mame-xmlparser 0.6.81.1

data/lib/xmldigest.rb

@@ -0,0 +1 @@
+require 'xml/dom/digest'

data/lib/xmlencoding-ja.rb

@@ -0,0 +1,11 @@
+## -*- Ruby -*-
+## Sample XMLEncoding class for Japanese (EUC-JP, Shift_JIS)
+## 1998 by yoshidam
+##
+## Usage:
+##    require 'xml/encoding-ja'
+##    include XMLEncoding_ja
+
+require 'xml/encoding-ja'
+
+XMLEncoding_ja = XML::Encoding_ja

data/lib/xmltree.rb

@@ -0,0 +1 @@
+require 'xml/dom/core'

data/lib/xmltreebuilder-ja.rb

@@ -0,0 +1,9 @@
+## -*- Ruby -*-
+## Tree builder class for Japanese encoding
+## 1998 by yoshidam
+
+require 'xml/dom/builder-ja'
+
+module XML
+  JapaneseTreeBuilder = DOM::JapaneseBuilder
+end

data/lib/xmltreebuilder.rb

@@ -0,0 +1 @@
+require 'xml/dom/builder'

data/lib/xmltreevisitor.rb

@@ -0,0 +1 @@
+require 'xml/dom/visitor'

data/samples/buildertest.rb

@@ -0,0 +1,47 @@
+#! /usr/local/bin/ruby -Ku
+
+require 'xml/dom/builder'
+require 'nkf'
+#require 'uconv'
+
+class XMLRetry<Exception; end
+
+#def Uconv.unknown_unicode_handler(u)
+#  return '??'
+##  return "#[#{format('%04x', u)}]"
+#end
+
+class EUCTreeBuilder < XML::DOM::Builder
+#  def nameConverter(str)
+#    Uconv.u8toeuc(str)
+#  end
+#  def cdataConverter(str)
+#    Uconv.u8toeuc(str)
+#  end
+end
+
+builder = EUCTreeBuilder.new
+def builder.unknownEncoding(e)
+  raise XMLRetry, e
+end
+
+xml = $<.read
+
+begin
+  tree = builder.parse(xml)
+rescue XMLRetry
+  newencoding = nil
+  e = $!.to_s
+  if e =~ /^iso-2022-jp$/i
+    xml = NKF.nkf("-Je", xml)
+    newencoding = "EUC-JP"
+  end
+  builder = EUCTreeBuilder.new(0, newencoding)
+  retry
+rescue XML::Parser::Error
+  line = builder.line
+  print "#{$0}: #{$!} (in line #{line})\n"
+  exit 1
+end
+#print tree.to_s.gsub(/\#\[([0-9a-f]{4})\]/, "&#x\\1;"), "\n"
+tree.dump

data/samples/buildertest2.rb

@@ -0,0 +1,50 @@
+#! /usr/local/bin/ruby -Ku
+
+require 'xml/dom/builder'
+require 'nkf'
+#require 'uconv'
+
+class XMLRetry<Exception; end
+
+#def Uconv.unknown_unicode_handler(u)
+#  return '??'
+##  return "#[#{format('%04x', u)}]"
+#end
+
+class EUCTreeBuilder < XML::DOM::Builder
+#  def nameConverter(str)
+#    Uconv.u8toeuc(str)
+#  end
+#  def cdataConverter(str)
+#    Uconv.u8toeuc(str)
+#  end
+end
+
+builder = EUCTreeBuilder.new(1)
+def builder.unknownEncoding(e)
+  raise XMLRetry, e
+end
+
+xml = $<.read
+
+begin
+  tree = builder.parse(xml)
+rescue XMLRetry
+  newencoding = nil
+  e = $!.to_s
+  if e =~ /^iso-2022-jp$/i
+    xml = NKF.nkf("-Je", xml)
+    newencoding = "EUC-JP"
+  end
+  builder = EUCTreeBuilder.new(1, newencoding)
+  retry
+rescue XML::Parser::Error
+  line = builder.line
+  print "#{$0}: #{$!} (in line #{line})\n"
+  exit 1
+end
+#print tree.to_s.gsub(/\#\[([0-9a-f]{4})\]/, "&#x\\1;"), "\n"
+tree.documentElement.normalize
+tree.dump
+#print tree
+

data/samples/digesttest.rb

@@ -0,0 +1,26 @@
+#! /usr/local/bin/ruby
+
+## DOMHASH test
+## 1999 by yoshidam
+##
+## Namespace support required
+##
+## Apr 20, 1999 Change for draft-hiroshi-dom-hash-01.txt
+##
+
+require 'xml/dom/builder'
+require 'xml/dom/digest'
+
+p = XML::DOM::Builder.new(0, nil, ":") ## nssep must be ':'
+if p.respond_to?(:setParamEntityParsing)
+  p.setParamEntityParsing(XML::Parser::PARAM_ENTITY_PARSING_UNLESS_STANDALONE)
+end
+begin
+  tree = p.parse($<.read, true)
+rescue XML::ParserError
+  print "#{$<.filename}:#{p.line}: #{$!}\n"
+  exit
+end
+tree.documentElement.normalize
+tree.documentElement.getDigest.each_byte { |c|; print "%02X" % c }
+print "\n"

data/samples/digesttest2.rb

@@ -0,0 +1,192 @@
+#! /usr/local/bin/ruby
+
+## DOMHASH without DOM tree
+## 1999 by yoshidam
+##
+## Namespace support required
+##
+
+require 'xml/parser'
+require 'md5'
+#require 'uconv'
+
+module XML
+  class ExtEntParser < Parser
+    def initialize(parent, *rest)
+      super
+      @parent = parent
+    end
+
+    def startElement(name, attr)
+      @parent.startElement(name, attr)
+    end
+
+    def endElement(name)
+      @parent.endElement(name)
+    end
+
+    def character(data)
+      @parent.character(data)
+    end
+
+    def processingInstruction(target, data)
+      @parent.processingInstruction(target, data)
+    end
+
+    def comment(data)
+      @parent.comment(data)
+    end
+
+    def externalEntityRef(context, base, systemId, publicId)
+      extp = ExtEntParser.new(self, context)
+      begin
+        tree = extp.parse(open(systemId).read)
+      rescue XML::ParserError
+        raise XML::ParserError.new("#{systemId}(#{extp.line}): #{$!}")
+      rescue Errno::ENOENT
+        raise Errno::ENOENT.new("#{$!}")
+      end
+      extp.done
+    end
+  end
+
+  class DigestParser < Parser
+    NODE_NODE = 0
+    ELEMENT_NODE = 1
+    ATTRIBUTE_NODE = 2
+    TEXT_NODE = 3
+    CDATA_SECTION_NODE = 4
+    ENTITY_REFERENCE_NODE = 5
+    ENTITY_NODE = 6
+    PROCESSING_INSTRUCTION_NODE = 7
+    COMMENT_NODE  = 8
+    DOCUMENT_NODE = 9
+    DOCUMENT_TYPE_NODE = 10
+    DOCUMENT_FRAGMENT_NODE = 11
+    NOTATION_NODE = 12
+
+    def initialize(*rest)
+      super
+      @elem_stack = []
+      @elem_data = [ "#document", [], [] ]
+      @text = ''
+      @root = nil
+    end
+
+    ## convert UTF-8 into UTF-16BE
+    def tou16(str)
+#      Uconv.u16swap(Uconv.u8tou16(str))
+      str.unpack("U*").pack("n*")
+    end
+
+    ## create digest value for the text node
+    def textDigest(text)
+      MD5.new([TEXT_NODE].pack("N") + tou16(text)).digest
+    end
+
+    ## create digest value for the element  node
+    def elementDigest(name, attrs, children)
+      MD5.new([ELEMENT_NODE].pack("N") +
+              tou16(name) +
+              "\0\0" +
+              [attrs.length].pack("N") +
+              attrs.join +
+              [children.length].pack("N") +
+              children.join).digest
+    end
+
+    ## create digest value for the attribute node
+    def attrDigest(name, value)
+      MD5.new([ATTRIBUTE_NODE].pack("N") +
+              tou16(name) + "\0\0" + tou16(value)).digest
+    end
+
+    def processingInstructionDigest(target, data)
+      MD5.new([PROCESSING_INSTRUCTION_NODE].pack("N") +
+              tou16(target) + "\0\0" + tou16(data)).digest
+    end
+
+    ## flush a bufferd text
+    def flushText
+      if @text.length > 0
+        @elem_data[2].push(textDigest(@text))
+        @text = ''
+      end
+    end
+
+    ## start element handler
+    def startElement(name, attr)
+      flushText
+      @elem_stack.push(@elem_data)
+      attr_digests = []
+      attr_array = attr.sort {|a, b|
+        tou16(a[0]) <=> tou16(b[0])
+      }
+      attr_array.each {|a|
+        attr_digests.push(attrDigest(a[0], a[1]))
+      }
+      @elem_data = [name, attr_digests, []]
+    end
+
+    ## end element handler
+    def endElement(name)
+      flushText
+      digest = elementDigest(*@elem_data)
+      @elem_data = @elem_stack.pop
+      @elem_data[2].push(digest)
+
+      ## digest for root element
+      if @elem_stack.length == 0
+        @root = digest
+      end
+    end
+
+    ## character data handler
+    def character(data)
+      ## Character data must be concatenated because expat split a text
+      ## node into some fragments.
+      @text << data
+    end
+
+    ## PI handler
+    def processingInstruction(target, data)
+      flushText
+      @elem_data[2].push(processingInstructionDigest(target, data))
+    end
+
+    ## comment handler
+    def comment(data)
+      flushText
+      ## ignore comment node
+    end
+
+    def externalEntityRef(context, base, systemId, publicId)
+      extp = ExtEntParser.new(self, context)
+      begin
+        tree = extp.parse(open(systemId).read)
+      rescue XML::ParserError
+        raise XML::ParserError.new("#{systemId}(#{extp.line}): #{$!}")
+      rescue Errno::ENOENT
+        raise Errno::ENOENT.new("#{$!}")
+      end
+      extp.done
+    end
+
+    def getRootDigest
+      @root
+    end
+  end
+end
+
+p = XML::DigestParser.new(nil, ":") ## nssep must be ':'
+if p.respond_to?(:setParamEntityParsing)
+  p.setParamEntityParsing(XML::Parser::PARAM_ENTITY_PARSING_UNLESS_STANDALONE)
+end
+begin
+  p.parse($<.read)
+rescue XML::ParserError
+  print "#{$<.filename}:#{p.line}: #{$!}\n"
+  exit
+end
+p.getRootDigest.each_byte { |c|; print "%02X" % c }
+print "\n"

data/samples/doctype.rb

@@ -0,0 +1,40 @@
+#! /usr/local/bin/ruby
+
+require 'xml/parser'
+
+p = XML::Parser.new(nil, "!")
+p.setReturnNSTriplet(true);
+
+def p.startDoctypeDecl(name, sysid, pubid, has_internal)
+  p(["startDoctypeDecl", sysid, pubid, has_internal])
+end
+
+def p.endDoctypeDecl
+  p(["endDoctypeDecl"])
+end
+
+def p.elementDecl(name, model)
+  p(["elementDecl", name, model])
+end
+
+def p.attlistDecl(elname, attname, att_type, dflt, isreq)
+  p(["attlistDecl", elname, attname, att_type, dflt, isreq])
+end
+
+def p.xmlDecl(version, enc, standalone)
+  p(["xmlDecl", version, enc, standalone])
+end
+
+def p.entityDecl(name, param, value, base, sysid, pubid, notation)
+  p(["entityDecl", name, param, value, base, sysid, pubid, notation])
+end
+
+def p.startElement(name, attrs)
+  p(["startElement", name, attrs])
+end
+
+def p.endElement(name)
+  p(["endElement", name])
+end
+
+p.parse($<.read)

data/samples/doctype.xml

@@ -0,0 +1,21 @@
+<?xml version="1.0"?>
+<!DOCTYPE hoge PUBLIC "-//Hoge Version 1.0//JA"
+                      "http://www.yoshidam.net/hoge/1.0/hoge.dtd" [
+<!ELEMENT hoge (hoge:a|hoge:b)>
+<!ATTLIST hoge aaa (AAAA|BBBB|CCCC) 'AAAA'
+               bbb CDATA #IMPLIED
+               ccc ID #IMPLIED
+               ddd NMTOKEN 'aaa'
+               eee NMTOKENS #REQUIRED>
+<!ELEMENT hoge:a EMPTY >
+<!ELEMENT hoge:b ANY >
+<!ELEMENT hoge:c (hoge:b?)>
+<!ELEMENT d (#PCDATA|hoge:a|hoge:b|hoge:c)*>
+<!ELEMENT e (hoge:a|hoge:b|hoge:c)?>
+<!ENTITY aaa "AAAAAAA">
+]>
+
+<hoge:ahoge xmlns:hoge="http://www.yoshidam.net/hoge" eee='a b c'>
+<aaa:aaa xmlns:aaa="http://aaaa" aaa:hoge="aaa"/>
+hoge
+</hoge:ahoge>

data/samples/doctypei.rb

@@ -0,0 +1,19 @@
+#! /usr/local/bin/ruby
+
+require 'xml/parser'
+
+p = XML::Parser.new(nil, " ")
+p.setReturnNSTriplet(true);
+
+def p.startDoctypeDecl(name, sysid, pubid, has_internal) end
+def p.endDoctypeDecl() end
+def p.elementDecl(name, model) end
+def p.attlistDecl(name, elname, attname, att_type, dflt, isreq) end
+def p.xmlDecl(version, enc, standalone) end
+def p.entityDecl(name, param, value, base, sysid, pubid, notation) end
+def p.startElement(name, attrs) end
+def p.endElement(name) end
+
+p.parse($<.read) do |t, n, d|
+  p([t, n, d])
+end

data/samples/document.dtd

@@ -0,0 +1,77 @@
+<!-- <!DOCTYPE document [ -->
+
+<!ENTITY % coreattrs "id ID #IMPLIED">
+<!ENTITY % phrase "em|code">
+<!ENTITY % special "link|image">
+<!ENTITY % inline "#PCDATA|%phrase;|%special;">
+<!ENTITY % block "para|rpara|list|pre|table">
+<!ENTITY % flow "%inline;|%block;">
+
+<!ELEMENT document (title,(index|contents|section|%block;)*)>
+<!ATTLIST document %coreattrs;>
+
+<!ELEMENT em (%inline;)*>
+<!ATTLIST em %coreattrs;>
+
+<!ELEMENT code (%inline;)*>
+<!ATTLIST code %coreattrs;>
+
+<!ELEMENT title (%inline;)*>
+<!ATTLIST title %coreattrs;>
+
+<!ELEMENT para (%inline;)*>
+<!ATTLIST para %coreattrs;>
+
+<!ELEMENT rpara (%inline;)*>
+<!ATTLIST rpara %coreattrs;>
+
+<!ELEMENT contents (item)+>
+<!ATTLIST contents %coreattrs;>
+
+<!ELEMENT index (link)+>
+<!ATTLIST index %coreattrs;>
+
+<!ELEMENT pre (%inline;)*>
+<!ATTLIST pre %coreattrs;>
+
+<!ELEMENT section (heading,(%block;|subsection)*)>
+<!ATTLIST section %coreattrs;>
+
+<!ELEMENT subsection (heading,(%block;)*)>
+<!ATTLIST subsection %coreattrs;>
+
+<!ELEMENT heading (%inline;)*>
+<!ATTLIST heading %coreattrs;>
+
+<!ELEMENT link (%inline;)*>
+<!ATTLIST link %coreattrs;>
+<!ATTLIST link xml:link CDATA #FIXED 'simple'
+               href CDATA #REQUIRED>
+
+<!ELEMENT image EMPTY>
+<!ATTLIST image %coreattrs;
+                xml:link CDATA #FIXED 'simple'
+                href CDATA #REQUIRED
+                title CDATA #REQUIRED
+                role CDATA #IMPLIED
+                show CDATA #IMPLIED>
+
+<!ELEMENT list (item)+>
+<!ATTLIST list %coreattrs;>
+
+<!ELEMENT item (%flow;)*>
+<!ATTLIST item %coreattrs;>
+
+<!ELEMENT table (caption?,tr+)>
+<!ATTLIST table %coreattrs;>
+
+<!ELEMENT tr (td)+>
+<!ATTLIST tr %coreattrs;>
+
+<!ELEMENT td (%flow;)*>
+<!ATTLIST td %coreattrs;>
+
+<!ELEMENT caption (%inline;)*>
+<!ATTLIST caption %coreattrs;>
+
+<!-- ]> -->