[scrap] 신규의원 후보중에 찾기. 아직 후보목록이 mongodb에 없어서 테스트는 하지 않음

API/candidate.py

@@ -88,4 +88,4 @@ def fetch_all_data(
     if args.save_method == "excel":
         save_to_excel(data_list, args.sgTypecode, is_elected=False)
     elif args.save_method == "mongo":
-        save_to_mongo(data_list, args.sgTypecode)
+        save_to_mongo(data_list, args.sgTypecode, "local_councilor_cand")

API/elected.py

@@ -85,4 +85,4 @@ def fetch_all_data(
     if args.save_method == "excel":
         save_to_excel(data_list, args.sgTypecode, is_elected=True)
     elif args.save_method == "mongo":
-        save_to_mongo(data_list, args.sgTypecode)
+        save_to_mongo(data_list, args.sgTypecode, "local_councilor")

API/utils.py

@@ -36,10 +36,9 @@ def get_district_id(sd_name: str, wiw_name: str) -> Optional[int]:
     return LOCAL_METRO_ID_MAP[(sd_name, wiw_name)]
 
 
-def save_to_mongo(data: List[dict], sgTypecode: str) -> None:
+def save_to_mongo(data: List[dict], sgTypecode: str, where: str) -> None:
     db = client["council"]
-    main_collection = db["local_councilor"]
-    # main_collection = db["test"]
+    main_collection = db[where]
 
     # TODO: Support other types of councils
     if sgTypecode in ["6", "9"]:

scrap/utils/database.py

@@ -33,6 +33,7 @@ def save_to_database(record: ScrapResult) -> bool:
     # TODO: DB에 없던 새로운 의원 핸들링
     update_councilors(collection, cid, other_councilors)
     remove_councilors(collection, resigned_councilors)
+    add_councilors(collection, cid, new_councilors, str(record.council_type))
 
     return True
 
@@ -96,3 +97,32 @@ def remove_councilors(collection, councilors: List[Councilor]) -> None:
         collection.delete_one(
             {"localId": councilor["localId"], "name": councilor["name"]}
         )
+
+def add_councilors(collection, cid: int, councilors: List[Councilor], council_type: str) -> None:
+    """
+    신규 의원을 검사합니다. 기존 비례대표의 사퇴로 새 비례대표가 바로 추가되었을
+    가능성을 보고, 아닐 시에는 사용자에게 보고합니다.
+    :param collection: MongoDB 컬렉션
+    :param councilors: 업데이트할 의원 리스트
+    :param council_type: local_councilor 등등, 의원의 종류 
+    """
+    db = client[str(MongoDBSecrets.database_name)]
+    candcoll = db[council_type + "_cand"] # 예: local_councilor_cand
+    for councilor in councilors:
+        # find the councilor in candidate
+        query = {"localId": cid, "name": councilor.name}
+        all_documents = list(candcoll.find(query))
+        assert(len(all_documents) <= 1, "신규의원을 같은 의회의 후보중에 찾는데, 동명이인이 있네요.")
+        if len(all_documents) == 0:
+            # no candidate found, report to user
+            print(f"신규의원 {councilor.name}을 찾을 수 없습니다.")
+            continue
+        found_document = all_documents[0]
+        age = found_document.get("age")
+        # birthday = found_document.get("birthday")
+        councilor.age = age
+        collection.update_one(
+            {"localId": cid, "name": councilor.name},
+            {"$set": asdict(councilor)},
+            upsert=False,
+        )