Lighten S3 tasks params and source desc params

Rename to deprecated

Improve test

Read paths from tree

Store file tree in range params

18 files changed, 610 insertions, 34 deletions

diff --git a/CMakeLists.darwin.txt b/CMakeLists.darwin.txt
index ec4fdd1cade..e9df897de34 100644
--- a/CMakeLists.darwin.txt
+++ b/CMakeLists.darwin.txt
@@ -778,6 +778,7 @@ add_subdirectory(ydb/library/yql/providers/pq/common)
 add_subdirectory(ydb/library/yql/providers/pq/expr_nodes)
 add_subdirectory(ydb/library/yql/providers/s3/provider)
 add_subdirectory(ydb/library/yql/providers/s3/expr_nodes)
+add_subdirectory(ydb/library/yql/providers/s3/range_helpers)
 add_subdirectory(ydb/library/yql/providers/ydb/provider)
 add_subdirectory(ydb/public/lib/experimental)
 add_subdirectory(ydb/library/yql/providers/ydb/expr_nodes)
@@ -1353,6 +1354,7 @@ add_subdirectory(ydb/library/yql/providers/common/http_gateway/mock)
 add_subdirectory(ydb/library/yql/providers/common/structured_token/ut)
 add_subdirectory(ydb/library/yql/providers/pq/gateway/dummy)
 add_subdirectory(ydb/library/yql/providers/s3/path_generator/ut)
+add_subdirectory(ydb/library/yql/providers/s3/range_helpers/ut)
 add_subdirectory(ydb/library/yql/udfs/common/stat/ut)
 add_subdirectory(ydb/library/yql/udfs/common/topfreq/ut)
 add_subdirectory(ydb/public/sdk/cpp/client/extensions/discovery_mutator/ut)
diff --git a/CMakeLists.linux.txt b/CMakeLists.linux.txt
index 80027f208a4..872b06e43ed 100644
--- a/CMakeLists.linux.txt
+++ b/CMakeLists.linux.txt
@@ -782,6 +782,7 @@ add_subdirectory(ydb/library/yql/providers/pq/common)
 add_subdirectory(ydb/library/yql/providers/pq/expr_nodes)
 add_subdirectory(ydb/library/yql/providers/s3/provider)
 add_subdirectory(ydb/library/yql/providers/s3/expr_nodes)
+add_subdirectory(ydb/library/yql/providers/s3/range_helpers)
 add_subdirectory(ydb/library/yql/providers/ydb/provider)
 add_subdirectory(ydb/public/lib/experimental)
 add_subdirectory(ydb/library/yql/providers/ydb/expr_nodes)
@@ -1374,6 +1375,7 @@ add_subdirectory(ydb/library/yql/providers/common/http_gateway/mock)
 add_subdirectory(ydb/library/yql/providers/common/structured_token/ut)
 add_subdirectory(ydb/library/yql/providers/pq/gateway/dummy)
 add_subdirectory(ydb/library/yql/providers/s3/path_generator/ut)
+add_subdirectory(ydb/library/yql/providers/s3/range_helpers/ut)
 add_subdirectory(ydb/library/yql/udfs/common/stat/ut)
 add_subdirectory(ydb/library/yql/udfs/common/topfreq/ut)
 add_subdirectory(ydb/public/sdk/cpp/client/extensions/discovery_mutator/ut)
diff --git a/ydb/library/yql/providers/s3/actors/CMakeLists.linux.txt b/ydb/library/yql/providers/s3/actors/CMakeLists.linux.txt
index 73edb9f3464..e828c67edbf 100644
--- a/ydb/library/yql/providers/s3/actors/CMakeLists.linux.txt
+++ b/ydb/library/yql/providers/s3/actors/CMakeLists.linux.txt
@@ -29,6 +29,7 @@ target_link_libraries(providers-s3-actors PUBLIC
   common-token_accessor-client
   providers-s3-compressors
   providers-s3-proto
+  providers-s3-range_helpers
   providers-s3-serializations
   yql-public-types
   clickhouse_client_udf
diff --git a/ydb/library/yql/providers/s3/actors/yql_s3_read_actor.cpp b/ydb/library/yql/providers/s3/actors/yql_s3_read_actor.cpp
index 1d0cc239cdb..67ef5f0a8c3 100644
--- a/ydb/library/yql/providers/s3/actors/yql_s3_read_actor.cpp
+++ b/ydb/library/yql/providers/s3/actors/yql_s3_read_actor.cpp
@@ -33,8 +33,8 @@
 
 #include <ydb/library/yql/providers/s3/compressors/factory.h>
 #include <ydb/library/yql/providers/s3/proto/range.pb.h>
+#include <ydb/library/yql/providers/s3/range_helpers/path_list_reader.h>
 #include <ydb/library/yql/providers/s3/serializations/serialization_interval.h>
-#include <ydb/library/yql/providers/common/provider/yql_provider_names.h>
 
 #include <library/cpp/actors/core/actor_bootstrapped.h>
 #include <library/cpp/actors/core/actor_coroutine.h>
@@ -48,7 +48,8 @@
 
 namespace NYql::NDq {
 
-using namespace NActors;
+using namespace ::NActors;
+using namespace ::NYql::NS3Details;
 
 namespace {
 
@@ -105,9 +106,6 @@ struct TEvPrivate {
     };
 };
 
-using TPath = std::tuple<TString, size_t>;
-using TPathList = std::vector<TPath>;
-
 class TRetryParams {
 public:
     TRetryParams(const std::shared_ptr<NS3::TRetryConfig>& retryConfig)
@@ -673,28 +671,10 @@ std::pair<NYql::NDq::IDqComputeActorAsyncInput*, IActor*> CreateS3ReadActor(
     const std::shared_ptr<NS3::TRetryConfig>& retryConfig)
 {
     const IFunctionRegistry& functionRegistry = *holderFactory.GetFunctionRegistry();
-    std::unordered_map<TString, size_t> map(params.GetPath().size());
-    for (auto i = 0; i < params.GetPath().size(); ++i)
-        map.emplace(params.GetPath().Get(i).GetPath(), params.GetPath().Get(i).GetSize());
 
     TPathList paths;
     ui64 startPathIndex = 0;
-    if (const auto taskParamsIt = taskParams.find(S3ProviderName); taskParamsIt != taskParams.cend()) {
-        NS3::TRange range;
-        TStringInput input(taskParamsIt->second);
-        range.Load(&input);
-        startPathIndex = range.GetStartPathIndex();
-        for (auto i = 0; i < range.GetPath().size(); ++i) {
-            const auto& path = range.GetPath().Get(i);
-            auto it = map.find(path);
-            YQL_ENSURE(it != map.end());
-            paths.emplace_back(path, it->second);
-        }
-    } else {
-        for (auto i = 0; i < params.GetPath().size(); ++i) {
-            paths.emplace_back(params.GetPath().Get(i).GetPath(), params.GetPath().Get(i).GetSize());
-        }
-    }
+    ReadPathsList(params, taskParams, paths, startPathIndex);
 
     const auto token = secureParams.Value(params.GetToken(), TString{});
     const auto credentialsProviderFactory = CreateCredentialsProviderFactoryForStructuredToken(credentialsFactory, token);
diff --git a/ydb/library/yql/providers/s3/proto/range.proto b/ydb/library/yql/providers/s3/proto/range.proto
index ce018fbe9a4..975c0d559ec 100644
--- a/ydb/library/yql/providers/s3/proto/range.proto
+++ b/ydb/library/yql/providers/s3/proto/range.proto
@@ -4,6 +4,14 @@ option cc_enable_arenas = true;
 package NYql.NS3;
 
 message TRange {
-    repeated string Path = 2;
+    repeated string DeprecatedPath = 2; // deprecated
     uint64 StartPathIndex = 3;
+    repeated TPath Paths = 4;
+
+    message TPath {
+        string Name = 1;
+        repeated TPath Children = 2;
+        uint64 Size = 3;
+        bool Read = 4; // Read this path
+    }
 }
diff --git a/ydb/library/yql/providers/s3/proto/source.proto b/ydb/library/yql/providers/s3/proto/source.proto
index 857ef69262a..232abb19cd1 100644
--- a/ydb/library/yql/providers/s3/proto/source.proto
+++ b/ydb/library/yql/providers/s3/proto/source.proto
@@ -11,7 +11,7 @@ message TPath {
 message TSource {
     string Url = 1;
     string Token = 2;
-    repeated TPath Path = 3;
+    repeated TPath DeprecatedPath = 3; // deprecated
     optional string RowType = 4;
     optional string Format = 5;
     map<string, string> Settings = 6;
diff --git a/ydb/library/yql/providers/s3/provider/CMakeLists.txt b/ydb/library/yql/providers/s3/provider/CMakeLists.txt
index 5bbc6501ecd..62694c69ca3 100644
--- a/ydb/library/yql/providers/s3/provider/CMakeLists.txt
+++ b/ydb/library/yql/providers/s3/provider/CMakeLists.txt
@@ -43,6 +43,7 @@ target_link_libraries(providers-s3-provider PUBLIC
   providers-result-expr_nodes
   providers-s3-expr_nodes
   providers-s3-proto
+  providers-s3-range_helpers
   library-yql-utils
 )
 target_sources(providers-s3-provider PRIVATE
diff --git a/ydb/library/yql/providers/s3/provider/yql_s3_dq_integration.cpp b/ydb/library/yql/providers/s3/provider/yql_s3_dq_integration.cpp
index 5d0b9c45a1c..2f3b05faa6f 100644
--- a/ydb/library/yql/providers/s3/provider/yql_s3_dq_integration.cpp
+++ b/ydb/library/yql/providers/s3/provider/yql_s3_dq_integration.cpp
@@ -10,6 +10,7 @@
 #include <ydb/library/yql/providers/s3/proto/range.pb.h>
 #include <ydb/library/yql/providers/s3/proto/sink.pb.h>
 #include <ydb/library/yql/providers/s3/proto/source.pb.h>
+#include <ydb/library/yql/providers/s3/range_helpers/file_tree_builder.h>
 #include <ydb/library/yql/utils/log/log.h>
 
 namespace NYql {
@@ -18,6 +19,8 @@ using namespace NNodes;
 
 namespace {
 
+using namespace NYql::NS3Details;
+
 class TS3DqIntegration: public TDqIntegrationBase {
 public:
     TS3DqIntegration(TS3State::TPtr state)
@@ -27,13 +30,16 @@ public:
 
     ui64 Partition(const TDqSettings&, size_t maxPartitions, const TExprNode& node, TVector<TString>& partitions, TString*, TExprContext&, bool) override {
         TString cluster;
-        std::vector<std::vector<TString>> parts;
+        std::vector<std::vector<std::pair<TString, ui64>>> parts;
         if (const TMaybeNode<TDqSource> source = &node) {
             cluster = source.Cast().DataSource().Cast<TS3DataSource>().Cluster().Value();
             const auto settings = source.Cast().Settings().Cast<TS3SourceSettingsBase>();
             parts.reserve(settings.Paths().Size());
             for (auto i = 0u; i < settings.Paths().Size(); ++i)
-                parts.emplace_back(std::vector<TString>(1U, settings.Paths().Item(i).Path().StringValue()));
+                parts.emplace_back(1U,
+                    std::pair(
+                        settings.Paths().Item(i).Path().StringValue(),
+                        FromString<ui64>(settings.Paths().Item(i).Size().Value())));
         }
 
         if (maxPartitions && parts.size() > maxPartitions) {
@@ -62,7 +68,9 @@ public:
         for (const auto& part : parts) {
             NS3::TRange range;
             range.SetStartPathIndex(startIdx);
-            std::for_each(part.cbegin(), part.cend(), [&range, &startIdx](const TString& path) { range.AddPath(path); ++startIdx; });
+            TFileTreeBuilder builder;
+            std::for_each(part.cbegin(), part.cend(), [&builder, &startIdx](const std::pair<TString, ui64>& f) { builder.AddPath(f.first, f.second); ++startIdx; });
+            builder.Save(&range);
 
             partitions.emplace_back();
             TStringOutput out(partitions.back());
@@ -170,11 +178,6 @@ public:
             const auto& paths = settings.Paths();
             YQL_ENSURE(paths.Size() > 0);
             const TStructExprType* extraColumnsType = paths.Item(0).ExtraColumns().Ref().GetTypeAnn()->Cast<TStructExprType>();
-            for (auto i = 0U; i < paths.Size(); ++i) {
-                const auto p = srcDesc.AddPath();
-                p->SetPath(paths.Item(i).Path().StringValue());
-                p->SetSize(FromString<ui64>(paths.Item(i).Size().Value()));
-            }
 
             if (const auto mayParseSettings = settings.Maybe<TS3ParseSettings>()) {
                 const auto parseSettings = mayParseSettings.Cast();
diff --git a/ydb/library/yql/providers/s3/range_helpers/CMakeLists.txt b/ydb/library/yql/providers/s3/range_helpers/CMakeLists.txt
new file mode 100644
index 00000000000..7e22842e685
--- /dev/null
+++ b/ydb/library/yql/providers/s3/range_helpers/CMakeLists.txt
@@ -0,0 +1,24 @@
+
+# This file was gererated by the build system used internally in the Yandex monorepo.
+# Only simple modifications are allowed (adding source-files to targets, adding simple properties
+# like target_include_directories). These modifications will be ported to original
+# ya.make files by maintainers. Any complex modifications which can't be ported back to the
+# original buildsystem will not be accepted.
+
+
+
+add_library(providers-s3-range_helpers)
+target_compile_options(providers-s3-range_helpers PRIVATE
+  -DUSE_CURRENT_UDF_ABI_VERSION
+)
+target_link_libraries(providers-s3-range_helpers PUBLIC
+  contrib-libs-cxxsupp
+  yutil
+  providers-common-provider
+  providers-s3-proto
+  library-yql-utils
+)
+target_sources(providers-s3-range_helpers PRIVATE
+  ${CMAKE_SOURCE_DIR}/ydb/library/yql/providers/s3/range_helpers/file_tree_builder.cpp
+  ${CMAKE_SOURCE_DIR}/ydb/library/yql/providers/s3/range_helpers/path_list_reader.cpp
+)
diff --git a/ydb/library/yql/providers/s3/range_helpers/file_tree_builder.cpp b/ydb/library/yql/providers/s3/range_helpers/file_tree_builder.cpp
new file mode 100644
index 00000000000..166e097b984
--- /dev/null
+++ b/ydb/library/yql/providers/s3/range_helpers/file_tree_builder.cpp
@@ -0,0 +1,41 @@
+#include "file_tree_builder.h"
+
+namespace NYql::NS3Details {
+
+void TFileTreeBuilder::AddPath(const TString& path, ui64 fileSize) {
+    const auto parts = SplitPath(path);
+    std::map<TString, TPath>* currentChildren = &Roots;
+    for (size_t i = 0, size = parts.size(); i < size; ++i) {
+        TPath& p = (*currentChildren)[parts[i]];
+        if (i == size - 1) { // last
+            Y_VERIFY(p.FileSize == 0);
+            Y_VERIFY(!p.Read);
+            p.FileSize = fileSize;
+            p.Read = true;
+        } else {
+            currentChildren = &p.Children;
+        }
+    }
+}
+
+void TFileTreeBuilder::Save(NS3::TRange* range) const {
+    for (const auto& [n, p] : Roots) {
+        SaveImpl(range->AddPaths(), n, p);
+    }
+}
+
+void TFileTreeBuilder::SaveImpl(NS3::TRange::TPath* path, const TString& name, const TPath& srcPath) const {
+    path->SetName(name);
+    path->SetSize(srcPath.FileSize);
+    path->SetRead(srcPath.Read);
+    for (const auto& [n, p] : srcPath.Children) {
+        SaveImpl(path->AddChildren(), n, p);
+    }
+}
+
+std::vector<TString> TFileTreeBuilder::SplitPath(const TString& path) {
+    std::vector<TString> parts = StringSplitter(path).Split('/');
+    return parts;
+}
+
+} // namespace NYql::NS3Details
diff --git a/ydb/library/yql/providers/s3/range_helpers/file_tree_builder.h b/ydb/library/yql/providers/s3/range_helpers/file_tree_builder.h
new file mode 100644
index 00000000000..363c60f78bd
--- /dev/null
+++ b/ydb/library/yql/providers/s3/range_helpers/file_tree_builder.h
@@ -0,0 +1,30 @@
+#pragma once
+#include <ydb/library/yql/providers/s3/proto/range.pb.h>
+
+#include <util/generic/string.h>
+#include <util/string/split.h>
+
+#include <map>
+
+namespace NYql::NS3Details {
+
+class TFileTreeBuilder {
+    struct TPath {
+        ui64 FileSize = 0;
+        bool Read = false;
+        std::map<TString, TPath> Children;
+    };
+
+public:
+    void AddPath(const TString& path, ui64 fileSize);
+    void Save(NS3::TRange* range) const;
+
+private:
+    void SaveImpl(NS3::TRange::TPath* path, const TString& name, const TPath& srcPath) const;
+    static std::vector<TString> SplitPath(const TString& path);
+
+private:
+    std::map<TString, TPath> Roots;
+};
+
+} // namespace NYql::NS3Details
diff --git a/ydb/library/yql/providers/s3/range_helpers/file_tree_builder_ut.cpp b/ydb/library/yql/providers/s3/range_helpers/file_tree_builder_ut.cpp
new file mode 100644
index 00000000000..8644c9acc01
--- /dev/null
+++ b/ydb/library/yql/providers/s3/range_helpers/file_tree_builder_ut.cpp
@@ -0,0 +1,163 @@
+#include "file_tree_builder.h"
+#include "path_list_reader.h"
+
+#include <ydb/library/yql/providers/common/provider/yql_provider_names.h>
+
+#include <library/cpp/testing/unittest/registar.h>
+
+namespace NYql::NS3Details {
+
+Y_UNIT_TEST_SUITE(S3FileTreeBuilderTest) {
+    THashMap<TString, TString> MakeParams(const NS3::TRange& range) {
+        TStringBuilder str;
+        range.Save(&str.Out);
+        THashMap<TString, TString> map;
+        map[S3ProviderName] = str;
+        return map;
+    }
+
+    Y_UNIT_TEST(Simple) {
+        TFileTreeBuilder b;
+        b.AddPath("name", 42);
+
+        NS3::TRange range;
+        b.Save(&range);
+
+        UNIT_ASSERT_VALUES_EQUAL(range.PathsSize(), 1);
+        UNIT_ASSERT_VALUES_EQUAL(range.GetPaths(0).GetName(), "name");
+        UNIT_ASSERT_VALUES_EQUAL(range.GetPaths(0).GetSize(), 42);
+        UNIT_ASSERT_VALUES_EQUAL(range.GetPaths(0).ChildrenSize(), 0);
+    }
+
+    Y_UNIT_TEST(Interesting) {
+        TFileTreeBuilder b;
+        b.AddPath("name", 42);
+        b.AddPath("root/folder/file", 100500);
+        b.AddPath("root2/file", 10);
+        b.AddPath("root2", 42);
+        b.AddPath("root/folder/other_file", 22);
+        b.AddPath("root/file/", 12);
+
+        NS3::TRange range;
+        b.Save(&range);
+
+        UNIT_ASSERT_VALUES_EQUAL(range.PathsSize(), 3);
+
+        // name
+        UNIT_ASSERT_VALUES_EQUAL(range.GetPaths(0).GetName(), "name");
+        UNIT_ASSERT_VALUES_EQUAL(range.GetPaths(0).GetSize(), 42);
+        UNIT_ASSERT_VALUES_EQUAL(range.GetPaths(0).ChildrenSize(), 0);
+
+        // root
+        UNIT_ASSERT_VALUES_EQUAL(range.GetPaths(1).GetName(), "root");
+        UNIT_ASSERT_VALUES_EQUAL(range.GetPaths(1).GetSize(), 0);
+        UNIT_ASSERT_VALUES_EQUAL(range.GetPaths(1).ChildrenSize(), 2);
+
+        // root/file/
+        UNIT_ASSERT_VALUES_EQUAL(range.GetPaths(1).GetChildren(0).ChildrenSize(), 1);
+        UNIT_ASSERT_VALUES_EQUAL(range.GetPaths(1).GetChildren(0).GetName(), "file");
+        UNIT_ASSERT_VALUES_EQUAL(range.GetPaths(1).GetChildren(0).GetChildren(0).GetName(), "");
+        UNIT_ASSERT_VALUES_EQUAL(range.GetPaths(1).GetChildren(0).GetChildren(0).GetSize(), 12);
+        UNIT_ASSERT_VALUES_EQUAL(range.GetPaths(1).GetChildren(0).GetChildren(0).ChildrenSize(), 0);
+
+        // root/folder
+        UNIT_ASSERT_VALUES_EQUAL(range.GetPaths(1).GetChildren(1).GetName(), "folder");
+        UNIT_ASSERT_VALUES_EQUAL(range.GetPaths(1).GetChildren(1).GetSize(), 0);
+        UNIT_ASSERT_VALUES_EQUAL(range.GetPaths(1).GetChildren(1).ChildrenSize(), 2);
+
+        // root/folder/file
+        UNIT_ASSERT_VALUES_EQUAL(range.GetPaths(1).GetChildren(1).GetChildren(0).GetName(), "file");
+        UNIT_ASSERT_VALUES_EQUAL(range.GetPaths(1).GetChildren(1).GetChildren(0).GetSize(), 100500);
+        UNIT_ASSERT_VALUES_EQUAL(range.GetPaths(1).GetChildren(1).GetChildren(0).ChildrenSize(), 0);
+
+        // root/folder/other_file
+        UNIT_ASSERT_VALUES_EQUAL(range.GetPaths(1).GetChildren(1).GetChildren(1).GetName(), "other_file");
+        UNIT_ASSERT_VALUES_EQUAL(range.GetPaths(1).GetChildren(1).GetChildren(1).GetSize(), 22);
+        UNIT_ASSERT_VALUES_EQUAL(range.GetPaths(1).GetChildren(1).GetChildren(1).ChildrenSize(), 0);
+
+        // root2
+        UNIT_ASSERT_VALUES_EQUAL(range.GetPaths(2).GetName(), "root2");
+        UNIT_ASSERT_VALUES_EQUAL(range.GetPaths(2).GetSize(), 42);
+        UNIT_ASSERT_VALUES_EQUAL(range.GetPaths(2).ChildrenSize(), 1);
+
+        // root2/file
+        UNIT_ASSERT_VALUES_EQUAL(range.GetPaths(2).GetChildren(0).GetName(), "file");
+        UNIT_ASSERT_VALUES_EQUAL(range.GetPaths(2).GetChildren(0).GetSize(), 10);
+        UNIT_ASSERT_VALUES_EQUAL(range.GetPaths(2).GetChildren(0).ChildrenSize(), 0);
+    }
+
+    Y_UNIT_TEST(PassesFileWithZeroSize) {
+        TFileTreeBuilder b;
+        b.AddPath("name", 0);
+
+        NS3::TRange range;
+        b.Save(&range);
+
+        TPathList paths;
+        ui64 startPathIndex = 0;
+        ReadPathsList({}, MakeParams(range), paths, startPathIndex);
+
+        UNIT_ASSERT_VALUES_EQUAL(paths.size(), 1);
+        UNIT_ASSERT_VALUES_EQUAL(std::get<TString>(paths[0]), "name");
+        UNIT_ASSERT_VALUES_EQUAL(std::get<ui64>(paths[0]), 0);
+    }
+
+    Y_UNIT_TEST(DeserializesManySlashes) {
+        TFileTreeBuilder b;
+        b.AddPath("a///b", 42);
+
+        NS3::TRange range;
+        b.Save(&range);
+
+        TPathList paths;
+        ui64 startPathIndex = 0;
+        ReadPathsList({}, MakeParams(range), paths, startPathIndex);
+
+        UNIT_ASSERT_VALUES_EQUAL(paths.size(), 1);
+        UNIT_ASSERT_VALUES_EQUAL(std::get<TString>(paths[0]), "a///b");
+        UNIT_ASSERT_VALUES_EQUAL(std::get<ui64>(paths[0]), 42);
+    }
+
+    Y_UNIT_TEST(DeserializesTrailingSlash) {
+        TFileTreeBuilder b;
+        b.AddPath("root/name//", 3);
+        b.AddPath("root/name/", 0);
+
+        NS3::TRange range;
+        b.Save(&range);
+
+        TPathList paths;
+        ui64 startPathIndex = 0;
+        ReadPathsList({}, MakeParams(range), paths, startPathIndex);
+
+        UNIT_ASSERT_VALUES_EQUAL(paths.size(), 2);
+        UNIT_ASSERT_VALUES_EQUAL(std::get<TString>(paths[0]), "root/name/");
+        UNIT_ASSERT_VALUES_EQUAL(std::get<ui64>(paths[0]), 0);
+
+        UNIT_ASSERT_VALUES_EQUAL(std::get<TString>(paths[1]), "root/name//");
+        UNIT_ASSERT_VALUES_EQUAL(std::get<ui64>(paths[1]), 3);
+    }
+
+    Y_UNIT_TEST(DeserializesLeadingSlash) {
+        TFileTreeBuilder b;
+        b.AddPath("/root/name", 3);
+        b.AddPath("/", 42);
+
+        NS3::TRange range;
+        b.Save(&range);
+
+        TPathList paths;
+        ui64 startPathIndex = 0;
+        ReadPathsList({}, MakeParams(range), paths, startPathIndex);
+
+        UNIT_ASSERT_VALUES_EQUAL(paths.size(), 2);
+
+        UNIT_ASSERT_VALUES_EQUAL(std::get<TString>(paths[0]), "/");
+        UNIT_ASSERT_VALUES_EQUAL(std::get<ui64>(paths[0]), 42);
+
+        UNIT_ASSERT_VALUES_EQUAL(std::get<TString>(paths[1]), "/root/name");
+        UNIT_ASSERT_VALUES_EQUAL(std::get<ui64>(paths[1]), 3);
+    }
+}
+
+} // namespace NYql::NS3Details
diff --git a/ydb/library/yql/providers/s3/range_helpers/path_list_reader.cpp b/ydb/library/yql/providers/s3/range_helpers/path_list_reader.cpp
new file mode 100644
index 00000000000..4813dfa12ac
--- /dev/null
+++ b/ydb/library/yql/providers/s3/range_helpers/path_list_reader.cpp
@@ -0,0 +1,59 @@
+#include "path_list_reader.h"
+
+#include <ydb/library/yql/providers/common/provider/yql_provider_names.h>
+#include <ydb/library/yql/utils/yql_panic.h>
+
+#include <util/stream/str.h>
+
+namespace NYql::NS3Details {
+
+static void BuildPathsFromTree(const google::protobuf::RepeatedPtrField<NYql::NS3::TRange::TPath>& children, TPathList& paths, TString& currentPath, size_t currentDepth = 0) {
+    if (children.empty()) {
+        return;
+    }
+    if (currentDepth) {
+        currentPath += '/';
+    }
+    for (const auto& path : children) {
+        const size_t prevSize = currentPath.size();
+        currentPath += path.GetName();
+        if (path.GetRead()) {
+            paths.emplace_back(currentPath, path.GetSize());
+        }
+        BuildPathsFromTree(path.GetChildren(), paths, currentPath, currentDepth + 1);
+        currentPath.resize(prevSize);
+    }
+}
+
+void ReadPathsList(const NS3::TSource& sourceDesc, const THashMap<TString, TString>& taskParams, TPathList& paths, ui64& startPathIndex) {
+    if (const auto taskParamsIt = taskParams.find(S3ProviderName); taskParamsIt != taskParams.cend()) {
+        NS3::TRange range;
+        TStringInput input(taskParamsIt->second);
+        range.Load(&input);
+        startPathIndex = range.GetStartPathIndex();
+
+        // Modern way
+        if (range.PathsSize()) {
+            TString buf;
+            BuildPathsFromTree(range.GetPaths(), paths, buf);
+            return;
+        }
+
+        std::unordered_map<TString, size_t> map(sourceDesc.GetDeprecatedPath().size());
+        for (auto i = 0; i < sourceDesc.GetDeprecatedPath().size(); ++i)
+            map.emplace(sourceDesc.GetDeprecatedPath().Get(i).GetPath(), sourceDesc.GetDeprecatedPath().Get(i).GetSize());
+
+        for (auto i = 0; i < range.GetDeprecatedPath().size(); ++i) {
+            const auto& path = range.GetDeprecatedPath().Get(i);
+            auto it = map.find(path);
+            YQL_ENSURE(it != map.end());
+            paths.emplace_back(path, it->second);
+        }
+    } else {
+        for (auto i = 0; i < sourceDesc.GetDeprecatedPath().size(); ++i) {
+            paths.emplace_back(sourceDesc.GetDeprecatedPath().Get(i).GetPath(), sourceDesc.GetDeprecatedPath().Get(i).GetSize());
+        }
+    }
+}
+
+} // namespace NYql::NS3Details
diff --git a/ydb/library/yql/providers/s3/range_helpers/path_list_reader.h b/ydb/library/yql/providers/s3/range_helpers/path_list_reader.h
new file mode 100644
index 00000000000..9be8c7814ba
--- /dev/null
+++ b/ydb/library/yql/providers/s3/range_helpers/path_list_reader.h
@@ -0,0 +1,18 @@
+#pragma once
+#include <ydb/library/yql/providers/s3/proto/source.pb.h>
+#include <ydb/library/yql/providers/s3/proto/range.pb.h>
+
+#include <util/generic/hash.h>
+#include <util/generic/string.h>
+
+#include <utility>
+#include <vector>
+
+namespace NYql::NS3Details {
+
+using TPath = std::tuple<TString, size_t>;
+using TPathList = std::vector<TPath>;
+
+void ReadPathsList(const NS3::TSource& sourceDesc, const THashMap<TString, TString>& taskParams, TPathList& paths, ui64& startPathIndex);
+
+} // namespace NYql::NS3Details
diff --git a/ydb/library/yql/providers/s3/range_helpers/path_list_reader_ut.cpp b/ydb/library/yql/providers/s3/range_helpers/path_list_reader_ut.cpp
new file mode 100644
index 00000000000..a54135e6561
--- /dev/null
+++ b/ydb/library/yql/providers/s3/range_helpers/path_list_reader_ut.cpp
@@ -0,0 +1,133 @@
+#include "path_list_reader.h"
+
+#include <ydb/library/yql/providers/common/provider/yql_provider_names.h>
+
+#include <library/cpp/testing/unittest/registar.h>
+
+#include <util/string/builder.h>
+
+namespace NYql::NS3Details {
+
+Y_UNIT_TEST_SUITE(PathListReaderTest) {
+    THashMap<TString, TString> MakeParams(const NS3::TRange& range) {
+        TStringBuilder str;
+        range.Save(&str.Out);
+        THashMap<TString, TString> map;
+        map[S3ProviderName] = str;
+        return map;
+    }
+
+    Y_UNIT_TEST(ReadsFilesListFromSourceSettings) {
+        NS3::TSource src;
+        {
+            auto* p = src.AddDeprecatedPath();
+            p->SetPath("my/path");
+            p->SetSize(100500);
+        }
+        {
+            auto* p = src.AddDeprecatedPath();
+            p->SetPath("other/path");
+            p->SetSize(1);
+        }
+
+        TPathList paths;
+        ui64 startPathIndex = 42;
+        ReadPathsList(src, {}, paths, startPathIndex);
+
+        UNIT_ASSERT_VALUES_EQUAL(startPathIndex, 42);
+        UNIT_ASSERT_VALUES_EQUAL(paths.size(), 2);
+
+        UNIT_ASSERT_VALUES_EQUAL(std::get<TString>(paths[0]), "my/path");
+        UNIT_ASSERT_VALUES_EQUAL(std::get<ui64>(paths[0]), 100500);
+
+        UNIT_ASSERT_VALUES_EQUAL(std::get<TString>(paths[1]), "other/path");
+        UNIT_ASSERT_VALUES_EQUAL(std::get<ui64>(paths[1]), 1);
+    }
+
+    Y_UNIT_TEST(ReadsFilesListFromParamsAndSourceSettings) {
+        NS3::TSource src;
+        {
+            auto* p = src.AddDeprecatedPath();
+            p->SetPath("my/path");
+            p->SetSize(100500);
+        }
+        {
+            auto* p = src.AddDeprecatedPath();
+            p->SetPath("other/path");
+            p->SetSize(1);
+        }
+
+        NS3::TRange range;
+        range.SetStartPathIndex(42);
+        range.AddDeprecatedPath("my/path");
+
+        TPathList paths;
+        ui64 startPathIndex = 0;
+        ReadPathsList(src, MakeParams(range), paths, startPathIndex);
+
+        UNIT_ASSERT_VALUES_EQUAL(startPathIndex, 42);
+        UNIT_ASSERT_VALUES_EQUAL(paths.size(), 1);
+
+        UNIT_ASSERT_VALUES_EQUAL(std::get<TString>(paths[0]), "my/path");
+        UNIT_ASSERT_VALUES_EQUAL(std::get<ui64>(paths[0]), 100500);
+    }
+
+    NYql::NS3::TRange::TPath* SetPath(NYql::NS3::TRange::TPath* path, const TString& name = {}, ui64 size = 0, bool read = false) {
+        path->SetName(name);
+        path->SetSize(size);
+        path->SetRead(read);
+        return path;
+    }
+
+    Y_UNIT_TEST(ReadsFilesListFromTreeParams) {
+        NS3::TSource src;
+        {
+            auto* p = src.AddDeprecatedPath();
+            p->SetPath("my/path");
+            p->SetSize(100500);
+        }
+
+        NS3::TRange range;
+        range.SetStartPathIndex(42);
+        range.AddDeprecatedPath("my/path"); // We shouldn't react on this
+
+        {
+            auto* root = SetPath(range.AddPaths(), "root", 1, true);
+            {
+                auto* folder = SetPath(root->AddChildren(), "folder");
+                SetPath(folder->AddChildren(), "f1", 42, true);
+                SetPath(folder->AddChildren(), "f2", 100500, true);
+            }
+            SetPath(root->AddChildren(), "f3", 0, true);
+            SetPath(root->AddChildren(), "nothing"); // Shouldn't be processed.
+        }
+        {
+            auto* root2 = SetPath(range.AddPaths(), "root2");
+            SetPath(root2->AddChildren(), "f4", 42, true);
+        }
+
+        TPathList paths;
+        ui64 startPathIndex = 0;
+        ReadPathsList(src, MakeParams(range), paths, startPathIndex);
+
+        UNIT_ASSERT_VALUES_EQUAL(startPathIndex, 42);
+        UNIT_ASSERT_VALUES_EQUAL(paths.size(), 5);
+
+        UNIT_ASSERT_VALUES_EQUAL(std::get<TString>(paths[0]), "root");
+        UNIT_ASSERT_VALUES_EQUAL(std::get<ui64>(paths[0]), 1);
+
+        UNIT_ASSERT_VALUES_EQUAL(std::get<TString>(paths[1]), "root/folder/f1");
+        UNIT_ASSERT_VALUES_EQUAL(std::get<ui64>(paths[1]), 42);
+
+        UNIT_ASSERT_VALUES_EQUAL(std::get<TString>(paths[2]), "root/folder/f2");
+        UNIT_ASSERT_VALUES_EQUAL(std::get<ui64>(paths[2]), 100500);
+
+        UNIT_ASSERT_VALUES_EQUAL(std::get<TString>(paths[3]), "root/f3");
+        UNIT_ASSERT_VALUES_EQUAL(std::get<ui64>(paths[3]), 0);
+
+        UNIT_ASSERT_VALUES_EQUAL(std::get<TString>(paths[4]), "root2/f4");
+        UNIT_ASSERT_VALUES_EQUAL(std::get<ui64>(paths[4]), 42);
+    }
+}
+
+} // namespace NYql::NS3Details
diff --git a/ydb/library/yql/providers/s3/range_helpers/ut/CMakeLists.darwin.txt b/ydb/library/yql/providers/s3/range_helpers/ut/CMakeLists.darwin.txt
new file mode 100644
index 00000000000..690425de126
--- /dev/null
+++ b/ydb/library/yql/providers/s3/range_helpers/ut/CMakeLists.darwin.txt
@@ -0,0 +1,46 @@
+
+# This file was gererated by the build system used internally in the Yandex monorepo.
+# Only simple modifications are allowed (adding source-files to targets, adding simple properties
+# like target_include_directories). These modifications will be ported to original
+# ya.make files by maintainers. Any complex modifications which can't be ported back to the
+# original buildsystem will not be accepted.
+
+
+
+add_executable(ydb-library-yql-providers-s3-range_helpers-ut)
+target_compile_options(ydb-library-yql-providers-s3-range_helpers-ut PRIVATE
+  -DUSE_CURRENT_UDF_ABI_VERSION
+)
+target_include_directories(ydb-library-yql-providers-s3-range_helpers-ut PRIVATE
+  ${CMAKE_SOURCE_DIR}/ydb/library/yql/providers/s3/range_helpers
+)
+target_link_libraries(ydb-library-yql-providers-s3-range_helpers-ut PUBLIC
+  contrib-libs-cxxsupp
+  yutil
+  library-cpp-cpuid_check
+  cpp-testing-unittest_main
+  providers-s3-range_helpers
+  providers-common-provider
+)
+target_link_options(ydb-library-yql-providers-s3-range_helpers-ut PRIVATE
+  -Wl,-no_deduplicate
+  -Wl,-sdk_version,10.15
+  -fPIC
+  -fPIC
+)
+target_sources(ydb-library-yql-providers-s3-range_helpers-ut PRIVATE
+  ${CMAKE_SOURCE_DIR}/ydb/library/yql/providers/s3/range_helpers/file_tree_builder_ut.cpp
+  ${CMAKE_SOURCE_DIR}/ydb/library/yql/providers/s3/range_helpers/path_list_reader_ut.cpp
+)
+add_test(
+  NAME
+  ydb-library-yql-providers-s3-range_helpers-ut
+  COMMAND
+  ydb-library-yql-providers-s3-range_helpers-ut
+  --print-before-suite
+  --print-before-test
+  --fork-tests
+  --print-times
+  --show-fails
+)
+vcs_info(ydb-library-yql-providers-s3-range_helpers-ut)
diff --git a/ydb/library/yql/providers/s3/range_helpers/ut/CMakeLists.linux.txt b/ydb/library/yql/providers/s3/range_helpers/ut/CMakeLists.linux.txt
new file mode 100644
index 00000000000..df529a3d408
--- /dev/null
+++ b/ydb/library/yql/providers/s3/range_helpers/ut/CMakeLists.linux.txt
@@ -0,0 +1,52 @@
+
+# This file was gererated by the build system used internally in the Yandex monorepo.
+# Only simple modifications are allowed (adding source-files to targets, adding simple properties
+# like target_include_directories). These modifications will be ported to original
+# ya.make files by maintainers. Any complex modifications which can't be ported back to the
+# original buildsystem will not be accepted.
+
+
+
+add_executable(ydb-library-yql-providers-s3-range_helpers-ut)
+target_compile_options(ydb-library-yql-providers-s3-range_helpers-ut PRIVATE
+  -DUSE_CURRENT_UDF_ABI_VERSION
+)
+target_include_directories(ydb-library-yql-providers-s3-range_helpers-ut PRIVATE
+  ${CMAKE_SOURCE_DIR}/ydb/library/yql/providers/s3/range_helpers
+)
+target_link_libraries(ydb-library-yql-providers-s3-range_helpers-ut PUBLIC
+  contrib-libs-cxxsupp
+  yutil
+  cpp-malloc-tcmalloc
+  libs-tcmalloc-no_percpu_cache
+  library-cpp-cpuid_check
+  cpp-testing-unittest_main
+  providers-s3-range_helpers
+  providers-common-provider
+)
+target_link_options(ydb-library-yql-providers-s3-range_helpers-ut PRIVATE
+  -ldl
+  -lrt
+  -Wl,--no-as-needed
+  -fPIC
+  -fPIC
+  -lpthread
+  -lrt
+  -ldl
+)
+target_sources(ydb-library-yql-providers-s3-range_helpers-ut PRIVATE
+  ${CMAKE_SOURCE_DIR}/ydb/library/yql/providers/s3/range_helpers/file_tree_builder_ut.cpp
+  ${CMAKE_SOURCE_DIR}/ydb/library/yql/providers/s3/range_helpers/path_list_reader_ut.cpp
+)
+add_test(
+  NAME
+  ydb-library-yql-providers-s3-range_helpers-ut
+  COMMAND
+  ydb-library-yql-providers-s3-range_helpers-ut
+  --print-before-suite
+  --print-before-test
+  --fork-tests
+  --print-times
+  --show-fails
+)
+vcs_info(ydb-library-yql-providers-s3-range_helpers-ut)
diff --git a/ydb/library/yql/providers/s3/range_helpers/ut/CMakeLists.txt b/ydb/library/yql/providers/s3/range_helpers/ut/CMakeLists.txt
new file mode 100644
index 00000000000..fc7b1ee73ce
--- /dev/null
+++ b/ydb/library/yql/providers/s3/range_helpers/ut/CMakeLists.txt
@@ -0,0 +1,13 @@
+
+# This file was gererated by the build system used internally in the Yandex monorepo.
+# Only simple modifications are allowed (adding source-files to targets, adding simple properties
+# like target_include_directories). These modifications will be ported to original
+# ya.make files by maintainers. Any complex modifications which can't be ported back to the
+# original buildsystem will not be accepted.
+
+
+if (APPLE)
+  include(CMakeLists.darwin.txt)
+elseif (UNIX AND NOT APPLE)
+  include(CMakeLists.linux.txt)
+endif()