rgen 0.3.0 → 0.4.0

data/redist/xmlscan/samples/chibixml.rb

@@ -0,0 +1,105 @@
+#
+# samples/chibixml.rb - xmlscan with ChibiXML
+#
+#   Copyright (C) UENO Katsuhiro 2002
+#
+# $Id: chibixml.rb,v 1.1 2002/09/28 10:52:41 katsu Exp $
+#
+
+require 'xmlscan/parser'
+require 'chibixml'
+
+
+module XMLScan
+
+  module ChibiXML
+
+    class Visitor
+
+      include XMLScan::Visitor
+      include ::ChibiXML
+
+      attr_reader :document
+
+      def on_start_document
+        @document = createDocument
+        @current = @document
+      end
+
+      def on_doctype(root, pubid, sysid)
+        @current.doctype = Node.new(DOCUMENT_TYPE_NODE, root)
+      end
+
+      def on_comment(str)
+        @current.appendChild @document.createComment(str)
+      end
+
+      def on_pi(target, pi)
+        @current.appendChild @document.createProcessingInstruction(target,pi)
+      end
+
+      def on_chardata(str)
+        @current.appendChild @document.createTextNode(str)
+      end
+
+      def on_entityref(ref)
+        @current.appendChild @document.createEntityReference(ref)
+      end
+
+      def on_stag(name)
+        element = @document.createElement(name)
+        @current.appendChild element
+        @current = element
+      end
+
+      def on_attribute(*)
+        @attr = ''
+      end
+
+      def on_attr_value(str)
+        @attr << str
+      end
+
+      def on_attr_charref(str)
+        @attr << [str].pack('U')
+      end
+
+      def on_attribute_end(name)
+        @current.setAttribute name, @attr
+      end
+
+      def on_etag(*)
+        @current = @current.parentNode
+      end
+
+    end
+
+
+    def self.parse(src)
+      visitor = Visitor.new
+      XMLScan::XMLParser.new(visitor).parse src
+      visitor.document
+    end
+
+    def self.scan(src)
+      visitor = Visitor.new
+      XMLScan::XMLScanner.new(visitor).parse src
+      visitor.document
+    end
+
+  end
+
+end
+
+
+
+
+
+if $0 == __FILE__ then
+  src = ARGF
+  def src.path; filename; end
+  t1 = Time.times.utime
+  XMLScan::ChibiXML.parse src
+  t2 = Time.times.utime
+  STDERR.printf "%2.3f sec\n", t2 - t1
+end

data/redist/xmlscan/samples/getxmlchar.rb

@@ -0,0 +1,122 @@
+#!/usr/bin/ruby
+#
+# getxmlchar.rb - get XML-valid characters from XML 1.0 Specification.
+#
+#   Copyright (C) UENO Katsuhiro 2002
+#
+# $Id: getxmlchar.rb,v 1.2 2002/09/27 02:33:40 katsu Exp $
+#
+
+require 'xmlscan/parser'
+
+
+class String
+
+  def decamel
+    gsub(/[a-z](?=[A-Z])|[A-Z](?=([A-Z][a-z])|((?=[a-z])))/){ |s|
+      ($1 ? s : s.downcase) + ($2 || '_')
+    }
+  end
+
+end
+
+
+
+class ProductionGetter
+
+  include XMLScan::Visitor
+
+  def initialize(targets)
+    @targets = targets
+    @parser = XMLScan::XMLParser.new(self)
+    @elem = nil
+    @content = nil
+    @current = nil
+    @production = {}
+  end
+
+  attr_reader :production
+
+  def parse(src)
+    @parser.parse src
+  end
+
+  def on_stag(name)
+    if name == 'lhs' or name == 'rhs' then
+      @elem = name
+      @content = ''
+    end
+  end
+
+  def on_chardata(str)
+    if @content then
+      @content << str
+    end
+  end
+
+  def on_cdata(str)
+    on_chardata str
+  end
+
+  def on_etag(name)
+    case name
+    when 'lhs' then
+      if @targets.include? @content then
+        if $VERBOSE then
+          STDERR.puts "`#{@content}' is found at line #{@parser.lineno}."
+        end
+        @current = @content
+      end
+    when 'rhs' then
+      if @current then
+        @production[@current] = @content
+      end
+      @current = nil
+    end
+    @elem = nil
+    @content = nil
+  end
+
+end
+
+
+
+
+
+def ARGF.path
+  filename
+end
+
+Productions = %w(Char BaseChar Ideographic CombiningChar Digit Extender)
+
+p = ProductionGetter.new(Productions)
+p.parse ARGF
+
+#   0x0041..0x005A,  0x0061..0x007A,  0x00C0..0x00D6,  0x00D8..0x00F6,
+# |-^--+----|----+---^|----+----|----+^---|----+----|--^-+----|----+----|
+# 0 2                19               36               53
+
+Productions.each { |name|
+  prod = p.production[name]
+  unless prod then
+    print "#{name} = []    # not found\n"
+  else
+    ranges = []
+    prod.scan(/\#x([0-9A-Fa-f]+)(?:-\#x([0-9A-Fa-f]+))?/) { |rstart,rend|
+      rend = rend || rstart
+      ranges.push Range.new(rstart.hex, rend.hex)
+    }
+    ranges.sort! { |a,b| a.begin <=> b.begin }
+    print "#{name.decamel} = ["
+    column = 0
+    ranges.each { |i|
+      column += 1
+      print "\n  " if column == 1
+      s = sprintf("0x%04X..0x%04X,", i.begin, i.end)
+      s = s.ljust(17) if column < 4
+      print s
+      column = 0 if column >= 4
+    }
+    print "\n]\n"
+  end
+}

data/redist/xmlscan/samples/rexml.rb

@@ -0,0 +1,159 @@
+#
+# samples/rexml.rb - xmlscan with REXML
+#
+#   Copyright (C) UENO Katsuhiro 2002
+#
+# $Id: rexml.rb,v 1.1 2002/09/29 15:55:31 katsu Exp $
+#
+
+require 'xmlscan/parser'
+require 'rexml/document'
+
+
+module XMLScan
+
+  module REXML
+
+    class Visitor
+
+      include XMLScan::Visitor
+      include ::REXML
+
+      def initialize(context = {})
+        @document = Document.new(nil, context)
+      end
+
+      attr_reader :document
+
+
+      def on_start_document
+        @current = @document
+      end
+
+
+      def on_xmldecl
+        @version = nil
+        @encoding = nil
+        @standalone = nil
+      end
+
+      def on_xmldecl_version(str)
+        @version = str
+      end
+
+      def on_xmldecl_encoding(str)
+        @encoding = str
+      end
+
+      def on_xmldecl_standalone(str)
+        @standalone = str
+      end
+
+      def on_xmldecl_end
+        @document.add XMLDecl.new(@version, @encoding, @standalone)
+      end
+
+
+      def on_doctype(root, pubid, sysid)
+        if pubid then
+          external_id = 'PUBLIC'
+        else
+          external_id = 'SYSTEM'
+        end
+        if pubid then
+          if /"/ =~ pubid then
+            external_id <<  " '#{pubid}'"
+          else
+            external_id << %' "#{pubid}"'
+          end
+        end
+        if sysid then
+          if /"/ =~ sysid then
+            external_id <<  " '#{sysid}'"
+          else
+            external_id << %' "#{sysid}"'
+          end
+        end
+        @current.add DocType.new(root, external_id)
+      end
+
+
+      def on_comment(str)
+        @current.add Comment.new(str)
+      end
+
+      def on_pi(target, pi)
+        @current.add Instruction.new(target, pi)
+      end
+
+
+      def on_stag(name)
+        @current = Element.new(name, @current, @current.context)
+      end
+
+      def on_attribute(name)
+        @attr = ''
+      end
+
+      def on_attr_value(str)
+        @attr << str
+      end
+
+      def on_attr_entityref(ref)
+        @attr << "&#{ref};"
+      end
+
+      def on_attr_charref(code)
+        @attr << [code].pack('U')
+      end
+
+      def on_attribute_end(name)
+        @current.add_attribute name, @attr
+      end
+
+      def on_etag(name)
+        @current = @current.parent
+      end
+
+
+      def on_chardata(str)
+        @current.add_text str
+      end
+
+      def on_cdata(str)
+        @current.add CData.new(str)
+      end
+
+      def on_entityref(ref)
+        @current.add_text "&#{ref};"
+      end
+
+      def on_charref(code)
+        @current.add_text "&#{code};"
+      end
+
+    end
+
+
+    def self.parse(src)
+      visitor = Visitor.new
+      XMLScan::XMLParser.new(visitor).parse src
+      visitor.document
+    end
+
+  end
+
+end
+
+
+
+
+
+if $0 == __FILE__ then
+  src = ARGF
+  def src.path; filename; end
+  t1 = Time.times.utime
+  XMLScan::REXML.parse src
+  t2 = Time.times.utime
+  STDERR.printf "%2.3f sec\n", t2 - t1
+end

data/redist/xmlscan/samples/xmlbench.rb

@@ -0,0 +1,88 @@
+#!/usr/bin/ruby
+#
+# samples/xmlbench.rb
+#
+#   Copyright (C) UENO Katsuhiro 2002
+#
+# $Id: xmlbench.rb,v 1.6 2003/03/16 02:34:53 katsu Exp $
+#
+
+require 'samples/xmlbench/xmlbench-lib'
+require 'benchmark'
+
+$KCODE = 'U'
+
+
+source = ARGV.shift
+
+unless source then
+  STDERR.print "#$0: no source document is given.\n"
+  exit 1
+end
+
+#XMLBench.weight_limit = 1000
+if ARGV.empty? then
+  subjects = XMLBench.setup_all
+else
+  subjects = XMLBench.setup(*ARGV)
+end
+
+
+
+class BenchmarkRunner
+
+  def initialize(source)
+    @source = source
+    @bench = []
+  end
+
+  def add_bench(*subjects)
+    @bench.concat subjects
+  end
+
+  BenchTitles = [
+    [ '** File **',   proc { |f| f }, ],
+    [ '** Array **',  proc { |f| f.readlines } ],
+    [ '** String **', proc { |f| f.read } ],
+  ]
+
+  def run
+    file = File.open(@source)
+    source = nil
+    pid = nil
+    begin
+      Benchmark.bm(30) { |x|
+        BenchTitles.each { |title,sourcemaker|
+          puts title
+          @bench.each { |i|
+            file.rewind
+            pid = fork {
+              begin
+                source = sourcemaker.call(file)
+                GC.start
+                x.report(i.name, '') { i.parse source }
+              rescue Exception => e
+                puts "  #{e.class.name}"
+              ensure
+                file.close
+              end
+            }
+            Process.waitpid(pid)
+            pid = nil
+            puts "  killed at status #{$?}" unless $? == 0
+          }
+        }
+      }
+    ensure
+      Process.waitpid pid if pid rescue nil
+      file.close
+    end
+    self
+  end
+
+end
+
+
+bench = BenchmarkRunner.new(source)
+bench.add_bench(*subjects)
+bench.run