R/extract_records.R

In robboyd/soaR:

#' extract records from GBIF and format them for use in soaR
#'
#' This function is used to scrape GBIF for occurrence records for a chosen taxonomic group, on a chosen continent.
#' @param continent String. E.g. "south_america", "europe". Do not specify unless country is NULL.
#' @param country string. E.g. "BR" for Brasil, "AR" for Argentina, and "CL" for Chile
#' @param year string. year(s) for which to extract data. For a range, format as e.g. "1995,2010". For a single year, as e.g. "2005".
#' @param taxon numeric. GBIF taxonomic key.
#' @param write logical
#' @param outPath string. Directory in which to save outputs.
#' @param outName string. Name of file to be saved if write_output = TRUE.
#' @param degrade logical. Whether or not to remove duplicates from the data (which may be repeat visits).
#' @export
#' @examples

extract_records <- function(roster) {

  if (!is.na(roster$continent) & !is.na(roster$country)) {

    stop("Can't specify both country and continent")

  }

  nrecs <- rgbif::occ_count(taxonKey = roster$taxa,
                            georeferenced = TRUE,
                            country = roster$country,
                            year = roster$year,
                            type = "count",
                            publishingCountry = NULL)

  if (nrecs <= 100000) {

    if (!is.na(roster$continent)) {

      x <- rgbif::occ_data(hasCoordinate = T,
                           hasGeospatialIssue = F,
                           continent = as.character(roster$continent),
                           taxonKey = roster$taxa,
                           year = roster$year,
                           limit = 99999)

    } else if (!is.na(roster$country)) {

      x <- rgbif::occ_data(hasCoordinate = T,
                           hasGeospatialIssue = F,
                           country = as.character(roster$country),
                           taxonKey = roster$taxa,
                           year = roster$year,
                           limit = 99999)
    } else {

      stop("Must specify one of country or continent")

    }

    ## take the second element which is the "data" as opposed to the metadata

    dat <- data.frame(x[2])

    ## check whether there are any records

    if ("data.species" %in% names(dat)) {

      ## if there are create a data frame with variables of interest

      fields <- c("data.species", "data.scientificName", "data.decimalLongitude",
                        "data.decimalLatitude", "data.year", "data.eventDate", "data.verbatimEventDate",
                        "data.country", "data.continent",
                        "data.basisOfRecord", "data.datasetKey",
                        "data.coordinateUncertaintyInMeters",
                        "data.country", "data.bibliographicCitation")

      names <- c("species","group","lon","lat","year", "Date", "originalDate", "country","continent","basisOfRecord",
                 "UUID", "spatialUncertainty", "country", "ref")

      keep <- which(fields %in% colnames(dat))

      out <- dat[, fields[keep]]

      colnames(out) <- names[keep]

      if (length(dat[,1]) == 199000) {
        warning("Reached max number of outputs, but more data is available.")
      }

      ## remove recods not identified to species level

      #out <- out[!is.na(out$species),]

      out$identifier <- roster$identifier

      if (roster$degrade == TRUE) {

        if (any(duplicated(out))) {
          out <- out[-which(duplicated(out)), ]
        }

      }


      nSpec <- length(unique(out$species))
      nRec <- length(out[,1])

      attr(dat, "nSpec") <- as.numeric(nSpec)
      attr(dat, "nRec") <- as.numeric(nRec)

    } else {
      out <- 0
      warning("This query produced zero records")
    }

    if (roster$write == TRUE) {

      write.csv(out,
                paste0(roster$outPath, roster$identifier,
                       "_", roster$country, ".csv"),
                row.names = F)

    }

    if (roster$write == FALSE) {

      return (out)

    }

    print(paste("Extraction completed for", roster$outName, "in", roster$country))

  } else {

    warning(paste0("There are more than 100,000 records for", roster$outName, "in", roster$country))

  }

}