feat: add table metadata reader

Author: wgtmac

src/iceberg/json_internal.cc

@@ -1209,4 +1209,20 @@ Result<std::unique_ptr<TableMetadata>> TableMetadataFromJson(const nlohmann::jso
   return table_metadata;
 }
 
+Result<nlohmann::json> FromJsonString(const std::string& json_string) {
+  try {
+    return nlohmann::json::parse(json_string);
+  } catch (const std::exception& e) {
+    return JsonParseError("Failed to parse JSON string: {}", e.what());
+  }
+}
+
+Result<std::string> ToJsonString(const nlohmann::json& json) {
+  try {
+    return json.dump();
+  } catch (const std::exception& e) {
+    return JsonParseError("Failed to serialize to JSON string: {}", e.what());
+  }
+}
+
 }  // namespace iceberg

src/iceberg/json_internal.h

@@ -244,4 +244,16 @@ nlohmann::json ToJson(const TableMetadata& table_metadata);
 /// \return A `TableMetadata` object or an error if the conversion fails.
 Result<std::unique_ptr<TableMetadata>> TableMetadataFromJson(const nlohmann::json& json);
 
+/// \brief Deserialize a JSON string into a `nlohmann::json` object.
+///
+/// \param json_string The JSON string to deserialize.
+/// \return A `nlohmann::json` object or an error if the deserialization fails.
+Result<nlohmann::json> FromJsonString(const std::string& json_string);
+
+/// \brief Serialize a `nlohmann::json` object into a JSON string.
+///
+/// \param json The `nlohmann::json` object to serialize.
+/// \return A JSON string or an error if the serialization fails.
+Result<std::string> ToJsonString(const nlohmann::json& json);
+
 }  // namespace iceberg

src/iceberg/table_metadata.cc

@@ -23,10 +23,15 @@
 #include <ranges>
 #include <string>
 
+#include <nlohmann/json.hpp>
+
+#include "iceberg/file_io.h"
+#include "iceberg/json_internal.h"
 #include "iceberg/partition_spec.h"
 #include "iceberg/result.h"
 #include "iceberg/schema.h"
 #include "iceberg/sort_order.h"
+#include "iceberg/util/macros.h"
 namespace iceberg {
 
 std::string ToString(const SnapshotLogEntry& entry) {
@@ -69,4 +74,35 @@ Result<std::shared_ptr<SortOrder>> TableMetadata::SortOrder() const {
   return *iter;
 }
 
+Result<MetadataFileCodecType> TableMetadataUtil::FromFileName(
+    std::string_view file_name) {
+  if (file_name.find(".metadata.json") == std::string::npos) {
+    return InvalidArgument("{} is not a valid metadata file", file_name);
+  }
+
+  // We have to be backward-compatible with .metadata.json.gz files
+  if (file_name.ends_with(".metadata.json.gz")) {
+    return MetadataFileCodecType::kGzip;
+  }
+
+  std::string_view file_name_without_suffix =
+      file_name.substr(0, file_name.find_last_of(".metadata.json"));
+  if (file_name_without_suffix.ends_with(".gz")) {
+    return MetadataFileCodecType::kGzip;
+  }
+  return MetadataFileCodecType::kNone;
+}
+
+Result<std::unique_ptr<TableMetadata>> TableMetadataUtil::Read(
+    FileIO& io, const std::string& location, std::optional<size_t> length) {
+  ICEBERG_ASSIGN_OR_RAISE(auto codec_type, FromFileName(location));
+  if (codec_type == MetadataFileCodecType::kGzip) {
+    return NotImplemented("Reading gzip-compressed metadata files is not supported yet");
+  }
+
+  ICEBERG_ASSIGN_OR_RAISE(auto content, io.ReadFile(location, length));
+  ICEBERG_ASSIGN_OR_RAISE(auto json, FromJsonString(content));
+  return TableMetadataFromJson(json);
+}
+
 }  // namespace iceberg

src/iceberg/table_metadata.h

@@ -24,6 +24,7 @@
 
 #include <memory>
 #include <string>
+#include <string_view>
 #include <unordered_map>
 #include <vector>
 
@@ -140,4 +141,29 @@ ICEBERG_EXPORT std::string ToString(const SnapshotLogEntry& entry);
 /// \brief Returns a string representation of a MetadataLogEntry
 ICEBERG_EXPORT std::string ToString(const MetadataLogEntry& entry);
 
+/// \brief The codec type of the table metadata file.
+ICEBERG_EXPORT enum class MetadataFileCodecType {
+  kNone,
+  kGzip,
+};
+
+/// \brief Utility class for table metadata
+struct ICEBERG_EXPORT TableMetadataUtil {
+  /// \brief Get the codec type from the table metadata file name.
+  ///
+  /// \param file_name The name of the table metadata file.
+  /// \return The codec type of the table metadata file.
+  static Result<MetadataFileCodecType> FromFileName(std::string_view file_name);
+
+  /// \brief Read the table metadata file.
+  ///
+  /// \param io The file IO to use to read the table metadata.
+  /// \param location The location of the table metadata file.
+  /// \param length The optional length of the table metadata file.
+  /// \return The table metadata.
+  static Result<std::unique_ptr<TableMetadata>> Read(class FileIO& io,
+                                                     const std::string& location,
+                                                     std::optional<size_t> length);
+};
+
 }  // namespace iceberg