cs373-housedowning

Presentation for team House Downing on Game of Thrones IDB

Transcript

1. Charlina Hung, Mallory Koai, Oliver Lee, Ariel Sibille, Troy Varney
2. Front End Team
3. Back End Team
4. Overview & Use Cases ❖ Database of Game of Thrones lore ❖ Difficult to keep track of ➢ Characters ➢ Locations ➢ Events ➢ Organizations ➢ Episodes
5. Tools
6. Front End ❖ No javascript frameworks ❖ Originally loaded blank template page and populated data content by using javascript to call our API ❖ The javascript meant our pages loaded slowly; switched to rendering templates
7. Front End ❖ Visual representation with toggle for more detailed option ❖ Bootstrap ❖ Hover-over effects ❖ Outsourcing to Hodor ❖ Twitter widget
8. API - Pillars, Basic Form ❖ /api/<pillar> ➢ grabs a json list of all the items in the pillar ❖ /api/<pillar>/<id> ➢ grabs a single item Getting a single item yields more information about that one item than getting all the items in the pillar
9. API - Pillars, Example ❖ /api/character [ {"last_name": "Baratheon", "gender": "Male", "religion": "Faith of the Seven", "char_id": 1, "first_name": "Robert", "status": "Deceased"}, ..... ] ❖ /api/character/1 { "affiliations": [... ], "alias": "Usurper", "char_id": 1, "description": ".....", "events": [ ... ], "first_name": "Robert", "gender": "Male", "last_name": "Baratheon", "religion": "Faith of the Seven", "status": "Deceased", "visited": [.... ] }
10. API - Search ❖ Single endpoint with request argument ➢ /api/search?query=<query terms> ❖ Returns a JSON object with ➢ Number of matches ➢ Search term ➢ a list of results
11. Back End Flask-SQLAlchemy models on Postgres DB
12. Back End - Search How we did it ❖ Tokenized search string and built multiple queries for each pillar ❖ Convert results into generic JSON and merge results between pillars ❖ Allows matching on names before descriptions
13. Back End - Testing ❖ Used SQLite for testing models ➢ Orders of magnitude faster using an in-memory only database ➢ Very important when running 70+ tests ❖ Difficulties in running the tests on the live site ➢ Flask-SQLAlchemy doesn’t support sharing models between databases ❖ Used the subprocess module to run the tests in their own process and capture results ❖ In the process of switching to a JSON based test result page
14. Demo http://valar-morghulis.charuri.me/
15. Site Critique What did we do well? ❖ Site is consistent and responsive ❖ API is uniform and clean ❖ API is thoroughly tested
16. Site Critique What did we learn? ❖ Javascript (we should have used Angular!) ❖ Rackspace sign-up ➢ Accidentally signed up for the wrong flavor ❖ Javascript is slow, rendering the template and sending a static page is much faster ➢ (since fixed)
17. Site Critique What could we have done better? ❖ API calls include redundant information ❖ Pagination of pillars and search results ❖ Use junction tables in the searches ❖ Search changes: blacklist grammatical articles, expose more data ❖ Nicer URLs for site pages ➢ ‘/characters/Robert Baratheon’ instead of ‘/characters/1’ ❖ We used javascript to pull data instead of rendering templates ➢ Site was slow and not indexable by external search engines ➢ This was fixed since turn in by using template rendering
18. Site Critique What puzzles us? ❖ Flask-SQLAlchemy not allowing sharing of models between databases ➢ Made testing difficult and non-intuitive ❖ Some features don’t work on some browsers
19. Pythian Critique What did they do well? ❖ Navbar looks nice ❖ Plenty of data ❖ Integration with Google Maps is well-done ❖ Very nice transitions when sorting their pillars ❖ “Featured” items from pillars is a nice way of encouraging exploration of their site. What did we learn from their site? ❖ Random sorting with animation looks impressive
20. Pythian Critique What can they do better? ❖ The font is a bit too small in several places ❖ The athlete names don’t stand out in the athlete list ❖ Page goes mobile too fast ❖ API doesn’t expose a way to get individual elements from the pillars ❖ Pagination would have really helped because they have so much information
21. Pythian Critique What puzzles us about their site? ❖ Inconsistency across pages ❖ Some text doesn’t seem centered in their corresponding elements ❖ Empty spaces in lists depending on sorting