Add application file

Dockerfile

@@ -0,0 +1,33 @@
+# base image for Python
+FROM python:3.9-slim
+
+# set working directory in container 컨테이너 내 작업 디렉토리 설정
+WORKDIR /app
+
+# systemp package update & install dependancies
+RUN apt-get update && apt-get install -y \
+    build-essential \
+    && apt-get clean \
+    && rm -rf /var/lib/apt/lists/*
+
+# copy requirements.txt
+COPY requirements.txt .
+
+# dependancy
+RUN pip install --no-cache-dir -r requirements.txt
+
+# copy app source code
+COPY . .
+
+# Expose the port the app runs on
+EXPOSE 7860
+
+# Define environment variable
+ENV FLASK_APP=app.py
+
+# Run the application
+CMD ["flask", "run", "--host=0.0.0.0", "--port=7860"]
+## Hugging Face Spaces는 포트 7860을 사용
+#ENV PORT 7860
+## set command
+#CMD ["python", "app.py"]

app.py

@@ -0,0 +1,83 @@
+from flask import Flask, request, jsonify
+from transformers import AutoTokenizer, AutoModelForSequenceClassification
+import torch
+import os
+
+
+# Flask 앱 초기화
+app = Flask(__name__)
+
+# Load the BERTweet model and tokenizer
+tokenizer = AutoTokenizer.from_pretrained("finiteautomata/bertweet-base-sentiment-analysis")
+model = AutoModelForSequenceClassification.from_pretrained("finiteautomata/bertweet-base-sentiment-analysis")
+
+LABELS = ['Negative', 'Neutral', 'Positive']
+
+def analyze_sentiment(text):
+    # Converts text into tokens that the model can process
+    # return_tensors="pt": Returns PyTorch tensors
+    # truncation=True: Cuts text if it's too long
+    # padding=True: Adds padding to make sequences uniform length
+    inputs = tokenizer(text, return_tensors="pt", truncation=True, padding=True)
+
+    # Predict with the model
+    # Passes tokenized input through the model
+    # Extracts logits (raw prediction scores) from model output
+    outputs = model(**inputs)
+    logits = outputs.logits
+
+    # Sentiment Classification:
+    # torch.argmax(): Finds index of highest score 가장 높은 값의 인덱스를 추출하여 감성 레이블을 선택
+    # Maps this index to a sentiment label using LABELS dictionary
+    # dim=1: Operates along rows, .item(): Converts tensor to Python scalar
+    predicted_class = torch.argmax(logits, dim=1).item()
+    sentiment = LABELS[predicted_class]
+
+    return sentiment
+
+# API endpoint to analyze a text
+@app.route('/analyze', methods=['POST'])
+def analyze():
+    data = request.get_json()
+
+    if 'text' not in data:
+        return jsonify({'error': 'No text provided'}), 400
+
+    # call 감성 분석 function for 입력 텍스트
+    text = data['text']
+    sentiment = analyze_sentiment(text)
+
+    return jsonify({'text': text, 'sentiment': sentiment})
+
+# API endpoint to analyze a list of texts
+@app.route('/analyze_texts', methods=['POST'])
+def analyze_texts():
+    data = request.get_json()
+
+    if 'texts' not in data:
+        return jsonify({'error': 'No texts provided'}), 400
+
+    # Perform analysis on each text in the list, process each one, and return a list of sentiment responses
+    texts = data['texts']
+    results = [{'text': text, 'sentiment': analyze_sentiment(text)} for text in texts]
+
+    # Count negative sentiments
+    negative_count = sum(1 for result in results if result['sentiment'] == 'Negative')
+
+    # Determine if it's a risk (more than half are negative)
+    risk = negative_count > (len(texts) / 2)
+
+    # Return the results as a JSON response
+    # return jsonify(results)
+    # Return the results and risk status as a JSON response
+    return jsonify({
+        'results': results,
+        'Negative Sentiment?': risk
+    })
+
+# run
+
+if __name__ == '__main__':
+    # 환경 변수 PORT를 우선 사용, 없으면 기본값 5000 사용
+    #port = int(os.getenv('PORT', 5001))
+    app.run(host='0.0.0.0', port=7860)

requirements.txt

@@ -0,0 +1,3 @@
+flask
+torch
+transformers

test.http

@@ -0,0 +1,9 @@
+### Sentiment Analysis API test
+
+POST http://0.0.0.0:5001/analyze_texts
+Content-Type: application/json
+
+{
+  "texts": ["I love using Flask for building APIs!", "not working", "Very happy"]
+}
+