This is an experimental blogpost with the goal of improving myself on the design of system programs.

The following words are a flow that tries to recapture the process that I follow when I program.

17.11

This is the tracklist:

Tracklist 1 Waniyetula - Lindis Farne 2 Missus Beastly - Steel's Electric 3 Vikings Invasion - Shadow Boogie 4 Live - Fly Like A Bird 5 Arktis - Speed Boogie 6 Level π - Hubble's Dream - Dream Without End 7 Emma Myldenberger - RAA 8 Skyline - Beautiful Lady 9 Guru Guru - Oxymoron

The problem

Yesterday I got an insider tip that probably http://isole.ecn.org/ might be taken down and I remembered that I had an old archive of the entire site. Actually I am an amateur data hoarder so I keep archives/backups of pretty much everything I find amusing.

I wanted to update the archive so I used httrack. I found the following problems:

1. too slow (doesn't even use 1/20 of my connection)
2. doesn't understand that if I want to mirror ecn.org I may want isole.ecn.org as well
3. if I configure httrack to archive everything basically I hit some external websites (like google.com) and it starts archiving that as well
4. I can't configure it to avoid any external link because I may need them

A perfect scraper for me would: 1. record a list of every external website it hit so that later I can archive them as well 2. allow me to scrape subdomains as well 3. configure a level of archiving for external websites

The approach

Right now httrack does a perfect job of storing the webpages with a coherent folder layout and offline browsing capabilities. Let's see how it does that.

On the original page I can found: <link rel='stylesheet' id='avia-google-webfont' href='//fonts.googleapis.com/css?family=Lato:300,400,700' type='text/css' media='all'/>

When httrack saves it: <link rel='stylesheet' id='avia-google-webfont' href='../fonts.googleapis.com/css8f84.css?family=Lato:300,400,700' type='text/css' media='all'/>