Exemple #1
0
static int inputsuffixarray(bool map,
                            Suffixarray *suffixarray,
                            unsigned int demand,
                            const GtStr *indexname,
                            Verboseinfo *verboseinfo,
                            GtError *err)
{
  bool haserr = false;
  Seqpos totallength = 0;

  gt_error_check(err);
  initsuffixarray(suffixarray);
  suffixarray->encseq = mapencodedsequence(true,
                                           indexname,
                                           (demand & SARR_ESQTAB) ? true
                                                                  : false,
                                           (demand & SARR_DESTAB) ? true
                                                                  : false,
                                           (demand & SARR_SDSTAB) ? true
                                                                  : false,
                                           (demand & SARR_SSPTAB) ? true
                                                                  : false,
                                           verboseinfo,
                                           err);
  if (suffixarray->encseq == NULL)
  {
    haserr = true;
  } else
  {
    totallength = getencseqtotallength(suffixarray->encseq);
  }
  if (!haserr)
  {
    haserr = scanprjfileuintkeys(suffixarray,indexname,verboseinfo,err);
  }
  if (!haserr && (demand & SARR_SUFTAB))
  {
    if (map)
    {
      suffixarray->suftab = genericmaptable(indexname,
                                            SUFTABSUFFIX,
                                            (unsigned long) (totallength+1),
                                            sizeof (Seqpos),
                                            err);
      if (suffixarray->suftab == NULL)
      {
        haserr = true;
      }
    } else
    {
      INITBufferedfile(indexname,&suffixarray->suftabstream,Seqpos,
                       SUFTABSUFFIX);
    }
    if (!haserr && !suffixarray->longest.defined)
    {
      gt_error_set(err,"longest not defined");
      haserr = true;
    }
  }
  if (!haserr && (demand & SARR_LCPTAB))
  {
    if (map)
    {
      suffixarray->lcptab = genericmaptable(indexname,
                                            LCPTABSUFFIX,
                                            (unsigned long) (totallength+1),
                                            sizeof (GtUchar),
                                            err);
      if (suffixarray->lcptab == NULL)
      {
        haserr = true;
      }
    } else
    {
      INITBufferedfile(indexname,&suffixarray->lcptabstream,GtUchar,
                       LCPTABSUFFIX);
      if (!haserr &&
          fseek(suffixarray->lcptabstream.fp,(long) sizeof (GtUchar),SEEK_SET))
      {
        gt_error_set(err,"fseek(esastream) failed: %s",strerror(errno));
        haserr = true;
      }
    }
    if (!haserr && !suffixarray->numoflargelcpvalues.defined)
    {
      gt_error_set(err,"numoflargelcpvalues not defined");
      haserr = true;
    }
    if (!haserr && suffixarray->numoflargelcpvalues.valueseqpos > 0)
    {
      if (map)
      {
        suffixarray->llvtab
          = genericmaptable(indexname,
                            LARGELCPTABSUFFIX,
                            (unsigned long) suffixarray->numoflargelcpvalues.
                            valueseqpos,
                            sizeof (Largelcpvalue),
                            err);
        if (suffixarray->llvtab == NULL)
        {
          haserr = true;
        }
      } else
      {
        INITBufferedfile(indexname,&suffixarray->llvtabstream,Largelcpvalue,
                         LARGELCPTABSUFFIX);
      }
    }
  }
  if (!haserr && (demand & SARR_BWTTAB))
  {
    if (map)
    {
      suffixarray->bwttab = genericmaptable(indexname,
                                            BWTTABSUFFIX,
                                            (unsigned long) (totallength+1),
                                            sizeof (GtUchar),
                                            err);
      if (suffixarray->bwttab == NULL)
      {
        haserr = true;
      }
    } else
    {
      INITBufferedfile(indexname,&suffixarray->bwttabstream,GtUchar,
                       BWTTABSUFFIX);
    }
  }
  if (!haserr && (demand & SARR_BCKTAB))
  {
    if (map)
    {
      suffixarray->bcktab = mapbcktab(indexname,
                                      getencseqAlphabetnumofchars(suffixarray->
                                                                  encseq),
                                      suffixarray->prefixlength,
                                      err);
      if (suffixarray->bcktab == NULL)
      {
        haserr = true;
      }
    } else
    {
      gt_error_set(err,"cannot stream bcktab");
      haserr = true;
    }
  }
  if (haserr)
  {
    freesuffixarray(suffixarray);
  }
  return haserr ? -1 : 0;
}
Exemple #2
0
int runidxlocali(const IdxlocaliOptions *idxlocalioptions,GtError *err)
{
  Genericindex *genericindex = NULL;
  bool haserr = false;
  Verboseinfo *verboseinfo;
  const Encodedsequence *encseq = NULL;

  verboseinfo = newverboseinfo(idxlocalioptions->verbose);

  if (idxlocalioptions->doonline)
  {
    encseq = mapencodedsequence (true,
                                 idxlocalioptions->indexname,
                                 true,
                                 false,
                                 false,
                                 true,
                                 verboseinfo,
                                 err);
    if (encseq == NULL)
    {
      haserr = true;
    }
  } else
  {
    genericindex = genericindex_new(idxlocalioptions->indexname,
                                    idxlocalioptions->withesa,
                                    idxlocalioptions->withesa ||
                                    idxlocalioptions->docompare,
                                    false,
                                    true,
                                    0,
                                    verboseinfo,
                                    err);
    if (genericindex == NULL)
    {
      haserr = true;
    } else
    {
      encseq = genericindex_getencseq(genericindex);
    }
  }
  if (!haserr)
  {
    GtSeqIterator *seqit;
    const GtUchar *query;
    unsigned long querylen;
    char *desc = NULL;
    int retval;
    Limdfsresources *limdfsresources = NULL;
    const AbstractDfstransformer *dfst;
    SWdpresource *swdpresource = NULL;
    Showmatchinfo showmatchinfo;
    Processmatch processmatch;
    void *processmatchinfoonline, *processmatchinfooffline;
    Storematchinfo storeonline, storeoffline;

    if (idxlocalioptions->docompare)
    {
      processmatch = storematch;
      initstorematch(&storeonline,encseq);
      initstorematch(&storeoffline,encseq);
      processmatchinfoonline = &storeonline;
      processmatchinfooffline = &storeoffline;
    } else
    {
      processmatch = showmatch;
      showmatchinfo.encseq = encseq;
      showmatchinfo.characters = getencseqAlphabetcharacters(encseq);
      showmatchinfo.wildcardshow = getencseqAlphabetwildcardshow(encseq);
      showmatchinfo.showalignment = idxlocalioptions->showalignment;
      processmatchinfoonline = processmatchinfooffline = &showmatchinfo;
    }
    if (idxlocalioptions->doonline || idxlocalioptions->docompare)
    {
      swdpresource = newSWdpresource(idxlocalioptions->matchscore,
                                     idxlocalioptions->mismatchscore,
                                     idxlocalioptions->gapextend,
                                     idxlocalioptions->threshold,
                                     idxlocalioptions->showalignment,
                                     processmatch,
                                     processmatchinfoonline);
    }
    dfst = locali_AbstractDfstransformer();
    if (!idxlocalioptions->doonline || idxlocalioptions->docompare)
    {
      gt_assert(genericindex != NULL);
      limdfsresources = newLimdfsresources(genericindex,
                                           true,
                                           0,
                                           0,    /* maxpathlength */
                                           true, /* keepexpandedonstack */
                                           processmatch,
                                           processmatchinfooffline,
                                           NULL, /* processresult */
                                           NULL, /* processresult info */
                                           dfst);
    }
    seqit = gt_seqiterator_new(idxlocalioptions->queryfiles, err);
    if (!seqit)
      haserr = true;
    if (!haserr)
    {
      gt_seqiterator_set_symbolmap(seqit, getencseqAlphabetsymbolmap(encseq));
      for (showmatchinfo.queryunit = 0; /* Nothing */;
           showmatchinfo.queryunit++)
      {
        retval = gt_seqiterator_next(seqit,
                                     &query,
                                     &querylen,
                                     &desc,
                                     err);
        if (retval < 0)
        {
          haserr = true;
          break;
        }
        if (retval == 0)
        {
          break;
        }
        printf("process sequence " Formatuint64_t " of length %lu\n",
                PRINTuint64_tcast(showmatchinfo.queryunit),querylen);
        if (idxlocalioptions->doonline || idxlocalioptions->docompare)
        {
          multiapplysmithwaterman(swdpresource,encseq,query,querylen);
        }
        if (!idxlocalioptions->doonline || idxlocalioptions->docompare)
        {
          indexbasedlocali(limdfsresources,
                           idxlocalioptions->matchscore,
                           idxlocalioptions->mismatchscore,
                           idxlocalioptions->gapstart,
                           idxlocalioptions->gapextend,
                           idxlocalioptions->threshold,
                           query,
                           querylen,
                           dfst);
        }
        if (idxlocalioptions->docompare)
        {
          checkandresetstorematch(showmatchinfo.queryunit,
                                  &storeonline,&storeoffline);
        }
        gt_free(desc);
      }
      if (limdfsresources != NULL)
      {
        freeLimdfsresources(&limdfsresources,dfst);
      }
      if (swdpresource != NULL)
      {
        freeSWdpresource(swdpresource);
        swdpresource = NULL;
      }
      gt_seqiterator_delete(seqit);
    }
    if (idxlocalioptions->docompare)
    {
      freestorematch(&storeonline);
      freestorematch(&storeoffline);
    }
  }
  if (genericindex == NULL)
  {
    gt_assert(encseq != NULL);
    encodedsequence_free((Encodedsequence **) &encseq);
  } else
  {
    genericindex_delete(genericindex);
  }
  freeverboseinfo(&verboseinfo);
  return haserr ? -1 : 0;
}