- What is Feed Analyser?
- How does it work?
- How do I run this?
- Running with a GNU Guix container
- Suggestions/ Help Wanted
What is Feed Analyser?
This aggregates content from different channels(for now only twitter and github commits are properly supported) and displays them in one place. In most hackathons, organisations and sometimes events, content spans different places e.g. Discord, slack, IRC, mailing list, issue trackers etc. This project aims to aggregate news from different places and display them in one place.
For now, the project supports twitter and github.
It also allows you to vote on the feeds. You can view a demo here
How does it work?
The program scrapes content from twitter and github and stores the data into redis queues with a score. You can then read from the redis queue.
There's also an in-built provision for voting on queues, whereby each vote updates the zscore of an individual item thereby increasing or decreasing the likelihood of it being displayed on a page.
Why are you storing things to Redis? Why not just generate pages straight from them?
We'd like to extend the program later so that it can provide a source of aggregate data to other programs. To achieve this, you'd need an intermediary data store, and for this, Redis was chosen.
How do I run this?
If you are fetching data from twitter, ensure you have Python3 and twint installed as this is used to fetch data from twitter without using twitter's restrictive API. And abide by the Twitter terms of service and usage policies when using that data!
Configure the feed reader
See configuration examples.
Start the twitter/ gh fetching daemon:
# Replace example.conf.rkt with your own conf file with the correct settings racket newsfeed/update-feed.rkt -c etc/biohackathon.conf.rkt
The worker daemon will fetch data from twitter and github on a set time frame as set in the configuration - typically every 6 hours or so - and feed the results to redis. Note the feed-prefix that distinguishes the actual feeds.
Run the voting 'server'
The voting server responds to requests from a web service. It serves
# Replace the conf with your own conf file with the correct settings racket newsfeed/voting-server.rkt -c etc/biohackathon.conf.rkt
Running with a GNU Guix container
Right now, since the feedanalyser uses redis-racket– which throws errors in Racket v7.9, you could use guix time-machine with a channels file to set things up!
First, the channels files looks(tried and tested so it works) something like:
(list (channel (name 'guix-bioinformatics) (url "https://github.com/BonfaceKilz/guix-bioinformatics") (branch "master") (commit "ad741c19d6a7bfc16610fa25689398243deaa7a5")) (channel (name 'guix) (url "https://git.savannah.gnu.org/git/guix.git") (branch "master") (commit "1f39175d1a030877b034a0ba85ef94b987b50b3e")))
The above channel definitions contains the working versions of racket and python-twint prior the v7.9 bump in upstream Guix.
After setting up the channels file, create a guix container with
racket and other useful tools ("expose"
it's required by one of the Racket packages. The -N flag allows you
to connect to the internet, and to also connect to redis which runs
outside your container):
./pre-inst-env guix time-machine -C channels.scm -- environment -N -C --expose=/usr/share/zoneinfo --ad-hoc racket python-twint nss-certs coreutils
Now export the SSL_CERTS so that you can use Racket over the internet:
export SSL_CERT_DIR="$GUIX_ENVIRONMENT/etc/ssl/certs" && export SSL_CERT_FILE="$GUIX_ENVIRONMENT/etc/ssl/certs/ca-certificates.crt"
Finally, the usual(inside the env):
raco pkg install
Now you can run the feed and associated voting server using the commands:
# Start the feed daemon in the background racket newsfeed/update-feed.rkt -c etc/biohackathon.conf.rkt & # Start the voting server. Be sure to configure things properly! racket newsfeed/voting-server.rkt -c etc/biohackathon.conf.rkt
PS: You could set racket up in it's own profile by running something like:
./pre-inst-env guix time-machine -C channels.scm -- install racket python-twint -p ~/opt/racket
Suggestions/ Help Wanted
Fetch data from Slack
Come up with an algorithm to sieve out noise(Machine Learning)
Expose content using an API
Come up with a sane deployment process
[x] Better Parsing from Twitter
Add package and dependencies to Guix
Integrate to GN2
Fetch Pull Requests and merges from GitHub and display them
Fetch Data from Slack
Make a lib layer and add project to upstream to racket-lang.org
Fetch Data from IRC
Add NLP to make more meaningful data
This tool is published under the GPLv3. See LICENSE.