R/clr_desc_corpus.R

In jaytimm/corpuslingr:

#' Provides basic frequency stats for an annotated corpus.
#'
#' Summarizes token frequency, type frequency, # of sentences by corpus/text/(optionally) genre.
#' @name clr_desc_corpus
#' @return A dataframe
#' @import data.table
#'
#'
#' @export
#' @rdname clr_desc_corpus
clr_desc_corpus <- function (corp,
                             doc ='id',
                             sent='sid',
                             tok='word',
                             upos='upos',
                             genre=NULL) {

  x <- corp
  if ("meta" %in% names(x)) x <- x$corpus
  if (!is.data.frame(x)) {x <- rbindlist(x)}
  x <- as.data.table(x)

  byText <- x[upos!="PUNCT", list(textLength=.N,textType=length(unique(get(tok))),textSent=length(unique(get(sent)))), by=doc]

  byText <- byText[order(as.numeric(gsub('[A-Za-z]+','',get(doc))))]

  corpus <- x[upos!="PUNCT", list(n_docs=length(unique(get(doc))),textLength=.N,textType=length(unique(get(tok))),textSent=length(unique(paste(get(doc),get(sent), sep=""))))]

  if (is.null(genre) == FALSE){
  setDT(corp$meta)
  x <- x[corp$meta, on=c("doc_id"), nomatch=0]

  byGenre <- x[upos!="PUNCT", list(n_docs=length(unique(get(doc))),textLength=.N,textType=length(unique(get(tok))),textSent=length(unique(paste(get(doc),get(sent), sep="")))), by=genre]

  out <- list("text" = byText, "corpus" = corpus, "genre" = byGenre)
  return(out)

    } else{
  out <- list("text" = byText, "corpus" = corpus)
  return(out)}
}