Spark SQL中时间戳条件约束与字符串判空

SELECT 
    -- 成员id
    distinct_id AS member_id, 
    -- 浏览时长
    get_json_object(all_json, '$.properties.event_duration') AS brow_duration,
    -- 浏览退出时间
    from_unixtime(CAST(get_json_object(all_json, '$.time') AS DOUBLE) / 1000) AS out_time,
    -- 内容编号
    regexp_extract(get_json_object(all_json, '$.properties.$url_query'), 'id=([^&]*)', 1) AS number
FROM {args['TABLE_NAME']} 
WHERE event = '$MPPageLeave' AND LENGTH(distinct_id)<30 
    AND get_json_object(all_json, '$.time') is not null 
    AND get_json_object(all_json, '$.properties.$url_query') is not null
    AND regexp_extract(get_json_object(all_json, '$.properties.$url_query'), 'id=([^&]*)', 1) is not null
    AND TRIM(regexp_extract(get_json_object(all_json, '$.properties.$url_query'), 'id=([^&]*)', 1)) != ''
    AND created_at >= unix_timestamp('2026-01-20 00:00:00')
    AND created_at < unix_timestamp('2026-01-21 00:00:00');

参考

• pyspark.sql.functions.unix_timestamp