Question

以下是pandas数据框中的一列。我想获得所有独特的价值，如风格，颜色，味道，大小，香味名称......

Style: Lovenest - Newborn Pillow|||Color: Gray #Style and Color from this row
Style: Baby Calendula Face Cream #Style from this row
Color: Brown #Color from this row
Color: Matrix|||Item Package Quantity: 1 #Color & Item Package Quantity from this row
Color: Matrix|||Item Package Quantity: 1 #Color & Item Package Quantity from this row
Flavor: Baby Colic Babies Magic Tea|||Size: 1 Pack #Flavor & Size from this row
Scent Name: Sensitive|||Size: 100 Count (Pack of 6) #Scent Name & Size from this row
Scent Name: Sensitive|||Size: 100 Count (Pack of 6) #Scent Name & Size from this row

In [3]: df['variations'].head()
Out[3]: 
0    Style: Lovenest - Newborn Pillow|||Color: Gray
1    Style: Lovenest - Newborn Pillow|||Color: Gray
2    Style: Lovenest - Newborn Pillow|||Color: Gray
3    Style: Lovenest - Newborn Pillow|||Color: Gray
4    Flavor: Baby Colic Babies Magic Tea|||Size: 1 Pack
Name: variations, dtype: object

预期输出[＆＃39;样式＆＃39;＆＃39;颜色＆＃39;，＆＃39;风味＆＃39;＆＃39;尺寸＆＃39;]

Answer 1

以下代码可以起作用：

df_new = df['variations'].apply(lambda x: pd.Series({x.split(':')[0]:x.split(':')[1] for x in x.split('|||')}) if pd.notnull(x) else '')

df_new的列名是唯一的。

Answer 2

您可以使用字符串方法按某种模式拆分值。

稍后将每个字符串拆分为键值对并取出关键部分。

In[1]: df['variations'].str.split('\|\|\|').apply(lambda items: [item.split(':')[0] for item in items])
Out[1]: 
0                     [Style, Color]
1                     [Style, Color]
2                     [Style, Color]
3                     [Style, Color]
4                     [Style, Color]
Name: item, dtype: object

编辑：我看到你改变了输入和预期输出。如果你试图在列中获取所有键的集合，那么你可以在pandas中完成所有操作，而不必逐行迭代它。

In[1]: keys_list = df['variations'].apply(lambda x: [y.split(': ')[0] for y in x.split('|||')]).tolist()
In[2]: list(set([key for keys in keys_list for key in keys]))
Out[2]: ['Flavor', 'Item Package Quantity', 'Size', 'Color', 'Style', 'Scent Name']

Answer 3

In [25]: data = []
...: for x in df['variations']:
...:     if pd.notnull(x):
...:         d =  {x.split(':')[0]:x.split(':')[1] for x in x.split('|||')}
...:         vals = d.keys()
...:         data.extend(vals)
...:         
...: print list(set(data))
['Style', 'Material', 'Number of Items', 'Pattern', .....

如何使用pandas或regex提取以下内容

3 个答案: