api.py

from flask import Flask, request, send_file, jsonify
from transformers import DistilBertForSequenceClassification, DistilBertTokenizer
import pandas as pd
import torch
import tempfile
import os
import re
from collections import Counter
from flask_cors import CORS

app = Flask(__name__)
CORS(app)

# Load model from Hugging Face Hub
model_name = "AbdoIR/x-sentiment-analysis/fine_tuned_model"
model = DistilBertForSequenceClassification.from_pretrained(model_name)
tokenizer = DistilBertTokenizer.from_pretrained(model_name)

device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
model.to(device)
model.eval()

# Predict sentiment
def predict_sentiment(texts):
  encodings = tokenizer(texts, truncation=True, padding=True, max_length=128, return_tensors="pt")
  encodings = {key: val.to(device) for key, val in encodings.items()}
  with torch.no_grad():
      outputs = model(**encodings)
      predictions = torch.argmax(outputs.logits, dim=1)
  sentiment_map = {0: "Negative", 1: "Neutral", 2: "Positive"}
  return [sentiment_map[p.item()] for p in predictions]

# Top frequent words
def get_top_words(texts, n=30):
  all_words = []
  for text in texts:
      tokens = re.findall(r'\b\w{3,}\b', str(text).lower())
      all_words.extend(tokens)
  counter = Counter(all_words)
  most_common = counter.most_common(n)
  return pd.DataFrame(most_common, columns=['word', 'count'])

# POST /predict
@app.route('/predict', methods=['POST'])
def predict():
  if 'file' not in request.files:
      return jsonify({'error': 'No file uploaded'}), 400

  file = request.files['file']
  try:
      df = pd.read_csv(file)
  except Exception:
      try:
          file.seek(0)
          df = pd.read_excel(file)
      except Exception:
          return jsonify({'error': 'Unable to read the file'}), 400

  if 'content' in df.columns:
      text_col = 'content'
  elif 'tweet' in df.columns:
      text_col = 'tweet'
  else:
      return jsonify({'error': 'No "content" or "tweet" column found'}), 400

  texts = df[text_col].astype(str).tolist()
  df['sentiment'] = predict_sentiment(texts)

  top_words_df = get_top_words(texts)

  temp_dir = tempfile.mkdtemp()

  sentiment_path = os.path.join(temp_dir, 'final_data.csv')
  df.to_csv(sentiment_path, index=False)

  words_path = os.path.join(temp_dir, 'word_frequent.csv')
  top_words_df.to_csv(words_path, index=False)

  return jsonify({
      'sentiment_file': f'/download?file={sentiment_path}',
      'top_words_file': f'/download?file={words_path}',
      'sentiment_data': df.to_dict(orient='records'),
      'top_words_data': top_words_df.to_dict(orient='records')
  })

# GET /download
@app.route('/download')
def download():
  file_path = request.args.get('file')
  if not file_path or not os.path.exists(file_path):
      return jsonify({'error': 'File not found'}), 404
  return send_file(file_path, as_attachment=True)

if __name__ == '__main__':
  app.run(host="0.0.0.0", port=5000, debug=True)