From 3ee7f87f583e7687b6d4c3be0b91aa9253e591b8 Mon Sep 17 00:00:00 2001
From: Abi Hunter <abihunter@gmail.com>
Date: Fri, 2 Feb 2024 15:45:26 -0500
Subject: [PATCH 1/2] add test for dedup config validation

---
 pdgstaging/ConfigManager.py | 14 +++++++++++++-
 tests/test_deduplicator.py  | 16 ++++++++++++++++
 2 files changed, 29 insertions(+), 1 deletion(-)
 create mode 100644 tests/test_deduplicator.py

diff --git a/pdgstaging/ConfigManager.py b/pdgstaging/ConfigManager.py
index 212c59f..d0bbfb7 100644
--- a/pdgstaging/ConfigManager.py
+++ b/pdgstaging/ConfigManager.py
@@ -1471,4 +1471,16 @@ 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
\ No newline at end of file
+        return rgb_hex
+
+
+def validate_dedup_rules(gdf, config):
+    gdf_attrs = gdf.attrs
+    allowed_comp_operators = ['smaller', 'larger']
+    for rule in config:
+        if rule[0] not in gdf_attrs.keys():
+            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}')
diff --git a/tests/test_deduplicator.py b/tests/test_deduplicator.py
new file mode 100644
index 0000000..9159578
--- /dev/null
+++ b/tests/test_deduplicator.py
@@ -0,0 +1,16 @@
+import geopandas as gpd
+import shapely
+from pdgstaging.ConfigManager import ConfigManager
+
+def test_validate_dedup_rules():
+    tests = {
+        "valid_rule1": {
+            "data": gpd.DataFrame({
+                "col1": ["a", "b"],
+                "geometry": [shapely.Point(1, 1), shapely.Point(1,2)]
+            }),
+            "rules": [["col1", "smaller"]]
+        }
+    }
+    for test in tests:
+        ConfigManager.validate_dedup_rules(test["data"], test["rules"])

From 0c42b6f2fd381b63e75828b6658a66b91e9074df Mon Sep 17 00:00:00 2001
From: Abi Hunter <abihunter@gmail.com>
Date: Mon, 5 Feb 2024 10:51:58 -0500
Subject: [PATCH 2/2] add dedup_rule validation and test for same

---
 pdgstaging/ConfigManager.py | 23 +++++++++++++----------
 tests/test_deduplicator.py  |  5 +++--
 2 files changed, 16 insertions(+), 12 deletions(-)

diff --git a/pdgstaging/ConfigManager.py b/pdgstaging/ConfigManager.py
index d0bbfb7..b78172f 100644
--- a/pdgstaging/ConfigManager.py
+++ b/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):
@@ -1474,13 +1487,3 @@ def color_list_from_cmaps(cmap_name):
         return rgb_hex
 
 
-def validate_dedup_rules(gdf, config):
-    gdf_attrs = gdf.attrs
-    allowed_comp_operators = ['smaller', 'larger']
-    for rule in config:
-        if rule[0] not in gdf_attrs.keys():
-            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}')
diff --git a/tests/test_deduplicator.py b/tests/test_deduplicator.py
index 9159578..d931621 100644
--- a/tests/test_deduplicator.py
+++ b/tests/test_deduplicator.py
@@ -2,15 +2,16 @@
 import shapely
 from pdgstaging.ConfigManager import ConfigManager
 
+
 def test_validate_dedup_rules():
     tests = {
         "valid_rule1": {
-            "data": gpd.DataFrame({
+            "data": gpd.GeoDataFrame({
                 "col1": ["a", "b"],
                 "geometry": [shapely.Point(1, 1), shapely.Point(1,2)]
             }),
             "rules": [["col1", "smaller"]]
         }
     }
-    for test in tests:
+    for test in tests.values():
         ConfigManager.validate_dedup_rules(test["data"], test["rules"])