LWX/ai_teaching_system/scripts/seed_data.py

# -*- coding: utf-8 -*-
"""初始化初中化学题库与知识图谱数据"""
import sys
import os
sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))

from app import (
    app,
    db,
    Question,
    KnowledgePoint,
    KnowledgeRelation,
    CurriculumTopic,
    ChemistryQuestionBank,
)


def seed_questions():
    """初中化学题库示例"""
    if Question.query.first():
        print("题库已有数据，跳过")
        return
    data = [
        ("下列物质中，不属于电解质的是？", "NaCl", "CH₃COOH", "Cu (金属铜)", "NaOH", "C", "电解质"),
        ("下列关于化学反应速率的说法正确的是？", "反应速率是指反应物浓度的减少", "反应速率是指生成物浓度的增加",
         "反应速率可以用单位时间内反应物浓度的减少来表示", "反应速率与反应物的性质无关", "C", "化学反应速率"),
        ("下列属于氧化还原反应的是？", "CaCO₃ 高温分解", "NaOH + HCl = NaCl + H₂O",
         "2H₂ + O₂ 点燃 2H₂O", "AgNO₃ + NaCl = AgCl↓ + NaNO₃", "C", "氧化还原反应"),
        ("物质的量的单位是？", "g", "mol", "L", "m³", "B", "物质的量"),
        ("元素周期表中，同周期元素从左到右，原子半径变化规律是？", "逐渐增大", "逐渐减小",
         "先增大后减小", "不变", "B", "元素周期律"),
        ("下列物质能在氧气中燃烧并产生大量白烟的是？", "碳", "硫", "磷", "铁丝", "C", "氧气性质"),
        ("原子核由什么组成？", "质子和电子", "质子和中子", "中子和电子", "质子、中子和电子", "B", "原子结构"),
        ("酸溶液能使紫色石蕊试液变？", "红色", "蓝色", "无色", "紫色", "A", "酸碱指示剂"),
        ("催化剂在化学反应中的作用是？", "增加反应物", "减少生成物", "改变化学反应速率", "改变反应条件", "C", "催化剂"),
        ("离子反应发生的条件是？", "有沉淀、气体或水生成", "有催化剂", "加热", "加压", "A", "离子反应"),
    ]
    for content, a, b, c, d, correct, kp in data:
        db.session.add(Question(
            content=content, option_a=a, option_b=b, option_c=c, option_d=d,
            correct_option=correct, knowledge_point=kp or None
        ))
    db.session.commit()
    print("已插入 %d 道题目" % len(data))


def seed_curriculum_topics():
    """初始化课程主题"""
    if CurriculumTopic.query.first():
        print("curriculum_topics 已有数据，跳过")
        return
    topics = [
        dict(name="物质分类", phase="初中", chapter="第一章 走进化学世界",
             description="区分纯净物、混合物、单质、化合物等"),
        dict(name="原子结构", phase="初中", chapter="第二章 走进原子世界",
             description="认识质子、中子、电子及核外电子排布"),
        dict(name="化学方程式", phase="初中", chapter="基本概念与化学方程式",
             description="化学反应的符号表示及配平"),
        dict(name="化学实验基础", phase="初中", chapter="化学实验基本方法",
             description="常用仪器、操作规范、安全意识"),
        dict(name="溶液与浓度", phase="初中", chapter="溶液",
             description="溶液配制、溶质质量分数等"),
        dict(name="金属及其化合物", phase="初中", chapter="金属活动性顺序",
             description="金属与酸、盐溶液的反应与置换"),
        dict(name="酸碱盐", phase="初中", chapter="酸碱盐",
             description="常见酸碱盐的性质及反应"),
        dict(name="气体制取与性质", phase="初中", chapter="常见气体",
             description="氧气、二氧化碳等的制取与性质"),
        dict(name="微观粒子模型", phase="初中", chapter="微观与宏观",
             description="分子、原子、离子等微观示意图"),
        dict(name="化学安全与环保", phase="初中", chapter="化学与生活",
             description="安全操作、事故应急与环保意识"),
    ]
    db.session.bulk_save_objects([CurriculumTopic(**t) for t in topics])
    db.session.commit()
    print("已插入 %d 个课程主题" % len(topics))


def seed_chemistry_question_bank():
    """多题型统一题库（每类大约 10~20 题，后续可自行扩展）"""
    if ChemistryQuestionBank.query.first():
        print("chemistry_question_bank 已有数据，跳过")
        return

    questions = []

    # 1. 基础概念辨析题（choice）约 50 题
    choice_data = [
        dict(
            difficulty="easy",
            topic="物质分类",
            content="下列物质属于纯净物的是：A.空气 B.矿泉水 C.蒸馏水 D.泥水",
            answer_schema={
                "options": {"A": "空气", "B": "矿泉水", "C": "蒸馏水", "D": "泥水"},
                "correct": ["C"],
            },
            hint="纯净物只有一种微粒组成",
        ),
        dict(
            difficulty="medium",
            topic="原子结构",
            content="原子核中一定含有的粒子是：A.质子 B.中子 C.电子 D.离子",
            answer_schema={
                "options": {"A": "质子", "B": "中子", "C": "电子", "D": "离子"},
                "correct": ["A"],
            },
            hint="注意氢原子的特殊性",
        ),
        dict(
            difficulty="easy",
            topic="化学方程式",
            content="催化剂在化学反应中的作用是：A.提高产物质量 B.改变反应速率 C.增加反应物浓度 D.减少能量消耗",
            answer_schema={
                "options": {
                    "A": "提高产物质量",
                    "B": "改变反应速率",
                    "C": "增加反应物浓度",
                    "D": "减少能量消耗",
                },
                "correct": ["B"],
            },
            hint="只改变速率，不改反应前后物质总能量",
        ),
        dict(
            difficulty="easy",
            topic="物质分类",
            content="下列属于金属单质的是：A.Fe B.Fe₂O₃ C.NaCl D.H₂O",
            answer_schema={
                "options": {"A": "Fe", "B": "Fe₂O₃", "C": "NaCl", "D": "H₂O"},
                "correct": ["A"],
            },
            hint="元素与化合物的区分",
        ),
        dict(
            difficulty="easy",
            topic="物质分类",
            content="下列属于混合物的是：A.液氯 B.蒸馏水 C.干冰 D.生理盐水",
            answer_schema={
                "options": {
                    "A": "液氯",
                    "B": "蒸馏水",
                    "C": "干冰",
                    "D": "生理盐水",
                },
                "correct": ["D"],
            },
            hint="盐溶解在水中形成混合物",
        ),
        dict(
            difficulty="medium",
            topic="酸碱盐",
            content="能使紫色石蕊溶液变红的是：A.NaCl溶液 B.NaOH溶液 C.HCl溶液 D.蒸馏水",
            answer_schema={
                "options": {
                    "A": "NaCl溶液",
                    "B": "NaOH溶液",
                    "C": "HCl溶液",
                    "D": "蒸馏水",
                },
                "correct": ["C"],
            },
            hint="酸使石蕊变红，碱使石蕊变蓝",
        ),
        dict(
            difficulty="easy",
            topic="酸碱盐",
            content="下列属于碱的是：A.NaCl B.NaOH C.H₂SO₄ D.CO₂",
            answer_schema={
                "options": {"A": "NaCl", "B": "NaOH", "C": "H₂SO₄", "D": "CO₂"},
                "correct": ["B"],
            },
            hint="含有金属阳离子和 OH⁻ 的化合物",
        ),
        dict(
            difficulty="medium",
            topic="金属及其化合物",
            content="在金属活动性顺序中，能从 CuSO₄ 溶液中置换出铜的金属是：A.Cu B.Ag C.Fe D.Hg",
            answer_schema={
                "options": {"A": "Cu", "B": "Ag", "C": "Fe", "D": "Hg"},
                "correct": ["C"],
            },
            hint="Fe 比 Cu 活泼，可置换",
        ),
        dict(
            difficulty="easy",
            topic="化学安全与环保",
            content="下列行为有利于环境保护的是：A.露天焚烧垃圾 B.乱倒废酸 C.使用无磷洗衣粉 D.随意排放含重金属废水",
            answer_schema={
                "options": {
                    "A": "露天焚烧垃圾",
                    "B": "乱倒废酸",
                    "C": "使用无磷洗衣粉",
                    "D": "随意排放含重金属废水",
                },
                "correct": ["C"],
            },
            hint="无磷洗衣粉可减轻水体富营养化",
        ),
        dict(
            difficulty="easy",
            topic="化学安全与环保",
            content="下列做法中符合化学实验室安全要求的是：A.品尝药品判断其性质 B.用鼻子直接凑近药品闻气味 C.用手扇动少量气体再闻 D.把药品带回家继续做实验",
            answer_schema={
                "options": {
                    "A": "品尝药品判断其性质",
                    "B": "用鼻子直接凑近药品闻气味",
                    "C": "用手扇动少量气体再闻",
                    "D": "把药品带回家继续做实验",
                },
                "correct": ["C"],
            },
            hint="标准闻气体方法",
        ),
    ]

    # 补充自动生成的基础选择题示例，使 choice 题型接近 50 题
    base_choice_count = len(choice_data)
    for i in range(base_choice_count + 1, 51):
        choice_data.append(dict(
            difficulty="easy" if i <= 14 else "medium",
            topic="物质分类" if i % 2 == 0 else "酸碱盐",
            content=f"【选择题示例{i}】下列说法中正确的是？",
            answer_schema={
                "options": {
                    "A": "这是一个占位示例选项 A",
                    "B": "这是一个占位示例选项 B（正确）",
                    "C": "这是一个占位示例选项 C",
                    "D": "这是一个占位示例选项 D",
                },
                "correct": ["B"],
            },
            hint="用于系统演示的占位选择题，可后续替换为真实题目",
        ))

    for q in choice_data:
        questions.append(ChemistryQuestionBank(
            type="choice",
            difficulty=q["difficulty"],
            topic=q["topic"],
            content=q["content"],
            answer_schema=q["answer_schema"],
            hint=q.get("hint"),
        ))

    # 2. 化学方程式书写题（equation）约 50 题
    equation_data = [
        dict(
            difficulty="easy",
            topic="气体制取与性质",
            content="写出实验室用双氧水制氧气的化学方程式：",
            answer_schema={
                "expected_equation": "2H2O2 → 2H2O + O2↑",
                "conditions": "MnO2 作催化剂",
            },
            hint="注意配平和气体符号",
        ),
        dict(
            difficulty="easy",
            topic="金属及其化合物",
            content="写出铁在氧气中燃烧的化学方程式：",
            answer_schema={
                "expected_equation": "3Fe + 2O2 → Fe3O4",
                "conditions": "点燃",
            },
            hint="生成黑色固体 Fe3O4",
        ),
        dict(
            difficulty="easy",
            topic="气体制取与性质",
            content="写出碳酸钙与盐酸反应的化学方程式：",
            answer_schema={
                "expected_equation": "CaCO3 + 2HCl → CaCl2 + H2O + CO2↑",
                "conditions": "常温",
            },
            hint="实验室制备 CO2",
        ),
        dict(
            difficulty="easy",
            topic="酸碱盐",
            content="写出盐酸与氢氧化钠溶液反应的化学方程式：",
            answer_schema={
                "expected_equation": "HCl + NaOH → NaCl + H2O",
                "conditions": "溶液",
            },
            hint="典型中和反应",
        ),
        dict(
            difficulty="medium",
            topic="金属及其化合物",
            content="写出锌与稀盐酸反应的化学方程式：",
            answer_schema={
                "expected_equation": "Zn + 2HCl → ZnCl2 + H2↑",
                "conditions": "常温",
            },
            hint="金属与酸反应生成盐和氢气",
        ),
        dict(
            difficulty="medium",
            topic="金属及其化合物",
            content="写出铁与硫酸铜溶液反应的化学方程式：",
            answer_schema={
                "expected_equation": "Fe + CuSO4 → FeSO4 + Cu",
                "conditions": "常温",
            },
            hint="金属活动性顺序应用",
        ),
        dict(
            difficulty="medium",
            topic="酸碱盐",
            content="写出碳酸钠溶液与盐酸反应的化学方程式：",
            answer_schema={
                "expected_equation": "Na2CO3 + 2HCl → 2NaCl + H2O + CO2↑",
                "conditions": "溶液",
            },
            hint="注意 CO2 气体放出",
        ),
        dict(
            difficulty="medium",
            topic="酸碱盐",
            content="写出石灰水与二氧化碳反应的化学方程式：",
            answer_schema={
                "expected_equation": "Ca(OH)2 + CO2 → CaCO3↓ + H2O",
                "conditions": "微量 CO2",
            },
            hint="澄清石灰水变浑浊",
        ),
    ]

    # 补充自动生成的方程式题示例，使 equation 题型接近 50 题
    base_equation_count = len(equation_data)
    for i in range(base_equation_count + 1, 51):
        equation_data.append(dict(
            difficulty="easy" if i <= 14 else "medium",
            topic="化学方程式",
            content=f"【方程式示例{i}】写出一个常见的初中化学反应方程式：",
            answer_schema={
                "expected_equation": "示例：2H2 + O2 → 2H2O",
                "conditions": "可根据课堂情境自行设定",
            },
            hint="本题为占位示例，可后续替换为具体反应方程式",
        ))

    for q in equation_data:
        questions.append(ChemistryQuestionBank(
            type="equation",
            difficulty=q["difficulty"],
            topic=q["topic"],
            content=q["content"],
            answer_schema=q["answer_schema"],
            hint=q.get("hint"),
        ))

    # 3. 实验操作与安全常识题（experiment）约 50 题
    experiment_data = [
        dict(
            difficulty="easy",
            topic="化学实验基础",
            content="稀释浓硫酸的正确操作是？",
            answer_schema={
                "scenario": "稀释浓硫酸",
                "correct_operation": "将浓硫酸沿器壁慢慢倒入水中并不断搅拌",
                "key_point": "酸入水防飞溅",
            },
            hint="牢记“酸入水”原则",
        ),
        dict(
            difficulty="easy",
            topic="化学实验基础",
            content="检查装置气密性的常用方法是？",
            answer_schema={
                "scenario": "检查装置气密性",
                "correct_operation": "封闭系统后轻轻挤压或捂热容器，观察导管末端气泡变化",
                "key_point": "压强变化观察气泡",
            },
            hint="不能用明火直接检查",
        ),
        dict(
            difficulty="medium",
            topic="化学实验基础",
            content="蒸发结晶停止加热的合适时机是？",
            answer_schema={
                "scenario": "蒸发结晶",
                "correct_operation": "蒸发皿中出现大量晶体、溶液尚未全部蒸干时停止加热",
                "key_point": "防止暴沸和晶体飞溅",
            },
            hint="不能把水分完全蒸干",
        ),
        dict(
            difficulty="easy",
            topic="化学安全与环保",
            content="酒精灯着火后，应采用哪种方法灭火最安全？",
            answer_schema={
                "scenario": "酒精灯起火",
                "correct_operation": "用湿抹布或灯帽盖灭，隔绝空气",
                "key_point": "禁止向燃烧的灯内添加酒精",
            },
            hint="突出隔绝空气",
        ),
        dict(
            difficulty="easy",
            topic="化学安全与环保",
            content="强酸不慎溅到皮肤上，正确的处理方式是？",
            answer_schema={
                "scenario": "强酸溅到皮肤",
                "correct_operation": "立即用大量清水冲洗，再根据情况就医",
                "key_point": "大量清水冲洗稀释",
            },
            hint="不要先用碱液中和",
        ),
        dict(
            difficulty="medium",
            topic="化学实验基础",
            content="用量筒量取液体时，视线应如何放置？",
            answer_schema={
                "scenario": "量筒读数",
                "correct_operation": "视线与凹液面最低处所在水平线相平",
                "key_point": "避免视差",
            },
            hint="读弯月面最低点",
        ),
        dict(
            difficulty="medium",
            topic="化学实验基础",
            content="过滤操作中，滤纸应如何贴紧漏斗？",
            answer_schema={
                "scenario": "过滤",
                "correct_operation": "将滤纸润湿后贴紧漏斗内壁，过滤时液面低于滤纸边缘",
                "key_point": "防止液体从滤纸外流下",
            },
            hint="常见过滤规范",
        ),
        dict(
            difficulty="medium",
            topic="化学安全与环保",
            content="若不慎吸入较多氯气，正确的紧急处理方式是？",
            answer_schema={
                "scenario": "吸入有毒气体",
                "correct_operation": "迅速离开现场到空气新鲜处，必要时就医",
                "key_point": "远离气体来源并就医",
            },
            hint="强调撤离与就医",
        ),
    ]

    # 补充自动生成的实验与安全题示例，使 experiment 题型接近 50 题
    base_experiment_count = len(experiment_data)
    for i in range(base_experiment_count + 1, 51):
        experiment_data.append(dict(
            difficulty="easy" if i <= 14 else "medium",
            topic="化学实验基础" if i % 2 == 0 else "化学安全与环保",
            content=f"【实验与安全示例{i}】下列哪一项更符合规范的化学实验操作或安全要求？",
            answer_schema={
                "scenario": "实验室规范操作占位情境",
                "correct_operation": "选择符合课本要求的规范操作步骤",
                "key_point": "遵守实验室安全与操作规范",
            },
            hint="本题为占位示例，用于丰富题库数量，可后续替换为具体实验情境",
        ))

    for q in experiment_data:
        questions.append(ChemistryQuestionBank(
            type="experiment",
            difficulty=q["difficulty"],
            topic=q["topic"],
            content=q["content"],
            answer_schema=q["answer_schema"],
            hint=q.get("hint"),
        ))

    # 4. 计算类习题（calculation）约 50 题
    calculation_data = [
        dict(
            difficulty="medium",
            topic="溶液与浓度",
            content="将 20g 质量分数为 10% 的 NaCl 溶液稀释至 5%，需加水多少克？",
            answer_schema={
                "formula": "w = m溶质 / m溶液 × 100%",
                "steps": [
                    "计算溶质质量：m溶质 = 20g × 10% = 2g",
                    "设稀释后溶液质量为 m2，有 2 / m2 = 5% ⇒ m2 = 40g",
                    "需加水质量 = 40g - 20g = 20g",
                ],
                "answer": 20.0,
                "unit": "g",
            },
            hint="注意溶质质量不变",
        ),
        dict(
            difficulty="medium",
            topic="气体制取与性质",
            content="电解 18g 水可制得氢气多少克？(相对分子质量：H2O=18，H2=2)",
            answer_schema={
                "formula": "2H2O → 2H2 + O2",
                "steps": [
                    "由方程式知：36g 水生成 4g 氢气",
                    "18g 水生成 x 克氢气，x = 4 × 18 / 36 = 2g",
                ],
                "answer": 2.0,
                "unit": "g",
            },
            hint="由化学方程式确定质量比",
        ),
        dict(
            difficulty="medium",
            topic="溶液与浓度",
            content="向 100g 质量分数为 5% 的 NaCl 溶液中加入 10g NaCl 固体，所得溶液的质量分数约为多少？",
            answer_schema={
                "formula": "w = m溶质 / m溶液 × 100%",
                "steps": [
                    "原溶质质量：m1 = 100g × 5% = 5g",
                    "加入溶质 10g，溶质总质量 m溶质 = 15g",
                    "溶液总质量 m溶液 = 100g + 10g = 110g",
                    "w ≈ 15 / 110 × 100% ≈ 13.6%",
                ],
                "answer": 13.6,
                "unit": "%",
            },
            hint="既改变分子也改变分母",
        ),
        dict(
            difficulty="medium",
            topic="溶液与浓度",
            content="配制 200g 质量分数为 10% 的 NaCl 溶液，需要 NaCl 和水各多少克？",
            answer_schema={
                "formula": "w = m溶质 / m溶液 × 100%",
                "steps": [
                    "m溶质 = 200g × 10% = 20g",
                    "m溶剂(水) = 200g - 20g = 180g",
                ],
                "answer": {"solute": 20.0, "solvent": 180.0},
                "unit": {"solute": "g", "solvent": "g"},
            },
            hint="先求溶质，再求溶剂",
        ),
    ]

    # 补充自动生成的计算题示例，使 calculation 题型接近 50 题
    base_calculation_count = len(calculation_data)
    for i in range(base_calculation_count + 1, 51):
        calculation_data.append(dict(
            difficulty="medium",
            topic="溶液与浓度" if i % 2 == 0 else "气体制取与性质",
            content=f"【计算题示例{i}】根据给出的相对分子质量和化学方程式，完成一题质量或质量分数计算。",
            answer_schema={
                "formula": "可结合题干给出的方程式与数据进行比例计算",
                "steps": [
                    "写出对应的化学方程式并配平",
                    "根据物质的量关系列出质量比或浓度表达式",
                    "代入数据计算并保留合适有效数字",
                ],
                "answer": None,
                "unit": "",
            },
            hint="本题为占位示例，用于训练计算思路，可后续替换为具体数值题",
        ))

    for q in calculation_data:
        questions.append(ChemistryQuestionBank(
            type="calculation",
            difficulty=q["difficulty"],
            topic=q["topic"],
            content=q["content"],
            answer_schema=q["answer_schema"],
            hint=q.get("hint"),
        ))

    # 5. 物质推断题（inference）约 50 题
    inference_data = [
        dict(
            difficulty="medium",
            topic="金属及其化合物",
            content=(
                "A 为红色金属，B 能使澄清石灰水变浑浊，反应关系为："
                "A + O₂ → B；B + Ca(OH)₂ → C。试推断 A、B、C 分别是什么物质？"
            ),
            answer_schema={
                "clues": "A 为红色金属；B 能使澄清石灰水变浑浊",
                "reaction_chain": "A + O2 → B; B + Ca(OH)2 → C",
                "answer": ["Cu", "CO2", "CaCO3"],
            },
            hint="结合颜色和使石灰水变浑浊的气体",
        ),
        dict(
            difficulty="medium",
            topic="金属及其化合物",
            content=(
                "X 与稀盐酸反应放出无色无味气体 Y，Y 能在空气中点燃呈淡蓝色火焰；"
                "金属 Z 能将 X 从其盐溶液中置换出来。试推断 X、Y、Z 分别可能是什么？"
            ),
            answer_schema={
                "clues": "X 与酸反应放氢气；Z 能置换 X",
                "reaction_chain": "X + 2HCl → XCl2 + H2↑; Z + XCl2 → ZCl2 + X",
                "answer": ["Zn", "H2", "Fe"],
            },
            hint="结合金属活动性顺序",
        ),
        dict(
            difficulty="medium",
            topic="酸碱盐",
            content=(
                "某白色固体 A 能溶于水，溶液能使紫色石蕊变红；"
                "A 溶液与一种无色溶液 B 反应生成白色沉淀 C。"
                "已知 B 中含有 Ba²⁺，推断 A、B、C 可能是什么？"
            ),
            answer_schema={
                "clues": "A 溶液显酸性；B 中含 Ba2+；生成白色沉淀",
                "reaction_chain": "2H+ + SO4(2-) + Ba2+ → BaSO4↓ + 2H+",
                "answer": ["H2SO4 或 NaHSO4 等酸性物质", "BaCl2 溶液", "BaSO4"],
            },
            hint="考虑生成难溶的硫酸钡沉淀",
        ),
    ]

    # 补充自动生成的物质推断题示例，使 inference 题型接近 50 题
    base_inference_count = len(inference_data)
    for i in range(base_inference_count + 1, 51):
        inference_data.append(dict(
            difficulty="medium",
            topic="金属及其化合物" if i % 2 == 0 else "酸碱盐",
            content=(
                f"【物质推断示例{i}】已知某些物质在反应中会生成气体或沉淀，"
                f"并且与金属活动性顺序或酸碱中和有关，请根据给出的线索推断相关物质。"
            ),
            answer_schema={
                "clues": "结合题干给出的颜色变化、气体放出或沉淀生成等信息",
                "reaction_chain": "围绕金属活动性顺序或酸碱中和建立推断链",
                "answer": [],
            },
            hint="本题为占位示例，用于训练推理链条，可后续替换为具体情境题",
        ))

    for q in inference_data:
        questions.append(ChemistryQuestionBank(
            type="inference",
            difficulty=q["difficulty"],
            topic=q["topic"],
            content=q["content"],
            answer_schema=q["answer_schema"],
            hint=q.get("hint"),
        ))

    # 6. 微观示意图分析题（diagram）约 50 题
    diagram_data = [
        dict(
            difficulty="easy",
            topic="微观粒子模型",
            content="下图为水分子分解示意图，该图可以解释哪一类化学变化？",
            answer_schema={
                "diagram_url": "img/molecule_H2O.png",
                "description": "水分子电解分解为氢气和氧气的微观示意",
                "analysis_question": "该图示可解释何种反应？",
                "key_insight": "水发生化学变化，分子分裂生成新物质",
            },
            hint="注意“分子分裂”为化学变化特征",
        ),
        dict(
            difficulty="easy",
            topic="微观粒子模型",
            content="下图为 NaCl 溶解电离的微观示意图，该溶液能导电的根本原因是？",
            answer_schema={
                "diagram_url": "img/ionization_NaCl.png",
                "description": "NaCl 晶体在水中电离为 Na⁺ 和 Cl⁻",
                "analysis_question": "为何溶液能导电？",
                "key_insight": "溶液中有大量可以自由移动的离子",
            },
            hint="区分金属导电与溶液导电机理",
        ),
    ]

    # 补充自动生成的微观示意图分析题示例，使 diagram 题型接近 50 题
    base_diagram_count = len(diagram_data)
    for i in range(base_diagram_count + 1, 51):
        diagram_data.append(dict(
            difficulty="easy" if i <= 14 else "medium",
            topic="微观粒子模型",
            content=f"【微观示意图示例{i}】根据给出的粒子示意图，判断该图反映的是物质状态变化、化学变化或电离过程之一。",
            answer_schema={
                "diagram_url": "img/example_micro_diagram.png",
                "description": "占位用微观粒子示意图，可表示分子间距、离子或原子的变化",
                "analysis_question": "图中反映的是哪一类微观变化？",
                "key_insight": "通过粒子种类、数目和排列方式的变化来判断物理或化学变化",
            },
            hint="本题为占位示例，可后续替换为具体教材中的微观粒子示意图",
        ))

    for q in diagram_data:
        questions.append(ChemistryQuestionBank(
            type="diagram",
            difficulty=q["difficulty"],
            topic=q["topic"],
            content=q["content"],
            answer_schema=q["answer_schema"],
            hint=q.get("hint"),
        ))

    db.session.bulk_save_objects(questions)
    db.session.commit()
    print("已向 chemistry_question_bank 插入 %d 道多题型习题" % len(questions))


def seed_chemistry_hard_questions():
    """为多题型题库补充 hard 难度习题，每种题型不少于 10 道

    已有的 hard 题会被保留，只在数量不足的题型上补齐到 10 道。
    """
    TARGET_COUNT = 10
    new_questions = []

    # 1. choice 题型：补足到 10 道 hard
    existing_choice = ChemistryQuestionBank.query.filter_by(
        difficulty="hard", type="choice"
    ).count()
    if existing_choice < TARGET_COUNT:
        for i in range(existing_choice + 1, TARGET_COUNT + 1):
            content = (
                f"【Hard 选择题 {i}】某反应前后有关物质的量或化合价发生变化，"
                f"下列说法中正确的是？"
            )
            answer_schema = {
                "options": {
                    "A": "只要有气体生成就是氧化还原反应",
                    "B": "只要有沉淀生成就是氧化还原反应",
                    "C": "判断是否为氧化还原反应要看元素化合价是否变化",
                    "D": "化学方程式两边原子个数不同也可能是氧化还原反应",
                },
                "correct": ["C"],
            }
            new_questions.append(ChemistryQuestionBank(
                type="choice",
                difficulty="hard",
                topic="化学方程式",
                content=content,
                answer_schema=answer_schema,
                hint="掌握氧化还原反应的本质：元素化合价发生变化",
            ))

    # 2. equation 题型：补足到 10 道 hard
    existing_equation = ChemistryQuestionBank.query.filter_by(
        difficulty="hard", type="equation"
    ).count()
    if existing_equation < TARGET_COUNT:
        for i in range(existing_equation + 1, TARGET_COUNT + 1):
            content = (
                f"【Hard 方程式 {i}】写出一个既体现金属活动性顺序，"
                f"又能体现沉淀生成的离子反应的配平化学方程式。"
            )
            answer_schema = {
                "expected_equation": "示例：Fe + CuSO4 → FeSO4 + Cu",
                "conditions": "常温，溶液中",
            }
            new_questions.append(ChemistryQuestionBank(
                type="equation",
                difficulty="hard",
                topic="化学方程式",
                content=content,
                answer_schema=answer_schema,
                hint="结合金属活动性顺序与难溶盐沉淀的知识，自拟一个合理的反应方程式",
            ))

    # 3. experiment 题型：补足到 10 道 hard
    existing_experiment = ChemistryQuestionBank.query.filter_by(
        difficulty="hard", type="experiment"
    ).count()
    if existing_experiment < TARGET_COUNT:
        for i in range(existing_experiment + 1, TARGET_COUNT + 1):
            content = (
                f"【Hard 实验与安全 {i}】在设计验证某气体为 CO2 的实验时，"
                f"如何合理选择试剂、装置并避免误判？"
            )
            answer_schema = {
                "scenario": "验证某无色气体是否为 CO2",
                "correct_operation": (
                    "选用澄清石灰水检验，并设置对照实验；"
                    "控制气体通入速度，防止生成 Ca(HCO3)2 造成再次澄清导致误判"
                ),
                "key_point": "有对照、控制变量，考虑条件变化对现象的影响",
            }
            new_questions.append(ChemistryQuestionBank(
                type="experiment",
                difficulty="hard",
                topic="化学实验基础",
                content=content,
                answer_schema=answer_schema,
                hint="思考如何通过对照与控制变量排除干扰因素",
            ))

    # 4. calculation 题型：补足到 10 道 hard
    existing_calculation = ChemistryQuestionBank.query.filter_by(
        difficulty="hard", type="calculation"
    ).count()
    if existing_calculation < TARGET_COUNT:
        for i in range(existing_calculation + 1, TARGET_COUNT + 1):
            content = (
                f"【Hard 计算题 {i}】某 100g NaCl 溶液，质量分数为 15%，"
                f"蒸发部分水并加入 10g NaCl 固体后，溶液质量分数变为 25%，"
                f"求蒸发水的质量。"
            )
            answer_schema = {
                "formula": "w = m溶质 / m溶液 × 100%",
                "steps": [
                    "原溶质质量 m1 = 100g × 15% = 15g",
                    "加入固体 NaCl 10g 后，溶质总质量 m溶质 = 25g",
                    "设蒸发水质量为 x，则溶液终质量 m2 = 100 + 10 - x = 110 - x",
                    "由 25 / (110 - x) × 100% = 25%，得 25 / (110 - x) = 0.25",
                    "解得 110 - x = 100 ⇒ x = 10(g)",
                ],
                "answer": 10.0,
                "unit": "g",
            }
            new_questions.append(ChemistryQuestionBank(
                type="calculation",
                difficulty="hard",
                topic="溶液与浓度",
                content=content,
                answer_schema=answer_schema,
                hint="同时注意溶质和溶剂质量的变化，建立方程求解",
            ))

    # 5. inference 题型：补足到 10 道 hard
    existing_inference = ChemistryQuestionBank.query.filter_by(
        difficulty="hard", type="inference"
    ).count()
    if existing_inference < TARGET_COUNT:
        for i in range(existing_inference + 1, TARGET_COUNT + 1):
            content = (
                f"【Hard 物质推断 {i}】某白色固体 A 溶于水后溶液呈碱性，"
                f"与稀盐酸反应放出无色无味气体 B，B 能在空气中点燃呈淡蓝色火焰；"
                f"A 溶液与硫酸铜溶液反应生成蓝色沉淀 C。推断 A、B、C 分别是什么。"
            )
            answer_schema = {
                "clues": (
                    "溶液呈碱性且与稀盐酸反应放出氢气；"
                    "能与 CuSO4 生成蓝色沉淀"
                ),
                "reaction_chain": (
                    "A + 2HCl → B↑ + 相应盐；"
                    "A 溶液 + CuSO4 → C↓ + 其他产物"
                ),
                "answer": ["A：NaOH 或类似强碱", "B：H2", "C：Cu(OH)2"],
            }
            new_questions.append(ChemistryQuestionBank(
                type="inference",
                difficulty="hard",
                topic="酸碱盐",
                content=content,
                answer_schema=answer_schema,
                hint="结合气体性质和沉淀颜色，多步推断相关物质",
            ))

    # 6. diagram 题型：补足到 10 道 hard
    existing_diagram = ChemistryQuestionBank.query.filter_by(
        difficulty="hard", type="diagram"
    ).count()
    if existing_diagram < TARGET_COUNT:
        for i in range(existing_diagram + 1, TARGET_COUNT + 1):
            content = (
                f"【Hard 微观示意图 {i}】给出某反应前后微观粒子示意图："
                f"反应前为双原子分子与单原子金属颗粒的混合，"
                f"反应后生成新的离子晶体。根据图示判断该反应类型并说明理由。"
            )
            answer_schema = {
                "diagram_url": "img/hard_example_micro_diagram.png",
                "description": "金属与酸或盐溶液反应生成新晶体和气体/离子的示意图",
                "analysis_question": "判断反应类型（置换、复分解等），并从微观角度解释",
                "key_insight": "从粒子种类和数目变化判断是否有新物质生成及反应类型",
            }
            new_questions.append(ChemistryQuestionBank(
                type="diagram",
                difficulty="hard",
                topic="微观粒子模型",
                content=content,
                answer_schema=answer_schema,
                hint="结合微观粒子变化与反应类型的宏观特征进行分析",
            ))

    if not new_questions:
        print("chemistry_question_bank 各题型的 hard 难度习题数量已达到或超过 10 道，跳过")
        return

    db.session.bulk_save_objects(new_questions)
    db.session.commit()
    print("已向 chemistry_question_bank 追加 %d 道 hard 难度习题" % len(new_questions))


def seed_knowledge_points():
    """从题目与知识图谱中同步知识点到 knowledge_points 表"""
    if KnowledgePoint.query.first():
        return
    names = set()
    for (kp,) in db.session.query(Question.knowledge_point).filter(
        Question.knowledge_point.isnot(None),
        Question.knowledge_point != "",
    ).distinct().all():
        if kp:
            names.add(kp)
    for s, p, o in [
        ("电解质", "定义", "在水溶液或熔融状态下能够导电的化合物"),
        ("氧化还原反应", "特征", "有元素化合价发生变化的反应"),
        ("物质的量", "单位", "mol（摩尔）"),
        ("原子", "结构", "原子核和核外电子"),
    ]:
        names.add(s)
        if isinstance(o, str):
            names.add(o)
    for name in names:
        if not KnowledgePoint.query.filter_by(name=name).first():
            db.session.add(KnowledgePoint(name=name))
    db.session.commit()
    print("已同步 %d 个知识点到 knowledge_points" % len(names))


def seed_knowledge_graph():
    """初中化学知识图谱三元组"""
    if KnowledgeRelation.query.first():
        print("知识图谱已有数据，跳过")
        return
    data = [
        ("电解质", "定义", "在水溶液或熔融状态下能够导电的化合物"),
        ("电解质", "包含", "酸、碱、盐"),
        ("非电解质", "定义", "在水溶液和熔融状态下都不导电的化合物"),
        ("非电解质", "举例", "蔗糖、酒精"),
        ("氧化还原反应", "特征", "有元素化合价发生变化的反应"),
        ("氧化还原反应", "包含", "氧化反应和还原反应"),
        ("物质的量", "单位", "mol（摩尔）"),
        ("物质的量", "表示", "含有一定数目粒子的集合体"),
        ("元素周期律", "内容", "元素的性质随原子序数递增呈周期性变化"),
        ("元素周期律", "表现", "原子半径、金属性、非金属性等"),
        ("离子反应", "条件", "有沉淀、气体或弱电解质生成"),
        ("化学反应速率", "定义", "单位时间内反应物或生成物浓度的变化"),
        ("催化剂", "特点", "能改变化学反应速率而本身质量和化学性质不变"),
        ("原子", "结构", "原子核和核外电子"),
        ("原子核", "组成", "质子和中子"),
    ]
    for s, p, o in data:
        db.session.add(KnowledgeRelation(subject=s, predicate=p, obj=o))
    db.session.commit()
    print("已插入 %d 条知识图谱三元组" % len(data))


def main():
    with app.app_context():
        seed_questions()
        seed_knowledge_graph()
        seed_knowledge_points()
        seed_curriculum_topics()
        seed_chemistry_question_bank()
        seed_chemistry_hard_questions()
    print("初始化完成")


if __name__ == "__main__":
    main()