● 在生产上,我们需要自己手动建立索引和映射,是为了更好的管理索引,就像数据库的建表数据一样。
● 创建索引的语法:
PUT /index
{
"settings":{...},
"mappings":{
"properties":{
...
}
},
"aliases":{
"default_index":{}
}
}
示例:
● 创建索引:
PUT /my_index
{
"settings": {
"number_of_shards": 1,
"number_of_replicas": 1
},
"mappings": {
"properties": {
"field1":{
"type": "text"
},
"field2":{
"type": "text"
}
}
},
"aliases": {
"default_index": {}
}
}
插入数据:
PUT /my_index/_doc/1
{
"field1":"java",
"field2":"js"
}
查询数据:
GET /my_index/_doc/1
GET /default_index/_doc/1
语法:查询索引:
GET /index
语法:查询索引中的映射信息
GET /index/_mapping
示例:
GET /my_index
GET /my_index/_mapping
GET /my_index/_settings
语法:
PUT /index/_settings
{
"index" : {
"number_of_replicas" : 5
}
}
示例:修改副本分片数
PUT /my_index/_settings
{
"index" : {
"number_of_replicas" : 5
}
}
语法:删除索引
DELETE /index
语法:删除多个索引
DELETE /index1,index2
DELETE /index*
语法:删除全部索引(危险)
DELETE /_all
示例:删除索引
DELETE /book
示例:删除多个索引
DELETE /book,my_index
示例:
DELETE /book*
示例:删除所有索引
DELETE /_all
为了安全起见,防止恶意删除索引,删除的时候必须指定索引名,我们可以在elasticsearch.yml中配置action.destructive_requires_name: true。
● 分词器有三个组件,分别为:character filter,tokenizer和token filter。
● 默认的分词器是:standard 分词器。
● standard tokenizer:以单词边界进行切分。
● standard token filter:什么都不做。
● lowercase token filter:将所有字母都转换为小写。
● stop token filter(默认被禁用):移除停用词,比如a、the、it等等。
● 启用english停用词token filter:
PUT /my_index
{
"settings": {
"analysis": {
"analyzer": {
"es_std":{
"type":"standard",
"stopwords":"_english_"
}
}
}
}
}
使用标准分词器测试分词:
GET /my_index/_analyze
{
"analyzer": "standard",
"text": "a dog is in the house"
}
使用english停用词测试分词:
GET /my_index/_analyze
{
"analyzer": "es_std",
"text":"a dog is in the house"
}
● 定制自己的分词器:
PUT /my_index
{
"settings": {
"analysis": {
"char_filter": {
//&替换为and
"&_to_and": {
"type": "mapping",
"mappings": ["&=> and"]
}
},
"filter": {
//停用词--遇到这些词语会忽略分词
"my_stopwords": {
"type": "stop",
"stopwords": ["the", "a"]
}
},
"analyzer": {
//自定义分词器
"my_analyzer": {
"type": "custom",
//字符过滤器
"char_filter": ["html_strip", "&_to_and"],
//以单词边界进行切分
"tokenizer": "standard",
//再进行过滤
"filter": ["lowercase", "my_stopwords"]
}
}
}
}
}
测试自定义分词器:
GET /my_index/_analyze
{
"analyzer": "my_analyzer",
"text": "tom&jerry are a friend in the house, <a>, HAHA!!"
}
设置字段使用自定义分词器:
PUT /my_index/_mapping/
{
"properties": {
"content": {
"type": "text",
"analyzer": "my_analyzer"
}
}
}
● type是一个Index中用来区分类似的数据的。这些类似的数据,可能有不同的fields,而且有不同的属性来控制索引和分词器的建立。
● field的value,在底层的Lucene中建立索引的时候,全都是opaque bytes类型,是不区分类型的。
● Lucene是没有type的概念的,在document(文档)中,实际上是将type作为document(文档)的field来存储的,即_type,ES通过_type来进行type的过滤和筛选。
● 一个Index中的多个type,实际上是放在一起存储的,因此同一个Index下,不能有多个type重名,因为那样是无法处理的。
● 比如:创建索引:
{
"goods": {
"mappings": {
"electronic_goods": {
"properties": {
"name": {
"type": "string",
},
"price": {
"type": "double"
},
"service_period": {
"type": "string"
}
}
},
"fresh_goods": {
"properties": {
"name": {
"type": "string",
},
"price": {
"type": "double"
},
"eat_period": {
"type": "string"
}
}
}
}
}
}
插入数据:
PUT /goods/electronic_goods/1
{
"name": "小米空调",
"price": 1999.0,
"service_period": "one year"
}
PUT /goods/fresh_goods/1
{
"name": "澳洲龙虾",
"price": 199.0,
"eat_period": "one week"
}
ES文档在底层的存储是这个样子的:
{
"goods": {
"mappings": {
"_type": {
"type": "text",
"index": "false"
},
"name": {
"type": "text"
}
"price": {
"type": "double"
}
"service_period": {
"type": "text"
},
"eat_period": {
"type": "text"
}
}
}
}
那么插入的数据在底层的数据存储格式:
{
"_type": "electronic_goods",
"name": "小米空调",
"price": 1999.0,
"service_period": "one year",
"eat_period": ""
}
{
"_type": "fresh_goods",
"name": "澳洲龙虾",
"price": 199.0,
"service_period": "",
"eat_period": "one week"
}
● 同一个索引下,不同type的数据存储其他type的field的大量空值,会造成资源浪费。
● 所以,不同类型的数据,要放在不同的索引中。
● ES9中,将彻底删除type。
语法:
PUT /index
{
"settings": {...},
"mappings": {
"dynamic": "xxx",
"properties": {
"filed1": {
"type": ""
},
"filed2": {
"type": "",
"dynamic": "xxx"
},
...
}
},
"aliases": {
"default_index": {}
}
}
dynamic:
● 示例:
● 创建索引
PUT /my_index
{
"mappings": {
"dynamic": "strict",
"properties": {
"title": {
"type": "text"
},
"address": {
"type": "object",
"dynamic": "true"
}
}
}
}
测试插入数据
PUT /my_index/_doc/1
{
"title": "my article",
"content": "this is my article",
"address": {
"province": "guangdong",
"city": "guangzhou"
}
}
如果是false情况下,content陌生字段不会被索引,我们在进行全局搜索时,不会搜索到content字段,但是如果搜索别的字段,搜索到该条记录时,还是会携带返回content字段的。
查询my_index的底层mapping映射,也可以发现其中并没有为content字段建立映射。
GET /my_index/_mapping
{
"my_index" : {
"mappings" : {
"dynamic" : "false",
"properties" : {
"address" : {
"dynamic" : "true",
"properties" : {
"city" : {
"type" : "text",
"fields" : {
"keyword" : {
"type" : "keyword",
"ignore_above" : 256
}
}
},
"province" : {
"type" : "text",
"fields" : {
"keyword" : {
"type" : "keyword",
"ignore_above" : 256
}
}
}
}
},
"title" : {
"type" : "text"
}
}
}
}
}
● 日期探测(Date Detection):默认会按照一定格式识别date,比如yyyy-MM-dd,但是如果某个field先过来一个"2019-11-11"的值,就会自动被dynamic mapping识别成date,后面如果在过来一个"hello world"之类的值,就会报错。可以手动关闭某个type的date detection,如果有需要,自己手动指定某个field为date类型。
● 语法:
PUT /index
{
"mappings": {
"date_detection": false,
"properties": {
"filed1": {
"type": ""
},
"filed2": {
"type": ""
},
...
}
}
}
示例:关闭日期探测
PUT /my_index
{
"mappings": {
"date_detection": false,
"properties": {
"title": {
"type": "text"
},
"address": {
"type": "object",
"dynamic": "true"
}
}
}
}
测试插入数据:
PUT /my_index/_doc/1
{
"title": "my article",
"content": "this is my article",
"address": {
"province": "guangdong",
"city": "guangzhou"
},
"post_date":"2019-09-10"
}
● 自定义日期格式:dynamic_date_formats可以定制自己的格式。
● 语法:
PUT /index
{
"mappings": {
"dynamic_date_formats": ["MM/dd/yyyy"],
"properties": {
"filed1": {
"type": ""
},
"filed2": {
"type": ""
},
...
}
}
}
示例:自定义日期格式
PUT /my_index
{
"mappings": {
"dynamic_date_formats": ["MM/dd/yyyy"]
}
}
测试插入数据:
PUT /my_index/_doc/1
{
"create_date": "09/25/2019"
}
● 数字探测(Numeric Detection):虽然JSON支持本机浮点和整数数据类型,但是某些应用程序或语言有时候可能将数字呈现为字符串。通常正确的解决方案是显示的映射这些字段,那么就可以启用数字检测(默认情况下禁用)来自动完成这些操作。
● 语法:
PUT /index
{
"mappings": {
"numeric_detection": true,
"properties": {
"filed1": {
"type": ""
},
"filed2": {
"type": ""
},
...
}
}
}
示例:开启数字探测
PUT /my_index
{
"mappings": {
"numeric_detection": true
}
}
测试插入数据:
PUT /my_index/_doc/1
{
"my_float": "1.0",
"my_integer": "1"
}
● 动态映射模板允许我们自定义映射,这些映射可以应用到动态添加的字段。
● 语法:
PUT index
{
"mappings": {
"dynamic_templates": [
{
"template_name": {
... match conditions ...
"mapping": { ... }
}
},
...
]
}
}
● 示例:
● 定制自己的映射模块:
PUT /my_index
{
"mappings": {
"dynamic_templates": [
{
"en": {
"match": "*_en",
"match_mapping_type": "string",
"mapping": {
"type": "text",
"analyzer": "english"
}
}
}
]
}
}
插入数据:
PUT /my_index/_doc/1
{
"title": "this is my first article"
}
PUT /my_index/_doc/2
{
"title_en": "this is my first article"
}
搜索:
GET /my_index/_search?q=article
GET /my_index/_search?q=is
title没有匹配到任何的动态模块,默认就是standard分词器,不会过滤掉停用词,像is会进入到倒排索引,用is来搜索是可以搜索到的。
title_en匹配到了动态模块,就是english分词器,会过滤掉停用词,is这种停用词会被过滤掉,用is来搜索是搜索不到的。
模板写法:
PUT my_index
{
"mappings": {
"dynamic_templates": [
{
"integers": {
"match_mapping_type": "long",
"mapping": {
"type": "integer"
}
}
},
{
"strings": {
"match_mapping_type": "string",
"mapping": {
"type": "text",
"fields": {
"raw": {
"type": "keyword",
"ignore_above": 256
}
}
}
}
}
]
}
}
模板参数:
"match": "long_*",
"unmatch": "*_text",
"match_mapping_type": "string",
"path_match": "name.*",
"path_unmatch": "*.middle",
"match_pattern": "regex",
"match": "^profit_\d+$"
● 动态模块的应用场景:
● ①结构化搜索:
○ 默认情况下,ElasticSearch将字符串字段映射为带有子关键字字段的文本字段。但是,如果只对结构化内容进行索引,而对全文检索不感兴趣,则可以仅将字段映射为关键字。注意:为了能够搜索这些字段,必须搜索索引的时候用完全相同的值。
{
"strings_as_keywords": {
"match_mapping_type": "string",
"mapping": {
"type": "keyword"
}
}
}
● ②仅搜索:
○ 如果只关系字符串字段的全文检索,并且不打算对字符串字段进行聚合、排序或精确搜索,可以将其映射为文本字段。
{
"strings_as_text": {
"match_mapping_type": "string",
"mapping": {
"type": "text"
}
}
}
● ③norms :不关心评分。
○ norms是指标时间的评分因素。如果不关心评分,比如:不按评分对文档进行排序,则可以在索引中禁用这些评分因子的存储以便节省一些空间。
{
"strings_as_keywords": {
"match_mapping_type": "string",
"mapping": {
"type": "text",
"norms": false,
"fields": {
"keyword": {
"type": "keyword",
"ignore_above": 256
}
}
}
}
}
● 场景:index被确定以后,一个field的设置是不能被修改的,如果要修改一个field,应该重新按照新的mapping,建立新的index,并将数据批量查询出来,重新用bulk api写入到index中。
● 批量查询的时候,建议采用scroll api,并且采用多线程并发的方式来重建索引数据,每次scroll就查询执行日期的一段数据,交给一个线程即可。
示例:
①一开始,依据动态映射(dynamic mapping)插入数据,但是不小心有些数据是诸如"2019-11-11"之类的日期 格式,所以会被映射为date类型,实际上它是string类型。
PUT /my_index/_doc/1
{
"title": "2019-11-11"
}
PUT /my_index/_doc/2
{
"title": "2019-11-12"
}
②当后期向索引中插入string类型的title值的时候,会报错。
PUT /my_index/_doc/3
{
"title": "this is my first article"
}
报错:
{
"error" : {
"root_cause" : [
{
"type" : "mapper_parsing_exception",
"reason" : "failed to parse field [title] of type [date] in document with id '3'. Preview of field's value: 'this is my first article'"
}
],
"type" : "mapper_parsing_exception",
"reason" : "failed to parse field [title] of type [date] in document with id '3'. Preview of field's value: 'this is my first article'",
"caused_by" : {
"type" : "illegal_argument_exception",
"reason" : "failed to parse date field [this is my first article] with format [strict_date_optional_time||epoch_millis]",
"caused_by" : {
"type" : "date_time_parse_exception",
"reason" : "Failed to parse with all enclosed parsers"
}
}
},
"status" : 400
}
③此时,想修改title的类型,是不可能的:
PUT /my_index/_mapping
{
"properties":{
"title":{
"type":"text"
}
}
}
报错:
{
"error" : {
"root_cause" : [
{
"type" : "illegal_argument_exception",
"reason" : "mapper [title] cannot be changed from type [date] to [text]"
}
],
"type" : "illegal_argument_exception",
"reason" : "mapper [title] cannot be changed from type [date] to [text]"
},
"status" : 400
}
④此时,唯一的办法,就是重建索引,即重新建立一个新的索引,将旧索引的数据查询出来,再导入新的索引。
⑤如果说就索引的名字是old_index,新索引的名字是new_index。终端Java应用已经在使用old_index操作了,难道我们需要停止Java应用,修改使用的index为new_index,再重新启动Java应用?这个过程中,必然会导致Java应用停机,可用性也降低。
⑥给索引起一个别名,Java应用指向这个别名,那么此时Java应用指向的是旧索引。
PUT /my_index/_alias/prod_index
⑦创建新的Index,调整title的类型为string:
PUT /my_index_new
{
"mappings": {
"properties": {
"title": {
"type": "text"
}
}
}
}
⑧使用scroll api将数据批量查询出来:
GET /my_index/_search?scroll=1m
{
"query": {
"match_all": {}
},
"size": 1
}
返回:
{
"_scroll_id" : "FGluY2x1ZGVfY29udGV4dF91dWlkDXF1ZXJ5QW5kRmV0Y2gBFnRRN1doOHVhUktXQllGZGlHRjN3bFEAAAAAAAANOhZzenoyWWlMd1I2T1kyMWtzMDExYVl3",
"took" : 0,
"timed_out" : false,
"_shards" : {
"total" : 1,
"successful" : 1,
"skipped" : 0,
"failed" : 0
},
"hits" : {
"total" : {
"value" : 1,
"relation" : "eq"
},
"max_score" : 1.0,
"hits" : [
{
"_index" : "my_index",
"_type" : "_doc",
"_id" : "1",
"_score" : 1.0,
"_source" : {
"title" : "2019-11-11"
}
}
]
}
}
⑨采用bulk api将scoll查出来的一批数据,批量写入新索引:
POST /_bulk
{ "index": { "_index": "my_index_new", "_id": "1" }}
{ "title": "2019-11-11" }
⑩反复循环8~9,查询一批批数据出现,采用bulk api将每一批数据批量写入到新索引。
⑪将prod_index这个alias切换到my_index_new上,Java应用会直接通过index别名使用新的索引中的数据,Java应用程序不需要停机,高可用。
POST /_aliases
{
"actions": [
{ "remove": { "index": "my_index", "alias": "prod_index" }},
{ "add": { "index": "my_index_new", "alias": "prod_index" }}
]
}
⑫直接通过pro_index别名查询:
GET /prod_index/_search
对索引进行别名操作:
PUT /my_index_v1/_alias/my_index
● Java客户端对别名(my_index)进行操作。
● 重建索引后,切换v1到v2:
POST /_aliases
{
"actions": [
{ "remove": { "index": "my_index_v1", "alias": "my_index" }},
{ "add": { "index": "my_index_v2", "alias": "my_index" }}
]
}