Skip to content

Latest commit

 

History

History
executable file
·
68 lines (40 loc) · 3.45 KB

README.md

File metadata and controls

executable file
·
68 lines (40 loc) · 3.45 KB
Raw

Quick, Draw Data API

In order to self-host, it's easiest to use Google Cloud Platform. Create a GCP project if do not have one already for your projct. The following instructions assume you have your GCP Project ID handy.

Copy the Quick, Draw! data

The bucket contains 191.63 GB of doodles. Assuming multiregional storage, and at the time of writing this, your cost would be ($0.026 * 191.63) = $4.98 per month. Storage and transferring between GCS buckets may vary based on your bucket location (it's free to transfer in the same region, bucket to bucket). You can find more details here.

  1. In your GCP Project, go to Storage > Browser and create a bucket with a unique id of your choosing (leave all other default settings)
  2. Go to Storage > Transfer
  3. Select "Google Cloud Storage bucket" as the source, and enter gs://quickdraw-data-complete and the Cloud Storage bucket source. Press continue.
  4. Enter gs://YOUR_BUCKET_ID where YOUR_BUCKET_ID is the unique id you chose in the first step. Leave all other default settings and press continue.
  5. Leave the default transfer settings to run now, and press "Create". The files will now transfer over.

Running the API locally

  1. Run npm install in this directory

  2. In the root of this API directory, create a config.json file replacing the values below (GCS_BUCKET_ID is the bucket you created when copying the data over in the above step) :

     {
   "projectId": "YOUR_PROJECT_ID",
   "bucketId": "GCS_BUCKET_ID"
 }

NOTE: If you secure your bucket to certain roles, you can add your service account credential key in this config file by adding: "keyFilename": "./PATH_TO_SERVICE_ACCOUNT_KEY"

You can now run npm start and the server should run locally on your machine at http://localhost:8080. You can then try sending a request using the instructions below.

Send a request

Choose your local or production server:

# If you're running locally, you won't need an API key.
$ export ENDPOINTS_HOST=http://localhost:8080

$ export ENDPOINTS_HOST=https://YOUR_PROJECT_ID.appspot.com
$ export ENDPOINTS_KEY=AIza...

Send the request:

$ curl -vv -H 'Content-Type: application/json' "${ENDPOINTS_HOST}/drawing/cat/count?key=${ENDPOINTS_KEY}"

If you're running locally, you won't need an API key.

Deploying the API

  1. You will need to make sure you have Python 2.7.x installed as well as the Google Cloud SDK on your local machine.

  2. Change __app.yaml to app.yaml and inside the file, update name: YOUR_PROJECT_ID.appspot.com replacing YOUR_PROJECT_ID with the ID of the project you created in GCP where the files live.

  3. Change __openapi-appengine.yaml to openapi-appengine.yaml and inside the file, update host: "YOUR_PROJECT_ID.appspot.com" replacing YOUR_PROJECT_ID with the same project ID.

  4. Run npm run deploy-spec to deploy your API definition file

  5. Run npm run deploy to deploy your application code

License

The component & API fall under the Apache 2.0 license.

This data is made available by Google, Inc. under the Creative Commons Attribution 4.0 International license.