Python - Endpoint : Can the function not return an array of dictionaries ?

[mathmari]

Hey!

I am trying to write a code for a server in Python and I got stuck.

I gave as an input a csv file and using pandas we get a dictionary where the titles are the keys and the inputs are the values.

From that we get the below :

I have written the below endpoint to get all the islands that belog to the specific municipality :

@app.route("/municipality/<municipality>",methods=["GET"])
def get_island_municipality(municipality): 
    dict_to_send = {} 
    dict_to_send["data"] = get_islands_municipality(municipality)
    dict_to_send["message"]=f'These are the islands that belong to {municipality}'
    return jsonify(dict_to_send)

and the function I have used is the below one :

def get_islands_municipality(municipality): 
    islands = []  
    for i in range(len(dataframe["Municipality"])) : 
        islands_dict = {} 
        if dataframe["Municipality"][i] == municipality :  
            islands_dict["Island's Name"] = dataframe["Island's Name"][i] 
            islands_dict["Municipality"] = dataframe["Municipality"][i] 
            islands_dict["Amount of Hotels"] = dataframe["Amount of Hotels"][i]  
            islands_dict["Amount of Beaches"] = dataframe["Amount of Beaches"][i]  
            islands.append(islands_dict)
    return islands

but I get an error, I suppose that the error relates to the type of my return. I return an array of dictionaries, right? Is that not allowed? The error that I get is "TypeError: Object of type int64 is not JSON serializable".

:unsure:

 

[Jameson]

I had to look this up but I found a couple threads on StackExhange which reference this error and they both say that this is due to the int64 type. That is a numpy type, so the fix seems to be converting it to a Python int before dumping to JSON. I would try modifying your function to something like this maybe.

def get_islands_municipality(municipality):
    islands = [] 
    for i in range(len(dataframe["Municipality"])) :
        islands_dict = {}
        if dataframe["Municipality"][i] == municipality : 
            islands_dict["Island's Name"] = dataframe["Island's Name"][i]
            islands_dict["Municipality"] = dataframe["Municipality"][i]
            islands_dict["Amount of Hotels"] = int(dataframe["Amount of Hotels"][i])
            islands_dict["Amount of Beaches"] = int(dataframe["Amount of Beaches"][i])
            islands.append(islands_dict)
    return islands

 

[mathmari]

I had to look this up but I found a couple threads on StackExhange which reference this error and they both say that this is due to the int64 type. That is a numpy type, so the fix seems to be converting it to a Python int before dumping to JSON. I would try modifying your function to something like this maybe.

def get_islands_municipality(municipality):
    islands = []
    for i in range(len(dataframe["Municipality"])) :
        islands_dict = {}
        if dataframe["Municipality"][i] == municipality :
            islands_dict["Island's Name"] = dataframe["Island's Name"][i]
            islands_dict["Municipality"] = dataframe["Municipality"][i]
            islands_dict["Amount of Hotels"] = int(dataframe["Amount of Hotels"][i])
            islands_dict["Amount of Beaches"] = int(dataframe["Amount of Beaches"][i])
            islands.append(islands_dict)
    return islands

Now it works properly! :geek: Does this happen because in the cvs filrthe numbers are not defined to beof type int? Or why does this happen?
So every time we have numbers in a csv file and we want to return these values we have to make them int first? :unsure:

 

[Jameson]

I think so unfortunately. Here are the types that the JSON format can work with and it looks like pandas has default behavior to use int64 and float64. You could write a function to loop over your columns and column types and convert any int64 dtypes to int as well as float64 to float.