Trim non-breaking spaces with xml_text(trim = TRUE)

Fixes #151

NEWS.md

@@ -1,5 +1,7 @@
 # xml2 1.0.0.9000
 
+* Trim non-breaking spaces in `xml_text(trim = TRUE)` (#151).
+
 * Allow setting non-character attributes (values are coerced to characters). (@sjp, #117, #122).
 
 * Fixed return value in call to vapply in xml_integer.xml_nodeset. (@ddiez, #146, #147).

src/xml2_node.cpp

@@ -37,8 +37,9 @@ void node_set_name(XPtrNode node, std::string value) {
 CharacterVector node_text(XPtrNode node, bool trim) {
   std::string text = Xml2String(xmlNodeGetContent(node.checked_get())).asStdString();
 
-  if (trim)
-    boost::algorithm::trim(text);
+  if (trim) {
+    boost::algorithm::trim_if(text, (! boost::algorithm::is_print()) || boost::algorithm::is_space());
+  }
 
   return asCharacterVector(text.c_str());
 }

tests/testthat/test-xml_text.R

@@ -25,10 +25,10 @@ test_that("xml_text works properly with xml_nodeset objects", {
     c("This is some text. This is some nested text.", "This is some nested text."))
 })
 
-test_that("xml_text trims whitespace if requested", {
-  x <- read_xml("<p>   Some text    </p>")
+test_that("xml_text trims whitespace if requested, including non-breaking spaces", {
+  x <- read_html("<p>   Some text    &nbsp;</p>")
   expect_identical(xml_text(x),
-    "   Some text    ")
+    "   Some text     ")
 
   expect_identical(xml_text(x, trim = TRUE),
     "Some text")