Create multiple-columns pandas dataframe from list-CodePudding

I can't figure out how to create pandas dataframe (multiple-columns) from list. Some lines contains character ">" at the beggining. I want them to be column headers. Number of lines after each header is not the same.

My list:

>header
a
b
>header2
c
d
e
f
>header3
g
h
i

Dataframe I want to create:

>header1   >header2   >header3
a           c          g
b           d          h
            e          i
            f

CodePudding user response：

Simply iterate through lines and match the headers with '>'. The challenge though is to create a df from a dictionary of lists with unequal size.

# The given list
lines = [">header", "a", "b", ">header2", "c", "d", "e", "f", ">header3", "g", "h", "i"]

# Iterate through the lines and create a sublist for each header
data = {}
column = ''
for line in lines:
    if line.startswith('>'):
        column = line
        data[column] = []
        continue
    data[column].append(line)

# Create the DataFrame
df = pd.DataFrame.from_dict(data,orient='index').T

output:

  >header >header2 >header3
0       a        c        g
1       b        d        h
2    None        e        i
3    None        f     None

CodePudding user response：

I'm assuming you have a text with this list. You can use str.splitlines() to split it and then construct the dataframe with help of itertools.zip_longest:

from itertools import zip_longest

text = '''\
>header
a
b
>header2
c
d
e
f
>header3
g
h
i'''

current, data = None, {}
for line in text.splitlines():
    line = line.strip()
    if line.startswith('>'):
        current = line
    else:
        data.setdefault(current, []).append(line)

df = pd.DataFrame(zip_longest(*data.values(), fillvalue=''), columns=list(data))
print(df)

Prints:

  >header >header2 >header3
0       a        c        g
1       b        d        h
2                e        i
3                f