Improve logic for identifying photos in csv-normaliser

Readme.md

@@ -35,7 +35,7 @@ The "short" named is created by taking the electorate name and removing the word
 The file can be generated using the `jq` tool and the VEC data:
 
 ```
-jq '[.[] | {name: .electorateName, electorateId: .electorateId, shortName: .parentElectorateName | match("(.*?)(?:(?: Rural)?(?: City| Shire) Council)").captures[0].string, parentElectorateId: .parentElectorateId, councilName: .parentElectorateName }] | group_by(.parentElectorateId) | map({shortName: .[0].shortName, slug: .[0].shortName | ascii_downcase | split(" ") | join("-"), councilName: .[0].councilName, wardNames: . | map(.name) }) | sort_by(.shortName)' "VEC Data\wards.json" > council_names.json
+jq '[.[] | {name: .electorateName, electorateId: .electorateId, shortName: .parentElectorateName | match("(.*?)(?:(?: Rural)?(?: City| Shire| Borough) Council)").captures[0].string, parentElectorateId: .parentElectorateId, councilName: .parentElectorateName }] | group_by(.parentElectorateId) | map({shortName: .[0].shortName, slug: .[0].shortName | ascii_downcase | split(" ") | join("-"), councilName: .[0].councilName, wardNames: . | map(.name) }) | sort_by(.shortName)' "VEC Data\wards.json" > council_names.json
 ```
 
 ## lga-links-filter

council_names.json

@@ -786,6 +786,14 @@
       "Beaufort"
     ]
   },
+  {
+    "shortName": "Queenscliffe",
+    "slug": "queenscliffe",
+    "councilName": "Queenscliffe Borough Council",
+    "wardNames": [
+      "Unsubdivided"
+    ]
+  },
   {
     "shortName": "South Gippsland",
     "slug": "south-gippsland",

csv-normaliser/main.php

@@ -96,19 +96,23 @@ if (($handle = fopen($inputFile, "r")) !== FALSE) {
 
             //print("Adding candidate " . $candidateName . " to ". $currentWard . "\n");
 
-            $name_split = explode(" ", str_replace(",", "", str_replace("'", "_", $data[1])));
+            $name_split = array_values(array_filter(explode(" ", str_replace(",", "", str_replace("'", "_", $data[1]))), function($value) { return !is_null($value) && $value !== ''; }));
 
             $name_patterns = [
-                implode(".*", $name_split),
-                implode(".*", array_reverse($name_split)),
+                ".*" . implode(".*", $name_split) . ".*",
+                ".*" . implode(".*", array_reverse($name_split)) . ".*",
+                "^" . $name_split[array_key_last($name_split)] . ".*",
+                "^" . $name_split[0] . ".*"
             ];
 
-            $regex_groups = array_map(function($x) { return "(?:.*" . $x . ".*)"; }, $name_patterns);
+            $regex_groups = array_map(function($x) { return "(?:" . $x . ")"; }, $name_patterns);
 
             $regex_pattern = "/" . implode("|", $regex_groups) . "/i";
 
             $picture = "";
             foreach ($mediaFiles as $mediaFile) {
+                if ($mediaFile == ".") continue;
+                if ($mediaFile == "..") continue;
                 if (preg_match($regex_pattern, $mediaFile)) {
                     $picture = $mediaFile;
                     break;

map-generator/src.js

@@ -1,7 +1,7 @@
 import polylabel from 'polylabel';
 
 function normaliseCouncilName(str) {
-    const regex = /(.*?)(?:(?: Rural)?(?: City| Shire) Council)/g;
+    const regex = /(.*?)(?:(?: Rural)?(?: City| Shire| Borough) Council)/g;
     const matches = str.matchAll(regex);
 
     // If we get a match, convert to slug format