From 3d3fbaedcebf342d088c46eec7773dfa88011ae4 Mon Sep 17 00:00:00 2001
From: jez <>
Date: Tue, 7 Aug 2007 21:29:27 +0000
Subject: [PATCH] pass through skippedEntity and processingInstruction

---
 include/SAX/filter/PYXWriter.hpp              | 114 ++++++++++++++++++
 .../SAX/filter/WhitespaceStripperFilter.hpp   |  41 +++++++
 include/SAX/filter/Writer.h                   |   4 +
 3 files changed, 159 insertions(+)
 create mode 100644 include/SAX/filter/PYXWriter.hpp
 create mode 100644 include/SAX/filter/WhitespaceStripperFilter.hpp

diff --git a/include/SAX/filter/PYXWriter.hpp b/include/SAX/filter/PYXWriter.hpp
new file mode 100644
index 00000000..d7349748
--- /dev/null
+++ b/include/SAX/filter/PYXWriter.hpp
@@ -0,0 +1,114 @@
+#ifndef ARABICA_SAX_PYX_WRITER_H
+#define ARABICA_SAX_PYX_WRITER_H
+
+#include <SAX/ArabicaConfig.h>
+#include <SAX/helpers/XMLFilterImpl.h>
+#include <XML/UnicodeCharacters.h>
+#include <ostream>
+
+namespace SAX {
+
+template<class string_type>
+class PYXWriter : public basic_XMLFilterImpl<string_type>
+{
+  public:
+    typedef string_type stringT;
+    typedef PYXWriter<stringT> PYXWriterT;
+    typedef typename string_type::value_type charT;
+    typedef typename string_type::traits_type traitsT;
+    typedef std::basic_ostream<charT, traitsT> ostreamT;
+    typedef basic_XMLReader<stringT> XMLReaderT;
+    typedef basic_XMLFilterImpl<stringT> XMLFilterT;
+    typedef typename basic_XMLFilterImpl<stringT>::AttributesT AttributesT;
+    typedef Arabica::Unicode<charT> UnicodeT;
+  private:
+    using XMLFilterT::getParent;
+
+  public:
+    PYXWriter(ostreamT& stream) :
+      XMLFilterT(),
+      stream_(&stream)
+    {
+    } // PYXWriter
+
+    PYXWriter(ostreamT& stream, XMLReaderT& parent) :
+      XMLFilterT(parent),
+      stream_(&stream)
+    {
+    } // PYXWriter
+
+  public:
+    // ContentHandler
+    virtual void startElement(const stringT& namespaceURI, const stringT& localName,
+                              const stringT& qName, const AttributesT& atts);
+    virtual void endElement(const stringT& namespaceURI, const stringT& localName,
+                            const stringT& qName);
+    virtual void characters(const stringT& ch);
+    virtual void processingInstruction(const stringT& target, const stringT& data);
+
+  private: 
+    void escape(const stringT& ch);
+
+    ostreamT* stream_;
+}; // class PYXWriter
+
+template<class string_type>
+void PYXWriter<string_type>::startElement(
+                              const stringT& namespaceURI, const stringT& localName,
+                              const stringT& qName, const AttributesT& atts)
+{ 
+  *stream_ << UnicodeT::LEFT_PARENTHESIS << localName << std::endl;
+  for(int i = 0; i < atts.getLength(); ++i)
+  {
+    *stream_ << UnicodeT::CAPITAL_A << atts.getLocalName(i) 
+             << UnicodeT::SPACE;
+    escape(atts.getValue(i));
+    *stream_ << std::endl; 
+  }
+
+  XMLFilterT::startElement(namespaceURI, localName, qName, atts);
+} // startElement
+
+template<class string_type>
+void PYXWriter<string_type>::endElement(
+                            const stringT& namespaceURI, const stringT& localName,
+                            const stringT& qName)
+{
+  *stream_ << UnicodeT::RIGHT_PARENTHESIS << localName << std::endl;
+
+  XMLFilterT::endElement(namespaceURI, localName, qName);
+} // endElement
+
+template<class string_type>
+void PYXWriter<string_type>::characters(const stringT& ch)
+{
+  *stream_ << UnicodeT::HYPHEN_MINUS;
+  escape(ch);
+  *stream_ << std::endl;
+
+  XMLFilterT::characters(ch);
+} // characters
+
+template<class string_type>
+void PYXWriter<string_type>::processingInstruction(const stringT& target, const stringT& data)
+{
+  *stream_ << UnicodeT::QUESTION_MARK << target
+           << UnicodeT::SPACE << data
+           << std::endl;
+
+  XMLFilterT::processingInstruction(target, data);
+} // processingInstruction
+
+template<class string_type>
+void PYXWriter<string_type>::escape(const stringT& ch)
+{
+  for(typename stringT::const_iterator s = ch.begin(), se = ch.end(); s != se; ++s)
+    if(*s == UnicodeT::LINE_FEED)
+      *stream_ << UnicodeT::BACK_SLASH << UnicodeT::LOWERCASE_N;
+    else
+      *stream_ << *s;
+} // escape
+
+} // namespace SAX
+
+#endif
diff --git a/include/SAX/filter/WhitespaceStripperFilter.hpp b/include/SAX/filter/WhitespaceStripperFilter.hpp
new file mode 100644
index 00000000..7a62bea9
--- /dev/null
+++ b/include/SAX/filter/WhitespaceStripperFilter.hpp
@@ -0,0 +1,41 @@
+#ifndef WHITESPACE_STRIPPER_FILTER_HPP
+#define WHITESPACE_STRIPPER_FILTER_HPP
+
+#include <SAX/helpers/XMLFilterImpl.h>
+
+#include <Utils/normalize_whitespace.hpp>
+
+namespace SAX
+{
+
+template<class string_type, class string_adaptor = Arabica::default_string_adaptor<string_type> >
+class WhitespaceStripper : public SAX::basic_XMLFilterImpl<string_type>
+{
+  public:
+    typedef string_type stringT;
+    typedef SAX::basic_XMLFilterImpl<stringT> baseT;
+    typedef SAX::basic_XMLReader<stringT> XMLReaderT;
+
+    WhitespaceStripper() : 
+      baseT() 
+    { 
+    } // WhitespaceStripper
+
+    WhitespaceStripper(XMLReaderT& parent) :
+      baseT(parent)
+    {
+    } // WhitespaceStripper    
+
+    virtual void characters(const stringT& ch)
+    { 
+      baseT::characters(Arabica::string::normalize_whitespace<string_type, string_adaptor>(ch));
+    } // characters
+
+    virtual void ignorableWhitespace(const stringT& ch)
+    {
+    } // ignorableWhitespace
+}; // class WhitespaceStripper
+
+} // namespace SAX
+#endif
+// end of file
diff --git a/include/SAX/filter/Writer.h b/include/SAX/filter/Writer.h
index e33f80ce..4192f7b7 100644
--- a/include/SAX/filter/Writer.h
+++ b/include/SAX/filter/Writer.h
@@ -286,6 +286,8 @@ void basic_Writer<string_type>::processingInstruction(const stringT& target, con
     *stream_ << UnicodeT::QUESTION_MARK
              << UnicodeT::GREATER_THAN_SIGN;
   }
+
+  XMLFilterT::processingInstruction(target, data);
 } // processingInstruction
 
 template<class string_type>
@@ -293,6 +295,8 @@ void basic_Writer<string_type>::skippedEntity(const stringT& name)
 {
   if(!isDtd(name))
     *stream_ << UnicodeT::AMPERSAND << name << UnicodeT::SEMI_COLON;
+
+  XMLFilterT::skippedEntity(name);
 } // skippedEntity
 
 template<class string_type>