Option to write uncompressed catalogs

[aulemahal]

Is your feature request related to a problem? Please describe.
The default csv file written by cat.serialize is compressed with gzip. This is quite cool for large catalogs, but it looses some interoperability with low-level editing tools (text editors) as it's now a binary file.

Describe the solution you'd like
An option to control the compression of the catalog file. It could even be interesting for some to control the codec used? Like a compression argument, the same as pd.read_csv, but without the "infer" option, since the catalog filename is not passed explicitly. So one of { ‘gzip’, ‘bz2’, ‘zip’, ‘xz’, None}.

Describe alternatives you've considered
Serializing the catalog myself with cat.df.to_csv, but that's just not ideal.

[aulemahal]

EDIT: In fact, there could simply be a **csv_kwargs passed directly to df.to_csv?
I am ok with changing the default value of compression and injecting index=False, I just think it could be interesting to have finer control.

[andersy005]

EDIT: In fact, there could simply be a **csv_kwargs passed directly to df.to_csv?
I just think it could be interesting to have finer control.

👍🏽

How about adding a to_csv_kwargs argument to

intake-esm/intake_esm/core.py

Lines 362 to 367 in 40fe3a7

	@pydantic.validate_arguments
	def serialize(
	self,
	name: pydantic.StrictStr,
	directory: typing.Union[pydantic.DirectoryPath, pydantic.StrictStr] = None,
	catalog_type: str = 'dict',

and passing these keyword arguments down to

intake-esm/intake_esm/cat.py

Line 118 in 40fe3a7

def save(self, name: str, *, directory: str = None, catalog_type: str = 'dict') -> None:

?