This is an HTTP front-end to the Eros library. Eros is a text censorship library, that I wrote.
If the server receives a GET request, it returns an HTML representation of this file.
It takes an input string via POST, and returns some data in JSON mapping each phraselist to the score for the input string. The JSON is compressed.
I fed the server the GPL, here are the results, which have been prettified, and alphabetized by key.
{
"chat": 0,
"conspiracy": 0,
"drug-advocacy": 0,
"forums": 0,
"gambling": 0,
"games": 0,
"gore": 0,
"id-theft": 0,
"illegal-drugs": 0,
"intolerance": 0,
"legal-drugs": 0,
"malware": 0,
"music": 0,
"news": 0,
"nudism": 0,
"peer2peer": 0,
"personals": 0,
"pornography": 20,
"proxies": 0,
"secret-societies": 0,
"self-labeling": 0,
"sport": 30,
"translation": 0,
"upstream-filter": 0,
"violence": 0,
"warez-hacking": 0,
"weapons": 0,
"webmail": 0
}
The home page of PornHub yields this result:
{
"chat": 0,
"conspiracy": 0,
"drug-advocacy": 0,
"forums": 0,
"gambling": 0,
"games": 0,
"gore": 0,
"id-theft": 0,
"illegal-drugs": 0,
"intolerance": 0,
"legal-drugs": 0,
"malware": 0,
"music": 0,
"news": 0,
"nudism": 0,
"peer2peer": 0,
"personals": 0,
"pornography": 8790,
"proxies": 0,
"secret-societies": 0,
"self-labeling": 0,
"sport": 0,
"translation": 0,
"upstream-filter": 0,
"violence": 0,
"warez-hacking": 0,
"weapons": 0,
"webmail": 0
}
This is the command I used
curl www.pornhub.com | curl -d @- localhost:8000 | prettify.rb
This uses my JSON prettifier script.
Edit - This bug was fixed in version 0.6.
There's a fundamental flaw in the algorithm where if it gets a lot of data with
a lot of flagged phrases, it takes a long time to calculate the result. The
server times out after 30 seconds. I haven't figured out how to get
multiprocessing, so this will remain a bug until I do.
This won't be a problem for typical usage, only if you try to send paragraphs of
dirty data in one string.
Typical usage is for SMS messages, which are at most 160 characters long.