XML replaced with xml2, fix #98

ropensci · Feb 3, 2016 · 75b96e5 · 75b96e5
1 parent 698719a
commit 75b96e5
Show file tree

Hide file tree

Showing 7 changed files with 13 additions and 15 deletions.
diff --git a/DESCRIPTION b/DESCRIPTION
@@ -21,14 +21,12 @@ Imports:
     methods,
     stats,
     jsonlite,
-    XML,
     httr (>= 1.0.0),
     plyr,
     bibtex,
     dplyr,
     R6
 Suggests:
     testthat,
-    roxygen2 (>= 4.0.1),
     knitr
 RoxygenNote: 5.0.1
diff --git a/NAMESPACE b/NAMESPACE
@@ -41,9 +41,6 @@ export(get_styles)
 export(id_converter)
 export(pmid2doi)
 importFrom(R6,R6Class)
-importFrom(XML,xmlAttrs)
-importFrom(XML,xmlParse)
-importFrom(XML,xpathSApply)
 importFrom(bibtex,read.bib)
 importFrom(dplyr,bind_rows)
 importFrom(dplyr,rbind_all)
@@ -69,3 +66,6 @@ importFrom(plyr,llply)
 importFrom(plyr,rbind.fill)
 importFrom(stats,setNames)
 importFrom(utils,modifyList)
+importFrom(xml2,read_xml)
+importFrom(xml2,xml_attr)
+importFrom(xml2,xml_find_all)
diff --git a/R/cr_citation_count.r b/R/cr_citation_count.r
@@ -23,16 +23,14 @@ cr_citation_count <- function(doi, url = "http://www.crossref.org/openurl/",
 	key = "cboettig@ropensci.org", ...)
 {
   ## Assemble a url query such as:
-  #http://www.crossref.org/openurl/?id=doi:10.3998/3336451.0009.101&noredirect=true&pid=API_KEY&format=unixref
-  args <- list(id = paste("doi:", doi, sep=""), pid = as.character(key),
+  args <- list(id = paste("doi:", doi, sep = ""), pid = as.character(key),
                noredirect = as.logical(TRUE))
-#  args$format=as.character("unixref")
   cite_count <- GET(url, query = args, ...)
   stop_for_status(cite_count)
-  ans <- xmlParse(ct_utf8(cite_count))
-  if(get_attr(ans, "status") == "unresolved") NA else as.numeric(get_attr(ans, "fl_count"))
+  ans <- xml2::read_xml(ct_utf8(cite_count))
+  if (get_attr(ans, "status") == "unresolved") NA else as.numeric(get_attr(ans, "fl_count"))
 }
 
 get_attr <- function(xml, attr){
-  xpathSApply(xml, sprintf("//*[@%s]", attr), function(x) xmlAttrs(x)[[attr]])
+  xml2::xml_attr(xml2::xml_find_all(xml, sprintf("//*[@%s]", attr)), attr)
 }
diff --git a/R/cr_ft_text.R b/R/cr_ft_text.R
@@ -34,6 +34,7 @@
 #' # xml and plain text links
 #' out <- cr_works(filter=c(has_full_text = TRUE))
 #' dois <- out$data$DOI
+#' cr_ft_links(dois[1], "pdf")
 #' cr_ft_links(dois[2], "xml")
 #' cr_ft_links(dois[1], "plain")
 #' cr_ft_links(dois[1], "all")
@@ -227,7 +228,7 @@ cr_auth <- function(url, type) {
 getTEXT <- function(x, type, auth, ...){
   res <- GET(x, auth, ...)
   switch(type,
-         xml = XML::xmlParse(ct_utf8(res)),
+         xml = xml2::read_xml(ct_utf8(res)),
          plain = ct_utf8(res))
 }
 

diff --git a/R/rcrossref-package.R b/R/rcrossref-package.R
@@ -44,7 +44,7 @@
 #' @importFrom stats setNames
 #' @importFrom dplyr rbind_all bind_rows tbl_df
 #' @importFrom R6 R6Class
-#' @importFrom XML xmlParse xpathSApply xmlAttrs
+#' @importFrom xml2 read_xml xml_attr xml_find_all
 #' @importFrom bibtex read.bib
 #' @importFrom httr GET POST stop_for_status content_type_json accept_json content
 #' write_disk accept http_status parse_url add_headers config

diff --git a/R/zzz.R b/R/zzz.R
@@ -38,8 +38,8 @@ get_err <- function(x) {
   if (x$headers$`content-type` == "text/plain") {
     tmp <- xx
   } else if (x$headers$`content-type` == "text/html") {
-    html <- XML::htmlParse(xx)
-    tmp <- XML::xpathSApply(html, '//h3[@class="info"]', xmlValue)
+    html <- xml2::read_html(xx)
+    tmp <- xml2::xml_text(xml2::xml_find_one(html, '//h3[@class="info"]'))
   } else if (x$headers$`content-type` == "application/json;charset=UTF-8") {
     tmp <- jsonlite::fromJSON(xx, FALSE)
   } else {

diff --git a/man/cr_ft_text.Rd b/man/cr_ft_text.Rd