ARROW-4142: [Java] JDBC Array -> Arrow ListVector

java/adapter/jdbc/src/main/java/org/apache/arrow/adapter/jdbc/JdbcToArrow.java

@@ -184,7 +184,6 @@ public static VectorSchemaRoot sqlToArrow(ResultSet resultSet, BaseAllocator all
  */
  public static VectorSchemaRoot sqlToArrow(ResultSet resultSet, Calendar calendar) throws SQLException, IOException {
  Preconditions.checkNotNull(resultSet, "JDBC ResultSet object can not be null");
-
  return sqlToArrow(resultSet, new JdbcToArrowConfig(new RootAllocator(Integer.MAX_VALUE), calendar, false));
  }
 
@@ -220,6 +219,7 @@ public static VectorSchemaRoot sqlToArrow(ResultSet resultSet, JdbcToArrowConfig
  throws SQLException, IOException {
  Preconditions.checkNotNull(resultSet, "JDBC ResultSet object can not be null");
  Preconditions.checkNotNull(config, "The configuration cannot be null");
+ Preconditions.checkArgument(config.isValid(), "The configuration must be valid");
 
  VectorSchemaRoot root = VectorSchemaRoot.create(
  JdbcToArrowUtils.jdbcToArrowSchema(resultSet.getMetaData(), config), config.getAllocator());

java/adapter/jdbc/src/main/java/org/apache/arrow/adapter/jdbc/JdbcToArrowConfig.java

@@ -18,6 +18,7 @@
 package org.apache.arrow.adapter.jdbc;
 
 import java.util.Calendar;
+import java.util.Map;
 
 import org.apache.arrow.memory.BaseAllocator;
 
@@ -28,16 +29,29 @@
  * <p>
  * The allocator is used to construct the {@link org.apache.arrow.vector.VectorSchemaRoot},
  * and the calendar is used to define the time zone of any {@link org.apahe.arrow.vector.pojo.ArrowType.Timestamp}
- * fields that are created during the conversion.
+ * fields that are created during the conversion. Neither field may be <code>null</code>.
  * </p>
  * <p>
- * Neither field may be <code>null</code>.
+ * If the <code>includeMetadata</code> flag is set, the Arrow field metadata will contain information
+ * from the corresponding {@link java.sql.ResultSetMetaData} that was used to create the
+ * {@link org.apache.arrow.vector.types.pojo.FieldType} of the corresponding
+ * {@link org.apache.arrow.vector.FieldVector}.
+ * </p>
+ * <p>
+ * If there are any {@link java.sql.Types#ARRAY} fields in the {@link java.sql.ResultSet}, the corresponding
+ * {@link JdbcFieldInfo} for the array's contents must be defined here. Unfortunately, the sub-type
+ * information cannot be retrieved from all JDBC implementations (H2 for example, returns
+ * {@link java.sql.Types#NULL} for the array sub-type), so it must be configured here. The column index
+ * or name can be used to map to a {@link JdbcFieldInfo}, and that will be used for the conversion.
  * </p>
  */
 public final class JdbcToArrowConfig {
+
  private Calendar calendar;
  private BaseAllocator allocator;
  private boolean includeMetadata;
+ private Map<Integer, JdbcFieldInfo> arraySubTypesByColumnIndex;
+ private Map<String, JdbcFieldInfo> arraySubTypesByColumnName;
 
  /**
  * Constructs a new configuration from the provided allocator and calendar. The <code>allocator</code>
@@ -48,18 +62,21 @@ public final class JdbcToArrowConfig {
  * @param calendar The calendar to use when constructing Timestamp fields and reading time-based results.
  * @param includeMetadata Whether to include JDBC field metadata in the Arrow Schema Field metadata.
  */
- JdbcToArrowConfig(BaseAllocator allocator, Calendar calendar, boolean includeMetadata) {
+ public JdbcToArrowConfig(BaseAllocator allocator, Calendar calendar, boolean includeMetadata) {
  Preconditions.checkNotNull(allocator, "Memory allocator cannot be null");
 
  this.allocator = allocator;
  this.calendar = calendar;
  this.includeMetadata = includeMetadata;
+ this.arraySubTypesByColumnIndex = null;
+ this.arraySubTypesByColumnName = null;
  }
 
  /**
  * The calendar to use when defining Arrow Timestamp fields
  * and retrieving {@link Date}, {@link Time}, or {@link Timestamp}
  * data types from the {@link ResultSet}, or <code>null</code> if not converting.
+ *
  * @return the calendar.
  */
  public Calendar getCalendar() {
@@ -82,4 +99,78 @@ public BaseAllocator getAllocator() {
  public boolean shouldIncludeMetadata() {
  return includeMetadata;
  }
+
+ /**
+ * Sets whether to include JDBC ResultSet field metadata in the Arrow Schema field metadata.
+ *
+ * @param includeMetadata Whether to include or exclude JDBC metadata in the Arrow Schema field metadata.
+ * @return This instance of the <code>JdbcToArrowConfig</code>, for chaining.
+ */
+ public JdbcToArrowConfig setIncludeMetadata(boolean includeMetadata) {
+ this.includeMetadata = includeMetadata;
+ return this;
+ }
+
+ /**
+ * Returns the array sub-type {@link JdbcFieldInfo} defined for the provided column index.
+ *
+ * @param index The {@link java.sql.ResultSetMetaData} column index of an {@link java.sql.Types#ARRAY} type.
+ * @return The {@link JdbcFieldInfo} for that array's sub-type, or <code>null</code> if not defined.
+ */
+ public JdbcFieldInfo getArraySubTypeByColumnIndex(int index) {
+ if (arraySubTypesByColumnIndex == null) {
+ return null;
+ } else {
+ return arraySubTypesByColumnIndex.get(index);
+ }
+ }
+
+ /**
+ * Sets the mapping of column-index-to-{@link JdbcFieldInfo} used for columns of type {@link java.sql.Types#ARRAY}.
+ *
+ * @param map The mapping.
+ * @return This instance of the <code>JdbcToArrowConfig</code>, for chaining.
+ */
+ public JdbcToArrowConfig setArraySubTypeByColumnIndexMap(Map<Integer, JdbcFieldInfo> map) {
+ this.arraySubTypesByColumnIndex = map;
+ return this;
+ }
+
+ /**
+ * Returns the array sub-type {@link JdbcFieldInfo} defined for the provided column name.
+ *
+ * @param index The {@link java.sql.ResultSetMetaData} column name of an {@link java.sql.Types#ARRAY} type.
+ * @return The {@link JdbcFieldInfo} for that array's sub-type, or <code>null</code> if not defined.
+ */
+ public JdbcFieldInfo getArraySubTypeByColumnName(String name) {
+ if (arraySubTypesByColumnName == null) {
+ return null;
+ } else {
+ return arraySubTypesByColumnName.get(name);
+ }
+ }
+
+ /**
+ * Sets the mapping of column-name-to-{@link JdbcFieldInfo} used for columns of type {@link java.sql.Types#ARRAY}.
+ *
+ * @param map The mapping.
+ * @return This instance of the <code>JdbcToArrowConfig</code>, for chaining.
+ */
+ public JdbcToArrowConfig setArraySubTypeByColumnNameMap(Map<String, JdbcFieldInfo> map) {
+ this.arraySubTypesByColumnName = map;
+ return this;
+ }
+
+ /**
+ * Whether this configuration is valid. The configuration is valid when:
+ * <ul>
+ * <li>A memory allocator is provided.</li>
+ * <li>A calendar is provided.</li>
+ * </ul>
+ *
+ * @return Whether this configuration is valid.
+ */
+ public boolean isValid() {
+ return (calendar != null) && (allocator != null);
+ }
 }