import gradio as gr
from transformers import AutoTokenizer, AutoModelWithLMHead 
import gc
import os
import pandas as pd
import huggingface_hub
from huggingface_hub import Repository

HF_TOKEN = os.environ.get("HF_TOKEN")
DATASET_NAME = "emotion_detection"
DATASET_REPO_URL = f"https://huggingface.co/datasets/pragnakalp/{DATASET_NAME}"
DATA_FILENAME = "emotion_detection_logs.csv"
DATA_FILE = os.path.join("emotion_detection_logs", DATA_FILENAME)
DATASET_REPO_ID = "pragnakalp/emotion_detection"
print("is none?", HF_TOKEN is None)

sentences_value = """Raj loves Simran.\nLast year I lost my Dog.\nI bought a new phone!\nShe is scared of cockroaches.\nWow! I was not expecting that.\nShe got mad at him."""
cwd = os.getcwd()
model_path = os.path.join(cwd)
tokenizer = AutoTokenizer.from_pretrained("mrm8488/t5-base-finetuned-emotion")
model_base = AutoModelWithLMHead.from_pretrained(model_path)

def get_emotion(text):

    # input_ids = tokenizer.encode(text + '</s>', return_tensors='pt')
    input_ids = tokenizer.encode(text, return_tensors='pt')
    output = model_base.generate(input_ids=input_ids,
                                 max_length=2)

    dec = [tokenizer.decode(ids) for ids in output]
    label = dec[0]
    gc.collect()
    return label

def generate_emotion(article):
    sen_list = article
    sen_list = sen_list.split('\n')
    sen_list_temp = sen_list[0:]
    print(sen_list_temp)
    results_dict = []
    results = []

    for sen in sen_list_temp:
        if(sen.strip()):
            cur_result = get_emotion(sen)

            results.append(cur_result)
            results_dict.append(
                {
                    'sentence': sen,
                    'emotion': cur_result
                }
            )

    # result = {
    #     'result': results_dict,
    # }
    result = {'Input':sen_list_temp, 'Detected Emotion':results}
    gc.collect()
    add_csv = [results_dict]
    with open(DATA_FILE, "a") as f:
        writer = csv.writer(f)
        # write the data
        writer.writerow(add_csv)
        commit_url = repo.push_to_hub()
        print("commit data   :",commit_url)

    return pd.DataFrame(result)
"""
Save generated details
"""
# def save_data_and_sendmail(article,generated_questions,num_que,result):
#     try:
#         hostname = {}
#         hostname = get_device_ip_address()
#         url = 'https://pragnakalpdev35.pythonanywhere.com/HF_space_que_gen'
#         # url = 'http://pragnakalpdev33.pythonanywhere.com/HF_space_question_generator'
#         myobj = {'article': article,'total_que': num_que,'gen_que':result,'ip_addr':hostname.get("ip_addr",""),'host':hostname.get("host","")}
#         x = requests.post(url, json = myobj) 
#         add_csv = [article, generated_questions, num_que]
#         with open(DATA_FILE, "a") as f:
#             writer = csv.writer(f)
#             # write the data
#             writer.writerow(add_csv)
#             commit_url = repo.push_to_hub()
#             print("commit data   :",commit_url)
#     # except Exception as e:
#     #     return "Error while storing data -->" + e
        
#     # try:  
#         # with open(DATA_FILE, "r") as file:
#         #     data = json.load(file)
#         # data.append(entry)
#         # with open(DATA_FILE, "w") as file:
#         #     json.dump(data, file)
#         #     commit_url = repo.push_to_hub()
#     except Exception as e:
#         return "Error while sending mail" + e
        
#     return "Successfully save data"

inputs = gr.Textbox(value=sentences_value,lines=10, label="Sentences",elem_id="inp_div")
outputs = [gr.Dataframe(row_count = (2, "dynamic"), col_count=(2, "fixed"), label="Here is the Result", headers=["Input","Detected Emotion"])]

demo = gr.Interface(
    generate_emotion,
    inputs,
    outputs,
    title="Emotion Detection",
    description="Feel free to give your feedback", 
    css=".gradio-container {background-color: lightgray} #inp_div {background-color: #FB3D5;}"
)
demo.launch()