Geojson cog support (#10)

* add handling of geojson and cog files

src/eodash_catalog/endpoints.py

@@ -10,7 +10,7 @@
 
 import requests
 from dateutil import parser
-from pystac import Catalog, Collection, Item, Link, SpatialExtent, Summaries
+from pystac import Asset, Catalog, Collection, Item, Link, SpatialExtent, Summaries
 from pystac_client import Client
 
 from eodash_catalog.sh_endpoint import get_SH_token
@@ -23,6 +23,7 @@
 from eodash_catalog.thumbnails import generate_thumbnail
 from eodash_catalog.utils import (
     Options,
+    create_geojson_from_bbox,
     create_geojson_point,
     generate_veda_cog_link,
     retrieveExtentFromWMSWMTS,
@@ -764,3 +765,58 @@ def handle_custom_endpoint(
         collection_config,
     )
     return collection
+
+
+def handle_raw_source(
+    catalog_config: dict,
+    endpoint_config: dict,
+    collection_config: dict,
+    catalog: Catalog,
+) -> Collection:
+    collection = get_or_create_collection(
+        catalog, collection_config["Name"], collection_config, catalog_config, endpoint_config
+    )
+    if len(endpoint_config.get("TimeEntries", [])) > 0:
+        for time_entry in endpoint_config["TimeEntries"]:
+            extra_fields = {}
+            if "DataProjection" in endpoint_config:
+                extra_fields["proj:epsg"] = endpoint_config["DataProjection"]
+            assets = {}
+            media_type = "application/geo+json"
+            style_type = "text/vector-styles"
+            if endpoint_config["Name"] == "COG source":
+                style_type = "text/cog-styles"
+                media_type = "image/tiff"
+            for a in time_entry["Assets"]:
+                assets[a["Identifier"]] = Asset(
+                    href=a["File"], roles=["data"], media_type=media_type, extra_fields=extra_fields
+                )
+            bbox = endpoint_config.get("Bbox", [-180, -85, 180, 85])
+            item = Item(
+                id=time_entry["Time"],
+                bbox=bbox,
+                properties={},
+                geometry=create_geojson_from_bbox(bbox),
+                datetime=parser.isoparse(time_entry["Time"]),
+                assets=assets,
+            )
+            ep_st = endpoint_config["Style"]
+            style_link = Link(
+                rel="style",
+                target=ep_st
+                if ep_st.startswith("http")
+                else f"{catalog_config['assets_endpoint']}/{ep_st}",
+                media_type=style_type,
+                extra_fields={
+                    "asset:keys": list(assets),
+                },
+            )
+            item.add_link(style_link)
+            if "DataProjection" in endpoint_config:
+                collection.extra_fields["proj:epsg"] = endpoint_config["DataProjection"]
+            link = collection.add_item(item)
+            link.extra_fields["datetime"] = time_entry["Time"]
+            link.extra_fields["assets"] = [a["File"] for a in time_entry["Assets"]]
+    add_collection_information(catalog_config, collection, collection_config)
+    collection.update_extent_from_items()
+    return collection

src/eodash_catalog/generate_indicators.py

@@ -20,6 +20,7 @@
     handle_collection_only,
     handle_custom_endpoint,
     handle_GeoDB_endpoint,
+    handle_raw_source,
     handle_SH_endpoint,
     handle_SH_WMS_endpoint,
     handle_VEDA_endpoint,
@@ -240,6 +241,10 @@ def process_collection_file(
                         collection_config,
                         catalog,
                     )
+                elif resource["Name"] in ["COG source", "GeoJSON source"]:
+                    collection = handle_raw_source(
+                        catalog_config, resource, collection_config, catalog
+                    )
                 else:
                     raise ValueError("Type of Resource is not supported")
                 if collection:

src/eodash_catalog/utils.py

@@ -35,6 +35,21 @@ def create_geojson_point(lon: int | float, lat: int | float) -> dict[str, Any]:
     return {"type": "Feature", "geometry": point, "properties": {}}
 
 
+def create_geojson_from_bbox(bbox: list[float | int]) -> dict:
+    coordinates = [
+        [bbox[0], bbox[1]],
+        [bbox[2], bbox[1]],
+        [bbox[2], bbox[3]],
+        [bbox[0], bbox[3]],
+        [bbox[0], bbox[1]],
+    ]
+    polygon = {"type": "Polygon", "coordinates": [coordinates]}
+
+    feature = {"type": "Feature", "geometry": polygon, "properties": {}}
+    feature_collection = {"type": "FeatureCollection", "features": [feature]}
+    return feature_collection
+
+
 def retrieveExtentFromWMSWMTS(
     capabilities_url: str, layer: str, version: str = "1.1.1", wmts: bool = False
 ) -> tuple[list[float], list[str]]:

tests/test_generate.py

@@ -111,3 +111,47 @@ def test_baselayers_and_overlays_added(catalog_output_folder):
         ]
         assert len(baselayer_links) == 1
         assert len(overlay_links) == 1
+
+
+def test_geojson_dataset_handled(catalog_output_folder):
+    collection_name = "crop_forecast_at"
+    root_collection_path = os.path.join(catalog_output_folder, collection_name)
+    child_collection_path = os.path.join(root_collection_path, collection_name)
+    child_child_collection_path = os.path.join(child_collection_path, collection_name)
+    item_dir = os.path.join(child_child_collection_path, "2024")
+    item_paths = os.listdir(item_dir)
+    assert len(item_paths) == 1
+    with open(os.path.join(child_collection_path, "collection.json")) as fp:
+        collection_json = json.load(fp)
+        geojson_links = [
+            link
+            for link in collection_json["links"]
+            if (link.get("rel", "") == "item" and len(link.get("assets", [])) > 0)
+        ]
+        # geojson link with assets exists
+        assert len(geojson_links) > 0
+        # and has a correct value
+        assert (
+            geojson_links[0]["assets"][0]
+            == "https://raw.githubusercontent.com/eodash/eodash_catalog/main/tests/regional_forecast.json"
+        )
+        # epsg code saved on collection
+        assert collection_json.get("proj:epsg", "") == 3035
+    # epsg code is saved to item
+    with open(os.path.join(item_dir, item_paths[0])) as fp:
+        item_json = json.load(fp)
+        assert item_json["assets"]["vector_data"]["type"] == "application/geo+json"
+        assert item_json["collection"] == collection_name
+
+
+def test_cog_dataset_handled(catalog_output_folder):
+    collection_name = "solar_energy"
+    root_collection_path = os.path.join(catalog_output_folder, collection_name)
+    child_collection_path = os.path.join(root_collection_path, collection_name)
+    child_child_collection_path = os.path.join(child_collection_path, collection_name)
+    item_dir = os.path.join(child_child_collection_path, "2023")
+    item_paths = os.listdir(item_dir)
+    with open(os.path.join(item_dir, item_paths[0])) as fp:
+        item_json = json.load(fp)
+        assert item_json["assets"]["solar_power"]["type"] == "image/tiff"
+        assert item_json["collection"] == collection_name

tests/testing-catalogs/testing.yaml

@@ -8,3 +8,5 @@ assets_endpoint: "https://raw.githubusercontent.com/eurodatacube/eodash-assets/m
 collections:
   - test_wms_no_time
   - test_indicator
+  - test_CROPOMAT1
+  - test_see_solar_energy

tests/testing-collections/test_CROPOMAT1.yaml

@@ -0,0 +1,51 @@
+Name: crop_forecast_at
+Title: Austria yield
+EodashIdentifier: CROPOMAT1
+Subtitle: Yield forecast under "what-if" scenarios
+Description: ""
+Themes:
+  - agriculture
+  - economy
+Tags:
+  - Crop
+  - Yield
+  - Water
+  - Forecast
+DataSource:
+  Spaceborne:
+    Satellite:
+      - Sentinel-2
+      - CLMS HR-VPP Vegetation Indices
+      - CLMS HR-VPP Seasonal Trajectory 
+  InSitu:
+    - Meteorological data from local Meteorological offices (AT)
+Agency:
+  - ESA
+Provider:
+  - Name: CropOM
+    Url: https://cropom.com
+  - Name: Danube Data Cube
+    Url: https://ddc.cropom.com
+  - Name: Copernicus Land Monitoring Service
+    Url: https://land.copernicus.eu/en/dataset-catalog
+  - Name: FAO
+    Url: https://www.fao.org/home/en
+License:
+  - Name: Commercial license
+    Url: https://cropom.com/terms_and_conditions.pdf
+References:
+  - Name: Danube Data Cube documentation
+    Url: https://doc.cropom.com
+  - Name: AquaCrop
+    Url: https://www.fao.org/aquacrop/en/
+Resources:
+    - Name: GeoJSON source
+      Style: crop_forecast_CropOM/style_yield.json
+      Bbox: [9.27, 46.2, 17.3, 49.2]
+      DataProjection: 3035
+      TimeEntries:
+        - Time: "20240101"
+          Assets:
+            - Identifier: vector_data
+              File: "https://raw.githubusercontent.com/eodash/eodash_catalog/main/tests/regional_forecast.json"
+Legend: crop_forecast_CropOM/cm_legend.png

tests/testing-collections/test_see_solar_energy.yaml

@@ -0,0 +1,25 @@
+Name: solar_energy
+Title: Solar energy cog example
+EodashIdentifier: SEE
+Description: testing 
+Themes:
+  - placeholder
+Tags:
+  - placeholder
+DataSource:
+  Spaceborne:
+    Satellite:
+      - Sentinel-2
+Agency:
+  - ESA
+Resources:
+    - Name: COG source
+      Style: test_assets/cog_style.json
+      Bbox: [11, 46.5, 15.5, 48.9]
+      TimeEntries:
+        - Time: "2023"
+          Assets: 
+            - Identifier: solar_power
+              File: https://eox-gtif-public.s3.eu-central-1.amazonaws.com/DHI/v2/SolarPowerPotential_Annual_COG_clipped_3857_fixed.tif
+            - Identifier: wsf
+              File: https://eox-gtif-public.s3.eu-central-1.amazonaws.com/DHI/WSF_EucDist_Austria_3857_COG_fix.tif