|
| 1 | +""" |
| 2 | + |
| 3 | +@file : 012-bs+selenium爬数据.py |
| 4 | + |
| 5 | +@author : xiaolu |
| 6 | + |
| 7 | +@time : 2019年10月23日 |
| 8 | + |
| 9 | +""" |
| 10 | +from selenium import webdriver |
| 11 | +import time |
| 12 | +import random |
| 13 | +import pandas as pd |
| 14 | +import requests |
| 15 | +from bs4 import BeautifulSoup |
| 16 | +import re |
| 17 | + |
| 18 | + |
| 19 | +def save_data(content, i): |
| 20 | + with open('{}corpus.data'.format(i), 'w', errors='ignore') as f: |
| 21 | + f.write(content) |
| 22 | + |
| 23 | + |
| 24 | +def spider(driver, lists): |
| 25 | + # 1. 刷新首页 |
| 26 | + i = 0 |
| 27 | + for list in lists: |
| 28 | + driver.get(list) |
| 29 | + # 随机停几秒 |
| 30 | + i += 1 |
| 31 | + time.sleep(random.randint(5, 10)) |
| 32 | + # /html/body/div[7]/div[1]/div[1]/h1 |
| 33 | + # data = driver.find_element_by_xpath('/html/body/div[7]/div[1]/div[1]/h1/text()')[0] |
| 34 | + data = driver.find_element_by_xpath('/html/body/div[7]/div[1]/div[1]/div[2]').text |
| 35 | + save_data(data, i) |
| 36 | + |
| 37 | + |
| 38 | +def crawl_link(): |
| 39 | + ''' |
| 40 | + 爬去链接 |
| 41 | + :return: |
| 42 | + ''' |
| 43 | + headers = {'User-Agent': 'Mozilla/5.0 (Windows NT 10.0; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/72.0.3626.121 Safari/537.36'} |
| 44 | + url = 'https://www.duanwenxue.com/sanwen/youmei/list_8.html' |
| 45 | + response = requests.get(url, headers) |
| 46 | + html = response.text |
| 47 | + |
| 48 | + soup = BeautifulSoup(html, features='lxml') |
| 49 | + |
| 50 | + # <a target="_blank" href="/article/4881118.html">小城记忆</a> |
| 51 | + links = soup.find_all('a', {'href': re.compile('/article/.*')}) |
| 52 | + total_links = ['https://www.duanwenxue.com' + _['href'] for _ in links] |
| 53 | + print(total_links) |
| 54 | + |
| 55 | + return total_links |
| 56 | + |
| 57 | + |
| 58 | +if __name__ == '__main__': |
| 59 | + # 实例化webdriver |
| 60 | + lists = crawl_link() |
| 61 | + driver = webdriver.Chrome(r'D:\learn-install\python3.6\chromedriver.exe') |
| 62 | + spider(driver, lists) |
0 commit comments