Add a new HTMLEntry

stloyd · stloyd · commit 254aacfffd4e · 2025-11-02T16:14:50.000+01:00
diff --git a/src/lib/types/src/Flow/Types/Type/Native/String/StringTypeChecker.php b/src/lib/types/src/Flow/Types/Type/Native/String/StringTypeChecker.php
@@ -4,8 +4,7 @@
 
 namespace Flow\Types\Type\Native\String;
 
-use Dom\HTMLDocument;
-use Flow\Types\Value\Uuid;
+use Flow\Types\Value\{HTMLDocument, Uuid};
 
 final readonly class StringTypeChecker
 {
@@ -101,53 +100,12 @@ public function isFloat() : bool
             return false;
         }
 
-        // scientific notation
-        if (\is_numeric($this->string) && (\str_contains($this->string, 'e') || \str_contains($this->string, 'E'))) {
-
-        }
-
         return \is_numeric($this->string) && \str_contains($this->string, '.');
     }
 
     public function isHTML() : bool
     {
-        if ($this->string === '') {
-            return false;
-        }
-
-        if ('<' !== $this->string[0]) {
-            return false;
-        }
-
-        if (\preg_match('/(<!doctype(.+?)>)?<html(.+?)>(.+?)<\/html>/im', $this->string) === 1) {
-            if (\class_exists('\Dom\HTMLDocument', false)) {
-                $options = \LIBXML_HTML_NOIMPLIED;
-
-                if (defined('Dom\HTML_NO_DEFAULT_NS')) {
-                    $options |= constant('\Dom\HTML_NO_DEFAULT_NS');
-                }
-
-                $doc = HTMLDocument::createFromString($this->string, $options);
-
-                return \in_array(\strtolower((string) $doc->documentElement->tagName), ['doctype', 'html', 'html5'], true);
-            }
-
-            try {
-                \libxml_use_internal_errors(true);
-
-                $doc = new \DOMDocument();
-                $result = @$doc->loadHTML($this->string);
-
-                return (bool) $result;
-            } catch (\Exception) {
-                return false;
-            } finally {
-                \libxml_clear_errors(); // Clear any errors if needed
-                \libxml_use_internal_errors(false); // Restore standard error handling
-            }
-        }
-
-        return false;
+        return HTMLDocument::isValid($this->string);
     }
 
     public function isInteger() : bool
diff --git a/src/lib/types/src/Flow/Types/Value/HTMLDocument.php b/src/lib/types/src/Flow/Types/Value/HTMLDocument.php
@@ -34,7 +34,7 @@ public function __construct(string|object $value)
         // Cut all new lines and tabs
         $value = trim(str_replace(["\n", "\t"], '', $value));
 
-        if (!$this->isValid($value)) {
+        if (!self::isValid($value)) {
             throw new InvalidArgumentException('Invalid HTML document given: ' . var_export($value, true));
         }
 
@@ -46,6 +46,19 @@ public static function fromString(string $value) : self
         return new self($value);
     }
 
+    public static function isValid(string $value) : bool
+    {
+        if ('' === $value) {
+            return false;
+        }
+
+        if ('<' !== $value[0]) {
+            return false;
+        }
+
+        return \preg_match(self::HTML_ALIKE_REGEX, $value) === 1;
+    }
+
     public function __toString() : string
     {
         return $this->toString();
@@ -60,13 +73,4 @@ public function toString() : string
     {
         return $this->value;
     }
-
-    private function isValid(string $value) : bool
-    {
-        if ('' === $value) {
-            return false;
-        }
-
-        return \preg_match(self::HTML_ALIKE_REGEX, $value) === 1;
-    }
 }
diff --git a/src/lib/types/tests/Flow/Types/Tests/Unit/Type/Native/String/StringTypeCheckerTest.php b/src/lib/types/tests/Flow/Types/Tests/Unit/Type/Native/String/StringTypeCheckerTest.php
@@ -55,7 +55,8 @@ public function test_detecting_float() : void
 
     public function test_detecting_html() : void
     {
-        self::assertTrue((new StringTypeChecker('<html lang="en"><body><div><span>1</span></div></body></html>'))->isHTML());
+        self::assertTrue((new StringTypeChecker('<!DOCTYPE html><html><head></head><body><div>baz</div></body></html>'))->isHTML());
+        self::assertFalse((new StringTypeChecker('<html lang="en"><body><div><span>1</span></div></body></html>'))->isHTML());
         self::assertFalse((new StringTypeChecker('not html'))->isHTML());
     }
 

Original file line number	Diff line number	Diff line change
`@@ -55,7 +55,8 @@ public function test_detecting_float() : void`
`55`	`55`
`56`	`56`	`public function test_detecting_html() : void`
`57`	`57`	`{`
`58`		`- self::assertTrue((new StringTypeChecker('<html lang="en"><body><div><span>1</span></div></body></html>'))->isHTML());`
	`58`	`+ self::assertTrue((new StringTypeChecker('<!DOCTYPE html><html><head></head><body><div>baz</div></body></html>'))->isHTML());`
	`59`	`+ self::assertFalse((new StringTypeChecker('<html lang="en"><body><div><span>1</span></div></body></html>'))->isHTML());`
`59`	`60`	`self::assertFalse((new StringTypeChecker('not html'))->isHTML());`
`60`	`61`	`}`
`61`	`62`