Create korean_rules.py

korean_rules.py

@@ -0,0 +1,188 @@
+"""
+korean_rules.py
+Pure-Python, deterministic Korean grammar rule engine.
+No ML. Uses Unicode Hangul decomposition for batchim detection.
+"""
+
+HANGUL_BASE   = 0xAC00
+INITIAL_COUNT = 21 * 28   # 588 per initial
+MEDIAL_COUNT  = 28
+
+FINALS = [
+    None,'ㄱ','ㄲ','ㄳ','ㄴ','ㄵ','ㄶ','ㄷ',
+    'ㄹ','ㄺ','ㄻ','ㄼ','ㄽ','ㄾ','ㄿ','ㅀ',
+    'ㅁ','ㅂ','ㅄ','ㅅ','ㅆ','ㅇ','ㅈ','ㅊ',
+    'ㅋ','ㅌ','ㅍ','ㅎ',
+]
+RIEUL = 'ㄹ'
+
+
+def _last_hangul(word):
+    for ch in reversed(word):
+        if '\uAC00' <= ch <= '\uD7A3':
+            return ch
+    return ''
+
+def _batchim(syl):
+    if not syl or not ('\uAC00' <= syl <= '\uD7A3'):
+        return None
+    return FINALS[(ord(syl) - HANGUL_BASE) % 28]
+
+def has_batchim(word):
+    return _batchim(_last_hangul(word)) is not None
+
+def has_batchim_no_rieul(word):
+    b = _batchim(_last_hangul(word))
+    return b is not None and b != RIEUL
+
+
+class KoreanRuleEngine:
+
+    # ── Particles ─────────────────────────────────────────────────────────────
+
+    def get_topic_marker(self, noun):
+        return '은' if has_batchim_no_rieul(noun) else '는'
+
+    def get_subject_marker(self, noun):
+        return '이' if has_batchim_no_rieul(noun) else '가'
+
+    def get_object_marker(self, noun):
+        return '을' if has_batchim(noun) else '를'
+
+    def get_copula(self, noun):
+        return '이에요' if has_batchim(noun) else '예요'
+
+    def get_negative_marker(self, noun):
+        """Full negative copula: '이 아니에요' or '가 아니에요'."""
+        m = '이' if has_batchim_no_rieul(noun) else '가'
+        return f'{m} 아니에요'
+
+    def attach_topic_marker(self, noun):
+        return noun + self.get_topic_marker(noun)
+
+    def attach_subject_marker(self, noun):
+        return noun + self.get_subject_marker(noun)
+
+    def attach_object_marker(self, noun):
+        return noun + self.get_object_marker(noun)
+
+    def attach_copula(self, noun):
+        return noun + self.get_copula(noun)
+
+    def attach_negative_copula(self, noun):
+        return noun + self.get_negative_marker(noun)
+
+    # ── Indirect quotation ────────────────────────────────────────────────────
+
+    def conjugate_indirect_quote(self, verb_stem, form, tense='present',
+                                  is_adjective=False):
+        """
+        form: statement | command | neg_command |
+              request_me | request_other | question | suggestion
+        tense: past | present | future
+        is_adjective: True for adjectives/있다/없다 — uses plain +다고 in present
+        """
+        s = verb_stem
+        if form == 'statement':
+            if tense == 'past':
+                suffix = '았' if self._needs_a(s) else '었'
+                return s + suffix + '다고'
+            if tense == 'future':
+                return (self._attach_batchim(s, 'ㄹ') + ' 거라고'
+                        if not has_batchim(s) else s + '을 거라고')
+            # present
+            if is_adjective:
+                return s + '다고'   # adjective/있다/없다: stem+다고
+            return (self._attach_batchim(s, 'ㄴ') + '다고'
+                    if not has_batchim(s) else s + '는다고')
+        if form == 'command':
+            return s + ('라고' if not has_batchim(s) else '으라고')
+        if form == 'neg_command':
+            return s + '지 말라고'
+        if form == 'request_me':
+            return s + self._vowel(s) + ' 달라고'
+        if form == 'request_other':
+            return s + self._vowel(s) + ' 주라고'
+        if form == 'question':
+            return self._drop_rieul(s) + '냐고'
+        if form == 'suggestion':
+            return s + '자고'
+        return s + '다고'
+
+    def conjugate_regret(self, verb_stem, negative=False):
+        if negative:
+            return verb_stem + '지 말 걸 그랬다'
+        if not has_batchim(verb_stem):
+            return self._attach_batchim(verb_stem, 'ㄹ') + ' 걸 그랬다'
+        return verb_stem + '을 걸 그랬다'
+
+    # ── Validation ────────────────────────────────────────────────────────────
+
+    def validate_token_order(self, submitted, correct):
+        return list(submitted) == list(correct)
+
+    def validate_particle(self, word, chosen, ptype):
+        fn = {
+            'topic':    self.get_topic_marker,
+            'subject':  self.get_subject_marker,
+            'object':   self.get_object_marker,
+            'copula':   self.get_copula,
+            'negative': self.get_negative_marker,
+        }.get(ptype)
+        return fn is not None and chosen == fn(word)
+
+    def get_hint(self, word, ptype):
+        syl  = _last_hangul(word)
+        b    = _batchim(syl) if syl else None
+        desc = f"ends with consonant '{b}'" if b else "ends with a vowel sound"
+        ans  = {
+            'topic':    '은' if (b and b != RIEUL) else '는',
+            'subject':  '이' if (b and b != RIEUL) else '가',
+            'object':   '을' if b else '를',
+            'copula':   '이에요' if b else '예요',
+            'negative': ('이 아니에요' if (b and b != RIEUL) else '가 아니에요'),
+        }.get(ptype, '?')
+        return f"'{word}' {desc} → use {ans}"
+
+    # ── Internal ──────────────────────────────────────────────────────────────
+
+    def _needs_a(self, stem):
+        syl = _last_hangul(stem)
+        if not syl:
+            return False
+        medial = ((ord(syl) - HANGUL_BASE) // 28) % 21
+        return medial in (0, 8)
+
+    def _vowel(self, stem):
+        return '아' if self._needs_a(stem) else '어'
+
+    def _drop_rieul(self, stem):
+        syl = _last_hangul(stem)
+        if syl and _batchim(syl) == RIEUL:
+            code    = ord(syl) - HANGUL_BASE
+            initial = code // INITIAL_COUNT
+            medial  = (code % INITIAL_COUNT) // 28
+            return stem[:-1] + chr(HANGUL_BASE + initial * INITIAL_COUNT + medial * 28)
+        return stem
+
+    def _attach_batchim(self, stem, jamo):
+        """
+        Attach a jamo final consonant to the last syllable of stem.
+        e.g. '가' + 'ㄴ' → '간',  '가' + 'ㄹ' → '갈'
+        Only works when last syllable has no existing batchim.
+        """
+        FINAL_IDX = {f: i for i, f in enumerate(FINALS) if f}
+        idx = FINAL_IDX.get(jamo)
+        if idx is None:
+            return stem + jamo   # fallback: just concatenate
+        syl = _last_hangul(stem)
+        if not syl or _batchim(syl) is not None:
+            return stem + jamo   # already has batchim
+        code    = ord(syl) - HANGUL_BASE
+        initial = code // INITIAL_COUNT
+        medial  = (code % INITIAL_COUNT) // 28
+        new_syl = chr(HANGUL_BASE + initial * INITIAL_COUNT + medial * 28 + idx)
+        return stem[:-1] + new_syl
+
+
+rule_engine = KoreanRuleEngine()