Parquet: Introducing more bad_data for testing

mapleFU · Aug 15, 2024 · 060e0d5 · 060e0d5
1 parent f518d6b
commit 060e0d5
Show file tree

Hide file tree

Showing 3 changed files with 11 additions and 4 deletions.
diff --git a/cpp/build-support/fuzzing/generate_corpuses.sh b/cpp/build-support/fuzzing/generate_corpuses.sh
@@ -56,4 +56,5 @@ rm -rf ${CORPUS_DIR}
 ${OUT}/parquet-arrow-generate-fuzz-corpus ${CORPUS_DIR}
 # Add Parquet testing examples
 cp ${ARROW_CPP}/submodules/parquet-testing/data/*.parquet ${CORPUS_DIR}
+cp ${ARROW_CPP}/submodules/parquet-testing/bad_data/*.parquet ${CORPUS_DIR}
 ${ARROW_CPP}/build-support/fuzzing/pack_corpus.py ${CORPUS_DIR} ${OUT}/parquet-arrow-fuzz_seed_corpus.zip
diff --git a/cpp/src/parquet/arrow/arrow_reader_writer_test.cc b/cpp/src/parquet/arrow/arrow_reader_writer_test.cc
@@ -5298,14 +5298,20 @@ TEST(TestArrowReadWrite, MultithreadedWrite) {
 
 TEST(TestArrowReadWrite, FuzzReader) {
   constexpr size_t kMaxFileSize = 1024 * 1024 * 1;
-  {
-    auto path = test::get_data_file("PARQUET-1481.parquet", /*is_good=*/false);
+  auto check_bad_file = [](const std::string& file_name) {
+    SCOPED_TRACE(file_name);
+    auto path = test::get_data_file(file_name, /*is_good=*/false);
     PARQUET_ASSIGN_OR_THROW(auto source, ::arrow::io::MemoryMappedFile::Open(
                                              path, ::arrow::io::FileMode::READ));
     PARQUET_ASSIGN_OR_THROW(auto buffer, source->Read(kMaxFileSize));
     auto s = internal::FuzzReader(buffer->data(), buffer->size());
     ASSERT_NOT_OK(s);
-  }
+  };
+  check_bad_file("PARQUET-1481.parquet");
+  check_bad_file("ARROW-GH-41317.parquet");
+  check_bad_file("ARROW-GH-41321.parquet");
+  check_bad_file("ARROW-RS-GH-6229-LEVELS.parquet");
+  check_bad_file("ARROW-RS-GH-6229-DICTHEADER.parquet");
   {
     auto path = test::get_data_file("alltypes_plain.parquet", /*is_good=*/true);
     PARQUET_ASSIGN_OR_THROW(auto source, ::arrow::io::MemoryMappedFile::Open(

diff --git a/cpp/submodules/parquet-testing b/cpp/submodules/parquet-testing
+43 −0		.asf.yaml
+ −		bad_data/ARROW-GH-41317.parquet
+ −		bad_data/ARROW-GH-41321.parquet
+ −		bad_data/ARROW-RS-GH-6229-DICTHEADER.parquet
+ −		bad_data/ARROW-RS-GH-6229-LEVELS.parquet
+9 −1		bad_data/README.md
+41 −0		data/README.md
+ −		data/column_chunk_key_value_metadata.parquet
+ −		data/incorrect_map_schema.parquet
+ −		data/sort_columns.parquet