Merge pull request #73 from phug-php/feature/issue-69-interpolations-…

…in-comments Ignore interpolations in comments
phug-php · May 16, 2020 · 39195e9 · 39195e9
2 parents 5ace1d7 + b53a717
commit 39195e9
Show file tree

Hide file tree

Showing 8 changed files with 87 additions and 47 deletions.
diff --git a/src/Phug/Lexer/Lexer/AbstractToken.php b/src/Phug/Lexer/Lexer/AbstractToken.php
@@ -42,4 +42,9 @@ public function markAsHandled()
     {
         $this->handled = true;
     }
+
+    public function __toString()
+    {
+        return '['.get_class($this).']';
+    }
 }
diff --git a/src/Phug/Lexer/Lexer/Analyzer/LineAnalyzer.php b/src/Phug/Lexer/Lexer/Analyzer/LineAnalyzer.php
@@ -122,12 +122,6 @@ public function getLines()
     public function getFlatLines()
     {
         return array_map(function ($line) {
-            foreach ($line as $chunk) {
-                if ($chunk instanceof TokenInterface) {
-                    $this->state->throwException('Unexpected '.get_class($chunk).' inside raw text.');
-                }
-            }
-
             return implode('', $line);
         }, $this->lines);
     }

diff --git a/src/Phug/Lexer/Lexer/Scanner/CommentScanner.php b/src/Phug/Lexer/Lexer/Scanner/CommentScanner.php
@@ -34,25 +34,26 @@ public function scan(State $state)
         }
 
         yield $state->endToken($token);
+
         $line = $reader->readUntilNewLine();
         $lines = $line === '' ? [] : [[$line]];
 
         /** @var TextToken $token */
         $token = $state->createToken(TextToken::class);
 
         $analyzer = new LineAnalyzer($state, $reader, $lines);
+        $analyzer->disallowInterpolation();
         $analyzer->analyze(false);
         $lines = $analyzer->getFlatLines();
 
         if (end($lines) === '') {
             array_pop($lines);
         }
-        $token->setValue(implode("\n", $lines));
 
-        //TODO: As it seems, this is the only TextToken that will actually contain newlines, thus Stat->endToken will
-        // end up with a wrong line offset. This is why endToken is not applied at all here and only the start
-        // position will be kept
-        $token->getSourceLocation()->setOffsetLength(1); //Let it have at least 1 length for debugging
+        $lines = implode("\n", $lines);
+        $token->setValue($lines);
+        $token->getSourceLocation()->setOffsetLength(mb_strlen($lines));
+
         yield $token;
 
         if ($analyzer->hasNewLine()) {

diff --git a/src/Phug/Lexer/Lexer/Scanner/InterpolationScanner.php b/src/Phug/Lexer/Lexer/Scanner/InterpolationScanner.php
@@ -18,6 +18,35 @@
 
 class InterpolationScanner implements ScannerInterface
 {
+    protected $interpolationChars = [
+        'tagInterpolation' => ['[', ']'],
+        'interpolation'    => ['{', '}'],
+    ];
+
+    protected $regExp;
+
+    public function __construct()
+    {
+        $interpolations = [];
+        $backIndex = 2;
+
+        foreach ($this->interpolationChars as $name => list($start, $end)) {
+            $start = preg_quote($start, '/');
+            $end = preg_quote($end, '/');
+            $interpolations[] = $start.'(?<'.$name.'>'.
+                '(?>"(?:\\\\[\\S\\s]|[^"\\\\])*"|\'(?:\\\\[\\S\\s]|[^\'\\\\])*\'|[^'.
+                $start.$end.
+                '\'"]++|(?-'.$backIndex.'))*+'.
+                ')'.$end;
+            $backIndex++;
+        }
+
+        $this->regExp = '(?<text>.*?)'.
+            '(?<!\\\\)'.
+            '(?<escape>#|!(?='.preg_quote($this->interpolationChars['interpolation'][0], '/').'))'.
+            '(?<wrap>'.implode('|', $interpolations).')';
+    }
+
     protected function throwEndOfLineExceptionIf(State $state, $condition)
     {
         if ($condition) {
@@ -103,21 +132,9 @@ public function scan(State $state)
     {
         $reader = $state->getReader();
 
-        //TODO: $state->endToken
-        while ($reader->match(
-            '(?<text>.*?)'.
-            '(?<!\\\\)'.
-            '(?<escape>#|!(?=\{))(?<wrap>'.
-                '\\[(?<tagInterpolation>'.
-                    '(?>"(?:\\\\[\\S\\s]|[^"\\\\])*"|\'(?:\\\\[\\S\\s]|[^\'\\\\])*\'|[^\\[\\]\'"]++|(?-2))*+'.
-                ')\\]|'.
-                '\\{(?<interpolation>'.
-                    '(?>"(?:\\\\[\\S\\s]|[^"\\\\])*"|\'(?:\\\\[\\S\\s]|[^\'\\\\])*\'|[^{}\'"]++|(?-3))*+'.
-                ')\\}'.
-            ')'
-        )) {
+        while ($reader->match($this->regExp)) {
             $text = $reader->getMatch('text');
-            $text = preg_replace('/\\\\([#!]\\[|#\\{)/', '$1', $text);
+            $text = preg_replace('/\\\\([#!]\\[|#{)/', '$1', $text);
 
             if (mb_strlen($text) > 0) {
                 /** @var TextToken $token */
@@ -142,6 +159,7 @@ public function scan(State $state)
                 /** @var TextToken $token */
                 $token = $state->createToken(TextToken::class);
                 $token->setValue("\n");
+
                 yield $token;
             }
         }

diff --git a/tests/Phug/AbstractLexerTest.php b/tests/Phug/AbstractLexerTest.php
@@ -49,7 +49,7 @@ protected function filterTokenClass($className)
         }
     }
 
-    protected function assertTokens($expression, array $classNames, Lexer $lexer = null)
+    protected function assertTokens($expression, array $classNames, Lexer $lexer = null, &$tokens = [])
     {
         $lexer = $lexer ?: $this->lexer;
         $tokens = iterator_to_array($lexer->lex($expression));

diff --git a/tests/Phug/Lexer/Scanner/CommentScannerTest.php b/tests/Phug/Lexer/Scanner/CommentScannerTest.php
@@ -146,6 +146,38 @@ public function testCommentInIndent()
         ]);
     }
 
+    /**
+     * @covers \Phug\Lexer\Analyzer\LineAnalyzer::getFlatLines
+     */
+    public function testInterpolationInComment()
+    {
+        $comment = implode("\n", [
+            '',
+            '  p.',
+            '    go to #[a(href=\'\') page]',
+        ]);
+
+        $this->assertTokens("//$comment", [
+            CommentToken::class,
+            TextToken::class,
+        ], null, $tokens);
+
+        /** @var TextToken $text */
+        $text = $tokens[1];
+
+        $this->assertSame($comment, $text->getValue());
+
+        $this->assertTokens("//-$comment", [
+            CommentToken::class,
+            TextToken::class,
+        ], null, $tokens);
+
+        /** @var TextToken $text */
+        $text = $tokens[1];
+
+        $this->assertSame($comment, $text->getValue());
+    }
+
     /**
      * @covers \Phug\Lexer\Scanner\CommentScanner
      * @covers \Phug\Lexer\Scanner\CommentScanner::scan

diff --git a/tests/Phug/Lexer/Scanner/InterpolationScannerTest.php b/tests/Phug/Lexer/Scanner/InterpolationScannerTest.php
@@ -2,8 +2,6 @@
 
 namespace Phug\Test\Lexer\Scanner;
 
-use Phug\Lexer\Analyzer\LineAnalyzer;
-use Phug\Lexer\State;
 use Phug\Lexer\Token\ExpressionToken;
 use Phug\Lexer\Token\IndentToken;
 use Phug\Lexer\Token\InterpolationEndToken;
@@ -13,9 +11,7 @@
 use Phug\Lexer\Token\TagInterpolationStartToken;
 use Phug\Lexer\Token\TagToken;
 use Phug\Lexer\Token\TextToken;
-use Phug\Reader;
 use Phug\Test\AbstractLexerTest;
-use Phug\Util\SourceLocation;
 
 class InterpolationScannerTest extends AbstractLexerTest
 {
@@ -166,22 +162,6 @@ public function testScan()
         ]);
     }
 
-    /**
-     * @covers                   \Phug\Lexer\Analyzer\LineAnalyzer::<public>
-     * @expectedException        \Phug\LexerException
-     * @expectedExceptionMessage Failed to lex: Unexpected Phug\Lexer\Token\InterpolationStartToken inside raw text.
-     */
-    public function testTokenInLineAnalyzer()
-    {
-        $input = 'p #{42}';
-        $analyzer = new LineAnalyzer(new State($this->lexer, $input, []), new Reader($input), [
-            [
-                new InterpolationStartToken(new SourceLocation('foo.pug', 12, 43)),
-            ],
-        ]);
-        $analyzer->getFlatLines();
-    }
-
     /**
      * @covers \Phug\Lexer\Scanner\InterpolationScanner::scanTagInterpolation
      * @covers \Phug\Lexer\Scanner\InterpolationScanner::throwEndOfLineExceptionIf
@@ -192,7 +172,7 @@ public function testNewLineInTagInterpolation()
         $this->expectMessageToBeThrown('End of line was reached with no closing bracket for interpolation.');
 
         $input = "p #[em\n]";
-        $tokens = iterator_to_array($this->lexer->lex($input));
+        iterator_to_array($this->lexer->lex($input));
     }
 
     /**

diff --git a/tests/Phug/Lexer/Token/InterpolationStartTokenTest.php b/tests/Phug/Lexer/Token/InterpolationStartTokenTest.php
@@ -23,4 +23,14 @@ public function testEnd()
         $start->setEnd($end);
         self::assertSame($end, $start->getEnd());
     }
+
+    /**
+     * @covers ::__toString
+     */
+    public function testStringification()
+    {
+        $start = new InterpolationStartToken();
+
+        self::assertSame('['.InterpolationStartToken::class.']', "$start");
+    }
 }