Sheerka-Old/tests/parsers/test_UnrecognizedNodeParser.py

from core.builtin_concepts import ParserResultConcept, BuiltinConcepts
from core.concept import Concept
from core.tokenizer import Tokenizer, TokenKind
from parsers.BaseNodeParser import ConceptNode, UnrecognizedTokensNode, SourceCodeWithConceptNode, SourceCodeNode
from parsers.BnfNodeParser import BnfNodeParser
from parsers.SequenceNodeParser import SequenceNodeParser
from parsers.SyaNodeParser import SyaNodeParser
from parsers.UnrecognizedNodeParser import UnrecognizedNodeParser
from tests.TestUsingMemoryBasedSheerka import TestUsingMemoryBasedSheerka
from tests.parsers.parsers_utils import compute_expected_array, get_node, CC, UTN, CNC, CN, SCWC, \
    compare_with_test_object, SCN


def get_input_nodes_from(my_concepts_map, full_expr, *args):
    def _get_real_node(n):
        if isinstance(n, CC):
            concept = n.concept or Concept.update_from(my_concepts_map[n.concept_key])
            for k, v in n.compiled.items():
                concept.get_compiled()[k] = _get_real_node(v)
            return concept

        if isinstance(n, UTN):
            return UnrecognizedTokensNode(n.start, n.end, full_expr_as_tokens[n.start: n.end + 1])

        if isinstance(n, (CNC, CN)):
            concept = n.concept if hasattr(n, "concept") and n.concept else \
                Concept().update_from(my_concepts_map[n.concept_key])
            tokens = full_expr_as_tokens[n.start: n.end + 1]
            if hasattr(n, "compiled"):
                for k, v in n.compiled.items():
                    concept.get_compiled()[k] = _get_real_node(v)
            return ConceptNode(concept, n.start, n.end, tokens)

        if isinstance(n, SCWC):
            n.first = _get_real_node(n.first)
            n.last = _get_real_node(n.last)
            n.content = tuple(_get_real_node(nn) for nn in n.content)
            return SourceCodeWithConceptNode(n.first, n.last, list(n.content)).pseudo_fix_source()

        if isinstance(n, (UnrecognizedTokensNode, ConceptNode, SourceCodeNode, SourceCodeWithConceptNode)):
            return n

        raise NotImplementedError()

    res = []
    full_expr_as_tokens = list(Tokenizer(full_expr))
    tokens_for_get_node = [token.value for token in full_expr_as_tokens if token.type != TokenKind.EOF]
    for arg in args:
        node = get_node(my_concepts_map, tokens_for_get_node, arg)
        res.append(_get_real_node(node))

    return res


concepts_map = {
    "5params": Concept("5params").def_var("a").def_var("b").def_var("c").def_var("d").def_var("e"),
    "plus": Concept("a plus b", body="a + b").def_var("a").def_var("b"),
    "mult": Concept("a mult b", body="a * b").def_var("a").def_var("b"),
    "one": Concept("one", body="1"),
    "two": Concept("two", body="2"),
    "three": Concept("three", body="3"),
    "twenties": Concept("twenties", definition="'twenty' (one|two)=unit", body="20 + unit").def_var("unit"),
    "hello_atom": Concept("hello one"),
    "hello_sya": Concept("hello a").def_var("a"),
    "greetings_a": Concept("greetings a").def_var("a"),
    "greetings_b": Concept("greetings b").def_var("b"),

}


class TestUnrecognizedNodeParser(TestUsingMemoryBasedSheerka):
    shared_ontology = None

    @classmethod
    def setup_class(cls):
        init_test_helper = cls().init_test(cache_only=False, ontology="#TestUnrecognizedNodeParser#")
        sheerka, context, *updated = init_test_helper.with_concepts(*concepts_map.values(), create_new=True).unpack()
        for i, concept_name in enumerate(concepts_map):
            concepts_map[concept_name] = updated[i]

        sheerka.set_is_greater_than(context,
                                    BuiltinConcepts.PRECEDENCE,
                                    concepts_map["mult"],
                                    concepts_map["plus"], 'Sya')

        cls.shared_ontology = sheerka.get_ontology(context)
        sheerka.pop_ontology(context)

    def init_parser(self, my_concepts_map=None, **kwargs):
        if my_concepts_map is None:
            sheerka, context = self.init_test().unpack()
            sheerka.add_ontology(context, self.shared_ontology)
        else:
            sheerka, context, *updated = self.init_test().with_concepts(*my_concepts_map.values(), **kwargs).unpack()
            for i, pair in enumerate(my_concepts_map):
                my_concepts_map[pair] = updated[i]

        parser = UnrecognizedNodeParser()
        return sheerka, context, parser

    def test_i_can_validate_a_valid_concept_node(self):
        sheerka, context, parser = self.init_parser()
        node = get_input_nodes_from(concepts_map, "one", "one")[0]

        res = UnrecognizedNodeParser().validate_concept_node(context, node)

        assert res.status
        assert res.body.concept == concepts_map["one"]

    def test_i_can_validate_concept_unrecognized_tokens(self):
        sheerka, context, parser = self.init_parser()
        node = get_input_nodes_from(
            concepts_map,
            "5params one two three twenty one 1 + 2 one plus two mult three",
            CNC("5params",
                a=" one ",
                b=" two three ",
                c=" twenty one ",
                d=UTN(" 1 + 2 ", 12, 18),
                e=" one plus two mult three"))[0]

        res = UnrecognizedNodeParser().validate_concept_node(context, node)

        assert res.status

        concept = res.body.concept
        assert concept == concepts_map["5params"]

        assert len(concept.get_compiled()["a"]) == 1
        assert sheerka.isinstance(concept.get_compiled()["a"][0], BuiltinConcepts.RETURN_VALUE)
        assert concept.get_compiled()["a"][0].status
        assert concept.get_compiled()["a"][0].who == "parsers." + SequenceNodeParser.NAME
        compare_with_test_object(concept.get_compiled()["a"][0].body.body, [CN("one", "one", 1, 1)])

        assert len(concept.get_compiled()["b"]) == 1
        assert sheerka.isinstance(concept.get_compiled()["b"][0], BuiltinConcepts.RETURN_VALUE)
        assert concept.get_compiled()["b"][0].status
        assert concept.get_compiled()["b"][0].who == "parsers." + SequenceNodeParser.NAME
        compare_with_test_object(concept.get_compiled()["b"][0].body.body,
                                 [CN("two", "two", 1, 1), CN("three", "three", 3, 3)])

        assert len(concept.get_compiled()["c"]) == 1
        assert sheerka.isinstance(concept.get_compiled()["c"][0], BuiltinConcepts.RETURN_VALUE)
        assert concept.get_compiled()["c"][0].status
        assert concept.get_compiled()["c"][0].who == "parsers." + BnfNodeParser.NAME
        expected_nodes = compute_expected_array(
            concepts_map,
            " twenty one ",
            [CNC("twenties", "twenty one", unit="one")])
        compare_with_test_object(concept.get_compiled()["c"][0].body.body, expected_nodes)

        assert len(concept.get_compiled()["d"]) == 1
        assert sheerka.isinstance(concept.get_compiled()["d"][0], BuiltinConcepts.RETURN_VALUE)
        assert concept.get_compiled()["d"][0].status
        assert concept.get_compiled()["d"][0].who == "parsers.Python"
        assert concept.get_compiled()["d"][0].body.source == " 1 + 2 "

        assert len(concept.get_compiled()["e"]) == 1
        assert sheerka.isinstance(concept.get_compiled()["e"][0], BuiltinConcepts.RETURN_VALUE)
        assert concept.get_compiled()["e"][0].status
        assert concept.get_compiled()["e"][0].who == "parsers." + SyaNodeParser.NAME
        expected_nodes = compute_expected_array(
            concepts_map,
            " one plus two mult three ",
            [CNC("plus", a="one", b=CC("mult", a="two", b="three"))],
            exclude_body=True)

        compare_with_test_object(concept.get_compiled()["e"][0].body.body, expected_nodes)

        # # sanity check, I can evaluate the concept
        # evaluated = sheerka.evaluate_concept(self.get_context(sheerka, eval_body=True), concept)
        # assert evaluated.key == concept.key
        # assert evaluated.get_value("a") ==

    def test_i_can_validate_with_recursion(self):
        sheerka, context, parser = self.init_parser()

        node = get_input_nodes_from(
            concepts_map,
            "1 plus 2 mult twenty two",
            CNC("plus",
                a="1 ",
                b=CC("mult", a=" 2 ", b=" twenty two")))[0]

        res = UnrecognizedNodeParser().validate_concept_node(context, node)

        assert res.status
        assert res.body.concept == concepts_map["plus"]
        assert len(res.body.concept.get_compiled()["a"]) == 1
        assert res.body.concept.get_compiled()["a"][0].status
        assert res.body.concept.get_compiled()["a"][0].who == "parsers.Python"
        assert res.body.concept.get_compiled()["a"][0].body.source == "1 "

        assert res.body.concept.get_compiled()["b"] == concepts_map["mult"]
        assert sheerka.isinstance(res.body.concept.get_compiled()["b"].get_compiled()["a"][0],
                                  BuiltinConcepts.RETURN_VALUE)
        assert res.body.concept.get_compiled()["b"].get_compiled()["a"][0].status
        assert res.body.concept.get_compiled()["b"].get_compiled()["a"][0].who == "parsers.Python"
        assert res.body.concept.get_compiled()["b"].get_compiled()["a"][0].body.source == " 2 "

        assert sheerka.isinstance(res.body.concept.get_compiled()["b"].get_compiled()["b"][0],
                                  BuiltinConcepts.RETURN_VALUE)
        assert res.body.concept.get_compiled()["b"].get_compiled()["b"][0].status
        assert res.body.concept.get_compiled()["b"].get_compiled()["b"][0].who == "parsers.Bnf"
        expected_nodes = compute_expected_array(
            concepts_map,
            " twenty two",
            [CNC("twenties", "twenty two", unit="two")])
        compare_with_test_object(res.body.concept.get_compiled()["b"].get_compiled()["b"][0].body.body, expected_nodes)

    def test_i_can_validate_and_evaluate_a_concept_node_with_python(self):
        sheerka, context, parser = self.init_parser()

        node = get_input_nodes_from(
            concepts_map,
            "one plus 1 + 1",
            CNC("plus",
                a=UTN("one "),
                b=UTN("1 + 1")))[0]

        res = UnrecognizedNodeParser().validate_concept_node(context, node)

        assert res.status
        concept_found = res.body.concept
        compiled = concept_found.get_compiled()
        assert concept_found == concepts_map["plus"]
        assert len(compiled["a"]) == 1
        assert sheerka.isinstance(compiled["a"][0], BuiltinConcepts.RETURN_VALUE)
        assert compiled["a"][0].body.body[0].concept.id == concepts_map["one"].id

        assert len(compiled["b"]) == 1
        assert sheerka.isinstance(compiled["b"][0], BuiltinConcepts.RETURN_VALUE)
        assert compiled["b"][0].status
        assert compiled["b"][0].who == "parsers.Python"
        assert compiled["b"][0].body.source == "1 + 1"

        # # evaluate
        context = self.get_context(sheerka, eval_body=True)
        evaluated = sheerka.evaluate_concept(context, res.body.concept)
        assert evaluated.body == 3

    def test_i_can_validate_and_evaluate_concept_when_bnf_concept(self):
        sheerka, context, parser = self.init_parser()
        node = get_input_nodes_from(
            concepts_map,
            "one plus twenty one",
            CNC("plus",
                a=UTN("one "),
                b=UTN("twenty one")))[0]

        res = UnrecognizedNodeParser().validate_concept_node(context, node)

        assert res.status
        concept_found = res.body.concept
        compiled = concept_found.get_compiled()
        assert concept_found == concepts_map["plus"]
        assert len(compiled["a"]) == 1
        assert sheerka.isinstance(compiled["a"][0], BuiltinConcepts.RETURN_VALUE)
        assert compiled["a"][0].body.body[0].concept.id == concepts_map["one"].id

        assert len(compiled["b"]) == 1
        assert compiled["b"][0].status
        assert compiled["b"][0].who == "parsers.Bnf"

        # evaluate
        context = self.get_context(sheerka, eval_body=True)
        evaluated = sheerka.evaluate_concept(context, res.body.concept)
        assert evaluated.body == 22

    def test_i_can_parse_and_evaluate_unrecognized_python_node(self):
        sheerka, context, parser = self.init_parser()

        expression = "1 + 1"
        nodes = get_input_nodes_from(concepts_map, expression, UTN(expression))
        parser_input = ParserResultConcept("parsers.xxx", source=expression, value=nodes)

        res = parser.parse(context, parser_input)
        parser_result = res.body
        actual_nodes = res.body.body

        assert res.status
        assert sheerka.isinstance(parser_result, BuiltinConcepts.PARSER_RESULT)
        assert parser_result.source == expression
        assert len(actual_nodes) == 1
        compare_with_test_object(actual_nodes[0], SCN(expression, 0, 4))

    def test_i_cannot_parse_unrecognized_python_that_looks_like_concept(self):
        sheerka, context, parser = self.init_parser()

        expression = "fake_concept_name"  # as it's not a concept, it will be recognized as python node
        nodes = get_input_nodes_from(concepts_map, expression, UTN(expression))
        parser_input = ParserResultConcept("parsers.xxx", source=expression, value=nodes)

        res = parser.parse(context, parser_input)
        parser_result = res.body
        actual_nodes = res.body.body

        assert not res.status
        assert sheerka.isinstance(parser_result, BuiltinConcepts.PARSER_RESULT)
        assert parser_result.source == expression
        assert len(actual_nodes) == 1
        assert actual_nodes[0] == nodes[0]

    def test_i_can_parse_unrecognized_bnf_concept_node(self):

        sheerka, context, parser = self.init_parser()

        expression = "twenty one"
        nodes = get_input_nodes_from(concepts_map, expression, UTN(expression))
        parser_input = ParserResultConcept("parsers.xxx", source=expression, value=nodes)

        res = parser.parse(context, parser_input)
        parser_result = res.body
        actual_nodes = res.body.body

        assert res.status
        assert sheerka.isinstance(parser_result, BuiltinConcepts.PARSER_RESULT)
        assert parser_result.source == expression
        assert len(actual_nodes) == 1
        expected_array = compute_expected_array(
            concepts_map,
            expression, [CNC("twenties", source=expression, unit="one")])
        compare_with_test_object(actual_nodes, expected_array)

    def test_i_can_parse_unrecognized_sya_concept_node(self):
        sheerka, context, parser = self.init_parser()

        expression = "one plus two mult three"
        nodes = get_input_nodes_from(concepts_map, expression, UTN(expression))
        parser_input = ParserResultConcept("parsers.xxx", source=expression, value=nodes)

        res = parser.parse(context, parser_input)
        parser_result = res.body
        actual_nodes = res.body.body

        assert res.status
        assert sheerka.isinstance(parser_result, BuiltinConcepts.PARSER_RESULT)
        assert parser_result.source == expression
        assert len(actual_nodes) == 1

        expected_array = compute_expected_array(
            concepts_map,
            expression, [CNC("plus",
                             a="one",
                             b=CC("mult", source="two mult three", a="two", b="three"))],
            exclude_body=True)
        compare_with_test_object(actual_nodes, expected_array)

    def test_i_can_parse_unrecognized_source_code_with_concept_node(self):
        sheerka, context, parser = self.init_parser()

        expression = "desc(a plus b)"
        source_code_concepts = SCWC("desc(", ")", CNC("plus", a=UTN("a"), b=UTN("b")))
        nodes = get_input_nodes_from(concepts_map, expression, source_code_concepts)
        parser_input = ParserResultConcept("parsers.xxx", source=expression, value=nodes)

        res = parser.parse(context, parser_input)
        parser_result = res.body
        actual_nodes = res.body.body

        assert not res.status  # status is False to let PythonWithConceptParser validate the code
        assert sheerka.isinstance(parser_result, BuiltinConcepts.PARSER_RESULT)
        assert parser_result.source == expression
        assert len(actual_nodes) == 1
        assert actual_nodes[0].nodes[
            0].concept.get_hints().is_evaluated  # 'a plus b' is recognized as concept definition

    def test_i_can_parse_unrecognized_source_code_with_concept_node_when_var_in_short_term_memory(self):
        sheerka, context, parser = self.init_parser()

        expression = "desc(a plus b)"
        source_code_concepts = SCWC("desc(", ")", CNC("plus", a=UTN("a"), b=UTN("b")))
        nodes = get_input_nodes_from(concepts_map, expression, source_code_concepts)
        parser_input = ParserResultConcept("parsers.xxx", source=expression, value=nodes)

        context.add_to_short_term_memory("a", 1)  # -> a plus b is now an instance of the concept
        res = parser.parse(context, parser_input)
        parser_result = res.body
        actual_nodes = res.body.body

        assert not res.status  # status is False to let PythonWithConceptParser validate the code
        assert sheerka.isinstance(parser_result, BuiltinConcepts.PARSER_RESULT)
        assert parser_result.source == expression
        assert len(actual_nodes) == 1
        assert not actual_nodes[0].nodes[0].concept.get_hints().is_evaluated  # 'a plus b' need to be evaluated

    def test_i_can_parse_unrecognized_sya_concept_that_references_source_code(self):
        sheerka, context, parser = self.init_parser()

        expression = "hello get_user_name(twenty one)"
        tmp_node = CNC("hello_sya",
                       "hello get_user_name(twenty one)",
                       a=SCWC("get_user_name(", ")", CNC("twenties", "twenty one", unit="one")))
        nodes = get_input_nodes_from(concepts_map, expression, tmp_node)
        parser_input = ParserResultConcept("parsers.xxx", source=expression, value=nodes)

        res = parser.parse(context, parser_input)
        parser_result = res.body
        actual_nodes = res.body.body

        assert res.status
        assert sheerka.isinstance(parser_result, BuiltinConcepts.PARSER_RESULT)
        assert parser_result.source == expression
        assert len(actual_nodes) == 1

        expected_array = compute_expected_array(
            concepts_map,
            expression, [CN("hello_sya", "hello get_user_name(twenty one)")],
            exclude_body=True)
        compare_with_test_object(actual_nodes, expected_array)
        assert isinstance(actual_nodes[0].concept.get_compiled()["a"], list)
        assert sheerka.isinstance(actual_nodes[0].concept.get_compiled()["a"][0], BuiltinConcepts.RETURN_VALUE)

    def test_i_can_parse_sequences(self):
        sheerka, context, parser = self.init_parser()

        expression = "one plus two three"
        sequence = get_input_nodes_from(concepts_map, expression,
                                        CNC("plus", a="one", b="two"),
                                        UTN(" three", 5, 6))
        parser_input = ParserResultConcept("parsers.xxx", source="one plus two three", value=sequence)

        res = parser.parse(context, parser_input)
        actual_nodes = res.body.body

        assert res.status

        expected_array = compute_expected_array(
            concepts_map,
            expression, [
                CNC("plus", a="one", b="two"),
                CN("three", start=6, end=6)])
        compare_with_test_object(actual_nodes, expected_array)

    def test_i_can_parse_when_multiple_atom_and_sya(self):
        sheerka, context, parser = self.init_parser()
        expression = "two hello one three"
        nodes = get_input_nodes_from(concepts_map, expression,
                                     "two", UTN("hello one"), "three")
        parser_input = ParserResultConcept("parsers.xxx", source=expression, value=nodes)

        res = parser.parse(context, parser_input)
        assert len(res) == 2
        assert res[0].status
        assert res[1].status

        actual_nodes0 = res[0].body.body
        expected_0 = compute_expected_array(concepts_map, expression, [
            CN("two", start=0, end=0),
            CN("hello_atom", source="hello one", start=2, end=4),
            CN("three", start=6, end=6)])
        compare_with_test_object(actual_nodes0, expected_0)

        actual_nodes1 = res[1].body.body
        expected_1 = compute_expected_array(concepts_map, expression, [
            CN("two", start=0, end=0),
            CNC("hello_sya", "hello one", start=2, end=4, a="one"),
            CN("three", start=6, end=6)],
                                            exclude_body=True)
        compare_with_test_object(actual_nodes1, expected_1)

    def test_i_can_parse_when_multiple_sya_concepts(self):
        sheerka, context, parser = self.init_parser()
        expression = "greetings two"
        nodes = get_input_nodes_from(concepts_map, expression, UTN("greetings two"))

        parser_input = ParserResultConcept("parsers.xxx", source=expression, value=nodes)

        res = parser.parse(context, parser_input)
        assert len(res) == 2
        assert res[0].status
        assert res[1].status

        actual_nodes0 = res[0].body.body
        expected_0 = compute_expected_array(concepts_map, expression, [
            CNC("greetings_a", source="greetings two", start=0, end=2, a="two")], exclude_body=True)
        compare_with_test_object(actual_nodes0, expected_0)

        actual_nodes1 = res[1].body.body
        expected_1 = compute_expected_array(concepts_map, expression, [
            CNC("greetings_b", source="greetings two", start=0, end=2, b="two")], exclude_body=True)
        compare_with_test_object(actual_nodes1, expected_1)

    def test_i_cannot_parse_when_some_unrecognized_remain(self):
        sheerka, context, parser = self.init_parser()
        expression = "twenty one + one"
        nodes = get_input_nodes_from(concepts_map, expression, UTN("twenty "), "one", " + ", ("one", 1))

        parser_input = ParserResultConcept("parsers.xxx", source=expression, value=nodes)
        res = parser.parse(context, parser_input)

        assert not res.status
        assert res.body.body == nodes

    def test_i_cannot_parse_when_i_cannot_validate(self):
        sheerka, context, parser = self.init_parser(concepts_map, create_new=True)
        expression = "one plus unknown tokens"
        nodes = get_input_nodes_from(concepts_map, expression,
                                     CNC("plus", a="one ", b=" unknown tokens"))

        parser_input = ParserResultConcept("parsers.xxx", source="six", value=nodes)
        res = parser.parse(context, parser_input)

        assert not res.status
        assert sheerka.isinstance(res.body, BuiltinConcepts.ERROR)

    def test_i_cannot_parse_when_unrecognized(self):
        sheerka, context, parser = self.init_parser(concepts_map, create_new=True)
        expression = "unknown tokens"
        nodes = get_input_nodes_from(concepts_map, expression, UTN(expression))

        parser_input = ParserResultConcept("parsers.xxx", source="six", value=nodes)
        res = parser.parse(context, parser_input)
        actual_nodes = res.body.body

        assert not res.status
        assert actual_nodes == nodes