Add resumable US policy news collectors and provenance
Collect official crypto and US policy documents with deterministic screening, full bodies and attachments, bounded retries, and durable history cursors. Preserve publication precision and source identity while keeping new collectors disabled. Add real Congress, GovInfo, DOJ and Federal Register fixtures, offline capture replay, isolated database validation, and rollout instructions. Validate 982 backend tests, frontend checks, and 57 deployment control tests.
This commit is contained in:
eezd committed
1 parent
dc8a283571
commit
89c729092c
31 files changed
+2199
-23
No files matched your search
@@ -0,0 +1,171 @@
|
||||
"""Official US news catalog and deterministic collection policy."""
|
||||
import hashlib
|
||||
import json
|
||||
import re
|
||||
from dataclasses import dataclass
|
||||
from datetime import UTC, datetime
|
||||
from urllib.parse import urljoin, urlsplit, urlunsplit
|
||||
|
||||
from parsel import Selector
|
||||
|
||||
|
||||
@dataclass(frozen=True)
|
||||
class OfficialSource:
|
||||
name: str
|
||||
entries: tuple[str, ...]
|
||||
article_path: str
|
||||
body: str
|
||||
hosts: tuple[str, ...]
|
||||
interval: int = 300
|
||||
|
||||
|
||||
SOURCES = {
|
||||
'sec': OfficialSource('SEC·证券监管', ('https://www.sec.gov/newsroom/press-releases', 'https://www.sec.gov/newsroom/speeches-statements', 'https://www.sec.gov/rules-regulations/rulemaking-activity', 'https://www.sec.gov/rules-regulations/self-regulatory-organization-rulemaking'), r'/(?:newsroom/(?:press-releases|speeches-statements)/[^/]+|rules-regulations/.*\d.*)', '.node-details-layout__main-region__content .field--name-body', ('www.sec.gov',)),
|
||||
'cftc': OfficialSource('CFTC·衍生品监管', ('https://www.cftc.gov/PressRoom/PressReleases', 'https://www.cftc.gov/PressRoom/SpeechesTestimony', 'https://www.cftc.gov/LawRegulation/CFTCStaffLetters/letters.htm'), r'/(?:PressRoom/(?:PressReleases/\d+-\d+|SpeechesTestimony/(?!index)[^/]+)|csl/)', 'article .field--name-body', ('www.cftc.gov',)),
|
||||
'treasury': OfficialSource('美国财政部', ('https://home.treasury.gov/news/press-releases',), r'/news/press-releases/[a-z]+\d+/?$', '.field--name-field-news-body', ('home.treasury.gov', 'www.treasury.gov', 'ofac.treasury.gov')),
|
||||
'ofac': OfficialSource('OFAC·金融制裁', ('https://ofac.treasury.gov/recent-actions',), r'/recent-actions/\d{8}[^/]*$', '.field--name-body, .body-content', ('ofac.treasury.gov', 'home.treasury.gov')),
|
||||
'fincen': OfficialSource('FinCEN·反洗钱', ('https://www.fincen.gov/news/press-releases', 'https://www.fincen.gov/news/enforcement-actions', 'https://www.fincen.gov/resources/advisoriesbulletinsfact-sheets'), r'/(?:news/news-releases/[^/]+|resources/advisories/[^/]+|news/enforcement-actions/[^/]+)', 'article .field--name-body', ('www.fincen.gov',)),
|
||||
'whitehouse': OfficialSource('白宫·政策', ('https://www.whitehouse.gov/presidential-actions/', 'https://www.whitehouse.gov/fact-sheets/', 'https://www.whitehouse.gov/briefings-statements/', 'https://bidenwhitehouse.archives.gov/briefing-room/'), r'/(?:presidential-actions|fact-sheets|briefings-statements|briefing-room/[^/]+)/202[5-9]/\d{2}/\d{2}/|/(?:presidential-actions|fact-sheets|briefings-statements)/202[5-9]/\d{2}/[^/]+', '.wp-block-post-content, .body-content, .entry-content', ('www.whitehouse.gov', 'bidenwhitehouse.archives.gov')),
|
||||
'congress': OfficialSource('美国国会·立法', ('https://api.congress.gov/v3/bill?format=json&limit=250', 'https://api.congress.gov/v3/committee-meeting?format=json&limit=250'), r'/bill/', '', ('api.congress.gov', 'www.congress.gov', 'www.govinfo.gov'), 900),
|
||||
'federal_register': OfficialSource('Federal Register·规则', ('https://www.federalregister.gov/api/v1/documents.json',), r'/documents/', '', ('www.federalregister.gov', 'public-inspection.federalregister.gov', 'www.govinfo.gov'), 900),
|
||||
'occ': OfficialSource('OCC·银行与稳定币', tuple(f'https://www.occ.gov/news-events/newsroom/news-issuances-by-year/{kind}/{year}-{kind}.html' for year in range(datetime.now(UTC).year, 2024, -1) for kind in ('news-releases', 'bulletins')), r'/news-issuances/(?:news-releases|bulletins|interpretations-and-precedents)/202[5-9]/[^/]+', '.occgov-issuance-content', ('www.occ.gov', 'occ.gov', 'www.occ.treas.gov', 'occ.treas.gov')),
|
||||
'doj': OfficialSource('DOJ·加密机构执法', ('https://www.justice.gov/api/v1/press_releases.json?pagesize=50&sort=date&direction=DESC&fields=uuid,date',), r'/(?:opa|usao[^/]*|archives/opa)/(?:pr|speech)/[^/]+', '.field--name-body, .field--name-field-pr-body', ('www.justice.gov',)),
|
||||
'fed': OfficialSource('美联储·政策与讲话', tuple(f'https://www.federalreserve.gov/newsevents/{kind}/{year}-{suffix}.htm' for year in range(datetime.now(UTC).year, 2024, -1) for kind, suffix in (('pressreleases', 'press'), ('speeches', 'speeches'), ('testimony', 'testimony'))) + ('https://www.federalreserve.gov/monetarypolicy/fomccalendars.htm', 'https://www.federalreserve.gov/releases/h41/releaseDates.json'), r'/(?:newsevents/(?:pressreleases|speech|testimony)/[^/]+\.htm|monetarypolicy/(?:fomcminutes|fomcprojtabl)\d+\.htm|releases/h41/\d+/)', '#article, .col-xs-12.col-sm-8.col-md-8, #content', ('www.federalreserve.gov',), 60),
|
||||
}
|
||||
RULE_VERSION = 'us-policy-v1'
|
||||
CRYPTO = re.compile(r'\b(?:crypto(?:currency|currencies|[ -]assets?)?|digital[ -]assets?|virtual[ -]currenc(?:y|ies)|stablecoins?|bitcoin|ethereum|blockchain|tokeniz(?:ed|ation))\b', re.I)
|
||||
POLICY = re.compile(r'\b(?:regulat\w*|rules?|guidance|legislation|acts?|sanctions?|licenses?|executive order|monetary policy|federal funds|interest rates?|balance sheet|liquidity|quantitative|refunding|borrowing|debt limit|tariffs?|fiscal|budget|appropriation|tax reform)\b', re.I)
|
||||
INSTITUTION = re.compile(r'\b(?:exchanges?|banks?|financial institutions?|payment processors?|custodians?|stablecoin issuers?|binance|coinbase|tether|circle|kraken|bybit|bitfinex)\b', re.I)
|
||||
MACRO = re.compile(r'\b(?:monetary policy|federal funds|interest rates?|fomc|balance sheet|liquidity|quantitative|refunding|borrowing estimates|debt (?:limit|ceiling)|tariffs?|fiscal|budget|appropriation|tax reform|financial sanctions)\b', re.I)
|
||||
|
||||
|
||||
def selection(source: str, title: str, text: str) -> list[str]:
|
||||
content = f'{title}\n{text}'
|
||||
crypto = bool(CRYPTO.search(content))
|
||||
if crypto and (source not in ('doj', 'ofac', 'fincen') or POLICY.search(title) or INSTITUTION.search(content)):
|
||||
return ['crypto_policy' if POLICY.search(content) else 'crypto_institution']
|
||||
if source not in ('doj', 'cftc', 'sec') and MACRO.search(content):
|
||||
return ['us_macro_policy']
|
||||
if source == 'ofac' and INSTITUTION.search(content) and POLICY.search(content):
|
||||
return ['financial_sanctions']
|
||||
return []
|
||||
|
||||
|
||||
def official_url(source: str, url: str, base: str = '') -> str:
|
||||
value = urlsplit(urljoin(base, url))
|
||||
if value.scheme != 'https' or value.hostname not in SOURCES[source].hosts or value.username or value.password:
|
||||
raise ValueError('链接不是该来源的官方 HTTPS 地址')
|
||||
if re.search(r'(?:api_key|token|key)=', value.query, re.I):
|
||||
raise ValueError('官方链接含凭据参数')
|
||||
return urlunsplit((value.scheme, value.netloc, value.path, value.query, ''))
|
||||
|
||||
|
||||
def identity(url: str) -> str:
|
||||
return hashlib.sha256(url.encode()).hexdigest()
|
||||
|
||||
|
||||
def published(value: str) -> tuple[datetime, str]:
|
||||
if not isinstance(value, str):
|
||||
raise ValueError('官方材料缺少可识别的发布时间')
|
||||
value = value.strip().replace('\xa0', ' ')
|
||||
if re.fullmatch(r'\d{4}-\d{2}-\d{2}', value):
|
||||
return datetime.fromisoformat(value).replace(tzinfo=UTC), 'date'
|
||||
try:
|
||||
dt = datetime.fromisoformat(value.replace('Z', '+00:00'))
|
||||
if dt.tzinfo:
|
||||
return dt.astimezone(UTC), 'datetime'
|
||||
except ValueError:
|
||||
pass
|
||||
match = re.search(r'([A-Z][a-z]+\.? \d{1,2}, \d{4}|\d{1,2}/\d{1,2}/\d{4})', value)
|
||||
if match:
|
||||
for fmt in ('%B %d, %Y', '%b %d, %Y', '%m/%d/%Y'):
|
||||
try:
|
||||
return datetime.strptime(match[0].replace('.', ''), fmt).replace(tzinfo=UTC), 'date'
|
||||
except ValueError:
|
||||
pass
|
||||
raise ValueError('官方材料缺少可识别的发布时间')
|
||||
|
||||
|
||||
def validate_body(content: str) -> None:
|
||||
"""Reject error documents without requiring full-page HTML around fragments."""
|
||||
if content.lstrip().startswith('<?xml'):
|
||||
from lxml import etree
|
||||
try:
|
||||
etree.fromstring(content.encode(), etree.XMLParser(resolve_entities=False, no_network=True))
|
||||
except etree.XMLSyntaxError as exc:
|
||||
raise ValueError('官方 XML 正文无法解析') from exc
|
||||
sel = Selector(content)
|
||||
heading = ' '.join(sel.css('title, h1').xpath('normalize-space(.)').getall())
|
||||
text = ' '.join(sel.xpath('//text()[not(ancestor::script or ancestor::style)]').getall()).strip()
|
||||
if re.search(r'just a moment|access denied|verify (?:you are|your)|sign in|log in|captcha|page not found|service unavailable|bad gateway', heading, re.I) or sel.css('#challenge-form, #cf-challenge-running, script[src*="/cdn-cgi/challenge"]'):
|
||||
raise ValueError('官方正文返回验证或登录页面')
|
||||
if re.search(r'enable javascript and cookies to continue', text, re.I):
|
||||
raise ValueError('官方正文返回验证页面')
|
||||
if len(text) < 40:
|
||||
raise ValueError('官方正文为空或过短')
|
||||
|
||||
|
||||
def article(source: str, html: str, url: str) -> dict:
|
||||
sel = Selector(html)
|
||||
validate_body(html)
|
||||
selectors = {
|
||||
'fed': ('#article h3.title', '#article h3', '#article h2'),
|
||||
'treasury': ('h2.uswds-page-title',),
|
||||
'cftc': ('article h1:not(.press-release-title)',),
|
||||
}.get(source, ('.main-content__main h1', '#main_content h1', 'main h1', 'h1'))
|
||||
title = next((value for css in selectors if (value := sel.css(css).xpath('normalize-space(.)').get())), None)
|
||||
if source == 'fed' and '/releases/h41/' in url:
|
||||
title = 'Federal Reserve Balance Sheet — H.4.1'
|
||||
if not title:
|
||||
raise ValueError('官方正文缺少标题')
|
||||
values = sel.css('meta[property="article:published_time"]::attr(content), meta[name="DC.date.issued"]::attr(content), meta[name="dcterms.date"]::attr(content), time::attr(datetime)').getall()
|
||||
for script in sel.css('script[type="application/ld+json"]::text').getall():
|
||||
try:
|
||||
data = json.loads(script)
|
||||
nodes = data if isinstance(data, list) else data.get('@graph', [data])
|
||||
values += [n['datePublished'] for n in nodes if isinstance(n, dict) and n.get('datePublished')]
|
||||
except (ValueError, TypeError, AttributeError):
|
||||
continue
|
||||
values += sel.css('.node-details-layout__main-region__content, .date, .date-display-single, .release-date, .article__time, .published-date, .field--name-field-date, .field--name-field-release-date, .press-release-date').xpath('normalize-space(.)').getall()
|
||||
values += re.findall(r'(?:January|February|March|April|May|June|July|August|September|October|November|December) \d{1,2}, \d{4}', ' '.join(sel.css('main, article, #content').xpath('string(.)').getall()))[:3]
|
||||
# OFAC action identifiers encode the publisher's day, not the fetch time.
|
||||
if source == 'fed' and (m := re.search(r'/releases/h41/(\d{4})(\d{2})(\d{2})', url)):
|
||||
values.insert(0, '-'.join(m.groups()))
|
||||
if source == 'ofac' and (m := re.search(r'/recent-actions/(\d{4})(\d{2})(\d{2})', url)):
|
||||
values.insert(0, '-'.join(m.groups()))
|
||||
at = precision = None
|
||||
for value in values:
|
||||
try:
|
||||
at, precision = published(value)
|
||||
break
|
||||
except ValueError:
|
||||
continue
|
||||
if at is None:
|
||||
raise ValueError('官方材料缺少可识别的发布时间')
|
||||
body = sel.css(SOURCES[source].body)
|
||||
if not body:
|
||||
raise ValueError('官方正文结构未识别')
|
||||
for node in body.css('script, style, nav, header, footer, form'):
|
||||
node.root.getparent().remove(node.root)
|
||||
body = sorted(body, key=lambda node: len(node.xpath('string(.)').get()), reverse=True)
|
||||
content = body[0].get()
|
||||
text = ' '.join(body[0].xpath('.//text()').getall()).strip()
|
||||
if len(text) < 40:
|
||||
raise ValueError('官方正文为空或过短')
|
||||
attachments = []
|
||||
for link in body[0].css('a[href]'):
|
||||
href = link.attrib['href']
|
||||
label = link.xpath('normalize-space(.)').get()
|
||||
if re.search(r'\.pdf(?:\?|$)|/media/\d+/(?:download|dl)|/csl/[^/]+/download', href, re.I):
|
||||
try:
|
||||
link_url = official_url(source, href, url)
|
||||
except ValueError:
|
||||
candidate = urljoin(url, href)
|
||||
if urlsplit(candidate).scheme == 'https':
|
||||
attachments.append({'url': candidate, 'title': label, 'status': 'unsupported_host', 'error': '附件不在该来源的官方域名范围内，未下载'})
|
||||
continue
|
||||
if link_url not in [a['url'] for a in attachments]:
|
||||
attachments.append({'url': link_url, 'title': label, 'status': 'pending'})
|
||||
return dict(title=title, content_html=content, published_at=at,
|
||||
extra={'language': 'en', 'content_scope': 'article', 'published_at_precision': precision,
|
||||
'published_date': at.date().isoformat(), 'rule_version': RULE_VERSION, 'attachments': attachments})
|
||||
Reference in new issue
Block a user