Fix SAFE/S2GM band selection picking up auxiliary rasters (AOT/WVP/TCI).

opensr_utils/pipeline.py

@@ -353,7 +353,9 @@ def safe_band_name(path):
             used = set(lr_file_dict.values())
             for path in jp2s:
                 if path not in used:
-                    lr_file_dict[safe_band_name(path)] = path
+                    name = safe_band_name(path)
+                    if name.startswith("B") and name[1:].isalnum():
+                        lr_file_dict[name] = path
             if not lr_file_dict:
                 raise FileNotFoundError(f"No usable SAFE bands found in {self.root}.")
 
@@ -407,7 +409,11 @@ def safe_band_name(path):
                     lr_file_dict[label] = by_name[band_name]
             used = set(lr_file_dict.values())
             for band_name, path in by_name.items():
-                if path not in used:
+                if (
+                    path not in used
+                    and band_name.startswith("B")
+                    and band_name[1:].isalnum()
+                ):
                     lr_file_dict[band_name] = path
             self.image_meta["lr_file_dict"] = lr_file_dict
 

tests/test_pipeline_helpers.py

@@ -438,3 +438,33 @@ def test_write_to_file_creates_placeholder_from_prediction_index(
     with rasterio.open(runner.placeholder_filepath) as src:
         assert (src.width, src.height, src.count) == (9, 6, 2)
         assert src.dtypes == ("uint16", "uint16")
+
+
+def test_safe_band_selection_excludes_auxiliary_rasters(write_raster, tmp_path):
+    safe = tmp_path / "S2A_TEST.SAFE"
+    granule = safe / "GRANULE" / "L1C_TEST" / "IMG_DATA" / "R10m"
+    granule.mkdir(parents=True)
+    for band in ("B02", "B03", "B04", "B08", "AOT", "WVP", "TCI"):
+        write_raster(
+            granule / f"T_TEST_{band}_10m.jp2",
+            data=np.ones((1, 16, 16), dtype=np.uint16),
+            width=16,
+            height=16,
+        )
+    (safe / "manifest.safe").write_text("")
+
+    runner = pipeline.large_file_processing(
+        root=str(safe),
+        model=None,
+        window_size=(8, 8),
+        factor=2,
+        overlap=4,
+        eliminate_border_px=0,
+        device="cpu",
+        batch_size=1,
+        num_workers=0,
+        prefetch_factor=0,
+    )
+
+    assert runner.image_meta["bands"] == 4
+    assert set(runner.image_meta["lr_file_dict"]) == {"R", "G", "B", "NIR"}