/*****************************************************************************
#   Copyright (C) 1994-2008 by David Gordon.
#   All rights reserved.                           
#                                                                           
#   This software is part of a beta-test version of the Consed/Autofinish
#   package.  It should not be redistributed or
#   used for any commercial purpose, including commercially funded
#   sequencing, without written permission from the author and the
#   University of Washington.
#   
#   This software is provided ``AS IS'' and any express or implied
#   warranties, including, but not limited to, the implied warranties of
#   merchantability and fitness for a particular purpose, are disclaimed.
#   In no event shall the authors or the University of Washington be
#   liable for any direct, indirect, incidental, special, exemplary, or
#   consequential damages (including, but not limited to, procurement of
#   substitute goods or services; loss of use, data, or profits; or
#   business interruption) however caused and on any theory of liability,
#   whether in contract, strict liability, or tort (including negligence
#   or otherwise) arising in any way out of the use of this software, even
#   if advised of the possibility of such damage.
#
#   Building Consed from source is error prone and not simple which is
#   why I provide executables.  Due to time limitations I cannot
#   provide any assistance in building Consed.  Even if you do not
#   modify the source, you may introduce errors due to using a
#   different version of the compiler, a different version of motif,
#   different versions of other libraries than I used, etc.  For this
#   reason, if you discover Consed bugs, I can only offer help with
#   those bugs if you first reproduce those bugs with an executable
#   provided by me--not an executable you have built.
# 
#   Modifying Consed is also difficult.  Although Consed is modular,
#   some modules are used by many other modules.  Thus making a change
#   in one place can have unforeseen effects on many other features.
#   It may takes months for you to notice these other side-effects
#   which may not seen connected at all.  It is not feasable for me to
#   provide help with modifying Consed sources because of the
#   potentially huge amount of time involved.
#
#*****************************************************************************/
#include    "myAgrepAndFilter.h"
#include    "myAgrep.h"
#include    "assert.h"
#include    "abs.h"


void myAgrepAndFilter( char* szText,
                       const int nTextLength,
                       char* szQuery,
                       const int nQueryLength,
                       const int nMaxErrors,
                       RWTValOrderedVector<int>* pMatchPositions,
                       RWTValOrderedVector<int>* pMatchScores,
                       RWTValOrderedVector<bool>* pUseMatches ) {


   myAgrep( szText,
            nTextLength,
            szQuery,
            nQueryLength,
            nMaxErrors,
            pMatchPositions,
            pMatchScores );

   
   // now need to filter these

   assert( pMatchPositions->length() == pMatchScores->length() );

   // start by saying use all matches

   int n;
   for( n = 0; n < pMatchPositions->length(); ++n ) {
      pUseMatches->insert( true );
   }


   for( n = 0; n < pMatchPositions->length(); ++n ) {
      int nPos = (*pMatchPositions)[ n ];

      // now I'd like to find a cluster for this match

      int nBeginningOfCluster = n;
      int nEndOfCluster = n;
      int n2;
      for( n2 = n + 1; n2 < pMatchPositions->length(); ++n2 ) {
         
         if ( ABS( (*pMatchPositions)[ n2 ] - nPos ) <= 2*nMaxErrors ) 
            nEndOfCluster = n2;
         else
            break; // reached end of cluster
      }


      // within the cluster, find the lowest scoring match 
      // (lowest means best because we are measuring mismatches 

      int nScoreOfLowestScoringMatch = nMaxErrors + 1; 
                   // so the first time will be found
                                                        
      int nPositionOfLowestScoringMatch;
      for( n2 = nBeginningOfCluster; n2 <= nEndOfCluster; ++n2 ) {
         if ( (*pMatchScores)[ n2 ] < nScoreOfLowestScoringMatch ) {
            nPositionOfLowestScoringMatch = n2;
            nScoreOfLowestScoringMatch = (*pMatchScores)[ n2 ];
         }
      }


      for( n2 = nBeginningOfCluster; n2 <= nEndOfCluster; ++n2 ) {
         if ( n2 != nPositionOfLowestScoringMatch )
            (*pUseMatches)[n2] = false;

      }

      // I could try to advance n over the cluster, but I not sure if 
      // I would miss some...

   }
}