Clean html tags | R | https://intip.in/FnaD

cleanScript <- function(htmlString) {
  return( gsub( "#<script(.*?)>(.*?)</script>#is", "", htmlString ) )
}

cleanStyle <- function(htmlString) {
  return( gsub( "#<style(.*?)>(.*?)</style>#is", "", htmlString ) )
}

cleanTags <- function(htmlString) {
  return(gsub("<.*?>", "", htmlString))
}


cleanSekrip <- function(htmlString) {
  return( gsub( '>.*$', '></div>', htmlString ) )
}


cleanKabeh <- function(htmlString) {
  htmlString <- gsub( "#<script(.*?)>(.*?)</script>#is", "", htmlString )
  htmlString <- gsub( "#<style(.*?)>(.*?)</style>#is", "", htmlString )
  return( gsub("<.*?>", "", htmlString) )
}

cleanKabeh <- function(htmlString) {
  return( gsub("<.*?>", "", gsub( "#<style(.*?)>(.*?)</style>#is", "", gsub( \<(?:[^:]+:)?script\>.*?\<\/(?:[^:]+:)?script\>, "", htmlString ) ) ) )
}





library(XML)
doc <- htmlParse(article,asText=TRUE)
styleNodes <- getNodeSet(doc, "//style")
styleNodes <- getNodeSet(doc, "//script")
removeNodes(styleNodes)
doc