quiz/app/matcher.py

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]

WORD_TO_NUM = {
    "null": "0", "zero": "0",
    "eins": "1", "ein": "1", "one": "1",
    "zwei": "2", "two": "2",
    "drei": "3", "three": "3",
    "vier": "4", "four": "4",
    "fuenf": "5", "five": "5",
    "sechs": "6", "six": "6",
    "sieben": "7", "seven": "7",
    "acht": "8", "eight": "8",
    "neun": "9", "nine": "9",
    "zehn": "10", "ten": "10",
    "elf": "11", "eleven": "11",
    "zwoelf": "12", "twelve": "12",
    "dreizehn": "13", "thirteen": "13",
    "vierzehn": "14", "fourteen": "14",
    "fuenfzehn": "15", "fifteen": "15",
    "sechzehn": "16", "sixteen": "16",
    "siebzehn": "17", "seventeen": "17",
    "achtzehn": "18", "eighteen": "18",
    "neunzehn": "19", "nineteen": "19",
    "zwanzig": "20", "twenty": "20",
    "fuenfzig": "50", "fifty": "50",
    "hundert": "100", "einhundert": "100", "one hundred": "100",
    "einhundertfuenfzig": "150", "one hundred fifty": "150", "one hundred and fifty": "150"
}

NUM_TO_WORDS = {}
for w, n in WORD_TO_NUM.items():
    NUM_TO_WORDS.setdefault(n, []).append(w)

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

    # Also consider digit conversion of user input if a word was typed
    user_variants = [clean_user]
    if clean_user in WORD_TO_NUM:
        user_variants.append(WORD_TO_NUM[clean_user])

    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)

    # Expand any numeric target with corresponding number words and vice-versa
    expanded_targets = list(targets)
    for t in targets:
        if t in NUM_TO_WORDS:
            expanded_targets.extend(NUM_TO_WORDS[t])
        if t in WORD_TO_NUM:
            expanded_targets.append(WORD_TO_NUM[t])

    for target in expanded_targets:
        if not target:
            continue
        for u in user_variants:
            if u == target:
                return True
            # For short answers (1-3 chars, e.g. numbers or abbreviations), require exact match
            if len(target) <= 3 or len(u) <= 3:
                continue
            # For medium words (4-6 chars), allow max 1 typo
            max_allowed = 1 if len(target) <= 6 else 2
            dist = levenshtein_distance(u, target)
            if dist <= max_allowed:
                return True

            # Support compound answers connected by 'und' / 'and' (e.g. "Mantikor und Feuerkrabbe")
            for sep in (" und ", " and "):
                if sep in target:
                    parts = [p.strip() for p in target.split(sep) if len(p.strip()) >= 3]
                    if len(parts) == 2:
                        u_words = u.split()
                        matched_both = True
                        for part in parts:
                            if part in u:
                                continue
                            # Check fuzzy match against tokens or reversed order
                            part_matched = any(levenshtein_distance(w, part) <= (1 if len(part) <= 6 else 2) for w in u_words)
                            if not part_matched:
                                matched_both = False
                                break
                        if matched_both:
                            return True

    return False