nokogiri 1.10.3 → 1.12.5

data/lib/nokogiri/xml/document.rb

@@ -1,3 +1,8 @@
+# coding: utf-8
+# frozen_string_literal: true
+
+require 'pathname'
+
 module Nokogiri
   module XML
     ##
@@ -9,11 +14,12 @@ module Nokogiri
     # Nokogiri::XML::Searchable#xpath
     #
     class Document < Nokogiri::XML::Node
-      # I'm ignoring unicode characters here.
-      # See http://www.w3.org/TR/REC-xml-names/#ns-decl for more details.
+      # See http://www.w3.org/TR/REC-xml-names/#ns-decl for more details. Note that we're not
+      # attempting to handle unicode characters partly because libxml2 doesn't handle unicode
+      # characters in NCNAMEs.
       NCNAME_START_CHAR = "A-Za-z_"
-      NCNAME_CHAR       = NCNAME_START_CHAR + "\\-.0-9"
-      NCNAME_RE         = /^xmlns(:[#{NCNAME_START_CHAR}][#{NCNAME_CHAR}]*)?$/
+      NCNAME_CHAR       = NCNAME_START_CHAR + "\\-\\.0-9"
+      NCNAME_RE         = /^xmlns(?::([#{NCNAME_START_CHAR}][#{NCNAME_CHAR}]*))?$/
 
       ##
       # Parse an XML file.
@@ -43,9 +49,11 @@ module Nokogiri
       #
       def self.parse string_or_io, url = nil, encoding = nil, options = ParseOptions::DEFAULT_XML
         options = Nokogiri::XML::ParseOptions.new(options) if Integer === options
-        # Give the options to the user
+
         yield options if block_given?
 
+        url ||= string_or_io.respond_to?(:path) ? string_or_io.path : nil
+
         if empty_doc?(string_or_io)
           if options.strict?
             raise Nokogiri::XML::SyntaxError.new("Empty document")
@@ -55,12 +63,17 @@ module Nokogiri
         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
+                if string_or_io.is_a?(Pathname)
+                  # resolve the Pathname to the file and open it as an IO object, see #2110
+                  string_or_io = string_or_io.expand_path.open
+                  url ||= string_or_io.path
+                end
+
+                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?
@@ -68,42 +81,142 @@ module Nokogiri
         return doc
       end
 
+      ##
+      # @!method wrap(java_document)
+      # @!scope class
+      #
+      # Create a {Document} using an existing Java DOM document object.
+      #
+      # The returned {Document} shares the same underlying data structure as the Java object, so
+      # changes in one are reflected in the other.
+      #
+      # @param java_document [Java::OrgW3cDom::Document]
+      # @return [Nokogiri::XML::Document]
+      # @note This method is only available when running JRuby.
+      # @note The class +Java::OrgW3cDom::Document+ is also accessible as +org.w3c.dom.Document+.
+      # @see #to_java
+
+      ##
+      # @!method to_java()
+      #
+      # Returns the underlying Java DOM document object for the {Document}.
+      #
+      # The returned Java object shares the same underlying data structure as the {Document}, so
+      # changes in one are reflected in the other.
+      #
+      # @return [Java::OrgW3cDom::Document]
+      # @note This method is only available when running JRuby.
+      # @note The class +Java::OrgW3cDom::Document+ is also accessible as +org.w3c.dom.Document+.
+      # @see .wrap
+
+
       # A list of Nokogiri::XML::SyntaxError found when parsing a document
       attr_accessor :errors
 
+      # When true, reparented elements without a namespace will inherit their new parent's
+      # namespace (if one exists). Defaults to +false+.
+      #
+      # @example Default behavior of namespace inheritance
+      #   xml = <<~EOF
+      #           <root xmlns:foo="http://nokogiri.org/default_ns/test/foo">
+      #             <foo:parent>
+      #             </foo:parent>
+      #           </root>
+      #         EOF
+      #   doc = Nokogiri::XML(xml)
+      #   parent = doc.at_xpath("//foo:parent", "foo" => "http://nokogiri.org/default_ns/test/foo")
+      #   parent.add_child("<child></child>")
+      #   doc.to_xml
+      #   # => <?xml version="1.0"?>
+      #   #    <root xmlns:foo="http://nokogiri.org/default_ns/test/foo">
+      #   #      <foo:parent>
+      #   #        <child/>
+      #   #      </foo:parent>
+      #   #    </root>
+      #
+      # @example Setting namespace inheritance to +true+
+      #   xml = <<~EOF
+      #           <root xmlns:foo="http://nokogiri.org/default_ns/test/foo">
+      #             <foo:parent>
+      #             </foo:parent>
+      #           </root>
+      #         EOF
+      #   doc = Nokogiri::XML(xml)
+      #   doc.namespace_inheritance = true
+      #   parent = doc.at_xpath("//foo:parent", "foo" => "http://nokogiri.org/default_ns/test/foo")
+      #   parent.add_child("<child></child>")
+      #   doc.to_xml
+      #   # => <?xml version="1.0"?>
+      #   #    <root xmlns:foo="http://nokogiri.org/default_ns/test/foo">
+      #   #      <foo:parent>
+      #   #        <foo:child/>
+      #   #      </foo:parent>
+      #   #    </root>
+      #
+      # @return [Boolean]
+      #
+      # @since v1.12.4
+      attr_accessor :namespace_inheritance
+
       def initialize *args # :nodoc:
         @errors     = []
         @decorators = nil
+        @namespace_inheritance = false
       end
 
       ##
-      # Create an element with +name+, and optionally setting the content and attributes.
+      # Create a new +Element+ with +name+ sharing GC lifecycle with the document, optionally
+      # setting contents or attributes.
+      #
+      # Arguments may be passed to initialize the element:
+      # - a +Hash+ argument will be used to set attributes
+      # - a non-Hash object that responds to +#to_s+ will be used to set the new node's contents
+      #
+      # A block may be passed to mutate the node.
+      #
+      # @param name [String]
+      # @param contents_or_attrs [#to_s,Hash]
+      # @yieldparam node [Nokogiri::XML::Element]
+      # @return [Nokogiri::XML::Element]
+      #
+      # @example An empty element without attributes
+      #   doc.create_element("div")
+      #   # => <div></div>
+      #
+      # @example An element with contents
+      #   doc.create_element("div", "contents")
+      #   # => <div>contents</div>
       #
-      #   doc.create_element "div" # <div></div>
-      #   doc.create_element "div", :class => "container" # <div class='container'></div>
-      #   doc.create_element "div", "contents" # <div>contents</div>
-      #   doc.create_element "div", "contents", :class => "container" # <div class='container'>contents</div>
-      #   doc.create_element "div" { |node| node['class'] = "container" } # <div class='container'></div>
+      # @example An element with attributes
+      #   doc.create_element("div", {"class" => "container"})
+      #   # => <div class='container'></div>
       #
-      def create_element name, *args, &block
+      # @example An element with contents and attributes
+      #   doc.create_element("div", "contents", {"class" => "container"})
+      #   # => <div class='container'>contents</div>
+      #
+      # @example Passing a block to mutate the element
+      #   doc.create_element("div") { |node| node["class"] = "blue" if before_noon? }
+      #
+      def create_element(name, *contents_or_attrs, &block)
         elm = Nokogiri::XML::Element.new(name, self, &block)
-        args.each do |arg|
+        contents_or_attrs.each do |arg|
           case arg
           when Hash
-            arg.each { |k,v|
+            arg.each do |k, v|
               key = k.to_s
               if key =~ NCNAME_RE
-                ns_name = key.split(":", 2)[1]
-                elm.add_namespace_definition ns_name, v
+                ns_name = Regexp.last_match(1)
+                elm.add_namespace_definition(ns_name, v)
               else
                 elm[k.to_s] = v.to_s
               end
-            }
+            end
           else
             elm.content = arg
           end
         end
-        if ns = elm.namespace_definitions.find { |n| n.prefix.nil? or n.prefix == '' }
+        if ns = elm.namespace_definitions.find { |n| n.prefix.nil? || (n.prefix == '') }
           elm.namespace = ns
         end
         elm
@@ -251,30 +364,14 @@ module Nokogiri
       end
       alias :<< :add_child
 
-      ##
-      # +JRuby+
-      # Wraps Java's org.w3c.dom.document and returns Nokogiri::XML::Document
-      def self.wrap document
-        raise "JRuby only method" unless Nokogiri.jruby?
-        return wrapJavaDocument(document)
-      end
-
-      ##
-      # +JRuby+
-      # Returns Java's org.w3c.dom.document of this Document.
-      def to_java
-        raise "JRuby only method" unless Nokogiri.jruby?
-        return toJavaDocument()
-      end
-
       private
+
       def self.empty_doc? string_or_io
         string_or_io.nil? ||
           (string_or_io.respond_to?(:empty?) && string_or_io.empty?) ||
           (string_or_io.respond_to?(:eof?) && string_or_io.eof?)
       end
 
-      # @private
       IMPLIED_XPATH_CONTEXTS = [ '//'.freeze ].freeze # :nodoc:
 
       def inspect_attributes

data/lib/nokogiri/xml/document_fragment.rb

@@ -1,3 +1,4 @@
+# frozen_string_literal: true
 module Nokogiri
   module XML
     class DocumentFragment < Nokogiri::XML::Node
@@ -140,6 +141,10 @@ module Nokogiri
         document.errors = things
       end
 
+      def fragment(data)
+        document.fragment(data)
+      end
+
       private
 
       # fix for issue 770
@@ -149,12 +154,6 @@ module Nokogiri
           %Q{xmlns#{prefix}="#{namespace.href}"}
         end.join ' '
       end
-
-      def coerce data
-        return super unless String === data
-
-        document.fragment(data).children
-      end
     end
   end
 end

data/lib/nokogiri/xml/dtd.rb

@@ -1,3 +1,4 @@
+# frozen_string_literal: true
 module Nokogiri
   module XML
     class DTD < Nokogiri::XML::Node

data/lib/nokogiri/xml/element_content.rb

@@ -1,3 +1,4 @@
+# frozen_string_literal: true
 module Nokogiri
   module XML
     ###

data/lib/nokogiri/xml/element_decl.rb

@@ -1,3 +1,4 @@
+# frozen_string_literal: true
 module Nokogiri
   module XML
     class ElementDecl < Nokogiri::XML::Node

data/lib/nokogiri/xml/entity_decl.rb

@@ -1,3 +1,4 @@
+# frozen_string_literal: true
 module Nokogiri
   module XML
     class EntityDecl < Nokogiri::XML::Node

data/lib/nokogiri/xml/entity_reference.rb

@@ -1,3 +1,4 @@
+# frozen_string_literal: true
 module Nokogiri
   module XML
     class EntityReference < Nokogiri::XML::Node

data/lib/nokogiri/xml/namespace.rb

@@ -1,3 +1,4 @@
+# frozen_string_literal: true
 module Nokogiri
   module XML
     class Namespace

data/lib/nokogiri/xml/node/save_options.rb

@@ -1,3 +1,4 @@
+# frozen_string_literal: true
 module Nokogiri
   module XML
     class Node
@@ -33,7 +34,7 @@ module Nokogiri
           DEFAULT_HTML = FORMAT | NO_DECLARATION | NO_EMPTY_TAGS | AS_HTML
         end
         # the default for XHTML document
-        DEFAULT_XHTML = FORMAT | NO_DECLARATION | NO_EMPTY_TAGS | AS_XHTML
+        DEFAULT_XHTML = FORMAT | NO_DECLARATION | AS_XHTML
 
         # Integer representation of the SaveOptions
         attr_reader :options