Optimizations in parser

- array_walk is slower than a foreach, so buffer now uses foreach.
- By caching some of the inline node parsers we can strip out about 9% of
  the overhead in the inline parser, as we do not have to check them for every token.

Author: jaapio

packages/guides-restructured-text/src/RestructuredText/Parser/Buffer.php

@@ -14,7 +14,6 @@
 namespace phpDocumentor\Guides\RestructuredText\Parser;
 
 use function array_pop;
-use function array_walk;
 use function count;
 use function implode;
 use function ltrim;

packages/guides-restructured-text/src/RestructuredText/Parser/InlineParser.php

@@ -16,8 +16,11 @@
 use Exception;
 use phpDocumentor\Guides\Nodes\Inline\PlainTextInlineNode;
 use phpDocumentor\Guides\Nodes\InlineCompoundNode;
+use phpDocumentor\Guides\RestructuredText\Parser\Productions\InlineRules\CachableInlineRule;
 use phpDocumentor\Guides\RestructuredText\Parser\Productions\InlineRules\InlineRule;
 
+use function array_filter;
+use function array_key_exists;
 use function usort;
 
 /** @internal */
@@ -26,11 +29,21 @@ class InlineParser
     /** @var InlineRule[] */
     private array $rules;
 
+    /** @var array<InlineLexer::*, CachableInlineRule> */
+    private array $cache = [];
+
     /** @param iterable<InlineRule> $inlineRules */
     public function __construct(iterable $inlineRules)
     {
-        $this->rules = [...$inlineRules];
+        $this->rules = array_filter([...$inlineRules], static fn ($rule) => $rule instanceof CachableInlineRule === false);
         usort($this->rules, static fn (InlineRule $a, InlineRule $b): int => $a->getPriority() > $b->getPriority() ? -1 : 1);
+        foreach ($inlineRules as $rule) {
+            if (!($rule instanceof CachableInlineRule)) {
+                continue;
+            }
+
+            $this->cache[$rule->getToken()] = $rule;
+        }
     }
 
     public function parse(string $content, BlockContext $blockContext): InlineCompoundNode
@@ -44,7 +57,9 @@ public function parse(string $content, BlockContext $blockContext): InlineCompou
         while ($lexer->token !== null) {
             foreach ($this->rules as $inlineRule) {
                 $node = null;
-                if ($inlineRule->applies($lexer)) {
+                if (array_key_exists($lexer->token->type ?? -1, $this->cache)) {
+                    $node = $this->cache[$lexer->token->type]->apply($blockContext, $lexer);
+                } elseif ($inlineRule->applies($lexer)) {
                     $node = $inlineRule->apply($blockContext, $lexer);
                 }
 

packages/guides-restructured-text/src/RestructuredText/Parser/Productions/InlineRules/AnonymousReferenceRule.php

@@ -28,8 +28,13 @@
  *
  * @see https://docutils.sourceforge.io/docs/ref/rst/restructuredtext.html#anonymous-hyperlinks
  */
-final class AnonymousReferenceRule extends ReferenceRule
+final class AnonymousReferenceRule extends ReferenceRule implements CachableInlineRule
 {
+    public function getToken(): int
+    {
+        return InlineLexer::ANONYMOUSE_REFERENCE;
+    }
+
     public function applies(InlineLexer $lexer): bool
     {
         return $lexer->token?->type === InlineLexer::ANONYMOUSE_REFERENCE;

packages/guides-restructured-text/src/RestructuredText/Parser/Productions/InlineRules/CachableInlineRule.php

@@ -0,0 +1,22 @@
+<?php
+
+declare(strict_types=1);
+
+/**
+ * This file is part of phpDocumentor.
+ *
+ * For the full copyright and license information, please view the LICENSE
+ * file that was distributed with this source code.
+ *
+ * @link https://phpdoc.org
+ */
+
+namespace phpDocumentor\Guides\RestructuredText\Parser\Productions\InlineRules;
+
+use phpDocumentor\Guides\RestructuredText\Parser\InlineLexer;
+
+interface CachableInlineRule extends InlineRule
+{
+    /** @return InlineLexer::* */
+    public function getToken(): int;
+}

packages/guides-restructured-text/src/RestructuredText/Parser/Productions/InlineRules/LiteralRule.php

@@ -23,8 +23,13 @@
 /**
  * Rule for literals such as ``something``
  */
-final class LiteralRule extends AbstractInlineRule
+final class LiteralRule extends AbstractInlineRule implements CachableInlineRule
 {
+    public function getToken(): int
+    {
+        return InlineLexer::LITERAL;
+    }
+
     public function applies(InlineLexer $lexer): bool
     {
         return $lexer->token?->type === InlineLexer::LITERAL;

packages/guides-restructured-text/src/RestructuredText/Parser/Productions/InlineRules/NamedReferenceRule.php

@@ -28,8 +28,13 @@
  *
  * @see https://docutils.sourceforge.io/docs/ref/rst/restructuredtext.html#hyperlink-references
  */
-final class NamedReferenceRule extends ReferenceRule
+final class NamedReferenceRule extends ReferenceRule implements CachableInlineRule
 {
+    public function getToken(): int
+    {
+        return InlineLexer::NAMED_REFERENCE;
+    }
+
     public function applies(InlineLexer $lexer): bool
     {
         return $lexer->token?->type === InlineLexer::NAMED_REFERENCE;

packages/guides-restructured-text/src/RestructuredText/Parser/Productions/InlineRules/NbspRule.php

@@ -20,8 +20,13 @@
 /**
  * Rule to parse for non-breaking spaces: a~b
  */
-final class NbspRule extends ReferenceRule
+final class NbspRule extends AbstractInlineRule implements CachableInlineRule
 {
+    public function getToken(): int
+    {
+        return InlineLexer::NBSP;
+    }
+
     public function applies(InlineLexer $lexer): bool
     {
         return $lexer->token?->type === InlineLexer::NBSP;

packages/guides-restructured-text/src/RestructuredText/Parser/Productions/InlineRules/StandaloneEmailRule.php

@@ -26,8 +26,13 @@
  *
  * @see https://docutils.sourceforge.io/docs/ref/rst/restructuredtext.html#standalone-hyperlinks
  */
-final class StandaloneEmailRule extends ReferenceRule
+final class StandaloneEmailRule extends ReferenceRule implements CachableInlineRule
 {
+    public function getToken(): int
+    {
+        return InlineLexer::EMAIL;
+    }
+
     public function applies(InlineLexer $lexer): bool
     {
         return $lexer->token?->type === InlineLexer::EMAIL;

packages/guides-restructured-text/src/RestructuredText/Parser/Productions/InlineRules/StandaloneHyperlinkRule.php

@@ -26,8 +26,13 @@
  *
  * @see https://docutils.sourceforge.io/docs/ref/rst/restructuredtext.html#standalone-hyperlinks
  */
-final class StandaloneHyperlinkRule extends ReferenceRule
+final class StandaloneHyperlinkRule extends ReferenceRule implements CachableInlineRule
 {
+    public function getToken(): int
+    {
+        return InlineLexer::HYPERLINK;
+    }
+
     public function applies(InlineLexer $lexer): bool
     {
         return $lexer->token?->type === InlineLexer::HYPERLINK;

packages/guides-restructured-text/tests/benchmarks/BufferBench.php

@@ -2,6 +2,15 @@
 
 declare(strict_types=1);
 
+/**
+ * This file is part of phpDocumentor.
+ *
+ * For the full copyright and license information, please view the LICENSE
+ * file that was distributed with this source code.
+ *
+ * @link https://phpdoc.org
+ */
+
 namespace phpDocumentor\Guides\RestructuredText;
 
 use PhpBench\Attributes\Iterations;
@@ -21,7 +30,6 @@ public function __construct()
         $this->buffer->push(' Final line.');
     }
 
-
     #[Revs([1000, 10_000])]
     #[Iterations(5)]
     public function benchGetLines(): void