org.apache.iceberg.spark.procedures

Class CreateChangelogViewProcedure

java.lang.Object

org.apache.iceberg.spark.procedures.CreateChangelogViewProcedure

All Implemented Interfaces:

Procedure

public class CreateChangelogViewProcedure
extends java.lang.Object

A procedure that creates a view for changed rows.

The procedure removes the carry-over rows by default. If you want to keep them, you can set "remove_carryovers" to be false in the options.

The procedure doesn't compute the pre/post update images by default. If you want to compute them, you can set "compute_updates" to be true in the options.

Carry-over rows are the result of a removal and insertion of the same row within an operation because of the copy-on-write mechanism. For example, given a file which contains row1 (id=1, data='a') and row2 (id=2, data='b'). A copy-on-write delete of row2 would require erasing this file and preserving row1 in a new file. The changelog table would report this as (id=1, data='a', op='DELETE') and (id=1, data='a', op='INSERT'), despite it not being an actual change to the table. The procedure finds the carry-over rows and removes them from the result.

Pre/post update images are converted from a pair of a delete row and an insert row. Identifier columns are used for determining whether an insert and a delete record refer to the same row. If the two records share the same values for the identity columns they are considered to be before and after states of the same row. You can either set identifier fields in the table schema or input them as the procedure parameters. Here is an example of pre/post update images with an identifier column(id). A pair of a delete row and an insert row with the same id:

• (id=1, data='a', op='DELETE')
• (id=1, data='b', op='INSERT')

will be marked as pre/post update images:

• (id=1, data='a', op='UPDATE_BEFORE')
• (id=1, data='b', op='UPDATE_AFTER')

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
protected static class	org.apache.iceberg.spark.procedures.BaseProcedure.Builder<T extends org.apache.iceberg.spark.procedures.BaseProcedure>

Field Summary

Fields

Modifier and Type	Field and Description
protected static org.apache.spark.sql.types.DataType	STRING_ARRAY
protected static org.apache.spark.sql.types.DataType	STRING_MAP

Method Summary

Modifier and Type	Method and Description
protected SparkActions	actions()
static SparkProcedures.ProcedureBuilder	builder()
org.apache.spark.sql.catalyst.InternalRow[]	call(org.apache.spark.sql.catalyst.InternalRow args) Executes this procedure.
protected void	closeService() Closes this procedure's executor service if a new one was created with executorService(int, String).
java.lang.String	description() Returns the description of this procedure.
protected java.util.concurrent.ExecutorService	executorService(int threadPoolSize, java.lang.String nameFormat) Starts a new executor service which can be used by this procedure in its work.
protected org.apache.spark.sql.Dataset<org.apache.spark.sql.Row>	loadRows(org.apache.spark.sql.connector.catalog.Identifier tableIdent, java.util.Map<java.lang.String,java.lang.String> options)
protected SparkTable	loadSparkTable(org.apache.spark.sql.connector.catalog.Identifier ident)
protected <T> T	modifyIcebergTable(org.apache.spark.sql.connector.catalog.Identifier ident, java.util.function.Function<Table,T> func)
protected org.apache.spark.sql.catalyst.InternalRow	newInternalRow(java.lang.Object... values)
org.apache.spark.sql.types.StructType	outputType() Returns the type of rows produced by this procedure.
ProcedureParameter[]	parameters() Returns the input parameters of this procedure.
protected void	refreshSparkCache(org.apache.spark.sql.connector.catalog.Identifier ident, org.apache.spark.sql.connector.catalog.Table table)
protected org.apache.spark.sql.SparkSession	spark()
protected org.apache.spark.sql.connector.catalog.TableCatalog	tableCatalog()
protected Spark3Util.CatalogAndIdentifier	toCatalogAndIdentifier(java.lang.String identifierAsString, java.lang.String argName, org.apache.spark.sql.connector.catalog.CatalogPlugin catalog)
protected org.apache.spark.sql.connector.catalog.Identifier	toIdentifier(java.lang.String identifierAsString, java.lang.String argName)
protected <T> T	withIcebergTable(org.apache.spark.sql.connector.catalog.Identifier ident, java.util.function.Function<Table,T> func)

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

STRING_MAP

protected static final org.apache.spark.sql.types.DataType STRING_MAP

STRING_ARRAY

protected static final org.apache.spark.sql.types.DataType STRING_ARRAY

Method Detail

builder

public static SparkProcedures.ProcedureBuilder builder()

parameters

public ProcedureParameter[] parameters()

Description copied from interface: Procedure

Returns the input parameters of this procedure.

outputType

public org.apache.spark.sql.types.StructType outputType()

Description copied from interface: Procedure

Returns the type of rows produced by this procedure.

call

public org.apache.spark.sql.catalyst.InternalRow[] call(org.apache.spark.sql.catalyst.InternalRow args)

Description copied from interface: Procedure

Executes this procedure.

Spark will align the provided arguments according to the input parameters defined in Procedure.parameters() either by position or by name before execution.

Implementations may provide a summary of execution by returning one or many rows as a result. The schema of output rows must match the defined output type in Procedure.outputType().

Parameters:

args - input arguments

Returns:

the result of executing this procedure with the given arguments

description

public java.lang.String description()

Description copied from interface: Procedure

Returns the description of this procedure.

spark

protected org.apache.spark.sql.SparkSession spark()

actions

protected SparkActions actions()

tableCatalog

protected org.apache.spark.sql.connector.catalog.TableCatalog tableCatalog()

modifyIcebergTable

protected <T> T modifyIcebergTable(org.apache.spark.sql.connector.catalog.Identifier ident,
                                   java.util.function.Function<Table,T> func)

withIcebergTable

protected <T> T withIcebergTable(org.apache.spark.sql.connector.catalog.Identifier ident,
                                 java.util.function.Function<Table,T> func)

toIdentifier

protected org.apache.spark.sql.connector.catalog.Identifier toIdentifier(java.lang.String identifierAsString,
                                                                         java.lang.String argName)

toCatalogAndIdentifier

protected Spark3Util.CatalogAndIdentifier toCatalogAndIdentifier(java.lang.String identifierAsString,
                                                                 java.lang.String argName,
                                                                 org.apache.spark.sql.connector.catalog.CatalogPlugin catalog)

loadSparkTable

protected SparkTable loadSparkTable(org.apache.spark.sql.connector.catalog.Identifier ident)

loadRows

protected org.apache.spark.sql.Dataset<org.apache.spark.sql.Row> loadRows(org.apache.spark.sql.connector.catalog.Identifier tableIdent,
                                                                          java.util.Map<java.lang.String,java.lang.String> options)

refreshSparkCache

protected void refreshSparkCache(org.apache.spark.sql.connector.catalog.Identifier ident,
                                 org.apache.spark.sql.connector.catalog.Table table)

newInternalRow

protected org.apache.spark.sql.catalyst.InternalRow newInternalRow(java.lang.Object... values)

closeService

protected void closeService()

Closes this procedure's executor service if a new one was created with executorService(int, String). Does not block for any remaining tasks.

executorService

protected java.util.concurrent.ExecutorService executorService(int threadPoolSize,
                                                               java.lang.String nameFormat)

Starts a new executor service which can be used by this procedure in its work. The pool will be automatically shut down if withIcebergTable(Identifier, Function) or modifyIcebergTable(Identifier, Function) are called. If these methods are not used then the service can be shut down with closeService() or left to be closed when this class is finalized.

Parameters:

threadPoolSize - number of threads in the service

nameFormat - name prefix for threads created in this service

Returns:

the new executor service owned by this procedure