Generate candidates-elected.csv files from results.json.

results/gen-elected.php

@@ -1,8 +1,6 @@
 <?php
 
-//require_once("parse_generic_csv.php");
+$options = getopt("", ["candidates-files:", "results-file:"]);
-
-$options = getopt("", ["candidates-files:"]);
 
 if (isset($options['candidates-files'])) {
     $candidates_files = $options['candidates-files'];
@@ -11,10 +9,50 @@ if (isset($options['candidates-files'])) {
     exit(1);
 }
 
+if (isset($options['results-file'])) {
+    $results_file = $options['results-file'];
+    $results_string = file_get_contents($results_file);
+
+    if ($results_string !== FALSE) {
+        $results = json_decode($results_string, true);
+    } else {
+        error_log("Error opening results.json.");
+        exit(1);
+    }
+} else {
+    error_log("Error: Missing required option '--results-file'.");
+    exit(1);
+}
+
 function trim_sluggify($input) {
     return strtolower(str_replace(' ', '-', trim($input)));
 }
 
+function match_words($words, $list) {
+    /* Match database names to VEC names */
+    $max_score = 0;
+    $best_match = "no match";
+    foreach ($list as $possible_match) {
+        $aa = preg_split("/[^a-z]/", strtolower($words));
+        $bb = preg_split("/[^a-z]/", strtolower($possible_match));
+
+        $score_sum = 0;
+        foreach ($aa as $a) {
+            foreach ($bb as $b) {
+                similar_text($a, $b, $score);
+                if ($score > 70) $score_sum += $score;
+                else $score_sum -= 1;
+            }
+        }
+
+        if ($score_sum > $max_score) {
+            $max_score = $score_sum;
+            $best_match = $possible_match;
+        }
+    }
+    return array($max_score, $best_match);
+}
+
 $candidates_files = explode(" ", $candidates_files);
 
 /* Generate dictionary of candidates and LGAs */
@@ -30,6 +68,8 @@ foreach ($candidates_files as $file) {
         exit(1);
     }
 
+    $candidate_data[$config['councilName']]['_filename'] = $file;
+
     if (($handle = fopen($file, "r")) !== FALSE) {
         $headers = fgetcsv($handle);
         while (($data = fgetcsv($handle)) !== FALSE) {
@@ -37,122 +77,66 @@ foreach ($candidates_files as $file) {
             foreach ($headers as $key => $value) {
                 $candidate[$value] = $data[$key];
             }
-            $candidate['Council'] = $config['councilName'];
             $name_slug = trim_sluggify($candidate['Candidate Name']);
-            $candidate_data[$name_slug] = $candidate;
+            $candidate_data[$config['councilName']][$name_slug] = $candidate;
         }
     }
 }
 
-print_r($candidate_data);
+$vec_lga_names = [];
+foreach ($results as $lga => $data) {
+    $vec_lga_names[] = $lga;
+}
 
-/* Get list of elected candidates */
+function was_elected($candidate, $vec_wards) {
+    foreach ($vec_wards as $vec_candidates) {
+        list($score, $match) = match_words($candidate, $vec_candidates);
+        if ($score > 100) return true;
+    }
+    return false;
+}
 
-//$lga_list = [];
+$header = ["Ward", "Candidate Name", "Elected"];
-/* Generate dictionary of LGAs and Wards */
-//foreach ($config_files as $config_file) {
-//    $config_string = file_get_contents($config_file);
-//    if ($config_string !== FALSE) {
-//        $config = json_decode($config_string, true);
-//    } else {
-//        error_log("Error opening config.json.");
-//        exit(1);
-//    }
-//    $config['config-file'] = $config_file;
-//    $lga_list[] = $config;
-//}
 
-/* Match user typed LGA/Ward to our database */
+foreach ($candidate_data as $lga => $db_candidates) {
-//match_lga($candidate_data, $lga_list);
+    /* Find LGA in results dict */
+    list($score, $vec_lga_name) = match_words($lga, $vec_lga_names);
+    $vec_wards = $results[$vec_lga_name];
 
-$header = ["Ward", "Candidate Name", "Rating", "Pledge", "Picture"];
+    $elected = [];
+    /* Go through database candidates and build list of elected candidates */
+    foreach ($db_candidates as $key => $value) {
+        if ($key === '_filename') {
+            $output_file = dirname($value)."/candidates-elected.csv";
+            continue;
+        }
+        if (was_elected($value['Candidate Name'], $vec_wards)) {
+            $elected[] = $value;
+        }
+    }
 
-/* Generate candidates-generic.csv */
+    /* Don't create file if none were elected. */
-//foreach ($lga_list as $lga) {
+    if (count($elected) === 0) continue;
-//    $lga_candidates = array_filter($candidate_data, function ($candidate) use ($lga) {
+
-//        return $candidate['match_lga'] === $lga['slug'];
+    if (($handle = fopen($output_file, "w")) === FALSE) {
-//    });
+        error_log('Error opening output file');
-//
+        exit(1);
-//    if (count($lga_candidates) === 0) continue;
+    }
-//
+
-//    remove_duplicates($lga_candidates);
+    if (fputcsv($handle, $header) === FALSE) {
-//
+        error_log('Error writing headers to output file');
-//    $dir = dirname($lga['config-file']);
+        exit(3);
-//    $dir_files = scandir($dir);
+    }
-//    $output_file = $dir."/candidates-generic.csv";
+
-//    $override_file = $dir."/candidates-override.csv";
+    foreach ($elected as $candidate) {
-//
+        $line = array($candidate['Ward'], $candidate['Candidate Name'], "y");
-//    if (($handle = fopen($output_file, "w")) === FALSE) {
+        if (fputcsv($handle, $line) === FALSE) {
-//        error_log('Error opening output file');
+            error_log('Error writing candidate to output file');
-//        exit(1);
+            exit(3);
-//    }
+        }
-//
+    }
-//    if (fputcsv($handle, $header) === FALSE) {
+
-//        error_log('Error writing headers to output file');
+    fclose($handle);
-//        exit(3);
+}
-//    }
-//
-//    $lines = [];
-//    foreach ($lga_candidates as $candidate) {
-//        /* Add extension to photo hash */
-//        if (strlen($candidate['Photo'])) {
-//            foreach ($dir_files as $file) {
-//                if (preg_match("/\.json$/", $file)) continue;
-//                if (strstr($file, $candidate['Photo'])) {
-//                    $candidate['Photo'] = $file;
-//                }
-//            }
-//        }
-//
-//        $lines[] = [
-//            $candidate['match_ward'],
-//            $candidate['Name'],
-//            $candidate['Score'],
-//            $candidate['Pledge'],
-//            $candidate['Photo'],
-//        ];
-//    }
-//
-//    /* Apply overrides if they exist */
-//    $overrides = [];
-//    if (file_exists($override_file)) {
-//        if (($ovr_handle = fopen($override_file, "r")) !== FALSE) {
-//            $headers = fgetcsv($ovr_handle);
-//            while (($data = fgetcsv($ovr_handle)) !== FALSE) {
-//                $override = [];
-//                foreach ($headers as $key => $value) {
-//                    $override[$value] = $data[$key];
-//                }
-//                $overrides[] = $override;
-//            }
-//            fclose($ovr_handle);
-//        } else {
-//            error_log('Error opening overrides file');
-//            exit(3);
-//        }
-//    }
-//
-//    foreach ($overrides as $override) {
-//        foreach ($lines as $line_key => $line) {
-//            $match_index = array_search($override['Match Field'], $header); 
-//            $replace_index = array_search($override['Replace Field'], $header); 
-//            if ($line[$match_index] === $override['Match Value']) {
-//                if ($replace_index !== false)
-//                    $lines[$line_key][$replace_index] = $override['Replace Value'];
-//                else /* If 'Replace Field' is not matched - delete this entry */
-//                    $lines[$line_key]['Delete'] = 'y';
-//            }
-//        }
-//    }
-//
-//    foreach ($lines as $line) {
-//        if (isset($line['Delete'])) continue;
-//        if (fputcsv($handle, $line) === FALSE) {
-//            error_log('Error writing candidate to output file');
-//            exit(3);
-//        }
-//    }
-//    fclose($handle);
-//}
 
 exit(0);

update-elected.sh

@@ -18,4 +18,5 @@ for folder in "$DATA_PATH"/*; do
     fi
 done
 
-php results/gen-elected.php --candidates-files "${candidates_files[*]}"
+php results/gen-elected.php --candidates-files "${candidates_files[*]}" \
+                            --results-file $DATA_PATH/results.json