Sinew collects structured data from web sites (screen scraping). It provides a Ruby DSL built for crawling, a robust caching system, and integration with Nokogiri. Though small, this project is the culmination of years of effort based on crawling systems built at several different companies.
Sinew requires Ruby 1.9, HTML Tidy and Curl.
Sinew is distributed as a ruby gem:
gem install sinewHere's an example for collecting Amazon's bestseller list:
# get the url
get "http://www.amazon.com/gp/bestsellers/books/ref=sv_b_3"
# use nokogiri to find books
noko.css(".zg_itemRow").each do |item|
# pull out the stuff we care about using nokogiri
row = { }
row[:url] = item.css(".zg_title a").first[:href]
row[:title] = item.css(".zg_title")
row[:img] = item.css(".zg_itemImage_normal img").first[:src]
# append a row to the csv
csv_emit(row)
endIf you paste this into a file called bestsellers.sinew and run sinew bestsellers.sinew, it will create a bestsellers.csv file containing the url, title and img for each bestseller.
I'm not an expert on Mechanize, but this question has come up repeatedly and I'll try to address it. Mechanize is a great toolkit and it's better for some situations. Briefly:
- Sinew caches all HTTP requests on disk. That makes it possible to iterate quickly. Crawl once and then continue to work on your recipe. Run the recipe over and over while you tune your CSS selectors and regular expressions.
- Sinew runs responses through HTML Tidy. This cleans up dirty HTML and makes it easier to parse in many cases, especially if you have to fallback to regular expressions instead of Nokogiri. Unfortunately, this is a common use case in my experience.
- Sinew outputs CSV files. It does exactly one thing and it does it well - Sinew crawls a site and outputs a CSV file. Mechanize is a more general toolkit.
Full docs are in the wiki: