From 45389c4e48ff44bd924f3b02690b2c28d7dac320 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Hannes=20M=C3=BChleisen?= <hannes@muehleisen.org>
Date: Fri, 7 Oct 2022 11:10:02 +0200
Subject: [PATCH] Making sure parquet column readers return the expected amount
 of rows, fixes #4903

---
 data/parquet-testing/bug4903.parquet    | Bin 0 -> 4576 bytes
 extension/parquet/parquet_reader.cpp    |  17 +++++++++--------
 test/sql/copy/parquet/parquet_4903.test |   9 +++++++++
 3 files changed, 18 insertions(+), 8 deletions(-)
 create mode 100644 data/parquet-testing/bug4903.parquet
 create mode 100644 test/sql/copy/parquet/parquet_4903.test

diff --git a/data/parquet-testing/bug4903.parquet b/data/parquet-testing/bug4903.parquet
new file mode 100644
index 0000000000000000000000000000000000000000..2979df6993832e482683962e12f40573e777000f
GIT binary patch
literal 4576
zcmds5O=ufO6kcuDvTVt+Vtb}*5Umc?;({?P#D*GCQW{bdN)V;=P)gLbcdcx)UOBrf
z6Pq@N5`xJghZJ(?p@-t;V1f><&n3C^5PZnNgj{k<Lon$j=N$TGMmsaB)mo*drDQ>?
z`FY>>erD#)lj<LR_C8Wj8C7PH28}GrmUZpflW!GW`vcxs!J6Dx6s34lg}0U=(?k$Z
zU!kuhc<K7$Rr}XxKV5D%^vR9wg(rGqgwiHPlX9SFZjY%eLA#fjjCHOxt{qNq|K^09
z@9!)3|4ewJ=)?<Ehi`Zy>q(nDAbtv+maHE=<6N<=y6>#mdhT@FH$$i8QG5H&07c~+
zW?>H65nvSBf+%5=_VWh{M3}s(ZBieAk&R@Ud;nfS??^Ug>UyS5fwTpXQjOV=fp(p~
zk~-hIv;{iJo9PsNNhcX3AAswqEa}uh2;k(6HQa5nJ`5FRseXmMle)(!I8VTIM~OZR
zMgEd$G6C=+nh?4d00hsl&U_^E1ymCA%!*XovQylu7;FLDfakTC?m583uhe)HPm)zs
z6*a@-5`%Z+@_!MFt=Vg_)UkurjdtPy&7$|D9iBT+mJ}YHx9B0}JuwbK_;YqrB2e^=
z)ah5%CCzWOLS!gNt*Uz7@j~18Ot;}!aI9ct#B!Q;5c2-GV+En_c+E!8UR`yrRN;UU
zIn6WIY|<e#1=0!D%@v#D^1<fXa?5Rmo9kR`ENrd2_J++y2~^Xu2ik0_Vc9FLN%as!
zz<7~whi%`ZB#ppndM0#j$N->oN}luDu4^v4wlc5kF%ThZn0_-rql|KBrPcOA=EDS?
z<-A}!=q1h-BUnR{&NC&B>x7$3Y>2uYMP%HxEQeed5KYSr&1k$t$<|HZTx-Mt^2DlZ
zHUnZkmZVo&9>m^+H42c5vdTPM#;y~DNIgpO-*glV9H*Yxux;qF{1<^08m0JY`h*y_
z>{YYvh5?DW1`9aE;)%8C^7N~>ja(V_{(Ls4OviRSQsFM!!dH!4O|Ej-9VMzDI8w&-
zOc@or6yk_bf*(};qY47GWC4>Lq1=D3^f7kc5ITF0A=UgPJ`X}QQrpmUz-k=hJ5Au|
zF{Q%4XYhjz%+>V1l?*kk?+kH&e@H|ycG!6c_lF>cN7gmQgRHPiP7jo06<-8_8dNsD
z#f`ByER@Gn2-m(h4BACSL!%+}i3`6-n;<3WP`VAg1+qW~p!8eRxH}?@ie((0PPlP*
zM}SR~JT%f$pBT40T1HBDd-Y9hl&pqNfn;rg6-34%fV$gyCT2pEy#ki(DE1=u05H(4
z$9cX+DV%+hCRMPAKQtcaYuyXYxLp`T)RSa6$?XCsIT%PHCuPoXAQ*0q4KjlzagMz;
z#<f2bSiLv7bNPot{$`Qu(&gMvlNJZuEOG_=DFB+hIAC7@ZjAF4UrM<s=mn87Cv69k
z^=^z0+z#6%?$SR|n6_8W;&aA!iO1p5@wD{YNJPUUp<(C5vHVEvGD5#|f;;f?c=Xa1
zB*ZTT@x2MY;4{4!ue6%vaNnC4xSE|w9)$#3+ES5Y-I?T`zBq!S-#QTZ3jyDqO1B^p
zWP;tD;;Nq?ft4aiP|pSG4*a_jUEERx<7e^?;#!_c@_h;ndMeN0v&K^a>EJTby2+ar
z9#m>KgF6|cgKMHV7cF{2rIhanZ(>cES*)A><+dHpowLKrhV2JXSyrlFutv2qv)p!E
it5W^M3Fkg-dCiYoYimxpWCj<i?<$AzTe=Sa?feB2d8WYt

literal 0
HcmV?d00001

diff --git a/extension/parquet/parquet_reader.cpp b/extension/parquet/parquet_reader.cpp
index 123d100c3239..0b697de5f200 100644
--- a/extension/parquet/parquet_reader.cpp
+++ b/extension/parquet/parquet_reader.cpp
@@ -258,7 +258,7 @@ unique_ptr<ColumnReader> ParquetReader::CreateReaderRecursive(const FileMetaData
 
 	if (!s_ele.__isset.type) { // inner node
 		if (s_ele.num_children == 0) {
-			throw std::runtime_error("Node has no children but should");
+			throw InvalidInputException("Node has no children but should");
 		}
 		child_list_t<LogicalType> child_types;
 		vector<unique_ptr<ColumnReader>> child_readers;
@@ -575,7 +575,7 @@ uint64_t ParquetReader::GetGroupCompressedSize(ParquetReaderScanState &state) {
 
 	if (total_compressed_size != 0 && calc_compressed_size != 0 &&
 	    (idx_t)total_compressed_size != calc_compressed_size) {
-		throw std::runtime_error("mismatch between calculated compressed size and reported compressed size");
+		throw InvalidInputException("mismatch between calculated compressed size and reported compressed size");
 	}
 
 	return total_compressed_size ? total_compressed_size : calc_compressed_size;
@@ -918,7 +918,7 @@ bool ParquetReader::ScanInternal(ParquetReaderScanState &state, DataChunk &resul
 			double scan_percentage = (double)(to_scan_compressed_bytes) / total_row_group_span;
 
 			if (to_scan_compressed_bytes > total_row_group_span) {
-				throw std::runtime_error(
+				throw InvalidInputException(
 				    "Malformed parquet file: sum of total compressed bytes of columns seems incorrect");
 			}
 
@@ -1042,11 +1042,12 @@ bool ParquetReader::ScanInternal(ParquetReaderScanState &state, DataChunk &resul
 				continue;
 			}
 
-			//			std::cout << "Reading nofilter for col " <<
-			// root_reader->GetChildReader(file_col_idx)->Schema().name
-			//<< "\n";
-			root_reader->GetChildReader(file_col_idx)
-			    ->Read(result.size(), filter_mask, define_ptr, repeat_ptr, result.data[out_col_idx]);
+			auto rows_read = root_reader->GetChildReader(file_col_idx)
+			                     ->Read(result.size(), filter_mask, define_ptr, repeat_ptr, result.data[out_col_idx]);
+			if (rows_read != result.size()) {
+				throw InvalidInputException("Mismatch in parquet read for column %llu, expected %llu rows, got %llu",
+				                            file_col_idx, result.size(), rows_read);
+			}
 		}
 	}
 
diff --git a/test/sql/copy/parquet/parquet_4903.test b/test/sql/copy/parquet/parquet_4903.test
new file mode 100644
index 000000000000..778576be11ed
--- /dev/null
+++ b/test/sql/copy/parquet/parquet_4903.test
@@ -0,0 +1,9 @@
+# name: test/sql/copy/parquet/parquet_4903.test
+# description: Issue #4442: Parquet reader converts timestamp to i64 *sometimes*
+# group: [parquet]
+
+require parquet
+
+# file is corrupt
+statement error
+SELECT type_param_constraints FROM 'data/parquet-testing/bug4903.parquet' limit 10