mammuth / alexa-mensa-munich Goto Github PK

View Code? Open in Web Editor NEW

10.0 3.0 0.0 1.94 MB

Alexa skill for Munich's canteens :spaghetti: :fork_and_knife: :apple:

License: Apache License 2.0

JavaScript 100.00%

canteen mensa tum lmu alexa open-source alexa-skill munich university

alexa-mensa-munich's Introduction

Mensa München Alexa Skill

An Amazon Alexa skill that tells you the menu of all canteens in Munich. The skill uses the API of openmensa.org

Status

WIP
Not all canteens supported yet.
German only. English will follow
Buggy as shit 🙈

Usage

"Alexa, öffne Mensa München"
"Alexa, frage Mensa München was gibt es am Mittwoch in Garching?"
"Alexa, frage Mensa München nach dem Speiseplan der Mensa Leopoldstraße"
"Mahlzeiten Stammgelände"
"Speiseplan der Mensa in Großhadern am Freitag"
"Mensa Leopoldstraße"

Installation

Install the "Mensa München" skill from the Alexa Skill store https://www.amazon.de/Maximilian-Muth-Mensa-M%C3%BCnchen/dp/B06XYY4Q93/

Supported Canteens

TUM Mensa Garching
Leopoldstraße
Arcisstraße (theoretically, but alexa always understands something different which makes it unusable...)
Martinsried
Großhadern

Currently it only supports 'Mensen', not StuCafe's (#15)

Development

There are some open issues, just pick one and open a PR. I'd love to build this together with others, so go for it! 💪

alexa-mensa-munich's People

Contributors

Stargazers

Watchers

alexa-mensa-munich's Issues

Support English

English speech output
Moment.js locale
English utterances

Finetune utterances

utterances are not really thought through and could probably be improved.

Support all Munich canteens

Use a custom slot for all available canteens in Munich. Currently the canteen in Garching is hardcoded.

How to develop locally?

What tools can be used to test and run the skill without uploading it. Any changes to the project structure needed?

One-shot questions not working on device

"Alexa, frage Mensa München was es morgen zu essen gibt" seems not to work

Add a test suite

Find out how to properly test a skill
German and English
...

Improve canteen custom slot

Currently there are only a couple of canteens in the custom slot type.

How does it behave if the user says "Mensa in Garching" instead of "Mensa Garching"?
Currently canteens are determined by string comparison. Should be improved.

Flash Briefing capabilities

It might be useful to have this skill as a flash briefing, which tells you the meals of the most used (or configured) canteen.

Add support for StuCafe's

Alexa doesn't understand StuCafe and thinks its stuff like "Stuck Kaffee".

Missing:
145 StuCafe Boltzmannstraße, Maschinenwesen
230 StuCafe Adalbertstraße
136 StuBistro Schellingstraße
232 StuCafe Karlstraße

Related #6

Fallback to most used canteen if none is provided

I guess most people have only one or two canteens they're visiting regularly. As a result it might make sense to store the canteens that were questioned. Maybe we should also count how often each canteen was accessed.

This is useful, because we could then automatically suggest this canteen in case the user forget to mention it and eg. just asked "What to eat today?".

Add FMI Bistro

Maybe really write a parser for openmensa. See https://github.com/mammuth/TUM.sexy/issues/35

Crash when there are no meals at the requested day

Scenario:

It's friday
Say "Alexa, frage Mensa München 'was gibt es morgen zu essen'"?

API returns just nothing (nice design, openmensa 🙈):
https://openmensa.org/api/v2/canteens/139/days/2017-03-25/meals

The skill doesn't handle an empty response properly.

Which open mensa categories should be included?

Seems like there is not only a 'Tagesgericht X', 'Aktionsgericht X', but also eg. 'Biogericht X'.

How to handle 'Aktion'? It is sometimes a real meal, but most of the time just a supplement.

Eg. https://openmensa.org/api/v2/canteens/139/days/2017-03-23/meals

Add fancy interjections

https://developer.amazon.com/public/solutions/alexa/alexa-skills-kit/docs/speechcon-reference
Those funny / living interjections could be used to improve the UX by providing a more interesting character to our skill.

Currently US-only. But we could download & host the sounds and play them with the audio player.

Optimize error handling

Give the user better feedback what exactly did not work.

No canteen detected
No canteen matching

Maybe even ask for the wished one. Is there a session needed therefor?

Current behavior: If the skill can't detect a mensa, it simply triggers the help intent

Which API to use?

I just started with using openmensa.org, but there might be some alternatives, mainly the API of the TUM Campus App (it's undocumented, check out the Android client for details: https://github.com/TCA-Team/TumCampusApp).

API related issues: #11, #1