🥚 Allow using XpySTACAssetReader without xpystac when engine!=stac (#100

)

* 🥚 Allow using XpySTACAssetReader without xpystac when engine!=stac

A little hidden feature to use `read_from_xpystac` with other engines (e.g. `netcdf4`, `h5netcdf`) without having to install `xpystac`! Added a unit test using `engine="rasterio"` which is technically deprecated, but works without having to install extra dependencies.

* ✏️ Remove mention of the STAC_URL environment variable

The STAC_URL environment variable never actually worked, see stac-utils/pystac-client#317. Also updated some links on the walkthrough to point to torchdata 0.6.1.

docs/walkthrough.md

@@ -84,14 +84,14 @@ This is how the Sentinel-2 image looks like over Singapore on 15 Jan 2022.
 
 ![Sentinel-2 image over Singapore on 20220115](https://planetarycomputer.microsoft.com/api/data/v1/item/preview.png?collection=sentinel-2-l2a&item=S2A_MSIL2A_20220115T032101_R118_T48NUG_20220115T170435&assets=visual&asset_bidx=visual%7C1%2C2%2C3&nodata=0)
 
-## 1️⃣ Construct [DataPipe](https://github.com/pytorch/data/tree/v0.4.0#what-are-datapipes) 📡
+## 1️⃣ Construct [DataPipe](https://github.com/pytorch/data/tree/v0.6.1#what-are-datapipes) 📡
 
 A torch `DataPipe` is a way of composing data (rather than inheriting data).
 Yes, I don't know what it really means either, so here's some extra reading.
 
 🔖 References:
 - https://pytorch.org/blog/pytorch-1.11-released/#introducing-torchdata
-- https://github.com/pytorch/data/tree/v0.4.0#what-are-datapipes
+- https://github.com/pytorch/data/tree/v0.6.1#what-are-datapipes
 - https://realpython.com/inheritance-composition-python
 
 ### Create an Iterable 📏

zen3geo/datapipes/pystac_client.py

@@ -39,8 +39,7 @@ class PySTACAPISearcherIterDataPipe(IterDataPipe):
           provided Collections will be searched.
 
     catalog_url : str
-        The URL of a STAC Catalog. If not specified, this will use the
-        ``STAC_URL`` environment variable.
+        The URL of a STAC Catalog.
 
     kwargs : Optional
         Extra keyword arguments to pass to

zen3geo/datapipes/xpystac.py

@@ -122,7 +122,7 @@ def __init__(
         engine: str = "stac",
         **kwargs: Optional[Dict[str, Any]]
     ) -> None:
-        if xpystac is None:
+        if xpystac is None and engine == "stac":
             raise ModuleNotFoundError(
                 "Package `xpystac` is required to be installed to use this datapipe. "
                 "Please use `pip install xpystac` "

zen3geo/tests/test_datapipes_xpystac.py

@@ -6,16 +6,16 @@
 
 from zen3geo.datapipes import XpySTACAssetReader
 
-pystac = pytest.importorskip("pystac")
-xpystac = pytest.importorskip("xpystac")
-
 
 # %%
 def test_xpystac_asset_reader_cog():
     """
     Ensure that XpySTACAssetReader works to read in a pystac.Asset object
     stored as a Cloud-Optimized GeoTIFF and output to an xarray.Dataset object.
     """
+    pystac = pytest.importorskip("pystac")
+    xpystac = pytest.importorskip("xpystac")
+
     item_url: str = "https://github.com/stac-utils/pystac/raw/v1.7.1/tests/data-files/raster/raster-sentinel2-example.json"
     asset: pystac.Asset = pystac.Item.from_file(href=item_url).assets["overview"]
     assert asset.media_type == pystac.MediaType.COG
@@ -43,6 +43,9 @@ def test_xpystac_asset_reader_zarr():
     Ensure that XpySTACAssetReader works to read in a pystac.Asset object
     stored as a Zarr file and output to an xarray.Dataset object.
     """
+    pystac = pytest.importorskip("pystac")
+    xpystac = pytest.importorskip("xpystac")
+
     collection_url: str = "https://planetarycomputer.microsoft.com/api/stac/v1/collections/daymet-daily-hi"
     asset: pystac.Asset = pystac.Collection.from_file(href=collection_url).assets[
         "zarr-https"
@@ -65,3 +68,35 @@ def test_xpystac_asset_reader_zarr():
     assert dataset.rio.bounds() == (-5802750.0, -622500.0, -5518750.0, -38500.0)
     assert dataset.rio.resolution() == (1000.0, -1000.0)
     assert dataset.rio.grid_mapping == "lambert_conformal_conic"
+
+
+def test_xpystac_asset_reader_geotiff_without_xpystac():
+    """
+    Ensure that XpySTACAssetReader works to read in a GeoTIFF file and output
+    to an xarray.Dataset object, even when xpystac is not installed.
+
+    Note that `engine="rasterio"` has been removed in xarray v2023.04.0, see
+    https://github.com/pydata/xarray/pull/7671. So, this test will need to be
+    updated once we change to require an xarray verson greater than 2023.04.0.
+    Only included this test to check an alternative to `engine="stac"` that
+    did not require installing extra required dependencies like `netcdf4` or
+    `h5netcdf`.
+    """
+    tif_url: str = "https://github.com/corteva/rioxarray/raw/0.14.1/test/test_data/input/cint16.tif"
+
+    dp = IterableWrapper(iterable=[tif_url])
+
+    # Using class constructors
+    dp_xpystac = XpySTACAssetReader(source_datapipe=dp, engine="rasterio")
+    # Using functional form (recommended)
+    dp_xpystac = dp.read_from_xpystac(engine="rasterio")
+
+    assert len(dp_xpystac) == 1
+    it = iter(dp_xpystac)
+    dataset = next(it)
+
+    assert dataset.sizes == {"band": 1, "x": 100, "y": 100}
+    assert dataset.band_data.dtype == "complex64"
+    assert dataset.rio.bounds() == (0.0, 100.0, 100.0, 0.0)
+    assert dataset.rio.resolution() == (1.0, 1.0)
+    assert dataset.rio.crs == "EPSG:4326"