fix: Complex test name normalization (#139)

* Add edge test cases

* Update how dataset and test name are matched

* Use group names for better regex readability

* Repalce multiple non-word character by single hyphen

* Update tests expected output

* Remove useless import

* Move data providers together

Author: flohw

src/Browser/Test/LegacyExtension.php

@@ -113,15 +113,19 @@ private static function saveBrowserStates(string $test, string $type): void
 
     private static function normalizeTestName(string $name): string
     {
+        if (!strchr($name, 'with data set')) {
+            return \strtr($name, '\\:', '-_');
+        }
+
         // Try to match for a numeric data set index. If it didn't, match for a string one.
-        if (!\preg_match('#^([\w:\\\]+)(.+\#(\d+))#', $name, $matches)) {
-            \preg_match('#^([\w:\\\]+)(.+"([^"]+)")?#', $name, $matches);
+        if (!\preg_match('#^(?<test>[\w:\\\]+) with data set \#(?<dataset>\d+)#', $name, $matches)) {
+            \preg_match('#^(?<test>[\w:\\\]+) with data set "(?<dataset>.*)"#', $name, $matches);
         }
 
-        $normalized = \strtr($matches[1], '\\:', '-_');
+        $normalized = \strtr($matches['test'], '\\:', '-_');
 
-        if (isset($matches[3])) {
-            $normalized .= '__data-set-'.\strtr($matches[3], '\\: ', '-_-');
+        if (isset($matches['dataset'])) {
+            $normalized .= '__data-set-'.preg_replace('/\W+/', '-', $matches['dataset']);
         }
 
         return $normalized;

tests/NormalizationTest.php

@@ -17,6 +17,26 @@
 
 final class NormalizationTest extends TestCase
 {
+    /**
+     * @test
+     * @dataProvider namesProvider
+     * @dataProvider edgeCaseTestNames
+     */
+    public function can_normalize_test_names(string $testName, string $expectedOutput): void
+    {
+        $browser = $this->createMock(Browser::class);
+        $browser
+            ->expects(self::once())
+            ->method('saveCurrentState')
+            ->with($expectedOutput);
+
+        $extension = new LegacyExtension();
+        $extension->executeBeforeFirstTest();
+        $extension->executeBeforeTest($testName);
+        $extension::registerBrowser($browser);
+        $extension->executeAfterTestError($testName, '', 0);
+    }
+
     public static function namesProvider(): \Generator
     {
         $baseTemplate = 'error_'.__METHOD__;
@@ -52,22 +72,32 @@ public static function namesProvider(): \Generator
         ];
     }
 
-    /**
-     * @test
-     * @dataProvider namesProvider
-     */
-    public function can_normalize_test_names(string $testName, string $normalizedName): void
+    public static function edgeCaseTestNames(): \Generator
     {
-        $browser = $this->createMock(Browser::class);
-        $browser
-            ->expects(self::once())
-            ->method('saveCurrentState')
-            ->with($normalizedName);
-
-        $extension = new LegacyExtension();
-        $extension->executeBeforeFirstTest();
-        $extension->executeBeforeTest($testName);
-        $extension::registerBrowser($browser);
-        $extension->executeAfterTestError($testName, '', 0);
+        $baseTemplate = \strtr('error_'.__METHOD__."__data-set-", '\\:', '-_');
+        yield 'self within moustache' => [
+            'test name' => __METHOD__.' with data set "te{{self}}st" (test set)',
+            'expected output' => $baseTemplate.'te-self-st__0',
+        ];
+        yield 'double quoted with space' => [
+            'test name' => __METHOD__.' with data set "_self.env.setCache("uri://host.net:2121") _self.env.loadTemplate("other-host")" (test set)',
+            'expected output' => $baseTemplate.'_self-env-setCache-uri-host-net-2121-_self-env-loadTemplate-other-host-__0',
+        ];
+        yield 'double quotes in moustache' => [
+            'test name' => __METHOD__.' with data set "te{{_self.env.registerUndefinedFilterCallback("exec")}}{{_self.env.getFilter("id")}}st"',
+            'expected output' => $baseTemplate.'te-_self-env-registerUndefinedFilterCallback-exec-_self-env-getFilter-id-st__0',
+        ];
+        yield 'escaped simple quote' => [
+            'test name' => __METHOD__.' with data set "te{{\'/etc/passwd\'|file_excerpt(1,30)}}st"',
+             'expected output' => $baseTemplate.'te-etc-passwd-file_excerpt-1-30-st__0',
+         ];
+        yield 'single quote for array index access' => [
+            'test name' => __METHOD__.' with data set "te{{[\'id\']|filter(\'system\')}}st"',
+            'expected output' => $baseTemplate.'te-id-filter-system-st__0',
+        ];
+        yield 'numeric array access' => [
+            'test name' => __METHOD__.' with data set "te{{[0]|reduce(\'system\',\'id\')}}st"',
+            'expected output' => $baseTemplate.'te-0-reduce-system-id-st__0',
+        ];
     }
 }