From 59d474139ccbde18df7ea1c46efae833067156ce Mon Sep 17 00:00:00 2001
From: Bikramjeet Vig <bikramjeet@fb.com>
Date: Tue, 10 Dec 2024 22:29:22 -0800
Subject: [PATCH] fix(fuzzer): Ensure common dictionary generation skips
 row_number column (#11821)

Summary:
Ensure that the row_number column, used for lining up results with the
reference database, is not treated as a candidate for wrapping by the
fuzzer's feature that probabilistically adds a common dictionary layer.
This prevents verification mismatches caused by the column being
wrapped and losing its intended purpose.

Pull Request resolved: https://github.com/facebookincubator/velox/pull/11821

Reviewed By: kagamiori

Differential Revision: D67071154

Pulled By: bikramSingh91

fbshipit-source-id: 8051d264873fb74bc27064a64027ec72448131ad
---
 .../fuzzer/ExpressionFuzzerVerifier.cpp          | 16 +++++++---------
 .../expression/fuzzer/ExpressionFuzzerVerifier.h |  8 ++++----
 2 files changed, 11 insertions(+), 13 deletions(-)

diff --git a/velox/expression/fuzzer/ExpressionFuzzerVerifier.cpp b/velox/expression/fuzzer/ExpressionFuzzerVerifier.cpp
index ddae30f11f0e..1ede9f9cccb9 100644
--- a/velox/expression/fuzzer/ExpressionFuzzerVerifier.cpp
+++ b/velox/expression/fuzzer/ExpressionFuzzerVerifier.cpp
@@ -321,17 +321,15 @@ void ExpressionFuzzerVerifier::retryWithTry(
   }
 }
 
-RowVectorPtr ExpressionFuzzerVerifier::fuzzInputWithRowNumber(
-    VectorFuzzer& fuzzer,
-    const RowTypePtr& type) {
-  auto rowVector = fuzzer.fuzzInputRow(type);
-  auto names = type->names();
+RowVectorPtr ExpressionFuzzerVerifier::appendRowNumberColumn(
+    RowVectorPtr& inputRow) {
+  auto names = asRowType(inputRow->type())->names();
   names.push_back("row_number");
 
-  auto& children = rowVector->children();
+  auto& children = inputRow->children();
   velox::test::VectorMaker vectorMaker{pool_.get()};
   children.push_back(vectorMaker.flatVector<int64_t>(
-      rowVector->size(), [&](auto row) { return row; }));
+      inputRow->size(), [&](auto row) { return row; }));
 
   return vectorMaker.rowVector(names, children);
 }
@@ -375,10 +373,10 @@ void ExpressionFuzzerVerifier::go() {
 
     std::vector<core::TypedExprPtr> plans = std::move(expressions);
 
-    auto rowVector = fuzzInputWithRowNumber(*vectorFuzzer_, inputType);
-
+    auto rowVector = vectorFuzzer_->fuzzInputRow(inputType);
     InputRowMetadata inputRowMetadata =
         generateInputRowMetadata(rowVector, *vectorFuzzer_);
+    rowVector = appendRowNumberColumn(rowVector);
 
     auto resultVectors = generateResultVectors(plans);
     ResultOrError result;
diff --git a/velox/expression/fuzzer/ExpressionFuzzerVerifier.h b/velox/expression/fuzzer/ExpressionFuzzerVerifier.h
index c03de0de2b3d..aaf6d7d5cdce 100644
--- a/velox/expression/fuzzer/ExpressionFuzzerVerifier.h
+++ b/velox/expression/fuzzer/ExpressionFuzzerVerifier.h
@@ -201,10 +201,10 @@ class ExpressionFuzzerVerifier {
       const RowVectorPtr& rowVector,
       VectorFuzzer& vectorFuzzer);
 
-  // Fuzzes the input vector of type with an additional row number column.
-  RowVectorPtr fuzzInputWithRowNumber(
-      VectorFuzzer& fuzzer,
-      const RowTypePtr& type);
+  // Appends an additional row number column called 'row_number' at the end of
+  // the 'inputRow'. This column is then used to line up rows when comparing
+  // results against a reference database.
+  RowVectorPtr appendRowNumberColumn(RowVectorPtr& inputRow);
 
   const Options options_;