/*
** PROGRAM NAME: WORDCNT.CPL
** -------------
**  Program Source Code File (.CPL)
**  Concordance(tm) Information Retrieval System, Professional Edition
**
** COPYRIGHT:
** ----------
**  (C) Copyright Dataflight Software, Inc. 1992. All Rights Reserved.
**
** SYNOPSIS: Counts total occurrence of all words in database.
** ---------
**
*/

/* Uniform width of document number stored in temporary btree file. */
short WIDTH = 10;

/* Error codes. */
short DISKFULLERROR = 1;
short REINDEXERROR  = 2;
short USEREXITERROR = 3;
short NOHITSERROR   = 4;
short OPENDBERROR   = 5;
short TEMPFILEERROR = 6;
short IVTFILEERROR  = 7;
short DCTFILEERROR  = 8;
short QUERYZEROERR  = 9;
short ZERODOCERROR  = 10;

main()
{
int db;

   switch(talley(db)) {
        case DISKFULLERROR:
                                Message("Disk full encountered while writing temporary file.");
                                break;
        case USEREXITERROR:
                                break;
        case OPENDBERROR  :
                                Message("Please open a database before running word count.");
                                break;
        case TEMPFILEERROR:
                                Message("Couldn't create temporary .LST file. ");
                                break;
        case DCTFILEERROR :
                                Message("Couldn't open database dictionary.");
                                break;
        case IVTFILEERROR :
                                Message("Couldn't open database .IVT file.");
                                break;
        case ZERODOCERROR :
                                Message("This search located zero documents, nothing to do.");
                                break;
        case 0:
                                Message("This database dictionary has been tallied.");
                                break;
   }

} /* main() */

/****************************************************************
* Name:         talley                                          *
* Synopsis:     Talley total number of occurrences in database. *
****************************************************************/
talley(int db)
{
int   error,                    /* Error control variable.                 */
      i,                        /* Misc loop variable.                     */
      offset,                   /* Offset into field of word.              */
      invtHandle,               /* Handle to opened .IVT file.             */
      treeHandle,               /* Handle to opened .DCT file.             */
      wordNumber,               /* Status variable for progress display.   */
      blocks;                   /* Number of blocks in progress display.   */
char  word[33],                 /* Word to weight, at "offset" in "field." */
      treeName[128];            /* Name of temporary weightedTreeHandle.   */
float occurrences,              /* Total occurrences of all words.         */
      percent,                  /* Progress percentage through database.   */
      totalWords;               /* Total count of word in dictionary.      */
text  screen;                   /* Saved screen, restored on exit.         */
int   DOCUMENTOFFSET,
      WORDOFFSET = 4;
   cursoroff();

   if (db.documents < 0)
      if (getfile("Open Database","*.DCB",treeName) == CR)
         db = opendb(treeName);

   /* Do some initial error checking to ensure that a database  */
   /* is open, that documents were located in a query, and that */
   /* a search was actually performed.                          */
   if (db.documents < 0)
      error = OPENDBERROR;
   else {
      if (count(db) <= 0)
         error = ZERODOCERROR;
      else {
         /* Open the inverted text file for processing. */
         if ((invtHandle = open(db.database+".IVT","r")) == EOF)
            error = IVTFILEERROR;
         else {

            /* Open the database dictionary for processing */
            if ((treeHandle = btopen(db.database+".DCT")) == EOF) {
               close(invtHandle);
               error = DCTFILEERROR;
            }
         }
      }
   }

   if (error == FALSE) {
      /* Initialize the screen for status display */
      /* while we are processing the data base.   */
      screen = save(8,19,17,61);
      blocks = 37;
      scroll(8,19,14,59,0,0,MenuColor_);
      box(8,19,16,61,"DS",MenuColor_);
      puts(09,24,"ÄÄ Database Occurrence Tally ÄÄ",MenuColor_);
      puts(10,21,"Processing word",MenuColor_);
      puts(11,21,"Occurrence count",MenuColor_);
      puts(12,21,"Dictionary processed",MenuColor_);
      puts(14,21,rep('°',blocks),MenuColor_);
      puts(10,57,"1",MenuColor_);
      puts(12,57,"0%",MenuColor_);

      /* Cycle through each word retrieving hit count. */
      totalWords = btcount(treeHandle);
      for(i = btfirst(treeHandle,word,offset); i == 0; i = btnext(treeHandle,word,offset)) {

         seek(invtHandle,offset+WORDOFFSET,'B');
         read(invtHandle,i,sizeof(i));
         occurrences = occurrences + i;

         /* Update the status display. */
         wordNumber = wordNumber + 1;
         percent = wordNumber/totalWords;
         puts(10,36,pad(word,'R',22),MenuColor_);
         puts(11,48,str(occurrences,10,0,','),MenuColor_);
         puts(12,48,str(percent*100.0,10),MenuColor_);
         puts(14,21,rep('²',blocks*percent),MenuColor_);

         /* See if the user wants to exit the processing loop early. */
         if (keypress()) {
           if (getkey() == ESC)
              if (Message("Cancel dictionary talley? Y/N") == 'Y') {
                 error = USEREXITERROR;
                 break;
              }
         }
      }

      /* Close the .IVT and .DCT files, we are finished with them. */
      close(invtHandle);
      btclose(treeHandle);

      /* Restore the user's screen. */
      if (error == FALSE) 
         getkey();      
      restore(8,19,screen);
   }
   
   return(error);

} /* talley() */

/****************************************************************
* Name:         Message                                         *
* Synopsis:     Displays error message and waits for key.       *
****************************************************************/
Message(text message)
{
text screen;
int key;
   cursoroff();
   screen = save(10,13,13,69);
   box(10,13,13,69,"DS", MenuColor_);
   puts(11,14,pad(message,'C',53),MenuColor_);
   if ((key = getkey()) & 255)
      key = asc(upper(chr(key)));
   restore(10,13,screen);
   return(key);
} /* Message() */
/****************************************************************
* Name:         Path                                            *
* Synopsis:     Returns the path up to the last \ and without   *
*               the file's name.                                *
****************************************************************/
Path(text dosPath)
{
int i, j;

   if ((i = match(dosPath,":",1)) == 0)
      i = match(dosPath,"\",1);
   while(j = match(dosPath,"\",i+1))
      i = j;
   return(substr(dosPath,1,i));

} /* Path() */
/****************************************************************
* Global Variable Declarations and Initialization               *
****************************************************************/

int     CTRLPGUP = 33792;
short   LEFT = 19200,
        RIGHT = 19712,
        UP = 18432,
        DOWN = 20480,
        HOME = 18176,
        END = 20224,
        PGUP = 18688,
        PGDN = 20736,
        CTRLPGDN = 30208,
        F2 = 15360,
        F3 = 15616,
        F4 = 15872,
        F5 = 16128,
        F6 = 16384,
        F7 = 16640,
        F8 = 16896,
        F9 = 17152,
        F10 = 17408;
char    ESC = 27,
        CTRLP = 16,
        EOF = -1,
        FALSE = 0,
        TRUE = 1,
        CR = 13,
        LF = 10;