[python] add all supported dtypes

py-polars/pypolars/datatypes.py

@@ -61,6 +61,62 @@ class Date64:
     pass
 
 
+class Time32Millisecond:
+    pass
+
+
+class Time32Second:
+    pass
+
+
+class Time64Nanosecond:
+    pass
+
+
+class Time64Microsecond:
+    pass
+
+
+class DurationNanosecond:
+    pass
+
+
+class DurationMicrosecond:
+    pass
+
+
+class DurationMillisecond:
+    pass
+
+
+class DurationSecond:
+    pass
+
+
+class IntervalDayTime:
+    pass
+
+
+class IntervalYearMonth:
+    pass
+
+
+class TimestampNanosecond:
+    pass
+
+
+class TimestampMicrosecond:
+    pass
+
+
+class TimestampMillisecond:
+    pass
+
+
+class TimestampSecond:
+    pass
+
+
 # Don't change the order of these!
 dtypes = [
     Int8,
@@ -78,6 +134,20 @@ class Date64:
     LargeList,
     Date32,
     Date64,
+    Time32Millisecond,
+    Time32Second,
+    Time64Nanosecond,
+    Time64Microsecond,
+    DurationNanosecond,
+    DurationMicrosecond,
+    DurationMillisecond,
+    DurationSecond,
+    IntervalDayTime,
+    IntervalYearMonth,
+    TimestampNanosecond,
+    TimestampMicrosecond,
+    TimestampMillisecond,
+    TimestampSecond,
 ]
 DTYPE_TO_FFINAME = {
     Int8: "i8",
@@ -92,9 +162,23 @@ class Date64:
     Float64: "f64",
     Bool: "bool",
     Utf8: "str",
-    LargeList: "largelist",
+    LargeList: "large_list",
     Date32: "date32",
     Date64: "date64",
+    Time32Millisecond: "time32_millisecond",
+    Time32Second: "time32_second",
+    Time64Nanosecond: "time64_nanosecond",
+    Time64Microsecond: "time64_microsecond",
+    DurationNanosecond: "duration_nanosecond",
+    DurationMicrosecond: "duration_microsecond",
+    DurationMillisecond: "duration_millisecond",
+    DurationSecond: "duration_second",
+    IntervalDayTime: "interval_daytime",
+    IntervalYearMonth: "interval_yearmonth",
+    TimestampNanosecond: "timestamp_nanosecond",
+    TimestampMicrosecond: "timestamp_microsecond",
+    TimestampMillisecond: "timestamp_millisecond",
+    TimestampSecond: "timestamp_second",
 }
 
 

py-polars/pypolars/frame.py

@@ -92,7 +92,7 @@ def from_parquet(
     @staticmethod
     def from_ipc(file: Union[str, BinaryIO]) -> DataFrame:
         """
-        Read into a DataFrame from Arrow IPC stream format.
+        Read into a DataFrame from Arrow IPC stream format. This is also called the feather format.
 
         Parameters
         ----------
@@ -107,9 +107,25 @@ def from_ipc(file: Union[str, BinaryIO]) -> DataFrame:
         self._df = PyDataFrame.from_ipc(file)
         return self
 
+    @staticmethod
+    def from_feather(file: Union[str, BinaryIO]) -> DataFrame:
+        """
+        Read into a DataFrame from Arrow IPC stream format. This is also called the feather format.
+
+        Parameters
+        ----------
+        file
+            Path to a file or a file like object.
+
+        Returns
+        -------
+        DataFrame
+        """
+        return DataFrame.from_ipc(file)
+
     def to_csv(
         self,
-        path: str,
+        file: Union[TextIO, str],
         batch_size: int = 100000,
         has_headers: bool = True,
         delimiter: str = ",",
@@ -119,7 +135,7 @@ def to_csv(
 
         Parameters
         ----------
-        path
+        file
             write location
         batch_size
             Size of the write buffer. Increase to have faster io.
@@ -128,20 +144,33 @@ def to_csv(
         delimiter
             Space elements with this symbol.
         """
-        self._df.to_csv(path, batch_size, has_headers, ord(delimiter))
+        self._df.to_csv(file, batch_size, has_headers, ord(delimiter))
+
+    def to_ipc(self, file: Union[BinaryIO, str], batch_size):
+        """
+        Write to Arrow IPC binary stream, or a feather file.
+
+        Parameters
+        ----------
+        file
+            write location
+        batch_size
+            Size of the write buffer. Increase to have faster io.
+        """
+        self._df.to_ipc(file, batch_size)
 
-    def to_ipc(self, path: str, batch_size):
+    def to_feather(self, file: Union[BinaryIO, str], batch_size):
         """
-        Write to Arrow IPC binary stream.
+        Write to Arrow IPC binary stream, or a feather file.
 
         Parameters
         ----------
-        path
+        file
             write location
         batch_size
             Size of the write buffer. Increase to have faster io.
         """
-        self._df.to_ipc(path, batch_size)
+        self.to_ipc(file, batch_size)
 
     def __str__(self) -> str:
         return self._df.as_str()

py-polars/src/datatypes.rs

@@ -1,4 +1,4 @@
-use polars::datatypes::ArrowDataType;
+use polars::datatypes::{ArrowDataType, IntervalUnit, TimeUnit};
 
 // Don't change the order of these!
 #[repr(u8)]
@@ -18,6 +18,20 @@ pub enum DataType {
     LargeList,
     Date32,
     Date64,
+    Time32Millisecond,
+    Time32Second,
+    Time64Nanosecond,
+    Time64Microsecond,
+    DurationNanosecond,
+    DurationMicrosecond,
+    DurationMillisecond,
+    DurationSecond,
+    IntervalDayTime,
+    IntervalYearMonth,
+    TimestampNanosecond,
+    TimestampMicrosecond,
+    TimestampMillisecond,
+    TimestampSecond,
 }
 
 impl From<&ArrowDataType> for DataType {
@@ -39,6 +53,20 @@ impl From<&ArrowDataType> for DataType {
             ArrowDataType::LargeList(_) => LargeList,
             ArrowDataType::Date32(_) => Date32,
             ArrowDataType::Date64(_) => Date64,
+            ArrowDataType::Time32(TimeUnit::Millisecond) => Time32Millisecond,
+            ArrowDataType::Time32(TimeUnit::Second) => Time32Second,
+            ArrowDataType::Time64(TimeUnit::Nanosecond) => Time64Nanosecond,
+            ArrowDataType::Time64(TimeUnit::Microsecond) => Time64Microsecond,
+            ArrowDataType::Interval(IntervalUnit::DayTime) => IntervalDayTime,
+            ArrowDataType::Interval(IntervalUnit::YearMonth) => IntervalYearMonth,
+            ArrowDataType::Duration(TimeUnit::Nanosecond) => DurationNanosecond,
+            ArrowDataType::Duration(TimeUnit::Microsecond) => DurationMicrosecond,
+            ArrowDataType::Duration(TimeUnit::Millisecond) => DurationMillisecond,
+            ArrowDataType::Duration(TimeUnit::Second) => DurationSecond,
+            ArrowDataType::Timestamp(TimeUnit::Nanosecond, _) => TimestampNanosecond,
+            ArrowDataType::Timestamp(TimeUnit::Microsecond, _) => TimestampMicrosecond,
+            ArrowDataType::Timestamp(TimeUnit::Millisecond, _) => TimestampMillisecond,
+            ArrowDataType::Timestamp(TimeUnit::Second, _) => TimestampSecond,
             dt => panic!(format!("datatype: {:?} not supported", dt)),
         }
     }