Question

I have around 5 GB of html data which I want to process to find links to a set of websites and perform some additional filtering. Right now I use simple regexp for each site and iterate over them, searching for matches. In my case links can be outside of “a” tags and be not well formed in many ways(like “\n” in the middle of link) so I try to grab as much “links” as I can and check them later in other scripts(so no BeatifulSoup\lxml\etc). The problem is that my script is pretty slow, so I am thinking about any ways to speed it up. I am writing a set of test to check different approaches, but hope to get some advices 🙂

Right now I am thinking about getting all links without filtering first(maybe using C module or standalone app, which doesn’t use regexp but simple search to get start and end of every link) and then using regexp to match ones I need.

Answer 1

Ways out.

• Parallelise
• Profile your code to see where the bottleneck is. The result are often surprising.
• Use a single regexp (concatenate using |) rather than multiple ones.