调用Gemini API遇403(API_KEY_HTTP_REFERRER_BLOCKED)错误求助

问题描述 #

部署在IP为158.38.176.102的服务器上的Python Flask项目调用Gemini API时，出现以下403错误：

"error": "403 Requests from referer are blocked. [reason:
"API_KEY_HTTP_REFERRER_BLOCKED"
domain:
"googleapis.com"
metadata {
key: "service"
value:
"generativelanguage.googleapis.com"
}
metadata {
key:
"consumer"
value:
"projects/883817471146"
}
]"

项目核心代码如下：

from flask import Flask, jsonify, request
from flask_restful import reqparse
from dotenv import load_dotenv
from PIL import Image
import google.generativeai as genai
import io

genai.configure(api_key="Example")

app = Flask(__name__)

def get_gemini_response(category, title, description, image):
    model = genai.GenerativeModel('gemini-1.5-flash')
    # Combining category, title, and description into a prompt
    input_text = f"Category: {category}\nTitle: {title}\nDescription: {description}"

    if input_text:
        response = model.generate_content([f"User: {input_text}", image])
    else:
        response = model.generate_content(image)
    return response.text

@app.route('/check-category-post', methods=['POST'])
def evaluate_image():
    try:
        # Extract form data and file
        category = request.form.get('category')
        title = request.form.get('title')
        description = request.form.get('description')
        file = request.files.get('image')

        print("Category:", category)
        print("Title:", title)
        print("Description:", description)
        print("File:", file) 

        if not category or not title or not description or not file:
            return jsonify({"error": "Category, title, description, and image file are required"}), 400

        # Open the image
        image = Image.open(io.BytesIO(file.read()))

        # Pass the category, title, description, and image to the Gemini model
        response_text = get_gemini_response(category, title, description, image)

        return jsonify({"response": response_text}), 200
    except Exception as e:
        print("Error:", str(e))  # Log the error for debugging
        return jsonify({"error": str(e)}), 500

@app.route('/check-category-post', methods=['POST'])
def check_category_post():
    # Log headers and raw data for debugging
    print("Headers:", request.headers)
    print("Raw Data:", request.data)

    # Set up reqparse
    parser = reqparse.RequestParser()
    parser.add_argument('cpf', type=str, required=True, location='args')  # Expecting args
    args = parser.parse_args()

    return jsonify({"message": "Success", "data": args}), 200

if __name__ == '__main__':
    app.run(host='0.0.0.0', port=5000, debug=True)

Google Cloud控制台API设置截图：

解决方案 #

这个错误是因为你的Gemini API密钥设置了HTTP referer限制，但服务器端调用API时没有携带符合要求的referer头，或者当前请求的来源不在允许列表中。

方法1：调整API密钥的referer限制设置 #

1. 登录Google Cloud控制台，找到对应项目projects/883817471146的API密钥。
2. 在密钥的"应用限制"部分，针对generativelanguage.googleapis.com服务：
   • 如果不需要referer限制，选择"无限制"（不推荐生产环境使用）。
   • 如果需要保留限制，添加你的服务器IP或绑定域名到允许列表，比如https://158.38.176.102/*。
3. 保存设置后，等待几分钟让配置生效。

方法2：在请求中手动添加合法的referer头 #

使用google.generativeai SDK时，可以通过配置客户端选项添加referer头：

import google.generativeai as genai
from google.api_core.client_options import ClientOptions

# 配置客户端选项，添加合法的referer
client_options = ClientOptions(api_endpoint="generativelanguage.googleapis.com")
client_options.headers = {"Referer": "https://158.38.176.102/"}

genai.configure(
    api_key="你的实际API密钥",
    client_options=client_options
)

额外注意事项 #

• 代码中存在两个相同路径/check-category-post的POST路由，Flask只会加载最后定义的check_category_post函数，导致前面的evaluate_image无法生效，需修改其中一个路由路径避免冲突。
• 不要硬编码API密钥，使用dotenv加载环境变量保障密钥安全：

  import os
  load_dotenv()
  genai.configure(api_key=os.getenv("GEMINI_API_KEY"))
  

内容的提问来源于stack exchange，提问作者Shehbaz Khanani