backend/app/core/keywords.py

"""
Chinese keywords for game review topic detection.
Used in hybrid approach (Keywords + ML Sentiment).

Categories based on common topics in Steam game reviews.
Seed keywords will be expanded using the expand_keywords pipeline.

Structure: topic -> {single_char, compound, phrase}
- single_char: standalone Chinese characters (1 char, prone to false positives)
- compound: multi-char Chinese words or short English words
- phrase: multi-word phrases (EN or ZH)
"""

TOPIC_KEYWORDS: dict[str, dict[str, list[str]]] = {
    # =========================================================================
    # CORE GAMEPLAY - 核心玩法
    # =========================================================================
    "Gameplay": {
        "single_char": ["刷", "肝"],
        "compound": [
            "玩法", "游戏性", "机制", "战斗", "任务", "关卡",
            "探索", "技能", "装备", "gameplay",
        ],
        "phrase": ["战斗系统"],
    },

    "Fun": {
        "single_char": ["爽", "烂"],
        "compound": [
            # Positive
            "好玩", "有趣", "上瘾", "神作", "佳作", "精品",
            "沉浸", "过瘾", "带感", "回血", "爽游",
            "解压", "杀时间",
            # Negative
            "无聊", "枯燥", "乏味", "垃圾", "辣鸡", "粪作",
            "失望", "无趣",
        ],
        "phrase": [
            "电子伟哥", "治好了", "精神时光屋", "时光屋",
            "电子阳痿", "电子ed",
        ],
    },

    "Difficulty": {
        "single_char": [],
        "compound": [
            "难度", "简单", "困难", "硬核",
            "劝退", "手残", "新手", "上手",
            "souls", "魂类",
        ],
        "phrase": ["太难", "太简单"],
    },

    # =========================================================================
    # TECHNICAL - 技术
    # =========================================================================
    "Performance": {
        "single_char": ["卡"],
        "compound": [
            "优化", "卡顿", "帧率", "帧数", "流畅", "掉帧",
            "丝滑", "显卡", "显存", "延迟",
            "fps", "cpu", "gpu",
        ],
        "phrase": [
            "稳60", "锁60", "解锁帧率", "吃配置", "带不动",
            "PPT效果", "幻灯片", "帧生成", "输入延迟", "帧数不稳",
        ],
    },

    "Bugs": {
        "single_char": [],
        "compound": [
            "闪退", "崩溃", "卡死", "报错", "存档",
            "黑屏", "进不去", "打不开", "未响应", "无响应",
            "弹窗", "坏档", "掉线",
            "bug", "bugs",
        ],
        "phrase": [
            "存档损坏", "无法保存", "卡加载",
            "加载失败", "连不上",
        ],
    },

    # =========================================================================
    # AUDIO-VISUAL - 视听
    # =========================================================================
    "Graphics": {
        "single_char": [],
        "compound": [
            "画面", "画质", "特效", "建模", "贴图",
            "美术", "风格", "场景", "光影",
            "4k", "hdr",
        ],
        "phrase": [],
    },

    "Sound": {
        "single_char": [],
        "compound": [
            "音乐", "音效", "配音", "配乐", "声音",
            "原声",
            "bgm", "ost",
        ],
        "phrase": ["中文配音"],
    },

    # =========================================================================
    # CONTENT & VALUE - 内容与价值
    # =========================================================================
    "Content": {
        "single_char": [],
        "compound": [
            "内容", "时长", "流程", "耐玩", "通关",
            "主线", "支线", "收集", "小时", "体量",
            "注水", "重复", "换皮", "多周目",
            "dlc",
        ],
        "phrase": [
            "素材复用", "拖时长", "强行延长", "通关后",
        ],
    },

    "Monetization": {
        "single_char": [],
        "compound": [
            # ex-Price
            "价格", "定价", "值得", "不值", "贵", "便宜",
            "打折", "史低", "入手", "白嫖", "性价比",
            # ex-Microtransactions
            "氪金", "内购", "充值", "抽卡", "648",
            "课金", "首充", "月卡", "战令", "季票",
            "开箱", "箱子", "钥匙", "保底", "抽奖",
            "p2w",
        ],
        "phrase": [
            "通行证", "pay to win",
        ],
    },

    # =========================================================================
    # MULTIPLAYER & COMMUNITY - 多人与社区
    # =========================================================================
    "Multiplayer": {
        "single_char": [],
        "compound": [
            "联机", "多人", "匹配", "服务器", "延迟",
            "掉线", "开黑", "组队", "单机", "野排", "车队",
            "单排", "组排", "路人", "挂机",
            "pvp", "pve", "coop",
        ],
        "phrase": [
            "坑比", "猪队友", "送人头",
        ],
    },

    "Community": {
        "single_char": [],
        "compound": [
            "社区", "玩家", "汉化",
            "官方", "民间",
            "mod", "mods",
        ],
        "phrase": ["创意工坊"],
    },

    # =========================================================================
    # CONTROLS & UI - 操控与界面
    # =========================================================================
    "Controls": {
        "single_char": [],
        "compound": [
            "操作", "手感", "手柄", "键鼠", "键盘",
            "摇杆", "触发", "键位", "改键",
            "死区", "陀螺仪", "扳机", "震动",
        ],
        "phrase": [
            "自定义键位", "辅助瞄准", "触觉反馈", "自适应扳机",
        ],
    },

    "UI": {
        "single_char": [],
        "compound": [
            "界面", "菜单", "字幕", "字体",
            "中文", "汉化",
            "ui", "hud",
        ],
        "phrase": [],
    },

    # =========================================================================
    # STORY & NARRATIVE - 剧情
    # =========================================================================
    "Story": {
        "single_char": [],
        "compound": [
            "剧情", "故事", "人物", "角色", "结局",
            "剧本", "叙事", "世界观", "背景", "喂屎",
            "烂尾", "降智", "工具人", "脸谱化",
            "剧情杀", "都合主义",
            "npc",
        ],
        "phrase": ["逻辑硬伤"],
    },

    # =========================================================================
    # DEVELOPER SUPPORT - 开发支持
    # =========================================================================
    "Support": {
        "single_char": [],
        "compound": [
            "更新", "修复", "维护", "开发商", "官方",
            "补丁", "版本",
        ],
        "phrase": [],
    },

    "Localization": {
        "single_char": [],
        "compound": [
            "本地化", "汉化", "翻译", "机翻", "缺字", "乱码",
            "繁体", "简体",
        ],
        "phrase": [
            "语言支持", "中文支持", "无中文", "不支援中文",
            "文本质量", "字幕翻译", "界面翻译",
        ],
    },

    # =========================================================================
    # REFINEMENT - 打磨
    # =========================================================================
    "Polish": {
        "single_char": [],
        "compound": [
            "打磨", "精致", "粗糙", "用心", "敷衍", "细节",
            "诚意", "偷懒", "不用心", "精良", "精美",
        ],
        "phrase": ["粗制滥造"],
    },

    # =========================================================================
    # RETENTION - 留存
    # =========================================================================
    "Retention": {
        "single_char": [],
        "compound": [
            # Positive (High Retention)
            "推荐", "安利", "入正", "入坑", "必玩",
            "神作", "年度", "满分",
            # Negative (Churn)
            "退款", "卸载", "弃坑", "劝退", "不推荐",
            "避雷", "踩雷", "退坑",
            "回坑", "出坑", "已弃",
        ],
        "phrase": [
            "坚持玩", "每天玩", "停不下来", "刷了",
            "已退", "退款了",
        ],
    },
}

# =============================================================================
# EXCLUSIONS (Context-aware filtering)
# =============================================================================
# Words to exclude when they appear in certain contexts.
# Format: "keyword": ["context_word1", "context_word2"]

EXCLUSIONS = {
    # "fps" as genre (FPS shooter) vs performance (60 fps)
    "fps": ["射击", "枪战", "第一人称"],
    # Empty for now - will be expanded based on false positives
}