Return genera listed by theplantlist.org. Provide a vector of family names with argument 'family' to restrict to one or more families.

tpl_genera.R

tpl_genera <- function (family=NULL) {
  require(RCurl)
  require(XML)
  require(plyr)
  if (!is.null(family)) {
    if (!exists('tpl_families') || !is.function(tpl_families)) {
      stop(paste('If providing family names, the function "tpl_families" is required.',
                 'See https://gist.github.com/johnbaums/6958504.', sep='\n'),                    
           call.=FALSE)
    }
    families <- tpl_families()
    if (all(!tolower(family) %in% tolower(families$family))) {
      stop(paste('Requested families not found on TPL.',
                 'Use tpl_families() to list plant families indexed by TPL.'),
           call.=FALSE)
    }
    if (any(!tolower(family) %in% tolower(families$family))) {
      warning(sprintf('Requested families not found on TPL: %s.\n%s', 
                      paste(family[!tolower(family) %in% tolower(families)], 
                            collapse=', '),
                      'Use tpl_families() to list plant families indexed by TPL.'),
              call.=FALSE)
    }
    families <- families[tolower(families$family) %in% tolower(family), ]
    families$urls <- file.path('http://www.theplantlist.org/browse',
                               substring(families$group, 1, 1),
                               families$family, '')
    get.genera <- function(u) {
      doc <- getURL(u)
      doc <- htmlParse(doc)
      xpathSApply(doc, "//ul[@id='nametree']//a", xmlValue)  
    }
    genera <- llply(families$urls, get.genera, .progress='text')
    data.frame(family=factor(rep(families$family, sapply(genera, length))),
               genus=unlist(genera), stringsAsFactors=FALSE)
  } else {
    doc <- getURL('http://www.theplantlist.org/browse/-/-/')
    doc <- htmlParse(doc)
    ns.genera <- getNodeSet(doc, "//ul[@id='nametree']//a")  
    genera <- xmlSApply(ns.genera, xmlValue)  
    ns.family <- getNodeSet(doc, "//ul[@id='nametree']//i")
    family <- factor(xmlSApply(ns.family, xmlValue))
    data.frame(family=family, genus=genera)
  }
}

# Examples
all.genera <- tpl_genera()
some.genera <- tpl_genera(c("Boweniaceae", "Davalliaceae", 
                            "Myrtaceae", "Poaceae", "Rapateaceae"))