viirya commented on code in PR #50430:
URL: https://github.com/apache/arrow/pull/50430#discussion_r3639216870
##########
python/pyarrow/array.pxi:
##########
@@ -3618,18 +3611,65 @@ cdef class MapArray(ListArray):
Concrete class for Arrow arrays of a map data type.
"""
- cdef object _getitem_py(self, int64_t i):
+ cdef object _getitem_py(self, int64_t i, object maps_as_pydicts):
cdef CListArray* arr = <CListArray*> self.ap
+ cdef bint as_dicts = maps_as_pydicts is not None
+ if as_dicts and maps_as_pydicts != "lossy" and maps_as_pydicts !=
"strict":
+ # Matches MapScalar.as_py, which validates before the null check.
+ raise ValueError(
+ "Invalid value for 'maps_as_pydicts': "
+ + "valid values are 'lossy', 'strict' or `None` (default). "
+ + f"Received {maps_as_pydicts!r}."
+ )
if arr.IsNull(i):
return None
if self._children_cache is None:
self._children_cache = (self.keys, self.items)
cdef Array keys = <Array> (<tuple> self._children_cache)[0]
cdef Array items = <Array> (<tuple> self._children_cache)[1]
cdef int64_t j, start = arr.value_offset(i), end = arr.value_offset(i
+ 1)
- # Matches MapScalar.as_py with the default maps_as_pydicts=None:
- # an association list of (key, value) tuples.
- return [(keys._getitem_py(j), items._getitem_py(j)) for j in
range(start, end)]
+ if not as_dicts:
+ # Matches MapScalar.as_py with the default maps_as_pydicts=None:
+ # an association list of (key, value) tuples.
+ return [
+ (keys._getitem_py(j, None), items._getitem_py(j,
maps_as_pydicts))
+ for j in range(start, end)
+ ]
+ # Convert all keys first (as MapScalar.as_py does via keys()) and
+ # detect duplicates before converting any value, so that the 'lossy'
+ # warnings and the 'strict' KeyError are emitted at the same point as
+ # in MapScalar.as_py even when a later value conversion raises.
+ cdef int64_t count = end - start
+ cdef list keys_py = [keys._getitem_py(j, None) for j in range(start,
end)]
+ cdef dict result = {}
+ cdef int64_t k
+ cdef bint no_dups
+ try:
+ no_dups = len(set(keys_py)) == count
+ except TypeError:
+ # Unhashable keys (e.g. struct or list keys): the per-key loop
+ # below reproduces MapScalar.as_py exactly, raising TypeError at
+ # the same membership test as the Scalar path does.
+ no_dups = False
+ if no_dups:
+ for k in range(count):
+ result[keys_py[k]] = items._getitem_py(start + k,
maps_as_pydicts)
+ return result
+ # Duplicate or unhashable keys: per-key loop matching MapScalar.as_py
+ # exactly.
+ for k in range(count):
+ key = keys_py[k]
+ if key in result:
+ if maps_as_pydicts == "strict":
+ raise KeyError(
+ "Converting to Python dictionary is not supported in
strict mode "
+ f"when duplicate keys are present (duplicate key was
'{key}')."
+ )
+ else:
+ warnings.warn(
+ f"Encountered key '{key}' which was already
encountered.")
+ result[key] = items._getitem_py(start + k, maps_as_pydicts)
+ return result
Review Comment:
You're right, and it measures that way too — benchmarked both variants on
1M-row maps (best of 5, `lossy`): unique str keys ×3: 223→178 ms (-20%), unique
int keys ×3: 142→122 ms (-14%), str keys ×8: 118→112 ms, with duplicates:
118→103 ms (-12%). The temporary set plus the extra traversal cost more than
the membership tests. Applied your suggestion in e309a6f768 — it also removes
the unhashable-key special case, since the per-key loop handles that naturally
at the same point as `MapScalar.as_py`. Thanks!
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
