LQAdminPlatform/src/app/utils/vector_utils.py

import hashlib
import math
import requests
import logging
from typing import List
from app.core.config import config_handler

logger = logging.getLogger(__name__)

# Read config
EMBEDDING_BASE_URL = config_handler.get("admin_app", "EMBEDDING_BASE_URL", "")
EMBEDDING_MODEL = config_handler.get("admin_app", "EMBEDDING_MODEL", "")
EMBEDDING_API_KEY = config_handler.get("admin_app", "EMBEDDING_API_KEY", "dummy")

def text_to_vector_algo(text: str, dim: int = 768) -> List[float]:
    """
    调用 Embedding API 生成向量。
    如果 API 调用失败，返回全 0 向量 (长度为 dim)。
    注意：返回的向量维度取决于模型，可能与传入的 dim 不一致。
    """
    if not text:
        return [0.0] * dim

    # 如果没有配置 URL，回退到原来的哈希算法 (或者直接报错，视需求而定)
    # 这里为了防止完全无法运行，保留一个简单的 fallback，但打个 warning
    if not EMBEDDING_BASE_URL:
        logger.warning("未配置 EMBEDDING_BASE_URL，使用 Dummy Hash 向量")
        return _dummy_hash_vector(text, dim)

    try:
        url = f"{EMBEDDING_BASE_URL}/embeddings"
        headers = {
            "Content-Type": "application/json",
            "Authorization": f"Bearer {EMBEDDING_API_KEY}"
        }
        payload = {
            "input": text,
            "model": EMBEDDING_MODEL
        }
        
        # 简单重试机制
        response = requests.post(url, json=payload, headers=headers, timeout=30)
        response.raise_for_status()
        data = response.json()
        
        # 兼容 OpenAI 格式
        if 'data' in data and len(data['data']) > 0:
            embedding = data['data'][0]['embedding']
            return embedding
        else:
            logger.error(f"Embedding API 响应格式错误: {data}")
            return [0.0] * dim
            
    except Exception as e:
        logger.error(f"Embedding API 调用失败: {e}")
        return [0.0] * dim

def _dummy_hash_vector(text: str, dim: int) -> List[float]:
    """原有的特征哈希算法，作为 fallback"""
    vector = [0.0] * dim
    tokens = list(text) 
    for token in tokens:
        hash_obj = hashlib.md5(token.encode('utf-8'))
        hash_val = int(hash_obj.hexdigest(), 16)
        idx = hash_val % dim
        vector[idx] += 1.0
    magnitude = math.sqrt(sum(x*x for x in vector))
    if magnitude > 0:
        vector = [x / magnitude for x in vector]
    return vector