From 9038ee2efaec9debd73e1b2edb492aece7806c6c Mon Sep 17 00:00:00 2001
From: eezd <eezd@outlook.com>
Date: Thu, 5 Nov 2026 13:43:06 +0800
Subject: [PATCH] Parse SEC official release datelines with September
 abbreviations

---
 backend/src/alphapulse/core/us_news.py |  3 +++
 backend/tests/test_us_official_news.py | 10 ++++++++++
 2 files changed, 13 insertions(+)

diff --git a/backend/src/alphapulse/core/us_news.py b/backend/src/alphapulse/core/us_news.py
index 8a38d25..5258e49 100644
--- a/backend/src/alphapulse/core/us_news.py
+++ b/backend/src/alphapulse/core/us_news.py
@@ -69,6 +69,7 @@ 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', ' ')
+    value = re.sub(r'\bSept(?=\.? \d)', 'Sep', value)
     if re.fullmatch(r'\d{4}-\d{2}-\d{2}', value):
         return datetime.fromisoformat(value).replace(tzinfo=UTC), 'date'
     try:
@@ -120,6 +121,8 @@ def article(source: str, html: str, url: str, *, release_date: str | None = None
     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()
+    if source == 'sec':
+        values += sel.css('.field--name-dynamic-twig-fieldnode-press-release-lead-in').xpath('normalize-space(.)').getall()
     for script in sel.css('script[type="application/ld+json"]::text').getall():
         try:
             data = json.loads(script)
diff --git a/backend/tests/test_us_official_news.py b/backend/tests/test_us_official_news.py
index 06b1af3..b967415 100644
--- a/backend/tests/test_us_official_news.py
+++ b/backend/tests/test_us_official_news.py
@@ -1376,3 +1376,13 @@ def test_doj_captioned_photo_is_not_a_document_gap(trading):
     item, = asyncio.run(collect())
     assert item['extra']['attachments'] == []
     assert not obj.cursor.get('attachment_retries')
+
+
+def test_sec_uses_official_dateline_not_last_reviewed_date():
+    body = '''<main><h1>SEC regulatory announcement</h1>
+    <div class="field--name-dynamic-twig-fieldnode-press-release-lead-in">Washington D.C., Sept. 30, 2026 —</div>
+    <div class="node-details-layout__main-region__content"><div class="field--name-body">The Commission proposes digital asset market rules for regulated financial institutions.</div></div>
+    <div class="date-modified">Last Reviewed or Updated: November 5, 2026</div></main>'''
+    item = article('sec', body, 'https://www.sec.gov/newsroom/press-releases/example')
+    assert item['published_at'] == datetime(2026, 9, 30, tzinfo=UTC)
+    assert item['extra']['published_at_precision'] == 'date'