app.py

# standard libraries
import os

# third party libraries
import gradio as gr
import validators
from validators import ValidationFailure
import gspread
from google.oauth2.service_account import Credentials 
from pydrive.auth import GoogleAuth
from pydrive.drive import GoogleDrive

import etl
from generic import matched_super_generic

cred_filename = 'heliumhealth-a05d595e5991.json'
# cred_file_path = os.path.join('credentials', cred_filename)

# google auth scopes
scopes = ['https://www.googleapis.com/auth/spreadsheets',
          'https://www.googleapis.com/auth/drive']

# create credentials 
credentials = Credentials.from_service_account_file(
    cred_filename, scopes=scopes )

# authorize google spreadsheet
gc = gspread.authorize(credentials)

gauth = GoogleAuth()
drive = GoogleDrive(gauth)


def prescription_quality_check(gs, sheet_type):
    """
    - Loads, transforms prescription data 
    - Performs RX Norm check for prescription data 

    Args:
        gs (Object): authenticated google sheet instance
        sheet_type (str): sheet type always: prescription
    """
    # load data from sheet
    pres_data, super_data = etl.load_data(gs, sheet_type)

    # process data for use
    pres_data, super_data = etl.preprocess_data(pres_data, super_data, sheet_type)

    # run the matching algorithm
    pres_data = pres_data.apply(matched_super_generic, 
                                axis=1, 
                                args=[
                                    super_data, 
                                    'RX Norm [Super Generic]',
                                    'SUPER_GENERIC', 
                                    'match'])

    # post process data
    pres_data = pres_data[['Unstructured Name', 'RX Norm [Super Generic]', 'match', 'Number']]
    
    # write output to sheet 
    etl.output_data(gs, pres_data, sheet_type)


def diagnosis_quality_check(gs, sheet_type):
    """
    - Loads, transforms prescription data 
    - Performs RX Norm check for prescription data 

    Args:
        gs (Object): google-sheet instance 
        sheet_type (str): sheet type always diagnosis
    """
    # load data from sheet
    diag_data, super_data = etl.load_data(gs, sheet_type)
    # process data for use
    diag_data, super_data = etl.preprocess_data(diag_data, super_data, sheet_type)
    
    
    # run the matching algorithm
    diag_data = diag_data.apply(matched_super_generic, 
                                axis=1, 
                                args=[
                                    super_data, 
                                    'ICD10 Diagnosis',
                                    'Name', 
                                    'match'])

    # post processing 
    diag_data = diag_data[['Unstructured Name', 'ICD10 Diagnosis', 'match', 'Number']]

    # write output to sheet 
    etl.output_data(gs, diag_data, sheet_type)

def quality_check_main(sheet_type, sheet_url):
    """
   main check function

    Args:
        sheet_type (str): sheet type either: prescription or diagnosis 
        sheet_url (url): url to data sheet

    Raises:
        gr.exceptions.Error: Value error for invalid urls 

    Returns:
        message (str): success or failure messages.
    """
    result = validators.url(sheet_url)

    # open the google sheet for reading 
    gs = gc.open_by_url(sheet_url)
    
    # Check if the input is valid url
    # ToDO: Error message displayed is not explanatory - Fix it 
    if isinstance(result, ValidationFailure):
        raise gr.exceptions.Error(message="Please enter a valid URL")
    
    if sheet_type == 'diagnosis':
        diagnosis_quality_check(gs, sheet_type)
    elif sheet_type == 'prescription':
        prescription_quality_check(gs, sheet_type)

    output_text = f"{sheet_type} Q/A successfully. Please check the result sheet for output"
    return output_text



with gr.Blocks() as demo:
  gr.Markdown(
      """
      # Quality Assurance App

      ## Instructions

      * Add IAM User to sheet you want to test
      * Gsheet tabs required for diagnosis:  
        * Diagnosis, ICD10 Diagnosis, resultd
      * Gsheet tabs required for prescription:  
        * Prescriptions, Super_generic_Prescriptions, result
      * Data headers required for diagnosis
        * Unstructured Name, ICD10 Diagnosis, Number
      * Data headers required for prescription
        * Unstructured Name, RX Norm [Super Generic], Number
      """
  )
  # inputs
  sheet_type = gr.Dropdown(['prescription', 'diagnosis'], label="QA Type")
  url = gr.Textbox(label="URL", placeholder="Enter sheet url ...")
  # outputs
  output = gr.Textbox(label="Output Box")
  
  run_btn = gr.Button("Run")
  run_btn.click(
      fn=quality_check_main,
      inputs=[
          sheet_type,
          url
          ],
      outputs=output, 
      api_name="quality_check"
  )

demo.launch()