Github user BryanCutler commented on a diff in the pull request:
https://github.com/apache/spark/pull/18664#discussion_r144167503
--- Diff: python/pyspark/sql/types.py ---
@@ -1619,11 +1619,47 @@ def to_arrow_type(dt):
arrow_type = pa.decimal(dt.precision, dt.scale)
elif type(dt) == StringType:
arrow_type = pa.string()
+ elif type(dt) == DateType:
+ arrow_type = pa.date32()
+ elif type(dt) == TimestampType:
+ arrow_type = pa.timestamp('us', tz='UTC')
else:
raise TypeError("Unsupported type in conversion to Arrow: " +
str(dt))
return arrow_type
+def _check_localize_series_timestamps(s):
+ from pandas.types.common import is_datetime64_dtype
+ # TODO: handle nested timestamps?
+ if is_datetime64_dtype(s.dtype):
+ # TODO: pyarrow.Column.to_pandas keeps data in UTC but removes
timezone
--- End diff --
It seems like `pyarrow.Column.to_pandas()` produces a different timestamp
series than `pyarrow.Table.to_pandas()` to get a DataFrame and then accessing
the timestamp column. The former keeps timestamps in UTC and removes the
timezone, so a different conversion was required to get the data right. Not
sure if this is an Arrow bug or not, need to look into it more.
---
---------------------------------------------------------------------
To unsubscribe, e-mail: reviews-unsubscr...@spark.apache.org
For additional commands, e-mail: reviews-h...@spark.apache.org
