# Licensed under a 3-clause BSD style license - see LICENSE.rst
from __future__ import absolute_import, division, print_function, unicode_literals
import sys
import numpy as np
from astropy.table import Table
from astropy.units import Quantity
from astropy.coordinates import Angle, SkyCoord
from astropy.time import Time
from astropy.utils import lazyproperty
from ..utils.scripts import make_path
from ..utils.time import time_relative_to_ref
__all__ = [
'ObservationTable',
]
[docs]class ObservationTable(Table):
"""Observation table.
This is an `~astropy.table.Table` sub-class, with a few
convenience methods. The format of the observation table
is described in :ref:`dataformats_observation_lists`.
"""
@classmethod
[docs] def read(cls, filename, **kwargs):
"""Read an observation table from file.
Parameters
----------
filename : `~gammapy.extern.pathlib.Path`, str
Filename
"""
filename = make_path(filename)
return super(ObservationTable, cls).read(str(filename), **kwargs)
@property
def pointing_radec(self):
"""Pointing positions as ICRS (`~astropy.coordinates.SkyCoord`)"""
return SkyCoord(self['RA_PNT'], self['DEC_PNT'], unit='deg', frame='icrs')
@property
def pointing_galactic(self):
"""Pointing positions as Galactic (`~astropy.coordinates.SkyCoord`)"""
return SkyCoord(self['GLON_PNT'], self['GLAT_PNT'], unit='deg', frame='galactic')
@lazyproperty
def _index_dict(self):
"""Dict containing row index for all obs ids"""
# TODO: Switch to http://docs.astropy.org/en/latest/table/indexing.html once it is more stable
temp = (zip(self['OBS_ID'], np.arange(len(self))))
return dict(temp)
[docs] def get_obs_idx(self, obs_id):
"""Get row index for given ``obs_id``.
Raises KeyError if observation is not available.
Parameters
----------
obs_id : int, list
observation ids
Returns
-------
idx : list
indices corresponding to obs_id
"""
idx = [self._index_dict[key] for key in np.atleast_1d(obs_id)]
return idx
[docs] def select_obs_id(self, obs_id):
"""Get `~gammapy.data.ObservationTable` containing only ``obs_id``.
Raises KeyError if observation is not available.
Parameters
----------
obs_id: int, list
observation ids
"""
return self[self.get_obs_idx(obs_id)]
[docs] def summary(self, file=None):
"""Info string (str)"""
if not file:
file = sys.stdout
print('Observation table:', file=file)
if 'OBSERVATORY_NAME' in self.meta:
obs_name = self.meta['OBSERVATORY_NAME']
print('Observatory name: {}'.format(obs_name), file=file)
print('Number of observations: {}'.format(len(self)), file=file)
# TODO: clean this up. Make those properties?
# ontime = Quantity(self['ONTIME'].sum(), self['ONTIME'].unit)
#
# ss += 'Total observation time: {}\n'.format(ontime)
# livetime = Quantity(self['LIVETIME'].sum(), self['LIVETIME'].unit)
# ss += 'Total live time: {}\n'.format(livetime)
# dtf = 100. * (1 - livetime / ontime)
# ss += 'Average dead time fraction: {:5.2f}%\n'.format(dtf)
# time_ref = time_ref_from_dict(self.meta)
# time_ref_unit = time_ref_from_dict(self.meta).format
# ss += 'Time reference: {} {}'.format(time_ref, time_ref_unit)
#
# return ss
[docs] def select_linspace_subset(self, num):
"""Select subset of observations.
This is mostly useful for testing, if you want to make
the analysis run faster.
Parameters
----------
num : int
Number of samples to select.
Returns
-------
table : `ObservationTable`
Subset observation table (a copy).
"""
indices = np.linspace(start=0, stop=len(self), num=num, endpoint=False)
# Round down to nearest integer
indices = indices.astype('int')
return self[indices]
[docs] def select_range(self, selection_variable, value_range, inverted=False):
"""Make an observation table, applying some selection.
Generic function to apply a 1D box selection (min, max) to a
table on any variable that is in the observation table and can
be casted into a `~astropy.units.Quantity` object.
If the range length is 0 (min = max), the selection is applied
to the exact value indicated by the min value. This is useful
for selection of exact values, for instance in discrete
variables like the number of telescopes.
If the inverted flag is activated, the selection is applied to
keep all elements outside the selected range.
Parameters
----------
selection_variable : str
Name of variable to apply a cut (it should exist on the table).
value_range : `~astropy.units.Quantity`-like
Allowed range of values (min, max). The type should be
consistent with the selection_variable.
inverted : bool, optional
Invert selection: keep all entries outside the (min, max) range.
Returns
-------
obs_table : `~gammapy.data.ObservationTable`
Observation table after selection.
"""
value_range = Quantity(value_range)
# read values into a quantity in case units have to be taken into account
value = Quantity(self[selection_variable])
mask = (value_range[0] <= value) & (value < value_range[1])
if np.allclose(value_range[0].value, value_range[1].value):
mask = (value_range[0] == value)
if inverted:
mask = np.invert(mask)
return self[mask]
[docs] def select_time_range(self, selection_variable, time_range, inverted=False):
"""Make an observation table, applying a time selection.
Apply a 1D box selection (min, max) to a
table on any time variable that is in the observation table.
It supports both fomats: absolute times in
`~astropy.time.Time` variables and [MET]_.
If the inverted flag is activated, the selection is applied to
keep all elements outside the selected range.
Parameters
----------
selection_variable : str
Name of variable to apply a cut (it should exist on the table).
time_range : `~astropy.time.Time`
Allowed time range (min, max).
inverted : bool, optional
Invert selection: keep all entries outside the (min, max) range.
Returns
-------
obs_table : `~gammapy.data.ObservationTable`
Observation table after selection.
"""
if self.meta['TIME_FORMAT'] == 'absolute':
# read times into a Time object
time = Time(self[selection_variable])
else:
# transform time to MET
time_range = time_relative_to_ref(time_range, self.meta)
# read values into a quantity in case units have to be taken into account
time = Quantity(self[selection_variable])
mask = (time_range[0] <= time) & (time < time_range[1])
if inverted:
mask = np.invert(mask)
return self[mask]
[docs] def select_observations(self, selection=None):
"""Select subset of observations.
Returns a new observation table representing the subset.
There are 3 main kinds of selection criteria, according to the
value of the **type** keyword in the **selection** dictionary:
- sky regions (boxes or circles)
- time intervals (min, max)
- intervals (min, max) on any other parameter present in the
observation table, that can be casted into an
`~astropy.units.Quantity` object
Allowed selection criteria are interpreted using the following
keywords in the **selection** dictionary under the **type** key.
- ``sky_box`` and ``sky_circle`` are 2D selection criteria acting
on sky coordinates
- ``sky_box`` is a squared region delimited by the **lon** and
**lat** keywords: both tuples of format (min, max); uses
`~gammapy.catalog.select_sky_box`
- ``sky_circle`` is a circular region centered in the coordinate
marked by the **lon** and **lat** keywords, and radius **radius**;
uses `~gammapy.catalog.select_sky_circle`
in each case, the coordinate system can be specified by the **frame**
keyword (built-in Astropy coordinate frames are supported, e.g.
``icrs`` or ``galactic``); an aditional border can be defined using
the **border** keyword
- ``time_box`` is a 1D selection criterion acting on the observation
start time (**TSTART**); the interval is set via the
**time_range** keyword; uses
`~gammapy.data.ObservationTable.select_time_range`
- ``par_box`` is a 1D selection criterion acting on any
parameter defined in the observation table that can be casted
into an `~astropy.units.Quantity` object; the parameter name
and interval can be specified using the keywords **variable** and
**value_range** respectively; min = max selects exact
values of the parameter; uses
`~gammapy.data.ObservationTable.select_range`
In all cases, the selection can be inverted by activating the
**inverted** flag, in which case, the selection is applied to keep all
elements outside the selected range.
A few examples of selection criteria are given below.
Parameters
----------
selection : dict
Dictionary with a few keywords for applying selection cuts.
Returns
-------
obs_table : `~gammapy.data.ObservationTable`
Observation table after selection.
Examples
--------
>>> selection = dict(type='sky_box', frame='icrs',
... lon=Angle([150, 300], 'deg'),
... lat=Angle([-50, 0], 'deg'),
... border=Angle(2, 'deg'))
>>> selected_obs_table = obs_table.select_observations(selection)
>>> selection = dict(type='sky_circle', frame='galactic',
... lon=Angle(0, 'deg'),
... lat=Angle(0, 'deg'),
... radius=Angle(5, 'deg'),
... border=Angle(2, 'deg'))
>>> selected_obs_table = obs_table.select_observations(selection)
>>> selection = dict(type='time_box',
... time_range=Time(['2012-01-01T01:00:00', '2012-01-01T02:00:00']))
>>> selected_obs_table = obs_table.select_observations(selection)
>>> selection = dict(type='par_box', variable='ALT',
... value_range=Angle([60., 70.], 'deg'))
>>> selected_obs_table = obs_table.select_observations(selection)
>>> selection = dict(type='par_box', variable='OBS_ID',
... value_range=[2, 5])
>>> selected_obs_table = obs_table.select_observations(selection)
>>> selection = dict(type='par_box', variable='N_TELS',
... value_range=[4, 4])
>>> selected_obs_table = obs_table.select_observations(selection)
"""
from ..catalog import select_sky_box, select_sky_circle
if 'inverted' not in selection.keys():
selection['inverted'] = False
if selection['type'] == 'sky_circle':
lon = selection['lon']
lat = selection['lat']
radius = selection['radius'] + selection['border']
return select_sky_circle(
self, lon_cen=lon, lat_cen=lat, radius=radius,
frame=selection['frame'], inverted=selection['inverted']
)
elif selection['type'] == 'sky_box':
lon = selection['lon']
lat = selection['lat']
border = selection['border']
lon = Angle([lon[0] - border, lon[1] + border])
lat = Angle([lat[0] - border, lat[1] + border])
return select_sky_box(
self, lon_lim=lon, lat_lim=lat,
frame=selection['frame'], inverted=selection['inverted']
)
elif selection['type'] == 'time_box':
return self.select_time_range(
'TSTART', selection['time_range'], selection['inverted']
)
elif selection['type'] == 'par_box':
return self.select_range(
selection['variable'], selection['value_range'], selection['inverted']
)
else:
raise ValueError('Invalid selection type: {}'.format(selection['type']))