Splitting a string on the first space - regex

Question

I'd like to split a vector of character strings (people's names) into two columns (vectors). The problem is some people have a 'two word' last name. I'd like to split the first and last names into two columns. I can slit out and take the first names using the code below but the last name eludes me. (look at obs 29 in the sample set below to get an idea as the Ford has a "last name" of Pantera L that must be kept together)
What I have attempted to do so far;
x<-rownames(mtcars)
unlist(strsplit(x, " .*"))
What I'd like it to look like:
MANUF MAKE
27 Porsche 914-2
28 Lotus Europa
29 Ford Pantera L
30 Ferrari Dino
31 Maserati Bora
32 Volvo 142E

Answer 1

The regular expression rexp matches the word at the start of the string, an optional space, then the rest of the string. The parenthesis are subexpressions accessed as backreferences \\1 and \\2.
rexp <- "^(\\w+)\\s?(.*)$"
y <- data.frame(MANUF=sub(rexp,"\\1",x), MAKE=sub(rexp,"\\2",x))
tail(y)
# MANUF MAKE
# 27 Porsche 914-2
# 28 Lotus Europa
# 29 Ford Pantera L
# 30 Ferrari Dino
# 31 Maserati Bora
# 32 Volvo 142E

Answer 2

For me, Hadley's colsplit function in the reshape2 package is the most intuitive for this purpose. Joshua's way is more general (ie can be used wherever a regex could be used) and flexible (if you want to change the specification); but the colsplit function is perfectly suited to this specific setting:
library(reshape2)
y <- colsplit(x," ",c("MANUF","MAKE"))
tail(y)
# MANUF MAKE
#27 Porsche 914-2
#28 Lotus Europa
#29 Ford Pantera L
#30 Ferrari Dino
#31 Maserati Bora
#32 Volvo 142E

Answer 3

Here are two approaches:
1) strsplit. This approach uses only functions in the core of R and no complex regular expressions. Replace the first space with a semicolon (using sub and not gsub), strsplit on the semicolon and then rbind it into a 2 column matrix:
mat <- do.call("rbind", strsplit(sub(" ", ";", x), ";"))
colnames(mat) <- c("MANUF", "MAKE")
2) strapply in gsubfn package Here is a one-liner using strapply in the gsubfn package. The two parenthesized portions of the regular expression capture the desired first and second columns respectively and the function (which is specified in formula notation -- its the same as specifying function(x, y) c(MANUF = x, MAKE = y)) grabs them and adds names. The simplify=rbind argument is to used to turn it into a matrix as in the prior solution.
library(gsubfn)
mat <- strapply(x, "(\\S+)\\s+(.*)", ~ c(MANUF = x, MAKE = y), simplify = rbind)
Note: In either case a "character" matrix, mat, is returned. If a data frame of "character" columns is desired then add this:
DF <- as.data.frame(mat, stringsAsFactors = FALSE)
Omit the stringsAsFactors argument if "factor" columns are wanted.

Answer 4

Yet another way of doing it:
str_split from stringr will handle the split, but returns it in a different form (a list, like strsplit does). Manipulating into the correct form is straightforward though.
library(stringr)
split_x <- str_split(x, " ", 2)
(y <- data.frame(
MANUF = sapply(split_x, head, n = 1),
MAKE = sapply(split_x, tail, n = 1)
))
Or, as Hadley mentioned in the comments, with str_split_fixed.
y <- as.data.frame(str_split_fixed(x, " ", 2))
colnames(y) <- c("MANUF", "MAKE")
y

Answer 5

If you can do pattern and group matching, I'd try something like this (untested):
\s+(.*)\s+(.*)

Answer 6

You can also use tidyr::extract after converting your vector into a data frame first - I think this would also be the more modern version of older solutions with reshape2
library(tidyr)
## first convert into a data frame
x <- data.frame(x = rownames(mtcars))
## use extract, and for example Joshua's regex
res <- extract(x, col = x, into = c("MANUF", "MAKE"), regex = "^(\\w+)\\s?(.*)$")
head(res)
#> MANUF MAKE
#> 1 Mazda RX4
#> 2 Mazda RX4 Wag
#> 3 Datsun 710
#> 4 Hornet 4 Drive
#> 5 Hornet Sportabout
#> 6 Valiant

Answer 7

I think searching for [^\s]+ would work. Untested.