Merge commit 'e088a1c892e97436fdd3263c6cfc6df61e47a365'

NAMESPACE

@@ -71,15 +71,6 @@ S3method(xml_add_parent,xml_nodeset)
 S3method(xml_add_sibling,xml_missing)
 S3method(xml_add_sibling,xml_node)
 S3method(xml_add_sibling,xml_nodeset)
-S3method(xml_attr,xml_missing)
-S3method(xml_attr,xml_node)
-S3method(xml_attr,xml_nodeset)
-S3method(xml_attrs,xml_missing)
-S3method(xml_attrs,xml_node)
-S3method(xml_attrs,xml_nodeset)
-S3method(xml_double,xml_missing)
-S3method(xml_double,xml_node)
-S3method(xml_double,xml_nodeset)
 S3method(xml_find_all,xml_missing)
 S3method(xml_find_all,xml_node)
 S3method(xml_find_all,xml_nodeset)
@@ -95,25 +86,13 @@ S3method(xml_find_lgl,xml_nodeset)
 S3method(xml_find_num,xml_missing)
 S3method(xml_find_num,xml_node)
 S3method(xml_find_num,xml_nodeset)
-S3method(xml_integer,xml_missing)
-S3method(xml_integer,xml_node)
-S3method(xml_integer,xml_nodeset)
-S3method(xml_length,xml_missing)
-S3method(xml_length,xml_node)
-S3method(xml_length,xml_nodeset)
-S3method(xml_name,xml_missing)
-S3method(xml_name,xml_node)
-S3method(xml_name,xml_nodeset)
 S3method(xml_ns,xml_document)
 S3method(xml_ns,xml_missing)
 S3method(xml_ns,xml_node)
 S3method(xml_ns,xml_nodeset)
 S3method(xml_parent,xml_missing)
 S3method(xml_parent,xml_node)
 S3method(xml_parent,xml_nodeset)
-S3method(xml_path,xml_missing)
-S3method(xml_path,xml_node)
-S3method(xml_path,xml_nodeset)
 S3method(xml_remove,xml_missing)
 S3method(xml_remove,xml_node)
 S3method(xml_remove,xml_nodeset)
@@ -132,12 +111,6 @@ S3method(xml_set_attrs,xml_nodeset)
 S3method(xml_set_name,xml_missing)
 S3method(xml_set_name,xml_node)
 S3method(xml_set_name,xml_nodeset)
-S3method(xml_text,xml_missing)
-S3method(xml_text,xml_node)
-S3method(xml_text,xml_nodeset)
-S3method(xml_type,xml_missing)
-S3method(xml_type,xml_node)
-S3method(xml_type,xml_nodeset)
 S3method(xml_url,xml_missing)
 S3method(xml_url,xml_node)
 S3method(xml_url,xml_nodeset)

NEWS.md

@@ -1,6 +1,7 @@
 # xml2 (development version)
 
 * Remove unused dependencies on glue, withr and lifecycle (@mgirlich).
+* `print()` is faster for very long `xml_nodeset` inputs (#366, @michaelchirico).
 
 # xml2 1.3.5
 

R/cpp11.R

@@ -64,24 +64,24 @@ ns_lookup <- function(doc_sxp, node_sxp, prefix_sxp) {
   .Call(`_xml2_ns_lookup`, doc_sxp, node_sxp, prefix_sxp)
 }
 
-node_name <- function(node_sxp, nsMap) {
-  .Call(`_xml2_node_name`, node_sxp, nsMap)
+node_name <- function(x, nsMap) {
+  .Call(`_xml2_node_name`, x, nsMap)
 }
 
 node_set_name <- function(node_sxp, value) {
   .Call(`_xml2_node_set_name`, node_sxp, value)
 }
 
-node_text <- function(node_sxp) {
-  .Call(`_xml2_node_text`, node_sxp)
+node_text <- function(x) {
+  .Call(`_xml2_node_text`, x)
 }
 
-node_attr <- function(node_sxp, name_sxp, missing_sxp, nsMap_sxp) {
-  .Call(`_xml2_node_attr`, node_sxp, name_sxp, missing_sxp, nsMap_sxp)
+node_attr <- function(x, name_sxp, missing_sxp, nsMap_sxp) {
+  .Call(`_xml2_node_attr`, x, name_sxp, missing_sxp, nsMap_sxp)
 }
 
-node_attrs <- function(node_sxp, nsMap_sxp) {
-  .Call(`_xml2_node_attrs`, node_sxp, nsMap_sxp)
+node_attrs <- function(x, nsMap_sxp) {
+  .Call(`_xml2_node_attrs`, x, nsMap_sxp)
 }
 
 node_set_attr <- function(node_sxp, name_sxp, value, nsMap) {
@@ -96,8 +96,8 @@ node_children <- function(node_sxp, only_node_sxp) {
   .Call(`_xml2_node_children`, node_sxp, only_node_sxp)
 }
 
-node_length <- function(node_sxp, only_node_sxp) {
-  .Call(`_xml2_node_length`, node_sxp, only_node_sxp)
+node_length <- function(x, only_node_sxp) {
+  .Call(`_xml2_node_length`, x, only_node_sxp)
 }
 
 node_has_children <- function(node_sxp, only_node_sxp) {
@@ -116,16 +116,16 @@ node_parent <- function(node_sxp) {
   .Call(`_xml2_node_parent`, node_sxp)
 }
 
-node_path <- function(node_sxp) {
-  .Call(`_xml2_node_path`, node_sxp)
+node_path <- function(x) {
+  .Call(`_xml2_node_path`, x)
 }
 
 nodes_duplicated <- function(nodes) {
   .Call(`_xml2_nodes_duplicated`, nodes)
 }
 
-node_type <- function(node_sxp) {
-  .Call(`_xml2_node_type`, node_sxp)
+node_type <- function(x) {
+  .Call(`_xml2_node_type`, x)
 }
 
 node_copy <- function(node_sxp) {

R/xml_attr.R

@@ -60,31 +60,7 @@
 #' xml_attrs(doc) <- c("b:id" = "one", "f:id" = "two", "id" = "three")
 #' xml_set_attrs(doc, c("b:id" = "one", "f:id" = "two", "id" = "three"))
 xml_attr <- function(x, attr, ns = character(), default = NA_character_) {
-  UseMethod("xml_attr")
-}
-
-#' @export
-xml_attr.xml_missing <- function(x, attr, ns = character(), default = NA_character_) {
-  default
-}
-
-#' @export
-xml_attr.xml_node <- function(x, attr, ns = character(),
-                              default = NA_character_) {
-  node_attr(x$node, attr, as.character(default), ns)
-}
-
-#' @export
-xml_attr.xml_nodeset <- function(x, attr, ns = character(),
-                                 default = NA_character_) {
-  vapply(
-    x,
-    xml_attr,
-    attr = attr,
-    default = default,
-    ns = ns,
-    FUN.VALUE = character(1)
-  )
+  node_attr(x, attr, as.character(default), ns)
 }
 
 #' @export
@@ -96,22 +72,7 @@ xml_has_attr <- function(x, attr, ns = character()) {
 #' @export
 #' @rdname xml_attr
 xml_attrs <- function(x, ns = character()) {
-  UseMethod("xml_attrs")
-}
-
-#' @export
-xml_attrs.xml_missing <- function(x, ns = character()) {
-  NA_character_
-}
-
-#' @export
-xml_attrs.xml_node <- function(x, ns = character()) {
-  node_attrs(x$node, nsMap_sxp = ns)
-}
-
-#' @export
-xml_attrs.xml_nodeset <- function(x, ns = character()) {
-  lapply(x, xml_attrs, ns = ns)
+  node_attrs(x, nsMap_sxp = ns)
 }
 
 #' @param value character vector of new value.

R/xml_children.R

@@ -100,26 +100,7 @@ xml_parent.xml_nodeset <- function(x) {
 #' @export
 #' @rdname xml_children
 xml_length <- function(x, only_elements = TRUE) {
-  UseMethod("xml_length")
-}
-
-#' @export
-xml_length.xml_missing <- function(x, only_elements = TRUE) {
-  0L
-}
-
-#' @export
-xml_length.xml_node <- function(x, only_elements = TRUE) {
-  node_length(x$node, only_elements)
-}
-
-#' @export
-xml_length.xml_nodeset <- function(x, only_elements = TRUE) {
-  if (length(x) == 0) {
-    return(0L)
-  }
-
-  vapply(x, xml_length, only_elements = only_elements, FUN.VALUE = integer(1))
+  node_length(x, only_elements)
 }
 
 #' @export

R/xml_name.R

@@ -18,22 +18,7 @@
 #' z <- xml_children(y)
 #' xml_name(xml_children(y))
 xml_name <- function(x, ns = character()) {
-  UseMethod("xml_name")
-}
-
-#' @export
-xml_name.xml_missing <- function(x, ns = character()) {
-  NA_character_
-}
-
-#' @export
-xml_name.xml_nodeset <- function(x, ns = character()) {
-  vapply(x, xml_name, ns = ns, FUN.VALUE = character(1))
-}
-
-#' @export
-xml_name.xml_node <- function(x, ns = character()) {
-  node_name(x$node, ns)
+  node_name(x, ns)
 }
 
 #' Modify the (tag) name of an element

R/xml_nodeset.R

@@ -38,6 +38,23 @@ as.character.xml_nodeset <- function(x, ...) {
   xml_nodeset(NextMethod())
 }
 
+#' Wrapper for encodeString() that takes width into consideration
+#'
+#' encodeString() is relatively expensive to run (see #366), so
+#'   avoid doing so to very wide inputs by first trimming inputs
+#'   to approximately the correct width, then encoding. A second
+#'   round of truncation occurs after encoding to account for
+#'   any newly-inserted characters bumping an input too wide.
+#' @noRd
+encode_with_width <- function(x, width) {
+  truncate_raw <- nchar(x) > width
+  x[truncate_raw] <- substr(x[truncate_raw], 1L, width - 3L)
+  x <- encodeString(x)
+  truncate_encoded <- truncate_raw | nchar(x) > width
+  x[truncate_encoded] <- paste(substr(x[truncate_encoded], 1L, width - 3L), "...")
+  x
+}
+
 show_nodes <- function(x, width = getOption("width"), max_n = 20) {
   stopifnot(inherits(x, "xml_nodeset"))
 
@@ -46,20 +63,16 @@ show_nodes <- function(x, width = getOption("width"), max_n = 20) {
     return()
   }
 
-  if (n > max_n) {
+  trunc <- n > max_n
+  if (trunc) {
     n <- max_n
     x <- x[seq_len(n)]
-    trunc <- TRUE
-  } else {
-    trunc <- FALSE
   }
 
   label <- format(paste0("[", seq_len(n), "]"), justify = "right")
-  contents <- encodeString(vapply(x, as.character, FUN.VALUE = character(1)))
+  contents <- vapply(x, as.character, FUN.VALUE = character(1L))
 
-  desc <- paste0(label, " ", contents)
-  needs_trunc <- nchar(desc) > width
-  desc[needs_trunc] <- paste(substr(desc[needs_trunc], 1, width - 3), "...")
+  desc <- encode_with_width(paste(label, contents), width)
 
   cat(desc, sep = "\n")
   if (trunc) {

R/xml_path.R

@@ -10,20 +10,5 @@
 #' x <- read_xml("<foo><bar><baz /></bar><baz /></foo>")
 #' xml_path(xml_find_all(x, ".//baz"))
 xml_path <- function(x) {
-  UseMethod("xml_path")
-}
-
-#' @export
-xml_path.xml_missing <- function(x) {
-  NA_character_
-}
-
-#' @export
-xml_path.xml_node <- function(x) {
-  node_path(x$node)
-}
-
-#' @export
-xml_path.xml_nodeset <- function(x) {
-  vapply(x, xml_path, FUN.VALUE = character(1))
+  node_path(x)
 }

R/xml_text.R

@@ -22,27 +22,16 @@
 #' xml_integer(xml_find_all(x, "//@x"))
 #' @export
 xml_text <- function(x, trim = FALSE) {
-  UseMethod("xml_text")
-}
-
-#' @export
-xml_text.xml_missing <- function(x, trim = FALSE) {
-  NA_character_
-}
-
-#' @export
-xml_text.xml_node <- function(x, trim = FALSE) {
-  res <- node_text(x$node)
+  res <- node_text(x)
   if (isTRUE(trim)) {
-    res <- sub("^[[:space:]\u00a0]+", "", res)
-    res <- sub("[[:space:]\u00a0]+$", "", res)
+    res <- trim_text(res)
   }
   res
 }
 
-#' @export
-xml_text.xml_nodeset <- function(x, trim = FALSE) {
-  vapply(x, xml_text, trim = trim, FUN.VALUE = character(1))
+trim_text <- function(x) {
+  x <- sub("^[[:space:]\u00a0]+", "", x)
+  sub("[[:space:]\u00a0]+$", "", x)
 }
 
 #' @rdname xml_text
@@ -94,46 +83,11 @@ xml_text.xml_nodeset <- function(x, trim = FALSE) {
 #' @rdname xml_text
 #' @export
 xml_double <- function(x) {
-  UseMethod("xml_double")
-}
-
-#' @export
-xml_double.xml_missing <- function(x) {
-  NA_real_
-}
-
-#' @export
-xml_double.xml_node <- function(x) {
   as.numeric(xml_text(x))
 }
 
-#' @export
-xml_double.xml_nodeset <- function(x) {
-  vapply(x, xml_double, numeric(1))
-}
-
-#' @export
-xml_integer <- function(x) {
-  UseMethod("xml_integer")
-}
-
-#' @export
-xml_integer.xml_missing <- function(x) {
-  NA_integer_
-}
-
 #' @rdname xml_text
 #' @export
 xml_integer <- function(x) {
-  UseMethod("xml_integer")
-}
-
-#' @export
-xml_integer.xml_node <- function(x) {
   as.integer(xml_text(x))
 }
-
-#' @export
-xml_integer.xml_nodeset <- function(x) {
-  vapply(x, xml_integer, integer(1))
-}