Upload 7 files

.dockerignore

@@ -0,0 +1,11 @@
+venv
+__pycache__
+*.pyc
+*.pyo
+*.pyd
+.Python
+.env
+.git
+.gitignore
+.huggingface
+.cache

Dockerfile.prod

@@ -0,0 +1,36 @@
+# Production Dockerfile: pre-download model and run with gunicorn
+FROM python:3.11-slim as builder
+
+WORKDIR /app
+
+RUN apt-get update && apt-get install -y --no-install-recommends \
+    build-essential \
+    libxml2-dev libxslt1-dev zlib1g-dev libjpeg-dev libpng-dev git curl && \
+    rm -rf /var/lib/apt/lists/*
+
+COPY requirements.txt ./
+RUN python -m pip install --upgrade pip
+RUN pip install --no-cache-dir -r requirements.txt
+
+# Pre-download the model into the cache
+COPY download_model.py ./
+RUN python download_model.py
+
+# Runtime image
+FROM python:3.11-slim
+WORKDIR /app
+
+RUN apt-get update && apt-get install -y --no-install-recommends \
+    libxml2-dev libxslt1-dev zlib1g-dev libjpeg-dev libpng-dev && rm -rf /var/lib/apt/lists/*
+
+COPY --from=builder /root/.cache /root/.cache
+COPY requirements.txt ./
+RUN python -m pip install --upgrade pip
+RUN pip install --no-cache-dir -r requirements.txt
+
+COPY . /app
+
+ENV MOCK_MODE=0
+EXPOSE 5000
+
+CMD ["gunicorn", "-w", "2", "-b", "0.0.0.0:5000", "app:app"]

Procfile

@@ -0,0 +1 @@
+web: gunicorn -w 2 -b 0.0.0.0:$PORT app:app

README.md

@@ -0,0 +1,60 @@
+# Misinfo Backend (Production-ready)
+
+A Flask backend providing `/analyze` to check text or article URLs for possible misinformation.
+
+Key improvements for production
+
+- Pinned dependencies in `requirements.txt`.
+- `/health` endpoint for readiness checks.
+- Production Dockerfile (`Dockerfile.prod`) that pre-downloads the HF model into the image.
+- Uses `gunicorn` as the WSGI server (configured in `Procfile` for Heroku-like platforms).
+
+Quick local dev
+
+1. Create venv and install dev requirements:
+```powershell
+python -m venv venv
+.\venv\Scripts\Activate.ps1
+python -m pip install -r requirements.txt
+```
+
+2. Run fast dev/mock mode:
+```powershell
+$env:MOCK_MODE=1; python app.py
+```
+
+3. Health check:
+```powershell
+Invoke-RestMethod -Uri 'http://127.0.0.1:5000/health' -Method Get
+```
+
+Docker (production image)
+
+Build production image (this pre-downloads model during image build):
+```powershell
+docker build -f Dockerfile.prod -t misinfo-backend:prod .
+```
+
+Run:
+```powershell
+docker run -p 5000:5000 --env MOCK_MODE=0 misinfo-backend:prod
+```
+
+Heroku / Platform.sh / similar
+
+- Use the provided `Procfile` and push the repo. Set environment variable `MOCK_MODE=0` to enable real model.
+
+Notes
+
+- The image pre-downloads the model into `/root/.cache`. For large-scale production, consider hosting the model in a model server or using a smaller / distilled model.
+- Consider using a GPU-equipped instance if you need faster inference.
+
+Files added/changed
+
+- `requirements.txt` (pinned)
+- `app.py` (health endpoint, prod-friendly)
+- `download_model.py` (pre-download helper)
+- `Dockerfile.prod` (production dockerfile)
+- `.dockerignore`
+- `Procfile`
+

app.py

@@ -0,0 +1,73 @@
+from flask import Flask, request, jsonify
+import os
+import sys
+from newspaper import Article
+
+app = Flask(__name__)
+
+# Configuration from environment
+MOCK_MODE = os.environ.get("MOCK_MODE") == "1" or ("--mock" in sys.argv)
+DEBUG = os.environ.get("DEBUG", "0") == "1"
+
+# Load model at import time so Docker build can pre-download using a helper script
+model = None
+if not MOCK_MODE:
+    try:
+        from transformers import pipeline
+        model = pipeline("text-classification", model="mrm8488/bert-tiny-finetuned-fake-news-detection")
+    except Exception:
+        MOCK_MODE = True
+
+
+@app.route("/health", methods=["GET"])
+def health():
+    return jsonify({"status": "ok", "mock_mode": MOCK_MODE})
+
+
+@app.route("/analyze", methods=["POST"])
+def analyze():
+    data = request.get_json() or {}
+    text = data.get("text")
+    url = data.get("url")
+
+    if url:
+        try:
+            article = Article(url)
+            article.download()
+            article.parse()
+            text = article.text
+        except Exception as e:
+            return jsonify({"error": "Failed to extract article from URL", "details": str(e)}), 400
+
+    if not text:
+        return jsonify({"error": "No text provided"}), 400
+
+    if MOCK_MODE:
+        lowered = text.lower()
+        if "fake" in lowered or "misinform" in lowered or "hoax" in lowered:
+            label = "MISINFO"
+            score = 0.92
+            verdict = "Possibly Misinformation"
+        else:
+            label = "REAL"
+            score = 0.85
+            verdict = "Authentic"
+        authenticity_score = round(score * 100, 2)
+    else:
+        result = model(text)[0]
+        label = result["label"]
+        score = result["score"]
+        verdict = "Authentic" if label == "REAL" else "Possibly Misinformation"
+        authenticity_score = round(score * 100, 2)
+
+    return jsonify({
+        "verdict": verdict,
+        "authenticity_score": authenticity_score,
+        "model_label": label,
+    })
+
+
+if __name__ == "__main__":
+    port = int(os.environ.get("PORT", "5000"))
+    # In production (gunicorn) this block isn't used; it's useful for local dev.
+    app.run(debug=DEBUG, host="0.0.0.0", port=port)

download_model.py

@@ -0,0 +1,8 @@
+from transformers import pipeline
+
+def download():
+    # instantiate pipeline to force model download into cache
+    pipeline("text-classification", model="mrm8488/bert-tiny-finetuned-fake-news-detection")
+
+if __name__ == '__main__':
+    download()

requirements.txt

@@ -0,0 +1,6 @@
+flask==2.3.3
+transformers==4.34.0
+newspaper3k==0.2.8
+lxml_html_clean==0.3.0
+gunicorn==20.1.0
+torch