删除单个 Excel 工作表中的多个标题答案

【问题标题】：Removing multiple headers in a single excel sheet删除单个 Excel 工作表中的多个标题
【发布时间】：2022-02-10 00:20:21
【问题描述】：

类似于Remove header row in Excel using pandas

我希望从 Excel 工作表中删除标题。

使用相同的示例但已编辑：

Company ABC

Account Name 1
Account No.1
Description    group     pair          amount    ...  result
value1         value1    value1        value1    ...  value1
value2         value2    value2        value2    ...  value2
totals                               sum values

Account Name 2
Account No.2
Description    group     pair          amount    ...  result
value3         value3    value3        value3    ...  value3
value4         value4    value4        value4    ...  value4
totals                               sum values

Sales
00234
Description    group     pair          amount    ...  result
value5         value5    value5        value5    ...  value5
value6         value6    value6        value6    ...  value6
totals                               sum values

Inventory
00012345
Description    group     pair          amount    ...  result
value7         value7    value7        value7    ...  value7
value8         value8    value8        value8    ...  value8
value9         value9    value9        value9    ...  value9
totals                               sum values

我想把它连接起来变成像

cabinet_name   group     pair          amount    ...  result
value1         value1    value1        value1    ...  value1
value2         value2    value2        value2    ...  value2
value3         value3    value3        value3    ...  value3
value4         value4    value4        value4    ...  value4
value5         value5    value5        value5    ...  value5
value6         value6    value6        value6    ...  value6
value7         value7    value7        value7    ...  value7
value8         value8    value8        value8    ...  value8
value9         value9    value9        value9    ...  value9

我已经设法通过跳过行来删除顶部标题，例如（skiprows = 4）。但是，这仍然会留下像这样附加的其他标题和总计：

cabinet_name   group     pair          amount    ...  result
value1         value1    value1        value1    ...  value1
value2         value2    value2        value2    ...  value2
totals                               sum values
Account Name 2
Account No.2
cabinet_name   group     pair          amount    ...  result
value3         value3    value3        value3    ...  value3
value4         value4    value4        value4    ...  value4
totals                               sum values
Sales
00234
Description    group     pair          amount    ...  result
value5         value5    value5        value5    ...  value5
value6         value6    value6        value6    ...  value6
totals                               sum values
Inventory
00012345
Description    group     pair          amount    ...  result
value7         value7    value7        value7    ...  value7
value8         value8    value8        value8    ...  value8
value9         value9    value9        value9    ...  value9
totals                               sum values

如果有人能告诉我如何用 pandas 清理这张表，我将不胜感激，因为我在网上看到的只是一张 Excel 表中的一张表。

如果我遗漏了什么，请告诉我，我很乐意编辑这个问题。

我认为这可能有用，

我在excel中清理文件的常规过程是先删除前4行，只留下

cabinet_name   group     pair          amount    ...  result
value1         value1    value1        value1    ...  value1
value2         value2    value2        value2    ...  value2
totals         *blank*   *blank*       sum values

Account Name 2
Account No.2
cabinet_name   group     pair          amount    ...  result
value3         value3    value3        value3    ...  value3
value4         value4    value4        value4    ...  value4
totals         *blank*   *blank*       sum values

Account Name 3
Account No.3
cabinet_name   group     pair          amount    ...  result
value5         value5    value5        value5    ...  value5
value6         value6    value6        value6    ...  value6
totals         *blank*   *blank*       sum values

然后我将过滤 group 或 pair 以在所述列中查找空白值并将其删除。

这是

的结果

print(df_total.head(8).to_dict())

import datetime
from numpy import nan
{'Date': {0: nan, 1: datetime.datetime(2021, 1, 1, 0, 0), 2: datetime.datetime(2021, 1, 1, 0, 0), 3: datetime.datetime(2021, 1, 29, 0, 0), 4: datetime.datetime(2021, 1, 31, 0, 0), 5: 
'Totals', 6: 'Net difference', 7: nan}, 
'Journal number': {0: nan, 1: 'AX009473', 2: 'AX009473', 3: 'AX003312', 4: 'AX009641', 5: nan, 6: nan, 7: nan}, 
'Voucher': {0: nan, 1: 'TSPN-2021-3', 2: 'TSPN-2021-3', 3: 'GBJ-2021-1', 4: 'VIT-2021-1', 5: nan, 6: nan, 7: nan}, 
'Posting type': {0: nan, 1: nan, 2: nan, 3: 'Ledger journal', 4: 'Ledger journal', 5: nan, 6: nan, 7: nan}, 
'Ledger account': {0: nan, 1: '00388211', 2: '00388211', 3: '00388211', 4: '00388211', 5: nan, 6: nan, 7: nan}, 
'Description': {0: nan, 1: nan, 2: nan, 3: 'DISBERSMENT FOR PETROL', 4: 'TAXI FAIR', 5: nan, 6: nan, 7: nan}, 
'Unnamed: 6': {0: nan, 1: nan, 2: nan, 3: nan, 4: nan, 5: nan, 6: nan, 7: nan}, 
'Unnamed: 7': {0: nan, 1: 'SGD', 2: 'USD', 3: 'SGD', 4: 'SGD', 5: nan, 6: nan, 7: nan}, 
'Amount in transaction currency': {0: 'Debit', 1: 13.55, 2: 0, 3: 0, 4: 5, 5: nan, 6: nan, 7: nan}, 'Unnamed: 9': {0: 'Credit', 1: 0, 2: 25, 3: 52, 4: 0, 5: nan, 6: nan, 7: nan}, 
'Amount in accounting currency': {0: 'Debit', 1: 13.55, 2: 0, 3: 0, 4: 5, 5: 18.55, 6: nan, 7: nan}, 'Unnamed: 11': {0: 'Credit', 1: 0, 2: 33.42, 3: 52, 4: 0, 5: 85.42, 6: 66.87, 7: nan}, 
'Amount in reporting currency': {0: 'Debit', 1: 13.55, 2: 0, 3: 0, 4: 5, 5: 18.55, 6: nan, 7: nan}, 'Unnamed: 13': {0: 'Credit', 1: 0, 2: 33.42, 3: 52, 4: 0, 5: 85.42, 6: nan, 7: nan}, 
'Unnamed: 14': {0: nan, 1: nan, 2: nan, 3: nan, 4: nan, 5: nan, 6: nan, 7: nan}, 'Unnamed: 15': {0: nan, 1: nan, 2: nan, 3: nan, 4: nan, 5: nan, 6: nan, 7: nan}, 'Unnamed: 16': {0: nan, 1: nan, 2: nan, 3: nan, 4: nan, 5: nan, 6: nan, 7: nan}}

创建了我所拥有的样本 excel

转换数据的 excel

【问题讨论】：

stackoverflow.com/questions/36688022/…
@AndrewRyan 谢谢！马上去看看
嗯，这仍然只适用于每张 1 张桌子。我目前在同一张表中有多个表。除非我对这个问题的理解是错误的......您能否进一步详细说明？ :)

标签： python pandas dataframe append concatenation

【解决方案1】：

因此，我们可以通过将第一行添加到列名来稍微修改列名（这更正了名称以“未命名”开头的所需列）。然后过滤掉以“Unnamed”开头的列名（其余不需要的），然后使用“Date”列，创建一个掩码并过滤DataFrame：

df = df.rename(columns={**{f'Unnamed: {i}': j for i,j in zip((9,11,13), 
                                                             ('Amount in transaction currency',
                                                              'Amount in accounting currency',
                                                              'Amount in reporting currency'))}, 
                        **{'Unnamed: 7': 'Currency'}})

df.columns = [f'{col}_{first}' if first==first else col for col, first in zip(df.columns, df.loc[0])]
df = df[df.columns[~df.columns.str.startswith('Unnamed')]]
date_filter = df['Date'].apply(isinstance, args=(datetime.datetime,))
df = df[date_filter]

上面的代码用一个显式的循环来构造列表来修改列名：

cols = {'Unnamed: 7': 'Currency', 'Unnamed: 9': 'Amount in transaction currency', 
        'Unnamed: 11': 'Amount in accounting currency', 'Unnamed: 13': 'Amount in reporting currency'}
df = df.rename(columns=cols)

another_cols = []
for col, first in zip(df.columns, df.loc[0]):
    if first==first:
        another_cols.append(f'{col}_{first}')
    else:
        another_cols.append(col)
df.columns = another_cols
df = df[df.columns[~df.columns.str.startswith('Unnamed')]]
date_filter = df['Date'].apply(isinstance, args=(datetime.datetime,))
df = df[date_filter]

输出：

                  Date Journal number      Voucher    Posting type  \
1  2021-01-01 00:00:00       AX009473  TSPN-2021-3             NaN   
2  2021-01-01 00:00:00       AX009473  TSPN-2021-3             NaN   
3  2021-01-29 00:00:00       AX003312   GBJ-2021-1  Ledger journal   
4  2021-01-31 00:00:00       AX009641   VIT-2021-1  Ledger journal   

  Ledger account             Description Currency  \
1       00388211                     NaN      SGD   
2       00388211                     NaN      USD   
3       00388211  DISBERSMENT FOR PETROL      SGD   
4       00388211               TAXI FAIR      SGD   

  Amount in transaction currency_Debit Amount in transaction currency_Credit  \
1                                13.55                                     0   
2                                    0                                    25   
3                                    0                                    52   
4                                    5                                     0   

  Amount in accounting currency_Debit Amount in accounting currency_Credit  \
1                               13.55                                    0   
2                                   0                                33.42   
3                                   0                                   52   
4                                   5                                    0   

  Amount in reporting currency_Debit Amount in reporting currency_Credit  
1                              13.55                                   0  
2                                  0                               33.42  
3                                  0                                  52  
4                                  5                                   0

【讨论】：

@HOA ** 是字典解包操作符。我们通过将两个字典解包到另一个字典中来组合两个字典。它用于编辑一些列名； rename 方法可以使用字典重命名列。
非常感谢恩克！这真的帮助我理解了你的代码！赞！如果我们可以在链接中连接，是否有可能？

【解决方案2】：

据推测，前 3 行以及带有“总计”和“总和值”的行都有空单元格，因此dropna 应该消除这些行。然后drop_duplicates 和keep=False 参数应该删除重复的列名：

out = df.replace('', np.nan).replace(' ', np.nan).dropna().drop_duplicates()

我无法测试它，因为您的 read_clipboard 抱怨您的数据格式。

【讨论】：

【解决方案3】：

您可以尝试像这样删除不需要的行。

df= df[~df['first_column_name'].str.startswith(('Company name','Account Name','Account No.','cabinet_name'))]

【讨论】：

谢谢，马上试试！
嗨！感谢您的回答，帐号名称和帐号是唯一不同的。有解决方案吗？我将编辑我的问题以使其更明显
str.startswith(()) 检查并删除以“帐号”开头的行。所以它会删除所有以“Account No”开头的行。因此，“帐号”之后的内容无关紧要。希望对你有帮助
您好，这是个好主意！但是，帐户名称和编号是一个唯一字段，我希望通过不对名称进行硬编码来自动化该过程，例如Inventory 001324 另一个表是 Sales 1100 它没有特定的可识别名称