medrxivr_extensions.R

all_caps <- function(x) {
  
  x_v <- stringr::str_to_lower(x) %>%
  stringr::str_split(stringr::boundary()) %>%
  unlist()

for (position in 1:nchar(x)) {
  x_v[position] <- paste0("[",stringr::str_to_upper(x_v[position]),x_v[position],"]") 
}
  
  x_v <- paste0(x_v, collapse = "")
  
  return(x_v)

}

fix_caps <- function(x){
  
  unlist(lapply(x, function(x){
  
  if (grepl("\\[",x)==TRUE) {
    return(x)
  }
  
  x_v <- stringr::str_to_lower(x) %>%
    stringr::str_split(stringr::boundary()) %>%
    unlist()
  
  x_v[1] <- paste0("[",stringr::str_to_upper(x_v[1]),x_v[1],"]")

  x_v <- paste0(x_v, collapse = "")
  
  return(x_v)}
  
   ))
  
}


fix_wildcard <- function(x) {
  
  x <- stringr::str_replace_all(x,"\\*","([[:alpha:]])")
  
  return(x)
  
}


fix_near <- function(x) {
  
  x <- stringr::str_replace_all(x,"\\s?[Nn][Ee][Aa][Rr](\\d)\\s?","NEAR\\1") %>%
    stringr::str_replace_all("NEAR(\\d)","(\\\\s+)([[:graph:]]+\\\\s+){0,\\1}")
  
  return(x)
  
}

library(testthat)

expect_true(fix_near("sysNEAR2rev") == "sys(\\s+)([[:graph:]]+\\s+){0,2}rev")
expect_true(fix_near("sys Near2 rev") == "sys(\\s+)([[:graph:]]+\\s+){0,2}rev")
expect_false(fix_near("sys  Near2 rev") == "sys(\\s+)([[:graph:]]+\\s+){0,2}rev")
expect_true(fix_near("sys near2 rev") == "sys(\\s+)([[:graph:]]+\\s+){0,2}rev")
expect_false(fix_near("sys near rev") == "sys(\\s+)([[:graph:]]+\\s+){0,2}rev")