[AssetMapper] Detect import with a sequence parser

symfony · fabpot · Jan 25, 2025 · Nov 27, 2024 · Jan 25, 2025 · 720c38746083b0578ca8f7deee96f30378edb3d4
commit 720c38746083b0578ca8f7deee96f30378edb3d4
diff --git a/src/Symfony/Component/AssetMapper/Compiler/JavaScriptImportPathCompiler.php b/src/Symfony/Component/AssetMapper/Compiler/JavaScriptImportPathCompiler.php
@@ -13,6 +13,7 @@
 
 use Psr\Log\LoggerInterface;
 use Symfony\Component\AssetMapper\AssetMapperInterface;
+use Symfony\Component\AssetMapper\Compiler\Parser\JavascriptSequenceParser;
 use Symfony\Component\AssetMapper\Exception\CircularAssetsException;
 use Symfony\Component\AssetMapper\Exception\RuntimeException;
 use Symfony\Component\AssetMapper\ImportMap\ImportMapConfigReader;
@@ -61,15 +62,13 @@ public function __construct(
 
     public function compile(string $content, MappedAsset $asset, AssetMapperInterface $assetMapper): string
     {
-        return preg_replace_callback(self::IMPORT_PATTERN, function ($matches) use ($asset, $assetMapper, $content) {
-            $fullImportString = $matches[0][0];
+        $jsParser = new JavascriptSequenceParser($content);
 
-            // Ignore matches that did not capture import statements
-            if (!isset($matches[1][0])) {
-                return $fullImportString;
-            }
+        return preg_replace_callback(self::IMPORT_PATTERN, function ($matches) use ($asset, $assetMapper, $jsParser) {
+            $fullImportString = $matches[0][0];
 
-            if ($this->isCommentedOut($matches[0][1], $content)) {
+            $jsParser->parseUntil($matches[0][1]);
+            if (!$jsParser->isExecutable()) {
                 return $fullImportString;
             }
 
@@ -146,33 +145,6 @@ private function handleMissingImport(string $message, ?\Throwable $e = null): vo
         };
     }
 
-    /**
-     * Simple check for the most common types of comments.
-     *
-     * This is not a full parser, but should be good enough for most cases.
-     */
-    private function isCommentedOut(mixed $offsetStart, string $fullContent): bool
-    {
-        $lineStart = strrpos($fullContent, "\n", $offsetStart - \strlen($fullContent));
-        $lineContentBeforeImport = substr($fullContent, $lineStart, $offsetStart - $lineStart);
-        $firstTwoChars = substr(ltrim($lineContentBeforeImport), 0, 2);
-        if ('//' === $firstTwoChars) {
-            return true;
-        }
-
-        if ('/*' === $firstTwoChars) {
-            $commentEnd = strpos($fullContent, '*/', $lineStart);
-            // if we can't find the end comment, be cautious: assume this is not a comment
-            if (false === $commentEnd) {
-                return false;
-            }
-
-            return $offsetStart < $commentEnd;
-        }
-
-        return false;
-    }
-
     private function findAssetForBareImport(string $importedModule, AssetMapperInterface $assetMapper): ?MappedAsset
     {
         if (!$importMapEntry = $this->importMapConfigReader->findRootImportMapEntry($importedModule)) {

diff --git a/src/Symfony/Component/AssetMapper/Compiler/Parser/JavascriptSequenceParser.php b/src/Symfony/Component/AssetMapper/Compiler/Parser/JavascriptSequenceParser.php
@@ -0,0 +1,187 @@
+<?php
+
+/*
+ * This file is part of the Symfony package.
+ *
+ * (c) Fabien Potencier <[email protected]>
+ *
+ * For the full copyright and license information, please view the LICENSE
+ * file that was distributed with this source code.
+ */
+
+namespace Symfony\Component\AssetMapper\Compiler\Parser;
+
+/**
+ * Parses JavaScript content to identify sequences of strings, comments, etc.
+ *
+ * @author Simon André <[email protected]>
+ *
+ * @internal
+ */
+final class JavascriptSequenceParser
+{
+    private const STATE_DEFAULT = 0;
+    private const STATE_COMMENT = 1;
+    private const STATE_STRING = 2;
+
+    private int $cursor = 0;
+
+    private int $contentEnd;
+
+    private string $pattern;
+
+    private int $currentSequenceType = self::STATE_DEFAULT;
+
+    private ?int $currentSequenceEnd = null;
+
+    private const COMMENT_SEPARATORS = [
+        '/*',   // Multi-line comment
+        '//',   // Single-line comment
+        '"',    // Double quote
+        '\'',   // Single quote
+        '`',    // Backtick
+    ];
+
+    public function __construct(
+        private readonly string $content,
+    ) {
+        $this->contentEnd = \strlen($content);
+
+        $this->pattern ??= '/'.implode('|', array_map(
+            fn (string $ch): string => preg_quote($ch, '/'),
+            self::COMMENT_SEPARATORS
+        )).'/';
+    }
+
+    public function isString(): bool
+    {
+        return self::STATE_STRING === $this->currentSequenceType;
+    }
+
+    public function isExecutable(): bool
+    {
+        return self::STATE_DEFAULT === $this->currentSequenceType;
+    }
+
+    public function isComment(): bool
+    {
+        return self::STATE_COMMENT === $this->currentSequenceType;
+    }
+
+    public function parseUntil(int $position): void
+    {
+        if ($position > $this->contentEnd) {
+            throw new \RuntimeException('Cannot parse beyond the end of the content.');
+        }
+        if ($position < $this->cursor) {
+            throw new \RuntimeException('Cannot parse backwards.');
+        }
+
+        while ($this->cursor <= $position) {
+            // Current CodeSequence ?
+            if (null !== $this->currentSequenceEnd) {
+                if ($this->currentSequenceEnd > $position) {
+                    $this->cursor = $position;
+
+                    return;
+                }
+
+                $this->cursor = $this->currentSequenceEnd;
+                $this->setSequence(self::STATE_DEFAULT, null);
+            }
+
+            preg_match($this->pattern, $this->content, $matches, \PREG_OFFSET_CAPTURE, $this->cursor);
+            if (!$matches) {
+                $this->endsWithSequence(self::STATE_DEFAULT, $position);
+
+                return;
+            }
+
+            $matchPos = (int) $matches[0][1];
+            $matchChar = $matches[0][0];
+
+            if ($matchPos > $position) {
+                $this->setSequence(self::STATE_DEFAULT, $matchPos - 1);
+                $this->cursor = $position;
+
+                return;
+            }
+
+            // Multi-line comment
+            if ('/*' === $matchChar) {
+                if (false === $endPos = strpos($this->content, '*/', $matchPos + 2)) {
+                    $this->endsWithSequence(self::STATE_COMMENT, $position);
+
+                    return;
+                }
+
+                $this->cursor = min($endPos + 2, $position);
+                $this->setSequence(self::STATE_COMMENT, $endPos + 2);
+                continue;
+            }
+
+            // Single-line comment
+            if ('//' === $matchChar) {
+                if (false === $endPos = strpos($this->content, "\n", $matchPos + 2)) {
+                    $this->endsWithSequence(self::STATE_COMMENT, $position);
+
+                    return;
+                }
+
+                $this->cursor = min($endPos + 1, $position);
+                $this->setSequence(self::STATE_COMMENT, $endPos + 1);
+                continue;
+            }
+
+            // Single-line string
+            if ('"' === $matchChar || "'" === $matchChar) {
+                if (false === $endPos = strpos($this->content, $matchChar, $matchPos + 1)) {
+                    $this->endsWithSequence(self::STATE_STRING, $position);
+
+                    return;
+                }
+                while (false !== $endPos && '\\' == $this->content[$endPos - 1]) {
+                    $endPos = strpos($this->content, $matchChar, $endPos + 1);
+                }
+
+                $this->cursor = min($endPos + 1, $position);
+                $this->setSequence(self::STATE_STRING, $endPos + 1);
+                continue;
+            }
+
+            // Multi-line string
+            if ('`' === $matchChar) {
+                if (false === $endPos = strpos($this->content, $matchChar, $matchPos + 1)) {
+                    $this->endsWithSequence(self::STATE_STRING, $position);
+
+                    return;
+                }
+                while (false !== $endPos && '\\' == $this->content[$endPos - 1]) {
+                    $endPos = strpos($this->content, $matchChar, $endPos + 1);
+                }
+
+                $this->cursor = min($endPos + 1, $position);
+                $this->setSequence(self::STATE_STRING, $endPos + 1);
+            }
+        }
+    }
+
+    /**
+     * @param int<self::STATE_*> $type
+     */
+    private function endsWithSequence(int $type, int $cursor): void
+    {
+        $this->cursor = $cursor;
+        $this->currentSequenceType = $type;
+        $this->currentSequenceEnd = $this->contentEnd;
+    }
+
+    /**
+     * @param int<self::STATE_*> $type
+     */
+    private function setSequence(int $type, ?int $end = null): void
+    {
+        $this->currentSequenceType = $type;
+        $this->currentSequenceEnd = $end;
+    }
+}
diff --git a/src/Symfony/Component/AssetMapper/Tests/Compiler/JavaScriptImportPathCompilerTest.php b/src/Symfony/Component/AssetMapper/Tests/Compiler/JavaScriptImportPathCompilerTest.php
@@ -290,15 +290,6 @@ public static function provideCompileTests(): iterable
             'expectedJavaScriptImports' => [],
         ];
 
-        yield 'multi_line_comment_with_no_end_parsed_for_safety' => [
-            'input' => <<<EOF
-                const fun;
-                    /* comment import("./other.js");
-                EOF
-            ,
-            'expectedJavaScriptImports' => ['/assets/other.js' => ['lazy' => true, 'asset' => 'other.js', 'add' => true]],
-        ];
-
         yield 'multi_line_comment_with_no_end_found_eventually_ignored' => [
             'input' => <<<EOF
                 const fun;