kobold-ai
74 lines
2.5 KiB
Python
Executable File
74 lines
2.5 KiB
Python
Executable File
import re
|
|
import unicodedata
|
|
import json
|
|
|
|
def normalize_text(text: str) -> str:
|
|
if not text:
|
|
return ""
|
|
text = text.lower().strip()
|
|
# Replace German umlauts
|
|
text = text.replace("ä", "ae").replace("ö", "oe").replace("ü", "ue").replace("ß", "ss")
|
|
# Remove accents
|
|
nfkd = unicodedata.normalize("NFKD", text)
|
|
text = "".join([c for c in nfkd if not unicodedata.combining(c)])
|
|
# Remove special punctuation like quotes, dots, dashes, commas
|
|
text = re.sub(r"[^\w\s]", "", text)
|
|
# Collapse whitespace
|
|
text = re.sub(r"\s+", " ", text).strip()
|
|
return text
|
|
|
|
def levenshtein_distance(s1: str, s2: str) -> int:
|
|
if len(s1) < len(s2):
|
|
return levenshtein_distance(s2, s1)
|
|
if len(s2) == 0:
|
|
return len(s1)
|
|
previous_row = list(range(len(s2) + 1))
|
|
for i, c1 in enumerate(s1):
|
|
current_row = [i + 1]
|
|
for j, c2 in enumerate(s2):
|
|
insertions = previous_row[j + 1] + 1
|
|
deletions = current_row[j] + 1
|
|
substitutions = previous_row[j] + (c1 != c2)
|
|
current_row.append(min(insertions, deletions, substitutions))
|
|
previous_row = current_row
|
|
return previous_row[-1]
|
|
|
|
def is_answer_correct(user_input: str, canonical: str, accepted_raw: str | None = None, canonical_en: str | None = None) -> bool:
|
|
clean_user = normalize_text(user_input)
|
|
if not clean_user:
|
|
return False
|
|
|
|
targets = [normalize_text(canonical)]
|
|
if canonical_en:
|
|
clean_en = normalize_text(canonical_en)
|
|
if clean_en and clean_en not in targets:
|
|
targets.append(clean_en)
|
|
|
|
if accepted_raw:
|
|
try:
|
|
parsed = json.loads(accepted_raw)
|
|
if isinstance(parsed, list):
|
|
targets.extend([normalize_text(str(x)) for x in parsed])
|
|
except Exception:
|
|
# Fallback comma-separated
|
|
for part in accepted_raw.split(","):
|
|
part_clean = normalize_text(part)
|
|
if part_clean:
|
|
targets.append(part_clean)
|
|
|
|
for target in targets:
|
|
if not target:
|
|
continue
|
|
if clean_user == target:
|
|
return True
|
|
# For short answers (1-3 chars, e.g. numbers or abbreviations), require exact match
|
|
if len(target) <= 3:
|
|
continue
|
|
# For medium words (4-6 chars), allow max 1 typo
|
|
max_allowed = 1 if len(target) <= 6 else 2
|
|
dist = levenshtein_distance(clean_user, target)
|
|
if dist <= max_allowed:
|
|
return True
|
|
|
|
return False
|