hpricot 0.6.164 → 0.7

data/lib/hpricot/tag.rb

@@ -2,66 +2,55 @@ module Hpricot
   # :stopdoc:
 
   class Doc
-    attr_accessor :children
-    def initialize(children = [], options = {})
-      @children = children ? children.each { |c| c.parent = self }  : []
-      @options = options
-    end
     def output(out, opts = {})
-      @children.each do |n|
+      children.each do |n|
         n.output(out, opts)
-      end
+      end if children
       out
     end
     def make(input = nil, &blk)
-      Hpricot.make(input, @options, &blk)
+      Hpricot.make(input, @options, &blk).children
     end
     def altered!; end
+    def inspect_tree
+      children.map { |x| x.inspect_tree }.join if children
+    end
   end
 
   class BaseEle
-    attr_accessor :raw_string, :parent
     def html_quote(str)
       "\"" + str.gsub('"', '\\"') + "\""
     end
     def if_output(opts)
-      if opts[:preserve] and not @raw_string.nil?
-        @raw_string
+      if opts[:preserve] and not raw_string.nil?
+        raw_string
       else
         yield opts
       end
     end
     def pathname; self.name end
     def altered!
-      @raw_string = nil
-    end
-    def self.alterable(*fields)
-      attr_accessor(*fields)
-      fields.each do |f|
-        define_method("#{f}=") do |v|
-          altered!
-          instance_variable_set("@#{f}", v)
-        end
-      end
+      clear_raw
+    end
+    def inspect_tree(depth = 0)
+      %{#{" " * depth}} + self.class.name.split(/::/).last.downcase + "\n"
     end
   end
 
   class Elem
-    attr_accessor :stag, :etag, :children
-    def initialize(stag, children=nil, etag=nil)
-      @stag, @etag = stag, etag
-      @children = children ? children.each { |c| c.parent = self }  : []
-    end
-    def empty?; @children.empty? end
-    [:name, :raw_attributes, :parent, :altered!].each do |m|
-      [m, "#{m}="].each { |m2| define_method(m2) { |*a| [@etag, @stag].inject { |_,t| t.send(m2, *a) if t and t.respond_to?(m2) } } }
+    def initialize tag, attrs = nil, children = nil, etag = nil
+      self.name, self.raw_attributes, self.children, self.etag =
+        tag, attrs, children, etag
     end
+    def empty?; children.nil? or children.empty? end
     def attributes
       if raw_attributes
         raw_attributes.inject({}) do |hsh, (k, v)|
           hsh[k] = Hpricot.uxs(v)
           hsh
         end
+      else
+        {}
       end
     end
     def to_plain_text
@@ -79,151 +68,128 @@ module Hpricot
     end
     def pathname; self.name end
     def output(out, opts = {})
-      if empty? and ElementContent[@stag.name] == :EMPTY
-        @stag.output(out, opts.merge(:style => :empty))
-      else
-        @stag.output(out, opts)
-        @children.each { |n| n.output(out, opts) }
-        if @etag
-          @etag.output(out, opts)
-        elsif !opts[:preserve]
-          ETag.new(@stag.name).output(out, opts)
+      out <<
+        if_output(opts) do
+          "<#{name}#{attributes_as_html}" +
+            ((empty? and not etag) ? " /" : "") +
+            ">"
         end
+      if children
+        children.each { |n| n.output(out, opts) }
+      end
+      if etag
+        etag.output(out, opts)
+      elsif !opts[:preserve] && !empty?
+        out <<
+          if_output(opts) do
+            "</#{name}>"
+          end
       end
       out
     end
-  end
-
-  class STag < BaseEle
-    def initialize(name, attributes=nil)
-      @name = name.to_s
-      @raw_attributes = attributes || {}
-    end
-    alterable :name, :raw_attributes
     def attributes_as_html
-      if @raw_attributes
-        @raw_attributes.map do |aname, aval|
+      if raw_attributes
+        raw_attributes.map do |aname, aval|
           " #{aname}" +
             (aval ? "=#{html_quote aval}" : "")
         end.join
       end
     end
-    def output(out, opts = {})
-      out <<
-        if_output(opts) do
-          "<#{@name}#{attributes_as_html}" +
-            (opts[:style] == :empty ? " /" : "") +
-            ">"
-        end
+    def inspect_tree(depth = 0)
+      %{#{" " * depth}} + name + "\n" +
+        (children ? children.map { |x| x.inspect_tree(depth + 1) }.join : "")
     end
   end
 
-  class ETag < BaseEle
-    def initialize(qualified_name)
-      @name = qualified_name.to_s
-    end
-    alterable :name
+  class ETag
+    def initialize name; self.name = name end
     def output(out, opts = {})
       out <<
         if_output(opts) do
-          "</#{@name}>"
+          "</#{name}>"
         end
     end
   end
 
-  class BogusETag < ETag
+  class BogusETag
     def output(out, opts = {}); out << if_output(opts) { '' }; end
   end
 
-  class Text < BaseEle
-    def initialize(text)
-      @content = text
-    end
-    alterable :content
+  class Text
+    def initialize content; self.content = content end
     def pathname; "text()" end
     def to_s
-      Hpricot.uxs(@content)
+      Hpricot.uxs(content)
     end
     alias_method :inner_text, :to_s
     alias_method :to_plain_text, :to_s
+    def << str; self.content << str end
     def output(out, opts = {})
       out <<
         if_output(opts) do
-          @content
+          content.to_s
         end
     end
   end
 
-  class CData < Text
+  class CData
+    def initialize content; self.content = content end
     alias_method :to_s, :content
     alias_method :to_plain_text, :content
     def output(out, opts = {})
       out <<
         if_output(opts) do
-          "<![CDATA[#@content]]>"
+          "<![CDATA[#{content}]]>"
         end
     end
   end
 
-  class XMLDecl < BaseEle
-    def initialize(version, encoding, standalone)
-      @version, @encoding, @standalone = version, encoding, standalone
-    end
-    alterable :version, :encoding, :standalone
+  class XMLDecl
     def pathname; "xmldecl()" end
     def output(out, opts = {})
       out <<
         if_output(opts) do
-          "<?xml version=\"#{@version}\"" +
-            (@encoding ? " encoding=\"#{encoding}\"" : "") +
-            (@standalone != nil ? " standalone=\"#{standalone ? 'yes' : 'no'}\"" : "") +
+          "<?xml version=\"#{version}\"" +
+            (encoding ? " encoding=\"#{encoding}\"" : "") +
+            (standalone != nil ? " standalone=\"#{standalone ? 'yes' : 'no'}\"" : "") +
             "?>"
         end
     end
   end
 
-  class DocType < BaseEle
-    def initialize(target, pubid, sysid)
-      @target, @public_id, @system_id = target, pubid, sysid
+  class DocType
+    def initialize target, pub, sys
+      self.target, self.public_id, self.system_id = target, pub, sys
     end
-    alterable :target, :public_id, :system_id
     def pathname; "doctype()" end
     def output(out, opts = {})
       out <<
         if_output(opts) do
-          "<!DOCTYPE #{@target} " +
-            (@public_id ? "PUBLIC \"#{@public_id}\"" : "SYSTEM") +
-            (@system_id ? " #{html_quote(@system_id)}" : "") + ">"
+          "<!DOCTYPE #{target} " +
+            (public_id ? "PUBLIC \"#{public_id}\"" : "SYSTEM") +
+            (system_id ? " #{html_quote(system_id)}" : "") + ">"
         end
     end
   end
 
-  class ProcIns < BaseEle
-    def initialize(target, content)
-      @target, @content = target, content
-    end
+  class ProcIns
     def pathname; "procins()" end
-    alterable :target, :content
     def output(out, opts = {})
       out << 
         if_output(opts) do
-          "<?#{@target}" +
-           (@content ? " #{@content}" : "") +
+          "<?#{target}" +
+           (content ? " #{content}" : "") +
            "?>"
         end
     end
   end
 
-  class Comment < BaseEle
-    def initialize(content)
-      @content = content
-    end
+  class Comment
     def pathname; "comment()" end
-    alterable :content
     def output(out, opts = {})
       out <<
         if_output(opts) do
-          "<!--#{@content}-->"
+          "<!--#{content}-->"
         end
     end
   end

data/lib/hpricot/traverse.rb

@@ -26,7 +26,7 @@ module Hpricot
       if parent and parent.respond_to? :make
         parent.make(input, &blk)
       else
-        Hpricot.make(input, &blk)
+        Hpricot.make(input, &blk).children
       end
     end
 
@@ -51,7 +51,7 @@ module Hpricot
         return i if (x.respond_to?(:name) and name == x.name) or
           (x.text? and name == "text()")
         i += 1
-      end
+      end if children
       -1
     end
 
@@ -146,16 +146,20 @@ module Hpricot
     # Builds a string from the text contained in this node.  All
     # HTML elements are removed.
     def to_plain_text
-      if respond_to? :children
+      if respond_to?(:children) and children
         children.map { |x| x.to_plain_text }.join.strip.gsub(/\n{2,}/, "\n\n")
+      else
+        ""
       end
     end
 
     # Builds a string from the text contained in this node.  All
     # HTML elements are removed.
     def inner_text
-      if respond_to? :children
+      if respond_to?(:children) and children
         children.map { |x| x.inner_text }.join
+      else
+        ""
       end
     end
     alias_method :innerText, :inner_text
@@ -172,8 +176,10 @@ module Hpricot
         end
         reparent self.children
       else
-        if respond_to? :children
+        if respond_to?(:children) and children
           children.map { |x| x.output("") }.join
+        else
+          ""
         end
       end
     end
@@ -207,7 +213,7 @@ module Hpricot
         parent.children.each do |e|
           id = sim if e == self
           sim += 1 if e.pathname == self.pathname
-        end
+        end if parent.children
         p = File.join(parent.xpath, self.pathname)
         p += "[#{id+1}]" if sim >= 2
         p
@@ -224,7 +230,7 @@ module Hpricot
         parent.children.each do |e|
           id = sim if e == self
           sim += 1 if e.pathname == self.pathname
-        end
+        end if parent.children
         p = parent.css_path
         p = p ? "#{p} > #{self.pathname}" : self.pathname
         p += ":nth(#{id})" if sim >= 2
@@ -489,13 +495,13 @@ module Hpricot
 
     # +each_child+ iterates over each child.
     def each_child(&block) # :yields: child_node
-      children.each(&block)
+      children.each(&block) if children
       nil
     end
 
     # +each_child_with_index+ iterates over each child.
     def each_child_with_index(&block) # :yields: child_node, index
-      children.each_with_index(&block)
+      children.each_with_index(&block) if children
       nil
     end
 
@@ -626,7 +632,7 @@ module Hpricot
   # :stopdoc:
   module Doc::Trav
     def traverse_all_element(&block)
-      children.each {|c| c.traverse_all_element(&block) }
+      children.each {|c| c.traverse_all_element(&block) } if children
     end
     def xpath
       "/"
@@ -639,7 +645,7 @@ module Hpricot
   module Elem::Trav
     def traverse_all_element(&block)
       yield self
-      children.each {|c| c.traverse_all_element(&block) }
+      children.each {|c| c.traverse_all_element(&block) } if children
     end
   end
 
@@ -651,14 +657,14 @@ module Hpricot
 
   module Doc::Trav
     def traverse_some_element(name_set, &block)
-      children.each {|c| c.traverse_some_element(name_set, &block) }
+      children.each {|c| c.traverse_some_element(name_set, &block) } if children
     end
   end
 
   module Elem::Trav
     def traverse_some_element(name_set, &block)
       yield self if name_set.include? self.name
-      children.each {|c| c.traverse_some_element(name_set, &block) }
+      children.each {|c| c.traverse_some_element(name_set, &block) } if children
     end
   end
 
@@ -797,7 +803,7 @@ module Hpricot
   module Doc::Trav
     def root
       es = []
-      children.each {|c| es << c if c.elem? }
+      children.each {|c| es << c if c.elem? } if children
       raise Hpricot::Error, "no element" if es.empty?
       raise Hpricot::Error, "multiple top elements" if 1 < es.length
       es[0]

data/test/files/boingboing.html

@@ -2263,4 +2263,4 @@ Why are so many drawings from earlier centuries so deliciously weird? Here are a
 -->
     </div>
 </body>
-</html>
+</html>

data/test/nokogiri-bench.rb

@@ -0,0 +1,64 @@
+#!/usr/bin/env ruby
+require 'rubygems'
+require 'open-uri'
+require 'hpricot'
+require 'nokogiri'
+require 'benchmark'
+
+content = File.read("test/files/boingboing.html")
+
+N = 100
+
+unless Gem.loaded_specs['hpricot'].version > Gem::Version.new('0.6.161')
+  abort "** Use higher than Hpricot 0.6.161!"
+end
+
+puts "Hpricot #{Gem.loaded_specs['hpricot'].version} vs. Nokogiri #{Gem.loaded_specs['nokogiri'].version}"
+hdoc = Hpricot(content)
+ndoc = Nokogiri.Hpricot(content)
+
+Benchmark.bm do |x|
+  x.report('hpricot:doc') do
+    N.times do
+      Hpricot(content)
+    end
+  end
+
+  x.report('nokogiri:doc') do
+    N.times do
+      Nokogiri.Hpricot(content)
+    end
+  end
+end
+
+Benchmark.bm do |x|
+  x.report('hpricot:xpath') do
+    N.times do
+      info = hdoc.search("//a[@name='027906']").first.inner_text
+      url = hdoc.search("h3[text()='College kids reportedly taking more smart drugs']").first.inner_text
+    end
+  end
+
+  x.report('nokogiri:xpath') do
+    N.times do
+      info = ndoc.search("//a[@name='027906']").first.inner_text
+      url = ndoc.search("h3[text()='College kids reportedly taking more smart drugs']").first.inner_text
+    end
+  end
+end
+
+Benchmark.bm do |x|
+  x.report('hpricot:css') do
+    N.times do
+      info = hdoc.search('form input[@checked]').first
+      url = hdoc.search('td spacer').first.inner_text
+    end
+  end
+
+  x.report('nokogiri:css') do
+    N.times do
+      info = ndoc.search('form input[@checked]').first
+      url = ndoc.search('td spacer').first.inner_text
+    end
+  end
+end