No Description

Eliezer Croitoru 6311641a03 Added LICENSE. 2 months ago
test 4a88d42af3 Initial Commit 2 months ago
LICENSE 6311641a03 Added LICENSE. 2 months ago
README.md 4a88d42af3 Initial Commit 2 months ago
REFRENCES.md 4a88d42af3 Initial Commit 2 months ago
cache.go b46da56dc8 Fixed cople tiny things. 2 months ago
category.go 4a88d42af3 Initial Commit 2 months ago
httputils.go 4a88d42af3 Initial Commit 2 months ago
json.go 4a88d42af3 Initial Commit 2 months ago
main.go b46da56dc8 Fixed cople tiny things. 2 months ago
utils.go 4a88d42af3 Initial Commit 2 months ago

README.md

RedWood external dynamic classification portal

Since the RedWood proxy and Squid lacks some dynamicity. In squid this dynamicity can be completed with external tools like ICAP service. After I spoke to Andy about the integration of my drbl-peer library inside RedWood with simplified the concept. Instead of giving to RedWood do all the work we split the logic into two.

  • Redwood part
  • External tools

RedWood part

Add an option to "classification service http/https url based api". The RedWood proxy will send the url towards the classification service and expects json response. The json response shold either be blank ie "{}" or a dictionary of categories: int-weight. For example RedWood will send to the url "https://service/check-for-abusive-content/?url=http://www.youtube.com/watch?v=videoID..." RedWood excpects to accept a list of matched categoreis and their weight. For example if this video has ofensive/abusive nature the response will be "{ "abuse": 1000}" If it fits malware and shopping it will return "{"malware": 1000, "shopping": 1000}"

And the RedWood Admin will just need to put the right rules in the regualr categories directories structure.

External serivice

Need to be able to work as an http service that can validate urls and maybe enforce access rules. The basic sketch will be a "query" only service. It will have only one end point "128" or "1024" and will verify the classification of both malware and abusive content match It will use the drbl-peer library to first check only the hostname of the url to classifiy.

Caching

It might contain couple cache features in the future for 1-3 seconds.

Planned YouTube videos classification

I am planning to add a specific option that will allow to have a youtube videos, images, playlist and users ID classification.