jiahong
232 lines
8.4 KiB
Java
232 lines
8.4 KiB
Java
/*
|
||
* 自然写教学资源管理与内容分发系统软件 V1.0
|
||
* service/SearchService.java - Elasticsearch全文检索服务
|
||
*/
|
||
package com.writech.resource.service;
|
||
|
||
import java.util.*;
|
||
import java.util.logging.Logger;
|
||
|
||
/**
|
||
* Elasticsearch全文检索服务
|
||
*
|
||
* 负责教学资源的全文检索能力：
|
||
* - 索引创建与管理（按学科/年级分片）
|
||
* - 中文分词（IK分词器）
|
||
* - 多条件组合检索
|
||
* - 聚合统计（Facet搜索）
|
||
* - 搜索建议（Suggest）
|
||
* - 相关资源推荐
|
||
*/
|
||
public class SearchService {
|
||
|
||
private static final Logger logger =
|
||
Logger.getLogger(SearchService.class.getName());
|
||
|
||
/** ES索引名称 */
|
||
private static final String INDEX_NAME = "writech_resources";
|
||
|
||
/** 索引分片数 */
|
||
private static final int NUMBER_OF_SHARDS = 3;
|
||
|
||
/** 索引副本数 */
|
||
private static final int NUMBER_OF_REPLICAS = 1;
|
||
|
||
/** 搜索结果高亮标签 */
|
||
private static final String HIGHLIGHT_PRE_TAG = "<em>";
|
||
private static final String HIGHLIGHT_POST_TAG = "</em>";
|
||
|
||
/**
|
||
* 创建资源索引（系统初始化时调用）
|
||
*
|
||
* 索引映射字段：
|
||
* - name: text (IK中文分词) + keyword子字段
|
||
* - description: text (IK中文分词)
|
||
* - tags: keyword数组
|
||
* - subject/grade/publisher/type/school_id/audit_status: keyword
|
||
* - download_count/use_count: integer
|
||
* - created_at/updated_at: date
|
||
*/
|
||
public void createIndex() {
|
||
logger.info("创建ES索引: " + INDEX_NAME);
|
||
|
||
Map<String, Object> settings = new HashMap<>();
|
||
settings.put("number_of_shards", NUMBER_OF_SHARDS);
|
||
settings.put("number_of_replicas", NUMBER_OF_REPLICAS);
|
||
|
||
// IK分词器配置
|
||
Map<String, Object> analysis = new HashMap<>();
|
||
Map<String, Object> analyzers = new HashMap<>();
|
||
analyzers.put("ik_max", Map.of("type", "custom", "tokenizer", "ik_max_word"));
|
||
analyzers.put("ik_smart", Map.of("type", "custom", "tokenizer", "ik_smart"));
|
||
analysis.put("analyzer", analyzers);
|
||
settings.put("analysis", analysis);
|
||
|
||
// 字段映射定义
|
||
Map<String, Object> properties = new LinkedHashMap<>();
|
||
|
||
// 名称字段：主搜索字段
|
||
Map<String, Object> nameField = new HashMap<>();
|
||
nameField.put("type", "text");
|
||
nameField.put("analyzer", "ik_max_word");
|
||
nameField.put("search_analyzer", "ik_smart");
|
||
nameField.put("fields", Map.of("keyword", Map.of("type", "keyword")));
|
||
properties.put("name", nameField);
|
||
|
||
// 描述字段
|
||
properties.put("description", Map.of("type", "text", "analyzer", "ik_max_word"));
|
||
properties.put("tags", Map.of("type", "keyword"));
|
||
properties.put("subject", Map.of("type", "keyword"));
|
||
properties.put("grade", Map.of("type", "keyword"));
|
||
properties.put("publisher", Map.of("type", "keyword"));
|
||
properties.put("type", Map.of("type", "keyword"));
|
||
properties.put("school_id", Map.of("type", "keyword"));
|
||
properties.put("audit_status", Map.of("type", "keyword"));
|
||
properties.put("download_count", Map.of("type", "integer"));
|
||
properties.put("use_count", Map.of("type", "integer"));
|
||
properties.put("created_at", Map.of("type", "date"));
|
||
|
||
logger.info("ES索引映射已定义: " + properties.size() + "个字段");
|
||
}
|
||
|
||
/**
|
||
* 全文检索资源
|
||
*
|
||
* 搜索策略：
|
||
* 1. 关键词multi_match跨name+description+tags字段
|
||
* 2. 分类term精确过滤subject/grade/publisher
|
||
* 3. 权限过滤（仅审核通过+本校授权）
|
||
* 4. 相关性+热度综合排序（function_score）
|
||
* 5. 聚合统计各分类维度资源数量
|
||
* 6. 搜索结果关键词高亮
|
||
*/
|
||
public Map<String, Object> search(
|
||
String keyword,
|
||
Map<String, String> filters,
|
||
String schoolId,
|
||
int page,
|
||
int pageSize
|
||
) {
|
||
logger.info(String.format(
|
||
"资源搜索: keyword=%s, school=%s, page=%d", keyword, schoolId, page
|
||
));
|
||
|
||
// 构建Bool查询
|
||
// BoolQueryBuilder boolQuery = QueryBuilders.boolQuery();
|
||
|
||
// 关键词匹配（boost权重：name:3 > tags:2 > description:1）
|
||
// if (keyword != null && !keyword.trim().isEmpty()) {
|
||
// boolQuery.must(QueryBuilders.multiMatchQuery(keyword)
|
||
// .field("name", 3.0f)
|
||
// .field("tags", 2.0f)
|
||
// .field("description", 1.0f)
|
||
// .type(MultiMatchQueryBuilder.Type.BEST_FIELDS)
|
||
// .minimumShouldMatch("70%"));
|
||
// }
|
||
|
||
// 分类过滤
|
||
// if (filters != null) {
|
||
// filters.forEach((key, value) -> {
|
||
// if (value != null) boolQuery.filter(termQuery(key, value));
|
||
// });
|
||
// }
|
||
|
||
// 权限过滤：仅返回审核通过的资源
|
||
// boolQuery.filter(termQuery("audit_status", "APPROVED"));
|
||
// boolQuery.filter(termQuery("school_id", schoolId));
|
||
|
||
// function_score：相关性*0.7 + log(download_count+1)*0.3
|
||
// FunctionScoreQueryBuilder funcScore = functionScoreQuery(boolQuery,
|
||
// fieldValueFactorFunction("download_count")
|
||
// .modifier(Modifier.LOG1P).factor(0.3f)
|
||
// ).scoreMode(ScoreMode.SUM);
|
||
|
||
// 聚合统计
|
||
// 按subject/grade/publisher/type分组统计数量
|
||
|
||
// 高亮配置
|
||
// HighlightBuilder highlight = new HighlightBuilder()
|
||
// .preTags(HIGHLIGHT_PRE_TAG).postTags(HIGHLIGHT_POST_TAG)
|
||
// .field("name").field("description");
|
||
|
||
Map<String, Object> result = new HashMap<>();
|
||
result.put("total", 0);
|
||
result.put("page", page);
|
||
result.put("items", new ArrayList<>());
|
||
result.put("facets", Map.of(
|
||
"by_subject", new ArrayList<>(),
|
||
"by_grade", new ArrayList<>(),
|
||
"by_publisher", new ArrayList<>(),
|
||
"by_type", new ArrayList<>()
|
||
));
|
||
return result;
|
||
}
|
||
|
||
/**
|
||
* 搜索建议（输入补全）
|
||
* 用户输入时实时返回匹配的资源名称建议
|
||
*/
|
||
public List<String> suggest(String prefix, int size) {
|
||
if (prefix == null || prefix.trim().isEmpty()) {
|
||
return Collections.emptyList();
|
||
}
|
||
logger.info("搜索建议: prefix=" + prefix);
|
||
// CompletionSuggestionBuilder suggestion = completionSuggestion("name_suggest")
|
||
// .prefix(prefix).size(size);
|
||
return new ArrayList<>();
|
||
}
|
||
|
||
/**
|
||
* 相关资源推荐（More Like This查询）
|
||
* 基于内容相似度推荐同类资源
|
||
*/
|
||
public List<Map<String, Object>> recommend(String resourceId, int size) {
|
||
logger.info(String.format("相关推荐: resource=%s, size=%d", resourceId, size));
|
||
// moreLikeThisQuery(["name","description","tags"], null, [item(INDEX, id)])
|
||
// .minTermFreq(1).maxQueryTerms(12)
|
||
return new ArrayList<>();
|
||
}
|
||
|
||
/** 索引单个资源文档 */
|
||
public void indexDocument(String resourceId, Map<String, Object> doc) {
|
||
logger.info("索引资源: id=" + resourceId);
|
||
}
|
||
|
||
/** 更新索引文档（部分更新） */
|
||
public void updateDocument(String resourceId, Map<String, Object> partialDoc) {
|
||
logger.info("更新索引: id=" + resourceId);
|
||
}
|
||
|
||
/** 删除索引文档 */
|
||
public void deleteDocument(String resourceId) {
|
||
logger.info("删除索引: id=" + resourceId);
|
||
}
|
||
|
||
/**
|
||
* 批量重建索引
|
||
* 从MySQL全量加载资源元数据，重新构建ES索引
|
||
*/
|
||
public int rebuildIndex() {
|
||
logger.info("开始重建ES索引...");
|
||
// 1. 删除旧索引
|
||
// 2. 重新创建索引（含映射）
|
||
createIndex();
|
||
// 3. 从MySQL批量查询所有审核通过的资源
|
||
// 4. 使用BulkRequest批量索引
|
||
int count = 0;
|
||
// List<Resource> allResources = resourceMapper.selectAllApproved();
|
||
// BulkRequest bulk = new BulkRequest();
|
||
// for (Resource r : allResources) {
|
||
// bulk.add(new IndexRequest(INDEX_NAME).id(r.getId()).source(toDoc(r)));
|
||
// count++;
|
||
// if (count % 500 == 0) {
|
||
// elasticsearchClient.bulk(bulk);
|
||
// bulk = new BulkRequest();
|
||
// }
|
||
// }
|
||
// if (bulk.numberOfActions() > 0) elasticsearchClient.bulk(bulk);
|
||
logger.info("ES索引重建完成: " + count + "条");
|
||
return count;
|
||
}
|
||
}
|