ARROW-15134: [GLib] Add GArrow{Month,DayTime,MonthDayNano}IntervalDat…

…aType

Closes #11975 from okadakk/arrow-glib-add-tinterval-data-type

Lead-authored-by: okadakk <k.suke.jp1990@gmail.com>
Co-authored-by: Sutou Kouhei <kou@clear-code.com>
Signed-off-by: Sutou Kouhei <kou@clear-code.com>

c_glib/arrow-glib/basic-data-type.cpp

@@ -97,6 +97,18 @@ G_BEGIN_DECLS
  * #GArrowTime64DataType is a class for the number of microseconds or
  * nanoseconds since midnight in the 64-bit signed integer data type.
  *
+ * #GArrowIntervalDataType is an abstract class for interval related
+ * data type such as #GArrowMonthIntervalDataType.
+ *
+ * #GArrowMonthIntervalDataType is a class for the month intarval data
+ * type.
+ *
+ * #GArrowDayTimeIntervalDataType is a class for the day time intarval
+ * data type.
+ *
+ * #GArrowMonthDayNanoIntervalDataType is a class for the month day
+ * nano intarval data type.
+ *
  * #GArrowDecimalDataType is a base class for the decimal data types.
  *
  * #GArrowDecimal128DataType is a class for the 128-bit decimal data type.
@@ -1287,6 +1299,140 @@ garrow_time64_data_type_new(GArrowTimeUnit unit, GError **error)
 }
 
 
+G_DEFINE_ABSTRACT_TYPE(GArrowIntervalDataType,
+                       garrow_interval_data_type,
+                       GARROW_TYPE_TEMPORAL_DATA_TYPE)
+
+static void
+garrow_interval_data_type_init(GArrowIntervalDataType *object)
+{
+}
+
+static void
+garrow_interval_data_type_class_init(GArrowIntervalDataTypeClass *klass)
+{
+}
+
+/**
+ * garrow_interval_data_type_get_interval_type:
+ * @type: The #GArrowIntervalDataType.
+ *
+ * Returns: The interval type of the given @type.
+ *
+ * Since: 7.0.0
+ */
+GArrowIntervalType
+garrow_interval_data_type_get_interval_type(GArrowIntervalDataType *type)
+{
+  const auto arrow_data_type = garrow_data_type_get_raw(GARROW_DATA_TYPE(type));
+  const auto arrow_interval_type =
+    std::static_pointer_cast<arrow::IntervalType>(arrow_data_type);
+  return garrow_interval_type_from_raw(arrow_interval_type->interval_type());
+}
+
+
+G_DEFINE_TYPE(GArrowMonthIntervalDataType,
+              garrow_month_interval_data_type,
+              GARROW_TYPE_INTERVAL_DATA_TYPE)
+static void
+garrow_month_interval_data_type_init(GArrowMonthIntervalDataType *object)
+{
+}
+
+static void
+garrow_month_interval_data_type_class_init(
+  GArrowMonthIntervalDataTypeClass *klass)
+{
+}
+
+/**
+ * garrow_month_interval_data_type_new:
+ *
+ * Returns: The newly created month interval data type.
+ *
+ * Since: 7.0.0
+ */
+GArrowMonthIntervalDataType *
+garrow_month_interval_data_type_new(void)
+{
+  auto arrow_data_type = arrow::month_interval();
+
+  auto data_type = g_object_new(GARROW_TYPE_MONTH_INTERVAL_DATA_TYPE,
+                                "data-type", &arrow_data_type,
+                                NULL);
+  return GARROW_MONTH_INTERVAL_DATA_TYPE(data_type);
+}
+
+
+G_DEFINE_TYPE(GArrowDayTimeIntervalDataType,
+              garrow_day_time_interval_data_type,
+              GARROW_TYPE_INTERVAL_DATA_TYPE)
+
+static void
+garrow_day_time_interval_data_type_init(GArrowDayTimeIntervalDataType *object)
+{
+}
+
+static void
+garrow_day_time_interval_data_type_class_init(
+  GArrowDayTimeIntervalDataTypeClass *klass)
+{
+}
+
+/**
+ * garrow_day_time_interval_data_type_new:
+ *
+ * Returns: The newly created day time interval data type.
+ *
+ * Since: 7.0.0
+ */
+GArrowDayTimeIntervalDataType *
+garrow_day_time_interval_data_type_new(void)
+{
+  auto arrow_data_type = arrow::day_time_interval();
+
+  auto data_type = g_object_new(GARROW_TYPE_DAY_TIME_INTERVAL_DATA_TYPE,
+                                "data-type", &arrow_data_type,
+                                NULL);
+  return GARROW_DAY_TIME_INTERVAL_DATA_TYPE(data_type);
+}
+
+
+G_DEFINE_TYPE(GArrowMonthDayNanoIntervalDataType,
+              garrow_month_day_nano_interval_data_type,
+              GARROW_TYPE_INTERVAL_DATA_TYPE)
+
+static void
+garrow_month_day_nano_interval_data_type_init(
+  GArrowMonthDayNanoIntervalDataType *object)
+{
+}
+
+static void
+garrow_month_day_nano_interval_data_type_class_init(
+  GArrowMonthDayNanoIntervalDataTypeClass *klass)
+{
+}
+
+/**
+ * garrow_month_day_nano_interval_data_type_new:
+ *
+ * Returns: The newly created month day nano interval data type.
+ *
+ * Since: 7.0.0
+ */
+GArrowMonthDayNanoIntervalDataType *
+garrow_month_day_nano_interval_data_type_new(void)
+{
+  auto arrow_data_type = arrow::month_day_nano_interval();
+
+  auto data_type = g_object_new(GARROW_TYPE_MONTH_DAY_NANO_INTERVAL_DATA_TYPE,
+                                "data-type", &arrow_data_type,
+                                NULL);
+  return GARROW_MONTH_DAY_NANO_INTERVAL_DATA_TYPE(data_type);
+}
+
+
 G_DEFINE_ABSTRACT_TYPE(GArrowDecimalDataType,
                        garrow_decimal_data_type,
                        GARROW_TYPE_FIXED_SIZE_BINARY_DATA_TYPE)

c_glib/arrow-glib/basic-data-type.h

@@ -457,6 +457,73 @@ GArrowTime64DataType *garrow_time64_data_type_new      (GArrowTimeUnit unit,
                                                         GError **error);
 
 
+#define GARROW_TYPE_INTERVAL_DATA_TYPE (garrow_interval_data_type_get_type())
+G_DECLARE_DERIVABLE_TYPE(GArrowIntervalDataType,
+                         garrow_interval_data_type,
+                         GARROW,
+                         INTERVAL_DATA_TYPE,
+                         GArrowTimeDataType)
+struct _GArrowIntervalDataTypeClass
+{
+  GArrowTimeDataTypeClass parent_class;
+};
+
+GARROW_AVAILABLE_IN_7_0
+GArrowIntervalType
+garrow_interval_data_type_get_interval_type(GArrowIntervalDataType *type);
+
+
+#define GARROW_TYPE_MONTH_INTERVAL_DATA_TYPE   \
+  (garrow_month_interval_data_type_get_type())
+G_DECLARE_DERIVABLE_TYPE(GArrowMonthIntervalDataType,
+                         garrow_month_interval_data_type,
+                         GARROW,
+                         MONTH_INTERVAL_DATA_TYPE,
+                         GArrowIntervalDataType)
+struct _GArrowMonthIntervalDataTypeClass
+{
+  GArrowIntervalDataTypeClass parent_class;
+};
+
+GARROW_AVAILABLE_IN_7_0
+GArrowMonthIntervalDataType *
+garrow_month_interval_data_type_new(void);
+
+
+#define GARROW_TYPE_DAY_TIME_INTERVAL_DATA_TYPE         \
+  (garrow_day_time_interval_data_type_get_type())
+G_DECLARE_DERIVABLE_TYPE(GArrowDayTimeIntervalDataType,
+                         garrow_day_time_interval_data_type,
+                         GARROW,
+                         DAY_TIME_INTERVAL_DATA_TYPE,
+                         GArrowIntervalDataType)
+struct _GArrowDayTimeIntervalDataTypeClass
+{
+  GArrowIntervalDataTypeClass parent_class;
+};
+
+GARROW_AVAILABLE_IN_7_0
+GArrowDayTimeIntervalDataType *
+garrow_day_time_interval_data_type_new(void);
+
+
+#define GARROW_TYPE_MONTH_DAY_NANO_INTERVAL_DATA_TYPE \
+  (garrow_month_day_nano_interval_data_type_get_type())
+G_DECLARE_DERIVABLE_TYPE(GArrowMonthDayNanoIntervalDataType,
+                         garrow_month_day_nano_interval_data_type,
+                         GARROW,
+                         MONTH_DAY_NANO_INTERVAL_DATA_TYPE,
+                         GArrowIntervalDataType)
+struct _GArrowMonthDayNanoIntervalDataTypeClass
+{
+  GArrowIntervalDataTypeClass parent_class;
+};
+
+GARROW_AVAILABLE_IN_7_0
+GArrowMonthDayNanoIntervalDataType *
+garrow_month_day_nano_interval_data_type_new(void);
+
+
 #define GARROW_TYPE_DECIMAL_DATA_TYPE (garrow_decimal_data_type_get_type())
 G_DECLARE_DERIVABLE_TYPE(GArrowDecimalDataType,
                          garrow_decimal_data_type,

c_glib/arrow-glib/type.cpp

@@ -60,12 +60,8 @@ garrow_type_from_raw(arrow::Type::type type)
     return GARROW_TYPE_DOUBLE;
   case arrow::Type::type::STRING:
     return GARROW_TYPE_STRING;
-  case arrow::Type::type::LARGE_STRING:
-    return GARROW_TYPE_LARGE_STRING;
   case arrow::Type::type::BINARY:
     return GARROW_TYPE_BINARY;
-  case arrow::Type::type::LARGE_BINARY:
-    return GARROW_TYPE_LARGE_BINARY;
   case arrow::Type::type::FIXED_SIZE_BINARY:
     return GARROW_TYPE_FIXED_SIZE_BINARY;
   case arrow::Type::type::DATE32:
@@ -79,29 +75,39 @@ garrow_type_from_raw(arrow::Type::type type)
   case arrow::Type::type::TIME64:
     return GARROW_TYPE_TIME64;
   case arrow::Type::type::INTERVAL_MONTHS:
-    return GARROW_TYPE_INTERVAL_MONTHS;
+    return GARROW_TYPE_MONTH_INTERVAL;
   case arrow::Type::type::INTERVAL_DAY_TIME:
-    return GARROW_TYPE_INTERVAL_DAY_TIME;
+    return GARROW_TYPE_DAY_TIME_INTERVAL;
   case arrow::Type::type::DECIMAL128:
     return GARROW_TYPE_DECIMAL128;
   case arrow::Type::type::DECIMAL256:
     return GARROW_TYPE_DECIMAL256;
   case arrow::Type::type::LIST:
     return GARROW_TYPE_LIST;
-  case arrow::Type::type::LARGE_LIST:
-    return GARROW_TYPE_LARGE_LIST;
   case arrow::Type::type::STRUCT:
     return GARROW_TYPE_STRUCT;
-  case arrow::Type::type::MAP:
-    return GARROW_TYPE_MAP;
-  case arrow::Type::type::EXTENSION:
-    return GARROW_TYPE_EXTENSION;
   case arrow::Type::type::SPARSE_UNION:
     return GARROW_TYPE_SPARSE_UNION;
   case arrow::Type::type::DENSE_UNION:
     return GARROW_TYPE_DENSE_UNION;
   case arrow::Type::type::DICTIONARY:
     return GARROW_TYPE_DICTIONARY;
+  case arrow::Type::type::MAP:
+    return GARROW_TYPE_MAP;
+  case arrow::Type::type::EXTENSION:
+    return GARROW_TYPE_EXTENSION;
+  case arrow::Type::type::FIXED_SIZE_LIST:
+    return GARROW_TYPE_FIXED_SIZE_LIST;
+  case arrow::Type::type::DURATION:
+    return GARROW_TYPE_DURATION;
+  case arrow::Type::type::LARGE_STRING:
+    return GARROW_TYPE_LARGE_STRING;
+  case arrow::Type::type::LARGE_BINARY:
+    return GARROW_TYPE_LARGE_BINARY;
+  case arrow::Type::type::LARGE_LIST:
+    return GARROW_TYPE_LARGE_LIST;
+  case arrow::Type::type::INTERVAL_MONTH_DAY_NANO:
+    return GARROW_TYPE_MONTH_DAY_NANO_INTERVAL;
   default:
     return GARROW_TYPE_NA;
   }
@@ -140,3 +146,18 @@ garrow_time_unit_to_raw(GArrowTimeUnit unit)
     return arrow::TimeUnit::type::SECOND;
   }
 }
+
+GArrowIntervalType
+garrow_interval_type_from_raw(arrow::IntervalType::type type)
+{
+  switch (type) {
+  case arrow::IntervalType::type::MONTHS:
+    return GARROW_INTERVAL_TYPE_MONTH;
+  case arrow::IntervalType::type::DAY_TIME:
+    return GARROW_INTERVAL_TYPE_DAY_TIME;
+  case arrow::IntervalType::type::MONTH_DAY_NANO:
+    return GARROW_INTERVAL_TYPE_MONTH_DAY_NANO;
+  default:
+    return GARROW_INTERVAL_TYPE_MONTH;
+  }
+}

c_glib/arrow-glib/type.h

@@ -48,8 +48,8 @@ G_BEGIN_DECLS
  *   Default unit millisecond.
  * @GARROW_TYPE_TIME32: Exact time encoded with int32, supporting seconds or milliseconds
  * @GARROW_TYPE_TIME64: Exact time encoded with int64, supporting micro- or nanoseconds
- * @GARROW_TYPE_INTERVAL_MONTHS: YEAR_MONTH interval in SQL style.
- * @GARROW_TYPE_INTERVAL_DAY_TIME: DAY_TIME interval in SQL style.
+ * @GARROW_TYPE_MONTH_INTERVAL: YEAR_MONTH interval in SQL style.
+ * @GARROW_TYPE_DAY_TIME_INTERVAL: DAY_TIME interval in SQL style.
  * @GARROW_TYPE_DECIMAL128: Precision- and scale-based decimal
  *   type with 128-bit. Storage type depends on the parameters.
  * @GARROW_TYPE_DECIMAL256: Precision- and scale-based decimal
@@ -67,6 +67,7 @@ G_BEGIN_DECLS
  * @GARROW_TYPE_LARGE_STRING: 64bit offsets UTF-8 variable-length string.
  * @GARROW_TYPE_LARGE_BINARY: 64bit offsets Variable-length bytes (no guarantee of UTF-8-ness).
  * @GARROW_TYPE_LARGE_LIST: A list of some logical data type with 64-bit offsets.
+ * @GARROW_TYPE_MONTH_DAY_NANO_INTERVAL: MONTH_DAY_NANO interval in SQL style.
  *
  * They are corresponding to `arrow::Type::type` values.
  */
@@ -92,8 +93,8 @@ typedef enum {
   GARROW_TYPE_TIMESTAMP,
   GARROW_TYPE_TIME32,
   GARROW_TYPE_TIME64,
-  GARROW_TYPE_INTERVAL_MONTHS,
-  GARROW_TYPE_INTERVAL_DAY_TIME,
+  GARROW_TYPE_MONTH_INTERVAL,
+  GARROW_TYPE_DAY_TIME_INTERVAL,
   GARROW_TYPE_DECIMAL128,
   GARROW_TYPE_DECIMAL256,
   GARROW_TYPE_LIST,
@@ -107,7 +108,8 @@ typedef enum {
   GARROW_TYPE_DURATION,
   GARROW_TYPE_LARGE_STRING,
   GARROW_TYPE_LARGE_BINARY,
-  GARROW_TYPE_LARGE_LIST
+  GARROW_TYPE_LARGE_LIST,
+  GARROW_TYPE_MONTH_DAY_NANO_INTERVAL,
 } GArrowType;
 
 /**
@@ -126,4 +128,22 @@ typedef enum {
   GARROW_TIME_UNIT_NANO
 } GArrowTimeUnit;
 
+/**
+ * GArrowIntervalType:
+ * @GARROW_INTERVAL_TYPE_MONTH: A number of months.
+ * @GARROW_INTERVAL_TYPE_DAY_TIME: A number of days and
+ *   milliseconds (fraction of day).
+ * @GARROW_INTERVAL_TYPE_MONTH_DAY_NANO: A number of months, days and
+ *   nanoseconds between two dates.
+ *
+ * They are corresponding to `arrow::IntervalType::type` values.
+ *
+ * Since 7.0.0
+ */
+typedef enum {
+  GARROW_INTERVAL_TYPE_MONTH,
+  GARROW_INTERVAL_TYPE_DAY_TIME,
+  GARROW_INTERVAL_TYPE_MONTH_DAY_NANO,
+} GArrowIntervalType;
+
 G_END_DECLS

c_glib/arrow-glib/type.hpp

@@ -26,3 +26,4 @@
 GArrowType garrow_type_from_raw(arrow::Type::type type);
 GArrowTimeUnit garrow_time_unit_from_raw(arrow::TimeUnit::type unit);
 arrow::TimeUnit::type garrow_time_unit_to_raw(GArrowTimeUnit unit);
+GArrowIntervalType garrow_interval_type_from_raw(arrow::IntervalType::type type);