def generate_all(self):
self.check_casegen()
generated_examples = []
if self.randgen is not None:
generated_examples.append(Example())
aliases = self.parser \
.get_definition(self.name, UnitType.alias) \
.generate_all(self.variation_name, self.arg_value)
generated_examples.extend(aliases)
if self.leading_space:
for (i, ex) in enumerate(generated_examples):
if may_get_leading_space(ex.text):
generated_examples[i].text = ' ' + ex.text
if self.casegen:
tmp_buffer = []
for ex in generated_examples:
if may_change_leading_case(ex.text):
tmp_buffer.append(
Example(with_leading_lower(ex.text), ex.entities))
tmp_buffer.append(
Example(with_leading_upper(ex.text), ex.entities))
else:
tmp_buffer.append(ex)
generated_examples = tmp_buffer
return generated_examples
def generate_all(self):
generated_examples = []
if self.randgen is not None:
generated_examples.append("")
if self.casegen:
generated_examples.append(with_leading_lower(self.words))
generated_examples.append(with_leading_upper(self.words))
else:
generated_examples.append(self.words)
if self.leading_space:
for (i, ex) in enumerate(generated_examples):
if may_get_leading_space(ex):
generated_examples[i] = ' ' + ex
result = [Example(ex) for ex in generated_examples]
return result
def generate_all(self):
self.check_casegen()
generated_examples = []
if self.randgen:
generated_examples.append(Example())
for choice in self.choices:
current_examples = []
for token in choice:
current_token_all_generations = token.generate_all()
if len(current_examples) <= 0:
current_examples = [
gen for gen in current_token_all_generations
]
else:
current_examples = [
Example(partial_example.text + gen.text,
partial_example.entities + gen.entities)
for partial_example in current_examples
for gen in current_token_all_generations
]
generated_examples.extend(current_examples)
if self.leading_space:
for (i, ex) in enumerate(generated_examples):
if may_get_leading_space(ex.text):
generated_examples[i].text = ' ' + ex.text
if self.casegen:
tmp_buffer = []
for ex in generated_examples:
tmp_buffer.append(
Example(with_leading_lower(ex.text), ex.entities))
tmp_buffer.append(
Example(with_leading_upper(ex.text), ex.entities))
return generated_examples
def test_change(self):
strings = ["test", "\tindentation", "Several words", "IN CAPITAL?"]
upper_str = ["Test", "\tIndentation", "Several words", "IN CAPITAL?"]
for (i, s) in enumerate(strings):
assert with_leading_upper(s) == upper_str[i]
def test_no_change(self):
strings = ["12", " ", "???", "(test)"]
for s in strings:
assert with_leading_upper(s) == s
def test_empty(self):
assert with_leading_upper("") == ""
def generate_all(self, variation_name=None, arg_value=None):
if (arg_value is not None
and arg_value not in self.arg_values_encountered):
# Memorize arg value
self.arg_values_encountered.append(arg_value)
generated_examples = []
relevant_rules = self.rules
if variation_name is not None:
if variation_name in self.variations:
relevant_rules = self.variations[variation_name]
else:
raise SyntaxError("Couldn't find variation '" +
str(variation_name) + "' for slot '" +
str(self.name) + "'")
if not relevant_rules: # No rules
if variation_name is None:
raise SyntaxError("No rules could be found for "+self.type+" '"+
self.name+"'")
else:
raise SyntaxError("No rules could be found for "+self.type+" '"+
self.name+"' (variation: '"+variation_name+"'")
for rule in relevant_rules:
examples_from_current_rule = []
for sub_unit_rule in rule:
sub_unit_possibilities = \
sub_unit_rule.generate_all()
if len(examples_from_current_rule) <= 0:
examples_from_current_rule = sub_unit_possibilities
else:
tmp_buffer = []
for ex in examples_from_current_rule:
for possibility in sub_unit_possibilities:
tmp_buffer.append(
Example(
ex.text + possibility.text,
ex.entities + possibility.entities, # this is a list
)
)
examples_from_current_rule = tmp_buffer
# Replace `arg` inside generated sentences
if arg_value is not None and self.modifiers.argument_name is not None:
for ex in examples_from_current_rule:
ex.text = self._replace_arg(ex.text, arg_value)
for entity in ex.entities:
entity["text"] = self._replace_arg(entity["text"],
arg_value)
entity["value"] = self._replace_arg(entity["value"],
arg_value)
# Apply casegen
if self.modifiers.casegen and self.can_have_casegen():
tmp_examples = []
for ex in examples_from_current_rule:
(lower_ex, upper_ex) = (deepcopy(ex), deepcopy(ex))
lower_ex.text = with_leading_lower(lower_ex.text)
upper_ex.text = with_leading_upper(upper_ex.text)
if lower_ex != upper_ex:
tmp_examples.append(lower_ex)
tmp_examples.append(upper_ex)
else:
tmp_examples.append(ex)
examples_from_current_rule = tmp_examples
# Add the entity in the list
slot_value = rule[0].name
if not isinstance(rule[0], DummySlotValRuleContent):
slot_value = None
else: # Replace the argument by its value if needed
slot_value = self._replace_arg(slot_value, arg_value)
for ex in examples_from_current_rule:
if slot_value is not None:
ex.entities.append({
"slot-name": self.name,
"text": ex.text[:],
"value": slot_value,
})
else:
ex.entities.append({
"slot-name": self.name,
"text": ex.text[:],
"value": ex.text[:],
})
generated_examples.extend(examples_from_current_rule)
for ex in generated_examples:
ex.text = ENTITY_MARKER + ex.text
return generated_examples
