from enum import Enum
from typing import Optional
from pydantic import BaseModel, Field

class Environment(str, Enum):
    """环境标签，枚举值受约束"""
    PRODUCTION = "生产"
    STAGING = "预发"
    TEST = "测试"

class ServerCI(BaseModel):
    """服务器配置项的结构化 Schema"""
    ci_type: str = Field(description="配置项类型，固定为 '服务器'")
    manufacturer: Optional[str] = Field(None, description="厂商：Dell、HP、华为")
    model: Optional[str] = Field(None, description="型号：R750、DL380")
    ip: Optional[str] = Field(None, description="IP 地址，IPv4 格式")
    cpu_cores: Optional[int] = Field(None, description="CPU 核数")
    memory_gb: Optional[int] = Field(None, description="内存大小（GB）")
    environment: Optional[Environment] = Field(None, description="环境标签")
    location: Optional[str] = Field(None, description="机房位置")

from langchain_openai import ChatOpenAI
from langchain_core.prompts import ChatPromptTemplate
import os
import json

# DeepSeek 兼容 OpenAI 接口，换 base_url 和 model 即可
llm = ChatOpenAI(
    model="deepseek-chat",
    temperature=0,
    base_url="https://api.deepseek.com",
    api_key=os.environ["DEEPSEEK_API_KEY"],
)
# DeepSeek 不支持默认的 json_schema 方式，必须指定 method="function_calling"
structured_llm = llm.with_structured_output(ServerCI, method="function_calling")

prompt = ChatPromptTemplate.from_messages([
    ("system", """从 CMDB 配置项描述中提取结构化属性。

内部缩写（必须遵循）：
- prod / stg / test 对应 生产 / 预发 / 测试
- 192 机房 = 北京 192 号机房
- R750 = Dell PowerEdge R750

规则：
- 描述中未提及的字段填 None，禁止编造
- 数值字段返回数字，不能带单位"""),
    ("human", "{description}"),
])

chain = prompt | structured_llm

# 同一台机器的两种写法
for desc in ["nginx 生产服务器，192 机房，8 核 32G",
             "prod-web-01, Dell R750, 10.0.1.5, 192机房, 8C32G"]:
    result = chain.invoke({"description": desc})
    print(json.dumps(result.model_dump(mode="json"), indent=2, ensure_ascii=False))

def is_likely_duplicate(ci_a: ServerCI, ci_b: ServerCI) -> tuple[bool, str]:
    """规则层：拿到两个结构化 CI，判定是否重复。
    确定性函数，不涉及 LLM。"""
    # 规则 1：IP 相同 = 极可能重复
    if ci_a.ip and ci_a.ip == ci_b.ip:
        return True, f"IP 相同（{ci_a.ip}）"
    # 规则 2：两边都有型号 + 同型号 + 同机房 + 同环境 = 疑似重复
    if (ci_a.model and ci_b.model
            and ci_a.model == ci_b.model
            and ci_a.location == ci_b.location
            and ci_a.environment == ci_b.environment):
        return True, f"同型号同机房同环境（{ci_a.model}/{ci_a.location}）"
    return False, "无重复特征"

# 场景：两条记录厂商型号都不同，但 IP 相同
# 规则层只看结构化属性，IP 相同即判定为疑似重复
r1 = ServerCI(ci_type="服务器", manufacturer="Dell", model="R750",
              ip="10.0.1.5", cpu_cores=8, memory_gb=32,
              environment=Environment.PRODUCTION, location="北京 192 号机房")
r2 = ServerCI(ci_type="服务器", manufacturer="HP", model="DL380",
              ip="10.0.1.5", cpu_cores=16, memory_gb=64,
              environment=Environment.PRODUCTION, location="北京 192 号机房")

print(is_likely_duplicate(r1, r2))