Minor S5 tweaks and Sync with Latest HTML5lib
This commit is contained in:
Jacques Distler
parent
dbed460843
commit
81d3cdc8e4
81 changed files with 9887 additions and 1687 deletions
|
|
@ -3,153 +3,151 @@ module HTML5
|
|||
module TreeWalkers
|
||||
|
||||
module TokenConstructor
|
||||
def error(msg)
|
||||
return {:type => "SerializeError", :data => msg}
|
||||
def error(msg)
|
||||
{:type => "SerializeError", :data => msg}
|
||||
end
|
||||
|
||||
def normalize_attrs(attrs)
|
||||
attrs.to_a
|
||||
end
|
||||
|
||||
def empty_tag(name, attrs, has_children=false)
|
||||
error(_("Void element has children")) if has_children
|
||||
{:type => :EmptyTag, :name => name, :data => normalize_attrs(attrs)}
|
||||
end
|
||||
|
||||
def start_tag(name, attrs)
|
||||
{:type => :StartTag, :name => name, :data => normalize_attrs(attrs)}
|
||||
end
|
||||
|
||||
def end_tag(name)
|
||||
{:type => :EndTag, :name => name, :data => []}
|
||||
end
|
||||
|
||||
def text(data)
|
||||
if data =~ /\A([#{SPACE_CHARACTERS.join('')}]+)/m
|
||||
yield({:type => :SpaceCharacters, :data => $1})
|
||||
data = data[$1.length .. -1]
|
||||
return if data.empty?
|
||||
end
|
||||
|
||||
def normalizeAttrs(attrs)
|
||||
attrs.to_a
|
||||
if data =~ /([#{SPACE_CHARACTERS.join('')}]+)\Z/m
|
||||
yield({:type => :Characters, :data => data[0 ... -$1.length]})
|
||||
yield({:type => :SpaceCharacters, :data => $1})
|
||||
else
|
||||
yield({:type => :Characters, :data => data})
|
||||
end
|
||||
end
|
||||
|
||||
def emptyTag(name, attrs, hasChildren=false)
|
||||
error(_("Void element has children")) if hasChildren
|
||||
return({:type => :EmptyTag, :name => name, \
|
||||
:data => normalizeAttrs(attrs)})
|
||||
end
|
||||
def comment(data)
|
||||
{:type => :Comment, :data => data}
|
||||
end
|
||||
|
||||
def startTag(name, attrs)
|
||||
return {:type => :StartTag, :name => name, \
|
||||
:data => normalizeAttrs(attrs)}
|
||||
end
|
||||
def doctype(name, public_id, system_id, correct=nil)
|
||||
{:type => :Doctype, :name => name, :public_id => public_id, :system_id => system_id, :correct => correct}
|
||||
end
|
||||
|
||||
def endTag(name)
|
||||
return {:type => :EndTag, :name => name, :data => []}
|
||||
end
|
||||
def unknown(nodeType)
|
||||
error(_("Unknown node type: ") + nodeType.to_s)
|
||||
end
|
||||
|
||||
def text(data)
|
||||
if data =~ /\A([#{SPACE_CHARACTERS.join('')}]+)/m
|
||||
yield({:type => :SpaceCharacters, :data => $1})
|
||||
data = data[$1.length .. -1]
|
||||
return if data.empty?
|
||||
end
|
||||
|
||||
if data =~ /([#{SPACE_CHARACTERS.join('')}]+)\Z/m
|
||||
yield({:type => :Characters, :data => data[0 ... -$1.length]})
|
||||
yield({:type => :SpaceCharacters, :data => $1})
|
||||
else
|
||||
yield({:type => :Characters, :data => data})
|
||||
end
|
||||
end
|
||||
|
||||
def comment(data)
|
||||
return {:type => :Comment, :data => data}
|
||||
end
|
||||
|
||||
def doctype(name)
|
||||
return {:type => :Doctype, :name => name, :data => name.upcase() == "HTML"}
|
||||
end
|
||||
|
||||
def unknown(nodeType)
|
||||
return error(_("Unknown node type: ") + nodeType.to_s)
|
||||
end
|
||||
|
||||
def _(str)
|
||||
str
|
||||
end
|
||||
def _(str)
|
||||
str
|
||||
end
|
||||
end
|
||||
|
||||
class Base
|
||||
include TokenConstructor
|
||||
|
||||
def initialize(tree)
|
||||
@tree = tree
|
||||
@tree = tree
|
||||
end
|
||||
|
||||
def each
|
||||
raise NotImplementedError
|
||||
raise NotImplementedError
|
||||
end
|
||||
|
||||
alias walk each
|
||||
end
|
||||
|
||||
class NonRecursiveTreeWalker < TreeWalkers::Base
|
||||
def node_details(node)
|
||||
raise NotImplementedError
|
||||
end
|
||||
def node_details(node)
|
||||
raise NotImplementedError
|
||||
end
|
||||
|
||||
def first_child(node)
|
||||
raise NotImplementedError
|
||||
end
|
||||
def first_child(node)
|
||||
raise NotImplementedError
|
||||
end
|
||||
|
||||
def next_sibling(node)
|
||||
raise NotImplementedError
|
||||
end
|
||||
def next_sibling(node)
|
||||
raise NotImplementedError
|
||||
end
|
||||
|
||||
def parent(node)
|
||||
raise NotImplementedError
|
||||
end
|
||||
def parent(node)
|
||||
raise NotImplementedError
|
||||
end
|
||||
|
||||
def each
|
||||
currentNode = @tree
|
||||
while currentNode != nil
|
||||
details = node_details(currentNode)
|
||||
hasChildren = false
|
||||
def each
|
||||
current_node = @tree
|
||||
while current_node != nil
|
||||
details = node_details(current_node)
|
||||
has_children = false
|
||||
|
||||
case details.shift
|
||||
when :DOCTYPE
|
||||
yield doctype(*details)
|
||||
case details.shift
|
||||
when :DOCTYPE
|
||||
yield doctype(*details)
|
||||
|
||||
when :TEXT
|
||||
text(*details) {|token| yield token}
|
||||
when :TEXT
|
||||
text(*details) {|token| yield token}
|
||||
|
||||
when :ELEMENT
|
||||
name, attributes, hasChildren = details
|
||||
if VOID_ELEMENTS.include?(name)
|
||||
yield emptyTag(name, attributes.to_a, hasChildren)
|
||||
hasChildren = false
|
||||
else
|
||||
yield startTag(name, attributes.to_a)
|
||||
end
|
||||
|
||||
when :COMMENT
|
||||
yield comment(details[0])
|
||||
|
||||
when :DOCUMENT, :DOCUMENT_FRAGMENT
|
||||
hasChildren = true
|
||||
|
||||
when nil
|
||||
# ignore (REXML::XMLDecl is an example)
|
||||
|
||||
else
|
||||
yield unknown(details[0])
|
||||
end
|
||||
|
||||
firstChild = hasChildren ? first_child(currentNode) : nil
|
||||
if firstChild != nil
|
||||
currentNode = firstChild
|
||||
else
|
||||
while currentNode != nil
|
||||
details = node_details(currentNode)
|
||||
if details.shift == :ELEMENT
|
||||
name, attributes, hasChildren = details
|
||||
yield endTag(name) if !VOID_ELEMENTS.include?(name)
|
||||
end
|
||||
|
||||
if @tree == currentNode
|
||||
currentNode = nil
|
||||
else
|
||||
nextSibling = next_sibling(currentNode)
|
||||
if nextSibling != nil
|
||||
currentNode = nextSibling
|
||||
break
|
||||
end
|
||||
|
||||
currentNode = parent(currentNode)
|
||||
end
|
||||
end
|
||||
end
|
||||
when :ELEMENT
|
||||
name, attributes, has_children = details
|
||||
if VOID_ELEMENTS.include?(name)
|
||||
yield empty_tag(name, attributes.to_a, has_children)
|
||||
has_children = false
|
||||
else
|
||||
yield start_tag(name, attributes.to_a)
|
||||
end
|
||||
|
||||
when :COMMENT
|
||||
yield comment(details[0])
|
||||
|
||||
when :DOCUMENT, :DOCUMENT_FRAGMENT
|
||||
has_children = true
|
||||
|
||||
when nil
|
||||
# ignore (REXML::XMLDecl is an example)
|
||||
|
||||
else
|
||||
yield unknown(details[0])
|
||||
end
|
||||
|
||||
first_child = has_children ? first_child(current_node) : nil
|
||||
if first_child != nil
|
||||
current_node = first_child
|
||||
else
|
||||
while current_node != nil
|
||||
details = node_details(current_node)
|
||||
if details.shift == :ELEMENT
|
||||
name, attributes, has_children = details
|
||||
yield end_tag(name) if !VOID_ELEMENTS.include?(name)
|
||||
end
|
||||
|
||||
if @tree == current_node
|
||||
current_node = nil
|
||||
else
|
||||
next_sibling = next_sibling(current_node)
|
||||
if next_sibling != nil
|
||||
current_node = next_sibling
|
||||
break
|
||||
end
|
||||
|
||||
current_node = parent(current_node)
|
||||
end
|
||||
end
|
||||
end
|
||||
end
|
||||
end
|
||||
end
|
||||
|
||||
end
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue