//  /*--------------------------------------------------------*\
//  |   Main Program: Tag_Cloud_Generator
//  |*--------------------------------------------------------*|
//  |   Date: 2009 June 02
//  |   Author: Michael Kusold & Marko Skugor
//  |
//  |   Brief User's Manual:
//  |   Allows the user to create a html tag cloud with 
//  |   input done by redirection.
//  |
//  \*--------------------------------------------------------*/

///-------------------------------------------------------------
/// Global Context ---------------------------------------------
///-------------------------------------------------------------

#include "RESOLVE_Foundation.h"

#include "CT/Word_Occurrence/Kernel.h"
#include "CT/Sorting_Machine/Kernel_3_C.h"
#include "CI/Text/Are_In_Order_1.h"

///-------------------------------------------------------------
/// Instantiations ---------------------------------------------
///-------------------------------------------------------------

concrete_instance
class Word_Occurrence_Map :
    instantiates
        Word_Occurrence_Kernel <Text>
{};

//--------------------------------------------------------------

concrete_instance
class Word_Occur_Rec :
    instantiates
        Record <
    Text,
    Integer
    >
{};

field_name (Word_Occur_Rec, 0, Text, word);
field_name (Word_Occur_Rec, 1, Integer, count);

///-------------------------------------------------------------
/// Utility Classes - ------------------------------------------
///-------------------------------------------------------------

concrete_instance
utility_class Word_Text_Are_In_Order_1 :
    implements
	abstract_instance General_Are_In_Order <Word_Occur_Rec>
{
public:

    /*!
	math definition ARE_IN_ORDER (
		x: string of character,
		y: string of character
	    ): boolean is
	    x <= y
    !*/

    utility_function_body Boolean Are_In_Order (
	    preserves Word_Occur_Rec& x,
	    preserves Word_Occur_Rec& y
	)
    {
	return (x[word] <= y[word]);
    };

};

//--------------------------------------------------------------

concrete_instance
utility_class Word_Count_Are_In_Order_1 :
    implements
	abstract_instance General_Are_In_Order <Word_Occur_Rec>
{
public:

    /*!
	math definition ARE_IN_ORDER (
		x: string of character,
		y: string of character
	    ): boolean is
	    x <= y
    !*/

    utility_function_body Boolean Are_In_Order (
	    preserves Word_Occur_Rec& x,
	    preserves Word_Occur_Rec& y
	)
    {
	return (x[count] <= y[count]);
    };

};

//--------------------------------------------------------------

concrete_instance
class Sorting_Machine_Text :
    instantiates
         Sorting_Machine_Kernel_3_C <
            Word_Occur_Rec,
            Word_Text_Are_In_Order_1
        >
{};

//--------------------------------------------------------------

concrete_instance
class Sorting_Machine_Count :
    instantiates
         Sorting_Machine_Kernel_3_C <
            Word_Occur_Rec,
            Word_Count_Are_In_Order_1
        >
{};

///-------------------------------------------------------------
/// Global operations ------------------------------------------
///-------------------------------------------------------------

//  global_procedure Get_Line (
//         preserves Character_IStream& input,
// 	produces Text line
// 	);

//  global_procedure Get_Word (
//         alters Text& line,
// 	produces Text word
//      );

 global_procedure Get_Next_Word (
        alters Character_IStream& ins,
        produces Text word
         );

 global_procedure Remove_Non_White (
        alters Text& line
     );
 global_function Boolean Is_Common_Word (
        preserves Text& word
     );

///-------------------------------------------------------------
/// Global operation bodies ------------------------------------
///-------------------------------------------------------------

//  procedure_body Get_Line (
//      preserves Character_IStream& input,
//      produces Text line
//      )
// {
//     object Boolean end = false;
//     while(end == false)
//     {
// 	object Character ch;
	
// 	input.Remove(0,ch);
// 	if(ch=='\n')
// 	{
// 	    end = true;
// 	}
// 	else
// 	{
// 	    line.Add(line.Length(), ch);
// 	}
	
//     }
// }

//  procedure_body Get_Word (
//        alters Text& line,
//        produces Text word
//      )
// {
//     object Boolean end = false;
    
//     while(end == false)
//     {
// 	object Character ch;
	
// 	line.Remove(0,ch);
// 	if(ch=='\n' || ch=='\t' || ch == ' ')
// 	{
// 	    end = true;
// 	    Remove_Non_White(line);
// 	}
// 	else
// 	{
// 	    To_Lower(ch);
// 	    word.Add(word.Length(),ch);
// 	}
//     }
// }

global_procedure_body Get_Next_Word (
        alters Character_IStream& ins,
        produces Text& word
    )
{
    object Boolean done, in_word;

    word.Clear (); // make sure word starts empty

    while (not ins.At_EOS () and not done)
    {
        object Character ch;

        ins.Read (ch); // input one character at a time

        if (Is_Alpha (ch)) // ch is a letter
        {
            in_word = true; // we are now inside a word
            word.Add (word.Length (), ch);
        }
        else // ch is not a letter
        {
            if (in_word) // we were inside a word, so we must be done
            {
                done = true;
            }
        }
    }
}



//  procedure_body Remove_Non_White (
//        alters Text& line
//      )
// {
//     object Boolean end = false;
    
//     while(end==false)
//     {
// 	object Character ch;

// 	line.Remove(0,ch);
	
// 	if(ch=='\n' && ch=='\t' && ch==' ')
// 	{
// 	}
// 	else
// 	{
// 	    line.Add(0,ch);
// 	    end = true;
// 	}
//     }
// }

 function_body Boolean Is_Common_Word (
        preserves Text& word
     )
{
    object Boolean answer = false;
    object Text word_copy = word;

    if(word =="a")
    {
	answer = true;
    }
    else if(word =="and")
    {
	answer = true;
    }
    else if(word == "all")
    {
	answer = true;
    }
    else if(word == "am")
    {
	answer = true;
    }
    else if(word == "an")
    {
	answer = true;
    }
    else if(word == "and")
    {
	answer = true;
    }
    else if(word == "any")
    {
	answer = true;
    }
    else if(word == "are")
    {
	answer = true;
    }
    else if(word == "as")
    {
	answer = true;
    }
    else if(word == "at")
    {
	answer = true;
    }
    else if(word == "be")
    {
	answer = true;
    }
    else if(word == "been")
    {
	answer = true;
    }
    else if(word == "but")
    {
	answer = true;
    }
    else if(word == "by")
    {
	answer = true;
    }
    else if(word == "can")
    {
	answer = true;
    }
    else if(word == "do")
    {
	answer = true;
    }
    else if(word == "for")
    {
	answer = true;
    }
    else if(word == "from")
    {
	answer = true;
    }
    else if(word == "had")
    {
	answer = true;
    }
    else if(word == "has")
    {
	answer = true;
    }
    else if(word == "have")
    {
	answer = true;
    }
    else if(word == "he")
    {
	answer = true;
    }
    else if(word == "her")
    {
	answer = true;
    }
    else if(word == "him")
    {
	answer = true;
    }
    else if(word == "his")
    {
	answer = true;
    }
    else if(word == "i")
    {
	answer = true;
    }
    else if(word == "if")
    {
	answer = true;
    }
    else if(word == "in")
    {
	answer = true;
    }
    else if(word == "is")
    {
	answer = true;
    }
    else if(word == "it")
    {
	answer = true;
    }
    else if(word == "me")
    {
	answer = true;
    } 
    else if(word == "my")
    {
	answer = true;
    }
    else if(word == "not")
    {
	answer = true;
    }
    else if(word == "now")
    {
	answer = true;
    }
    else if(word == "of")
    {
	answer = true;
    }
    else if(word == "on")
    {
	answer = true;
    }
    else if(word == "s")
    {
	answer = true;
    }
    else if(word == "say")
    {
	answer = true;
    } 
    else if(word == "she")
    {
	answer = true;
    }
    else if(word == "should")
    {
	answer = true;
    }
    else if(word == "so")
    {
	answer = true;
    }
    else if(word == "some")
    {
	answer = true;
    }
    else if(word == "t")
    {
	answer = true;
    }
    else if(word == "that")
    {
	answer = true;
    }
    else if(word == "the")
    {
	answer = true;
    }
    else if(word == "there")
    {
	answer = true;
    }
    else if(word == "they")
    {
	answer = true;
    }
    else if(word == "this")
    {
	answer = true;
    }
    else if(word == "to")
    {
	answer = true;
    }
    else if(word == "very")
    {
	answer = true;
    }
    else if(word == "was")
    {
	answer = true;
    }
    else if(word == "we")
    {
	answer = true;
    }
    else if(word == "what")
    {
	answer = true;
    }
    else if(word == "who")
    {
	answer = true;
    }
    else if(word == "will")
    {
	answer = true;
    }
    else if(word == "with")
    {
	answer = true;
    }
    else if(word == "would")
    {
	answer = true;
    }
    else if(word == "you")
    {
	answer = true;
    }
    else if(word == "your")
    {
	answer = true;
    }
    return answer;
}

    
///-------------------------------------------------------------
/// Interface --------------------------------------------------
///-------------------------------------------------------------

program_body main ()
{

    object Word_Occurrence_Map word_occur, final;
   //  object Sorting_Machine_Text sort_word_occur;
//     object Sorting_Machine_Integer sort_final;
    object Character_OStream outs;
    object Character_IStream ins;
    object Character ch;
    object Integer top_words, i;
    object Text tba, line;

    ins.Open_External ("");
    outs.Open_External ("");

    //Set the number of words you want outputted here
    top_words = 100;
    
    //Output Header information
    outs << "<html>\n"
	 << "<head>\n"
	 << "<title>Tag Cloud</title>\n"
	 << "<link href=\"tagcloud.css\" rel=\"stylesheet\" type=\"text/css\">\n"
	 << "</head>\n"
	 << "<body>\n";
    
    //Get Author
    ins >> line;
    
    //Output Top X Words in Author's
    outs << "<h2>Top " << top_words << " Words in " << line << "'s ";
    
    //Get Title
    ins >> line;

    //Output "Title"
    outs << "\"" << line << "\"</h2>\n"
	 << "<hr>\n"
	 << "<div class=\"cdiv\">\n"
	 << "<p class=\"cbox\">\n";

    //Read all words in the Text into word_occur that !Is_Common_Word
    while (not ins.At_EOS ())
    {
	Get_Next_Word(ins,tba);
// 	ins >> line;
// 	while(line.Length()>0)
// 	{
// 	    Get_Word(line,tba);
	    
 	    if(!Is_Common_Word(tba))
 	    {
 		word_occur.Add(tba);
 	    }
// 	}
    }
    

    //Sort based on number of times that word appears

    
    //Remove the first "top_words"
    //Add them to Word_Occurrence called final
	i.Clear();
    while(i<top_words)
    {
	object Integer count;
	object Text word;

	word_occur.Remove_Any(word,count);
//	final.Add(word,count);
	i++;
    }
//	word_occur.Clear();
	
    //Sort final based on Alphabetical


    //Remove, compute frequency, and output to tagcloud
	i.Clear();
	while(i<top_words)
	{
	    object Integer occur,  total=final.Total();
	    object Real freq;

//	    final.Remove(tba,occur); 
	    freq = To_Real(occur/total);
	    if(freq >=.95)
	    {
		outs << "<span style=\"cursor:default\" class=\"f48\" title=\"count: "
		     << occur
		     << "\">"
		     << tba
		     << "</span>";
	    }
	    else if( freq < .95 && freq >= .90)
	    {
		outs << "<span style=\"cursor:default\" class=\"f46\" title=\"count: "
		     << occur
		     << "\">"
		     << tba
		     << "</span>";
	    }
	    else if( freq < .90 && freq >= .85)
	    {
		outs << "<span style=\"cursor:default\" class=\"f44\" title=\"count: "
		     << occur
		     << "\">"
		     << tba
		     << "</span>";
	    }
	    else if( freq < .85 && freq >= .80)
	    {
		outs << "<span style=\"cursor:default\" class=\"f42\" title=\"count: "
		     << occur
		     << "\">"
		     << tba
		     << "</span>";
	    }
	    else if( freq < .80 && freq >= .75)
	    {
		outs << "<span style=\"cursor:default\" class=\"f40\" title=\"count: "
		     << occur
		     << "\">"
		     << tba
		     << "</span>";
	    }
	    else if( freq < .75 && freq >= .70)
	    {
		outs << "<span style=\"cursor:default\" class=\"f38\" title=\"count: "
		     << occur
		     << "\">"
		     << tba
		     << "</span>";
	    }
	    else if( freq < .70 && freq >= .65)
	    {
		outs << "<span style=\"cursor:default\" class=\"f36\" title=\"count: "
		     << occur
		     << "\">"
		     << tba
		     << "</span>";
	    }
	    else if( freq < .65 && freq >= .60)
	    {
		outs << "<span style=\"cursor:default\" class=\"f34\" title=\"count: "
		     << occur
		     << "\">"
		     << tba
		     << "</span>";
	    }
	    else if( freq < .60 && freq >= .55)
	    {
		outs << "<span style=\"cursor:default\" class=\"f32\" title=\"count: "
		     << occur
		     << "\">"
		     << tba
		     << "</span>";
	    }
	    else if( freq < .55 && freq >= .50)
	    {
		outs << "<span style=\"cursor:default\" class=\"f30\" title=\"count: "
		     << occur
		     << "\">"
		     << tba
		     << "</span>";
	    }
	    else if( freq < .50 && freq >= .45)
	    {
		outs << "<span style=\"cursor:default\" class=\"f28\" title=\"count: "
		     << occur
		     << "\">"
		     << tba
		     << "</span>";
	    }
	    else if( freq < .45 && freq >= .40)
	    {
		outs << "<span style=\"cursor:default\" class=\"f26\" title=\"count: "
		     << occur
		     << "\">"
		     << tba
		     << "</span>";
	    }
	    else if( freq < .40 && freq >= .35)
	    {
		outs << "<span style=\"cursor:default\" class=\"f24\" title=\"count: "
		     << occur
		     << "\">"
		     << tba
		     << "</span>";
	    }
	    else if( freq < .35 && freq >= .30)
	    {
		outs << "<span style=\"cursor:default\" class=\"f22\" title=\"count: "
		     << occur
		     << "\">"
		     << tba
		     << "</span>";
	    }
	    else if( freq < .30 && freq >= .25)
	    {
		outs << "<span style=\"cursor:default\" class=\"f20\" title=\"count: "
		     << occur
		     << "\">"
		     << tba
		     << "</span>";
	    }
	    else if( freq < .25 && freq >= .20)
	    {
		outs << "<span style=\"cursor:default\" class=\"f18\" title=\"count: "
		     << occur
		     << "\">"
		     << tba
		     << "</span>";
	    }
	    else if( freq < .20 && freq >= .15)
	    {
		outs << "<span style=\"cursor:default\" class=\"f16\" title=\"count: "
		     << occur
		     << "\">"
		     << tba
		     << "</span>";
	    }
	    else if( freq < .15 && freq >= .10)
	    {
		outs << "<span style=\"cursor:default\" class=\"f14\" title=\"count: "
		     << occur
		     << "\">"
		     << tba
		     << "</span>";
	    }
	    else if( freq < .10 && freq >= .05)
	    {
		outs << "<span style=\"cursor:default\" class=\"f12\" title=\"count: "
		     << occur
		     << "\">"
		     << tba
		     << "</span>";
	    }
	    else if( freq < .05 && freq >= .00)
	    {
		outs << "<span style=\"cursor:default\" class=\"f10\" title=\"count: "
		     << occur
		     << "\">"
		     << tba
		     << "</span>";
	    }
	    else
	    {
		outs << "<h1>Your code is FUBAR</h1>";
	    }
	    i++;
	}

	outs << "</p>\n"
	     << "</div>\n"
	     << "</body>\n"
	     << "</html>";
     
    ins.Close_External ();
    outs.Close_External ();
}