From dc5ef4f73f76e68a60590d7203f82b3ebecc6fb0 Mon Sep 17 00:00:00 2001
From: hyukjinkwon <gurwls223@gmail.com>
Date: Tue, 17 Nov 2015 11:03:32 +0900
Subject: [PATCH] [SPARK-11694][FOLLOW-UP] Clean up imports, use a common
 function for metadata and add a test for FIXED_LEN_BYTE_ARRAY

---
 .../test/resources/dec-in-fixed-len.parquet   | Bin 0 -> 460 bytes
 .../datasources/parquet/ParquetIOSuite.scala  |  42 +++++++-----------
 2 files changed, 15 insertions(+), 27 deletions(-)
 create mode 100644 sql/core/src/test/resources/dec-in-fixed-len.parquet

diff --git a/sql/core/src/test/resources/dec-in-fixed-len.parquet b/sql/core/src/test/resources/dec-in-fixed-len.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..6ad37d5639511cdb430f33fa6165eb70cd9034c0
GIT binary patch
literal 460
zcmZuu%SyvQ6rI#oO3~7Vw<M577-(rBhCY%QsvCDMM2ff&RLXQ_qS2&}JbX~P^8;M^
zF@As_;wSV2ypakj4s*|)nRD+w2WEJ9+Jzzr5XZ0s<i`^E`5i+sb9l*Ng-|%p;SFKK
z$l)0_K?U^^3?LXaoJLHwgp{&cr#fDUp%hEC-r4;t0u(g!Xe*hq_->T&8yPVu5U4^0
z3hK5WJW4SN<zyrzZ&YyJ4yhZOuIbIP4ppsDBif%Tz=HaRg1cEh_(wbE45I03dm3L`
zY{pR2`jWSjd1#@Jc_vdUo<3NSOqpOQQ*ReEeVR!+ph+5MJf$Y}0~v~h4lWjS!lIn|
z<rOQ|9cVm=SlH@z%w8LxCs`P>aflr_N}BXrgbo(V<<*j?`)dokQEarvSr7`N-{ZFH
k+I`P<B=gu_$KicP*phobcO9E!_Pwqv+T=tbWQhOx35yqQV*mgE

literal 0
HcmV?d00001

diff --git a/sql/core/src/test/scala/org/apache/spark/sql/execution/datasources/parquet/ParquetIOSuite.scala b/sql/core/src/test/scala/org/apache/spark/sql/execution/datasources/parquet/ParquetIOSuite.scala
index a148facd056a..177ab42f7767 100644
--- a/sql/core/src/test/scala/org/apache/spark/sql/execution/datasources/parquet/ParquetIOSuite.scala
+++ b/sql/core/src/test/scala/org/apache/spark/sql/execution/datasources/parquet/ParquetIOSuite.scala
@@ -17,8 +17,6 @@
 
 package org.apache.spark.sql.execution.datasources.parquet
 
-import java.util.Collections
-
 import org.apache.parquet.column.{Encoding, ParquetProperties}
 
 import scala.collection.JavaConverters._
@@ -33,7 +31,7 @@ import org.apache.parquet.example.data.{Group, GroupWriter}
 import org.apache.parquet.hadoop._
 import org.apache.parquet.hadoop.api.WriteSupport
 import org.apache.parquet.hadoop.api.WriteSupport.WriteContext
-import org.apache.parquet.hadoop.metadata.{CompressionCodecName, FileMetaData, ParquetMetadata}
+import org.apache.parquet.hadoop.metadata.CompressionCodecName
 import org.apache.parquet.io.api.RecordConsumer
 import org.apache.parquet.schema.{MessageType, MessageTypeParser}
 
@@ -243,15 +241,9 @@ class ParquetIOSuite extends QueryTest with ParquetTest with SharedSQLContext {
       """.stripMargin)
 
     withTempPath { location =>
-      val extraMetadata = Map.empty[String, String].asJava
-      val fileMetadata = new FileMetaData(parquetSchema, extraMetadata, "Spark")
       val path = new Path(location.getCanonicalPath)
-      val footer = List(
-        new Footer(path, new ParquetMetadata(fileMetadata, Collections.emptyList()))
-      ).asJava
-
-      ParquetFileWriter.writeMetadataFile(sparkContext.hadoopConfiguration, path, footer)
-
+      val conf = sparkContext.hadoopConfiguration
+      writeMetadata(parquetSchema, path, conf)
       val errorMessage = intercept[Throwable] {
         sqlContext.read.parquet(path.toString).printSchema()
       }.toString
@@ -267,20 +259,14 @@ class ParquetIOSuite extends QueryTest with ParquetTest with SharedSQLContext {
         |}
       """.stripMargin)
 
+    val expectedSparkTypes = Seq(StringType, BinaryType)
+
     withTempPath { location =>
-      val extraMetadata = Map.empty[String, String].asJava
-      val fileMetadata = new FileMetaData(parquetSchema, extraMetadata, "Spark")
       val path = new Path(location.getCanonicalPath)
-      val footer = List(
-        new Footer(path, new ParquetMetadata(fileMetadata, Collections.emptyList()))
-      ).asJava
-
-      ParquetFileWriter.writeMetadataFile(sparkContext.hadoopConfiguration, path, footer)
-
-      val jsonDataType = sqlContext.read.parquet(path.toString).schema(0).dataType
-      assert(jsonDataType === StringType)
-      val bsonDataType = sqlContext.read.parquet(path.toString).schema(1).dataType
-      assert(bsonDataType === BinaryType)
+      val conf = sparkContext.hadoopConfiguration
+      writeMetadata(parquetSchema, path, conf)
+      val sparkTypes = sqlContext.read.parquet(path.toString).schema.map(_.dataType)
+      assert(sparkTypes === expectedSparkTypes)
     }
   }
 
@@ -607,10 +593,12 @@ class ParquetIOSuite extends QueryTest with ParquetTest with SharedSQLContext {
       sqlContext.range(1 << 4).select('id % 10 cast DecimalType(10, 2) as 'i64_dec))
   }
 
-  // TODO Adds test case for reading dictionary encoded decimals written as `FIXED_LEN_BYTE_ARRAY`
-  // The Parquet writer version Spark 1.6 and prior versions use is `PARQUET_1_0`, which doesn't
-  // provide dictionary encoding support for `FIXED_LEN_BYTE_ARRAY`.  Should add a test here once
-  // we upgrade to `PARQUET_2_0`.
+  test("read dictionary encoded decimals written as FIXED_LEN_BYTE_ARRAY") {
+    checkAnswer(
+      // Decimal column in this file is encoded using plain dictionary
+      readResourceParquetFile("dec-in-fixed-len.parquet"),
+      sqlContext.range(1 << 4).select('id % 10 cast DecimalType(10, 2) as 'fixed_len_dec))
+  }
 }
 
 class JobCommitFailureParquetOutputCommitter(outputPath: Path, context: TaskAttemptContext)