Playwright: 比 Puppeteer 更好用的浏览器自动化工具

pip install playwright==1.8.0a1  # 很奇怪，必须指定版本，不指定会安装到一个古老的版本
python -m playwright install  # 安装浏览器，此处国内网络可能会有问题（你懂的），请自行解决

from playwright.sync_api import sync_playwright as playwright

with playwright() as pw:
    webkit = pw.webkit.launch(headless=False)
    context = webkit.new_context()  # 需要创建一个 context
    page = context.new_page()  # 创建一个新的页面
    page.goto("https://www.apple.com")
    print(page.content())
    webkit.close()

context = browser.new_context(
  user_agent='My user agent'
)
context = browser.new_context(
  viewport={ 'width': 1280, 'height': 1024 }
)
context = browser.new_context(
    http_credentials={"username": "bill", "password": "pa55w0rd"}
)

# new_context 其他几个比较有用的选项：
ignore_https_errors=False
proxy={"server": "http://example.com:3128", "bypass": ".example.com", "username": "", "password": ""}
extra_http_headers={"X-Header": ""}

# hook 新建页面中的 Math.random 函数，总是返回 42
context.add_init_script(script="Math.random = () => 42;")
# 或者写在一个文件里
context.add_init_script(path="preload.js")

page.goto(url, referer="", timeout=30, wait_until="domcontentloaded|load|networkidle")

page.wait_for_event("event", event_predict, timeout)
page.wait_for_function(js_function)
page.wait_for_load_state(state="domcontentloaded|load|networkidle", timeout)
page.wait_for_selector(selector, timeout)
page.wait_for_timeout(timeout)  # 不推荐使用

# selector 指的是 CSS 等表达式
page.click(selector)
page.fill(selector, value)  # 在 input 中填充值

# 例子
page.click("#search")

page.url  # url
page.title()  # title
page.content()  # 获取页面全文
page.inner_text(selector)  # element.inner_text()
page.inner_html(selector)
page.text_content(selector)
page.get_attribute(selector, attr)

# eval_on_selector 用于获取 DOM 中的值
page.eval_on_selector(selector, js_expression)
# 比如：
search_value = page.eval_on_selector("#search", "el => el.value")

# evaluate 用于获取页面中 JS 中的数据，比如说可以读取 window 中的值
result = page.evaluate("([x, y]) => Promise.resolve(x * y)", [7, 8])
print(result) # prints "56"

# 通过文本选择元素，这是 Playwright 自定义的一种表达式
page.click("text=login")

# 直接通过 id 选择
page.click("id=login")

# 通过 CSS 选择元素
page.click("#search")
# 除了常用的 CSS 表达式外，Playwright 还支持了几个新的伪类
# :has 表示包含某个元素的元素
page.click("article:has(div.prome)")
# :is 用来对自身做断言
page.click("button:is(:text('sign in'), :text('log in'))")
# :text 表示包含某个文本的元素
page.click("button:text('Sign in')")  # 包含
page.click("button:text-is('Sign is')")  # 严格匹配
page.click("button:text-matches('\w+')")  # 正则
# 还可以根据方位匹配
page.click("button:right-of(#search)")  # 右边
page.click("button:left-of(#search)")  # 左边
page.click("button:above(#search)")  # 上边
page.click("button:below(#search)")  # 下边
page.click("button:near(#search)")  # 50px 之内的元素

# 通过 XPath 选择
page.click("//button[@id='search'])")
# 所有 // 或者 .. 开头的表达式都会默认为 XPath 表达式

page.click('css=nav >> text=Login')

# 保存状态
import json
storage = context.storage_state()
with open("state.json", "w") as f:
    f.write(json.dumps(storage))

# 加载状态
with open("state.json") as f:
    storage_state = json.loads(f.read())
context = browser.new_context(storage_state=storage_state)

def log_request(intercepted_request):
    print("a request was made:", intercepted_request.url)
page.on("request", log_request)
# sometime later...
page.remove_listener("request", log_request)

from playwright.sync_api import sync_playwright as playwright

def run(pw):
    browser = pw.webkit.launch()
    page = browser.new_page()
    # Subscribe to "request" and "response" events.
    page.on("request", lambda request: print(">>", request.method, request.url))
    page.on("response", lambda response: print("<<", response.status, response.url))
    page.goto("https://example.com")
    browser.close()

with playwright() as pw:
    run(pw)

context = browser.new_context()
page = context.new_page()
# route 的参数默认是通配符，也可以传递编译好的正则表达式对象
context.route("**/*.{png,jpg,jpeg}", lambda route: route.abort())
context.route(re.compile(r"(\.png$)|(\.jpg$)"), lambda route: route.abort())
page.goto("https://example.com")
browser.close()

context = browser.new_context(
    proxy={"server": "http://example.com:3128", "bypass": ".example.com", "username": "", "password": ""}
)

python -m playwright codegen http://example.com/