clean clear-text artist names from collaborations and secondary artists

clean_artistnames.R

require(stringi)

spelling_variants <- function(x, remove_collabs=F, remove_parentheses=T) {
  qualifiers = c(" feat .*", " feat[.].*", " ft.*", " ft[.].*"," featuring.*"," vs[.].*"," vs.*"," versus.*"," with.*","[-].*"," / .*",
                 "/.*","[|].*", "[[].*[]]", "[)].*", ";.*","[+].*","[&] .*","[&].*",",.*"," and .*", " con .*", " e .*", " et .*",
                 " x .*")
  
  # remove articles (a, the)
  ret = gsub(" a ", "", tolower(str_trim(x)))
  
  removes = c("the ", # article
              "^[(][[:digit:]]+[)]", # digits wrapped in parentheses at beginning of string
              "^[#][[:digit:]]{1,5}[[:punct:]|[:space:]]", # digits lead by # at beginning of string
              "^[0][[:digit:]]{1,2}[[:punct:]|[:space:]]", # digits lead by 0 at beginning of string
              "^[[:digit:]]{1,3}[[:punct:]|[:space:]]", # digits at the beginning of a string if they are followed by a punctuation mark (e.g., 12 - Hello keeps Hello)
              "[[:digit:]]{2}[[:punct:]][[:digit:]]{2}[[:punct:]][[:digit:]]{2,4}") # date-type objects
              
  for (rem in removes) ret=str_trim(gsub(rem,'',ret))
  
  # Remove dash if occuring within first three characters (e.g., "JAY-Z")
  ret = gsub('(^[a-z]{0,3})([-])', '\\1', ret)
  
  # Remove special characters (e.g., - Jodeli -) at the beginning and end of a string
  for (i in 1:5) {
    first_char = substr(ret,1,1)
    last_char = substr(stri_reverse(ret),1,1)
    eval_condition = grepl('[[:punct:]]{2}', paste0(first_char,last_char))
    start_char = ifelse(eval_condition==T, 2,1)
    end_char = ifelse(eval_condition==T, nchar(ret)-1,nchar(ret))
    ret=str_trim(substr(ret,start_char,end_char))
  }
  
  if(remove_parentheses==T) {
    # Remove content from parentheses, unless it's at the beginning of a string
    ret = gsub('(?<!^)[(].*','', ret,perl=T)
  }
  
  # Remove dash/+/[ at beginning of string
  ret = gsub('^[-]|^[+]|^[[]','', ret)
  
  
  if(remove_collabs==T) {
    # remove collaborations
    for (qual in qualifiers) ret=sub(qual, "", str_trim(ret))
  }
  
  ret = str_trim(sub("30 ", "thirty ", ret))
  
  ret = str_trim(gsub("[^0-9A-Za-z ]", "", ret))
  for (i in 1:5) ret = str_trim(gsub("  ", " ", ret)) # remove white space inbetween characters
  return(ret)
}