Merge pull request #184 from protegrity/av_sequncer_decryption_017

Refactor to move Parquet and AttributeMap logic out of encryption_sequencer

Author: avalerio-tkd

CMakeLists.txt

@@ -139,7 +139,7 @@ target_link_libraries(dbps_common_lib PUBLIC tcb_span)
 add_library(dbps_server_lib STATIC 
   src/server/encryption_sequencer.cpp
   src/server/auth_utils.cpp
-  src/server/decoding_utils.cpp
+  src/server/parquet_utils.cpp
   src/server/compression_utils.cpp
   src/server/encryptors/basic_encryptor.cpp
 )
@@ -278,14 +278,14 @@ if(BUILD_TESTS)
     gtest_main
   )
 
-  # Decoding utils tests
-  add_executable(decoding_utils_test src/server/decoding_utils_test.cpp)
-  target_link_libraries(decoding_utils_test
+  # Parquet utils tests
+  add_executable(parquet_utils_test src/server/parquet_utils_test.cpp)
+  target_link_libraries(parquet_utils_test
     dbps_server_lib
     dbps_common_lib
     gtest_main
   )
-  target_include_directories(decoding_utils_test PRIVATE src/server)
+  target_include_directories(parquet_utils_test PRIVATE src/server)
 
   # Bytes utils tests
   add_executable(bytes_utils_test src/common/bytes_utils_test.cpp)
@@ -484,7 +484,7 @@ if(BUILD_TESTS)
       value_encryption_utils_test
       typed_list_values_test
       encryption_sequencer_test
-      decoding_utils_test
+      parquet_utils_test
       bytes_utils_test
       compression_utils_test
       basic_encryptor_test
@@ -505,7 +505,7 @@ if(BUILD_TESTS)
   gtest_discover_tests(value_encryption_utils_test)
   gtest_discover_tests(typed_list_values_test)
   gtest_discover_tests(encryption_sequencer_test)
-  gtest_discover_tests(decoding_utils_test)
+  gtest_discover_tests(parquet_utils_test)
   gtest_discover_tests(bytes_utils_test)
   gtest_discover_tests(compression_utils_test)
   gtest_discover_tests(basic_encryptor_test)

src/common/bytes_utils.h

@@ -21,9 +21,14 @@
 #include <cstdint>
 #include <limits>
 #include <cstring>
+#include <map>
+#include <string>
+#include <variant>
+#include <cassert>
 #include "exceptions.h"
 
-// Little-endian helpers reused across modules
+// Utility functions for little-endian number reading and writing.
+
 inline void append_u32_le(std::vector<uint8_t>& out, uint32_t v) {
     out.push_back(static_cast<uint8_t>(v & 0xFF));
     out.push_back(static_cast<uint8_t>((v >> 8) & 0xFF));
@@ -69,6 +74,8 @@ inline uint32_t read_u32_le(const std::vector<uint8_t>& in, size_t offset) {
         (static_cast<uint32_t>(in[offset + 3]) << 24);
 }
 
+// Utility functions for splitting and joining byte vectors.
+
 struct SplitBytesPair {
     std::vector<uint8_t> leading;
     std::vector<uint8_t> trailing;
@@ -171,3 +178,61 @@ inline SplitBytesPair SplitWithLengthPrefix(const std::vector<uint8_t>& bytes) {
 
     return result;
 }
+
+// Utility functions for creating an AttributesMap
+
+// Common alias for converted encoding attributes used across modules.
+// Numeric values are captured as int32_t.
+using AttributesMap = std::map<std::string, std::variant<int32_t, bool, std::string>>;
+
+inline const std::string& GetRequiredAttribute(
+    const std::map<std::string, std::string>& attributes,
+    const std::string& key) {
+    auto it = attributes.find(key);
+    if (it == attributes.end()) {
+        throw InvalidInputException("Required encoding attribute [" + key + "] is missing");
+    }
+    return it->second;
+}
+
+inline int32_t AddIntAttribute(
+    AttributesMap& out,
+    const std::map<std::string, std::string>& attributes,
+    const std::string& key) {
+    const std::string& value = GetRequiredAttribute(attributes, key);
+    try {
+        int32_t value_int = static_cast<int32_t>(std::stol(value));
+        assert(value_int >= 0);
+        out[key] = value_int;
+        return value_int;
+    } catch (const std::exception& e) {
+        throw InvalidInputException(
+            "Failed to convert [" + key + "] with value [" + value + "] to int: " + e.what());
+    }
+}
+
+inline bool AddBoolAttribute(
+    AttributesMap& out,
+    const std::map<std::string, std::string>& attributes,
+    const std::string& key) {
+    const std::string& value = GetRequiredAttribute(attributes, key);
+    if (value == "true") {
+        out[key] = true;
+        return true;
+    } else if (value == "false") {
+        out[key] = false;
+        return false;
+    } else {
+        throw InvalidInputException(
+            "Failed to convert [" + key + "] with value [" + value + "] to bool");
+    }
+}
+
+inline std::string AddStringAttribute(
+    AttributesMap& out,
+    const std::map<std::string, std::string>& attributes,
+    const std::string& key) {
+    const std::string& value = GetRequiredAttribute(attributes, key);
+    out[key] = value;
+    return value;
+}

src/common/bytes_utils_test.cpp

@@ -19,6 +19,7 @@
 #include "exceptions.h"
 
 #include <vector>
+#include <variant>
 #include <gtest/gtest.h>
 
 TEST(BytesUtils, Split_Normal) {
@@ -199,4 +200,43 @@ TEST(BytesUtils, JoinWithLengthPrefixAndSplit_RoundTrip) {
 
     EXPECT_EQ(leading, parsed.leading);
     EXPECT_EQ(trailing, parsed.trailing);
+}
+
+TEST(BytesUtils, AttributesMap_AddString) {
+    std::map<std::string, std::string> attrs{{"page_type", "DATA_PAGE_V1"}};
+    AttributesMap out;
+
+    auto v = AddStringAttribute(out, attrs, "page_type");
+    EXPECT_EQ("DATA_PAGE_V1", v);
+    EXPECT_EQ("DATA_PAGE_V1", std::get<std::string>(out.at("page_type")));
+
+    EXPECT_THROW(AddStringAttribute(out, attrs, "missing_key"), InvalidInputException);
+}
+
+TEST(BytesUtils, AttributesMap_AddInt) {
+    std::map<std::string, std::string> attrs{{"data_page_num_values", "10"}};
+    AttributesMap out;
+
+    auto v = AddIntAttribute(out, attrs, "data_page_num_values");
+    EXPECT_EQ(10, v);
+    EXPECT_EQ(10, std::get<int32_t>(out.at("data_page_num_values")));
+
+    std::map<std::string, std::string> bad_attrs{{"data_page_num_values", "abc"}};
+    EXPECT_THROW(AddIntAttribute(out, bad_attrs, "data_page_num_values"), InvalidInputException);
+}
+
+TEST(BytesUtils, AttributesMap_AddBool) {
+    std::map<std::string, std::string> attrs_true{{"page_v2_is_compressed", "true"}};
+    AttributesMap out;
+    auto v_true = AddBoolAttribute(out, attrs_true, "page_v2_is_compressed");
+    EXPECT_TRUE(v_true);
+    EXPECT_TRUE(std::get<bool>(out.at("page_v2_is_compressed")));
+
+    std::map<std::string, std::string> attrs_false{{"page_v2_is_compressed", "false"}};
+    auto v_false = AddBoolAttribute(out, attrs_false, "page_v2_is_compressed");
+    EXPECT_FALSE(v_false);
+    EXPECT_FALSE(std::get<bool>(out.at("page_v2_is_compressed")));
+
+    std::map<std::string, std::string> bad_attrs{{"page_v2_is_compressed", "maybe"}};
+    EXPECT_THROW(AddBoolAttribute(out, bad_attrs, "page_v2_is_compressed"), InvalidInputException);
 }

src/common/value_encryption_utils_test.cpp

@@ -21,7 +21,6 @@
 #include <stdexcept>
 #include <vector>
 #include <cstdint>
-#include <variant>
 #include <string>
 #include <array>
 #include <limits>