Pandas read_csv读取TXT文件列分割不正确的解决办法

问题描述 #

我有一个TXT文件，尝试用以下代码读取：

df = pd.read_csv(path, sep='\t', encoding='utf-8')

但列未正确分割，得到的结果如下（忽略索引）：

"",t,I_m [A] 4.002,Tm [Nm],V_bus [V] 5.005,n [rpm] 0.010,T_inv [ºC] 7.034,T_acu [%] 4.019,f_m [Hz] 5.001,V_m [V] 5.002,P_m [kW] 5.003,n_ref [rpm] 3.001,I_m_par [A] 4.002,P_m_par [W] 5.003,Unnamed: 13
0,12:00:52,0.0,0.0,560.0,0.0,51.0,0.0,0.0,0.0,0.0,0.0,0.0,NaN,NaN
1,12:00:52,0.0,0.0,560.0,0.0,51.0,0.0,0.0,0.0,0.0,0.0,0.0,NaN,NaN
2,12:00:52,0.0,0.0,560.0,0.0,51.0,0.0,0.0,0.0,0.0,0.0,0.0,NaN,NaN
3,12:00:52,0.0,0.0,561.0,0.0,51.0,0.0,0.0,0.0,0.0,0.0,0.0,NaN,NaN
4,12:00:52,0.0,0.0,561.0,0.0,51.0,0.0,0.0,0.0,0.0,0.0,0.0,NaN,NaN

解决方案 #

1. 自动识别空白分隔符 #

你的TXT文件实际分隔符大概率是多个空格而非制表符，直接用delim_whitespace=True参数让Pandas自动识别任意空白字符（空格、制表符等）作为分隔符：

df = pd.read_csv(path, delim_whitespace=True, encoding='utf-8')

2. 用正则匹配任意数量空格 #

如果明确是空格分隔，也可以用正则表达式匹配任意数量的空格作为分隔符：

df = pd.read_csv(path, sep='\s+', encoding='utf-8')

3. 清理表头冗余内容 #

从结果看，表头带有多余的数字后缀（比如I_m [A] 4.002），如果不需要这些内容，读取后可进一步处理表头：

# 读取文件
df = pd.read_csv(path, delim_whitespace=True, encoding='utf-8')
# 清理表头，保留括号及之前的核心名称
df.columns = [col.split(' ')[0] if '[' in col else col for col in df.columns]

内容的提问来源于stack exchange，提问作者pepopi1