Merge pull request #54635 from Avogar/cluster-func-names-with-spaces

Fix schema inference for *Cluster functions for column names with spaces

src/Core/NamesAndTypes.cpp

@@ -240,4 +240,19 @@ size_t NamesAndTypesList::getPosByName(const std::string &name) const noexcept
     return pos;
 }
 
+String NamesAndTypesList::toNamesAndTypesDescription() const
+{
+    WriteBufferFromOwnString buf;
+    bool first = true;
+    for (const auto & name_and_type : *this)
+    {
+        if (!std::exchange(first, false))
+            writeCString(", ", buf);
+        writeBackQuotedString(name_and_type.name, buf);
+        writeChar(' ', buf);
+        writeString(name_and_type.type->getName(), buf);
+    }
+    return buf.str();
+}
+
 }

src/Core/NamesAndTypes.h

@@ -122,6 +122,8 @@ public:
 
     /// Try to get column position by name, returns number of columns if column isn't found
     size_t getPosByName(const std::string & name) const noexcept;
+
+    String toNamesAndTypesDescription() const;
 };
 
 using NamesAndTypesLists = std::vector<NamesAndTypesList>;

src/Databases/DatabaseDictionary.cpp

@@ -105,7 +105,7 @@ ASTPtr DatabaseDictionary::getCreateTableQueryImpl(const String & table_name, Co
 
         auto names_and_types = StorageDictionary::getNamesAndTypes(ExternalDictionariesLoader::getDictionaryStructure(*load_result.config));
         buffer << "CREATE TABLE " << backQuoteIfNeed(getDatabaseName()) << '.' << backQuoteIfNeed(table_name) << " (";
-        buffer << StorageDictionary::generateNamesAndTypesDescription(names_and_types);
+        buffer << names_and_types.toNamesAndTypesDescription();
         buffer << ") Engine = Dictionary(" << backQuoteIfNeed(table_name) << ")";
     }
 

src/Storages/IStorageCluster.cpp

@@ -77,7 +77,7 @@ Pipe IStorageCluster::read(
     const bool add_agg_info = processed_stage == QueryProcessingStage::WithMergeableState;
 
     if (!structure_argument_was_provided)
-        addColumnsStructureToQuery(query_to_send, StorageDictionary::generateNamesAndTypesDescription(storage_snapshot->metadata->getColumns().getAll()), context);
+        addColumnsStructureToQuery(query_to_send, storage_snapshot->metadata->getColumns().getAll().toNamesAndTypesDescription(), context);
 
     RestoreQualifiedNamesVisitor::Data data;
     data.distributed_table = DatabaseAndTableWithAlias(*getTableExpression(query_info.query->as<ASTSelectQuery &>(), 0));

src/Storages/StorageDictionary.cpp

@@ -39,7 +39,7 @@ namespace
             {
                 throw Exception(ErrorCodes::THERE_IS_NO_COLUMN, "Not found column {} {} in dictionary {}. There are only columns {}",
                                 column.name, column.type->getName(), backQuote(dictionary_name),
-                                StorageDictionary::generateNamesAndTypesDescription(dictionary_names_and_types));
+                                dictionary_names_and_types.toNamesAndTypesDescription());
             }
         }
     }
@@ -80,20 +80,6 @@ NamesAndTypesList StorageDictionary::getNamesAndTypes(const DictionaryStructure
     return dictionary_names_and_types;
 }
 
-
-String StorageDictionary::generateNamesAndTypesDescription(const NamesAndTypesList & list)
-{
-    WriteBufferFromOwnString ss;
-    bool first = true;
-    for (const auto & name_and_type : list)
-    {
-        if (!std::exchange(first, false))
-            ss << ", ";
-        ss << name_and_type.name << ' ' << name_and_type.type->getName();
-    }
-    return ss.str();
-}
-
 StorageDictionary::StorageDictionary(
     const StorageID & table_id_,
     const String & dictionary_name_,

src/Storages/StorageDictionary.h

@@ -81,7 +81,6 @@ public:
     std::shared_ptr<const IDictionary> getDictionary() const;
 
     static NamesAndTypesList getNamesAndTypes(const DictionaryStructure & dictionary_structure);
-    static String generateNamesAndTypesDescription(const NamesAndTypesList & list);
 
     bool isDictionary() const override { return true; }
     void shutdown() override;

tests/queries/0_stateless/02876_s3_cluster_schema_inference_names_with_spaces.reference

@@ -0,0 +1,2 @@
+a b c	Nullable(UInt8)					
+42

tests/queries/0_stateless/02876_s3_cluster_schema_inference_names_with_spaces.sql

@@ -0,0 +1,6 @@
+-- Tags: no-fasttest
+-- Tag no-fasttest: Depends on AWS
+
+desc s3Cluster(test_cluster_one_shard_three_replicas_localhost, 'http://localhost:11111/test/02876.parquet');
+select * from s3Cluster(test_cluster_one_shard_three_replicas_localhost, 'http://localhost:11111/test/02876.parquet');
+