knowledge_base.py

"""
垃圾 DNA 知识库 - 物品生命周期溯源系统
Garbage DNA Knowledge Base - Item Lifecycle Traceability

提供每件垃圾的：材质构成、降解时间、回收去向、趣味知识和处理建议
"""

import json
import os
import logging
from typing import Optional, Dict, List

logger = logging.getLogger(__name__)

_DATA_FILE = os.path.join(os.path.dirname(__file__), "data", "item_dna.json")

# 中文别名 -> 英文 key 映射
_ITEM_ALIASES = {
    "塑料瓶": "plastic bottle for drinking",
    "玻璃瓶": "glass bottle for drinking",
    "易拉罐": "aluminum can for soda or beer",
    "纸张纸板": "paper and cardboard for recycling",
    "纸箱": "cardboard box for packaging",
    "报纸杂志": "newspaper and magazines",
    "衣物织物": "clothing and textiles",
    "塑料袋": "plastic bag for shopping",
    "金属制品": "metal tools and hardware",
    "电子产品": "electronics and wires with circuit",
    "书本": "books and notebooks for school",
    "玻璃罐": "glass jar with lid",
    "牛奶盒": "milk carton or tetra pak drink box",
    "干净纸杯": "clean paper coffee cup",
    "铁罐": "iron and steel food cans",
    "塑料容器": "plastic container for food storage",
    "鞋包": "shoes and bags made of fabric",
    "瓶盖": "plastic bottle cap",
    "电池": "battery cell for electronics",
    "灯泡": "light bulb glass transparent",
    "药品": "medicine pills and tablets",
    "油漆桶": "paint can with metal handle",
    "化妆品": "nail polish bottle small",
    "食物残渣": "food waste and leftovers on plate",
    "水果果皮": "fruit and fruit peels organic waste",
    "茶叶咖啡渣": "tea leaves and coffee grounds wet",
    "蛋壳": "egg shell broken pieces",
    "鱼骨鸡骨": "fish bones and chicken bones",
    "花草植物": "flowers and plants wilted",
    "泡沫塑料": "styrofoam and foam packaging white",
    "口罩": "disposable face mask blue",
    "陶瓷": "ceramics and pottery plate or bowl",
    "纸巾": "tissue paper and napkins used",
    "烟蒂": "cigarette butt with filter",
    "尿不湿/卫生巾": "diaper and sanitary pads",
    "泡面桶": "instant noodle cup styrofoam cup with lid",
    "外卖盒": "takeout food container plastic box",
    "奶茶杯": "bubble tea cup with plastic lid and straw",
    "一次性筷子": "disposable wooden chopsticks pair",
    "吸管": "plastic drinking straw",
    "碎玻璃": "broken glass not recyclable shards",
    "泡沫饭盒": "disposable foam lunch box",
    "湿巾": "wet wipes in package",
    "保鲜膜": "plastic wrap and cling film",
    "碎瓷器": "broken porcelain ceramic shards",
    "木筷": "wooden chopsticks reusable",
    "牙签": "wooden toothpick small",
    "棉签": "cotton swab for ears",
    "创可贴": "bandage and adhesive tape medical",
    "纸碗": "paper bowl disposable",
    "木材家具": "wooden furniture or wood pieces",
    "杀虫剂瓶": "pesticide spray bottle",
    "温度计": "mercury thermometer glass",
    "荧光灯管": "fluorescent tube light",
    "尘土": "dust and dirt pile",
}


class KnowledgeBase:
    """垃圾 DNA 知识库 - 从 JSON 文件加载"""

    def __init__(self):
        self._dna: Dict[str, dict] = {}

    def load(self):
        """从 JSON 文件加载 DNA 数据"""
        if not os.path.exists(_DATA_FILE):
            logger.warning(f"[DNA] 知识库文件不存在: {_DATA_FILE}")
            return

        try:
            with open(_DATA_FILE, "r", encoding="utf-8") as f:
                self._dna = json.load(f)
            logger.info(f"[DNA] 已加载 {len(self._dna)} 条知识记录")
        except Exception as e:
            logger.error(f"[DNA] 加载知识库失败: {e}")

    def query(self, item_name: str = None, item_name_zh: str = None) -> Optional[Dict]:
        """根据物品名称查询 DNA 信息

        匹配策略:
        1. 精确匹配英文 key
        2. 精确匹配中文别名
        3. 模糊匹配: 中文名包含关系
        4. 模糊匹配: 英文 key 包含关系
        """
        if not item_name and not item_name_zh:
            return None

        # 1. 精确匹配英文 key
        if item_name and item_name in self._dna:
            return self._dna[item_name]

        # 2. 精确匹配中文别名
        if item_name_zh and item_name_zh in _ITEM_ALIASES:
            key = _ITEM_ALIASES[item_name_zh]
            if key in self._dna:
                return self._dna[key]

        # 3. 模糊匹配中文
        if item_name_zh:
            for alias, key in _ITEM_ALIASES.items():
                if alias in item_name_zh or item_name_zh in alias:
                    if key in self._dna:
                        return self._dna[key]

        # 4. 模糊匹配英文
        if item_name:
            for key in self._dna:
                if item_name in key or key in item_name:
                    return self._dna[key]

        return None

    def all_items(self) -> List[str]:
        return list(self._dna.keys())

    @property
    def count(self) -> int:
        return len(self._dna)


# 全局单例
_kb: Optional[KnowledgeBase] = None


def get_knowledge_base() -> KnowledgeBase:
    """获取知识库单例"""
    global _kb
    if _kb is None:
        _kb = KnowledgeBase()
        _kb.load()
    return _kb


def query_dna(item_name: str = None, item_name_zh: str = None) -> Optional[Dict]:
    """便捷查询接口"""
    return get_knowledge_base().query(item_name, item_name_zh)


if __name__ == "__main__":
    logging.basicConfig(level=logging.INFO)
    kb = get_knowledge_base()
    print(f"知识库已加载 {kb.count} 条记录")

    tests = [
        ("plastic bottle for drinking", "塑料瓶"),
        ("aluminum can for soda or beer", "易拉罐"),
        ("battery cell for electronics", "电池"),
        (None, "未知物品"),
    ]
    for en, zh in tests:
        result = kb.query(en, zh)
        if result:
            print(f"  [{zh}] 材质: {result['material'][:30]}...")
            print(f"  降解: {result['decomposition_desc'][:40]}...")
            print(f"  趣味: {result['fun_facts'][0]}")
        else:
            print(f"  [{zh}] 未找到 DNA 信息")