write function that gets canonical path based on available public query vars on site

Author: dannyvankooten

src/class-burst-importer.php

@@ -22,7 +22,7 @@ public static function show_page(): void
 
          // grab start date
         global $wpdb;
-        $date_start = $wpdb->get_var("SELECT MIN(date) FROM {$wpdb->prefix}burst_summary;");
+        $date_start = $wpdb->get_var("SELECT MIN(date) FROM {$wpdb->prefix}burst_summary");
         $date_end = $wpdb->get_var("SELECT MAX(date) FROM {$wpdb->prefix}burst_summary");
 
         ?>

src/class-pageview-aggregator.php

@@ -284,7 +284,7 @@ public function clean_url(string $url): string
             parse_str($query_str, $params);
 
             // strip all but the following query parameters from the URL
-            $allowed_params = [ 'page_id', 'p', 'cat', 'product' ];
+            $allowed_params = [ 'page_id', 'p', 'cat', 'tag', 'product' ];
             $new_params     = array_intersect_key($params, array_flip($allowed_params));
             $new_query_str  = http_build_query($new_params);
             $new_url        = substr($url, 0, $pos + 1) . $new_query_str;

src/class-script-loader.php

@@ -63,6 +63,30 @@ private static function get_post_id(): string
         return "-1";
     }
 
+    public static function get_canonical_path(): string
+    {
+        global $wp;
+
+        $base_path = parse_url(home_url('/'), PHP_URL_PATH);
+        $request_path = $_SERVER['REQUEST_URI'];
+
+        if (str_starts_with($request_path, $base_path)) {
+            $request_path = substr($request_path, strlen($base_path));
+        }
+
+        $question_mark_pos = strpos($request_path, '?');
+        if ($question_mark_pos !== false) {
+            $query_string = parse_url($request_path, PHP_URL_QUERY);
+            $params = [];
+            parse_str($query_string, $params);
+            $new_params     = array_intersect_key($params, array_flip($wp->public_query_vars));
+            $new_query_str  = http_build_query($new_params);
+            $request_path        = rtrim(substr($request_path, 0, $question_mark_pos + 1) . $new_query_str, '?');
+        }
+
+        return '/' . ltrim($request_path, '/');
+    }
+
     private static function get_tracker_url(): string
     {
         if (\defined('KOKO_ANALYTICS_CUSTOM_ENDPOINT') && KOKO_ANALYTICS_CUSTOM_ENDPOINT) {
@@ -89,7 +113,7 @@ public static function print_js_object()
             'url'   => self::get_tracker_url(),
             'site_url' => get_home_url(),
 
-            'path' => '/' . ltrim(add_query_arg($wp->query_vars, $wp->request), '/'),
+            'path' => self::get_canonical_path(),
 
             // ID of the current post (if singular post type), path name otherwise
             'post_id'       => (string) self::get_post_id(),

tests/FunctionsTest.php

@@ -25,15 +25,14 @@ public function testExtractPageviewData(): void
 
        // complete but invalid
         $this->assertEquals(extract_pageview_data(['p' => '', 'nv' => '', 'up' => '']), []);
-        $this->assertEquals(extract_pageview_data(['p' => 'x', 'nv' => '2', 'up' => '3']), []);
         $this->assertEquals(extract_pageview_data(['p' => '1', 'nv' => 'x', 'up' => '3']), []);
         $this->assertEquals(extract_pageview_data(['p' => '1', 'nv' => '2', 'up' => 'x']), []);
         $this->assertEquals(extract_pageview_data(['p' => '1', 'nv' => '2', 'up' => '3', 'r' => 'not an url']), []);
 
         // complete and valid
-        foreach (
-            [
+        foreach ([
             [['p' => '1', 'nv' => '2', 'up' => '3'], ['p', null, 1, 2, 3, '']],
+            [['p' => '/', 'nv' => '2', 'up' => '3'], ['p', null, '/', 2, 3, '']],
             [['p' => '1', 'nv' => '2', 'up' => '3', 'r' => ''], ['p', null, 1, 2, 3, '']],
             [['p' => '1', 'nv' => '2', 'up' => '3', 'r' => 'https://www.kokoanalytics.com'], ['p', null, 1, 2, 3, 'https://www.kokoanalytics.com']],
 

tests/ScriptLoaderTest.php

@@ -6,6 +6,7 @@
 
 use KokoAnalytics\Script_Loader;
 use PHPUnit\Framework\TestCase;
+use stdClass;
 
 final class ScriptLoaderTest extends TestCase
 {
@@ -14,4 +15,19 @@ public function testCanInstantiate(): void
         $i = new Script_Loader();
         self::assertTrue($i instanceof Script_Loader);
     }
+
+    public function test_get_canonical_path(): void
+    {
+        $GLOBALS['wp'] = new stdClass;
+        $GLOBALS['wp']->public_query_vars = ['p', 'page_id', 'tag', 'cat'];
+
+        $_SERVER['REQUEST_URI'] = '/';
+        self::assertEquals(Script_Loader::get_canonical_path(), '/');
+
+        $_SERVER['REQUEST_URI'] = '/about/?utm_medium=Email';
+        self::assertEquals(Script_Loader::get_canonical_path(), '/about/');
+
+        $_SERVER['REQUEST_URI'] = '/about/?utm_medium=Email&foo=bar';
+        self::assertEquals(Script_Loader::get_canonical_path(), '/about/');
+    }
 }

tests/benchmarks/functions.php

@@ -10,5 +10,7 @@ function bench(Closure $fn, $iterations = 1000)
         $fn();
     }
     $time_end = microtime(true);
+
+    // in microseconds
     return round(($time_end - $time_start) * 1000 * 1000, 2);
 }

tests/mocks.php

@@ -151,6 +151,11 @@ function _deprecated_function($function, $version, $replacement)
 {
 }
 
+function home_url(string $path = '/')
+{
+    return 'http://localhost:8080' . $path;
+}
+
 class wpdb_mock
 {
     public $prefix = '';