YQ-3892 fix data race in json filters (#11827)

Author: GrigoriyPA

ydb/core/fq/libs/row_dispatcher/common.cpp

@@ -21,6 +21,10 @@ class TPureCalcProgramFactory : public IPureCalcProgramFactory {
         return it->second;
     }
 
+    TGuard<TMutex> LockFactory() const override {
+        return Guard(FactoryMutex);
+    }
+
 private:
     void CreateFactory(const TSettings& settings) {
         ProgramFactories.insert({settings, NYql::NPureCalc::MakeProgramFactory(
@@ -31,6 +35,7 @@ class TPureCalcProgramFactory : public IPureCalcProgramFactory {
 
 private:
     std::map<TSettings, NYql::NPureCalc::IProgramFactoryPtr> ProgramFactories;
+    TMutex FactoryMutex;
 };
 
 } // anonymous namespace

ydb/core/fq/libs/row_dispatcher/common.h

@@ -1,6 +1,7 @@
 #pragma once
 
 #include <util/generic/ptr.h>
+#include <util/system/mutex.h>
 
 #include <ydb/library/yql/public/purecalc/common/fwd.h>
 
@@ -18,6 +19,9 @@ class IPureCalcProgramFactory : public TThrRefBase {
 
 public:
     virtual NYql::NPureCalc::IProgramFactoryPtr GetFactory(const TSettings& settings) const = 0;
+
+    // Before creating purecalc program factory should be locked
+    virtual TGuard<TMutex> LockFactory() const = 0;
 };
 
 IPureCalcProgramFactory::TPtr CreatePureCalcProgramFactory();

ydb/core/fq/libs/row_dispatcher/json_filter.cpp

@@ -279,9 +279,13 @@ class TJsonFilter::TImpl {
         TCallback callback,
         IPureCalcProgramFactory::TPtr pureCalcProgramFactory,
         const IPureCalcProgramFactory::TSettings& factorySettings)
-        : Sql(GenerateSql(whereFilter, factorySettings)) {
+        : PureCalcProgramFactory(pureCalcProgramFactory)
+        , Sql(GenerateSql(whereFilter, factorySettings)) {
         Y_ENSURE(columns.size() == types.size(), "Number of columns and types should by equal");
 
+        // Shared factory may change during compilation, so it should be locked
+        auto guard = pureCalcProgramFactory->LockFactory();
+
         // Program should be stateless because input values
         // allocated on another allocator and should be released
         LOG_ROW_DISPATCHER_DEBUG("Creating program...");
@@ -304,6 +308,12 @@ class TJsonFilter::TImpl {
         return Sql;
     }
 
+    ~TImpl() {
+        auto guard = PureCalcProgramFactory->LockFactory();
+        InputConsumer.Reset();
+        Program.Reset();
+    }
+
 private:
     TString GenerateSql(const TString& whereFilter, const IPureCalcProgramFactory::TSettings& factorySettings) {
         TStringStream str;
@@ -317,6 +327,7 @@ class TJsonFilter::TImpl {
     }
 
 private:
+    const IPureCalcProgramFactory::TPtr PureCalcProgramFactory;
     THolder<NYql::NPureCalc::TPushStreamProgram<TFilterInputSpec, TFilterOutputSpec>> Program;
     THolder<NYql::NPureCalc::IConsumer<TInputType>> InputConsumer;
     const TString Sql;