Detect the tracked page URL from REQUEST_URI instead of PATH_INFO (#154)

* Detect the tracked page URL from REQUEST_URI, not PATH_INFO (#141)

getCurrentScriptName() builds the auto-detected page URL (the path
between host and query string). It preferred $_SERVER['PATH_INFO'],
which only holds the trailing path-info segment — so with front-
controller / path-info routing (e.g. /dir1/page handled by
dir1/index.php) the tracker recorded a truncated '/page' instead of
'/dir1/page'.

REQUEST_URI already contains the full requested path (PATH_INFO is
always just a suffix of it), so use it as the source and drop PATH_INFO
entirely; SCRIPT_NAME stays as the fallback when REQUEST_URI is absent.
This also aligns the primary source with Matomo core's Url helper.

Reported in #141.

* Drop the ticket number from an inline test comment

Keep issue references in the CHANGELOG and git history, not in code.

Author: sgiehl

CHANGELOG.md

@@ -38,6 +38,7 @@ Attention: this is a major release with breaking changes.
 - Event and content tracking requests now send `&ca=1` (custom action), so Matomo no longer falls back to recording them as page views if the handling plugin is disabled (#80).
 - The `cip` (override IP) tracking parameter is now URL-encoded like every other value (#151).
 - No longer calls the deprecated `curl_close()` (it was already a no-op on the supported PHP versions) (#149).
+- Auto-detection of the tracked page URL now uses `REQUEST_URI` as the source instead of `PATH_INFO`. With front-controller / path-info routing (e.g. `/dir1/page` handled by `dir1/index.php`), `PATH_INFO` only holds the trailing `/page`, so the tracker previously recorded a truncated URL; it now records the full requested path. `PATH_INFO` is no longer used at all (`SCRIPT_NAME` remains the fallback when `REQUEST_URI` is unavailable) (#141).
 
 ### Added
 - PHPStan static analysis at max level (`phpstan.neon.dist`) and the Matomo coding standard via PHP_CodeSniffer (`phpcs.xml.dist`), both enforced for every pull request through GitHub Actions.

MatomoTracker.php

@@ -2621,24 +2621,26 @@ protected function getCookieMatchingName(string $name): string|false
     }
 
     /**
-     * If current URL is "http://example.org/dir1/dir2/index.php?param1=value1&param2=value2"
-     * will return "/dir1/dir2/index.php"
+     * Returns the path portion of the URL the visitor requested (everything between the host and
+     * the query string). For "http://example.org/dir1/dir2/index.php?param1=value1" this returns
+     * "/dir1/dir2/index.php"; for a front-controller URL such as "http://example.org/dir1/page"
+     * (where "/page" is handled by dir1/index.php) it returns "/dir1/page".
+     *
+     * The full request path is taken from REQUEST_URI. PATH_INFO is deliberately not used: it only
+     * holds the trailing path-info segment (e.g. "/page"), so it would drop the directory/script
+     * prefix and yield a truncated URL. SCRIPT_NAME is the fallback when REQUEST_URI is unavailable.
      *
      * @ignore
      */
     protected static function getCurrentScriptName(): string
     {
         $url = '';
-        if (!empty($_SERVER['PATH_INFO'])) {
-            $url = self::toStringValue($_SERVER['PATH_INFO']);
-        } else {
-            if (!empty($_SERVER['REQUEST_URI'])) {
-                $requestUri = self::toStringValue($_SERVER['REQUEST_URI']);
-                if (($pos = strpos($requestUri, '?')) !== false) {
-                    $url = substr($requestUri, 0, $pos);
-                } else {
-                    $url = $requestUri;
-                }
+        if (!empty($_SERVER['REQUEST_URI'])) {
+            $requestUri = self::toStringValue($_SERVER['REQUEST_URI']);
+            if (($pos = strpos($requestUri, '?')) !== false) {
+                $url = substr($requestUri, 0, $pos);
+            } else {
+                $url = $requestUri;
             }
         }
         if (empty($url) && isset($_SERVER['SCRIPT_NAME'])) {

tests/Unit/MatomoTrackerTest.php

@@ -1848,17 +1848,24 @@ public function testGetCurrentScriptName(): void
         unset($_SERVER['PATH_INFO'], $_SERVER['REQUEST_URI'], $_SERVER['SCRIPT_NAME']);
         $this->assertSame('/', TestableMatomoTracker::callGetCurrentScriptName());
 
+        // SCRIPT_NAME is only the fallback when REQUEST_URI is unavailable.
         $_SERVER['SCRIPT_NAME'] = 'script.php';
         $this->assertSame('/script.php', TestableMatomoTracker::callGetCurrentScriptName());
 
+        // REQUEST_URI is the primary source; the query string is stripped.
         $_SERVER['REQUEST_URI'] = '/dir/page.php?query=1';
         $this->assertSame('/dir/page.php', TestableMatomoTracker::callGetCurrentScriptName());
 
         $_SERVER['REQUEST_URI'] = '/dir/other.php';
         $this->assertSame('/dir/other.php', TestableMatomoTracker::callGetCurrentScriptName());
 
-        $_SERVER['PATH_INFO'] = '/path/info';
-        $this->assertSame('/path/info', TestableMatomoTracker::callGetCurrentScriptName());
+        // Front-controller / path-info routing: with a request for /dir1/page handled by
+        // dir1/index.php, PATH_INFO is only "/page". The full requested path must still be tracked,
+        // so REQUEST_URI wins and PATH_INFO is ignored (previously it truncated the URL to "/page").
+        $_SERVER['REQUEST_URI'] = '/dir1/page';
+        $_SERVER['PATH_INFO'] = '/page';
+        $_SERVER['SCRIPT_NAME'] = '/dir1/index.php';
+        $this->assertSame('/dir1/page', TestableMatomoTracker::callGetCurrentScriptName());
     }
 
     public function testGetCurrentQueryStringAndUrl(): void
@@ -1871,7 +1878,8 @@ public function testGetCurrentQueryStringAndUrl(): void
 
         $_SERVER['HTTPS'] = 'on';
         $_SERVER['HTTP_HOST'] = 'matomo.example';
-        $_SERVER['PATH_INFO'] = '/page';
+        unset($_SERVER['PATH_INFO']);
+        $_SERVER['REQUEST_URI'] = '/page';
         $this->assertSame('https://matomo.example/page?a=b&c=d', TestableMatomoTracker::callGetCurrentUrl());
     }