Making sure parquet column readers return the expected amount of rows, …

…fixes duckdb#4903
MSRG · Oct 7, 2022 · 45389c4 · 45389c4
1 parent 307477b
commit 45389c4
Show file tree

Hide file tree

Showing 3 changed files with 18 additions and 8 deletions.
diff --git a/data/parquet-testing/bug4903.parquet b/data/parquet-testing/bug4903.parquet
diff --git 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
@@ -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