From d9cb183b2acbaeafbf4d916891ba1408f7c9e015 Mon Sep 17 00:00:00 2001
From: Matt Way <mattyway@gmail.com>
Date: Thu, 15 Aug 2024 22:10:03 +1000
Subject: [PATCH] Implement CSV normaliser logic

---
 csv-normaliser/main.php | 169 +++++++++++++++++++++++++++++-----------
 1 file changed, 122 insertions(+), 47 deletions(-)

diff --git a/csv-normaliser/main.php b/csv-normaliser/main.php
index 39cec97..ae1d4f2 100644
--- a/csv-normaliser/main.php
+++ b/csv-normaliser/main.php
@@ -1,69 +1,144 @@
 <?php
-require_once "page_renderer.php";
+$options = getopt("", ["input:", "output:", "media:"]);
 
-$options = getopt("", ["council-file:", "candidates-file:", "media-file:"]);
-
-if (isset($options['council-file'])) {
-    $councilFileContents = file_get_contents($options['council-file']);
+if (isset($options['input'])) {
+    $inputFile = $options['input'];
 } else {
-    error_log("Error: Missing required option '--council-file'.");
+    error_log("Error: Missing required option '--input'.");
     exit(1);
 }
 
-$councilData = json_decode($councilFileContents, true);
-
-// Check for decoding errors
-if (json_last_error() !== JSON_ERROR_NONE) {
-    error_log('Error decoding council file: ' . json_last_error_msg());
-    exit(1);
-}
-
-if (isset($options['candidates-file'])) {
-    $candidatesFile = $options['candidates-file'];
+if (isset($options['output'])) {
+    $outputFile = $options['output'];
 } else {
-    error_log("Error: Missing required option '--candidates-file'.");
+    error_log("Error: Missing required option '--output'.");
     exit(1);
 }
 
-// Convert CSV into an array of dictionaries. Use the header as the key in the dictionary.
-$candidateData = [];
-if (($handle = fopen($candidatesFile, "r")) !== FALSE) {
-    $headers = fgetcsv($handle);
+if (isset($options['media'])) {
+    $mediaFolder = $options['media'];
+} else {
+    error_log("Error: Missing required option '--media'.");
+    exit(1);
+}
+
+$mediaFiles = scandir($mediaFolder);
+if ($mediaFiles === FALSE) {
+    error_log("Failed to list files in media folder");
+    exit(1);
+}
+
+$candidates = [];
+if (($handle = fopen($inputFile, "r")) !== FALSE) {
+    $currentWard = null;
+    $currentLine = 0;
     while (($data = fgetcsv($handle)) !== FALSE) {
-        $candidate = [];
-        foreach ($headers as $key => $value) {
-            $candidate[$value] = $data[$key];
+        $currentLine++;
+        //echo var_dump($data);
+        if ($data[0] == "Ward") {
+            // CSV contains ward names in uppercase, convert them to a more readable form
+            $currentWard = ucwords(strtolower($data[1]));
+
+            // Handle some special cases where the above logic doesn't match the expected names
+            // Note that we cannot just convert every letter after a '-' character to uppercase
+            // because there are some ward names like "Bulleke-bek"
+            if ($currentWard == "Warrk-warrk") {
+                $currentWard = "Warrk-Warrk";
+            }
+            if ($currentWard == "Djirri-djirri") {
+                $currentWard = "Djirri-Djirri";
+            }
+            if ($currentWard == "Coastal-promontory") {
+                $currentWard = "Coastal-Promontory";
+            }
+        }
+        if ($data[0] == "Candidate") {
+            if ($currentWard == null) {
+                error_log("No ward found, skipping data on line " . $currentLine);
+                continue;
+            }
+
+            $candidateName = $data[1];
+
+            if ($candidateName == " example name") {
+                error_log("Skipping line ". $currentLine);
+                continue;
+            }
+
+            print("Adding candidate to " . $candidateName . " to ". $currentWard . "\n");
+
+            $name_split = explode(" ", $data[1]);
+
+            $name_patterns = [
+                implode(".*", $name_split),
+                implode(".*", array_reverse($name_split)),
+            ];
+
+            $regex_groups = array_map(function($x) { return "(?:.*" . $x . ".*)"; }, $name_patterns);
+
+            $regex_pattern = "/" . implode("|", $regex_groups) . "/i";
+
+            $picture = "";
+            foreach ($mediaFiles as $mediaFile) {
+                if (preg_match($regex_pattern, $mediaFile)) {
+                    $picture = $mediaFile;
+                    break;
+                }
+            }
+            if ($picture === "") {
+                print("Failed to identify picture for " . $candidateName);
+            }
+
+            array_push(
+                $candidates, 
+                [
+                    "Ward" => $currentWard,
+                    "Candidate Name" => $candidateName,
+                    "Rating" => $data[2],
+                    "Picture" => $picture
+                ]
+            );
         }
-        $candidateData[] = $candidate;
     }
     fclose($handle);
 } else {
-    error_log('Error opening candidates file');
+    error_log('Error opening input file');
     exit(1);
 }
 
-$candidateData = array_filter($candidateData, function ($candidate) use ($councilData) {
-    return isset($candidate["Council"]) && $candidate["Council"] === $councilData['shortName'];
-});
-
-if (empty($candidateData)) {
-    error_log("Failed to load any candidates for " . $councilData['shortName']);
-}
-
-if (isset($options['media-file'])) {
-    $mediaFileContents = file_get_contents($options['media-file']);
-} else {
-    error_log("Error: Missing required option '--media-file'.");
-    exit(1);
-}
-
-$mediaData = json_decode($mediaFileContents, true);
-
-$renderer = new SPLPageRenderer();
-$pageContent = $renderer->renderCouncilPage($councilData, $candidateData, $mediaData);
-if ($pageContent === null) {
+if (empty($candidates)) {
+    error_log("Failed to find any candidates");
     exit(2);
 }
 
-echo $pageContent;
+if (($handle = fopen($outputFile, "w")) !== FALSE) {
+    $headers = array(
+        "Ward",
+        "Candidate Name",
+        "Rating",
+        "Picture"
+    );
+    if (fputcsv($handle, $headers) === FALSE) {
+        error_log('Error writing headers to output file');
+        exit(3);
+    }
+    foreach ($candidates as $candidate) {
+        $fields = array(
+            $candidate["Ward"],
+            $candidate["Candidate Name"],
+            $candidate["Rating"],
+            $candidate["Picture"]
+        );
+        if (fputcsv($handle, $fields) === FALSE) {
+            error_log('Error writing candidate to output file');
+            exit(3);
+        }
+    }
+} else {
+    error_log('Error opening output file');
+    exit(1);
+}
+
+print("Data written to " . $outputFile);
+
 exit(0);