FN가이드_펀드가이드_기준가격 크롤링

from selenium import webdriver

import chromedriver_autoinstaller

from selenium.webdriver.common.keys import Keys

from selenium.webdriver.common.by import By

import time

import csv

from selenium.common.exceptions import NoSuchElementException

from selenium.webdriver.chrome.options import Options

from pykrx import stock

import pandas as pd

from pandas_datareader import data

from datetime import datetime

from datetime import timedelta

list_result = []

# 크롬브라우저 실행

driver = webdriver.Chrome('')

driver.get('https://www.fundguide.net/Fund/SimpleSearch')

time.sleep(1)

# 검색어 입력

a = driver.find_element(By.CSS_SELECTOR,'#search_keyword')

a.send_keys('웰컴공모주코스닥벤처기업증권투자신탁[주식혼합-파생형] 종류 A')

time.sleep(1)

# 검색 버튼 클릭

b = driver.find_element(By.CSS_SELECTOR,'#btnSearch')

driver.execute_script('arguments[0].click();', b)

time.sleep(1)

# 펀드코드로 펀드선택 클릭 (data-fund_cd)

c = driver.find_element(By.CSS_SELECTOR,'a[data-fund_cd="K55244C54535"]')

driver.execute_script('arguments[0].click();', c)

time.sleep(1)

# 성과 버튼 클릭

d = driver.find_element(By.CSS_SELECTOR,'#tabFundMenu > ul > li:nth-child(2) > button')

driver.execute_script('arguments[0].click();', d)

time.sleep(1)

# 날짜 및 기준가 수집 (현재 기준 1년 = 25page)

for i in range(25):

   

    for row in range(1, 11):

        # 날짜

        date = driver.find_elements(By.CSS_SELECTOR,f'#tbl_rate > tbody > tr:nth-child({row}) > td.taC')

        var1 = date[0].text if date else ''

        # 기준가

        price = driver.find_elements(By.CSS_SELECTOR,f'#tbl_rate > tbody > tr:nth-child({row}) > td:nth-child(2)')

        var2 = price[0].text if price else ''

       

        list_result.append([var1, var2])

       

    # 다음 페이지 버튼 클릭

    e = driver.find_element(By.CSS_SELECTOR,'#tabPaging_info > div > tr > button.btn--paging.btn--next')

    driver.execute_script('arguments[0].click();', e)

   

    time.sleep(1)

# 코로나19 버튼, 3년 버튼 클릭

f = driver.find_element(By.CSS_SELECTOR,'#btnRatetime > button:nth-child(2) > span')

driver.execute_script('arguments[0].click();', f)

g = driver.find_element(By.CSS_SELECTOR,'#radio_rate > span:nth-child(5) > label > span')

driver.execute_script('arguments[0].click();', g)

time.sleep(1)

# 날짜 및 기준가 수집 (코로나 기준 3년 = 124page)

for i in range(124):

   

    for row in range(1, 11):

        # 날짜

        date = driver.find_elements(By.CSS_SELECTOR,f'#tbl_rate > tbody > tr:nth-child({row}) > td.taC')

        var1 = date[0].text if date else ''

        # 기준가

        price = driver.find_elements(By.CSS_SELECTOR,f'#tbl_rate > tbody > tr:nth-child({row}) > td:nth-child(2)')

        var2 = price[0].text if price else ''

       

        list_result.append([var1, var2])

       

    # 다음 페이지 버튼 클릭

    e = driver.find_element(By.CSS_SELECTOR,'#tabPaging_info > div > tr > button.btn--paging.btn--next')

    driver.execute_script('arguments[0].click();', e)

   

    time.sleep(1)

# 리스트를 데이터프레임으로 변환

df = pd.DataFrame(list_result, columns=['date', 'price'])

df.to_excel('output.xlsx', index=False)

print('done')