Observe API service

[batpad]

Service that actions can call to talk to the Observe API.

Currently, implements a base callAPI function that other API methods can call - handles setting the authorization header and passing parameters for GET and POST requests.

Sets up a simple test to call the getProfile action that currently just console.logs the user profile information fetched from the API.

Next steps:

• Wire this up to the action to submit traces, and sketch out what the actions, reducers and error handling looks like.

cc @vgeorge @geohacker

[vgeorge]

@batpad, developmentseed/observe-dashboard#16 has a working implementation of async actions based the framework conceived by Daniel. It also includes a wrapper function to make authenticated requests seamlessly and a basic example on how to read data and handle errors.

[batpad]

Have wired up the POST to post new traces to the API. It currently stores errors in an errors array in the trace object in the state. We can probably refine the error types and messages when dealing with displaying traces on the list page.

I'll add some tests for different status responses from the API. Let me know if there are thoughts / questions on the code so far, @geohacker @vgeorge @sethvincent .

TODO:

• Tests for error responses from the API
• Clean up extraneous console.logs

[sethvincent]

@batpad I pushed a commit with some small changes needed to get things running.

I can login fine. It seems like I'm getting errors sending photos to the api:

response data {"error": "Bad Request", "message": "Invalid request payload input", "statusCode": 400} 400

Not sure yet if this is an api problem or an app problem. I can follow up with more debugging tomorrow.

[batpad]

@sethvincent oops sorry I wasn't more clear -- this PR only includes the actions to POST Traces. If the structure looks okay and that works, I thought I'd work on photos in a separate PR.

[batpad]

@sethvincent thanks for the catch on persisting the token.

[sethvincent]

@batpad oh ok that makes sense. I'll take another quick look. I feel like what's here is pretty great.

[sethvincent]

I'm getting an error when posting to observe-api:

LOG  response data {"error": "Bad Request", "message": "Invalid request payload input", "statusCode": 400} 400
LOG  error [TypeError: undefined is not an object (evaluating 'data.properties.id')]

I think because properties.id isn't defined in the validation for the traces POST endpoint? I'm not really sure that makes sense though. I don't see an error logged on the api output. @vgeorge @batpad

[sethvincent]

My original idea was not correct. The error I was seeing was because the properties object doesn't exist sometimes when adding the id.

[batpad]

Hmmm it seems a bit strange that the properties object would not exist - this maybe better to ensure a bit further up when the trace is being created. Let me look a bit closer. Thanks for the catch / fix.

[sethvincent]

I resolved the above error and it seems now that if a trace description is an empty string that i get an Invalid request payload input error. I'm not sure the empty string is the problem, but it definitely works so far when I include a description. This seems like it may be more on the API side.

A higher-level issue I've noticed is that if the server restarts after making changes the next time I try to post a trace the app will crash and stay in that state until i delete it and reinstall. Presumably something similar would happen anytime the access token would expire? We may need to make an issue for this and tackle it in a follow-up pr if not in this one.

[batpad]

I resolved the above error and it seems now that if a trace description is an empty string that i get an Invalid request payload input error. I'm not sure the empty string is the problem, but it definitely works so far when I include a description. This seems like it may be more on the API side.

This looks fixed by developmentseed/observe-api#32

A higher-level issue I've noticed is that if the server restarts after making changes the next time I try to post a trace the app will crash and stay in that state until i delete it and reinstall. Presumably something similar would happen anytime the access token would expire? We may need to make an issue for this and tackle it in a follow-up pr if not in this one.

I have made an issue for this here: #132

I'm a bit conflicted on whether to tackle this in this PR or as the next follow-up. I can't quite wrap my head around the cleanest way to handle token expiry, and also think it may need some UI pieces figured out.

I'm going to hold off on merging this just yet though, because this is a pretty bad issue, and we should think this through.

Thanks a lot for the fixes and review, @sethvincent .

[batpad]

I added tests for the reducer, and also changed to using a single status variable to hold the pending, uploading, or uploaded status of the feature, similar to edits.

@geohacker - if this looks good, I think we can merge. And then I can start working on the API token expiry in a separate PR.

[geohacker]

I tested this on iOS and seemed to work well. I think we should test again after building the Your Traces screen. I got a promise rejection warning when I tried to 'Get Observe API profile' in the settings -- might be something minor but just noting.