codegay

python3 + selenium + chrome浏览器

# -*- coding: utf-8 -*-
"""
Created on Sun Dec  3 00:40:35 2017
@author: codegay
python3 & 编辑器 == spyder
"""

from time import sleep
from selenium import webdriver
# appdata = environ["LOCALAPPDATA"] + r"\Google\Chrome\User Data"
appdata = "c:/chrome-selenium"
options = webdriver.ChromeOptions()

options.add_argument("user-data-dir=" + appdata)
# options.add_argument("headless")
options.add_argument("disable-gpu")

ddurl = "http://dd.ma/mE0NUHSi"

chrome = webdriver.Chrome(chrome_options=options)
chrome.get(ddurl)
sleep(10)
try:
    geturl = chrome.find_element_by_link_text("点击这里进入")
except:
    print("异常错误或者找不到元素")
    exit()

url = geturl.get_property("href")

print("url:", url)

chrome.get(url)
复制代码

codegay

优化了一下。通过id和tag定位，这样不用等待JS的那个倒计时。

# -*- coding: utf-8 -*-
"""
Created on Sun Dec  3 00:40:35 2017
@author: codegay
python3 & 编辑器 == spyder
"""

from time import sleep
from selenium import webdriver
# appdata = environ["LOCALAPPDATA"] + r"\Google\Chrome\User Data"
appdata = "c:/chrome-selenium"
options = webdriver.ChromeOptions()

options.add_argument("user-data-dir=" + appdata)
# options.add_argument("headless")
# options.add_argument("disable-gpu")

ddurl = "http://dd.ma/mE0NUHSi"

chrome = webdriver.Chrome(chrome_options=options)
chrome.get(ddurl)
sleep(1)
try:
    geturl = chrome.find_element_by_id("btn_open")
    geturl = geturl.find_element_by_tag_name("a")
except Exception as e:
    print("异常错误或者找不到元素:")
    print(e)
    exit()

url = geturl.get_property("href")

print("url:", url)

chrome.get(url)
复制代码

codegay

nim语言写的

import httpclient
import re
import os

proc getlink*(url:string):string {.discardable.} =
    var client = newHttpClient()
    var link: array[1,string]
    var ddmaContent:string
    ddmaContent = client.getContent(url)
    client.close()

    if find(ddmaContent,re"""link1\"\s+href=\"(.+)\"""",link,1) != -1 :
        echo "link:",link[0]
        result = link[0]
    else:
        result = ""

proc startlink*(url:string):int {.discardable.} =
    result = execShellCmd("start " & url)

startlink(getlink("http://dd.ma/mE0NUHSi"))
复制代码