refactoring

core/change_parser.py

@@ -0,0 +1,85 @@
+"""
+core/change_parser.py — 선수 교체 이벤트 파싱
+
+교체 텍스트에서 선수명, 포지션, 교체 유형 등을 추출합니다.
+"""
+from __future__ import annotations
+
+import re
+from typing import Any
+
+from core.config_loader import position_to_defense_no
+
+
+def extract_change_actor(text: str) -> tuple[str | None, int | None, str]:
+    """교체 텍스트의 왼쪽(actor)에서 역할, 타순, 이름 추출
+
+    '5번타자 문보경' → ('batter', 5, '문보경')
+    '투수 임찬규' → ('투수', None, '임찬규')
+    """
+    lhs = (text or "").split(" : ", 1)[0].strip()
+
+    batter_match = re.search(r"(\d+)번타자\s+(.+)$", lhs)
+    if batter_match:
+        return "batter", int(batter_match.group(1)), batter_match.group(2).strip()
+
+    roles = (
+        "대타", "대주자",
+        "1루주자", "2루주자", "3루주자", "주자",
+        "투수", "포수", "1루수", "2루수", "3루수",
+        "유격수", "좌익수", "중견수", "우익수",
+    )
+    for role in roles:
+        if lhs.startswith(role + " "):
+            return role, None, lhs[len(role):].strip()
+
+    return None, None, lhs
+
+
+def is_merged_pitcher_substitution(actor_role: str | None, in_role: str | None) -> bool:
+    """야수→투수 교체인지 확인 (투수가 DH로 전환되는 병합 교체)"""
+    field_roles = {"포수", "1루수", "2루수", "3루수", "유격수", "좌익수", "중견수", "우익수"}
+    return actor_role in field_roles and in_role == "투수"
+
+
+def normalize_change_event(change_event: dict[str, Any]) -> dict[str, Any]:
+    """교체 이벤트를 정규화
+
+    텍스트 파싱 → actor_name, out_player, in_player, change_type 등 추출
+    """
+    if change_event.get("actor_name") or change_event.get("player_name"):
+        return change_event
+
+    text = change_event.get("text") or ""
+    normalized = dict(change_event)
+    normalized["change_type"] = "position_change" if "수비위치 변경" in text else "substitution"
+    actor_role, bat_order, actor_name = extract_change_actor(text)
+    normalized["actor_role"] = actor_role
+    normalized["actor_name"] = actor_name
+    if bat_order is not None:
+        normalized["bat_order"] = bat_order
+
+    if normalized["change_type"] == "position_change":
+        rhs = text.split(" : ", 1)[1] if " : " in text else ""
+        normalized["player_name"] = actor_name
+        normalized["to_position"] = rhs.split("(으)로", 1)[0].strip()
+        return normalized
+
+    rhs = text.split(" : ", 1)[1] if " : " in text else ""
+    rhs = rhs.split("(으)로 교체", 1)[0].strip()
+    in_role, _, in_name = extract_change_actor(rhs)
+    normalized["out_player"] = actor_name
+    normalized["in_player"] = in_name
+    normalized["in_role"] = in_role
+
+    pos_defense = position_to_defense_no()
+    if is_merged_pitcher_substitution(actor_role, in_role):
+        normalized["change_type"] = "merged_pitcher_substitution"
+        normalized["player_name"] = actor_name
+        normalized["to_position"] = "지명타자"
+        normalized["pitcher_in_player"] = in_name
+        return normalized
+
+    if in_role in pos_defense:
+        normalized["to_position"] = in_role
+    return normalized