make sure array mask is propagated after applying offset to band

mapchete_eo/io/assets.py

@@ -32,17 +32,17 @@
 
 
 class STACRasterBandProperties(BaseModel):
-    nodata: NodataVal = None
+    nodata: Optional[NodataVal] = None
     data_type: Optional[str] = None
     scale: float = 1.0
     offset: float = 0.0
 
     @staticmethod
     def from_asset(
         asset: pystac.Asset,
-        nodataval: NodataVal = None,
+        nodataval: Optional[NodataVal] = None,
     ) -> STACRasterBandProperties:
-        if asset.extra_fields.get("raster:offset") is not None:
+        if asset.extra_fields.get("raster:offset", {}):
             properties = dict(
                 offset=asset.extra_fields.get("raster:offset"),
                 scale=asset.extra_fields.get("raster:scale"),
@@ -87,16 +87,21 @@ def asset_to_np_array(
     )
 
     logger.debug("reading asset %s and indexes %s ...", asset, indexes)
-    data = read_raster(
+    array = read_raster(
         inp=path,
         indexes=indexes,
         grid=grid,
         resampling=resampling.name,
         dst_nodata=band_properties.nodata,
-    ).data
-
+    ).array
     if apply_offset and band_properties.offset:
-        data_type = band_properties.data_type or data.dtype
+        logger.debug(
+            "apply offset %s and scale %s to asset %s",
+            band_properties.offset,
+            band_properties.scale,
+            asset,
+        )
+        data_type = band_properties.data_type or array.dtype
 
         # determine value range for the target data_type
         clip_min, clip_max = dtype_ranges[str(data_type)]
@@ -105,18 +110,17 @@ def asset_to_np_array(
         if clip_min == band_properties.nodata:
             clip_min += 1
 
-        data[:] = (
+        array[~array.mask] = (
             (
-                ((data * band_properties.scale) + band_properties.offset)
+                ((array[~array.mask] * band_properties.scale) + band_properties.offset)
                 / band_properties.scale
             )
             .round()
             .clip(clip_min, clip_max)
             .astype(data_type, copy=False)
             .data
         )
-
-    return data
+    return array
 
 
 def get_assets(

mapchete_eo/search/config.py

@@ -1,3 +1,7 @@
+import logging
+
+
+from contextlib import contextmanager
 from typing import Optional, Dict, Any
 
 from mapchete.path import MPath, MPathLike
@@ -61,3 +65,46 @@ def deprecate_max_cloud_cover(cls, values: Dict[str, Any]) -> Dict[str, Any]:
         ),
     )
     search_index: Optional[MPathLike] = None
+
+
+@contextmanager
+def patch_invalid_assets():
+    """
+    Context manager/decorator to fix pystac crash on malformed assets (strings instead of dicts).
+
+    """
+    try:
+        from pystac.extensions.file import FileExtensionHooks
+    except ImportError:  # pragma: no cover
+        yield
+        return
+
+    logger = logging.getLogger(__name__)
+
+    _original_migrate = FileExtensionHooks.migrate
+
+    def _safe_migrate(self, obj, version, info):
+        if "assets" in obj and isinstance(obj["assets"], dict):
+            bad_keys = []
+            for key, asset in obj["assets"].items():
+                if not isinstance(asset, dict):
+                    logger.debug(
+                        "Removing malformed asset '%s' (type %s) from item %s",
+                        key,
+                        type(asset),
+                        obj.get("id", "unknown"),
+                    )
+                    bad_keys.append(key)
+
+            for key in bad_keys:
+                del obj["assets"][key]
+
+        return _original_migrate(self, obj, version, info)
+
+    # Apply patch
+    FileExtensionHooks.migrate = _safe_migrate
+    try:
+        yield
+    finally:
+        # Restore original
+        FileExtensionHooks.migrate = _original_migrate

mapchete_eo/search/stac_search.py

@@ -15,7 +15,7 @@
 
 from mapchete_eo.product import blacklist_products
 from mapchete_eo.search.base import CollectionSearcher, StaticCollectionWriterMixin
-from mapchete_eo.search.config import StacSearchConfig
+from mapchete_eo.search.config import StacSearchConfig, patch_invalid_assets
 from mapchete_eo.settings import mapchete_eo_settings
 from mapchete_eo.types import TimeRange
 
@@ -138,14 +138,16 @@ def _search_chunks(
                     query=query,
                 )
 
-        for search in _searches():
-            for item in search.items():
-                if item.get_self_href() in self.blacklist:  # pragma: no cover
-                    logger.debug(
-                        "item %s found in blacklist and skipping", item.get_self_href()
-                    )
-                    continue
-                yield item
+        with patch_invalid_assets():
+            for search in _searches():
+                for item in search.items():
+                    if item.get_self_href() in self.blacklist:  # pragma: no cover
+                        logger.debug(
+                            "item %s found in blacklist and skipping",
+                            item.get_self_href(),
+                        )
+                        continue
+                    yield item
 
     @cached_property
     def default_search_params(self):

tests/conftest.py

@@ -400,15 +400,13 @@ def s2_l2a_safe_metadata(s2_testdata_dir):
     )
 
 
-@pytest.mark.remote
 @pytest.fixture(scope="session")
 def s2_l2a_metadata_remote():
     return guess_s2metadata_from_metadata_xml(
         "s3://sentinel-s2-l2a/tiles/51/K/XR/2020/7/31/0/metadata.xml"
     )
 
 
-@pytest.mark.remote
 @pytest.fixture(scope="session")
 def s2_l2a_roda_metadata_remote():
     """Same content as s2_l2a_metadata_remote, but hosted on different server."""
@@ -417,7 +415,6 @@ def s2_l2a_roda_metadata_remote():
     )
 
 
-@pytest.mark.remote
 @pytest.fixture(scope="session")
 def s2_l2a_roda_metadata_jp2_masks_remote():
     """From about 2022 on, ahte masks are now encoded as JP2 (rasters), not as GMLs (features)."""
@@ -426,7 +423,6 @@ def s2_l2a_roda_metadata_jp2_masks_remote():
     )
 
 
-@pytest.mark.remote
 @pytest.fixture()
 def s2_l2a_earthsearch_xml_remote():
     """Metadata used by Earth-Search V1 endpoint"""
@@ -435,7 +431,6 @@ def s2_l2a_earthsearch_xml_remote():
     )
 
 
-@pytest.mark.remote
 @pytest.fixture()
 def s2_l2a_earthsearch_xml_remote_broken():
     """Metadata used by Earth-Search V1 endpoint"""
@@ -444,14 +439,12 @@ def s2_l2a_earthsearch_xml_remote_broken():
     )
 
 
-@pytest.mark.remote
 @pytest.fixture(scope="session")
 def s2_l2a_earthsearch_remote(s2_l2a_earthsearch_remote_item):
     """Metadata used by Earth-Search V1 endpoint"""
     return guess_s2metadata_from_item(s2_l2a_earthsearch_remote_item)
 
 
-@pytest.mark.remote
 @pytest.fixture(scope="session")
 def s2_l2a_earthsearch_remote_item():
     """Metadata used by Earth-Search V1 endpoint"""