Merge pull request #487 from duckontheweb/fix/stac-object-identification

Fix identification of invalid objects with stac_version field

Author: Jon Duckworth

CHANGELOG.md

@@ -15,6 +15,10 @@
 
 ### Fixed
 
+- Bug in `pystac.serialization.identify_stac_object_type` where invalid objects with
+  `stac_version == 1.0.0` were incorrectly identified as Catalogs
+  ([#487](https://github.com/stac-utils/pystac/pull/487))
+
 ### Removed
 
 - `STAC_IO` class in favor of `StacIO`. This was deprecated in v1.0.0-beta.1 and has

pystac/serialization/identify.py

@@ -78,7 +78,7 @@ class STACVersionRange:
 
     def __init__(
         self,
-        min_version: Union[str, STACVersionID] = "0.4.0",
+        min_version: Union[str, STACVersionID] = "0.8.0",
         max_version: Optional[Union[str, STACVersionID]] = None,
     ):
         if isinstance(min_version, str):
@@ -180,17 +180,34 @@ def identify_stac_object_type(
     Args:
         json_dict : The dict of JSON to identify.
     """
-    # Try to identify using 'type' property, if present
-    if "type" in json_dict:
-        # Try to find 'type' property in known STACObjectType values
-        for t in pystac.STACObjectType:
-            if json_dict["type"].lower() == t.value.lower():
-                return t
-
+    stac_version = (
+        STACVersionID(json_dict["stac_version"])
+        if "stac_version" in json_dict
+        else None
+    )
     obj_type = json_dict.get("type")
 
+    # Try to identify using 'type' property for v1.0.0-rc.1 and higher
+    introduced_type_attribute = STACVersionID("1.0.0-rc.1")
+    if stac_version is not None and stac_version >= introduced_type_attribute:
+
+        # Since v1.0.0-rc.1 requires a "type" field for all STAC objects, any object
+        # that is missing this attribute is not a valid STAC object.
+        if obj_type is None:
+            return None
+
+        # Try to match the "type" attribute
+        if obj_type == pystac.STACObjectType.CATALOG:
+            return pystac.STACObjectType.CATALOG
+        elif obj_type == pystac.STACObjectType.COLLECTION:
+            return pystac.STACObjectType.COLLECTION
+        elif obj_type == pystac.STACObjectType.ITEM:
+            return pystac.STACObjectType.ITEM
+        else:
+            return None
+
     # For pre-1.0 objects for version 0.8.* or later 'stac_version' must be present
-    if "stac_version" in json_dict:
+    if stac_version is not None:
         # Pre-1.0 STAC objects with 'type' == "Feature" are Items
         if obj_type == "Feature":
             return pystac.STACObjectType.ITEM

pystac/serialization/migrate.py

@@ -16,8 +16,7 @@
 def _migrate_catalog(
     d: Dict[str, Any], version: STACVersionID, info: STACJSONDescription
 ) -> None:
-    if version < "0.8":
-        d["stac_extensions"] = list(info.extensions)
+    d["type"] = pystac.STACObjectType.CATALOG
 
 
 def _migrate_collection_summaries(
@@ -35,7 +34,7 @@ def _migrate_collection_summaries(
 def _migrate_collection(
     d: Dict[str, Any], version: STACVersionID, info: STACJSONDescription
 ) -> None:
-    _migrate_catalog(d, version, info)
+    d["type"] = pystac.STACObjectType.COLLECTION
     _migrate_collection_summaries(d, version, info)
 
 

pystac/stac_object.py

@@ -15,9 +15,9 @@ class STACObjectType(str, Enum):
     def __str__(self) -> str:
         return str(self.value)
 
-    CATALOG = "CATALOG"
-    COLLECTION = "COLLECTION"
-    ITEM = "ITEM"
+    CATALOG = "Catalog"
+    COLLECTION = "Collection"
+    ITEM = "Feature"
 
 
 class STACObject(ABC):