godfreyhe commented on code in PR #20363:
URL: https://github.com/apache/flink/pull/20363#discussion_r935226448
##########
flink-table/flink-table-api-java/src/main/java/org/apache/flink/table/api/internal/TableEnvironmentImpl.java:
##########
@@ -1894,4 +1926,319 @@ public TableImpl createTable(QueryOperation
tableOperation) {
public String explainPlan(InternalPlan compiledPlan, ExplainDetail...
extraDetails) {
return planner.explainPlan(compiledPlan, extraDetails);
}
+
+ private TableResultInternal analyzeTable(AnalyzeTableOperation operation)
+ throws TableNotPartitionedException, TableNotExistException,
PartitionNotExistException,
+ TablePartitionedException {
+ CatalogTable table =
+
catalogManager.getTable(operation.getTableIdentifier()).get().getTable();
+ ResolvedSchema schema =
+
table.getUnresolvedSchema().resolve(catalogManager.getSchemaResolver());
+ List<Column> columns =
+ operation.getColumns().stream()
+ .map(c -> schema.getColumn(c).get())
+ .collect(Collectors.toList());
+ Catalog catalog =
+
catalogManager.getCatalog(operation.getTableIdentifier().getCatalogName()).get();
+ ObjectPath objectPath = operation.getTableIdentifier().toObjectPath();
+
+ if (table.isPartitioned()) {
+ List<CatalogPartitionSpec> targetPartitions =
+
operation.getPartitionSpecs().orElse(catalog.listPartitions(objectPath));
+ for (CatalogPartitionSpec partitionSpec : targetPartitions) {
+ String statSql =
+ generateAnalyzeSql(operation.getTableIdentifier(),
partitionSpec, columns);
+ TableResult tableResult = executeSql(statSql);
+ List<Row> result =
CollectionUtil.iteratorToList(tableResult.collect());
+ Preconditions.checkArgument(result.size() == 1);
+ Row row = result.get(0);
+ CatalogTableStatistics tableStat =
convertToTableStatistics(row);
+ catalog.alterPartitionStatistics(objectPath, partitionSpec,
tableStat, false);
+ if (!columns.isEmpty()) {
+ CatalogColumnStatistics columnStat =
convertToColumnStatistics(row, columns);
+ catalog.alterPartitionColumnStatistics(
+ objectPath, partitionSpec, columnStat, false);
+ }
+ }
+ } else {
+ String statSql =
generateAnalyzeSql(operation.getTableIdentifier(), null, columns);
+ TableResult tableResult = executeSql(statSql);
+ List<Row> result =
CollectionUtil.iteratorToList(tableResult.collect());
+ Preconditions.checkArgument(result.size() == 1);
+ Row row = result.get(0);
+ CatalogTableStatistics tableStat = convertToTableStatistics(row);
+ catalog.alterTableStatistics(objectPath, tableStat, false);
+ if (!columns.isEmpty()) {
+ CatalogColumnStatistics columnStat =
convertToColumnStatistics(row, columns);
+ catalog.alterTableColumnStatistics(objectPath, columnStat,
false);
+ }
+ }
+ return TableResultImpl.TABLE_RESULT_OK;
+ }
+
+ private String generateAnalyzeSql(
+ ObjectIdentifier tableIdentifier,
+ @Nullable CatalogPartitionSpec partitionSpec,
+ List<Column> columns) {
+ Optional<ContextResolvedTable> optionalCatalogTable =
+ catalogManager.getTable(tableIdentifier);
+ Preconditions.checkArgument(
Review Comment:
I think it's necessary, because we should there is only one row in the result
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
