ElasticSearch 是构建在Apache Lucene之上的的搜索引擎服务,开源(Apache2协议),分布式,RESTful。安装方便,使用简单。
官方站点:http://www.elasticsearch.com/
中文站点:http://es-cn.medcl.net/
安装
必须先安装Java环境,并设置 JAVA_HOME => C:\Program Files\Java\jdk1.6.0_18
elasticsearch-rtf 中文入门集成包 https://github.com/medcl/elasticsearch-rtf 使用git签出,下载到本地。windows下,执行bin下面的elasticsearch.bat。linux下,执行bin下面或者service下面elasticsearch。
Pyes https://github.com/aparo/pyes 更多客户端
Bottle http://bottlepy.org/docs/dev/
角色关系对照
elasticsearch 跟 MySQL 中定义资料格式的角色关系对照表如下
MySQL | helasticsearch |
---|---|
database | index |
table | type |
row | document |
field | field |
索引映射
#创建索引
$ curl -XPUT http://localhost:9200/test-index
#创建Mapping
$ curl -XPUT http://localhost:9200/test-index/test-type/_mapping -d '{
"properties" : {
"name" : { "type" : "string" }
}
}'
@route('/indexsetting/')
def indexmapping():
"""索引映射"""
conn = ES('127.0.0.1:9200')
conn.debug_dump = True
try:
#删除索引
conn.delete_index("test-index")
except:
pass
#创建索引
conn.create_index("test-index")
mapping = {
u'id': {'store': 'yes',
'type': u'integer'},
u'author': {'boost': 1.0,
'index': 'not_analyzed',
'store': 'yes',
'type': u'string'},
u'published': {'boost': 1.0,
'index': 'not_analyzed',
'store': 'yes',
'type': u'datetime'},
u'url': {'store': 'yes',
'type': u'string'},
u'title': {'boost': 1.0,
'index': 'analyzed',
'store': 'yes',
'type': u'string'},
u'content': {'boost': 1.0,
'index': 'analyzed',
'store': 'yes',
'type': u'string',
"term_vector" : "with_positions_offsets"}
}
#索引映射
conn.put_mapping("test-type", {'properties':mapping}, ["test-index"])
return "索引映射"
索引
#索引
$ curl -XPUT http://localhost:9200/test-index/test-type/1 -d '{
"user": "kimchy",
"post_date": "2009-11-15T13:12:00",
"message": "Trying out elasticsearch, so far so good?"
}'
#获取
$ curl -XGET http://localhost:9200/test-index/test-type/1
#删除
$ curl -XDELETE 'http://localhost:9200/test-index/test-type/1'
@route('/indextest/')
def indexTest():
"""索引测试"""
conn = ES('127.0.0.1:9200')
for item in Data().getData():
#添加索引
conn.index(item,"test-index", "test-type",item['id'])
#索引优化
conn.optimize(["test-index"])
#删除索引内容
conn.delete("test-index", "test-type", 2668090)
#更新索引内容
model = conn.get("test-index", "test-type", 2667371)
model["title"]="标题修改测试"
conn.update(model,"test-index", "test-type",2667371)
#刷新索引
conn.refresh(["test-index"])
q = MatchAllQuery()
results = conn.search(query = q,indices="test-index", doc_types="test-type")
# for r in results:
# print r
return template('default.tpl', list=results,count=len(results))
搜索
#lucene语法方式的查询
$ curl -XGET http://localhost:9200/test-index/test-type/_search?q=user:kimchy
#query DSL方式查询
$ curl -XGET http://localhost:9200/test-index/test-type/_search -d '{
"query" : {
"term" : { "user": "kimchy" }
}
}'
#query DSL方式查询
$ curl -XGET http://localhost:9200/test-index/_search?pretty=true -d '{
"query" : {
"range" : {
"post_date" : {
"from" : "2009-11-15T13:00:00",
"to" : "2009-11-15T14:30:00"
}
}
}
}'
#查找全部索引内容
$ curl -XGET http://localhost:9200/test-index/test-type/_search?pretty=true
@route('/search/')
@route('/search/<searchkey>')
def search(searchkey=u"关键算法"):
"""索引搜索"""
conn = ES('127.0.0.1:9200')
#TextQuery会对searchkey进行分词
qtitle = TextQuery("title", searchkey)
qcontent = TextQuery("content", searchkey)
#发布时间大于"2012-9-2 22:00:00"
qpublished=RangeQuery(ESRangeOp("published", "gt", datetime(2012, 9, 2, 22, 0, 0)))
h = HighLighter(['<b>'], ['</b>'], fragment_size=500)
#多字段搜索(must=>and,should=>or),高亮,结果截取(分页),排序
q = Search(BoolQuery(must=[qpublished],should=[qtitle,qcontent]), highlight=h, start=0, size=3, sort={'id': {'order': 'asc'}})
q.add_highlight("title")
q.add_highlight("content")
results = conn.search(query = q,indices="test-index", doc_types="test-type")
list=[]
for r in results:
if(r._meta.highlight.has_key("title")):
r['title']=r._meta.highlight[u"title"][0]
if(r._meta.highlight.has_key("content")):
r['content']=r._meta.highlight[u"content"][0]
list.append(r)
return template('search.tpl', list=list,count=results.total)
设置
#创建索引,并设置分片和副本参数
$ curl -XPUT http://localhost:9200/elasticsearch/ -d '{
"settings" : {
"number_of_shards" : 2,
"number_of_replicas" : 3
}
}'
其他
#分词
curl -XGET 'http://localhost:9200/test-index/_analyze?text=中华人民共和国'