refactor: Add SchemaById and SnapshotById to TableMetadata

Author: gty404

src/iceberg/table.cc

@@ -90,14 +90,7 @@ Result<std::shared_ptr<Snapshot>> Table::current_snapshot() const {
 }
 
 Result<std::shared_ptr<Snapshot>> Table::SnapshotById(int64_t snapshot_id) const {
-  auto iter = std::ranges::find_if(metadata_->snapshots,
-                                   [this, &snapshot_id](const auto& snapshot) {
-                                     return snapshot->snapshot_id == snapshot_id;
-                                   });
-  if (iter == metadata_->snapshots.end()) {
-    return NotFound("Snapshot with ID {} is not found", snapshot_id);
-  }
-  return *iter;
+  return metadata_->SnapshotById(snapshot_id);
 }
 
 const std::vector<std::shared_ptr<Snapshot>>& Table::snapshots() const {

src/iceberg/table_metadata.cc

@@ -47,11 +47,16 @@ std::string ToString(const MetadataLogEntry& entry) {
 }
 
 Result<std::shared_ptr<Schema>> TableMetadata::Schema() const {
-  auto iter = std::ranges::find_if(schemas, [this](const auto& schema) {
-    return schema->schema_id() == current_schema_id;
+  return SchemaById(current_schema_id);
+}
+
+Result<std::shared_ptr<Schema>> TableMetadata::SchemaById(
+    const std::optional<int32_t>& schema_id) const {
+  auto iter = std::ranges::find_if(schemas, [schema_id](const auto& schema) {
+    return schema->schema_id() == schema_id;
   });
   if (iter == schemas.end()) {
-    return NotFound("Current schema is not found");
+    return NotFound("Schema with ID {} is not found", schema_id.value_or(-1));
   }
   return *iter;
 }
@@ -77,11 +82,15 @@ Result<std::shared_ptr<SortOrder>> TableMetadata::SortOrder() const {
 }
 
 Result<std::shared_ptr<Snapshot>> TableMetadata::Snapshot() const {
-  auto iter = std::ranges::find_if(snapshots, [this](const auto& snapshot) {
-    return snapshot->snapshot_id == current_snapshot_id;
+  return SnapshotById(current_snapshot_id);
+}
+
+Result<std::shared_ptr<Snapshot>> TableMetadata::SnapshotById(int64_t snapshot_id) const {
+  auto iter = std::ranges::find_if(snapshots, [snapshot_id](const auto& snapshot) {
+    return snapshot->snapshot_id == snapshot_id;
   });
   if (iter == snapshots.end()) {
-    return NotFound("Current snapshot with ID {} is not found", current_snapshot_id);
+    return NotFound("Snapshot with ID {} is not found", snapshot_id);
   }
   return *iter;
 }

src/iceberg/table_metadata.h

@@ -123,12 +123,17 @@ struct ICEBERG_EXPORT TableMetadata {
 
   /// \brief Get the current schema, return NotFoundError if not found
   Result<std::shared_ptr<Schema>> Schema() const;
+  /// \brief Get the current schema by ID, return NotFoundError if not found
+  Result<std::shared_ptr<iceberg::Schema>> SchemaById(
+      const std::optional<int32_t>& schema_id) const;
   /// \brief Get the current partition spec, return NotFoundError if not found
   Result<std::shared_ptr<PartitionSpec>> PartitionSpec() const;
   /// \brief Get the current sort order, return NotFoundError if not found
   Result<std::shared_ptr<SortOrder>> SortOrder() const;
   /// \brief Get the current snapshot, return NotFoundError if not found
   Result<std::shared_ptr<Snapshot>> Snapshot() const;
+  /// \brief Get the snapshot of this table with the given id
+  Result<std::shared_ptr<iceberg::Snapshot>> SnapshotById(int64_t snapshot_id) const;
 
   friend bool operator==(const TableMetadata& lhs, const TableMetadata& rhs);
 };

src/iceberg/table_scan.cc

@@ -34,8 +34,8 @@
 namespace iceberg {
 
 // implement FileScanTask
-FileScanTask::FileScanTask(std::shared_ptr<DataFile> file)
-    : data_file_(std::move(file)) {}
+FileScanTask::FileScanTask(std::shared_ptr<DataFile> data_file)
+    : data_file_(std::move(data_file)) {}
 
 const std::shared_ptr<DataFile>& FileScanTask::data_file() const { return data_file_; }
 
@@ -94,32 +94,13 @@ Result<std::unique_ptr<TableScan>> TableScanBuilder::Build() {
     return InvalidArgument("No snapshot ID specified for table {}",
                            table_metadata->table_uuid);
   }
-  auto iter = std::ranges::find_if(
-      table_metadata->snapshots,
-      [id = *snapshot_id](const auto& snapshot) { return snapshot->snapshot_id == id; });
-  if (iter == table_metadata->snapshots.end()) {
-    return NotFound("Snapshot with ID {} is not found", *snapshot_id);
-  }
-  context_.snapshot = *iter;
+  ICEBERG_ASSIGN_OR_RAISE(context_.snapshot, table_metadata->SnapshotById(*snapshot_id));
 
   if (!context_.projected_schema) {
     const auto& snapshot = context_.snapshot;
     auto schema_id =
         snapshot->schema_id ? snapshot->schema_id : table_metadata->current_schema_id;
-    if (!schema_id) {
-      return InvalidArgument("No schema ID found in snapshot {} for table {}",
-                             snapshot->snapshot_id, table_metadata->table_uuid);
-    }
-
-    const auto& schemas = table_metadata->schemas;
-    const auto it = std::ranges::find_if(schemas, [id = *schema_id](const auto& schema) {
-      return schema->schema_id() == id;
-    });
-    if (it == schemas.end()) {
-      return InvalidArgument("Schema {} in snapshot {} is not found",
-                             *snapshot->schema_id, snapshot->snapshot_id);
-    }
-    const auto& schema = *it;
+    ICEBERG_ASSIGN_OR_RAISE(auto schema, table_metadata->SchemaById(schema_id));
 
     if (column_names_.empty()) {
       context_.projected_schema = schema;
@@ -139,6 +120,9 @@ Result<std::unique_ptr<TableScan>> TableScanBuilder::Build() {
       context_.projected_schema =
           std::make_shared<Schema>(std::move(projected_fields), schema->schema_id());
     }
+  } else if (!column_names_.empty()) {
+    return InvalidArgument(
+        "Cannot specify column names when a projected schema is provided");
   }
 
   return std::make_unique<DataTableScan>(std::move(context_), file_io_);

src/iceberg/table_scan.h

@@ -45,7 +45,7 @@ class ICEBERG_EXPORT ScanTask {
 /// \brief Task representing a data file and its corresponding delete files.
 class ICEBERG_EXPORT FileScanTask : public ScanTask {
  public:
-  explicit FileScanTask(std::shared_ptr<DataFile> file);
+  explicit FileScanTask(std::shared_ptr<DataFile> data_file);
 
   /// \brief The data file that should be read by this scan task.
   const std::shared_ptr<DataFile>& data_file() const;