一、timestamp字段与int字段相加

（一）场景

change_time字段是timestamp字段，代表一个红绿灯周期的开始时间（先是绿灯、再是黄灯、最后红灯），而green是int字段，代表绿灯的秒数，现在要求出绿灯的结束时间。即change_time字段+green字段

（二）timestamp字段与int字段无法直接相加，可以先把change_time字段转为时间戳，然后和green字段相加，最后再转为日期

样例：from_unixtime(unix_timestamp(change_time, 'yyyy-MM-dd HH:mm:ss') + green) AS new_timestamp

（三）SQL语句

 成功！！！

二、with语句与insert结合使用

（一）场景

在DWS层中，对多层SQL使用with语句嵌套查询，然后insert插入数据。如果直接把insert放在with语句上面，那么就会如下报错

（二）报错

org.apache.hadoop.hive.ql.parse.ParseException:line 2:0 cannot recognize input near 'with' 'a1' 'as' in statement

（三）解决方法

把insert放在with的后面，select的前面

（四）SQL语句

with a1 as(
selectb1.site_id, b1.phase_id, b1.phase_start, b1.program_id, b1.lane_direction, b1.device_direction,b1.min_gree_end, b1.phase_end, b1.team_id, b1.name,t8.device_no,t9.lane_num lane_no
from  dws_pass as b1
left join hurys_dc_dwd.dwd_radar_config as t8
on t8.direction=b1.device_direction and t8.device_no=b1.device_no  --得到真正的雷达编号字段
left join hurys_dc_dwd.dwd_radar_lane as t9
on t9.device_no=b1.device_no and t9.lane_direction=b1.lane_direction  --得到车道编号字段
group by b1.site_id, b1.phase_id, b1.phase_start, b1.program_id, b1.lane_direction, b1.device_direction, b1.min_gree_end, b1.phase_end, b1.team_id, b1.name, t8.device_no, t9.lane_num)
insert  overwrite  table  dws_pass_sparetime_1hour  partition(day)
selecta1.site_id,phase_id,program_id,phase_start,min_gree_end,phase_end,a1.device_no,team_id,name,t10.create_time,concat(substr(create_time, 1, 14), '00:00') start_time,a1.lane_no,section_no,coil_no,device_direction direction,lane_direction,target_id,target_type,drive_in_time,day
from a1
left join hurys_dc_dwd.dwd_pass as t10
on t10.device_no=a1.device_no and t10.lane_no=a1.lane_noand t10.create_time between a1.min_gree_end and a1.phase_end
group by a1.site_id, phase_id, program_id, phase_start, min_gree_end, phase_end, a1.device_no, team_id, name, t10.create_time, a1.lane_no, section_no, coil_no, device_direction, lane_direction, target_id, target_type, drive_in_time, day
;