def test_sync_playwright_multiple_times():
with sync_playwright() as pw1:
assert pw1.chromium
with pytest.raises(Error) as exc:
with sync_playwright() as pw2:
assert pw1.chromium == pw2.chromium
assert "Can only run one Playwright at a time." in exc.value.message
def run(self):
with sync_playwright() as playwright:
browser = getattr(playwright, browser_name).launch()
# This should not throw ^^.
browser.new_page()
browser.close()
result.append("Success")
def run_with_playwright(rh, sheet):
print("Running RPA Challenge with Playwright")
with sync_playwright() as p:
browser = p.chromium.launch(headless=True)
page = browser.newPage()
page.goto("http://www.rpachallenge.com")
page.click('text="Start"')
for row in sheet:
myeval(
page, '//input[@ng-reflect-name="labelFirstName"]', row["First Name"]
)
myeval(page, '//input[@ng-reflect-name="labelLastName"]', row["Last Name"])
myeval(
page,
'//input[@ng-reflect-name="labelCompanyName"]',
row["Company Name"],
)
myeval(
page, '//input[@ng-reflect-name="labelRole"]', row["Role in Company"]
)
myeval(page, '//input[@ng-reflect-name="labelAddress"]', row["Address"])
myeval(page, '//input[@ng-reflect-name="labelEmail"]', row["Email"])
myeval(
page, '//input[@ng-reflect-name="labelPhone"]', str(row["Phone Number"])
)
page.evaluate("""document.querySelector('input[type="submit"]').click();""")
result = page.evaluate(
"""() => {return document.querySelector('.message2').textContent;}"""
)
rh.check_for_new_record(page, result)
browser.close()
def get_wicksly_info(tracking_number):
secrets = Secrets()
USER_NAME = secrets.get_secret("credentials")["PLATFORM_USERNAME"]
PASSWORD = secrets.get_secret("credentials")["PLATFORM_PASS"]
with sync_playwright() as playwright:
browser = playwright.chromium.launch(headless=False)
wicksly_context = browser.newContext()
w_page = wicksly_context.newPage()
w_page.goto(os.environ.get('PLATFORM_URL'))
w_page.fill('#id_username', USER_NAME)
w_page.fill('#id_password', PASSWORD)
with w_page.expect_navigation():
w_page.click('input[type=submit]')
w_page.goto(os.environ.get('PLATFORM_URL'))
w_page.fill('#searchbar', tracking_number)
with w_page.expect_navigation():
w_page.click('text=Search')
w_page.querySelectorAll('table#result_list tbody tr a')[0].click()
street = w_page.getAttribute('#id_street1', 'value')
city = w_page.getAttribute('#id_city', 'value')
state = w_page.getAttribute('#id_state', 'value')
postal_code = w_page.getAttribute('#id_postal_code', 'value')
return {
'street': street,
'city': city,
'state': state,
'postal_code': postal_code
}
def process_tang(tang_values):
if not tang_values:
return
with sync_playwright() as playwright:
browser_type = playwright.chromium
executablePath = r'ms-playwright\chromium-833159\chrome-win\chrome.exe'
browser = browser_type.launch(headless=False,
executablePath=executablePath)
context = browser.newContext()
# Open new page
page = context.newPage()
page.setViewportSize(1920, 1040)
# Go to http://tang.csair.com/WebFace/Tang.WebFace.Cargo/AgentAwbBrower.aspx?menuID=1
page.goto(
"http://tang.csair.com/WebFace/Tang.WebFace.Cargo/AgentAwbBrower.aspx?menuID=1"
)
for waybills_number in tang_values:
prefix = waybills_number.split('-')[0]
suffix = waybills_number.split('-')[-1]
# Click input[name="ctl00$ContentPlaceHolder1$txtPrefix"]
page.click("input[name=\"ctl00$ContentPlaceHolder1$txtPrefix\"]")
# Fill input[name="ctl00$ContentPlaceHolder1$txtPrefix"]
page.press("input[name=\"ctl00$ContentPlaceHolder1$txtPrefix\"]",
'Control+A')
page.press("input[name=\"ctl00$ContentPlaceHolder1$txtPrefix\"]",
'Backspace')
page.fill("input[name=\"ctl00$ContentPlaceHolder1$txtPrefix\"]",
prefix)
# Click input[name="ctl00$ContentPlaceHolder1$txtNo"]
page.click("input[name=\"ctl00$ContentPlaceHolder1$txtNo\"]")
# Fill input[name="ctl00$ContentPlaceHolder1$txtNo"]
page.press("input[name=\"ctl00$ContentPlaceHolder1$txtNo\"]",
'Control+A')
page.press("input[name=\"ctl00$ContentPlaceHolder1$txtNo\"]",
'Backspace')
page.fill("input[name=\"ctl00$ContentPlaceHolder1$txtNo\"]",
suffix)
# Press Enter
page.click("//input[@id='btnSearch']")
# assert page.url == "http://tang.csair.com/WebFace/Tang.WebFace.Cargo/AgentAwbBrower.aspx?menuID=1"
page.waitForLoadState(state="domcontentloaded")
time.sleep(10)
page.screenshot(path=f'{hx_path}/%s.png' % waybills_number)
# Close page
page.close()
# ---------------------
context.close()
browser.close()
def page():
with sync_playwright() as playwright:
browser = playwright.chromium.launch()
context = browser.newContext()
page = context.newPage()
yield page
page.close()
context.close()
browser.close()
def get_playwright():
global playwright
if playwright is None:
try:
playwright = sync_playwright().start()
except Exception as e:
raise e
return playwright
def page():
with sync_playwright() as play:
if os.getenv('DOCKER_RUN') or os.getenv('GITHUB_RUN'):
browser = play.chromium.launch(headless=True,
args=['--no-sandbox'])
else:
browser = play.chromium.launch(headless=False)
page = browser.newPage()
global PAGE
PAGE = page
yield page
browser.close()
class play_wright(unittest.TestCase):
playwright = sync_playwright().start()
browser = playwright.chromium.launch(headless=False)
page = browser.newContext().newPage()
def test_a(self):
self.page.goto("https://www.baidu.com")
self.page.type("input[id=kw]", "Playwright GitHub")
self.page.waitForTimeout(1000)
self.page.click("input[id=su]")
self.page.waitForTimeout(3000)
text = self.page.querySelector('a[name="tj_login"]').textContent()
print(text)
self.page.waitForTimeout(3000)
self.page.goBack()
page2 = self.browser.newContext().newPage()
page2.goto('https://google.com')
self.page.bringToFront()
self.page.waitForTimeout(3000)
def test_c(self):
self.page.setViewportSize(width=1920, height=1080)
self.page.fill("input[id=kw]", '')
self.page.type("input[id=kw]", "ps5")
self.page.waitForTimeout(1000)
self.page.click("input[id=su]")
self.page.waitForTimeout(3000)
self.page.keyboard.down(key="ArrowDown")
if self.page.waitForSelector("//a[contains(text(),'下一页 >')]", 5000,
'visible'):
self.page.waitForTimeout(3000)
self.page.click("//a[contains(text(),'下一页 >')]")
else:
self.page.keyboard.down(key="ArrowDown")
# self.page.querySelector('a[name="tj_login"]').waitForElementState('visible',5000)
self.page.querySelector('a[name="tj_login"]').textContent()
self.page.querySelector('a[name="tj_login"]')
# self.page.waitForTimeout(3000)
self.page.querySelector(
"//a[contains(text(),'下一页 >')]").scrollIntoViewIfNeeded()
# self.page.waitForTimeout(2000)
self.page.querySelector(
"//a[contains(text(),'下一页 >')]").scrollIntoViewIfNeeded()
self.page.waitForTimeout(3000)
self.page.context.on()
# if self.page.waitForSelector('a[name="tj_login"]', 5000,'visible'):
# self.page.click('a[name="tj_login"]')
# self.page.waitForTimeout(3000)
self.page.close()
def test_d(self):
self.page.setViewportSize(width=1920, height=1080)
def browser():
pw_context = sync_playwright()
pw = pw_context.__enter__()
browser = getattr(pw, "chromium").connect(
wsEndpoint='wss://chrome.headlesstesting.com?token=[YOUR-TOKEN]')
browser._close = browser.close
def _handle_close() -> None:
browser._close()
pw_context.__exit__(None, None, None)
browser.close = _handle_close
return browser
def browser():
with sync_playwright() as p:
try:
video_location = f"{os.path.dirname(os.path.abspath(__file__))}"
new_browser = p.chromium.launch(headless=False)
page = new_browser.newPage(
viewport={"width": 1680, "height": 1080},
videosPath=f"{video_location}"
)
yield page
new_browser.close()
p.stop()
except Exception as e:
print(e)
pytest.skip("Exception")
def launch(**kwargs: Dict[Any, Any]) -> Browser:
headful_option = pytestconfig.getoption("--headful")
launch_options = {**browser_type_launch_args, **kwargs}
if headful_option:
launch_options["headless"] = False
pw = sync_playwright().start()
browser = getattr(pw, browser_name).launch(**launch_options)
browser._close = browser.close
def _handle_close() -> None:
browser._close()
pw.stop()
browser.close = _handle_close
return browser
def main():
"""Set up playwright and print ballot details."""
with sync_playwright() as playwright:
browser = playwright.chromium.launch()
context = browser.newContext()
page = context.newPage()
details = get_details(page, INPUTS)
page.close()
context.close()
browser.close()
for field, value in details._asdict().items():
print(f"{field:10} {value}")
def launch(**kwargs: Dict[Any, Any]) -> Browser:
headful_option = pytestconfig.getoption("--headful")
launch_options = {**launch_arguments, **kwargs}
if headful_option:
launch_options["headless"] = False
pw_context = sync_playwright()
pw = pw_context.__enter__()
browser = getattr(pw, browser_name).launch(**launch_options)
browser._close = browser.close
def _handle_close() -> None:
browser._close()
pw_context.__exit__(None, None, None)
browser.close = _handle_close
return browser
def get_recent_call_history(phone_number, password, national_id):
with sync_playwright() as p:
browser = p.chromium.launch(headless=False)
page = browser.newPage()
# login
page.goto('https://myais.ais.co.th/login')
page.type("input[name='a mobile input']",phone_number)
page.click("button[type=submit]")
page.click("button[id=btnLoginFormPassword]")
page.type("input[id='txtLoginFormPassword']", password)
page.click("button[id=btnLoginFormSubmitLogin2]")
# access page
page.waitForSelector("myais-sidebar-menu:nth-child(4) .menu-content")
page.goto("https://myais.ais.co.th/balance/usage")
#fill national id
page.type("input[id='input-idcard']", national_id)
# XHR **/invoice
url = '**/invoices'
with page.expect_response(url):
page.click("button[type=submit]")
response = page.waitForResponse(url)
invoices = response.json()
logger.debug(invoices)
# page.on('request', print)
# in case you want other bill hisory
# .check-box .check , .check-box img
# .check-box img = [bill1]
# .billing-row-detail:nth-child(3) .check [bill2]
# .billing-row-detail:nth-child(4) .check [bill3]
# XHR call history(**/filter)
url = "**/usages/billed/group/VOICE/chargedetail/filter"
with page.expect_response(url):
page.click("#btnApproveUsageDetailSeaching")
response = page.waitForResponse(url)
call_usage = response.json()
logger.debug(call_usage)
browser.close()
return call_usage
def extract_leads(search_url):
with sync_playwright() as p:
generated_data = []
browser = p.webkit.launch(headless=False)
context = browser.newContext()
page = context.newPage()
login(page)
page.goto(search_url)
page.click('.search-results__pagination-next-button')
page.waitForResponse('**people**')
page.click('.search-results__pagination-previous-button')
first_response = page.waitForResponse('**people**')
json_response = first_response.json()
generated_data = generated_data + process_json(json_response)
while has_next_button(page):
generated_data = crawl_page(page, generated_data)
output_csv(generated_data)
def process_cargo(cargo_values):
if not cargo_values:
return
with sync_playwright() as playwright:
browser_type = playwright.chromium
executablePath = r'ms-playwright\chromium-833159\chrome-win\chrome.exe'
browser = browser_type.launch(headless=False,
executablePath=executablePath)
context = browser.newContext()
# Open new page
page = context.newPage()
page.setViewportSize(1920, 1040)
# Go to https://cargo.china-airlines.com/CCNetv2/content/manage/ShipmentTracking.aspx
page.goto(
"https://cargo.china-airlines.com/CCNetv2/content/manage/ShipmentTracking.aspx"
)
for waybills_number in cargo_values:
suffix = waybills_number.split('-')[-1]
# Click input[name="ctl00$ContentPlaceHolder1$txtAwbNum"]
page.click("input[name=\"ctl00$ContentPlaceHolder1$txtAwbNum\"]")
# Fill input[name="ctl00$ContentPlaceHolder1$txtAwbNum"]
page.press("input[name=\"ctl00$ContentPlaceHolder1$txtAwbNum\"]",
'Control+A')
page.press("input[name=\"ctl00$ContentPlaceHolder1$txtAwbNum\"]",
'Backspace')
page.fill("input[name=\"ctl00$ContentPlaceHolder1$txtAwbNum\"]",
suffix)
# Click input[name="ctl00$ContentPlaceHolder1$btnSearch"]
page.click("input[name=\"ctl00$ContentPlaceHolder1$btnSearch\"]")
# assert page.url == "https://cargo.china-airlines.com/CCNetv2/content/manage/ShipmentTracking.aspx"
page.waitForLoadState(state="networkidle")
page.screenshot(path=f'{hx_path}/%s.png' % waybills_number)
# Close page
page.close()
# ---------------------
context.close()
browser.close()
def get_talks_schedule():
charlas = []
with sync_playwright() as p:
browser = p.chromium.launch()
page = browser.newPage()
# Visit Schedule Page
page.goto('https://eventos.python.org.ar/events/pyconar2020/schedule')
# Get all divs with the schedule of the day
days = page.querySelectorAll('css=div .well')
for day in days:
# Get the day header text
dia = day.querySelector('h2').innerText().strip()
print(dia)
dia = dia.split('- ')[1] # Día 1 - 16/11/2020 --> 16/11/2020
# Get all talks from that day.
charlas_divs = day.querySelectorAll('css=div .fc-content')
for charla in charlas_divs:
"""
Expected html
<div class="fc-time" data-start="19:00" data-full="7:00 PM - 8:00 PM">
<span>19:00 - 20:00</span>
</div>
<div class="fc-title">¿Por que no se hacen mas tests?</div>
"""
# For each talk, get its schedule and its title
horario = charla.querySelector('.fc-time').innerText()
start, end = horario.split(' - ')
start_date, end_date = parse_dates(dia, start, end)
titulo = charla.querySelector('.fc-title').innerText()
print('Charla ', titulo)
charlas.append({
'day': str(dia),
'titulo': titulo,
'start': start_date.isoformat(),
'end': end_date.isoformat(),
})
browser.close()
return charlas
def cookie_loggin(self):
'''自动登录操作'''
self.judge_cookie() # 首先判断cookie是否已获取，是否过期
print("自动登录开始...")
# 加启动配置
playwright = sync_playwright().start()
#无头浏览器模式
browser = playwright.chromium.launch()
#打开浏览器模式
#browser = playwright.chromium.launch(headless=False,slowMo=200)
context = browser.newContext()
#设置防爬的参数
context.addInitScript(source="""
const newProto = navigator.__proto__;
delete newProto.webdriver;
navigator.__proto__ = newProto;
""")
#设置超时时间为30s
context.setDefaultTimeout(30000)
page = context.newPage()
#打开权力运行系统登陆网址
page.goto(self.log_url, waitUntil="load")
context.clear_cookies() # 清除旧cookies
# with可以上下文管理上文进行设置部署，下文进行处理，然后把处理的结果赋值给变量（cookie）
with open(os.path.join(self.cookie_path, self.cookie_name),
'r') as cookief:
#使用json读取cookies 注意读取的是文件 所以用load而不是loads
cookieslist = json.load(cookief)
# 方法1删除该字段
cookies_dict = dict()
for cookie in cookieslist:
#该字段有问题所以删除就可以,浏览器打开后记得刷新页面 有的网页注入cookie后仍需要刷新一下
if 'expiry' in cookie:
del cookie['expiry']
cookies_dict[cookie['name']] = cookie['value']
context.add_cookies(cookies_dict)
page.goto(self.home_url, waitUntil='load')
#等待10s查看网页是否cookies登录
time.sleep(10)
page.close()
context.close()
browser.close()
print("浏览器退出")
def playwright_start(login_url):
playwright = sync_playwright().start()
#无头浏览器模式
#browser = playwright.chromium.launch()
#打开浏览器模式
browser = playwright.chromium.launch(headless=False, slowMo=200)
context = browser.newContext()
#设置防爬的参数
context.addInitScript(source="""
const newProto = navigator.__proto__;
delete newProto.webdriver;
navigator.__proto__ = newProto;
""")
#设置超时时间为30s
context.setDefaultTimeout(30000)
page = context.newPage()
# page.on("request", lambda request: print("Request %s" % request.url))
# page.on("requestfinished", lambda request: print("Request finished %s" % request.url))
# page.on("response",lambda response: print("Response %s"% (response.text())))
page.goto(login_url, waitUntil="networkidle")
with open(r'.\date\1.html', 'w', encoding='utf-8') as f:
f.write(page.content())
# deptname_list=page.querySelectorAll('//div[@class="pub-people"]/ul[1]/div/li[@class="depItem"]/a')
# for deptname in deptname_list:
# dept_name=deptname.getAttribute("title")
# dept_url=deptname.getAttribute("href")
html = etree.HTML(page.content())
deptname_list = html.xpath(
'//div[@class="pub-people"]/ul[1]/div/li[@class="depItem"]')
for deptname in deptname_list:
dept_name = deptname.xpath('.//a/@title')[0]
dept_url = deptname.xpath('.//a/@href')[0]
content = {
"单位名称": dept_name,
"网站链接": dept_url,
}
print(content)
contents.append(content)
page.close()
context.close()
browser.close()
def get_cookie(self):
'''登录获取cookie'''
playwright = sync_playwright().start()
#无头浏览器模式
browser = playwright.chromium.launch()
#打开浏览器模式
#browser = playwright.chromium.launch(headless=False,slowMo=200)
context = browser.newContext()
#设置防爬的参数
context.addInitScript(source="""
const newProto = navigator.__proto__;
delete newProto.webdriver;
navigator.__proto__ = newProto;
""")
#设置超时时间为30s
context.setDefaultTimeout(30000)
page = context.newPage()
#打开权力运行系统登陆网址
page.goto(self.log_url, waitUntil="load")
page.fill("input[name=\"userNameHeader\"]", USENAME)
page.fill("input[name=\"password\"]", PASSWORD)
page.check("input[id=\"recordAccount\"]")
with page.expect_navigation():
page.click('//a[@class="btn-login"]')
print("登陆成功")
page.waitForSelector(
r'//div[@class="lw-navContent"]/ul/li/a[contains(text(), "权力事项管理")]'
)
page.click(
r'//div[@class="lw-navContent"]/ul/li/a[contains(text(), "权力事项管理")]'
)
print("进入权力事项管理")
with open(os.path.join(self.cookie_path, self.cookie_name),
'w') as cookief: # 创建文本覆盖保存cookie
# 将cookies保存为json格式
cookief.write(json.dumps(context.cookies()))
print("cookies保存完成")
page.close()
context.close()
browser.close()
from playwright import sync_playwright
with sync_playwright() as p:
for browser_type in [p.chromium, p.firefox, p.webkit]:
browser = browser_type.launch(headless=False)
page = browser.newPage()
page.goto('http://whatsmyuseragent.org/')
page.screenshot(path=f'screenshots/example-{browser_type.name}.png')
browser.close()
def run(excel_file):
with sync_playwright() as playwright:
browser_type = playwright.chromium
executablePath = r'ms-playwright\chromium-833159\chrome-win\chrome.exe'
browser = browser_type.launch(headless=False,
executablePath=executablePath)
context = browser.newContext()
# Open new page
page = context.newPage()
page.setViewportSize(1920, 1040)
logger.info(' - 开始登陆网站')
# Go to http://uniner.synology.me:88/LKT/index.php
page.goto("http://uniner.synology.me:88/LKT/index.php")
# Go to http://uniner.synology.me:88/LKT/index.php?module=Login
page.goto("http://uniner.synology.me:88/LKT/index.php?module=Login")
# Click input[name="login"]
page.click("input[name=\"login\"]")
# Fill input[name="login"]
page.fill("input[name=\"login\"]", "laiketui")
# Click input[name="pwd"]
page.click("input[name=\"pwd\"]")
# Fill input[name="pwd"]
page.fill("input[name=\"pwd\"]", "laiketui")
# Press Enter
# with page.expect_navigation(url="http://uniner.synology.me:88/LKT/index.php?module=AdminLogin"):
with page.expect_navigation():
page.press("input[name=\"pwd\"]", "Enter")
logger.info(' - 网站登录成功')
page.waitForLoadState('networkidle')
# Click //dt[normalize-space(.)='商品管理']
page.click("//dt[normalize-space(.)='商品管理']")
# Click text="商品列表"
page.click("text=\"商品列表\"")
frame_product = page.frames[2]
# result = input('是否删除之前商品信息 (y/n) :')
logger.info(' - 开始自动删除历史数据')
while True:
# if result != 'y':
# break
if frame_product.textContent(
'//div[@class="mt-20"]/div/table/tbody').strip():
frame_product.click("//label")
frame_product.click("text=/.*删除.*/")
frame_product.click("text=\"确认\"")
frame_product.click("text=\"确认\"")
else:
logger.info(' - 历史商品信息全部删除成功')
time.sleep(2)
break
logger.info(' - 10s后开始自动录入数据...')
time.sleep(10)
logger.info(' - 开始读取并处理Excel表格数据')
df = pd.read_excel(excel_file)
logger.info(' - 开始自动上传商品信息')
for i in df.index:
sku = df.loc[i, 'sku']
title = df.loc[i, '标题']
s_price = int(df.loc[i, '售价'])
m_price = int(df.loc[i, '市场价'])
img_path = df.loc[i, '图片路径']
# Click text=/.*发布商品.*/
frame_product.click("text=/.*发布商品.*/")
frame_product.waitForLoadState('networkidle')
# Fill input[name="product_title"]
frame_product.fill("input[name=\"product_title\"]", title) # 商品标题
# Fill input[name="subtitle"]
frame_product.fill("input[name=\"subtitle\"]", title) # 副标题
frame_product.selectOption('//*[@id="product_classId"]',
'-7-28-') # 商品类别
frame_product.selectOption('//*[@id="brand_classId"]', '9') # 商品品牌
frame_product.click('//*[@id="image"]')
frame_product.click("text=/.*本地上传.*/")
# Upload 003dd00372c3f98156512c62afe559cc.jpg
frame_product.setInputFiles("input[name=\"imgFile\"]",
img_path) # 商品主图
frame_product.click("text=/.*确定.*/")
# Upload 微信图片_20201208093221.jpg
frame_product.setInputFiles("input[name=\"imgurls[]\"]",
img_path) # 商品展示图
# Fill input[name="initial[cbj]"]
frame_product.fill("input[name=\"initial[cbj]\"]",
f'{int(s_price * 0.5)}') # 成本价
# Fill input[name="initial[yj]"]
frame_product.fill("input[name=\"initial[yj]\"]",
f'{m_price}') # 原价
# Fill input[name="initial[sj]"]
frame_product.fill("input[name=\"initial[sj]\"]",
f'{s_price}') # 售价
frame_product.selectOption('//*[@id="unit"]', '双') # 单位
# Fill input[name="initial[kucun]"]
frame_product.fill("input[name=\"initial[kucun]\"]",
f'{random.randint(1000, 10000)}') # 库存
# Fill input[placeholder="请输入属性名称"]
frame_product.fill("input[placeholder=\"请输入属性名称\"]",
f'{sku}') # sku
frame_product.click(
f'//label[@for="sex-{random.randint(1, 4)}"]') # 类型
frame_product.hover('//*[@id="form1"]//input[@name="Submit"]')
frame_product.fill('//input[@id="volumeId"]',
f'{random.randint(500, 5000)}') # 拟定销量
frame_product.selectOption('//select[@id="freightId"]',
f'{random.randint(1, 2)}') # 运费设置
frame_ueditor_0 = page.frames[4]
frame_ueditor_0.addScriptTag(
content='document.body.innerHTML="{}"'.format('无'))
page.once("dialog", lambda dialog: click_alert(dialog))
frame_product.click(
'//*[@id="form1"]//input[@name="Submit"]') # 提交
msg = '第%s款产品发布成功！' % (i + 1)
logger.info(' - %s' % msg)
else:
page.click("text=\"商品列表\"")
time.sleep(2)
logger.info(' - 商品信息全部上传完成')
time.sleep(2)
logger.info(' - 自动录入完成，60后自动关闭程序')
time.sleep(56)
logger.info(' - 程序结束')
# Close page
page.close()
# ---------------------
context.close()
browser.close()
def main():
with sync_playwright() as playwright:
run(playwright)
get_repositories(url)
def run(tracking_number, shipment_info):
ORIGIN_COMPANY = os.environ.get('ORIGIN_COMPANY')
ORIGIN_STREET = os.environ.get('ORIGIN_STREET')
ORIGIN_ZIP = os.environ.get('ORIGIN_ZIP')
ORIGIN_CITY = os.environ.get('ORIGIN_CITY')
ORIGIN_STATE = os.environ.get('ORIGIN_STATE')
ORIGIN_PHONE = os.environ.get('ORIGIN_PHONE')
CONTACT_EMAIL = os.environ.get('CONTACT_EMAIL')
CONTACT_FIRST = os.environ.get('CONTACT_FIRST')
CONTACT_LAST = os.environ.get('CONTACT_LAST')
with sync_playwright() as playwright:
browser = playwright.chromium.launch(headless=False)
context = browser.newContext()
page = context.newPage()
page.goto('https://www.usps.com/help/missing-mail.htm')
page.click("text=\"Start Your Form\"")
page.fill("input[type=\"text\"]", tracking_number)
with page.expect_navigation():
page.click("text=\"Track\"")
page.click("text=\"Start\"")
page.click("a[role=\"button\"]")
page.click(
"//a[normalize-space(.)='Package' and normalize-space(@title)='Package' and normalize-space(@role)='menuitemradio']"
)
page.click(
"//a[normalize-space(.)='--None--' and normalize-space(@role)='button']"
)
# Click //a[normalize-space(.)='First-Class Mail®' and normalize-space(@title)='First-Class Mail®' and normalize-space(@role)='menuitemradio']
page.click(
"//a[normalize-space(.)='First-Class Mail®' and normalize-space(@title)='First-Class Mail®' and normalize-space(@role)='menuitemradio']"
)
# Click //a[normalize-space(.)='--None--' and normalize-space(@role)='button']
page.click(
"//a[normalize-space(.)='--None--' and normalize-space(@role)='button']"
)
# Click //a[normalize-space(.)='Business' and normalize-space(@title)='Business' and normalize-space(@role)='menuitemradio']
page.click(
"//a[normalize-space(.)='Business' and normalize-space(@title)='Business' and normalize-space(@role)='menuitemradio']"
)
# Click //a[normalize-space(.)='--None--' and normalize-space(@role)='button']
page.click(
"//a[normalize-space(.)='--None--' and normalize-space(@role)='button']"
)
# Click (//a[normalize-space(.)='Residence' and normalize-space(@title)='Residence' and normalize-space(@role)='menuitemradio'])[2]
page.click(
"(//a[normalize-space(.)='Residence' and normalize-space(@title)='Residence' and normalize-space(@role)='menuitemradio'])[2]"
)
# Check //div[normalize-space(.)='None']/input[normalize-space(@type)='checkbox']
page.check(
"//div[normalize-space(.)='None']/input[normalize-space(@type)='checkbox']"
)
# Click //div[normalize-space(.)='Unknown' and normalize-space(@role)='option']
page.click(
"//div[normalize-space(.)='Unknown' and normalize-space(@role)='option']"
)
# Click //lightning-primitive-icon
page.click("//lightning-primitive-icon")
# Click input[type="text"]
page.click("input[type=\"text\"]")
# Fill input[type="text"]
page.fill("input[type=\"text\"]", "15")
# Click text="Next"
page.click("text=\"Next\"")
page.waitForSelector(
"//label[span[text()='Street Address']]/following-sibling::input")
page.fill(
"//label[span[text()='Street Address']]/following-sibling::input",
ORIGIN_STREET)
# Click //div[normalize-space(.)='City']/input[normalize-space(@type)='text']
page.click(
"//div[normalize-space(.)='City']/input[normalize-space(@type)='text']"
)
# Fill //div[normalize-space(.)='City']/input[normalize-space(@type)='text']
page.fill(
"//div[normalize-space(.)='City']/input[normalize-space(@type)='text']",
ORIGIN_CITY)
page.click(
"//div[normalize-space(@role)='list']/div[4]/div[1][normalize-space(@role)='listitem']/div/div/div/div/div[1]/div/div/a[normalize-space(.)='--None--' and normalize-space(@role)='button']"
)
page.click(
f"//a[normalize-space(.)='{ORIGIN_STATE}' and normalize-space(@title)='{ORIGIN_STATE}' and normalize-space(@role)='menuitemradio']"
)
page.click(
"//div[normalize-space(.)='ZIP Code™']/input[normalize-space(@type)='text']"
)
# Click //div[normalize-space(.)='ZIP Code™']/input[normalize-space(@type)='text']
page.click(
"//div[normalize-space(.)='ZIP Code™']/input[normalize-space(@type)='text']"
)
# Fill //div[normalize-space(.)='ZIP Code™']/input[normalize-space(@type)='text']
page.fill(
"//div[normalize-space(.)='ZIP Code™']/input[normalize-space(@type)='text']",
ORIGIN_ZIP)
page.waitForSelector(
"//div[normalize-space(.)='Street Address*']/input[normalize-space(@type)='text']"
)
# Click //div[normalize-space(.)='Street Address*']/input[normalize-space(@type)='text']
page.click(
"//div[normalize-space(.)='Street Address*']/input[normalize-space(@type)='text']"
)
# Fill //div[normalize-space(.)='Street Address*']/input[normalize-space(@type)='text']
page.fill(
"//div[normalize-space(.)='Street Address*']/input[normalize-space(@type)='text']",
shipment_info['street'])
# Click (//div[normalize-space(@role)='listitem']/div/div/div[normalize-space(.)='Apt/Suite/Other']/input[normalize-space(@type)='text'])[2]
page.click(
"(//div[normalize-space(@role)='listitem']/div/div/div[normalize-space(.)='Apt/Suite/Other']/input[normalize-space(@type)='text'])[2]"
)
page.click(
"//div[normalize-space(.)='City*']/input[normalize-space(@type)='text']"
)
page.fill(
"//div[normalize-space(.)='City*']/input[normalize-space(@type)='text']",
shipment_info['city'])
page.click(
"(//div[normalize-space(@role)='listitem']/div/div/div/div/div[1]/div/div/a[normalize-space(.)='--None--' and normalize-space(@role)='button'])[2]"
)
page.querySelectorAll(
f"//a[starts-with(@title, '{shipment_info['state']}') and normalize-space(@role)='menuitemradio']"
)[1].click()
page.click(
"//div[normalize-space(.)='ZIP Code™*']/input[normalize-space(@type)='text']"
)
page.fill(
"//div[normalize-space(.)='ZIP Code™*']/input[normalize-space(@type)='text']",
shipment_info['postal_code'])
page.click("text=\"Next\"")
page.waitForSelector(
"//div[normalize-space(.)='First Name*']/input[normalize-space(@type)='text']"
)
page.fill(
"//div[normalize-space(.)='First Name*']/input[normalize-space(@type)='text']",
CONTACT_FIRST)
page.fill(
"//div[normalize-space(.)='Middle Name']/input[normalize-space(@type)='text']",
"")
page.press(
"//div[normalize-space(.)='Middle Name']/input[normalize-space(@type)='text']",
"Tab")
page.fill(
"//div[normalize-space(.)='Last Name*']/input[normalize-space(@type)='text']",
CONTACT_LAST)
page.press(
"//div[normalize-space(.)='Last Name*']/input[normalize-space(@type)='text']",
"Tab")
page.fill(
"(//div[normalize-space(@role)='listitem']/div/div/div[normalize-space(.)='Street Address*']/input[normalize-space(@type)='text'])[2]",
ORIGIN_STREET)
page.click(
"(//div[normalize-space(@role)='listitem']/div/div/div[normalize-space(.)='City*']/input[normalize-space(@type)='text'])[2]"
)
page.fill(
"(//div[normalize-space(@role)='listitem']/div/div/div[normalize-space(.)='City*']/input[normalize-space(@type)='text'])[2]",
ORIGIN_CITY)
page.click(
"(//div[normalize-space(@role)='listitem']/div/div/div/div/div[1]/div/div/a[normalize-space(.)='--None--' and normalize-space(@role)='button'])[2]"
)
page.click(
f"(//a[normalize-space(.)='{ORIGIN_STATE}' and normalize-space(@title)='{ORIGIN_STATE}' and normalize-space(@role)='menuitemradio'])[3]"
)
page.click(
"(//div[normalize-space(@role)='listitem']/div/div/div[normalize-space(.)='ZIP Code™*']/input[normalize-space(@type)='text'])[2]"
)
page.fill(
"(//div[normalize-space(@role)='listitem']/div/div/div[normalize-space(.)='ZIP Code™*']/input[normalize-space(@type)='text'])[2]",
ORIGIN_ZIP)
page.click(
"//div[normalize-space(.)='Email*']/input[normalize-space(@type)='text']"
)
page.fill(
"//div[normalize-space(.)='Email*']/input[normalize-space(@type)='text']",
CONTACT_EMAIL)
page.press(
"//div[normalize-space(.)='Email*']/input[normalize-space(@type)='text']",
"Tab")
page.fill(
"//div[normalize-space(.)='Retype Email*']/input[normalize-space(@type)='text']",
CONTACT_EMAIL)
page.press(
"//div[normalize-space(.)='Retype Email*']/input[normalize-space(@type)='text']",
"Tab")
page.fill("input[type=\"tel\"]", ORIGIN_PHONE)
page.click(
"(//div[normalize-space(@role)='listitem']/div/div/div/div/div[1]/div/div/a[normalize-space(.)='--None--' and normalize-space(@role)='button'])[2]"
)
page.click(
"(//a[normalize-space(.)='Business' and normalize-space(@title)='Business' and normalize-space(@role)='menuitemradio'])[3]"
)
page.click(
"//div[normalize-space(.)='Company']/input[normalize-space(@type)='text']"
)
page.fill(
"//div[normalize-space(.)='Company']/input[normalize-space(@type)='text']",
ORIGIN_COMPANY)
page.click("//div[normalize-space(.)='PreviousFinishCancel']")
page.click("text=\"Finish\"")
page.click("text=\"Submit\"")
page.close()
context.close()
browser.close()
def visit(browser, method, url):
playwright = sync_playwright().start()
browsers = playwright.chromium.launch(headless=False)
page = browsers.newPage()
page.goto(url)
def playwright() -> Generator[Any, None, None]:
pw = sync_playwright().start()
yield pw
pw.stop()
page2.goto("https://webkit.org")
print("Page2 tile: %s" % page2.title())
print("Pages in context: %d" % len(context.pages))
print("\nQuerying body...")
body1 = page1.query_selector("body")
assert body1
print("Body text %s" % body1.text_content())
print("Closing page1...")
page1.close()
print("Pages in context: %d" % len(context.pages))
print("Navigating page2 to https://cnn.com...")
page2.goto("https://cnn.com")
print("Page2 main frame url: %s" % page2.main_frame.url)
print("Page2 tile: %s" % page2.title())
print("Frames in page2: %d" % len(page2.frames))
print("Pages in context: %d" % len(context.pages))
print("Closing context...")
context.close()
print("Contexts in browser: %d" % len(browser.contexts))
print("Closing browser")
browser.close()
if __name__ == "__main__":
with playwright.sync_playwright() as p:
main(p)
import logging
import random
import string
import requests
# Import Detection From Stealth
from .get_acrawler import get_acrawler
from playwright import sync_playwright
try:
playwright = sync_playwright().start()
except Exception as e:
raise e
async_support = False
logger = logging.getLogger(__name__)
def get_playwright():
return playwright
class browser:
def __init__(
self,
**kwargs,
):
self.debug = kwargs.get("debug", False)
self.proxy = kwargs.get("proxy", None)
self.api_url = kwargs.get("api_url", None)
def playwright():
with sync_playwright() as p:
yield p
