Using Firebug for scraping¶
This document explains how to use Firebug (a Firefox add-on) to make the scraping process easier and more fun. For other useful Firefox add-ons see Useful Firefox add-ons for scraping. There are some caveats with using Firefox add-ons to inspect pages, see Caveats with inspecting the live browser DOM.
Firebug comes with a very useful feature called Inspect Element which allows you to inspect the HTML code of the different page elements just by hovering your mouse over them. Otherwise you would have to search for the tags manually through the HTML body which can be a very tedious task.
In the following screenshot you can see the Inspect Element tool in action.
At first sight, we can see that the directory is divided in categories, which are also divided in subcategories.
However, it seems that there are more subcategories than the ones being shown in this page, so we’ll keep looking:
As expected, the subcategories contain links to other subcategories, and also links to actual websites, which is the purpose of the directory.
Extracting the data¶
Now we’re going to write the code to extract data from those pages.
With the help of Firebug, we’ll take a look at some page containing links to websites (say http://directory.google.com/Top/Arts/Awards/) and find out how we can extract those links using XPath selectors. We’ll also use the Scrapy shell to test those XPath’s and make sure they work as we expect.
As you can see, the page markup is not very descriptive: the elements don’t
class or any attribute that clearly identifies them, so
we’‘ll use the ranking bars as a reference point to select the data to extract
when we construct our XPaths.
After using FireBug, we can see that each link is inside a
td tag, which is
itself inside a
tr tag that also contains the link’s ranking bar (in
So we can select the ranking bar, then find its parent (the
tr), and then
finally, the link’s
td (which contains the data we want to scrape).
This results in the following XPath:
It’s important to use the Scrapy shell to test these complex XPath expressions and make sure they work as expected.
Basically, that expression will look for the ranking bar’s
td element, and
then select any
td element who has a descendant
a element whose
href attribute contains the string
Of course, this is not the only XPath, and maybe not the simpler one to select
that data. Another approach could be, for example, to find any
that have that grey colour of the links,
Finally, we can write our
def parse_category(self, response): hxs = HtmlXPathSelector(response) # The path to website links in directory page links = hxs.select('//td[descendant::a[contains(@href, "#pagerank")]]/following-sibling::td/font') for link in links: item = DirectoryItem() item['name'] = link.select('a/text()').extract() item['url'] = link.select('a/@href').extract() item['description'] = link.select('font/text()').extract() yield item
Be aware that you may find some elements which appear in Firebug but
not in the original HTML, such as the typical case of
or tags which Therefer in page HTML sources may on Firebug inspects the live DOM