R/bx_extract.R

In biorxivr: Search and Download Papers from the bioRxiv Preprint Server

#' Extract data from search results
#' @description Extract information about each paper in search results, returning author info, abstract and download metrics
#' @param bxso search results from bx_search()
#' @examples \dontrun{
#' bxEco <- bx_search("ecology",limit=5) 
#' bxEcoData <- bx_extract(bxEco)
#' ## See a plot of abstract views
#' plot(bxEcoData[[1]],type="abs")
#' 
#' ## See a plot of downloads
#' plot(bxEcoData[[1]],type="dl)
#' 
#' }
#' @details The metrics field will return a month by month data frame with abstract views and pdf downloads
#' @return a list holding the details about all the papers from search results. Each element in the list is an S3 object that holds all the salient details about the paper
#' @export
#' @import XML
bx_extract <- function(bxso){
  
  p <- sapply(bxso$URL,bx_extract_single,simplify=F)
  return(p)
}


#'Extract data from a single record
#'@description Generate an S3 object that represents a single paper
#'@param bxso_url the URL of a single biorxiv paper
#'@return a single S3 object with all the salient details about a paper
#'@export

bx_extract_single <- function(bxso_url){
  
  xData <- getURL(bxso_url)
  pgRes <- htmlParse(xData)
  meta.names <- xpathApply(pgRes, "//meta[@name]",xmlGetAttr,"name")
  meta.data <- xpathApply(pgRes, "//meta[@name]",xmlGetAttr,"content")# DC.Contributor"
  names(meta.data) <- unlist(meta.names)
  nlist <- unlist(meta.names)
  paper <- list()
  authors <- list()
authors$names <- unname(unlist(meta.data[which(nlist == "DC.Contributor")]))
authors$email <- unname(unlist(meta.data[which(nlist == "citation_author_email")]))
  paper$title <- unname(unlist(meta.data[which(nlist == "DC.Title")]))
  paper$abstract <- unname(unlist(meta.data[which(nlist == "DC.Description")]))
  paper$date <- unname(unlist(meta.data[which(nlist == "DC.Date")]))
  paper$DOI <- unname(unlist(meta.data[which(nlist == "DC.Identifier")]))
  paper$fulltext_url <- unname(unlist(meta.data[which(nlist == "citation_pdf_url")]))

  xData2 <- getURL(paste(bxso_url,".article-metrics"))
  metricRes <- htmlParse(xData2)
  ## This returns a value for each element and we'll convert it to a matrix
  #metrics <- unlist(xpathApply(metricRes, "//tbody/tr/td",xmlValue))
  ## If there are no metrics, at this point  the metrics dataframe will be null, hence this if statement
#if(!is.null(metrics)){
  #metrics <- data.frame(matrix(metrics,ncol=3,nrow=length(metrics)/3, byrow=T))
  #metrics[,3] <- as.numeric(as.character(metrics[,3]))
  #metrics[,2] <- as.numeric(as.character(metrics[,2]))
#metrics[,1] <- do.call(c,lapply(strsplit(as.character(metrics[,1])," "),function(x) {as.Date(strptime(paste(paste(x,collapse="-"),"01",sep="-"),"%b-%Y-%d"))}))
  #colnames(metrics) <- c("date","Abstract","PDF")
#}
  
    
return(structure(list(authors = authors, paper = paper),#, 
      #  metrics = metrics)
       class = "biorxiv_paper"))
}