code_01.R

x = sample(LETTERS, 10000, replace = TRUE)

# Method 1
start = Sys.time()
result1 = rep(NA, length(x))
for(i in 1:length(x)) {
  tmp = x[-c(1:i)]
  result1[i] = tmp[tmp %in% c("A", "B", "C")][1]
}
end = Sys.time()
time1 = end - start

# Method 2
library(Rcpp)

cppFunction('CharacterVector nextABC(CharacterVector x) {
  int n = x.size();
  CharacterVector result(n);
  for(int i = 0; i < n; i++) {
    for(int j = i+1; j < n; j++) {
      if(x[j] == "A" || x[j] == "B" || x[j] == "C") {
        result[i] = x[j];
        break;
      }
    }
  }
  return result;
}')

start = Sys.time()

result2 = nextABC(x)
result2[result2 == ""] = NA

end = Sys.time()
time2 = end - start

# Method 3
start = Sys.time()
goods = which(x %in% c("A","B","C"))
nexti = sapply(1:length(x), function(i) min(goods[goods>i],length(x)+1))
result3 = x[nexti]
end = Sys.time()
time3 = end - start

# Method 4
library(dplyr)

## 
## Attaching package: 'dplyr'

## The following objects are masked from 'package:stats':
## 
##     filter, lag

## The following objects are masked from 'package:base':
## 
##     intersect, setdiff, setequal, union

library(tidyr)
start = Sys.time()
df = as.data.frame(x, stringsAsFactors = FALSE)
result4 = 
  df %>% 
  mutate(new = rev(ifelse(x %in% c("A","B","C"), x, NA))) %>% 
  fill(new) %>% 
  mutate(new = rev(new)) %>% 
  pull(new)
end = Sys.time()
time4 = end - start

# Method 5
start = Sys.time()
abc = x%in%c("A","B","C")
qq = cumsum(c(T,abc))[1:length(abc)]
result5 = x[which(abc)[qq]]
end = Sys.time()
time5 = end - start

time1 # 'for' loop

## Time difference of 2.819571 secs

time2 # Rcpp

## Time difference of 0.008967876 secs

time3 # sapply

## Time difference of 0.1276851 secs

time4 # dplyr

## Time difference of 0.08501935 secs

time5 # cumsum

## Time difference of 0.003102779 secs

all.equal(result1, result2, result3, result4, result5)

## [1] TRUE