Ask HN: Anyone here do Web mining/harvesting/scraping?

9 points·Zelgius··6 comments

I've seen a bunch of software solutions out there. WebQL lets you write SQL-like code to harvest websites. There are also GUI solutions like modenza, fetch and kapow.

Does anyone here work with these kinds of solutions and what do you consider to be the best approach and why?

Thanks!

View on HN

I have done some webscraping with twisted and BeautifulSoup. But it was just for a few small sites.

That was when I found http://scrapy.org/ , a complete python framework for writing webcrawlers. It looks promissing, but I haven't used it so far.

We do a lot of scraping and the best we've found, really, is beautifulsoup and being good at python. There's no UI or easy tool that's going to replace that.

Actually WebQL is pretty amazing. A simple program like select * from links within http://www.cnn.com to an advanced 10 page script to pull and merge multiple sources. The code is real easy to read and if you have any kind of SQL background very easy to learn. We're just always keeping our eyes out for something better/cheaper (ie free) :)

I would recommend hosted crawling/scrapping solutions where you don't need to dedicate your own hardware resources. Also monitoring and maintaining the scrappers/harvesters can be a pain. Shameless Plug: I work @ beevolve (www.beevolve.com) and we provide hosted crawling and scrapping solutions. PM me if you are interested.

I use a mixture of cURL and PHP's DOM and simpleXML functions to scrape what I need.

I've done it with Java, HttpClient and regular expressions.

Reply on news.ycombinator.com