Python API from OpenAPI endpoints

[banesullivan]

We need to (automatically?) generate a Python API from the OpenAPI endpoints. Or at least provide a way to access those endpoints from a Python client.

One viable option is: https://github.com/OpenAPITools/openapi-generator

Another approach would be to manually create the API... I'm actually leaning towards this to make sure we can design the API in a way that is user friendly and have clear documentation

First, we need a few clear use cases, likely coming from ndsampler

[banesullivan]

Another route is: https://github.com/swagger-api/swagger-codegen

[banesullivan]

We need to be able to perform the following tasks from the Python client:

• Search raster data for a given ROI (pass GeoJSON or bounding box coordinates) and other parameters in the search filter endpoint
  • Add additional cloud coverage param when Add cloud coverage to RasterMetaEntry #304 / Add cloud_cover raster meta field #310 lands
• Download original imagery
  • Hits download_image_entry_file
• Use bounding box to extract ROI from image and download that ROI
  • Requires #295

cc @aashish24 @AlmightyYakob

[banesullivan]

Use case workflow

1. Login into RGD via the Python client
2. Perform a search that returns a list of results
3. Grab 1 RasterEntry from list of results
4. Load raster with rasterio and show/plot with matplotlib
5. Use GeoJSON/BBox to subsample raster (hits the POST endpoint)
6. Wait until subsampling is completed
7. Fetch subsampled raster and view next to original raster

Psuedocode

See demos/GeoData-API.ipynb for inspiration.

import rasterio
from rasterio.plot import show
import matplotlib.pyplot as plt

# Define ROI
points = np.array([[-107.08212524738178, 39.01040379702808], 
                   [-106.96182164246767, 39.03110215679572], 
                   [-106.90895466037738, 38.98387516880551], 
                   [-106.9805540376965, 38.91038429753703], 
                   [-107.07130208569401, 38.952157178475225], 
                   [-107.08212524738178, 39.01040379702808]])

shape = {"type": "Polygon", 
         "coordinates": [points.tolist(), ]}

...

q = GeoData.search(geojson=shape, **params)
raster = q[0]  # choose raster from search results

f = GeoData.fetch_raster_file(raster)

src = rasterio.open(f)
show(src) # plots raster

sub = GeoData.subsample_raster(raster, geojson=shape)
src_sub = rasterio.open(sub)

# Show two in same plot
show(src, cmap='Greys')
show(src_sub)