brec-analysis-tools/blindbox_analysis.py

import argparse
from datetime import timedelta
import os
import traceback
from typing import Any, TypedDict
import pandas as pd

from danmaku_reader import DanmakuReader
from structs import Percentage

class BlindboxItemData(TypedDict):
    price: int
    probability: float

class BlindboxInfo(TypedDict):
    price: int
    items: dict[str, BlindboxItemData]

blindbox_config: dict[str, BlindboxInfo] = {
    "星月盲盒": {
        "price": 50,
        "items": {
            "落樱缤纷": {
                "price": 600,
                "probability": 0.75
            },
            "星河入梦": {
                "price": 199,
                "probability": 1
            },
            "冲鸭": {
                "price": 99,
                "probability": 10.5
            },
            "少女祈祷": {
                "price": 66,
                "probability": 20
            },
            "情书": {
                "price": 52,
                "probability": 23.15
            },
            "星与月": {
                "price": 25,
                "probability": 24
            },
            "小蛋糕": {
                "price": 15,
                "probability": 20.6
            }
        }
    },
    "幸运盲盒": {
        "price": 50,
        "items": {
            "幸运泡泡": {
                "price": 15,
                "probability": 49.6
            },
            "星光铃铛": {
                "price": 52,
                "probability": 42.2
            },
            "梦雾纸签": {
                "price": 100,
                "probability": 5
            },
            "福灵小兽": {
                "price": 200,
                "probability": 2.4
            },
            "星愿花园": {
                "price": 600,
                "probability": 0.8
            }
        }
    },
    "心动盲盒": {
        "price": 150,  # 请根据实际盲盒价格填写
        "items": {
            "浪漫城堡": {
                "price": 22330,
                "probability": 0.04
            },
            "蛇形护符": {
                "price": 2000,
                "probability": 0.08
            },
            "时空之站": {
                "price": 1000,
                "probability": 0.12
            },
            "绮彩权杖": {
                "price": 400,
                "probability": 3.7
            },
            "爱心抱枕": {
                "price": 160,
                "probability": 45.56
            },
            "棉花糖": {
                "price": 90,
                "probability": 44.5
            },
            "电影票": {
                "price": 20,
                "probability": 6
            }
        }
    },
    "至尊盲盒": {
        "price": 1000,  # 请根据实际盲盒价格调整
        "items": {
            "奇幻之城": {
                "price": 32000,
                "probability": 0.6
            },
            "金蛇献福": {
                "price": 5000,
                "probability": 0.2
            },
            "蛇形护符": {
                "price": 2000,
                "probability": 1.45
            },
            "星际启航": {
                "price": 1010,
                "probability": 42
            },
            "许愿精灵": {
                "price": 888,
                "probability": 34
            },
            "绮彩权杖": {
                "price": 400,
                "probability": 19
            },
            "璀璨钻石": {
                "price": 200,
                "probability": 2.75
            }
        }
    }
}

gift_to_blindbox: dict[str, str] = {}
for box_name, box_info in blindbox_config.items():
    for item_name in box_info["items"].keys():
        gift_to_blindbox[item_name] = box_name

def analyze_blindbox_file(filepath: str) -> pd.DataFrame:
    """分析单个盲盒数据文件"""
    if not os.path.isfile(filepath):
        raise FileNotFoundError(f"文件未找到: {filepath}")

    danmaku_data = DanmakuReader(filepath)

    datalist = {
        "uid": [],
        "username": [],
        "blindbox_name": [],
        "blindbox_price": [],
        "item_name": [],
        "item_price": [],
        "item_probability": [],
        "profit": [],
        "time": []
    }

    for gift_info in danmaku_data.gift_list:
        gift_name = gift_info.giftname
        if gift_name in gift_to_blindbox:
            box_name = gift_to_blindbox[gift_name]
            box_info = blindbox_config[box_name]
            item_info = box_info["items"][gift_name]
            item_time = danmaku_data.record_info.start_time + timedelta(seconds=gift_info.timestamp)

            datalist["uid"].append(gift_info.uid)
            datalist["username"].append(gift_info.user)
            datalist["blindbox_name"].append(box_name)
            datalist["blindbox_price"].append(box_info["price"])
            datalist["item_name"].append(gift_name)
            datalist["item_price"].append(item_info["price"])
            datalist["item_probability"].append(item_info["probability"])
            datalist["profit"].append(item_info["price"] - box_info["price"])
            datalist["time"].append(item_time.strftime("%Y-%m-%d %H:%M:%S"))

    return pd.DataFrame(datalist)

def analysis_file(filepath: str) -> pd.DataFrame:
    """分析单个盲盒数据文件"""
    try:
        print(f"正在分析文件: {filepath}")
        df = analyze_blindbox_file(filepath)
        df.sort_values(by="time", inplace=True)
        df.reset_index(drop=True, inplace=True)
        return df
    except Exception as e:
        print(f"分析文件时出错: {e}")
        traceback.print_exc()
        return pd.DataFrame()

def analysis_directory(directory: str) -> pd.DataFrame:
    """分析指定目录下所有的盲盒数据"""
    df = pd.DataFrame()
    for filename in os.listdir(directory):
        if filename.endswith('.xml'):
            filepath = os.path.join(directory, filename)
            try:
                print(f"正在分析文件: {filepath}")
                sub_df = analyze_blindbox_file(filepath)
                if df.empty:
                    df = sub_df
                elif not sub_df.empty:
                    df = pd.concat([df, sub_df], ignore_index=True)
            except Exception as e:
                print(f"分析文件时出错: {e}")
                traceback.print_exc()
    
    df.sort_values(by="time", inplace=True)
    df.reset_index(drop=True, inplace=True)
    return df

def profit_statistic(df: pd.DataFrame) -> dict[str, dict[str, Any]]:
    """对分析结果进行统计"""
    if df.empty:
        return {}
    
    # 统计每个盲盒中的各个物品的数量和总利润
    stats = {}
    grouped = df.groupby("blindbox_name")
    for box_name, group in grouped:
        box_config = blindbox_config.get(box_name, {})
        total_count = group["item_name"].count()
        total_investment = (group["blindbox_price"]).sum()
        total_revenue = (group["item_price"]).sum()
        total_profit = group["profit"].sum()

        box_stats = {
            "总数": total_count,
            "总投入": total_investment,
            "总收益": total_revenue,
            "总利润": total_profit,
            "盈亏比例": Percentage(total_profit / total_investment if total_investment > 0 else 0)
        }
        
        # 统计每个物品的出现概率
        item_counts = group["item_name"].value_counts()
        item_stats = {}
        for item_name, count in item_counts.items():
            item_config = box_config.get("items", {}).get(item_name, {})
            item_stats[item_name] = {
                "出现次数": count,
                "出现概率": Percentage(count / total_count if total_count > 0 else 0),
                "预期概率": Percentage(item_config.get("probability", 0) / 100),
            }

        sorted_item_stats = {}
        for item_name in box_config.get("items", {}).keys():
            if item_name in item_stats:
                sorted_item_stats[item_name] = item_stats[item_name]
            else:
                sorted_item_stats[item_name] = {
                    "出现次数": 0,
                    "出现概率": Percentage(0),
                    "预期概率": Percentage(box_config["items"][item_name].get("probability", 0) / 100),
                }

        box_stats["物品统计"] = sorted_item_stats

        stats[box_name] = box_stats

    return stats

def run_statistics(df: pd.DataFrame) -> dict[str, dict[str, Any]]:
    """运行统计分析"""
    stats = {}

    # 统计整体数据
    total_stats = profit_statistic(df)
    stats["整体"] = total_stats

    # 统计去掉每个用户每天第一次开盲盒的数据
    def remove_first_record_per_group(group):
        """去掉每组的第一条记录"""
        return group.iloc[1:]  # 从第二条开始返回

    # 确保时间列是datetime格式
    if 'time' in df.columns:
        df['time'] = pd.to_datetime(df['time'])
        # 提取日期部分（去掉时间）
        df['date'] = df['time'].dt.date
    # 按日期、用户ID、盲盒名称分组
    grouped = df.groupby(['date', 'username', 'blindbox_name'])
    # 去掉每组的第一条数据
    df_filtered = grouped.apply(remove_first_record_per_group).reset_index(drop=True)
    filtered_stats = profit_statistic(df_filtered)
    stats["去除保底"] = filtered_stats

    # 统计周五以外的数据
    non_friday_df = df[pd.to_datetime(df["time"]).dt.weekday != 4]
    non_friday_stats = profit_statistic(non_friday_df)
    stats["非周五"] = non_friday_stats
    
    # 统计周五的数据
    friday_df = df[pd.to_datetime(df["time"]).dt.weekday == 4]
    friday_stats = profit_statistic(friday_df)
    stats["周五"] = friday_stats

    return stats

def print_tree(tree_data: dict[str, Any], indent: int = 0, current_depth: int = 0):
    """以树状结构打印统计结果"""
    indent_str = "  " * indent
    for key, value in tree_data.items():
        if isinstance(value, dict):
            if current_depth == 0:
                print("=" * 40)
                print(f" {key}")
                print("=" * 40)
                print_tree(value, indent, current_depth + 1)
            else:
                print(f"{indent_str}{key}:")
                print_tree(value, indent + 1, current_depth + 1)
        else:
            print(f"{indent_str}{key}: {value}")

if __name__ == "__main__":
    arg_parser = argparse.ArgumentParser(description="盲盒数据分析工具")
    arg_parser.add_argument('-f', '--file', type=str, help='要分析的盲盒数据文件路径', action='append')
    arg_parser.add_argument('-d', '--directory', type=str, help='要分析的盲盒数据文件夹路径', action='append')
    arg_parser.add_argument('-i', '--inputcsv', type=str, help='从上次导出的分析结果继续分析', action='append', default=[])
    arg_parser.add_argument('-o', '--output', type=str, help='分析结果输出文件路径')

    args = arg_parser.parse_args()

    if (not args.file and not args.directory and not args.inputcsv):
        arg_parser.print_help()
        exit(1)
    elif not args.output and not args.inputcsv:
        print("请指定输出文件路径")
        arg_parser.print_help()
        exit(1)

    result_df = pd.DataFrame()
    if args.file:
        for file_path in args.file:
            sub_df = analysis_file(file_path)
            if result_df.empty:
                result_df = sub_df
            elif not sub_df.empty:
                result_df = pd.concat([result_df, sub_df], ignore_index=True)

    if args.directory:
        for dir_path in args.directory:
            sub_df = analysis_directory(dir_path)
            if result_df.empty:
                result_df = sub_df
            elif not sub_df.empty:
                result_df = pd.concat([result_df, sub_df], ignore_index=True)

    if args.inputcsv:
        for csv_path in args.inputcsv:
            if os.path.isfile(csv_path):
                try:
                    print(f"正在导入文件: {csv_path}")
                    sub_df = pd.read_csv(csv_path, encoding='utf-8-sig')
                    if result_df.empty:
                        result_df = sub_df
                    elif not sub_df.empty:
                        result_df = pd.concat([result_df, sub_df], ignore_index=True)
                except Exception as e:
                    print(f"导入文件时出错: {e}")
                    traceback.print_exc()
            else:
                print(f"文件未找到: {csv_path}")

    if not result_df.empty:
        result_df.to_csv(args.output, index=False, encoding='utf-8-sig')

        # 运行统计分析
        stats = run_statistics(result_df)
        print("\n统计结果:")
        print_tree(stats)
    else:
        print("未找到任何盲盒数据")