/* Copyright (C) 2003 MySQL AB This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version. This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You should have received a copy of the GNU General Public License along with this program; if not, write to the Free Software Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA */ /***************************************************************************** * Name: NdbDictionary.hpp * Include: * Link: * Author: Jonas Oreland * Date: 2003-05-14 * Version: 0.1 * Description: Data dictionary support * Documentation: * Adjust: 2003-05-14 Jonas Oreland First version. ****************************************************************************/ #ifndef NdbDictionary_H #define NdbDictionary_H #include class Ndb; struct charset_info_st; typedef struct charset_info_st CHARSET_INFO; /** * @class NdbDictionary * @brief Data dictionary class * * This class supports all schema data definition and enquiry such as: * -# Creating tables (Dictionary::createTable) and table columns * -# Dropping tables (Dictionary::dropTable) * -# Creating secondary indexes (Dictionary::createIndex) * -# Dropping secondary indexes (Dictionary::dropIndex) * -# Enquiries about tables * (Dictionary::getTable, Table::getNoOfColumns, * Table::getPrimaryKey, and Table::getNoOfPrimaryKeys) * -# Enquiries about indexes * (Dictionary::getIndex, Index::getNoOfColumns, * and Index::getColumn) * * NdbDictionary has several help (inner) classes: * -# NdbDictionary::Table for creating tables * -# NdbDictionary::Column for creating table columns * -# NdbDictionary::Index for creating secondary indexes * * See @ref ndbapi_example4.cpp for details of usage. */ class NdbDictionary { public: /** * @class Object * @brief Meta information about a database object (a table, index, etc) */ class Object { public: /** * Status of object */ enum Status { New, ///< The object only exist in memory and ///< has not been created in the NDB Kernel Changed, ///< The object has been modified in memory ///< and has to be commited in NDB Kernel for ///< changes to take effect Retrieved, ///< The object exist and has been read ///< into main memory from NDB Kernel Invalid, ///< The object has been invalidated ///< and should not be used Altered ///< Table has been altered in NDB kernel ///< but is still valid for usage }; /** * Get status of object */ virtual Status getObjectStatus() const = 0; /** * Get version of object */ virtual int getObjectVersion() const = 0; /** * Object type */ enum Type { TypeUndefined = 0, ///< Undefined SystemTable = 1, ///< System table UserTable = 2, ///< User table (may be temporary) UniqueHashIndex = 3, ///< Unique un-ordered hash index HashIndex = 4, ///< Non-unique un-ordered hash index UniqueOrderedIndex = 5, ///< Unique ordered index OrderedIndex = 6, ///< Non-unique ordered index HashIndexTrigger = 7, ///< Index maintenance, internal IndexTrigger = 8, ///< Index maintenance, internal SubscriptionTrigger = 9,///< Backup or replication, internal ReadOnlyConstraint = 10 ///< Trigger, internal }; /** * Object state */ enum State { StateUndefined = 0, ///< Undefined StateOffline = 1, ///< Offline, not usable StateBuilding = 2, ///< Building, not yet usable StateDropping = 3, ///< Offlining or dropping, not usable StateOnline = 4, ///< Online, usable StateBackup = 5, ///< Online, being backuped, usable StateBroken = 9 ///< Broken, should be dropped and re-created }; /** * Object store */ enum Store { StoreUndefined = 0, ///< Undefined StoreTemporary = 1, ///< Object or data deleted on system restart StorePermanent = 2 ///< Permanent. logged to disk }; /** * Type of fragmentation. * * This parameter specifies how data in the table or index will * be distributed among the db nodes in the cluster.
* The bigger the table the more number of fragments should be used. * Note that all replicas count as same "fragment".
* For a table, default is FragAllMedium. For a unique hash index, * default is taken from underlying table and cannot currently * be changed. */ enum FragmentType { FragUndefined = 0, ///< Fragmentation type undefined or default FragSingle = 1, ///< Only one fragment FragAllSmall = 2, ///< One fragment per node, default FragAllMedium = 3, ///< two fragments per node FragAllLarge = 4 ///< Four fragments per node. }; }; class Table; // forward declaration /** * @class Column * @brief Represents an column in an NDB Cluster table * * Each column has a type. The type of a column is determind by a number * of type specifiers. * The type specifiers are: * - Builtin type * - Array length or max length * - Precision and scale */ class Column { public: /** * The builtin column types */ enum Type { Undefined=0,///< Undefined Tinyint, ///< 8 bit. 1 byte signed integer, can be used in array Tinyunsigned, ///< 8 bit. 1 byte unsigned integer, can be used in array Smallint, ///< 16 bit. 2 byte signed integer, can be used in array Smallunsigned, ///< 16 bit. 2 byte unsigned integer, can be used in array Mediumint, ///< 24 bit. 3 byte signed integer, can be used in array Mediumunsigned,///< 24 bit. 3 byte unsigned integer, can be used in array Int, ///< 32 bit. 4 byte signed integer, can be used in array Unsigned, ///< 32 bit. 4 byte unsigned integer, can be used in array Bigint, ///< 64 bit. 8 byte signed integer, can be used in array Bigunsigned, ///< 64 Bit. 8 byte signed integer, can be used in array Float, ///< 32-bit float. 4 bytes float, can be used in array Double, ///< 64-bit float. 8 byte float, can be used in array Olddecimal, ///< MySQL < 5.0 signed decimal, Precision, Scale Char, ///< Len. A fixed array of 1-byte chars Varchar, ///< Max len Binary, ///< Len Varbinary, ///< Max len Datetime, ///< Precision down to 1 sec (sizeof(Datetime) == 8 bytes ) Date, ///< Precision down to 1 day(sizeof(Date) == 4 bytes ) Blob, ///< Binary large object (see NdbBlob) Text, ///< Text blob Time = 25, ///< Time without date Year = 26, ///< Year 1901-2155 (1 byte) Timestamp = 27, ///< Unix time Olddecimalunsigned = 28 }; /** * @name General * @{ */ /** * Constructor * @param name Name of column */ Column(const char * name = ""); /** * Copy constructor * @param column Column to be copied */ Column(const Column& column); ~Column(); /** * Set name of column * @param name Name of the column */ void setName(const char * name); /** * Get name of column * @return Name of the column */ const char* getName() const; /** * Set whether column is nullable or not */ void setNullable(bool); /** * Get if the column is nullable or not */ bool getNullable() const; /** * Set that column is part of primary key */ void setPrimaryKey(bool); /** * Check if column is part of primary key */ bool getPrimaryKey() const; /** * Get number of column (horizontal position within table) */ int getColumnNo() const; /** * Check if column is equal to some other column * @param column Column to compare with * @return true if column is equal to some other column otherwise false. */ bool equal(const Column& column) const; /** @} *******************************************************************/ /** * @name Type Specifiers * @{ */ /** * Set type of column * @param type Type of column * * @note setType resets all column attributes * to (type dependent) defaults and should be the first * method to call. Default type is Unsigned. */ void setType(Type type); /** * Get type of column */ Type getType() const; /** * Set precision of column. * @note Only applicable for decimal types */ void setPrecision(int); /** * Get precision of column. * @note Only applicable for decimal types */ int getPrecision() const; /** * Set scale of column. * @note Only applicable for decimal types */ void setScale(int); /** * Get scale of column. * @note Only applicable for decimal types */ int getScale() const; /** * Set length for column * Array length for column or max length for variable length arrays. */ void setLength(int length); /** * Get length for column * Array length for column or max length for variable length arrays. */ int getLength() const; /** * For Char or Varchar or Text, set or get MySQL CHARSET_INFO. This * specifies both character set and collation. See get_charset() * etc in MySQL. (The cs is not "const" in MySQL). */ void setCharset(CHARSET_INFO* cs); CHARSET_INFO* getCharset() const; /** * For blob, set or get "inline size" i.e. number of initial bytes * to store in table's blob attribute. This part is normally in * main memory and can be indexed and interpreted. */ void setInlineSize(int size); int getInlineSize() const; /** * For blob, set or get "part size" i.e. number of bytes to store in * each tuple of the "blob table". Can be set to zero to omit parts * and to allow only inline bytes ("tinyblob"). */ void setPartSize(int size); int getPartSize() const; /** * For blob, set or get "stripe size" i.e. number of consecutive * parts to store in each node group. */ void setStripeSize(int size); int getStripeSize() const; /** * Get size of element */ int getSize() const; /** * Set distribution key * * A distribution key is a set of attributes which are used * to distribute the tuples onto the NDB nodes. * The distribution key uses the NDB Cluster hashing function. * * An example where this is useful is TPC-C where it might be * good to use the warehouse id and district id as the distribution key. * This would place all data for a specific district and warehouse * in the same database node. * * Locally in the fragments the full primary key * will still be used with the hashing algorithm. * * @param enable If set to true, then the column will be part of * the distribution key. */ void setDistributionKey(bool enable); /** * Check if column is part of distribution key * @see setDistributionKey */ bool getDistributionKey() const; /** @} *******************************************************************/ #ifndef DOXYGEN_SHOULD_SKIP_INTERNAL void setTupleKey(bool); bool getTupleKey() const; void setDistributionGroup(bool, int bits = 16); bool getDistributionGroup() const; int getDistributionGroupBits() const; void setIndexOnlyStorage(bool); bool getIndexOnlyStorage() const; const Table * getBlobTable() const; /** * @name ODBC Specific methods * @{ */ void setAutoIncrement(bool); bool getAutoIncrement() const; void setAutoIncrementInitialValue(Uint64 val); void setDefaultValue(const char*); const char* getDefaultValue() const; /** @} *******************************************************************/ static const Column * FRAGMENT; static const Column * ROW_COUNT; static const Column * COMMIT_COUNT; #endif private: friend class NdbRecAttr; friend class NdbColumnImpl; class NdbColumnImpl & m_impl; Column(NdbColumnImpl&); Column& operator=(const Column&); }; #ifndef DOXYGEN_SHOULD_SKIP_INTERNAL /** * ??? */ typedef Column Attribute; #endif /** * @brief Represents a table in NDB Cluster * * TableSize
* When calculating the data storage one should add the size of all * attributes (each attributeconsumes at least 4 bytes) and also an overhead * of 12 byte. Variable size attributes (not supported yet) will have a * size of 12 bytes plus the actual data storage parts where there is an * additional overhead based on the size of the variable part.
* An example table with 5 attributes: * one 64 bit attribute, one 32 bit attribute, * two 16 bit attributes and one array of 64 8 bits. * This table will consume * 12 (overhead) + 8 + 4 + 2*4 (4 is minimum) + 64 = 96 bytes per record. * Additionally an overhead of about 2 % as page headers and waste should * be allocated. Thus, 1 million records should consume 96 MBytes * plus the overhead 2 MByte and rounded up to 100 000 kBytes.
* */ class Table : public Object { public: /** * @name General * @{ */ /** * Constructor * @param name Name of table */ Table(const char * name = ""); /** * Copy constructor * @param table Table to be copied */ Table(const Table& table); virtual ~Table(); /** * Assignment operator, deep copy * @param table Table to be copied */ Table& operator=(const Table&); /** * Name of table * @param name Name of table */ void setName(const char * name); /** * Get table name */ const char * getName() const; /** * Get table id */ int getTableId() const; /** * Add a column definition to a table * @note creates a copy */ void addColumn(const Column &); /** * Get column definition via name. * @return null if none existing name */ const Column* getColumn(const char * name) const; /** * Get column definition via index in table. * @return null if none existing name */ Column* getColumn(const int attributeId); /** * Get column definition via name. * @return null if none existing name */ Column* getColumn(const char * name); /** * Get column definition via index in table. * @return null if none existing name */ const Column* getColumn(const int attributeId) const; /** @} *******************************************************************/ /** * @name Storage * @{ */ /** * If set to false, then the table is a temporary * table and is not logged to disk. * * In case of a system restart the table will still * be defined and exist but will be empty. * Thus no checkpointing and no logging is performed on the table. * * The default value is true and indicates a normal table * with full checkpointing and logging activated. */ void setLogging(bool); /** * @see NdbDictionary::Table::setLogging. */ bool getLogging() const; /** * Set fragmentation type */ void setFragmentType(FragmentType); /** * Get fragmentation type */ FragmentType getFragmentType() const; /** * Set KValue (Hash parameter.) * Only allowed value is 6. * Later implementations might add flexibility in this parameter. */ void setKValue(int kValue); /** * Get KValue (Hash parameter.) * Only allowed value is 6. * Later implementations might add flexibility in this parameter. */ int getKValue() const; /** * Set MinLoadFactor (Hash parameter.) * This value specifies the load factor when starting to shrink * the hash table. * It must be smaller than MaxLoadFactor. * Both these factors are given in percentage. */ void setMinLoadFactor(int); /** * Get MinLoadFactor (Hash parameter.) * This value specifies the load factor when starting to shrink * the hash table. * It must be smaller than MaxLoadFactor. * Both these factors are given in percentage. */ int getMinLoadFactor() const; /** * Set MaxLoadFactor (Hash parameter.) * This value specifies the load factor when starting to split * the containers in the local hash tables. * 100 is the maximum which will optimize memory usage. * A lower figure will store less information in each container and thus * find the key faster but consume more memory. */ void setMaxLoadFactor(int); /** * Get MaxLoadFactor (Hash parameter.) * This value specifies the load factor when starting to split * the containers in the local hash tables. * 100 is the maximum which will optimize memory usage. * A lower figure will store less information in each container and thus * find the key faster but consume more memory. */ int getMaxLoadFactor() const; /** @} *******************************************************************/ /** * @name Other * @{ */ /** * Get number of columns in the table */ int getNoOfColumns() const; /** * Get number of primary keys in the table */ int getNoOfPrimaryKeys() const; /** * Get name of primary key */ const char* getPrimaryKey(int no) const; /** * Check if table is equal to some other table */ bool equal(const Table&) const; /** * Get frm file stored with this table */ const void* getFrmData() const; Uint32 getFrmLength() const; /** * Set frm file to store with this table */ void setFrm(const void* data, Uint32 len); /** * Set table object type */ void setObjectType(Object::Type type); /** * Get table object type */ Object::Type getObjectType() const; /** * Get object status */ virtual Object::Status getObjectStatus() const; /** * Get object version */ virtual int getObjectVersion() const; /** @} *******************************************************************/ #ifndef DOXYGEN_SHOULD_SKIP_DEPRECATED void setStoredTable(bool x) { setLogging(x); } bool getStoredTable() const { return getLogging(); } int getRowSizeInBytes() const ; int createTableInDb(Ndb*, bool existingEqualIsOk = true) const ; #endif private: friend class NdbTableImpl; class NdbTableImpl & m_impl; Table(NdbTableImpl&); }; /** * @class Index * @brief Represents an index in an NDB Cluster */ class Index : public Object { public: /** * Constructor * @param name Name of index */ Index(const char * name = ""); virtual ~Index(); /** * Set the name of an index */ void setName(const char * name); /** * Get the name of an index */ const char * getName() const; /** * Define the name of the table to be indexed */ void setTable(const char * name); /** * Get the name of the table being indexed */ const char * getTable() const; /** * Get the number of columns in the index */ unsigned getNoOfColumns() const; /** * Get the number of columns in the index * Depricated, use getNoOfColumns instead. */ int getNoOfIndexColumns() const; /** * Get a specific column in the index */ const Column * getColumn(unsigned no) const ; /** * Get a specific column name in the index * Depricated, use getColumn instead. */ const char * getIndexColumn(int no) const ; /** * Add a column to the index definition * Note that the order of columns will be in * the order they are added (only matters for ordered indexes). */ void addColumn(const Column & c); /** * Add a column name to the index definition * Note that the order of indexes will be in * the order they are added (only matters for ordered indexes). */ void addColumnName(const char * name); /** * Add a column name to the index definition * Note that the order of indexes will be in * the order they are added (only matters for ordered indexes). * Depricated, use addColumnName instead. */ void addIndexColumn(const char * name); /** * Add several column names to the index definition * Note that the order of indexes will be in * the order they are added (only matters for ordered indexes). */ void addColumnNames(unsigned noOfNames, const char ** names); /** * Add several column names to the index definition * Note that the order of indexes will be in * the order they are added (only matters for ordered indexes). * Depricated, use addColumnNames instead. */ void addIndexColumns(int noOfNames, const char ** names); /** * Represents type of index */ enum Type { Undefined = 0, ///< Undefined object type (initial value) UniqueHashIndex = 3, ///< Unique un-ordered hash index ///< (only one currently supported) HashIndex = 4, ///< Non-unique un-ordered hash index UniqueOrderedIndex = 5, ///< Unique ordered index OrderedIndex = 6 ///< Non-unique ordered index }; /** * Set index type of the index */ void setType(Type type); /** * Get index type of the index */ Type getType() const; /** * Enable/Disable index storage on disk * * @param enable If enable is set to true, then logging becomes enabled * * @see NdbDictionary::Table::setLogging * * @note Non-logged indexes are rebuilt at system restart. * @note Ordered index does not currently support logging. */ void setLogging(bool enable); /** * Check if index is set to be stored on disk * * @see NdbDictionary::Index::setLogging */ bool getLogging() const; #ifndef DOXYGEN_SHOULD_SKIP_DEPRECATED void setStoredIndex(bool x) { setLogging(x); } bool getStoredIndex() const { return getLogging(); } #endif /** * Get object status */ virtual Object::Status getObjectStatus() const; /** * Get object version */ virtual int getObjectVersion() const; private: friend class NdbIndexImpl; class NdbIndexImpl & m_impl; Index(NdbIndexImpl&); }; /** * @brief Represents an Event in NDB Cluster * */ class Event : public Object { public: enum TableEvent { TE_INSERT=1, TE_DELETE=2, TE_UPDATE=4, TE_ALL=7 }; enum EventDurability { ED_UNDEFINED = 0, #if 0 // not supported ED_SESSION = 1, // Only this API can use it // and it's deleted after api has disconnected or ndb has restarted ED_TEMPORARY = 2, // All API's can use it, // But's its removed when ndb is restarted #endif ED_PERMANENT = 3 // All API's can use it, // It's still defined after a restart }; Event(const char *name); virtual ~Event(); void setName(const char *); void setTable(const char *); void addTableEvent(const TableEvent); void setDurability(const EventDurability); void addColumn(const Column &c); void addEventColumn(unsigned attrId); void addEventColumn(const char * columnName); void addEventColumns(int n, const char ** columnNames); /** * Get object status */ virtual Object::Status getObjectStatus() const; /** * Get object version */ virtual int getObjectVersion() const; void print(); private: friend class NdbEventImpl; friend class NdbEventOperationImpl; class NdbEventImpl & m_impl; Event(NdbEventImpl&); }; /** * @class Dictionary * @brief Dictionary for defining and retreiving meta data */ class Dictionary { public: /** * @class List * @brief Structure for retrieving lists of object names */ struct List { /** * @struct Element * @brief Object to be stored in an NdbDictionary::Dictionary::List */ struct Element { unsigned id; ///< Id of object Object::Type type; ///< Type of object Object::State state; ///< State of object Object::Store store; ///< How object is stored char * database; ///< In what database the object resides char * schema; ///< What schema the object is defined in char * name; ///< Name of object Element() : id(0), type(Object::TypeUndefined), state(Object::StateUndefined), store(Object::StoreUndefined), database(0), schema(0), name(0) { } }; unsigned count; ///< Number of elements in list Element * elements; ///< Pointer to array of elements List() : count(0), elements(0) {} ~List() { if (elements != 0) { for (unsigned i = 0; i < count; i++) { delete[] elements[i].database; delete[] elements[i].schema; delete[] elements[i].name; elements[i].name = 0; } delete[] elements; count = 0; elements = 0; } } }; /** * @name General * @{ */ /** * Fetch list of all objects, optionally restricted to given type. */ int listObjects(List & list, Object::Type type = Object::TypeUndefined); /** * Get the latest error * * @return Error object. */ const struct NdbError & getNdbError() const; /** @} *******************************************************************/ /** * @name Tables * @{ */ /** * Create defined table given defined Table instance * @param Table instance to create * @return 0 if successful otherwise -1. */ int createTable(const Table &); /** * Drop table given retrieved Table instance * @param Table instance to drop * @return 0 if successful otherwise -1. */ int dropTable(Table &); /** * Drop table given table name * @param name Name of table to drop * @return 0 if successful otherwise -1. */ int dropTable(const char * name); /** * Alter defined table given defined Table instance * @param Table Table to alter * @return -2 (incompatible version)
* -1 general error
* 0 success */ int alterTable(const Table &); /** * Get table with given name, NULL if undefined * @param name Name of table to get * @return table if successful otherwise NULL. */ const Table * getTable(const char * name); #ifndef DOXYGEN_SHOULD_SKIP_INTERNAL /** * Invalidate cached table object * @param name Name of table to invalidate */ void invalidateTable(const char * name); #endif /** * Remove table/index from local cache */ void removeCachedTable(const char * table); void removeCachedIndex(const char * index, const char * table); /** @} *******************************************************************/ /** * @name Indexes * @{ */ /** * Create index given defined Index instance * @param Index to create * @return 0 if successful otherwise -1. */ int createIndex(const Index &); /** * Drop index with given name * @param indexName Name of index to drop. * @param tableName Name of table that index belongs to. * @return 0 if successful otherwise -1. */ int dropIndex(const char * indexName, const char * tableName); /** * Drop index the defined Index instance * @param Index to drop * @return 0 if successful otherwise -1. */ int dropIndex(const Index &); /** * Get index with given name, NULL if undefined * @param indexName Name of index to get. * @param tableName Name of table that index belongs to. * @return index if successful, otherwise 0. */ const Index * getIndex(const char * indexName, const char * tableName); /** * Get index with given name, NULL if undefined * @param indexName Name of index to get. * @param Table instance table that index belongs to. * @return index if successful, otherwise 0. */ const Index * getIndex(const char * indexName, const Table & table); #ifndef DOXYGEN_SHOULD_SKIP_INTERNAL /** * Invalidate cached index object */ void invalidateIndex(const char * indexName, const char * tableName); #endif /** * Fetch list of indexes of given table. * @param list Reference to list where to store the listed indexes * @param tableName Name of table that index belongs to. * @return 0 if successful, otherwise -1 */ int listIndexes(List & list, const char * tableName); /** @} *******************************************************************/ /** * @name Events * @{ */ /** * Create event given defined Event instance * @param Event to create * @return 0 if successful otherwise -1. */ int createEvent(const Event &); /** * Drop event with given name * @param eventName Name of event to drop. * @return 0 if successful otherwise -1. */ int dropEvent(const char * eventName); /** * Get event with given name. * @param eventName Name of event to get. * @return an Event if successful, otherwise NULL. */ const Event * getEvent(const char * eventName); /** @} *******************************************************************/ protected: Dictionary(Ndb & ndb); ~Dictionary(); private: friend class NdbDictionaryImpl; friend class UtilTransactions; friend class NdbBlob; class NdbDictionaryImpl & m_impl; Dictionary(NdbDictionaryImpl&); const Table * getIndexTable(const char * indexName, const char * tableName); public: const Table * getTable(const char * name, void **data); void set_local_table_data_size(unsigned sz); }; }; class NdbOut& operator <<(class NdbOut& out, const NdbDictionary::Column& col); #endif