hpricot 0.7-x86-mswin32

data/lib/hpricot/inspect.rb

@@ -0,0 +1,103 @@
+require 'pp'
+
+module Hpricot
+  # :stopdoc:
+  class Elements
+    def pretty_print(q)
+      q.object_group(self) { super }
+    end
+    alias inspect pretty_print_inspect
+  end
+
+  class Doc
+    def pretty_print(q)
+      q.object_group(self) { children.each {|elt| q.breakable; q.pp elt } }
+    end
+    alias inspect pretty_print_inspect
+  end
+
+  module Leaf
+    def pretty_print(q)
+      q.group(1, '{', '}') {
+        q.text self.class.name.sub(/.*::/,'').downcase
+        if rs = raw_string
+          rs.scan(/[^\r\n]*(?:\r\n?|\n|[^\r\n]\z)/) {|line|
+            q.breakable
+            q.pp line
+          }
+        elsif self.respond_to? :to_s
+          q.breakable
+          q.text self.to_s
+        end
+      }
+    end
+    alias inspect pretty_print_inspect
+  end
+
+  class Elem
+    def pretty_print(q)
+      if empty?
+        q.group(1, '{emptyelem', '}') {
+          q.breakable; pretty_print_stag q
+        }
+      else
+        q.group(1, "{elem", "}") {
+          q.breakable; pretty_print_stag q
+          if children
+            children.each {|elt| q.breakable; q.pp elt }
+          end
+          if etag
+            q.breakable; q.pp etag
+          end
+        }
+      end
+    end
+    def pretty_print_stag(q)
+      q.group(1, '<', '>') {
+        q.text name
+
+        if raw_attributes
+          raw_attributes.each {|n, t|
+            q.breakable
+            if t
+              q.text "#{n}=\"#{Hpricot.uxs(t)}\""
+            else
+              q.text n
+            end
+          }
+        end
+      }
+    end
+    alias inspect pretty_print_inspect
+  end
+
+  class ETag
+    def pretty_print(q)
+      q.group(1, '</', '>') {
+        q.text name
+      }
+    end
+    alias inspect pretty_print_inspect
+  end
+
+  class Text
+    def pretty_print(q)
+      q.text content.dump
+    end
+  end
+
+  class BogusETag
+    def pretty_print(q)
+      q.group(1, '{', '}') {
+        q.text self.class.name.sub(/.*::/,'').downcase
+        if rs = raw_string
+          q.breakable
+          q.text rs
+        else
+          q.text "</#{name}>"
+        end
+      }
+    end
+  end
+  # :startdoc:
+end

data/lib/hpricot/modules.rb

@@ -0,0 +1,38 @@
+module Hpricot
+  class Name; include Hpricot end
+  class Context; include Hpricot end
+
+  # :stopdoc:
+  module Tag; include Hpricot end
+    class ETag; include Tag end
+  # :startdoc:
+
+  module Node; include Hpricot end
+    module Container; include Node end
+      class Doc; include Container end
+      class Elem; include Container end
+
+    module Leaf; include Node end
+      class Text; include Leaf end
+      class XMLDecl; include Leaf end
+      class DocType; include Leaf end
+      class ProcIns; include Leaf end
+      class Comment; include Leaf end
+      class BogusETag; include Leaf end
+
+  module Traverse end
+  module Container::Trav; include Traverse end
+  module Leaf::Trav; include Traverse end
+  class Doc;       module Trav; include Container::Trav end; include Trav end
+  class Elem;      module Trav; include Container::Trav end; include Trav end
+  class CData;     module Trav; include Leaf::Trav      end; include Trav end
+  class Text;      module Trav; include Leaf::Trav      end; include Trav end
+  class XMLDecl;   module Trav; include Leaf::Trav      end; include Trav end
+  class DocType;   module Trav; include Leaf::Trav      end; include Trav end
+  class ProcIns;   module Trav; include Leaf::Trav      end; include Trav end
+  class Comment;   module Trav; include Leaf::Trav      end; include Trav end
+  class BogusETag; module Trav; include Leaf::Trav      end; include Trav end
+
+  class Error < StandardError; end
+end
+

data/lib/hpricot/parse.rb

@@ -0,0 +1,38 @@
+require 'hpricot/htmlinfo'
+
+def Hpricot(input = nil, opts = {}, &blk)
+  Hpricot.make(input, opts, &blk)
+end
+
+module Hpricot
+  # Exception class used for any errors related to deficiencies in the system when
+  # handling the character encodings of a document.
+  class EncodingError < StandardError; end
+
+  # Hpricot.parse parses <i>input</i> and return a document tree.
+  # represented by Hpricot::Doc.
+  def Hpricot.parse(input = nil, opts = {}, &blk)
+    make(input, opts, &blk)
+  end
+
+  # Hpricot::XML parses <i>input</i>, disregarding all the HTML rules
+  # and returning a document tree.
+  def Hpricot.XML(input = nil, opts = {}, &blk)
+    opts.merge! :xml => true
+    make(input, opts, &blk)
+  end
+
+  # :stopdoc:
+
+  def Hpricot.make(input = nil, opts = {}, &blk)
+    if blk
+      doc = Hpricot.build(&blk)
+      doc.instance_variable_set("@options", opts)
+      doc
+    else
+      Hpricot.scan(input, opts)
+    end
+  end
+
+  # :startdoc:
+end

data/lib/hpricot/tag.rb

@@ -0,0 +1,198 @@
+module Hpricot
+  # :stopdoc:
+
+  class Doc
+    def output(out, opts = {})
+      children.each do |n|
+        n.output(out, opts)
+      end if children
+      out
+    end
+    def make(input = nil, &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
+    def html_quote(str)
+      "\"" + str.gsub('"', '\\"') + "\""
+    end
+    def if_output(opts)
+      if opts[:preserve] and not raw_string.nil?
+        raw_string
+      else
+        yield opts
+      end
+    end
+    def pathname; self.name end
+    def altered!
+      clear_raw
+    end
+    def inspect_tree(depth = 0)
+      %{#{" " * depth}} + self.class.name.split(/::/).last.downcase + "\n"
+    end
+  end
+
+  class Elem
+    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
+      if self.name == 'br'
+        "\n"
+      elsif self.name == 'p'
+        "\n\n" + super + "\n\n"
+      elsif self.name == 'a' and self.has_attribute?('href')
+        "#{super} [#{self['href']}]"
+      elsif self.name == 'img' and self.has_attribute?('src')
+        "[img:#{self['src']}]"
+      else
+        super
+      end
+    end
+    def pathname; self.name end
+    def 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
+    def attributes_as_html
+      if raw_attributes
+        raw_attributes.map do |aname, aval|
+          " #{aname}" +
+            (aval ? "=#{html_quote aval}" : "")
+        end.join
+      end
+    end
+    def inspect_tree(depth = 0)
+      %{#{" " * depth}} + name + "\n" +
+        (children ? children.map { |x| x.inspect_tree(depth + 1) }.join : "")
+    end
+  end
+
+  class ETag
+    def initialize name; self.name = name end
+    def output(out, opts = {})
+      out <<
+        if_output(opts) do
+          "</#{name}>"
+        end
+    end
+  end
+
+  class BogusETag
+    def output(out, opts = {}); out << if_output(opts) { '' }; end
+  end
+
+  class Text
+    def initialize content; self.content = content end
+    def pathname; "text()" end
+    def to_s
+      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.to_s
+        end
+    end
+  end
+
+  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}]]>"
+        end
+    end
+  end
+
+  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'}\"" : "") +
+            "?>"
+        end
+    end
+  end
+
+  class DocType
+    def initialize target, pub, sys
+      self.target, self.public_id, self.system_id = target, pub, sys
+    end
+    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)}" : "") + ">"
+        end
+    end
+  end
+
+  class ProcIns
+    def pathname; "procins()" end
+    def output(out, opts = {})
+      out << 
+        if_output(opts) do
+          "<?#{target}" +
+           (content ? " #{content}" : "") +
+           "?>"
+        end
+    end
+  end
+
+  class Comment
+    def pathname; "comment()" end
+    def output(out, opts = {})
+      out <<
+        if_output(opts) do
+          "<!--#{content}-->"
+        end
+    end
+  end
+
+  # :startdoc:
+end

data/lib/hpricot/tags.rb

@@ -0,0 +1,164 @@
+module Hpricot
+
+  FORM_TAGS = [ :form, :input, :select, :textarea ]
+  SELF_CLOSING_TAGS = [ :base, :meta, :link, :hr, :br, :param, :img, :area, :input, :col ]
+
+  # Common sets of attributes.
+  AttrCore = [:id, :class, :style, :title]
+  AttrI18n = [:lang, 'xml:lang'.intern, :dir]
+  AttrEvents = [:onclick, :ondblclick, :onmousedown, :onmouseup, :onmouseover, :onmousemove, 
+      :onmouseout, :onkeypress, :onkeydown, :onkeyup]
+  AttrFocus = [:accesskey, :tabindex, :onfocus, :onblur]
+  AttrHAlign = [:align, :char, :charoff]
+  AttrVAlign = [:valign]
+  Attrs = AttrCore + AttrI18n + AttrEvents
+
+  # All the tags and attributes from XHTML 1.0 Strict
+  class XHTMLStrict
+    class << self
+      attr_accessor :tags, :tagset, :forms, :self_closing, :doctype
+    end
+    @doctype = ["-//W3C//DTD XHTML 1.0 Strict//EN", "DTD/xhtml1-strict.dtd"]
+    @tagset = {
+      :html => AttrI18n + [:id, :xmlns],
+      :head => AttrI18n + [:id, :profile],
+      :title => AttrI18n + [:id],
+      :base => [:href, :id],
+      :meta => AttrI18n + [:id, :http, :name, :content, :scheme, 'http-equiv'.intern],
+      :link => Attrs + [:charset, :href, :hreflang, :type, :rel, :rev, :media],
+      :style => AttrI18n + [:id, :type, :media, :title, 'xml:space'.intern],
+      :script => [:id, :charset, :type, :src, :defer, 'xml:space'.intern],
+      :noscript => Attrs,
+      :body => Attrs + [:onload, :onunload],
+      :div => Attrs,
+      :p => Attrs,
+      :ul => Attrs,
+      :ol => Attrs,
+      :li => Attrs,
+      :dl => Attrs,
+      :dt => Attrs,
+      :dd => Attrs,
+      :address => Attrs,
+      :hr => Attrs,
+      :pre => Attrs + ['xml:space'.intern],
+      :blockquote => Attrs + [:cite],
+      :ins => Attrs + [:cite, :datetime],
+      :del => Attrs + [:cite, :datetime],
+      :a => Attrs + AttrFocus + [:charset, :type, :name, :href, :hreflang, :rel, :rev, :shape, :coords],
+      :span => Attrs,
+      :bdo => AttrCore + AttrEvents + [:lang, 'xml:lang'.intern, :dir],
+      :br => AttrCore,
+      :em => Attrs,
+      :strong => Attrs,
+      :dfn => Attrs,
+      :code => Attrs,
+      :samp => Attrs,
+      :kbd => Attrs,
+      :var => Attrs,
+      :cite => Attrs,
+      :abbr => Attrs,
+      :acronym => Attrs,
+      :q => Attrs + [:cite],
+      :sub => Attrs,
+      :sup => Attrs,
+      :tt => Attrs,
+      :i => Attrs,
+      :b => Attrs,
+      :big => Attrs,
+      :small => Attrs,
+      :object => Attrs + [:declare, :classid, :codebase, :data, :type, :codetype, :archive, :standby, :height, :width, :usemap, :name, :tabindex],
+      :param => [:id, :name, :value, :valuetype, :type],
+      :img => Attrs + [:src, :alt, :longdesc, :height, :width, :usemap, :ismap],
+      :map => AttrI18n + AttrEvents + [:id, :class, :style, :title, :name],
+      :area => Attrs + AttrFocus + [:shape, :coords, :href, :nohref, :alt],
+      :form => Attrs + [:action, :method, :enctype, :onsubmit, :onreset, :accept, :accept],
+      :label => Attrs + [:for, :accesskey, :onfocus, :onblur],
+      :input => Attrs + AttrFocus + [:type, :name, :value, :checked, :disabled, :readonly, :size, :maxlength, :src, :alt, :usemap, :onselect, :onchange, :accept],
+      :select => Attrs + [:name, :size, :multiple, :disabled, :tabindex, :onfocus, :onblur, :onchange],
+      :optgroup => Attrs + [:disabled, :label],
+      :option => Attrs + [:selected, :disabled, :label, :value],
+      :textarea => Attrs + AttrFocus + [:name, :rows, :cols, :disabled, :readonly, :onselect, :onchange],
+      :fieldset => Attrs,
+      :legend => Attrs + [:accesskey],
+      :button => Attrs + AttrFocus + [:name, :value, :type, :disabled],
+      :table => Attrs + [:summary, :width, :border, :frame, :rules, :cellspacing, :cellpadding],
+      :caption => Attrs,
+      :colgroup => Attrs + AttrHAlign + AttrVAlign + [:span, :width],
+      :col => Attrs + AttrHAlign + AttrVAlign + [:span, :width],
+      :thead => Attrs + AttrHAlign + AttrVAlign,
+      :tfoot => Attrs + AttrHAlign + AttrVAlign,
+      :tbody => Attrs + AttrHAlign + AttrVAlign,
+      :tr => Attrs + AttrHAlign + AttrVAlign,
+      :th => Attrs + AttrHAlign + AttrVAlign + [:abbr, :axis, :headers, :scope, :rowspan, :colspan],
+      :td => Attrs + AttrHAlign + AttrVAlign + [:abbr, :axis, :headers, :scope, :rowspan, :colspan],
+      :h1 => Attrs,
+      :h2 => Attrs,
+      :h3 => Attrs,
+      :h4 => Attrs,
+      :h5 => Attrs,
+      :h6 => Attrs
+    }
+
+    @tags = @tagset.keys
+    @forms = @tags & FORM_TAGS
+    @self_closing = @tags & SELF_CLOSING_TAGS
+  end
+
+  # Additional tags found in XHTML 1.0 Transitional
+  class XHTMLTransitional
+    class << self
+      attr_accessor :tags, :tagset, :forms, :self_closing, :doctype
+    end
+    @doctype = ["-//W3C//DTD XHTML 1.0 Transitional//EN", "DTD/xhtml1-transitional.dtd"]
+    @tagset = XHTMLStrict.tagset.merge \
+      :strike => Attrs,
+      :center => Attrs,
+      :dir => Attrs + [:compact], 
+      :noframes => Attrs,
+      :basefont => [:id, :size, :color, :face], 
+      :u => Attrs,
+      :menu => Attrs + [:compact], 
+      :iframe => AttrCore + [:longdesc, :name, :src, :frameborder, :marginwidth, :marginheight, :scrolling, :align, :height, :width],
+      :font => AttrCore + AttrI18n + [:size, :color, :face],
+      :s => Attrs,
+      :applet => AttrCore + [:codebase, :archive, :code, :object, :alt, :name, :width, :height, :align, :hspace, :vspace],
+      :isindex => AttrCore + AttrI18n + [:prompt]
+
+    # Additional attributes found in XHTML 1.0 Transitional
+    { :script => [:language],
+      :a => [:target],
+      :td => [:bgcolor, :nowrap, :width, :height],
+      :p => [:align],
+      :h5 => [:align],
+      :h3 => [:align],
+      :li => [:type, :value],
+      :div => [:align],
+      :pre => [:width],
+      :body => [:background, :bgcolor, :text, :link, :vlink, :alink],
+      :ol => [:type, :compact, :start],
+      :h4 => [:align],
+      :h2 => [:align],
+      :object => [:align, :border, :hspace, :vspace],
+      :img => [:name, :align, :border, :hspace, :vspace],
+      :link => [:target],
+      :legend => [:align],
+      :dl => [:compact],
+      :input => [:align],
+      :h6 => [:align],
+      :hr => [:align, :noshade, :size, :width],
+      :base => [:target],
+      :ul => [:type, :compact],
+      :br => [:clear],
+      :form => [:name, :target],
+      :area => [:target],
+      :h1 => [:align]
+    }.each do |k, v|
+        @tagset[k] += v
+    end
+
+    @tags = @tagset.keys
+    @forms = @tags & FORM_TAGS
+    @self_closing = @tags & SELF_CLOSING_TAGS
+  end
+
+end