将列表列表解析为字典到熊猫 DataFrame 时忽略错误

Question

如何告诉 pandas 忽略列表中不正确的字典项？

为简单起见，如果我有来自上一个问题的第一版解决方案的案例：

L =[['Manufacturer: Hyundai',
  'Model: Tucson',
  'Mileage: 258000 km',
  'Registered: 07/2019'],
 ['Manufacturer: Mazda',
  'Model: 6',
  'Year: 2014',
  'Registered: 07/2019',
  'Comfort',
  'Safety']]

df = pd.DataFrame([dict(y.split(':') for y in x) for x in L])
print (df)

第二个 dict 项有 2 个最后一项缺少值（'Comfort' 和 'Safety'），但它们也缺少“：”，因此 pandas 正在抛出：

ValueError: dictionary update sequence element #5 has length 1; 2 is required

如何告诉 pandas 忽略这些类型的错误并继续解析列表？

Answer 1

只需添加一点if 条件。

pd.DataFrame([
    dict(y.split(':') for y in x if ':' in y) for x in L])

  Manufacturer     Mileage    Model Registered   Year
0      Hyundai   258000 km   Tucson    07/2019    NaN
1        Mazda         NaN        6    07/2019   2014

如果您想将这些值包含为 NaN，请在推导式中将 if 更改为 if-else。

pd.DataFrame([
    dict(y.split(':') if ':' in y else (y, np.nan) for y in x) for x in L])


   Comfort Manufacturer     Mileage    Model Registered  Safety   Year
0      NaN      Hyundai   258000 km   Tucson    07/2019     NaN    NaN
1      NaN        Mazda         NaN        6    07/2019     NaN   2014

Answer 2

如果没有: 的值是键，则可以添加if-else：

df = pd.DataFrame([dict(y.split(':') if ':' in y else (y, np.nan) for y in x) for x in L])
print (df)
       Comfort Manufacturer     Mileage    Model Registered  Safety   Year
0      NaN      Hyundai   258000 km   Tucson    07/2019     NaN    NaN
1      NaN        Mazda         NaN        6    07/2019     NaN   2014