So this is the Final Blog post. GsoC was Super fun I learned a lot interacted with cool people, made a network of awesome developers, worked under the best mentors. You can check my work product here
https://gist.github.com/abhaykatheria/9b4c5186f06f8ba9ede4bb59d254644c
It was an awesome experience thanks to PSF and my mentors for such an awesome opportunity :)
I will keep on contrubuting to PSF after gsoc.
Cheers, have a good one...
I finalised my work this week and prepared for final evaluation also helped test the multistage analysis patch by ForgetMe17.
I will try to implement the license parser after gsoc is over I will still be contributing to tern.
none
So I took this week off to prepare for college tests meanwhile i have been working on license extraction from copyrights string i found a package called scancode toolkit which have a sub-module called licensedcode which can help in this objective but since its in early production phase i am coming across a lot of bugs like scancode library does not work on py3.7or py3.8. and its api module is also buggy so i am looking into it. I hope I will finish this by the end of this week.:P
So Rose opened up a pr which will allow tern to set the environment variables. Which really shrinks the size of shell scripts required to get metadata. I also finalized my pr for metadata collection.
Waiting for the pr to be approved by mentors :)
Still stuck with parsing copyrights to obtain license.
Hi everyone a little late on this blog as I have to go back to college to get my stuff back. I passed the second evaluation, yay,. Till now I have been able to get the module metadata except the licenses. The licenses can be collected in 2 ways one is through making bash scripts to find and collect licenses from the module directory. Since licenses files dont have a set structure or a convention that's why its getting difficult and a foolproof guarantee cant be given. Another way is through making an in house copyright to License parser. For debian licenses we have debut used in tern I have to come up with similar kind of parser. I also have one idea of training a model from data collected by using github's api but thats just overkill.