Cannot perform search about serverless-photo-recognition HOT 11 CLOSED

To fix this you'll need to go to the console, bring up the api, and select the POST method. Click the little pencil icon and reselect the Lambda function -- that will add the necessary permissions for it to be run. I'm looking for ways to automate this step (I might have found it). You'll need to redeploy the api once you reselect the Lambda function.

from serverless-photo-recognition.

Ok. Thanks -- not sure I would have come up with that on my own :)

I made the change in the API console as you describe and I'm still getting {"message":"Unauthorized"} when I attempt the curl statement with the pool-id and id token from the Cognito Startup webpage -- same when I run from the API Gateway Test.

However, I added to the setup script to sign-up and confirm a user in the same cognito identity pool created in the setup script (I'll share that in a pull request soon). From there I begin the authflow to fetch the id token. Using the id token and the identity pool, I'm uploading a picture and running the curl command -- authorization works just fine, the lambda function returns without errors. Here's the response received:

{"message":"Success","pictures":[]}

Getting closer -- I'm looking at the SearchPhotosHandler.kt file to see if I can track this down . . . no errors are being reported in the log file for the function.

Let me know what you think?

Also, fwiw, after I made the change to the API as you had suggested, I inadvertently forgot to redeploy the API -- and I received the success message. Perhaps the API doesnt have to be re-deployed after all??

from serverless-photo-recognition.

from serverless-photo-recognition.

The user is created with the same coginito id as the one being used in the S3 URI -- thats how it should be correct? I've confirmed via the console that the created user is in cognito pool.

I'll open a pull request so you can see how the user is being created and how I'm using it.

Thanks

from serverless-photo-recognition.

I can't replicate the isssue -- searching with the correct id works for me (as long as you follow the process of re-enabling the Lambda function, which I'm trying to automate).

from serverless-photo-recognition.

I've pushed the fix to grant api gateway the permissions to kick off the lambda function...you no longer need to reselect the function once deployed.

from serverless-photo-recognition.

fwiw, here's my experience -- running on the awslabs/master branch

Startup and login to Cognito Starter web app, copy 'Cognito id' from 'My Profile' tab

Copy an image to the S3 bucket:

 ~/Projects/AWS-SPR/setup (master)*$ aws s3 cp new-york.jpg s3://rekognition-20170226092810/usercontent/<Cognito ID copied and pasted from Cognito Starter My Profile tab>/

upload: ./new-york.jpg to s3://rekognition-20170226092810/usercontent/<snip>/new-york.jpg

Run the curl command using the ID Token from the JWT Token tab on the Cognito Starter web app

 ~/Projects/AWS-SPR/setup (master)*$ curl -X POST -H "Authorization: <ID Token copied and pasted from Cognito Starter webpage>" -H "search-key: building" -H "Cache-Control: no-cache" "https://f61jdzdb2d.execute-api.us-east-1.amazonaws.com/prod/picture/search/"

{"message":"Unauthorized"}

from serverless-photo-recognition.

Have you modified the values in the ~/aws-cognito-angular2-quickstart/src/environments/environment.ts file to reflect the newly-created cognito setup?

from serverless-photo-recognition.

Frankly, seeing how difficult it is, I'm thinking of taking some of the code you have to enable SRP and user creation automation to run through the whole scenario without needing the angular project.

from serverless-photo-recognition.

from serverless-photo-recognition.

from serverless-photo-recognition.