From 7fc33facb36bf545d26318d6a3ebcbc8f04c6a71 Mon Sep 17 00:00:00 2001
From: zhousipei <zsp233@qq.com>
Date: Thu, 19 Oct 2023 20:34:56 +0800
Subject: [PATCH] fix partition column repeated read

(cherry picked from commit 726dc84c6709ad52b7d2618a72effc0364fc70aa)
---
 .../boostkit/spark/jni/OrcColumnarBatchJniReader.java       | 3 +--
 .../sql/execution/datasources/orc/OmniOrcFileFormat.scala   | 6 +++---
 2 files changed, 4 insertions(+), 5 deletions(-)

diff --git a/omnioperator/omniop-spark-extension/java/src/main/java/com/huawei/boostkit/spark/jni/OrcColumnarBatchJniReader.java b/omnioperator/omniop-spark-extension/java/src/main/java/com/huawei/boostkit/spark/jni/OrcColumnarBatchJniReader.java
index 128ff6ca1..1cbc8c050 100644
--- a/omnioperator/omniop-spark-extension/java/src/main/java/com/huawei/boostkit/spark/jni/OrcColumnarBatchJniReader.java
+++ b/omnioperator/omniop-spark-extension/java/src/main/java/com/huawei/boostkit/spark/jni/OrcColumnarBatchJniReader.java
@@ -251,7 +251,6 @@ public class OrcColumnarBatchJniReader {
     }
 
     public int next(Vec[] vecList) {
-        int vectorCnt = vecList.length;
         int[] typeIds = new int[realColsCnt];
         long[] vecNativeIds = new long[realColsCnt];
         long rtn = recordReaderNext(recordReader, batchReader, typeIds, vecNativeIds);
@@ -259,7 +258,7 @@ public class OrcColumnarBatchJniReader {
             return 0;
         }
         int nativeGetId = 0;
-        for (int i = 0; i < vectorCnt; i++) {
+        for (int i = 0; i < realColsCnt; i++) {
             if (colsToGet[i] != 0) {
                 continue;
             }
diff --git a/omnioperator/omniop-spark-extension/java/src/main/scala/org/apache/spark/sql/execution/datasources/orc/OmniOrcFileFormat.scala b/omnioperator/omniop-spark-extension/java/src/main/scala/org/apache/spark/sql/execution/datasources/orc/OmniOrcFileFormat.scala
index 7325635ff..286a6e66b 100644
--- a/omnioperator/omniop-spark-extension/java/src/main/scala/org/apache/spark/sql/execution/datasources/orc/OmniOrcFileFormat.scala
+++ b/omnioperator/omniop-spark-extension/java/src/main/scala/org/apache/spark/sql/execution/datasources/orc/OmniOrcFileFormat.scala
@@ -100,8 +100,8 @@ class OmniOrcFileFormat extends FileFormat with DataSourceRegister with Serializ
         }
 
         val (requestedColIds, canPruneCols) = resultedColPruneInfo.get
-        val resultSchemaString = OrcUtils.orcResultSchemaString(canPruneCols,
-          dataSchema, resultSchema, partitionSchema, conf)
+        val requiredSchemaString = OrcUtils.orcResultSchemaString(canPruneCols,
+          dataSchema, requiredSchema, partitionSchema, conf)
         assert(requestedColIds.length == requiredSchema.length,
           "[BUG] requested column IDs do not match required schema")
         val taskConf = new Configuration(conf)
@@ -125,7 +125,7 @@ class OmniOrcFileFormat extends FileFormat with DataSourceRegister with Serializ
         SparkMemoryUtils.init()
           batchReader.initialize(fileSplit, taskAttemptContext)
           batchReader.initBatch(
-            TypeDescription.fromString(resultSchemaString),
+            TypeDescription.fromString(requiredSchemaString),
             resultSchema.fields,
             requestedDataColIds,
             requestedPartitionColIds,
-- 
Gitee