course code sorting

course_code_sort.R

library(dplyr)
library(tidyr)
library(stringr)

sort_course_codes <- function(course_codes) {
  course_with_sorting <- data_frame(course_code = course_codes) %>%
    mutate(course_code_cleaned = str_trim(tolower(str_replace_all(
      course_code, "[(\\[\\{][^(\\[\\{]+[)\\]\\}]", ""
    )))) %>%
    separate(
      course_code_cleaned,
      into = c("alphnum_l", "alphnum_r"),
      sep = "[-,[:space:]]",
      extra = "merge",
      fill = "right",
      remove = F
    ) %>%
    mutate(
      num_l = as.numeric(str_extract(alphnum_l, "^[[:digit:]\\.]+")),
      alph_l = str_replace(alphnum_l, "^[[:digit:]\\.]*", ""),
      num_r = as.numeric(str_extract(alphnum_r, "^[[:digit:]\\.]+")),
      alph_r = str_replace(alphnum_r, "^[[:digit:]\\.]*", "")
    ) %>%
    replace_na(list(num_l = Inf, alph_l = "", num_r = Inf, alph_r = "")) %>%
    arrange(num_l, alph_l, num_r, alph_r, course_code_cleaned, course_code)

  course_with_sorting$course_code
}

course_codes = c(
  "195", "106x", "241", "113-C", "123", "219", "104z", "143",
  "104", "B", "159", "163", "163.2", "130x", "180", "20 [x-reg]",
  "185", "150", "158", "212", "121a", "A-26", "A-30", "200.01"
)

sorted_course_codes <- sort_course_codes(course_codes)

invisible(sapply(sorted_course_codes, message))

output

20 [x-reg]
104
104z
106x
113-C
121a
123
130x
143
150
158
159
163
163.2
180
185
195
200.01
212
219
241
A-26
A-30
B