python - 在Python中使用panda读取文件时忽略空DataFrame

Question

我有一个这样的txt文件:

`Empty DataFrame 
 Columns: [0, 1, 2, 3, 4]
 Index: []
 Empty DataFrame
 Columns: [0, 1, 2, 3, 4]
 Index: []
                       0                         1                           2  \
46   RNA/4v6p.csv,46AA/U/551    RNA/4v6p.csv,46AA/A/33         RNA/4v6p.csv,46WW_cis   
47   RNA/4v6p.csv,46AA/G/550    RNA/4v6p.csv,46AA/C/34         RNA/4v6p.csv,46WW_cis   
48   RNA/4v6p.csv,46AA/A/553    RNA/4v6p.csv,46AA/U/30         RNA/4v6p.csv,46WW_cis   
49   RNA/4v6p.csv,46AA/U/552    RNA/4v6p.csv,46AA/A/33         RNA/4v6p.csv,46WW_cis   
50   RNA/4v6p.csv,46AA/U/1199   RNA/4v6p.csv,46AA/G/1058       RNA/4v6p.csv,46WW_cis   

     3   4  
46 NaN NaN  
47 NaN NaN  
48 NaN NaN  
49 NaN NaN  
50 NaN NaN`

我想将其读入一个 3 列的数组中。现在我尝试使用 pd.read_csv(self.filename,delim_whitespace=True) ，但这在尝试读取空 DataFrame 部分时给我带来了很多错误。如何让程序忽略这部分？

编辑: 最佳解决方案是如果我的文件中没有空数据帧。该文件是在很多文件中查找的结果，其中有一些是空的。我以为我已经通过给出异常来过滤空文件，这样在空文件中搜索的效果就不会存储在结果中。我想我是以错误的方式做的。有人可以纠正我吗？

from numpy import numpy.mean as nm
def find_same_direction_chain(self, results):
         separation= lambda x: pd.Series([i for i in x.split('/')])
         left_chain=self.data[0].apply(separation)
         right_chain=self.data[1].apply(separation)
         i=1
         try:
            while i<len(self.data[:])-5:
                if nm(left_chain[2][i:i+3])>=nm(left_chain[2][i+2:i+5])  and nm(right_chain[2][i:i+3])>=nm(right_chain[2][i+2:i+5]) and len(self.data[:])>0:   
                    if nm(left_chain[2][i+2:i+5])>=nm(left_chain[2][i+4:i+7])  and nm(right_chain[2][i+2:i+5])>=nm(right_chain[2][i+4:i+7]):   
                        results.chains.append(str(self.filename+", "+str(i)+self.data[0:3][i:i+5]))

                else: pass
                i+=1
         except ValueError:
                    results.bin.append(self.filename)
         except TypeError:
                    results.data_structure_error.append(self.filename)

Answer 1

您可以使用:

import pandas as pd
import io

temp=u"""Empty DataFrame 
 Columns: [0, 1, 2, 3, 4]
 Index: []
 Empty DataFrame
 Columns: [0, 1, 2, 3, 4]
 Index: []
                       0                         1                           2  \
46   RNA/4v6p.csv,46AA/U/551    RNA/4v6p.csv,46AA/A/33         RNA/4v6p.csv,46WW_cis   
47   RNA/4v6p.csv,46AA/G/550    RNA/4v6p.csv,46AA/C/34         RNA/4v6p.csv,46WW_cis   
48   RNA/4v6p.csv,46AA/A/553    RNA/4v6p.csv,46AA/U/30         RNA/4v6p.csv,46WW_cis   
49   RNA/4v6p.csv,46AA/U/552    RNA/4v6p.csv,46AA/A/33         RNA/4v6p.csv,46WW_cis   
50   RNA/4v6p.csv,46AA/U/1199   RNA/4v6p.csv,46AA/G/1058       RNA/4v6p.csv,46WW_cis   

     3   4  
46 NaN NaN  
47 NaN NaN  
48 NaN NaN  
49 NaN NaN  
50 NaN NaN"""

#after testing replace io.StringIO(temp) to filename
df = pd.read_csv(io.StringIO(temp), delim_whitespace=True, names=range(7))

#remove rows with NaN in columns 0 - 3
df = df.dropna(subset=[0,1,2,3])

#remove rows where first column contains text 'Columns'
df = df[~df.iloc[:,0].str.contains('Columns')] 

#shift first row
df.iloc[0,:] = df.iloc[0,:].shift(-3)

#set first column to index
df = df.set_index(df.iloc[:,0])
#remove unnecessary columns
df = df.drop([0,4,5,6], axis=1)
print df
                           1                         2                      3
0                                                                            
46   RNA/4v6p.csv,46AA/U/551    RNA/4v6p.csv,46AA/A/33  RNA/4v6p.csv,46WW_cis
47   RNA/4v6p.csv,46AA/G/550    RNA/4v6p.csv,46AA/C/34  RNA/4v6p.csv,46WW_cis
48   RNA/4v6p.csv,46AA/A/553    RNA/4v6p.csv,46AA/U/30  RNA/4v6p.csv,46WW_cis
49   RNA/4v6p.csv,46AA/U/552    RNA/4v6p.csv,46AA/A/33  RNA/4v6p.csv,46WW_cis
50  RNA/4v6p.csv,46AA/U/1199  RNA/4v6p.csv,46AA/G/1058  RNA/4v6p.csv,46WW_cis

或者使用 read_csv 中的 skiprows 解决方案:

#after testing replace io.StringIO(temp) to filename
df = pd.read_csv(io.StringIO(temp), delim_whitespace=True, names=range(7), skiprows=6)

#remove rows with NaN
df = df.dropna(subset=[0,1,2,3])

#shift first row
df.iloc[0,:] = df.iloc[0,:].shift(-3)

#set first column to index
df = df.set_index(df.iloc[:,0])
#remove unnecessary columns
df = df.drop([0,4,5,6], axis=1)
print df
                           1                         2                      3
0                                                                            
46   RNA/4v6p.csv,46AA/U/551    RNA/4v6p.csv,46AA/A/33  RNA/4v6p.csv,46WW_cis
47   RNA/4v6p.csv,46AA/G/550    RNA/4v6p.csv,46AA/C/34  RNA/4v6p.csv,46WW_cis
48   RNA/4v6p.csv,46AA/A/553    RNA/4v6p.csv,46AA/U/30  RNA/4v6p.csv,46WW_cis
49   RNA/4v6p.csv,46AA/U/552    RNA/4v6p.csv,46AA/A/33  RNA/4v6p.csv,46WW_cis
50  RNA/4v6p.csv,46AA/U/1199  RNA/4v6p.csv,46AA/G/1058  RNA/4v6p.csv,46WW_cis

编辑:

您可以尝试更改(我没有示例数据，因此未经测试):

results.chains.append(str(self.filename+", "+str(i)+self.data[0:3][i:i+5]))

至:

if len(self.data[0:3][i:i+5]) > 0:                      
    results.chains.append(str(self.filename+", "+str(i)+self.data[0:3][i:i+5]))