匹配原文

This commit is contained in:
yuxin-pc 2026-01-20 11:07:54 +08:00
parent 0b2abd342a
commit ee9473a6c3
2 changed files with 6 additions and 3 deletions

View File

@ -76,6 +76,7 @@ def main():
WHERE es_srcname IN ({placeholders}) WHERE es_srcname IN ({placeholders})
AND es_urlname IS NOT NULL AND es_urlname IS NOT NULL
AND es_urlname != '' AND es_urlname != ''
AND es_loadtime > '2026-01-16 10:40:00'
""" """
cursor.execute(query, TARGET_SRCNAMES) cursor.execute(query, TARGET_SRCNAMES)
records = cursor.fetchall() records = cursor.fetchall()

View File

@ -63,6 +63,7 @@ def get_chinese_records(cursor) -> List[Tuple]:
WHERE es_srcname IN ({placeholders}) WHERE es_srcname IN ({placeholders})
AND es_urltitle IS NOT NULL AND TRIM(es_urltitle) != '' AND es_urltitle IS NOT NULL AND TRIM(es_urltitle) != ''
AND es_urltime IS NOT NULL AND es_urltime IS NOT NULL
AND es_loadtime > '2026-01-16 10:40:00'
""" """
cursor.execute(query, TARGET_SRCNAMES) cursor.execute(query, TARGET_SRCNAMES)
return cursor.fetchall() return cursor.fetchall()
@ -73,11 +74,12 @@ def get_foreign_candidates_by_time(cursor, pub_time) -> List[Tuple]:
获取同一发布时间的所有外文候选记录要求 es_abstract 不为空 获取同一发布时间的所有外文候选记录要求 es_abstract 不为空
""" """
query = """ query = """
SELECT es_sid, es_abstract, es_urltitle, es_urlcontent SELECT es_sid, es_title, es_urltitle, es_urlcontent
FROM indeximos FROM indeximos
WHERE es_urltime = %s WHERE es_urltime = %s
AND es_abstract IS NOT NULL AND TRIM(es_abstract) != '' AND es_title IS NOT NULL AND TRIM(es_title) != ''
AND es_urlcontent IS NOT NULL AND es_urlcontent IS NOT NULL
AND es_loadtime > '2026-01-16 10:40:00'
""" """
cursor.execute(query, (pub_time,)) cursor.execute(query, (pub_time,))
return cursor.fetchall() return cursor.fetchall()