Upload 6 files

Dockerfile

@@ -0,0 +1,11 @@
+FROM python:3.10-slim
+
+WORKDIR /app
+
+COPY requirements.txt .
+RUN pip install --no-cache-dir -r requirements.txt
+
+COPY . .
+
+EXPOSE 8000
+CMD ["uvicorn", "app:app", "--host", "0.0.0.0", "--port", "8000"]

README.md

@@ -1,11 +1,34 @@
----
-title: Magbert Ner Api
-emoji: 🏃
-colorFrom: yellow
-colorTo: indigo
-sdk: docker
-pinned: false
-short_description: magbert-ner-fr-api
----
-
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+# Private BERT NER API on Hugging Face Space
+
+This repo contains a private FastAPI-based NER API built on a Hugging Face BERT model.
+
+## Features
+- Private Hugging Face model loading
+- FastAPI REST endpoint for NER
+- Token-based authentication (`X-API-KEY`)
+- Rate limiting per IP using `slowapi`
+- Usage tracking per API key
+
+## Run locally
+```bash
+export HF_TOKEN=your_hf_token
+export X-API-KEY=your-secret-key
+uvicorn app:app --reload
+```
+
+## Example request
+```bash
+curl -X POST http://localhost:8000/ner \
+ -H "X-API-KEY: your-secret-key" \
+ -H "Content-Type: application/json" \
+ -d '{"text": "Barack Obama was the president of the United States."}'
+```
+
+## Deployment
+Can be deployed to Hugging Face Spaces (with Docker option enabled) or any cloud with Docker support.
+
+### Hugging Face Space Setup
+- Make sure to enable **Docker** runtime
+- Add two secrets:
+  - `HF_TOKEN`
+  - `X-API-KEY`

api_keys.json

@@ -0,0 +1,7 @@
+{
+	"users": {
+		"test-key": {
+			"usage_count": 0
+		}
+	}
+}

app.py

@@ -0,0 +1,70 @@
+from fastapi import FastAPI, Depends, HTTPException, Request
+from fastapi.security import APIKeyHeader
+from pydantic import BaseModel
+from model import ner_pipeline
+import logging
+import time
+import json
+import os
+import secrets
+from slowapi import Limiter, _rate_limit_exceeded_handler
+from slowapi.util import get_remote_address
+from slowapi.errors import RateLimitExceeded
+
+logging.basicConfig(level=logging.INFO)
+logger = logging.getLogger(__name__)
+
+app = FastAPI()
+limiter = Limiter(key_func=get_remote_address)
+app.state.limiter = limiter
+app.add_exception_handler(RateLimitExceeded, _rate_limit_exceeded_handler)
+
+API_KEY_FILE = "api_keys.json"
+token_header = APIKeyHeader(name="X-API-KEY")
+
+if os.path.exists(API_KEY_FILE):
+    with open(API_KEY_FILE, "r") as f:
+        API_KEYS_STORE = json.load(f)
+else:
+    API_KEYS_STORE = {"users": {}}
+
+API_KEYS = API_KEYS_STORE.get("users", {})
+ADMIN_KEY = os.getenv("ADMIN_KEY")
+
+class TextRequest(BaseModel):
+    text: str
+
+class RegisterRequest(BaseModel):
+    label: str = "user"
+
+def save_keys():
+    API_KEYS_STORE["users"] = API_KEYS
+    with open(API_KEY_FILE, "w") as f:
+        json.dump(API_KEYS_STORE, f, indent=2)
+
+def verify_token(x_api_key: str = Depends(token_header)):
+    if x_api_key not in API_KEYS:
+        raise HTTPException(status_code=403, detail="Unauthorized")
+    return x_api_key
+
+@app.post("/ner")
+@limiter.limit("10/minute")
+def ner_predict(request: TextRequest, api_key: str = Depends(verify_token), req: Request = None):
+    logger.info("Received NER request from IP: %s", get_remote_address(req))
+    predictions = ner_pipeline(request.text)
+    API_KEYS[api_key]["usage_count"] = API_KEYS[api_key].get("usage_count", 0) + 1
+    save_keys()
+    return {
+        "entities": predictions,
+        "usage": API_KEYS[api_key]["usage_count"]
+    }
+
+@app.post("/register_user")
+def register_user(request: RegisterRequest, x_api_key: str = Depends(token_header)):
+    if x_api_key != ADMIN_KEY:
+        raise HTTPException(status_code=403, detail="Admin access required")
+
+    new_key = secrets.token_urlsafe(32)
+    API_KEYS[new_key] = {"usage_count": 0, "label": request.label}
+    save_keys()
+    return {"message": "User registered", "api_key": new_key}

model.py

@@ -0,0 +1,10 @@
+import os
+from transformers import AutoTokenizer, AutoModelForTokenClassification, pipeline
+
+HF_TOKEN = os.getenv("HF_TOKEN")
+MODEL_ID = "TypicaAI/magbert-ner"
+
+tokenizer = AutoTokenizer.from_pretrained(MODEL_ID, use_auth_token=HF_TOKEN)
+model = AutoModelForTokenClassification.from_pretrained(MODEL_ID, use_auth_token=HF_TOKEN)
+
+ner_pipeline = pipeline("ner", model=model, tokenizer=tokenizer, aggregation_strategy="first")

requirements.txt

@@ -0,0 +1,5 @@
+transformers
+fastapi
+uvicorn
+python-dotenv
+slowapi