def collect_gen_expr(statement_info):
expressions = []
# Join all of the constant expressions into a single GeneralExpression.
constant_names = []
lines = []
for names, node in statement_info.constant_expressions:
lines.append(unparse(node))
# Inform the statement of the variables it is writing to.
constant_names.extend(names)
if lines:
ge = GeneralExpression()
ge.code = ''.join(lines)
expressions.append(ge)
for node in statement_info.general_assignments:
ge = GeneralExpression()
ge.code = unparse(node)
# Now check this expression for any inputs that are provided by the
# import statements we had. If so, add that import statement to the
# object and remove that input.
imported_inputs = []
for iv in ge.inputs:
node = statement_info.import_statements.get(iv.binding, None)
if node is not None:
ge.import_statements.append(unparse(node))
imported_inputs.append(iv)
for iv in imported_inputs:
ge.inputs.remove(iv)
expressions.append(ge)
return expressions
def add_celem(self,var_name,active_experiment):
""" Add a new output to the gfunc specification.
The new celem is named as the first available name in the context
with var_name root.
Each celem is an instance of GeneralExpression to create it in a
way that can be added "as is" to the execution model.
"""
o_var = OutputVariable()
o_var.name = var_name
o_var.name = find_next_binding(o_var,active_experiment)
c_el = GeneralExpression(code=o_var.name)
c_el.outputs = [o_var]
self.curr_elemts.append(c_el)
def add_function_object_to_model(self, item, x=None, y=None):
""" Add the double clicked or dropped FunctionCall object to
the code/canvas.
The added function is also marked as selected.
If "New Function" is clicked, we hand back a FunctionCall based
on a LocalFunctionInfo. Otherwise, we hand back a FunctionCall
based on a PythonFunctionInfo.
# fixme: We need to add LocalFunctionInfo objects to the
# FunctionLibrary somehow.
"""
# Prevent the execution of the code when adding a new block.
self.project.active_experiment.exec_model.allow_execute = False
group = False
if item == NEW_EXPR_ENTRY:
node = GeneralExpression()
elif item == NEW_FUNCTION_ENTRY:
exp = self.project.active_experiment
base_name = 'new_function'
func_name = exp.exec_model.generate_unique_function_name(base_name = base_name)
# This should create a LocalFunctionInfo...
# fixme: Generate a unique name that isn't on the canvas.
code_template = "def %(name)s(a, b):\n" \
" return x, y\n"
code = code_template % {'name':func_name}
# fixme: Short term for testing. Remove imports in future and
# replace with FunctionCall UI.
function = LocalFunctionInfo(code=code)
traits_class = self.match_function_to_has_traits_class(item.name)
node = FunctionCall.from_callable_object(function, traits_class, exp)
elif item == NEW_LOOP_ENTRY:
group = True
exp = self.project.active_experiment
base_name = 'group'
group_name = exp.exec_model.generate_unique_function_name(base_name = base_name)
selection = SelGType()
is_ok = selection.configure_traits(kind='modal')
if is_ok:
group_type = selection.check_list[0]
lsp = GroupSpec(type=group_type,active_experiment=self.project.active_experiment)
node = FunctionCallGroup(lsp, gname=group_name);
else:
return
else:
function = PythonFunctionInfo(name=item.name,
module=item.module)
traits_class = self.match_function_to_has_traits_class(item.name)
node = FunctionCall.from_callable_object(function, traits_class, self.project.active_experiment)
if group:
res = node.configure_traits(kind="livemodal")
# FIXME: It will disappear when the creation of loops will be
# properly managed ("graphically") in the canvas
node.update_from_UI()
its_OK = res
else:
# Bring up the dialog box to edit it
res = node.edit_traits(kind="modal")
its_OK = res.result
if its_OK:
t_in = time.time()
self.add_function_to_execution_model(node, x, y)
t_out = time.time()
print '%f seconds: add func to execution model' % (t_out-t_in)
self.select_function_on_canvas(node)
self.project.active_experiment.exec_model.allow_execute = True
return
