# Licensed under a 3-clause BSD style license - see LICENSE.rst
"""Source catalog and object base classes."""
from collections import OrderedDict
import copy
import numpy as np
from astropy.utils import lazyproperty
from ..utils.array import _is_int
from ..utils.table import table_row_to_dict
from .utils import skycoord_from_table
__all__ = ["SourceCatalog", "SourceCatalogObject"]
[docs]class SourceCatalogObject:
"""Source catalog object.
This class can be used directly, but it's mostly used as a
base class for the other source catalog classes.
The catalog data on this source is stored in the `source.data`
attribute as on OrderedDict.
The source catalog object is decoupled from the source catalog,
it doesn't hold a reference back to it.
The catalog table row index is stored in `_table_row_index` though,
because it can be useful for debugging or display.
"""
_source_name_key = "Source_Name"
_source_index_key = "catalog_row_index"
def __init__(self, data, data_extended=None):
self.data = data
if data_extended:
self.data_extended = data_extended
@property
def name(self):
"""Source name (str)"""
name = self.data[self._source_name_key]
return name.strip()
@property
def index(self):
"""Row index of source in catalog (int)"""
return self.data[self._source_index_key]
@property
def _data_python_dict(self):
"""Convert ``data`` into a Python dict that only contains
Python data types, i.e. is readily JSON or YAML serialisable.
Quantity unit information is stripped.
This is mainly used at the moment to pass the data to
the gamma-sky.net webpage.
"""
out = OrderedDict()
for key, value in self.data.items():
if isinstance(value, int):
out_val = value
else:
# This works because almost all values in ``data``
# are Numpy objects, and ``tolist`` works for Numpy
# arrays and scalars.
out_val = np.asarray(value).tolist()
out[key] = out_val
return out
@property
def position(self):
"""Source position (`~astropy.coordinates.SkyCoord`)."""
return skycoord_from_table(self.data)
[docs]class SourceCatalog:
"""Generic source catalog.
This class can be used directly, but it's mostly used as a
base class for the other source catalog classes.
This is a thin wrapper around `~astropy.table.Table`,
which is stored in the ``catalog.table`` attribute.
Parameters
----------
table : `~astropy.table.Table`
Table with catalog data.
source_name_key : str
Column with source name information
source_name_alias : tuple of str
Columns with source name aliases. This will allow accessing the source
row by alias names as well.
"""
source_object_class = SourceCatalogObject
# TODO: at the moment these are duplicated in SourceCatalogObject.
# Should we share them somehow?
_source_index_key = "catalog_row_index"
def __init__(self, table, source_name_key="Source_Name", source_name_alias=()):
self.table = table
self._source_name_key = source_name_key
self._source_name_alias = source_name_alias
def __str__(self):
s = self.description
s += " with {} objects.".format(len(self.table))
return s
@lazyproperty
def _name_to_index_cache(self):
# Make a dict for quick lookup: source name -> row index
names = dict()
for idx, row in enumerate(self.table):
name = row[self._source_name_key]
names[name.strip()] = idx
for alias_column in self._source_name_alias:
for alias in row[alias_column].split(","):
if not alias == "":
names[alias.strip()] = idx
return names
[docs] def row_index(self, name):
"""Look up row index of source by name.
Parameters
----------
name : str
Source name
Returns
-------
index : int
Row index of source in table
"""
index = self._name_to_index_cache[name]
row = self.table[index]
# check if name lookup is correct other wise recompute _name_to_index_cache
possible_names = [row[self._source_name_key]]
for alias_column in self._source_name_alias:
possible_names += row[alias_column].split(",")
if name not in possible_names:
self.__dict__.pop("_name_to_index_cache")
index = self._name_to_index_cache[name]
return index
[docs] def source_name(self, index):
"""Look up source name by row index.
Parameters
----------
index : int
Row index of source in table
"""
source_name_col = self.table[self._source_name_key]
name = source_name_col[index]
return name.strip()
def __getitem__(self, key):
"""Get source by name.
Parameters
----------
key : str or int
Source name or row index
Returns
-------
source : `SourceCatalogObject`
An object representing one source.
Notes
-----
At the moment this can raise KeyError, IndexError and ValueError
for invalid keys. Should we always raise KeyError to simplify this?
"""
if isinstance(key, str):
index = self.row_index(key)
elif _is_int(key):
index = key
else:
msg = "Key must be source name string or row index integer. "
msg += "Type not understood: {}".format(type(key))
raise ValueError(msg)
return self._make_source_object(index)
def _make_source_object(self, index):
"""Make one source object.
Parameters
----------
index : int
Row index
Returns
-------
source : `SourceCatalogObject`
Source object
"""
data = table_row_to_dict(self.table[index])
data[self._source_index_key] = index
try:
name_extended = data["Extended_Source_Name"].strip()
idx = self._lookup_extended_source_idx[name_extended]
data_extended = table_row_to_dict(self.extended_sources_table[idx])
except KeyError:
data_extended = None
source = self.source_object_class(data, data_extended)
return source
@lazyproperty
def _lookup_extended_source_idx(self):
names = [_.strip() for _ in self.extended_sources_table["Source_Name"]]
idx = range(len(names))
return dict(zip(names, idx))
@property
def _data_python_list(self):
"""Convert catalog into a Python list that only contains
Python data types, i.e. is readily JSON or YAML serialisable.
Quantity unit information is stripped.
This is mainly used at the moment to pass the data to
the gamma-sky.net webpage.
"""
return [source._data_python_dict for source in self]
@property
def positions(self):
"""Source positions (`~astropy.coordinates.SkyCoord`)."""
return skycoord_from_table(self.table)
[docs] def copy(self):
"""Copy catalog"""
return copy.deepcopy(self)