Improve template minifier

composer.json

@@ -66,6 +66,7 @@
         "magento/magento-composer-installer": ">=0.1.11",
         "magento/zendframework1": "dev-master as 1.14.6",
         "monolog/monolog": "^2.3",
+        "nikic/php-parser": "~4.4.0",
         "pelago/emogrifier": "^5.0.0",
         "php-amqplib/php-amqplib": "~3.0.0",
         "phpseclib/mcrypt_compat": "2.0.0",

lib/internal/Magento/Framework/View/Template/Html/Minifier.php

@@ -9,6 +9,7 @@
 
 use Magento\Framework\App\Filesystem\DirectoryList;
 use Magento\Framework\Filesystem;
+use Magento\Framework\View\Template\Html\Minifier\Php;
 
 class Minifier implements MinifierInterface
 {
@@ -116,49 +117,94 @@ public function minify($file)
         $dir = dirname($file);
         $fileName = basename($file);
         $content = $this->readFactory->create($dir)->readFile($fileName);
-        //Storing Heredocs
-        $heredocs = [];
-        $content = preg_replace_callback(
-            '/<<<([A-z]+).*?\1;/ims',
-            function ($match) use (&$heredocs) {
-                $heredocs[] = $match[0];
+        $heredocs = null;
 
-                return '__MINIFIED_HEREDOC__' .(count($heredocs) - 1);
-            },
+        // Safely minify PHP code and remove single-line PHP comments by using a parser.
+        if (null !== $content) {
+            $parser = (new \PhpParser\ParserFactory())->create(\PhpParser\ParserFactory::PREFER_PHP7);
+
+            /**
+             * Prevent problems with deeply nested ASTs if Xdebug is enabled.
+             * @see https://github.com/nikic/PHP-Parser/blob/v4.4.0/doc/2_Usage_of_basic_components.markdown#bootstrapping
+             */
+            $nestingLevelConfigValue = ini_get('xdebug.max_nesting_level');
+
+            if (false !== $nestingLevelConfigValue) {
+                ini_set('xdebug.max_nesting_level', '3000');
+            }
+
+            try {
+                $ast = $parser->parse($content);
+
+                $traverser = new \PhpParser\NodeTraverser();
+                $traverser->addVisitor(new Php\NodeVisitor());
+                $ast = $traverser->traverse($ast);
+
+                $prettyPrinter = new Php\PrettyPrinter();
+                $content = $prettyPrinter->prettyPrintFile($ast);
+                $heredocs = $prettyPrinter->getDelayedHeredocs();
+            } catch (\Error $error) {
+                // Some PHP code is seemingly invalid, or too complex.
+            } finally {
+                if (false !== $nestingLevelConfigValue) {
+                    ini_set('xdebug.max_nesting_level', $nestingLevelConfigValue);
+                }
+            }
+        }
+
+        // Stash the heredocs now if the template could not be parsed.
+        if (null === $heredocs) {
+            $content = preg_replace_callback(
+                '/<<<([A-z]+).*?\1\s*;/ims',
+                function ($match) use (&$heredocs) {
+                    $heredocs[] = $match[0];
+
+                    return '__MINIFIED_HEREDOC__' .(count($heredocs) - 1);
+                },
             ($content ?? '')
-        );
+            );
+        }
+
+        // Remove insignificant spaces before closing HTML tags
+        // (preserve one space after ]]>, and all spaces inside <pre> and <textarea> tags).
         $content = preg_replace(
-            '#(?<!]]>)\s+</#',
+            '#(?<!]]>)\s+</(?!(?>textarea|pre)\b)#',
             '</',
+            // Remove redundant spaces after PHP tags that do not start with a print or condition statement,
+            // and that do not contain any "?".
             preg_replace(
                 '#((?:<\?php\s+(?!echo|print|if|elseif|else)[^\?]*)\?>)\s+#',
                 '$1 ',
+                // Remove single space in empty non-inline tags.
                 preg_replace(
                     '#(?<!' . implode('|', $this->inlineHtmlTags) . ')\> \<#',
                     '><',
+                    // Remove redundant spaces outside of tags in which they are relevant.
                     preg_replace(
                         '#(?ix)(?>[^\S ]\s*|\s{2,})(?=(?:(?:[^<]++|<(?!/?(?:textarea|pre|script)\b))*+)'
                         . '(?:<(?>textarea|pre|script)\b|\z))#',
                         ' ',
+                        // Remove single-line comments in <script> tags, except for <![CDATA[ and ]]>.
+                        // Do nothing if the "//" part is seemingly part of a string / URL / RegExp.
                         preg_replace(
-                            '#(?<!:|\\\\|\'|"|/)//(?!/)(?!\s*\<\!\[)(?!\s*]]\>)[^\n\r]*#',
+                            '#(?<!:|\\\\|\\\|\'|"|/)//(?!/)(?!\s*\<\!\[)(?!\s*]]\>)[^\n\r]*'
+                            . '(?!(?:(?:[^<]++|<(?!/?(?:script)\b))*+)(?:<(?>script)\b|\z))#',
                             '',
+                            // Remove commented single-line PHP tags in <script> tags.
+                            // Do nothing if the "//" part is seemingly part of a URL / RegExp.
                             preg_replace(
-                                '#(?<!:|\'|")//[^\n\r<]*(\?\>)#',
-                                ' $1',
-                                preg_replace(
-                                    '#(?<!:)//[^\n\r]*(\<\?php)[^\n\r]*(\s\?\>)[^\n\r]*#',
-                                    '',
-                                    ($content ?? '')
-                                )
+                                '#(?<!:|\\\)//[^\n\r]*(\<\?(php|=))[^\n\r]*(\s\?\>)[^\n\r]*'
+                                . '(?!(?:(?:[^<]++|<(?!/?(?:script)\b))*+)(?:<(?>script)\b|\z))#',
+                                '',
+                                ($content ?? '')
                             )
                         )
                     )
                 )
             )
         );
 
-        //Restoring Heredocs
+        // Restore the stashed heredocs.
         $content = preg_replace_callback(
             '/__MINIFIED_HEREDOC__(\d+)/ims',
             function ($match) use ($heredocs) {
@@ -184,3 +230,4 @@ private function getRelativeGeneratedPath($sourcePath)
         return $this->filesystem->getDirectoryRead(DirectoryList::ROOT)->getRelativePath($sourcePath);
     }
 }
+

lib/internal/Magento/Framework/View/Template/Html/Minifier/Php/NodeVisitor.php

@@ -0,0 +1,84 @@
+<?php
+/**
+ * Copyright © Magento, Inc. All rights reserved.
+ * See COPYING.txt for license details.
+ */
+declare(strict_types=1);
+
+namespace Magento\Framework\View\Template\Html\Minifier\Php;
+
+use PhpParser\Comment;
+use PhpParser\Node;
+use PhpParser\NodeTraverser;
+use PhpParser\NodeVisitorAbstract;
+
+class NodeVisitor extends NodeVisitorAbstract
+{
+    /**
+     * @var Node[]
+     */
+    private $stack = [];
+
+    /**
+     * @var ?Node
+     */
+    private $previous;
+
+    public function beforeTraverse(array $nodes)
+    {
+        $this->stack = [];
+        $this->previous = null;
+    }
+
+    public function enterNode(Node $node)
+    {
+        if ($node instanceof Node\Stmt) {
+            // Mark isolated echo statements, to replace them later with short echo tags.
+            $parent = empty($this->stack) ?: $this->stack[count($this->stack) - 1];
+
+            if ($node instanceof Node\Stmt\InlineHTML) {
+                $node->setAttribute('parent', $parent);
+
+                if (
+                    ($this->previous instanceof Node\Stmt\Echo_)
+                    && ($previousHtmlStatement = $this->previous->getAttribute('previousHtmlStatement'))
+                    && ($previousHtmlStatement->getAttribute('parent') === $parent)
+                ) {
+                    $this->previous->setAttribute('isSingleEchoStatement', true);
+                    $previousHtmlStatement->setAttribute('hasSingleEchoStatementNext', true);
+                }
+            } elseif (
+                ($this->previous instanceof Node\Stmt\InlineHTML)
+                && ($this->previous->getAttribute('parent') === $parent)
+            ) {
+                $node->setAttribute('previousHtmlStatement', $this->previous);
+            }
+        }
+
+        $this->stack[] = $node;
+    }
+
+    public function leaveNode(Node $node)
+    {
+        $this->previous = $node;
+
+        array_pop($this->stack);
+
+        // Remove nodes that only contain non-doc comments.
+        if ($node instanceof Node\Stmt\Nop) {
+            $comments = $node->getComments();
+            $isSuperfluousNode = true;
+
+            foreach ($comments as $key => $comment) {
+                if ($comment instanceof Comment\Doc) {
+                    $isSuperfluousNode = false;
+                    break;
+                }
+            }
+
+            if ($isSuperfluousNode) {
+                return NodeTraverser::REMOVE_NODE;
+            }
+        }
+    }
+}

lib/internal/Magento/Framework/View/Template/Html/Minifier/Php/PrettyPrinter.php

@@ -0,0 +1,132 @@
+<?php
+/**
+ * Copyright © Magento, Inc. All rights reserved.
+ * See COPYING.txt for license details.
+ */
+declare(strict_types=1);
+
+namespace Magento\Framework\View\Template\Html\Minifier\Php;
+
+use PhpParser\Comment;
+use PhpParser\Node;
+use PhpParser\PrettyPrinter\Standard;
+
+class PrettyPrinter extends Standard
+{
+    /**
+     * @var string[]
+     */
+    private $delayedHeredocs = [];
+
+    protected function resetState()
+    {
+        $this->delayedHeredocs = [];
+        $this->indentLevel = 0;
+        $this->nl = '';
+        $this->origTokens = null;
+    }
+
+    protected function setIndentLevel(int $level)
+    {
+        // Ignore indentation.
+    }
+
+    protected function indent()
+    {
+        // Ignore indentation.
+    }
+
+    protected function outdent()
+    {
+        // Ignore indentation.
+    }
+
+    /**
+     * @param string $heredoc
+     * @return string
+     */
+    private function getHeredocPlaceholder(string $heredoc): string
+    {
+        $index = count($this->delayedHeredocs) + 1;
+
+        $this->delayedHeredocs[$index] = $this->handleMagicTokens($heredoc);
+
+        return '__MINIFIED_HEREDOC__' . $index;
+    }
+
+    protected function pScalar_String(Node\Scalar\String_ $node): string
+    {
+        $result = parent::pScalar_String($node);
+
+        return $node->getAttribute('kind') !== Node\Scalar\String_::KIND_HEREDOC
+            ? $result
+            : $this->getHeredocPlaceholder($result);
+    }
+
+    protected function pScalar_Encapsed(Node\Scalar\Encapsed $node): string
+    {
+        $result = parent::pScalar_Encapsed($node);
+
+        return $node->getAttribute('kind') !== Node\Scalar\String_::KIND_HEREDOC
+            ? $result
+            : $this->getHeredocPlaceholder($result);
+    }
+
+    protected function pCommaSeparated(array $nodes): string
+    {
+        return $this->pImplode($nodes, ',');
+    }
+
+    protected function pComments(array $comments): string
+    {
+        // Only preserve doc comments.
+        foreach ($comments as $key => $comment) {
+            if (!$comment instanceof Comment\Doc) {
+                unset($comments[$key]);
+            }
+        }
+
+        $formattedComments = [];
+
+        foreach ($comments as $comment) {
+            $formattedComments[] = str_replace("\n", '', $comment->getReformattedText());
+        }
+
+        // Add a space between doc comments to avoid occurrences of "//" that could later be misinterpreted.
+        return implode(' ', $formattedComments) . ' ';
+    }
+
+    protected function pExpr_Array(Node\Expr\Array_ $node): string
+    {
+        $node->setAttribute('kind', Node\Expr\Array_::KIND_SHORT);
+
+        return parent::pExpr_Array($node);
+    }
+
+    protected function pStmt_Echo(Node\Stmt\Echo_ $node): string
+    {
+        $output = $this->pCommaSeparated($node->exprs);
+
+        return $node->getAttribute('isSingleEchoStatement')
+            ? $output . ' '
+            : 'echo ' . $output . ';';
+    }
+
+    protected function pStmt_InlineHTML(Node\Stmt\InlineHTML $node): string
+    {
+        $newline = $node->getAttribute('hasLeadingNewline', true) ? "\n" : '';
+
+        return '?>'
+            . $newline
+            . $node->value
+            . ($node->getAttribute('hasSingleEchoStatementNext') ? '<?= ' : '<?php ');
+    }
+
+    /**
+     * @return string[]
+     */
+    public function getDelayedHeredocs(): array
+    {
+        return $this->delayedHeredocs;
+    }
+}