narf 0.5.1

data/lib/web/htmlparser/sgml-parser.rb~

@@ -0,0 +1,332 @@
+# A parser for SGML, using the derived class as static DTD.
+
+class SGMLParser
+
+  # Regular expressions used for parsing:
+  Interesting = /[&<]/
+  Incomplete = Regexp.compile('&([a-zA-Z][a-zA-Z0-9]*|#[0-9]*)?|' +
+                              '<([a-zA-Z][^<>]*|/([a-zA-Z][^<>]*)?|' +
+                              '![^<>]*)?')
+
+  Entityref = /&([a-zA-Z][-.a-zA-Z0-9]*)[^-.a-zA-Z0-9]/
+  Charref = /&#([0-9]+)[^0-9]/
+
+  Starttagopen = /<[>a-zA-Z]/
+  Endtagopen = /<\/[<>a-zA-Z]/
+  Endbracket = /[<>]/
+  Special = /<![^<>]*>/
+  Commentopen = /<!--/
+  Commentclose = /--[ \t\n]*>/
+  Tagfind = /[a-zA-Z][a-zA-Z0-9.-]*/
+  Attrfind = Regexp.compile('[\s,]*([a-zA-Z_][a-zA-Z_0-9.-]*)' +
+                            '(\s*=\s*' +
+                            "('[^']*'" +
+                            '|"[^"]*"' +
+                            '|[-~a-zA-Z0-9,./:+*%?!()_#=]*))?')
+
+  Entitydefs =
+    {'lt'=>'<', 'gt'=>'>', 'amp'=>'&', 'quot'=>'"', 'apos'=>'\''}
+
+  def initialize(verbose=false)
+    @verbose = verbose
+    reset
+  end
+
+  def reset
+    @rawdata = ''
+    @stack = []
+    @lasttag = '???'
+    @nomoretags = false
+    @literal = false
+  end
+
+  def has_context(gi)
+    @stack.include? gi
+  end
+
+  def setnomoretags
+    @nomoretags = true
+    @literal = true
+  end
+
+  def setliteral(*args)
+    @literal = true
+  end
+
+  def feed(data)
+    @rawdata << data
+    goahead(false)
+  end
+
+  def close
+    goahead(true)
+  end
+
+  def goahead(_end)
+    rawdata = @rawdata
+    i = 0
+    n = rawdata.length
+    while i < n
+      if @nomoretags
+        handle_data(rawdata[i..(n-1)])
+        i = n
+        break
+      end
+      j = rawdata.index(Interesting, i)
+      j = n unless j
+      if i < j
+        handle_data(rawdata[i..(j-1)])
+      end
+      i = j
+      break if (i == n)
+      if rawdata[i] == ?< #
+        if rawdata.index(Starttagopen, i) == i
+          if @literal
+            handle_data(rawdata[i, 1])
+            i += 1
+            next
+          end
+          k = parse_starttag(i)
+          break unless k
+          i = k
+          next
+        end
+        if rawdata.index(Endtagopen, i) == i
+          k = parse_endtag(i)
+          break unless k
+          i = k
+          @literal = false
+          next
+        end
+        if rawdata.index(Commentopen, i) == i
+          if @literal
+            handle_data(rawdata[i,1])
+            i += 1
+            next
+          end
+          k = parse_comment(i)
+          break unless k
+          i += k
+          next
+        end
+        if rawdata.index(Special, i) == i
+          if @literal
+            handle_data(rawdata[i, 1])
+            i += 1
+            next
+          end
+          k = parse_special(i)
+          break unless k
+          i += k
+          next
+        end
+      elsif rawdata[i] == ?& #
+        if rawdata.index(Charref, i) == i
+          i += $&.length
+          handle_charref($1)
+          i -= 1 unless rawdata[i-1] == ?;
+          next
+        end
+        if rawdata.index(Entityref, i) == i
+          i += $&.length
+          handle_entityref($1)
+          i -= 1 unless rawdata[i-1] == ?;
+          next
+        end
+      else
+        raise RuntimeError, 'neither < nor & ??'
+      end
+      # We get here only if incomplete matches but
+      # nothing else
+      match = rawdata.index(Incomplete, i)
+      unless match == i
+        handle_data(rawdata[i, 1])
+        i += 1
+        next
+      end
+      j = match + $&.length
+      break if j == n # Really incomplete
+      handle_data(rawdata[i..(j-1)])
+      i = j
+    end
+    # end while
+    if _end and i < n
+      handle_data(@rawdata[i..(n-1)])
+      i = n
+    end
+    @rawdata = rawdata[i..-1]
+  end
+
+  def parse_comment(i)
+    rawdata = @rawdata
+    if rawdata[i, 4] != '<!--'
+      raise RuntimeError, 'unexpected call to handle_comment'
+    end
+    match = rawdata.index(Commentclose, i)
+    return nil unless match
+    matched_length = $&.length
+    j = match
+    handle_comment(rawdata[i+4..(j-1)])
+    j = match + matched_length
+    return j-i
+  end
+
+  def parse_starttag(i)
+    rawdata = @rawdata
+    j = rawdata.index(Endbracket, i + 1)
+    return nil unless j
+    attrs = []
+    if rawdata[i+1] == ?> #
+      # SGML shorthand: <> == <last open tag seen>
+      k = j
+      tag = @lasttag
+    else
+      match = rawdata.index(Tagfind, i + 1)
+      unless match
+        raise RuntimeError, 'unexpected call to parse_starttag'
+      end
+      k = i + 1 + ($&.length)
+      tag = $&.downcase
+      @lasttag = tag
+    end
+    while k < j
+      break unless rawdata.index(Attrfind, k)
+      matched_length = $&.length
+      attrname, rest, attrvalue = $1, $2, $3
+      if not rest
+        attrvalue = '' # was: = attrname
+      elsif (attrvalue[0] == ?' && attrvalue[-1] == ?') or
+          (attrvalue[0] == ?" && attrvalue[-1] == ?")
+        attrvalue = attrvalue[1..-2]
+      end
+      attrs << [attrname.downcase, attrvalue]
+      k += matched_length
+    end
+    if rawdata[j] == ?> #
+      j += 1
+    end
+    finish_starttag(tag, attrs)
+    return j
+  end
+
+  def parse_endtag(i)
+    rawdata = @rawdata
+    j = rawdata.index(Endbracket, i + 1)
+    return nil unless j
+    tag = (rawdata[i+2..j-1].strip).downcase
+    if rawdata[j] == ?> #
+      j += 1
+    end
+    finish_endtag(tag)
+    return j
+  end
+
+  def finish_starttag(tag, attrs)
+    method = 'start_' + tag
+    if self.respond_to?(method)
+      @stack << tag
+      handle_starttag(tag, method, attrs)
+      return 1
+    else
+      method = 'do_' + tag
+      if self.respond_to?(method)
+        handle_starttag(tag, method, attrs)
+        return 0
+      else
+        unknown_starttag(tag, attrs)
+        return -1
+      end
+    end
+  end
+
+  def finish_endtag(tag)
+    if tag == ''
+      found = @stack.length - 1
+      if found < 0
+        unknown_endtag(tag)
+        return
+      end
+    else
+      unless @stack.include? tag
+        method = 'end_' + tag
+        unless self.respond_to?(method)
+          unknown_endtag(tag)
+        end
+        return
+      end
+      found = @stack.index(tag) #or @stack.length
+    end
+    while @stack.length > found
+      tag = @stack[-1]
+      method = 'end_' + tag
+      if respond_to?(method)
+        handle_endtag(tag, method)
+      else
+        unknown_endtag(tag)
+      end
+      @stack.pop
+    end
+  end
+
+  def parse_special(i)
+    rawdata = @rawdata
+    match = rawdata.index(Endbracket, i+1)
+    return nil unless match
+    matched_length = $&.length
+    handle_special(rawdata[i+1..(match-1)])
+    return match - i + matched_length
+  end
+
+  def handle_starttag(tag, method, attrs)
+    self.send(method, attrs)
+  end
+
+  def handle_endtag(tag, method)
+    self.send(method)
+  end
+
+  def report_unbalanced(tag)
+    if @verbose
+      print '*** Unbalanced </' + tag + '>', "\n"
+      print '*** Stack:', self.stack, "\n"
+    end
+  end
+
+  def handle_charref(name)
+    n = Integer(name)
+    if !(0 <= n && n <= 255)
+      unknown_charref(name)
+      return
+    end
+    handle_data(n.chr)
+  end
+
+  def handle_entityref(name)
+    table = Entitydefs
+    if table.include?(name)
+      handle_data(table[name])
+    else
+      unknown_entityref(name)
+      return
+    end
+  end
+
+  def handle_data(data)
+  end
+
+  def handle_comment(data)
+  end
+
+  def handle_special(data)
+  end
+
+  def unknown_starttag(tag, attrs)
+  end
+  def unknown_endtag(tag)
+  end
+  def unknown_charref(ref)
+  end
+  def unknown_entityref(ref)
+  end
+
+end

data/lib/web/htmltools/element.rb

@@ -0,0 +1,296 @@
+# Copyright::   Copyright (C) 2002, Ned Konz <ned@bike-nomad.com>
+# License::     Same as Ruby's
+# CVS ID:       $Id: element.rb,v 1.8 2002/06/04 01:55:59 ned Exp $
+
+# This module is a mix-in that provides parent/child behavior to real
+# Element classes. Because it defines <tt>each()</tt> and includes Enumerable,
+# you can iterate through a tree using the usual Enumerable methods.
+
+require 'web/htmltools/tags'
+
+module HTMLTree #:nodoc: all
+  module TreeElement
+    include Enumerable
+
+    protected
+
+    def initialize_tree_element(parent_or_nil = nil, contents_or_nil = nil)
+      @_content, @_parent = contents_or_nil, parent_or_nil
+      if parent_or_nil
+        parent_or_nil.add_child(self)
+      end
+    end
+
+    attr_accessor :_parent
+
+    public
+
+    # Add one or more children to this node.
+    def add_child(*children_to_add)
+      if can_have_children?
+        children_to_add.each do |child|
+          @_content << child
+          child._parent = self
+        end
+      else
+        raise(ArgumentError.exception('node cannot have children'))
+      end
+    end
+
+    alias_method(:add_children, :add_child)
+
+    # Remove one or more children from this node.
+    def remove_child(*children_to_remove)
+      if can_have_children?
+        children_to_remove.each do |child|
+          child._parent = nil if @_content.delete(child)
+        end
+      else
+        raise(ArgumentError.exception('node cannot have children'))
+      end
+    end
+
+    alias_method(:remove_children, :remove_child)
+
+    # Change my parent. Disconnects from prior parent, if any.
+    def parent=(parent_or_nil)
+      @_parent.remove_child(self) if @_parent
+      parent_or_nil.add_child(self) if parent_or_nil
+    end
+
+    # Return true if my content is a collection of Elements
+    # rather than actual data.
+    def can_have_children?
+      @_content.kind_of?(Array)
+    end
+
+    # Return a collection of my children. Returns an empty Array if I am a
+    # data element, just to keep other methods simple.
+    def children
+      can_have_children? ? @_content : []
+    end
+
+    # Return my content; either my children or my data.
+    def content
+      @_content
+    end
+
+    # Return my parent element.
+    def parent
+      @_parent
+    end
+
+    # Return the ultimate parent.
+    def root
+      @_parent ? self : @_parent.root
+    end
+
+    # Return true if I have any children.
+    def has_children?
+      children.size > 0
+    end
+
+    # Breadth-first iterator, required by Enumerable.
+    def each(&block)
+      block.call(self)
+      children.each { |ch| ch.each(&block) }
+    end
+
+    # Print out to $stdout (or given IO or String)
+    # a formatted dump of my structure.
+    def dump(indent=0, io=$stdout)
+      io << "  " * indent
+      io << self.to_s
+      io << "\n"
+      children.each { |ea| ea.dump(indent+1, io) }
+    end
+
+  end
+
+  # This is a Element that represents the whole document (and makes a
+  # scope for the DTD declaration)
+  class Document
+    include TreeElement
+
+    def initialize
+      initialize_tree_element(nil, [])
+    end
+
+    def to_s
+      ''
+    end
+
+    def each(&block)
+      children.each { |ch| ch.each(&block) }
+    end
+
+    def write(io)
+      children.each { |t| t.write(io) }
+    end
+
+    def tag
+      ''
+    end
+
+    # Return my child <html> node, if any.
+    def html_node
+      children.detect { |ea| ea.tag == 'html' }
+    end
+  end
+
+  # This is a TreeElement that represents tagged items in an HTML
+  # document.
+  class Element
+    include TreeElement
+
+    protected
+
+    # parent_or_nil::   TreeElement or nil
+    # tag_name::        String
+    def initialize(parent_or_nil = nil, tag_name = nil)
+      initialize_tree_element(parent_or_nil, [])
+      @_tag = tag_name
+      @_attributes = {}
+      @_attribute_order = []
+    end
+
+    public
+
+    def can_have_children?; true; end
+
+    # Return true if I'm data instead of a tag
+    def data?; false; end
+
+    def to_s
+      a = [ "<", tag ]
+      @_attribute_order.each { |k|
+        v = @_attributes[k]
+        a << " #{k.to_s}=\"#{v.to_s}\""
+      }
+      a << ">"
+      a.join('')
+    end
+
+    # Append an attribute. <tt>values</tt> are first flattened into an Array,
+    # then converted into strings.
+    #
+    # If there is a single attribute value, it will appear as a String,
+    # otherwise it will be an Array of Strings.
+    #
+    # Example:
+    #   element.add_attribute("width", "123")
+    #   element.add_attribute("value", [ "a", "b" ])
+    def add_attribute(name, *values)
+      values = values.flatten.collect { |ea| ea.to_s.strip }
+      name = name.downcase
+      if @_attributes.include?(name)
+        @_attributes[name] = @_attributes[name].to_a + values
+      else
+        @_attributes[name] = values.size > 1 ? values : values[0]
+      end
+      @_attribute_order << name
+      self
+    end
+
+    # Return my tag (should be a String)
+    def tag; @_tag; end
+
+    # Return an HTML::Tag for further information, or nil if this is an
+    # unknown tag.
+    def tag_info
+      begin
+        HTML::Tag.named(@_tag)
+      rescue NoSuchHTMLTagError
+        nil
+      end
+    end
+
+    # Return my attributes Hash.
+    def attributes; @_attributes; end
+
+    # Return the order of my attributes
+    def attribute_order; @_attribute_order; end
+
+    # Return the value of a single attribute (a String or Array).
+    def attribute(name); @_attributes[name]; end
+
+    # Return the value of a single attribute (a String or Array).
+    def [](name); attribute(name); end
+
+    # Replace an attribute.
+    def []=(name, *values)
+      @_attributes[name] = values.size > 1 ? values : values[0]
+      @_attribute_order.delete(name)
+      self
+    end
+
+    # Print me (and my descendents) on the given IO stream.
+    def write(io)
+      io << self
+      children.each { |t| t.write(io) }
+      unless tag_info.is_empty_element
+        io.puts( "</#{tag()}>" )
+      end
+    end
+
+  end
+
+  # This is a TreeElement that represents leaf data nodes (CDATA, scripts,
+  # comments, processing directives). It forwards unknown messages to the
+  # content element, so it otherwise behaves like a String.
+  class Data
+    include TreeElement
+
+    protected 
+
+    # parent_or_nil:: parent, TreeElement or nil
+    # str:: contents, String
+    def initialize(parent_or_nil = nil, str = '')
+      initialize_tree_element(parent_or_nil, str)
+    end
+
+    public
+
+    # Return true because I am a data Element.
+    def data?; true; end
+
+    # Return false because I have no children.
+    def can_have_children?; false; end
+
+    # Return an empty collection because I have no children.
+    def children; []; end
+
+    # Return my (empty) tag String.
+    def tag; ''; end
+
+    # Return my (empty) attributes Hash.
+    def attributes; {}; end
+
+    def to_s
+      @_content
+    end
+
+    # Print me on the given IO stream.
+    def write(io)
+      io << self
+    end
+
+    # Forward all other methods to my content, so I can otherwise behave
+    # like a String.
+    def method_missing(sym, *args)
+      @_content.method(sym).call(*args)
+    end
+  end
+
+  class Comment < Data
+    def to_s
+      '<!--' + @_content + '-->'
+    end
+  end
+
+  class Special < Data
+    def to_s
+      '<' + @_content + '>'
+    end
+  end
+end