Disable external compression for parquet

5 files changed, 27 insertions, 11 deletions

diff --git a/ydb/library/yql/providers/common/provider/yql_provider.cpp b/ydb/library/yql/providers/common/provider/yql_provider.cpp
index 809dfe35cbf..f13a717f2d1 100644
--- a/ydb/library/yql/providers/common/provider/yql_provider.cpp
+++ b/ydb/library/yql/providers/common/provider/yql_provider.cpp
@@ -1144,8 +1144,15 @@ void WriteStatistics(NYson::TYsonWriter& writer, bool totalOnly, const THashMap<
     writer.OnEndMap();
 }
 
-bool ValidateCompressionForInput(std::string_view compression, TExprContext& ctx) {
-    if (compression.empty() || IsIn(Compressions, compression)) {
+bool ValidateCompressionForInput(std::string_view format, std::string_view compression, TExprContext& ctx) {
+    if (compression.empty()) {
+        return true;
+    }
+    if (format == "parquet"sv) {
+        ctx.AddError(TIssue(TStringBuilder() << "External compression for parquet is not supported"));
+        return false;
+    }
+    if (IsIn(Compressions, compression)) {
         return true;
     }
     ctx.AddError(TIssue(TStringBuilder() << "Unknown compression: " << compression
@@ -1153,8 +1160,15 @@ bool ValidateCompressionForInput(std::string_view compression, TExprContext& ctx
     return false;
 }
 
-bool ValidateCompressionForOutput(std::string_view compression, TExprContext& ctx) {
-    if (compression.empty() || IsIn(Compressions, compression)) {
+bool ValidateCompressionForOutput(std::string_view format, std::string_view compression, TExprContext& ctx) {
+    if (compression.empty()) {
+        return true;
+    }
+    if (format == "parquet"sv) {
+        ctx.AddError(TIssue(TStringBuilder() << "External compression for parquet is not supported"));
+        return false;
+    }
+    if (IsIn(Compressions, compression)) {
         return true;
     }
     ctx.AddError(TIssue(TStringBuilder() << "Unknown compression: " << compression
diff --git a/ydb/library/yql/providers/common/provider/yql_provider.h b/ydb/library/yql/providers/common/provider/yql_provider.h
index b7541edaed0..1ee50a6b3d2 100644
--- a/ydb/library/yql/providers/common/provider/yql_provider.h
+++ b/ydb/library/yql/providers/common/provider/yql_provider.h
@@ -136,8 +136,8 @@ double GetDataReplicationFactor(const TExprNode& lambda, TExprContext& ctx);
 
 void WriteStatistics(NYson::TYsonWriter& writer, bool totalOnly, const THashMap<ui32, TOperationStatistics>& statistics);
 
-bool ValidateCompressionForInput(std::string_view compression, TExprContext& ctx);
-bool ValidateCompressionForOutput(std::string_view compression, TExprContext& ctx);
+bool ValidateCompressionForInput(std::string_view format, std::string_view compression, TExprContext& ctx);
+bool ValidateCompressionForOutput(std::string_view format, std::string_view compression, TExprContext& ctx);
 
 bool ValidateFormatForInput(std::string_view format, TExprContext& ctx);
 bool ValidateFormatForOutput(std::string_view format, TExprContext& ctx);
diff --git a/ydb/library/yql/providers/pq/provider/yql_pq_datasource_type_ann.cpp b/ydb/library/yql/providers/pq/provider/yql_pq_datasource_type_ann.cpp
index 2680ddb56b4..f3a1051802c 100644
--- a/ydb/library/yql/providers/pq/provider/yql_pq_datasource_type_ann.cpp
+++ b/ydb/library/yql/providers/pq/provider/yql_pq_datasource_type_ann.cpp
@@ -99,11 +99,12 @@ public:
             return TStatus::Error;
         }
 
-        if (!NCommon::ValidateFormatForInput(read.Format().Ref().Content(), ctx)) {
+        auto format = read.Format().Ref().Content();
+        if (!NCommon::ValidateFormatForInput(format, ctx)) {
             return TStatus::Error;
         }
 
-        if (!NCommon::ValidateCompressionForInput(read.Compression().Ref().Content(), ctx)) {
+        if (!NCommon::ValidateCompressionForInput(format, read.Compression().Ref().Content(), ctx)) {
             return TStatus::Error;
         }
 
diff --git a/ydb/library/yql/providers/s3/provider/yql_s3_datasink_type_ann.cpp b/ydb/library/yql/providers/s3/provider/yql_s3_datasink_type_ann.cpp
index cfc5293dac5..5ad6565cf34 100644
--- a/ydb/library/yql/providers/s3/provider/yql_s3_datasink_type_ann.cpp
+++ b/ydb/library/yql/providers/s3/provider/yql_s3_datasink_type_ann.cpp
@@ -107,7 +107,8 @@ private:
             return TStatus::Repeat;
         }
 
-        if (!EnsureAtom(*input->Child(TS3Target::idx_Format), ctx) || !NCommon::ValidateFormatForOutput(input->Child(TS3Target::idx_Format)->Content(), ctx)) {
+        const auto format = input->Child(TS3Object::idx_Format)->Content();
+        if (!EnsureAtom(*input->Child(TS3Target::idx_Format), ctx) || !NCommon::ValidateFormatForOutput(format, ctx)) {
             return TStatus::Error;
         }
 
@@ -126,7 +127,7 @@ private:
                         return false;
                     }
 
-                    return NCommon::ValidateCompressionForOutput(value.Content(), ctx);
+                    return NCommon::ValidateCompressionForOutput(format, value.Content(), ctx);
                 }
 
                 if (name == "partitionedby") {
diff --git a/ydb/library/yql/providers/s3/provider/yql_s3_datasource_type_ann.cpp b/ydb/library/yql/providers/s3/provider/yql_s3_datasource_type_ann.cpp
index 908aebd3bd7..c638bb4bf6a 100644
--- a/ydb/library/yql/providers/s3/provider/yql_s3_datasource_type_ann.cpp
+++ b/ydb/library/yql/providers/s3/provider/yql_s3_datasource_type_ann.cpp
@@ -505,7 +505,7 @@ public:
                     if (!ExtractSettingValue(setting.Tail(), "compression"sv, format, {}, ctx, compression)) {
                         return false;
                     }
-                    return NCommon::ValidateCompressionForInput(compression, ctx);
+                    return NCommon::ValidateCompressionForInput(format, compression, ctx);
                 }
 
                 if (name == "partitionedby"sv) {