代码工艺：写代码的好习惯

题目：

sql建表：

Create table If Not Exists Product (product_id int, product_name varchar(30))
Create table If Not Exists Sales (product_id int, period_start date, period_end date, average_daily_sales int)
Truncate table Product
insert into Product (product_id, product_name) values ('1', 'LC Phone ')
insert into Product (product_id, product_name) values ('2', 'LC T-Shirt')
insert into Product (product_id, product_name) values ('3', 'LC Keychain')
Truncate table Sales
insert into Sales (product_id, period_start, period_end, average_daily_sales) values ('1', '2019-01-25', '2019-02-28', '100')
insert into Sales (product_id, period_start, period_end, average_daily_sales) values ('2', '2018-12-01', '2020-01-01', '10')
insert into Sales (product_id, period_start, period_end, average_daily_sales) values ('3', '2019-12-01', '2020-01-31', '1')

分析：首先，我们看到题目后，肯定会感觉这个题很难，我看到的时候也是，这道题的难点是怎么把每个商品的每一年分开，我们这里可以使用递归，递归结束条件为period_end的年份，开始年份是period_start的年份，但是，我们这里要加一些字段，例如sales表中的所有字段，然后再根据特定的条件来计算每一年的天数，然后在乘以日销，在关联商品表获取商品名称，大概思路就是这个样子，这里文字不太好理解，直接看图解（这一篇写的很详细）：

sql实现：

with RECURSIVE
    year as (select product_id,
                    year(period_start) as year,
                    year(period_end)      a2,
                    period_start,
                    period_end,
                    average_daily_sales
             from Sales
             union
             select product_id, year + 1, a2, period_start, period_end, average_daily_sales
             from year
             where year < a2),        -- 递归输出每个商品的每一年，加上sales的字段
    t2 as (select product_id,
                  year,
                  if(year = a2 and year = year(period_start),
                     (datediff(period_end, period_start) + 1)*average_daily_sales,          -- 如果满足year = a2 and year = year(period_start)条件，然后计算period_end和period_start的差值
                     
                     if(year = a2, (datediff(period_end, date_format(concat(year, '-01-01'), '%Y-%m-%d'))+1)*average_daily_sales,           -- 如果满足year = a2 and year ！= year(period_start)条件，计算period_end和year+1的第一天的差值
                     
                        if(year != a2 and year = year(period_start),
                           (datediff(date_format(concat(year + 1, '-01-01'), '%Y-%m-%d'), period_start))*average_daily_sales,365*average_daily_sales))) as a -- 如果满足year != a2 and year = year(period_start) 返回year+的第一天与period_start的差值，否则返回365
                           
           from year)
select t2.product_id,product_name,date_format(concat(year,'-01-01'),'%Y') report_year ,a total_amount  
from t2,Product p1 where t2.product_id=p1.product_id
order by t2.product_id; -- 连接商品表获取商品姓名，然后返回需要字段，并转换类型

pandas例子：

data = [[1, 'LC Phone '], [2, 'LC T-Shirt'], [3, 'LC Keychain']]
product = pd.DataFrame(data, columns=['product_id', 'product_name']).astype({'product_id':'Int64', 'product_name':'object'})
data = [[1, '2019-01-25', '2019-02-28', 100], [2, '2018-12-01', '2020-01-01', 10], [3, '2019-12-01', '2020-01-31', 1]]
sales = pd.DataFrame(data, columns=['product_id', 'period_start', 'period_end', 'average_daily_sales']).astype({'product_id':'Int64', 'period_start':'datetime64[ns]', 'period_end':'datetime64[ns]', 'average_daily_sales':'Int64'})

pandas分析和sql思路一样，只不过是把递归改成了for循环

pandas实现：

import pandas as pd

def total_sales(product: pd.DataFrame, sales: pd.DataFrame) -> pd.DataFrame:
    sales['year']=sales['period_start'].dt.year.astype('Int64')
    sales['end'] = sales['period_end'].dt.year.astype('Int64')  -- 新增两列year和end，把开始时间和结束时间的年取出转换成int方便后面计算

    b = [[a,sales['end'].iloc[i],sales['product_id'].iloc[i],sales['period_start'].iloc[i],sales['period_end'].iloc[i],sales['average_daily_sales'].iloc[i]] for i in range(0, len(sales)) for a in range(sales['year'].iloc[i],sales['end'].iloc[i]+1)]  -- for循环输出每个商品的每年

    c=pd.DataFrame(b, columns=['year','end','product_id', 'period_start', 'period_end', 'average_daily_sales'])          -- 转换成dataframe对象并设置列名

-- 自定义函数
    def get_func(x):
    -- 如果满足year = a2 and year = year(period_start)条件，然后计算period_end和period_start的差值
        if x['year']==x['end'] and x['year']==int(str(x['period_start']).split('-')[0]):
            a=x['period_end']-x['period_start']
            a=str(a).split(' ')[0]
            # print(a)
            return (int(a)+1)*x['average_daily_sales']
   --  如果满足year = a2 and year ！= year(period_start)条件，计算period_end和year+1的第一天的差值
        elif x['year']==x['end'] and x['year']!=int(str(x['period_start']).split('-')[0]):
            a=x['period_end']-pd.to_datetime(str(x['year']))
            a=str(a).split(' ')[0]
            # print(a)
            return (int(a)+1)*x['average_daily_sales']
            
      --  如果满足year != a2 and year = year(period_start) 返回year+的第一天与period_start的差值，否则返回365
        elif x['year']!=x['end'] and x['year']==int(str(x['period_start']).split('-')[0]):
            a=pd.to_datetime(str(x['year']+1))-x['period_start']
            a=str(a).split(' ')[0]
            return (int(a))*x['average_daily_sales']
        else:
            return 365*x['average_daily_sales']
    c['total_amount']=c.apply(get_func,axis=1)  -- 调用自定义函数

    d=c.merge(product, on='product_id', how='inner')   --连接商品表获取表名

    d=d[['product_id','product_name','year','total_amount']].sort_values(by=['product_id','year'])   -- 取出需要的列，然后排序

    d.columns=['product_id','product_name','report_year','total_amount']  -- 修改列名
    d['report_year']=d['report_year'].astype('str')   -- 把year改成字符串类型
    return d