Providing information about frequencies present in a given dataset
Currently dataset.json or the API response to /datasets/{provider_code}/{dataset_code}
don't contain standardized information about the frequency of the data in the dataset. In some cases, if there exists a FREQ or FREQUENCY dimension, some information is available, but it is neither standardized (follows correctly the provider's conventions) nor reliable as some providers list more frequencies in the dimension than actually used in the dataset.
If it isn't too costly, we could also compute the number of series for a given frequency.
We need
- A list of all frequencies appearing in DBnomics. A fetcher encountering a new frequency should add to it
- A fetcher must provide a mapping between the code and names of frequencies declared by the provider and the standardized frequencies
- A fetcher must declare a frequency even if the information isn't provided in a dimension. Sometime, the frequency is in the name of the dataset or must be inferred from the date format.
- @frequency should be made explicitly by the fetcher rather than guessed from the periods format by the API