hive 学习系列五（hive 和elasticsearch 的交互，很详细哦，我又来吹liubi了）

Posted 2020-12-25 unnunique

hive 操作elasticsearch

一，从hive 表格向elasticsearch 导入数据

1，首先，创建elasticsearch 索引，索引如下

curl -XPUT ‘10.81.179.209:9200/zebra_info_demo?pretty‘ -H ‘Content-Type: application/json‘ -d‘
{
    "settings": {
        "number_of_shards":5,
        "number_of_replicas":2
    },
    "mappings": {
         "zebra_info": {
              "properties": {
                    "name" : {"type" : "text"},
                    "type": {"type": "text"},
                    "province": {"type": "text"},
                    "city": {"type": "text"},
                    "citycode": {"type": "text", "index": "no"},
                    "district": {"type": "text"},
                    "adcode": {"type": "text", "index": "no"},
                    "township": {"type": "text"},
                    "bausiness_circle": {"type": "text"},
                    "formatted_address": {"type": "text"},
                    "location": {"type": "geo_point"},
                    "extensions": {
                      "type": "nested",
                      "properties": {
                        "map_lat": {"type": "double", "index": "no"},
                        "map_lng": {"type": "double", "index": "no"},
                        "avg_price": {"type": "double", "index": "no"},
                        "shops": {"type":"short", "index": "no"},
                        "good_comments": {"type":"short", "index": "no"},
                        "lvl": {"type":"short", "index": "no"},
                        "leisure_type": {"type": "text", "index": "no"},
                        "fun_type": {"type": "text", "index": "no"},
                        "numbers": {"type": "short", "index": "no"}
                       }
                   }
             }
        }
    }
}
‘

2，查看elasticsearch版本，下载相应的elasticsearch-hive-hadoop jar 包

可以用如下命令查看elastic search 的版本
本文版本5.6.9

到如下maven 官网下载jar 包。
https://repo.maven.apache.org/maven2/org/elasticsearch/elasticsearch-hadoop-hive/
选择正确的版本即可。

3， 把下载下来的jar 包上传到hdfs 路径下。

本文jar 包路径，hdfs:///udf/elasticsearch-hadoop-hive-5.6.9.jar

4，哦了，建表，用起来

DELETE jars;
add jar hdfs:///udf/elasticsearch-hadoop-hive-5.6.9.jar;
drop table zebra_info_demo;
CREATE EXTERNAL  TABLE zebra_info_demo(
name string,
`type` string,
province double,
city string,
citycode string,
district string,
adcode string,
township string,
business_circle string,
formatted_address string,
location string,
extensions STRUCT<map_lat:double, map_lng:double, avg_price:double, shops:smallint, good_comments:smallint, lvl:smallint, leisure_type:STRING, fun_type:STRING, numbers:smallint>
)
STORED BY ‘org.elasticsearch.hadoop.hive.EsStorageHandler‘ 
TBLPROPERTIES(‘es.nodes‘ = ‘10.81.179.209:9200‘,
‘es.index.auto.create‘ = ‘false‘,
‘es.resource‘ = ‘zebra_info_demo/zebra_info‘,
‘es.read.metadata‘ = ‘true‘,
‘es.mapping.names‘ = ‘name:name, type:type, province:province, city:city, citycode:citycode, district:district, adcode:adcode, township:township, business_circle:business_circle, formatted_address:formatted_address, location:location, extensions:extensions‘);  

5, 往里面填充数据，就O了。

INSERT INTO TABLE zebra_info_demo
SELECT 
a.name,
a.brands,
a.province,
a.city,
null as citycode,
null as district,
null as adcode,
null as township,
a.business_circle,
null as formatted_address,
concat(a.map_lat, ‘, ‘, a.map_lng) as `location`,
named_struct(‘map_lat‘, cast(a.map_lat as double), ‘map_lng‘,cast(a.map_lng as double) ,‘avg_price‘, cast(0 as DOUBLE), ‘shops‘, 0S,  ‘good_comments‘, 0S, ‘lvl‘, cast(a.lv1 as SMALLINT), ‘leisure_type‘, ‘‘, ‘fun_type‘, ‘‘, ‘numbers‘, 0S) as extentions
from medicalsite_childclinic a;

运行结果：

二，已知elasticsearch 索引，然后，建立hive 表格和elasticsearch 进行交互。可以join 哦，一个字，liubi

1,先看一下索引和数据

已知索引如下：

curl -XPUT  ‘10.81.179.209:9200/join_tests?pretty‘ -H ‘Content-Type: application/json‘ -d‘
{
  "mappings": {
    "cities": {
      "properties": {
        "province": {
          "type": "string"
        },
        "city": {
          "type": "string"
        }
      }
    }
    }
  }
}
‘

curl -XPUT  ‘10.81.179.209:9200/join_tests1?pretty‘ -H ‘Content-Type: application/json‘ -d‘
{
  "mappings": {
    "shop": {
      "properties":{
        "name": {
          "type": "string"
        },
        "city": {
          "type": "string"
        }
      }
    }
   }
  }
}
‘

数据如下：

2，建立表格，写一堆有毒的sql 语句。

DELETE jars;
add jar hdfs:///udf/elasticsearch-hadoop-hive-5.6.9.jar;
create table join_tests(
    province string,
    city string
)STORED BY ‘org.elasticsearch.hadoop.hive.EsStorageHandler‘ 
TBLPROPERTIES(‘es.nodes‘ = ‘10.81.179.209:9200‘,
‘es.index.auto.create‘ = ‘false‘,
‘es.resource‘ = ‘join_tests/cities‘,
‘es.read.metadata‘ = ‘true‘,
‘es.mapping.names‘ = ‘province:province, city:city‘);

create table join_tests1(
    name string,
    city string
)STORED BY ‘org.elasticsearch.hadoop.hive.EsStorageHandler‘ 
TBLPROPERTIES(‘es.nodes‘ = ‘10.81.179.209:9200‘,
‘es.index.auto.create‘ = ‘false‘,
‘es.resource‘ = ‘join_tests1/shop‘,
‘es.read.metadata‘ = ‘true‘,
‘es.mapping.names‘ = ‘name:name, city:city‘);




SELECT 
    a.province,
    b.city,
    b.name
from join_tests a LEFT JOIN join_tests1 b on a.city = b.city;

3，运行结果

结束语

推荐一个useful 的工具， apache Hue, 可以用来管理hdfs 文件，hive 操作。mysql 操作等。