# -*- coding: utf-8 -*-

"""
脚本说明：
本脚本用于KANO需求分析，能够处理问卷数据并输出需求分类结果
"""

import re

import numpy
import pandas

from utils.pandas_extension import save_as_workbook

print("1 打开并读取Excel文件...", end="")

try:
    dataset = pandas.read_excel(io="KANO模型客户调研问卷.xlsx", sheet_name="问卷结果")

    # 选项范围
    alternatives = ["非常满意", "理应如此", "无所谓", "勉强接受", "不满意"]

    # 数据清洗
    dataset = (
        dataset.iloc[
            :, 3:
        ]  # 原始数据第一列为编号、第二列为提交人、第三列为提交时间，从第四列到最后一列为选项，删除第一列至第三列
        .loc[lambda dataframe: dataframe.nunique(axis=1) != 1]  # 删除相同选项的样本
        .map(
            lambda cell: cell if cell in alternatives else pandas.NA
        )  # 检查是否在选项范围，若不在选项范围则置为缺失值
        .dropna(axis="index", how="any")  # 删除缺失值
    )

    # 统计样本数
    samples_size = dataset.shape[0]
    # 若样本数为0则抛出异常
    if samples_size == 0:
        raise Exception("样本数为0")

    # 统计列数
    columns = dataset.shape[1]
    # 若列数非奇数则抛出异常
    if columns % 2 != 0:
        raise Exception("列数为奇数")

    print(f"已完成，样本数为{samples_size}")

except Exception as exception:
    print(f"发生异常：{str(exception)}")
    exit()

print("2 生成KANO评价结果...", end="")

# KANO评价结果
kano = pandas.DataFrame()

for column in range(columns):
    if column % 2 != 0:
        continue

    # 功能名称
    feature = re.search(
        pattern="【(?P<feature>.*?)】", string=dataset.columns[column]
    ).group("feature")

    # 生成某功能的选项分布
    distribution = pandas.crosstab(
        index=dataset.iloc[:, column],  # 行索引为如果该产品有{{功能名称}}的选项
        columns=dataset.iloc[:, column + 1],  # 列索引为如果该产品没有{{功能名称}}的选项
        rownames=[""],
        colnames=[""],
        normalize=True,  # 转为占比
    ).reindex(index=alternatives, columns=alternatives, fill_value=0)

    kano = pandas.concat(
        objs=[
            kano,
            pandas.DataFrame(
                [
                    {
                        "功能名称": feature,
                        "魅力属性": distribution.loc[
                            "非常满意", ["理应如此", "无所谓", "勉强接受"]
                        ].sum(),
                        "期望属性": distribution.loc["非常满意", "不满意"].sum(),
                        "必备属性": distribution.loc[
                            ["理应如此", "无所谓", "勉强接受"], "不满意"
                        ].sum(),
                        "无差异属性": distribution.loc[
                            ["理应如此", "无所谓", "勉强接受"],
                            ["理应如此", "无所谓", "勉强接受"],
                        ]
                        .sum()
                        .sum(),
                        "反向属性": distribution.loc[
                            ["理应如此", "无所谓", "勉强接受", "不满意"], "非常满意"
                        ].sum()
                        + distribution.loc[
                            ("不满意", ["理应如此", "无所谓", "勉强接受"])
                        ].sum(),
                        "可疑结果": distribution.loc["非常满意", "非常满意"]
                        + distribution.loc["不满意", "不满意"],
                    }
                ]
            ),
        ],
        ignore_index=True,
    )

# 计算better-worse分数，并贴标签
kano = (
    kano.assign(
        满意系数=lambda dataframe: (dataframe["魅力属性"] + dataframe["期望属性"])
        / (
            dataframe["魅力属性"]
            + dataframe["期望属性"]
            + dataframe["必备属性"]
            + dataframe["无差异属性"]
        ),
        不满意系数=lambda dataframe: (dataframe["必备属性"] + dataframe["期望属性"])
        / (
            dataframe["魅力属性"]
            + dataframe["期望属性"]
            + dataframe["必备属性"]
            + dataframe["无差异属性"]
        ),
    )
    .pipe(
        lambda dataframe: dataframe.assign(
            需求标签=numpy.select(
                condlist=[
                    (dataframe["满意系数"] >= dataframe["满意系数"].mean())
                    & (dataframe["不满意系数"] >= dataframe["不满意系数"].mean()),
                    (dataframe["满意系数"] >= dataframe["满意系数"].mean())
                    & (dataframe["不满意系数"] < dataframe["不满意系数"].mean()),
                    (dataframe["满意系数"] < dataframe["满意系数"].mean())
                    & (dataframe["不满意系数"] < dataframe["不满意系数"].mean()),
                    (dataframe["满意系数"] < dataframe["满意系数"].mean())
                    & (dataframe["不满意系数"] >= dataframe["不满意系数"].mean()),
                ],
                choicelist=[
                    "期望型需求(P2)",
                    "魅力型需求(P3)",
                    "无差异需求(P4)",
                    "必备型需求(P1)",
                ],
                default="未分类",
            ),
        )
    )
    .assign(
        魅力属性=lambda dataframe: dataframe["魅力属性"].apply(
            lambda cell: f"{cell:.2f}"
        ),
        期望属性=lambda dataframe: dataframe["期望属性"].apply(
            lambda cell: f"{cell:.2f}"
        ),
        必备属性=lambda dataframe: dataframe["必备属性"].apply(
            lambda cell: f"{cell:.2f}"
        ),
        无差异属性=lambda dataframe: dataframe["无差异属性"].apply(
            lambda cell: f"{cell:.2f}"
        ),
        反向属性=lambda dataframe: dataframe["反向属性"].apply(
            lambda cell: f"{cell:.2f}"
        ),
        可疑结果=lambda dataframe: dataframe["可疑结果"].apply(
            lambda cell: f"{cell:.2f}"
        ),
        满意系数=lambda dataframe: dataframe["满意系数"].apply(
            lambda cell: f"{cell:.2f}"
        ),
        不满意系数=lambda dataframe: dataframe["不满意系数"].apply(
            lambda cell: f"{cell:.2f}"
        ),
    )
)

save_as_workbook(worksheets=[("Sheet1", kano)], workbook_name="result.xlsx")

print("已完成")