|
|
@@ -23,7 +23,7 @@ class FwhDataRecycle:
|
|
|
|
|
|
PUBLISH_SUCCESS_STATUS = 2
|
|
|
|
|
|
- STAT_PERIOD = 2 * 24 * 3600
|
|
|
+ STAT_PERIOD = 3 * 24 * 3600
|
|
|
|
|
|
ARTICLE_ILLEGAL_CODE = 25012
|
|
|
|
|
|
@@ -247,10 +247,11 @@ class SaveFwhDataToDatabase(FwhDataRecycle):
|
|
|
from long_articles_group_send_result
|
|
|
where gh_id = %s and recycle_status = %s and create_time > %s;
|
|
|
"""
|
|
|
+ earliest_time = datetime.fromtimestamp(earliest_timestamp).strftime("%Y-%m-%d %H:%M:%S")
|
|
|
return self.long_articles_client.fetch(
|
|
|
fetch_query,
|
|
|
DictCursor,
|
|
|
- (gh_id, self.RECYCLE_SUCCESS_STATUS, earliest_timestamp),
|
|
|
+ (gh_id, self.RECYCLE_SUCCESS_STATUS, earliest_time),
|
|
|
)
|
|
|
|
|
|
def process_each_account_data(self, account_published_article_list):
|
|
|
@@ -404,3 +405,8 @@ class FwhDataExportTemp(FwhDataRecycle):
|
|
|
print(f"article {article['ContentUrl']} is not available, skip it")
|
|
|
df = pd.DataFrame(L)
|
|
|
df.to_csv("temp2.csv", index=False)
|
|
|
+
|
|
|
+
|
|
|
+if __name__ == "__main__":
|
|
|
+ article_list = SaveFwhDataToDatabase().get_stat_published_articles("gh_5e543853d8f0")
|
|
|
+ print(article_list)
|
luojunhui