文章目录
一、数据准备
创建索引结构
PUT my_index
{
"mappings": {
"properties": {
"name": {
"type": "text",
"fields": {
"keyword": {
"type": "keyword"
}
}
},
"age": {
"type": "text",
"fields": {
"keyword": {
"type": "keyword"
}
}
},
"interests": {
"type": "nested",
"properties": {
"name": {
"type": "text",
"fields": {
"keyword": {
"type": "keyword"
}
}
},
"level": {
"type": "text",
"fields": {
"keyword": {
"type": "keyword"
}
}
}
}
}
}
}
}
生成测试数据
POST my_index/_bulk
{ "index": { "_id": "1" }}
{ "name": "John Doe", "age": "25", "interests": [{ "name": "skiing interesting", "level": "8" }, { "name": "hiking interesting", "level": "6" }] }
{ "index": { "_id": "2" }}
{ "name": "Jane Smith", "age": "30", "interests": [{ "name": "reading interesting", "level": "9" }, { "name": "traveling interesting", "level": "7" }] }
{ "index": { "_id": "3" }}
{ "name": "Bob Johnson", "age": "40", "interests": [{ "name": "cooking interesting", "level": "5" }, { "name": "painting interesting", "level": "4" }] }
{ "index": { "_id": "4" }}
{ "name": "John Doe1", "age": "35", "interests": [{ "name": "fitness interesting", "level": "8" }, { "name": "hiking interesting", "level": "6" }] }
{ "index": { "_id": "5" }}
{ "name": "Jane Smith1", "age": "31", "interests": [{ "name": "fitness interesting", "level": "9" }, { "name": "traveling interesting", "level": "7" }] }
{ "index": { "_id": "6" }}
{ "name": "Bob Johnson1", "age": "44", "interests": [{ "name": "running interesting", "level": "5" }, { "name": "painting interesting", "level": "4" }] }
{ "index": { "_id": "7" }}
{ "name": "John Doe2", "age": "27", "interests": [{ "name": "hurlbat interesting", "level": "8" }, { "name": "hiking interesting", "level": "6" }] }
{ "index": { "_id": "8" }}
{ "name": "Jane Smith2", "age": "36", "interests": [{ "name": "game interesting", "level": "9" }, { "name": "hurlbat interesting", "level": "7" }] }
{ "index": { "_id": "9" }}
{ "name": "Bob Johnson2", "age": "42", "interests": [{ "name": "running interesting", "level": "5" }, { "name": "painting interesting", "level": "4" }] }
{ "index": { "_id": "10" }}
{ "name": "John Doe3", "age": "22", "interests": [{ "name": "game interesting", "level": "8" }, { "name": "hurlbat interesting", "level": "6" }] }
{ "index": { "_id": "11" }}
{ "name": "Jane Smith3", "age": "39", "interests": [{ "name": "hurlbat interesting", "level": "9" }, { "name": "traveling interesting", "level": "7" }] }
{ "index": { "_id": "12" }}
{ "name": "Bob Johnson4", "age": "49", "interests": [{ "name": "swimming interesting", "level": "5" }, { "name": "painting interesting", "level": "4" }] }
二、DSL高亮检索
1.非nested字段高亮
在查询中设置 highlight 参数，指定需要高亮显示的字段和高亮格式。例如：
GET /my_index/_search
{
"query": {
"match": {
"name": "Bob Johnson4"
}
},
"highlight": {
"fields": {
"name": {} // 高亮显示 name字段
},
"pre_tags": ["<em>"], // 高亮起始标签
"post_tags": ["</em>"] // 高亮结束标签
}
}
发送查询请求后，查询结果中会包含一个 highlight 字段，其中包含了高亮后的内容。例如：
{
"took" : 4,
"timed_out" : false,
"_shards" : {
"total" : 1,
"successful" : 1,
"skipped" : 0,
"failed" : 0
},
"hits" : {
"total" : {
"value" : 4,
"relation" : "eq"
},
"max_score" : 3.220356,
"hits" : [
{
"_index" : "my_index",
"_type" : "_doc",
"_id" : "12",
"_score" : 3.220356,
"_source" : {
"name" : "Bob Johnson4",
"age" : "49",
"interests" : [
{
"name" : "swimming interesting",
"level" : "5"
},
{
"name" : "painting interesting",
"level" : "4"
}
]
},
"highlight" : {
"name" : [
"<em>Bob</em> <em>Johnson4</em>"
]
}
},
{
"_index" : "my_index",
"_type" : "_doc",
"_id" : "3",
"_score" : 1.060872,
"_source" : {
"name" : "Bob Johnson",
"age" : "40",
"interests" : [
{
"name" : "cooking interesting",
"level" : "5"
},
{
"name" : "painting interesting",
"level" : "4"
}
]
},
"highlight" : {
"name" : [
"<em>Bob</em> Johnson"
]
}
},
{
"_index" : "my_index",
"_type" : "_doc",
"_id" : "6",
"_score" : 1.060872,
"_source" : {
"name" : "Bob Johnson1",
"age" : "44",
"interests" : [
{
"name" : "running interesting",
"level" : "5"
},
{
"name" : "painting interesting",
"level" : "4"
}
]
},
"highlight" : {
"name" : [
"<em>Bob</em> Johnson1"
]
}
},
{
"_index" : "my_index",
"_type" : "_doc",
"_id" : "9",
"_score" : 1.060872,
"_source" : {
"name" : "Bob Johnson2",
"age" : "42",
"interests" : [
{
"name" : "running interesting",
"level" : "5"
},
{
"name" : "painting interesting",
"level" : "4"
}
]
},
"highlight" : {
"name" : [
"<em>Bob</em> Johnson2"
]
}
}
]
}
}
可以看到，查询结果中的 highlight 字段中包含了高亮后的内容，用 <em> 和 </em> 包裹了关键词。在前端页面中，可以使用 CSS 样式来修改高亮显示的样式。
2.nested字段高亮
在嵌套查询时，如果使用普通高亮查询，会有一个问题，因为一个文档数据记录中可能会有多个对象，所以一条文档可能会有多个满足高亮的字段，而普通高亮只会把高亮的字段在数组中罗列出来，并不知道一条高亮字段属于文档中的哪个对象。在查询中设置 highlight 参数，指定需要高亮显示的字段和高亮格式。对于 nested 类型的字段，需要使用 inner_hits 子句来指定需要高亮的子字段。例如：
GET /my_index/_search
{
"query": {
"nested": {
"path": "interests",
"query": {
"bool": {
"must": [
{
"match": {
"interests.name": "hurlbat"
}
},
{
"match": {
"interests.level": "8"
}
}
]
}
},
"inner_hits": {
"_source": false, // inner_hits中的_source不会返回字段内容
"highlight": {
"fields": {
"interests.name": {}
},
"pre_tags": [
"<em>"
],
"post_tags": [
"</em>"
]
}
}
}
}
}
发送查询请求后，查询结果中的 inner_hits 字段中包含了高亮后的内容。例如：
{
"took" : 4,
"timed_out" : false,
"_shards" : {
"total" : 1,
"successful" : 1,
"skipped" : 0,
"failed" : 0
},
"hits" : {
"total" : {
"value" : 1,
"relation" : "eq"
},
"max_score" : 3.429597,
"hits" : [
{
"_index" : "my_index",
"_type" : "_doc",
"_id" : "7",
"_score" : 3.429597,
"_source" : {
"name" : "John Doe2",
"age" : "27",
"interests" : [
{
"name" : "hurlbat interesting",
"level" : "8"
},
{
"name" : "hiking interesting",
"level" : "6"
}
]
},
"inner_hits" : {
"interests" : {
"hits" : {
"total" : {
"value" : 1,
"relation" : "eq"
},
"max_score" : 3.429597,
"hits" : [
{
"_index" : "my_index",
"_type" : "_doc",
"_id" : "7",
"_nested" : { // 数组中的位置信息
"field" : "interests",// path
"offset" : 0 // 偏移量 这里说明是数组中第一个元素
},
"_score" : 3.429597,
"_source" : {
"level" : "8",
"name" : "hurlbat interesting"
},
"highlight" : {
"interests.name" : [
"<em>hurlbat</em> interesting"
]
}
}
]
}
}
}
}
]
}
}
可以看到，查询结果中的 inner_hits 字段中包含了高亮后的内容，用 <em> 和 </em> 包裹了关键词。在前端页面中，可以使用 CSS 样式来修改高亮显示的样式。
三、SpringBoot实现Elasticsearch检索高亮
SpringBoot集成Elasticsearch查看我另一篇博客SpringBoot集成Elasticsearch
由于高亮highlight是单独的属性，并没有体现在返回内容当中，所以就需要获取到highlight字段内容，然后替换掉检索结果_source中的字段内容。
封装一个ES组件
@Component
public class EsComponent {
@Resource
private ElasticsearchRestTemplate elasticsearchRestTemplate;
/**
* 检索
*
* @return
*/
public SearchHits<Map> search(Query query) {
final SearchHits<Map> searchHits = elasticsearchRestTemplate.search(query, Map.class, IndexCoordinates.of("my_index"));
return searchHits;
}
}
1.非nested字段高亮
public void nonNestedFieldsHighlight() {
// 1.构建检索Query
final HighlightBuilder highlightBuilder = new HighlightBuilder()
.preTags("<em>").postTags("</em>")
.highlighterType("unified")
.field("name");
final NativeSearchQueryBuilder nativeSearchQueryBuilder = new NativeSearchQueryBuilder()
.withQuery(QueryBuilders.matchQuery("name", "Bob Johnson4"))
.withTrackTotalHits(true)
.withHighlightBuilder(highlightBuilder);
// 2.检索
final SearchHits<Map> searchResponse = esComponent.search(nativeSearchQueryBuilder.build());
// 3.检索结果处理
final List<SearchHit<Map>> searchHits = searchResponse.getSearchHits();
for (SearchHit searchHit : searchHits) {
final Map<String, ?> highlightFields = searchHit.getHighlightFields();
//使用高亮替换检索内容
highlightFields.entrySet().forEach(entry -> ((Map) searchHit.getContent()).put(entry.getKey(), entry.getValue()));
}
searchHits.forEach(entry -> System.out.println(entry.getContent()));
}
2.nested字段高亮
/**
* nested字段高亮
*/
public void nestedFieldsHighlight() {
// 1.构建检索Query
InnerHitBuilder innerHitBuilder = new InnerHitBuilder();
HighlightBuilder highlightBuilder = new HighlightBuilder();
highlightBuilder.preTags("<span style='color:red'>").postTags("</span>");
//设置高亮的方法
highlightBuilder.highlighterType("plain");
//设置分段的数量不做限制
highlightBuilder.numOfFragments(0);
highlightBuilder.field("interests.name").field("interests.level");
innerHitBuilder.setHighlightBuilder(highlightBuilder);
innerHitBuilder.setName(UUID.randomUUID().toString());
innerHitBuilder.setFetchSourceContext(new FetchSourceContext(false));
final NestedQueryBuilder nestedQueryBuilder = QueryBuilders.nestedQuery(
"interests",
QueryBuilders.boolQuery()
.must(QueryBuilders.matchQuery("interests.name", "hurlbat"))
.must(QueryBuilders.matchQuery("interests.level", "8")),
ScoreMode.None);
nestedQueryBuilder.innerHit(innerHitBuilder);
final NativeSearchQueryBuilder nativeSearchQueryBuilder = new NativeSearchQueryBuilder()
.withQuery(nestedQueryBuilder)
.withTrackTotalHits(true);
// 2.检索
final SearchHits<Map> searchResponse = esComponent.search(nativeSearchQueryBuilder.build());
// 3.检索结果处理
final List<SearchHit<Map>> searchHits = searchResponse.getSearchHits();
for (SearchHit<Map> searchHit : searchHits) {
// Map<第一层属性,Map<第二层属性,value>>
final Map<String, ?> map = (Map<String, Map<String, String>>) searchHit.getContent();
final Map<String, SearchHits<?>> innerHitMap = searchHit.getInnerHits();
if (CollectionUtil.isNotEmpty(innerHitMap)) {
for (Map.Entry<String, SearchHits<?>> entry : innerHitMap.entrySet()) {
final String innerHitName = entry.getKey();
final List<? extends SearchHit<?>> highlightSearchHits = entry.getValue().getSearchHits();
if (CollectionUtil.isNotEmpty(highlightSearchHits)) {
final Map<String, List<String>> highlightFields = highlightSearchHits.get(0).getHighlightFields();
final NestedMetaData nestedMetaData = highlightSearchHits.get(0).getNestedMetaData();
if (CollectionUtil.isNotEmpty(highlightFields)) {
highlightFields.entrySet().forEach(e -> {
final String[] keys = e.getKey().split("\\.");
final List<String> value = e.getValue();
// map.get(keys[0])两种情况：一种是对象 一种是集合
final Object o = map.get(keys[0]);
if (o instanceof Map) {
((Map<String,String>) o).put(keys[1], value.get(0));
} else if (o instanceof List) {
((List<Map<String,String>>) o).get(nestedMetaData.getOffset()).put(keys[1], value.get(0));
}
});
}
}
}
}
}
searchHits.forEach(entry -> System.out.println(entry.getContent()));
}
1万+
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
