Scrapy at a glance

Scrapy is an application framework for crawling web sites and extracting structured data which can be used for a wide range of useful applications, like data mining, information processing or historical archival.

Even though Scrapy was originally designed for screen scraping (more precisely, web scraping), it can also be used to extract data using APIs (such as Amazon Associates Web Services) or as a general purpose web crawler.

The purpose of this document is to introduce you to the concepts behind Scrapy so you can get an idea of how it works and decide if Scrapy is what you need.

When you’re ready to start a project, you can start with the tutorial.

Pick a website

So you need to extract some information from a website, but the website doesn’t provide any API or mechanism to access that info programmatically. Scrapy can help you extract that information.

Let’s say we want to extract the URL, name, description and size of all torrent files added today in the Mininova site.

The list of all torrents added today can be found on this page:

Define the data you want to scrape

The first thing is to define the data we want to scrape. In Scrapy, this is done through Scrapy Items (Torrent files, in this case).

This would be our Item:

from scrapy.item import Item, Field

class TorrentItem(Item):
    url = Field()
    name = Field()
    description = Field()
    size = Field()

Write a Spider to extract the data

The next thing is to write a Spider which defines the start URL (http://www.mininova.org/today), the rules for following links and the rules for extracting the data from pages.

If we take a look at that page content we’ll see that all torrent URLs are like http://www.mininova.org/tor/NUMBER where NUMBER is an integer. We’ll use that to construct the regular expression for the links to follow: /tor/\d+.

We’ll use XPath for selecting the data to extract from the web page HTML source. Let’s take one of those torrent pages:

And look at the page HTML source to construct the XPath to select the data we want which is: torrent name, description and size.

By looking at the page HTML source we can see that the file name is contained inside a <h1> tag:

<h1>Home[2009][Eng]XviD-ovd</h1>

An XPath expression to extract the name could be:

//h1/text()

And the description is contained inside a <div> tag with id="description":

<h2>Description:</h2>

<div id="description">
"HOME" - a documentary film by Yann Arthus-Bertrand
<br/>
<br/>
***
<br/>
<br/>
"We are living in exceptional times. Scientists tell us that we have 10 years to change the way we live, avert the depletion of natural resources and the catastrophic evolution of the Earth's climate.

...

An XPath expression to select the description could be:

//div[@id='description']

Finally, the file size is contained in the second <p> tag inside the <div> tag with id=specifications:

<div id="specifications">

<p>
<strong>Category:</strong>
<a href="/cat/4">Movies</a> &gt; <a href="/sub/35">Documentary</a>
</p>

<p>
<strong>Total size:</strong>
699.79&nbsp;megabyte</p>

An XPath expression to select the file size could be:

//div[@id='specifications']/p[2]/text()[2]

For more information about XPath see the XPath reference.

Finally, here’s the spider code:

class MininovaSpider(CrawlSpider):

    name = 'mininova'
    allowed_domains = ['mininova.org']
    start_urls = ['http://www.mininova.org/today']
    rules = [Rule(SgmlLinkExtractor(allow=['/tor/\d+']), 'parse_torrent')]

    def parse_torrent(self, response):
        sel = Selector(response)
        torrent = TorrentItem()
        torrent['url'] = response.url
        torrent['name'] = sel.xpath("//h1/text()").extract()
        torrent['description'] = sel.xpath("//div[@id='description']").extract()
        torrent['size'] = sel.xpath("//div[@id='info-left']/p[2]/text()[2]").extract()
        return torrent

For brevity’s sake, we intentionally left out the import statements. The Torrent item is defined above.

Run the spider to extract the data

Finally, we’ll run the spider to crawl the site an output file scraped_data.json with the scraped data in JSON format:

scrapy crawl mininova -o scraped_data.json -t json

This uses feed exports to generate the JSON file. You can easily change the export format (XML or CSV, for example) or the storage backend (FTP or Amazon S3, for example).

You can also write an item pipeline to store the items in a database very easily.

Review scraped data

If you check the scraped_data.json file after the process finishes, you’ll see the scraped items there:

[{"url": "http://www.mininova.org/tor/2657665", "name": ["Home[2009][Eng]XviD-ovd"], "description": ["HOME - a documentary film by ..."], "size": ["699.69 megabyte"]},
# ... other items ...
]

You’ll notice that all field values (except for the url which was assigned directly) are actually lists. This is because the selectors return lists. You may want to store single values, or perform some additional parsing/cleansing to the values. That’s what Item Loaders are for.

What else?

You’ve seen how to extract and store items from a website using Scrapy, but this is just the surface. Scrapy provides a lot of powerful features for making scraping easy and efficient, such as:

  • Built-in support for selecting and extracting data from HTML and XML sources
  • Built-in support for cleaning and sanitizing the scraped data using a collection of reusable filters (called Item Loaders) shared between all the spiders.
  • Built-in support for generating feed exports in multiple formats (JSON, CSV, XML) and storing them in multiple backends (FTP, S3, local filesystem)
  • A media pipeline for automatically downloading images (or any other media) associated with the scraped items
  • Support for extending Scrapy by plugging your own functionality using signals and a well-defined API (middlewares, extensions, and pipelines).
  • Wide range of built-in middlewares and extensions for:
    • cookies and session handling
    • HTTP compression
    • HTTP authentication
    • HTTP cache
    • user-agent spoofing
    • robots.txt
    • crawl depth restriction
    • and more
  • Robust encoding support and auto-detection, for dealing with foreign, non-standard and broken encoding declarations.
  • Support for creating spiders based on pre-defined templates, to speed up spider creation and make their code more consistent on large projects. See genspider command for more details.
  • Extensible stats collection for multiple spider metrics, useful for monitoring the performance of your spiders and detecting when they get broken
  • An Interactive shell console for trying XPaths, very useful for writing and debugging your spiders
  • A System service designed to ease the deployment and run of your spiders in production.
  • A built-in Web service for monitoring and controlling your bot
  • A Telnet console for hooking into a Python console running inside your Scrapy process, to introspect and debug your crawler
  • Logging facility that you can hook on to for catching errors during the scraping process.
  • Support for crawling based on URLs discovered through Sitemaps
  • A caching DNS resolver

What’s next?

The next obvious steps are for you to download Scrapy, read the tutorial and join the community. Thanks for your interest!