How to load a weird sas file that was found in the wild, but refused to be read be haven::read_sas

load_weird_sas.R

library(tidyverse)
sas <- tempfile()
download.file("http://psych.colorado.edu/~carey/Courses/PSYC7291/DataSets/Documentation/InterestDataDoc.txt",
              sas)
sas <- read_lines(sas)
starts_with_spaces <- function(string){
  out <- seq_along(string)
  for (i in out) {
    if(grepl("^    ", string[i]))out[i] <- out[i-1]
  }
  return(as.factor(out))
}

get_datalines <- function(string){
  string <- string[(grep("DATALINES;", string)+1):length(string)]
  string <- string[-(which(map_lgl(string, identical, "")):length(string))]
  return(string)
}

get_input <- function(string) {
  found <- FALSE
  string <- str_squish(string)
  for (i in seq_along(string)) {
    if(grepl("^input ", string[i])){
      found <- TRUE
      start <- i
      end <- start
    } else if(found){
      if(!grepl(";", string[i])){
        end <- end + 1
      } else {
        end <- end + 1
        break
      }
    }
  }
  input <- string[start:end]
  input <- str_remove(input, "input ")
  input <- str_remove(input, ";")
  input <- str_split(input, " ")
  input <- flatten_chr(input)
  return(input)
}

sas_data <- sas %>% get_datalines() %>% split(., starts_with_spaces(.)) %>% 
  map(paste, collapse = " ") %>% 
  str_squish() %>% 
  str_split(" ") %>% 
  do.call(rbind, .)

colnames(sas_data) <- get_input(sas)
sas_data <- sas_data %>% as_tibble() %>% map_dfc(parse_number)