def __init__(self, svg_source, **kwargs):
from svg_model import svg_polygons_to_df, compute_shape_centers
extend = kwargs.pop('extend', .5)
self.electrode_channels = svg_polygons_to_channels(svg_source,
**kwargs)
# Read device layout from SVG file.
df_device = svg_polygons_to_df(svg_source)
#self.df_paths = scale_svg_frame(df_device)
self.df_paths = compute_shape_centers(df_device, 'path_id')
self.df_connected = extract_adjacent_paths(self.df_paths, extend)
self.df_connected['cost'] = 1
# The following returns one row per electrode, indexed by electrode path id
self.df_path_centers = (self.df_paths.drop_duplicates(['path_id'])
.set_index('path_id')[['x_center',
'y_center']])
(self.adjacency_matrix, self.indexed_paths,
self.path_indexes) = get_adjacency_matrix(self.df_connected)
self.df_indexed_path_centers = (self.df_path_centers
.loc[self.path_indexes.index]
.reset_index())
self.df_indexed_path_centers.rename(columns={'index': 'path_id'},
inplace=True)
self.df_connected_indexed = self.df_connected.copy()
self.df_connected_indexed['source'] = map(str, self.path_indexes
[self.df_connected
['source']])
self.df_connected_indexed['target'] = map(str, self.path_indexes
[self.df_connected
['target']])
self.df_paths_indexed = self.df_paths.copy()
self.df_paths_indexed['path_id'] = map(str, self.path_indexes
[self.df_paths.path_id])
self.graph = nx.Graph()
for index, row in self.df_connected.iterrows():
self.graph.add_edge(row['source'], row['target'],
cost=row['cost'])
def from_svg(cls, svg_filepath, **kwargs):
df_shapes = svg_polygons_to_df(svg_filepath)
return cls(df_shapes, 'path_id', **kwargs)
