From fc5f028c973f3983351c9914b676916983a371cf Mon Sep 17 00:00:00 2001
From: frank chen <frank.chen021@outlook.com>
Date: Fri, 18 Jun 2021 14:51:00 +0800
Subject: [PATCH] Add more tests to ORC format

---
 .../example/test_json_path_functions.orc      | Bin 0 -> 473 bytes
 .../druid/data/input/orc/OrcReaderTest.java   |  58 +++++++++++++++++-
 2 files changed, 57 insertions(+), 1 deletion(-)
 create mode 100644 extensions-core/orc-extensions/example/test_json_path_functions.orc

diff --git a/extensions-core/orc-extensions/example/test_json_path_functions.orc b/extensions-core/orc-extensions/example/test_json_path_functions.orc
new file mode 100644
index 0000000000000000000000000000000000000000..e3916d6a23308afc61f3c1a4573022ea9aab116b
GIT binary patch
literal 473
zcmeYdau#G@;9?VE;b074Fk@hNJn3mtDg&dUAfKD6o!+!5XJVWc4Kxo59$;XQ1}fy{
zVrF1qfT-36@`WLM9u8(9HV!@kAqj&31{07ekCT!am_erK+jX|3nwP&d3vUQu<YC}s
zU`WZ#HDqUCaAR-)l1>Z<xEUDyn1FhD85qhK%s1|m2MRkdFf0NZlk_-2IAOwrXG|A+
zgo+Yoc{(xp7<+p(WQuV{DxBq*lFoj~$6*$)k8xwB+X5>-gR@EpIhRB>n#>juW@i>K
zR!r86JO1d5%ijk*NA`T_IpV|60W|Z|<D`cPJP8kk9w&r6dYt5W@{Fkn(_|qTMjpXE
zF*yo1In4Cz_|g>3`0N-JIhhsaGcUQ<?Y}7T=GoZYXGIl60_Mmw@G&wxP7td&uW|I~
zL7fTe3zrHkI=DjR=t1QpXHFbZIvN)fG-c-WJdLB1CP&679SsRubL7aubw^GeR6TNn
zC#A_?IpbncZuVq`=BX0wi-qLHy}i=bEfknB<pD=R$Ahg&%qQJ8M1CwZ;NDpHfs?~i
zg5jr(!q>POA>Pw@?{4L;V_4Ym=ntRp9~%iJ2}X`CMgc~Nh6Vv8CI%i2cLVmOFU-vT
HLCz8YkPMkW

literal 0
HcmV?d00001

diff --git a/extensions-core/orc-extensions/src/test/java/org/apache/druid/data/input/orc/OrcReaderTest.java b/extensions-core/orc-extensions/src/test/java/org/apache/druid/data/input/orc/OrcReaderTest.java
index 093ae7ab87e..9b2b8e7e6b1 100644
--- a/extensions-core/orc-extensions/src/test/java/org/apache/druid/data/input/orc/OrcReaderTest.java
+++ b/extensions-core/orc-extensions/src/test/java/org/apache/druid/data/input/orc/OrcReaderTest.java
@@ -112,7 +112,11 @@ public class OrcReaderTest
             ImmutableList.of(
                 new JSONPathFieldSpec(JSONPathFieldType.PATH, "struct_list_struct_int", "$.middle.list[1].int1"),
                 new JSONPathFieldSpec(JSONPathFieldType.PATH, "struct_list_struct_intlist", "$.middle.list[*].int1"),
-                new JSONPathFieldSpec(JSONPathFieldType.PATH, "struct_list_struct_middleListLength", "$.middle.list.length()"),
+                new JSONPathFieldSpec(
+                    JSONPathFieldType.PATH,
+                    "struct_list_struct_middleListLength",
+                    "$.middle.list.length()"
+                ),
                 new JSONPathFieldSpec(JSONPathFieldType.PATH, "list_struct_string", "$.list[0].string1"),
                 new JSONPathFieldSpec(JSONPathFieldType.PATH, "map_struct_int", "$.map.chani.int1")
             )
@@ -257,6 +261,58 @@ public class OrcReaderTest
     }
   }
 
+  /**
+   * schema: struct<string1:string, list:array<int>, ts:timestamp>
+   * data:   {"dim1","[7,8,9]","2000-03-12 15:00:00"}
+   */
+  @Test
+  public void testJsonPathFunctions() throws IOException
+  {
+    final OrcInputFormat inputFormat = new OrcInputFormat(
+        new JSONPathSpec(
+            true,
+            ImmutableList.of(
+                new JSONPathFieldSpec(JSONPathFieldType.PATH, "min", "$.list.min()"),
+                new JSONPathFieldSpec(JSONPathFieldType.PATH, "max", "$.list.max()"),
+                new JSONPathFieldSpec(JSONPathFieldType.PATH, "avg", "$.list.avg()"),
+                new JSONPathFieldSpec(JSONPathFieldType.PATH, "len", "$.list.length()"),
+                new JSONPathFieldSpec(JSONPathFieldType.PATH, "sum", "$.list.sum()"),
+                new JSONPathFieldSpec(JSONPathFieldType.PATH, "stddev", "$.list.stddev()"),
+                new JSONPathFieldSpec(JSONPathFieldType.PATH, "append", "$.list.append(10)")
+            )
+        ),
+        null,
+        new Configuration()
+    );
+    final InputEntityReader reader = createReader(
+        new TimestampSpec("ts", "millis", null),
+        new DimensionsSpec(null),
+        inputFormat,
+        "example/test_json_path_functions.orc"
+    );
+    try (CloseableIterator<InputRow> iterator = reader.read()) {
+      int actualRowCount = 0;
+
+      while (iterator.hasNext()) {
+        final InputRow row = iterator.next();
+        actualRowCount++;
+
+        Assert.assertEquals("7.0", Iterables.getOnlyElement(row.getDimension("min")));
+        Assert.assertEquals("8.0", Iterables.getOnlyElement(row.getDimension("avg")));
+        Assert.assertEquals("9.0", Iterables.getOnlyElement(row.getDimension("max")));
+        Assert.assertEquals("24.0", Iterables.getOnlyElement(row.getDimension("sum")));
+        Assert.assertEquals("3", Iterables.getOnlyElement(row.getDimension("len")));
+
+        //deviation of [7,8,9] is 1/3, stddev is sqrt(1/3), approximately 0.8165
+        Assert.assertEquals(0.8165, Double.parseDouble(Iterables.getOnlyElement(row.getDimension("stddev"))), 0.0001);
+
+        //append is not supported
+        Assert.assertEquals(Collections.emptyList(), row.getDimension("append"));
+      }
+      Assert.assertEquals(1, actualRowCount);
+    }
+  }
+
   private InputEntityReader createReader(
       TimestampSpec timestampSpec,
       DimensionsSpec dimensionsSpec,