提取data.table中的字符串

Question

I have the following data.table called D .

                  ngram        
1          in_the_years          
2  the_years_thereafter        
3 years_thereafter_most 
4        he_wasn't_home        
5           how_are_you          
6    thereafter_most_of  

I need to add a few variables.

1. queryWord (the requirement is to extract the first 2 words) the following is my code

D[,queryWord:=strsplit(ngram,"_[^_]+$")[[1]],by=ngram]
                   ngram        queryWord
1          in_the_years           in_the
2  the_years_thereafter        the_years
3 years_thereafter_most years_thereafter
4        he_wasn't_home        he_wasn't
5           how_are_you          how_are
6    thereafter_most_of  thereafter_most

2. predict . The requirement is to extract the last word. The following is desired output

                   ngram        queryWord            predict
1          in_the_years           in_the             years
2  the_years_thereafter        the_years             thereafter
3 years_thereafter_most        years_thereafter      most
4        he_wasn't_home        he_wasn't             home 
5           how_are_you          how_are             you
6    thereafter_most_of  thereafter_most             of

For this purpose I wrote the following function

getLastTerm<-function(x){
              y<-strsplit(x,"_")
              y[[1]][length(y[[1]])]
}

getLasTerm("in_the_years","_") return "years" however is not working inside the data.table object D .

D[,predict:=getLastTerm(ngram)[[1]],by=ngram] 

Please I need help

Answer 1

Your get last term function only selects the first list. Try below.

getLastTerm <- function(x){
  y <- strsplit(x,"_")

  for (i in (1:6)) { 
    x[i] <- y[[i]][length(y[[i]])]
  }
  x
}


D$new <- getLastTerm(D$ngram)

Answer 2

Before adressing your actual question, you can simplify your first step to:

# option 1
D[, queryWord := strsplit(ngram,"_[^_]+$")][]
# option 2
D[, queryWord := sub('(.*)_.*$','\\1',ngram)][]

To get the predict -column, you don't need to write a special function. Using a combination of strsplit , lapply and last :

D[, predict := lapply(strsplit(D$ngram,"_"), last)][]

Or an even easier solution is using only sub :

D[, predict := sub('.*_(.*)$','\\1',ngram)][]

Both approaches give the following final result:

 > D ngram queryWord predict 1: in_the_years in_the years 2: the_years_thereafter the_years thereafter 3: years_thereafter_most years_thereafter most 4: he_wasn't_home he_wasn't home 5: how_are_you how_are you 6: thereafter_most_of thereafter_most of 

---

Used data:

D <- fread("ngram
in_the_years
the_years_thereafter
years_thereafter_most
he_wasn't_home
how_are_you
thereafter_most_of", header = TRUE)