From f9c151bfaef95ff4ce19d4054bdd9105d00d71bd Mon Sep 17 00:00:00 2001
From: Kim Taylor <kmtaylor@gmx.com>
Date: Sun, 22 Sep 2024 22:57:35 +1000
Subject: [PATCH] Start generating image download script.

---
 csv-generic/csv-to-json.php       | 20 ------------
 csv-generic/gen-generic.php       | 45 +++-----------------------
 csv-generic/gen-image-map.php     | 54 +++++++++++++++++++++++++++++++
 csv-generic/parse_generic_csv.php | 42 ++++++++++++++++++++++++
 get-generic.sh                    | 13 ++++++--
 5 files changed, 111 insertions(+), 63 deletions(-)
 delete mode 100644 csv-generic/csv-to-json.php
 create mode 100644 csv-generic/gen-image-map.php

diff --git a/csv-generic/csv-to-json.php b/csv-generic/csv-to-json.php
deleted file mode 100644
index 9bef38b..0000000
--- a/csv-generic/csv-to-json.php
+++ /dev/null
@@ -1,20 +0,0 @@
-<?php
-
-require_once("parse_generic_csv.php");
-
-$options = getopt("", ["generic-csv:"]);
-
-if (isset($options['generic-csv'])) {
-    $generic_csv = $options['generic-csv'];
-} else {
-    error_log("Error: Missing required option '--generic-csv'.");
-    exit(1);
-}
-
-$candidate_data = parse_generic_csv($generic_csv);
-
-$json_data = json_encode($candidate_data);
-
-print_r($json_data);
-
-exit(0);
diff --git a/csv-generic/gen-generic.php b/csv-generic/gen-generic.php
index 642c299..a0901ce 100644
--- a/csv-generic/gen-generic.php
+++ b/csv-generic/gen-generic.php
@@ -35,6 +35,9 @@ foreach ($config_files as $config_file) {
     $lga_list[] = $config;
 }
 
+/* Match user typed LGA/Ward to our database */
+match_lga($candidate_data, $lga_list);
+
 /* Calculate score for candidate */
 foreach ($candidate_data as &$candidate) {
     $score = 0;
@@ -48,46 +51,6 @@ foreach ($candidate_data as &$candidate) {
     $candidate['Score'] = $score;
 }
 
-foreach ($candidate_data as &$candidate) {
-    /* Match user typed LGA/Ward to our database */
-    $max_score = 0;
-    foreach ($lga_list as $lga) {
-        $aa = preg_split("/[^a-z]/", strtolower($candidate['LGA']));
-        $bb = preg_split("/[^a-z]/", $lga['slug']);
-
-        $score_sum = 0;
-        foreach ($aa as $a) {
-            foreach ($bb as $b) {
-                similar_text($a, $b, $score);
-                if ($score > 70) $score_sum += $score;
-                else $score_sum -= 10;
-            }
-        }
-
-        if ($score_sum > $max_score) {
-            $max_score = $score_sum;
-            $match_lga = $lga;
-        }
-    }
-
-    $max_score = 0;
-    foreach ($match_lga['wardNames'] as $ward) {
-        similar_text($ward, $candidate['Ward'], $score);
-        if ($score > $max_score) {
-            $max_score = $score;
-            $match_ward = $ward;
-        }
-    }
-
-    $candidate['match_lga'] = $match_lga['slug'];
-    $candidate['match_ward'] = $match_ward;
-}
-
-/* Get picture */
-foreach ($candidate_data as &$candidate) {
-    $candidate['match_picture'] = "";
-}
-
 $header = ["Ward", "Candidate Name", "Rating", "Picture"];
 
 /* Generate candidates-generic.csv */
@@ -115,7 +78,7 @@ foreach ($lga_list as $lga) {
             $candidate['match_ward'],
             $candidate['Name'],
             $candidate['Score'],
-            $candidate['match_picture'],
+            $candidate['Photo'],
         ];
 
         if (fputcsv($handle, $fields) === FALSE) {
diff --git a/csv-generic/gen-image-map.php b/csv-generic/gen-image-map.php
new file mode 100644
index 0000000..3c94c1b
--- /dev/null
+++ b/csv-generic/gen-image-map.php
@@ -0,0 +1,54 @@
+<?php
+
+require_once("parse_generic_csv.php");
+
+$options = getopt("", ["generic-csv:", "config-files:"]);
+
+if (isset($options['generic-csv'])) {
+    $generic_csv = $options['generic-csv'];
+} else {
+    error_log("Error: Missing required option '--generic-csv'.");
+    exit(1);
+}
+
+if (isset($options['config-files'])) {
+    $config_files = $options['config-files'];
+} else {
+    error_log("Error: Missing required option '--config-files'.");
+    exit(1);
+}
+
+$config_files = explode(" ", $config_files);
+$candidate_data = parse_generic_csv($generic_csv);
+
+$lga_list = [];
+/* 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[] = $config;
+}
+
+/* Match user typed LGA/Ward to our database */
+match_lga($candidate_data, $lga_list);
+
+$image_map = [];
+foreach ($candidate_data as $candidate) {
+    if (strlen($candidate['photo_url'])) {
+        $map['url'] = $candidate['photo_url'];
+        $map['match_lga'] = $candidate['match_lga'];
+        $image_map[$candidate['Photo']] = $map;
+    }
+}
+
+$json_data = json_encode($image_map);
+
+print_r($json_data);
+
+exit(0);
diff --git a/csv-generic/parse_generic_csv.php b/csv-generic/parse_generic_csv.php
index 4c68136..c90ee05 100644
--- a/csv-generic/parse_generic_csv.php
+++ b/csv-generic/parse_generic_csv.php
@@ -23,6 +23,11 @@ function parse_generic_csv($generic_csv) {
                     if (strstr($data[$key], "I pledge")) $data[$key] = "Yes";
                     else $data[$key] = "No";
                 }
+                
+                if ($value === "Photo") {
+                    $candidate['photo_url'] = $data[$key];
+                    $data[$key] = preg_filter("/.*id=/", "", $data[$key]);
+                }
 
                 if ($is_question) {
                     $candidate['q'.$question_no++] = $data[$key];
@@ -44,3 +49,40 @@ function parse_generic_csv($generic_csv) {
 
     return $candidate_data;
 }
+
+function match_lga(&$candidate_data, $lga_list) {
+    foreach ($candidate_data as &$candidate) {
+        /* Match user typed LGA/Ward to our database */
+        $max_score = 0;
+        foreach ($lga_list as $lga) {
+            $aa = preg_split("/[^a-z]/", strtolower($candidate['LGA']));
+            $bb = preg_split("/[^a-z]/", $lga['slug']);
+
+            $score_sum = 0;
+            foreach ($aa as $a) {
+                foreach ($bb as $b) {
+                    similar_text($a, $b, $score);
+                    if ($score > 70) $score_sum += $score;
+                    else $score_sum -= 10;
+                }
+            }
+
+            if ($score_sum > $max_score) {
+                $max_score = $score_sum;
+                $match_lga = $lga;
+            }
+        }
+
+        $max_score = 0;
+        foreach ($match_lga['wardNames'] as $ward) {
+            similar_text($ward, $candidate['Ward'], $score);
+            if ($score > $max_score) {
+                $max_score = $score;
+                $match_ward = $ward;
+            }
+        }
+
+        $candidate['match_lga'] = $match_lga['slug'];
+        $candidate['match_ward'] = $match_ward;
+    }
+}
diff --git a/get-generic.sh b/get-generic.sh
index dd1060a..28ef8ed 100755
--- a/get-generic.sh
+++ b/get-generic.sh
@@ -15,6 +15,15 @@ for folder in "$DATA_PATH"/*; do
     fi
 done
 
-#php csv-generic/csv-to-json.php --generic-csv $DATA_LOC/responses.csv
 
-php csv-generic/gen-generic.php --generic-csv $GENERIC_SURVEY --config-files "${config_files[*]}"
+image_map=$(php csv-generic/gen-image-map.php --generic-csv $GENERIC_SURVEY --config-files "${config_files[*]}")
+
+for key in $(jq -r 'keys[]' <<< $image_map) ; do
+    url=$(jq -r ".[\"$key\"][\"url\"]" <<< $image_map)
+    lga=$(jq -r ".[\"$key\"][\"match_lga\"]" <<< $image_map)
+    dst=$DATA_PATH/$lga/$key
+    echo wget $url -O $dst
+    break
+done
+
+#php csv-generic/gen-generic.php --generic-csv $GENERIC_SURVEY --config-files "${config_files[*]}"