feat: add find field (by id and name) support to schema

src/iceberg/schema.cc

@@ -19,13 +19,57 @@
 
 #include "iceberg/schema.h"
 
+#include <algorithm>
 #include <format>
+#include <functional>
 
 #include "iceberg/type.h"
 #include "iceberg/util/formatter.h"  // IWYU pragma: keep
+#include "iceberg/util/macros.h"
+#include "iceberg/util/visit_type.h"
 
 namespace iceberg {
 
+class IdToFieldVisitor {
+ public:
+  explicit IdToFieldVisitor(
+      std::unordered_map<int32_t, std::reference_wrapper<const SchemaField>>&
+          id_to_field);
+  Status Visit(const Type& type);
+  Status VisitNestedType(const Type& type);
+
+ private:
+  std::unordered_map<int32_t, std::reference_wrapper<const SchemaField>>& id_to_field_;
+};
+
+class NameToIdVisitor {
+ public:
+  explicit NameToIdVisitor(
+      std::unordered_map<std::string, int32_t, string_hash, std::equal_to<>>& name_to_id,
+      bool case_sensitive = true,
+      std::function<std::string(std::string_view)> quoting_func = {});
+  Status Visit(const ListType& type, const std::string& path,
+               const std::string& short_path);
+  Status Visit(const MapType& type, const std::string& path,
+               const std::string& short_path);
+  Status Visit(const StructType& type, const std::string& path,
+               const std::string& short_path);
+  Status Visit(const PrimitiveType& type, const std::string& path,
+               const std::string& short_path);
+  void Finish();
+
+ private:
+  std::string BuildPath(std::string_view prefix, std::string_view field_name,
+                        bool case_sensitive);
+
+ private:
+  bool case_sensitive_;
+  std::unordered_map<std::string, int32_t, string_hash, std::equal_to<>>& name_to_id_;
+  std::unordered_map<std::string, int32_t, string_hash, std::equal_to<>>
+      short_name_to_id_;
+  std::function<std::string(std::string_view)> quoting_func_;
+};
+
 Schema::Schema(std::vector<SchemaField> fields, std::optional<int32_t> schema_id)
     : StructType(std::move(fields)), schema_id_(schema_id) {}
 
@@ -44,4 +88,181 @@ bool Schema::Equals(const Schema& other) const {
   return schema_id_ == other.schema_id_ && fields_ == other.fields_;
 }
 
+Result<std::optional<std::reference_wrapper<const SchemaField>>> Schema::FindFieldByName(
+    std::string_view name, bool case_sensitive) const {
+  if (case_sensitive) {
+    ICEBERG_RETURN_UNEXPECTED(InitNameToIdMap());
+    auto it = name_to_id_.find(name);
+    if (it == name_to_id_.end()) return std::nullopt;
+    return FindFieldById(it->second);
+  }
+  ICEBERG_RETURN_UNEXPECTED(InitLowerCaseNameToIdMap());
+  auto it = lowercase_name_to_id_.find(StringUtils::ToLower(name));
+  if (it == lowercase_name_to_id_.end()) return std::nullopt;
+  return FindFieldById(it->second);
+}
+
+Status Schema::InitIdToFieldMap() const {
+  if (!id_to_field_.empty()) {
+    return {};
+  }
+  IdToFieldVisitor visitor(id_to_field_);
+  ICEBERG_RETURN_UNEXPECTED(VisitTypeInline(*this, &visitor));
+  return {};
+}
+
+Status Schema::InitNameToIdMap() const {
+  if (!name_to_id_.empty()) {
+    return {};
+  }
+  NameToIdVisitor visitor(name_to_id_, /*case_sensitive=*/true);
+  ICEBERG_RETURN_UNEXPECTED(
+      VisitTypeInline(*this, &visitor, /*path=*/"", /*short_path=*/""));
+  visitor.Finish();
+  return {};
+}
+
+Status Schema::InitLowerCaseNameToIdMap() const {
+  if (!lowercase_name_to_id_.empty()) {
+    return {};
+  }
+  NameToIdVisitor visitor(lowercase_name_to_id_, /*case_sensitive=*/false);
+  ICEBERG_RETURN_UNEXPECTED(
+      VisitTypeInline(*this, &visitor, /*path=*/"", /*short_path=*/""));
+  visitor.Finish();
+  return {};
+}
+
+Result<std::optional<std::reference_wrapper<const SchemaField>>> Schema::FindFieldById(
+    int32_t field_id) const {
+  ICEBERG_RETURN_UNEXPECTED(InitIdToFieldMap());
+  auto it = id_to_field_.find(field_id);
+  if (it == id_to_field_.end()) {
+    return std::nullopt;
+  }
+  return it->second;
+}
+
+IdToFieldVisitor::IdToFieldVisitor(
+    std::unordered_map<int32_t, std::reference_wrapper<const SchemaField>>& id_to_field)
+    : id_to_field_(id_to_field) {}
+
+Status IdToFieldVisitor::Visit(const Type& type) {
+  if (type.is_nested()) {
+    ICEBERG_RETURN_UNEXPECTED(VisitNestedType(type));
+  }
+  return {};
+}
+
+Status IdToFieldVisitor::VisitNestedType(const Type& type) {
+  const auto& nested = iceberg::internal::checked_cast<const NestedType&>(type);
+  const auto& fields = nested.fields();
+  for (const auto& field : fields) {
+    auto it = id_to_field_.try_emplace(field.field_id(), std::cref(field));
+    if (!it.second) {
+      return InvalidSchema("Duplicate field id found: {}", field.field_id());
+    }
+    ICEBERG_RETURN_UNEXPECTED(Visit(*field.type()));
+  }
+  return {};
+}
+
+NameToIdVisitor::NameToIdVisitor(
+    std::unordered_map<std::string, int32_t, string_hash, std::equal_to<>>& name_to_id,
+    bool case_sensitive, std::function<std::string(std::string_view)> quoting_func)
+    : name_to_id_(name_to_id),
+      case_sensitive_(case_sensitive),
+      quoting_func_(std::move(quoting_func)) {}
+
+Status NameToIdVisitor::Visit(const ListType& type, const std::string& path,
+                              const std::string& short_path) {
+  const auto& field = type.fields()[0];
+  std::string new_path = BuildPath(path, field.name(), case_sensitive_);
+  std::string new_short_path;
+  if (field.type()->type_id() == TypeId::kStruct) {
+    new_short_path = short_path;
+  } else {
+    new_short_path = BuildPath(short_path, field.name(), case_sensitive_);
+  }
+  auto it = name_to_id_.try_emplace(new_path, field.field_id());
+  if (!it.second) {
+    return InvalidSchema("Duplicate path found: {}, prev id: {}, curr id: {}",
+                         it.first->first, it.first->second, field.field_id());
+  }
+  short_name_to_id_.try_emplace(new_short_path, field.field_id());
+  ICEBERG_RETURN_UNEXPECTED(
+      VisitTypeInline(*field.type(), this, new_path, new_short_path));
+  return {};
+}
+
+Status NameToIdVisitor::Visit(const MapType& type, const std::string& path,
+                              const std::string& short_path) {
+  std::string new_path, new_short_path;
+  const auto& fields = type.fields();
+  for (const auto& field : fields) {
+    new_path = BuildPath(path, field.name(), case_sensitive_);
+    if (field.name() == MapType::kValueName &&
+        field.type()->type_id() == TypeId::kStruct) {
+      new_short_path = short_path;
+    } else {
+      new_short_path = BuildPath(short_path, field.name(), case_sensitive_);
+    }
+    auto it = name_to_id_.try_emplace(new_path, field.field_id());
+    if (!it.second) {
+      return InvalidSchema("Duplicate path found: {}, prev id: {}, curr id: {}",
+                           it.first->first, it.first->second, field.field_id());
+    }
+    short_name_to_id_.try_emplace(new_short_path, field.field_id());
+    ICEBERG_RETURN_UNEXPECTED(
+        VisitTypeInline(*field.type(), this, new_path, new_short_path));
+  }
+  return {};
+}
+
+Status NameToIdVisitor::Visit(const StructType& type, const std::string& path,
+                              const std::string& short_path) {
+  const auto& fields = type.fields();
+  std::string new_path, new_short_path;
+  for (const auto& field : fields) {
+    new_path = BuildPath(path, field.name(), case_sensitive_);
+    new_short_path = BuildPath(short_path, field.name(), case_sensitive_);
+    auto it = name_to_id_.try_emplace(new_path, field.field_id());
+    if (!it.second) {
+      return InvalidSchema("Duplicate path found: {}, prev id: {}, curr id: {}",
+                           it.first->first, it.first->second, field.field_id());
+    }
+    short_name_to_id_.try_emplace(new_short_path, field.field_id());
+    ICEBERG_RETURN_UNEXPECTED(
+        VisitTypeInline(*field.type(), this, new_path, new_short_path));
+  }
+  return {};
+}
+
+Status NameToIdVisitor::Visit(const PrimitiveType& type, const std::string& path,
+                              const std::string& short_path) {
+  return {};
+}
+
+std::string NameToIdVisitor::BuildPath(std::string_view prefix,
+                                       std::string_view field_name, bool case_sensitive) {
+  std::string quoted_name;
+  if (!quoting_func_) {
+    quoted_name = std::string(field_name);
+  } else {
+    quoted_name = quoting_func_(field_name);
+  }
+  if (case_sensitive) {
+    return prefix.empty() ? quoted_name : std::string(prefix) + "." + quoted_name;
+  }
+  return prefix.empty() ? StringUtils::ToLower(quoted_name)
+                        : std::string(prefix) + "." + StringUtils::ToLower(quoted_name);
+  ;
+}
+
+void NameToIdVisitor::Finish() {
+  for (auto&& it : short_name_to_id_) {
+    name_to_id_.try_emplace(it.first, it.second);
+  }
+}
+
 }  // namespace iceberg

src/iceberg/schema.h

@@ -29,8 +29,10 @@
 #include <vector>
 
 #include "iceberg/iceberg_export.h"
+#include "iceberg/result.h"
 #include "iceberg/schema_field.h"
 #include "iceberg/type.h"
+#include "iceberg/util/string_util.h"
 
 namespace iceberg {
 
@@ -54,13 +56,46 @@ class ICEBERG_EXPORT Schema : public StructType {
 
   [[nodiscard]] std::string ToString() const override;
 
+  /// \brief Find the SchemaField by field name.
+  ///
+  /// Short names for maps and lists are included for any name that does not conflict with
+  /// a canonical name. For example, a list, 'l', of structs with field 'x' will produce
+  /// short name 'l.x' in addition to canonical name 'l.element.x'. a map 'm', if its
+  /// value include a structs with field 'x' wil produce short name 'm.x' in addition to
+  /// canonical name 'm.value.x'
+  /// FIXME: Currently only handles ASCII lowercase conversion; extend to support
+  /// non-ASCII characters (e.g., using std::towlower or ICU)
+  [[nodiscard]] Result<std::optional<std::reference_wrapper<const SchemaField>>>
+  FindFieldByName(std::string_view name, bool case_sensitive = true) const;
+
+  /// \brief Find the SchemaField by field id.
+  [[nodiscard]] Result<std::optional<std::reference_wrapper<const SchemaField>>>
+  FindFieldById(int32_t field_id) const;
+
   friend bool operator==(const Schema& lhs, const Schema& rhs) { return lhs.Equals(rhs); }
 
+ private:
+  /// Mapping from field id to field.
+  mutable std::unordered_map<int32_t, std::reference_wrapper<const SchemaField>>
+      id_to_field_;
+  /// Mapping from field name to field id.
+  mutable std::unordered_map<std::string, int32_t, string_hash, std::equal_to<>>
+      name_to_id_;
+  /// Mapping from lowercased field name to field id
+  mutable std::unordered_map<std::string, int32_t, string_hash, std::equal_to<>>
+      lowercase_name_to_id_;
+
  private:
   /// \brief Compare two schemas for equality.
   [[nodiscard]] bool Equals(const Schema& other) const;
 
   const std::optional<int32_t> schema_id_;
+
+  // TODO(nullccxsy): Address potential concurrency issues in lazy initialization (e.g.,
+  // use std::call_once)
+  Status InitIdToFieldMap() const;
+  Status InitNameToIdMap() const;
+  Status InitLowerCaseNameToIdMap() const;
 };
 
 }  // namespace iceberg

src/iceberg/util/macros.h

@@ -19,10 +19,13 @@
 
 #pragma once
 
-#define ICEBERG_RETURN_UNEXPECTED(result)          \
-  if (!result) [[unlikely]] {                      \
-    return std::unexpected<Error>(result.error()); \
-  }
+#define ICEBERG_RETURN_UNEXPECTED(result)                         \
+  do {                                                            \
+    auto&& iceberg_temp_result = (result);                        \
+    if (!iceberg_temp_result) [[unlikely]] {                      \
+      return std::unexpected<Error>(iceberg_temp_result.error()); \
+    }                                                             \
+  } while (false);
 
 #define ICEBERG_ASSIGN_OR_RAISE_IMPL(result_name, lhs, rexpr) \
   auto&& result_name = (rexpr);                               \

src/iceberg/util/string_util.h

@@ -46,4 +46,15 @@ class ICEBERG_EXPORT StringUtils {
   }
 };
 
+/// \brief Transparent hash function that supports std::string_view as lookup key
+///
+/// Enables std::unordered_map to directly accept std::string_view lookup keys
+/// without creating temporary std::string objects, using C++20's transparent lookup.
+struct string_hash {
+  using hash_type = std::hash<std::string_view>;
+  using is_transparent = void;
+
+  std::size_t operator()(std::string_view str) const { return hash_type{}(str); }
+};
+
 }  // namespace iceberg