ashtonthomas / bento-wiki-spider

hack project

How many hops to Philosophy

https://bento-wiki-spider.herokuapp.com

Self Evaluation

The Good

• Well, it "works"
• Used Spark framework for the first time
• Deployment on Heroku via Maven
• Gulp + optimizations setup (see Bendystraw)
• Using Angular 1.5 front-end (not following all of my guidelines but it is something to look at..)
• I'm using Belly's style framework (Makes it a little prettier and shows just another integration)

The Bad

• Not fully complete - there are some criteria not met
• Logic: Edge cases are probably not handled completely
• Code organization could be improved
• Data persistence is not leveraged outside of the session
• Checks for infinite loops could be handled more appropriately
• No checks for known nodes/paths (potential performance optimization)
• No tests

This Repo

• The angular code can be found in /source
• The angular project uses brousalis/bendystraw a collection of gulp tasks, configurations
• The styling leverages bellycard/rolodex which is a sass library using functional css
• gulp build will compile front-end resources to /public
• The backend spark files can be found in src/main/java/sparq
• The meat of the logic is in util/PathTracer

Done

• Node, NPM, Bower, Gulp
• brousalis/bendystraw integration (deployment optimizations and more)
• bellycard/rolodex integration
• Base Angular setup with ui-router
• Angular communication with Spark backend
• Spark Setup
• Heroku deployment
• Use session for temp storage (sorry)
• Trace wiki route
• Display hops in angular app
• Configure heroku plugin to include static files (remove target from git repo)
• Use heroku maven plugin to deploy (remove public and target folders from repo)

Not Done

• Hibernate connection with Derby (failed here: #1)
• Need a better way to develop locally and be able to edit angular code and java code without rebuilding
• Find a way to build on deploy and not commit compiled resources on every change (this is so annoying)
• Show loading while parsing / or show progress (but that will take more changes)
• Configure maven to run gulp build

This is the start of the java hack project.

• Full-stack java-based app
• Web view that accepts a wikipedia url (starting point)
• recursively retrieve the first lowercase link in the main text until the Philosophy page is reached
• persist and present the path taken (include datastore)
• handle edge cases or special scenarios
• display sufficient design and organization of application
• will be mostly new technologies and I get to use an IDE again (not even sure what is on my personal computer...)

Technologies to tinker with:

• Spark
• Hibernate (nvm)
• Angular (1.5)
• MySQL (maybe use an in memory data store or filestore) - maybe Postgres if I'm deploying to Heroku
• Deploy to Heroku?

References:

• https://github.com/rayokota/generator-angular-spark
• https://sparktutorials.github.io/2016/06/26/ajax-without-writing-javascript.html
• https://blog.openshift.com/developing-single-page-web-applications-using-java-8-spark-mongodb-and-angularjs/
• https://sparktutorials.github.io/2015/08/24/spark-heroku.html
• https://devcenter.heroku.com/articles/connecting-to-relational-databases-on-heroku-with-java
• http://ramonaharrison.github.io/accesscode/java/http/wikipedia/2015/03/27/wikipedia-philosophy/
• https://github.com/ramonaharrison/wikipedia-philosophy

Running Deploying

brew // all the things (node, npm)
npm install -g bower
npm install --global gulp-cli

npm install
bower install
gulp build
mvn heroku:deploy

heroku open

// and when things catch fire
heroku logs --tail -a bento-wiki-spider

Or if you really need to get your hands dirty

heroku run bash -a bento-wiki-spider

For Angular Development

Just run gulp and then open: http://localhost:4567

When a dev server hangs on the port

kill -kill `lsof -t -i tcp:4567`