feat: add json serde for table metadata

Author: wgtmac

src/iceberg/json_internal.cc

@@ -24,6 +24,9 @@
 #include <nlohmann/json_fwd.hpp>
 
 #include "iceberg/result.h"
+#include "iceberg/snapshot.h"
+#include "iceberg/statistics_file.h"
+#include "iceberg/table_metadata.h"
 #include "iceberg/type_fwd.h"
 
 namespace iceberg {
@@ -38,16 +41,6 @@ namespace iceberg {
 /// \return A JSON object representing the `SortField` in the form of key-value pairs.
 nlohmann::json ToJson(const SortField& sort_field);
 
-/// \brief Serializes a `SortOrder` object to JSON.
-///
-/// This function converts a `SortOrder` object into a JSON representation.
-/// The resulting JSON includes the order ID and a list of `SortField` objects.
-/// Each `SortField` is serialized as described in the `ToJson(SortField)` function.
-///
-/// \param sort_order The `SortOrder` object to be serialized.
-/// \return A JSON object representing the `SortOrder` with its order ID and fields array.
-nlohmann::json ToJson(const SortOrder& sort_order);
-
 /// \brief Deserializes a JSON object into a `SortField` object.
 ///
 /// This function parses the provided JSON and creates a `SortField` object.
@@ -59,6 +52,16 @@ nlohmann::json ToJson(const SortOrder& sort_order);
 /// JSON is malformed or missing expected fields, an error will be returned.
 Result<std::unique_ptr<SortField>> SortFieldFromJson(const nlohmann::json& json);
 
+/// \brief Serializes a `SortOrder` object to JSON.
+///
+/// This function converts a `SortOrder` object into a JSON representation.
+/// The resulting JSON includes the order ID and a list of `SortField` objects.
+/// Each `SortField` is serialized as described in the `ToJson(SortField)` function.
+///
+/// \param sort_order The `SortOrder` object to be serialized.
+/// \return A JSON object representing the `SortOrder` with its order ID and fields array.
+nlohmann::json ToJson(const SortOrder& sort_order);
+
 /// \brief Deserializes a JSON object into a `SortOrder` object.
 ///
 /// This function parses the provided JSON and creates a `SortOrder` object.
@@ -74,44 +77,32 @@ Result<std::unique_ptr<SortOrder>> SortOrderFromJson(const nlohmann::json& json)
 ///
 /// \param[in] schema The Iceberg schema to convert.
 /// \return The JSON representation of the schema.
-nlohmann::json SchemaToJson(const Schema& schema);
-
-/// \brief Convert an Iceberg Type to JSON.
-///
-/// \param[in] type The Iceberg type to convert.
-/// \return The JSON representation of the type.
-nlohmann::json TypeToJson(const Type& type);
-
-/// \brief Convert an Iceberg SchemaField to JSON.
-///
-/// \param[in] field The Iceberg field to convert.
-/// \return The JSON representation of the field.
-nlohmann::json FieldToJson(const SchemaField& field);
-
-/// \brief Serializes a `SnapshotRef` object to JSON.
-///
-/// \param[in] snapshot_ref The `SnapshotRef` object to be serialized.
-/// \return A JSON object representing the `SnapshotRef`.
-nlohmann::json ToJson(const SnapshotRef& snapshot_ref);
-
-/// \brief Serializes a `Snapshot` object to JSON.
-///
-/// \param[in] snapshot The `Snapshot` object to be serialized.
-/// \return A JSON object representing the `snapshot`.
-nlohmann::json ToJson(const Snapshot& snapshot);
+nlohmann::json ToJson(const Schema& schema);
 
 /// \brief Convert JSON to an Iceberg Schema.
 ///
 /// \param[in] json The JSON representation of the schema.
 /// \return The Iceberg schema or an error if the conversion fails.
 Result<std::unique_ptr<Schema>> SchemaFromJson(const nlohmann::json& json);
 
+/// \brief Convert an Iceberg Type to JSON.
+///
+/// \param[in] type The Iceberg type to convert.
+/// \return The JSON representation of the type.
+nlohmann::json ToJson(const Type& type);
+
 /// \brief Convert JSON to an Iceberg Type.
 ///
 /// \param[in] json The JSON representation of the type.
 /// \return The Iceberg type or an error if the conversion fails.
 Result<std::unique_ptr<Type>> TypeFromJson(const nlohmann::json& json);
 
+/// \brief Convert an Iceberg SchemaField to JSON.
+///
+/// \param[in] field The Iceberg field to convert.
+/// \return The JSON representation of the field.
+nlohmann::json ToJson(const SchemaField& field);
+
 /// \brief Convert JSON to an Iceberg SchemaField.
 ///
 /// \param[in] json The JSON representation of the field.
@@ -129,18 +120,6 @@ Result<std::unique_ptr<SchemaField>> FieldFromJson(const nlohmann::json& json);
 /// pairs.
 nlohmann::json ToJson(const PartitionField& partition_field);
 
-/// \brief Serializes a `PartitionSpec` object to JSON.
-///
-/// This function converts a `PartitionSpec` object into a JSON representation.
-/// The resulting JSON includes the spec ID and a list of `PartitionField` objects.
-/// Each `PartitionField` is serialized as described in the `ToJson(PartitionField)`
-/// function.
-///
-/// \param partition_spec The `PartitionSpec` object to be serialized.
-/// \return A JSON object representing the `PartitionSpec` with its order ID and fields
-/// array.
-nlohmann::json ToJson(const PartitionSpec& partition_spec);
-
 /// \brief Deserializes a JSON object into a `PartitionField` object.
 ///
 /// This function parses the provided JSON and creates a `PartitionField` object.
@@ -153,6 +132,18 @@ nlohmann::json ToJson(const PartitionSpec& partition_spec);
 Result<std::unique_ptr<PartitionField>> PartitionFieldFromJson(
     const nlohmann::json& json);
 
+/// \brief Serializes a `PartitionSpec` object to JSON.
+///
+/// This function converts a `PartitionSpec` object into a JSON representation.
+/// The resulting JSON includes the spec ID and a list of `PartitionField` objects.
+/// Each `PartitionField` is serialized as described in the `ToJson(PartitionField)`
+/// function.
+///
+/// \param partition_spec The `PartitionSpec` object to be serialized.
+/// \return A JSON object representing the `PartitionSpec` with its order ID and fields
+/// array.
+nlohmann::json ToJson(const PartitionSpec& partition_spec);
+
 /// \brief Deserializes a JSON object into a `PartitionSpec` object.
 ///
 /// This function parses the provided JSON and creates a `PartitionSpec` object.
@@ -166,16 +157,90 @@ Result<std::unique_ptr<PartitionField>> PartitionFieldFromJson(
 Result<std::unique_ptr<PartitionSpec>> PartitionSpecFromJson(
     const std::shared_ptr<Schema>& schema, const nlohmann::json& json);
 
+/// \brief Serializes a `SnapshotRef` object to JSON.
+///
+/// \param[in] snapshot_ref The `SnapshotRef` object to be serialized.
+/// \return A JSON object representing the `SnapshotRef`.
+nlohmann::json ToJson(const SnapshotRef& snapshot_ref);
+
 /// \brief Deserializes a JSON object into a `SnapshotRef` object.
 ///
 /// \param[in] json The JSON object representing a `SnapshotRef`.
 /// \return A `SnapshotRef` object or an error if the conversion fails.
 Result<std::unique_ptr<SnapshotRef>> SnapshotRefFromJson(const nlohmann::json& json);
 
+/// \brief Serializes a `Snapshot` object to JSON.
+///
+/// \param[in] snapshot The `Snapshot` object to be serialized.
+/// \return A JSON object representing the `snapshot`.
+nlohmann::json ToJson(const Snapshot& snapshot);
+
 /// \brief Deserializes a JSON object into a `Snapshot` object.
 ///
 /// \param[in] json The JSON representation of the snapshot.
 /// \return A `Snapshot` object or an error if the conversion fails.
 Result<std::unique_ptr<Snapshot>> SnapshotFromJson(const nlohmann::json& json);
 
+/// \brief Serializes a `StatisticsFile` object to JSON.
+///
+/// \param statistics_file The `StatisticsFile` object to be serialized.
+/// \return A JSON object representing the `StatisticsFile`.
+nlohmann::json ToJson(const StatisticsFile& statistics_file);
+
+/// \brief Deserializes a JSON object into a `StatisticsFile` object.
+///
+/// \param json The JSON object representing a `StatisticsFile`.
+/// \return A `StatisticsFile` object or an error if the conversion fails.
+Result<std::unique_ptr<StatisticsFile>> StatisticsFileFromJson(
+    const nlohmann::json& json);
+
+/// \brief Serializes a `PartitionStatisticsFile` object to JSON.
+///
+/// \param partition_statistics_file The `PartitionStatisticsFile` object to be
+/// serialized. \return A JSON object representing the `PartitionStatisticsFile`.
+nlohmann::json ToJson(const PartitionStatisticsFile& partition_statistics_file);
+
+/// \brief Deserializes a JSON object into a `PartitionStatisticsFile` object.
+///
+/// \param json The JSON object representing a `PartitionStatisticsFile`.
+/// \return A `PartitionStatisticsFile` object or an error if the conversion fails.
+Result<std::unique_ptr<PartitionStatisticsFile>> PartitionStatisticsFileFromJson(
+    const nlohmann::json& json);
+
+/// \brief Serializes a `SnapshotLogEntry` object to JSON.
+///
+/// \param snapshot_log_entry The `SnapshotLogEntry` object to be serialized.
+/// \return A JSON object representing the `SnapshotLogEntry`.
+nlohmann::json ToJson(const SnapshotLogEntry& snapshot_log_entry);
+
+/// \brief Deserializes a JSON object into a `SnapshotLogEntry` object.
+///
+/// \param json The JSON object representing a `SnapshotLogEntry`.
+/// \return A `SnapshotLogEntry` object or an error if the conversion fails.
+Result<SnapshotLogEntry> SnapshotLogEntryFromJson(const nlohmann::json& json);
+
+/// \brief Serializes a `MetadataLogEntry` object to JSON.
+///
+/// \param metadata_log_entry The `MetadataLogEntry` object to be serialized.
+/// \return A JSON object representing the `MetadataLogEntry`.
+nlohmann::json ToJson(const MetadataLogEntry& metadata_log_entry);
+
+/// \brief Deserializes a JSON object into a `MetadataLogEntry` object.
+///
+/// \param json The JSON object representing a `MetadataLogEntry`.
+/// \return A `MetadataLogEntry` object or an error if the conversion fails.
+Result<MetadataLogEntry> MetadataLogEntryFromJson(const nlohmann::json& json);
+
+/// \brief Serializes a `TableMetadata` object to JSON.
+///
+/// \param table_metadata The `TableMetadata` object to be serialized.
+/// \return A JSON object representing the `TableMetadata`.
+nlohmann::json ToJson(const TableMetadata& table_metadata);
+
+/// \brief Deserializes a JSON object into a `TableMetadata` object.
+///
+/// \param json The JSON object representing a `TableMetadata`.
+/// \return A `TableMetadata` object or an error if the conversion fails.
+Result<std::unique_ptr<TableMetadata>> TableMetadataFromJson(const nlohmann::json& json);
+
 }  // namespace iceberg

src/iceberg/json_internal.h

@@ -19,6 +19,7 @@
 
 #pragma once
 
+#include <format>
 #include <string>
 
 #include "iceberg/expected.h"
@@ -60,4 +61,20 @@ using Result = expected<T, E>;
 
 using Status = Result<void>;
 
+/// \brief Create an unexpected error with kNotImplemented
+template <typename... Args>
+auto NotImplementedError(const std::format_string<Args...> fmt, Args&&... args)
+    -> unexpected<Error> {
+  return unexpected<Error>({.kind = ErrorKind::kNotImplemented,
+                            .message = std::format(fmt, std::forward<Args>(args)...)});
+}
+
+/// \brief Create an unexpected error with kJsonParseError
+template <typename... Args>
+auto JsonParseError(const std::format_string<Args...> fmt, Args&&... args)
+    -> unexpected<Error> {
+  return unexpected<Error>({.kind = ErrorKind::kJsonParseError,
+                            .message = std::format(fmt, std::forward<Args>(args)...)});
+}
+
 }  // namespace iceberg

src/iceberg/result.h

@@ -23,29 +23,26 @@
 
 namespace iceberg {
 
-bool BlobMetadata::Equals(const BlobMetadata& other) const {
-  return type == other.type && source_snapshot_id == other.source_snapshot_id &&
-         source_snapshot_sequence_number == other.source_snapshot_sequence_number &&
-         fields == other.fields && properties == other.properties;
-}
-
-std::string BlobMetadata::ToString() const {
+std::string ToString(const BlobMetadata& blob_metadata) {
   std::string repr = "BlobMetadata[";
   std::format_to(std::back_inserter(repr),
-                 "type='{}',sourceSnapshotId={},sourceSnapshotSequenceNumber={},", type,
-                 source_snapshot_id, source_snapshot_sequence_number);
+                 "type='{}',sourceSnapshotId={},sourceSnapshotSequenceNumber={},",
+                 blob_metadata.type, blob_metadata.source_snapshot_id,
+                 blob_metadata.source_snapshot_sequence_number);
   std::format_to(std::back_inserter(repr), "fields=[");
-  for (auto iter = fields.cbegin(); iter != fields.cend(); ++iter) {
-    if (iter != fields.cbegin()) {
+  for (auto iter = blob_metadata.fields.cbegin(); iter != blob_metadata.fields.cend();
+       ++iter) {
+    if (iter != blob_metadata.fields.cbegin()) {
       std::format_to(std::back_inserter(repr), ",{}", *iter);
     } else {
       std::format_to(std::back_inserter(repr), "{}", *iter);
     }
   }
   std::format_to(std::back_inserter(repr), "],properties=[");
-  for (auto iter = properties.cbegin(); iter != properties.cend(); ++iter) {
+  for (auto iter = blob_metadata.properties.cbegin();
+       iter != blob_metadata.properties.cend(); ++iter) {
     const auto& [key, value] = *iter;
-    if (iter != properties.cbegin()) {
+    if (iter != blob_metadata.properties.cbegin()) {
       std::format_to(std::back_inserter(repr), ",{}:{}", key, value);
     } else {
       std::format_to(std::back_inserter(repr), "{}:{}", key, value);
@@ -55,28 +52,32 @@ std::string BlobMetadata::ToString() const {
   return repr;
 }
 
-bool StatisticsFile::Equals(const StatisticsFile& other) const {
-  return snapshot_id == other.snapshot_id && path == other.path &&
-         file_size_in_bytes == other.file_size_in_bytes &&
-         file_footer_size_in_bytes == other.file_footer_size_in_bytes &&
-         blob_metadata == other.blob_metadata;
-}
-
-std::string StatisticsFile::ToString() const {
+std::string ToString(const StatisticsFile& statistics_file) {
   std::string repr = "StatisticsFile[";
   std::format_to(std::back_inserter(repr),
                  "snapshotId={},path={},fileSizeInBytes={},fileFooterSizeInBytes={},",
-                 snapshot_id, path, file_size_in_bytes, file_footer_size_in_bytes);
+                 statistics_file.snapshot_id, statistics_file.path,
+                 statistics_file.file_size_in_bytes,
+                 statistics_file.file_footer_size_in_bytes);
   std::format_to(std::back_inserter(repr), "blobMetadata=[");
-  for (auto iter = blob_metadata.cbegin(); iter != blob_metadata.cend(); ++iter) {
-    if (iter != blob_metadata.cbegin()) {
-      std::format_to(std::back_inserter(repr), ",{}", iter->ToString());
+  for (auto iter = statistics_file.blob_metadata.cbegin();
+       iter != statistics_file.blob_metadata.cend(); ++iter) {
+    if (iter != statistics_file.blob_metadata.cbegin()) {
+      std::format_to(std::back_inserter(repr), ",{}", ToString(*iter));
     } else {
-      std::format_to(std::back_inserter(repr), "{}", iter->ToString());
+      std::format_to(std::back_inserter(repr), "{}", ToString(*iter));
     }
   }
   repr += "]]";
   return repr;
 }
 
+std::string ToString(const PartitionStatisticsFile& partition_statistics_file) {
+  std::string repr = "PartitionStatisticsFile[";
+  std::format_to(std::back_inserter(repr), "snapshotId={},path={},fileSizeInBytes={},",
+                 partition_statistics_file.snapshot_id, partition_statistics_file.path,
+                 partition_statistics_file.file_size_in_bytes);
+  return repr;
+}
+
 }  // namespace iceberg