def test_match():
v1 = [1, 1, 2, 2, 3, 3]
v2 = [1, 2, 3]
v3 = [0, 1, 2]
c = [1]
assert match(v1, v2) == [0, 0, 1, 1, 2, 2]
assert match(v1, v2, incomparables=c) == [-1, -1, 1, 1, 2, 2]
assert match(v1, v3) == [1, 1, 2, 2, -1, -1]
def test_match():
v1 = [1, 1, 2, 2, 3, 3]
v2 = [1, 2, 3]
v3 = [0, 1, 2]
c = [1]
assert match(v1, v2) == [0, 0, 1, 1, 2, 2]
assert match(v1, v2, incomparables=c) == [-1, -1, 1, 1, 2, 2]
assert match(v1, v3) == [1, 1, 2, 2, -1, -1]
def train_position_scales(self, layout, layers):
"""
Compute ranges for the x and y scales
"""
_layout = layout.layout
panel_scales_x = layout.panel_scales_x
panel_scales_y = layout.panel_scales_y
# loop over each layer, training x and y scales in turn
for layer in layers:
data = layer.data
match_id = match(data['PANEL'], _layout['PANEL'])
if panel_scales_x:
x_vars = list(set(panel_scales_x[0].aesthetics) &
set(data.columns))
# the scale index for each data point
SCALE_X = _layout['SCALE_X'].iloc[match_id].tolist()
panel_scales_x.train(data, x_vars, SCALE_X)
if panel_scales_y:
y_vars = list(set(panel_scales_y[0].aesthetics) &
set(data.columns))
# the scale index for each data point
SCALE_Y = _layout['SCALE_Y'].iloc[match_id].tolist()
panel_scales_y.train(data, y_vars, SCALE_Y)
return self
def map(self, data, layout):
if not len(data):
data['PANEL'] = pd.Categorical(
[], categories=layout['PANEL'].cat.categories, ordered=True)
return data
vars = [x for x in self.rows + self.cols]
margin_vars = [
list(data.columns & self.rows),
list(data.columns & self.cols)
]
data = add_margins(data, margin_vars, self.margins)
facet_vals = eval_facet_vars(data, vars, self.plot.environment)
data, facet_vals = add_missing_facets(data, layout, vars, facet_vals)
# assign each point to a panel
if len(facet_vals) == 0:
# Special case of no facetting
data['PANEL'] = 1
else:
keys = join_keys(facet_vals, layout, vars)
data['PANEL'] = match(keys['x'], keys['y'], start=1)
data = data.sort_values('PANEL', kind='mergesort')
# matching dtype and
# the categories(panel numbers) for the data should be in the
# same order as the panels. i.e the panels are the reference,
# they "know" the right order
data['PANEL'] = pd.Categorical(
data['PANEL'],
categories=layout['PANEL'].cat.categories,
ordered=True)
data.reset_index(drop=True, inplace=True)
return data