如何在标头的行中使用带有更多分隔符的Pandas读取csv文件？

Question

I want to read a csv file using read_csv function from Pandas, the file has more delimiters in the rows that in the header. Pandas thinks the first columns are multi-index. The 'NAME' column can have an arbitrary number of delimiters and the affected column could be any one (we do not know which one is affected), even more than one.

I have tried to tune the key-word arguments of read_csv without success. I am using Python 3.7.0 and Pandas 0.25.0. However, Excel can read the file correctly.

import pandas

with open('test.csv', mode='w') as csv_file:
    csv_file.write('A,NAME,B\n')
    csv_file.write('a, Peter, Parker, b\n')

df = pandas.read_csv('test.csv', header=0, delimiter=',')
print(df)

Expected output:

   A            NAME   B
0  a   Peter, Parker   b

Actual output:

    A     NAME   B
a   Peter   Parker   b

Other example:

import pandas

with open('test.csv', mode='w') as csv_file:
    csv_file.write('A,NAME,B,PLACE\n')
    csv_file.write('a, Peter, Parker, b, Queens, New York City\n')

df = pandas.read_csv('test.csv', header=0, delimiter=',')
print(df)

Expected output:

   A            NAME   B                 PLACE
0  a   Peter, Parker   b Queens, New York City

Actual output:

                A NAME        B           PLACE
a  Peter   Parker    b   Queens   New York City

Answer 1

Isn't something like

df = pandas.read_csv('test.csv', header=0, delimiter=',')
df = df.reset_index()
df["NAME"] = df["A"] + ", " + df["NAME"]
df["A"] = df["Unnamed: 0"]
df = df.drop("Unnamed: 0", axis=1)

possible ? It is not completely answering this question but could do the trick for your df.

EDIT : Another possibility, if the file is also available in .xls/.xlsx format, pd.read_excel("name.xls") should solve your problem

Answer 2

A workaround:

with open('test.csv', mode='w') as csv_file:  
    csv_file.write('A,NAME,B\n')  
    csv_file.write('a, Peter, Parker, b\n')  
    csv_file.write('aa, John, Lee, Mary, bb\n')            

df=pd.DataFrame(columns=["A","NAMES","B"])                                                                           

with open("test.csv") as ff:  
   for line in ff:  
       A,N= line.split(",",maxsplit=1)  
       N,B= N.rsplit(",",maxsplit=1)  
       df.loc[len(df.index)]= [A.strip(),N.strip(),B.strip()] 

df.drop(0,axis="index")

    A            NAMES   B
1   a    Peter, Parker   b
2  aa  John, Lee, Mary  bb

Answer 3

# Read first line as a list using ',' as delimiter
with open('test.csv', 'r') as f:
  header = f.readline().replace('\n', '').split(',')
# Read file skipping first line (header) using two character delimiter ', '
df = pandas.read_csv('test.csv', skiprows = 1, header = None, delimiter=', ', engine = 'python')
header = header + ["missing-column"] # In your example file header has only 3 columns but data has 4
# Assign header list as dataframe columns names
df.columns = header
print(df)