Hive Struct to String转换(Hive Struct to String conversion)
我有一个包含结构的表 - 让我们说:
create external table table1 (
a int,
b STRUCT ,
e string
)
我在这张桌子上执行选择,得到类似的东西 -
1100 {“c”:12.3,“d”:45.6} str
但是当我将这些数据插入另一个表格时 -
create external table table2 (
a string,
b string,
c string
)
insert overwrite table table2
select a,b,c
from table1;
我得到以下奇怪的行为,表明在hive中struct和string之间的转换不能按预期工作
select * from table2;
会导致 -
1100 12.345.6 str
结果是结构中的一种奇怪的值连接,甚至在处理更复杂的结构时也会发生奇怪的事情
有没有办法阻止这种自动转换? 在这种情况下让hive抛出错误?
有没有一种干净的方法来改变这种自动转换以不同的方式工作?
I have a table that has a struct in it - let's say:
create external table table1 (
a int,
b STRUCT ,
e string
)
I perform select on this table and get something like -
1100 {"c":12.3,"d":45.6} str
But then when I insert this data to another table -
create external table table2 (
a string,
b string,
c string
)
insert overwrite table table2
select a,b,c
from table1;
I get the following strange behavior indicating that the conversion between struct and string in hive doesn't work as expected
select * from table2;
would result -
1100 12.345.6 str
the result is a kind of strange concatenation of values in the struct and even stranger things happen when dealing with more complex structs
Is there a way to prevent this auto conversion? Make hive throw an error in such cases?
Is there a clean way to change this auto conversion to work differently?
原文:https://stackoverflow.com/questions/40395868
更新时间:2020-03-05 15:03
最满意答案
我们无法阻止自动转换时直接insert overwrite table table2 select a,b,c from table1; 调用。 背后发生的事情是它只是从struct所有值。
更快捷的方式:
如果你的目的是从struct获取价值并将其存储为原始值,那么尝试如下,
create external table table2 (
a string,
b_c string,
b_d string,
c string
)
insert overwrite table table2
select a,b.c,b.e,c
from table1;
如果这有帮助,请告诉我。
We cannot prevent the auto conversion when directly insert overwrite table table2 select a,b,c from table1; called. Whats happening behind is it just concat all the values from struct.
Quicker way:
if your intention is to get value out of struct and store it as a primitive value then try like below,
create external table table2 (
a string,
b_c string,
b_d string,
c string
)
insert overwrite table table2
select a,b.c,b.e,c
from table1;
Let me know if this helps.
2016-11-03
相关问答
你只需要爆炸一次(连同横向视图)。 在爆炸之后,你可以使用一个新的列(在我的例子中称为prod_and_ts),它将是结构类型的。 然后,您可以解析此新结构列的product_id和timestamps成员以检索所需的结果。 SELECT
user_id,
prod_and_ts.product_id as product_id,
prod_and_ts.timestamps as timestamps
FROM
SampleTable
LATERAL VIEW
...
找到我的问题的一个解决方案: select
seq,
split(split(results,",")[0],':')[1] as offerId,
split(split(results,",")[1],':')[1] as businessName,
split(split(results,",")[2],':')[1] as businessGroup,
split(split(resul
...
select * from hbweather where key.usaf=400010 and key.wban=99999 and key.`date`='199906280000' limit 10;
select * from hbweather where key.usaf=400010 and key.wban=99999 and key.`date`='199906280000' limit 10;
我找到了一个解决方案: 使用Hive爆炸UDTF来爆炸结构数组,即为数组“features”中的每个结构创建第二个(临时)表,并为其创建第二个(临时)表。 CREATE TABLE tbl_exploded as
select main_id,
f.name as f_name,
f.value as f_value
from tbl
LATERAL VIEW explode(features) exploded_table as f
-- optionally filter here inst
...
select t.name
,count (e.col.n_nationkey) as count
,sum (e.col.n_nationkey) as sum
,count (distinct e.col.n_nationkey) as distinct_val
from temp t lateral view explode
...
map keys terminated by ',' create external table temp
(
regionkey smallint
,name string
,comment string
,nations array>
)
row format delimited
fields terminat
...
我终于找到了答案。 看来这是用于序列化/反序列化JSON的JAR的问题。 默认的(Apache)无法对我拥有的数据执行良好的工作。 我尝试了所有这些典型的JAR(在括号中,'ROW FORMAT SERDE'的类): hive-json-serde-0.2.jar(org.apache.hadoop.hive.contrib.serde2.JsonSerde) hive-serdes-1.0-SNAPSHOT.jar(com.cloudera.hive.serde.JSONSerDe) hive-
...
您忘记了deviceID:调用DeviceData.init(deviceID:type:attributes:)命名参数,并且您还忘记了Meta.init(currentPage:nextPage:deviceID)的currentPage和nextPage命名参数Meta.init(currentPage:nextPage:deviceID) 。 这是一个编译的示例: var exampleData = Device(
data: DeviceData(
deviceI
...
在Hive中,Map列的键必须是基元(即不是Struct)。 https://cwiki.apache.org/confluence/display/Hive/LanguageManual+Types#LanguageManualTypes-ComplexTypes 我强烈建议你不要将关键字作为结构。 在您的示例中,如果我不知道开始或结束,如何访问Map的值? 用户需要知道确切的开始和结束,并对表中的每一行进行更改吗? In Hive the key for a Map column must b
...
我们无法阻止自动转换时直接insert overwrite table table2 select a,b,c from table1; 调用。 背后发生的事情是它只是从struct所有值。 您可以编写通用UDF来使用struct ref: http : //www.dataiku.com/blog/2013/05/01/a-complete-guide-to-writing-hive-udf.html 更快捷的方式: 如果你的目的是从struct获取价值并将其存储为原始值,那么尝试如下, cre
...