[orc] branch branch-1.6 updated: ORC-688: Allow CHAR to be promoted to STRING (#575)

[pg...@apache.org]

This is an automated email from the ASF dual-hosted git repository.

pgaref pushed a commit to branch branch-1.6
in repository https://gitbox.apache.org/repos/asf/orc.git


The following commit(s) were added to refs/heads/branch-1.6 by this push:
     new fb18bbb  ORC-688: Allow CHAR to be promoted to STRING (#575)
fb18bbb is described below

commit fb18bbb02c08b8a0da5ad811210102cf7b421096
Author: Panagiotis Garefalakis <pg...@apache.org>
AuthorDate: Tue Dec 1 09:24:51 2020 +0000

    ORC-688: Allow CHAR to be promoted to STRING (#575)
    
    ### What changes were proposed in this pull request?
    
    Enable String conversion from varchar.
    Currently createStringConvertTreeReader in ConvertTreeReaderFactory is missing that opportunity
    
    ### Why are the changes needed?
    
    Modify createStringConvertTreeReader to handle char, varchar and string as part of the same group.
    
    ### How was this patch tested?
    
    TestSchemaEvolution.testCharToStringEvolution
    
    (cherry picked from commit 48cba3bafa7ef31f162f8c377d513f1c58a27876)
    Change-Id: Ibf1daa60665784c21dc7d25066d85e1b7193b155
---
 .../apache/orc/impl/ConvertTreeReaderFactory.java  |  7 +----
 .../org/apache/orc/impl/TestSchemaEvolution.java   | 32 ++++++++++++++++++++++
 2 files changed, 33 insertions(+), 6 deletions(-)

diff --git a/java/core/src/java/org/apache/orc/impl/ConvertTreeReaderFactory.java b/java/core/src/java/org/apache/orc/impl/ConvertTreeReaderFactory.java
index 6a8682d..bd6bac5 100644
--- a/java/core/src/java/org/apache/orc/impl/ConvertTreeReaderFactory.java
+++ b/java/core/src/java/org/apache/orc/impl/ConvertTreeReaderFactory.java
@@ -1914,14 +1914,9 @@ public class ConvertTreeReaderFactory extends TreeReaderFactory {
       return new DecimalFromStringGroupTreeReader(columnId, fileType, readerType, context);
 
     case CHAR:
-      return new StringGroupFromStringGroupTreeReader(columnId, fileType, readerType, context);
-
     case VARCHAR:
-      return new StringGroupFromStringGroupTreeReader(columnId, fileType, readerType, context);
-
     case STRING:
-      throw new IllegalArgumentException("No conversion of type " +
-          readerType.getCategory() + " to self needed");
+      return new StringGroupFromStringGroupTreeReader(columnId, fileType, readerType, context);
 
     case BINARY:
       return new BinaryTreeReader(columnId, context);
diff --git a/java/core/src/test/org/apache/orc/impl/TestSchemaEvolution.java b/java/core/src/test/org/apache/orc/impl/TestSchemaEvolution.java
index af30854..94c3b25 100644
--- a/java/core/src/test/org/apache/orc/impl/TestSchemaEvolution.java
+++ b/java/core/src/test/org/apache/orc/impl/TestSchemaEvolution.java
@@ -577,6 +577,38 @@ public class TestSchemaEvolution {
   }
 
   @Test
+  public void testCharToStringEvolution() throws IOException {
+    TypeDescription fileType = TypeDescription.fromString("struct<x:char(10)>");
+    TypeDescription readType = TypeDescription.fromString("struct<x:string>");
+    SchemaEvolution evo = new SchemaEvolution(fileType, readType, options);
+
+    TreeReaderFactory.Context treeContext =
+        new TreeReaderFactory.ReaderContext().setSchemaEvolution(evo);
+    TreeReaderFactory.TreeReader reader =
+        TreeReaderFactory.createTreeReader(readType, treeContext);
+
+    // Make sure the tree reader is built properly
+    assertEquals(TreeReaderFactory.StructTreeReader.class, reader.getClass());
+    TreeReaderFactory.TreeReader[] children =
+        ((TreeReaderFactory.StructTreeReader) reader).getChildReaders();
+    assertEquals(1, children.length);
+    assertEquals(ConvertTreeReaderFactory.StringGroupFromStringGroupTreeReader.class, children[0].getClass());
+
+    // Make sure that varchar behaves the same as char
+    fileType = TypeDescription.fromString("struct<x:varchar(10)>");
+    evo = new SchemaEvolution(fileType, readType, options);
+
+    treeContext = new TreeReaderFactory.ReaderContext().setSchemaEvolution(evo);
+    reader = TreeReaderFactory.createTreeReader(readType, treeContext);
+
+    // Make sure the tree reader is built properly
+    assertEquals(TreeReaderFactory.StructTreeReader.class, reader.getClass());
+    children = ((TreeReaderFactory.StructTreeReader) reader).getChildReaders();
+    assertEquals(1, children.length);
+    assertEquals(ConvertTreeReaderFactory.StringGroupFromStringGroupTreeReader.class, children[0].getClass());
+  }
+
+  @Test
   public void testStringToDecimalEvolution() throws Exception {
     testFilePath = new Path(workDir, "TestSchemaEvolution." +
       testCaseName.getMethodName() + ".orc");