symfony · smnandre · Mar 2, 2024 · Mar 2, 2024
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\JavascriptParser;
 use Symfony\Component\AssetMapper\Exception\CircularAssetsException;
 use Symfony\Component\AssetMapper\Exception\RuntimeException;
 use Symfony\Component\AssetMapper\ImportMap\ImportMapConfigReader;
@@ -30,16 +31,7 @@ final class JavaScriptImportPathCompiler implements AssetCompilerInterface
     /**
      * @see https://regex101.com/r/1iBAIb/1
      */
-    private const IMPORT_PATTERN = '/
-        ^
-            (?:\/\/.*)                     # Lines that start with comments
-        |
-            (?:
-                \'(?:[^\'\\\\\n]|\\\\.)*\'   # Strings enclosed in single quotes
-            |
-                "(?:[^"\\\\\n]|\\\\.)*"      # Strings enclosed in double quotes
-            )
-        |
+    private const PREVIOUS_IMPORT_PATTERN = '/
             (?:                            # Import statements (script captured)
                 import\s*
                     (?:
@@ -49,10 +41,11 @@ final class JavaScriptImportPathCompiler implements AssetCompilerInterface
             |
                 \bimport\(
             )
-            \s*[\'"`](\.\/[^\'"`\n]+|(\.\.\/)*[^\'"`\n]+)[\'"`]\s*[;\)]
-        ?
+            \s*[\'"`](\.\/[^\'"`\n]+|(\.\.\/)*[^\'"`\n]+)[\'"`]\s*[)]
     /mx';
 
+    private const IMPORT_PATTERN = '/import(?:{[^{}]+}|.*?)\s*(?:from)?\s*[\'"`](.*?)[\'"`]|import\(.*?\)/m';
+
     public function __construct(
         private readonly ImportMapConfigReader $importMapConfigReader,
         private readonly string $missingImportMode = self::MISSING_IMPORT_WARN,
@@ -62,15 +55,13 @@ 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];
+        $jsParser = JavascriptParser::create($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;
             }
 
@@ -142,33 +133,6 @@ 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/CodeSequence.php b/src/Symfony/Component/AssetMapper/Compiler/Parser/CodeSequence.php
@@ -0,0 +1,49 @@
+<?php
+
+/*
+ * This file is part of the Symfony package.
+ *
+ * (c) Fabien Potencier <fabien@symfony.com>
+ *
+ * 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;
+
+/**
+ * Represents a sequence of code (e.g. a string, a comment, a block of code).
+ *
+ * @author Simon André <smn.andre@gmail.com>
+ *
+ * @internal
+ */
+class CodeSequence implements \Stringable
+{
+    public function __construct(
+        private readonly string $type,
+        private readonly int $start,
+        private readonly int $end,
+    ) {
+    }
+
+    public function getType(): string
+    {
+        return $this->type;
+    }
+
+    public function getStart(): int
+    {
+        return $this->start;
+    }
+
+    public function getEnd(): int
+    {
+        return $this->end;
+    }
+
+    public function __toString(): string
+    {
+        return sprintf('%s [%d:%d]', $this->type, $this->start, $this->end);
+    }
+}
diff --git a/src/Symfony/Component/AssetMapper/Compiler/Parser/JavascriptParser.php b/src/Symfony/Component/AssetMapper/Compiler/Parser/JavascriptParser.php
@@ -0,0 +1,186 @@
+<?php
+
+/*
+ * This file is part of the Symfony package.
+ *
+ * (c) Fabien Potencier <fabien@symfony.com>
+ *
+ * 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 <smn.andre@gmail.com>
+ *
+ * @internal
+ */
+class JavascriptParser
+{
+    private const STATE_DEFAULT = 'DEFAULT';
+    private const STATE_COMMENT = 'COMMENT';
+    private const STATE_STRING = 'STRING';
+
+    private int $cursor = 0;
+
+    private int $contentEnd;
+
+    private string $pattern;
+
+    private ?CodeSequence $currentSequence = null;
+
+    /**
+     * @var CodeSequence[]
+     */
+    private array $sequences = [];
+
+    private function __construct(
+        private string $content,
+    ) {
+        $this->contentEnd = \strlen($content);
+
+        $chars = [
+            '/*',   // Multi-line comment
+            '//',   // Single-line comment
+            '"',    // Double quote
+            '\'',   // Single quote
+            '`',    // Backtick
+        ];
+        $this->pattern = '/'.implode('|', array_map(fn ($ch) => preg_quote($ch, '/'), $chars)).'/';
+    }
+
+    public static function create(string $content): self
+    {
+        return new self($content);
+    }
+
+    /**
+     * @return CodeSequence[]
+     */
+    public function getSequences(): array
+    {
+        return $this->sequences;
+    }
+
+    public function getCurrentSequence(): ?CodeSequence
+    {
+        return $this->currentSequence;
+    }
+
+    public function isExecutable(): bool
+    {
+        return self::STATE_DEFAULT === $this->currentSequence?->getType();
+    }
+
+    public function parseUntil(int $position): void
+    {
+        if ($position > $this->contentEnd) {
+            throw new \InvalidArgumentException('Cannot parse beyond the end of the content.');
+        }
+        if ($position < $this->cursor) {
+            throw new \InvalidArgumentException('Cannot parse backwards.');
+        }
+
+        while ($this->cursor <= $position) {
+            // Current CodeSequence ?
+            if (null !== $this->currentSequence) {
+                if ($this->currentSequence->getEnd() > $position) {
+                    $this->cursor = $position;
+
+                    return;
+                }
+
+                $this->cursor = $this->currentSequence->getEnd();
+                $this->currentSequence = null;
+            }
+
+            preg_match($this->pattern, $this->content, $matches, \PREG_OFFSET_CAPTURE, $this->cursor);
+            if (!$matches) {
+                $this->cursor = $position;
+                $this->pushSequence(self::STATE_DEFAULT, $this->cursor, $this->contentEnd);
+
+                return;
+            }
+
+            $matchPos = (int) $matches[0][1];
+            $matchChar = $matches[0][0];
+
+            if ($matchPos > $position) {
+                $this->pushSequence(self::STATE_DEFAULT, $this->cursor, $matchPos - 1);
+                $this->cursor = $position;
+
+                return;
+            }
+
+            // Multi-line comment
+            if ('/*' === $matchChar) {
+                if (false === $endPos = strpos($this->content, '*/', $matchPos + 2)) {
+                    $this->cursor = $position;
+                    $this->pushSequence(self::STATE_COMMENT, $matchPos, $this->contentEnd);
+
+                    return;
+                }
+
+                $this->cursor = min($endPos + 2, $position);
+                $this->pushSequence(self::STATE_COMMENT, $matchPos, $endPos + 2);
+                continue;
+            }
+
+            // Single-line comment
+            if ('//' === $matchChar) {
+                if (false === $endPos = strpos($this->content, "\n", $matchPos + 2)) {
+                    $this->cursor = $position;
+                    $this->pushSequence(self::STATE_COMMENT, $matchPos, $this->contentEnd);
+
+                    return;
+                }
+
+                $this->cursor = min($endPos + 1, $position);
+                $this->pushSequence(self::STATE_COMMENT, $matchPos, $endPos + 1);
+                continue;
+            }
+
+            // Single-line string
+            if ('"' === $matchChar || "'" === $matchChar) {
+                if (false === $endPos = strpos($this->content, $matchChar, $matchPos + 1)) {
+                    $this->cursor = $position;
+                    $this->pushSequence(self::STATE_STRING, $matchPos, $this->contentEnd);
+
+                    return;
+                }
+                while (false !== $endPos && '\\' == $this->content[$endPos - 1]) {
+                    $endPos = strpos($this->content, $matchChar, $endPos + 1);
+                }
+
+                $this->cursor = min($endPos + 1, $position);
+                $this->pushSequence(self::STATE_STRING, $matchPos, $endPos + 1);
+                continue;
+            }
+
+            // Multi-line string
+            if ('`' === $matchChar) {
+                if (false === $endPos = strpos($this->content, $matchChar, $matchPos + 1)) {
+                    $this->cursor = $position;
+                    $this->pushSequence(self::STATE_STRING, $matchPos, $this->contentEnd);
+
+                    return;
+                }
+                while (false !== $endPos && '\\' == $this->content[$endPos - 1]) {
+                    $endPos = strpos($this->content, $matchChar, $endPos + 1);
+                }
+
+                $this->cursor = min($endPos + 1, $position);
+                $this->pushSequence(self::STATE_STRING, $matchPos, $endPos + 1);
+                continue;
+            }
+        }
+    }
+
+    private function pushSequence(string $type, int $start, int $end): void
+    {
+        $this->sequences[] = $this->currentSequence = new CodeSequence($type, $start, $end);
+    }
+}
diff --git a/src/Symfony/Component/AssetMapper/Tests/Compiler/JavaScriptImportPathCompilerTest.php b/src/Symfony/Component/AssetMapper/Tests/Compiler/JavaScriptImportPathCompilerTest.php
@@ -274,15 +274,6 @@ 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;