import gradio as gr
import requests
import os
import json
from transformers import pipeline
import matplotlib.pyplot as plt
import numpy as np
import pandas as pd




bearer_token = os.environ.get("BEARER_TOKEN")
print(bearer_token)


search_url = "https://api.twitter.com/2/tweets/search/recent"



def bearer_oauth(r):
    """
    Method required by bearer token authentication.
    """

    r.headers["Authorization"] = f"Bearer {bearer_token}"
    r.headers["User-Agent"] = "v2RecentSearchPython"
    return r

def connect_to_endpoint(url, params):
    response = requests.get(url, auth=bearer_oauth, params=params)
    print(response.status_code)
    if response.status_code != 200:
        raise Exception(response.status_code, response.text)
    return response.json()


def fetch_tweets(tag):
    q = "\"" + tag + "\""
    query_params = {'query': q, 'tweet.fields': 'author_id', 'max_results': 100}
    json_response = connect_to_endpoint(search_url, query_params)
    #print(json.dumps(json_response, indent=4, sort_keys=True))
    phrases = []
    for entry in json_response["data"]:
      phrases.append(entry["text"])
    return phrases

pipe = pipeline("text-classification", model="mrm8488/distilroberta-finetuned-financial-news-sentiment-analysis")

def analyze_phrases(phrases):
  positive = 0
  positive_examples = {}
  negative = 0
  negative_examples = {}
  neutral = 0
  neutral_examples = {}
  outputs = pipe(phrases)
  for index, x in enumerate(outputs):
    if x['label'] == 'positive':
      positive += 1
      if positive <= 3:
        positive_examples[phrases[index]] =  x['score']
    elif x['label'] == 'neutral':
      neutral += 1
      if neutral <= 3:
        neutral_examples[phrases[index]] =  x['score']
    elif x['label'] == 'negative':
      negative += 1
      if negative <= 3:
        negative_examples[phrases[index]] =  x['score']
    else:
      pass
  counts = [positive, neutral, negative]
  return counts, positive_examples, neutral_examples, negative_examples


def calculate_sentiment(tag):
  phrases = fetch_tweets(tag)
  counts, positive_examples, neutral_examples, negative_examples = analyze_phrases(phrases)
  output = "positive: " + str(counts[0]) + "\n" + "neutral: " + str(counts[1]) + "\n" + "negative: " + str(counts[2])
  plt.style.use('_mpl-gallery-nogrid')


  # make data
  colors = ['green', 'yellow', 'red']
  labels = ["Positive", "Neutral", "Negative"]

  # plot
  fig, ax = plt.subplots(figsize=(10, 6))
  wedges, texts = ax.pie(counts, colors=colors, radius=3, center=(4, 4),
        wedgeprops={"linewidth": 1, "edgecolor": "white"}, labeldistance=1.05)

  # Create a legend
  ax.legend(wedges, labels, title="Categories", loc="center left", bbox_to_anchor=(1, 0, 0.5, 1))

  ax.set(xlim=(0, 8),
        ylim=(0, 8))
  print(positive_examples)
  html_content = ""
  positive_tweets = list(positive_examples.items())
  p_df = pd.DataFrame(positive_tweets, columns=["Tweet", "Confidence"])
  positive_table = p_df.to_html(index=False)

  neutral_tweets = list(neutral_examples.items())
  n_df = pd.DataFrame(neutral_tweets, columns=["Tweet", "Confidence"])
  neutral_table = n_df.to_html(index=False)

  negative_tweets = list(negative_examples.items())
  neg_df = pd.DataFrame(negative_tweets, columns=["Tweet", "Confidence"])
  negative_table = neg_df.to_html(index=False)

  html_content += f"<h2>Recent Positive Tweets</h2>" + positive_table
  html_content += f"<h2>Recent Negative Tweets</h2>" + negative_table
  
  
  return fig, html_content
    

iface = gr.Interface(fn=calculate_sentiment, inputs="text", outputs=["plot","html"])
iface.launch(debug=True)