参考
Elasticsearch: analyzer
一、测试
Must 一切都必须建立在里面
should如果在里面建立一个，should里面只有一个，所以用这个条件
term 精确匹配，不分词
match_phrase 精确匹配
会分词一般按空格分开 目标文档包括分词后的所有单词 文档位于同一位置 使用boost的场景，term、range、boost
Range gt、le 大于小于 y M w d h H m s
script脚本操作
//doc['price'] GET /mall/product/_search { "query": { "script": { "script": { "inline": "doc['price'].value > params.num", "params": {"num": 25} } } } } //ctx._source POST /mall/product/3/_update { "script": { "inline": "ctx._source.price = params.num", "params": {"num": 40} } } PUT /lulog/doc/2 { "log_size": 2 } POST /lulog/doc/_search { "query": { "match_all": {} }, "script_fields": { "total_size": { "script": { "inline": "int total=0;for(int i=0; i<doc['log_size'].length;i ){total = doc['log_size'][i];}return total" } } } } POST /lulog/doc/_search { "query": { "match_all": {} }, "script_fields": { "count": { "script": { "inline": "return doc['log_size'].value 1" } } } } 聚合：
Metric统计，min、max、sum、avg、stats、extended_stats
Bucket分类，terms，range，date_range，histogram，date_histogram，filter过滤
Pipeline对bucket统计分类，min_bucket，max_bucket，sum_bucket，avg_bucket，stats_bucket，extended_stats_bucket
Matrix计算两个数值字段之间的关系
GET incall_daily-2020.04.24/doc/_search { "aggs": { "asdjhasd": { "matrix_stats": { "fields": ["applause_count", "answer_count"] } } } } { "name": "applause_count", ///字段名称 "count": 154, //字段样本数量 "mean": 44.31818181818182, //平均值 "variance": 276.04842543077837, ///方差，偏离平均值的程度 "skewness": -1.60794152400977, //偏差，量化平均值附近的非对称分布 "kurtosis": 5.006387495325678, //峰度，量化分布形状 "covariance": { //协方差，描述一个字段数据随另一个字段数据变化程度的矩阵 "applause_count": 276.04842543077837, "answer_count": 290.8722519310755 }, "correlation": { //相关性，描述两个字段数据之间的分布关系，其协方差矩阵值为[-1， 1] "applause_count": 1.0, "answer_count": 0.9397723506362066 } }, { "name": "answer_count", "count": 154, "mean": 53.48051948051948, "variance": 347.035565741448, "skewness": -2.2011119165278625, "kurtosis": 6.583666847748493, "covariance": { "applause_count": 290.8722519310755, "answer_count": 347.035565741448 }, "correlation": { "applause_count": 0.9397723506362066, "answer_count": 1.0 } }
如果是聚合物，建议过滤数据filter，不要使用query
为啥？filter 我想要的只能匹配文档，query匹配后，还有一个相关的评分过程，所以filter效率更高
GET incall_daily*/doc/_search { "query": { "bool": { "must": [ { "match": { "parent_id": "17871" } } ] } }, "aggs": { "a1": { "date_histogram": { "field": "create_time", "interval": "2d" }, "aggs": { "a2": { "stats": { "field": "applause_count" } } } } } } GET incall_daily-2020.04.24/doc/_search { "aggs": { "f1": { "filter": { "term": { "user_id": "17871" } }, "aggs": { "a1": { "date_histogram": { "field": "create_time", "interval": "2d" }, "aggs": { "a2": { "terms": { "field": "applause_count" } } } } } } } } 二、查询篇
1、高亮查询
GET callcenter/_search { "query": { "match": { "content": "天翼" } }, "size": 30, "highlight": { "pre_tags": "<span>", "post_tags": "</span>", "fields": { "content": { "fragment_size": 20, //从content前20个0个单词的亮字段 } } } } 2.查询字符串
query_string，只有 AND OR，而且必须是大写 simple_query_string， 与 |或 -非 () * GET /callcenter/_search { "query": { "query_string": { "fields": ["title", "content"], "query": "天翼 AND (骚扰 OR 商务)" } } } GET /callcenter/_search { "query": { "query_string": { "fields": ["title", "content"], "query": "(商务 OR 直播) AND 商务" } } } GET callcenter/_search { "query": { "simple_query_string": { "query": "-商务 -工号 -外勤 -网吧 -天翼", "fields": ["title"] } }
}
相关性
GET callcenter/_search
{
"query": {
"boosting": {
"positive": {
"match": {
"content": "天翼商务"
}
},
"negative": {
"match": {
"content": "商务"
}
},
"negative_boost": 0.2
}
},
"size": 50
}
3、复合查询
该查询实现和上面的查询一样的效果，看得出，代码写的贼麻烦
GET callcenter/_search
{
"query": {
"bool": {
"must": [
{
"bool": {
"must_not": [
{"match": {"title": "商务"}},
{"match": {"title": "工号"}},
{"match": {"title": "外勤"}},
{"match": {"title": "网吧"}},
{"match": {"title": "天翼"}}
]
}
},
{
"bool": {
"must_not": [
{"match": {"content": "商务"}},
{"match": {"content": "工号"}},
{"match": {"content": "外勤"}},
{"match": {"content": "网吧"}},
{"match": {"content": "天翼"}}
]
}
}
]
}
}
}
4、精确查找
GET callcenter/_search
{
"query": {
"term": {
"title": {
"value": "商务"
}
}
}
}
GET callcenter/_search
{
"query": {
"match_phrase": {
"title": "商务"
}
}
}
GET callcenter/_search
{
"query": {
"terms": {
"title": [
"电信",
"号码"
]
}
}
}
5、分页查询
GET article/_search
{
"query": {
"match_all": {}
},
"sort": [
{
"date": {
"order": "desc"
}
}
],
"from": 20,
"size": 20
}
滚动翻页
GET article/_search?scroll=1m
{
"query": {
"match_all": {}
},
"sort": [
{
"date": {
"order": "desc"
}
}
],
"size": 10
}
GET _search/scroll
{
"scroll": "1m",
"scroll_id": "DnF1ZXJ5VGhlbkZldGNoBQAAAAAABdUMFk5uSzVIbGRuU1FTb1RoT0x5dTVyM2cAAAAAAAXVDRZObks1SGxkblNRU29UaE9MeXU1cjNnAAAAAAAF1Q4WTm5LNUhsZG5TUVNvVGhPTHl1NXIzZwAAAAAABdUPFk5uSzVIbGRuU1FTb1RoT0x5dTVyM2cAAAAAAAXVEBZObks1SGxkblNRU29UaE9MeXU1cjNn"
}
6、聚合查询
GET incall_daily-2020.04.24/_search
{
"aggs": {
"f": {
"range": {
"field": "create_time",
"ranges": [
{
"from": "2020-04-23T00:00:00.000Z"
}
]
},
"aggs": {
"d1": {
"date_histogram": {
"field": "create_time",
"interval": "1d",
"order": {
"_key": "desc"
}
},
"aggs": {
"d2": {
"terms": {
"field": "order_type.keyword",
"order": {
"d3": "desc"
}
},
"aggs": {
"d3": {
"sum": {
"field": "answer_count"
}
}
}
}
}
}
}
}
}
}
7、建议查询
精准程度上(Precision)看： Completion > Phrase > term
速度上，Completion是最快的
自动补全设计：优先取Completion，依次...
POST /lu_article/_search
{
"_source": ["title", "descript", "create_time"],
"suggest": {
"s1": {
"prefix": "cento",
"completion": {
"field": "suggest",
"fuzzy": {
}
}
},
"s2": {
"text": "elasticseqrch mybatls mysal",
"phrase": {
"field": "descript",
"highlight": {
"pre_tag": "<em>",
"post_tag": "</em>"
}
}
},
"s3": {
"text": "elasticseqrch mybatls mysal",
"term": {
"field": "descript"
}
}
}
}
##completion创建字段，插入值，自动补全，suggest可以是字符串、对象、数组，weight选填
"suggest": {
"type": "completion",
"analyzer": "ik_max_word"
},
PUT /lu_article/_doc/1
{
"suggest": [
{
"input": "lucene solr",
"weight": 1
},
{
"input": "lucene so cool",
"weight": 4
},
{
"input": "lucene elasticsearch",
"weight": 3
}
]
}
POST /lu_article/_search
{
"_source": ["title", "descript", "create_time", "suggest"],
"suggest": {
"s1": {
"prefix": "luce",
"completion": {
"field": "suggest"
}
}
}
}