在RHEL系统上部署并启动Spacy Displacy REST微服务

Got it, let's walk through exactly how to get spaCy Services up and running on your RHEL machine so you can hit that /ent endpoint locally. Here's a step-by-step breakdown that should cover everything you need:

1. Set up basic system dependencies #

First, make sure your RHEL system has the core tools installed. Open your terminal and run:

• For RHEL 8/9 (using dnf):
  sudo dnf install python3 python3-pip git -y
• For RHEL 7 (using yum):
  sudo yum install python3 python3-pip git -y

2. Use a virtual environment (highly recommended) #

Isolating dependencies avoids messing with your system's default Python setup. Here's how to set one up:

• Create a virtual environment:
  python3 -m venv spacy-services-env
• Activate it (you'll see the environment name in your terminal prompt once active):
  source spacy-services-env/bin/activate

3. Clone the spaCy Services repository #

Grab the code from the official repo:
git clone https://github.com/explosion/spacy-services

• Navigate into the cloned directory:
  cd spacy-services

4. Install required packages #

First, install all dependencies listed in the repo's requirements file:
pip install -r requirements.txt

Next, you'll need a spaCy language model to power named entity recognition. Let's use the lightweight default English model:
python -m spacy download en_core_web_sm

5. Start the API service #

Launch the server using Uvicorn (the ASGI server specified in the repo):
uvicorn main:app --host 0.0.0.0 --port 8000
You'll see output confirming the server is running on http://0.0.0.0:8000—this means it's accessible locally at http://localhost:8000.

6. Test the /ent endpoint #

Keep the server running in your first terminal, then open a new window to send a test POST request with curl:
curl -X POST -H "Content-Type: application/json" -d '{"text": "Apple is planning to acquire a UK-based startup for $2 billion"}' http://localhost:8000/ent

You’ll get a JSON response back listing all detected named entities (like "Apple" as ORG, "UK" as GPE, and "$2 billion" as MONEY).

Troubleshooting quick fixes #

• If you can’t reach the endpoint, check if RHEL's firewalld is blocking port 8000. Open it with:
  sudo firewall-cmd --add-port=8000/tcp --permanent
sudo firewall-cmd --reload
• If dependency installation fails, confirm your Python version is 3.7 or higher (spaCy requires this minimum version).
• If the model download times out, configure a proxy for pip if you’re behind a corporate firewall.

内容的提问来源于stack exchange，提问作者Mohit