Experimenting with toxicity.

composer/composer.json

@@ -46,7 +46,8 @@
     "prewk/xml-string-streamer": "^1.2",
     "sentry/sdk": "^3.1",
     "cweagans/composer-patches": "^1.7",
-    "stripe/stripe-php": "^16.2"
+    "stripe/stripe-php": "^16.2",
+    "stajor/perspectiveapi": "^2.0"
   },
   "repositories": [
     {

install/iznik.conf.php

@@ -38,6 +38,7 @@
 define('GOOGLE_CLIENT_SECRET', 'zzz');
 define('GOOGLE_APP_NAME', 'zzz');
 define('GOOGLE_SITE_VERIFICATION', 'zzz');
+define('GOOGLE_PERSPECTIVE_KEY', 'zzz');
 
 # Yahoo App keys
 define('YAHOO_APPID', 'zzz');

scripts/fix/fix_toxicity.php

@@ -0,0 +1,59 @@
+<?php
+
+namespace Freegle\Iznik;
+
+define('BASE_DIR', dirname(__FILE__) . '/../..');
+require_once(BASE_DIR . '/include/config.php');
+
+require_once(IZNIK_BASE . '/include/db.php');
+global $dbhr, $dbhm;
+
+$results = [];
+
+#$newsfeeds = $dbhr->preQuery("SELECT * FROM newsfeed WHERE message IS NOT NULL ORDER BY id DESC LIMIT 100;");
+$newsfeeds = $dbhr->preQuery("SELECT * FROM chat_messages WHERE message IS NOT NULL ORDER BY id DESC LIMIT 100;");
+
+foreach ($newsfeeds as $newsfeed) {
+    do {
+        $sleep = FALSE;
+
+        try {
+            $commentsClient = new \PerspectiveApi\CommentsClient(GOOGLE_PERSPECTIVE_KEY);
+            $commentsClient->comment(['text' => $newsfeed['message']]);
+            $commentsClient->languages(['en']);
+            $commentsClient->context(['entries' => ['text' => 'off-topic', 'type' => 'PLAIN_TEXT']]);
+            $commentsClient->requestedAttributes(['TOXICITY' => ['scoreType' => 'PROBABILITY', 'scoreThreshold' => 0]]);
+            $response = $commentsClient->analyze();
+            $tox = $response->attributeScores()['TOXICITY']['summaryScore']['value'];
+        } catch (\Exception $e) {
+            if (strpos($e->getMessage(), 'Quota exceeded')  !== FALSE) {
+                #error_log("...quota exceeded");
+                $sleep = TRUE;
+            } else {
+                throw $e;
+            }
+        }
+
+        if ($sleep) {
+            sleep(1);
+        }
+    } while ($sleep);
+
+    if ($tox > 0.3) {
+        #error_log("$tox for " . $newsfeed['id'] . " " . $newsfeed['message']);
+        $results[] = [
+            'id' => $newsfeed['id'],
+            'message' => $newsfeed['message'],
+            'tox' => $tox
+        ];
+    }
+}
+
+# Sort ascending by tox
+usort($results, function($a, $b) {
+    return $b['tox'] - $a['tox'];
+});
+
+foreach ($results as $result) {
+    echo ($result['tox'] * 100) . " " . $result['id'] . " " . $result['message'] . "\n";
+}