GtSeqIterator* gt_seq_iterator_sequence_buffer_new(const GtStrArray
                                                                   *filenametab,
                                                  GtError *err)
{
  GtSeqIterator *si;
  GtSequenceBuffer *sb = gt_sequence_buffer_new_guess_type(filenametab, err);
  if (!sb)
    return NULL;
  si = gt_seq_iterator_sequence_buffer_new_with_buffer(sb);
  gt_sequence_buffer_delete(sb); /* drop this reference */
  return si;
}
Ejemplo n.º 2
0
static int gt_seqtranslate_runner(int argc, const char **argv, int parsed_args,
                              void *tool_arguments, GT_UNUSED GtError *err)
{
  GtTranslateArguments *arguments = tool_arguments;
  GtSeqIterator *si = NULL;
  GtSequenceBuffer *sb = NULL;
  GtStrArray *infiles;
  int had_err = 0,
      rval,
      i;
  GtStr *translations[3];
  translations[0] = gt_str_new();
  translations[1] = gt_str_new();
  translations[2] = gt_str_new();

  gt_error_check(err);
  gt_assert(arguments);

  infiles = gt_str_array_new();
  for (i = parsed_args; i < argc; i++) {
    gt_str_array_add_cstr(infiles, argv[i]);
  }
  sb = gt_sequence_buffer_new_guess_type(infiles, err);
  if (!sb)
    had_err = -1;
  if (!had_err) {
    si = gt_seq_iterator_sequence_buffer_new_with_buffer(sb);
    if (!si)
      had_err = -1;
  }
  if (!had_err) {
    char *desc;
    const GtUchar *sequence;
    GtUword len;
    while (!had_err && (rval = gt_seq_iterator_next(si,
                                                   &sequence,
                                                   &len, &desc, err))) {
      if (rval < 0) {
        had_err = -1;
        break;
      }
      if (len < GT_CODON_LENGTH) {
        gt_warning("sequence '%s' is shorter than codon length of %d, skipping",
                   desc, GT_CODON_LENGTH);
      } else {
        had_err = gt_seqtranslate_do_translation(arguments, (char*) sequence,
                                                 len, desc,
                                                 translations, false, err);
        if (!had_err && arguments->reverse) {
          char *revseq = gt_cstr_dup_nt((char*) sequence, len);
          had_err = gt_reverse_complement(revseq, len, err);
          if (!had_err) {
            had_err = gt_seqtranslate_do_translation(arguments, revseq, len,
                                                  desc, translations, true,
                                                  err);
          }
          gt_free(revseq);
        }
      }
    }
  }
  gt_str_delete(translations[0]);
  gt_str_delete(translations[1]);
  gt_str_delete(translations[2]);
  gt_str_array_delete(infiles);
  gt_seq_iterator_delete(si);
  gt_sequence_buffer_delete(sb);
  return had_err;
}
Ejemplo n.º 3
0
static int gt_convertseq_runner(int argc, const char **argv, int parsed_args,
                              void *tool_arguments, GtError *err)
{
  GtConvertseqArguments *arguments = tool_arguments;
  int had_err = 0, i;
  GtFilelengthvalues *flv;
  GtSeqIterator *seqit;
  GtSequenceBuffer *sb = NULL;
  GtStrArray *files;
  const GtUchar *sequence;
  char *desc;
  GtUword len, j;
  off_t totalsize;
  gt_error_check(err);
  gt_assert(arguments != NULL);

  files = gt_str_array_new();
  for (i = parsed_args; i < argc; i++)
  {
    gt_str_array_add_cstr(files, argv[i]);
  }
  totalsize = gt_files_estimate_total_size(files);

  flv = gt_calloc((size_t) gt_str_array_size(files),
                  sizeof (GtFilelengthvalues));

  sb = gt_sequence_buffer_new_guess_type(files, err);
  if (!sb) {
    had_err = -1;
  }
  if (!had_err) {
    gt_sequence_buffer_set_filelengthtab(sb, flv);
    /* read input using seqiterator */
    seqit = gt_seq_iterator_sequence_buffer_new_with_buffer(sb);
    if (arguments->verbose)
    {
      gt_progressbar_start(gt_seq_iterator_getcurrentcounter(seqit,
                                                           (GtUint64)
                                                           totalsize),
                           (GtUint64) totalsize);
    }
    while (true)
    {
      GtUchar *seq = NULL;
      desc = NULL;
      j = 0UL;
      had_err = gt_seq_iterator_next(seqit, &sequence, &len, &desc, err);
      if (had_err != 1)
        break;
      if (arguments->revcomp) {
        GtUchar *newseq = gt_calloc((size_t) len+1, sizeof (GtUchar));
        memcpy(newseq, sequence, (size_t) len*sizeof (GtUchar));
        had_err = gt_reverse_complement((char*) newseq, len, err);
        if (had_err)
          break;
        seq = newseq;
      } else seq = (GtUchar*) sequence;

      if (!arguments->showseq) {
        bool in_wildcard = false;
        gt_file_xprintf(arguments->outfp, ">%s\n", desc);
        for (i = 0; (GtUword) i < len; i++) {
          if (arguments->reduce_wc_dna) {
            switch (seq[i]) {
              case 'a':
              case 'A':
              case 'c':
              case 'C':
              case 'g':
              case 'G':
              case 't':
              case 'u':
              case 'T':
              case 'U':
                in_wildcard = false;
                gt_file_xfputc((int) seq[i], arguments->outfp);
                j++;
                break;
              default:
                if (!in_wildcard) {
                  in_wildcard = true;
                  if (isupper((int) seq[i]))
                    gt_file_xfputc((int) 'N', arguments->outfp);
                  else
                    gt_file_xfputc((int) 'n', arguments->outfp);
                  j++;
                }
            }
          }
          else if (arguments->reduce_wc_prot) {
            switch (seq[i]) {
              case 'X':
              case 'B':
              case 'Z':
                if (!in_wildcard) {
                  in_wildcard = true;
                  gt_file_xfputc((int) 'N', arguments->outfp);
                  j++;
                }
                break;
              case 'x':
              case 'b':
              case 'z':
                if (!in_wildcard) {
                  in_wildcard = true;
                  gt_file_xfputc((int) 'n', arguments->outfp);
                  j++;
                }
                break;
              default:
                in_wildcard = false;
                gt_file_xfputc((int) seq[i], arguments->outfp);
                j++;
            }
          }
          else {
            gt_file_xfputc((int) seq[i], arguments->outfp);
            j++;
          }
          if (arguments->fastawidth > 0 && j % arguments->fastawidth == 0) {
            j = 0;
            gt_file_xprintf(arguments->outfp, "\n");
          }
        }
        if (arguments->fastawidth == 0 || len % arguments->fastawidth != 0)
            gt_file_xprintf(arguments->outfp, "\n");
      }
      if (arguments->revcomp) {
        gt_free(seq);
      }
    }
    if (arguments->showflv) {
      for (j=0;j<gt_str_array_size(files);j++) {
        fprintf(stderr, "file "GT_WU" (%s): "GT_WU"/"GT_WU"\n",
               j,
               gt_str_array_get(files, j),
               (GtUword) flv[j].length,
               (GtUword) flv[j].effectivelength);
      }
    }
    if (arguments->verbose)
    {
      gt_progressbar_stop();
    }
    gt_sequence_buffer_delete(sb);
    gt_seq_iterator_delete(seqit);
  }
  gt_str_array_delete(files);
  gt_free(flv);

  return had_err;
}