Tighten up fallback to per-block encryption on Encryption sequencer

src/server/compression_utils.cpp

@@ -23,6 +23,7 @@ using namespace dbps::enum_utils;
 
 namespace dbps::compression {
 
+// TODO(Issue #188): Add support for other compressions.
 std::vector<uint8_t> Compress(const std::vector<uint8_t>& bytes, CompressionCodec::type compression) {
     if (compression == CompressionCodec::UNCOMPRESSED) {
         return bytes;

src/server/encryption_sequencer.cpp

@@ -140,24 +140,49 @@ bool DataBatchEncryptionSequencer::ConvertAndEncrypt(const std::vector<uint8_t>&
         // Compress the joined encrypted bytes
         encrypted_result_ = Compress(joined_encrypted_bytes, encrypted_compression_);
 
-    } catch (const DBPSUnsupportedException& e) {
-        // If any stage is as of yet unsupported, default to whole payload (per-block) encryption
-        // (as opposed to per-value)
+    }
+    // If the sequence was interrupted by a DBPSUnsupportedException, allow fallback to per-block encryption
+    // but only for explicitly unsupported conditions.
+    // Any conditions that are already supported should not fallback. In those cases, the exception is re-thrown.
+    catch (const DBPSUnsupportedException& e) {
+
+        // Compression: Only UNCOMPRESSED and SNAPPY are supported
+        // TODO(Issue #188): Add support for other compressions.
+        const bool is_compression_supported = (compression_ == CompressionCodec::UNCOMPRESSED ||
+                                               compression_ == CompressionCodec::SNAPPY);
+
+        // Format: Only PLAIN is supported
+        // TODO(Issue #187): Add support for other encodings.
+        const bool is_format_supported = (format_ == Format::PLAIN);
+
+        // Page type: All are supported (DATA_PAGE_V1, DATA_PAGE_V2, DICTIONARY_PAGE)
+        const bool is_page_supported = true;
+
+        // Datatype: All datatypes are supported.
+        const bool is_datatype_supported = true;
+
+        if (is_compression_supported && is_format_supported && is_page_supported && is_datatype_supported) {
+            // All conditions are supported, therefore an DBPSUnsupportedException exception should not have happened. 
+            // Re-throw the exception.
+            throw;
+        }
+
+        // Fallback: Use per-block encryption for unsupported combinations.
         encrypted_result_ = encryptor_->EncryptBlock(plaintext);
         if (encrypted_result_.empty()) {
             error_stage_ = "encryption";
             error_message_ = "Failed to encrypt data";
             return false;
         }
-        // If the sequence was interrupted by a DBPSUnsupportedException at any point, use per block encryption.
-        // Set the encryption type to per block.
         encryption_metadata_[ENCRYPTION_MODE] = ENCRYPTION_PER_BLOCK;
         encryption_metadata_[DBPS_VERSION_KEY] = DBPS_VERSION;
         return true;
+
     } catch (const InvalidInputException& e) {
         // Throw the exception so it can be caught by the caller.
         throw;
     }
+
     // If the sequencer got here, it means the encryption of the values in the typed list finished successfully.
     // Set the encryption type to per value
     encryption_metadata_[ENCRYPTION_MODE] = ENCRYPTION_PER_VALUE;

src/server/encryptors/basic_encryptor.h

@@ -28,7 +28,6 @@
  * 
  * This implementation provides:
  * - Block encryption/decryption using XOR with key_id hash (same as current encryption_sequencer)
- * - Value encryption/decryption currently throws DBPSUnsupportedException (same as current encryption_sequencer)
  * 
  * This is a simple, default encryption implementation that can be replaced with more
  * sophisticated encryption providers (e.g., Protegrity) in the future.

src/server/parquet_utils.cpp

@@ -92,8 +92,8 @@ int CalculateLevelBytesLength(const std::vector<uint8_t>& raw,
         total_level_bytes = 0;
 
     } else {
-        // Unknown page type
-        throw DBPSUnsupportedException("Unsupported page type: " + page_type);
+        // Invalid page type
+        throw InvalidInputException("Invalid page type: " + page_type);
     }
 
     // Validate that the total level bytes before returning.
@@ -130,10 +130,10 @@ inline static size_t GetFixedElemSizeOrThrow(Type::type datatype, const std::opt
         case Type::UNDEFINED:
             return 1;
         case Type::BYTE_ARRAY:
-            throw DBPSUnsupportedException("BYTE_ARRAY is variable-length; not fixed-size");
+            throw InvalidInputException("BYTE_ARRAY is variable-length; not fixed-size");
         default:
-            throw DBPSUnsupportedException(
-                "Unsupported datatype: " + std::string(to_string(datatype)));
+            throw InvalidInputException(
+                "Invalid datatype. Only fixed-size types are supported: " + std::string(to_string(datatype)));
     }
 }
 
@@ -142,8 +142,16 @@ std::vector<RawValueBytes> SliceValueBytesIntoRawBytes(
     Type::type datatype,
     const std::optional<int>& datatype_length,
     Format::type format) {
+
+    // RLE_DICTIONARY is not supported for per-value operations since the values themselves are not present in the data,
+    // only references to them.
+    if (format == Format::RLE_DICTIONARY) {
+        throw DBPSUnsupportedException("Unsupported format: RLE_DICTIONARY is not supported for per-value operations");
+    }
+
+    // TODO(Issue #187): Add support for other encodings.
     if (format != Format::PLAIN) {
-        throw DBPSUnsupportedException("Unsupported format: " + std::string(to_string(format)));
+        throw DBPSUnsupportedException("On SliceValueBytesIntoRawBytes, unsupported format: " + std::string(to_string(format)));
     }
 
     // Variable-length BYTE_ARRAY: parse [4-byte len][bytes...] elements in order.
@@ -193,8 +201,16 @@ std::vector<uint8_t> CombineRawBytesIntoValueBytes(
     Type::type datatype,
     const std::optional<int>& datatype_length,
     Format::type format) {
+
+    // RLE_DICTIONARY is not supported for per-value operations since the values themselves are not present in the data,
+    // only references to them.
+    if (format == Format::RLE_DICTIONARY) {
+        throw DBPSUnsupportedException("Unsupported format: RLE_DICTIONARY is not supported for per-value operations");
+    }
+
+    // TODO(Issue #187): Add support for other encodings.
     if (format != Format::PLAIN) {
-        throw DBPSUnsupportedException("Unsupported format: " + std::string(to_string(format)));
+        throw DBPSUnsupportedException("On CombineRawBytesIntoValueBytes, unsupported format: " + std::string(to_string(format)));
     }
 
     if (datatype == Type::BYTE_ARRAY) {

src/server/parquet_utils_test.cpp

@@ -120,7 +120,7 @@ TEST(ParquetUtils, CalculateLevelBytesLength_UnknownPageType) {
     AttributesMap attribs = {
         {"page_type", std::string("UNKNOWN_PAGE_TYPE")}
     };
-    EXPECT_THROW(CalculateLevelBytesLength(raw, attribs), DBPSUnsupportedException);
+    EXPECT_THROW(CalculateLevelBytesLength(raw, attribs), InvalidInputException);
 }
 
 TEST(ParquetUtils, CalculateLevelBytesLength_InvalidTotalSize) {