Simple Download/Upload of Airtable Data into/from Python using Airtable API

airtable_helper.py

import requests
import json
import pandas as pd


def airtable_download(table, params_dict={}, api_key=None, base_id=None, record_id=None):
    """Makes a request to Airtable for all records from a single table.
        Returns data in dictionary format.

    Keyword Arguments:
    • table: set to table name
        ◦ see: https://support.airtable.com/hc/en-us/articles/360021333094#table
    • params_dict: desired parameters in dictionary format {parameter : value}
        ◦ example: {"maxRecords" : 20, "view" : "Grid view"}
        ◦ see "List Records" in API Documentation (airtable.com/api)
    • api_key: retrievable at https://airtable.com/account
        ◦ looks like "key●●●●●●●●●●●●●●"
    • base_id: retrievable at https://airtable.com/api for specific base
        ◦ looks like "app●●●●●●●●●●●●●●"
    • record_id: optional for single record lookups
        ◦ looks like "rec●●●●●●●●●●●●●●"
        """

    # Authorization Credentials
    if api_key is None:
        print("Enter Airtable API key. \n  *Find under Airtable Account Overview: https://airtable.com/account")
        api_key = input()
    headers = {"Authorization": "Bearer {}".format(api_key)}
    validate_airtable_kwargs(api_key, "API key", "key")

    # Locate Base
    if base_id is None:
        print("Enter Airtable Base ID. \n  *Find under Airtable API Documentation: https://airtable.com/api for specific base")
        base_id = input()
    url = 'https://api.airtable.com/v0/{}/'.format(base_id)
    path = url + table
    validate_airtable_kwargs(base_id, "Base ID", "app")

    # Validate Record ID
    if record_id is not None:
        validate_airtable_kwargs(record_id, "Record ID", "rec")

    # Format parameters for request
    constant_params = ()
    for parameter in params_dict:
        constant_params += ((parameter, params_dict[parameter]),)
    params = constant_params

    # Start with blank list of records
    airtable_records = []

    # Retrieve multiple records
    if record_id is None:
        run = True
        while run is True:
            response = requests.get(path, params=params, headers=headers)
            airtable_response = response.json()

            try:
                airtable_records += (airtable_response['records'])
            except:
                if 'error' in airtable_response:
                    identify_errors(airtable_response)
                    return airtable_response

            if 'offset' in airtable_response:
                run = True
                params = (('offset', airtable_response['offset']),) + constant_params
            else:
                run = False

    # Retrieve single record
    if record_id is not None:
        if params_dict != {}:
            print("⚠️ Caution: parameters are redundant for single record lookups. Consider removing `params_dict` argument.")
        path = "{}/{}".format(path, record_id)
        response = requests.get(path, headers=headers)
        airtable_response = response.json()

        if 'error' in airtable_response:
            identify_errors(airtable_response)
            return airtable_response

        airtable_records = [airtable_response]

    return airtable_records


def convert_to_dataframe(airtable_records):
    """Converts dictionary output from airtable_download() into a Pandas dataframe."""
    airtable_rows = []
    airtable_index = []
    for record in airtable_records:
        airtable_rows.append(record['fields'])
        airtable_index.append(record['id'])
    airtable_dataframe = pd.DataFrame(airtable_rows, index=airtable_index)
    return airtable_dataframe


def create_field_matching_dict(airtable_records, value_field, key_field = None, swap_pairs = False):
    """Uses airtable_download() output to create a dictionary that matches field values from
    the same record together. Useful for keeping track of relational data.
    
    If second_field is `None`, then the dictionary pairs will be {<record id>:value_field}.
    Otherwise, the dictionary pairx will be {key_field:value_field}.
    If swap_pairs is True, then dictionary pairs will be {value_field:<record id>(or key_field)}.
    """
    airtable_dict = {}
    for airtable_record in airtable_records:
        if key_field == None:
            key = airtable_record['id']
        else:
            key = airtable_record['fields'].get(key_field)
        value = airtable_record['fields'].get(value_field)
        if swap_pairs:
            airtable_dict.update({key : value})
        else:
            airtable_dict.update({value : key})
    return airtable_dict


def airtable_upload(table, upload_data, typecast = False, api_key = None, base_id = None, record_id = None):
    """Sends dictionary data to Airtable to add or update a record in a given table. 
        Returns new or updated record in dictionary format.
    
    Keyword arguments:
    • table: set to table name
        ◦ see: https://support.airtable.com/hc/en-us/articles/360021333094#table
    • upload_data: a dictionary of fields and corresponding values to upload in format {field : value}
        ◦ example: {"Fruit" : "Apple", "Quantity" : 20}
    • typecast: if set to true, Airtable will attempt "best-effort automatic data conversion from string values"
        • see: "Create Records" or "Update Records" in API Documentation, available at https://airtable.com/api for specific base
    • api_key: retrievable at https://airtable.com/account
        ◦ looks like "key●●●●●●●●●●●●●●"
    • base_id: retrievable at https://airtable.com/api for specific base
        ◦ looks like "app●●●●●●●●●●●●●●"
    • record_id: when included function will update specified record will be rather than creating a new record
        ◦ looks like "rec●●●●●●●●●●●●●●"
        """
    
    # Authorization Credentials
    if api_key == None:
        print("Enter Airtable API key. \n  *Find under Airtable Account Overview: https://airtable.com/account")
        api_key = input()
    headers = {"Authorization" : "Bearer {}".format(api_key),
              'Content-Type': 'application/json'}
    validate_airtable_kwargs(api_key, "API key", "key")

    # Locate Base
    if base_id == None:
        print("Enter Airtable Base ID. \n  *Find under Airtable API Documentation: https://airtable.com/api for specific base]")
        base_id = input()
    url = 'https://api.airtable.com/v0/{}/'.format(base_id)
    path = url + table
    validate_airtable_kwargs(base_id, "Base ID", "app")
    
    # Validate Record ID
    if record_id != None:
        validate_airtable_kwargs(record_id, "Record ID", "rec")
    
    # Validate upload_data
    if type(upload_data) != dict:
        print("❌ Error: `upload_data` is not a dictonary.")
        return

    # Create New Record
    if record_id == None:
        upload_dict = {"records": [{"fields" : upload_data}], "typecast" : typecast}
        upload_json = json.dumps(upload_dict)
        response = requests.post(path, data=upload_json, headers=headers)
        airtable_response = response.json()

    # Update Record
    if record_id != None:
        path = "{}/{}".format(path, record_id)
        upload_dict = {"fields" : upload_data, "typecast" : True}
        upload_json = json.dumps(upload_dict)
        response = requests.patch(path, data=upload_json, headers=headers)
        airtable_response = response.json()
    
    # Identify Errors
    if 'error' in airtable_response:
        identify_errors(airtable_response)
        
    return airtable_response


def upload_pandas_dataframe(pandas_dataframe, table, api_key, base_id):
    """Uploads a Pandas dataframe to Airtable. If Pandas index values are Airtable Record IDs, will attempt to update 
        record. Otherwise, will create new records."""
    pandas_dicts = pandas_dataframe.to_dict(orient="index")
    for pandas_dict in pandas_dicts:
        record_id = pandas_dict
        if validate_airtable_kwargs(str(record_id), "Record ID", "rec", print_messages=False) is False:
            record_id = None
        upload_data = pandas_dicts[pandas_dict]
        airtable_upload(table, upload_data, api_key=api_key, base_id=base_id, record_id=record_id)
    return


# Troubleshooting Functions
def validate_airtable_kwargs(kwarg, kwarg_name, prefix, char_length=17, print_messages=True):
    """Designed for use with airtable_download() and airtable_upload() functions.
        Checks `api_key`, `base_id` and `record_id` arguments to see if they conform to the expected Airtable API format.
        """
    valid_status = True
    if len(kwarg) != char_length:
        if print_messages is True:
            print("⚠️ Caution: {} not standard length. Make sure API key is {} characters long.".format(kwarg_name, char_length))
        valid_status = False
    if kwarg.startswith(prefix) is False:
        if print_messages is True:
            print("⚠️ Caution: {} doesn't start with `{}`.".format(kwarg_name, prefix))
        valid_status = False
    return valid_status


def identify_errors(airtable_response):
    """Designed for use with airtable_download() and airtable_upload() functions.
        Prints error responses from the Airtable API in an easy-to-read format.
        """
    if 'error' in airtable_response:
        try:
            print('❌ {} error: "{}"'.format(airtable_response['error']['type'], airtable_response['error']['message']))
        except:
            print("❌ Error: {}".format(airtable_response['error']))
    return

cc: @Airtable

Check out my tutorial for downloading Airtable data into Python on Towards Data Science.

Hi Kaleb
I was hoping if there is a tutorial link or sample step by step python code for updating/inserting/deleting rows in airtable iteratively as well as 1 time (similar to your code and tutorial for download)
-sanket zambre

Hi Kaleb
I was hoping if there is a tutorial link or sample step by step python code for updating/inserting/deleting rows in airtable iteratively as well as 1 time (similar to your code and tutorial for download)
-sanket zambre

Hi @sanky23 – thanks for your inquiry. You can preview a draft version of the step-by-step tutorial for uploading here at this link: https://medium.com/@kalebnyquist/airtable-python-made-possible-uploading-data-into-airtable-from-python-using-airtables-api-3075009abf98

Hopefully this is still helpful for you. Let me know what you think!

Thanks Kaleb
This is really helpful .I need help a bit even deeper looking at my case study.
I am comfortable now with updating/adding airtable record from python if we define json for new records or matching value hardcoded manually .

it will be really great if there are examples for say.1) if i have pandas dataframe of few rows which i have to insert into existing airtable table, based on matching condition on single or multiple columns or index of the dataframe ,2) another example of creating new table in airtable in existing base from pandas dataframe (struggling to convert dataframe into json required for airtable post request)

Thanks
Sanket Z

it will be really great if there are examples for say.1) if i have pandas dataframe of few rows which i have to insert into existing airtable table, based on matching condition on single or multiple columns or index of the dataframe

@sanky23, it might not be quite the solution what you're looking for, but I've added a function upload_pandas_dataframe() to the gist above. See if you can get it to work for you. I'll try and incorporate your question in the final draft of the tutorial when it is published.

,2) another example of creating new table in airtable in existing base from pandas dataframe (struggling to convert dataframe into json required for airtable post request)

This question requires working with Airtable's metadata API, which is still in limited release and you will have to register for. Learn more here: airtable.com/api/meta

Hi @KalebNyquist , I found your script from your medium article Airtable & Python Made Possible: Downloading Airtable Data into Python using Airtable’s API. It is not directly hosted on medium but embedded on "towardsdatascience" website. I was thinking to create something like this but this is well done. Thanks for sharing.

@KalebNyquist awesome funs mate. thanks for making this available.