Add a new HTMLEntry

src/adapter/etl-adapter-parquet/tests/Flow/ETL/Adapter/Parquet/Tests/Unit/RowsNormalizerTest.php

@@ -12,6 +12,8 @@ enum_entry,
     enum_schema,
     float_entry,
     float_schema,
+    html_entry,
+    html_schema,
     int_entry,
     int_schema,
     json_entry,
@@ -72,6 +74,7 @@ public function test_normalization_nullable_entries() : void
                 ),
                 enum_entry('enum', null),
                 xml_entry('xml', null),
+                html_entry('html', null),
             )
         );
         $schema = schema(
@@ -101,6 +104,7 @@ enum_entry('enum', null),
             ),
             enum_schema('enum', BackedStringEnum::class, true),
             xml_schema('xml', true),
+            html_schema('html', true),
         );
 
         self::assertEquals(
@@ -119,6 +123,7 @@ enum_schema('enum', BackedStringEnum::class, true),
                     'struct' => null,
                     'enum' => null,
                     'xml' => null,
+                    'html' => null,
                 ],
             ],
             (new RowsNormalizer())->normalize($rows, $schema)

src/core/etl/src/Flow/ETL/DSL/functions.php

@@ -165,6 +165,7 @@
 use Flow\ETL\Retry\RetryStrategy\{AnyThrowable, OnExceptionTypes};
 use Flow\ETL\Row\{Entries, EntryFactory, SortOrder};
 use Flow\ETL\Row\Entry\{BooleanEntry, DateEntry, DateTimeEntry, EnumEntry, FloatEntry, IntegerEntry, JsonEntry, ListEntry, MapEntry, StringEntry, StructureEntry, TimeEntry, UuidEntry, XMLElementEntry, XMLEntry};
+use Flow\ETL\Row\Entry\HTMLEntry;
 use Flow\ETL\Row\{Entry, EntryReference, Reference, References};
 use Flow\ETL\Row\Formatter\ASCIISchemaFormatter;
 use Flow\ETL\Schema\{Definition, Formatter\PHPFormatter\TypeFormatter, Formatter\PHPFormatter\ValueFormatter};
@@ -627,6 +628,15 @@ function xml_element_entry(string $name, \DOMElement|string|null $value, ?Metada
     return new XMLElementEntry($name, $value, $metadata);
 }
 
+/**
+ * @return Entry<?HTMLDocument>
+ */
+#[DocumentationDSL(module: Module::CORE, type: DSLType::ENTRY)]
+function html_entry(string $name, HTMLDocument|string|null $value, ?Metadata $metadata = null) : Entry
+{
+    return new HTMLEntry($name, $value, $metadata);
+}
+
 /**
  * @param Entry<mixed> ...$entries
  */

src/core/etl/src/Flow/ETL/Row/Entry/HTMLEntry.php

@@ -0,0 +1,119 @@
+<?php
+
+declare(strict_types=1);
+
+namespace Flow\ETL\Row\Entry;
+
+use function Flow\Types\DSL\{type_equals, type_html, type_optional};
+use Flow\ETL\Row\{Entry, Reference};
+use Flow\ETL\Schema\{Definition, Metadata};
+use Flow\Types\Type;
+use Flow\Types\Value\HTMLDocument;
+
+/**
+ * @implements Entry<?HTMLDocument>
+ */
+final class HTMLEntry implements Entry
+{
+    use EntryRef;
+
+    private Metadata $metadata;
+
+    /**
+     * @var Type<HTMLDocument>
+     */
+    private readonly Type $type;
+
+    private ?HTMLDocument $value;
+
+    public function __construct(
+        private readonly string $name,
+        HTMLDocument|string|null $value,
+        ?Metadata $metadata = null,
+    ) {
+        if (\is_string($value)) {
+            $this->value = HTMLDocument::fromString($value);
+        } else {
+            $this->value = $value;
+        }
+
+        $this->metadata = $metadata ?: Metadata::empty();
+        $this->type = type_html();
+    }
+
+    public function __toString() : string
+    {
+        return $this->toString();
+    }
+
+    public function definition() : Definition
+    {
+        return new Definition($this->name, $this->type, null === $this->value, $this->metadata);
+    }
+
+    public function duplicate() : self
+    {
+        return new self($this->name, $this->value ? clone $this->value : null, $this->metadata);
+    }
+
+    public function is(Reference|string $name) : bool
+    {
+        if ($name instanceof Reference) {
+            return $this->name === $name->name();
+        }
+
+        return $this->name === $name;
+    }
+
+    public function isEqual(Entry $entry) : bool
+    {
+        if (!$entry instanceof self || !$this->is($entry->name())) {
+            return false;
+        }
+
+        if (!type_equals($this->type, $entry->type)) {
+            return false;
+        }
+
+        return $entry->value()?->toString() === $this->value?->toString();
+    }
+
+    public function map(callable $mapper) : self
+    {
+        return new self($this->name, $mapper($this->value));
+    }
+
+    public function name() : string
+    {
+        return $this->name;
+    }
+
+    public function rename(string $name) : self
+    {
+        return new self($name, $this->value);
+    }
+
+    public function toString() : string
+    {
+        if (null === $this->value) {
+            return '';
+        }
+
+        return $this->value->toString();
+    }
+
+    public function type() : Type
+    {
+        return $this->type;
+    }
+
+    public function value() : ?HTMLDocument
+    {
+        return $this->value;
+    }
+
+    public function withValue(mixed $value) : self
+    {
+        return new self($this->name, type_optional($this->type())->assert($value), $this->metadata);
+    }
+}

src/core/etl/src/Flow/ETL/Row/EntryFactory.php

@@ -9,6 +9,7 @@
     datetime_entry,
     enum_entry,
     float_entry,
+    html_entry,
     int_entry,
     json_entry,
     json_object_entry,
@@ -51,6 +52,7 @@ enum_entry,
     UuidType,
     XMLElementType,
     XMLType};
+use Flow\Types\Type\Logical\HTMLType;
 use Flow\Types\Type\Native\{
     ArrayType,
     BooleanType,
@@ -171,6 +173,7 @@ public function createAs(string $entryName, mixed $value, Definition|Type $defin
                 NullType::class => StringEntry::fromNull($entryName, $metadata),
                 XMLType::class => xml_entry($entryName, null, $metadata),
                 XMLElementType::class => xml_element_entry($entryName, null, $metadata),
+                HTMLType::class => html_entry($entryName, null, $metadata),
                 default => throw new InvalidArgumentException("Can't convert value into type \"{$type->toString()}\""),
             };
         }
@@ -234,6 +237,10 @@ public function createAs(string $entryName, mixed $value, Definition|Type $defin
                 }
             }
 
+            if ($type instanceof HTMLType) {
+                return html_entry($entryName, type_optional($type)->cast($value), $metadata);
+            }
+
             if ($type instanceof XMLType) {
                 return xml_entry($entryName, type_optional($type)->cast($value), $metadata);
             }

src/core/etl/tests/Flow/ETL/Tests/Integration/DataFrame/DisplayTest.php

@@ -11,6 +11,7 @@ enum_entry,
     float_entry,
     from_array,
     from_rows,
+    html_entry,
     int_entry,
     json_entry,
     list_entry,
@@ -75,6 +76,7 @@ public function extract(FlowContext $context) : \Generator
                                 ),
                                 enum_entry('enum', BackedStringEnum::three),
                                 xml_entry('xml', '<xml><node id="123">test<foo>bar</foo></node></xml>'),
+                                html_entry('html', '<!DOCTYPE html><html lang="en"><head></head><body></body></html>'),
                             ),
                         );
                     }
@@ -84,15 +86,15 @@ enum_entry('enum', BackedStringEnum::three),
 
         self::assertCommandOutputIdentical(
             <<<'ASCIITABLE'
-+------+------------+-----+---------+----------------------+-------+----------------------+---------+-------------------+----------------------+-------+----------------------+
-|   id |      price | 100 | deleted |           created-at | phase |                array |    list |               map |                items |  enum |                  xml |
-+------+------------+-----+---------+----------------------+-------+----------------------+---------+-------------------+----------------------+-------+----------------------+
-| 1234 | 123.450000 | 100 |   false | 2020-07-13T15:00:00+ |       | [{"id":1,"status":"N | [1,2,3] | ["NEW","PENDING"] | {"item-id":"1","name | three | <xml><node id="123"> |
-| 1234 | 123.450000 | 100 |   false | 2020-07-13T15:00:00+ |       | [{"id":1,"status":"N | [1,2,3] | ["NEW","PENDING"] | {"item-id":"1","name | three | <xml><node id="123"> |
-| 1234 | 123.450000 | 100 |   false | 2020-07-13T15:00:00+ |       | [{"id":1,"status":"N | [1,2,3] | ["NEW","PENDING"] | {"item-id":"1","name | three | <xml><node id="123"> |
-| 1234 | 123.450000 | 100 |   false | 2020-07-13T15:00:00+ |       | [{"id":1,"status":"N | [1,2,3] | ["NEW","PENDING"] | {"item-id":"1","name | three | <xml><node id="123"> |
-| 1234 | 123.450000 | 100 |   false | 2020-07-13T15:00:00+ |       | [{"id":1,"status":"N | [1,2,3] | ["NEW","PENDING"] | {"item-id":"1","name | three | <xml><node id="123"> |
-+------+------------+-----+---------+----------------------+-------+----------------------+---------+-------------------+----------------------+-------+----------------------+
++------+------------+-----+---------+----------------------+-------+----------------------+---------+-------------------+----------------------+-------+----------------------+----------------------+
+|   id |      price | 100 | deleted |           created-at | phase |                array |    list |               map |                items |  enum |                  xml |                 html |
++------+------------+-----+---------+----------------------+-------+----------------------+---------+-------------------+----------------------+-------+----------------------+----------------------+
+| 1234 | 123.450000 | 100 |   false | 2020-07-13T15:00:00+ |       | [{"id":1,"status":"N | [1,2,3] | ["NEW","PENDING"] | {"item-id":"1","name | three | <xml><node id="123"> | <!DOCTYPE html><html |
+| 1234 | 123.450000 | 100 |   false | 2020-07-13T15:00:00+ |       | [{"id":1,"status":"N | [1,2,3] | ["NEW","PENDING"] | {"item-id":"1","name | three | <xml><node id="123"> | <!DOCTYPE html><html |
+| 1234 | 123.450000 | 100 |   false | 2020-07-13T15:00:00+ |       | [{"id":1,"status":"N | [1,2,3] | ["NEW","PENDING"] | {"item-id":"1","name | three | <xml><node id="123"> | <!DOCTYPE html><html |
+| 1234 | 123.450000 | 100 |   false | 2020-07-13T15:00:00+ |       | [{"id":1,"status":"N | [1,2,3] | ["NEW","PENDING"] | {"item-id":"1","name | three | <xml><node id="123"> | <!DOCTYPE html><html |
+| 1234 | 123.450000 | 100 |   false | 2020-07-13T15:00:00+ |       | [{"id":1,"status":"N | [1,2,3] | ["NEW","PENDING"] | {"item-id":"1","name | three | <xml><node id="123"> | <!DOCTYPE html><html |
++------+------------+-----+---------+----------------------+-------+----------------------+---------+-------------------+----------------------+-------+----------------------+----------------------+
 5 rows
 
 ASCIITABLE,