Bigquery UDF bq data move to memorystore by customizable query

Question

I would like to create a bigquery udf to transport the data from bigquery to memorystore. To do this I think I should follow 3 steps:

1. def get_column_from_table(keys) .... I should bring my main key here from bigquery

2. def get_field_value(field, value) if the 3rd element is an array, it should be broken into as many elements as the array has and added as many elements to the memorystore. Here I should reverse the process

   def get_field_value(field, value):

    element= get_column_from_table(key)
    keys = list(element.keys())
    uniq_keys = list(dict.fromkeys([x.partition(":")[0] for x in keys]))
    result = {}
    for uniq in uniq_keys:
        value = None
        for key in keys:
            if key.partition(":")[0] == uniq:
                if value is not None:
                    value = value + "\n" + element[key]
                else:
                    value = element[key]
        result[uniq] = value
    if field in result.keys():
        return result[field]
    else:
        return ""
   

3. def_post(): ....

x=hset(...)

It is not very clear to me how I should take the first step. Also, I intend for the query to be of the following type in the cloud function: select .....(key, field) If I only enter the key to retrieve all my data (and at the same time I should treat the data from the table that are null to be ignored) or to be able to add a certain column. Any sugestion/documentation/example please?