Python Scrapy Dynamic Web Sites

问题:

I am trying to scrape a very simple web page with the help of Scrapy and it's xpath selectors but for some reason the selectors I have do not work in Scrapy but they do work in other xpath utilities

I am trying to parse this snippet of html:

<select id="chapterMenu" name="chapterMenu">

<option value="/111-3640-1/20th-century-boys/chapter-1.html" selected="selected">Chapter 1: Friend</option>

<option value="/111-3641-1/20th-century-boys/chapter-2.html">Chapter 2: Karaoke</option>

<option value="/111-3642-1/20th-century-boys/chapter-3.html">Chapter 3: The Boy Who Bought a Guitar</option>

<option value="/111-3643-1/20th-century-boys/chapter-4.html">Chapter 4: Snot Towel</option>

<option value="/111-3644-1/20th-century-boys/chapter-5.html">Chapter 5: Night of the Science Room</option>

</select>

Scrapy parse_item code:

def parse_item(self, response):
    itemLoader = XPathItemLoader(item=MangaItem(), response=response)
    itemLoader.add_xpath('chapter', '//select[@id="chapterMenu"]/option[@selected="selected"]/text()')
    return itemLoader.load_item()

Scrapy does not extract any text from this but if I get the same xpath and html snippet and run it here it works just fine.

if I use this xpath:

//select[@id="chapterMenu"]

I get the correct element but when I try to access the options inside it does not get anything

回答1:

Scrapy only does a GET request for the url, it is not a web browser and therefore cannot run JavaScript. Because of this Scrapy alone will not be enough to scrape through dynamic web pages.

In addition you will need something like Selenium which basically gives you an interface to several web browsers and their functionalities, one of them being the ability to run JavaScript and get client side generated HTML.

Here is a snippet of how one can go about doing this:

from Project.items import SomeItem
from scrapy.contrib.linkextractors.sgml import SgmlLinkExtractor
from scrapy.contrib.spiders import CrawlSpider, Rule
from scrapy.selector import Selector
from selenium import webdriver
import time

class RandomSpider(CrawlSpider):

    name = 'RandomSpider'
    allowed_domains = ['random.com']
    start_urls = [
        'http://www.random.com'
    ]

    rules = (
        Rule(SgmlLinkExtractor(allow=('some_regex_here')), callback='parse_item', follow=True),
    )

    def __init__(self):
        CrawlSpider.__init__(self)
        # use any browser you wish
        self.browser = webdriver.Firefox() 

    def __del__(self):
        self.browser.close()

    def parse_item(self, response):
        item = SomeItem()
        self.browser.get(response.url)
        # let JavaScript Load
        time.sleep(3) 

        # scrape dynamically generated HTML
        hxs = Selector(text=self.browser.page_source) 
        item['some_field'] = hxs.select('some_xpath')
        return item

回答2:

I think I found the webpage you want to extract from, and the chapters are loaded after fetching some JSON data, based on a "mangaid" (that is available in a Javascript Array in the page.

So fetching the chapters is a matter of making a specific GET request to a specific /actions/selector/ endpoint. It's basically emulating what your browser's Javascript engine is doing.

You probably get better performance using this technique than Selenium, but it does involve (minor) Javascript parsing (no real interpretation needed).