Refactored Caching, Refactored BnfNodeParser, Introduced Sphinx

This commit is contained in:
2020-05-12 17:21:10 +02:00
parent 7d3a490bc5
commit 6e343ba996
110 changed files with 13865 additions and 7540 deletions
+44 -21
View File
@@ -2,18 +2,21 @@ import pytest
from core.builtin_concepts import BuiltinConcepts
from core.concept import Concept
from parsers.AtomNodeParser import AtomNodeParser
from parsers.BaseNodeParser import cnode, utnode, CNC
from parsers.BaseNodeParser import cnode, utnode, CNC, scnode, SCN
from tests.TestUsingMemoryBasedSheerka import TestUsingMemoryBasedSheerka
from tests.parsers.parsers_utils import compute_expected_array
class TestAtomsParser(TestUsingMemoryBasedSheerka):
def init_parser(self, concepts_map, **kwargs):
sheerka, context, *updated_concepts = self.init_concepts(singleton=True, *concepts_map.values(), **kwargs)
def init_parser(self, my_map, create_new=False, singleton=True):
sheerka, context, *updated_concepts = self.init_concepts(
*my_map.values(),
create_new=create_new,
singleton=singleton)
parser = AtomNodeParser()
parser.initialize(context, updated_concepts)
parser.init_from_concepts(context, updated_concepts)
return sheerka, context, parser
@@ -34,7 +37,7 @@ class TestAtomsParser(TestUsingMemoryBasedSheerka):
concepts_map = {
"foo": Concept("foo"),
"bar": Concept("bar"),
"twenties": Concept("twenties", definition="'twenty' ('one'|'two')=unit").def_prop("unit"),
"twenties": Concept("twenties", definition="'twenty' ('one'|'two')=unit").def_var("unit"),
}
sheerka, context, parser = self.init_parser(concepts_map)
@@ -57,7 +60,7 @@ class TestAtomsParser(TestUsingMemoryBasedSheerka):
concepts_map = {
"foo bar": Concept("foo bar"),
"one two three": Concept("one two three"),
"twenties": Concept("twenties", definition="'twenty' ('one'|'two')=unit").def_prop("unit"),
"twenties": Concept("twenties", definition="'twenty' ('one'|'two')=unit").def_var("unit"),
}
sheerka, context, parser = self.init_parser(concepts_map)
@@ -75,27 +78,27 @@ class TestAtomsParser(TestUsingMemoryBasedSheerka):
("foo bar suffixed one", False, ["foo bar", " suffixed ", "one"]),
("foo bar one prefixed", False, ["foo bar", "one", " prefixed"]),
("foo bar one infix two", False, ["foo bar", "one", " infix ", "two"]),
("foo bar 1 + 1", True, ["foo bar", " 1 + 1"]),
("foo bar 1 + 1", False, ["foo bar", " 1 + 1"]),
("foo bar twenty one", False, ["foo bar", " twenty ", "one"]),
("foo bar x$!#", False, ["foo bar", " x$!#"]),
("suffixed one foo bar", False, ["suffixed ", "one", "foo bar"]),
("one prefixed foo bar", False, ["one", " prefixed ", "foo bar"]),
("one infix two foo bar", False, ["one", " infix ", "two", "foo bar"]),
("1 + 1 foo bar", True, ["1 + 1 ", "foo bar"]),
("1 + 1 foo bar", False, ["1 + 1 ", "foo bar"]),
("twenty one foo bar", False, ["twenty ", "one", "foo bar"]),
("x$!# foo bar", False, ["x$!# ", "foo bar"]),
("func(one)", False, ["func(", "one", ")"]),
])
def test_i_can_parse_when_unrecognized(self, text, expected_status, expected):
concepts_map = {
"prefixed": Concept("a prefixed").def_prop("a"),
"suffixed": Concept("prefixed a").def_prop("a"),
"infix": Concept("a infix b").def_prop("a").def_prop("b"),
"prefixed": Concept("a prefixed").def_var("a"),
"suffixed": Concept("prefixed a").def_var("a"),
"infix": Concept("a infix b").def_var("a").def_var("b"),
"foo bar": Concept("foo bar"),
"one": Concept("one"),
"two": Concept("two"),
"twenties": Concept("twenties", definition="'twenty' ('one'|'two')=unit").def_prop("unit"),
"twenties": Concept("twenties", definition="'twenty' ('one'|'two')=unit").def_var("unit"),
}
sheerka, context, parser = self.init_parser(concepts_map)
@@ -116,13 +119,13 @@ class TestAtomsParser(TestUsingMemoryBasedSheerka):
])
def test_i_can_parse_when_surrounded_by_spaces(self, text, expected_status, expected):
concepts_map = {
"prefixed": Concept("a prefixed").def_prop("a"),
"suffixed": Concept("prefixed a").def_prop("a"),
"infix": Concept("a infix b").def_prop("a").def_prop("b"),
"prefixed": Concept("a prefixed").def_var("a"),
"suffixed": Concept("prefixed a").def_var("a"),
"infix": Concept("a infix b").def_var("a").def_var("b"),
"foo bar": Concept("foo bar"),
"one": Concept("one"),
"two": Concept("two"),
"twenties": Concept("twenties", definition="'twenty' ('one'|'two')=unit").def_prop("unit"),
"twenties": Concept("twenties", definition="'twenty' ('one'|'two')=unit").def_var("unit"),
}
sheerka, context, parser = self.init_parser(concepts_map)
@@ -201,8 +204,8 @@ class TestAtomsParser(TestUsingMemoryBasedSheerka):
])
def test_i_cannot_parse_concepts_with_property_or_bnf_or_unrecognized(self, text):
concepts_map = {
"foo": Concept("foo a").def_prop("a"),
"twenties": Concept("twenties", definition="'twenty' ('one'|'two')=unit").def_prop("unit"),
"foo": Concept("foo a").def_var("a"),
"twenties": Concept("twenties", definition="'twenty' ('one'|'two')=unit").def_var("unit"),
}
sheerka, context, parser = self.init_parser(concepts_map)
@@ -221,12 +224,12 @@ class TestAtomsParser(TestUsingMemoryBasedSheerka):
])
def test_i_can_parse_when_unrecognized_yield_multiple_values(self, text, expected):
concepts_map = {
"hello1": Concept("hello a").def_prop("a"),
"hello2": Concept("hello b").def_prop("b"),
"hello1": Concept("hello a").def_var("a"),
"hello2": Concept("hello b").def_var("b"),
"bar": Concept("bar")
}
sheerka, context, parser = self.init_parser(concepts_map, create_new=True)
sheerka, context, parser = self.init_parser(concepts_map, create_new=True, singleton=False)
list_of_res = parser.parse(context, text)
assert len(list_of_res) == len(expected)
@@ -239,3 +242,23 @@ class TestAtomsParser(TestUsingMemoryBasedSheerka):
expected_array = compute_expected_array(concepts_map, text, expected[1])
assert sheerka.isinstance(wrapper, BuiltinConcepts.PARSER_RESULT)
assert lexer_nodes == expected_array
@pytest.mark.parametrize("text, expected", [
("1 + twenty one", [SCN("1 + twenty "), "one"]),
("one + twenty one", ["one", SCN(" + twenty "), ("one", 1)]),
])
def test_source_code_found_must_be_considered_as_potential_false_positive(self, text, expected):
concepts_map = {
"one": Concept("one")
}
sheerka, context, parser = self.init_parser(concepts_map)
res = parser.parse(context, text)
wrapper = res.body
lexer_nodes = res.body.body
assert not res.status
expected_array = compute_expected_array(concepts_map, text, expected)
assert sheerka.isinstance(wrapper, BuiltinConcepts.PARSER_RESULT)
assert lexer_nodes == expected_array