From fff24136a1bc511f1924a591540543d983045354 Mon Sep 17 00:00:00 2001
From: Matt Way <mattyway@gmail.com>
Date: Mon, 23 Sep 2024 23:06:18 +1000
Subject: [PATCH] Improve logic for identifying photos in csv-normaliser

---
 csv-normaliser/main.php | 12 ++++++++----
 1 file changed, 8 insertions(+), 4 deletions(-)

diff --git a/csv-normaliser/main.php b/csv-normaliser/main.php
index df4e991..a3e39d7 100644
--- a/csv-normaliser/main.php
+++ b/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;