python中elasticsearch_dsl模块用法详解
elasticsearch-dsl是基于elasticsearch-py封装实现的，提供了更简便的操作elasticsearch的方法。
安装：
pip install elasticsearch_dsl
连接elasticsearch
from elasticsearch_dsl import connections, Search
es = connections.create_connection(hosts=["127.0.0.1:9200"], timeout=20)
print(es)
还可以通过alias给连接设置别名，后续可以通过别名来引用该连接，默认别名为default。
from elasticsearch_dsl import connections, Search
# 方式一：连接es
es = connections.create_connection(hosts=["127.0.0.1:9200"], timeout=20)
print(es)
# 方式二：连接es
connections.create_connection(alias="my_new_connection", hosts=["127.0.0.1:9200"], timeout=20)
elasticsearch_dsl.Search
search对象代表整个搜索请求，包括：queries、filters、aggregations、sort、pagination、additional parameters、associated client。
API被设置为可链接的即和用.连续操作。search对象是不可变的，除了聚合，对对象的所有更改都将导致创建包含该更改的浅表副本。
当初始化Search对象时，传递elasticsearch客户端作为using的参数
示例代码1：
from elasticsearch_dsl import connections, Search
# 方式一：连接es
es = connections.create_connection(hosts=["127.0.0.1:9200"], timeout=20)
# print(es)
# 方式二：连接es
connections.create_connection(alias="my_new_connection", hosts=["127.0.0.1:9200"], timeout=20)
# 不使用别名使用
res = Search(using=es).index("test_index").query()
# print(res)
for data in res:
print(data.to_dict())
print("*" * 100)
# 使用别名后这样使用
res2 = Search(using="my_new_connection").index('test_index').query()
# print(e)
for data in res2:
print(data.to_dict())
运行结果：
示例代码2：
from elasticsearch_dsl import connections, Search
# 方式一：连接es
es = connections.create_connection(hosts=["127.0.0.1:9200"], timeout=20)
# print(es)
# 不使用别名使用
res = Search(using=es).index("test_index").query()
# print(res)
for data in res:
print(data.to_dict())
print("*" * 100)
# 书写方式一：按条件查询数据
res2 = Search(using=es).index("test_index").query("match", name="张三") # 查询时注意分词器的使用
for data in res2:
print(data.to_dict())
print("*" * 100)
# 书写方式二：按条件查询数据
res3 = Search(using=es).index("test_index").query({"match": {"name": "张三"}})
for data in res3:
print(data.to_dict())
运行结果：
在上述执行execute方法将请求发送给elasticsearch：
response = res.execute()不需要执行execute()方法，迭代后可以通过to_dict()方法将Search对象序列化为一个dict对象，这样可以方便调试。
query方法
查询，参数可以是Q对象，也可以是query模块中的一些类，还可以是自已写上如何查询。
示例代码1：
from elasticsearch_dsl import connections, Search, Q
import time
# 方式一：连接es
es = connections.create_connection(hosts=["127.0.0.1:9200"], timeout=20)
# print(es)
res = Search(using=es, index="test_index").query().query() # 当调用.query()方法多次时，内部会使用&操作符
print(res.to_dict())
运行结果：
filter方法
在过滤上下文中添加查询，可以使用filter()函数来使之变的简单。
示例代码1：
from elasticsearch_dsl import connections, Search, Q
# 方式一：连接es
es = connections.create_connection(hosts=["127.0.0.1:9200"], timeout=20)
# print(es)
# res = Search(using=es).index("test_index").filter({"match": {"name": "北"}})
# res = Search(using=es).index("test_index").filter("terms", tags=["name", "id"])
res = Search(using=es).index("test_index").query("bool", filter=[
Q("terms", tags=["name", "id"])]) # 上面代码在背后会产生一个bool查询，并将指定的条件查询放入到filter分支
print(res)
for data in res:
print(data.to_dict())
示例代码2：
from elasticsearch_dsl import connections, Search, Q
import time
# 方式一：连接es
es = connections.create_connection(hosts=["127.0.0.1:9200"], timeout=20)
# print(es)
# 范围查询
# res = Search(using=es, index="test_index").filter("range", timestamp={"gte": 0, "lt": time.time()}).query({"match": {"name": "北"}})
res = Search(using=es, index="test_index").filter("range", id={"gte": 1, "lte": 4}).query({"match": {"name": "北"}})
print(res)
for data in res:
print(data.to_dict())
# 普通过滤
res2 = Search(using=es, index="test_index").filter("terms", id=["2", "4"]).execute()
print(res2)
for data in res2:
print(data.to_dict())
运行结果：
示例代码3：
from elasticsearch_dsl import connections, Search, Q
# 方式一：连接es
es = connections.create_connection(hosts=["127.0.0.1:9200"], timeout=20)
# print(es)
# 方式一
q = Q('range', age={"gte": 25, "lte": 27})
res = Search(using=es, index="account_info").query(q)
print(res.to_dict())
for data in res:
print(data.to_dict())
print("*" * 100)
# 方式二
q2 = Q('range', **{"age": {"gte": 25, "lte": 27}})
res2 = Search(using=es, index="account_info").query(q2)
print(res2.to_dict())
for data in res2:
print(data.to_dict())
运行结果：
index方法
指定索引
usring方法
指定哪个elasticsearch
elasticsearch_dsl.query
该库为所有的Elasticsearch查询类型都提供了类。以关键字参数传递所有的参数，最终会把参数序列化后传递给Elasticsearch，这意味着在原始查询和它对应的dsl之间有这一个清理的一对一的映射。
示例代码：
from elasticsearch_dsl import connections, Search, Q
from elasticsearch_dsl.query import MultiMatch, Match
# 方式一：连接es
es = connections.create_connection(hosts=["127.0.0.1:9200"], timeout=20)
# print(es)
# 相对与{"multi_match": {"query": "ha", "fields": ["firstname", "lastname"]}}
m1 = MultiMatch(query="Ha", fields=["firstname", "lastname"])
res = Search(using=es, index="test_index").query(m1)
print(res)
for data in res:
print(data.to_dict())
# 相当于{"match": {"firstname": {"query": "Hughes"}}}
m2 = Match(firstname={"query": "Hughes"})
res = Search(using=es, index="test_index").query(m2)
print(res)
for data in res:
print(data.to_dict())
elasticsearch_dsl.Q
使用快捷方式Q通过命名参数或者原始dict类型数据来构建一个查询实例。Q的格式一般是Q("查询类型", 字段="xxx")或Q("查询类型", query="xxx", fields=["字段1", "字段2"])
示例代码1：
from elasticsearch_dsl import connections, Search, Q
from elasticsearch_dsl.query import MultiMatch, Match
# 方式一：连接es
es = connections.create_connection(hosts=["127.0.0.1:9200"], timeout=20)
# print(es)
# q = Q("match", city="Summerfield")
q = Q("multi_match", query="Summerfield", fields=["city", "firstname"])
res = Search(using=es, index="test_index").query(q)
print(res)
for data in res:
print(data.to_dict())
查询对象可以通过逻辑运算符组合起来：
Q("match", title="python") | Q("match", title="django")
# {"bool": {"should": [...]}}
Q("match", title="python") & Q("match", title="django")
# {"bool": {"must": [...]}}
~Q("match", title="python")
# {"bool": {"must_not": [...]}}示例代码2：
from elasticsearch_dsl import connections, Search, Q
# 方式一：连接es
es = connections.create_connection(hosts=["127.0.0.1:9200"], timeout=20)
# print(es)
# q = Q("multi_match", query="123.244.101.255", fields=["clientip", "timestamp"])
q = Q('match', name='张') | Q("match", name="北")
res = Search(using=es, index="test_index").query(q)
# print(res)
for data in res:
print(data.to_dict(), data.name)
print("*" * 100)
q = Q('match', name='张') & Q("match", name="北")
res = Search(using=es, index="test_index").query(q)
# print(res)
for data in res:
print(data.to_dict(), data.name)
print("*" * 100)
q = ~Q('match', name='张')
res = Search(using=es, index="test_index").query(q)
# print(res)
for data in res:
print(data.to_dict(), data.name)
运行结果：
示例代码3：
from elasticsearch_dsl import connections, Search, Q
# 连接es
es = connections.create_connection(hosts=["127.0.0.1:9200"], timeout=20)
# print(es)
s = Search(using=es, index="account_info")
# constant_score内置属性
q = Q({"constant_score": {"filter": {"term": {"age": 25}}}})
res = s.query(q).execute()
for hit in res:
print(hit.to_dict())
print("*" * 100)
q2 = Q("bool", must=[Q("match", address="山")], should=[Q("match", gender="男"), Q("match", emplyer="AAA")], minimum_should_match=1)
res2 = s.query(q2).execute()
for hit in res2:
print(hit.to_dict())
运行结果：
嵌套类型
有时候需要引用一个在其他字段中的字段，例如多字段（title.keyword）或者在一个json文档中的address.city。为了方便，Q允许你使用双下划线‘__’代替关键词参数中的‘.’
示例代码：
from elasticsearch_dsl import connections, Search, Q
# 方式一：连接es
es = connections.create_connection(hosts=["127.0.0.1:9200"], timeout=20)
# print(es)
# res = Search(using=es, index="test_index").query("match", address__city="北京")
res = Search(using=es, index="test_index").filter("term", address__city="北京")
# print(res)
for data in res:
print(data.to_dict(), data.name)
查询
示例代码：
from elasticsearch_dsl import Search
from elasticsearch import Elasticsearch
# 连接es
es = Elasticsearch(hosts=["127.0.0.1:9200"], sniffer_timeout=60, timeout=30)
# 获取es中所有的索引
# 返回类型为字典，只返回索引名
index_name = es.cat.indices(format="json", h="index")
print(index_name)
# 查询多个索引
es_multi_index = Search(using=es, index=["personal_info_5000000", "grade", "test_index"])
print(es_multi_index.execute())
# 查询一个索引
es_one_index = Search(using=es, index="test_index")
print(es_one_index.execute())
print("*" * 100)
# 条件查询1
es_search1 = es_one_index.filter("range", id={"gte": 1, "lt": 5})
print(es_search1.execute())
# 条件查询2
es_search2 = es_one_index.filter("term", name="张")
print(es_search2.execute())
print("*" * 100)
# 结果转换为字典
es_search3 = es_search2.to_dict()
print(es_search3)
es_search4 = es_search2.execute().to_dict()
print(es_search4)
运行结果：
排序
示例代码：
from elasticsearch_dsl import connections, Search, A
# 连接es
es = connections.create_connection(hosts=["127.0.0.1:9200"], timeout=20)
# print(es)
s = Search(using=es, index="account_info")
res = s.query().sort('-age').execute()
# print(res)
for data in res:
print(data.to_dict())
运行结果：
分页
要指定from、size
示例代码：
from elasticsearch_dsl import connections, Search, A
# 连接es
es = connections.create_connection(hosts=["127.0.0.1:9200"], timeout=20)
# print(es)
s = Search(using=es, index="account_info")
res = s.query()[2: 5].execute() # {"from": 2, "size": 5}
# print(res)
for data in res:
print(data.to_dict())
运行结果：
要访问匹配的所有文档，可以使用scan()函数，scan()函数使用scan、scroll elasticsearch API，需要注意的是这种情况下结果是不会被排序的。
示例代码：
from elasticsearch_dsl import connections, Search
# 连接es
es = connections.create_connection(hosts=["127.0.0.1:9200"], timeout=20)
# print(es)
s = Search(using=es, index="account_info")
res = s.query()
# print(res)
for hit in res.scan():
print(hit.age, hit.address)
运行结果：
聚合
使用A快捷方式来定义一个聚合。为了实现聚合嵌套，你可以使用.bucket()、.metirc()以及.pipeline()方法。
bucket 即为分组，其中第一个参数是分组的名字，自己指定即可，第二个参数是方法，第三个是指定的field。
metric 也是同样，metric的方法有sum、avg、max、min等等，但是需要指出的是有两个方法可以一次性返回这些值，stats和extended_stats，后者还可以返回方差等值。
示例代码1：
from elasticsearch_dsl import connections, Search, A
# 连接es
es = connections.create_connection(hosts=["127.0.0.1:9200"], timeout=20)
# print(es)
s = Search(using=es, index="account_info")
a = A("terms", field="gender")
s.aggs.bucket("gender_terms", a)
res = s.execute()
# print(res)
for hit in res.aggregations.gender_terms:
print(hit.to_dict())
运行结果：
示例代码2：
from elasticsearch_dsl import connections, Search, A
# 连接es
es = connections.create_connection(hosts=["127.0.0.1:9200"], timeout=20)
# print(es)
s = Search(using=es, index="account_info")
s.aggs.bucket("per_gender", "terms", field="gender")
s.aggs["per_gender"].metric("sum_age", "sum", field="age")
s.aggs["per_gender"].bucket("terms_balance", "terms", field="balance")
res = s.execute()
# print(res)
for hit in res.aggregations.per_gender:
print(hit.to_dict())
运行结果：
示例代码3：
from elasticsearch_dsl import connections, Search, Q
# 连接es
es = connections.create_connection(hosts=["127.0.0.1:9200"], timeout=20)
# print(es)
s = Search(using=es, index="account_info")
res = s.aggs.bucket("aaa", "terms", field="gender").metric("avg_age", "avg", field="age")
print(res.to_dict())
运行结果：
示例代码4： 【聚合，内置排序】
from elasticsearch_dsl import connections, Search, Q
# 连接es
es = connections.create_connection(hosts=["127.0.0.1:9200"], timeout=20)
# print(es)
"""
{
'terms': {
'field': 'age',
'order': {
'_count': 'desc'
}
}
}
"""
s = Search(using=es, index="account_info")
res = s.aggs.bucket("agg_age", "terms", field="age", order={"_count": "desc"})
print(res.to_dict())
response = s.execute()
for hit in response.aggregations.agg_age:
print(hit.to_dict())
"""
{
'terms': {
'field': 'age',
'order': {
'_count': 'asc'
}
},
'aggs': {
'avg_age': {
'avg': {
'field': 'age'
}
}
}
}
"""
s2 = Search(using=es, index="account_info")
res2 = s2.aggs.bucket("agg_age", "terms", field="age", order={"_count": "asc"}).metric("avg_age", "avg", field="age")
print(res2.to_dict())
response = s2.execute()
for hit in response.aggregations.agg_age:
print(hit.to_dict())
运行结果：
示例代码5：
from elasticsearch_dsl import connections, Search, Q
# 连接es
es = connections.create_connection(hosts=["127.0.0.1:9200"], timeout=20)
# print(es)
"""
{
'aggs': {
'avg_age': {
'avg': {
'field': 'age'
}
}
}
}
"""
s = Search(using=es, index="account_info").query("range", age={"gte": 28})
res = s.aggs.metric("avg_age", "avg", field="age")
print(res.to_dict())
response = s.execute()
print(response)
for hit in response:
print(hit.to_dict())
运行结果：
高亮显示
示例代码： 【目前似乎没有效果，待验证】
from elasticsearch_dsl import connections, Search, Q
# 连接es
es = connections.create_connection(hosts=["127.0.0.1:9200"], timeout=20)
# print(es)
s = Search(using=es, index="test_index")
res = s.highlight("id").execute().to_dict()
print(res)
运行结果：
source限制返回字段
示例代码：
from elasticsearch_dsl import connections, Search, Q
# 连接es
es = connections.create_connection(hosts=["127.0.0.1:9200"], timeout=20)
# print(es)
s = Search(using=es, index="account_info")
res = s.query().execute()
for hit in res:
print(hit.to_dict())
# 限制返回字段
s2 = Search(using=es, index="account_info")
res2 = s2.query().source(['account_number', 'address']).execute()
for hit in res2:
print(hit.to_dict())
运行结果：
删除
调用Search对象上的delete方法而不是execute来实现删除匹配查询的文档
示例代码：
from elasticsearch_dsl import connections, Search, Q
# 连接es
es = connections.create_connection(hosts=["127.0.0.1:9200"], timeout=20)
# print(es)
s = Search(using=es, index="test_index")
res = s.query("match", name="张").delete()
print(res)
运行结果：
案例分析
创建索引：
from elasticsearch_dsl import Search
from elasticsearch import Elasticsearch
# 连接es
es = Elasticsearch(hosts=["127.0.0.1:9200"], sniffer_timeout=60, timeout=30)
body = {
"mappings": {
"properties": {
"account_number": {
"type": "integer"
},
"balance": {
"type": "integer"
},
"firstname": {
"type": "text"
},
"lastname": {
"type": "text"
},
"age": {
"type": "integer"
},
"gender": {
"type": "keyword"
},
"address": {
"type": "text"
},
"employer": {
"type": "text"
},
"email": {
"type": "text"
},
"province": {
"type": "text"
},
"state": {
"type": "text"
}
}
}
}
# 创建 index
es.indices.create(index="account_info", body=body)
查看索引：
使用kibana批量生成数据：
POST account_info/_bulk
{"index": {"_index":"account_info"}}
{"account_number":1,"balance":20,"firstname":"三","lastname":"张","age":25,"gender":"男","address":"北京朝阳","employer":"AAA","email":"[email protected]","province":"北京","state":"正常"}
{"index": {"_index":"account_info"}}
{"account_number":2,"balance":70,"firstname":"二","lastname":"张","age":26,"gender":"男","address":"北京海淀","employer":"AAA","email":"[email protected]","province":"北京","state":"正常"}
{"index": {"_index":"account_info"}}
{"account_number":3,"balance":80,"firstname":"四","lastname":"张","age":27,"gender":"女","address":"辽宁朝阳","employer":"BBB","email":"[email protected]","province":"辽宁","state":"正常"}
{"index": {"_index":"account_info"}}
{"account_number":4,"balance":60,"firstname":"五","lastname":"张","age":28,"gender":"男","address":"山东青岛","employer":"AAA","email":"[email protected]","province":"山东","state":"正常"}
{"index": {"_index":"account_info"}}
{"account_number":5,"balance":40,"firstname":"六","lastname":"张","age":29,"gender":"女","address":"山东济南","employer":"AAA","email":"[email protected]","province":"山东","state":"正常"}
{"index": {"_index":"account_info"}}
{"account_number":6,"balance":50,"firstname":"七","lastname":"张","age":30,"gender":"男","address":"河北唐山","employer":"BBB","email":"[email protected]","province":"河北","state":"正常"}
{"index": {"_index":"account_info"}}
{"account_number":7,"balance":30,"firstname":"一","lastname":"张","age":31,"gender":"女","address":"河北石家庄","employer":"AAA","email":"[email protected]","province":"河北","state":"正常"}查看生成的数据：
根据条件查询：
1.查询balance在40~70的信息
from elasticsearch_dsl import connections, Search, Q
# 连接es
es = connections.create_connection(hosts=["127.0.0.1:9200"], timeout=20)
# print(es)
s = Search(using=es, index="account_info")
# 查询balance在40~70的信息
q = Q("range", balance={"gte": 40, "lte": 70})
res = s.query(q)
for data in res:
print(data.to_dict())
print("共查到%d条数据" % res.count())
2.查询balance在40~70的男性信息
from elasticsearch_dsl import connections, Search, Q
# 连接es
es = connections.create_connection(hosts=["127.0.0.1:9200"], timeout=20)
# print(es)
s = Search(using=es, index="account_info")
# 查询balance在40~70的信息
q1 = Q("range", balance={"gte": 40, "lte": 70})
# 男性
q2 = Q("term", gender="男")
# and
q = q1 & q2
res = s.query(q)
for data in res:
print(data.to_dict())
print("共查到%d条数据" % res.count())
3.省份为北京、25或30岁的男性信息
from elasticsearch_dsl import connections, Search, Q
# 连接es
es = connections.create_connection(hosts=["127.0.0.1:9200"], timeout=20)
# print(es)
s = Search(using=es, index="account_info")
# 方式一：
# 省份为北京
q1 = Q("match", province="北京")
# 25或30岁的男性信息
q2 = Q("bool", must=[Q("terms", age=[25, 30]), Q("term", gender="男")])
# and
q = q1 & q2
res = s.query(q)
for data in res:
print(data.to_dict())
print("共查到%d条数据" % res.count())
print("*" * 100)
# 方式二
# 省份为北京
q1 = Q("match", province="北京")
# 25或30岁的信息
# q2 = Q("bool", must=[Q("terms", age=[25, 30]), Q("term", gender="男")])
q2 = Q("term", age=25) | Q("term", age=30)
# 男性
q3 = Q("term", gender="男")
res = s.query(q1).query(q2).query(q3) # 多次query就是& ==> and 操作
for data in res:
print(data.to_dict())
print("共查到%d条数据" % res.count())
4.地址中有“山”字，年龄不在25~28岁的女性信息
from elasticsearch_dsl import connections, Search, Q
# 连接es
es = connections.create_connection(hosts=["127.0.0.1:9200"], timeout=20)
# print(es)
s = Search(using=es, index="account_info")
# 地址中有“山”字且为女性
q1 = Q("match", address="山") & Q("match", gender="女")
# 年龄在25~28岁
q2 = ~Q("range", age={"gte": 25, "lte": 28})
# 使用filter过滤
# query和filter的前后关系都行
res = s.filter(q2).query(q1)
for data in res:
print(data.to_dict())
print("共查到%d条数据" % res.count())
5.根据年龄进行聚合，然后计算每个年龄的评价balance数值
示例代码：
from elasticsearch_dsl import connections, Search, A
# 连接es
es = connections.create_connection(hosts=["127.0.0.1:9200"], timeout=20)
# print(es)
s = Search(using=es, index="account_info")
# 先用年龄聚合，然后拿到返平均数
# size指定最大返回多少条数据，默认10条
# 实质上account的数据中，age分组没有100个这么多
q = A("terms", field="age", size=100).metric("age_per_balance", "avg", field="balance")
s.aggs.bucket("res", q)
# 执行并拿到返回值
response = s.execute()
# res是bucket指定的名字
# response.aggregations.to_dict是一个{'key': 25, 'doc_count': 1, 'age_per_balance': {'value': 20.0}}的数据，和用restful查询的一样
for data in response.aggregations.res:
print(data.to_dict())
运行结果：
6.根据年龄聚合，求25~28岁不同性别的balance值。
示例代码：
from elasticsearch_dsl import connections, Search, A
# 连接es
es = connections.create_connection(hosts=["127.0.0.1:9200"], timeout=20)
# print(es)
s = Search(using=es, index="account_info")
# 这次就用这种方法
# range 要注意指定ranges参数和from to
a1 = A("range", field="age", ranges={"from": 25, "to": 28})
a2 = A("terms", field="gender")
a3 = A("avg", field="balance")
s.aggs.bucket("res", a1).bucket("gender_group", a2).metric("balance_avg", a3)
# 执行并拿到返回值
response = s.execute()
# res是bucket指定的名字
for data in response.aggregations.res:
print(data.to_dict())
运行结果： 【注意：不包含年龄28的值】
总结：
- 假如是数组，如：bool的must、terms，那么就要
字段=[ ]- 假如是字典，如：range，那么就要
字段={xxx: yyy, .... }- 假如是单值，如：term、match，那么就要
字段=值- 假如查的是多个字段，如：multi_mathc，那么就要加上
query="要查的值", fields=["字段1", "字段2", ...]- 然后各个条件的逻辑关系，可以通过多次query和filter或直接用Q("bool", must=[Q...], should=[Q...])再加上& | ~表示
参考博文：python elasticsearch_dsl模块-云海天教程