Method fixFloatingPointValue

src/core/Internal.java:577–591

Returns a corrected value if this is a floating point value to fix. OpenTSDB used to encode all floating point values as `float' (4 bytes) but actually store them on 8 bytes, with 4 leading 0 bytes, and flags correctly stating the value was on 4 bytes. This function detects such values and r

(final byte flags,
                                              final byte[] value)

   * @throws IllegalDataException if the value is malformed.
   */
  public static byte[] fixFloatingPointValue(final byte flags,
                                              final byte[] value) {
    if (floatingPointValueToFix(flags, value)) {
      // The first 4 bytes should really be zeros.
      if (value[0] == 0 && value[1] == 0 && value[2] == 0 && value[3] == 0) {
        // Just keep the last 4 bytes.
        return new byte[] { value[4], value[5], value[6], value[7] };
      } else {  // Very unlikely.
        throw new IllegalDataException("Corrupted floating point value: "
          + Arrays.toString(value) + " flags=0x" + Integer.toHexString(flags)
          + " -- first 4 bytes are expected to be zeros.");
      }
    }
    return value;
  }
  
  /**
   * Determines if the qualifier is in milliseconds or not