Spark SQL中时间戳条件约束与字符串判空
SELECT-- 成员iddistinct_idASmember_id,-- 浏览时长get_json_object(all_json,$.properties.event_duration)ASbrow_duration,-- 浏览退出时间from_unixtime(CAST(get_json_object(all_json,$.time)ASDOUBLE)/1000)ASout_time,-- 内容编号regexp_extract(get_json_object(all_json,$.properties.$url_query),id([^]*),1)ASnumberFROM{args[TABLE_NAME]}WHEREevent$MPPageLeaveANDLENGTH(distinct_id)30ANDget_json_object(all_json,$.time)isnotnullANDget_json_object(all_json,$.properties.$url_query)isnotnullANDregexp_extract(get_json_object(all_json,$.properties.$url_query),id([^]*),1)isnotnullANDTRIM(regexp_extract(get_json_object(all_json,$.properties.$url_query),id([^]*),1))!ANDcreated_atunix_timestamp(2026-01-20 00:00:00)ANDcreated_atunix_timestamp(2026-01-21 00:00:00);参考pyspark.sql.functions.unix_timestamp
