nokogiri 1.5.10 → 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
     ##
@@ -5,15 +10,16 @@ module Nokogiri
     # XML documents.  The Document is created by parsing an XML document.
     # See Nokogiri::XML::Document.parse() for more information on parsing.
     #
-    # For searching a Document, see Nokogiri::XML::Node#css and
-    # Nokogiri::XML::Node#xpath
+    # For searching a Document, see Nokogiri::XML::Searchable#css and
+    # 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.
@@ -33,70 +39,184 @@ module Nokogiri
       # +block+ (optional) is passed a configuration object on which
       # parse options may be set.
       #
-      # When parsing untrusted documents, it's recommended that the
-      # +nonet+ option be used, as shown in this example code:
-      #
-      #   Nokogiri::XML::Document.parse(xml_string) { |config| config.nonet }
+      # By default, Nokogiri treats documents as untrusted, and so
+      # does not attempt to load DTDs or access the network. See
+      # Nokogiri::XML::ParseOptions for a complete list of options;
+      # and that module's DEFAULT_XML constant for what's set (and not
+      # set) by default.
       #
       # Nokogiri.XML() is a convenience method which will call this method.
       #
-      def self.parse string_or_io, url = nil, encoding = nil, options = ParseOptions::DEFAULT_XML, &block
-        options = Nokogiri::XML::ParseOptions.new(options) if Fixnum === options
-        # Give the options to the user
+      def self.parse string_or_io, url = nil, encoding = nil, options = ParseOptions::DEFAULT_XML
+        options = Nokogiri::XML::ParseOptions.new(options) if Integer === options
+
         yield options if block_given?
 
-        return new if empty_doc?(string_or_io)
+        url ||= string_or_io.respond_to?(:path) ? string_or_io.path : nil
 
-        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)
+        if empty_doc?(string_or_io)
+          if options.strict?
+            raise Nokogiri::XML::SyntaxError.new("Empty document")
+          else
+            return encoding ? new.tap { |i| i.encoding = encoding } : new
+          end
         end
 
+        doc = if string_or_io.respond_to?(:read)
+                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?
 
         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>
       #
-      #   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 contents
+      #   doc.create_element("div", "contents")
+      #   # => <div>contents</div>
       #
-      def create_element name, *args, &block
+      # @example An element with attributes
+      #   doc.create_element("div", {"class" => "container"})
+      #   # => <div class='container'></div>
+      #
+      # @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
@@ -233,10 +353,10 @@ module Nokogiri
       undef_method :namespace_definitions, :line, :add_namespace
 
       def add_child node_or_tags
-        raise "Document already has a root node" if root
+        raise "A document may not have multiple root nodes." if (root && root.name != 'nokogiri_text_wrapper') && !(node_or_tags.comment? || node_or_tags.processing_instruction?)
         node_or_tags = coerce(node_or_tags)
         if node_or_tags.is_a?(XML::NodeSet)
-          raise "Document cannot have multiple root nodes" if node_or_tags.size > 1
+          raise "A document may not have multiple root nodes." if node_or_tags.size > 1
           super(node_or_tags.first)
         else
           super
@@ -244,32 +364,15 @@ 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
 
-      def implied_xpath_context
-        "/"
-      end
+      IMPLIED_XPATH_CONTEXTS = [ '//'.freeze ].freeze # :nodoc:
 
       def inspect_attributes
         [:name, :children]

data/lib/nokogiri/xml/document_fragment.rb

@@ -1,3 +1,4 @@
+# frozen_string_literal: true
 module Nokogiri
   module XML
     class DocumentFragment < Nokogiri::XML::Node
@@ -25,6 +26,17 @@ module Nokogiri
         children.each { |child| child.parent = self }
       end
 
+      if Nokogiri.uses_libxml?
+        def dup
+          new_document = document.dup
+          new_fragment = self.class.new(new_document)
+          children.each do |child|
+            child.dup(1, new_document).parent = new_fragment
+          end
+          new_fragment
+        end
+      end
+
       ###
       # return the name for DocumentFragment
       def name
@@ -73,15 +85,43 @@ module Nokogiri
       end
 
       ###
-      # Search this fragment.  See Nokogiri::XML::Node#css
+      # call-seq: css *rules, [namespace-bindings, custom-pseudo-class]
+      #
+      # Search this fragment for CSS +rules+. +rules+ must be one or more CSS
+      # selectors. For example:
+      #
+      # For more information see Nokogiri::XML::Searchable#css
       def css *args
         if children.any?
-          children.css(*args)
+          children.css(*args) # 'children' is a smell here
         else
           NodeSet.new(document)
         end
       end
 
+      #
+      #  NOTE that we don't delegate #xpath to children ... another smell.
+      #  def xpath ; end
+      #
+
+      ###
+      # call-seq: search *paths, [namespace-bindings, xpath-variable-bindings, custom-handler-class]
+      #
+      # Search this fragment for +paths+. +paths+ must be one or more XPath or CSS queries.
+      #
+      # For more information see Nokogiri::XML::Searchable#search
+      def search *rules
+        rules, handler, ns, binds = extract_params(rules)
+
+        rules.inject(NodeSet.new(document)) do |set, rule|
+          set += if rule =~ Searchable::LOOKS_LIKE_XPATH
+                   xpath(*([rule, ns, handler, binds].compact))
+                 else
+                   children.css(*([rule, ns, handler].compact)) # 'children' is a smell here
+                 end
+        end
+      end
+
       alias :serialize :to_s
 
       class << self
@@ -92,6 +132,19 @@ module Nokogiri
         end
       end
 
+      # A list of Nokogiri::XML::SyntaxError found when parsing a document
+      def errors
+        document.errors
+      end
+
+      def errors= things # :nodoc:
+        document.errors = things
+      end
+
+      def fragment(data)
+        document.fragment(data)
+      end
+
       private
 
       # fix for issue 770
@@ -101,12 +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
@@ -12,10 +13,20 @@ module Nokogiri
         attributes.keys
       end
 
-      def each &block
-        attributes.each { |key, value|
-          block.call([key, value])
-        }
+      def each
+        attributes.each do |key, value|
+          yield([key, value])
+        end
+      end
+
+      def html_dtd?
+        name.casecmp('html').zero?
+      end
+
+      def html5_dtd?
+        html_dtd? &&
+          external_id.nil? &&
+          (system_id.nil? || system_id == 'about:legacy-compat')
       end
     end
   end

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

@@ -0,0 +1,19 @@
+# frozen_string_literal: true
+module Nokogiri
+  module XML
+    class EntityReference < Nokogiri::XML::Node
+      def children
+        # libxml2 will create a malformed child node for predefined
+        # entities. because any use of that child is likely to cause a
+        # segfault, we shall pretend that it doesn't exist.
+        #
+        # see https://github.com/sparklemotion/nokogiri/issues/1238 for details
+        NodeSet.new(document)
+      end
+
+      def inspect_attributes
+        [:name]
+      end
+    end
+  end
+end

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