"R거지", R the hard way

README.md

R거지, R the hard way

why?

files:

• main.R

• _library.R

• _read_csv.R

• _assign.R
  • R의 라이브러리 함수들은 종종 data frame의 key를 그냥 object처럼 넘겨버릴 때가 있음.
  • R의 변수를 key/value로 관리하고 싶어서 만듬.

how?

• stopifnot()
• assign()/get()
• function(){}
• source()

_assign.R

ASSIGN = function (.KEY, .VALUE) {
  stopifnot(is.character(.KEY))
  print(.KEY)
  stopifnot(is.function(.VALUE))
  print(.VALUE)
  assign(.KEY, .VALUE(get(.KEY)), inherits=TRUE)
}

_library.R

# _library.R
IS_LIBRARY_EXISTS = function(.NAME) {
  length(find.package(.NAME, quiet = TRUE)) != 0
}

LIBRARY = function(.NAME, .TYPE=getOption("pkgType")) {
  stopifnot(is.character(.NAME))
  
  if (!IS_LIBRARY_EXISTS(.NAME)) {
    install.packages(.NAME, type=.TYPE)
  }
  if (IS_LIBRARY_EXISTS(.NAME)) {
    suppressPackageStartupMessages({
      library(.NAME, character.only=TRUE)
    })
    TRUE
  } else {
    FALSE
  }
}

JAVA_LIBRARY = function(.NAME, .TYPE=getOption("pkgType"), .LIBJVM_SO='/Library/Java/JavaVirtualMachines/jdk1.8.0_112.jdk/Contents/Home/jre/lib/server/libjvm.dylib') {
  if (Sys.getenv('JAVA_HOME') == '') {  # FIXME.
    dyn.load(.LIBJVM_SO)
  }
  LIBRARY(.NAME, .TYPE)
}

_read_csv.R

RENAME_VARIABLE = function(.FILENAME) {
  stopifnot(is.character(.FILENAME))

  . = gsub('/', '_', .FILENAME)
  . = gsub('-', '_', .)
  .
}

READ_CSV = function(.FILENAME) {
  source('_library.R')
  stopifnot(LIBRARY('readr'))
  
  stopifnot(is.character(.FILENAME))

  . = RENAME_VARIABLE(.FILENAME)
  assign(., read_csv(file=.FILENAME), inherits=TRUE)
  .
}

main.R

# 0. initialize
rm(list=ls())

setwd('~/projects/LOVER/')

source('_read_csv.R')
key1 = READ_CSV('watcha-doc2vec-regression/reviews1_11.csv')
key2 = READ_CSV('watcha-doc2vec-regression/reviews12_27.csv')

source('_library.R')
stopifnot(LIBRARY('dplyr'))
database = full_join(get(key1), get(key2))
rm(key1, key2, list=c(key1, key2))

source('_assign.R')
ASSIGN('database', function (data) {
  . = subset(data, select=c('movie_title', 'text'))
  na.omit(.)
})

stopifnot(LIBRARY('KoNLP'))
useSejongDic()