unit tests fail intermittently upon catalog construction: "IndexError: list index out of range"

[bossie]

Lately the unit tests have been failing a couple of times while constructing the layer catalog. Simply restarting the build fixes them.

Specific cases include: tests.test_views.test_health_default[0.4.0-/health?mode=basic-expected3] and tests.test_api_result.test_apply_dimension_array_concat.

Jenkins output:

Error Message
failed on setup with "IndexError: list index out of range"

Stacktrace

request = <SubRequest 'backend_implementation' for <Function test_health_default[0.4.0-/health?mode=basic-expected3]>>

    @pytest.fixture
    def backend_implementation(request) -> 'GeoPySparkBackendImplementation':
        from openeogeotrellis.backend import GeoPySparkBackendImplementation
>       backend = GeoPySparkBackendImplementation()

tests/conftest.py:119: 
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 
openeogeotrellis/backend.py:274: in __init__
    catalog = get_layer_catalog(opensearch_enrich=True)
openeogeotrellis/layercatalog.py:755: in get_layer_catalog
    metadata = _merge_layers_with_common_name(metadata)
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 

metadata = {'AGERA5': {'_vito': {'data_source': {'data_glob': '/data/MEP/ECMWF/AgERA5/*/*/AgERA5_dewpoint-temperature_*.tif', 'da..., 'extent': {'spatial': {'bbox': [[-180, -56, 180, 83]]}, 'temporal': {'interval': [['2014-10-06', None]]}}, ...}, ...}

    def _merge_layers_with_common_name(metadata):
        common_names = set(m["common_name"] for m in metadata.values() if "common_name" in m)
        logger.debug(f"Creating merged collections for common names: {common_names}")
        for common_name in common_names:
            merged = {
                "id": common_name,
                "_vito": {"data_source": {
                    "type": "merged_by_common_name",
                    "common_name": common_name,
                    "merged_collections": [],
                }},
                "providers": [],
                "links": [],
                "extent": {"spatial": {"bbox": []}, "temporal": {"interval": []}},
            }
    
            for to_merge in (m for m in metadata.values() if m.get("common_name") == common_name):
                merged["_vito"]["data_source"]["merged_collections"].append(to_merge["id"])
                # Fill some fields with first hit
                for field in ["title", "description", "keywords", "version", "license", "cube:dimensions", "summaries"]:
                    if field not in merged and field in to_merge:
                        merged[field] = deepcopy(to_merge[field])
                # Fields to take union
                for field in ["providers", "links"]:
                    if isinstance(to_merge.get(field), list):
                        merged[field] += deepcopy(to_merge[field])
    
                # Take union of bands
                for band_dim in [k for k, v in to_merge["cube:dimensions"].items() if v["type"] == "bands"]:
                    if band_dim not in merged["cube:dimensions"]:
                        merged["cube:dimensions"][band_dim] = deepcopy(to_merge["cube:dimensions"][band_dim])
                    else:
                        for b in to_merge["cube:dimensions"][band_dim]["values"]:
                            if b not in merged["cube:dimensions"][band_dim]["values"]:
                                merged["cube:dimensions"][band_dim]["values"].append(b)
                for b in to_merge["summaries"]["eo:bands"]:
                    eob_names = [x["name"] for x in merged["summaries"]["eo:bands"]]
                    if b["name"] not in eob_names:
                        merged["summaries"]["eo:bands"].append(b)
                    else:
                        i = eob_names.index(b["name"])
                        merged["summaries"]["eo:bands"][i]["aliases"] = list(
                            set(merged["summaries"]["eo:bands"][i].get("aliases", []))
>                           | set(to_merge["summaries"]["eo:bands"][i].get("aliases", []))
                        )
E                       IndexError: list index out of range

openeogeotrellis/layercatalog.py:803: IndexError