From e89ac3db2e54178dfedd5341d8808c3baa3a6a6a Mon Sep 17 00:00:00 2001
From: fozy81 <fozy81@hotmail.com>
Date: Sat, 17 Jun 2023 17:25:16 +0100
Subject: [PATCH 1/2] fix: remove duplicate word plural forms

---
 ODSCategories.json | 19 +------------------
 merge_data.py      | 20 +++++++++++++++++++-
 2 files changed, 20 insertions(+), 19 deletions(-)

diff --git a/ODSCategories.json b/ODSCategories.json
index 3eefe34..89e06b3 100644
--- a/ODSCategories.json
+++ b/ODSCategories.json
@@ -17,7 +17,6 @@
       "memorial",
       "military",
       "museum",
-      "museums",
       "poems",
       "theatre"
    ],
@@ -76,7 +75,6 @@
       "shop",
       "shopping",
       "shopping centres",
-      "shops",
       "simd",
       "social security",
       "tenders",
@@ -106,7 +104,6 @@
       "council policies",
       "council tax",
       "councillors",
-      "councils",
       "data protection",
       "data zone lookup",
       "disability living allowance",
@@ -134,21 +131,17 @@
       "pupil",
       "school",
       "school meals",
-      "schools",
       "scqf",
       "sqa",
       "student"
    ],
    "Elections / Politics": [
       "community council",
-      "community councils",
       "council area",
       "councillor",
-      "councillors",
       "democracy",
       "elecorate",
       "election",
-      "elections",
       "electoral",
       "electorate",
       "local authority",
@@ -239,7 +232,6 @@
       "health board",
       "hospital",
       "hospital admissions",
-      "hospitals",
       "human services",
       "implant",
       "induced abortion",
@@ -304,20 +296,17 @@
    ],
    "Law and Licensing": [
       "court",
-      "courts",
       "crime",
       "law",
       "licence",
       "licenses",
       "licensing",
       "permit",
-      "permits",
       "police",
       "policing",
       "regulation",
-      "regulations",
       "tribunal",
-      "tribunals", "Fixed Penalty Notices"
+      "Fixed Penalty Notices"
    ],
    "Parks / Recreation": [
       "features of interest",
@@ -390,8 +379,6 @@
       "parks",
       "societies",
       "sport",
-      "sport",
-      "sports",
       "sports activities",
       "sports clubs",
       "sports facilities", "Community Centres"
@@ -402,7 +389,6 @@
       "cafes",
       "historic buildings",
       "hotel",
-      "hotels",
       "public toilets",
       "restaurants",
       "tourism",
@@ -420,7 +406,6 @@
       "commuting",
       "core paths",
       "cycle",
-      "cycles",
       "cycling",
       "electric vehicle",
       "electric vehicle charging points",
@@ -440,9 +425,7 @@
       "road closures",
       "road safety",
       "road works",
-      "roads",
       "salt bin",
-      "salt bins",
       "scotrail",
       "speed limits",
       "station",
diff --git a/merge_data.py b/merge_data.py
index f81e435..382f33f 100644
--- a/merge_data.py
+++ b/merge_data.py
@@ -229,7 +229,23 @@ def tidy_categories(categories_string):
     data["OriginalTags"] = data["OriginalTags"].apply(tidy_categories)
     data["ManualTags"] = data["ManualTags"].apply(tidy_categories)
 
-    ### Creating dataset categories for ODS
+    ### Creating dataset categories for ODS   
+    def remove_trailing_s(string):
+        """Remove trailing 's' from all words in string to remove requirement to search for pural categories in
+
+        Args:
+            string: String to remove trialing 's' from
+
+        Returns:
+            string: the resulting string, with trailing 's' removed from all words.
+        """
+        s = []
+        words = string.split()
+        for word in words:
+            s.append(re.sub('s$', "", word))
+        sentence = ' '.join(s)
+        return sentence
+
     def find_keyword(str_tofind, str_findin):
         """Finds if single word or phrase exists in string
 
@@ -240,6 +256,8 @@ def find_keyword(str_tofind, str_findin):
         Returns:
             boolean: True if match is found
         """
+        str_findin = remove_trailing_s(str_findin)
+        str_tofind = remove_trailing_s(str_tofind)
         if re.search(r"\b" + re.escape(str_tofind) + r"\b", str_findin, re.I):
             return True
         return False

From 9385b6394ce30f157237a1022e53946d9b27bf05 Mon Sep 17 00:00:00 2001
From: fozy81 <fozy81@hotmail.com>
Date: Sat, 17 Jun 2023 19:15:06 +0100
Subject: [PATCH 2/2] fix: remove plural duplicates

- remove trailing 's'  or ' S' (case insensitive)
---
 merge_data.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/merge_data.py b/merge_data.py
index 382f33f..3a5d922 100644
--- a/merge_data.py
+++ b/merge_data.py
@@ -242,7 +242,7 @@ def remove_trailing_s(string):
         s = []
         words = string.split()
         for word in words:
-            s.append(re.sub('s$', "", word))
+            s.append(re.sub('[Ss]$', "", word))
         sentence = ' '.join(s)
         return sentence