[scrap] Integrate the last commit #8aa00ca into current format. add f…

…ew councils
NewWays-TechForImpactKAIST · Nov 15, 2023 · dd5764c · dd5764c
1 parent d6ce8d9
commit dd5764c
Show file tree

Hide file tree

Showing 47 changed files with 857 additions and 1,649 deletions.
diff --git a/scrap/local_councils/__init__.py b/scrap/local_councils/__init__.py
@@ -2,9 +2,6 @@
 각 기초의회들의 크롤링 코드를 모아놓은 패키지입니다.
 광역자치단체 별로 폴더를 만들어서 관리합니다.
 """
-from .daejeon.daejeon import *
-from .ulsan import *
-from .daejeon import *
 import re
 from urllib.parse import urlparse
 from typing import List

diff --git a/scrap/local_councils/busan.py b/scrap/local_councils/busan.py
@@ -239,7 +239,7 @@ def scrap_36(url, cid, args: ArgsType = None) -> ScrapResult:
             party = "정당 정보 없음"
             party_info = name_tag.find_next("span", string="소속당  : ")
             if party_info:
-                party = party_info.parent.get_text(strip=True)[7:].strip()
+                party = party_info.parent.get_text(strip=True)[6:].strip()
 
             councilors.append(Councilor(name=name, jdName=party))
 
@@ -284,7 +284,7 @@ def scrap_38(url, cid, args: ArgsType = None) -> ScrapResult:
         party = "정당 정보 없음"
         party_info = profile.find("span", class_="bold", string="정당 : ")
         if party_info:
-            party = party_info.parent.get_text(strip=True)[5:].strip()
+            party = party_info.parent.get_text(strip=True)[4:].strip()
 
         councilors.append(Councilor(name=name, jdName=party))
 

diff --git a/scrap/local_councils/daegu.py b/scrap/local_councils/daegu.py
@@ -105,7 +105,7 @@ def scrap_46(url, cid, args: ArgsType = None) -> ScrapResult:
     for profile in soup.find_all("div", class_="profile"):
         name_tag = profile.find("em", class_="name")
         name = (
-            name_tag.get_text(strip=True).split()[0].strip() if name_tag else "이름 정보 없음"
+            name_tag.get_text(strip=True).split()[0].replace('의원', '').strip() if name_tag else "이름 정보 없음"
         )
 
         party = "정당 정보 없음"

diff --git a/scrap/local_councils/daejeon/daejeon.py b/scrap/local_councils/daejeon/daejeon.py
diff --git a/scrap/local_councils/gangwon.py b/scrap/local_councils/gangwon.py
@@ -1,5 +1,6 @@
 import os
 
+from scrap.local_councils import *
 from scrap.local_councils.basic import *
 from scrap.utils.requests import get_selenium, By
 from scrap.utils.utils import getPartyList
@@ -171,7 +172,7 @@ def scrap_118(
 
     for profile in soup.find_all("div", class_="person_info"):
         name_td = profile.find("th", string="성함/직위").find_next("td")
-        name = name_td.get_text(strip=True) if name_td else "이름 정보 없음"
+        name = name_td.get_text(strip=True).split()[0] if name_td else "이름 정보 없음"
 
         party_td = profile.find("th", string="정당").find_next("td")
         party = party_td.get_text(strip=True) if party_td else "정당 정보 없음"
@@ -290,7 +291,7 @@ def scrap_123(
     for profile in soup.find_all("div", class_="img_text_box"):
         name_li = profile.select_one("ul.bu li:contains('성명')")
         name = (
-            name_li.get_text(strip=True).replace("성명", "").strip()
+            name_li.get_text(strip=True).replace("성명", "").strip().split()[0]
             if name_li
             else "이름 정보 없음"
         )