Python-ElasticSearch客户端的封装（聚合查询、统计查询、全量数据）

---

ES Python客户端介绍

官方提供了两个客户端elasticsearch、elasticsearch-dsl

pip install elasticsearch

pip install elasticsearch-dsl

第二个是对第一个的封装，类似ORM操作数据库，可以.filter、.groupby，个人感觉很鸡肋，star数也不多。平时使用的时候一般会在kibana上测试，然后直接把query拷贝过来获取更多数据，所以这里做下第一个的封装。

封装代码

1. 封装后依然暴露了es，方便有特殊情况下使用
2. index一般很少改动，就直接放到对象中了，可以使用set_index修改
3. 常用的应该是get_doc和get_doc_scroll来获取少量和全量数据

代码测试时使用的是7.17.12版本，大于此版本可能由于官方改动出异常

pip install elasticsearch==7.17.12

es.py

import random
import string
from elasticsearch import Elasticsearch
from elasticsearch.helpers import bulk
from typing import List,Dict


class ESClient:

    def __init__(self, host="127.0.0.1",index="", http_auth = None):
        self.index = index
        if http_auth is None:
            self.es = Elasticsearch(hosts=host)
        else:
            self.es = Elasticsearch(hosts=host, http_auth=http_auth)
        print("success to connect " + host)

    def close(self):
        self.es.close()

    # 设置索引
    def set_index(self,index:str):
        self.index = index

    # 创建索引
    def create_index(self, index_name: str, mappings=None):
        res = self.es.indices.create(index=index_name, mappings=mappings)
        return res

    # 删除索引
    def delete_index(self, index_name: str):
        res = self.es.indices.delete(index=index_name)
        return res

    # 获取索引
    def get_index(self, index_name: str):
        res = self.es.indices.get(index=index_name)
        return res

    # 创建文档（单个）
    def create_doc(self,body, _id=''.join(random.sample(string.ascii_letters+string.ascii_uppercase+string.digits,20))):
        res = self.es.create(index=self.index, body=body, id=_id)
        return res

    # 创建文档(批量)
    def create_doc_bulk(self, docs: List[Dict]):
        actions = []
        for doc in docs:
            action = {
                "_index": self.index,
                "_op_type": "create",
                "_id": ''.join(random.sample(string.ascii_letters+string.ascii_uppercase+string.digits,20))
            }
            for k,v in doc.items():
                action[k] = v
            actions.append(action)
        res = bulk(client=self.es, actions=actions)
        return res

    # 删除文档
    def delete_doc(self, doc_id):
        res = self.es.delete(index=self.index, id=doc_id)
        return res

    # 更新文档
    def update_doc(self, doc_id, doc:Dict):
        body = {
            "doc" : doc
        }
        res = self.es.update(index=self.index, id=doc_id, body=body)
        return res

    # 分页获取超过100000的文档
    def get_doc_scroll(self,query:Dict):
        res = self.es.search(index=self.index,size=10000,body=query,search_type="query_then_fetch",scroll="5m")
        data_list = []
        hits = res.get("hits")
        scroll_id = res.get('_scroll_id')
        total_value = 0
        # total 可能为Dict或int
        if isinstance(hits.get('total'),Dict):
            total_value= hits.get('total').get('value')
        else:
            total_value = hits.get('total')

        if total_value>0:
            for data in hits.get('hits'):
                data_list.append(data.get('_source'))
        return scroll_id,data_list

    # 通过scroll_id分页获取后序文档
    def get_doc_by_scroll_id(self,scroll_id):
        page = self.es.scroll(scroll_id=scroll_id,scroll="5m")
        data_list = []
        scroll_id = page.get('_scroll_id')
        for data in page.get('hits').get('hits'):
            data_list.append(data)
        return scroll_id,data_list

    # 清空scroll_id，防止服务端不够用
    def clear_scroll(self,scroll_id):
        self.es.clear_scroll(scroll_id)

    # 获取索引的hits内容（一般用于获取文档id、总数）
    def get_doc_all(self):
        res = self.es.search(index=self.index)
        return res['hits']

    # 获取一个文档
    def get_doc_by_id(self, id_):
        res = self.es.get(index=self.index, id=id_)
        return res["_source"]

    # 获取所有文档的_source内容(小于100000)
    def get_doc(self,query:Dict,size:int=100000):
        query['size'] = size
        res = self.es.search(index=self.index,body=query)
        data_list = []
        hits = res.get("hits")
        total_value = 0
        # total 可能为Dict或int
        if isinstance(hits.get('total'), Dict):
            total_value = hits.get('total').get('value')
        else:
            total_value = hits.get('total')

        if total_value > 0:
            for data in hits.get('hits'):
                data_list.append(data.get('_source'))
        return data_list

    # 聚合查询（分组条件名为group_by,返回buckets）
    def get_doc_agg(self, query):
        res = self.es.search(index=self.index, body=query)
        return res['aggregations']['group_by'].get('buckets')

    # 统计查询（统计条件为stats_by，返回最值、平均值等）
    def get_doc_stats(self,query):
        res = self.es.search(index=self.index,body=query)
        return res['aggregations']["stats_by"]

测试代码

import unittest
from es import ESClient

cli = ESClient(host="http://10.28.144.3:9200",http_auth=["elastic","changeme"])
def test_create_index():
    res = cli.create_index(index_name="test")
    print(res)

def test_delete_index():
    res = cli.delete_index(index_name="test")
    print(res)

def test_get_index():
    res = cli.get_index(index_name="test")
    print(res)

def test_set_index():
    cli.set_index(index="test")

def test_create_doc():
    body = {
        "name": "lady_killer9",
        "age": 19
    }
    res = cli.create_doc(body=body)
    print(res)

def test_create_doc_bulk():
    from copy import deepcopy
    body = {
        "name": "lady_killer9"
    }
    users = []
    for i in range(100001):
        tmp = deepcopy(body)
        tmp["age"] = i
        users.append(tmp)
    res = cli.create_doc_bulk(docs=users)
    print(res)


def test_get_doc_all():
    res = cli.get_doc_all()
    print(res)


def test_get_doc_by_id():
    res = cli.get_doc_by_id("jHALXDQaENQZPM4C9EUt")
    print(res)

def test_get_doc():
    query = {
        "query": {
            "match_all": {

            }
        }
    }
    res = cli.get_doc(query=query,size=20)
    print(res)

def test_update_doc():
    body={
        "name": "lady_killer_after_update"
    }
    res = cli.update_doc(doc_id="jHALXDQaENQZPM4C9EUt",doc=body)
    print(res)


def test_delete_doc():
    res = cli.delete_doc(doc_id="jHALXDQaENQZPM4C9EUt")
    print(res)

def test_get_doc_agg():
    query = {
            "aggs": {
                "group_by": {
                    "terms": {
                        "field": "age"
                    }
                }
            }
    }
    res = cli.get_doc_agg(query=query)
    print(res)

def test_get_doc_stats():
    query = {
            "aggs": {
                "stats_by": {
                    "stats": {
                        "field": "age"
                    }
                }
            }
    }
    res = cli.get_doc_stats(query=query)
    print(res)

def test_get_doc_scroll():
    query = {
        "query": {
            "match_all": {}
        }
    }
    scroll_id,data_list = cli.get_doc_scroll(query=query)
    res = []
    while data_list:
        res.extend(data_list)
        scroll_id,data_list = cli.get_doc_by_scroll_id(scroll_id=scroll_id)
    print(len(res))


if __name__ == '__main__':
    # test_delete_index()
    test_create_index()
    test_get_index()
    # test_set_index()
    # test_create_doc()
    # test_create_doc_bulk()
    # test_get_doc_all()
    # test_update_doc()
    # test_get_doc_by_id()
    # test_get_doc()
    # test_delete_doc()
    # test_get_doc_agg()
    # test_get_doc_stats()
    # test_get_doc_scroll()
    cli.close()

测试截图

更多python相关内容：【python总结】python学习框架梳理

本人b站账号：一路狂飚的蜗牛

有问题请下方评论，转载请注明出处，并附有原文链接，谢谢！如有侵权，请及时联系。如果您感觉有所收获，自愿打赏，可选择支付宝18833895206（小于），您的支持是我不断更新的动力。

参考

github-elasticsearch
github-elasticsearch-dsl