%{
#include "sequencedb.h"
#include "hscore.h"
#include "complexsequence.h"
#include "complexevalset.h"
#include "randommodel.h"
#include "randomdb.h"
#include "protein.h"

%}


struct ProteinDB
boolean is_single_seq !def="FALSE"
boolean is_random_db  !def="FALSE"
ComplexSequence * single
SequenceDB * sdb
ComplexSequenceEvalSet * cses
RandomProteinDB * rnd 
boolean test_dna !def="FALSE" 
%info
A database of proteins. This takes
either a single sequence or a 
sequence database and allows a
method expecting protein complexsequences
to loop over it.

It also provides generic database indexing
for this


Horrible physical dependency in this
generated by the random protein db
requiring histogram and randommodel
stuff. Yuk!
%%

api
object ProteinDB
des free_ProteinDB
endobject
func new_ProteinDB_from_single_seq
func single_fasta_ProteinDB
func new_ProteinDB
endapi

%{
#include "proteindb.h"


%func 
shows the Hscore by the ProteinDB information

%arg
hs  High Score structure
ofp output file
%%
void show_Hscore_ProteinDB(Hscore * hs,FILE * ofp)
{
  int i;

  for(i=0;i<hs->len;i++)
    fprintf(ofp,"Query [%20s] Target [%20s] %d\n",hs->ds[i]->query->name,hs->ds[i]->target->name,hs->ds[i]->score);
}

%func
Gets Protein sequence out from
the proteindb using the information stored in
dataentry
%simple get_entry
%arg
prodb r ProteinDB database
de     r DataEntry information 
%%
Protein * get_Protein_from_ProteinDB(ProteinDB * prodb,DataEntry * de)
{
  Sequence * seq;

  if( prodb->is_single_seq == TRUE ) {
    return Protein_from_Sequence(hard_link_Sequence(prodb->single->seq));
  }

  /* we need to get out the Sequence from seqdb */

  seq = get_Sequence_from_SequenceDB(prodb->sdb,de);
  if( seq == NULL ) {
    warn("Cannot get entry for %s from Protein db",de->name);
    return NULL;
  }

  seq->type = SEQUENCE_PROTEIN; /* force to protein */

  return Protein_from_Sequence(seq);
}


%func
adds information to dataentry from ProteinDB

This information is the necessary information for
the proteindb to find this sequence later
%%
boolean dataentry_add_ProteinDB(DataEntry * de,ComplexSequence * cs,ProteinDB * prodb)
{
  if( cs == NULL || cs->seq == NULL ){
     warn("Adding a dataentry with a NULL complex sequence or null internal sequence. Nope!");
     return FALSE;
  }

  if( prodb->is_single_seq == FALSE)
     add_SequenceDB_info_DataEntry(prodb->sdb,de);

  de->name = stringalloc(cs->seq->name);

  return TRUE;
}


%func 
top level function which opens the protein database
%arg
prodb protein database
return_status w the status of the open from database.h
%%
ComplexSequence * init_ProteinDB(ProteinDB * prodb,int * return_status)
{
  ComplexSequence * cs;
  Sequence * seq;

  if( prodb->is_single_seq == TRUE) {
    *return_status = DB_RETURN_OK;
    return prodb->single;
  }

  seq = init_SequenceDB(prodb->sdb,return_status);

  if( seq == NULL || *return_status == DB_RETURN_ERROR || *return_status == DB_RETURN_END ) {
    return NULL; /** error already reported **/
  }

  if( prodb->test_dna == FALSE ) {
    seq->type = SEQUENCE_PROTEIN;
  } else {
    if( seq->type != SEQUENCE_PROTEIN ) {
      warn("For sequence %s, looks like a DNA sequence. Failing");
      *return_status = DB_RETURN_ERROR;
    }
  }

  cs = new_ComplexSequence(seq,prodb->cses);

  free_Sequence(seq);

  return cs;
}

%func
function which reloads the database
%arg
last previous complex sequence, will be freed
return_status w return_status of the load
%%
ComplexSequence * reload_ProteinDB(ComplexSequence * last,ProteinDB * prodb,int * return_status)
{
  ComplexSequence * cs;
  Sequence * seq;
    
  if( prodb->is_single_seq == TRUE ) {
    *return_status = DB_RETURN_END;
    return NULL;
  }

  /** free Complex Sequence **/
  if( last != NULL ) {
    free_ComplexSequence(last);
  }

  seq = reload_SequenceDB(NULL,prodb->sdb,return_status);

  if( seq == NULL || *return_status == DB_RETURN_ERROR || *return_status == DB_RETURN_END ) {
    return NULL; /** error already reported **/
  }
  if( prodb->test_dna == FALSE ) {
    seq->type = SEQUENCE_PROTEIN;
  } else {
    if( seq->type != SEQUENCE_PROTEIN ) {
      warn("For sequence %s, looks like a DNA sequence. Failing");
      *return_status = DB_RETURN_ERROR;
    }
  }

  cs = new_ComplexSequence(seq,prodb->cses);

  free_Sequence(seq);

  return cs;
}
  


%func
top level function which closes the protein database
%arg
cs last complex sequence 
prodb protein database
%%
boolean close_ProteinDB(ComplexSequence * cs,ProteinDB * prodb) 
{
  if( prodb->is_single_seq == TRUE ) {
    return TRUE;
  }

  if( cs != NULL) {
    free_ComplexSequence(cs);
  }

  return close_SequenceDB(NULL,prodb->sdb);
}

%func 
To make a new protein database
from a single Sequence with default amino acid mapping
%arg
seq sequence which as placed into ProteinDB structure.
%%
ProteinDB * new_ProteinDB_from_single_seq(Sequence * seq)
{
  ComplexSequenceEvalSet * cses;
  ComplexSequence * cs;

  cses = default_aminoacid_ComplexSequenceEvalSet();

  cs = new_ComplexSequence(seq,cses);

  free_ComplexSequenceEvalSet(cses);

  return new_ProteinDB_from_single_cseq(cs);
}
  

%func
To make a new protein database
from a single ComplexSequence
%arg
cs complex sequence which is held.
%%
ProteinDB * new_ProteinDB_from_single_cseq(ComplexSequence * cs)
{
  ProteinDB * out;


  out = ProteinDB_alloc();
  out->is_single_seq = TRUE;
  out->single = cs;
  
  return out;
}

%func
pre-packed single fasta protein database

%arg
filename name of fasta file
%%
ProteinDB * single_fasta_ProteinDB(char * filename)
{
  return new_ProteinDB(single_fasta_SequenceDB(filename),default_aminoacid_ComplexSequenceEvalSet());
}

  
%func
To make a new protein database
%arg
seqdb sequence database
cses  protein evaluation set
%%
ProteinDB * new_ProteinDB(SequenceDB * seqdb,ComplexSequenceEvalSet * cses)
{
  ProteinDB * out;

  /** should check sequence database **/

  if( seqdb == NULL ) {
     warn("Cannot make ProteinDB from NULL SequenceDB object");
     return NULL;
  }

  if( cses->type != SEQUENCE_PROTEIN ) {
    warn("You can't make a protein database with a non SEQUENCE_PROTEIN cses type [%d]",cses->type);
    return NULL;
  }


  out = ProteinDB_alloc();

  out->sdb  = seqdb;
  out->cses = cses;

  return out;
}

 
%}