归并分类：
结果归并 - 图1
归并流程：
结果归并 - 图2

书接上文

mybatis的execute执行完成后，将要进行结果的归并，故进入到handleResultSets方法进行探寻源码。
image.png
经过Debug，定位到ShardingPreparedStatement的getResultSet，其源码如下：

  1.     public ResultSet getResultSet() throws SQLException {
  2.         if (null != this.currentResultSet) {
  3.             return this.currentResultSet;
  4.         } else {
  5.             if (this.executionContext.getSqlStatementContext() instanceof SelectStatementContext || this.executionContext.getSqlStatementContext().getSqlStatement() instanceof DALStatement) {
  6.                 List<ResultSet> resultSets = this.getResultSets();
  7.                 MergedResult mergedResult = this.mergeQuery(this.getQueryResults(resultSets));
  8.                 this.currentResultSet = new ShardingResultSet(resultSets, mergedResult, this, this.executionContext);
  9.             }
  11.             return this.currentResultSet;
  12.         }
  13.     }

源码分析

经过寻找在getResultSet中找到mergeQuery的归并关键字，以下将围绕mergeQuery进行分析，其源码如下：

  1.     private MergedResult mergeQuery(List<QueryResult> queryResults) throws SQLException {
  2.         ShardingRuntimeContext runtimeContext = this.connection.getRuntimeContext();
  3.         MergeEngine mergeEngine = new MergeEngine(((ShardingRule)runtimeContext.getRule()).toRules(), runtimeContext.getProperties(), runtimeContext.getDatabaseType(), runtimeContext.getMetaData().getSchema());
  4.         return mergeEngine.merge(queryResults, this.executionContext.getSqlStatementContext());
  5.     }

追寻merge：
image.png
process：
image.png
追寻merge：

  1.     private Optional<MergedResult> merge(List<QueryResult> queryResults, SQLStatementContext sqlStatementContext) throws SQLException {
  2.         Iterator var3 = this.engines.entrySet().iterator();
  4.         Entry entry;
  5.         do {
  6.             if (!var3.hasNext()) {
  7.                 return Optional.empty();
  8.             }
  10.             entry = (Entry)var3.next();
  11.         } while(!(entry.getValue() instanceof ResultMergerEngine));
  13.         ResultMerger resultMerger = ((ResultMergerEngine)entry.getValue()).newInstance(this.databaseType, (BaseRule)entry.getKey(), this.properties, sqlStatementContext);
  14.         return Optional.of(resultMerger.merge(queryResults, sqlStatementContext, this.schemaMetaData));
  15.     }

到这继续追寻merge方法，其有三个实现：
image.png
这就涉及到了DAL和DQL的问题，这里默认是DQL，继续探寻:

  1.     public MergedResult merge(List<QueryResult> queryResults, SQLStatementContext sqlStatementContext, SchemaMetaData schemaMetaData) throws SQLException {
  2.         if (1 == queryResults.size()) {
  3.             return new IteratorStreamMergedResult(queryResults);
  4.         } else {
  5.             Map<String, Integer> columnLabelIndexMap = this.getColumnLabelIndexMap((QueryResult)queryResults.get(0));
  6.             SelectStatementContext selectStatementContext = (SelectStatementContext)sqlStatementContext;
  7.             selectStatementContext.setIndexes(columnLabelIndexMap);
  8.             MergedResult mergedResult = this.build(queryResults, selectStatementContext, columnLabelIndexMap, schemaMetaData);
  9.             return this.decorate(queryResults, selectStatementContext, mergedResult);
  10.         }
  11.     }

终于找到了它的核心方法build：

  1.     private MergedResult build(List<QueryResult> queryResults, SelectStatementContext selectStatementContext, Map<String, Integer> columnLabelIndexMap, SchemaMetaData schemaMetaData) throws SQLException {
  2.         if (this.isNeedProcessGroupBy(selectStatementContext)) {
  3.             return this.getGroupByMergedResult(queryResults, selectStatementContext, columnLabelIndexMap, schemaMetaData);
  4.         } else if (this.isNeedProcessDistinctRow(selectStatementContext)) {
  5.             this.setGroupByForDistinctRow(selectStatementContext);
  6.             return this.getGroupByMergedResult(queryResults, selectStatementContext, columnLabelIndexMap, schemaMetaData);
  7.         } else {
  8.             return (MergedResult)(this.isNeedProcessOrderBy(selectStatementContext) ? new OrderByStreamMergedResult(queryResults, selectStatementContext, schemaMetaData) : new IteratorStreamMergedResult(queryResults));
  9.         }
  10.     }

这里涉及到了流式归并和内存归并。

流式归并

其源码如下：

  1.     public OrderByStreamMergedResult(List<QueryResult> queryResults, SelectStatementContext selectStatementContext, SchemaMetaData schemaMetaData) throws SQLException {
  2.         this.orderByItems = selectStatementContext.getOrderByContext().getItems();
  3.         this.orderByValuesQueue = new PriorityQueue(queryResults.size());
  4.         this.orderResultSetsToQueue(queryResults, selectStatementContext, schemaMetaData);
  5.         this.isFirstNext = true;
  6.     }

orderResultSetsToQueue方法中将结果存储到一个队列中：

  1.     private void orderResultSetsToQueue(List<QueryResult> queryResults, SelectStatementContext selectStatementContext, SchemaMetaData schemaMetaData) throws SQLException {
  2.         Iterator var4 = queryResults.iterator();
  4.         while(var4.hasNext()) {
  5.             QueryResult each = (QueryResult)var4.next();
  6.             OrderByValue orderByValue = new OrderByValue(each, this.orderByItems, selectStatementContext, schemaMetaData);
  7.             if (orderByValue.next()) {
  8.                 this.orderByValuesQueue.offer(orderByValue);
  9.             }
  10.         }
  12.         this.setCurrentQueryResult(this.orderByValuesQueue.isEmpty() ? (QueryResult)queryResults.get(0) : ((OrderByValue)this.orderByValuesQueue.peek()).getQueryResult());
  13.     }

内存归并

  1.     protected MemoryMergedResult(T rule, SchemaMetaData schemaMetaData, SQLStatementContext sqlStatementContext, List<QueryResult> queryResults) throws SQLException {
  2.         List<MemoryQueryResultRow> memoryQueryResultRowList = this.init(rule, schemaMetaData, sqlStatementContext, queryResults);
  3.         this.memoryResultSetRows = memoryQueryResultRowList.iterator();
  4.         if (!memoryQueryResultRowList.isEmpty()) {
  5.             this.currentResultSetRow = (MemoryQueryResultRow)memoryQueryResultRowList.get(0);
  6.         }
  8.     }

显然，是直接存在private final Object[] data中的。

装饰归并

装饰归并详见：

  1.     private MergedResult decorate(List<QueryResult> queryResults, SelectStatementContext selectStatementContext, MergedResult mergedResult) throws SQLException {
  2.         PaginationContext paginationContext = selectStatementContext.getPaginationContext();
  3.         if (paginationContext.isHasPagination() && 1 != queryResults.size()) {
  4.             String trunkDatabaseName = DatabaseTypes.getTrunkDatabaseType(this.databaseType.getName()).getName();
  5.             if (!"MySQL".equals(trunkDatabaseName) && !"PostgreSQL".equals(trunkDatabaseName)) {
  6.                 if ("Oracle".equals(trunkDatabaseName)) {
  7.                     return new RowNumberDecoratorMergedResult(mergedResult, paginationContext);
  8.                 } else {
  9.                     return (MergedResult)("SQLServer".equals(trunkDatabaseName) ? new TopAndRowNumberDecoratorMergedResult(mergedResult, paginationContext) : mergedResult);
  10.                 }
  11.             } else {
  12.                 return new LimitDecoratorMergedResult(mergedResult, paginationContext);
  13.             }
  14.         } else {
  15.             return mergedResult;
  16.         }
  17.     }