Add summary function and command

[amontanez24]

No description provided.

[csala]

I added a few suggestions to improve the current implementation and interface

[csala]

I would make a few changes to this function signature:

• I would rename file_path to raw_csv_path (or results_csv_path)
• I would get the output path as an input argument that would default to None. If an output path is given, use it, otherwise write the output as re.sub('.csv$', '.xlsx', raw_csv_path)
• I would make the baseline a module-level dictionary, with data modalities as keys and baseline lists as values and also add a baselines argument which would default to None (meaning use the default one we just defined).

At the implementation level, I would make a sub-function called _make_summary which would receive:

• A name (which will be the data modality)
• The data already filtered by data modality
• The baselines list (list, not dict!)
• The writer

The overall process would be something similar to this:

def _add_summary(data, modality, baselines, writer):
    # Compute the summaries on the data and add the corresponding
    # sheets to the writer, using `{Sheet name} ({modality})` as names

def make_summary_spreadsheet(raw_csv_path, output_path=None, baselines=None):
    # Preprocess the data
    # Create and configure the writer
    for modality, subset in data.groupby('modality'):
        modality_baselines = baselines[modality]
        _add_summary(subset, modality, modality_baselines, writer)

    writer.save()

[csala]

@amontanez24 we should also add support to read from and write to S3 buckets, following an approach similar to the collect command.
This means we need to add aws_key and aws_secret arguments, and that we need to use the s3 module functions to read and write files (we may need to change something there to write the xlsx file?)

[csala]

If the results.py module is not being used anywhere else (which I think is not), I would copy whatever is necessary here and remove the module.

[csala]

Comment about how to decide whether to write to S3 or not, and how to do it.

[csala]

aws_key and aws_secret cannot be used to decide whether we are writing in S3 or not, since an S3 path can be passed without keys to let boto3 use the system-wide credentials.

To keep things simple, I would just not distinguish between a local or an S3 path here and always write to the BytesIO and pass it down to write_file, which will decide whether to write to a local or remote file based on the given path.

[amontanez24]

I see, I didn't know that. This is nice though since it will be cleaner in the end

[csala]

Good to go! Thanks @amontanez24 !