ExactConceptParser can now recognize concepts by their names
This commit is contained in:
@@ -255,6 +255,26 @@ class BaseParser:
|
||||
|
||||
return start, end
|
||||
|
||||
@staticmethod
|
||||
def merge_concepts(list_a, b):
|
||||
if not b:
|
||||
return list_a
|
||||
|
||||
list_b = b if isinstance(b, list) else [b]
|
||||
|
||||
if not list_a:
|
||||
return list_b
|
||||
|
||||
by_ids = {c.id for c in list_b}
|
||||
for c in list_b:
|
||||
if c.id in by_ids: # and c.metadata.is_evaluated == by_ids[c.id].metadata.is_evaluated:
|
||||
continue
|
||||
|
||||
list_a.append(c)
|
||||
by_ids.add(c.id)
|
||||
|
||||
return list_a
|
||||
|
||||
|
||||
class BaseTokenizerIterParser(BaseParser):
|
||||
|
||||
|
||||
@@ -455,7 +455,7 @@ class BnfNodeFirstTokenVisitor(ParsingExpressionVisitor):
|
||||
def visit_ConceptExpression(self, pe):
|
||||
concept = self.sheerka.get_by_key(pe.concept) if isinstance(pe.concept, str) else pe.concept
|
||||
if self.sheerka.is_known(concept):
|
||||
self.add_first_token(core.utils.str_concept(concept, skip_key=True))
|
||||
self.add_first_token(core.utils.str_concept(concept, drop_name=True))
|
||||
return self.STOP
|
||||
|
||||
def visit_StrMatch(self, pe):
|
||||
|
||||
@@ -3,7 +3,9 @@ import logging
|
||||
from core.builtin_concepts import ReturnValueConcept, BuiltinConcepts
|
||||
from core.concept import VARIABLE_PREFIX
|
||||
from core.tokenizer import Keywords, TokenKind, LexerError
|
||||
from core.utils import str_concept
|
||||
from parsers.BaseParser import BaseParser
|
||||
import core.builtin_helpers
|
||||
|
||||
|
||||
class ExactConceptParser(BaseParser):
|
||||
@@ -26,8 +28,8 @@ class ExactConceptParser(BaseParser):
|
||||
"""
|
||||
|
||||
context.log(f"Parsing '{parser_input}'", self.name)
|
||||
res = []
|
||||
sheerka = context.sheerka
|
||||
|
||||
try:
|
||||
words = self.get_words(parser_input)
|
||||
except LexerError as e:
|
||||
@@ -40,7 +42,7 @@ class ExactConceptParser(BaseParser):
|
||||
body = sheerka.new(BuiltinConcepts.NOT_FOR_ME, body=parser_input, reason=too_long)
|
||||
return sheerka.ret(self.name, False, body)
|
||||
|
||||
recognized = [] # keep track of the concepts founds
|
||||
already_recognized = [] # keep track of the concepts founds
|
||||
for combination in self.combinations(words):
|
||||
|
||||
concept_key = " ".join(combination)
|
||||
@@ -52,7 +54,7 @@ class ExactConceptParser(BaseParser):
|
||||
concepts = result if isinstance(result, list) else [result]
|
||||
|
||||
for concept in concepts:
|
||||
if concept.id in recognized:
|
||||
if concept in already_recognized:
|
||||
context.log(f"Recognized concept {concept} again. Skipping.", self.name)
|
||||
# example
|
||||
# if the input is foo a and a concept is defined as foo a
|
||||
@@ -65,38 +67,33 @@ class ExactConceptParser(BaseParser):
|
||||
for i, token in enumerate(combination):
|
||||
if token.startswith(VARIABLE_PREFIX):
|
||||
index = int(token[len(VARIABLE_PREFIX):])
|
||||
concept.def_var_by_index(index, words[i])
|
||||
value = words[i]
|
||||
concept.def_var_by_index(index, str_concept(value) if isinstance(value, tuple) else value)
|
||||
concept.metadata.need_validation = True
|
||||
if self.verbose_log.isEnabledFor(logging.DEBUG):
|
||||
prop_name = concept.metadata.variables[index][0]
|
||||
context.log(
|
||||
f"Added property {index}: {prop_name}='{words[i]}'.",
|
||||
f"Added variable {index}: {prop_name}='{words[i]}'.",
|
||||
self.name)
|
||||
|
||||
res.append(ReturnValueConcept(
|
||||
self.name,
|
||||
True,
|
||||
context.sheerka.new(
|
||||
BuiltinConcepts.PARSER_RESULT,
|
||||
parser=self,
|
||||
source=parser_input if isinstance(parser_input, str) else self.get_text_from_tokens(
|
||||
parser_input),
|
||||
body=concept,
|
||||
try_parsed=concept)))
|
||||
recognized.append(concept.id)
|
||||
already_recognized.append(concept)
|
||||
|
||||
if len(recognized) > 0:
|
||||
by_name = sheerka.resolve(self.get_input_as_text(parser_input))
|
||||
core.builtin_helpers.set_is_evaluated(by_name)
|
||||
recognized = self.merge_concepts(already_recognized, by_name)
|
||||
|
||||
if len(recognized) == 0:
|
||||
ret = sheerka.ret(self.name, False, sheerka.new(BuiltinConcepts.UNKNOWN_CONCEPT, body=parser_input))
|
||||
self.log_result(context, parser_input, ret)
|
||||
return ret
|
||||
else:
|
||||
res = [self.as_return_value(context, parser_input, c) for c in recognized]
|
||||
if len(res) == 1:
|
||||
self.log_result(context, parser_input, res[0])
|
||||
else:
|
||||
self.log_multiple_results(context, parser_input, res)
|
||||
return res
|
||||
return res
|
||||
|
||||
ret = sheerka.ret(self.name, False, sheerka.new(BuiltinConcepts.UNKNOWN_CONCEPT, body=parser_input))
|
||||
self.log_result(context, parser_input, ret)
|
||||
return ret
|
||||
|
||||
def get_words(self, text):
|
||||
tokens = self.get_input_as_tokens(text)
|
||||
res = []
|
||||
@@ -138,7 +135,17 @@ class ExactConceptParser(BaseParser):
|
||||
indices[j] = indices[j - 1] + 1
|
||||
res.add(self.get_tuple(pool, indices))
|
||||
|
||||
return res
|
||||
# remove all result that contains a token concepts
|
||||
# They are not valid entries, since a token concept MUST be replaced by a variable
|
||||
filtered = set()
|
||||
for combination in res:
|
||||
for entry in combination:
|
||||
if isinstance(entry, tuple):
|
||||
break
|
||||
else:
|
||||
filtered.add(combination)
|
||||
|
||||
return filtered
|
||||
|
||||
@staticmethod
|
||||
def get_tuple(pool, indices):
|
||||
@@ -158,3 +165,14 @@ class ExactConceptParser(BaseParser):
|
||||
value = pool[i]
|
||||
res.append(vars[value] if value in vars else value)
|
||||
return tuple(res)
|
||||
|
||||
def as_return_value(self, context, parser_input, concept):
|
||||
return ReturnValueConcept(
|
||||
self.name,
|
||||
True,
|
||||
context.sheerka.new(
|
||||
BuiltinConcepts.PARSER_RESULT,
|
||||
parser=self,
|
||||
source=parser_input if isinstance(parser_input, str) else self.get_text_from_tokens(parser_input),
|
||||
body=concept,
|
||||
try_parsed=concept))
|
||||
|
||||
Reference in New Issue
Block a user