Function composition in R - text mining application

odyssey.R

# Using 'Book I', the Odyssey from The Internet Classics Archive
# http://classics.mit.edu/Homer/odyssey.html

needed <- c("tm", "functional")
# install.packages(needed)
sapply(needed, require, character.only = TRUE)

text <- readLines("http://classics.mit.edu/Homer/odyssey.mb.txt")
str(text)
text[198]



# f . g x = f(g(x))

# Instead of consequently applying all cleaning functions: tolower(stripWhitespace(...)),
# function composition allows through 'Compose' function allows to simplify the process:
# (tolower . stripWhitespace . ...) ('input')
# 
# As of June 2017, only one-argument functions are allowed for composition

clean <- function(docs, pattern) {
  lapply(docs, function(txt) {
      repl <- function(t) gsub(pattern, " ", t)
      Compose(tolower, stripWhitespace, removePunctuation, repl)(txt)
  })
}

patterns <- "\n d5> <d1> \f \ ?"
text <- clean(text, patterns)

text[198]