8 files changed, 481 insertions, 0 deletions

diff --git a/src/inprogress/xmldocument.cpp b/src/inprogress/xmldocument.cpp
new file mode 100644
index 0000000..cb21826
--- /dev/null
+++ b/src/inprogress/xmldocument.cpp
@@ -0,0 +1,9 @@
+#include "xmldocument.h"
+
+Bu::XmlDocument::XmlDocument()
+{
+}
+
+Bu::XmlDocument::~XmlDocument()
+{
+}
diff --git a/src/inprogress/xmldocument.h b/src/inprogress/xmldocument.h
new file mode 100644
index 0000000..e16e3ea
--- /dev/null
+++ b/src/inprogress/xmldocument.h
@@ -0,0 +1,22 @@
+#ifndef XML_DOCUMENT_H
+#define XML_DOCUMENT_H
+
+#include <stdint.h>
+
+namespace Bu
+{
+        /**
+         *
+         */
+        class XmlDocument
+        {
+        public:
+                XmlDocument();
+                virtual ~XmlDocument();
+
+        private:
+
+        };
+}
+
+#endif
diff --git a/src/inprogress/xmlnode.cpp b/src/inprogress/xmlnode.cpp
new file mode 100644
index 0000000..58ef5c5
--- /dev/null
+++ b/src/inprogress/xmlnode.cpp
@@ -0,0 +1,9 @@
+#include "xmlnode.h"
+
+Bu::XmlNode::XmlNode()
+{
+}
+
+Bu::XmlNode::~XmlNode()
+{
+}
diff --git a/src/inprogress/xmlnode.h b/src/inprogress/xmlnode.h
new file mode 100644
index 0000000..cd9961a
--- /dev/null
+++ b/src/inprogress/xmlnode.h
@@ -0,0 +1,22 @@
+#ifndef XML_NODE_H
+#define XML_NODE_H
+
+#include <stdint.h>
+
+namespace Bu
+{
+        /**
+         *
+         */
+        class XmlNode
+        {
+        public:
+                XmlNode();
+                virtual ~XmlNode();
+
+        private:
+
+        };
+}
+
+#endif
diff --git a/src/inprogress/xmlreader.cpp b/src/inprogress/xmlreader.cpp
new file mode 100644
index 0000000..bd241cf
--- /dev/null
+++ b/src/inprogress/xmlreader.cpp
@@ -0,0 +1,267 @@
+#include "xmlreader.h"
+
+Bu::XmlReader::XmlReader( Bu::Stream &sIn ) :
+        sIn( sIn )
+{
+}
+
+Bu::XmlReader::~XmlReader()
+{
+}
+
+const char *Bu::XmlReader::lookahead( int nAmnt )
+{
+        if( sBuf.getSize() >= nAmnt )
+                return sBuf.getStr();
+
+        int nNew = nAmnt - sBuf.getSize();
+        char *buf = new char[nNew];
+        sIn.read( buf, nNew );
+        sBuf.append( buf );
+
+        return sBuf.getStr();
+}
+
+void Bu::XmlReader::burn( int nAmnt )
+{
+        if( sBuf.getSize() < nAmnt )
+        {
+                lookahead( nAmnt );
+        }
+
+        //sBuf.remove( nAmnt );
+}
+
+void Bu::XmlReader::checkString( const char *str, int nLen )
+{
+        if( !strncmp( str, lookahead( nLen ), nLen ) )
+        {
+                burn( nLen );
+                return;
+        }
+
+        throw Bu::ExceptionBase("Expected string '%s'", str );
+}
+
+Bu::XmlNode *Bu::XmlReader::read()
+{
+        prolog();
+}
+
+void Bu::XmlReader::prolog()
+{
+        XMLDecl();
+        Misc();
+}
+
+void Bu::XmlReader::XMLDecl()
+{
+        checkString("<?xml", 5 );
+        S();
+        VersionInfo();
+        EncodingDecl();
+        SDDecl();
+        Sq();
+        checkString("?>", 2 );
+}
+
+void Bu::XmlReader::Misc()
+{
+        for(;;)
+        {
+                S();
+                if( !strncmp("<!--", lookahead( 4 ), 4 ) )
+                {
+                        Comment();
+                }
+                else if( !strncmp("<?", lookahead( 2 ), 2 ) )
+                {
+                        PI();
+                }
+                else
+                {
+                        return;
+                }
+        }
+}
+
+void Bu::XmlReader::Comment()
+{
+        checkString("<!--", 4 );
+        for(;;)
+        {
+                unsigned char c = *lookahead(1);
+                if( c == '-' )
+                {
+                        if( lookahead(2)[1] == '-' )
+                        {
+                                checkString("-->", 3 );
+                                return;
+                        }
+                }
+                burn( 1 );
+        }
+}
+
+void Bu::XmlReader::PI()
+{
+        checkString("<?", 2 );
+        FString sName = Name();
+        printf("PI: %s\n---\n", sName.getStr() );
+        S();
+        for(int j = 0;; j++ )
+        {
+                if( !strncmp( "?>", lookahead(j+2)+j, 2 ) )
+                {
+                        burn( j+2 );
+                        return;
+                }
+        }
+}
+
+void Bu::XmlReader::S()
+{
+        for( int j = 0;; j++ )
+        {
+                char c = *lookahead( 1 );
+                if( c == 0x20 || c == 0x9 || c == 0xD || c == 0xA )
+                        continue;
+                if( j == 0 )
+                        throw ExceptionBase("Expected whitespace.");
+                return;
+        }
+}
+
+void Bu::XmlReader::Sq()
+{
+        for(;;)
+        {
+                char c = *lookahead( 1 );
+                if( c == 0x20 || c == 0x9 || c == 0xD || c == 0xA )
+                        continue;
+                return;
+        }
+}
+
+void Bu::XmlReader::VersionInfo()
+{
+        try
+        {
+                S();
+                checkString("version", 7 );
+        }
+        catch( ExceptionBase &e )
+        {
+                return;
+        }
+        Eq();
+        Bu::FString ver = AttValue();
+        if( ver != "1.1" )
+                throw ExceptionBase("Currently we only support xml version 1.1\n");
+}
+
+void Bu::XmlReader::Eq()
+{
+        Sq();
+        checkString("=", 1 );
+        Sq();
+}
+
+void Bu::XmlReader::EncodingDecl()
+{
+        S();
+        try
+        {
+                checkString("encoding", 8 );
+        }
+        catch( ExceptionBase &e )
+        {
+                return;
+        }
+
+        Eq();
+        AttValue();
+}
+
+void Bu::XmlReader::SDDecl()
+{
+        S();
+        try
+        {
+                checkString("standalone", 10 );
+        }
+        catch( ExceptionBase &e )
+        {
+                return;
+        }
+
+        Eq();
+        AttValue();
+}
+
+Bu::FString Bu::XmlReader::AttValue()
+{
+        char q = *lookahead(1);
+        if( q == '\"' )
+        {
+                for( int j = 2;; j++ )
+                {
+                        if( lookahead(j)[j-1] == '\"' )
+                        {
+                                Bu::FString ret( lookahead(j)+1, j-2 );
+                                burn( j );
+                                return ret;
+                        }
+                }
+        }
+        else if( q == '\'' )
+        {
+                for( int j = 2;; j++ )
+                {
+                        if( lookahead(j)[j-1] == '\'' )
+                        {
+                                Bu::FString ret( lookahead(j)+1, j-2 );
+                                burn( j );
+                                return ret;
+                        }
+                }
+        }
+
+        throw ExceptionBase("Excpected either \' or \".\n");
+}
+
+Bu::FString Bu::XmlReader::Name()
+{
+        unsigned char c = *lookahead( 1 );
+        if( c != ':' && c != '_' &&
+                (c < 'A' || c > 'Z') &&
+                (c < 'a' || c > 'z') &&
+                (c < 0xC0 || c > 0xD6 ) &&
+                (c < 0xD8 || c > 0xF6 ) &&
+                (c < 0xF8))
+        {
+                throw ExceptionBase("Invalid entity name starting character.");
+        }
+
+        for( int j = 1;; j++ )
+        {
+                unsigned char c = lookahead(j+1)[j];
+                if( isS( c ) )
+                {
+                        FString ret( lookahead(j+1), j+1 );
+                        burn( j+1 );
+                        return ret;
+                }
+                if( c != ':' && c != '_' && c != '-' && c != '.' && c != 0xB7 &&
+                        (c < 'A' || c > 'Z') &&
+                        (c < 'a' || c > 'z') &&
+                        (c < '0' || c > '9') && 
+                        (c < 0xC0 || c > 0xD6 ) &&
+                        (c < 0xD8 || c > 0xF6 ) &&
+                        (c < 0xF8))
+                {
+                        throw ExceptionBase("Invalid character in name.");
+                }
+        }
+}
+
diff --git a/src/inprogress/xmlreader.h b/src/inprogress/xmlreader.h
new file mode 100644
index 0000000..708a386
--- /dev/null
+++ b/src/inprogress/xmlreader.h
@@ -0,0 +1,121 @@
+#ifndef XML_READER_H
+#define XML_READER_H
+
+#include <stdint.h>
+#include "bu/stream.h"
+#include "bu/fstring.h"
+#include "bu/xmlnode.h"
+
+namespace Bu
+{
+        /**
+         * An Xml 1.1 reader.  I've decided to write this, this time, based on the
+         * official W3C reccomendation, now included with the source code.  I've
+         * named the productions in the parser states the same as in that document,
+         * which may make them easier to find, etc, although possibly slightly less
+         * optimized than writing my own reduced grammer.
+         *
+         * Below I will list differences between my parser and the official standard
+         * as I come up with them.
+         *  - Encoding and Standalone headings are ignored for the moment. (4.3.3,
+         *    2.9)
+         *  - The standalone heading attribute can have any standard whitespace
+         *    before it (the specs say only spaces, no newlines). (2.9)
+         *  - Since standalone is ignored, it is currently allowed to have any
+         *    value (should be restricted to "yes" or "no"). (2.9)
+         *  - Currently only UTF-8 / ascii are parsed.
+         *  - [optional] The content of comments is thrown away. (2.5)
+         *  - The content of processing instruction blocks is parsed properly, but
+         *    thrown away. (2.6)
+         */
+        class XmlReader
+        {
+        public:
+                XmlReader( Bu::Stream &sIn );
+                virtual ~XmlReader();
+
+                XmlNode *read();
+
+        private:
+                Bu::Stream &sIn;
+                Bu::FString sBuf;
+
+        private: // Helpers
+                const char *lookahead( int nAmnt );
+                void burn( int nAmnt );
+                void checkString( const char *str, int nLen );
+
+        private: // States
+                /**
+                 * The headers, etc.
+                 */
+                void prolog();
+
+                /**
+                 * The xml decleration (version, encoding, etc).
+                 */
+                void XMLDecl();
+
+                /**
+                 * Misc things, Includes Comments and PIData (Processing Instructions).
+                 */
+                void Misc();
+
+                /**
+                 * Comments
+                 */
+                void Comment();
+
+                /**
+                 * Processing Instructions
+                 */
+                void PI();
+
+                /**
+                 * Whitespace eater.
+                 */
+                void S();
+
+                /**
+                 * Optional whitespace eater.
+                 */
+                void Sq();
+
+                /**
+                 * XML Version spec
+                 */
+                void VersionInfo();
+
+                /**
+                 * Your basic equals sign with surrounding whitespace.
+                 */
+                void Eq();
+
+                /**
+                 * Read in an attribute value.
+                 */
+                FString AttValue();
+
+                /**
+                 * Read in the name of something.
+                 */
+                FString Name();
+
+                /**
+                 * Encoding decleration in the header
+                 */
+                void EncodingDecl();
+
+                /**
+                 * Standalone decleration in the header
+                 */
+                void SDDecl();
+
+                bool isS( unsigned char c )
+                {
+                        return ( c == 0x20 || c == 0x9 || c == 0xD || c == 0xA );
+                }
+        };
+}
+
+#endif
diff --git a/src/inprogress/xmlwriter.cpp b/src/inprogress/xmlwriter.cpp
new file mode 100644
index 0000000..23a5175
--- /dev/null
+++ b/src/inprogress/xmlwriter.cpp
@@ -0,0 +1,9 @@
+#include "xmlwriter.h"
+
+Bu::XmlWriter::XmlWriter()
+{
+}
+
+Bu::XmlWriter::~XmlWriter()
+{
+}
diff --git a/src/inprogress/xmlwriter.h b/src/inprogress/xmlwriter.h
new file mode 100644
index 0000000..796d6fb
--- /dev/null
+++ b/src/inprogress/xmlwriter.h
@@ -0,0 +1,22 @@
+#ifndef XML_WRITER_H
+#define XML_WRITER_H
+
+#include <stdint.h>
+
+namespace Bu
+{
+        /**
+         *
+         */
+        class XmlWriter
+        {
+        public:
+                XmlWriter();
+                virtual ~XmlWriter();
+
+        private:
+
+        };
+}
+
+#endif