get tags

dssm/get_tractate_data.py

@@ -19,6 +19,26 @@ def get_ndays_before_no_minus(n):
     return get_ndays_before_with_format(n, "%Y%m%d")
 
 
+def connect_doris(spark, table):
+    return spark.read.format("jdbc") \
+            .option("driver", "com.mysql.jdbc.Driver") \
+            .option("url", "jdbc:mysql://172.16.30.136:3306/doris_prod") \
+            .option("dbtable", table) \
+            .option("user", "doris") \
+            .option("password", "o5gbA27hXHHm") \
+            .load()
+
+
+def get_content_tag3(spark, card_type):
+    if card_type == "diary":
+        content_tag3 = connect_doris(spark, "strategy_content_tagv3_info")
+    elif card_type == "user_post":
+        content_tag3 = connect_doris(spark, "strategy_tractate_tagv3_info")
+    else:
+        content_tag3 = connect_doris(spark, "strategy_answer_tagv3_info")
+    return content_tag3
+
+
 def get_spark(app_name=""):
     sparkConf = SparkConf()
     sparkConf.set("spark.sql.crossJoin.enabled", True)
@@ -237,6 +257,9 @@ def get_exposure_data(spark, card_type, start, end):
 
 
 def get_card_feature_df(spark, card_type, yesterday):
+    content_tag3 = get_content_tag3(spark, card_type)
+    content_tag3.createOrReplaceTempView("content_tag3")
+
     reg = r"""^\\d+$"""
     sql = """
           SELECT CAST(card_id as INT) as card_id,