From 90faae59ec22588eb41b5a4d54c7d65f6493c4c8 Mon Sep 17 00:00:00 2001
From: JulianForeman <71847719+JulianForeman@users.noreply.github.com>
Date: Thu, 28 Nov 2024 10:26:14 -0800
Subject: [PATCH] Adding prep function that formats postal codes and validates
 them if asked to do so (#418)

---
 django/api/constants/constants.py             |  4 +-
 .../api/services/spreadsheet_uploader_prep.py | 39 ++++++++++++++++++-
 2 files changed, 41 insertions(+), 2 deletions(-)

diff --git a/django/api/constants/constants.py b/django/api/constants/constants.py
index 227dab2c..ffe20391 100644
--- a/django/api/constants/constants.py
+++ b/django/api/constants/constants.py
@@ -26,7 +26,8 @@
     location_checker,
     email_validator,
     validate_field_values,
-    region_checker
+    region_checker,
+    format_postal_codes
 )
 from api.services.resolvers import get_google_resolver
 from api.constants.misc import GER_VALID_FIELD_VALUES, ARC_VALID_FIELD_VALUES, LOCALITY_FEATURES_MAP, CVP_DATA_VALID_FIELD_VALUES
@@ -829,6 +830,7 @@ class CVPDataColumnMapping(Enum):
             {"function": location_checker, "columns": ["City"], "kwargs": {"columns_to_features_map": {"City": LOCALITY_FEATURES_MAP}, "indices_offset":2}},
             {"function": email_validator, "columns": ["Email"], "kwargs": {"indices_offset":2, "get_resolver": get_google_resolver}},
             {"function": validate_field_values, "columns": [], "kwargs": {"indices_offset":2, "fields_and_values": GER_VALID_FIELD_VALUES}},
+            {"function": format_postal_codes, "columns": ["Postal code"], "kwargs": {"indices_offset":2, "validate": True}}
         ]
     },
     "CVP Data": {
diff --git a/django/api/services/spreadsheet_uploader_prep.py b/django/api/services/spreadsheet_uploader_prep.py
index 8d8317bf..7e4db9a6 100644
--- a/django/api/services/spreadsheet_uploader_prep.py
+++ b/django/api/services/spreadsheet_uploader_prep.py
@@ -410,4 +410,41 @@ def region_checker(df, *columns, **kwargs):
                     }
                 }
         
-    return result
\ No newline at end of file
+    return result
+
+def format_postal_codes(df, *columns, **kwargs):
+    validate = kwargs.get('validate', False)
+    indices_offset = kwargs.get("indices_offset", 0)
+    
+    result = {}
+    
+    for column in columns:
+        series = df[column]
+        map_of_values_to_indices = get_map_of_values_to_indices(series, indices_offset)
+        invalid_groups = []
+
+        for value, indices in map_of_values_to_indices.items():
+            clean_value = value.replace(" ", "") if isinstance(value, str) else ""
+            
+            if len(clean_value) == 6:
+                formatted_value = clean_value[:3] + " " + clean_value[3:]
+                for index in indices:
+                    df.at[index - indices_offset, column] = formatted_value
+            elif validate:
+                if pd.isna(value) or value == "":
+                    value = "Empty"
+                invalid_groups.append({
+                    "Invalid Postal Code": value,
+                    "Rows": indices
+                })
+        
+        if validate and invalid_groups:
+            result[column] = {
+                "Invalid Postal Codes": {
+                    "Expected Type": "Postal Code should be formatted as (XXX XXX)",
+                    "Groups": invalid_groups,
+                    "Severity": "Warning"
+                }
+            }
+    
+    return result if validate else None