音楽チャートをスクレイピングした（ほぼ日記）

ただコードをおいておくためだけに。

import pandas as pd
import time
from selenium import webdriver
from datetime import date
from datetime import timedelta

driver = webdriver.Chrome("****/chromedriver")
driver.implicitly_wait(30)

url = "http://www.billboard-japan.com/charts/detail?a=sales&"
data = pd.DataFrame(columns = ["year", "title", "artist", "num"])
for i in range(60):
  d = date(2020, 10, 5) - timedelta(days = 28*i)
  day = d.strftime("year=%Y&month=%m&day=") + str(int(d.strftime("%d")))
  driver.get(url + day)
  title = driver.find_elements_by_class_name("musuc_title")
  artist = driver.find_elements_by_class_name("artist_name")
  num = driver.find_elements_by_class_name("num")
  if num == []:
      break
  for j in range(50):
    r = len(data.index)
    data.at[r, "year"] = d.strftime("%Y/%m/%d")
    data.at[r, "title"] = title[j].text
    data.at[r, "artist"] = artist[j].text
    data.at[r, "num"] = num[j].text
  print(">>> " + d.strftime("%Y/%m/%d") + " 読み込み完了")
  time.sleep(1)
driver.quit()

data.to_csv("****/data.csv")