Parser::CreateTableAsSelectStmt Class Reference

#include <ParserNode.h>

Inheritance diagram for Parser::CreateTableAsSelectStmt:

Collaboration diagram for Parser::CreateTableAsSelectStmt:

Public Member Functions
	CreateTableAsSelectStmt (const rapidjson::Value &payload)
	CreateTableAsSelectStmt (const std::string *table_name, const std::string *select_query, const bool is_temporary, const bool if_not_exists, std::list< NameValueAssign * > *s)
void	execute (const Catalog_Namespace::SessionInfo &session, bool read_only_mode) override
Public Member Functions inherited from Parser::InsertIntoTableAsSelectStmt
	InsertIntoTableAsSelectStmt (const rapidjson::Value &payload)
	InsertIntoTableAsSelectStmt (const std::string *table_name, const std::string *select_query, std::list< std::string * > *c)
void	populateData (QueryStateProxy, const TableDescriptor *td, bool validate_table, bool for_CTAS=false)
std::string &	get_table ()
std::string &	get_select_query ()
Public Member Functions inherited from Parser::DDLStmt
void	setColumnDescriptor (ColumnDescriptor &cd, const ColumnDef *coldef)
Public Member Functions inherited from Parser::Node
virtual	~Node ()

Private Attributes
bool	is_temporary_
bool	if_not_exists_
std::list< std::unique_ptr < NameValueAssign > >	storage_options_

Additional Inherited Members
Public Attributes inherited from Parser::InsertIntoTableAsSelectStmt
std::unique_ptr< QueryConnector >	leafs_connector_
Protected Attributes inherited from Parser::InsertIntoTableAsSelectStmt
std::vector< std::unique_ptr < std::string > >	column_list_
std::string	table_name_
std::string	select_query_

Detailed Description

Definition at line 1168 of file ParserNode.h.

Constructor & Destructor Documentation

Parser::CreateTableAsSelectStmt::CreateTableAsSelectStmt

(

const rapidjson::Value &

payload

)

Definition at line 4234 of file ParserNode.cpp.

References if_not_exists_, is_temporary_, json_bool(), Parser::anonymous_namespace{ParserNode.cpp}::parse_options(), and storage_options_.

    : InsertIntoTableAsSelectStmt(payload) {
  if (payload.HasMember("temporary")) {
    is_temporary_ = json_bool(payload["temporary"]);
  } else {
    is_temporary_ = false;
  }

  if (payload.HasMember("ifNotExists")) {
    if_not_exists_ = json_bool(payload["ifNotExists"]);
  } else {
    if_not_exists_ = false;
  }

  parse_options(payload, storage_options_);
}

Here is the call graph for this function:

Parser::CreateTableAsSelectStmt::CreateTableAsSelectStmt ( const std::string *  table_name, const std::string *  select_query, const bool  is_temporary, const bool  if_not_exists, std::list< NameValueAssign * > *  s  )

inline

Definition at line 1171 of file ParserNode.h.

References storage_options_.

      : InsertIntoTableAsSelectStmt(table_name, select_query, nullptr)
      , is_temporary_(is_temporary)
      , if_not_exists_(if_not_exists) {
    if (s) {
      for (const auto& e : *s) {
        storage_options_.emplace_back(e);
      }
      delete s;
    }
  }

Member Function Documentation

void Parser::CreateTableAsSelectStmt::execute ( const Catalog_Namespace::SessionInfo &  session, bool  read_only_mode  )

overridevirtual

Reimplemented from Parser::InsertIntoTableAsSelectStmt.

Definition at line 4251 of file ParserNode.cpp.

References Parser::anonymous_namespace{ParserNode.cpp}::acquire_query_table_locks(), CHECK, Catalog_Namespace::SessionInfo::checkDBAccessPrivileges(), Data_Namespace::CPU_LEVEL, query_state::QueryState::create(), bench_batch_streaming_ingest::create_table(), AccessPrivileges::CREATE_TABLE, DEFAULT_FRAGMENT_ROWS, DEFAULT_MAX_CHUNK_SIZE, DEFAULT_MAX_ROLLBACK_EPOCHS, DEFAULT_MAX_ROWS, DEFAULT_PAGE_SIZE, Data_Namespace::DISK_LEVEL, legacylockmgr::ExecutorOuterLock, TableDescriptor::fragmenter, TableDescriptor::fragPageSize, TableDescriptor::fragType, g_cluster, Catalog_Namespace::SessionInfo::get_currentUser(), Parser::StringLiteral::get_stringval(), Parser::anonymous_namespace{ParserNode.cpp}::get_table_definitions_for_ctas(), Catalog_Namespace::SessionInfo::getCatalog(), Parser::LocalQueryConnector::getColumnDescriptors(), legacylockmgr::LockMgr< MutexType, KeyType >::getMutex(), if_not_exists_, logger::INFO, Fragmenter_Namespace::INSERT_ORDER, is_temporary_, TableDescriptor::isView, kENCODING_DICT, kENCODING_GEOINT, TableDescriptor::keyMetainfo, sql_constants::kMaxNumericPrecision, Parser::InsertIntoTableAsSelectStmt::leafs_connector_, LOG, TableDescriptor::maxChunkSize, TableDescriptor::maxFragRows, TableDescriptor::maxRollbackEpochs, TableDescriptor::maxRows, TableDescriptor::nColumns, TableDescriptor::persistenceLevel, Parser::InsertIntoTableAsSelectStmt::populateData(), Parser::LocalQueryConnector::query(), Parser::InsertIntoTableAsSelectStmt::select_query_, Parser::anonymous_namespace{ParserNode.cpp}::serialize_key_metainfo(), STDLOG, storage_options_, Parser::InsertIntoTableAsSelectStmt::table_name_, TableDBObjectType, TableDescriptor::tableName, to_string(), TableDescriptor::userId, and Catalog_Namespace::UserMetadata::userId.

Referenced by heavydb.cursor.Cursor::executemany().

                                                           {
  if (read_only_mode) {
    throw std::runtime_error("CREATE TABLE invalid in read only mode.");
  }
  auto session_copy = session;
  auto session_ptr = std::shared_ptr<Catalog_Namespace::SessionInfo>(
      &session_copy, boost::null_deleter());
  auto query_state = query_state::QueryState::create(session_ptr, select_query_);
  auto stdlog = STDLOG(query_state);
  LocalQueryConnector local_connector;
  auto& catalog = session.getCatalog();
  bool create_table = nullptr == leafs_connector_;

  std::set<std::string> select_tables;
  if (create_table) {
    const auto execute_write_lock =
        heavyai::unique_lock<legacylockmgr::WrapperType<heavyai::shared_mutex>>(
            *legacylockmgr::LockMgr<heavyai::shared_mutex, bool>::getMutex(
                legacylockmgr::ExecutorOuterLock, true));

    // check access privileges
    if (!session.checkDBAccessPrivileges(DBObjectType::TableDBObjectType,
                                         AccessPrivileges::CREATE_TABLE)) {
      throw std::runtime_error("CTAS failed. Table " + table_name_ +
                               " will not be created. User has no create privileges.");
    }

    if (catalog.getMetadataForTable(table_name_) != nullptr) {
      if (if_not_exists_) {
        return;
      }
      throw std::runtime_error("Table " + table_name_ +
                               " already exists and no data was loaded.");
    }

    // only validate the select query so we get the target types
    // correctly, but do not populate the result set
    // we currently have exclusive access to the system so this is safe
    auto validate_result = local_connector.query(
        query_state->createQueryStateProxy(), select_query_, {}, true, false);

    auto column_descriptors_for_create =
        local_connector.getColumnDescriptors(validate_result, true);

    // some validation as the QE might return some out of range column types
    for (auto& cd : column_descriptors_for_create) {
      if (cd.columnType.is_decimal() &&
          cd.columnType.get_precision() > sql_constants::kMaxNumericPrecision) {
        throw std::runtime_error(cd.columnName + ": Precision too high, max " +
                                 std::to_string(sql_constants::kMaxNumericPrecision) +
                                 ".");
      }
    }

    TableDescriptor td;
    td.tableName = table_name_;
    td.userId = session.get_currentUser().userId;
    td.nColumns = column_descriptors_for_create.size();
    td.isView = false;
    td.fragmenter = nullptr;
    td.fragType = Fragmenter_Namespace::FragmenterType::INSERT_ORDER;
    td.maxFragRows = DEFAULT_FRAGMENT_ROWS;
    td.maxChunkSize = DEFAULT_MAX_CHUNK_SIZE;
    td.fragPageSize = DEFAULT_PAGE_SIZE;
    td.maxRows = DEFAULT_MAX_ROWS;
    td.maxRollbackEpochs = DEFAULT_MAX_ROLLBACK_EPOCHS;
    if (is_temporary_) {
      td.persistenceLevel = Data_Namespace::MemoryLevel::CPU_LEVEL;
    } else {
      td.persistenceLevel = Data_Namespace::MemoryLevel::DISK_LEVEL;
    }

    bool use_shared_dictionaries = true;
    bool force_geo_compression = true;

    if (!storage_options_.empty()) {
      for (auto& p : storage_options_) {
        if (boost::to_lower_copy<std::string>(*p->get_name()) ==
            "use_shared_dictionaries") {
          const StringLiteral* literal =
              dynamic_cast<const StringLiteral*>(p->get_value());
          if (nullptr == literal) {
            throw std::runtime_error(
                "USE_SHARED_DICTIONARIES must be a string parameter");
          }
          std::string val = boost::to_lower_copy<std::string>(*literal->get_stringval());
          use_shared_dictionaries = val == "true" || val == "1" || val == "t";
        } else if (boost::to_lower_copy<std::string>(*p->get_name()) ==
                   "force_geo_compression") {
          const StringLiteral* literal =
              dynamic_cast<const StringLiteral*>(p->get_value());
          if (nullptr == literal) {
            throw std::runtime_error("FORCE_GEO_COMPRESSION must be a string parameter");
          }
          std::string val = boost::to_lower_copy<std::string>(*literal->get_stringval());
          force_geo_compression = val == "true" || val == "1" || val == "t";
        } else {
          get_table_definitions_for_ctas(td, p, column_descriptors_for_create);
        }
      }
    }

    std::vector<SharedDictionaryDef> sharedDictionaryRefs;

    if (use_shared_dictionaries) {
      const auto source_column_descriptors =
          local_connector.getColumnDescriptors(validate_result, false);
      const auto mapping = catalog.getDictionaryToColumnMapping();

      for (auto& source_cd : source_column_descriptors) {
        const auto& ti = source_cd.columnType;
        if (ti.is_string()) {
          if (ti.get_compression() == kENCODING_DICT) {
            int dict_id = ti.get_comp_param();
            auto it = mapping.find(dict_id);
            if (mapping.end() != it) {
              const auto targetColumn = it->second;
              auto targetTable =
                  catalog.getMetadataForTable(targetColumn->tableId, false);
              CHECK(targetTable);
              LOG(INFO) << "CTAS: sharing text dictionary on column "
                        << source_cd.columnName << " with " << targetTable->tableName
                        << "." << targetColumn->columnName;
              sharedDictionaryRefs.emplace_back(
                  source_cd.columnName, targetTable->tableName, targetColumn->columnName);
            }
          }
        }
      }
    }

    if (force_geo_compression) {
      for (auto& cd_for_create : column_descriptors_for_create) {
        auto& ti = cd_for_create.columnType;
        if (ti.is_geometry() && ti.get_output_srid() == 4326) {
          // turn on GEOINT32 compression
          ti.set_compression(kENCODING_GEOINT);
          ti.set_comp_param(32);
        }
      }
    }

    // currently no means of defining sharding in CTAS
    td.keyMetainfo = serialize_key_metainfo(nullptr, sharedDictionaryRefs);

    catalog.createTable(td, column_descriptors_for_create, sharedDictionaryRefs, true);
    // TODO (max): It's transactionally unsafe, should be fixed: we may create object
    // w/o privileges
    SysCatalog::instance().createDBObject(
        session.get_currentUser(), td.tableName, TableDBObjectType, catalog);
  }

  // note there is a time where we do not have any executor outer lock here. someone could
  // come along and mess with the data or other tables.
  const auto execute_read_lock =
      heavyai::shared_lock<legacylockmgr::WrapperType<heavyai::shared_mutex>>(
          *legacylockmgr::LockMgr<heavyai::shared_mutex, bool>::getMutex(
              legacylockmgr::ExecutorOuterLock, true));

  auto locks = acquire_query_table_locks(
      catalog.name(), select_query_, query_state->createQueryStateProxy(), table_name_);
  const TableDescriptor* td = catalog.getMetadataForTable(table_name_);
  try {
    populateData(query_state->createQueryStateProxy(), td, false, true);
  } catch (...) {
    if (!g_cluster) {
      const TableDescriptor* created_td = catalog.getMetadataForTable(table_name_);
      if (created_td) {
        catalog.dropTable(created_td);
      }
    }
    throw;
  }
}

Here is the call graph for this function:

Here is the caller graph for this function:

Member Data Documentation

bool Parser::CreateTableAsSelectStmt::if_not_exists_

private

Definition at line 1192 of file ParserNode.h.

Referenced by CreateTableAsSelectStmt(), and execute().

bool Parser::CreateTableAsSelectStmt::is_temporary_

private

Definition at line 1191 of file ParserNode.h.

Referenced by CreateTableAsSelectStmt(), and execute().

std::list<std::unique_ptr<NameValueAssign> > Parser::CreateTableAsSelectStmt::storage_options_

private

Definition at line 1193 of file ParserNode.h.

Referenced by CreateTableAsSelectStmt(), and execute().

---

The documentation for this class was generated from the following files:

• /home/jenkins-slave/workspace/core-os-doxygen/Parser/ParserNode.h
• /home/jenkins-slave/workspace/core-os-doxygen/Parser/ParserNode.cpp