PowerPointCreator

// -------------------------------------------------------
// PowerPointCreator by Elm�Soft
// www.netcult.ch/elmue
// -------------------------------------------------------

using System;
using System.Collections;
using System.Text;

namespace PowerPointCreator
{
    /// <summary>
    /// This class parses XML and HTML code and corrects errors
    /// </summary>
    public class HtmlParser
    {
        private class HtmlNode
        {
            public string ms_Open = "";  // e.g. "<Font" or "</Font" or "<?XML"
            public string ms_Close = ""; // e.g. ">" or "/>" or "?>"
            public Hashtable mi_Attributes = new Hashtable();

            public void ToString(StringBuilder s_Html)
            {
                s_Html.Append(ms_Open);
                foreach (string s_Attrib in mi_Attributes.Keys)
                {
                    // The quotation marks are required for XML
                    s_Html.AppendFormat(" {0}=\"{1}\"", s_Attrib, mi_Attributes[s_Attrib]);
                }
                s_Html.Append(ms_Close);
            }
        }

        /// <summary>
        /// HTML --> XML
        /// This function avoids errors when loading HTML code into the XML parser.
        /// XML uses a much more strict syntax than HTML.
        /// A single <BR> is not possible in XML. All tags must be closed!
        /// A <P> without a following </P> cuases a XML syntax error exception.
        /// <FONT size=1> must be replaced with <FONT size="1">
        /// This is required to feed HTML code into the XML parser
        /// </summary>
        public static string HtmlToXml(string s_Html)
        {
            s_Html = Functions.Replace(s_Html, "<BR>", "<BR />");
            s_Html = Functions.Replace(s_Html, "</P>", "");
            s_Html = Functions.Replace(s_Html, "<P>", "<BR /><BR />");

            // This is required to avoid errors from the stupid XML parser:
            // "Undeclared entity 'nbsp'" exception.
            // See http://blogs.pingpoet.com/overflow/archive/2005/07/20/6607.aspx
            s_Html = Functions.Replace(s_Html, "&nbsp;", "#nbsp;");
        
            s_Html += "    "; // allows accessing s_Html[s32_Pos+2] without length check
            int s32_Pos = 0;

            StringBuilder s_Out = new StringBuilder(s_Html.Length * 2);
            while (true)
            {
                // find start of next tag
                int s32_Open = s_Html.IndexOf("<", s32_Pos);
                if (s32_Open < 0)
                {
                    // Append the rest of the html code
                    s_Out.Append(s_Html.Substring(s32_Pos));
                    break;
                }

                int s32_End = s32_Open - 1;

                // insert text which is contained between neighboured tags
                if (s32_End >= s32_Pos)
                {
                    s_Out.Append(s_Html.Substring(s32_Pos, s32_End - s32_Pos + 1));
                }

                // read the tag and all it's arguments (e.g. Color=red)
                s32_Pos = s32_Open;
                HtmlNode i_Node = GetNextTag(ref s_Html, ref s32_Pos);

                i_Node.ToString(s_Out);
            }
            return s_Out.ToString();
        }

        /// <summary>
        /// Retrieves the next Html tag
        /// Start position must be on the "<" character of the tag
        /// returns with position set behind the ">" character of the tag
        /// </summary>
        private static HtmlNode GetNextTag(ref string s_Html, ref int s32_Pos)
        {
            HtmlNode i_Node = new HtmlNode();

            int s32_First = s32_Pos;
            s32_Pos++;

            // Special case comments: don't parse !!   
            if (s_Html[s32_Pos] == '!' && s_Html[s32_Pos + 1] == '-' && s_Html[s32_Pos + 2] == '-')
            {
                s32_Pos = s_Html.IndexOf(">", s32_Pos);
                if (s32_Pos < 0)
                    s32_Pos = s_Html.Length;
                else
                    s32_Pos++;

                i_Node.ms_Open = s_Html.Substring(s32_First, s32_Pos - s32_First);
                return i_Node;
            }

            // count valid tag name characters up to first space or ">"
            while (s32_Pos + 1 < s_Html.Length && !Char.IsWhiteSpace(s_Html[s32_Pos]) && s_Html[s32_Pos] != '>')
            {
                s32_Pos++;
            }
            
            int s32_Last = s32_Pos;
            i_Node.ms_Open = s_Html.Substring(s32_First, s32_Last - s32_First);

            // read arguments
            bool b_Quoted = false;
            string s_Attr = null;
            while (s32_Pos < s_Html.Length)
            {
                if (!b_Quoted)
                {
                    int s32_End = 0;
                    if (s_Html[s32_Pos] == '>') s32_End = 1;
                    if (s_Html[s32_Pos] == '/' && s_Html[s32_Pos + 1] == '>') s32_End = 2;
                    if (s_Html[s32_Pos] == '?' && s_Html[s32_Pos + 1] == '>') s32_End = 2;

                    if (s32_End > 0 ||
                        (s_Attr == null && s_Html[s32_Pos] == '=') ||
                        (s_Attr != null && Char.IsWhiteSpace(s_Html[s32_Pos])))
                    {
                        string s_Part = s_Html.Substring(s32_Last, s32_Pos - s32_Last);
                        
                        // "\t \" test \" \r\n" ---> " test "
                        s_Part = s_Part.Trim().Trim(new char[]{'\"'});

                        if (s_Html[s32_Pos] == '=') s32_Pos++;

                        if (s_Attr == null)
                            s_Attr = s_Part;
                        else
                        {
                            i_Node.mi_Attributes[s_Attr] = s_Part;
                            s_Attr = null;
                        }
                        s32_Last = s32_Pos;
                    }

                    if (s32_End > 0)
                    {
                        i_Node.ms_Close = s_Html.Substring(s32_Pos, s32_End);
                        s32_Pos += s32_End;
                        break;
                    }
                }

                if (s_Html[s32_Pos] == '"')
                    b_Quoted = !b_Quoted;

                s32_Pos++;
            }
            return i_Node;
        }
    }
}