Missing "name" attribute under <Column> class

[adrivn]

After updating to 0.11 and running our pipelines as usual an error printed:

  File "C:\Users\adrivn\Projects\etl\config\functions.py", line 271, in write_to_table
    f'"{col_name}" {col.data_type.ddl()}' for col_name, col in pg_schema.columns
                                              ^^^^^^^^^^^^^
TypeError: cannot unpack non-iterable pgpq._pgpq.Column object

After inspection, I ran a test with dummy data and noticed that the Column class does have "data_type" and "nullable" attributes but "name" has gone missing, for some reason.

import polars as pl
from pgpq import ArrowToPostgresBinaryEncoder

test_sch = pl.from_dict({"a": [1,2,3], "words": ["hey", "test", "this"]})

dataset = test_sch.to_arrow()
encoder = ArrowToPostgresBinaryEncoder(dataset.schema)

pg_schema = encoder.schema()

first = next(iter(pg_schema.columns))
print(type(first), dir(first))

Prints:

<class 'pgpq._pgpq.Column'> ['__class__', '__delattr__', '__dir__', '__doc__', '__eq__', '__format__', '__ge__', '__getattribute__', '__getstate__', '__gt__', '__hash__', '__init__', '__init_subclass__', '__le__', '__lt__', '__module__', '__ne__', '__new__', '__reduce__', '__reduce_ex__', '__repr__', '__setattr__', '__sizeof__', '__str__', '__subclasshook__', 'data_type', 'nullable']

Am I looking at this the wrong way?

[nuno-faria]

I think this is caused by changes introduced in #59. PostgresSchema.columns went from columns: Vec<(String, Column)> to Vec<Column>, hence the first error. However, this was not updated in _pgpq.py, so the type hints won't be correct. As for the second point, there is a name stored in the Column but it is not exposed with a getter, so it also does not work.

@adriangb should PostgresSchema.columns be just a Vec<Column>, with each Column having a getter to the name?

[adriangb]

@adriangb should PostgresSchema.columns be just a Vec<Column>, with each Column having a getter to the name?

I think so - it would carry the same information as Vec<(String, Column)> but avoid a needless tuple.