ruby-web 1.1.1

data/lib/webunit/parser.rb

@@ -0,0 +1,229 @@
+=begin
+= WebUnit::Parser
+  Copyright(C) 2001 yuichi TAKAHASHI, Narushima Hironori.
+  $Id: parser.rb,v 1.5 2004/04/29 12:34:04 yuichi Exp $
+  Part of WebUnit::Parser is variant of html-parser
+=end
+
+require 'htmlrepair'
+require 'sgml-parser'
+require 'html-parser'
+require 'formatter'
+
+module WebUnit
+
+  class Parser < SGMLParser
+
+    include Utils
+
+    @@ignoretags = []
+    SingleTags = %w( meta br hr base link area ) # input, frame, img
+
+    def initialize()
+      super( NullFormatter.new )
+      @elem_stack = [ HtmlElem::new( nil, nil ) ]
+      @form_stack = []
+      @table_stack = []
+      self.ignore @@ignoretags
+    end
+
+    def feed( response )
+      @response = response
+      begin
+        $stderr.puts HTMLSplit.new( @response.body ).repair.to_s if $DEBUG
+        super( HTMLSplit.new( @response.body ).repair.to_s )
+      rescue
+        raise BadHtmlTags, $!.message + @elem_stack.inspect
+      end
+      @elem_stack.pop.children[0]
+    end
+
+    def handle_data(data)
+      @elem_stack.last.append data.squeeze( " \n" ).strip
+    end
+
+    def starttag( elem )
+      @elem_stack.last.append elem
+      @elem_stack.push elem
+    end
+
+    def endtag( tag )
+      elem = @elem_stack.pop
+      if tag != elem.tag
+        if SingleTags.include?(tag)
+          @elem_stack.push(elem)
+        else
+          puts "'#{tag}'(wait for '#{elem.tag.to_s}')" if $DEBUG
+          raise BadHtmlTags, "'#{tag}'(wait for '#{elem.tag.to_s}')"
+        end
+      end
+      elem
+    end
+
+    def unknown_starttag( tag, attrs )
+      elem = HtmlElem::new( tag, attrs_to_hash( attrs ) )
+      starttag( elem )
+    end
+
+    def unknown_endtag( tag )
+      endtag( tag )
+    end
+
+    # special start and end ( a, form, table, ... )
+
+    def start_a( attrs )
+      ah = attrs_to_hash attrs
+      unless ah["href"] =~ /^#/
+        ah["href"] = complete_url( ah["href"], @response.url )
+        ah["href"].gsub!( "&amp;", "&" )
+      end
+      elem = Link::new( ah )
+      @response.add_link( elem )
+      starttag( elem )
+    end
+
+    def start_form( attrs )
+      ah = attrs_to_hash attrs
+      ah["action"] = complete_url( ah["action"], @response.url )
+      ah["action"].gsub!( "&amp;", "&" )
+      elem = Form::new( ah )
+      @response.add_form elem
+      @form_stack.push elem
+      starttag( elem )
+    end
+    def end_form
+      endtag( 'form' )
+      @form_stack.pop
+    end
+
+    def start_select( attrs )
+      ah = attrs_to_hash attrs
+      elem = Select::new( ah )
+      @form_stack.last.add_param elem
+      starttag( elem )
+    end
+    def end_select
+      endtag( 'select' )
+      @form_stack.last.parameters.last.end_option
+    end
+
+    def start_option( attrs )
+      ah = attrs_to_hash attrs
+      elem = SelectOption::new( ah )
+      @form_stack.last.parameters.last.add_option elem
+      starttag( elem )
+    end
+
+    def start_textarea( attrs )
+      ah = attrs_to_hash attrs
+      elem = Textarea::new( ah )
+      @form_stack.last.add_param elem
+      starttag( elem )
+    end
+
+    def start_table( attrs )
+      ah = attrs_to_hash attrs
+      elem = Table::new( ah )
+      @response.add_table elem
+      @table_stack.push elem
+      starttag( elem )
+    end
+    def end_table
+      endtag( 'table' )
+      @table_stack.pop
+    end
+
+    def start_tr( attrs )
+      elem = TableRow::new( attrs_to_hash( attrs ) )
+      starttag( elem )
+      @table_stack.last.add_row( attrs_to_hash( attrs ) )
+    end
+
+    def start_th( attrs )
+      elem = TableCell::new( 'th', attrs_to_hash( attrs ) )
+      @table_stack.last.add_cell( elem )
+      starttag( elem )
+    end
+
+    def start_td( attrs )
+      elem = TableCell::new( 'td', attrs_to_hash( attrs ) )
+      @table_stack.last.add_cell( elem )
+      starttag( elem )
+    end
+
+    # do ( start only )
+    def dotag( tag, attrs )
+      elem = HtmlElem::new( tag, attrs_to_hash( attrs ) )
+      @elem_stack.last.children.push elem
+    end
+
+    SingleTags.each do |t|
+      eval "def do_#{t}( attrs ); dotag( '#{t}', attrs ); end"
+    end
+
+    # special do ( input, frame, ... )
+
+    def do_input( attrs )
+      ah = attrs_to_hash attrs
+      case ah['type']
+        when 'submit'
+          elem = InputSubmit::new( ah )
+        when 'reset'
+          elem = InputReset::new( ah )
+        when 'checkbox'
+          elem = InputCheckbox::new( ah )
+        when 'radio'
+          elem = InputRadio::new( ah )
+        when 'image'
+          elem = InputImage::new( ah )
+        when 'file'
+          elem = InputFile::new( ah )
+        else # text
+          elem = Input::new( ah )
+      end
+	  
+      @form_stack.last.add_param elem if @form_stack.last
+      @elem_stack.last.children.push elem
+    end
+
+    def do_frame( attrs )
+      ah = attrs_to_hash attrs
+      ah['src'] = complete_url( ah['src'], @response.url )
+      elem = Frame::new( ah )
+      @response.add_frame( elem )
+    end
+
+    def do_img( attrs )
+      ah = attrs_to_hash attrs
+      elem = Image::new( ah )
+      @response.add_image( elem )
+      @elem_stack.last.children.push elem
+    end
+
+    # ignore tag
+    def ignore( arr )
+      arr.each do |t|
+        self.instance_eval( "def start_#{t}( a ); end" )
+        self.instance_eval( "def end_#{t}; end" )
+        self.instance_eval( "def do_#{t}( a ); end" )
+      end
+    end
+
+    def self::ignore( arr )
+      @@ignoretags = arr
+    end
+
+    # utils
+
+    def attrs_to_hash( attrs )
+      h = {}
+      for a, v in attrs
+        v = v.split('"')[1]
+        h[a] = v ? v : ''
+      end
+      h
+    end
+
+  end
+
+end

data/lib/webunit/response.rb

@@ -0,0 +1,464 @@
+=begin
+= WebUnit::Response
+  Copyright(C) 2001 yuichi TAKAHASHI, Narushima Hironori.
+  $Id: response.rb,v 1.1.1.1 2003/01/21 11:50:40 yuichi Exp $
+=end
+
+require 'net/http'
+
+module WebUnit
+
+  class Response
+
+    include Utils
+
+    @@host = nil
+    @@port = nil
+    @@http = nil
+
+    attr_reader :url, :method, :response, :body, :code
+    attr_reader :tree, :links, :forms, :tables, :frames, :images
+    attr_accessor :username, :password
+
+=begin
+--- Response.get(url,data=nil)
+      return a Response.
+=end
+    def self::get( url, data=nil, up=nil )
+      r = self.new
+      r.username = up[0] if up
+      r.password = up[1] if up
+      r.init_http( url, "GET", data )
+    end
+
+=begin
+--- Response.post(url,data=nil)
+      return a Response.
+=end
+    def self::post( url, data=nil, up=nil )
+      r = self.new      
+      r.username = up[0] if up
+      r.password = up[1] if up
+      r.init_http( url, "POST", data )
+    end
+
+=begin
+--- Response.html( html)
+      return a Response.
+=end
+
+    def self::html( html, url=nil )
+      html = "<html>#{html}</html>" unless html =~ %r!^<html>!
+      self.new.init_html( html, url )
+    end
+
+    def self::reset
+      begin
+        @@http.finish if @@http
+      rescue IOError
+        @@http = nil
+      end
+      @@host = nil
+      @@port = nil
+      @@http = nil
+    end
+
+    def init_html( html, url )
+      @url = url
+      @method = 'html'
+      @body = html
+      feed
+      self
+    end
+
+	def do_connect(host, port)
+      if @@host != host or @@port != port
+        @@host = host
+        @@port = port
+        begin
+          @@http.finish if @@http
+        rescue IOError
+          @@http = nil
+        end
+        @@http = Net::HTTP::new( host, port )
+      end
+	end
+
+    def init_http( url, method, data=nil, retried=false, multipart=false)
+      url = $URLBASE + url unless url =~ %r!://!
+      @url = url
+      @method = method
+      header = {}
+	  
+      prot,host,port,path = parse_url( url )
+	  do_connect( host, port)
+	  
+      header.update( Cookies::header( path ) )
+      @username ||= nil; @password ||= nil
+      if @username != nil && @password != nil
+        header['authorization'] =
+        'Basic ' + ["#{@username}:#{@password}"].pack('m').gsub(/\s+/, '')
+      end
+	  
+      begin
+	    if multipart
+	        header['Content-type'] = "multipart/form-data,boundary=#{BOUNDARY}"
+			
+			#puts data
+			
+	        @response = @@http.post2( path, data.gsub("\n", "\r\n"), header)
+		else
+          case @method.upcase
+          when "GET" then
+            path << '?' + data if data
+            puts path if $DEBUG
+            @response = @@http.get2( path, header )
+          when "POST" then
+            if header['Content-type'] == nil
+              header['Content-type'] = "application/x-www-form-urlencoded"
+            end
+            @response = @@http.post2( path, data, header )
+          end
+		end
+      rescue Errno::EPIPE
+        @@host = nil
+        raise if retried
+        return self.init_http( url, method, data, true, multipart)
+      end
+	  
+      @body = @response.body
+      @code = @response.code
+      p @response['set-cookie'] if $DEBUG
+      Cookies::update( @response['set-cookie'] )
+      case @response['content-type']
+      when %r!text/html!
+        feed
+      when %r!text/xml!
+        return DomWalker::new( @response.body )
+      end
+	  
+      raise HttpNotFound if @code == "404"
+      self
+    end
+
+    def feed
+      @forms = []
+      @tables = []
+      @links = []
+      @frames = []
+      @images = []
+
+      @code ||= nil
+      if @code == '302'
+        @links.push Link.new( 'href' => @response['Location'] ) if @response.key? 'Location'
+      end
+      p = Parser.new
+      @tree = p.feed( self )
+      p.close
+      freeze
+    end
+
+    def add_link( link )
+      @links.push( link )
+    end
+
+    def add_form( form )
+      @forms.push( form )
+    end
+
+    def add_frame( frame )
+      @frames.push( frame )
+    end
+
+    def add_table( table )
+      @tables.push( table )
+    end
+
+    def add_image( image )
+      @images.push( image )
+    end
+
+    def freeze
+      @forms.each do |f|
+        f.freeze
+      end
+    end
+
+=begin
+--- Response#title
+      return title.
+=end
+
+    def title
+      @tree.find( 'title' )[0].data
+    end
+
+=begin
+--- Response#find( tag )
+      same as Htmlelem#find.
+=end
+
+    def find( tag )
+      @tree.find( tag )
+    end
+
+=begin
+--- Response#search( str )
+      same as Htmlelem#search.
+=end
+
+    def search( str )
+      @tree.search( str )
+    end
+
+=begin
+--- Response#readlink( str )
+      same as Htmlelem#find.
+=end
+
+    def readlink( str )
+      @tree.readlink( str )
+    end
+
+=begin
+--- Response#click( str )
+      at first Response#readlink, and then, Form#submit.
+=end
+
+    def click( str, x=nil, y=nil )
+      begin
+        @tree.readlink( str )
+      rescue ElemNotFound
+        matched = 0
+        @tree.find( 'input' ).each do |i|
+          if i.class == InputSubmit
+            matched = i.name if i.name == str || i.value == str
+          elsif i.class == InputImage
+            matched = i.name if i.name == str
+          end
+        end
+        if matched == 0
+          raise ElemNotFound
+        else
+          submit( matched, x, y )
+        end
+      end
+    end
+
+=begin
+--- Response#submit( button=nil,x=nil,y=nil)
+      short for Response.forms[0].submit.
+=end
+
+    def submit( button=nil, x=nil, y=nil )
+      @forms[0].submit( button, x, y )
+    end
+
+    def push( button=nil )
+      $stderr.puts "no longer support Response#push, use Response#submit"
+      submit
+    end
+
+=begin
+--- Response#param
+      short for Response.forms[0].params
+=end
+
+    def params
+      @forms[0].params
+    end
+
+=begin
+--- Response#image
+=end
+
+    def image
+      self.images[0]
+    end
+
+=begin
+--- Response#link
+      short for Response.links[0]
+=end
+
+    def link
+      @links[0]
+    end
+
+=begin
+--- Response#form
+      short for Response.forms[0]
+=end
+
+    def form
+      @forms[0]
+    end
+
+=begin
+--- Response#table
+      short for Response.tables[0]
+=end
+
+    def table
+      @tables[0]
+    end
+
+=begin
+--- Response#frame
+      short for Response.frames[0]
+=end
+
+    def frame
+      @frames[0]
+    end
+
+=begin
+--- Response#redirect
+=end
+
+    def redirect
+      if @code == '302'
+        @links[0].read
+      else
+        raise BadOperation
+      end
+    end
+
+=begin
+--- Response#opens
+=end
+
+    def opens
+      a = []
+      re_script = %r|<!--(.*?)// *-->|m
+      re_open = %r|open\( *['"]([^'"]*)['"], *['"]([^'"]*)['"] *\)|
+      body_rest = response.body
+      while re_script =~ body_rest
+        body_rest = $'
+        rest = $1
+        while re_open =~ rest
+          ah = {}
+          ah['file'] = $1
+          ah['target'] = $2
+          ah['url'] = complete_url( ah['file'], @url )
+          a << JSciriptOpenObject::new( ah )
+          rest = $'
+        end
+      end
+      a
+    end
+
+    def js_open_objs
+      $stderr.puts "no longer support Response#js_open_objs, use Response#opens"
+      opens
+    end
+
+=begin
+--- Response#open
+      short for Response.opens[0]
+=end
+
+    def open
+      opens[0]
+    end
+
+=begin
+--- Response#pbody
+=end
+
+    def pbody
+      puts "\n" + '-' * 72
+      puts self.format
+      puts '-' * 72
+    end
+
+=begin
+--- Response#format
+=end
+
+    def format
+      w = MemWriter.new
+      p = HTMLParser.new( AbstractFormatter.new( w ) )
+      p.feed( @body )
+      p.close
+      w.to_s
+    end
+
+  end
+
+  class MemWriter < NullWriter
+
+    def initialize
+      @body = ''
+      @maxcol = 72
+      super()
+      reset
+    end
+
+    def reset
+      @col = 0
+      @atbreak = false
+    end
+
+    def to_s
+      @body.squeeze( "\n" )
+    end
+
+    def send_paragraph(blankline)
+      @body << ("\n" + "\n"*blankline)
+      @col = 0
+      @atbreak = false
+    end
+
+    def send_line_break
+      @body << ("\n")
+      @col = 0
+      @atbreak = false
+    end
+
+    def send_hor_rule #(*args, **kw)
+      @body << ("\n")
+      @body << ('-'*@maxcol)
+      @body << ("\n")
+      @col = 0
+      @atbreak = false
+    end
+
+    def send_literal_data(data)
+      @body << (data)
+      i = data.rindex("\n")
+      if i
+        @col = 0
+        data = data[i+1..-1]
+        #data = string.expandtabs(data)
+        @col = @col + data.length
+        @atbreak = false
+      end
+    end
+
+    def send_flowing_data(data)
+      return if not data
+      atbreak = (@atbreak || (/^\s/ =~ data))
+      col = @col
+      maxcol = @maxcol
+      for word in data.split
+        if atbreak
+          if col + word.length >= maxcol
+            @body << ("\n")
+            col = 0
+          else
+            @body << (' ')
+            col = col + 1
+          end
+        end
+        @body << (word)
+        col = col + word.length
+        atbreak = true
+      end
+      @col = col
+      @atbreak = (/\s$/ =~ data)
+    end
+
+  end
+
+end