feat: add visit type support (#94)

Add a `TypeId` field to each concrete type to enable macro-based type
visitor support.

---------

Signed-off-by: Junwang Zhao <[email protected]>

Author: zhjwpku

LICENSE

@@ -267,3 +267,17 @@ https://github.com/aappleby/smhasher/blob/master/src/MurmurHash3.cpp
 
 MurmurHash3 was written by Austin Appleby, and is placed in the public
 domain. The author disclaims copyright to this source code.
+
+--------------------------------------------------------------------------------
+
+The file src/iceberg/util/checked_cast.h contains code adapted from
+
+https://github.com/apache/arrow/blob/main/cpp/src/arrow/util/checked_cast.h
+
+The file src/iceberg/util/visit_type.h contains code adapted from
+
+https://github.com/apache/arrow/blob/main/cpp/src/arrow/visit_type_inline.h
+
+Copyright: 2016-2025 The Apache Software Foundation.
+Home page: https://arrow.apache.org/
+License: https://www.apache.org/licenses/LICENSE-2.0

NOTICE

@@ -12,3 +12,7 @@ This product includes code from smhasher
  * MurmurHash3 was written by Austin Appleby, and is placed in the public
  * domain. The author hereby disclaims copyright to this source code.
  * https://github.com/aappleby/smhasher
+
+This product includes code from Apache Arrow
+ * Copyright 2016-2025 The Apache Software Foundation
+ * https://github.com/apache/arrow

src/iceberg/CMakeLists.txt

@@ -38,7 +38,8 @@ set(ICEBERG_SOURCES
     transform_function.cc
     type.cc
     util/murmurhash3_internal.cc
-    util/timepoint.cc)
+    util/timepoint.cc
+    util/unreachable.cc)
 
 set(ICEBERG_STATIC_BUILD_INTERFACE_LIBS)
 set(ICEBERG_SHARED_BUILD_INTERFACE_LIBS)

src/iceberg/type.cc

@@ -41,7 +41,7 @@ StructType::StructType(std::vector<SchemaField> fields) : fields_(std::move(fiel
   }
 }
 
-TypeId StructType::type_id() const { return TypeId::kStruct; }
+TypeId StructType::type_id() const { return kTypeId; }
 std::string StructType::ToString() const {
   std::string repr = "struct<\n";
   for (const auto& field : fields_) {
@@ -93,7 +93,7 @@ ListType::ListType(SchemaField element) : element_(std::move(element)) {
 ListType::ListType(int32_t field_id, std::shared_ptr<Type> type, bool optional)
     : element_(field_id, std::string(kElementName), std::move(type), optional) {}
 
-TypeId ListType::type_id() const { return TypeId::kList; }
+TypeId ListType::type_id() const { return kTypeId; }
 std::string ListType::ToString() const {
   // XXX: work around Clang/libc++: "<{}>" in a format string appears to get
   // parsed as {<>} or something; split up the format string to avoid that
@@ -146,7 +146,7 @@ MapType::MapType(SchemaField key, SchemaField value)
 
 const SchemaField& MapType::key() const { return fields_[0]; }
 const SchemaField& MapType::value() const { return fields_[1]; }
-TypeId MapType::type_id() const { return TypeId::kMap; }
+TypeId MapType::type_id() const { return kTypeId; }
 std::string MapType::ToString() const {
   // XXX: work around Clang/libc++: "<{}>" in a format string appears to get
   // parsed as {<>} or something; split up the format string to avoid that
@@ -192,33 +192,25 @@ bool MapType::Equals(const Type& other) const {
   return fields_ == map.fields_;
 }
 
-TypeId BooleanType::type_id() const { return TypeId::kBoolean; }
+TypeId BooleanType::type_id() const { return kTypeId; }
 std::string BooleanType::ToString() const { return "boolean"; }
-bool BooleanType::Equals(const Type& other) const {
-  return other.type_id() == TypeId::kBoolean;
-}
+bool BooleanType::Equals(const Type& other) const { return other.type_id() == kTypeId; }
 
-TypeId IntType::type_id() const { return TypeId::kInt; }
+TypeId IntType::type_id() const { return kTypeId; }
 std::string IntType::ToString() const { return "int"; }
-bool IntType::Equals(const Type& other) const { return other.type_id() == TypeId::kInt; }
+bool IntType::Equals(const Type& other) const { return other.type_id() == kTypeId; }
 
-TypeId LongType::type_id() const { return TypeId::kLong; }
+TypeId LongType::type_id() const { return kTypeId; }
 std::string LongType::ToString() const { return "long"; }
-bool LongType::Equals(const Type& other) const {
-  return other.type_id() == TypeId::kLong;
-}
+bool LongType::Equals(const Type& other) const { return other.type_id() == kTypeId; }
 
-TypeId FloatType::type_id() const { return TypeId::kFloat; }
+TypeId FloatType::type_id() const { return kTypeId; }
 std::string FloatType::ToString() const { return "float"; }
-bool FloatType::Equals(const Type& other) const {
-  return other.type_id() == TypeId::kFloat;
-}
+bool FloatType::Equals(const Type& other) const { return other.type_id() == kTypeId; }
 
-TypeId DoubleType::type_id() const { return TypeId::kDouble; }
+TypeId DoubleType::type_id() const { return kTypeId; }
 std::string DoubleType::ToString() const { return "double"; }
-bool DoubleType::Equals(const Type& other) const {
-  return other.type_id() == TypeId::kDouble;
-}
+bool DoubleType::Equals(const Type& other) const { return other.type_id() == kTypeId; }
 
 DecimalType::DecimalType(int32_t precision, int32_t scale)
     : precision_(precision), scale_(scale) {
@@ -230,57 +222,47 @@ DecimalType::DecimalType(int32_t precision, int32_t scale)
 
 int32_t DecimalType::precision() const { return precision_; }
 int32_t DecimalType::scale() const { return scale_; }
-TypeId DecimalType::type_id() const { return TypeId::kDecimal; }
+TypeId DecimalType::type_id() const { return kTypeId; }
 std::string DecimalType::ToString() const {
   return std::format("decimal({}, {})", precision_, scale_);
 }
 bool DecimalType::Equals(const Type& other) const {
-  if (other.type_id() != TypeId::kDecimal) {
+  if (other.type_id() != kTypeId) {
     return false;
   }
   const auto& decimal = static_cast<const DecimalType&>(other);
   return precision_ == decimal.precision_ && scale_ == decimal.scale_;
 }
 
-TypeId DateType::type_id() const { return TypeId::kDate; }
+TypeId DateType::type_id() const { return kTypeId; }
 std::string DateType::ToString() const { return "date"; }
-bool DateType::Equals(const Type& other) const {
-  return other.type_id() == TypeId::kDate;
-}
+bool DateType::Equals(const Type& other) const { return other.type_id() == kTypeId; }
 
-TypeId TimeType::type_id() const { return TypeId::kTime; }
+TypeId TimeType::type_id() const { return kTypeId; }
 std::string TimeType::ToString() const { return "time"; }
-bool TimeType::Equals(const Type& other) const {
-  return other.type_id() == TypeId::kTime;
-}
+bool TimeType::Equals(const Type& other) const { return other.type_id() == kTypeId; }
 
 bool TimestampType::is_zoned() const { return false; }
 TimeUnit TimestampType::time_unit() const { return TimeUnit::kMicrosecond; }
-TypeId TimestampType::type_id() const { return TypeId::kTimestamp; }
+TypeId TimestampType::type_id() const { return kTypeId; }
 std::string TimestampType::ToString() const { return "timestamp"; }
-bool TimestampType::Equals(const Type& other) const {
-  return other.type_id() == TypeId::kTimestamp;
-}
+bool TimestampType::Equals(const Type& other) const { return other.type_id() == kTypeId; }
 
 bool TimestampTzType::is_zoned() const { return true; }
 TimeUnit TimestampTzType::time_unit() const { return TimeUnit::kMicrosecond; }
-TypeId TimestampTzType::type_id() const { return TypeId::kTimestampTz; }
+TypeId TimestampTzType::type_id() const { return kTypeId; }
 std::string TimestampTzType::ToString() const { return "timestamptz"; }
 bool TimestampTzType::Equals(const Type& other) const {
-  return other.type_id() == TypeId::kTimestampTz;
+  return other.type_id() == kTypeId;
 }
 
-TypeId StringType::type_id() const { return TypeId::kString; }
+TypeId StringType::type_id() const { return kTypeId; }
 std::string StringType::ToString() const { return "string"; }
-bool StringType::Equals(const Type& other) const {
-  return other.type_id() == TypeId::kString;
-}
+bool StringType::Equals(const Type& other) const { return other.type_id() == kTypeId; }
 
-TypeId UuidType::type_id() const { return TypeId::kUuid; }
+TypeId UuidType::type_id() const { return kTypeId; }
 std::string UuidType::ToString() const { return "uuid"; }
-bool UuidType::Equals(const Type& other) const {
-  return other.type_id() == TypeId::kUuid;
-}
+bool UuidType::Equals(const Type& other) const { return other.type_id() == kTypeId; }
 
 FixedType::FixedType(int32_t length) : length_(length) {
   if (length < 0) {
@@ -289,20 +271,18 @@ FixedType::FixedType(int32_t length) : length_(length) {
 }
 
 int32_t FixedType::length() const { return length_; }
-TypeId FixedType::type_id() const { return TypeId::kFixed; }
+TypeId FixedType::type_id() const { return kTypeId; }
 std::string FixedType::ToString() const { return std::format("fixed({})", length_); }
 bool FixedType::Equals(const Type& other) const {
-  if (other.type_id() != TypeId::kFixed) {
+  if (other.type_id() != kTypeId) {
     return false;
   }
   const auto& fixed = static_cast<const FixedType&>(other);
   return length_ == fixed.length_;
 }
 
-TypeId BinaryType::type_id() const { return TypeId::kBinary; }
+TypeId BinaryType::type_id() const { return kTypeId; }
 std::string BinaryType::ToString() const { return "binary"; }
-bool BinaryType::Equals(const Type& other) const {
-  return other.type_id() == TypeId::kBinary;
-}
+bool BinaryType::Equals(const Type& other) const { return other.type_id() == kTypeId; }
 
 }  // namespace iceberg

src/iceberg/type.h

@@ -104,6 +104,7 @@ class ICEBERG_EXPORT NestedType : public Type {
 /// \brief A data type representing a struct with nested fields.
 class ICEBERG_EXPORT StructType : public NestedType {
  public:
+  constexpr static TypeId kTypeId = TypeId::kStruct;
   explicit StructType(std::vector<SchemaField> fields);
   ~StructType() override = default;
 
@@ -128,6 +129,7 @@ class ICEBERG_EXPORT StructType : public NestedType {
 /// \brief A data type representing a list of values.
 class ICEBERG_EXPORT ListType : public NestedType {
  public:
+  constexpr static const TypeId kTypeId = TypeId::kList;
   constexpr static const std::string_view kElementName = "element";
 
   /// \brief Construct a list of the given element.  The name of the child
@@ -157,6 +159,7 @@ class ICEBERG_EXPORT ListType : public NestedType {
 /// \brief A data type representing a dictionary of values.
 class ICEBERG_EXPORT MapType : public NestedType {
  public:
+  constexpr static const TypeId kTypeId = TypeId::kMap;
   constexpr static const std::string_view kKeyName = "key";
   constexpr static const std::string_view kValueName = "value";
 
@@ -194,6 +197,8 @@ class ICEBERG_EXPORT MapType : public NestedType {
 /// \brief A data type representing a boolean (true or false).
 class ICEBERG_EXPORT BooleanType : public PrimitiveType {
  public:
+  constexpr static const TypeId kTypeId = TypeId::kBoolean;
+
   BooleanType() = default;
   ~BooleanType() override = default;
 
@@ -207,6 +212,8 @@ class ICEBERG_EXPORT BooleanType : public PrimitiveType {
 /// \brief A data type representing a 32-bit signed integer.
 class ICEBERG_EXPORT IntType : public PrimitiveType {
  public:
+  constexpr static const TypeId kTypeId = TypeId::kInt;
+
   IntType() = default;
   ~IntType() override = default;
 
@@ -220,6 +227,8 @@ class ICEBERG_EXPORT IntType : public PrimitiveType {
 /// \brief A data type representing a 64-bit signed integer.
 class ICEBERG_EXPORT LongType : public PrimitiveType {
  public:
+  constexpr static const TypeId kTypeId = TypeId::kLong;
+
   LongType() = default;
   ~LongType() override = default;
 
@@ -234,6 +243,8 @@ class ICEBERG_EXPORT LongType : public PrimitiveType {
 ///   float.
 class ICEBERG_EXPORT FloatType : public PrimitiveType {
  public:
+  constexpr static const TypeId kTypeId = TypeId::kFloat;
+
   FloatType() = default;
   ~FloatType() override = default;
 
@@ -248,6 +259,8 @@ class ICEBERG_EXPORT FloatType : public PrimitiveType {
 ///   float.
 class ICEBERG_EXPORT DoubleType : public PrimitiveType {
  public:
+  constexpr static const TypeId kTypeId = TypeId::kDouble;
+
   DoubleType() = default;
   ~DoubleType() override = default;
 
@@ -261,6 +274,7 @@ class ICEBERG_EXPORT DoubleType : public PrimitiveType {
 /// \brief A data type representing a fixed-precision decimal.
 class ICEBERG_EXPORT DecimalType : public PrimitiveType {
  public:
+  constexpr static const TypeId kTypeId = TypeId::kDecimal;
   constexpr static const int32_t kMaxPrecision = 38;
 
   /// \brief Construct a decimal type with the given precision and scale.
@@ -288,6 +302,8 @@ class ICEBERG_EXPORT DecimalType : public PrimitiveType {
 ///   timezone or time.
 class ICEBERG_EXPORT DateType : public PrimitiveType {
  public:
+  constexpr static const TypeId kTypeId = TypeId::kDate;
+
   DateType() = default;
   ~DateType() override = default;
 
@@ -302,6 +318,8 @@ class ICEBERG_EXPORT DateType : public PrimitiveType {
 ///   reference to a timezone or date.
 class ICEBERG_EXPORT TimeType : public PrimitiveType {
  public:
+  constexpr static const TypeId kTypeId = TypeId::kTime;
+
   TimeType() = default;
   ~TimeType() override = default;
 
@@ -326,6 +344,8 @@ class ICEBERG_EXPORT TimestampBase : public PrimitiveType {
 ///   reference to a timezone.
 class ICEBERG_EXPORT TimestampType : public TimestampBase {
  public:
+  constexpr static const TypeId kTypeId = TypeId::kTimestamp;
+
   TimestampType() = default;
   ~TimestampType() override = default;
 
@@ -343,6 +363,8 @@ class ICEBERG_EXPORT TimestampType : public TimestampBase {
 ///   epoch in UTC.  A time zone or offset is not stored.
 class ICEBERG_EXPORT TimestampTzType : public TimestampBase {
  public:
+  constexpr static const TypeId kTypeId = TypeId::kTimestampTz;
+
   TimestampTzType() = default;
   ~TimestampTzType() override = default;
 
@@ -359,6 +381,8 @@ class ICEBERG_EXPORT TimestampTzType : public TimestampBase {
 /// \brief A data type representing an arbitrary-length byte sequence.
 class ICEBERG_EXPORT BinaryType : public PrimitiveType {
  public:
+  constexpr static const TypeId kTypeId = TypeId::kBinary;
+
   BinaryType() = default;
   ~BinaryType() override = default;
 
@@ -373,6 +397,8 @@ class ICEBERG_EXPORT BinaryType : public PrimitiveType {
 ///   (encoded in UTF-8).
 class ICEBERG_EXPORT StringType : public PrimitiveType {
  public:
+  constexpr static const TypeId kTypeId = TypeId::kString;
+
   StringType() = default;
   ~StringType() override = default;
 
@@ -386,6 +412,8 @@ class ICEBERG_EXPORT StringType : public PrimitiveType {
 /// \brief A data type representing a fixed-length bytestring.
 class ICEBERG_EXPORT FixedType : public PrimitiveType {
  public:
+  constexpr static const TypeId kTypeId = TypeId::kFixed;
+
   /// \brief Construct a fixed type with the given length.
   explicit FixedType(int32_t length);
   ~FixedType() override = default;
@@ -407,6 +435,8 @@ class ICEBERG_EXPORT FixedType : public PrimitiveType {
 ///   it is effectively a fixed(16).
 class ICEBERG_EXPORT UuidType : public PrimitiveType {
  public:
+  constexpr static const TypeId kTypeId = TypeId::kUuid;
+
   UuidType() = default;
   ~UuidType() override = default;