Unable to use presigned S3 URLs

[corral0824]

Hi Titiler community! First thanks for all the good work you have been up to. Titiler is an amazing tool.

DEPLOYMENT:

1. pgSTAC deployed locally.
2. STAC FAST API deployed locally
3. Titiler deployed locally
4. Filmdrop deployed locally.

CONTEXT:

I was firstly trying to set up Element84's Filmdrop in a STAC I have deployed locally and came across some issues when rendering COGs with Titiler. The URL that was not working was the following

http://127.0.0.1:8002/stac/tiles/9/137/203@1x.png?url=http://localhost:8000/collections/SATELLITE/items/ITEMID&assets=cog&color_formula=Gamma+RGB+2.2+Saturation+0.6+Sigmoidal+RGB+10+0.2

Note that I was running Titiler in port 8002 as STAC was running in port 8000.

As I could not make it work, I dag deep into the issue, trying to first visualize a preview using my local titiler deployment, and finally managed to at least load a preview using titiler, the following way

http://127.0.0.1:8001/collections/SATELLITE/items/ITEMID/preview?assets=cog&bidx=3&bidx=2&bidx=1&rescale=628,1999&rescale=940,2124&rescale=1000,2105&max_size=1024

I also proved that the presigned URLs allowed range requests by making the following curls

curl --output - -r 0-1000 "PRESIGNED URL"

However, when trying to visualize the cog using the documentations endpoints

http://127.0.0.1:8002/cog/WebMercatorQuad/tilejson.json?url={AWS_PRESIGNED_URL}

I get an error response

{"detail":"HTTP response code: 400"}

I managed to make it work and visualize a map with my COG using python and your working with COG example using httpx:

https://developmentseed.org/titiler/examples/notebooks/Working_with_CloudOptimizedGeoTIFF_simple/

However I don't manage to replicate your working with signed URLs example. For instance I have tried doing

`import base64
import requests

signed = "LONG_SIGNED_S3_URL"

b64 = base64.b64encode(signed.encode()).decode("utf-8")
value = f"b64:{b64}"

resp = requests.get(
"http://127.0.0.1:8002/cog/info",
params=[("url", value)]
)

print(resp.json())`

but this will return that the query is too long:

File name too long

QUESTIONS:

1. Is there a way to make this scenario work?
2. Is there a way to visualize my COGs served using a private AWS PRESIGNED URL using TITILER?
3. Ideally, I want to be able to consume this via http query. Is this a possibility?

Thanks for your attention and help.