copy of custom index does not align with original #7162

Activity

1. dcherian

   added

   bug

   topic-indexing

   on Oct 14, 2022
2. headtr1ck commented on Oct 14, 2022

   

   headtr1ck

   on Oct 14, 2022

   Collaborator

   More actions

   I don't think there is anything wrong with the deep copy implementation. Only before it was not deep copying the indexes and now it is.
   Can you check if index1.equals(index2) works if they are created the same?
3. headtr1ck commented on Oct 15, 2022

   

   headtr1ck

   on Oct 15, 2022

   Collaborator

   More actions

   Now that I think about it, Indexes.copy_indexes might also require some update that includes the memo argument.
   But not sure if that will solve the issue here.
4. benbovy commented on Oct 18, 2022

   

   benbovy

   on Oct 18, 2022

   Last edited by benbovy

   Member

   More actions

   Indexes.copy_indexes might also require some update that includes the memo argument. But not sure if that will solve the issue here.

   That's a possible cause. Alignment may fail early because .xindexes returns different mappings of coordinates vs. index objects. It's worth checking if after copying the dataset, copy.xindexes returns the same CRSIndex object for its "x", "y" and "spatial_ref" coordinates.

   EDIT: checking copy.xindexes.group_by_index() is more convenient.
5. benbovy commented on Oct 18, 2022

   

   benbovy

   on Oct 18, 2022

   Member

   More actions

   The refactored alignment logic could be improved (cf. #7002). The error raised in the method below is not very helpful.

   xarray/xarray/core/alignment.py

   Lines 294 to 333 in ab726c5

   	def assert_no_index_conflict(self) -> None:
   	"""Check for uniqueness of both coordinate and dimension names across all sets
   	of matching indexes.
   	We need to make sure that all indexes used for re-indexing or alignment
   	are fully compatible and do not conflict each other.
   	Note: perhaps we could choose less restrictive constraints and instead
   	check for conflicts among the dimension (position) indexers returned by
   	`Index.reindex_like()` for each matching pair of object index / aligned
   	index?
   	(ref: https://github.com/pydata/xarray/issues/1603#issuecomment-442965602)
   	"""
   	matching_keys = set(self.all_indexes) | set(self.indexes)
   	coord_count: dict[Hashable, int] = defaultdict(int)
   	dim_count: dict[Hashable, int] = defaultdict(int)
   	for coord_names_dims, _ in matching_keys:
   	dims_set: set[Hashable] = set()
   	for name, dims in coord_names_dims:
   	coord_count[name] += 1
   	dims_set.update(dims)
   	for dim in dims_set:
   	dim_count[dim] += 1
   	for count, msg in [(coord_count, "coordinates"), (dim_count, "dimensions")]:
   	dup = {k: v for k, v in count.items() if v > 1}
   	if dup:
   	items_msg = ", ".join(
   	f"{k!r} ({v} conflicting indexes)" for k, v in dup.items()
   	)
   	raise ValueError(
   	"cannot re-index or align objects with conflicting indexes found for "
   	f"the following {msg}: {items_msg}\n"
   	"Conflicting indexes may occur when\n"
   	"- they relate to different sets of coordinate and/or dimension names\n"
   	"- they don't have the same type\n"
   	"- they may be used to reindex data along common dimensions"
   	)
6. headtr1ck commented on Oct 25, 2022

   

   headtr1ck

   on Oct 25, 2022

   Collaborator

   More actions

   Could you create a MVCE without a custom index that I can use for debugging?
7. scottyhq commented on Mar 24, 2023

   

   scottyhq

   on Mar 24, 2023

   Last edited by scottyhq

   Contributor

   More actions

   Reviving this exploration with xarray : 2023.3.0 and I'm no longer seeing the traceback linked above, so I think this can be closed

   Original traceback from xr.align(copy, newds), which now succeeds:

   `File ~/mambaforge/envs/xarray-release/lib/python3.10/site-packages/xarray/core/indexes.py:490, in PandasIndex.equals(self, other)
       488 if not isinstance(other, PandasIndex):
       489     return False
   --> 490 return self.index.equals(other.index) and self.dim == other.dim
   
   AttributeError: 'PandasIndex' object has no attribute 'index'
8. dcherian commented on Mar 24, 2023

   

   dcherian

   on Mar 24, 2023

   ContributorAuthor

   More actions

   Thanks @scottyhq
9. dcherian

   closed this as completedon Mar 24, 2023