makeIndex.cc | searchcode

/GeneR/GeneR/src/makeIndex.cc

#
C++ | 512 lines | 356 code | 92 blank | 64 comment | 157 complexity | 3ea9ad7e66e841860193a4b845e03eba MD5 | raw file
Possible License(s): LGPL-2.0

/*! \file makeIndex.cc
 * 
 *  \date Created       : 08/07/04 
 *  \date Last Modified : Time-stamp: <2006-03-08 17:31:07 lucas>
 *
 *  For all sequence files we build index file for fast access to a 
 *  specific sequence. Index files are a table with 4 columns
 *  defined as follow:
 *  accno  deb_feature   deb_sequence   length_sequence
 *
 *  \brief make index for Fasta embl and Genbank file
 *  \version 1
 *  \author A. Lucas
 *  \note Licence: CeCIll
 */



#include <string>
#include <stdio.h>
#include <stdlib.h>
#include "makeIndex.h"
#include "GeneR_globals.h"

#include <Rinternals.h>
#include <Rdefines.h>


extern "C"
{
  void ixfasta(char **path,int *res)  ;
  void ixgbk(char **path,int *res)  ;
  void ixembl(char **path,char **index,int *res)  ;
  void delete_CR_infile (char ** filein , char ** fileout, int * err);


}

/** @brief Make an index for a Fasta file
 *
 *
 *   @param path File path
 *   @param res Integer return -1 if error; 1 if ok; -2 if "\r" found;
 *          -3 if accno too long -4 file not found;
 */
void ixfasta(char **path,int *res)  
{
  FILE *fichierin;
  FILE *fichierout;
  
  char * pathout;
  char c;
  int i,compteur_accno,ecriture,len,header;
  int deb_sequence,len_feature;

  len = strlen(*path);
  *res = -1;
  pathout = (char * ) malloc ( (len + 4)* sizeof(char) );
  strcpy(pathout,*path);
  pathout[len]='.';
  pathout[len+1]='i';
  pathout[len+2]='x';
  pathout[len+3]='\0';

  fichierin=fopen(*path,"r");
  fichierout=fopen(pathout,"w");
  if ((fichierin == NULL ) || (fichierout == NULL))
    {
      fprintf(stdout, "GeneR.so: error while opening file\n");
      *res = -4;
      return ;
    }

  /* On se place au debut du fichier */
  /*  rewind(fichierin);*/
  i=0;  

  ecriture=0;
  deb_sequence=0;
  header=0;
  len_feature = 0;
  compteur_accno=0;
  while((c = fgetc(fichierin)) != EOF)
    {
      i++;

      if(c == '\r')
	  *res = -2 ;

      /* On traite l'ent?te (header  ou feature) */
      if(header)
	{
	  len_feature++;
	  if(c == '\n')
	    {
	      header=0;
	      ecriture=0;	      
	    }
	  if(c == ' ')
	    ecriture=0;
	  if(c == '\r')
	    {
	      header=0;
	      ecriture=0;
	    }
	  if(c == '\t')
	    ecriture=0;
	  if(c == 10)
	    ecriture=0;
	  if(( (compteur_accno) > MAX_LEN_ACCNO) && (ecriture))
	    {
	      *res = -3;
	      ecriture=0;	      
	    }
	  if(ecriture)
	    {
		  compteur_accno++;
		  fputc(c,fichierout);		
	    }
	}

      if(c == '>' && (!header))
	{
	  if(i>1)
	    makeIndex::ixecritureligne(deb_sequence,compteur_accno,len_feature,i-deb_sequence,fichierout);
	  deb_sequence = i;
	  compteur_accno = 0;
	  ecriture=1;
	  header = 1;
	  len_feature = 0;
	}


    }

  makeIndex::ixecritureligne(deb_sequence,compteur_accno,len_feature,i-deb_sequence+1,fichierout);

  fclose(fichierin);
  fclose(fichierout);

  if(*res == -1)
    *res = 1;
}


/** @brief Make an index for a GenBank file
 *
 *
 *   @param path File path
 *   @param res Integer return -1 if error; 1 if ok; -2 if "\r" found;
 *          -3 if accno too long,  -4 file not found;
 */
void ixgbk(char **path,int *res)  
{
  FILE *fichierin;
  FILE *fichierout;
  
  char * pathout;
  char c,c0,c1,c2,c3,c4,c5,c6,c7,c8,c9;
  int i,compteur_accno,ecriture,len,header;
  int deb_sequence,len_feature;

  *res = -1;

  len = strlen(*path);
  pathout =(char * ) malloc ( (len + 4)* sizeof(char) );
  strcpy(pathout,*path);
  pathout[len]='.';
  pathout[len+1]='i';
  pathout[len+2]='x';
  pathout[len+3]='\0';

  c=c0=c1=c2=c3=c4=c5=c6=c7=c8=c9=0;
  fichierin=fopen(*path,"r");
  fichierout=fopen(pathout,"w");

  if ((fichierin == NULL ) || (fichierout == NULL))
    {
      fprintf(stdout, "GeneR.so: error while opening file\n");
      *res = -4;
      return ;
    }

  /* On se place au debut du fichier */
  /*  rewind(fichierin);*/
  i=0;  

  ecriture=0;
  deb_sequence=0;
  header=1;
  len_feature = 0;
  compteur_accno=0;
  while((c = fgetc(fichierin)) != EOF)
    {
      if(c == '\r')
	  *res = -2;

      c9=c8;
      c8=c7;
      c7=c6;       
      c6=c5;
      c5=c4;
      c4=c3;
      c3=c2;
      c2=c1;
      c1=c0;
      c0=c;
      i++;


      /* On traite l'ent?te (header  ou feature) */
      if(header)
	{
	  len_feature++;
	  if(((c6 == '\n')|| (c6 == '\r') ) && (c5 == 'O')&& (c4 == 'R')&& (c3 == 'I')&& (c2 == 'G')&& (c1 == 'I')&& (c  == 'N')  )
	    {

	      while((((c = fgetc(fichierin))!='\n') && (c != '\r') ) && c != EOF)
		{
		  i++;
		  len_feature++;
		}
	      header=0;
	      i++;
	      len_feature++;
	    }

	  if(((c9 == '\n') || (i < 10) || (c9 == '\r') ) && (c8 == 'A')&& (c7 == 'C')&& (c6 == 'C')&& (c5 == 'E')&& (c4 == 'S')&& (c3 == 'S') && (c2 == 'I')&&  (c1 == 'O')&& (c0  == 'N') )
	    {
	      ecriture=1;	      
	      while((c = fgetc(fichierin))==' ')
		{
		  i++;
		  len_feature++;
		}
	      i++;
	      len_feature++;
	      compteur_accno=0;
	      
	    }

	  if(ecriture)
	    {
	      if((c == ':') || (c == '\n') || (c == '\r') || (c == ' '))
		ecriture=0;
	      else
		{
		  fputc(c,fichierout);
		  compteur_accno++;
		}
	      if(compteur_accno > MAX_LEN_ACCNO)
		{
		  ecriture=0;
		  *res= -3;
		}
	    }
	 
	}

      if(((c2 == '\n') || (c2 == '\r')) && (c1 == '/') && (c0 == '/') )
	{
	  
	  while((((c = fgetc(fichierin))!='\n') && (c != '\r') )&& c != EOF)
	    {
	      i++;
	    }
	  i++;

	  if(i>2)
	    makeIndex::ixecritureligne(deb_sequence+1,compteur_accno,len_feature-1,i-deb_sequence,fichierout);
	  deb_sequence = i;
	  compteur_accno = 0;
	  ecriture=0;
	  header = 1;
	  len_feature = 0;
	}
    }

  if(!header)
    makeIndex::ixecritureligne(deb_sequence+1,compteur_accno,len_feature-1,i-deb_sequence,fichierout);

  fclose(fichierin);
  fclose(fichierout);

  if(*res == -1)
    *res = 1;

}

/** @brief Make an index for a EMBL file
 *   @param path File path
 *   @param index "d" or "x" if index file is ".ix" or ".id"
 *   @param res Integer return -1 if error; 1 if ok; -2 if "\r" found;
 *          -3 if accno too long,  -4 file not found;
 */
void ixembl(char **path,char **index,int *res)  
{
  FILE *fichierin;
  FILE *fichierout;
  
  char * pathout;
  char c,c0,c1,c2,c3,c4,c5,c6,c7,c8,c9;
  int i,compteur_accno,ecriture,len,header;
  int deb_sequence,len_feature;

  *res = -1;
  len = strlen(*path);
  pathout =(char * ) malloc ( (len + 4)* sizeof(char) );
  strcpy(pathout,*path);
  pathout[len]='.';
  pathout[len+1]='i';
  pathout[len+2]= **index;
  pathout[len+3]='\0';

  c=c0=c1=c2=c3=c4=c5=c6=c7=c8=c9=0;
  fichierin=fopen(*path,"r");
  fichierout=fopen(pathout,"w");

  if ((fichierin == NULL ) || (fichierout == NULL))
    {
      fprintf(stdout, "GeneR.so: error while opening file\n");
      *res = -4;
      return ;
    }

  /* On se place au debut du fichier */
  /*  rewind(fichierin);*/
  i=0;  

  ecriture=0;
  deb_sequence=0;
  header=1;
  len_feature = 0;
  compteur_accno=0;
  while((c = fgetc(fichierin)) != EOF)
    {

      if(c == '\r')
	  *res = -2;

      c9=c8;
      c8=c7;
      c7=c6;       
      c6=c5;
      c5=c4;
      c4=c3;
      c3=c2;
      c2=c1;
      c1=c0;
      c0=c;
      i++;


      /* On traite l'ent?te (header  ou feature) */
      if(header)
	{
	  len_feature++;
	  if(((c5 == '\n') || (c5 == '\r'))&&  (c4 == 'S')&&(c3 == 'Q')&&  (c2 == ' ')&& (c1 == ' ')&& (c0  == ' ') )

	    {

	      while((((c = fgetc(fichierin))!='\n') && (c != '\r') ) && c != EOF)
		{
		  i++;
		  len_feature++;
		}
	      header=0;
	      i++;
	      len_feature++;
	    }

	  if(((c6 == '\n')||(c6 == '\r')|| (i<7) )&&  (c5 == 'A')&&(c4 == 'C')&&  (c3 == ' ')&& (c2 == ' ')&& (c1  == ' ') )
	    {
	      ecriture=1;	      
	      compteur_accno=0;	     
	    }

	  if(ecriture)
	    {
	      if((c == ':') || (c == ';') || (c == '\n') || (c == '\r') || (c == ' '))
		ecriture=0;
	      else
		{
		  fputc(c,fichierout);
		  compteur_accno++;
		}
	      if(compteur_accno > MAX_LEN_ACCNO)
		{
		  *res = -3;
		  ecriture=0;
		}
	    }
	 
	}

      if(((c2 == '\n') ||(c2 == '\r')) && (c1 == '/') && (c0 == '/') )
	{
	  
	  while((((c = fgetc(fichierin))!='\n') && (c != '\r'))&& c != EOF)
	    {
	      i++;
	    }
	  i++;

	  if(i>2)
	    makeIndex::ixecritureligne(deb_sequence+1,compteur_accno,len_feature-1,i-deb_sequence,fichierout);
	  deb_sequence = i;
	  compteur_accno = 0;
	  ecriture=0;
	  header = 1;
	  len_feature = 0;
	}
    }

  if(!header)
    makeIndex::ixecritureligne(deb_sequence+1,compteur_accno,len_feature-1,i-deb_sequence,fichierout);

  fclose(fichierin);
  fclose(fichierout);

  if(*res == -1)
    *res = 1;
}




/**
 * Delete carriage return in the file.
 * \brief This function read filein and copy all to fileout except char "\r"
 *  it transfrom "\r" into "\n" if newline only delimited with "\r" otherwise, it just
 * delete "\r".
 * \param filein,fileout path to files
 * \param err: -1 problem while opening file, else 1
 */
void delete_CR_infile (char ** filein , char ** fileout, int * err)
{
  FILE *fichierin;
  FILE *fichierout;
  char c;
  int i=0;
  fichierin=fopen(*filein,"rb");
  fichierout=fopen(*fileout,"wb");


  if((fichierin == NULL) || (fichierout == NULL))
    {
      *err = -1;
      return ;
    }
    

  i=0;

  while((c = getc(fichierin)) != EOF)
    {
      i++;

      if(c == '\r')
	{
	  c = fgetc(fichierin);
	  if(c == EOF)
	    {
	      fputc('\n',fichierout);
	      break;
	    }
	  
	  if(c != '\n') /* case: newline delimited by "\r" (i.e mac ?) */
          fputc('\n',fichierout);	  

	  /* case: newline delimited by "\r\n" (i.e windows, "\r" char deleted ) */
      	}
      
      fputc(c,fichierout);	 

  
    }
  
  fclose(fichierin);
  fclose(fichierout);

}



namespace makeIndex{
  /** @brief Internal C function to write the end of a line in the index
   *  \note The accno should be written, function add spaces and all numbers
   *    to the current line of the file
   *
   *
   *   @param deb_sequence Begining of Sequence
   *   @param compteur_accno Lenght of Accno 
   *   @param len_feature Length of features
   *   @param len_sequence Length of sequence
   *   @param fichierout File out
   */

  int ixecritureligne(int deb_sequence,int compteur_accno,
			   int len_feature,int len_sequence,
			   FILE * fichierout)
  {
    int i;
    for(i=compteur_accno; i<=  MAX_LEN_ACCNO ; i++)
      fputc(' ',fichierout);
    
    return(fprintf(fichierout,"%10d %10d %8d\n",deb_sequence-1,
		   len_feature+deb_sequence,len_sequence));
  }


}