nokogiri 1.10.9 → 1.18.3

data/lib/nokogiri/xml/document.rb

@@ -1,132 +1,320 @@
+# coding: utf-8
+# frozen_string_literal: true
+
+require "pathname"
+
 module Nokogiri
   module XML
-    ##
-    # Nokogiri::XML::Document is the main entry point for dealing with
-    # 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::Searchable#css and
-    # Nokogiri::XML::Searchable#xpath
+    # Nokogiri::XML::Document is the main entry point for dealing with \XML documents. The Document
+    # is created by parsing \XML content from a String or an IO object. See
+    # Nokogiri::XML::Document.parse for more information on parsing.
     #
+    # Document inherits a great deal of functionality from its superclass Nokogiri::XML::Node, so
+    # please read that class's documentation as well.
     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}]*)?$/
+      NCNAME_CHAR       = NCNAME_START_CHAR + "\\-\\.0-9"
+      NCNAME_RE         = /^xmlns(?::([#{NCNAME_START_CHAR}][#{NCNAME_CHAR}]*))?$/
+
+      OBJECT_DUP_METHOD = Object.instance_method(:dup)
+      OBJECT_CLONE_METHOD = Object.instance_method(:clone)
+      private_constant :OBJECT_DUP_METHOD, :OBJECT_CLONE_METHOD
+
+      class << self
+        # call-seq:
+        #   parse(input) { |options| ... } => Nokogiri::XML::Document
+        #   parse(input, url:, encoding:, options:) => Nokogiri::XML::Document
+        #
+        # Parse \XML input from a String or IO object, and return a new XML::Document.
+        #
+        # 🛡 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.
+        #
+        # [Required Parameters]
+        # - +input+ (String | IO) The content to be parsed.
+        #
+        # [Optional Keyword Arguments]
+        # - +url:+ (String) The base URI for this document.
+        #
+        # - +encoding:+ (String) The name of the encoding that should be used when processing the
+        #   document. When not provided, the encoding will be determined based on the document
+        #   content.
+        #
+        # - +options:+ (Nokogiri::XML::ParseOptions) Configuration object that determines some
+        #   behaviors during parsing. See ParseOptions for more information. The default value is
+        #   +ParseOptions::DEFAULT_XML+.
+        #
+        # [Yields]
+        #   If a block is given, a Nokogiri::XML::ParseOptions object is yielded to the block which
+        #   can be configured before parsing. See Nokogiri::XML::ParseOptions for more information.
+        #
+        # [Returns] Nokogiri::XML::Document
+        def parse(
+          string_or_io,
+          url_ = nil, encoding_ = nil, options_ = XML::ParseOptions::DEFAULT_XML,
+          url: url_, encoding: encoding_, options: options_
+        )
+          options = Nokogiri::XML::ParseOptions.new(options) if Integer === options
+          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, "Empty document"
+            else
+              return encoding ? new.tap { |i| i.encoding = encoding } : new
+            end
+          end
+
+          doc = if string_or_io.respond_to?(:read)
+            # TODO: should we instead check for respond_to?(:to_path) ?
+            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?
+
+          doc
+        end
+
+        private
+
+        def 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
+      end
 
       ##
-      # Parse an XML file.
+      # :singleton-method: wrap
+      # :call-seq: wrap(java_document) → Nokogiri::XML::Document
       #
-      # +string_or_io+ may be a String, or any object that responds to
-      # _read_ and _close_ such as an IO, or StringIO.
+      # ⚠ This method is only available when running JRuby.
       #
-      # +url+ (optional) is the URI where this document is located.
+      # Create a Document using an existing Java DOM document object.
       #
-      # +encoding+ (optional) is the encoding that should be used when processing
-      # the document.
+      # The returned Document shares the same underlying data structure as the Java object, so
+      # changes in one are reflected in the other.
       #
-      # +options+ (optional) is a configuration object that sets options during
-      # parsing, such as Nokogiri::XML::ParseOptions::RECOVER. See the
-      # Nokogiri::XML::ParseOptions for more information.
+      # [Parameters]
+      # - `java_document` (Java::OrgW3cDom::Document)
+      #   (The class `Java::OrgW3cDom::Document` is also accessible as `org.w3c.dom.Document`.)
       #
-      # +block+ (optional) is passed a configuration object on which
-      # parse options may be set.
+      # [Returns] Nokogiri::XML::Document
       #
-      # 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.
+      # See also \#to_java
+
+      # :method: to_java
+      # :call-seq: to_java() → Java::OrgW3cDom::Document
       #
-      # Nokogiri.XML() is a convenience method which will call this method.
+      # ⚠ This method is only available when running JRuby.
       #
-      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?
-
-        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
+      # Returns the underlying Java DOM document object for this document.
+      #
+      # The returned Java object shares the same underlying data structure as this document, so
+      # changes in one are reflected in the other.
+      #
+      # [Returns]
+      #   Java::OrgW3cDom::Document
+      #   (The class `Java::OrgW3cDom::Document` is also accessible as `org.w3c.dom.Document`.)
+      #
+      # See also Document.wrap
 
-        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
+      # The errors found while parsing a document.
+      #
+      # [Returns] Array<Nokogiri::XML::SyntaxError>
+      attr_accessor :errors
 
-        # do xinclude processing
-        doc.do_xinclude(options) if options.xinclude?
+      # When `true`, reparented elements without a namespace will inherit their new parent's
+      # namespace (if one exists). Defaults to `false`.
+      #
+      # [Returns] Boolean
+      #
+      # *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>
+      #
+      # Since v1.12.4
+      attr_accessor :namespace_inheritance
 
-        return doc
+      def initialize(*args) # :nodoc: # rubocop:disable Lint/MissingSuper
+        @errors     = []
+        @decorators = nil
+        @namespace_inheritance = false
       end
 
-      # A list of Nokogiri::XML::SyntaxError found when parsing a document
-      attr_accessor :errors
+      #
+      # :call-seq:
+      #   dup → Nokogiri::XML::Document
+      #   dup(level) → Nokogiri::XML::Document
+      #
+      # Duplicate this node.
+      #
+      # [Parameters]
+      # - +level+ (optional Integer). 0 is a shallow copy, 1 (the default) is a deep copy.
+      # [Returns] The new Nokogiri::XML::Document
+      #
+      def dup(level = 1)
+        copy = OBJECT_DUP_METHOD.bind_call(self)
+        copy.initialize_copy_with_args(self, level)
+      end
 
-      def initialize *args # :nodoc:
-        @errors     = []
-        @decorators = nil
+      #
+      # :call-seq:
+      #   clone → Nokogiri::XML::Document
+      #   clone(level) → Nokogiri::XML::Document
+      #
+      # Clone this node.
+      #
+      # [Parameters]
+      # - +level+ (optional Integer). 0 is a shallow copy, 1 (the default) is a deep copy.
+      # [Returns] The new Nokogiri::XML::Document
+      #
+      def clone(level = 1)
+        copy = OBJECT_CLONE_METHOD.bind_call(self)
+        copy.initialize_copy_with_args(self, level)
       end
 
-      ##
-      # Create an element with +name+, and optionally setting the content and attributes.
+      # :call-seq:
+      #   create_element(name, *contents_or_attrs, &block) → Nokogiri::XML::Element
+      #
+      # Create a new Element with `name` belonging to this document, optionally setting contents or
+      # attributes.
+      #
+      # This method is _not_ the most user-friendly option if your intention is to add a node to the
+      # document tree. Prefer one of the Nokogiri::XML::Node methods like Node#add_child,
+      # Node#add_next_sibling, Node#replace, etc. which will both create an element (or subtree) and
+      # place it in the document tree.
+      #
+      # 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.
+      #
+      # [Parameters]
+      # - `name` (String)
+      # - `contents_or_attrs` (\#to_s, Hash)
+      # [Yields] `node` (Nokogiri::XML::Element)
+      # [Returns] Nokogiri::XML::Element
+      #
+      # *Example:* An empty element without attributes
+      #
+      #   doc.create_element("div")
+      #   # => <div></div>
+      #
+      # *Example:* An element with contents
       #
-      #   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>
+      #   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
       end
 
       # Create a Text Node with +string+
-      def create_text_node string, &block
-        Nokogiri::XML::Text.new string.to_s, self, &block
+      def create_text_node(string, &block)
+        Nokogiri::XML::Text.new(string.to_s, self, &block)
       end
 
       # Create a CDATA Node containing +string+
-      def create_cdata string, &block
-        Nokogiri::XML::CDATA.new self, string.to_s, &block
+      def create_cdata(string, &block)
+        Nokogiri::XML::CDATA.new(self, string.to_s, &block)
       end
 
       # Create a Comment Node containing +string+
-      def create_comment string, &block
-        Nokogiri::XML::Comment.new self, string.to_s, &block
+      def create_comment(string, &block)
+        Nokogiri::XML::Comment.new(self, string.to_s, &block)
       end
 
       # The name of this document.  Always returns "document"
       def name
-        'document'
+        "document"
       end
 
       # A reference to +self+
@@ -134,55 +322,61 @@ module Nokogiri
         self
       end
 
-      ##
-      # Recursively get all namespaces from this node and its subtree and
-      # return them as a hash.
+      # :call-seq:
+      #   collect_namespaces() → Hash<String(Namespace#prefix) ⇒ String(Namespace#href)>
       #
-      # For example, given this document:
+      # Recursively get all namespaces from this node and its subtree and return them as a
+      # hash.
       #
-      #   <root xmlns:foo="bar">
+      # ⚠ This method will not handle duplicate namespace prefixes, since the return value is a hash.
+      #
+      # Note that this method does an xpath lookup for nodes with namespaces, and as a result the
+      # order (and which duplicate prefix "wins") may be dependent on the implementation of the
+      # underlying XML library.
+      #
+      # *Example:* Basic usage
+      #
+      # Given this document:
+      #
+      #   <root xmlns="default" xmlns:foo="bar">
       #     <bar xmlns:hello="world" />
       #   </root>
       #
       # This method will return:
       #
-      #   { 'xmlns:foo' => 'bar', 'xmlns:hello' => 'world' }
+      #   {"xmlns:foo"=>"bar", "xmlns"=>"default", "xmlns:hello"=>"world"}
       #
-      # WARNING: this method will clobber duplicate names in the keys.
-      # For example, given this document:
+      # *Example:* Duplicate prefixes
+      #
+      # Given this document:
       #
       #   <root xmlns:foo="bar">
       #     <bar xmlns:foo="baz" />
       #   </root>
       #
-      # The hash returned will look like this: { 'xmlns:foo' => 'bar' }
-      #
-      # Non-prefixed default namespaces (as in "xmlns=") are not included
-      # in the hash.
+      # The hash returned will be something like:
       #
-      # Note that this method does an xpath lookup for nodes with
-      # namespaces, and as a result the order may be dependent on the
-      # implementation of the underlying XML library.
+      #   {"xmlns:foo" => "baz"}
       #
       def collect_namespaces
-        xpath("//namespace::*").inject({}) do |hash, ns|
-          hash[["xmlns",ns.prefix].compact.join(":")] = ns.href if ns.prefix != "xml"
-          hash
+        xpath("//namespace::*").each_with_object({}) do |ns, hash|
+          hash[["xmlns", ns.prefix].compact.join(":")] = ns.href if ns.prefix != "xml"
         end
       end
 
       # Get the list of decorators given +key+
-      def decorators key
-        @decorators ||= Hash.new
+      def decorators(key)
+        @decorators ||= {}
         @decorators[key] ||= []
       end
 
       ##
-      # Validate this Document against it's DTD.  Returns a list of errors on
+      # Validate this Document against its DTD.  Returns a list of errors on
       # the document or +nil+ when there is no DTD.
       def validate
-        return nil unless internal_subset
-        internal_subset.validate self
+        return unless internal_subset
+
+        internal_subset.validate(self)
       end
 
       ##
@@ -202,7 +396,7 @@ module Nokogiri
       #   ... which does absolutely nothing.
       #
       def slop!
-        unless decorators(XML::Node).include? Nokogiri::Decorators::Slop
+        unless decorators(XML::Node).include?(Nokogiri::Decorators::Slop)
           decorators(XML::Node) << Nokogiri::Decorators::Slop
           decorate!
         end
@@ -212,16 +406,17 @@ module Nokogiri
 
       ##
       # Apply any decorators to +node+
-      def decorate node
+      def decorate(node)
         return unless @decorators
-        @decorators.each { |klass,list|
+
+        @decorators.each do |klass, list|
           next unless node.is_a?(klass)
-          list.each { |moodule| node.extend(moodule) }
-        }
+
+          list.each { |mod| node.extend(mod) }
+        end
       end
 
-      alias :to_xml :serialize
-      alias :clone :dup
+      alias_method :to_xml, :serialize
 
       # Get the hash of namespaces on the root Nokogiri::XML::Node
       def namespaces
@@ -231,51 +426,85 @@ module Nokogiri
       ##
       # Create a Nokogiri::XML::DocumentFragment from +tags+
       # Returns an empty fragment if +tags+ is nil.
-      def fragment tags = nil
-        DocumentFragment.new(self, tags, self.root)
+      def fragment(tags = nil)
+        DocumentFragment.new(self, tags, root)
       end
 
       undef_method :swap, :parent, :namespace, :default_namespace=
       undef_method :add_namespace_definition, :attributes
       undef_method :namespace_definitions, :line, :add_namespace
 
-      def add_child node_or_tags
-        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?)
+      def add_child(node_or_tags)
+        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 "A document may not have multiple root nodes." if node_or_tags.size > 1
+
           super(node_or_tags.first)
         else
           super
         end
       end
-      alias :<< :add_child
+      alias_method :<<, :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)
+      # :call-seq:
+      #   xpath_doctype() → Nokogiri::CSS::XPathVisitor::DoctypeConfig
+      #
+      # [Returns] The document type which determines CSS-to-XPath translation.
+      #
+      # See XPathVisitor for more information.
+      def xpath_doctype
+        Nokogiri::CSS::XPathVisitor::DoctypeConfig::XML
       end
 
-      ##
-      # +JRuby+
-      # Returns Java's org.w3c.dom.document of this Document.
-      def to_java
-        raise "JRuby only method" unless Nokogiri.jruby?
-        return toJavaDocument()
+      #
+      #  :call-seq: deconstruct_keys(array_of_names) → Hash
+      #
+      #  Returns a hash describing the Document, to use in pattern matching.
+      #
+      #  Valid keys and their values:
+      #  - +root+ → (Node, nil) The root node of the Document, or +nil+ if the document is empty.
+      #
+      #  In the future, other keys may allow accessing things like doctype and processing
+      #  instructions. If you have a use case and would like this functionality, please let us know
+      #  by opening an issue or a discussion on the github project.
+      #
+      #  *Example*
+      #
+      #    doc = Nokogiri::XML.parse(<<~XML)
+      #      <?xml version="1.0"?>
+      #      <root>
+      #        <child>
+      #      </root>
+      #    XML
+      #
+      #    doc.deconstruct_keys([:root])
+      #    # => {:root=>
+      #    #      #(Element:0x35c {
+      #    #        name = "root",
+      #    #        children = [
+      #    #          #(Text "\n" + "  "),
+      #    #          #(Element:0x370 { name = "child", children = [ #(Text "\n")] }),
+      #    #          #(Text "\n")]
+      #    #        })}
+      #
+      #  *Example* of an empty document
+      #
+      #    doc = Nokogiri::XML::Document.new
+      #
+      #    doc.deconstruct_keys([:root])
+      #    # => {:root=>nil}
+      #
+      #  Since v1.14.0
+      #
+      def deconstruct_keys(keys)
+        { root: root }
       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:
+      IMPLIED_XPATH_CONTEXTS = ["//"].freeze # :nodoc:
 
       def inspect_attributes
         [:name, :children]