Add kerchunk and fastparquet to requirements; enhance xarray function…

… with dynamic kwargs

fly.toml

@@ -12,7 +12,7 @@ buildpacks = ["heroku/buildpack-python:0.17.0"]
 
 [[vm]]
 size = "shared-cpu-1x"
-memory = "2048mb"
+memory = "4096mb"
 
 
 [env]

requirements.txt

@@ -10,3 +10,5 @@ xarray>=2022.9
 zarr
 fastapi-cache2 @ git+https://github.com/andersy005/fastapi-cache@pydantic-v2-compat # for pydantic v2 compatilibity
 universal_pathlib
+kerchunk >=0.2.6
+fastparquet

src/app.py

@@ -2,6 +2,7 @@
 
 import os
 import traceback
+import typing
 from contextlib import asynccontextmanager
 
 import pydantic
@@ -15,6 +16,12 @@
 from .helpers import sanitize_url
 from .log import get_logger
 
+
+class XarrayOpenKwargs(pydantic.BaseModel):
+    engine: typing.Literal['zarr', 'kerchunk'] = 'zarr'
+    chunks: dict = {}
+
+
 origins = ['*']
 
 logger = get_logger()
@@ -75,8 +82,15 @@ async def xarray(
         description='URL to a zarr store',
         example='https://ncsa.osn.xsede.org/Pangeo/pangeo-forge/HadISST-feedstock/hadisst.zarr',
     ),
+    xarray_open_kwargs: str = Query(
+        default='{"engine": "zarr", "chunks": {}}',
+        description='Keyword arguments for xr.open_dataset',
+    ),
 ):
-    logger.info(f'🚀 Starting to process request for URL: {url}')
+    kwargs_model = XarrayOpenKwargs.model_validate_json(xarray_open_kwargs)
+    logger.info(
+        f'🚀 Starting to process request for URL: {url} with kwargs: {kwargs_model.model_dump()}'
+    )
 
     import time
 
@@ -93,7 +107,7 @@ async def xarray(
 
     try:
         logger.info(f'📂 Attempting to open dataset from URL: {url}')
-        ds = xr.open_dataset(sanitized_url, engine='zarr', chunks={})
+        ds = xr.open_dataset(sanitized_url, **kwargs_model.model_dump())
         logger.info('✅ Successfully opened dataset. Generating HTML representation.')
         html = ds._repr_html_().strip()