[arrow] branch main updated: GH-34703: [Python] Set copy=False explicitly when creating a pandas Series (#34593)

[alenka]

This is an automated email from the ASF dual-hosted git repository.

alenka pushed a commit to branch main
in repository https://gitbox.apache.org/repos/asf/arrow.git


The following commit(s) were added to refs/heads/main by this push:
     new 1140a53f51 GH-34703: [Python] Set copy=False explicitly when creating 
a pandas Series (#34593)
1140a53f51 is described below

commit 1140a53f51a35c37008d97d43b2d00cff4ae2caa
Author: Patrick Hoefler <61934744+ph...@users.noreply.github.com>
AuthorDate: Thu Mar 23 07:07:18 2023 -0400

    GH-34703: [Python] Set copy=False explicitly when creating a pandas Series 
(#34593)
    
    
    
    ### Rationale for this change
    
    pandas will change the default for creating a Series from an array (numpy, 
arrow, ...) to copy=True when Copy-on-Write is enabled. To avoid this when 
using it internally, we have to specify copy=False explicitly.
    
    ### What changes are included in this PR?
    
    Setting copy=False when creating a Series
    
    ### Are these changes tested?
    
    This is equivalent to the current default behavior, so no reason to add any 
additional tests.
    
    ### Are there any user-facing changes?
    
    no
    
    cc @ jorisvandenbossche
    * Closes: #34703
    
    Authored-by: Patrick Hoefler <61934744+ph...@users.noreply.github.com>
    Signed-off-by: Alenka Frim <frim.ale...@gmail.com>
---
 python/pyarrow/array.pxi        | 6 +++---
 python/pyarrow/pandas_compat.py | 4 ++--
 2 files changed, 5 insertions(+), 5 deletions(-)

diff --git a/python/pyarrow/array.pxi b/python/pyarrow/array.pxi
index 4f8f9b6bf9..778d59058f 100644
--- a/python/pyarrow/array.pxi
+++ b/python/pyarrow/array.pxi
@@ -1664,7 +1664,7 @@ cdef _array_like_to_pandas(obj, options, types_mapper):
     else:
         dtype = None
 
-    result = pandas_api.series(arr, dtype=dtype, name=name)
+    result = pandas_api.series(arr, dtype=dtype, name=name, copy=False)
 
     if (isinstance(original_type, TimestampType) and
             original_type.tz is not None and
@@ -3065,7 +3065,7 @@ cdef class ExtensionArray(Array):
         # pandas ExtensionDtype that implements conversion from pyarrow
         if hasattr(pandas_dtype, '__from_arrow__'):
             arr = pandas_dtype.__from_arrow__(self)
-            return pandas_api.series(arr)
+            return pandas_api.series(arr, copy=False)
 
         # otherwise convert the storage array with the base implementation
         return Array._to_pandas(self.storage, options, **kwargs)
@@ -3169,7 +3169,7 @@ cdef object get_values(object obj, bint* is_series):
         result = obj
         is_series[0] = False
     else:
-        result = pandas_api.series(obj).values
+        result = pandas_api.series(obj, copy=False).values
         is_series[0] = False
 
     return result
diff --git a/python/pyarrow/pandas_compat.py b/python/pyarrow/pandas_compat.py
index a6de60e87b..00e8613717 100644
--- a/python/pyarrow/pandas_compat.py
+++ b/python/pyarrow/pandas_compat.py
@@ -1015,9 +1015,9 @@ def _extract_index_level(table, result_table, field_name,
         values = values.copy()
 
     if isinstance(col.type, pa.lib.TimestampType) and col.type.tz is not None:
-        index_level = make_tz_aware(pd.Series(values), col.type.tz)
+        index_level = make_tz_aware(pd.Series(values, copy=False), col.type.tz)
     else:
-        index_level = pd.Series(values, dtype=values.dtype)
+        index_level = pd.Series(values, dtype=values.dtype, copy=False)
     result_table = result_table.remove_column(
         result_table.schema.get_field_index(field_name)
     )