Merge branch 'master' into fix/invalid-collection-tde-1216

.github/workflows/containers.yml

@@ -63,7 +63,7 @@ jobs:
 
       - name: Setup docker tags
         id: tags
-        uses: actions/github-script@d7906e4ad0b1822421a7e6a35d5ca353c962f410 # v6
+        uses: actions/github-script@60a0d83039c74a4aee543508d2ffcb1c3799cdea # v6
         with:
           result-encoding: string
           script: |

.github/workflows/release-please.yml

@@ -77,7 +77,7 @@ jobs:
 
       - name: Setup docker tags
         id: tags
-        uses: actions/github-script@d7906e4ad0b1822421a7e6a35d5ca353c962f410 # v6
+        uses: actions/github-script@60a0d83039c74a4aee543508d2ffcb1c3799cdea # v6
         with:
           result-encoding: string
           script: |

Dockerfile

@@ -1,4 +1,4 @@
-FROM ghcr.io/osgeo/gdal:ubuntu-small-3.9.0@sha256:d1a38af532e5d9e3991c4a6bddc2f2cb52644dc30a4eb8242101e8e23c3f83f6 as builder
+FROM ghcr.io/osgeo/gdal:ubuntu-small-3.9.0@sha256:d1a38af532e5d9e3991c4a6bddc2f2cb52644dc30a4eb8242101e8e23c3f83f6 AS builder
 
 # Avoid blocking `apt-get install` commands
 ARG DEBIAN_FRONTEND=noninteractive

pyproject.toml

@@ -57,5 +57,7 @@ pytest-dependency = "^0.5.1"
 pytest-mock = "^3.12.0"
 pytest-subtests = "*"
 shellcheck-py = "*"
+types-python-dateutil = "*"
 types-shapely = "*"
+types-urllib3 = "*"
 vulture = "^2.10"

scripts/collection_from_items.py

@@ -1,6 +1,8 @@
 import argparse
 import json
 import os
+from argparse import Namespace
+from typing import List
 
 import shapely.geometry
 import shapely.ops
@@ -17,10 +19,7 @@
 from scripts.stac.imagery.provider import Provider, ProviderRole
 
 
-# TODO: Refactor this function to avoid these pylint disable comments and improve testability
-# pylint: disable=too-many-locals
-# pylint: disable=too-many-statements
-def main() -> None:
+def parse_args(args: List[str] | None) -> Namespace:
     parser = argparse.ArgumentParser()
     parser.add_argument("--uri", dest="uri", help="s3 path to items and collection.json write location", required=True)
     parser.add_argument("--collection-id", dest="collection_id", help="Collection ID", required=True)
@@ -87,7 +86,12 @@ def main() -> None:
         "--concurrency", dest="concurrency", help="The number of files to limit concurrent reads", required=True, type=int
     )
 
-    arguments = parser.parse_args()
+    return parser.parse_args(args)
+
+
+# pylint: disable=too-many-locals
+def main(args: List[str] | None = None) -> None:
+    arguments = parse_args(args)
     uri = arguments.uri
 
     providers: list[Provider] = []

scripts/stac/imagery/capture_area.py

@@ -4,6 +4,7 @@
 from linz_logger import get_log
 from shapely import BufferCapStyle, BufferJoinStyle, to_geojson, union_all
 from shapely.geometry.base import BaseGeometry
+from shapely.ops import orient
 
 from scripts.logging.time_helper import time_in_ms
 
@@ -76,8 +77,11 @@ def merge_polygons(polygons: Sequence[BaseGeometry], buffer_distance: float) ->
     # Negative buffer back in the polygons
     union_unbuffered = union_buffered.buffer(-buffer_distance, cap_style=BufferCapStyle.flat, join_style=BufferJoinStyle.mitre)
     union_simplified = union_unbuffered.simplify(buffer_distance)
+    # Apply right-hand rule winding order (exterior rings should be counter-clockwise) to the geometry
+    # Ref: https://datatracker.ietf.org/doc/html/rfc7946#section-3.1.6
+    oriented_union_simplified = orient(union_simplified, sign=1.0)
 
-    return union_simplified
+    return oriented_union_simplified
 
 
 def generate_capture_area(polygons: Sequence[BaseGeometry], gsd: float) -> dict[str, Any]:

scripts/stac/imagery/tests/capture_area_test.py

@@ -1,4 +1,7 @@
-from shapely.geometry import Polygon
+from typing import cast
+
+from shapely import get_exterior_ring, is_ccw
+from shapely.geometry import MultiPolygon, Polygon, shape
 
 from scripts.stac.imagery.capture_area import generate_capture_area, merge_polygons, to_feature
 
@@ -121,3 +124,68 @@ def test_generate_capture_area_not_rounded() -> None:
     # This gap should not be covered in the capture-area
     # as the GSD is 0.2m * a buffer of 2 * 2 < 0.88m
     assert capture_area_expected != capture_area_result
+
+
+def test_capture_area_orientation_polygon() -> None:
+    # Test the orientation of the capture area
+    # The polygon capture area should be the same as the polygon and not reversed
+    polygons = []
+    polygons.append(
+        shape(
+            {
+                "type": "MultiPolygon",
+                "coordinates": [
+                    [
+                        [
+                            [174.673418475601466, -37.051277768264598],
+                            [174.673425023818197, -37.051550327851878],
+                            [174.673479832051271, -37.051280958541774],
+                            [174.673418475601466, -37.051277768264598],
+                        ]
+                    ]
+                ],
+            }
+        )
+    )
+    capture_area = generate_capture_area(polygons, 0.05)
+    assert is_ccw(get_exterior_ring(shape(capture_area["geometry"])))
+
+
+def test_capture_area_orientation_multipolygon() -> None:
+    # Test the orientation of the capture area
+    # The multipolygon capture area polygons should be the same as the polygon and not reversed
+    polygons = []
+    polygons.append(
+        shape(
+            {
+                "type": "MultiPolygon",
+                "coordinates": [
+                    [
+                        [
+                            [175.01786467173446, -36.80914409510033],
+                            [175.0154395531313, -36.80918523526587],
+                            [175.01524959114974, -36.80950706997984],
+                            [175.01312143626905, -36.81369683085496],
+                            [175.01801611134528, -36.81491172712502],
+                            [175.01786467173446, -36.80914409510033],
+                        ]
+                    ],
+                    [
+                        [
+                            [174.99635270693256, -36.809507300287486],
+                            [174.99160937900407, -36.80958686195598],
+                            [174.9916815185804, -36.80966722253352],
+                            [174.9909875028488, -36.810093453911804],
+                            [174.99113976763834, -36.815970642382126],
+                            [174.99336131537976, -36.81582975812777],
+                            [174.99645096077896, -36.81328982536313],
+                            [174.99635270693256, -36.809507300287486],
+                        ]
+                    ],
+                ],
+            }
+        )
+    )
+    capture_area = generate_capture_area(polygons, 0.05)
+    mp_geom = cast(MultiPolygon, shape(capture_area["geometry"]))
+    assert is_ccw(get_exterior_ring(mp_geom.geoms[0]))

scripts/stac/imagery/tests/collection_test.py

@@ -310,8 +310,7 @@ def test_capture_area_added(metadata: CollectionMetadata, subtests: SubTests) ->
 
     with subtests.test():
         assert collection.stac["assets"]["capture_area"]["file:checksum"] in (
-            "1220b15694be7495af38e0f70af67cfdc4f19b8bc415a2eb77d780e7a32c6e5b42c2",  # geos 3.11
-            "122040fc8700d5d2d04600f730e10677b19d33f3b1e43b02c7867f4cfc2101930863",  # geos 3.12
+            "1220369cd5d4179f5f68ca0fd9be70b9f66033fcc6bb2f3305c0ad977adc79d7ad53",  # geos 3.11 - geos 3.12 as yet untested
         )
 
     with subtests.test():

scripts/tests/collection_from_items_test.py

@@ -0,0 +1,60 @@
+from boto3 import client, resource
+from moto import mock_aws
+from moto.s3.responses import DEFAULT_REGION_NAME
+
+from scripts.collection_from_items import main
+from scripts.datetimes import utc_now
+from scripts.files.fs_s3 import write
+from scripts.json_codec import dict_to_json_bytes
+from scripts.stac.imagery.item import ImageryItem
+
+
+@mock_aws
+def test_should_create_collection_file() -> None:
+    # Mock AWS S3
+    s3 = resource("s3", region_name=DEFAULT_REGION_NAME)
+    boto3_client = client("s3", region_name=DEFAULT_REGION_NAME)
+    s3.create_bucket(Bucket="stacfiles")
+    # Create mocked STAC Item
+    item = ImageryItem("123", "./scripts/tests/data/empty.tiff", utc_now)
+    geometry = {
+        "type": "Polygon",
+        "coordinates": [[1799667.5, 5815977.0], [1800422.5, 5815977.0], [1800422.5, 5814986.0], [1799667.5, 5814986.0]],
+    }
+    bbox = (1799667.5, 5815977.0, 1800422.5, 5814986.0)
+    start_datetime = "2021-01-27T00:00:00Z"
+    end_datetime = "2021-01-27T00:00:00Z"
+    item.update_spatial(geometry, bbox)
+    item.update_datetime(start_datetime, end_datetime)
+    item.add_collection("abc")
+    write("s3://stacfiles/item.json", dict_to_json_bytes(item.stac))
+    # CLI arguments
+    args = [
+        "--uri",
+        "s3://stacfiles/",
+        "--collection-id",
+        "abc",
+        "--category",
+        "urban-aerial-photos",
+        "--region",
+        "hawkes-bay",
+        "--gsd",
+        "1m",
+        "--start-date",
+        "2023-09-20",
+        "--end-date",
+        "2023-09-20",
+        "--lifecycle",
+        "ongoing",
+        "--producer",
+        "Placeholder",
+        "--licensor",
+        "Placeholder",
+        "--concurrency",
+        "25",
+    ]
+    # Call script's main function
+    main(args)
+    # Verify collection.json has been created
+    resp = boto3_client.get_object(Bucket="stacfiles", Key="collection.json")
+    assert '"type": "Collection"' in resp["Body"].read().decode("utf-8")