有没有办法将重复项保存在 Hive 的收集集中，或者使用其他方法模拟 Hive 提供的那种聚合集合？我想将列中具有相同键的所有项目聚合到一个数组中，并带有重复项。
IE：
hash_id | num_of_cats
=====================
ad3jkfk 4
ad3jkfk 4
ad3jkfk 2
fkjh43f 1
fkjh43f 8
fkjh43f 8
rjkhd93 7
rjkhd93 4
rjkhd93 7
应该返回：
hash_agg | cats_aggregate
===========================
ad3jkfk Array<int>(4,4,2)
fkjh43f Array<int>(1,8,8)
rjkhd93 Array<int>(7,4,7)
尝试在 Hive 0.13.0 之后使用 COLLECT_LIST(col)
SELECT
hash_id, COLLECT_LIST(num_of_cats) AS aggr_set
FROM
tablename
WHERE
blablabla
GROUP BY
hash_id
;
没有内置任何东西，但是创建用户定义的函数，包括聚合，并没有那么糟糕。唯一粗糙的部分是试图使它们类型通用，但这里是一个收集示例。
package com.example;
import java.util.ArrayList;
import org.apache.hadoop.hive.ql.exec.UDFArgumentTypeException;
import org.apache.hadoop.hive.ql.metadata.HiveException;
import org.apache.hadoop.hive.ql.parse.SemanticException;
import org.apache.hadoop.hive.ql.udf.generic.AbstractGenericUDAFResolver;
import org.apache.hadoop.hive.ql.udf.generic.GenericUDAFEvaluator;
import org.apache.hadoop.hive.serde2.objectinspector.ObjectInspector;
import org.apache.hadoop.hive.serde2.objectinspector.ObjectInspectorFactory;
import org.apache.hadoop.hive.serde2.objectinspector.ObjectInspectorUtils;
import org.apache.hadoop.hive.serde2.objectinspector.PrimitiveObjectInspector;
import org.apache.hadoop.hive.serde2.objectinspector.StandardListObjectInspector;
import org.apache.hadoop.hive.serde2.typeinfo.TypeInfo;
public class CollectAll extends AbstractGenericUDAFResolver
{
@Override
public GenericUDAFEvaluator getEvaluator(TypeInfo[] tis)
throws SemanticException
{
if (tis.length != 1)
{
throw new UDFArgumentTypeException(tis.length - 1, "Exactly one argument is expected.");
}
if (tis[0].getCategory() != ObjectInspector.Category.PRIMITIVE)
{
throw new UDFArgumentTypeException(0, "Only primitive type arguments are accepted but " + tis[0].getTypeName() + " was passed as parameter 1.");
}
return new CollectAllEvaluator();
}
public static class CollectAllEvaluator extends GenericUDAFEvaluator
{
private PrimitiveObjectInspector inputOI;
private StandardListObjectInspector loi;
private StandardListObjectInspector internalMergeOI;
@Override
public ObjectInspector init(Mode m, ObjectInspector[] parameters)
throws HiveException
{
super.init(m, parameters);
if (m == Mode.PARTIAL1)
{
inputOI = (PrimitiveObjectInspector) parameters[0];
return ObjectInspectorFactory
.getStandardListObjectInspector((PrimitiveObjectInspector) ObjectInspectorUtils
.getStandardObjectInspector(inputOI));
}
else
{
if (!(parameters[0] instanceof StandardListObjectInspector))
{
inputOI = (PrimitiveObjectInspector) ObjectInspectorUtils
.getStandardObjectInspector(parameters[0]);
return (StandardListObjectInspector) ObjectInspectorFactory
.getStandardListObjectInspector(inputOI);
}
else
{
internalMergeOI = (StandardListObjectInspector) parameters[0];
inputOI = (PrimitiveObjectInspector) internalMergeOI.getListElementObjectInspector();
loi = (StandardListObjectInspector) ObjectInspectorUtils.getStandardObjectInspector(internalMergeOI);
return loi;
}
}
}
static class ArrayAggregationBuffer implements AggregationBuffer
{
ArrayList<Object> container;
}
@Override
public void reset(AggregationBuffer ab)
throws HiveException
{
((ArrayAggregationBuffer) ab).container = new ArrayList<Object>();
}
@Override
public AggregationBuffer getNewAggregationBuffer()
throws HiveException
{
ArrayAggregationBuffer ret = new ArrayAggregationBuffer();
reset(ret);
return ret;
}
@Override
public void iterate(AggregationBuffer ab, Object[] parameters)
throws HiveException
{
assert (parameters.length == 1);
Object p = parameters[0];
if (p != null)
{
ArrayAggregationBuffer agg = (ArrayAggregationBuffer) ab;
agg.container.add(ObjectInspectorUtils.copyToStandardObject(p, this.inputOI));
}
}
@Override
public Object terminatePartial(AggregationBuffer ab)
throws HiveException
{
ArrayAggregationBuffer agg = (ArrayAggregationBuffer) ab;
ArrayList<Object> ret = new ArrayList<Object>(agg.container.size());
ret.addAll(agg.container);
return ret;
}
@Override
public void merge(AggregationBuffer ab, Object o)
throws HiveException
{
ArrayAggregationBuffer agg = (ArrayAggregationBuffer) ab;
ArrayList<Object> partial = (ArrayList<Object>)internalMergeOI.getList(o);
for(Object i : partial)
{
agg.container.add(ObjectInspectorUtils.copyToStandardObject(i, this.inputOI));
}
}
@Override
public Object terminate(AggregationBuffer ab)
throws HiveException
{
ArrayAggregationBuffer agg = (ArrayAggregationBuffer) ab;
ArrayList<Object> ret = new ArrayList<Object>(agg.container.size());
ret.addAll(agg.container);
return ret;
}
}
}
然后在蜂巢中，只需发出add jar Whatever.jar;，CREATE TEMPORARY FUNCTION collect_all AS 'com.example.CollectAll';
您应该能够按预期使用它。
hive> SELECT hash_id, collect_all(num_of_cats) FROM test GROUP BY hash_id;
OK
ad3jkfk [4,4,2]
fkjh43f [1,8,8]
rjkhd93 [7,4,7]
值得注意的是，元素的顺序应该被认为是未定义的，因此如果您打算使用它来将信息输入 n_grams，您可能需要稍微扩展它以根据需要对数据进行排序。
修改了 Jeff Mc 的代码以删除输入必须是原始类型的限制（可能继承自 collect_set）。这个版本可以收集结构、映射和数组以及原语。
package com.example;
import java.util.ArrayList;
import org.apache.hadoop.hive.ql.exec.UDFArgumentTypeException;
import org.apache.hadoop.hive.ql.metadata.HiveException;
import org.apache.hadoop.hive.ql.parse.SemanticException;
import org.apache.hadoop.hive.ql.udf.generic.AbstractGenericUDAFResolver;
import org.apache.hadoop.hive.ql.udf.generic.GenericUDAFEvaluator;
import org.apache.hadoop.hive.serde2.objectinspector.ObjectInspector;
import org.apache.hadoop.hive.serde2.objectinspector.ObjectInspectorFactory;
import org.apache.hadoop.hive.serde2.objectinspector.ObjectInspectorUtils;
import org.apache.hadoop.hive.serde2.objectinspector.StandardListObjectInspector;
import org.apache.hadoop.hive.serde2.typeinfo.TypeInfo;
public class CollectAll extends AbstractGenericUDAFResolver
{
@Override
public GenericUDAFEvaluator getEvaluator(TypeInfo[] tis)
throws SemanticException
{
if (tis.length != 1)
{
throw new UDFArgumentTypeException(tis.length - 1, "Exactly one argument is expected.");
}
return new CollectAllEvaluator();
}
public static class CollectAllEvaluator extends GenericUDAFEvaluator
{
private ObjectInspector inputOI;
private StandardListObjectInspector loi;
private StandardListObjectInspector internalMergeOI;
@Override
public ObjectInspector init(Mode m, ObjectInspector[] parameters)
throws HiveException
{
super.init(m, parameters);
if (m == Mode.PARTIAL1)
{
inputOI = parameters[0];
return ObjectInspectorFactory
.getStandardListObjectInspector(ObjectInspectorUtils
.getStandardObjectInspector(inputOI));
}
else
{
if (!(parameters[0] instanceof StandardListObjectInspector))
{
inputOI = ObjectInspectorUtils
.getStandardObjectInspector(parameters[0]);
return (StandardListObjectInspector) ObjectInspectorFactory
.getStandardListObjectInspector(inputOI);
}
else
{
internalMergeOI = (StandardListObjectInspector) parameters[0];
inputOI = internalMergeOI.getListElementObjectInspector();
loi = (StandardListObjectInspector) ObjectInspectorUtils.getStandardObjectInspector(internalMergeOI);
return loi;
}
}
}
static class ArrayAggregationBuffer implements AggregationBuffer
{
ArrayList<Object> container;
}
@Override
public void reset(AggregationBuffer ab)
throws HiveException
{
((ArrayAggregationBuffer) ab).container = new ArrayList<Object>();
}
@Override
public AggregationBuffer getNewAggregationBuffer()
throws HiveException
{
ArrayAggregationBuffer ret = new ArrayAggregationBuffer();
reset(ret);
return ret;
}
@Override
public void iterate(AggregationBuffer ab, Object[] parameters)
throws HiveException
{
assert (parameters.length == 1);
Object p = parameters[0];
if (p != null)
{
ArrayAggregationBuffer agg = (ArrayAggregationBuffer) ab;
agg.container.add(ObjectInspectorUtils.copyToStandardObject(p, this.inputOI));
}
}
@Override
public Object terminatePartial(AggregationBuffer ab)
throws HiveException
{
ArrayAggregationBuffer agg = (ArrayAggregationBuffer) ab;
ArrayList<Object> ret = new ArrayList<Object>(agg.container.size());
ret.addAll(agg.container);
return ret;
}
@Override
public void merge(AggregationBuffer ab, Object o)
throws HiveException
{
ArrayAggregationBuffer agg = (ArrayAggregationBuffer) ab;
ArrayList<Object> partial = (ArrayList<Object>)internalMergeOI.getList(o);
for(Object i : partial)
{
agg.container.add(ObjectInspectorUtils.copyToStandardObject(i, this.inputOI));
}
}
@Override
public Object terminate(AggregationBuffer ab)
throws HiveException
{
ArrayAggregationBuffer agg = (ArrayAggregationBuffer) ab;
ArrayList<Object> ret = new ArrayList<Object>(agg.container.size());
ret.addAll(agg.container);
return ret;
}
}
}
从 hive 0.13 开始，有一个名为的内置 UDAFcollect_list()可以实现此目的。见这里。
查看 Brickhouse 收集 UDAF ( http://github.com/klout/brickhouse/blob/master/src/main/java/brickhouse/udf/collect/CollectUDAF.java )
它还支持收集成地图。Brickhouse 还包含许多有用的 UDF，它们不在标准 Hive 发行版中。
这是执行此工作的确切 hive 查询（仅在 hive > 0.13 中有效）：
SELECT hash_id, collect_set(num_of_cats) FROM GROUP BY hash_id;
对于它的价值（尽管我知道这是一篇较旧的帖子），Hive 0.13.0具有一个新的collect_list()函数，它不会重复数据删除。
收集结构的解决方法
假设你有一张桌子
tableWithStruct(
id string,
obj struct <a:string,b:string>)
现在创建另一个表
CREATE EXTERNAL TABLE tablename (
id string,
temp array<string>
)
ROW FORMAT DELIMITED
FIELDS TERMINATED BY '\t' COLLECTION ITEMS TERMINATED BY ',' MAP KEYS TERMINATED BY '|'
插入查询
insert into table tablename select id,collect(concat_ws('|',cast(obj.a as string),cast(obj.b as string)) from tableWithStruct group by id;
现在在与表名相同的位置创建另一个表
CREATE EXTERNAL TABLE tablename_final (
id string,
array_list array<struct<a:string,b:string>>
)
ROW FORMAT DELIMITED
FIELDS TERMINATED BY '\t' COLLECTION ITEMS TERMINATED BY ',' MAP KEYS TERMINATED BY '|'
当您从tablename_final中选择时，您将获得所需的输出
只是想知道 - 如果 n 声明 -
SELECT
hash_id, COLLECT_LIST(num_of_cats) AS aggr_set
FROM
tablename
WHERE
blablabla
GROUP BY
hash_id
;
我们想要对 num_of_cats 的元素进行排序和限制 - 怎么做？COZ 在大数据中，我们处理 PB 的数据。在这种情况下，我们可能不需要所有这些，但前 10 名或限制它。