Package org.geotools.data.duckdb

Class DuckDBDialect

Object
SQLDialect
BasicSQLDialect
DuckDBDialect
Direct Known Subclasses:
GeoParquetDialect
public class DuckDBDialect extends BasicSQLDialect
Base SQL Dialect for DuckDB-based datastores. Provides common DuckDB SQL functionality including spatial support.

This dialect implements the core SQL operations for DuckDB, with a focus on spatial functionality. It provides implementations for:

  • Geometry handling (WKB encoding/decoding, spatial function mapping)
  • SQL type conversions for geometry classes
  • Filter and query translation
  • Extension management (spatial, etc.)
  • Optimized bounds calculations
  • Geometry simplification for rendering

This base dialect is extended by format-specific dialects like GeoParquetDialect to provide specialized functionality for different data formats while sharing the common DuckDB handling code.

DuckDB is particularly well-suited for analytical workloads and includes excellent built-in support for spatial operations and columnar file formats like Parquet.

  • Constructor Details

    • DuckDBDialect

      protected DuckDBDialect(JDBCDataStore dataStore)

  • Method Details

    • encodeValue

      public void encodeValue(Object value, Class type, StringBuffer sql)
      Description copied from class: BasicSQLDialect
      Encodes a value in an sql statement.

      Subclasses may wish to override or extend this method to handle specific types. This default implementation does the following:

      1. The value is encoded via its Object.toString() representation.
      2. If type is a character type (extends CharSequence), it is wrapped in single quotes (').
      Overrides:
      encodeValue in class BasicSQLDialect

    • setScreenMapEnabled

      public void setScreenMapEnabled(boolean screenMapEnabled)

    • setSimplifyEnabled

      public void setSimplifyEnabled(boolean simplifyEnabled)

    • getDatabaseInitSql

      public List<String> getDatabaseInitSql()

    • getNameEscape

      public String getNameEscape()
      Description copied from class: SQLDialect
      Returns the string used to escape names.

      This value is used to escape any name in a query. This includes columns, tables, schemas, indexes, etc... If no escape is necessary this method should return the empty string, and never return null.

      This default implementation returns a single double quote ("), subclasses must override to provide a different espcape.

      Overrides:
      getNameEscape in class SQLDialect

    • escapeName

      public String escapeName(String name)
      Description copied from class: SQLDialect
      Surrounds a name with the SQL escape string.

      If the name contains the SQL escape string, the SQL escape string is duplicated.

      Overrides:
      escapeName in class SQLDialect

    • includeTable

      public boolean includeTable(String schemaName, String tableName, Connection cx) throws SQLException
      Description copied from class: SQLDialect
      Determines if the specified table should be included in those published by the datastore.

      This method returns true if the table should be published as a feature type, otherwise it returns false. Subclasses should override this method, this default implementation returns true.

      A database connection is provided to the dialect but it should not be closed. However any statements objects or result sets that are instantiated from it must be closed.

      Overrides:
      includeTable in class SQLDialect
      Parameters:
      schemaName - The schema of the table, might be null..
      tableName - The name of the table.
      cx - Database connection.
      Throws:
      SQLException

    • getDesiredTablesType

      public String[] getDesiredTablesType()
      Overrides:
      getDesiredTablesType in class SQLDialect
      Returns:
      Table types filtered from jdbc DatabaseMetaData

    • addSupportedHints

      protected void addSupportedHints(Set<Hints.Key> hints)
      Description copied from class: SQLDialect
      Add hints to the JDBC Feature Source. A subclass can override

      possible hints (but not limited to)

      Hints.GEOMETRY_GENERALIZATION Hints.GEOMETRY_SIMPLIFICATION

      Overrides:
      addSupportedHints in class SQLDialect

    • getMapping

      public Class<?> getMapping(ResultSet columnMetaData, Connection cx) throws SQLException
      It is called before the mappings registered in registerSqlTypeToClassMappings(Map) and are used to determine the mapping. Subclasses should implement as needed, the default implementation returns {@code null}.
      Overrides:
      getMapping in class SQLDialect
      Parameters:
      columnMetaData - The column metadata
      cx - The connection used to retrieve the metadata
      Returns:
      The class mapped to the to column, or null.
      Throws:
      SQLException

    • getParentMapping

      protected Class<?> getParentMapping(ResultSet columnMetaData, Connection cx) throws SQLException
      Throws:
      SQLException

    • registerSqlTypeToClassMappings

      public void registerSqlTypeToClassMappings(Map<Integer,Class<?>> mappings)
      Appends a mapping of Types.STRUCT to java.sql.Struct.class

      Support for struct is limited and results in GeoServer WMS and WFS to interpret it as a String literal. For example, a "bbox" Struct attribute value with SQL type name as STRUCT(xmin FLOAT, xmax FLOAT, ymin FLOAT, ymax FLOAT) will be encoded as <bbox>{xmin=-63.005005, xmax=-63.004997, ymin=-40.81569, ymax=-40.81568}</bbox> by virtue of DuckDBStruct.toString()

      Overrides:
      registerSqlTypeToClassMappings in class SQLDialect

    • registerSqlTypeNameToClassMappings

      public void registerSqlTypeNameToClassMappings(Map<String,Class<?>> mappings)
      Description copied from class: SQLDialect
      Registers the sql type name to java type mappings that the dialect uses when reading and writing objects to and from the database.

      Subclasses should extend (not override) this method to provide additional mappings, or to override mappings provided by this implementation. This implementation provides the following mappings:

      Overrides:
      registerSqlTypeNameToClassMappings in class SQLDialect

    • registerSqlTypeToSqlTypeNameOverrides

      public void registerSqlTypeToSqlTypeNameOverrides(Map<Integer,String> overrides)
      Description copied from class: SQLDialect
      Registers any overrides that should occur when mapping an integer sql type value to an underlying sql type name.

      The default implementation of this method does nothing. Subclasses should override in cases where:

      • database type metadata does not provide enough information to properly map
      • to support custom types (those not in Types)
      Overrides:
      registerSqlTypeToSqlTypeNameOverrides in class SQLDialect

    • registerClassToSqlMappings

      public void registerClassToSqlMappings(Map<Class<?>,Integer> mappings)
      Maps Geometry types to Types.OTHER and Struct to Types.STRUCT

      Support for struct is limited and results in GeoServer WMS and WFS to interpret it as a String literal. For example, a "bbox" Struct attribute value with SQL type name as STRUCT(xmin FLOAT, xmax FLOAT, ymin FLOAT, ymax FLOAT) will be encoded as <bbox>{xmin=-63.005005, xmax=-63.004997, ymin=-40.81569, ymax=-40.81568}</bbox> by virtue of DuckDBStruct.toString()

      Overrides:
      registerClassToSqlMappings in class SQLDialect

    • convertValue

      public Object convertValue(Object value, AttributeDescriptor ad)
      Description copied from class: SQLDialect
      Convertes the given value to the target type. The default implementation uses converters to match the attribute descriptor binding, subclasses can override.
      Overrides:
      convertValue in class SQLDialect

    • getGeometrySRID

      public Integer getGeometrySRID(String schemaName, String tableName, String columnName, Connection cx) throws SQLException
      Description copied from class: SQLDialect
      Returns the spatial reference system identifier (srid) for a particular geometry column.

      This method is given a direct connection to the database. The connection must not be closed. However any statements or result sets instantiated from the connection must be closed.

      In the event that the srid cannot be determined, this method should return null .

      Overrides:
      getGeometrySRID in class SQLDialect
      Parameters:
      schemaName - The database schema, could be null.
      tableName - The table, never null.
      columnName - The column name, never null
      cx - The database connection.
      Throws:
      SQLException

    • getSequenceForColumn

      public String getSequenceForColumn(String schemaName, String tableName, String columnName, Connection cx) throws SQLException
      Description copied from class: SQLDialect
      Determines the name of the sequence (if any) which is used to increment generate values for a table column.

      This method should return null if no such sequence exists.

      This method is given a direct connection to the database, but this connection should never be closed. However any statements or result sets instantiated from the connection must be closed.

      Overrides:
      getSequenceForColumn in class SQLDialect
      Parameters:
      schemaName - The schema name, this might be null.
      tableName - The table name.
      columnName - The column name.
      cx - The database connection.
      Throws:
      SQLException

    • getNextSequenceValue

      public Object getNextSequenceValue(String schemaName, String sequenceName, Connection cx) throws SQLException
      Description copied from class: SQLDialect
      Obtains the next value of a sequence, incrementing the sequence to the next state in the process.

      Implementations should determine the next value of a column for which values are automatically generated by the database.

      This method is given a direct connection to the database, but this connection should never be closed. However any statements or result sets instantiated from the connection must be closed.

      Implementations should handle the case where schemaName is null.

      Overrides:
      getNextSequenceValue in class SQLDialect
      Parameters:
      schemaName - The schema name, this might be null.
      sequenceName - The name of the sequence.
      cx - The database connection.
      Returns:
      The next value of the sequence, or null.
      Throws:
      SQLException

    • encodeNextSequenceValue

      public String encodeNextSequenceValue(String schemaName, String sequenceName)
      Description copied from class: SQLDialect
      Encodes how to get the next sequence value from the DB.

      Implementations should handle the case where schemaName is null.

      Overrides:
      encodeNextSequenceValue in class SQLDialect

    • getGeometryTypeName

      public String getGeometryTypeName(Integer type)
      Description copied from class: SQLDialect
      Returns the name of a geometric type based on its integer constant.

      The constant, type, is registered in SQLDialect.registerSqlTypeNameToClassMappings(Map).

      This default implementation returns null, subclasses should override.

      Overrides:
      getGeometryTypeName in class SQLDialect

    • getOptimizedBounds

      public List<ReferencedEnvelope> getOptimizedBounds(String schema, SimpleFeatureType featureType, Connection cx) throws SQLException, IOException
      Description copied from class: SQLDialect
      Returns the bounds of all geometry columns in the layer using any approach that proves to be faster than the plain bounds aggregation (e.g., better than the "plain select extent(geom) from table" on PostGIS), or null if none exists or the fast method has not been enabled (e.g., if the fast method is just an estimate of the bounds you probably want the user to enable it manually)
      Overrides:
      getOptimizedBounds in class SQLDialect
      Parameters:
      schema - The database schema, if any, or null
      featureType - The feature type containing the geometry columns whose bounds need to computed. Mind, it may be retyped and thus contain less geometry columns than the table
      Returns:
      a list of referenced envelopes (some of which may be null or empty)
      Throws:
      SQLException
      IOException

    • optimizedBounds

      protected ReferencedEnvelope optimizedBounds(SimpleFeatureType featureType, Connection cx) throws SQLException, IOException
      Throws:
      SQLException
      IOException

    • encodeGeometryEnvelope

      public void encodeGeometryEnvelope(String tableName, String geometryColumn, StringBuffer sql)
      Encodes the spatial extent function of a geometry column in a SELECT statement.

      This method must also be sure to properly encode the name of the column with the #encodeColumnName(String, StringBuffer) function.

      Specified by:
      encodeGeometryEnvelope in class SQLDialect
      See Also:

    • decodeGeometryEnvelope

      public Envelope decodeGeometryEnvelope(ResultSet rs, int column, Connection cx) throws SQLException, IOException
      Decodes the result of a spatial extent function in a SELECT statement.

      This method is given direct access to a result set. The column parameter is the index into the result set which contains the spatial extent value. The query for this value is build with the SQLDialect.encodeGeometryEnvelope(String, String, StringBuffer) method.

      This method must not read any other objects from the result set other then the one referenced by column.

      Specified by:
      decodeGeometryEnvelope in class SQLDialect
      Parameters:
      rs - A result set
      column - Index into the result set which points at the spatial extent value.
      cx - The database connection.
      Throws:
      SQLException
      IOException
      See Also:

    • encodeGeometryColumn

      public void encodeGeometryColumn(GeometryDescriptor gatt, String prefix, int srid, Hints hints, StringBuffer sql)
      Description copied from class: SQLDialect
      Encodes the name of a geometry column in a SELECT statement.

      This method should wrap the column name in any functions that are used to retrieve its value. For instance, often it is necessary to use the function asText, or asWKB when fetching a geometry.

      This method must also be sure to properly encode the name of the column with the SQLDialect.encodeColumnName(String, String, StringBuffer) function.

      Example: 

         
   sql.append( "asText(" );
   column( gatt.getLocalName(), sql );
   sql.append( ")" );
      Overrides:
      encodeGeometryColumn in class SQLDialect

    • encodeGeometryColumnInternal

      protected void encodeGeometryColumnInternal(GeometryDescriptor gatt, String prefix, Hints hints, boolean forceMulti, StringBuffer sql)

    • encodeColumnName

      public String encodeColumnName(String prefix, String raw)

    • encodeGeometryColumnGeneralized

      public void encodeGeometryColumnGeneralized(GeometryDescriptor gatt, String prefix, int srid, StringBuffer sql, Double distance)
      Overrides:
      encodeGeometryColumnGeneralized in class SQLDialect

    • encodeGeometryColumnSimplified

      public void encodeGeometryColumnSimplified(GeometryDescriptor gatt, String prefix, int srid, StringBuffer sql, Double distance)
      Overrides:
      encodeGeometryColumnSimplified in class SQLDialect

    • decodeGeometryValue

      public Geometry decodeGeometryValue(GeometryDescriptor descriptor, ResultSet rs, String column, GeometryFactory factory, Connection cx, Hints hints) throws IOException, SQLException
      Description copied from class: SQLDialect
      Decodes a geometry value from the result of a query.

      This method is given direct access to a result set. The column parameter is the index into the result set which contains the geometric value.

      An implementation should deserialize the value provided by the result set into Geometry object. For example, consider an implementation which deserializes from well known text: 

         String wkt = rs.getString( column );
   if ( wkt == null ) {
     return null;
   }
   return new WKTReader(factory).read( wkt );
      Note that implementations must handle null values.

      The factory parameter should be used to instantiate any geometry objects.

      Specified by:
      decodeGeometryValue in class SQLDialect
      Throws:
      IOException
      SQLException

    • parseWKB

      protected Geometry parseWKB(Blob blob) throws SQLException, IOException
      Throws:
      SQLException
      IOException

    • parseWKB

      protected Geometry parseWKB(GeometryFactory factory, Blob blob) throws SQLException, IOException
      Throws:
      SQLException
      IOException

    • encodeGeometryValue

      public void encodeGeometryValue(Geometry value, int dimension, int srid, StringBuffer sql) throws IOException
      Description copied from class: BasicSQLDialect
      Encodes a geometry value in an sql statement.

      An implementations should serialize value into some exchange format which will then be transported to the underlying database. For example, consider an implementation which converts a geometry into its well known text representation: 

         
   sql.append( "GeomFromText('" );
   sql.append( new WKTWriter().write( value ) );
   sql.append( ")" );

      The srid parameter is the spatial reference system identifier of the geometry, or 0 if not known.

      Attention should be paid to emtpy geometries (g.isEmtpy() == true) as they cannot be encoded in WKB and several databases fail to handle them property. Common treatment is to equate them to NULL

      Specified by:
      encodeGeometryValue in class BasicSQLDialect
      Throws:
      IOException

    • isConcreteGeometry

      protected boolean isConcreteGeometry(Class<?> binding)

    • encodePrimaryKey

      public void encodePrimaryKey(String column, StringBuffer sql)
      Description copied from class: SQLDialect
      Encodes the primary key definition in a CREATE TABLE statement.

      Subclasses should override this method if need be, the default implementation does the following: 

         
   encodeColumnName( column, sql );
   sql.append( " int PRIMARY KEY" );
      Overrides:
      encodePrimaryKey in class SQLDialect

    • isLimitOffsetSupported

      public boolean isLimitOffsetSupported()
      Description copied from class: SQLDialect
      Returns true if this dialect can encode both Query.getStartIndex() and Query.getMaxFeatures() into native SQL.
      Overrides:
      isLimitOffsetSupported in class SQLDialect

    • isAggregatedSortSupported

      public boolean isAggregatedSortSupported(String function)
      Description copied from class: SQLDialect
      Returns true if this dialect supports sorting together with the given aggregation function.
      Overrides:
      isAggregatedSortSupported in class SQLDialect

    • applyLimitOffset

      public void applyLimitOffset(StringBuffer sql, int limit, int offset)
      Description copied from class: SQLDialect
      Alters the query provided so that limit and offset are natively dealt with. This might mean simply appending some extra directive to the query, or wrapping it into a bigger one.
      Overrides:
      applyLimitOffset in class SQLDialect

    • createFilterToSQL

      public FilterToSQL createFilterToSQL()
      Description copied from class: BasicSQLDialect
      Creates the filter encoder to be used by the datastore when encoding query predicates.

      Sublcasses can override this method to return a subclass of FilterToSQL if need be.

      Overrides:
      createFilterToSQL in class BasicSQLDialect