geekan/scrapy-examples

scrapy-examples

Multifarious scrapy examples with integrated proxies and agents, which make you comfy to write a spider.

Don't use it to do anything illegal!

Real spider example: doubanbook

Tutorial

git clone https://github.com/geekan/scrapy-examples
cd scrapy-examples/doubanbook
scrapy crawl doubanbook

Depth

There are several depths in the spider, and the spider gets real data from depth2.

• Depth0: The entrance is http://book.douban.com/tag/
• Depth1: Urls like http://book.douban.com/tag/外国文学 from depth0
• Depth2: Urls like http://book.douban.com/subject/1770782/ from depth1

Example image

Avaiable Spiders

• tutorial
  • dmoz_item
  • douban_book
  • page_recorder
  • douban_tag_book
• doubanbook
• linkedin
• hrtencent
• sis
• zhihu
• alexa
  • alexa
  • alexa.cn

Advanced

• Use parse_with_rules to write a spider quickly.
  See dmoz spider for more details.

• Proxies

  • If you don't want to use proxy, just comment the proxy middleware in settings.
  • If you want to custom it, hack misc/proxy.py by yourself.

• Notice

  • Don't use parse as your method name, it's an inner method of CrawlSpider.

Advanced Usage

• Run ./startproject.sh <PROJECT> to start a new project.
  It will automatically generate most things, the only left things are:
  • PROJECT/PROJECT/items.py
  • PROJECT/PROJECT/spider/spider.py

Example to hack items.py and spider.py

Hacked items.py with additional fields url and description:

from scrapy.item import Item, Field

class exampleItem(Item):
    url = Field()
    name = Field()
    description = Field()

Hacked spider.py with start rules and css rules (here only display the class exampleSpider):

class exampleSpider(CommonSpider):
    name = "dmoz"
    allowed_domains = ["dmoz.org"]
    start_urls = [
        "http://www.dmoz.com/",
    ]
    # Crawler would start on start_urls, and follow the valid urls allowed by below rules.
    rules = [
        Rule(sle(allow=["/Arts/", "/Games/"]), callback='parse', follow=True),
    ]

    css_rules = {
        '.directory-url li': {
            '__use': 'dump', # dump data directly
            '__list': True, # it's a list
            'url': 'li > a::attr(href)',
            'name': 'a::text',
            'description': 'li::text',
        }
    }

    def parse(self, response):
        info('Parse '+response.url)
        # parse_with_rules is implemented here:
        #   https://github.com/geekan/scrapy-examples/blob/master/misc/spider.py
        self.parse_with_rules(response, self.css_rules, exampleItem)