2 - Use IDs instead of label values for category mapping?

[IdrissaD]

For now mapping is done like this:

"Theme": {
                "Flore": "Faune et flore",
                "Faune": "Faune et flore",
                "Géologie": "Eau et géologie",
                "Point de vue": "Forêt",
                "Eau": "Eau et géologie",
                "Histoire et patrimoine": "Histoire et culture",
                "Savoir-faire": "Histoire et culture",
                "Agropastoralisme": "Agriculture et élevage",
            },

As foreign keys in the API are 90% of the time only displayed as IDs in the main objects route (Trek, POI...), this functionment needs additional API requests in order to retrieve the value corresponding to those IDs.

Pros:

• no additional API requests;
• simplify mapping functions;
• robust to category label changes ("MTB" to "Mountain Bike" for instance), thus reduces maintenance load.

Cons:

• risk of false category data without error raised (if categories IDs change, or if the whole meaning of a category changes, e.g. "MTB" to "Canoe") whereas label value mapping would raise an error to any change.

[IdrissaD]

Idea: use ID to do the mapping, but keep the acutal functionment in order to raise warnings when categories labels change.
source_cat_to_gag_cat would be like this:

"PNE": {
    "Practice": {
        2: {
            "source_label": "MTB",
            "gag_label": "Mountainbike",
            "gag_id": "5"
        }
    }
}

If source_label is different from the label of the category with the same id in fk_api_values, a Warning is raised. It allows to be aware of and log this information somewhere, while not blocking the aggregation each time a category's spelling changes.

[IdrissaD]

Selected solution: IDs are used in the GAG app mapping, and categories' labels from last aggregation are displayed in the mapping web app for human control.