{XML,HTML}::Document.parse: handle Pathname arguments

Fixes sparklemotion#1821, sparklemotion#2110

lib/nokogiri.rb

@@ -3,6 +3,7 @@
 # Modify the PATH on windows so that the external DLLs will get loaded.
 
 require 'rbconfig'
+require 'pathname'
 
 if defined?(RUBY_ENGINE) && RUBY_ENGINE == "jruby"
   require 'nokogiri/jruby/dependencies'
@@ -54,6 +55,9 @@ class << self
     ###
     # Parse an HTML or XML document.  +string+ contains the document.
     def parse string, url = nil, encoding = nil, options = nil
+      # whackamole; see respective (XML|HTML)::Document
+      string = string.expand_path.open('rb') if string.is_a? Pathname
+
       if string.respond_to?(:read) ||
           /^\s*<(?:!DOCTYPE\s+)?html[\s>]/i === string[0, 512]
         # Expect an HTML indicator to appear within the first 512

lib/nokogiri/html/document.rb

@@ -1,4 +1,8 @@
 # frozen_string_literal: true
+
+# for the constant; pathname is standard lib
+require 'pathname'
+
 module Nokogiri
   module HTML
     class Document < Nokogiri::XML::Document
@@ -166,6 +170,10 @@ def parse string_or_io, url = nil, encoding = nil, options = XML::ParseOptions::
           # Give the options to the user
           yield options if block_given?
 
+          # coerce to binary filehandle if this is a Pathname object
+          string_or_io = string_or_io.expand_path.open('rb') if
+            string_or_io.is_a? Pathname
+
           if string_or_io.respond_to?(:encoding)
             unless string_or_io.encoding.name == "ASCII-8BIT"
               encoding ||= string_or_io.encoding.name

lib/nokogiri/xml/document.rb

@@ -1,4 +1,8 @@
 # frozen_string_literal: true
+
+# for the constant; pathname is standard lib
+require 'pathname'
+
 module Nokogiri
   module XML
     ##
@@ -56,12 +60,20 @@ def self.parse string_or_io, url = nil, encoding = nil, options = ParseOptions::
         end
 
         doc = if string_or_io.respond_to?(:read)
-          url ||= string_or_io.respond_to?(:path) ? string_or_io.path : nil
-          read_io(string_or_io, url, encoding, options.to_i)
-        else
-          # read_memory pukes on empty docs
-          read_memory(string_or_io, url, encoding, options.to_i)
-        end
+                # check url first cause this might get blown away
+                url ||= string_or_io.respond_to?(:path) ?
+                  string_or_io.path : nil
+
+                # coerce to binary filehandle if this is a Pathname object
+                string_or_io = string_or_io.expand_path.open('rb') if
+                  string_or_io.is_a? Pathname
+
+                # aaaand go
+                read_io(string_or_io, url, encoding, options.to_i)
+              else
+                # read_memory pukes on empty docs
+                read_memory(string_or_io, url, encoding, options.to_i)
+              end
 
         # do xinclude processing
         doc.do_xinclude(options) if options.xinclude?

test/test_nokogiri.rb

@@ -25,6 +25,24 @@ def test_atom_is_xml?
     assert !doc.html?
   end
 
+  def test_atom_from_pathname
+    # atom file is big enough to trip the input callback more than once
+
+    path = Pathname(XML_ATOM_FILE) # pathname should be already required
+
+    # XXX this behaviour should probably change
+    assert Nokogiri.parse(path).html?
+
+    # we explicitly say xml because of Nokogiri.parse behaviour
+    doc = Nokogiri.XML(path)
+
+    assert doc.xml?
+    assert !doc.html?
+
+    # wqe already know the second half of this works
+    assert_equal doc.to_xml, Nokogiri.parse(File.read(XML_ATOM_FILE)).to_xml
+  end
+
   def test_html?
     doc = Nokogiri.parse(File.read(HTML_FILE))
     assert !doc.xml?