Python's equivalent to R's grepl and dplyr filter

Question

I want to find a way to recreate the same command code from R in Python using something similar as the dplyr package. In R I would do this:

      library(dplyr)
  df <- data.frame(Countries=c('Brazil','Venezuela','Brazil, Colombia, Paraguay','Argentina','Peru','Andorra,Argentina,Chile,Uruguay'),
               Code=c(1,2,3,4,5,6))

df  %>% filter(grepl('(Brazil|Argentina)',Countries))

Or even:

    a=strsplit(as.character(df$Countries),',')
    a=lapply(a,FUN=function(t) gsub(" ","",t))
    ele=unlist(lapply(a,FUN=function(t) any(t%in%c('Brazil','Argentina'))))
    (df[ele,])

The output that I want:

                   Countries Code
1                     Brazil    1
2 Brazil, Colombia, Paraguay    3
3                  Argentina    4
4    Argentina,Chile,Uruguay    6

In Python I've tried this:

import pandas as pd
df = pd.DataFrame(dict(Countries=['Brazil','Venezuela','Brazil, Colombia, Paraguay','Argentina','Peru','Andorra,Argentina,Chile,Uruguay'], Code=[1,2,3,4,5,6]))

list_=['Brazil','Argentina']
print(df.loc[df['Countries'].isin(list_)])

But the output looks like:

   Countries  Code
0     Brazil     1
3  Argentina     4

score 1 · Answer 1 · answered Nov 09 '20 at 14:09

Seems like you're looking for the .str extension of pd.Series with an object dtype (essentially you can call pd.Series.str.... for a subset of pandas functions specifically for dealing with regular expression and other string based operations- however this only works if the array is of dtype "object".

mask = df["Countries"].str.contains("Brazil|Argentina")
subset = df.loc[mask]

print(subset)
                         Countries  Code
0                           Brazil     1
2       Brazil, Colombia, Paraguay     3
3                        Argentina     4
5  Andorra,Argentina,Chile,Uruguay     6

A neat way you can use this is by using the .join function on your list_ variable to join it into a single string usable by regular expression matching patterns.

list_=['Brazil','Argentina']
pattern = "|".join(list_) # Now we have "Brazil|Argentina" as a string

mask = df["Countries"].str.contains(pattern)
subset = df.loc[mask] # Same subset as the previous example

See the docs and other methods besides .str.contains at the documentation https://pandas.pydata.org/pandas-docs/stable/reference/api/pandas.Series.str.html

Python's equivalent to R's grepl and dplyr filter

1 Answers1