def load(self, nodes_file, node_types_file):
nodes_hf = h5py.File(nodes_file, 'r')
if 'nodes' not in nodes_hf.keys():
raise Exception('Could not find nodes in {}'.format(nodes_file))
nodes_group = nodes_hf['nodes']
self._network_name = nodes_group.attrs[
'network'] if 'network' in nodes_group.attrs.keys() else 'NA'
self._version = 'v0.1' # TODO: get the version number from the attributes
# Create Indices
self._nodes_index['node_gid'] = pd.Series(
nodes_group['node_gid'], dtype=nodes_group['node_gid'].dtype)
self._nodes_index['node_type_id'] = pd.Series(
nodes_group['node_type_id'],
dtype=nodes_group['node_type_id'].dtype)
self._nodes_index['node_group'] = pd.Series(
nodes_group['node_group'], dtype=nodes_group['node_group'].dtype)
self._nodes_index['node_group_index'] = pd.Series(
nodes_group['node_group_index'],
dtype=nodes_group['node_group_index'].dtype)
self._nodes_index.set_index(['node_gid'], inplace=True)
self._nrows = len(self._nodes_index)
# Save the node-types
self._node_types_table = tn.TypesTable(node_types_file, 'node_type_id')
# save pointers to the groups table
self._group_table = {
grp_id: Group(grp_id, grp_ptr, self._node_types_table)
for grp_id, grp_ptr in nodes_group.items()
if isinstance(grp_ptr, h5py.Group)
}
def load(self, nodes_file, node_types_file):
self._nodes_df = pd.read_csv(nodes_file, sep=' ', index_col=['node_id'])
self._node_types_table = tn.TypesTable(node_types_file, 'node_type_id')
self._nrows = len(self._nodes_df.index)
self._nodes_columns = tn.ColumnProperty.from_csv(self._nodes_df)
self._columns = self._nodes_columns + self._node_types_table.columns
def load(self, edges_file, edge_types_file):
edges_hf = h5py.File(edges_file, 'r')
self._edge_ptr_ds = edges_hf['edge_ptr']
self._num_syns_ds = edges_hf['num_syns']
self._src_gids_ds = edges_hf['src_gids']
# TODO: validate edge_types dataset keys
self._edge_types_ds = edges_hf['edge_types']
self._edge_types_table = tn.TypesTable(edge_types_file, 'edge_type_id')
self._index_len = len(self._edge_ptr_ds)
self._nrows = len(self._src_gids_ds)
def load(self, edges_file, edge_types_file):
edges_hf = h5py.File(edges_file, 'r')
if 'edges' not in edges_hf.keys():
raise Exception('Could not find edges in {}'.format(edges_file))
edges_group = edges_hf['edges']
# Preload the target index pointers into memory
self._target_index = pd.Series(
edges_group['index_pointer'],
dtype=edges_group['index_pointer'].dtype)
self._target_index_len = len(self._target_index)
# For the other index tables we only load in a file pointer
self._target_gid_ds = edges_group['target_gid']
if 'network' in self._target_gid_ds.attrs.keys():
self._target_network = self._target_gid_ds.attrs['network']
self._source_gid_ds = edges_group['source_gid']
if 'network' in self._source_gid_ds.attrs.keys():
self._source_network = self._source_gid_ds.attrs['network']
self._edge_type_ds = edges_group['edge_type_id']
self._edge_group_ds = edges_group['edge_group']
self._edge_group_index_ds = edges_group['edge_group_index']
self._nedges = len(self._edge_group_index_ds)
# Load in edge-types table
self._edge_types_table = tn.TypesTable(edge_types_file, 'edge_type_id')
# Load in the group properties
# TODO: look in attributes for group synonyms
# TODO: HDF5 group name will always be a string, but value in groups dataset will be an int.
self._group_table = {
grp_id: Group(grp_id, grp_ptr, self._edge_types_table)
for grp_id, grp_ptr in edges_group.items()
if isinstance(grp_ptr, h5py.Group)
}