引言
之前已经介绍了如何搭建elasticsearch服务端和简单的索引创建，和中文分词的支持。今天我们来说一说如何实现elasticsearch同时实现中文分词和pinyin分词。并且实现类似百度搜索栏的搜索建议的功能。
混合查询
实现混合查询有很多方式，这里介绍我认为是一个偷懒的方法，就是为你要拼音搜索的字段提供两个额外的字段，一个是全拼字段，一个是首字母缩写字段。我这里用的是官网的Employee的例子：
public class Employee implements Serializable { private String firstName; private String lastName; private String pinyin;//firstName全拼 private String header;//firstName首字母缩写 private int age; private String about; private List<String> interests; ....省略getter setter
接下来为index添加setting和mapping
XContentBuilder settings = XContentFactory.jsonBuilder(); settings.startObject() .startObject("analysis") .startObject("analyzer") .startObject("ik_analyzer").field("tokenizer","ik_smart") .endObject() .endObject() .endObject().endObject(); CreateIndexRequest createIndexRequest = new CreateIndexRequest(index).settings(settings); CreateIndexResponse createIndexResponse = esClient.admin().indices().create(createIndexRequest).get(); logger.info("Index:{} created,response:{}", index, JSON.toJSON(createIndexResponse)); XContentBuilder builder = XContentFactory.jsonBuilder(); builder.startObject() .startObject(type) .startObject("properties") .startObject("firstName").field("type", "string").field("analyzer","ik_smart") /* .field("search_analyzer","ik_smart").field("preserve_separators",false) .field("preserve_position_increments",false)*/ .endObject() .startObject("lastName").field("type", "string").field("analyzer","ik_smart") .endObject() .startObject("pinyin").field("type","string").field("analyzer","pinyin") .startObject() .startObject("header").field("type","string").field("analyzer","pinyin") .startObject("about").field("type", "string").field("analyzer","ik_smart") .endObject() .startObject("interests").field("type", "string").field("analyzer","ik_smart") .endObject() .endObject() .endObject() .endObject(); PutMappingRequest putMappingRequest = new PutMappingRequest(index); putMappingRequest.type(type); putMappingRequest.source(builder); PutMappingResponse putMappingResponse = esClient.admin().indices().putMapping(putMappingRequest).get(); logger.info("Mapping for `{}.{}` putted, response:{}", index, type, JSON.toJSON(putMappingResponse)); return true; } catch (Exception e) { logger.error("doCreateIndex", e); return false; }
添加几个测试用例，我这里直接用了批量插入索引的方法：
public Boolean bulkIndex(List<String> jsonList){ if(esIndexTypes.get(index)==null) { if(getMapping(index, indexType)) esIndexTypes.put(index,true); } BulkRequestBuilder bulkBuilder= esClient.prepareBulk(); for (String s : jsonList) { IndexRequestBuilder requestBuilder = esClient.prepareIndex(index, indexType) .setSource(s); bulkBuilder.add(requestBuilder); } BulkResponse bulkResponse = bulkBuilder.execute().actionGet(); logger.info("index:{} bulk request,:response:{}",index,JSON.toJSON(bulkResponse)); return true; } @org.junit.Test public void test(){ List<String> list1 = new ArrayList<>(10000); for (int i=0;i<10000;i++) { Employee employee = new Employee(); employee.setFirstName("告白气球"+i); employee.setPinyin("gaobaiqiqiu"+i); employee.setHeader("gbqq"); employee.setLastName("周杰伦,日记"); employee.setAbout("呜啦啦啦火车笛 " + " " + "随着奔腾的马蹄 " + " " + "小妹妹吹着口琴 " + " " + "夕阳下美了剪影 " + " " + "我用子弹写日记,我泡妞看电影"); employee.setAge(18); List<String> list = new ArrayList<String>(); list.add("喜欢打篮球"); list.add("在大晴天晒太阳"); list.add("泡妞看电影"); employee.setInterests(list); list1.add(JSON.toJSONString(employee)); } boolean index = esProxy.bulkIndex(list1); }
最后直接搜gaobaiqiqiu或gbqq搜出来的数据像这样：
[{"firstName":"告白气球","lastName":"周杰伦,日记","pinyin":"gaobaiqiqiu","about":"呜啦啦啦火车笛
随着奔腾的马蹄
小妹妹吹着口琴
夕阳下美了剪影
我用子弹写日记,我泡妞看电影","header":"gbqq","interests":["喜欢打篮球","在大晴天晒太阳","泡妞看电影"],"age":18}]
如果直接搜告白搜出来的数据像这样:
[{"firstName":"<span style="color:red">告白</span>气球","lastName":"周杰伦,日记","pinyin":"gaobaiqiqiu","about":"呜啦啦啦火车笛
随着奔腾的马蹄
小妹妹吹着口琴
夕阳下美了剪影
我用子弹写日记,我泡妞看电影","header":"gbqq","interests":["喜欢打篮球","在大晴天晒太阳","泡妞看电影"],"age":18}]
CompletionSuggestion查询建议
使用CompletionSuggestion时mapping需要改一下，实时推荐的字段type需要使用completion。
XContentBuilder builder = XContentFactory.jsonBuilder(); builder.startObject() .startObject(type) .startObject("properties") .startObject("firstName").field("type", "completion").field("analyzer","ik_smart") .field("search_analyzer","ik_smart").field("preserve_separators",false) .field("preserve_position_increments",false) .endObject() .startObject("lastName").field("type", "string").field("analyzer","ik_smart") .endObject() .startObject("pinyin").field("type","string").field("analyzer","pinyin") .startObject() .startObject("header").field("type","string").field("analyzer","pinyin") .startObject("about").field("type", "string").field("analyzer","ik_smart") .endObject() .startObject("interests").field("type", "string").field("analyzer","ik_smart") .endObject() .endObject() .endObject() .endObject();
查询的时候需要使用CompletionSuggestionBuilder.
public void searchSuggest(String str){ CompletionSuggestionBuilder suggestionBuilder = new CompletionSuggestionBuilder("firstName"); suggestionBuilder.analyzer("ik_smart"); suggestionBuilder.text(str); SearchResponse response = esClient.prepareSearch(index).setTypes(indexType).setQuery(QueryBuilders.matchAllQuery()) .suggest(new SuggestBuilder().addSuggestion("my-suggest-1",suggestionBuilder)).get(); Suggest suggest= response.getSuggest(); CompletionSuggestion suggestion = suggest.getSuggestion("my-suggest-1"); List<CompletionSuggestion.Entry> list = suggestion.getEntries(); for (int i = 0; i < list.size(); i++) { List<CompletionSuggestion.Entry.Option> options = list.get(i).getOptions(); for (int j = 0; j < options.size(); j++) { if (options.get(j) instanceof CompletionSuggestion.Entry.Option) { CompletionSuggestion.Entry.Option op = options.get(j); System.out.println(op.getScore()+"--"+op.getText()); } } } }
你也可以使用restAPI：http://192.168.10.xxx:9200/megacorp/_search?pretty这里megacorp是indexName,
{ "size": 0,
"suggest": {
"my-suggest-1": {
"prefix": "someone li",
"completion": {
"field": "firstName"
}
}
}
}
查询出来的结果:
{ "took": 12, "timed_out": false, "_shards": { "total": 5, "successful": 5, "failed": 0 }, "hits": { "total": 0, "max_score": 0, "hits": [] }, "suggest": { "blog-suggest": [ { "text": "someone li", "offset": 0, "length": 10, "options": [ { "text": "someone like you", "_index": "megacorp", "_type": "employee", "_id": "AV_doqcXKY206Vs3lcCO", "_score": 1, "_source": { "about": "呜啦啦啦火车笛 随着奔腾的马蹄 小妹妹吹着口琴 夕阳下美了剪影 我用子弹写日记,我泡妞看电影", "age": 18, "firstName": "someone like you", "interests": [ "喜欢打篮球", "在大晴天晒太阳", "泡妞看电影" ], "lastName": "周杰伦,日记" } } ] } ] } }
————————————————
版权声明：本文为CSDN博主「lance的java小菜」的原创文章，遵循CC 4.0 BY-SA版权协议，转载请附上原文出处链接及本声明。
原文链接：https://blog.csdn.net/nethackatschool/article/details/78594843