读取具有键值对的文本文件，并使用python pandas将每一行转换为一个字典

Question

I have a text file (one.txt) that contains an arbitrary number of key‐value pairs (where the key and value are separated by a = – eg 1=8 ). Here are some examples:

1=88|11=1438|15=KKK|45=00|45=00|21=66|86=a
4=13|11=1438|49=DDD|8=157.73|67=00|45=00|84=b|86=a
6=84|41=18|56=TTT|67=00|4=13|45=00|07=d

I need to create a DataFrame with a list of dictionaries, with each row as one dictionary in the list like so:

[{1:88,11:1438,15:kkk,45:7.7....},{4:13,11:1438....},{6:84,41:18,56:TTT...}]

df = pd.read_csv("input.txt",names=['text'],header=None)
data = df['text'].str.split("|")
names=[  y.split('=') for x in data for y in x]
ds=pd.DataFrame(names)
print ds

How can I create a dictionary for each line by splitting on the = symbol?

It should be one row and multiple columns. The DataFrame should have all keys as rows and values as columns.

Example:

1 11 15 45 21 86 4 49 8 67 84 6 41 56 45 07
88 1438 kkk 00 66 a
na 1438 na .....

Answer 1

I think performing a .pivot would work. Try this:

import pandas as pd

df = pd.read_csv("input.txt",names=['text'],header=None)
data = df['text'].str.split("|")
names=[  y.split('=') for x in data for y in x]
ds=pd.DataFrame(names)
ds = ds.pivot(columns=0).fillna('')

The .fillna('') removes the None values. If you'd like to replace with na you can use .fillna('na') .

Output:

ds.head()

   1
0 07   1    11   15 21 4 41  45 49 56 6 67 8 84 86
0     88
1         1438
2               KKK
3                            00
4                            00

For space I didn't print the entire dataframe, but it does column indexing based on the key and then values based on the values for each line (preserving the dict by line concept).