#!/usr/local/bin/thrift -java # # Thrift Service that the MetaStore is built on # include "thrift/fb303/if/fb303.thrift" namespace java org.apache.hadoop.hive.metastore.api namespace php metastore namespace cpp Apache.Hadoop.Hive const string DDL_TIME = "transient_lastDdlTime" struct Version { 1: string version, 2: string comments } struct FieldSchema { 1: string name, // name of the field 2: string type, // type of the field. primitive types defined above, specify list, map for lists & maps 3: string comment } struct Type { 1: string name, // one of the types in PrimitiveTypes or CollectionTypes or User defined types 2: optional string type1, // object type if the name is 'list' (LIST_TYPE), key type if the name is 'map' (MAP_TYPE) 3: optional string type2, // val type if the name is 'map' (MAP_TYPE) 4: optional list fields // if the name is one of the user defined types } // namespace for tables struct Database { 1: string name, 2: string description, 3: string locationUri, } // This object holds the information needed by SerDes struct SerDeInfo { 1: string name, // name of the serde, table name by default 2: string serializationLib, // usually the class that implements the extractor & loader 3: map parameters // initialization parameters } // sort order of a column (column name along with asc(1)/desc(0)) struct Order { 1: string col, // sort column name 2: i32 order // asc(1) or desc(0) } // this object holds all the information about physical storage of the data belonging to a table struct StorageDescriptor { 1: list cols, // required (refer to types defined above) 2: string location, // defaults to //tablename 3: string inputFormat, // SequenceFileInputFormat (binary) or TextInputFormat` or custom format 4: string outputFormat, // SequenceFileOutputFormat (binary) or IgnoreKeyTextOutputFormat or custom format 5: bool compressed, // compressed or not 6: i32 numBuckets, // this must be specified if there are any dimension columns 7: SerDeInfo serdeInfo, // serialization and deserialization information 8: list bucketCols, // reducer grouping columns and clustering columns and bucketing columns` 9: list sortCols, // sort order of the data in each bucket 10: map parameters // any user supplied key value hash } // table information struct Table { 1: string tableName, // name of the table 2: string dbName, // database name ('default') 3: string owner, // owner of this table 4: i32 createTime, // creation time of the table 5: i32 lastAccessTime, // last access time (usually this will be filled from HDFS and shouldn't be relied on) 6: i32 retention, // retention time 7: StorageDescriptor sd, // storage descriptor of the table 8: list partitionKeys, // partition keys of the table. only primitive types are supported 9: map parameters, // to store comments or any other user level parameters 10: string viewOriginalText, // original view text, null for non-view 11: string viewExpandedText, // expanded view text, null for non-view 12: string tableType // table type enum, e.g. EXTERNAL_TABLE } struct Partition { 1: list values // string value is converted to appropriate partition key type 2: string dbName, 3: string tableName, 4: i32 createTime, 5: i32 lastAccessTime, 6: StorageDescriptor sd, 7: map parameters } // index on a hive table is also another table whose columns are the subset of the base table columns along with the offset // this will automatically generate table (table_name_index_name) struct Index { 1: string indexName, // unique with in the whole database namespace 2: i32 indexType, // reserved 3: string tableName, 4: string dbName, 5: list colNames, // for now columns will be sorted in the ascending order 6: string partName // partition name } // schema of the table/query results etc. struct Schema { // column names, types, comments 1: list fieldSchemas, // delimiters etc 2: map properties } exception MetaException { 1: string message } exception UnknownTableException { 1: string message } exception UnknownDBException { 1: string message } exception AlreadyExistsException { 1: string message } exception InvalidObjectException { 1: string message } exception NoSuchObjectException { 1: string message } exception IndexAlreadyExistsException { 1: string message } exception InvalidOperationException { 1: string message } exception ConfigValSecurityException { 1: string message } /** * This interface is live. */ service ThriftHiveMetastore extends fb303.FacebookService { void create_database(1:Database database) throws(1:AlreadyExistsException o1, 2:InvalidObjectException o2, 3:MetaException o3) Database get_database(1:string name) throws(1:NoSuchObjectException o1, 2:MetaException o2) void drop_database(1:string name, 2:bool deleteData) throws(1:NoSuchObjectException o1, 2:InvalidOperationException o2, 3:MetaException o3) list get_databases(1:string pattern) throws(1:MetaException o1) list get_all_databases() throws(1:MetaException o1) // returns the type with given name (make seperate calls for the dependent types if needed) Type get_type(1:string name) throws(1:MetaException o1, 2:NoSuchObjectException o2) bool create_type(1:Type type) throws(1:AlreadyExistsException o1, 2:InvalidObjectException o2, 3:MetaException o3) bool drop_type(1:string type) throws(1:MetaException o1, 2:NoSuchObjectException o2) map get_type_all(1:string name) throws(1:MetaException o2) // Gets a list of FieldSchemas describing the columns of a particular table list get_fields(1: string db_name, 2: string table_name) throws (1: MetaException o1, 2: UnknownTableException o2, 3: UnknownDBException o3), // Gets a list of FieldSchemas describing both the columns and the partition keys of a particular table list get_schema(1: string db_name, 2: string table_name) throws (1: MetaException o1, 2: UnknownTableException o2, 3: UnknownDBException o3) // create a Hive table. Following fields must be set // tableName // database (only 'default' for now until Hive QL supports databases) // owner (not needed, but good to have for tracking purposes) // sd.cols (list of field schemas) // sd.inputFormat (SequenceFileInputFormat (binary like falcon tables or u_full) or TextInputFormat) // sd.outputFormat (SequenceFileInputFormat (binary) or TextInputFormat) // sd.serdeInfo.serializationLib (SerDe class name eg org.apache.hadoop.hive.serde.simple_meta.MetadataTypedColumnsetSerDe void create_table(1:Table tbl) throws(1:AlreadyExistsException o1, 2:InvalidObjectException o2, 3:MetaException o3, 4:NoSuchObjectException o4) // drops the table and all the partitions associated with it if the table has partitions // delete data (including partitions) if deleteData is set to true void drop_table(1:string dbname, 2:string name, 3:bool deleteData) throws(1:NoSuchObjectException o1, 2:MetaException o3) list get_tables(1: string db_name, 2: string pattern) throws (1: MetaException o1) list get_all_tables(1: string db_name) throws (1: MetaException o1) Table get_table(1:string dbname, 2:string tbl_name) throws (1:MetaException o1, 2:NoSuchObjectException o2) // alter table applies to only future partitions not for existing partitions void alter_table(1:string dbname, 2:string tbl_name, 3:Table new_tbl) throws (1:InvalidOperationException o1, 2:MetaException o2) // the following applies to only tables that have partitions Partition add_partition(1:Partition new_part) throws(1:InvalidObjectException o1, 2:AlreadyExistsException o2, 3:MetaException o3) Partition append_partition(1:string db_name, 2:string tbl_name, 3:list part_vals) throws (1:InvalidObjectException o1, 2:AlreadyExistsException o2, 3:MetaException o3) Partition append_partition_by_name(1:string db_name, 2:string tbl_name, 3:string part_name) throws (1:InvalidObjectException o1, 2:AlreadyExistsException o2, 3:MetaException o3) bool drop_partition(1:string db_name, 2:string tbl_name, 3:list part_vals, 4:bool deleteData) throws(1:NoSuchObjectException o1, 2:MetaException o2) bool drop_partition_by_name(1:string db_name, 2:string tbl_name, 3:string part_name, 4:bool deleteData) throws(1:NoSuchObjectException o1, 2:MetaException o2) Partition get_partition(1:string db_name, 2:string tbl_name, 3:list part_vals) throws(1:MetaException o1, 2:NoSuchObjectException o2) Partition get_partition_by_name(1:string db_name 2:string tbl_name, 3:string part_name) throws(1:MetaException o1, 2:NoSuchObjectException o2) // returns all the partitions for this table in reverse chronological order. // If max parts is given then it will return only that many. list get_partitions(1:string db_name, 2:string tbl_name, 3:i16 max_parts=-1) throws(1:NoSuchObjectException o1, 2:MetaException o2) list get_partition_names(1:string db_name, 2:string tbl_name, 3:i16 max_parts=-1) throws(1:MetaException o2) // get_partition*_ps methods allow filtering by a partial partition specification, // as needed for dynamic partitions. The values that are not restricted should // be empty strings. Nulls were considered (instead of "") but caused errors in // generated Python code. The size of part_vals may be smaller than the // number of partition columns - the unspecified values are considered the same // as "". list get_partitions_ps(1:string db_name 2:string tbl_name 3:list part_vals, 4:i16 max_parts=-1) throws(1:MetaException o1) list get_partition_names_ps(1:string db_name, 2:string tbl_name, 3:list part_vals, 4:i16 max_parts=-1) throws(1:MetaException o1) // changes the partition to the new partition object. partition is identified from the part values // in the new_part void alter_partition(1:string db_name, 2:string tbl_name, 3:Partition new_part) throws(1:InvalidOperationException o1, 2:MetaException o2) // gets the value of the configuration key in the metastore server. returns // defaultValue if the key does not exist. if the configuration key does not // begin with "hive", "mapred", or "hdfs", a ConfigValSecurityException is // thrown. string get_config_value(1:string name, 2:string defaultValue) throws(1:ConfigValSecurityException o1) // converts a partition name into a partition values array list partition_name_to_vals(1: string part_name) throws(1: MetaException o1) // converts a partition name into a partition specification (a mapping from // the partition cols to the values) map partition_name_to_spec(1: string part_name) throws(1: MetaException o1) } // For storing info about archived partitions in parameters // Whether the partition is archived const string IS_ARCHIVED = "is_archived", // The original location of the partition, before archiving. After archiving, // this directory will contain the archive. When the partition // is dropped, this directory will be deleted const string ORIGINAL_LOCATION = "original_location", // these should be needed only for backward compatibility with filestore const string META_TABLE_COLUMNS = "columns", const string META_TABLE_COLUMN_TYPES = "columns.types", const string BUCKET_FIELD_NAME = "bucket_field_name", const string BUCKET_COUNT = "bucket_count", const string FIELD_TO_DIMENSION = "field_to_dimension", const string META_TABLE_NAME = "name", const string META_TABLE_DB = "db", const string META_TABLE_LOCATION = "location", const string META_TABLE_SERDE = "serde", const string META_TABLE_PARTITION_COLUMNS = "partition_columns", const string FILE_INPUT_FORMAT = "file.inputformat", const string FILE_OUTPUT_FORMAT = "file.outputformat", const string META_TABLE_STORAGE = "storage_handler",