1. 페이지 접근

url = 'https://www.opinet.co.kr/searRgSelect.do'
driver = webdriver.Chrome("../driver/chromedriver.exe")
driver.get(url)

2. 서울 고정

seoul = driver.find_element_by_css_selector('#SIDO_NM0 > option:nth-child(2)')
seoul.click()

3. 구별 선택

gu_list_raw = driver.find_element_by_css_selector('#SIGUNGU_NM0') #부모태그
gu_list = gu_list_raw.find_elements_by_css_selector('option') 
gu_names = [option.get_attribute('value') for option in gu_list if option.get_attribute('value')] #if문은 공백제거역할

import pandas as pd #최종저장
import time
import googlemaps

gmaps_key = "AIzaSyC-aSVFR4qL3Jphc-MVw4Nmv_5YkjzQ7_k"
gmaps = googlemaps.Client(key=gmaps_key)

datas = []

#전체 구 가져오고, 개수에 맞게끔 순환
for i in range(len(gu_names)):
    gu_selecter = f'#SIGUNGU_NM0 > option:nth-child({i+2})'
    driver.find_element_by_css_selector(gu_selecter).click()
    station_items = driver.find_elements_by_css_selector('#body1 > tr') # 주유소 목록
    
    for idx in range(len(station_items)):
        detail_selector = f'#body1 > tr:nth-child({idx+1}) > td.rlist > a'
        driver.find_element_by_css_selector(detail_selector).click()
        #title = driver.find_element_by_css_selector('.header #os_nm').get_attribute('innerText') 
        #print(f'{v} 주유소 이름: ', title)
        #time.sleep(0.5)
        # 메인 데이터 수집

        name = driver.find_element_by_css_selector('.header #os_nm').get_attribute('innerText')
        gasoline = driver.find_element_by_css_selector('#b027_p').get_attribute('innerText')
        diesel = driver.find_element_by_css_selector('#d047_p').get_attribute('innerText')
        address = driver.find_element_by_css_selector('#rd_addr').get_attribute('innerText')
        brand = driver.find_element_by_css_selector('#poll_div_nm').get_attribute('innerText')
        cwsh_yn = 'N' if '_off' in driver.find_element_by_css_selector('.service #cwsh_yn').get_attribute('src').split('/')[-1] else 'Y'
        lpg_yn = 'N' if '_off' in driver.find_element_by_css_selector('.service #lpg_yn').get_attribute('src').split('/')[-1] else 'Y'
        maint_yn = 'N' if '_off' in driver.find_element_by_css_selector('.service #maint_yn').get_attribute('src').split('/')[-1] else 'Y'
        cvs_yn = 'N' if '_off' in driver.find_element_by_css_selector('.service #cvs_yn').get_attribute('src').split('/')[-1] else 'Y'
        sel24_yn = 'N' if '_off' in driver.find_element_by_css_selector('.service #sel24_yn').get_attribute('src').split('/')[-1] else 'Y'
        
        try: 
            driver.find_element_by_css_selector('#self_icon').get_attribute('alt')
            is_self = 'Y'

        except:
            is_self = 'N'


        # 구, 위도, 경도 한꺼번에 하겠음. sql할 때 한 코드안에 있어야 편함
        address = driver.find_element_by_css_selector('#rd_addr').get_attribute('innerText')
        
        #gu
        gu = address.split()[1]

        #lat, lng #주소값으로 검색함
        tmp = gmaps.geocode(address, language='ko') 
        lat = tmp[0].get('geometry')['location']['lat']
        lng = tmp[0].get('geometry')['location']['lng']

        datas.append({
            'name': name,
            'address': address,
            'brand': brand,
            'is_self': is_self,
            'gasoline': gasoline,
            'diesel': diesel,
            'car_wash': cwsh_yn,
            'charging_station': lpg_yn,
            'car_maintenance': maint_yn,
            'convenience_store': cvs_yn,
            '24_hour': sel24_yn,
            'gu': gu,
            'lat': lat,
            'lng': lng
        }) #위에서 변수로 찍어주고, 밑에서 이렇게 써주는게 유지보수가 편함. 위에서만 수정하면 반영됨

        time.sleep(0.2)
    time.sleep(0.5)

driver.quit()
df = pd.DataFrame(datas)
df.tail()

df.info()

csv 파일로 저장

import datetime
now = datetime.datetime.now()
nowDate = now.strftime('%Y%m%d')

df.to_csv(f'./oilstation_oneday_{nowDate}.csv', encoding='utf-8')

csv 파일 읽기

thousand=',' 로 하면 ,를 지우고 데이터타입을 int타입으로 인식함

stations = pd.read_csv('./oilstation_oneday_20240419.csv', encoding='utf-8', thousands=',', index_col=0)
stations

3. 주유소 정보 시각화

한글 설정

import matplotlib.pyplot as plt
import seaborn as sns
import platform
from matplotlib import font_manager, rc

get_ipython().run_line_magic("matplotlib", "inline")

f_path = r"C:\Windows\Fonts\NanumGothic.ttf" #tab
font_manager.FontProperties(fname=f_path).get_name()

rc("font", family="NanumGothic")

휘발유를 boxplot으로 그리기(seaborn 이용)

plt.figure(figsize=(12, 8))
sns.boxplot(x='is_self', y='gasoline', data=stations, palette='Set1')
plt.grid(True)
plt.show()

휘발유 브랜드 별 비교

plt.figure(figsize=(12, 8))
sns.boxplot(x='brand', y='gasoline', hue='is_self', data=stations, palette='Set1')
plt.grid(True)
plt.show()

휘발유 구별 비교

plt.figure(figsize=(18, 8))
sns.boxplot(x='gu', y='gasoline', data=stations, palette='Set1')
plt.grid(True)
plt.show()

4. 지도 시각화

import json
import folium

가장 비싼 주유소 10개

stations[['gu', 'name', 'is_self', 'gasoline']].sort_values(by='gasoline', ascending=False).head(10).reset_index(drop=True) #내림차순

가장 싼 주유소 10개

stations[['gu', 'name', 'is_self', 'gasoline']].sort_values(by='gasoline', ascending=True).head(10).reset_index(drop=True) #내림차순