def literal_match(self, literal: Literal, surface: str):
dtype = literal.datatype if hasattr(literal, "datatype") else None
literal, surface = str(literal).strip(), str(surface).strip()
score = 0.0
if dtype:
# Typed literals should match well
if str(dtype) == str(self.DATETIME):
try:
l = datetime.datetime.fromisoformat(literal).timestamp()
yearmatch = YEAR_PATTERN.match(surface)
if yearmatch:
year = int(yearmatch.groups()[0])
s = datetime.datetime(year, 1, 1).timestamp()
else:
try:
s = datetime.datetime.fromisoformat(
surface).timestamp()
except:
s = self._dateparse(surface).timestamp()
if s:
score = max(0, 1 - (abs(s - l) / (60 * 60 * 24 * 365)))
if score:
yield LiteralMatchResult(score, literal, dtype)
return
# else:
# log.debug(f"No date match ({l},{s}) = {score}")
except Exception as e:
pass
else:
try:
s = float(surface.replace(",", ""))
l = float(literal.replace(",", ""))
score = max(0, 1 - (abs(s - l) / max(abs(s), abs(l))))
if score > 0.95:
yield LiteralMatchResult(score, literal, dtype)
return
except Exception as e:
pass
score = bool(surface.lower() == literal.lower())
elif surface and literal:
# Strings may match approximately
if self.stringmatch == "jaccard":
stok, ltok = set(surface.lower().split()), set(
literal.lower().split())
if stok and ltok:
score = len(stok & ltok) / len(stok | ltok)
elif self.stringmatch == "levenshtein":
import Levenshtein
slow, llow = surface.lower(), literal.lower()
if slow and llow:
m = min(len(slow), len(llow))
score = max(0, (m - Levenshtein.distance(slow, llow)) / m)
if score:
yield LiteralMatchResult(score, literal, dtype)
