String splitting a dataframe with a vector as the pattern in R

Question

I have a dataframe that consists of multiple rows, and I would like to split every row into two components based off of elements of a vector (essentially run strsplit with a vector as the 'pattern') in R.

The dataframe (only one column) looks something like this:

     [,1]                
[1,] "apple please fuji" 
[2,] "pear help name"    
[3,] "banana me mango"   

Whereas my pattern vector could look like this: v <- c("please", "help", "me").

If possible, I would like my end output to be:

  df$name             df$part1  df$split  df$part2   
 "apple please fuji" "apple"    "please"  "fuji" 
 "pear help name"    "pear"     "help"    "name" 
 "banana me mango"   "banana"   "me"      "mango"

I would appreciate any help with the in-between step of being able to isolate components based on a vector, but if there is an even easier way to put it into a dataframe, that would be great!. Thank you so much!

Answer 1

Here are two methods in base R.

Start with a character vector:

text <- c("apple please fuji", "pear help name", "banana me mango")

Also, the desired variable names (for convenience)

varNames <- c("name", "part1", "split", "part2")

---

using regexec and regmatches

As an alternative, you can also use regular expressions with the regmatches / regexec combination to construct this dataset.

First, build a regular expression from v with paste.

myRegex <- paste0("^(.*) +(", paste(v, collapse="|"), ") +(.*)$")
myRegex
[1] "^(.*)(please|help|me)(.*)$"

setNames(do.call(rbind.data.frame, regmatches(text, regexec(myRegex, text))), varNames)

this returns the same as above

               name  part1  split part2
1 apple please fuji  apple please  fuji
2    pear help name   pear   help  name
3   banana me mango banana     me mango

---

using strsplit and do.call

First, split each element by v

tmp <- do.call(strsplit, list(text, split=v))
tmp
[[1]]
[1] "apple " " fuji" 

[[2]]
[1] "pear " " name"

[[3]]
[1] "banana " " mango" 

Now, rbind.data.frame these, which drops the second column, and returns a data.frame cbind the split and name variables, and then add names with setNames.

setNames(cbind(text, do.call(rbind.data.frame, tmp), v)[c(1, 2, 4, 2)], varNames)

this returns

               name   part1  split   part2
1 apple please fuji  apple  please  apple 
2    pear help name   pear    help   pear 
3   banana me mango banana      me banana 

---

Answer 2

This solution assumes the number of elements in v is equal to the number of rows in the dataframe. You can use separate from the tidyr package to create part1 and part2.

library(tidyverse)
df <- tibble(name = c("apple please fuji", "pear help name", "banana me mango"))
v <- c("please", "help", "me")

df %>% 
  separate(name, c("part1", "part2"), v, remove = FALSE) %>%
  add_column(split = v, .before = "part2")
#> # A tibble: 3 x 4
#>                name   part1  split  part2
#>               <chr>   <chr>  <chr>  <chr>
#> 1 apple please fuji  apple  please   fuji
#> 2    pear help name   pear    help   name
#> 3   banana me mango banana      me  mango

If you want to try and split each row using any element in v then you could try pasting v into a single pattern first before separating. I think something like this should work.

library(tidyverse)
library(stringr)
p <- paste0("\\b(?:", paste(v, collapse = "|"), ")\\b")
df %>% 
  separate(name, c("part1", "part2"), p, remove = FALSE) %>%
  mutate(split = str_extract(name, p)) %>%
  select(name, part1, split, part2)
#> # A tibble: 3 x 4
#>                name   part1  split  part2
#>               <chr>   <chr>  <chr>  <chr>
#> 1 apple please fuji  apple  please   fuji
#> 2    pear help name   pear    help   name
#> 3   banana me mango banana      me  mango

Answer 3

# Creating creating the df
name <- c("apple please fuji","pear help name","banana me mango")

# as.data.frame
df <- as.data.frame(name, stringsAsFactors = F)
# Initialize empty data frame. 
df_n <- data.frame()
# Loop through the original rows of the df. 
for(i in 1:nrow(df)){
  for(j in 1:nrow(df)){
    o <- strsplit(df$name, " ")[[i]][j]
  }
}
# rename and assign new df (df_n) changes to original df. 
df$part1 <- df_n$V1
df$part2 <- df_n$V2
df$part3 <- df_n$V3

print(df)