Change order of operations for get_tiles to speed things up (#72)

* Change order of operations for get_tiles to speed things up

* Fix formatting

* Fix bug with crs

* log python versions

* Remove proj:projjson, which is removed in rio-stac 0.10.0

.github/workflows/run-tests.yml

@@ -53,6 +53,7 @@ jobs:
             build-essential
           pip3 install --upgrade pip setuptools wheel poetry
           pip3 install .
+          pip3 freeze
 
       - name: Run tests
         run: |

dep_tools/grids.py

@@ -93,6 +93,7 @@ def get_tiles(
 
 def grid(
     resolution: int | float = 30,
+    simplify_tolerance: float = 0.1,
     crs=PACIFIC_EPSG,
     return_type: Literal["GridSpec", "GeoSeries", "GeoDataFrame"] = "GridSpec",
     intersect_with: GeoDataFrame | None = None,
@@ -121,16 +122,21 @@ def grid(
             return _intersect_grid(full_grid, intersect_with)
         else:
             gridspec = _gridspec(resolution, crs)
-            geometry = Geometry(loads(intersect_with.to_json()))
-            # This is a bit of a hack, but it works. Geometries that are transformed by the tiles_from_geopolygon
-            # are not valid, but doing the simplification and buffer fixes them.
-            buffer = 0.0 if buffer_distance is None else buffer_distance
-            fixed = (
-                geometry.to_crs(PACIFIC_EPSG, check_and_fix=True, wrapdateline=True)
-                .simplify(0.01)
-                .buffer(buffer)
+            simplified = (
+                intersect_with.to_crs(PACIFIC_EPSG)
+                .simplify(simplify_tolerance)
+                .to_frame()
+                .to_geo_dict()
             )
-            return gridspec.tiles_from_geopolygon(geopolygon=fixed)
+            geometry = Geometry(
+                simplified,
+                crs=PACIFIC_EPSG,
+            )
+            if buffer_distance is not None:
+                geometry = geometry.buffer(buffer_distance)
+            else:
+                geometry = geometry.buffer(0.0)
+            return gridspec.tiles_from_geopolygon(geopolygon=geometry)
 
     return {
         "GridSpec": _gridspec,

dep_tools/utils.py

@@ -77,7 +77,7 @@ def bbox_across_180(region: GeoDataFrame | GeoBox) -> BBOX | tuple[BBOX, BBOX]:
         geometry = region.to_crs(4326).geometry.make_valid().explode()
         geometry = geometry[
             geometry.geom_type.isin(["Polygon", "MultiPolygon"])
-        ].unary_union
+        ].union_all()
 
     geometry = _fix_geometry(geometry)
     bbox = antimeridian_bbox(geometry)

tests/test_get_stac_item.py

@@ -1,8 +1,6 @@
 import rioxarray
 
 from dep_tools.namers import LocalPath
-from dep_tools.stac_utils import StacCreator
-from dep_tools.writers import StacWriter
 
 from dep_tools.stac_utils import get_stac_item
 from pathlib import Path
@@ -41,7 +39,6 @@ def test_get_stac_item_properties(stac_item):
         "proj:bbox",
         "proj:shape",
         "proj:transform",
-        "proj:projjson",
     ]
     assert all([key in properties.keys() for key in keys])
 

tests/test_grid.py

@@ -14,7 +14,6 @@ def test_get_gadm():
 
 def test_get_tiles():
     # This takes 2 minutes to retrieve all the tiles. Keep it as a generator.
-    print("a")
     tiles = get_tiles(resolution=30)
 
     # Let's just check one
@@ -23,7 +22,6 @@ def test_get_tiles():
     assert type(tile) is tuple
     assert geobox.crs == PACIFIC_EPSG
 
-    print("b")
     # Check the count here, shouldn't take too long
     tiles = list(get_tiles(resolution=30, country_codes=["FJI"]))
     assert len(tiles) == 27