Multiple delimiters in column headers also separates the row values

问题

I had a some issue about defining multiple seperator when reading a file. It is originally solved in my previous post reading-files-with-multiple-delimiter-in-column-headers-and-skipping-some-rows thanks to @piRsquared

When I looked in detail to my real data I realized that some columns have .cd or .dvd extensions and when I applied the solution above they are also separated as a new column and the solution above started not to work!

b.txt

skip1
 A1| A2 |A3 |A4# A5# A6 A7| A8 , A9
1,2,3,4,5.cd,6,7,8.dvd,9
1,2,3,4,5.cd,6,7,8.dvd,9
1,2,3,4,5.cd,6,7,8.dvd,9

END123
Some other data starts from here

and read this b.txt file using the solution above

txt = open('b.txt').read().split('\nEND')[0]
pd.read_csv(
    pd.io.common.StringIO(txt),
    sep=r'\W+',
    skiprows=1,index_col=False, engine='python')

   A1  A2  A3  A4  A5  A6  A7  A8  A9
0   1   2   3   4   5  cd   6   7   8
1   1   2   3   4   5  cd   6   7   8
2   1   2   3   4   5  cd   6   7   8

A5 column should have the rows

5.cd
5.cd
5.cd

and same for the A9 column

8.dvd
8.dvd
8.dvd

and we should have A9 column but seems that it disappears due to this conflict.

EDIT:

I put almost similar identity to my real data

 skip rows
 A1| A2| A3|A4# A5#  |  A6 | A7  , A8,  A9  | A10 |
 1 | 2 | 3 |4 # 5 #  | 6.cd|7.dvd,   ,      | 10  | 
 1 | 2 | 3 |4 # 5 #  | 6.cd|     ,   ,   9  | 10  |
 1 | 2 | 3 |4 # 5 #  |     |7.dvd,   ,      | 10  |

END123
Some other data starts from here

and tried

txt = open('real_dat.txt').read().split('\nEND')[0]
_, h, txt = txt.split('\n', 2)
pat = r'[\|, ,#,\,]+'
names = re.split(pat, h.strip())

df=pd.read_csv(
    pd.io.common.StringIO(txt),
    names=names,skiprows=1,index_col=False,
    engine='python')

and got this output!

回答1:

Updated Answer
It was just easier to get rid of spaces... Let me know if this works

txt = open('b.txt').read().split('\nEND')[0] \
    .replace(' ', '').replace('|\n', '\n').split('\n', 1)[1]

pd.read_csv(
    pd.io.common.StringIO(txt),
    sep=r'#\||\||#|,',
    engine='python')

   A1  A2  A3  A4  A5    A6     A7  A8   A9  A10
0   1   2   3   4   5  6.cd  7.dvd NaN  NaN   10
1   1   2   3   4   5  6.cd    NaN NaN  9.0   10
2   1   2   3   4   5   NaN  7.dvd NaN  NaN   10

Old Answer

I used \W+ as a fast and easy way to parse what you showed. Below I used something more specific to the actual delimiters you need.

txt = open('b.txt').read().split('\nEND')[0]
pd.read_csv(
    pd.io.common.StringIO(txt),
    sep=r'[\|, ,#,\,]+',
    skiprows=1,index_col=False, engine='python')

   A1  A2  A3  A4    A5  A6  A7     A8  A9
0   1   2   3   4  5.cd   6   7  8.dvd   9
1   1   2   3   4  5.cd   6   7  8.dvd   9
2   1   2   3   4  5.cd   6   7  8.dvd   9

However, I still think this is a cleaner way to do it. Here, I separate the parsing of the header from the parsing of the rest of the data. That way, I assume the data should only be using , as a separator.

txt = open('b.txt').read().split('END')[0]
_, h, txt = txt.split('\n', 2)
pat = r'[\|, ,#,\,]+'
names = re.split(pat, h.strip())

pd.read_csv(
    pd.io.common.StringIO(txt),
    names=names, header=None,
    engine='python')

   A1  A2  A3  A4    A5  A6  A7     A8  A9
0   1   2   3   4  5.cd   6   7  8.dvd   9
1   1   2   3   4  5.cd   6   7  8.dvd   9
2   1   2   3   4  5.cd   6   7  8.dvd   9

来源：https://stackoverflow.com/questions/45699077/multiple-delimiters-in-column-headers-also-separates-the-row-values

标签

python

pandas

delimiter

delimited-text