如何让phantomJS webdriver等到加载特定的HTML元素然后返回page.source？

Question

我已经为Web爬网对象开发了以下代码。

它需要两个日期作为输入。然后创建这两个日期之间的日期列表，并将每个日期附加到包含位置的天气信息的网页URL。然后它将HTML数据表转换为Dataframe，之后将数据存储为存储中的csv文件（基本链接为：https://www.wunderground.com/history/daily/ir/mashhad/OIMM/date/2019-1-3，在本例中您可以看到日期为2019-1-3）：

from datetime import timedelta, date
from bs4 import BeautifulSoup
from selenium import webdriver
import pandas as pd
from furl import furl
import os
import time

class WebCrawler():
    def __init__(self, st_date, end_date):
        if not os.path.exists('Data'):
            os.makedirs('Data')
        self.path = os.path.join(os.getcwd(), 'Data')
        self.driver = webdriver.PhantomJS()
        self.base_url = 'https://www.wunderground.com/history/daily/ir/mashhad/OIMM/date/'
        self.st_date = st_date
        self.end_date = end_date

    def date_list(self):
        # Create list of dates between two dates given as inputs.
        dates = []
        total_days = int((self.end_date - self.st_date).days + 1)

        for i in range(total_days):
            date = self.st_date + timedelta(days=i)
            dates.append(date.strftime('%Y-%m-%d'))

        return dates

    def create_link(self, attachment):
        # Attach dates to base link
        f = furl(self.base_url)
        f.path /= attachment
        f.path.normalize()

        return f.url

    def open_link(self, link):
        # Opens link and visits page and returns html source code of page
        self.driver.get(link)
        html = self.driver.page_source

        return html

    def table_to_df(self, html):
        # Finds table of weather data and converts it into pandas dataframe and returns it
        soup = BeautifulSoup(html, 'lxml')
        table = soup.find("table",{"class":"tablesaw-sortable"})

        dfs = pd.read_html(str(table))
        df = dfs[0]

        return df

    def to_csv(self, name, df):
        # Save the dataframe as csv file in the defined path
        filename = name + '.csv'
        df.to_csv(os.path.join(self.path,filename), index=False)

这是我想要使用WebCrawler对象的方式：

date1 = date(2018, 12, 29)
date2 = date(2019, 1, 1)

# Initialize WebCrawler object
crawler = WebCrawler(st_date=date1, end_date=date2)
dates = crawler.date_list()

for day in dates:
    print('**************************')
    print('PROCESSING : ', day)
    link = crawler.create_link(day)
    print('WAITING... ')
    time.sleep(3)
    print('VISIT WEBPAGE ... ')
    html = crawler.open_link(link)
    print('DATA RETRIEVED ... ')
    df = crawler.table_to_df(html)
    print(df.head(3))
    crawler.to_csv(day, df)
    print('DATA SAVED ...')

发生的问题是循环的第一次迭代运行完美，但第二次循环停止时出现错误，表示No tables where found（发生在table = soup.find("table",{"class":"tablesaw-sortable"})行），这是因为页面源是由WebCrawler.open_link在网页完全加载网页内容之前返回的，包括表（包含天气信息）。网站也有可能拒绝请求，因为它使服务器太忙。

无论如何我们可以构建一个循环，一直试图打开链接，直到它找到表，或者至少等到表加载然后返回表？

Answer 1

您可以让selenium等待特定元素。在您的情况下，它将是类名为“tablesaw-sortable”的表。我强烈建议您使用CSS选择器来查找此元素，因为获取所有表元素的速度快且容易出错。

这是CSS选择器，为您预制table.tablesaw-sortable。将selenium设置为等待该元素加载。

资料来源：https://stackoverflow.com/a/26567563/4159473

Answer 2

我使用@mildmelon建议的https://stackoverflow.com/a/26567563/4159473解决方案重写代码，每次向服务器发送请求和请求页面源时，我也使用了一些延迟：

from datetime import timedelta, date
from bs4 import BeautifulSoup
from selenium import webdriver
from selenium.webdriver.support.ui import WebDriverWait
from selenium.webdriver.support import expected_conditions as EC
from selenium.webdriver.common.by import By
from selenium.common.exceptions import TimeoutException
import pandas as pd
from furl import furl
import os
import time
class WebCrawler():
    def __init__(self, st_date, end_date):
        if not os.path.exists('Data'):
            os.makedirs('Data')
        self.path = os.path.join(os.getcwd(), 'Data')
        self.driver = webdriver.PhantomJS()
        self.delay_for_page = 7
        self.base_url = 'https://www.wunderground.com/history/daily/ir/mashhad/OIMM/date/'
        self.st_date = st_date
        self.end_date = end_date

    def date_list(self):
        # Create list of dates between two dates given as inputs.
        dates = []
        total_days = int((self.end_date - self.st_date).days + 1)

        for i in range(total_days):
            date = self.st_date + timedelta(days=i)
            dates.append(date.strftime('%Y-%m-%d'))

        return dates

    def create_link(self, attachment):
        # Attach dates to base link
        f = furl(self.base_url)
        f.path /= attachment
        f.path.normalize()

        return f.url

    def open_link(self, link):
        # Opens link and visits page and returns html source code of page
        self.driver.get(link)
        myElem = WebDriverWait(self.driver, self.delay_for_page)\
        .until(EC.presence_of_element_located((By.CLASS_NAME, 'tablesaw-sortable')))


    def table_to_df(self, html):
        # Finds table of weather data and converts it into pandas dataframe and returns it
        soup = BeautifulSoup(html, 'lxml')
        table = soup.find("table",{"class":"tablesaw-sortable"})

        dfs = pd.read_html(str(table))
        df = dfs[0]

        return df

    def to_csv(self, name, df):
        # Save the dataframe as csv file in the defined path
        filename = name + '.csv'
        df.to_csv(os.path.join(self.path,filename), index=False)

date1 = date(2019, 2, 1)
date2 = date(2019, 3, 5)


# Initialize WebCrawler object
crawler = WebCrawler(st_date=date1, end_date=date2)
dates = crawler.date_list()
for day in few_dates:
    print('**************************')
    print('DATE : ', day)
    link = crawler.create_link(day)
    print('WAITING ....')
    print('')
    time.sleep(12)
    print('OPENING LINK ... ')

    try:
        crawler.open_link(link)
        html = crawler.driver.page_source
        print( "DATA IS FETCHED")
        df = crawler.table_to_df(html)
        print(df.head(3))
        crawler.to_csv(day, df)
        print('DATA SAVED ...')
    except TimeoutException:
        print( "NOT FETCHED ...!!!")

天气信息没有问题。我想每次请求之间的延迟会带来更好的性能。线myElem = WebDriverWait(self.driver, self.delay_for_page)\.until(EC.presence_of_element_located((By.CLASS_NAME, 'tablesaw-sortable')))也提高了速度。