Create app.py

app.py

@@ -0,0 +1,54 @@
+from fastapi import FastAPI, Request
+from fastapi.responses import HTMLResponse
+from fastapi.staticfiles import StaticFiles
+from fastapi.templating import Jinja2Templates
+import os
+from huggingface_hub import InferenceClient
+
+app = FastAPI()
+
+# Statische Dateien (CSS, JS) einbinden
+app.mount("/static", StaticFiles(directory="static"), name="static")
+templates = Jinja2Templates(directory="templates")
+
+# HF-Token aus Umgebungsvariable
+HF_TOKEN = os.environ.get("HF_TOKEN", "")
+
+@app.get("/", response_class=HTMLResponse)
+async def root(request: Request):
+    return templates.TemplateResponse("index.html", {"request": request})
+
+@app.post("/analyze")
+async def analyze_text(request: Request):
+    data = await request.json()
+    user_text = data.get("text", "")
+    
+    # Sentiment-Analyse durchführen
+    client = InferenceClient(token=HF_TOKEN)
+    sentiment = client.text_classification(
+        text=user_text,
+        model="cardiffnlp/twitter-roberta-base-sentiment"
+    )
+    
+    # Chat-Antwort generieren
+    messages = data.get("history", [])
+    messages.append({"role": "user", "content": user_text})
+    
+    # System-Prompt basierend auf Stimmung
+    best_sentiment = sorted(sentiment, key=lambda x: x["score"], reverse=True)[0]
+    if best_sentiment["label"] == "NEGATIVE" and best_sentiment["score"] > 0.6:
+        messages.append({"role": "system", "content": "Der Patient zeigt starke negative Emotionen – schlage Schuldprojektion oder Verdrängung vor."})
+    elif best_sentiment["label"] == "POSITIVE" and best_sentiment["score"] > 0.6:
+        messages.append({"role": "system", "content": "Der Patient wirkt übertrieben positiv – möglicherweise Abwehrmechanismus durch Kompensation."})
+    
+    # Chat-Antwort generieren
+    response = client.chat_completion(
+        model="gpt-3.5-turbo",
+        messages=messages
+    )
+    
+    return {
+        "reply": response.generated_text,
+        "toneLabel": best_sentiment["label"],
+        "toneScore": best_sentiment["score"]
+    }