Dedup rule validation

pdgstaging/ConfigManager.py

@@ -1433,6 +1433,19 @@ def list_updates(self):
 
         return updates
 
+    @staticmethod
+    def validate_dedup_rules(gdf, config):
+        gdf_cols = gdf.columns 
+        print("cols: " + str(gdf_cols))
+        allowed_comp_operators = ['smaller', 'larger']
+        for rule in config:
+            if rule[0] not in gdf_cols:
+                raise ValueError(
+                    f'invalid deduplication rule: property {rule[0]} does not exist in dataset')
+            if rule[1] not in allowed_comp_operators:
+                raise ValueError(
+                    f'invalid deduplication rule: second parameter must be one of {allowed_comp_operators}')
+
     @staticmethod
     def validate_palette(palette):
         if isinstance(palette, list):
@@ -1471,4 +1484,6 @@ def color_list_from_cmaps(cmap_name):
         pal_len = 10 if cmap.N > 10 else cmap.N
         rgb_vals = (cmap.discrete(pal_len).colors * 255).astype(int).tolist()
         rgb_hex = [f'#{i:02x}{j:02x}{k:02x}' for i, j, k in rgb_vals]
-        return rgb_hex
+        return rgb_hex
+
+

tests/test_deduplicator.py

@@ -0,0 +1,17 @@
+import geopandas as gpd
+import shapely
+from pdgstaging.ConfigManager import ConfigManager
+
+
+def test_validate_dedup_rules():
+    tests = {
+        "valid_rule1": {
+            "data": gpd.GeoDataFrame({
+                "col1": ["a", "b"],
+                "geometry": [shapely.Point(1, 1), shapely.Point(1,2)]
+            }),
+            "rules": [["col1", "smaller"]]
+        }
+    }
+    for test in tests.values():
+        ConfigManager.validate_dedup_rules(test["data"], test["rules"])