ES分页去重

wylc123 6月前 ⋅ 657 阅读
/**
     * 分组去重
     */
    public static Map<String, Object> searchGroupHistoryQuery(SearchHistoryQuery historyQuery, String esIndex) {
        RestHighLevelClient client = getRestHighLevelClient();
        Map<String, Object> result = new HashMap<>();
        try {
            List<Map<String, Object>> list = new ArrayList<>();
            SearchRequest searchRequest = new SearchRequest(esIndex);
            SearchSourceBuilder sourceBuilder = new SearchSourceBuilder();
            int currentPage = 1, pageSize = 10;
            if (historyQuery.getCurrentPage() != null) {
                currentPage = historyQuery.getCurrentPage();
            }
            if (historyQuery.getPageSize() != null) {
                pageSize = historyQuery.getPageSize();
            }

            sourceBuilder.from((currentPage - 1) * pageSize);
            sourceBuilder.size(pageSize);
            sourceBuilder.sort("createTime", SortOrder.DESC);

            BoolQueryBuilder boolQueryBuilder = QueryBuilders.boolQuery()
                    .must(QueryBuilders.termQuery("userCode", historyQuery.getUserCode()))
                    .must(QueryBuilders.termQuery("status", 1))
                    .mustNot(QueryBuilders.termQuery("query.keyword",""));
            if (historyQuery.getSearchType() != null) {
                TermQueryBuilder matchQueryBuilder = QueryBuilders.termQuery("searchType", historyQuery.getSearchType());
                boolQueryBuilder.must(matchQueryBuilder);
            }
            sourceBuilder.query(boolQueryBuilder);

            // 将重复数据折叠
            CollapseBuilder collapseBuilder = new CollapseBuilder("query.keyword");
            sourceBuilder.collapse(collapseBuilder);
            AggregationBuilder aggregation = AggregationBuilders.cardinality("total_size").field("query.keyword");
            sourceBuilder.aggregation(aggregation);
            // 设置聚合(去重)
            // 创建 terms 聚合
//            TermsAggregationBuilder termsAggregation = AggregationBuilders.terms("unique_values")
//                    .field("query.keyword")
//                    .size(pageSize);
//
//            // 在 terms 聚合内部创建 cardinality 聚合
////            CardinalityAggregationBuilder cardinalityAggregation = AggregationBuilders.cardinality("distinct_count")
////                    .field("query.keyword");
//
//            // 创建 top_hits 子聚合
//            TopHitsAggregationBuilder topHitsAggregation = AggregationBuilders.topHits("top_docs")
//                    .size(pageSize)
//                    .fetchSource(new String[]{"searchHistoryID","ip","productCode","productName","query","queryJson","resCode","resName","searchRange","searchType","uf","userCode","uv"}, null);
//
//            // 将 cardinality 聚合添加到 terms 聚合中
//            termsAggregation.subAggregation(topHitsAggregation);
//
//            // 将 terms 聚合添加到搜索源构建器中
//            sourceBuilder.aggregation(termsAggregation);

            searchRequest.source(sourceBuilder);
            SearchResponse searchResponse = client.search(searchRequest, RequestOptions.DEFAULT);
//            SearchResponse searchResponse = client.search(searchRequest, RequestOptions.DEFAULT);
            // 处理结果
//            Aggregation aggregation = searchResponse.getAggregations().get("unique_values");
//            if (aggregation instanceof ParsedStringTerms) {
//                ParsedStringTerms parsedTerms = (ParsedStringTerms) aggregation;
//                for (Terms.Bucket bucket : parsedTerms.getBuckets()) {
//                    Map<String, Object> sourceAsMap = new HashMap<>();
//                    String term = bucket.getKeyAsString();
//                    // 处理每个分组的信息
//                    ParsedTopHits topHits = bucket.getAggregations().get("top_docs");
//                    for (SearchHit hit : topHits.getHits()) {
//                        sourceAsMap = hit.getSourceAsMap();
//                    }
//                    sourceAsMap.put("query", term);
//                    list.add(sourceAsMap);
//                }
//            }
            for (SearchHit doc : searchResponse.getHits().getHits()) {
                Map<String, Object> sourceAsMap = doc.getSourceAsMap();
                list.add(sourceAsMap);
            }
//            int total = (int) searchResponse.getHits().getTotalHits().value;
//            int total = getTotal(historyQuery, esIndex);
            assert searchResponse != null;
            ParsedCardinality parsedCardinality = (ParsedCardinality) searchResponse.getAggregations().asList().get(0);
            int total = (int)parsedCardinality.getValue();
            result.put("data", list);
            result.put("total", total);
            result.put("totalPage", total == 0 ? 0 : (total % pageSize == 0 ? total / pageSize : (total / pageSize) + 1));
            result.put("pageSize", pageSize);
            result.put("currentPage", currentPage);
        } catch (Exception ex) {
            log.error("getSearchHistory error: ", ex);
        } finally {
            //closeClient(client);
        }

        return result;
    }

相关文章推荐

全部评论: 0

    我有话说: