process_data.py

import os
from pydantic import BaseModel, Field, validator, ValidationError
import gradio as gr
from openai import OpenAI
from typing import List, Dict, Any, Optional, Literal, Union
from enum import Enum
from gradio_toggle import Toggle
import json

from schema_classes import FarmActivities, Interactions, Trial, FarmActivitiesLite, PlantingLite, Log, Soil, Yield, InteractionsLite, TrialLite, Person, Treatment


# This API key must be in a "secret" in your environment. This is generated from OpenAI or the company's website that creates the model you wish to engage with. 
# To use other models, some other endpoints would need to slightly change
# As is, the endpoint used requires a model that is capable of OpenAI's structured outputs.
os.environ["OPENAI_API_KEY"] = os.getenv("OPENAI_API_KEY")
client = OpenAI()

# What the survey should look like: 

# Do you want to enter your text in one big block (free-form style) or in individual pieces (form-based style)? ###

# Free-form style means that a single JSON will be created from the block of text. This JSON will have a single level of nested that was created by the model. 
# Form-based style means that individual JSON pieces will be created from different pieces of text. You will have a few more prompts to enter in this case. These JSON pieces will be manually combined with code. This JSON will also have a single level of nested, that was manually created.
# Parse either one big block of text conditionally or individual pieces conditionally 
# values = big-block-input-text, individual-pieces-input-text

# actual long text value = onelonginputtext

# What model version do you want to process your input data?  ###
# Parse ['gpt-4o-mini-2024-07-18', 'gpt-4o-2024-08-06']

# Do you want to pre-process your data? We will feed your data to a conversational model prior to creating the schema, with a prompt of your choosing.
# Parse ['yes', 'no']

# Do you want to enter multiple pre-prompts? Or will you only be entering one? You can enter up to three separate pre-prompts at this time.
# Parse ['yes', 'no']

# Do you have a specific pre-processing style in mind? This is just for data collection purposes. (Allow to pick multiple)

# Not specified means you just want to put text in and it doesn't fit the other categories
# Summarization means you're going to ask the model to produce some sort of summary as your pre-processing step.
# Specific field extraction means you're going to ask the models to extract some details as your pre-processing step.
# Parse possibly multiple values ['not_specified', 'summarization', 'specific-field-extraction']

# Parse [1-3 pre_processing_prompts text boxes]

# Now it is time to enter your prompts. The number of prompts will be directly related to which input data form you've chosen. 
# This is the input data that will be parsed with this prompt: 

# Input data here ###
# This is a sample prompt that you can choose or create your own: 
    # These options if free-form with same text for each
    # Schema 1 prompts:
    #farm_prompt = "Extract the farm information."
    # Schema 2 prompts:
    #interactions_prompt = "Extract the interactions information."
    # Schema 3 prompts:
    #trial_prompt = "Extract the trial information."

    # Schema 1 prompts:
    #field_prompt = "Extract the field information."
    #plant_prompt = "Extract the planting information."
    #log_prompt = "Extract the log information."
    #soil_prompt = "Extract the soil information."
    #yield_prompt = "Extract the yield information."
    # Schema 2 prompts:
    #interaction_prompt = "Extract the interaction information"
    #person_prompt = "Please provide a list of people involved in this interaction, with each person's name, role, and any other relevant details."
    # Schema 3 prompts:
    #trial_prompt = "Extract the trial information"
    #treatment_prompt = "Please provide a list of different treatments (strips or blocks with the same conditions applied) performed by the partner."


# Parameters - The Levers and Flippers to be chosen 
# Use this for reference for now then work backwards
# Otter.ai-summary means that you've already pre-processed your input data using otter.ai and you don't ##### in this event it's just confusing don't include
#all_options = {
#    'model_version': ['gpt-4o-mini-2024-07-18 (Smaller version for faster responses)', 'gpt-4o-2024-08-06 (Latest GPT model with structured outputs)'],
#    'input_modality': ['free-text-input / single-JSON-creation (model creates entire JSON) / single-level-nested', 'form-text-input / stepwise-JSON-creation (individual pieces created then manually combined) / no-nesting (flat schema)'],
#    'pre_processing': ['yes', 'no'],
#    'pre_processing_multiple': ['yes', 'no'],
#    'pre_processing_specification': ['not_specified', 'summarization', 'specific-field-extraction'],
#    'prompting_style': ['no_specific_style', 'example_driven', 'role_specific', 'step_by_step', 'error_detection'],
#    'input_text' = ['whole_input_text': "value", 'input_text_pieces': ["piece_1": "value", "piece_2": "value"]],
#    'pre_processing_prompt' = ['pre_processing_prompts': ["prompt_1": "value", "prompt_2": "value"]],
#    'prompt' = ['prompts': ["prompt1": "value", "prompt2", "value"]]
#}

def generate_json(processed_data): 
    """
    Function to prompt OpenAI API to generate structured JSON output.

    Args: 

    Returns: 
        3 processed data-filled JSON objects: farm_pretty_json, interactions_pretty_json, trial_pretty_json
    """
    print("Generating JSON Whole!")
    input_text = processed_data["input_text"]
    model_version = processed_data["parameters"]["model_version"]

    farm_prompt = processed_data["prompts"]["firstschemaprompt"]
    interactions_prompt = processed_data["prompts"]["secondschemaprompt"]
    trial_prompt = processed_data["prompts"]["thirdschemaprompt"]

        
    try:
        #Call OpenAI API to generate structured output based on prompt
        
        farm_info_response = client.beta.chat.completions.parse(
            model=model_version,  # Use GPT model that supports structured output
            messages=[
                {"role": "system", "content": farm_prompt},
                {"role": "user", "content": input_text}
            ],
            response_format=FarmActivities,
        )

        farm_generated_json = farm_info_response.choices[0].message.parsed

        
        print("FARM JSON: ")
        print(farm_generated_json) # debugging
        farm_pretty_json = farm_generated_json.json()

        
        interactions_response = client.beta.chat.completions.parse(
            model=model_version,  # Use GPT model that supports structured output
            messages=[
                {"role": "system", "content": interactions_prompt},
                {"role": "user", "content": input_text}
            ],
            response_format=Interactions,
        )
            
        interactions_generated_json = interactions_response.choices[0].message.parsed

        print("INTERACTIONS JSON: ")
        print(interactions_generated_json) # debugging 2
        interactions_pretty_json = interactions_generated_json.json()


        trial_response = client.beta.chat.completions.parse(
            model=model_version,  # Use GPT model that supports structured output
            messages=[
                {"role": "system", "content": trial_prompt},
                {"role": "user", "content": input_text}
            ],
            response_format=Trial,
        )
        
        trial_generated_json = trial_response.choices[0].message.parsed

        print("TRIALS JSON: ")
        print(trial_generated_json) # debugging 3

        trial_pretty_json = trial_generated_json.json()

        return farm_pretty_json, interactions_pretty_json, trial_pretty_json

    except ValidationError as e:
        return {"error": str(e)}
    except Exception as e:
        return {"error": "Failed to generate valid JSON. " + str(e)}


# This is for the step-wise JSON creation
def generate_json_pieces(processed_data): 
    """
    This is primarily for one of the flippers, which allows each individual JSON section to be created individually, then concatenates them all together. 
    It is proposed that perhaps the individual calls to the model will be more robust than giving the model all the data at once. 

    Args: 
        Args: 

        parameters: (dict) All of the individual parameters and "flippers" 
            parameters["model_version"] = (str) what model should be used 
            parameters["chaining"] = (bool) whether or not the preprocessed input context should be chained (given to multiple models)
            parameters["context_pre_prompt"], parameters["summary_pre_prompt"], parameters["conversation_pre_prompt"], parameters["example_pre_prompt"] = (str) all of the pre-prompts, separated
            parameters["combined_pre_prompt"] = (str) concatenated individual pre-prompts
            parameters["additional_json_pieces_options"] = (str) "Explicit specific pieces" or "Parse from one big input text" to indicate whether it's many function calls on one input text or many function calls on smaller pieces of input texts

    Returns: 
        (str - json) A final combined JSON containing the data filled schema for Farm Activites
    """
    print("Generating JSON Pieces!")
    
    print("INPUT DATA")
    print(processed_data)
    
    model_version = processed_data["parameters"]["model_version"]

    print("Model Version")
    print(model_version)
    
    if processed_data["parameters"]["pre_process"] == "yes":
        print("Pre prompt is true")
        field_data_input = processed_data["input_text_pieces"]["pre_processed_pieces"]["field_data_input"]
        planting_data_input = processed_data["input_text_pieces"]["pre_processed_pieces"]["planting_data_input"]
        log_data_input = processed_data["input_text_pieces"]["pre_processed_pieces"]["log_data_input"]
        soil_data_input = processed_data["input_text_pieces"]["pre_processed_pieces"]["soil_data_input"] 
        yield_data_input = processed_data["input_text_pieces"]["pre_processed_pieces"]["yield_data_input"] 

        interaction_data_input = processed_data["input_text_pieces"]["pre_processed_pieces"]["interaction_data_input"]
        person_data_input = processed_data["input_text_pieces"]["pre_processed_pieces"]["person_data_input"]

        trial_data_input = processed_data["input_text_pieces"]["pre_processed_pieces"]["trial_data_input"]
        treatment_data_input = processed_data["input_text_pieces"]["pre_processed_pieces"]["treatment_data_input"]
    else:
        print("Pre prompt is false")
        field_data_input = processed_data["input_text_pieces"]["field_data_input"]
        planting_data_input = processed_data["input_text_pieces"]["planting_data_input"]
        log_data_input = processed_data["input_text_pieces"]["log_data_input"]
        soil_data_input = processed_data["input_text_pieces"]["soil_data_input"] 
        yield_data_input = processed_data["input_text_pieces"]["yield_data_input"] 

        interaction_data_input = processed_data["input_text_pieces"]["interaction_data_input"]
        person_data_input = processed_data["input_text_pieces"]["person_data_input"]

        trial_data_input = processed_data["input_text_pieces"]["trial_data_input"]
        treatment_data_input = processed_data["input_text_pieces"]["treatment_data_input"]


    # Fix these prompts for all 
    print("Setting prompts")
    field_prompt = processed_data["prompts"]["first_schema_prompt_one"]
    plant_prompt = processed_data["prompts"]["first_schema_prompt_two"]
    log_prompt = processed_data["prompts"]["first_schema_prompt_three"]
    soil_prompt = processed_data["prompts"]["first_schema_prompt_four"]
    yield_prompt = processed_data["prompts"]["first_schema_prompt_five"]
       
    interaction_prompt = processed_data["prompts"]["second_schema_prompt_one"]
    person_prompt = processed_data["prompts"]["second_schema_prompt_two"]
        
    trial_prompt = processed_data["prompts"]["third_schema_prompt_one"]
    treatment_prompt = processed_data["prompts"]["third_schema_prompt_two"]


    try:
        # Call OpenAI API to generate structured output based on prompt
        print("Getting all responses in pieces, starting with field response")

        # All of this is for the first schema for farm activities 
        print("Field prompt")
        print(field_prompt)

        print("Field data input")
        print(field_data_input)
        
        field_response = client.beta.chat.completions.parse(
            model=model_version,  # Use GPT model that supports structured output
            messages=[
                {"role": "system", "content": field_prompt},
                {"role": "user", "content": field_data_input}
            ],
            response_format=FarmActivitiesLite,
        )

        field_generated_json = field_response.choices[0].message.parsed
        print(type(field_generated_json))

        
        print("FIELD JSON: ")
        field_pretty_json = field_generated_json.dict()
        print(field_pretty_json) # debugging

        
        plant_response = client.beta.chat.completions.parse(
            model=model_version,  # Use GPT model that supports structured output
            messages=[
                {"role": "system", "content": plant_prompt},
                {"role": "user", "content": planting_data_input}
            ],
            response_format=PlantingLite,
        )

        plant_generated_json = plant_response.choices[0].message.parsed

        
        print("PLANT JSON: ")
        plant_pretty_json = plant_generated_json.dict()
        print(plant_pretty_json) # debugging

        log_response = client.beta.chat.completions.parse(
            model=model_version,  # Use GPT model that supports structured output
            messages=[
                {"role": "system", "content": log_prompt},
                {"role": "user", "content": log_data_input}
            ],
            response_format=Log,
        )

        log_generated_json = log_response.choices[0].message.parsed

        
        print("LOG JSON: ")
        log_pretty_json = log_generated_json.dict()
        print(log_pretty_json) # debugging

        soil_response = client.beta.chat.completions.parse(
            model=model_version,  # Use GPT model that supports structured output
            messages=[
                {"role": "system", "content": soil_prompt},
                {"role": "user", "content": soil_data_input}
            ],
            response_format=Soil,
        )

        soil_generated_json = soil_response.choices[0].message.parsed

        
        print("SOIL JSON: ")
        soil_pretty_json = soil_generated_json.dict()
        print(soil_pretty_json) # debugging

        yield_response = client.beta.chat.completions.parse(
            model=model_version,  # Use GPT model that supports structured output
            messages=[
                {"role": "system", "content": yield_prompt},
                {"role": "user", "content": yield_data_input}
            ],
            response_format=Yield,
        )

        yield_generated_json = yield_response.choices[0].message.parsed

        
        print("YIELD JSON: ")
        yield_pretty_json = yield_generated_json.dict()
        print(yield_pretty_json) # debugging

        plantings = {
            **plant_pretty_json,
            "logs": log_pretty_json,
            "soil": soil_pretty_json,
            "yield_": yield_pretty_json
        }
        
        farm_activities = {
            **field_pretty_json,
            "plantings": plantings
        }

        print("ADDED DICTS")
        print(farm_activities)
        print("FINAL JSON: ")
        final_pretty_farm_activity_json = json.dumps(farm_activities, indent=4)
        print(final_pretty_farm_activity_json)


        # This is for the second schema now, interactions
        print("Interaction prompt")
        print(interaction_prompt)

        print("Interaction data input")
        print(interaction_data_input)
        
        interaction_response = client.beta.chat.completions.parse(
            model=model_version,  # Use GPT model that supports structured output
            messages=[
                {"role": "system", "content": interaction_prompt},
                {"role": "user", "content": interaction_data_input}
            ],
            response_format=InteractionsLite,
        )

        interaction_generated_json = interaction_response.choices[0].message.parsed
        
        print("INTERACTION JSON: ")
        interaction_pretty_json = interaction_generated_json.dict()
        print(interaction_pretty_json) # debugging

        print("Person prompt")
        print(person_prompt)

        print("Person data input")
        print(person_data_input)
        
        person_response = client.beta.chat.completions.parse(
            model=model_version,  # Use GPT model that supports structured output
            messages=[
                {"role": "system", "content": person_prompt},
                {"role": "user", "content": person_data_input}
            ],
            response_format=Person,
        )

        person_generated_json = person_response.choices[0].message.parsed
        
        print("PERSON JSON: ")
        person_pretty_json = person_generated_json.dict()
        print(person_pretty_json) # debugging

        interactions = {
            **interaction_pretty_json,
            "people": person_pretty_json
        }

        print("ADDED DICTS 2")
        print(interactions)
        print("FINAL JSON: ")
        final_pretty_interactions_json = json.dumps(interactions, indent=4)
        print(final_pretty_interactions_json)

        # This is for the third schema now, trials
        print("Trial prompt")
        print(trial_prompt)

        print("Trial data input")
        print(trial_data_input)
        
        trial_response = client.beta.chat.completions.parse(
            model=model_version,  # Use GPT model that supports structured output
            messages=[
                {"role": "system", "content": trial_prompt},
                {"role": "user", "content": trial_data_input}
            ],
            response_format=TrialLite,
        )

        trial_generated_json = trial_response.choices[0].message.parsed
        
        print("TRIAL JSON: ")
        trial_pretty_json = trial_generated_json.dict()
        print(trial_pretty_json) # debugging

        print("Treatment prompt")
        print(treatment_prompt)

        print("Treatment data input")
        print(treatment_data_input)
        
        treatment_response = client.beta.chat.completions.parse(
            model=model_version,  # Use GPT model that supports structured output
            messages=[
                {"role": "system", "content": treatment_prompt},
                {"role": "user", "content": treatment_data_input}
            ],
            response_format=Treatment,
        )

        treatment_generated_json = treatment_response.choices[0].message.parsed
        
        print("TREATMENT JSON: ")
        treatment_pretty_json = treatment_generated_json.dict()
        print(treatment_pretty_json) # debugging

        trials = {
            **trial_pretty_json,
            "treatments": treatment_pretty_json
        }

        print("ADDED DICTS 3")
        print(trials)
        print("TREATMENT JSON: ")
        final_pretty_trials_json = json.dumps(trials, indent=4)
        print(final_pretty_trials_json)

        return final_pretty_farm_activity_json, final_pretty_interactions_json, final_pretty_trials_json
    except Exception as e:
        return {"error": "Failed to generate valid JSON. " + str(e)}
    

def pre_processing(processed_data):
    """
    In the event there's a pre-prompt, process the pre-prompts and input text accordingly

    Args: 

    Returns: 
        (dict) input_data
        input_data["input_context"] = (str) the text which should be used as context or "EMPTY" to indicate there is no context
        input_data["input_text"] = (str) input text
    """
    print("Starting preprocessing")


    pre_processing_list = [processed_data.get('parameters', {}).get('preprocessingprompt1', None), processed_data.get('parameters', {}).get('preprocessingprompt2', None), processed_data.get('parameters', {}).get('preprocessingprompt3', None)]

    print("Preprocessing list")
    print(pre_processing_list)

    print("Model Version")
    model_version = processed_data["parameters"]["preprocessmodelversion"]
    print(model_version)
    
    if processed_data["inputstyle"] == "individual-pieces-input-text":
        print("Stepwise Creation")
        processed_data["input_text_pieces"]["pre_processed_pieces"] = {}

        for text_label, text_body in processed_data["input_text_pieces"].items():
                    if 'data_input' in text_label:
                        for pre_prompt in pre_processing_list:
                            if pre_prompt:
                                print("Text Label")
                                print(text_label)
                                print("Prompt followed by data entered")
                                print(pre_prompt)
                                print(text_body)
                                response = client.chat.completions.create(
                                                model=model_version,
                                                messages=[
                                                    {"role": "system", "content": pre_prompt},
                                                    {"role": "user", "content": text_body}
                                                ]
                                            )
                                            
                                response_text = response.choices[0].message.content
                                print("Response text")
                                print(response_text)
                                
                                processed_data["input_text_pieces"]["pre_processed_pieces"][text_label] = response_text
                                print("PRE PROCESSED INPUT TEXT FOR TEXT LABEL: ")
                                print(text_label)
                                print("RESPONSE THAT WILL BE USED: ")
                                print(response_text)

        return processed_data
        
    if processed_data["inputstyle"] == "big-block-input-text":
        
        input_text = processed_data["input_text"]

        for pre_prompt in pre_processing_list:
            try:
                print("Pre-Processing: ")
                if pre_prompt:
                    print("Prompt: ")
                    print(pre_prompt)
                    print("Input Text: ")
                    print(input_text)
                    print("Model: ")
                    print(model_version)
                    
                    response = client.chat.completions.create(
                        model=model_version,
                        messages=[
                            {"role": "system", "content": pre_prompt},
                            {"role": "user", "content": input_text}
                        ]
                    )
                    
                    response_text = response.choices[0].message.content
                    
                    print("Response Text: ")
                    print(response_text)
                    
                    input_text = response_text
                    print("RESPONSE THAT WILL BE USED: ")
                    print(response_text)
                
            except Exception as e:
                print(f"Failed to parse response as JSON. Error was: {e}")

        processed_data["input_text"] = input_text
        return processed_data
        
    
def process_specifications(processed_data):
    """
    Once the parameters and data are processed, do the pre-processing and then generate JSONs

    Args: 

    Returns: 
        3 processed data-filled JSON objects: farm_pretty_json, interactions_pretty_json, trial_pretty_json
    """
    print("Processing specifications")
    print("here is the processed data thus far")

    print(processed_data)
    
    # here is where parsing and other things will happen before 

    if processed_data["inputstyle"] == "individual-pieces-input-text":
        print("You are continuing with step-wise creation with your individual text pieces")
        if processed_data["parameters"]["pre_process"] == "yes":
            print("You are continuing with pre_prompt processing")
            processed_input = pre_processing(processed_data)
        else: 
            print("You have elsed into no pre-processing")
            processed_input = processed_data
        return generate_json_pieces(processed_input)
    elif processed_data["inputstyle"] == "big-block-input-text":
        print("You are elifing into single json creation")
        if processed_data["parameters"]["pre_process"] == "yes":
            print("You are preprocessing now")
            processed_input = pre_processing(processed_data)
        else:
            print("You do not have any preprocessing now")
            processed_input = processed_data
        return generate_json(processed_input)
    
def parse_survey_stack(data):
    """
    Parse the incoming data from the survey stack survey

    Args: 
        data: (json) JSON retrieved from surveystack API after retrieving survey info/details

    Returns: 
        processed_data
            processed_data["input_text"] = (str) the raw input text
    """
    print("PROCESSING SURVEY STACK DATA")
    processed_data = {}

    print("JUST PRINTING OUT THE DATA FOR YA")

    print(data)

    
    processed_data["inputstyle"] = data[0]['data']['inputstyle']['value'][0]
    print("STEPWISE?: " + str(processed_data["inputstyle"]))

    
    processed_data["prompts"] = {}
    
    if processed_data["inputstyle"] == "individual-pieces-input-text":
        print("IN THE STEP")
        farm_management_inputs = data[0]['data']['input_data']['interview']
        print("FARM MANAGEMENT INPUTS" + str(farm_management_inputs))
        
        processed_data["input_text_pieces"] = {}
        processed_data["input_text_pieces"]["field_data_input"] = farm_management_inputs.get('field_data_input', {}).get('value', None)
        processed_data["input_text_pieces"]["planting_data_input"] = farm_management_inputs.get('planting_data_input', {}).get('value', None)
        processed_data["input_text_pieces"]["log_data_input"] = farm_management_inputs.get('log_data_input', {}).get('value', None)
        processed_data["input_text_pieces"]["soil_data_input"] = farm_management_inputs.get('soil_data_input', {}).get('value', None)
        processed_data["input_text_pieces"]["yield_data_input"] = farm_management_inputs.get('yield_data_input', {}).get('value', None)
        processed_data["input_text"] = "EMPTY"

        print("NEXT SCHEMA INPUTS")
        interactions_inputs = data[0]['data']['summary']
        print("INTERACTIONS INPUTS" + str(interactions_inputs))
        processed_data["input_text_pieces"]["interaction_data_input"] = interactions_inputs.get('interaction_data_input', {}).get('value', None)
        processed_data["input_text_pieces"]["person_data_input"] = interactions_inputs.get('person_data_input', {}).get('value', None)

        print("NEXT SCHEMA INPUTS 2")
        trials_inputs = data[0]['data']['trial']
        print("TRIALS INPUTS" + str(trials_inputs))
        processed_data["input_text_pieces"]["trial_data_input"] = trials_inputs.get('trial_data_input', {}).get('value', None)
        processed_data["input_text_pieces"]["treatment_data_input"] = trials_inputs.get('treatment_data_input', {}).get('value', None)

        print("NOW ONTO THE PROMPTS")
        input_item = data[0]['data']['sections_prompt']
        print("PROMPTS INPUTS FOR PIECES" + str(input_item))
        processed_data["prompts"]["first_schema_prompt_one"] = input_item.get('firstschemaprompt1', {}).get('value')
        processed_data["prompts"]["first_schema_prompt_two"] = input_item.get('firstschemaprompt2', {}).get('value')
        processed_data["prompts"]["first_schema_prompt_three"] = input_item.get('firstschemaprompt3', {}).get('value')
        processed_data["prompts"]["first_schema_prompt_four"] = input_item.get('firstschemaprompt4', {}).get('value')
        processed_data["prompts"]["first_schema_prompt_five"] = input_item.get('firstschemaprompt5', {}).get('value')
        processed_data["prompts"]["second_schema_prompt_one"] = input_item.get('secondschemaprompt1', {}).get('value')
        processed_data["prompts"]["second_schema_prompt_two"] = input_item.get('secondschemaprompt2', {}).get('value')
        processed_data["prompts"]["third_schema_prompt_one"] = input_item.get('thirdschemaprompt1', {}).get('value')
        processed_data["prompts"]["third_schema_prompt_two"] = input_item.get('thirdschemaprompt2', {}).get('value')

    elif processed_data["inputstyle"] == "big-block-input-text":
        print("IN THE SINGLE")
        processed_data["input_text"] = data[0]['data']['onelonginputtext']['value']
        input_item = data[0]['data']['schema_prompt']
        print("INPUT TEXT SINGLE BLOCK" + str(processed_data["input_text"]))
        print("INPUT ITEM" + str(input_item))

        print("Input Item")
        print(input_item)

        print("Processed input text")
        print(processed_data["input_text"])

        print("NOW ONTO THE PROMPTS from Input Item")
        processed_data["prompts"]["firstschemaprompt"] = input_item.get('firstschemaprompt', {}).get('value')
        processed_data["prompts"]["secondschemaprompt"] = input_item.get('secondschemaprompt', {}).get('value')
        processed_data["prompts"]["thirdschemaprompt"] = input_item.get('thirdschemaprompt', {}).get('value')
        

    # parameters
    parameter_data = data[0]['data']['parameters']
    print("PARAMETERS" + str(parameter_data))
    
    processed_data["parameters"] = {}
    processed_data["parameters"]["modelversion"] = parameter_data.get('modelversion', {}).get('value')[0]
    processed_data["parameters"]["preprocessdata"] = parameter_data.get('preprocessdata', {}).get('value')[0]
    processed_data["parameters"]["promptstyle"] = parameter_data.get('promptstyle', {}).get('value') 

    if processed_data["parameters"]["preprocessdata"] == "yes":
        processed_data["parameters"]["preprocessmodelversion"] = parameter_data.get('preprocessmodelversion', {}).get('value')[0] 
        processed_data["parameters"]["multiplepreprompts"] =  parameter_data.get('multiplepreprompts', {}).get('value')
        processed_data["parameters"]["prepromptstyle"] = parameter_data.get('prepromptstyle', {}).get('value')
        processed_data["parameters"]["preprocessingprompt1"] = parameter_data.get('preprocessingprompt1', {}).get('value') 
        processed_data["parameters"]["preprocessingprompt2"] = parameter_data.get('preprocessingprompt2', {}).get('value')
        processed_data["parameters"]["preprocessingprompt3"] = parameter_data.get('preprocessingprompt3', {}).get('value')
        
    print("RETURNING DATA")
    print(processed_data)
        
    return processed_data