Class DelimitedOutputFormatter

shell/impala_shell/shell_output.py:83–116

class DelimitedOutputFormatter(object):
  def __init__(self, field_delim="\t"):
    if field_delim:
      if sys.version_info.major > 2:
        # strings do not have a 'decode' method in python 3
        field_delim_bytes = bytearray(field_delim, 'utf-8')
        self.field_delim = field_delim_bytes.decode('unicode_escape')
      else:
        # csv.writer in python2 requires an ascii string delimiter
        self.field_delim = field_delim.decode('unicode_escape').encode('ascii', 'ignore')
      # IMPALA-8652, the delimiter should be a 1-character string and verified already
      assert len(self.field_delim) == 1

  def format(self, rows):
    """Returns string containing UTF-8-encoded representation of the table data."""
    # csv.writer expects a file handle to the input.
    temp_buffer = StringIO()
    writer = csv.writer(temp_buffer, delimiter=self.field_delim,
                        lineterminator='\n', quoting=csv.QUOTE_MINIMAL)
    for row in rows:
      if sys.version_info.major == 2:
        row = [val.encode('utf-8', 'replace') if isinstance(val, unicode)  # noqa: F821
            else val for val in row]
      writer.writerow(row)
    # The CSV writer produces an extra newline. Strip that extra newline (and
    # only that extra newline). csv wraps newlines for data values in quotes,
    # so rstrip will be limited to the extra newline.
    if sys.version_info.major == 2:
      # Python 2 is in encoded Unicode bytes, so this needs to be a bytes \n.
      rows = temp_buffer.getvalue().rstrip(b'\n')
    else:
      rows = temp_buffer.getvalue().rstrip('\n')
    temp_buffer.close()
    return rows


class VerticalOutputFormatter(DelimitedOutputFormatter):