feat(java): support delete rows from the dataset (#3498)

Author: yanghua

java/core/lance-jni/src/blocking_dataset.rs

@@ -861,6 +861,23 @@ fn inner_take(
     Ok(**byte_array)
 }
 
+#[no_mangle]
+pub extern "system" fn Java_com_lancedb_lance_Dataset_nativeDelete(
+    mut env: JNIEnv,
+    java_dataset: JObject,
+    predicate: JString,
+) {
+    ok_or_throw_without_return!(env, inner_delete(&mut env, java_dataset, predicate))
+}
+
+fn inner_delete(env: &mut JNIEnv, java_dataset: JObject, predicate: JString) -> Result<()> {
+    let predicate_str = predicate.extract(env)?;
+    let mut dataset_guard =
+        unsafe { env.get_rust_field::<_, _, BlockingDataset>(java_dataset, NATIVE_DATASET) }?;
+    RT.block_on(dataset_guard.inner.delete(&predicate_str))?;
+    Ok(())
+}
+
 //////////////////////////////
 // Schema evolution Methods //
 //////////////////////////////

java/core/src/main/java/com/lancedb/lance/Dataset.java

@@ -391,6 +391,20 @@ public void close() throws IOException {
 
   private native byte[] nativeTake(List<Long> indices, List<String> columns);
 
+  /**
+   * Delete rows of data by predicate.
+   *
+   * @param predicate the predicate to delete
+   */
+  public void delete(String predicate) {
+    try (LockManager.WriteLock writeLock = lockManager.acquireWriteLock()) {
+      Preconditions.checkArgument(nativeDatasetHandle != 0, "Dataset is closed");
+      nativeDelete(predicate);
+    }
+  }
+
+  private native void nativeDelete(String predicate);
+
   /**
    * Gets the URI of the dataset.
    *

java/core/src/test/java/com/lancedb/lance/DatasetTest.java

@@ -531,4 +531,39 @@ void testCalculateDataSize() {
       }
     }
   }
+
+  @Test
+  void testDeleteRows() {
+    String testMethodName = new Object() {}.getClass().getEnclosingMethod().getName();
+    String datasetPath = tempDir.resolve(testMethodName).toString();
+    try (RootAllocator allocator = new RootAllocator(Long.MAX_VALUE)) {
+      TestUtils.SimpleTestDataset testDataset =
+          new TestUtils.SimpleTestDataset(allocator, datasetPath);
+      dataset = testDataset.createEmptyDataset();
+
+      try (Dataset dataset2 = testDataset.write(1, 5)) {
+        // Initially there are 5 rows
+        assertEquals(5, dataset2.countRows());
+
+        // Delete rows where id > 2 (should delete id=3, id=4)
+        dataset2.delete("id > 2");
+
+        // Now verify we have 3 rows left (id=0, id=1, id=2)
+        assertEquals(3, dataset2.countRows());
+
+        // Verify the rows that remain
+        assertEquals(0, dataset2.countRows("id > 2"));
+        assertEquals(3, dataset2.countRows("id <= 2"));
+
+        // Delete another row
+        dataset2.delete("id = 1");
+
+        // Now verify we have 2 rows left (id=0, id=2)
+        assertEquals(2, dataset2.countRows());
+        assertEquals(1, dataset2.countRows("id = 0"));
+        assertEquals(1, dataset2.countRows("id = 2"));
+        assertEquals(0, dataset2.countRows("id = 1"));
+      }
+    }
+  }
 }