Extract a portion of a string in R - r

Question

I have a dataframe with ~200,000 rows and I need to extract a portion of a string from one of the columns. I have tried split and str_split and not getting at what I need.
I need to cut/split/explode on the / forward slash and only return the first field.
From the following sample I need only subdomain.domain.org
subdomain.domain.org/Site/SiteXYZ/Users/Usergroup
I have tried:
splitA <- str_split(my_data$OrganizationalUnit, pattern = "/", simplify = TRUE)
which returns a dataframe with multiple columns, one for each string between the delimiters
head(splitA)
[,1] [,2] [,3] [,4] [,5]
[1,] "subdomain.domain.org" "Site" "SiteXYZ" "Users" "Usergroup"
and
splitB <- str_split(my_data$OrganizationalUnit, pattern = "/", n = '2', simplify = TRUE)
which returns a dataframe with multiple columns, one for each string between the delimiters
head(splitB)
[,1] [,2]
[1,] "subdomain.domain.org" "Site/SiteXYZ/Users/Usergroup"
and
splitC <- **str_split**(my_data$OrganizationalUnit, pattern = "/", n = '1', simplify = TRUE)
which returns a dataframe with one column, that looks like the original
head(splitC)
[,1]
[1,] "subdomain.domain.org/Site/SiteXYZ/Users/Usergroup"
My end goal is to either:
Extract this string as part of larger queries
Add a column with this field
or add this as a column upon csv import
Any help will be much appreciated.
Thank you,
-Jacob

Answer 1

You were very close, if I understood right what you are after. As stated in the help file of ?str_split(), the simplify argument means:
"If FALSE, the default, returns a list of character vectors. If TRUE returns a character matrix."
If you are dealing with a character vector of domains, you can use simplify=TRUE and extract the first column of the matrix with (...)[, 1]. In any case, it is more efficient to use a combination of str_sub() and str_locate().
library(tidyverse)
x <- c("subdomain.domain.org/Site/SiteXYZ/Users/Usergroup",
"othersubdomain.domain.com/Site/SiteXYZ/Users/Usergroup",
"yetanothersubdomain.domain.com/Site/SiteXYZ/Users")
str_sub(x, start = 1, str_locate(string = x, "/")[, 1]-1)
# -1 otherwise / is kept in resulting string
str_split(x, pattern = "/", n = '2', simplify = TRUE)[, 1]
[1] "subdomain.domain.org" "othersubdomain.domain.com" "yetanothersubdomain.domain.com"
In a data.frame
If it is a column in a data.frame, you can use mutate() from {dplyr}:
library(tidyverse)
df <- tibble(domain = x,
y = rnorm(length(x))
)
df_extracted_domain <- df %>% mutate(
domain_suffix = str_sub(x, start = 1, str_locate(string = x, "/")[,1]-1)
)
in case you are not familiar with tidyverse and specially the pipe operator %>%, just read it as "and then". so the above line you read:
take df, "and then" mutate df (by adding a new variable, called domain_suffix). The mutated data.frame is assigned to a new object (or overwritten, if you call them the same)
benchmarking
y <- rep(x, 1e3)
bm <- rbenchmark::benchmark(
str_split = {
str_split(y, pattern = "/", n = '2', simplify = TRUE)[,1]
},
str_sub = {
str_sub(y, start = 1, str_locate(string = x, "/")-1)
},
replications = 1000
)
# test replications elapsed relative user.self sys.self user.child sys.child
# 1 str_split 1000 1.4 7.1 1.4 0.011 0 0
# 2 str_sub 1000 0.2 1.0 0.2 0.000 0 0