From c594ffc6333e206fedc26e0292fad7564edb38b0 Mon Sep 17 00:00:00 2001
From: Maximilian Linhoff <maximilian.linhoff@tu-dortmund.de>
Date: Thu, 24 Oct 2024 12:14:02 +0200
Subject: [PATCH] Set null island as fallback reference location in
 Subarray.from_hdf, fixes #2626

---
 docs/changes/2627.bugfix.rst            |  6 +++++
 src/ctapipe/instrument/subarray.py      |  8 ++++++-
 src/ctapipe/tools/tests/test_process.py | 29 ++++++++++++++++++++++++-
 3 files changed, 41 insertions(+), 2 deletions(-)
 create mode 100644 docs/changes/2627.bugfix.rst

diff --git a/docs/changes/2627.bugfix.rst b/docs/changes/2627.bugfix.rst
new file mode 100644
index 00000000000..4ed358e1ad8
--- /dev/null
+++ b/docs/changes/2627.bugfix.rst
@@ -0,0 +1,6 @@
+Fix that hdf5 files created with older versions of ctapipe, e.g.
+the public dataset created with 0.17 can be read by ctapipe-process.
+These files are missing the subarray reference location, which was
+introduced in later versions of ctapipe. A dummy location (lon=0, lat=0)
+is used for these now, the same value is already used for simtel files
+lacking this information.
diff --git a/src/ctapipe/instrument/subarray.py b/src/ctapipe/instrument/subarray.py
index e69e34f77f3..eafd56c2a95 100644
--- a/src/ctapipe/instrument/subarray.py
+++ b/src/ctapipe/instrument/subarray.py
@@ -751,7 +751,6 @@ def from_hdf(cls, path, focal_length_choice=FocalLengthKind.EFFECTIVE):
 
         positions = np.column_stack([layout[f"pos_{c}"] for c in "xyz"])
 
-        reference_location = None
         name = layout.meta.get("SUBARRAY", "Unknown")
 
         if "reference_itrs_x" in layout.meta:
@@ -760,6 +759,13 @@ def from_hdf(cls, path, focal_length_choice=FocalLengthKind.EFFECTIVE):
                 y=layout.meta["reference_itrs_y"] * u.m,
                 z=layout.meta["reference_itrs_z"] * u.m,
             )
+        else:
+            # "null island" dummy location for backwards compatibility with old files
+            reference_location = EarthLocation(
+                lon=0 * u.deg,
+                lat=0 * u.deg,
+                height=0 * u.m,
+            )
 
         return cls(
             name=str(name),
diff --git a/src/ctapipe/tools/tests/test_process.py b/src/ctapipe/tools/tests/test_process.py
index 769574c71f4..17ba5ae845d 100644
--- a/src/ctapipe/tools/tests/test_process.py
+++ b/src/ctapipe/tools/tests/test_process.py
@@ -1,5 +1,4 @@
 #!/usr/bin/env python3
-
 """
 Test ctapipe-process on a few different use cases
 """
@@ -488,3 +487,31 @@ def test_only_trigger_and_simulation(tmp_path):
         assert len(events) == 7
         assert "tels_with_trigger" in events.colnames
         assert "true_energy" in events.colnames
+
+
+@pytest.mark.parametrize(
+    ("input_url", "args"),
+    [
+        pytest.param(
+            "dataset://gamma_diffuse_dl2_train_small.dl2.h5",
+            ["--no-write-images", "--max-events=5"],
+            id="0.17",
+        )
+    ],
+)
+def test_on_old_file(input_url, args, tmp_path):
+    config = resource_file("stage1_config.json")
+
+    output_path = tmp_path / "test.dl1.h5"
+    run_tool(
+        ProcessorTool(),
+        argv=[
+            f"--config={config}",
+            f"--input={input_url}",
+            f"--output={output_path}",
+            "--overwrite",
+            *args,
+        ],
+        cwd=tmp_path,
+        raises=True,
+    )