Tag parsing (#215)

* Remove any trailing comments before parsing the tags * Disallow whitespace in tags * Don't consume all of line when parsing tags * Change error to deprecation
Behat · Jul 12, 2021 · 3f94bb5 · 3f94bb5
1 parent fbd6baa
commit 3f94bb5
Show file tree

Hide file tree

Showing 6 changed files with 64 additions and 4 deletions.
diff --git a/phpunit.xml.dist b/phpunit.xml.dist
@@ -3,6 +3,7 @@
 <phpunit backupGlobals="false"
          backupStaticAttributes="false"
          colors="true"
+         convertDeprecationsToExceptions="true"
          convertErrorsToExceptions="true"
          convertNoticesToExceptions="true"
          convertWarningsToExceptions="true"

diff --git a/src/Behat/Gherkin/Filter/TagFilter.php b/src/Behat/Gherkin/Filter/TagFilter.php
@@ -31,6 +31,13 @@ class TagFilter extends ComplexFilter
     public function __construct($filterString)
     {
         $this->filterString = trim($filterString);
+
+       if(preg_match('/\s/u', $this->filterString)) {
+            trigger_error(
+                "Tags with whitespace are deprecated and may be removed in a future version",
+                E_USER_DEPRECATED
+            );
+       }
     }
 
     /**

diff --git a/src/Behat/Gherkin/Lexer.php b/src/Behat/Gherkin/Lexer.php
@@ -180,6 +180,15 @@ protected function consumeLine()
         $this->trimmedLine = null;
     }
 
+    /**
+     * Consumes first part of line from input without incrementing the line number
+     */
+    protected function consumeLineUntil(int $trimmedOffset)
+    {
+        $this->line = mb_substr(ltrim($this->line), $trimmedOffset, null, 'utf-8');
+        $this->trimmedLine = null;
+    }
+
     /**
      * Returns trimmed version of line.
      *
@@ -515,16 +524,23 @@ protected function scanTableRow()
     protected function scanTags()
     {
         $line = $this->getTrimmedLine();
+
         if (!isset($line[0]) || '@' !== $line[0]) {
             return null;
         }
 
+        if(preg_match('/^(?<line>.*)\s+#.*$/', $line, $matches)) {
+            ['line' => $line] = $matches;
+            $this->consumeLineUntil(mb_strlen($line, 'utf-8'));
+        } else {
+            $this->consumeLine();
+        }
+
         $token = $this->takeToken('Tag');
         $tags = explode('@', mb_substr($line, 1, mb_strlen($line, 'utf8') - 1, 'utf8'));
         $tags = array_map('trim', $tags);
         $token['tags'] = $tags;
 
-        $this->consumeLine();
 
         return $token;
     }

diff --git a/src/Behat/Gherkin/Parser.php b/src/Behat/Gherkin/Parser.php
@@ -608,6 +608,9 @@ protected function parsePyString()
     protected function parseTags()
     {
         $token = $this->expectTokenType('Tag');
+
+        $this->guardTags($token['tags']);
+
         $this->tags = array_merge($this->tags, $token['tags']);
 
         $possibleTransitions = array(
@@ -644,6 +647,20 @@ protected function popTags()
         return $tags;
     }
 
+    /**
+     * Checks the tags fit the required format
+     *
+     * @param string[] $tags
+     */
+    protected function guardTags(array $tags)
+    {
+        foreach ($tags as $tag) {
+            if (preg_match('/\s/', $tag)) {
+                trigger_error('Whitespace in tags is deprecated, found "$tag"', E_USER_DEPRECATED);
+            }
+        }
+    }
+
     /**
      * Parses next text line & returns it.
      *

diff --git a/tests/Behat/Gherkin/Cucumber/CompatibilityTest.php b/tests/Behat/Gherkin/Cucumber/CompatibilityTest.php
@@ -39,12 +39,14 @@ class CompatibilityTest extends TestCase
         'escaped_pipes.feature' => 'Feature description has wrong whitespace captured',
         'incomplete_scenario.feature' => 'Wrong background parsing when there are no steps',
         'incomplete_background_2.feature' => 'Wrong background parsing when there are no steps',
-        'tags.feature' => 'Tags followed by comments not parsed correctly'
     ];
 
     private $parsedButShouldNotBe = [
         'invalid_language.feature' => 'Invalid language is silently ignored',
-        'whitespace_in_tags.feature' => 'Whitespace in tags is tolerated',
+    ];
+
+    private $deprecatedInsteadOfParseError = [
+        'whitespace_in_tags.feature' => '/Whitespace in tags is deprecated/',
     ];
 
     /**
@@ -96,7 +98,12 @@ public function testBadFeaturesDoNotParse(\SplFileInfo $file)
             $this->markTestIncomplete($this->parsedButShouldNotBe[$file->getFilename()]);
         }
 
-        $this->expectException(ParserException::class);
+        if (isset($this->deprecatedInsteadOfParseError[$file->getFilename()])) {
+            $this->expectDeprecation();
+            $this->expectDeprecationMessageMatches($this->deprecatedInsteadOfParseError[$file->getFilename()]);
+        } else {
+            $this->expectException(ParserException::class);
+        }
         $gherkinFile = $file->getPathname();
         $feature = $this->parser->parse(file_get_contents($gherkinFile), $gherkinFile);
     }

diff --git a/tests/Behat/Gherkin/Filter/TagFilterTest.php b/tests/Behat/Gherkin/Filter/TagFilterTest.php
@@ -268,4 +268,16 @@ public function testFilterFeatureWithTaggedExamples()
         /** @noinspection PhpUndefinedMethodInspection */
         $this->assertEquals(array($exampleTableNode3), $scenarioInterfaces[1]->getExampleTables());
     }
+
+    public function testFilterWithWhitespaceIsDeprecated()
+    {
+        $this->expectDeprecation();
+        $tagFilter = new TagFilter('@tag with space');
+        $scenario = new ScenarioNode(null, ['tag with space'], array(), null, 2);
+        $feature = new FeatureNode(null, null, [], null, [$scenario], null, null, null, 1);
+
+        $scenarios = $tagFilter->filterFeature($feature)->getScenarios();
+
+        $this->assertEquals([$scenario], $scenarios);
+    }
 }