eezd
e8ca4daa0e
Collect official RSS and Atom summaries, preserve historical captures, and poll macro releases with bounded cadence and Retry-After handling. Separate news collectors from the economic calendar worker. Track statistical periods, publication stages, first observations, revisions, and current official series. Include explicit migration files and isolated archive, parsing, and import tests; no migration or import is executed by this commit. Keep dependency declarations and lockfile together for PDF archives and the upcoming captcha integration. Verify the staged snapshot with 90 focused tests.
209 lines
10 KiB
Python
209 lines
10 KiB
Python
"""导入已抓取的官方历史摘要及原始响应；独立任务不改变实时采集状态。
|
|
|
|
用法：python -m alphapulse.crawler.import_history SOURCE CAPTURES.jsonl.gz --start ISO时间
|
|
每行包含一次真实 HTTP 响应（source/url/status/headers/body/fetched_at）及解析后的 articles。
|
|
"""
|
|
|
|
import argparse
|
|
import base64
|
|
import gzip
|
|
import hashlib
|
|
import json
|
|
import re
|
|
import uuid
|
|
from datetime import UTC, datetime
|
|
from html import escape
|
|
from pathlib import Path
|
|
from typing import Literal
|
|
|
|
from pydantic import AwareDatetime, BaseModel, Field, HttpUrl
|
|
from scrapy import Request, Spider, signals
|
|
from scrapy.crawler import Crawler, CrawlerProcess
|
|
from scrapy.http import Response
|
|
from scrapy.utils.project import get_project_settings
|
|
from sqlalchemy import select, update
|
|
from twisted.python.failure import Failure
|
|
|
|
from alphapulse.core.db import SessionLocal
|
|
from alphapulse.core.official_sources import FEEDS, LANGUAGES
|
|
from alphapulse.crawler.events import write_logs
|
|
from alphapulse.crawler.middlewares import RawResponseMiddleware
|
|
from alphapulse.crawler.spiders.base import RunState
|
|
from alphapulse.models import CrawlRun, NewsItem, RawResponse
|
|
|
|
SOURCES = ("coindesk", "fed", "bea", "bls", *FEEDS)
|
|
|
|
|
|
class ArchivedArticle(BaseModel):
|
|
guid: str = Field(min_length=1)
|
|
url: HttpUrl
|
|
title: str = Field(min_length=1)
|
|
summary: str
|
|
published_at: AwareDatetime
|
|
source_updated_at: AwareDatetime | None = None
|
|
author: str | None = None
|
|
tags: list[str] = Field(default_factory=list)
|
|
language: Literal["en", "de", "fr"] | None = None
|
|
|
|
|
|
class Capture(BaseModel):
|
|
source: Literal["coindesk", "fed", "bea", "bls", "ecb", "eurostat", "boe", "ons", "destatis", "insee"]
|
|
url: HttpUrl
|
|
status: Literal[200]
|
|
headers: dict[str, str]
|
|
body: str
|
|
body_encoding: Literal["utf-8", "base64"] = "utf-8"
|
|
fetched_at: AwareDatetime
|
|
articles: list[ArchivedArticle]
|
|
|
|
|
|
def article_item(source: str, article: ArchivedArticle, capture: Capture, raw_id: int) -> dict:
|
|
content_html = f"<p>{escape(article.summary)}</p>" if article.summary else ""
|
|
return {
|
|
"source_code": source, "external_id": hashlib.sha256(article.guid.encode()).hexdigest(),
|
|
"url": str(article.url), "title": article.title, "summary": article.summary,
|
|
"content_html": content_html, "content_text": article.summary,
|
|
"published_at": article.published_at, "source_updated_at": article.source_updated_at,
|
|
"author": article.author, "tags": article.tags, "is_paywalled": False,
|
|
"extra": {"language": article.language or LANGUAGES.get(source, "en"), "content_scope": "public_summary", "historical_backfill": True,
|
|
"archive_url": str(capture.url), "archive_fetched_at": capture.fetched_at.isoformat()},
|
|
"raw_response_id": raw_id,
|
|
}
|
|
|
|
|
|
def identity_url(url: str) -> str:
|
|
# BEA 的 RSS GUID 同时使用 bea.gov 和 www.bea.gov；回补按原文链接识别已有文章。
|
|
url = url.replace("https://bea.gov/", "https://www.bea.gov/").rstrip("/")
|
|
# 官网跳转后的公报与 RSS 发布页属于同一公告；只统一导入去重键，保留原始链接。
|
|
eurostat = re.fullmatch(r"https://ec\.europa\.eu/eurostat/(?:en/)?web/products-euro-indicators/w/([^/?#]+)", url)
|
|
if eurostat:
|
|
return f"https://ec.europa.eu/eurostat/product?code={eurostat[1]}"
|
|
ons = re.fullmatch(r"https://www\.ons\.gov\.uk/.*/bulletins/(consumerpriceinflation|uklabourmarket|gdpfirstquarterlyestimateuk|quarterlynationalaccounts)/([^/?#]+)", url)
|
|
if ons:
|
|
series = {"consumerpriceinflation": "consumerpriceinflationuk", "quarterlynationalaccounts": "gdpquarterlynationalaccountsuk"}.get(ons[1], ons[1])
|
|
return f"https://www.ons.gov.uk/releases/{series}{ons[2]}"
|
|
return url
|
|
|
|
|
|
class HistoryImportSpider(Spider):
|
|
name = "history_import"
|
|
custom_settings = {
|
|
"SCHEDULER": "scrapy.core.scheduler.Scheduler",
|
|
"DUPEFILTER_CLASS": "scrapy.dupefilters.RFPDupeFilter",
|
|
}
|
|
|
|
def __init__(self, source: str, captures: str, start: datetime, **kwargs):
|
|
super().__init__(**kwargs)
|
|
self.name = source
|
|
self.captures = Path(captures)
|
|
self.start_at = start
|
|
self.end_at = datetime.now(UTC)
|
|
self.run = RunState(uuid.uuid4(), "backfill", self.end_at, start)
|
|
self.completed = False
|
|
self.known_ids: set[str] = set()
|
|
self.known_urls: set[str] = set()
|
|
|
|
@classmethod
|
|
def from_crawler(cls, crawler: Crawler, *args, **kwargs):
|
|
spider = super().from_crawler(crawler, *args, **kwargs)
|
|
crawler.signals.connect(spider.on_opened, signal=signals.spider_opened)
|
|
crawler.signals.connect(spider.on_error, signal=signals.spider_error)
|
|
crawler.signals.connect(spider.on_error, signal=signals.item_error)
|
|
crawler.signals.connect(spider.on_closed, signal=signals.spider_closed)
|
|
return spider
|
|
|
|
async def on_opened(self):
|
|
async with SessionLocal() as session:
|
|
rows = (await session.execute(
|
|
select(NewsItem.external_id, NewsItem.url).where(NewsItem.source_code == self.name)
|
|
)).all()
|
|
self.known_ids = {row[0] for row in rows}
|
|
self.known_urls = {identity_url(row[1]) for row in rows}
|
|
session.add(CrawlRun(id=self.run.id, source_code=self.name, trigger="backfill",
|
|
started_at=self.run.started_at, status="running"))
|
|
await session.commit()
|
|
|
|
async def start(self):
|
|
try:
|
|
async for item in self.import_captures():
|
|
yield item
|
|
except Exception as exc:
|
|
# start 的异常不走 spider_error 信号，必须让任务状态和退出码反映失败。
|
|
self.run.errors.append(repr(exc))
|
|
raise
|
|
|
|
async def import_captures(self):
|
|
raw_writer = RawResponseMiddleware(self.crawler)
|
|
with gzip.open(self.captures, "rt", encoding="utf-8") as file:
|
|
for line in file:
|
|
capture = Capture.model_validate_json(line)
|
|
if capture.source != self.name:
|
|
raise ValueError("历史响应来源与导入任务不一致")
|
|
request = Request(str(capture.url), meta={"run_id": self.run.id})
|
|
# 官方 PDF 归档需逐字节保存，不能经由文本解码破坏原始响应。
|
|
body = base64.b64decode(capture.body, validate=True) if capture.body_encoding == "base64" else capture.body.encode()
|
|
response = Response(str(capture.url), status=capture.status, headers=capture.headers,
|
|
body=body, request=request)
|
|
digest = hashlib.sha256(response.body).hexdigest()
|
|
async with SessionLocal() as session:
|
|
raw_id = await session.scalar(select(RawResponse.id).where(
|
|
RawResponse.source_code == self.name, RawResponse.url == response.url,
|
|
RawResponse.body_sha256 == digest, RawResponse.fetched_at == capture.fetched_at,
|
|
).limit(1))
|
|
if raw_id is None:
|
|
await raw_writer.process_response(request, response)
|
|
raw_id = request.meta["raw_response_id"]
|
|
async with SessionLocal() as session:
|
|
await session.execute(update(RawResponse).where(RawResponse.id == raw_id)
|
|
.values(fetched_at=capture.fetched_at))
|
|
await session.commit()
|
|
self.run.requests += 1
|
|
for article in capture.articles:
|
|
if not self.start_at <= article.published_at < self.end_at:
|
|
continue
|
|
item = article_item(self.name, article, capture, raw_id)
|
|
identity = identity_url(item["url"])
|
|
if item["external_id"] in self.known_ids or identity in self.known_urls:
|
|
continue
|
|
self.known_ids.add(item["external_id"])
|
|
self.known_urls.add(identity)
|
|
yield item
|
|
|
|
def on_error(self, failure: Failure):
|
|
self.run.errors.append(str(failure.value))
|
|
|
|
async def on_closed(self, reason: str):
|
|
status = "failed" if self.run.errors else ("success" if reason == "finished" else "interrupted")
|
|
self.completed = status == "success"
|
|
async with SessionLocal() as session:
|
|
await session.execute(update(CrawlRun).where(CrawlRun.id == self.run.id).values(
|
|
finished_at=datetime.now(UTC), status=status, requests=self.run.requests,
|
|
items_seen=self.run.items_seen, items_new=self.run.items_new, items_revised=self.run.items_revised,
|
|
error="\n".join(self.run.errors)[:4000] if self.run.errors else None,
|
|
))
|
|
await session.commit()
|
|
await write_logs([(self.name, self.run.id, "INFO" if status == "success" else "ERROR",
|
|
f"历史摘要导入 {status}：响应 {self.run.requests}，新增 {self.run.items_new}，修订 {self.run.items_revised}")])
|
|
|
|
|
|
def main():
|
|
parser = argparse.ArgumentParser(description=__doc__)
|
|
parser.add_argument("source", choices=SOURCES)
|
|
parser.add_argument("captures", type=Path)
|
|
parser.add_argument("--start", required=True, type=datetime.fromisoformat)
|
|
args = parser.parse_args()
|
|
if args.start.tzinfo is None or args.start >= datetime.now(UTC):
|
|
parser.error("起始时间须带时区且早于当前时间")
|
|
process = CrawlerProcess(get_project_settings())
|
|
crawler = process.create_crawler(HistoryImportSpider)
|
|
process.crawl(crawler, source=args.source, captures=str(args.captures), start=args.start)
|
|
process.start()
|
|
if not crawler.spider.completed or crawler.spider.run.items_seen != crawler.spider.run.items_new:
|
|
raise SystemExit("历史导入存在失败或意外修订，请检查采集日志")
|
|
print(json.dumps({"source": args.source, "run_id": str(crawler.spider.run.id),
|
|
"requests": crawler.spider.run.requests, "inserted": crawler.spider.run.items_new}))
|
|
|
|
|
|
if __name__ == "__main__":
|
|
main()
|