Test for filtering records across multiple blocks

Author: Steven Willis

parquet-avro/src/test/java/parquet/avro/TestSpecificReadWrite.java

@@ -76,6 +76,17 @@ public void testFilterMatchesMultiple() throws IOException {
     assertNull(reader.read());
   }
 
+  @Test
+  public void testFilterMatchesMultipleBlocks() throws IOException {
+    Path path = writeCarsToParquetFile(10000, CompressionCodecName.UNCOMPRESSED, false, DEFAULT_BLOCK_SIZE/64, DEFAULT_PAGE_SIZE/64);
+    ParquetReader<Car> reader = new AvroParquetReader<Car>(path, column("make", equalTo("Volkswagen")));
+    for (int i = 0; i < 10000; i++) {
+      assertEquals(getVwPolo().toString(), reader.read().toString());
+      assertEquals(getVwPassat().toString(), reader.read().toString());
+    }
+    assertNull(reader.read());
+  }
+
   @Test
   public void testFilterWithDictionary() throws IOException {
     Path path = writeCarsToParquetFile(1,CompressionCodecName.UNCOMPRESSED,true);
@@ -159,6 +170,10 @@ public void testAvroReadSchema() throws IOException {
   }
 
   private Path writeCarsToParquetFile( int num, CompressionCodecName compression, boolean enableDictionary) throws IOException {
+    return writeCarsToParquetFile(num, compression, enableDictionary, DEFAULT_BLOCK_SIZE, DEFAULT_PAGE_SIZE);
+  }
+
+  private Path writeCarsToParquetFile( int num, CompressionCodecName compression, boolean enableDictionary, int blockSize, int pageSize) throws IOException {
     File tmp = File.createTempFile(getClass().getSimpleName(), ".tmp");
     tmp.deleteOnExit();
     tmp.delete();
@@ -169,7 +184,7 @@ private Path writeCarsToParquetFile( int num, CompressionCodecName compression,
     Car bmwMini  = getBmwMini();
 
     ParquetWriter<Car> writer = new AvroParquetWriter<Car>(path,Car.SCHEMA$, compression,
-        DEFAULT_BLOCK_SIZE, DEFAULT_PAGE_SIZE, enableDictionary);
+        blockSize, pageSize, enableDictionary);
     for (int i = 0; i < num; i++) {
       writer.write(vwPolo);
       writer.write(vwPassat);