add margin unit tests

tests/conftest.py

@@ -9,6 +9,7 @@
 from hipscat.catalog.association_catalog.partition_join_info import PartitionJoinInfo
 from hipscat.catalog.catalog_info import CatalogInfo
 from hipscat.catalog.dataset.base_catalog_info import BaseCatalogInfo
+from hipscat.catalog.margin_cache import MarginCacheCatalogInfo
 from hipscat.inspection.almanac import Almanac
 from hipscat.pixel_math import HealpixPixel
 
@@ -126,7 +127,7 @@ def source_catalog_info_with_extra() -> dict:
 
 
 @pytest.fixture
-def margin_cache_catalog_info() -> dict:
+def margin_cache_catalog_info_data() -> dict:
     return {
         "catalog_name": "test_margin",
         "catalog_type": "margin",
@@ -189,6 +190,23 @@ def catalog_info(catalog_info_data) -> CatalogInfo:
     return CatalogInfo(**catalog_info_data)
 
 
+@pytest.fixture
+def margin_catalog_info(margin_cache_catalog_info_data) -> MarginCacheCatalogInfo:
+    return MarginCacheCatalogInfo(**margin_cache_catalog_info_data)
+
+
+@pytest.fixture
+def margin_catalog_pixels() -> List[HealpixPixel]:
+    return [
+        HealpixPixel(0, 4), HealpixPixel(1, 44), HealpixPixel(1, 45), HealpixPixel(1, 46), HealpixPixel(1, 47)
+    ]
+
+
+@pytest.fixture
+def margin_catalog_path(test_data_dir) -> str:
+    return os.path.join(test_data_dir, "small_sky_xmatch_margin")
+
+
 @pytest.fixture
 def catalog_pixels() -> List[HealpixPixel]:
     return [HealpixPixel(1, 0), HealpixPixel(1, 1), HealpixPixel(2, 8)]

tests/data/small_sky_xmatch_margin/catalog_info.json

@@ -0,0 +1,7 @@
+{
+    "catalog_name": "small_sky_xmatch_margin",
+    "catalog_type": "margin",
+    "total_rows": 26,
+    "primary_catalog": "small_sky_xmatch",
+    "margin_threshold": 7200
+}

tests/data/small_sky_xmatch_margin/provenance_info.json

@@ -0,0 +1,28 @@
+{
+    "catalog_name": "small_sky_xmatch_margin",
+    "catalog_type": "margin",
+    "total_rows": 26,
+    "primary_catalog": "data/small_sky_xmatch",
+    "margin_threshold": 7200,
+    "version": "0.1.8",
+    "generation_date": "2024.01.11",
+    "tool_args": {
+        "tool_name": "hipscat_import",
+        "version": "0.1.3.dev12+gf50adee",
+        "runtime_args": {
+            "catalog_name": "small_sky_xmatch_margin",
+            "output_path": "data/small_sky_xmatch_margin",
+            "output_artifact_name": "small_sky_xmatch_margin",
+            "tmp_dir": "",
+            "overwrite": false,
+            "dask_tmp": "",
+            "dask_n_workers": 1,
+            "dask_threads_per_worker": 1,
+            "catalog_path": "data/small_sky_xmatch_margin/small_sky_xmatch_margin",
+            "tmp_path": "data/small_sky_xmatch_margin/small_sky_xmatch_margin/intermediate",
+            "input_catalog_path": "data/small_sky_xmatch",
+            "margin_threshold": 7200,
+            "margin_order": 2
+        }
+    }
+}

tests/hipscat/catalog/dataset/test_catalog_info_factory.py

@@ -50,8 +50,8 @@ def test_create_catalog_info_source(source_catalog_info, source_catalog_info_wit
     assert isinstance(catalog_info, SourceCatalogInfo)
 
 
-def test_create_catalog_info_margin_cache(margin_cache_catalog_info):
-    catalog_info = create_catalog_info(margin_cache_catalog_info)
+def test_create_catalog_info_margin_cache(margin_cache_catalog_info_data):
+    catalog_info = create_catalog_info(margin_cache_catalog_info_data)
     assert catalog_info.catalog_name == "test_margin"
     assert isinstance(catalog_info, BaseCatalogInfo)
     assert isinstance(catalog_info, MarginCacheCatalogInfo)

tests/hipscat/catalog/margin_cache/test_margin_cache_catalog_info.py

@@ -8,16 +8,16 @@
 from hipscat.io import file_io
 
 
-def test_margin_cache_catalog_info(margin_cache_catalog_info, assert_catalog_info_matches_dict):
-    info = MarginCacheCatalogInfo(**margin_cache_catalog_info)
-    assert_catalog_info_matches_dict(info, margin_cache_catalog_info)
+def test_margin_cache_catalog_info(margin_cache_catalog_info_data, assert_catalog_info_matches_dict):
+    info = MarginCacheCatalogInfo(**margin_cache_catalog_info_data)
+    assert_catalog_info_matches_dict(info, margin_cache_catalog_info_data)
 
 
-def test_str(margin_cache_catalog_info):
+def test_str(margin_cache_catalog_info_data):
     correct_string = ""
-    for name, value in margin_cache_catalog_info.items():
+    for name, value in margin_cache_catalog_info_data.items():
         correct_string += f"  {name} {value}\n"
-    cat_info = MarginCacheCatalogInfo(**margin_cache_catalog_info)
+    cat_info = MarginCacheCatalogInfo(**margin_cache_catalog_info_data)
     assert str(cat_info) == correct_string
 
 
@@ -38,29 +38,29 @@ def test_read_from_file(margin_cache_catalog_info_file, assert_catalog_info_matc
         assert_catalog_info_matches_dict(catalog_info, catalog_info_json)
 
 
-def test_required_fields_missing(margin_cache_catalog_info):
+def test_required_fields_missing(margin_cache_catalog_info_data):
     required_fields = ["primary_catalog", "margin_threshold"]
     for required_field in required_fields:
         assert required_field in MarginCacheCatalogInfo.required_fields
     for field in required_fields:
-        init_data = margin_cache_catalog_info.copy()
+        init_data = margin_cache_catalog_info_data.copy()
         init_data[field] = None
         with pytest.raises(ValueError, match=field):
             MarginCacheCatalogInfo(**init_data)
 
 
-def test_type_missing(margin_cache_catalog_info):
-    init_data = margin_cache_catalog_info.copy()
+def test_type_missing(margin_cache_catalog_info_data):
+    init_data = margin_cache_catalog_info_data.copy()
     init_data["catalog_type"] = None
     catalog_info = MarginCacheCatalogInfo(**init_data)
     assert catalog_info.catalog_type == CatalogType.MARGIN
 
 
-def test_wrong_type(margin_cache_catalog_info, catalog_info_data):
+def test_wrong_type(margin_cache_catalog_info_data, catalog_info_data):
     with pytest.raises(TypeError, match="unexpected"):
         MarginCacheCatalogInfo(**catalog_info_data)
 
     with pytest.raises(ValueError, match=f"{CatalogType.MARGIN}"):
-        init_data = margin_cache_catalog_info.copy()
+        init_data = margin_cache_catalog_info_data.copy()
         init_data["catalog_type"] = CatalogType.OBJECT
         MarginCacheCatalogInfo(**init_data)

tests/hipscat/catalog/margin_cache/test_margin_catalog.py

@@ -0,0 +1,77 @@
+import json
+import os
+
+import pytest
+
+from hipscat.catalog import MarginCatalog, CatalogType, PartitionInfo
+from hipscat.pixel_tree.pixel_node_type import PixelNodeType
+
+
+def test_init_catalog(margin_catalog_info, margin_catalog_pixels):
+    catalog = MarginCatalog(
+        margin_catalog_info, margin_catalog_pixels
+    )
+    assert catalog.catalog_name == margin_catalog_info.catalog_name
+    assert catalog.get_healpix_pixels() == margin_catalog_pixels
+    assert catalog.catalog_info == margin_catalog_info
+
+    assert len(catalog.get_healpix_pixels()) == len(margin_catalog_pixels)
+    for hp_pixel in catalog.get_healpix_pixels():
+        assert hp_pixel in margin_catalog_pixels
+        assert hp_pixel in catalog.pixel_tree
+        assert catalog.pixel_tree[hp_pixel].node_type == PixelNodeType.LEAF
+
+
+def test_wrong_catalog_type(margin_catalog_info, margin_catalog_pixels):
+    margin_catalog_info.catalog_type = CatalogType.OBJECT
+    with pytest.raises(ValueError, match="catalog_type"):
+        MarginCatalog(margin_catalog_info, margin_catalog_pixels)
+
+
+def test_wrong_catalog_info_type(catalog_info, margin_catalog_pixels):
+    catalog_info.catalog_type = CatalogType.MARGIN
+    with pytest.raises(TypeError, match="catalog_info"):
+        MarginCatalog(catalog_info, margin_catalog_pixels)
+
+
+def test_read_from_file(margin_catalog_path, margin_catalog_pixels):
+    catalog = MarginCatalog.read_from_hipscat(margin_catalog_path)
+    assert catalog.on_disk
+    assert catalog.catalog_path == margin_catalog_path
+    assert len(catalog.get_healpix_pixels()) == len(margin_catalog_pixels)
+    assert catalog.get_healpix_pixels() == margin_catalog_pixels
+
+    info = catalog.catalog_info
+    assert info.catalog_name == "small_sky_xmatch_margin"
+    assert info.catalog_type == CatalogType.MARGIN
+    assert info.primary_catalog == "small_sky_xmatch"
+    assert info.margin_threshold == 7200
+
+
+def test_empty_directory(tmp_path, margin_cache_catalog_info_data, margin_catalog_pixels):
+    """Test loading empty or incomplete data"""
+    ## Path doesn't exist
+    with pytest.raises(FileNotFoundError):
+        MarginCatalog.read_from_hipscat(os.path.join("path", "empty"))
+
+    catalog_path = os.path.join(tmp_path, "empty")
+    os.makedirs(catalog_path, exist_ok=True)
+
+    ## Path exists but there's nothing there
+    with pytest.raises(FileNotFoundError, match="catalog info"):
+        MarginCatalog.read_from_hipscat(catalog_path)
+
+    ## catalog_info file exists - getting closer
+    file_name = os.path.join(catalog_path, "catalog_info.json")
+    with open(file_name, "w", encoding="utf-8") as metadata_file:
+        metadata_file.write(json.dumps(margin_cache_catalog_info_data))
+
+    with pytest.raises(FileNotFoundError, match="metadata"):
+        MarginCatalog.read_from_hipscat(catalog_path)
+
+    ## Now we create the needed _metadata and everything is right.
+    part_info = PartitionInfo.from_healpix(margin_catalog_pixels)
+    part_info.write_to_metadata_files(catalog_path=catalog_path)
+
+    catalog = MarginCatalog.read_from_hipscat(catalog_path)
+    assert catalog.catalog_name == margin_cache_catalog_info_data["catalog_name"]