Python: Regex or Dictionary

问题

I have a DataFrame Column with one long string I would like to Parse. I am new to regex and have not worked with it yet. What I have below only returns the first name.. at best. I am wondering if parsing this string is easier for regex or creating a dictionary to iterate through. Here is what I have at the moment. The order is not always the same (C,W,D,G,UTIL) and I will be writing a for loop to iterate through multiple rows just like this one.

import pandas as pd
import numpy as np
import re

df = pd.DataFrame(data=np.array([['C Mark Scheifele C Pierre-Luc Dubois UTIL Zach Parise W Mats Zuccarello W Oliver Bjorkstrand W Nick Foligno D Ryan Suter D Seth Jones G Devan Dubnyk'],['UTIL Kyle Connor C Pierre-Luc Dubois C Boone Jenner W Mats Zuccarello W Oliver Bjorkstrand W Nick Foligno D Ryan Suter D Seth Jones G Devan Dubnyk']]), columns=['Lineup'])

df['C1'] = re.findall(r" C \w+",str(df['Lineup']))
df['C2'] = re.findall(r'C \w+',str(df['Lineup']))
df['W1'] = re.findall(r'W \w+',str(df['Lineup']))
df['W2'] = re.findall(r'W \w+',str(df['Lineup']))
df['W3'] = re.findall(r'W \w+',str(df['Lineup']))
df['D1'] = re.findall(r'D \w+',str(df['Lineup']))
df['D1'] = re.findall(r'D \w+',str(df['Lineup']))
df['G']= re.findall(r'G \w+',str(df['Lineup']))
df['UTIL'] = re.findall(r'UTIL \w+',str(df['Lineup']))

I am looking for storing these values into the DF.

df['C1'] = Mark Scheifele df['C2'] = Pierre-Luc Dubois df['W1'] = Mats Zuccarello df['W2'] = Oliver Bjorkstrand df['W3'] = Nick Foligno df['D1'] = Ryan Suter df['D2'] = Seth Jones df['G']= Devan Dubnyk df['UTIL'] = Zach Parise

RESULT DATAFRAME df_result = pd.DataFrame(data=np.array([['Mark Scheifele','Pierre-Luc Dubois','Mats Zuccarello','Oliver Bjorkstrand','Nick Foligno','Ryan Suter','Seth Jones','Devan Dubnyk','Zach Parise'],['Boone Jenner','Pierre-Luc Dubois','Mats Zuccarello','Oliver Bjorkstrand','Nick Foligno','Ryan Suter','Seth Jones','Devan Dubnyk','Kyle Connor']]), columns=['C1','C2','W1','W2','W3','D1','D2','G','UTIL'])

回答1:

import pandas as pd
import numpy as np
import re
def calc_col(col):
    '''This function takes a string,
    finds the upper case letters or words placed as delimeter,
    converts it to a list,
    adds a number to the list elements if recurring.
    Eg. input list :['W','W','W','D','D','G','C','C','UTIL']
    o/p list: ['W1','W2','W3','D1','D2','G','C1','C2','UTIL']
    '''
    col_list = re.findall(" ?([A-Z]+) ", col)
    col_list2 = []
    for i in col_list:
        cnt = col_list.count(i)
        if cnt == 1:
            col_list2.append(i)
        if cnt > 1:
            if i in " ".join(col_list2):
                continue;
            col_list2 += [i+str(k) for k in range(1,cnt+1)] 
    return col_list2

df = pd.DataFrame(data=np.array([['C Mark Scheifele C Pierre-Luc Dubois UTIL Zach Parise W Mats Zuccarello W Oliver Bjorkstrand W Nick Foligno D Ryan Suter D Seth Jones G Devan Dubnyk'],['UTIL Kyle Connor C Pierre-Luc Dubois C Boone Jenner W Mats Zuccarello W Oliver Bjorkstrand W Nick Foligno D Ryan Suter D Seth Jones G Devan Dubnyk']]), columns=['Lineup'])
extr_row = df['Lineup'].replace(to_replace =" ?[A-Z]+ ", value="\n", regex = True) #split the rows on 

df_final = pd.DataFrame(columns = sorted(calc_col(df['Lineup'].iloc[0]))) #Create an empty data frame df3 with sorted columns

for i in range(len(extr_row)): #traverse all the rows in the original dataframe and append the formatted rows to df3
    df_temp = pd.DataFrame((extr_row.values[i].split("\n")[1:])).T
    df_temp.columns = calc_col(df['Lineup'].iloc[i])
    df_temp= df_temp[sorted(df_temp)]
    df_final = df_final.append(df_temp)
df_final.reset_index(drop = True, inplace = True)
df_final

Please see the picture below for the final data frame. This should work for any number of rows:

回答2:

This version will give you the ability to have random orders, lengths (varying counts of ids and more. However, it relies on the indicator that a completely capitalised word is an id.

import pandas as pd

def get_df(string):

    result = [[key, f"{string[i + 1]} {string[i + 2]}"] for i, key in enumerate(string) if key.isupper()]

    occurs = {}

    for data in result:
        if data[0] not in occurs:
            occurs[data[0]] = 1
            data[0] = f"{data[0]}1"
        else:
            occurs[data[0]] += 1
            data[0] = f"{data[0]}{occurs[data[0]]}"

    return pd.DataFrame(data=[[i[1] for i in result]], columns=[i[0] for i in result])

data = ['C Mark Scheifele C Pierre-Luc Dubois UTIL Zach Parise W Mats Zuccarello W Oliver Bjorkstrand W Nick Foligno D Ryan Suter \
         D Seth Jones G Devan Dubnyk','UTIL Kyle Connor C Pierre-Luc Dubois C Boone Jenner W Mats Zuccarello W Oliver Bjorkstrand \
         W Nick Foligno D Ryan Suter D Seth Jones G Devan Dubnyk']


for i in data:
    print(get_df(i.split()))

Try this if you want to append the returned data frames together, hopefully returns the same data you're aiming for.

df = pd.DataFrame()

for i in data:
    df = df.append(get_df(i.split()))
    print(get_df(i.split()))


                  C1                 C2          D1          D2            G1        UTIL1               W1                  W2            W3
0     Mark Scheifele  Pierre-Luc Dubois  Ryan Suter  Seth Jones  Devan Dubnyk  Zach Parise  Mats Zuccarello  Oliver Bjorkstrand  Nick Foligno
0  Pierre-Luc Dubois       Boone Jenner  Ryan Suter  Seth Jones  Devan Dubnyk  Kyle Connor  Mats Zuccarello  Oliver Bjorkstrand  Nick Foligno

来源：https://stackoverflow.com/questions/59890489/python-regex-or-dictionary

标签

python

regex

dictionary

parsing