ARROW-18152: [Python] DataFrame Interchange Protocol for pyarrow Table#14613

Closed

AlenkaF wants to merge 29 commits intoapache:masterfrom

AlenkaF:ARROW-18152

Commits on Nov 28, 2022

Initial sceleton for interchange package
AlenkaF
committed
Add a dataframe (PyArrowTableXchg) class methods
AlenkaF
committed
Add a subpackage for testing interchange protocol, add a test for Table.__dataframe__ and do some minor corrections
AlenkaF
committed
Add column (PyArrowColumn) class methods
AlenkaF
committed
Add buffer (PyArrowBuffer) class methods, some changes and main tests
AlenkaF
committed
Make changes to buffer, column and dataframe classes
AlenkaF
committed
Make changes to from_dataframe.py skeleton
AlenkaF
committed
Add extra tests and make minor corrections
AlenkaF
committed
Run linter
AlenkaF
committed
Make changes to the code to make pa.Table -> pd.DataFrame work for int, float with missing values
AlenkaF
committed
Correct linter error and add a check for TypedDict import
AlenkaF
committed
Use len(...) for the size of the pa.Array/pa.ChunkedArray
AlenkaF
committed
Add missing annotations import and remove TypedDict leftover
AlenkaF
committed
Remove bool bit_width check
AlenkaF
committed
Change buffer representation of boolean arrays
AlenkaF
committed
Remove dataframe protocol abstract classes and move the docstrings and necessary defenitions to separate implementation files
AlenkaF
committed
Add missing changes to the class names and references
AlenkaF
committed
Add ColumnNullType = non nullable for columns without missing values
AlenkaF
committed
Correct test error after describe_null() change
AlenkaF
committed
Change DtypeKind to be imported from column.py
AlenkaF
committed
Add change for string dtype and bitmask - not sure about it though
AlenkaF
committed
Add a change for dictionary arrays
AlenkaF
committed
Add corrections for timestamp dtype
AlenkaF
committed
Change size() to size
AlenkaF
committed
Add schema to empty record batch and keep the number of chukes fixed to n_chunks
AlenkaF
committed
Add offset for sliced array with a test and use datetime instead of pandas timestamp in the tests
AlenkaF
committed
Fix linter errors
AlenkaF
committed
Add a skip for the test using from_dataframe() added in pandas versions < 1.5.0
AlenkaF
committed
Make changes to the from_dataframe.py skeleton
AlenkaF
committed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

ARROW-18152: [Python] DataFrame Interchange Protocol for pyarrow Table#14613

ARROW-18152: [Python] DataFrame Interchange Protocol for pyarrow Table#14613
AlenkaF wants to merge 29 commits intoapache:masterfrom
AlenkaF:ARROW-18152

Commits on Nov 28, 2022

Initial sceleton for interchange package

Add a dataframe (PyArrowTableXchg) class methods

Add a subpackage for testing interchange protocol, add a test for Table.dataframe and do some minor corrections

Add column (PyArrowColumn) class methods

Add buffer (PyArrowBuffer) class methods, some changes and main tests

Make changes to buffer, column and dataframe classes

Make changes to from_dataframe.py skeleton

Add extra tests and make minor corrections

Run linter

Make changes to the code to make pa.Table -> pd.DataFrame work for int, float with missing values

Correct linter error and add a check for TypedDict import

Use len(...) for the size of the pa.Array/pa.ChunkedArray

Add missing annotations import and remove TypedDict leftover

Remove bool bit_width check

Change buffer representation of boolean arrays

Remove dataframe protocol abstract classes and move the docstrings and necessary defenitions to separate implementation files

Add missing changes to the class names and references

Add ColumnNullType = non nullable for columns without missing values

Correct test error after describe_null() change

Change DtypeKind to be imported from column.py

Add change for string dtype and bitmask - not sure about it though

Add a change for dictionary arrays

Add corrections for timestamp dtype

Change size() to size

Add schema to empty record batch and keep the number of chukes fixed to n_chunks

Add offset for sliced array with a test and use datetime instead of pandas timestamp in the tests

Fix linter errors

Add a skip for the test using from_dataframe() added in pandas versions < 1.5.0

Make changes to the from_dataframe.py skeleton