Make XML metadata optional, extract from HDF if XML not found

CHANGELOG.md

@@ -9,6 +9,7 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 ### Changed
 
 - Item IDs no longer contain the production datetime ([#88](https://github.com/stactools-packages/modis/pull/88))
+- Make XML metadata optional - extract metadata from HDF file if XML is not available ([#XX](https://github.com/stactools-packages/modis/pull/XX))
 
 ### Fixed
 

src/stactools/modis/builder.py

@@ -110,7 +110,13 @@ def add_hdf_or_xml_href(
             xml_href = f"{href}.xml"
         else:
             raise ValueError(f"Invalid HDF or XML href: {href}")
-        self.add_xml_asset(xml_href)
+
+        # Add XML asset if it exists, otherwise extract metadata from HDF
+        if os.path.exists(xml_href):
+            self.add_xml_asset(xml_href)
+        else:
+            self.metadata = Metadata.from_hdf_href(hdf_href, self.read_href_modifier)
+
         self.add_hdf_asset(
             hdf_href, cog_directory=cog_directory, create_cogs=create_cogs
         )

src/stactools/modis/metadata.py

@@ -1,13 +1,16 @@
 import datetime
 import os.path
+import warnings
 from dataclasses import dataclass
 from typing import Any, Callable, Dict, List, Optional, Tuple
 
 import fsspec
 import numpy as np
+import rasterio
 from lxml import etree
 from rasterio import Affine
 from rasterio.crs import CRS
+from rasterio.errors import NotGeoreferencedWarning
 from shapely.geometry import shape
 from stactools.core.io import ReadHrefModifier
 from stactools.core.io.xml import XmlElement
@@ -232,6 +235,8 @@ def from_cog_tags(cls, cog_tags: Dict[str, str]) -> "Metadata":
         geometry, bbox = cls._geometry_and_bbox(
             collection, horizontal_tile, vertical_tile
         )
+        qa_percent = cog_tags.get("QAPERCENTNOTPRODUCEDCLOUD")
+        qa_percent_not_produced_cloud = int(qa_percent) if qa_percent else None
         return Metadata(
             id=os.path.splitext(cog_tags["LOCALGRANULEID"])[0],
             product=product,
@@ -242,16 +247,42 @@ def from_cog_tags(cls, cog_tags: Dict[str, str]) -> "Metadata":
             end_datetime=end_datetime,
             created=None,
             updated=None,
-            qa_percent_not_produced_cloud=int(cog_tags["QAPERCENTNOTPRODUCEDCLOUD"]),
+            qa_percent_not_produced_cloud=qa_percent_not_produced_cloud,
             qa_percent_cloud_cover=None,
             horizontal_tile=horizontal_tile,
             vertical_tile=vertical_tile,
-            tile_id=cog_tags["TileID"],
+            tile_id=cog_tags.get("TileID", ""),
             platforms=sorted(list(platforms)),
             instruments=sorted(list(instruments)),
             collection=collection,
         )
 
+    @classmethod
+    def from_hdf_href(
+        cls, href: str, read_href_modifier: Optional[ReadHrefModifier] = None
+    ) -> "Metadata":
+        """Reads metadata from an HDF file when XML is not available.
+
+        Args:
+            href (str): The href of the HDF file
+            read_href_modifier (Optional[Callable[[str], str]]): Optional
+                function to modify the read href
+
+        Returns:
+            Metadata: Information that will map to Item attributes.
+        """
+        if read_href_modifier:
+            read_href = read_href_modifier(href)
+        else:
+            read_href = href
+
+        with warnings.catch_warnings():
+            warnings.simplefilter("ignore", category=NotGeoreferencedWarning)
+            with rasterio.open(read_href) as dataset:
+                hdf_tags = dataset.tags()
+
+        return cls.from_cog_tags(hdf_tags)
+
     @property
     def datetime(self) -> Optional[datetime.datetime]:
         """Returns a single nominal datetime for this metadata file.

tests/test_stac.py

@@ -207,6 +207,35 @@ def test_raster_footprint_geometry() -> None:
         item.validate()
 
 
+def test_create_item_from_hdf_without_xml() -> None:
+    """Test that an item can be created from an HDF file when XML is not available.
+
+    This tests the fallback to extracting metadata directly from the HDF file
+    when the accompanying XML metadata file is not present.
+    """
+    hdf_file = "MOD10A2.A2022033.h09v05.061.2022042050729.hdf"
+    source_hdf_path = test_data.get_path(f"data-files/{hdf_file}")
+
+    with TemporaryDirectory() as temporary_directory:
+        # Copy only the HDF file (not the XML) to ensure XML is not available
+        temp_hdf_path = os.path.join(temporary_directory, hdf_file)
+        shutil.copyfile(source_hdf_path, temp_hdf_path)
+
+        # Verify XML does not exist in temp directory
+        temp_xml_path = f"{temp_hdf_path}.xml"
+        assert not os.path.exists(temp_xml_path), "XML file should not exist"
+
+        # Create item from HDF only - should extract metadata from HDF
+        item = stactools.modis.stac.create_item(temp_hdf_path)
+
+        # Verify item was created with correct metadata
+        assert item is not None
+        assert item.id.startswith("MOD10A2.A2022033.h09v05")
+        assert "hdf" in item.assets
+        assert "metadata" not in item.assets  # XML asset should not be present
+        item.validate()
+
+
 @pytest.mark.parametrize("file_name", PROJECTION_EDGE_FILES)
 def test_raster_footprint_at_projection_edge(file_name: str) -> None:
     path = test_data.get_path(file_name)