1717
1818package org .apache .arrow .adapter .jdbc ;
1919
20- import static org .apache .arrow .vector .types .FloatingPointPrecision .DOUBLE ;
21- import static org .apache .arrow .vector .types .FloatingPointPrecision .SINGLE ;
22-
23- import java .sql .Types ;
2420import java .util .Calendar ;
2521import java .util .Map ;
2622import java .util .function .Function ;
2723
2824import org .apache .arrow .memory .BufferAllocator ;
2925import org .apache .arrow .util .Preconditions ;
30- import org .apache .arrow .vector .types .DateUnit ;
31- import org .apache .arrow .vector .types .TimeUnit ;
3226import org .apache .arrow .vector .types .pojo .ArrowType ;
3327
3428/**
5549 */
5650public final class JdbcToArrowConfig {
5751
52+ public static final int DEFAULT_TARGET_BATCH_SIZE = 1024 ;
53+ public static final int NO_LIMIT_BATCH_SIZE = -1 ;
5854 private final Calendar calendar ;
5955 private final BufferAllocator allocator ;
6056 private final boolean includeMetadata ;
6157 private final boolean reuseVectorSchemaRoot ;
6258 private final Map <Integer , JdbcFieldInfo > arraySubTypesByColumnIndex ;
6359 private final Map <String , JdbcFieldInfo > arraySubTypesByColumnName ;
64-
65- public static final int DEFAULT_TARGET_BATCH_SIZE = 1024 ;
66- public static final int NO_LIMIT_BATCH_SIZE = -1 ;
67-
6860 /**
6961 * The maximum rowCount to read each time when partially convert data.
7062 * Default value is 1024 and -1 means disable partial read.
@@ -82,7 +74,7 @@ public final class JdbcToArrowConfig {
8274 /**
8375 * Constructs a new configuration from the provided allocator and calendar. The <code>allocator</code>
8476 * is used when constructing the Arrow vectors from the ResultSet, and the calendar is used to define
85- * Arrow Timestamp fields, and to read time-based fields from the JDBC <code>ResultSet</code>.
77+ * Arrow Timestamp fields, and to read time-based fields from the JDBC <code>ResultSet</code>.
8678 *
8779 * @param allocator The memory allocator to construct the Arrow vectors with.
8880 * @param calendar The calendar to use when constructing Timestamp fields and reading time-based results.
@@ -99,7 +91,7 @@ public final class JdbcToArrowConfig {
9991 /**
10092 * Constructs a new configuration from the provided allocator and calendar. The <code>allocator</code>
10193 * is used when constructing the Arrow vectors from the ResultSet, and the calendar is used to define
102- * Arrow Timestamp fields, and to read time-based fields from the JDBC <code>ResultSet</code>.
94+ * Arrow Timestamp fields, and to read time-based fields from the JDBC <code>ResultSet</code>.
10395 *
10496 * @param allocator The memory allocator to construct the Arrow vectors with.
10597 * @param calendar The calendar to use when constructing Timestamp fields and reading time-based results.
@@ -134,6 +126,8 @@ public final class JdbcToArrowConfig {
134126 * <li>TIMESTAMP --> ArrowType.Timestamp(TimeUnit.MILLISECOND, calendar timezone)</li>
135127 * <li>CLOB --> ArrowType.Utf8</li>
136128 * <li>BLOB --> ArrowType.Binary</li>
129+ * <li>ARRAY --> ArrowType.List</li>
130+ * <li>STRUCT --> ArrowType.Struct</li>
137131 * <li>NULL --> ArrowType.Null</li>
138132 * </ul>
139133 */
@@ -157,64 +151,7 @@ public final class JdbcToArrowConfig {
157151
158152 // set up type converter
159153 this .jdbcToArrowTypeConverter = jdbcToArrowTypeConverter != null ? jdbcToArrowTypeConverter :
160- fieldInfo -> {
161- final String timezone ;
162- if (calendar != null ) {
163- timezone = calendar .getTimeZone ().getID ();
164- } else {
165- timezone = null ;
166- }
167-
168- switch (fieldInfo .getJdbcType ()) {
169- case Types .BOOLEAN :
170- case Types .BIT :
171- return new ArrowType .Bool ();
172- case Types .TINYINT :
173- return new ArrowType .Int (8 , true );
174- case Types .SMALLINT :
175- return new ArrowType .Int (16 , true );
176- case Types .INTEGER :
177- return new ArrowType .Int (32 , true );
178- case Types .BIGINT :
179- return new ArrowType .Int (64 , true );
180- case Types .NUMERIC :
181- case Types .DECIMAL :
182- int precision = fieldInfo .getPrecision ();
183- int scale = fieldInfo .getScale ();
184- return new ArrowType .Decimal (precision , scale , 128 );
185- case Types .REAL :
186- case Types .FLOAT :
187- return new ArrowType .FloatingPoint (SINGLE );
188- case Types .DOUBLE :
189- return new ArrowType .FloatingPoint (DOUBLE );
190- case Types .CHAR :
191- case Types .NCHAR :
192- case Types .VARCHAR :
193- case Types .NVARCHAR :
194- case Types .LONGVARCHAR :
195- case Types .LONGNVARCHAR :
196- case Types .CLOB :
197- return new ArrowType .Utf8 ();
198- case Types .DATE :
199- return new ArrowType .Date (DateUnit .DAY );
200- case Types .TIME :
201- return new ArrowType .Time (TimeUnit .MILLISECOND , 32 );
202- case Types .TIMESTAMP :
203- return new ArrowType .Timestamp (TimeUnit .MILLISECOND , timezone );
204- case Types .BINARY :
205- case Types .VARBINARY :
206- case Types .LONGVARBINARY :
207- case Types .BLOB :
208- return new ArrowType .Binary ();
209- case Types .ARRAY :
210- return new ArrowType .List ();
211- case Types .NULL :
212- return new ArrowType .Null ();
213- default :
214- // no-op, shouldn't get here
215- return null ;
216- }
217- };
154+ jdbcFieldInfo -> JdbcToArrowUtils .getArrowTypeFromJdbcType (jdbcFieldInfo , calendar );
218155 }
219156
220157 /**
@@ -230,6 +167,7 @@ public Calendar getCalendar() {
230167
231168 /**
232169 * The Arrow memory allocator.
170+ *
233171 * @return the allocator.
234172 */
235173 public BufferAllocator getAllocator () {
0 commit comments