From b08803aa8943d005f889f15234fddd271772d5f6 Mon Sep 17 00:00:00 2001
From: mwish <maplewish117@gmail.com>
Date: Thu, 16 May 2024 23:08:19 +0800
Subject: [PATCH] Update test and err message

---
 cpp/src/parquet/column_reader.cc      |  2 +-
 cpp/src/parquet/column_reader_test.cc | 55 ++++++++-------------------
 cpp/submodules/parquet-testing        |  2 +-
 3 files changed, 18 insertions(+), 41 deletions(-)

diff --git a/cpp/src/parquet/column_reader.cc b/cpp/src/parquet/column_reader.cc
index a048aee9edfc1..a053971f95624 100644
--- a/cpp/src/parquet/column_reader.cc
+++ b/cpp/src/parquet/column_reader.cc
@@ -103,7 +103,7 @@ inline void CheckNumberDecoded(int64_t number_decoded, int64_t expected) {
 }
 
 constexpr std::string_view kErrorRepDefLevelNotMatchesNumValues =
-    "Number of decoded rep / def levels did less than num_values in page_header";
+    "Number of decoded rep / def levels did more or less than num_values in page_header";
 constexpr std::string_view kErrorRepDefLevelInEqual =
     "Number of decoded rep / def levels did not match";
 
diff --git a/cpp/src/parquet/column_reader_test.cc b/cpp/src/parquet/column_reader_test.cc
index 4c51a6e2cea1c..998f1c37be652 100644
--- a/cpp/src/parquet/column_reader_test.cc
+++ b/cpp/src/parquet/column_reader_test.cc
@@ -437,17 +437,16 @@ TEST_F(TestPrimitiveReader, TestReadValuesMissing) {
 TEST_F(TestPrimitiveReader, DefLevelNotExpected) {
   max_def_level_ = 1;
   max_rep_level_ = 0;
-  std::vector<bool> values(1, false);
-  // storing def-levels less than value in page-header
-  {
-    std::vector<int16_t> input_def_levels(1, 1);
+
+  auto do_check = [&](const std::vector<int16_t>& input_def_levels, int num_values) {
+    std::vector<bool> values(num_values, false);
     NodePtr type = schema::Boolean("a", Repetition::OPTIONAL);
     const ColumnDescriptor descr(type, max_def_level_, max_rep_level_);
 
     // The data page falls back to plain encoding
     std::shared_ptr<ResizableBuffer> dummy = AllocateBuffer();
     std::shared_ptr<DataPageV1> data_page = MakeDataPage<BooleanType>(
-        &descr, values, /*num_values=*/3, Encoding::PLAIN, /*indices=*/{},
+        &descr, values, /*num_values=*/num_values, Encoding::PLAIN, /*indices=*/{},
         /*indices_size=*/0, /*def_levels=*/input_def_levels, max_def_level_,
         /*rep_levels=*/{},
         /*max_rep_level=*/max_rep_level_);
@@ -456,7 +455,7 @@ TEST_F(TestPrimitiveReader, DefLevelNotExpected) {
     auto reader = static_cast<BoolReader*>(reader_.get());
     ASSERT_TRUE(reader->HasNext());
 
-    constexpr int batch_size = 3;
+    constexpr int batch_size = 10;
     std::vector<int16_t> def_levels(batch_size, 0);
     std::vector<int16_t> rep_levels(batch_size, 0);
     bool values_out[batch_size];
@@ -467,42 +466,20 @@ TEST_F(TestPrimitiveReader, DefLevelNotExpected) {
                             &values_read);
         },
         ParquetException,
-        ::testing::Property(&ParquetException::what,
-                            ::testing::HasSubstr("Number of decoded rep / def levels did "
-                                                 "less than num_values in page_header")));
+        ::testing::Property(
+            &ParquetException::what,
+            ::testing::HasSubstr("Number of decoded rep / def levels did "
+                                 "more or less than num_values in page_header")));
+  };
+  // storing def-levels less than value in page-header
+  {
+    std::vector<int16_t> input_def_levels(1, 1);
+    do_check(input_def_levels, /*num_values=*/3);
   }
-  //  storing def-levels more than value in page-header
+  // storing def-levels more than value in page-header
   {
     std::vector<int16_t> input_def_levels(2, 1);
-    NodePtr type = schema::Boolean("a", Repetition::OPTIONAL);
-    const ColumnDescriptor descr(type, max_def_level_, max_rep_level_);
-
-    // The data page falls back to plain encoding
-    std::shared_ptr<ResizableBuffer> dummy = AllocateBuffer();
-    std::shared_ptr<DataPageV1> data_page = MakeDataPage<BooleanType>(
-        &descr, values, /*num_values=*/1, Encoding::PLAIN, /*indices=*/{},
-        /*indices_size=*/0, /*def_levels=*/input_def_levels, max_def_level_,
-        /*rep_levels=*/{},
-        /*max_rep_level=*/max_rep_level_);
-    pages_.push_back(data_page);
-    InitReader(&descr);
-    auto reader = static_cast<BoolReader*>(reader_.get());
-    ASSERT_TRUE(reader->HasNext());
-
-    constexpr int batch_size = 3;
-    std::vector<int16_t> def_levels(batch_size, 0);
-    std::vector<int16_t> rep_levels(batch_size, 0);
-    bool values_out[batch_size];
-    int64_t values_read;
-    EXPECT_THROW_THAT(
-        [&]() {
-          reader->ReadBatch(batch_size, def_levels.data(), rep_levels.data(), values_out,
-                            &values_read);
-        },
-        ParquetException,
-        ::testing::Property(&ParquetException::what,
-                            ::testing::HasSubstr("Number of decoded rep / def levels did "
-                                                 "less than num_values in page_header")));
+    do_check(input_def_levels, /*num_values=*/1);
   }
 }
 
diff --git a/cpp/submodules/parquet-testing b/cpp/submodules/parquet-testing
index 74278bc4a1122..1ba34478f535c 160000
--- a/cpp/submodules/parquet-testing
+++ b/cpp/submodules/parquet-testing
@@ -1 +1 @@
-Subproject commit 74278bc4a1122d74945969e6dec405abd1533ec3
+Subproject commit 1ba34478f535c89382263c42c675a9af4f57f2dd