From 097d30f507463c3114bd154271260e58d1114110 Mon Sep 17 00:00:00 2001
From: yan ma <yan.ma@intel.com>
Date: Thu, 15 Aug 2024 00:14:34 +0800
Subject: [PATCH] fix parsing array struct when repetition_type is 'REPEATED'

---
 velox/dwio/parquet/reader/ParquetReader.cpp    |  17 +++++++++++++++++
 .../examples/proto-struct-with-array.parquet   | Bin 0 -> 1576 bytes
 2 files changed, 17 insertions(+)
 create mode 100644 velox/dwio/parquet/tests/examples/proto-struct-with-array.parquet

diff --git a/velox/dwio/parquet/reader/ParquetReader.cpp b/velox/dwio/parquet/reader/ParquetReader.cpp
index 635366f19f136..d5b9c6995b9d4 100644
--- a/velox/dwio/parquet/reader/ParquetReader.cpp
+++ b/velox/dwio/parquet/reader/ParquetReader.cpp
@@ -464,6 +464,23 @@ std::unique_ptr<ParquetTypeWithId> ReaderBase::getParquetColumnInfo(
               maxDefine,
               isOptional,
               isRepeated);
+        } else {
+          // Row type
+          auto type =
+              createRowType(children, isFileColumnNamesReadAsLowerCase());
+          return std::make_unique<ParquetTypeWithId>(
+              std::move(type),
+              std::move(children),
+              curSchemaIdx,
+              maxSchemaElementIdx,
+              ParquetTypeWithId::kNonLeaf, // columnIdx,
+              std::move(name),
+              std::nullopt,
+              std::nullopt,
+              maxRepeat,
+              maxDefine,
+              isOptional,
+              isRepeated);
         }
       } else {
         // Row type
diff --git a/velox/dwio/parquet/tests/examples/proto-struct-with-array.parquet b/velox/dwio/parquet/tests/examples/proto-struct-with-array.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..325a8370ad20ec31010fde0c816895da9f5dbd27
GIT binary patch
literal 1576
zcmcIlO>5gg5S?rsql%Icnq5|qgD{N<#x?oCY2!n{ZAEKv64iDOJsH{F!~)4uB-v0(
z@BJM;_owufA5^-l4@Z{ekVAVh>zOxi-n`LDMyq>_0q^0x8b<t3#u5MmfXxk=|MG~0
zr8vb80MFte#Wcm26MS4HDz@=<n&cImf-GVxYLf6(akYj^QU8<=MTK(IF_MCmUmvwT
z3qr$pLw7j4srN<`=9ur?>74l?^SjJrp%q&06h8-y4iMdSJ@MDH4c~HjX3j($=&sN1
zW|q&!OYxG3d&~^8@dlzhDa$1b0`rz(6tkBD*J153G=T1;gzF$B0g1WSKnPOy6<d?`
zF;W^(mXreZ7ziR~<V@qfY8L`|00NkHUyKt0tE8%3aNr~cmLp(pz6jVPO>M$~KQ|W5
z5A#DO!$$Zca>TK`;67WBib&?m76{4rqTmNgwH)UCc)*uPhjcg;fc!=Tfl{N?GyS_6
z3+tZPeSOS=k#BjS>(f75Q`2EhwX*hMsK_@Kv&ZT;SydBky3mDR6_J}cL*_TtV}7>H
zA+wumr}b9v46coS`}(TY;qmaR$9wg^82X@n)jvIvzps*~J`|Fl<jK4Gr_b%K{{G42
zxdmpGpRSS{daM6r^3;}OgIVg=dFo<QF7;lo^+8XD86<YsAG@_Z%kgZsX!(et3(Ui=
z-sEK!+xeelE#2p=Tj#5?iWc!NWC4A#d^bk683govIT-TS=F%2kuDvcm<)rqN<HLLl
Iz~JAAzx9@|oB#j-

literal 0
HcmV?d00001