[Data Liberation] Add EPub to Blocks converter

packages/playground/data-liberation/bootstrap.php

@@ -61,6 +61,7 @@
 
 require_once __DIR__ . '/src/entity-readers/WP_Entity_Reader.php';
 require_once __DIR__ . '/src/entity-readers/WP_HTML_Entity_Reader.php';
+require_once __DIR__ . '/src/entity-readers/WP_EPub_Entity_Reader.php';
 require_once __DIR__ . '/src/entity-readers/WP_WXR_Entity_Reader.php';
 require_once __DIR__ . '/src/entity-readers/WP_Directory_Tree_Entity_Reader.php';
 

packages/playground/data-liberation/src/block-markup/WP_HTML_To_Blocks.php

@@ -28,36 +28,43 @@ class WP_HTML_To_Blocks implements WP_Block_Markup_Converter {
 
 	private $state       = self::STATE_READY;
 	private $block_stack = array();
-	private $html;
+	private $markup_processor;
 	private $ignore_text            = false;
 	private $in_ephemeral_paragraph = false;
 	private $block_markup           = '';
 	private $metadata               = array();
+	private $last_error             = null;
 
-	public function __construct( $html ) {
-		$this->html = new \WP_HTML_Processor( $html );
+	public function __construct( $markup_processor ) {
+		$this->markup_processor = $markup_processor;
 	}
 
 	public function convert() {
 		if ( self::STATE_READY !== $this->state ) {
 			return false;
 		}
 
-		while ( $this->html->next_token() ) {
-			switch ( $this->html->get_token_type() ) {
+		while ( $this->markup_processor->next_token() ) {
+			switch ( $this->markup_processor->get_token_type() ) {
 				case '#text':
 					if ( $this->ignore_text ) {
 						break;
 					}
-					$this->append_rich_text( htmlspecialchars( $this->html->get_modifiable_text() ) );
+					$this->append_rich_text( htmlspecialchars( $this->markup_processor->get_modifiable_text() ) );
 					break;
 				case '#tag':
 					$this->handle_tag();
 					break;
 			}
 		}
 
+		if ( $this->markup_processor->get_last_error() ) {
+			$this->last_error = $this->markup_processor->get_last_error();
+			return false;
+		}
+
 		$this->close_ephemeral_paragraph();
+
 		return true;
 	}
 
@@ -77,12 +84,12 @@ public function get_block_markup() {
 	}
 
 	private function handle_tag() {
-		$html          = $this->html;
-		$tag           = $html->get_tag();
+		$html          = $this->markup_processor;
+		$tag           = strtoupper( $html->get_tag() );
 		$tag_lowercase = strtolower( $tag );
 
-		$is_tag_opener = ! $html->is_tag_closer();
-		if ( ! $html->expects_closer() ) {
+		$is_void_tag = ! $html->expects_closer() && ! $html->is_tag_closer();
+		if ( $is_void_tag ) {
 			switch ( $tag ) {
 				case 'META':
 					$key   = $html->get_attribute( 'name' );
@@ -110,7 +117,7 @@ private function handle_tag() {
 					//        Just insert an HTML block or what?
 					break;
 			}
-		} elseif ( $is_tag_opener ) {
+		} elseif ( ! $html->is_tag_closer() ) {
 			switch ( $tag ) {
 				// Block elements
 				case 'SCRIPT':
@@ -304,7 +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':
@@ -392,4 +399,8 @@ private function close_ephemeral_paragraph() {
 			$this->in_ephemeral_paragraph = false;
 		}
 	}
+
+	public function get_last_error() {
+		return $this->last_error;
+	}
 }

packages/playground/data-liberation/src/entity-readers/WP_EPub_Entity_Reader.php

@@ -0,0 +1,123 @@
+<?php
+
+use WordPress\Zip\WP_Zip_Filesystem;
+
+/**
+ * https://www.w3.org/AudioVideo/ebook/
+ *
+ * An EPUB Publication is transported as a single file (a "portable document") that contains:
+ * * a Package Document (OPF file) which specifies all the Publication's constituent content documents and their required resources, defines a reading order  and associates Publication-level metadata and navigation information.
+ *    * A metadata element including and/or referencing metadata applicable to the entire Publication and particular resources within it.
+ *    * A manifest element: identifies (via IRI) and describes (via MIME media type) the set of resources that constitute the EPUB Publication.
+ *    * A spine element : defines the default reading order of the Publication. (An ordered list of Publication Resources (EPUB Content Documents).
+ *    * A Bindings element defines a set of custom handlers for media types not supported by EPUB3. If the Reading System cannot support the specific media type, it could use scripting fallback if supported.
+ * * all Content Documents
+ * * all other required resources for processing the Publication.
+ *
+ * The OCF Container is packaged into a physical single ZIP file containing:
+ * * Mime Type file: application/epub+zip.
+ * * META-INF folder (container file which points to the location of the .opf file), signatures, encryption, rights, are xml files
+ * * OEBPS folder stores the book content .(opf, ncx, html, svg, png, css, etc. files)
+ */
+class WP_EPub_Entity_Reader extends WP_Entity_Reader {
+
+	protected $zip;
+	protected $finished = false;
+	protected $current_post_id;
+	protected $remaining_html_files;
+	protected $current_html_reader;
+	protected $last_error;
+	public function __construct( WP_Zip_Filesystem $zip, $first_post_id = 1 ) {
+		$this->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;
+	}
+}

packages/playground/data-liberation/src/entity-readers/WP_HTML_Entity_Reader.php

@@ -1,22 +1,21 @@
 <?php
 
-use WordPress\Data_Liberation\Block_Markup\WP_HTML_To_Blocks;
-
 /**
  * Converts a single HTML file into a stream of WordPress entities.
  *
  * @TODO: Support post meta.
  */
 class WP_HTML_Entity_Reader extends WP_Entity_Reader {
 
-	protected $html;
+	protected $html_processor;
 	protected $entities;
 	protected $finished = false;
 	protected $post_id;
+	protected $last_error;
 
-	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;
 	}
 
 	public function next_entity() {
@@ -36,8 +35,9 @@ 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;
 		}
 
@@ -90,6 +90,6 @@ public function is_finished(): bool {
 	}
 
 	public function get_last_error(): ?string {
-		return null;
+		return $this->last_error;
 	}
 }