Scrapy is following and scraping non-allowed links

问题

I have a CrawlSpider set up to following certain links and scrape a news magazine where the links to each issue follow the following URL scheme:

http://example.com/YYYY/DDDD/index.htm where YYYY is the year and DDDD is the three or four digit issue number.

I only want issues 928 onwards, and have my rules below. I don't have any problem connecting to the site, crawling links, or extracting items (so I didn't include the rest of my code). The spider seems determined to follow non-allowed links. It is trying to scrape issues 377, 398, and more, and follows the "culture.htm" and "feature.htm" links. This throws a lot of errors and isn't terribly important but it requires a lot of cleaning of the data. Any suggestions as to what is going wrong?

class crawlerNameSpider(CrawlSpider):
name = 'crawler'
allowed_domains = ["example.com"]
start_urls = ["http://example.com/issues.htm"]

rules = (
        Rule(SgmlLinkExtractor(allow = ('\d\d\d\d/(92[8-9]|9[3-9][0-9]|\d\d\d\d)/index\.htm', )), follow = True),
        Rule(SgmlLinkExtractor(allow = ('fr[0-9].htm', )), callback = 'parse_item'),
        Rule(SgmlLinkExtractor(allow = ('eg[0-9]*.htm', )), callback = 'parse_item'),
        Rule(SgmlLinkExtractor(allow = ('ec[0-9]*.htm', )), callback = 'parse_item'),
        Rule(SgmlLinkExtractor(allow = ('op[0-9]*.htm', )), callback = 'parse_item'),
        Rule(SgmlLinkExtractor(allow = ('sc[0-9]*.htm', )), callback = 'parse_item'),
        Rule(SgmlLinkExtractor(allow = ('re[0-9]*.htm', )), callback = 'parse_item'),
        Rule(SgmlLinkExtractor(allow = ('in[0-9]*.htm', )), callback = 'parse_item'),
        Rule(SgmlLinkExtractor(deny = ('culture.htm', )), ),
        Rule(SgmlLinkExtractor(deny = ('feature.htm', )), ),
    )

EDIT: I fixed this using a much simpler regex fot 2009, 2010, 2011, but I am still curious why the above doesn't work if anyone has any suggestions.

回答1:

You need to pass deny arguments to SgmlLinkExtractor which collects links to follow. And you don't need to create so many Rule's if they call one function parse_item. I would write your code as:

rules = (
        Rule(SgmlLinkExtractor(
                    allow = ('\d\d\d\d/(92[8-9]|9[3-9][0-9]|\d\d\d\d)/index\.htm', ),
                    deny = ('culture\.htm', 'feature\.htm'),
                    ), 
            follow = True
        ),
        Rule(SgmlLinkExtractor(
                allow = (
                    'fr[0-9].htm', 
                    'eg[0-9]*.htm',
                    'ec[0-9]*.htm',
                    'op[0-9]*.htm',
                    'sc[0-9]*.htm',
                    're[0-9]*.htm',
                    'in[0-9]*.htm',
                    )
                ), 
                callback = 'parse_item',
        ),
    )

If it's real url patterns in rules you are using to parse_item, it can be simplified to this:

 Rule(SgmlLinkExtractor(
                allow = ('(fr|eg|ec|op|sc|re|in)[0-9]*\.htm', ), 
                callback = 'parse_item',
        ),
 )

来源：https://stackoverflow.com/questions/8537687/scrapy-is-following-and-scraping-non-allowed-links

标签

python

screen-scraping

scrapy

web-crawler