Merge branch 'feature' into 'dev'

2024/6/7 原发溯源-关键词匹配逻辑调整、事件影响力更新结果导出表头调整 test

See merge request !533

src/main/java/com/zhiwei/brandkbs2/easyexcel/dto/ExportBytedanceEventComputeResultDTO.java

@@ -16,15 +16,15 @@ public class ExportBytedanceEventComputeResultDTO {
     @ExcelProperty("事件名")
     private String name;
 
-    @ExcelProperty("渠道指标")
+    @ExcelProperty("综合影响力")
+    private Double finalInf;
+
+    @ExcelProperty("渠道指数")
     private Double inf;
 
-    @ExcelProperty("流量指标")
+    @ExcelProperty("流量指数")
     private Double RMI;
 
-    @ExcelProperty("综合影响力")
-    private Double finalInf;
-
     @ExcelProperty("传播量")
     private Integer dataCount;
 }

src/main/java/com/zhiwei/brandkbs2/service/impl/MarkDataServiceImpl.java

@@ -103,6 +103,8 @@ public class MarkDataServiceImpl implements MarkDataService {
 
     private static final String[] EVENT_FETCH_SOURCE = new String[]{"id", "ind_full_text", "c5", "real_source", "source", "mtime", "time", "url", "mtag", "foreign", "brandkbs_mark_cache_maps"};
 
+    private static final String ORIGIN_SEARCH_PATTERN = "[\\p{P}+~$`^=|<>～｀＄＾＋＝｜＜＞￥×\\s\u200B\u200C\u200D\u00A0\u0020\u3000\\d]";
+
     @Value("${istarshine.addIStarShineKSData.url}")
     private String addIStarShineKSDataUrl;
 
@@ -3515,7 +3517,6 @@ public class MarkDataServiceImpl implements MarkDataService {
         try {
             BoolQueryBuilder query = getAllIndexSearchOriginQuery(dto);
             String[] indexes = esClientDao.getAllIndexes(dto.getStartTime(), dto.getEndTime());
-            Long total = esClientDao.count(indexes, query, null);
             // platform query
             if (Objects.equals("其他", dto.getPlatforms().get(0))){
                 List<String> platforms = getWholeSearchPlatformPercentage(dto).stream().map(json -> json.getString("platform")).collect(Collectors.toList());
@@ -3526,9 +3527,10 @@ public class MarkDataServiceImpl implements MarkDataService {
                 query.must(EsQueryTools.assemblePlatformQuery(GlobalPojo.PLATFORMS.stream()
                         .filter(platform -> Objects.equals(dto.getPlatforms().get(0), platform.getName())).collect(Collectors.toList())));
             }
+            Long total = esClientDao.count(indexes, query, null);
             int page = dto.getPage();
             int pageSize = dto.getPageSize();
-            FieldSortBuilder sort = new FieldSortBuilder("time").order(SortOrder.ASC);
+            FieldSortBuilder sort = new FieldSortBuilder(GenericAttribute.ES_TIME).order(SortOrder.DESC);
             SearchHits hits = esClientDao.searchHits(indexes, query, null, null, sort, (page - 1) * pageSize, pageSize, null);
             List<JSONObject> list = new ArrayList<>();
             for (SearchHit hit : hits) {
@@ -3537,7 +3539,7 @@ public class MarkDataServiceImpl implements MarkDataService {
                 jsonObject.put("platform", baseMap.getPlatform());
                 jsonObject.put("source", baseMap.getSource());
                 jsonObject.put("realSource", baseMap.getRealSource());
-                jsonObject.put("title", baseMap.getTitle());
+                jsonObject.put("title", baseMap.getTitleNullOptionalContent());
                 jsonObject.put("url", baseMap.getUrl());
                 jsonObject.put("time", baseMap.getTime());
                 list.add(jsonObject);
@@ -3549,17 +3551,45 @@ public class MarkDataServiceImpl implements MarkDataService {
         return null;
     }
 
+    /**
+     * 原发溯源大库es查询
+     * @param dto
+     * @return
+     */
     private BoolQueryBuilder getAllIndexSearchOriginQuery(MarkSearchDTO dto) {
+        BoolQueryBuilder query = QueryBuilders.boolQuery();
+        // time
         dto.setEndTime(System.currentTimeMillis());
         dto.setStartTime(dto.getEndTime() - Constant.ONE_MONTH);
-        // keyword
+        query.must(QueryBuilders.rangeQuery("time").gte(dto.getStartTime()).lte(dto.getEndTime()));
+        String keyword = dto.getKeyword();
+        // 特殊符号、数字替换空格
+        String processedKeyword = Tools.canonicalKeyword(keyword.replaceAll(ORIGIN_SEARCH_PATTERN, " "));
         BoolQueryBuilder keywordQuery = QueryBuilders.boolQuery();
-        // 使用matchPhrasePrefixQuery，保证单词被截取后也能查询
-        keywordQuery.must(QueryBuilders.matchPhrasePrefixQuery(GenericAttribute.ES_IND_FULL_TEXT, dto.getKeyword()));
+        // keyword条件一： 空格为且条件
+        if (StringUtils.isNotEmpty(processedKeyword)) {
+            BoolQueryBuilder keywordQuery1 = QueryBuilders.boolQuery();
+            String[] words = processedKeyword.split(" +");
+            for (int i = 0; i < words.length; i++) {
+                BoolQueryBuilder wordQuery = QueryBuilders.boolQuery();
+                if (i != words.length - 1) {
+                    keywordQuery1.must(wordQuery.must(QueryBuilders.matchPhraseQuery(GenericAttribute.ES_IND_FULL_TEXT, words[i])));
+                } else {
+                    keywordQuery1.must(wordQuery.must(QueryBuilders.matchPhrasePrefixQuery(GenericAttribute.ES_IND_FULL_TEXT, words[i])));
+                }
+            }
+            keywordQuery.should(keywordQuery1);
+        }
+        // keyword条件二
+        BoolQueryBuilder keywordQuery2 = QueryBuilders.boolQuery();
+        keywordQuery2.must(QueryBuilders.matchPhrasePrefixQuery(GenericAttribute.ES_IND_FULL_TEXT, keyword));
+        // 两个keyword条件满足一个即可
+        keywordQuery.should(keywordQuery2).minimumShouldMatch(1);
+        query.must(keywordQuery);
         // platform
         BoolQueryBuilder platformQuery = EsQueryTools.assemblePlatformQuery(GlobalPojo.PLATFORMS);
-        return QueryBuilders.boolQuery().must(keywordQuery).must(platformQuery)
-                .must(QueryBuilders.rangeQuery("time").gte(dto.getStartTime()).lte(dto.getEndTime()));
+        query.must(platformQuery);
+        return query;
     }
 
     private JSONObject getDailyReportSummary(JSONObject yuqingAmount, JSONObject yuqingEmotion, JSONObject platformPercentage){