[Python] 純文本查看 復制代碼# -*- coding: UTF-8 -*-

from docx import Document #安裝 pip install python-docx

import re

from selenium import webdriver

from selenium.common.exceptions import NoSuchElementException

import time

# 凡是出現(xiàn)sleep的，都是因為網(wǎng)絡等原因加載過慢，需要等一等

# 登錄云盤

def login(driver,username,password):

orgin_url = 'https://pan.baidu.com/'

driver.get(orgin_url)

try:

time.sleep(5)

elem_static = driver.find_element_by_id("TANGRAM__PSP_4__footerULoginBtn")

elem_static.click()

time.sleep(0.5)

elem_username = driver.find_element_by_id("TANGRAM__PSP_4__userName")

elem_username.clear()

elem_username.send_keys(username)

elem_userpas = driver.find_element_by_id("TANGRAM__PSP_4__password")

elem_userpas.clear()

elem_userpas.send_keys(password)

elem_submit = driver.find_element_by_id("TANGRAM__PSP_4__submit")

elem_submit.click()

time.sleep(5)

except NoSuchElementException:

pass

# 將加密分享的文件保存到自己云盤的目錄下

def extract(driver,srcurl,srcpwd):

driver.get(srcurl)

try:

getpwd = driver.find_element_by_id("mwxxPOD") #("esDEV5")

getpwd.send_keys(srcpwd)

getButton = driver.find_element_by_link_text("提取文件")

getButton.click()

time.sleep(10)

# 目前有兩種情況

# 一：分享文件是一壓縮包

# 二：分享的是一路徑

try:# 全選（情況二）

selectall = driver.find_element_by_class_name("zbyDdwb")

selectall.click()

except NoSuchElementException:

file_name = "no_zbyDdwb.png"

driver.save_screenshot(file_name)

driver.get_screenshot_as_file(file_name)

pass

savetodisk = driver.find_element_by_link_text("保存到網(wǎng)盤")

savetodisk.click()

time.sleep(5)

# AA 為指定的自己網(wǎng)盤保存路徑

selectdir = driver.find_element_by_xpath("//span[@node-path='/AA']")

selectdir.click()

enter = driver.find_element_by_link_text("確定")

enter.click()

time.sleep(2)

except NoSuchElementException:

file_name = "no_such_element.png"

driver.get_screenshot_as_file(file_name)

pass

def read_docx(path):

fulltext=[]

list1=[]

listUrl=[]

listpwd=[]

#生成word對象，并指向word文件

doc=Document(path)

#讀取對象的全部內(nèi)容，不包括圖片和表格

paras=doc.paragraphs

#將每行數(shù)據(jù)存入列表

for p in paras:

fulltext.append(p.text)

for i, element in enumerate(fulltext):

if element.find("https:")!=-1:

list1.append(i)

for j in range(len(list1)):

k =list1[j]

link = re.findall(r'.*(https.*)',fulltext[k].replace(u"\xa0",u"").replace(" ",""))

listUrl.append(link[0])

pwd = re.findall('[a-zA-Z0-9]{4}',fulltext[k+1])

listpwd.append(pwd[0])

#print(len(listUrl),len(listpwd))

return listUrl,listpwd

# 調(diào)用執(zhí)行

def doWork():

# 存放鏈接的文件

path=r'文件地址***.docx'

listUrl,listpwd= read_docx(path)

driver = webdriver.Chrome(executable_path='D:\Program10\python37\chromedriver\chromedriver')

login(driver,"網(wǎng)盤帳戶名***","密碼***")

for index in range(len(listUrl)):

srcurl = listUrl[index]

srcpwd = listpwd[index]

extract(driver,srcurl,srcpwd)

driver.quit()

if __name__ == '__main__':

doWork()

總結(jié)

以上是生活随笔為你收集整理的python资源百度云_Python Selenium 百度云分享链接资源 批量保存的全部內(nèi)容，希望文章能夠幫你解決所遇到的問題。

如果覺得生活随笔網(wǎng)站內(nèi)容還不錯，歡迎將生活随笔推薦給好友。