, , , etc.
- */
- }
- break;
+ break;
+ }
}
}
@@ -337,7 +311,7 @@ private function should_preserve_tag_in_rich_text( $tag ) {
}
private function is_at_inline_code_element() {
- $breadcrumbs = $this->html->get_breadcrumbs();
+ $breadcrumbs = $this->markup_processor->get_breadcrumbs();
foreach ( $breadcrumbs as $tag ) {
switch ( $tag ) {
case 'A':
@@ -363,12 +337,12 @@ private function is_at_inline_code_element() {
*
* @param string $html The HTML snippet to append.
*/
- private function append_html( $html ) {
+ private function append_rich_text( $html ) {
$html = trim( $html );
if ( empty( $html ) ) {
return;
}
- // Make sure two subsequent append_html() calls don't merge the text.
+ // Make sure two subsequent append_text() calls don't merge the text.
$html .= ' ';
$this->ensure_open_block();
$this->block_markup .= $html;
@@ -425,4 +399,8 @@ private function close_ephemeral_paragraph() {
$this->in_ephemeral_paragraph = false;
}
}
+
+ public function get_last_error() {
+ return $this->last_error;
+ }
}
diff --git a/packages/playground/data-liberation/src/entity-readers/WP_EPub_Entity_Reader.php b/packages/playground/data-liberation/src/entity-readers/WP_EPub_Entity_Reader.php
new file mode 100644
index 0000000000..db7b8b9df3
--- /dev/null
+++ b/packages/playground/data-liberation/src/entity-readers/WP_EPub_Entity_Reader.php
@@ -0,0 +1,123 @@
+zip = $zip;
+ $this->current_post_id = $first_post_id;
+ }
+
+ public function next_entity() {
+ if ( $this->last_error ) {
+ return false;
+ }
+
+ if ( $this->finished ) {
+ return false;
+ }
+
+ if ( null === $this->remaining_html_files ) {
+ $path = false;
+ foreach ( array( '/OEBPS', '/EPUB' ) as $path_candidate ) {
+ if ( $this->zip->is_dir( $path_candidate ) ) {
+ $path = $path_candidate;
+ break;
+ }
+ }
+ if ( false === $path ) {
+ _doing_it_wrong( __METHOD__, 'The EPUB file did not contain any HTML files.', '1.0.0' );
+ $this->finished = true;
+ return false;
+ }
+
+ $files = $this->zip->ls( $path );
+ if ( false === $files ) {
+ _doing_it_wrong( __METHOD__, 'The EPUB file did not contain any HTML files.', '1.0.0' );
+ $this->finished = true;
+ return false;
+ }
+ $this->remaining_html_files = array();
+ foreach ( $files as $file ) {
+ if ( str_ends_with( $file, '.xhtml' ) || str_ends_with( $file, '.html' ) ) {
+ $this->remaining_html_files[] = $path . '/' . $file;
+ }
+ }
+ }
+
+ while ( true ) {
+ if ( null !== $this->current_html_reader ) {
+ if (
+ ! $this->current_html_reader->is_finished() &&
+ $this->current_html_reader->next_entity()
+ ) {
+ return true;
+ }
+ if ( $this->current_html_reader->get_last_error() ) {
+ _doing_it_wrong(
+ __METHOD__,
+ 'The EPUB file did not contain any HTML files.',
+ '1.0.0'
+ );
+ $this->finished = true;
+ return false;
+ }
+ }
+
+ if ( count( $this->remaining_html_files ) === 0 ) {
+ $this->finished = true;
+ return false;
+ }
+
+ $html_file = array_shift( $this->remaining_html_files );
+ $html = $this->zip->read_file( $html_file );
+ $this->current_html_reader = new WP_HTML_Entity_Reader(
+ WP_XML_Processor::create_from_string( $html ),
+ $this->current_post_id
+ );
+ if ( $this->current_html_reader->get_last_error() ) {
+ $this->last_error = $this->current_html_reader->get_last_error();
+ return false;
+ }
+ ++$this->current_post_id;
+ }
+
+ return false;
+ }
+
+ public function get_entity() {
+ return $this->current_html_reader->get_entity();
+ }
+
+ public function is_finished(): bool {
+ return $this->finished;
+ }
+
+ public function get_last_error(): ?string {
+ return $this->last_error;
+ }
+}
diff --git a/packages/playground/data-liberation/src/entity-readers/WP_HTML_Entity_Reader.php b/packages/playground/data-liberation/src/entity-readers/WP_HTML_Entity_Reader.php
index 95923ef390..f500d698de 100644
--- a/packages/playground/data-liberation/src/entity-readers/WP_HTML_Entity_Reader.php
+++ b/packages/playground/data-liberation/src/entity-readers/WP_HTML_Entity_Reader.php
@@ -5,18 +5,7 @@
*/
class WP_HTML_Entity_Reader extends WP_Entity_Reader {
- /**
- * The HTML document to convert.
- *
- * @var string
- */
- protected $html;
-
- /**
- * The emitted entities.
- *
- * @var array
- */
+ protected $html_processor;
protected $entities;
/**
@@ -32,16 +21,11 @@ class WP_HTML_Entity_Reader extends WP_Entity_Reader {
* @var int
*/
protected $post_id;
+ protected $last_error;
- /**
- * Constructs the reader.
- *
- * @param string $html The HTML document to convert.
- * @param int $post_id The ID to use as `post_id` of the emitted post entity.
- */
- public function __construct( $html, $post_id ) {
- $this->html = $html;
- $this->post_id = $post_id;
+ public function __construct( $html_processor, $post_id ) {
+ $this->html_processor = $html_processor;
+ $this->post_id = $post_id;
}
/**
@@ -66,8 +50,9 @@ public function next_entity() {
}
// We did not read any entities yet. Let's convert the HTML document into entities.
- $converter = new WP_HTML_To_Blocks( $this->html );
+ $converter = new WP_HTML_To_Blocks( $this->html_processor );
if ( false === $converter->convert() ) {
+ $this->last_error = $converter->get_last_error();
return false;
}
@@ -135,6 +120,6 @@ public function is_finished(): bool {
* @return string|null The last error, or null if there was no error.
*/
public function get_last_error(): ?string {
- return null;
+ return $this->last_error;
}
}
diff --git a/packages/playground/data-liberation/src/xml-api/WP_XML_Processor.php b/packages/playground/data-liberation/src/xml-api/WP_XML_Processor.php
index 881e689020..b6b2a7669e 100644
--- a/packages/playground/data-liberation/src/xml-api/WP_XML_Processor.php
+++ b/packages/playground/data-liberation/src/xml-api/WP_XML_Processor.php
@@ -1558,7 +1558,6 @@ private function parse_next_tag() {
* See https://www.w3.org/TR/xml11.xml/#sec-cdata-sect
*/
if (
- ! $this->is_closing_tag &&
$doc_length > $this->token_starts_at + 8 &&
'[' === $xml[ $this->token_starts_at + 2 ] &&
'C' === $xml[ $this->token_starts_at + 3 ] &&
@@ -1583,6 +1582,59 @@ private function parse_next_tag() {
return true;
}
+ /*
+ * Identify DOCTYPE nodes.
+ *
+ * See https://www.w3.org/TR/xml11.html/#dtd
+ */
+ if (
+ $doc_length > $this->token_starts_at + 8 &&
+ 'D' === $xml[ $at + 2 ] &&
+ 'O' === $xml[ $at + 3 ] &&
+ 'C' === $xml[ $at + 4 ] &&
+ 'T' === $xml[ $at + 5 ] &&
+ 'Y' === $xml[ $at + 6 ] &&
+ 'P' === $xml[ $at + 7 ] &&
+ 'E' === $xml[ $at + 8 ]
+ ) {
+ $at += 9;
+ // Skip whitespace.
+ $at += strspn( $this->xml, " \t\f\r\n", $at );
+
+ if ( $doc_length <= $at ) {
+ $this->mark_incomplete_input( 'Unclosed DOCTYPE declaration.' );
+
+ return false;
+ }
+
+ // @TODO: Expose the "name" value instead of skipping it like that
+ $at += $this->parse_name( $at );
+
+ // Skip whitespace.
+ $at += strspn( $this->xml, " \t\f\r\n", $at );
+
+ if ( $doc_length <= $at ) {
+ $this->mark_incomplete_input( 'Unclosed DOCTYPE declaration.' );
+ return false;
+ }
+
+ if ( $this->xml[ $at ] !== '>' ) {
+ $this->last_error = self::ERROR_SYNTAX;
+ _doing_it_wrong(
+ __METHOD__,
+ __( 'Unsupported DOCTYPE syntax. Only a simple is supported.' ),
+ 'WP_VERSION'
+ );
+ return false;
+ }
+
+ $closer_at = $at;
+ $this->parser_state = self::STATE_DOCTYPE_NODE;
+ $this->token_length = $closer_at + 1 - $this->token_starts_at;
+ $this->bytes_already_parsed = $closer_at + 1;
+ return true;
+ }
+
/*
* Anything else here is either unsupported at this point or invalid
* syntax. See the class-level @TODO annotations for more information.
@@ -2471,6 +2523,22 @@ public function get_tag() {
return null;
}
+ /**
+ * Indicates if the currently matched tag is expected to be closed.
+ * Returns true for tag openers (
) and false for empty elements (![]()
) and tag closers (
).
+ *
+ * This method exists to provide a consistent interface with WP_HTML_Processor.
+ *
+ * @return bool Whether the tag is expected to be closed.
+ */
+ public function expects_closer() {
+ if ( self::STATE_MATCHED_TAG !== $this->parser_state ) {
+ return false;
+ }
+
+ return $this->is_tag_opener() && ! $this->is_empty_element();
+ }
+
/**
* Indicates if the currently matched tag is an empty element tag.
*
@@ -2604,6 +2672,9 @@ public function get_token_name() {
case self::STATE_CDATA_NODE:
return '#cdata-section';
+ case self::STATE_DOCTYPE_NODE:
+ return '#doctype';
+
case self::STATE_XML_DECLARATION:
return '#xml-declaration';
@@ -3030,10 +3101,11 @@ private function step_in_prolog( $node_to_process = self::PROCESS_NEXT_NODE ) {
$this->last_error = self::ERROR_SYNTAX;
_doing_it_wrong( __METHOD__, 'Unexpected token type in prolog stage.', 'WP_VERSION' );
}
-
return $this->step();
- case '#xml-declaration':
+ // @TODO: Fail if there's more than one or if was found before the XML declaration token.
+ case '#doctype':
case '#comment':
+ case '#xml-declaration':
case '#processing-instructions':
return true;
case '#tag':
@@ -3393,6 +3465,18 @@ private function mark_incomplete_input(
*/
const STATE_CDATA_NODE = 'STATE_CDATA_NODE';
+ /**
+ * Parser DOCTYPE Node State.
+ *
+ * Indicates that the parser has found a DOCTYPE declaration and it's possible
+ * to read and modify its modifiable text.
+ *
+ * @since WP_VERSION
+ *
+ * @access private
+ */
+ const STATE_DOCTYPE_NODE = 'STATE_DOCTYPE_NODE';
+
/**
* Indicates that the parser has found an XML processing instruction.
*
diff --git a/packages/playground/data-liberation/tests/WPEPubEntityReaderTests.php b/packages/playground/data-liberation/tests/WPEPubEntityReaderTests.php
new file mode 100644
index 0000000000..c6bf17248c
--- /dev/null
+++ b/packages/playground/data-liberation/tests/WPEPubEntityReaderTests.php
@@ -0,0 +1,40 @@
+next_entity() ) {
+ $data = $reader->get_entity()->get_data();
+ $entities[] = [
+ 'type' => $reader->get_entity()->get_type(),
+ 'data' => $data,
+ ];
+ }
+ $this->assertNull( $reader->get_last_error() );
+ $this->assertEquals( 3, count($entities) );
+ $this->assertGreaterThan( 100, strlen($entities[0]['data']['content']) );
+ $this->assertGreaterThan( 1000, strlen($entities[1]['data']['content']) );
+ $this->assertGreaterThan( 1000, strlen($entities[2]['data']['content']) );
+ echo $entities[2]['data']['content'];
+ }
+
+ public function epub_byte_reader_data_provider() {
+ return [
+ 'Local file' => [
+ \WordPress\ByteReader\WP_File_Reader::create( __DIR__ . '/fixtures/epub-entity-reader/childrens-literature.epub' )
+ ],
+ 'Remote file' => [
+ \WordPress\ByteReader\WP_Remote_File_Ranged_Reader::create( 'https://github.com/IDPF/epub3-samples/releases/download/20230704/childrens-literature.epub' )
+ ],
+ ];
+ }
+
+}
diff --git a/packages/playground/data-liberation/tests/WPHTMLEntityReaderTests.php b/packages/playground/data-liberation/tests/WPHTMLEntityReaderTests.php
index 0385dc8358..c8e32fc573 100644
--- a/packages/playground/data-liberation/tests/WPHTMLEntityReaderTests.php
+++ b/packages/playground/data-liberation/tests/WPHTMLEntityReaderTests.php
@@ -13,7 +13,7 @@ public function test_entity_reader() {
It is our pleasure to announce that WordPress 6.8 was released
Last week, WordPress 6.8 was released.
HTML; - $reader = new WP_HTML_Entity_Reader( $html, 1 ); + $reader = new WP_HTML_Entity_Reader( new WP_HTML_Processor( $html ), 1 ); $entities = []; while ( $reader->next_entity() ) { $data = $reader->get_entity()->get_data(); diff --git a/packages/playground/data-liberation/tests/WPHTMLToBlocksTests.php b/packages/playground/data-liberation/tests/WPHTMLToBlocksTests.php index 41d6ba8ae8..d3daef742e 100644 --- a/packages/playground/data-liberation/tests/WPHTMLToBlocksTests.php +++ b/packages/playground/data-liberation/tests/WPHTMLToBlocksTests.php @@ -16,7 +16,7 @@ public function test_metadata_extraction() {WordPress 6.8 was released
Last week, WordPress 6.8 was released. This release includes a new default theme, a new block editor experience, and a new block library. It also includes a new block editor experience, and a new block library.
HTML; - $converter = new WP_HTML_To_Blocks( $html ); + $converter = new WP_HTML_To_Blocks( new WP_HTML_Processor( $html ) ); $converter->convert( $html ); $metadata = $converter->get_all_metadata(); $expected_metadata = [ @@ -35,7 +35,7 @@ public function test_metadata_extraction() { * @dataProvider provider_test_conversion */ public function test_html_to_blocks_conversion( $html, $expected ) { - $converter = new WP_HTML_To_Blocks( $html ); + $converter = new WP_HTML_To_Blocks( new WP_HTML_Processor( $html ) ); $converter->convert( $html ); $blocks = $converter->get_block_markup(); @@ -89,16 +89,12 @@ public function provider_test_conversion() { ], 'Formatted text' => [ 'html' => 'Bold and Italic
', - 'expected' => "Bold and Italic
" + 'expected' => "Bold and Italic
" ], 'A blockquote' => [ 'html' => 'A simple blockquote', 'expected' => "
A simple blockquote" ], - 'A an tag' => [ - 'html' => '', - 'expected' => "<input type="text" value="A simple input"> " - ], 'A table' => [ 'html' => <<
