pgsql_lib.c

序列对齐 Compare a protein sequence to a protein sequence database or a DNA sequence to a DNA sequenc

    /* have get_desc, copy it */
    else {
      strcpy(m_fptr->sql_getdesc,bps);
      *bp=';'; /* replace ; */
      bps = bp+1;
      while(isspace(*bps)) bps++;
    }
  }
  else {
    fprintf(stderr," cannot find getdesc field:\n%s\n",tmp_str);
    goto error_r;
  }

  if ((bp=strchr(bps,';'))!=NULL) { *bp='\0';}

  if ((m_fptr->sql_getseq=calloc(strlen(bps)+1,sizeof(char)))==NULL) {
    fprintf(stderr, " cannot allocate space for database name [%d], %s\n",
	    strlen(bps),bps);
    goto error_r;
  }

  if (strlen(bps) > 0) {
    strcpy(m_fptr->sql_getseq,bps);
  }
  else {
    fprintf(stderr," cannot find getseq field:\n%s\n",tmp_str);
    return NULL;
  }
  if (bp!=NULL) *bp=';';

  /* now do the fetch */

  res = PQexec(m_fptr->pgsql_conn,"BEGIN;");
  if (PQresultStatus(res) != PGRES_COMMAND_OK) {
    fprintf(stderr,"*** Error %s - BEGIN failed:\n",
	    PQerrorMessage(conn));
    PQclear(res);
    goto error_r;
  }
  PQclear(res);

  res = PQexec(m_fptr->pgsql_conn, m_fptr->sql_query);
  if (PQresultStatus(res) != PGRES_COMMAND_OK) {
    fprintf(stderr,"*** Error %d:%s - query failed:\n%s\n",
	    PQresultStatus(res),PQerrorMessage(conn), m_fptr->sql_query);
    PQclear(res);
    goto error_r;
  }
  PQclear(res);
  m_fptr->pgsql_res=NULL;

  return m_fptr;

 error_r:
  free(m_fptr->sql_getseq);
  free(m_fptr->sql_getdesc);
  free(m_fptr->sql_query);
  free(m_fptr);
  free(sql_db);
  return NULL;
}

struct lmf_str *
pgsql_reopen(struct lmf_str *m_fptr) {
  m_fptr->sql_reopen = 1;
  return m_fptr;
}

void
pgsql_closelib(struct lmf_str *m_fptr) {

  if (m_fptr == NULL) return;
  if (m_fptr->pgsql_res != NULL) PQclear(m_fptr->pgsql_res);
  PQfinish(m_fptr->pgsql_conn);
  m_fptr->sql_reopen=0;
}

/*
static char *sql_seq = NULL, *sql_seqp;
static int sql_seq_len;
*/

int
pgsql_getlib( unsigned char *seq,
	      int maxs,
	      char *libstr,
	      int n_libstr,
	      fseek_t *libpos,
	      int *lcont,
	      struct lmf_str *lm_fd,
	      long *l_off)
{
  register unsigned char *cp, *seqp;
  register int *ap;
  unsigned char *seqm, *seqm1;
  PGresult *res;

  char *bp;
  /*   int l_start, l_stop, len; */

  seqp = seq;
  seqm = &seq[maxs-9];
  seqm1 = seqm-1;

  ap = lm_fd->sascii;

#ifdef SUPERFAMNUM
  sfnum[0]=nsfnum = 0;
#endif

  if (*lcont==0) {
    /* get a row, with UID, sequence */
    *l_off = 1;

    /* check to see if we already have a valid result */
    if (lm_fd->pgsql_res==NULL) {
      res = PQexec(lm_fd->pgsql_conn,"FETCH next_seq");
      if (PQresultStatus(res) != PGRES_TUPLES_OK) {
	fprintf(stderr,"*** Error %s - getlib FETCH failed:\n%s\n",
		PQerrorMessage(lm_fd->pgsql_conn), lm_fd->sql_query);
	PQclear(res);
	lm_fd->pgsql_res = NULL;
	*lcont = 0;
	*seqp = EOSEQ;
	return -1;
      }
    }
    else {res = lm_fd->pgsql_res;}

    if (PQntuples(res)>0) {
      lm_fd->pgsql_res = res;
      *libpos=(fseek_t)atol(PQgetvalue(res,0,0));
	
      *l_off = 1;
      if (PQnfields(res) > 2 && (bp=strchr(PQgetvalue(res,0,2),'@'))!=NULL &&
	  !strncmp(bp+1,"C:",2)) sscanf(bp+3,"%ld",l_off);

      lm_fd->sql_seqp = PQgetvalue(res,0,1);
    
      /* because of changes in pgsql_ranlib(), it is essential that
         libstr return the unique identifier; thus we must use
         sql_row[0], not sql_row[2]. Using libstr as the UID allows
         one to use any UID, not just numeric ones.  *libpos is not
         used for pgsql libraries.
      */

      if (n_libstr <= MAX_UID) {
	/* the normal case returns only GID/sequence */
	strncpy(libstr,PQgetvalue(res,0,0),MAX_UID-1);
	libstr[MAX_UID-1]='\0';
      }
      else {
	/* here we do not use the UID in libstr, because we are not
           going back into the db */
	/* the PVM case also returns a long description */
	if (PQnfields(res)>2) {
	  strncpy(libstr,PQgetvalue(res,0,2),n_libstr-1);
	}
	else {
	  strncpy(libstr,PQgetvalue(res,0,0),n_libstr-1);
	}
	libstr[n_libstr-1]='\0';
      }
    }
    else {
      PQclear(lm_fd->pgsql_res);
      lm_fd->pgsql_res=NULL;
      *lcont = 0;
      *seqp = EOSEQ;
      return -1;
    }
  }

  for (cp=(unsigned char *)lm_fd->sql_seqp; seqp<seqm1 && *cp; ) {
    if ((*seqp++=ap[*cp++])<NA &&
	(*seqp++=ap[*cp++])<NA &&
	(*seqp++=ap[*cp++])<NA &&
	(*seqp++=ap[*cp++])<NA &&
	(*seqp++=ap[*cp++])<NA &&
	(*seqp++=ap[*cp++])<NA &&
	(*seqp++=ap[*cp++])<NA &&
	(*seqp++=ap[*cp++])<NA &&
	(*seqp++=ap[*cp++])<NA &&
	(*seqp++=ap[*cp++])<NA) continue;
    --seqp;
    if (*(cp-1)==0) break;
  }
  lm_fd->sql_seqp = (char *)cp;

  if (seqp>=seqm1) (*lcont)++;
  else {
    *lcont=0;
    PQclear(lm_fd->pgsql_res);
    lm_fd->pgsql_res = NULL;
  }

  *seqp = EOSEQ;
  /*   if ((int)(seqp-seq)==0) return 1; */
  return (int)(seqp-seq);
}

void
pgsql_ranlib(char *str,
	     int cnt,
	     fseek_t libpos,
	     char *libstr,
	     struct lmf_str *lm_fd
	     )
{
  char tmp_query[1024], tmp_val[20];
  PGresult *res;
  char *bp;

  str[0]='\0';

  /* put the UID into the query string - cannot use sprintf because of
     "%' etc */

  /*   sprintf(tmp_query,lm_fd->sql_getdesc,libpos); */

  if ((bp=strchr(lm_fd->sql_getdesc,'#'))==NULL) {
    fprintf(stderr, "no KEY position in %s\n",lm_fd->sql_getdesc);
    goto next1;
  }
  else {
    *bp = '\0';
    strncpy(tmp_query,lm_fd->sql_getdesc,sizeof(tmp_query));
    tmp_query[sizeof(tmp_query)-1]='\0';
    /*    sprintf(tmp_val,"%ld",(long)libpos); */
    strncat(tmp_query,libstr,sizeof(tmp_query)-1);
    strncat(tmp_query,bp+1,sizeof(tmp_query)-1);
    *bp='#';
    lm_fd->lpos = libpos;
  }

  /*  fprintf(stderr," requesting: %s\n",tmp_query); */

  if (lm_fd->pgsql_res !=NULL) {
    PQclear(lm_fd->pgsql_res);
    lm_fd->pgsql_res = NULL;
  }

  res = PQexec(lm_fd->pgsql_conn,tmp_query);
  if (PQresultStatus(res) != PGRES_TUPLES_OK) {
    lm_fd->pgsql_res = NULL;

    sprintf(str,"gi|%ld ***Error - query failed***",(long)libpos);
    fprintf(stderr,"*** Error %s - ranlib DESC failed:\n%s\n",
	    PQerrorMessage(lm_fd->pgsql_conn), tmp_query);
    PQclear(res);
    goto next1;
  }

  if (PQntuples(res)<=0) {
/*     fprintf(stderr,"*** Error = use result failed\n%s\n", 
	   pgsql_error(lm_fd->pgsql_conn)); */
    sprintf(str,"gi|%ld ***use result failed***",(long)libpos);
    goto next0;
  }

  if (PQgetvalue(res,0,1)!= NULL) strncpy(str,PQgetvalue(res,0,1),cnt-1);
  else strncpy(str,PQgetvalue(res,0,0),cnt-1);
  str[cnt-1]='\0';
  /* change this later to support multiple row returns */
  /*
  while (strlen(str) < cnt-1 &&
	 (lm_fd->sql_row = pgsql_fetch_row(lm_fd->pgsql_res))!=NULL) {
    strncat(str," ",cnt-2-strlen(str));
    if (lm_fd->sql_row[1]!=NULL) 
      strncat(str,lm_fd->sql_row[1],cnt-2-strlen(str));
    else break;
  }
  */

  str[cnt-1]='\0';
  if ((bp = strchr(str,'\r'))!=NULL) *bp='\0';
  if ((bp = strchr(str,'\n'))!=NULL) *bp='\0';

 next0:
  PQclear(res);
 next1: 
  lm_fd->pgsql_res = NULL;

  /* get the sequence, set up for pgsql_getseq() */
  /* put the UID into the query string */

  if ((bp=strchr(lm_fd->sql_getseq,'#'))==NULL) {
    fprintf(stderr, "no GID position in %s\n",lm_fd->sql_getseq);
    return;
  }
  else {
    *bp = '\0';
    strncpy(tmp_query,lm_fd->sql_getseq,sizeof(tmp_query));
    tmp_query[sizeof(tmp_query)-1]='\0';
    /*    sprintf(tmp_val,"%ld",(long)libpos); */
    strncat(tmp_query,libstr,sizeof(tmp_query));
    strncat(tmp_query,bp+1,sizeof(tmp_query));
    *bp='#';
  }

  res = PQexec(lm_fd->pgsql_conn,tmp_query);
  if (PQresultStatus(res) != PGRES_TUPLES_OK) {
    PQclear(res);
    lm_fd->pgsql_res = NULL;
    fprintf(stderr,"*** Error - ranlib SEQ failed:\n%s\n%s\n",tmp_query,
	    PQerrorMessage(lm_fd->pgsql_conn));
    exit(1);
  }
  else {
    lm_fd->pgsql_res = res;
  }
}