# -*- Mode: python; tab-width: 4; indent-tabs-mode:nil; -*-
# vim: tabstop=4 expandtab shiftwidth=4 softtabstop=4
#
# MDAnalysis --- https://www.mdanalysis.org
# Copyright (c) 2006-2017 The MDAnalysis Development Team and contributors
# (see the file AUTHORS for the full list of names)
#
# Released under the Lesser GNU Public Licence, v2.1 or any higher version
#
# Please cite your use of MDAnalysis in published work:
#
# R. J. Gowers, M. Linke, J. Barnoud, T. J. E. Reddy, M. N. Melo, S. L. Seyler,
# D. L. Dotson, J. Domanski, S. Buchoux, I. M. Kenney, and O. Beckstein.
# MDAnalysis: A Python package for the rapid analysis of molecular dynamics
# simulations. In S. Benthall and S. Rostrup editors, Proceedings of the 15th
# Python in Science Conference, pages 102-109, Austin, TX, 2016. SciPy.
# doi: 10.25080/majora-629e541a-00e
#
# N. Michaud-Agrawal, E. J. Denning, T. B. Woolf, and O. Beckstein.
# MDAnalysis: A Toolkit for the Analysis of Molecular Dynamics Simulations.
# J. Comput. Chem. 32 (2011), 2319--2327, doi:10.1002/jcc.21787
#
#

# cython: embedsignature=True


"""
Timestep Class --- :mod:`MDAnalysis.coordinates.timestep`
============================================================

Derive other Timestep, classes from the classes in this module.
The derived classes must follow the :ref:`Trajectory API`.

Timestep
--------

A :class:`Timestep` holds information for the current time frame in
the trajectory. It is one of the central data structures in
MDAnalysis.

.. class:: Timestep

   .. automethod:: __init__
   .. automethod:: from_coordinates
   .. automethod:: from_timestep
   .. autoattribute:: n_atoms
   .. attribute:: frame

      frame number (0-based)

      .. versionchanged:: 0.11.0
         Frames now 0-based; was 1-based


   .. autoattribute:: time
   .. autoattribute:: dt
   .. autoattribute:: positions
   .. autoattribute:: velocities
   .. autoattribute:: forces
   .. autoattribute:: has_positions
   .. autoattribute:: has_velocities
   .. autoattribute:: has_forces
   .. attribute:: _pos

      :class:`numpy.ndarray` of dtype :class:`~numpy.float32` of shape
      (*n_atoms*, 3) and internal C order, holding the raw
      cartesian coordinates (in MDAnalysis units, i.e. Å).

      .. Note::

         Normally one does not directly access :attr:`_pos` but uses
         the :attr:`~MDAnalysis.core.groups.AtomGroup.positions`
         attribute of a :class:`~MDAnalysis.core.groups.AtomGroup` but
         sometimes it can be faster to directly use the raw
         coordinates. Any changes to this array are immediately
         reflected in the atom positions. If the frame is written to a new
         trajectory then the coordinates are changed. If a new
         trajectory frame is loaded, then *all* contents of
         :attr:`_pos` are overwritten.

   .. attribute:: _velocities

      :class:`numpy.ndarray` of dtype :class:`~numpy.float32`. of shape
      (*n_atoms*, 3), holding the raw velocities (in MDAnalysis
      units, i.e., Å/ps).

      .. Note::

         Normally velocities are accessed through the
         :attr:`Timestep.velocities` attribute or the
         :attr:`~MDAnalysis.core.groups.AtomGroup.velocities`
         attribute of an :class:`~MDAnalysis.core.groups.AtomGroup`

         :attr:`~Timestep._velocities` only exists if the :attr:`has_velocities`
         flag is ``True``.

      .. versionadded:: 0.7.5

   .. attribute:: _forces

      :class:`numpy.ndarray` of dtype :class:`~numpy.float32`. of shape
      (*n_atoms*, 3), holding the forces

      :attr:`~Timestep._forces` only exists if :attr:`has_forces`
      is ``True``.

      .. versionadded:: 0.11.0
         Added as optional to :class:`Timestep`

   .. autoattribute:: dtype
   .. autoattribute:: dimensions
   .. autoattribute:: triclinic_dimensions
   .. autoattribute:: volume
   .. attribute:: data

      :class:`dict` that holds arbitrary per Timestep data

      .. versionadded:: 0.11.0

   .. automethod:: __getitem__
   .. automethod:: __eq__
   .. automethod:: __iter__
   .. automethod:: copy
   .. automethod:: copy_slice

"""


from ..lib.util import Namespace
from ..exceptions import NoDataError
from . import core
from libc.stdint cimport uint64_t
import weakref
import warnings
import copy
import numbers

import numpy as np
cimport numpy as cnp
cnp.import_array()


cdef class Timestep:
    """Timestep data for one frame

    :Methods:

      ``ts = Timestep(n_atoms)``

         create a timestep object with space for n_atoms


    .. versionchanged:: 0.11.0
       Added :meth:`from_timestep` and :meth:`from_coordinates` constructor
       methods.
       :class:`Timestep` init now only accepts integer creation.
       :attr:`n_atoms` now a read only property.
       :attr:`frame` now 0-based instead of 1-based.
       Attributes `status` and `step` removed.
    .. versionchanged:: 2.0.0
       Timestep now can be (un)pickled. Weakref for Reader
       will be dropped.
       Timestep now stores in to numpy array memory in 'C' order rather than
       'F' (Fortran).
    .. versionchanged:: 2.3.0
        Timestep is now a Cython extension type.
        All arrays are now forced to be C contiguous using the NumPy C API.
    """

    order = 'C'

    def __cinit__(self, uint64_t n_atoms, **kwargs):
        """Initialise C++ level parameters of a Timestep

        Parameters
        ----------
        n_atoms : uint64
          The total number of atoms this Timestep describes


        .. versionadded:: 2.3.0
           Initialise C++ level parameters
        """
        # c++ level objects
        self._n_atoms = n_atoms
        self.frame = -1

        # NOTE  This is currently hardcoded to match MDA always casting to F32
        # meaning the DTYPE set in the args is not respected.
        # to fix remove hardcode with introspection of a dtype kwarg following
        # discussion of appropriate casting rules
        self._typenum = cnp.NPY_FLOAT32

        self._has_positions = False
        self._has_velocities = False
        self._has_forces = False

        # track whether array has been allocated correct size
        self._positions_alloc = False
        self._velocities_alloc = False
        self._forces_alloc = False

        # use this to create numpy zeros and empties of the right shape using
        # NumPy C API
        self._particle_dependent_dim[0] = self._n_atoms
        self._particle_dependent_dim[1] = 3

        # unitcell uses a temp only as its size can vary
        cdef cnp.npy_intp unitcell_dim_tmp[1]
        unitcell_dim_tmp[0] = 6

        # use temps so that we don't have to allocate a bunch of empty
        # arrays of large size, eg for vel and frc
        cdef cnp.npy_intp particle_dependent_dim_tmp[2]
        particle_dependent_dim_tmp[0] = 0
        particle_dependent_dim_tmp[1] = 0

        # these must be initialised, we initialise size 0
        self._unitcell = cnp.PyArray_ZEROS(
            1, unitcell_dim_tmp, self._typenum, 0)
        self._pos = cnp.PyArray_EMPTY(
            2, particle_dependent_dim_tmp, self._typenum, 0)
        self._velocities = cnp.PyArray_EMPTY(
            2, particle_dependent_dim_tmp, self._typenum, 0)
        self._forces = cnp.PyArray_EMPTY(
            2, particle_dependent_dim_tmp, self._typenum, 0)

    def __init__(self, uint64_t n_atoms, **kwargs):
        """Create a Timestep, representing a frame of a trajectory

        Parameters
        ----------
        n_atoms : uint64
          The total number of atoms this Timestep describes
        positions : bool, optional
          Whether this Timestep has position information [``True``]
        velocities : bool (optional)
          Whether this Timestep has velocity information [``False``]
        forces : bool (optional)
          Whether this Timestep has force information [``False``]
        reader : Reader (optional)
          A weak reference to the owning Reader.  Used for
          when attributes require trajectory manipulation (e.g. dt)
        dt : float (optional)
          The time difference between frames (ps).  If :attr:`time`
          is set, then `dt` will be ignored.
        time_offset : float (optional)
          The starting time from which to calculate time (in ps)


        .. versionchanged:: 0.11.0
           Added keywords for `positions`, `velocities` and `forces`.
           Can add and remove position/velocity/force information by using
           the ``has_*`` attribute.
        .. versionchanged:: 2.3.0
           Added the `dtype` attribute hardcoded to :class:`~numpy.float32`.
        """
        # hardcoded
        self._dtype = np.float32

        self.data = {}

        for att in ('dt', 'time_offset'):
            try:
                self.data[att] = kwargs[att]
            except KeyError:
                pass
        try:
            # do I have a hook back to the Reader?
            # can possibly __cinit__ this with PyWeakRefNew
            self._reader = weakref.ref(kwargs['reader'])
        except KeyError:
            pass

        self.has_positions = kwargs.get('positions', True)
        self.has_velocities = kwargs.get('velocities', False)
        self.has_forces = kwargs.get('forces', False)

        # set up aux namespace for adding auxiliary data
        self.aux = Namespace()

    def __dealloc__(self):
        pass

    @property
    def n_atoms(self):
        """A read only view of the number of atoms this Timestep has

        .. versionchanged:: 0.11.0
           Changed to read only property
        """
        # In future could do some magic here to make setting n_atoms
        # resize the coordinate arrays, but
        # - not sure if that is ever useful
        # - not sure how to manage existing data upon extension
        return self._n_atoms

    @property
    def dtype(self):
        """The NumPy dtype of the timestep, all arrays in the timestep will
            have this dtype. Currently hardcoded to :class:`~numpy.float32`.

        .. versionadded:: 2.3.0
           Added dtype
        """
        return self._dtype

    @property
    def has_positions(self):
        """A boolean of whether this Timestep has position data
        This can be changed to ``True`` or ``False`` to allocate space for
        or remove the data.

        .. versionadded:: 0.11.0
        """
        return self._has_positions

    @has_positions.setter
    def has_positions(self, val):
        if val and not self._has_positions:
            if self._positions_alloc:  # already allocated
                # Setting this will always zero fill position data
                # ie
                # True -> False -> True will wipe data from first True state
                cnp.PyArray_FILLWBYTE(self._pos, 0)
                self._has_positions = True
            else:  # first time, we need to allocate to correct shape
                self._pos = cnp.PyArray_ZEROS(
                    2, self._particle_dependent_dim, self._typenum, 0)
                self._has_positions = True
                self._positions_alloc = True
        elif not val:
            # Unsetting val won't delete the numpy array
            self._has_positions = False

    @property
    def has_velocities(self):
        """A boolean of whether this Timestep has velocity data
        This can be changed to ``True`` or ``False`` to allocate space for
        or remove the data.

        .. versionadded:: 0.11.0
        """
        return self._has_velocities

    @has_velocities.setter
    def has_velocities(self, val):
        if val and not self._has_velocities:
            if self._velocities_alloc:  # already allocated
                # Setting this will always zero fill velocity data
                # ie
                # True -> False -> True will wipe data from first True state
                cnp.PyArray_FILLWBYTE(self._velocities, 0)
                self._has_velocities = True
            else:  # first time, we need to allocate to correct shape
                self._velocities = cnp.PyArray_ZEROS(
                    2, self._particle_dependent_dim, self._typenum, 0)
                self._has_velocities = True
                self._velocities_alloc = True
        elif not val:
            # Unsetting val won't delete the numpy array
            self._has_velocities = False

    @property
    def has_forces(self):
        """A boolean of whether this Timestep has force data
        This can be changed to ``True`` or ``False`` to allocate space for
        or remove the data.

        .. versionadded:: 0.11.0
        """
        return self._has_forces

    @has_forces.setter
    def has_forces(self, val):
        if val and not self._has_forces:
            if self._forces_alloc:  # already allocated
                # Setting this will always zero fill force data
                # ie
                # True -> False -> True will wipe data from first True state
                cnp.PyArray_FILLWBYTE(self._forces, 0)
                self._has_forces = True
            else:  # first time, we need to allocate to correct shape
                self._forces = cnp.PyArray_ZEROS(
                    2, self._particle_dependent_dim, self._typenum, 0)
                self._has_forces = True
                self._forces_alloc = True
        elif not val:
            # Unsetting val won't delete the numpy array
            self._has_forces = False

    @property
    def positions(self):
        """A record of the positions of all atoms in this Timestep
        Setting this attribute will add positions to the Timestep if they
        weren't originally present.

        Returns
        -------
        positions : numpy.ndarray with dtype numpy.float32
               position data of shape ``(n_atoms, 3)`` for all atoms

        Raises
        ------
        :exc:`MDAnalysis.exceptions.NoDataError`
               if the Timestep has no position data


        .. versionchanged:: 0.11.0
           Now can raise :exc:`NoDataError` when no position data present
        """
        if self._has_positions:
            return self._pos
        else:
            raise NoDataError("This Timestep has no position information")

    @positions.setter
    def positions(self,  new_positions):
        self.has_positions = True
        if cnp.PyArray_Check(new_positions):  # is it an array?
            cnp.PyArray_CopyInto(self._pos, new_positions)
            # copy into target, handles dtype conversion
        else:
            self._pos[:] = new_positions

    @property
    def _x(self):
        """A view onto the x dimension of position data

        .. versionchanged:: 0.11.0
           Now read only
        """
        return self.positions[:, 0]

    @property
    def _y(self):
        """A view onto the y dimension of position data

        .. versionchanged:: 0.11.0
           Now read only
        """
        return self.positions[:, 1]

    @property
    def _z(self):
        """A view onto the z dimension of position data

        .. versionchanged:: 0.11.0
           Now read only
        """
        return self.positions[:, 2]

    @property
    def dimensions(self):
        """View of unitcell dimensions (*A*, *B*, *C*, *alpha*, *beta*, *gamma*)

        lengths *a*, *b*, *c* are in the MDAnalysis length unit (Å), and
        angles are in degrees.
        """
        if (self._unitcell[:3] == 0).all():
            return None
        else:
            return self._unitcell

    @dimensions.setter
    def dimensions(self,  new_dimensions):
        if new_dimensions is None:
            self._unitcell[:] = 0
        else:
            self._unitcell[:] = new_dimensions

    @property
    def volume(self):
        """volume of the unitcell"""
        if self.dimensions is None:
            return 0
        else:
            return core.box_volume(self.dimensions)

    @property
    def triclinic_dimensions(self):
        """The unitcell dimensions represented as triclinic vectors

        Returns
        -------
        numpy.ndarray
             A (3, 3) numpy.ndarray of unit cell vectors

        Examples
        --------
        The unitcell for a given system can be queried as either three
        vectors lengths followed by their respective angle, or as three
        triclinic vectors.

          >>> ts.dimensions
          array([ 13.,  14.,  15.,  90.,  90.,  90.], dtype=float32)
          >>> ts.triclinic_dimensions
          array([[ 13.,   0.,   0.],
                 [  0.,  14.,   0.],
                 [  0.,   0.,  15.]], dtype=float32)

        Setting the attribute also works::

          >>> ts.triclinic_dimensions = [[15, 0, 0], [5, 15, 0], [5, 5, 15]]
          >>> ts.dimensions
          array([ 15.        ,  15.81138802,  16.58312416,  67.58049774,
                  72.45159912,  71.56504822], dtype=float32)

        See Also
        --------
        :func:`MDAnalysis.lib.mdamath.triclinic_vectors`


        .. versionadded:: 0.11.0
        """
        if self.dimensions is None:
            return None
        else:
            return core.triclinic_vectors(self.dimensions)

    @triclinic_dimensions.setter
    def triclinic_dimensions(self, new_dimensions):
        """Set the unitcell for this Timestep as defined by triclinic vectors
        .. versionadded:: 0.11.0
        """
        if new_dimensions is None:
            self.dimensions = None
        else:
            self.dimensions = core.triclinic_box(*new_dimensions)

    @property
    def velocities(self):
        """A record of the velocities of all atoms in this Timestep
        Setting this attribute will add velocities to the Timestep if they
        weren't originally present.

        Returns
        -------
        velocities : numpy.ndarray with dtype numpy.float32
               velocity data of shape ``(n_atoms, 3)`` for all atoms

        Raises
        ------
        :exc:`MDAnalysis.exceptions.NoDataError`
               if the Timestep has no velocity data


        .. versionadded:: 0.11.0
        """
        if self._has_velocities:
            return self._velocities
        else:
            raise NoDataError("This Timestep has no velocities information")

    @velocities.setter
    def velocities(self,  new_velocities):
        self.has_velocities = True
        if cnp.PyArray_Check(new_velocities):  # is it an array?
            cnp.PyArray_CopyInto(self._velocities, new_velocities)
            # copy into target, handles dtype conversion
        else:
            self._velocities[:] = new_velocities

    @property
    def forces(self):
        """A record of the forces of all atoms in this Timestep
        Setting this attribute will add forces to the Timestep if they
        weren't originally present.

        Returns
        -------

        forces : numpy.ndarray with dtype numpy.float32
               force data of shape ``(n_atoms, 3)`` for all atoms
        Raises
        ------

        :exc:`MDAnalysis.exceptions.NoDataError`
               if the Timestep has no force data
        .. versionadded:: 0.11.0
        """
        if self._has_forces:
            return self._forces
        else:
            raise NoDataError("This Timestep has no force information")

    @forces.setter
    def forces(self,  new_forces):
        self.has_forces = True
        if cnp.PyArray_Check(new_forces):  # is it an array?
            cnp.PyArray_CopyInto(self._forces, new_forces)
            # copy into target, handles dtype conversion
        else:
            self._forces[:] = new_forces

    @classmethod
    def from_timestep(cls, Timestep other, **kwargs):
        """Create a copy of another Timestep, in the format of this Timestep

        .. versionadded:: 0.11.0
        """
        ts = cls(other.n_atoms,
                 positions=other.has_positions,
                 velocities=other.has_velocities,
                 forces=other.has_forces,
                 **kwargs)
        ts.frame = other.frame
        if other.dimensions is not None:
            ts.dimensions = other.dimensions.copy(order=cls.order)
        try:
            ts.positions = other.positions.copy(order=cls.order)
        except NoDataError:
            pass
        try:
            ts.velocities = other.velocities.copy(order=cls.order)
        except NoDataError:
            pass
        try:
            ts.forces = other.forces.copy(order=cls.order)
        except NoDataError:
            pass

        try:
            other._reader = weakref.ref(ts._reader())
        except TypeError:  # TypeError from calling None weakref
            pass

        ts.data = copy.deepcopy(other.data)

        return ts

    @classmethod
    def from_coordinates(cls,
                         positions=None,
                         velocities=None,
                         forces=None,
                         **kwargs):
        """Create an instance of this Timestep, from coordinate data

        Can pass position, velocity and force data to form a Timestep.

        .. versionadded:: 0.11.0
        """
        has_positions = positions is not None
        has_velocities = velocities is not None
        has_forces = forces is not None

        lens = [len(a) for a in [positions, velocities, forces]
                if a is not None]
        if not lens:
            raise ValueError("Must specify at least one set of data")
        n_atoms = max(lens)
        # Check arrays are matched length?
        if not all(val == n_atoms for val in lens):
            raise ValueError("Lengths of input data mismatched")

        ts = cls(n_atoms,
                 positions=has_positions,
                 velocities=has_velocities,
                 forces=has_forces,
                 **kwargs)
        if has_positions:
            ts.positions = positions
        if has_velocities:
            ts.velocities = velocities
        if has_forces:
            ts.forces = forces

        return ts

    def __eq__(self, other):
        """Compare with another Timestep

        .. versionadded:: 0.11.0
        """
        if not isinstance(other, Timestep):
            return NotImplemented

        if not self.frame == other.frame:
            return False

        if not self.n_atoms == other.n_atoms:
            return False

        if not self.has_positions == other.has_positions:
            return False
        if self.has_positions:
            if not (self.positions == other.positions).all():
                return False

        if self.dimensions is None:
            if other.dimensions is not None:
                return False
        else:
            if other.dimensions is None:
                return False
            if not (self.dimensions == other.dimensions).all():
                return False

        if not self.has_velocities == other.has_velocities:
            return False
        if self.has_velocities:
            if not (self.velocities == other.velocities).all():
                return False

        if not self.has_forces == other.has_forces:
            return False
        if self.has_forces:
            if not (self.forces == other.forces).all():
                return False

        return True

    # __ne__ is defers to __eq__ and inverts

    def __getitem__(self, atoms):
        """Get a selection of coordinates

        ``ts[i]``

           return coordinates for the i'th atom (0-based)

        ``ts[start:stop:skip]``

           return an array of coordinates, where start, stop and skip
           correspond to atom indices,
           :attr:`MDAnalysis.core.groups.Atom.index` (0-based)
        """
        if isinstance(atoms, numbers.Integral):
            return self._pos[atoms]
        elif isinstance(atoms, (slice, cnp.ndarray)):
            return self._pos[atoms]
        else:
            raise TypeError

    def __getattr__(self, attr):
        # special-case timestep info
        if attr in ('velocities', 'forces', 'positions'):
            raise NoDataError('This Timestep has no ' + attr)
        err = "{selfcls} object has no attribute '{attr}'"
        raise AttributeError(err.format(selfcls=type(self).__name__,
                                        attr=attr))

    def __len__(self):
        return self.n_atoms

    def __iter__(self):
        """Iterate over coordinates

        ``for x in ts``

        iterate of the coordinates, atom by atom
        """
        for i in range(self.n_atoms):
            yield self[i]

    def __repr__(self):
        desc = "< Timestep {0}".format(self.frame)
        if self.dimensions is not None:
            tail = " with unit cell dimensions {0} >".format(self.dimensions)
        else:
            tail = " >"
        return desc + tail

    def copy(self):
        """Make an independent ("deep") copy of the whole :class:`Timestep`."""
        return self.__deepcopy__()

    def __deepcopy__(self):
        return self.from_timestep(self)

    def __getstate__(self):
        """Make a dictionary of the class state to pickle Timestep instance.

           Must be done manually as Extension types do not have the __dict__
           class attribute and we use a non-trivial `__cinit__`. This means
           that cython cannot automatically  generate an `__reduce__` method
           for us.

        .. versionadded:: 2.3.0
        """
        state = {
            "frame": self.frame,
            "_n_atoms": self._n_atoms,
            "_has_positions": self._has_positions,
            "_has_velocities": self._has_velocities,
            "_has_forces": self._has_forces,

            "_unitcell": self._unitcell,
            "_pos": self._pos,
            "_velocities": self._velocities,
            "_forces": self._forces,

            "_dtype": self._dtype,
            "data": self.data,
            "aux": self.aux,
            "dt": self.dt
        }
        return state

    def __getnewargs_ex__(self):
        """Specify arguments to use in `__cinit__` and `__init__` to use in
           unpickling of timestep instance

        .. versionchanged:: 2.3.0
           removed implementations that use `__dict__` class attribute

        """
        return (self.n_atoms,), {}

    def __setstate__(self, state):
        """Restore class from `state` dictionary in unpickling of Timestep
           instance

        .. versionchanged:: 2.3.0
           removed implementations that use `__dict__` class attribute

        """
        self.frame = state["frame"]
        self._n_atoms = state["_n_atoms"]
        self.has_positions = state["_has_positions"]
        self._has_velocities = state["_has_velocities"]
        self._has_forces = state["_has_forces"]
        self._unitcell = state["_unitcell"]
        self._pos = state["_pos"]
        self._velocities = state["_velocities"]
        self._forces = state["_forces"]
        self._dtype = state["_dtype"]
        self.data = state["data"]
        self.aux = state["aux"]

    def copy_slice(self, sel):
        """Make a new `Timestep` containing a subset of the original `Timestep`.

        Parameters
        ----------
        sel : array_like or slice
            The underlying position, velocity, and force arrays are sliced
            using a :class:`list`, :class:`slice`, or any array-like.

        Returns
        -------
        :class:`Timestep`
            A `Timestep` object of the same type containing all header
            information and all atom information relevant to the selection.

        Note
        ----
        The selection must be a 0 based :class:`slice` or array of the atom indices
        in this :class:`Timestep`

        Example
        -------
        Using a Python :class:`slice` object::

           new_ts = ts.copy_slice(slice(start, stop, step))

        Using a list of indices::

           new_ts = ts.copy_slice([0, 2, 10, 20, 23])


        .. versionadded:: 0.8
        .. versionchanged:: 0.11.0
           Reworked to follow new Timestep API.  Now will strictly only
           copy official attributes of the Timestep.

        """
        # Detect the size of the Timestep by doing a dummy slice
        try:
            pos = self.positions[sel, :]
        except NoDataError:
            # It's cool if there's no Data, we'll live
            pos = None
        except Exception:
            errmsg = ("Selection type must be compatible with slicing the "
                      "coordinates")
            raise TypeError(errmsg) from None
        try:
            vel = self.velocities[sel, :]
        except NoDataError:
            vel = None
        except Exception:
            errmsg = ("Selection type must be compatible with slicing the "
                      "coordinates")
            raise TypeError(errmsg) from None
        try:
            force = self.forces[sel, :]
        except NoDataError:
            force = None
        except Exception:
            errmsg = ("Selection type must be compatible with slicing the "
                      "coordinates")
            raise TypeError(errmsg) from None

        new_TS = self.__class__.from_coordinates(
            positions=pos,
            velocities=vel,
            forces=force)

        new_TS.dimensions = self.dimensions

        new_TS.frame = self.frame

        try:
            new_TS._reader = weakref.ref(self._reader())
        except TypeError:  # TypeError from calling None weakref
            pass
        new_TS.data = copy.deepcopy(self.data)

        return new_TS

    @property
    def dt(self):
        """The time difference in ps between timesteps

        Note
        ----
        This defaults to 1.0 ps in the absence of time data


        .. versionadded:: 0.11.0
        """
        try:
            return self.data['dt']
        except KeyError:
            pass
        try:
            dt = self.data['dt'] = self._reader()._get_dt()
            return dt
        # TypeError from calling None weakref
        # AttributeError from ._get_dt()
        except (TypeError, AttributeError):
            pass

        warnings.warn("Reader has no dt information, set to 1.0 ps")
        return 1.0

    @dt.setter
    def dt(self, new_dt):
        self.data['dt'] = new_dt

    @dt.deleter
    def dt(self):
        del self.data['dt']

    @property
    def time(self):
        """The time in ps of this timestep

        This is calculated as::

          time = ts.data['time_offset'] + ts.time

        Or, if the trajectory doesn't provide time information::

          time = ts.data['time_offset'] + ts.frame * ts.dt

        .. versionadded:: 0.11.0
        """
        offset = self.data.get('time_offset', 0)
        try:
            return self.data['time'] + offset
        except KeyError:
            return self.dt * self.frame + offset

    @time.setter
    def time(self, new_time):
        self.data['time'] = new_time

    @time.deleter
    def time(self):
        del self.data['time']