Class: XMLScan::XMLParser

Inherits:

XMLScanner

• Object
• XMLScanner
• XMLScan::XMLParser

Defined in:

lib/xmlscan/parser.rb

Direct Known Subclasses

XMLParserNS

Defined Under Namespace

Classes: AttributeChecker

Constant Summary

Constants included from XMLScan::XMLScanner::OptRegexp

XMLScan::XMLScanner::OptRegexp::RE_ENCODINGS, XMLScan::XMLScanner::OptRegexp::RE_ENCODING_OPTIONS, XMLScan::XMLScanner::OptRegexp::UTFSTR

Instance Attribute Summary

Attributes inherited from XMLScanner

#optkey

Instance Method Summary

• #parse ⇒ Object
• #scan_content(s) ⇒ Object

Methods inherited from XMLScanner

apply_option, apply_options, #initialize, #lineno, new, #opt_encoding, #parse_document, #path, provided_options, #source

Constructor Details

This class inherits a constructor from XMLScan::XMLScanner

Instance Method Details

#parse ⇒ Object

# File 'lib/xmlscan/parser.rb', line 36

def parse(*)
  @elem = []
  @attr = AttributeChecker.new
  @standalone = nil
  super
end

#scan_content(s) ⇒ Object

# File 'lib/xmlscan/parser.rb', line 163

def scan_content(s)
  elem = @elem  # for speed
  src = @src  # for speed
  found_root_element = false

  begin

    # -- first start tag --
    elem.clear
    found_stag = false

    while s and not found_stag
      if (c = s[0]) == ?< then
        if (c = s[1]) == ?/ then
          # should be a parse error
          scan_etag s
        elsif c == ?! then
          if s[2] == ?- and s[3] == ?- then
            scan_comment s
          elsif /\A<!\[CDATA\[/n =~ s then
            parse_error "CDATA section is found outside of root element"
            scan_cdata $'
          else
            scan_bang_tag s
          end
        elsif c == ?? then
          scan_pi s
        else
          found_root_element = true
          found_stag = true
          scan_stag s
        end
      else
        parse_error "content of element is found outside of root element"
        scan_chardata s
      end
      s = src.get
    end

    if not found_root_element and not found_stag then
      parse_error "no root element was found"
    end

    # -- contents --
    while s and not elem.empty?
      if (c = s[0]) == ?< then
        if (c = s[1]) == ?/ then
          scan_etag s
        elsif c == ?! then
          if s[2] == ?- and s[3] == ?- then
            scan_comment s
          elsif /\A<!\[CDATA\[/n =~ s then
            scan_cdata $'
          else
            scan_bang_tag s
          end
        elsif c == ?? then
          scan_pi s
        else
          scan_stag s
        end
      else
        scan_chardata s
      end
      s = src.get
    end

    unless elem.empty? then
      while name = elem.pop
        parse_error "unclosed element `#{name}' meets EOF"
        @visitor.on_etag name
      end
    end

    # -- epilogue --
    finish = true

    while s
      if (c = s[0]) == ?< then
        if (c = s[1]) == ?/ then
          finish = false    # content out of root element
          break
        elsif c == ?! then
          if s[2] == ?- and s[3] == ?- then
            scan_comment s
          else
            finish = false  # content out of root element
            break
          end
        elsif c == ?? then
          scan_pi s
        else
          parse_error "another root element is found"  # stag
          finish = false
          break
        end
      else
        if s.strip.empty? then
          on_prolog_space s
        else
          finish = false    # content out of root element
          break
        end
      end
      s = src.get
    end

  end until finish

end