bug: The translate method of strings is not working as expected for pandas

[mesejo]

What happened?

The following code raises a TypeError exception:

import pandas as pd

import ibis

ibis.options.interactive = True
ibis.set_backend("pandas")

df = pd.DataFrame({
    'fruit': ['apple', 'banana', 'orange', 'kiwi', 'mango'],
    'to_column': ['red', 'ted', 'pet', 'bed', 'sed']
})

table = ibis.memtable(df)

result = table.fruit.translate('abc', table.to_column).execute()
print(result)

Note that using a column for Translate is not documented nor implemented for DuckDB, but it is implemented in here for pandas.

What would be the solution? Delete these methods or fix them?

What version of ibis are you using?

dev

What backend(s) are you using, if any?

pandas

Relevant log output

Traceback (most recent call last):
  File "/Users/daniel.mesejo/PycharmProjects/ibis/try_substr.py", line 15, in <module>
    result = table.fruit.translate('abc', table.to_column).execute()
  File "/Users/daniel.mesejo/PycharmProjects/ibis/ibis/expr/types/core.py", line 303, in execute
    return self._find_backend(use_default=True).execute(
  File "/Users/daniel.mesejo/PycharmProjects/ibis/ibis/backends/pandas/__init__.py", line 299, in execute
    return execute_and_reset(node, params=params, **kwargs)
  File "/Users/daniel.mesejo/PycharmProjects/ibis/ibis/backends/pandas/core.py", line 482, in execute_and_reset
    result = execute(
  File "/Users/daniel.mesejo/opt/anaconda3/envs/ibis-dev/lib/python3.10/site-packages/multipledispatch/dispatcher.py", line 278, in __call__
    return func(*args, **kwargs)
  File "/Users/daniel.mesejo/PycharmProjects/ibis/ibis/backends/pandas/trace.py", line 136, in traced_func
    return func(*args, **kwargs)
  File "/Users/daniel.mesejo/PycharmProjects/ibis/ibis/backends/pandas/core.py", line 428, in main_execute
    return execute_with_scope(
  File "/Users/daniel.mesejo/PycharmProjects/ibis/ibis/backends/pandas/core.py", line 218, in execute_with_scope
    result = execute_until_in_scope(
  File "/Users/daniel.mesejo/PycharmProjects/ibis/ibis/backends/pandas/trace.py", line 136, in traced_func
    return func(*args, **kwargs)
  File "/Users/daniel.mesejo/PycharmProjects/ibis/ibis/backends/pandas/core.py", line 350, in execute_until_in_scope
    result = execute_node(
  File "/Users/daniel.mesejo/opt/anaconda3/envs/ibis-dev/lib/python3.10/site-packages/multipledispatch/dispatcher.py", line 278, in __call__
    return func(*args, **kwargs)
  File "/Users/daniel.mesejo/PycharmProjects/ibis/ibis/backends/pandas/trace.py", line 136, in traced_func
    return func(*args, **kwargs)
  File "/Users/daniel.mesejo/PycharmProjects/ibis/ibis/backends/pandas/execution/strings.py", line 336, in execute_series_translate_scalar_series
    table = to_string.map(lambda y, x=from_string: str.maketrans(x=x, y=y))
  File "/Users/daniel.mesejo/opt/anaconda3/envs/ibis-dev/lib/python3.10/site-packages/pandas/core/series.py", line 4393, in map
    new_values = self._map_values(arg, na_action=na_action)
  File "/Users/daniel.mesejo/opt/anaconda3/envs/ibis-dev/lib/python3.10/site-packages/pandas/core/base.py", line 924, in _map_values
    new_values = map_f(values, mapper)
  File "pandas/_libs/lib.pyx", line 2834, in pandas._libs.lib.map_infer
  File "/Users/daniel.mesejo/PycharmProjects/ibis/ibis/backends/pandas/execution/strings.py", line 336, in <lambda>
    table = to_string.map(lambda y, x=from_string: str.maketrans(x=x, y=y))
TypeError: str.maketrans() takes no keyword arguments

Code of Conduct

• I agree to follow this project's Code of Conduct

[cpcloud]

@mesejo Thanks for the issue!

I think we should try to avoid deleting methods without a strong justification or equivalent functionality.

Is the fix for the pandas backend as simple as removing the caller's explicit keyword arguments?

[mesejo]

Removing the caller's explicit keyword arguments makes the function run, but I don't know if the output makes sense. For the original code, it produces:

0     apple
1    banana
2    orange
3      kiwi
4     mango
Name: Translate(fruit, 'abc', to_column), dtype: object

it also passes the pd.Series:

table = to_string.map(lambda y, x=from_string: str.maketrans(x, y))

to Series.str.translate, the expected input is a dictionary made by str.make_trans.

[mesejo]

Just as a side note, translate with two strings as input works:

result = table.fruit.translate('abc', 'efg').execute()
print(result)

Output

0     epple
1    fenene
2    orenge
3      kiwi
4     mengo
Name: Translate(fruit, 'abc', 'efg'), dtype: object