From dfb38131b2b7765e88545d4da304d37f390b8fb5 Mon Sep 17 00:00:00 2001
From: "ondrej@mirtes.cz" <ondrej@mirtes.cz>
Date: Mon, 26 Oct 2026 17:23:11 +0000
Subject: [PATCH] Do not walk to the end of a PHPDoc for every generic

TypeParser::isHtml() looked for the end tag of every "<name>" by walking the
rest of the tokens, so a PHPDoc with many generics like list<string> took
quadratic time. The iterator now remembers once where each "/name>" is
written, and the walk only runs when one is written ahead.

Amp-Thread-ID: https://ampcode.com/threads/T-01a17820-94d1-75c8-9206-2e802d9c0556
---
 src/Parser/TokenIterator.php | 33 +++++++++++++++++++++++++++++++++
 src/Parser/TypeParser.php    |  5 +++++
 2 files changed, 38 insertions(+)

diff --git a/src/Parser/TokenIterator.php b/src/Parser/TokenIterator.php
index 043639f1..259a6bdd 100644
--- a/src/Parser/TokenIterator.php
+++ b/src/Parser/TokenIterator.php
@@ -9,7 +9,9 @@
 use function assert;
 use function count;
 use function in_array;
+use function preg_match_all;
 use function strlen;
+use function strpos;
 use function substr;
 
 class TokenIterator
@@ -34,6 +36,9 @@ class TokenIterator
 	/** @var list<int>|null */
 	private ?array $offsets = null;
 
+	/** @var array<string, int>|null the last index of a token containing "/name>", by the name */
+	private ?array $endTagIndexes = null;
+
 	/**
 	 * @param list<array{string, int, int}> $tokens
 	 */
@@ -390,6 +395,34 @@ public function hasTokenImmediatelyAfter(int $pos, int $expectedTokenType): bool
 		return false;
 	}
 
+	/**
+	 * Whether a token at the current index or after it contains "/$name>",
+	 * which every end tag of an HTML tag of that name does.
+	 *
+	 * The tokens are searched once, so that asking about every "<" of a
+	 * PHPDoc does not walk to its end every time.
+	 *
+	 * @internal
+	 */
+	public function hasEndTagCandidateAhead(string $name): bool
+	{
+		if ($this->endTagIndexes === null) {
+			$this->endTagIndexes = [];
+			foreach ($this->tokens as $index => $token) {
+				$value = $token[Lexer::VALUE_OFFSET];
+				if (strpos($value, '/') === false || preg_match_all('~/([^/>]++)>~', $value, $matches) === 0) {
+					continue;
+				}
+
+				foreach ($matches[1] as $candidate) {
+					$this->endTagIndexes[$candidate] = $index;
+				}
+			}
+		}
+
+		return ($this->endTagIndexes[$name] ?? -1) >= $this->index;
+	}
+
 	public function getDetectedNewline(): ?string
 	{
 		return $this->newline;
diff --git a/src/Parser/TypeParser.php b/src/Parser/TypeParser.php
index a2db4fce..348c56d1 100644
--- a/src/Parser/TypeParser.php
+++ b/src/Parser/TypeParser.php
@@ -419,6 +419,11 @@ public function isHtml(TokenIterator $tokens): bool
 			return false;
 		}
 
+		// every end tag the walk below looks for contains this
+		if (!$tokens->hasEndTagCandidateAhead($htmlTagName)) {
+			return false;
+		}
+
 		$endTag = '</' . $htmlTagName . '>';
 		$endTagSearchOffset = - strlen($endTag);