Merge pull request #315 from microsoft/modis-platform-property

MODIS: Add platform property back in if its missing

datasets/modis/misc.py

@@ -0,0 +1,18 @@
+# Add back in the platform property which NASA removed from their XML on March 13 2024
+# On the MODIS side terra is distributed as MOD and aqua as MYD,
+# but Within MPC both are distributed as MODxxx
+def add_platform_field(item, href, logger):
+    if ("platform" not in item.properties) or (item.properties["platform"] == ""):
+        logger.debug("platform field missing, filling it in based on original xml href")
+        try:
+            if href.split('/')[4][0:3] == "MOD":
+                item.properties["platform"] = "terra"
+            elif href.split('/')[4][0:3] == "MYD":
+                item.properties["platform"] = "aqua"
+            elif href.split('/')[4][0:3] == "MCD":
+                item.properties["platform"] = "terra,aqua"
+            else:
+                logger.warning("href did not contain MOD/MYD/MCD in the usual spot")
+        except Exception as e:
+            logger.warning(f"href did not contain MOD/MYD/MCD in the usual spot, got error: {e}")
+    return item

datasets/modis/modis.py

@@ -9,6 +9,7 @@
 from azure.core.exceptions import ResourceNotFoundError
 from stactools.core.utils.antimeridian import Strategy
 from stactools.modis.file import File
+from misc import add_platform_field
 
 from pctasks.core.models.task import WaitTaskResult
 from pctasks.core.storage import StorageFactory
@@ -86,4 +87,6 @@ def create_item(
         item.assets["metadata"].href = file.xml_href
         item.assets["metadata"].href = file.xml_href
 
+        item = add_platform_field(item, file.xml_href, logger)
+
         return [item]

datasets/modis/tests/test_platform_field.py

@@ -0,0 +1,23 @@
+import logging
+from datasets.modis.misc import add_platform_field
+from pystac import Item
+from datetime import date
+
+logger = logging.getLogger(__name__)
+
+def test_terra():
+    href = 'https://modiseuwest.blob.core.windows.net/modis-061/MOD14A2/19/05/2024033/MOD14A2.A2024033.h19v05.061.2024044033426.hdf.xml'
+    item = Item(id="MOD14A2.A2024033.h19v05.061.2024044033426", properties={}, geometry=None, bbox=None, datetime=date.today())
+    item = add_platform_field(item, href, logger)
+    assert item.properties["platform"] == "terra"
+
+def test_aqua():
+    href = 'https://modiseuwest.blob.core.windows.net/modis-061/MYD14A2/20/05/2024033/MYD14A2.A2024033.h20v05.061.2024043110139.hdf.xml'
+    item = Item(id="MOD14A2.A2024033.h19v05.061.2024044033426", properties={}, geometry=None, bbox=None, datetime=date.today())
+    item = add_platform_field(item, href, logger)
+    assert item.properties["platform"] == "aqua"
+
+def test_malformed_href():
+    href = 'asdasdaasdasdads'
+    item = Item(id="MOD14A2.A2024033.h19v05.061.2024044033426", properties={}, geometry=None, bbox=None, datetime=date.today())
+    item = add_platform_field(item, href, logger)