ODPS语句集锦

-- DROP TABLE micro_operation_content_pool_day_temp;
CREATE TABLE IF NOT EXISTS micro_operation_content_pool_day_temp
(
    channel string COMMENT '渠道'
    ,content_id BIGINT COMMENT '内容id'
    ,createTime BIGINT COMMENT '插入时间'
    ,content_pool_id BIGINT COMMENT '数据池id，如首页推荐池，精选频道推荐池'
    ,data_source STRING COMMENT '数据来源，如点将台场景666'
    ,data_version STRING COMMENT '数据版本'
)
PARTITIONED BY 
(
    ds STRING
)
LIFECYCLE 3
;


INSERT OVERWRITE TABLE micro_operation_content_pool_day_temp PARTITION(ds='${bizdate}')
SELECT  "data_pool_diff" as channel
        ,content_id
        ,unix_timestamp(getdate()) as createTime
        ,content_pool_id
        ,data_source
        ,data_version
FROM    (
            SELECT  content_id
                    ,content_pool_id
                    ,data_source
                    ,data_version
                    ,MAX(data_version) OVER (PARTITION BY content_pool_id,data_source) AS max_data_version
            FROM    (
                        SELECT  id
                                ,bi_udf:bi_get_json_object(features,'$.refer_id') AS content_id
                                ,bi_udf:bi_get_json_object(features, '$.scene_id') AS content_pool_id
                                ,bi_udf:bi_get_json_object(features,'$.data_source') AS data_source
                                ,bi_udf:bi_get_json_object(features, '$.data_version') AS data_version
                                ,bi_udf:bi_get_json_object(features,'$.type') AS data_source_type
                        FROM    tbods.s_macross_feed
                        WHERE   ds = '${bizdate}'
                        AND     biz_type = 503
                        AND     status = 0
                    ) a
            WHERE   a.data_source_type = 1
            AND     a.data_source IS NOT NULL
            AND     a.data_version IS NOT NULL
            AND     a.content_id IS NOT NULL
        ) b
WHERE   data_version = max_data_version
;


-- SELECT temp.content_id from 
--     micro_operation_content_pool_day_temp temp LEFT OUTER JOIN tbods.s_motor_content s 
--     on  temp.content_id = s.id 
--     WHERE temp.content_id != NULL  
--         and s.id = NULL 
--         and temp.ds="${bizdate}"
--         and s.ds="${bizdate}";

CREATE TABLE IF NOT EXISTS data_pool_diff_day
(
    channel string COMMENT '渠道'
    ,content_id BIGINT COMMENT '内容id'
    ,createTime BIGINT COMMENT '插入时间'
)
PARTITIONED BY 
(
    ds STRING
)
LIFECYCLE 3;

INSERT OVERWRITE TABLE data_pool_diff_day PARTITION(ds='${bizdate}')
SELECT 
        channel
        ,content_id
        ,createTime    
FROM (
    SELECT a.content_id as content_id,b.id as id from 
        (SELECT * from micro_operation_content_pool_day_temp WHERE ds=="${bizdate}") a 
            LEFT OUTER JOIN 
        (select * from tbods.s_motor_content WHERE ds=="${bizdate}") b
            on a.content_id = b.id
    ) A
WHERE A.content_id is not null and A.id is NULL; 


SELECT * from  data_pool_diff_day WHERE ds="${bizdate}";

CREATE TABLE IF NOT EXISTS sxpt_scene_content_check_complete (   
    create_time DATETIME    
)
PARTITIONED BY (
	ds STRING
);

Alter table sxpt_scene_content_check_complete drop if exists partition(ds='${bizdate}');

INSERT overwrite TABLE sxpt_scene_content_check_complete partition (ds='${bizdate}')  VALUES (getdate());

SELECT * from micro_data.sxpt_scene_content_check_complete where ds='20180407';

CREATE TABLE IF NOT EXISTS garbage_relation_add_daily (
	user_id bigint,
	account_id bigint,
	tag_list string 
)
PARTITIONED BY (
	ds STRING
);


Alter table garbage_relation_add_daily drop if exists partition(ds='${bizdate}');

INSERT overwrite TABLE garbage_relation_add_daily partition (ds='${bizdate}')
select  * 
from 
	micro_data.relation_clean_result t
where 
	(t.user_id % 180) = (datediff(to_date('${bizdate}', 'yyyymmdd'), '2017-08-15 00:00:00','dd') % 180)
	and  instr(t.tag_list, ',') != 0;
	

 

SELECT   
        A.content_id
        ,A.pic_url
        ,A.result_pic_url
FROM (
        SELECT  a.content_id as content_id,
                b.content_id as id,
                b.pic_url as pic_url,
                b.result_pic_url as result_pic_url
        FROM 
            (
                SELECT content_id FROM beehive.sxpt_scene_9207 WHERE ds="20180418"
                UNION SELECT content_id FROM beehive.sxpt_scene_8835 WHERE ds="20180418"

                UNION SELECT content_id FROM beehive.sxpt_scene_8984 WHERE ds="20180419"
                UNION SELECT content_id FROM beehive.sxpt_scene_8718 WHERE ds="20180419"
                UNION SELECT content_id FROM beehive.sxpt_scene_8994 WHERE ds="20180419"
                UNION SELECT content_id FROM beehive.sxpt_scene_9004 WHERE ds="20180419"
                UNION SELECT content_id FROM beehive.sxpt_scene_9229 WHERE ds="20180419"
                UNION SELECT content_id FROM beehive.sxpt_scene_9237 WHERE ds="20180419"
            ) a
            LEFT OUTER JOIN 
            (select * from luban.pixel_need_1x1_cover_pic_result WHERE ds="20180418") b
            on a.content_id = b.content_id
        
) A 
WHERE A.content_id is not null and A.id is NOT NULL

drop table if exists relation_clean_final;
create table if not exists relation_clean_final(
    user_id  bigint,
    account_id  bigint,
    tag_list  String
);


insert overwrite table relation_clean_final
select 
    t1.src_link_id,
    t1.dest_link_id,
    t1.tag_list
from 
    secdm.mdl_mz_weitao_eyi_links_fdt  t1 
join 
(
select 
    user_id
from
    relation_clean_num
where 
    num > 100 ) t2
on t1.ds= 20170702  and t1.src_link_id = t2.user_id;







CREATE TABLE IF NOT EXISTS garbage_relation_add_daily (
    user_id bigint,
    account_id bigint,
    tag_list string 
)
PARTITIONED BY (
    ds STRING
);




INSERT INTO TABLE garbage_relation_add_daily partition (ds='${bizdate}')
select  * 
from relation_clean_final t
where t.account_id % 180 = ('${bizdate}' - 20170815) % 1