No Results

FDA Recalls Application


Food Recalls

A review of the U.S. Food and Drug Administration’s safety site shows that since the beginning of 2015, product recalls, withdrawals and safety alerts are almost a daily occurrence in America. With the frequency of recalls increasing every year, consumers need more transparency when it comes to recalls. By placing recall data on an immutable ledger such as Algorand, an environment of accountability and transparency is created for consumers, the FDA, and businesses.

By encoding the FDA’s API results in a transaction notefield using Python, we are able to successfully store critical pieces of information such as the UPC codes for recalled consumer drug products. After the recall data is encoded and stored on Algorand, a specific prefix is used to index the recall information.

This application focuses on the dates of 12/01/20 to 12/02/20 for food recalls specifically; the prefix ‘FDAF’ is used for the indexer portion. This application can encompass other date timelines and incorporate food, drug, or medical device recalls.


  • Python 3.0 or above
  • Python Algorand SDK module
  • Python requests module
  • Dockerized Algorand sandbox environment

Additional Resources of Use


Step 0

  • Get the code:

$ git clone
$ cd FDA_Recalls_App`

Step 1

  • Install algosdk module: $ pip install py-algorand-sdk
  • Install requests module: $ pip install requests

Step 2

  • Locate parse_into_json() function from file
  • Set date1 to desired year, month and date as follows YYYYMMDD
  • Set date2 to a later date with same format, YYYYMMDD
  • Set recall_type to either "food", "drug" or "device"

def parse_into_json():
    # dates (YYYYMMDD) and recall type(food, device, drug)
    date1 = 20201201
    date2 = 20201202
    recall_type = "food"

Step 3

  • Load Dockerized sandbox environment
  • Run $ ./sandbox up in the command line of your terminal within your installed sandbox folder
  • Once sandbox nightly is loaded, find the three listed private accounts or simply run command $ ./sandbox goal account list in the terminal

Sandbox Account List Image

  • Copy the first account address, which will resemble a 58 character address like VAJPCCDYERNMNJC6VDLAR5FCDD3OCFRBRETRTOBIG3YTBCMUG4JZXQ7J3U
  • Run command $ ./sandbox goal account export -a <paste 58 character address here>

Sandbox Account Export Image

  • After running the $ ./sandbox goal account export -a <paste 58 character address here> command, copy the 25 word passphrase/key for the first account

Step 4

  • Locate send_note() function from file

def send_note():
    algod_address = "http://localhost:4001"
    algod_token = "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa"
    algod_client = algod.AlgodClient(algod_token, algod_address)

    passphrase = ""
    private_key = mnemonic.to_private_key(passphrase)
    my_address = mnemonic.to_public_key(passphrase)
    print(f'My address: {my_address}')
    params = algod_client.suggested_params()
    # comment out the next two (2) lines to use suggested fees
    params.flat_fee = True
    params.fee = 1000
    json_note = parse_into_json()
    note = json_note.encode()
    receiver = ""

  • Set passphrase = "<paste 25 word passphrase/ key here>"
  • Go back to sandbox account list and copy the second or third 58 character account address,
  • Set receiver = "<second or third account address from sandbox account list here>"

Step 5

  • Return to parse_into_json() function from file
  • Make a debugger break for line 69 which contains parse_final_value = int(url_Number) - 1

    # find number of results with index starting at 0
    parse_final_value = int(url_Number) - 1

  • Debug to find url_Number value

Debug url_Number image

  • Scroll down to results_location and set this value equal to 0 or any number less than or equal to int(url_Number) - 1

# Add prefix FDAF to specified set of results, 0 to parse_final_value
    results_location = 0
    result = results_only[results_location]

    json_note = json.dumps(result)
    json_note = "FDAF" + json_note
    return json_note

  • Based on what recall_type is set in Step 2, change prefix from "FDAF", which represents FDA Food Recalls, to a prefix suited for the other recall types such as "FDADR" for drug recalls or "FDADE" for medical devices recalls

json_note = "<prefix here>" + json_note`

Step 6

  • Run file
  • Verify transaction is sent by seeing the decoded note in the terminal

Decoded Note Verification Image

  • If more transactions are to be sent, simply change the results_location value in parse_into_json() to another number equal to or less than int(url_Number) - 1 and run file again

# Add prefix FDAF to specified set of results, 0 to parse_final_value
    results_location = 10
    result = results_only[results_location]

    json_note = json.dumps(result)
    json_note = "FDAF" + json_note
    return json_note

Step 7

  • Locate file
  • Referring to the prefix set in Step 5, change note_prefix = 'FDAF'.encode() to note_prefix = '<your prefix>'.encode()
  • Run file

import base64
import json
from algosdk.v2client import indexer

myindexer = indexer.IndexerClient(
    indexer_token="", indexer_address="http://localhost:8980")
note_prefix = 'FDAF'.encode()
response = myindexer.search_transactions(note_prefix=note_prefix)
print("note_prefix = " + json.dumps(response, indent=2, sort_keys=True))

Final Step

  • Indexer should now have successfully indexed all sent transactions with a notefield encoded with '<your prefix>' and the FDA recall data

Indexer Image

Video Overview

Conclusion/ Future Improvements

Please contribute your ideas about this application on GitHub or reach out to me on the Dev Forums with any questions or about working with Algorand. My goal for this application is to bring more transparency for FDA recalls. Some improvement ideas are listed as follows:

  • FDA could send transaction with data encoded in notefield to an FDA receiver account
  • Build front-end interface for customers to create an Algorand account and scan receipts looking for recalls
  • Customer locates recall item and scans/inputs UPC(or UDI for Medical Recalls)
  • Indexer finds customer’s recall item and confirms customer’s recall item is Ongoing
  • Customer’s recall item’s UPC/UDI code is verified through third-party oracle
  • Companies can send refund transactions with a notefield containing the recall item and it’s UPC/UDI code
  • Companies can track all their recalls on Algorand’s blockchain
  • Companies could also track recall refunds using their own indexer







Algorand SDK

February 03, 2021