Tile serving example with FastAPI

[banesullivan]

This is a simple example to show how easy dynamic tile serving can be with large-image - this is effectively all that localtileserver does

import io

from fastapi import FastAPI
from ipyleaflet import Map, TileLayer
import large_image
from starlette.responses import StreamingResponse

from server_thread import ServerThread

src = large_image.open(
    "/Users/bane/Desktop/rasters/TC_NG_SFBay_US_Geo.tif", projection="EPSG:3857", encoding="PNG"
)

app = FastAPI()


@app.get("/metadata")
async def get_metadata():
    return src.getMetadata()


@app.get("/tile/{z}/{x}/{y}.png")
async def get_tile(z: int, x: int, y: int):
    tile_binary = src.getTile(x, y, z)
    return StreamingResponse(io.BytesIO(tile_binary), media_type="image/png")


# Launch app in a background thread
server = ServerThread(app)

t = TileLayer(
    url=f"http://{server.host}:{server.port}/tile/{{z}}/{{x}}/{{y}}.png",
    attribution="booya!",
)

m = Map(zoom=8, center=[37.8751050766018, -122.19290725886823])
m.add_layer(t)
m

[vincentsarago]

that's really cool 🙏

[banesullivan]

FYI, I've been playing around with Tornado in large-image, and with JupyterLab, we can avoid server-thread: girder/large_image#1065

[vincentsarago]

nice, I'm using server-thread in rio-viz https://github.com/developmentseed/rio-viz/blob/main/rio_viz/app.py#L821-L840

https://github.com/developmentseed/rio-viz/tree/main#in-notebook-environment

[banesullivan]

I really wish ipyleaflet could just accept a Python callable (e.g. the tile() method) and send the image binary contents via its comms

[banesullivan]

Oh that's awesome!! I hadn't seen rio-viz, will definitely try it out

[vincentsarago]

well we could also use geotiff js instead of python... but where is the fun of that !