From 8a53d07b93813ae969d0d266f4ef46f65979b354 Mon Sep 17 00:00:00 2001
From: Ariana Barzinpour <ariana.barzinpour@ga.gov.au>
Date: Tue, 3 Sep 2024 01:59:44 +0000
Subject: [PATCH] preserve accessory assets

---
 odc/stac/_mdtools.py          | 12 ++++++++++++
 odc/stac/eo3/_eo3converter.py |  1 +
 odc/stac/model.py             |  3 +++
 tests/test_mdtools.py         | 13 +++++++++++++
 4 files changed, 29 insertions(+)

diff --git a/odc/stac/_mdtools.py b/odc/stac/_mdtools.py
index 6afbe3f..c1fe4e7 100644
--- a/odc/stac/_mdtools.py
+++ b/odc/stac/_mdtools.py
@@ -660,6 +660,7 @@ def parse_item(
     band2grid = template.band2grid
     has_proj = False if template.has_proj is False else has_proj_ext(item)
     _assets = item.assets
+    _acc_names = list(_assets.keys())
 
     _grids: Dict[str, GeoBox] = {}
     bands: Dict[BandKey, RasterSource] = {}
@@ -681,6 +682,9 @@ def _get_grid(grid_name: str, asset: pystac.asset.Asset) -> GeoBox:
         asset = _assets.get(asset_name)
         if asset is None:
             continue
+        # the assets that aren't bands should be accessories
+        if asset_name in _acc_names:
+            _acc_names.remove(asset_name)
 
         grid_name = band2grid.get(asset_name, "default")
         geobox: Optional[GeoBox] = _get_grid(grid_name, asset) if has_proj else None
@@ -711,6 +715,13 @@ def _get_grid(grid_name: str, asset: pystac.asset.Asset) -> GeoBox:
             driver_data=driver_data,
         )
 
+    accessories = {
+        name: {
+            "path": _assets[name].href
+        }
+        for name in _acc_names
+    }
+        
     md = item.common_metadata
     return ParsedItem(
         item.id,
@@ -720,6 +731,7 @@ def _get_grid(grid_name: str, asset: pystac.asset.Asset) -> GeoBox:
         datetime=item.datetime,
         datetime_range=(md.start_datetime, md.end_datetime),
         href=item.get_self_href(),
+        accessories=accessories,
     )
 
 
diff --git a/odc/stac/eo3/_eo3converter.py b/odc/stac/eo3/_eo3converter.py
index 8ab7e6e..050a1df 100644
--- a/odc/stac/eo3/_eo3converter.py
+++ b/odc/stac/eo3/_eo3converter.py
@@ -220,6 +220,7 @@ def _to_dataset(
         "properties": dicttoolz.keymap(
             lambda k: STAC_TO_EO3_RENAMES.get(k, k), properties
         ),
+        "accessories": item.accessories,
         "lineage": {},
     }
 
diff --git a/odc/stac/model.py b/odc/stac/model.py
index cc4a14c..746b9ba 100644
--- a/odc/stac/model.py
+++ b/odc/stac/model.py
@@ -226,6 +226,9 @@ class ParsedItem(Mapping[BandIdentifier, RasterSource]):
 
     href: Optional[str] = None
     """Self link from stac item."""
+    
+    accessories: dict[str, Any] = field(default_factory=dict)
+    """Additional assets"""
 
     def geoboxes(self, bands: BandQuery = None) -> Tuple[GeoBox, ...]:
         """
diff --git a/tests/test_mdtools.py b/tests/test_mdtools.py
index b0bbb16..0070c9d 100644
--- a/tests/test_mdtools.py
+++ b/tests/test_mdtools.py
@@ -389,6 +389,19 @@ def test_parse_item_no_proj(sentinel_stac_ms: pystac.item.Item):
     assert _auto_load_params([xx] * 3) is None
 
 
+def test_accessories_preserved(ga_landsat_stac: pystac.item.Item):
+    item0 = ga_landsat_stac
+    item = pystac.Item.from_dict(item0.to_dict())
+
+    md = extract_collection_metadata(item, STAC_CFG)
+
+    xx = parse_item(item, md)
+    assert len(xx.accessories) == 3
+    assert xx.accessories.get("thumbnail:nbart")
+    assert xx.accessories.get("checksum:sha1")
+    assert xx.accessories.get("metadata:processor")
+
+
 @pytest.fixture
 def parsed_item_s2(sentinel_stac_ms: pystac.item.Item):
     (item,) = parse_items([sentinel_stac_ms], STAC_CFG)