BUG: GH38801 avoid attribute error with pyarrow >=0.16.0 and <1.0.0 (#38803)

Co-authored-by: Joris Van den Bossche <jorisvandenbossche@gmail.com>

Author: Ada Draginda

ci/deps/actions-37-locale.yaml

@@ -30,7 +30,7 @@ dependencies:
  - openpyxl
  - pandas-gbq
  - google-cloud-bigquery>=1.27.2 # GH 36436
- - pyarrow>=0.17
+ - pyarrow=0.17 # GH 38803
  - pytables>=3.5.1
  - scipy
  - xarray=0.12.3

doc/source/whatsnew/v1.2.1.rst

@@ -37,6 +37,7 @@ I/O
 
 - Bumped minimum fastparquet version to 0.4.0 to avoid ``AttributeError`` from numba (:issue:`38344`)
 - Bumped minimum pymysql version to 0.8.1 to avoid test failures (:issue:`38344`)
+- Fixed ``AttributeError`` with PyArrow versions [0.16.0, 1.0.0) (:issue:`38801`)
 
 -
 -

pandas/core/arrays/string_arrow.py

@@ -30,13 +30,12 @@
 except ImportError:
  pa = None
 else:
- # our min supported version of pyarrow, 0.15.1, does not have a compute
- # module
- try:
+ # PyArrow backed StringArrays are available starting at 1.0.0, but this
+ # file is imported from even if pyarrow is < 1.0.0, before pyarrow.compute
+ # and its compute functions existed. GH38801
+ if LooseVersion(pa.__version__) >= "1.0.0":
  import pyarrow.compute as pc
- except ImportError:
- pass
- else:
+
  ARROW_CMP_FUNCS = {
  "eq": pc.equal,
  "ne": pc.not_equal,

pandas/tests/io/test_parquet.py

@@ -898,7 +898,7 @@ def test_timezone_aware_index(self, pa, timezone_aware_date_list):
  # this use-case sets the resolution to 1 minute
  check_round_trip(df, pa, check_dtype=False)
 
- @td.skip_if_no("pyarrow", min_version="0.17")
+ @td.skip_if_no("pyarrow", min_version="1.0.0")
  def test_filter_row_groups(self, pa):
  # https://github.com/pandas-dev/pandas/issues/26551
  df = pd.DataFrame({"a": list(range(0, 3))})