1

I am trying to build a python script and deploy it as an HTTP function/Serverless cloud function on Pivotal cloud foundry or GKE, but I have gone through several articles and most of them mention using an SA and download Json key, setup env variable to JSON key location and run the script.

But how can I provide local downloaded JSON key file when I deploy it on cloud?

I have gone through below links but I couldn't understand as I am new to GCP, can anyone provide me an elaborated anws on how can I achieve this?

  1. Google Cloud Vision API - Python
  2. Google cloud vision api- OCR
  3. https://cloud.google.com/vision/docs/quickstart-client-libraries#client-libraries-usage-python
Aryyyy26
  • 11
  • 3

2 Answers2

3

According to docs, during function execution, Cloud Functions uses the service account PROJECT_ID@appspot.gserviceaccount.com as its identity. For instance, when making requests to Google Cloud Platform services using the Google Cloud Client Libraries, Cloud Functions can automatically obtain and use tokens to authorize the services this identity has permissions to use.

By default, the runtime service account has the Editor role, which lets it access many GCP services. In your case, you will need to enable the Vision API and grant the default service account with necessary permissions. Check out the Function Identity Docs for more details.

Pubudu Sitinamaluwa
  • 958
  • 7
  • 21
0

Instead of a SA json file, you could use an api key if that's easier for you. However, if you use an api key, you will only be able to send image bytes or specify a public image.

Brendan
  • 1,017
  • 5
  • 7