dict)
completed_documents_safe: bool = False
cooldown_until: datetime | None = None
document_retries_started: set[str] = field(default_factory=set)
@property
def full_scan(self) -> bool:
@@ -188,6 +192,7 @@ class NewsSpider(RedisSpider):
def _on_item_error(self, failure: Failure) -> None:
if self.run:
self.run.completed_documents_safe = False
self.run.errors.append(f"新闻处理或落库失败:{failure.value!r}")
async def _on_closed(self) -> None:
@@ -238,12 +243,21 @@ class NewsSpider(RedisSpider):
if history_saved:
cursor["history"] = run.official_cursor.get("history", {})
source.history_cursor = cursor
elif failed and run.official_cursor is not None and run.completed_documents_safe:
# Retain completed reads and explicit gaps, never failed lane positions.
cursor = dict(source.history_cursor or {})
for key in ('checked', 'document_retries', 'attachment_retries', 'bill_retries'):
if key in run.official_cursor:
cursor[key] = run.official_cursor[key]
source.history_cursor = cursor
if history_saved:
source.history_position = run.history_position
source.history_done = run.history_done
if failed:
source.consecutive_failures += 1
delay = min(BACKOFF_BASE_SECONDS * 2 ** (source.consecutive_failures - 1), BACKOFF_MAX_SECONDS)
if run.cooldown_until:
delay = max(delay, int((run.cooldown_until - now).total_seconds()) + 1)
source.paused_until = now + timedelta(seconds=delay)
source.last_error = run.errors[0][:1000]
else:
@@ -293,10 +307,22 @@ class NewsSpider(RedisSpider):
headers=getattr(self, "request_headers", None),
)
def block_until(self, until: datetime) -> None:
self.run.blocked = True
self.run.cooldown_until = max(self.run.cooldown_until or until, until)
def on_request_error(self, failure: Failure) -> None:
status = failure.value.response.status if failure.check(HttpError) else None
if status in (403, 429):
self.run.blocked = True
now = datetime.now(UTC)
self.block_until(now + timedelta(seconds=BACKOFF_MAX_SECONDS))
retry_after = failure.value.response.headers.get('Retry-After', b'').decode('ascii', errors='ignore')
if retry_after:
try:
until = now + timedelta(seconds=int(retry_after)) if retry_after.isdigit() else parsedate_to_datetime(retry_after)
self.block_until(until)
except (ValueError, TypeError, OverflowError):
self.logger.warning('忽略无效 Retry-After 响应头')
message = f"请求失败 {failure.request.url}:{status or repr(failure.value)}"
self.run.errors.append(message)
self.logger.error(message)