saveStitchedTileGrid: handle items instead of files #1111

Author: bossie

openeogeotrellis/geopysparkdatacube.py

@@ -1919,22 +1919,43 @@ def add_gdalinfo_objects(assets_original):
                 elif stitch:
                     if tile_grid:
                         _log.info("save_result save_stitched_tile_grid")
-                        tiles = self._save_stitched_tile_grid(max_level, save_filename, tile_grid, crop_bounds,
-                                                              zlevel=zlevel, filename_prefix=filename_prefix)
-
-                        # noinspection PyProtectedMember
-                        return add_gdalinfo_objects(
-                            {
-                                str(pathlib.Path(tile._1()).name): {
-                                    "href": tile._1(),
-                                    "bbox": to_latlng_bbox(tile._2()),
-                                    "geometry": mapping(Polygon.from_bounds(*to_latlng_bbox(tile._2()))),
+                        java_items = self._save_stitched_tile_grid(
+                            max_level,
+                            save_filename,
+                            tile_grid,
+                            crop_bounds,
+                            zlevel=zlevel,
+                            filename_prefix=filename_prefix,
+                        )
+
+                        items = {}
+
+                        for java_item in java_items:
+                            bbox = java_item.bbox()
+                            assets = {}
+
+                            for asset_key, asset in java_item.assets().items():
+                                assets[asset_key] = {
+                                    "href": asset.path(),
+                                    "bbox": to_latlng_bbox(bbox),
+                                    "geometry": mapping(Polygon.from_bounds(*to_latlng_bbox(bbox))),
                                     "type": "image/tiff; application=geotiff",
                                     "roles": ["data"],
                                 }
-                                for tile in tiles
+
+                            assets = add_gdalinfo_objects(assets)
+
+                            item = {
+                                "id": java_item.id(),
+                                "properties": {"datetime": java_item.datetime()},
+                                "geometry": mapping(Polygon.from_bounds(*to_latlng_bbox(bbox))),
+                                "bbox": to_latlng_bbox(bbox),
+                                "assets": assets,
                             }
-                        )
+
+                            items[java_item.id()] = item
+
+                        return items
                     else:
                         _log.info("save_result save_stitched")
                         java_item = self._save_stitched(max_level, save_filename, crop_bounds, zlevel=zlevel)

tests/test_batch_result.py

@@ -2937,8 +2937,11 @@ def test_unified_asset_keys_spatial_geotiff(
 
     assert len(items) == 1
     # single item ID can be anything (no spatial or temporal references)
-    assert job_metadata["start_datetime"] == "2025-04-01T00:00:00Z"  # top-level rather than on Item
-    assert job_metadata["end_datetime"] == "2025-04-21T00:00:00Z"  # ditto
+
+    # at job-level rather than on Item
+    assert items[0]["properties"].get("datetime") is None
+    assert job_metadata["start_datetime"] == "2025-04-01T00:00:00Z"
+    assert job_metadata["end_datetime"] == "2025-04-21T00:00:00Z"
 
     assert set(items[0]["assets"].keys()) == expected_asset_keys
 
@@ -3000,7 +3003,84 @@ def test_unified_asset_keys_stitch_geotiff(tmp_path):
 
     assert len(items) == 1
     # single item ID can be anything (no spatial or temporal references)
-    assert job_metadata["start_datetime"] == "2025-04-01T00:00:00Z"  # top-level rather than on Item
-    assert job_metadata["end_datetime"] == "2025-04-21T00:00:00Z"  # ditto
+
+    # at job-level rather than on Item
+    assert items[0]["properties"].get("datetime") is None
+    assert job_metadata["start_datetime"] == "2025-04-01T00:00:00Z"
+    assert job_metadata["end_datetime"] == "2025-04-21T00:00:00Z"
 
     assert set(items[0]["assets"].keys()) == {"openEO"}
+
+
+def test_unified_asset_keys_stitch_tile_grid(tmp_path):
+    process_graph = {
+        "load2": {
+            "process_id": "load_collection",
+            "arguments": {
+                "bands": [
+                    "Flat:0",
+                    "Flat:1",
+                    "Flat:2",
+                ],
+                "id": "TestCollection-LonLat16x16",
+                "spatial_extent": {
+                    "west": 0,
+                    "south": 50,
+                    "east": 2,
+                    "north": 51,
+                },
+                "temporal_extent": ["2025-04-01", "2025-04-21"],
+            },
+        },
+        "save1": {
+            "process_id": "save_result",
+            "arguments": {
+                "data": {"from_node": "load2"},
+                "format": "GTIFF",
+                "options": {
+                    "stitch": True,
+                    "tile_grid": "wgs84-1degree",
+                },
+            },
+            "result": True,
+        },
+    }
+
+    process = {
+        "process_graph": process_graph,
+    }
+
+    job_dir = tmp_path
+    metadata_file = job_dir / "job_metadata.json"
+
+    run_job(
+        process,
+        output_file=job_dir / "out",
+        metadata_file=metadata_file,
+        api_version="2.0.0",
+        job_dir=job_dir,
+        dependencies=[],
+    )
+
+    tiff_files = {file for file in os.listdir(job_dir) if file.endswith(".tiff")}
+    assert tiff_files == {
+        "openEO-N50E000.tiff",
+        "openEO-N50E001.tiff",
+    }
+
+    with open(metadata_file) as f:
+        job_metadata = json.load(f)
+
+    items = job_metadata["items"]
+    print(f"items={json.dumps(items, indent=2)}")
+
+    assert len(items) == 2
+    assert len({item["id"] for item in items}) == 2
+
+    # at job-level rather than on Item
+    assert {item["properties"].get("datetime") for item in items} == {None}
+    assert job_metadata["start_datetime"] == "2025-04-01T00:00:00Z"
+    assert job_metadata["end_datetime"] == "2025-04-21T00:00:00Z"
+
+    for item in items:
+        assert set(item["assets"].keys()) == {"openEO"}