def _transform(self, sampled_parameters, parameters):
#TODO:: add some more useful debug logging
temp = {}
for par in parameters:
# only keep uncertainties that exist in this model
try:
value = sampled_parameters[par.name]
except KeyError:
if par.default is not None:
value = par.default
else:
ema_logging.debug('{} not found'.format(par.name))
continue
multivalue = False
if isinstance(par, CategoricalParameter):
category = par.categories[value]
value = category.value
if category.multivalue == True:
multivalue = True
values = value
# translate uncertainty name to variable name
for i, varname in enumerate(par.variable_name):
# a bit hacky implementation, investigate some kind of
# zipping of variable_names and values
if multivalue:
value = values[i]
temp[varname] = value
sampled_parameters.data = temp
def set_fig_to_bw(fig,
style=HATCHING,
line_style='continuous',
all_colors=None):
"""
TODO it would be nice if for lines you can select either markers, gray
scale, or simple black
Take each axes in the figure and transform its content to black and white.
Lines are tranformed based on different line styles. Fills such as can
be used in `meth`:envelopes are gray-scaled. Heathmaps are also gray-scaled.
derived from and expanded for my use from:
http://stackoverflow.com/questions/7358118/matplotlib-black-white-colormap-with-dashes-dots-etc
Parameters
----------
fig : figure
the figure to transform to black and white
style : {HATCHING, GREYSCALE}
parameter controlling how collections are transformed.
line_style: str
linestyle to use for converting, can be continuous, black
or None
"""
all_colors = _identify_colors(fig)
if len(all_colors) > len(bw_mapping):
mapping_cycle = itertools.cycle(bw_mapping)
ema_logging.warning(
'more colors used than provided in B&W mapping, cycling over mapping'
)
else:
mapping_cycle = bw_mapping
colormap = dict(zip(all_colors, mapping_cycle))
ema_logging.debug(colormap.keys())
max_shade = 0.9
for i, color in enumerate(colormap.keys()):
relative_color = max_shade * ((i + 1) / len(all_colors))
colormap[color]['fill'] = str(relative_color)
for ax in fig.get_axes():
set_ax_lines_bw(ax, colormap)
set_ax_patches_bw(ax, colormap)
set_ax_collections_to_bw(ax, style, colormap)
set_ax_legend_to_bw(ax, style, colormap)
set_legend_to_bw(fig.legends, style, colormap)
def set_fig_to_bw(fig, style=HATCHING,
line_style='continuous', all_colors=None):
"""
TODO it would be nice if for lines you can select either markers, gray
scale, or simple black
Take each axes in the figure and transform its content to black and white.
Lines are tranformed based on different line styles. Fills such as can
be used in `meth`:envelopes are gray-scaled. Heathmaps are also gray-scaled.
derived from and expanded for my use from:
http://stackoverflow.com/questions/7358118/matplotlib-black-white-colormap-with-dashes-dots-etc
Parameters
----------
fig : figure
the figure to transform to black and white
style : {HATCHING, GREYSCALE}
parameter controlling how collections are transformed.
line_style: str
linestyle to use for converting, can be continuous, black
or None
"""
all_colors = _identify_colors(fig)
if len(all_colors) > len(bw_mapping):
mapping_cycle = itertools.cycle(bw_mapping)
ema_logging.warning(
'more colors used than provided in B&W mapping, cycling over mapping')
else:
mapping_cycle = bw_mapping
colormap = dict(zip(all_colors, mapping_cycle))
ema_logging.debug(colormap.keys())
max_shade = 0.9
for i, color in enumerate(colormap.keys()):
relative_color = max_shade * ((i + 1) / len(all_colors))
colormap[color]['fill'] = str(relative_color)
for ax in fig.get_axes():
set_ax_lines_bw(ax, colormap)
set_ax_patches_bw(ax, colormap)
set_ax_collections_to_bw(ax, style, colormap)
set_ax_legend_to_bw(ax, style, colormap)
set_legend_to_bw(fig.legends, style, colormap)
def test_log_messages(self):
ema_logging.log_to_stderr(ema_logging.DEBUG)
with mock.patch('ema_workbench.util.ema_logging._logger') as mocked_logger:
message = 'test message'
ema_logging.debug(message)
mocked_logger.debug.assert_called_with(message)
ema_logging.info(message)
mocked_logger.info.assert_called_with(message)
ema_logging.warning(message)
mocked_logger.warning.assert_called_with(message)
ema_logging.error(message)
mocked_logger.error.assert_called_with(message)
ema_logging.exception(message)
mocked_logger.exception.assert_called_with(message)
ema_logging.critical(message)
mocked_logger.critical.assert_called_with(message)
def _transform(self, sampled_parameters, parameters):
# TODO:: add some more useful debug logging
temp = {}
for par in parameters:
# only keep uncertainties that exist in this model
try:
value = sampled_parameters[par.name]
except KeyError:
if par.default is not None:
value = par.default
else:
ema_logging.debug("{} not found".format(par.name))
continue
if (isinstance(par, CategoricalParameter)) or (isinstance(par, IntegerParameter)):
value = int(value)
multivalue = False
if isinstance(par, CategoricalParameter):
value = par.cat_for_index(value)
category = par.categories[value]
value = category.value
if category.multivalue == True:
multivalue = True
values = value
# translate uncertainty name to variable name
for i, varname in enumerate(par.variable_name):
# a bit hacky implementation, investigate some kind of
# zipping of variable_names and values
if multivalue:
value = values[i]
temp[varname] = value
sampled_parameters.data = temp
def test_log_messages(self):
ema_logging.log_to_stderr(ema_logging.DEBUG)
with mock.patch(
'ema_workbench.util.ema_logging._logger') as mocked_logger:
message = 'test message'
ema_logging.debug(message)
mocked_logger.debug.assert_called_with(message)
ema_logging.info(message)
mocked_logger.info.assert_called_with(message)
ema_logging.warning(message)
mocked_logger.warning.assert_called_with(message)
ema_logging.error(message)
mocked_logger.error.assert_called_with(message)
ema_logging.exception(message)
mocked_logger.exception.assert_called_with(message)
ema_logging.critical(message)
mocked_logger.critical.assert_called_with(message)
def _transform(self, sampled_parameters, parameters):
if not parameters:
# no parameters defined, so nothing to transform, mainly
# useful for manual specification of scenario / policy
# without having to define uncertainties / levers
return
temp = {}
for par in parameters:
# only keep uncertainties that exist in this model
try:
value = sampled_parameters[par.name]
except KeyError:
if par.default is not None:
value = par.default
else:
ema_logging.debug('{} not found'.format(par.name))
continue
multivalue = False
if isinstance(par, CategoricalParameter):
if par.multivalue == True:
multivalue = True
values = value
# translate uncertainty name to variable name
for i, varname in enumerate(par.variable_name):
# a bit hacky implementation, investigate some kind of
# zipping of variable_names and values
if multivalue:
value = values[i]
temp[varname] = value
sampled_parameters.data = temp
