class-blockswithmetadataentityreader.php
6 days ago
class-databasecontententityreader.php
6 days ago
class-databaserowsentityreader.php
6 days ago
class-entityreaderiterator.php
6 days ago
class-epubentityreader.php
6 days ago
class-filesystementityreader.php
6 days ago
class-htmlentityreader.php
6 days ago
class-wxrentityreader.php
6 days ago
interface-entity-reader.php
6 days ago
class-epubentityreader.php
197 lines
| 1 | <?php |
| 2 | |
| 3 | namespace WordPress\DataLiberation\EntityReader; |
| 4 | |
| 5 | use WordPress\DataLiberation\DataFormatConsumer\BlocksWithMetadata; |
| 6 | use WordPress\DataLiberation\DataFormatConsumer\MarkupProcessorConsumer; |
| 7 | use WordPress\Filesystem\Filesystem; |
| 8 | use WordPress\XML\XMLProcessor; |
| 9 | |
| 10 | use function WordPress\Filesystem\wp_join_unix_paths; |
| 11 | |
| 12 | /** |
| 13 | * EPub specification: https://www.w3.org/AudioVideo/ebook/ |
| 14 | * |
| 15 | * An EPUB Publication is transported as a single file (a "portable document") that contains: |
| 16 | * * 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. |
| 17 | * * A metadata element including and/or referencing metadata applicable to the entire Publication and particular resources within it. |
| 18 | * * A manifest element: identifies (via IRI) and describes (via MIME media type) the set of resources that constitute the EPUB Publication. |
| 19 | * * A spine element : defines the default reading order of the Publication. (An ordered list of Publication Resources (EPUB Content Documents). |
| 20 | * * 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. |
| 21 | * * all Content Documents |
| 22 | * * all other required resources for processing the Publication. |
| 23 | * |
| 24 | * The OCF Container is packaged into a physical single ZIP file containing: |
| 25 | * * Mime Type file: application/epub+zip. |
| 26 | * * META-INF folder (container file which points to the location of the .opf file), signatures, encryption, rights, are xml files |
| 27 | * * OEBPS folder stores the book content .(opf, ncx, html, svg, png, css, etc. files) |
| 28 | */ |
| 29 | class EPubEntityReader implements EntityReader { |
| 30 | |
| 31 | protected $zip; |
| 32 | protected $finished = false; |
| 33 | protected $current_post_id; |
| 34 | protected $remaining_html_files; |
| 35 | protected $current_html_reader; |
| 36 | protected $manifest; |
| 37 | protected $manifest_path; |
| 38 | |
| 39 | public function __construct( Filesystem $zip, $first_post_id = 1 ) { |
| 40 | $this->zip = $zip; |
| 41 | $this->current_post_id = $first_post_id; |
| 42 | } |
| 43 | |
| 44 | public function next_entity() { |
| 45 | if ( $this->finished ) { |
| 46 | return false; |
| 47 | } |
| 48 | |
| 49 | if ( null === $this->remaining_html_files ) { |
| 50 | if ( false === $this->parse_manifest() ) { |
| 51 | _doing_it_wrong( __METHOD__, 'The EPUB file did not contain a manifest.', '1.0.0' ); |
| 52 | $this->finished = true; |
| 53 | |
| 54 | return false; |
| 55 | } |
| 56 | |
| 57 | $this->remaining_html_files = array(); |
| 58 | foreach ( $this->manifest['items'] as $item ) { |
| 59 | if ( 'application/xhtml+xml' !== $item['media-type'] ) { |
| 60 | continue; |
| 61 | } |
| 62 | if ( 'nav' === ( $item['properties'] ?? '' ) ) { |
| 63 | continue; |
| 64 | } |
| 65 | $this->remaining_html_files[] = wp_join_unix_paths( |
| 66 | dirname( $this->manifest_path ), |
| 67 | $item['href'] |
| 68 | ); |
| 69 | } |
| 70 | } |
| 71 | |
| 72 | while ( true ) { |
| 73 | if ( null !== $this->current_html_reader ) { |
| 74 | if ( |
| 75 | ! $this->current_html_reader->is_finished() && |
| 76 | $this->current_html_reader->next_entity() |
| 77 | ) { |
| 78 | return true; |
| 79 | } |
| 80 | } |
| 81 | |
| 82 | if ( 0 === count( $this->remaining_html_files ) ) { |
| 83 | $this->finished = true; |
| 84 | |
| 85 | return false; |
| 86 | } |
| 87 | |
| 88 | $html_file = array_shift( $this->remaining_html_files ); |
| 89 | $html = $this->zip->get_contents( $html_file ); |
| 90 | $converter = new MarkupProcessorConsumer( |
| 91 | XMLProcessor::create_from_string( $html ) |
| 92 | ); |
| 93 | $blocks_with_meta = $converter->consume(); |
| 94 | $meta = $blocks_with_meta->get_all_metadata(); |
| 95 | if ( ! array_key_exists( 'post_name', $meta ) ) { |
| 96 | $meta['post_name'] = array( |
| 97 | basename( $html_file, '.xhtml' ), |
| 98 | ); |
| 99 | } |
| 100 | if ( ! array_key_exists( 'post_title', $meta ) ) { |
| 101 | $meta['post_title'] = array( |
| 102 | basename( $html_file, '.xhtml' ), |
| 103 | ); |
| 104 | } |
| 105 | $meta['post_type'] = array( 'page' ); |
| 106 | $meta['post_status'] = array( 'publish' ); |
| 107 | $meta['link'] = array( 'file://' . $html_file ); |
| 108 | $blocks_with_meta = new BlocksWithMetadata( |
| 109 | $blocks_with_meta->get_block_markup(), |
| 110 | $meta |
| 111 | ); |
| 112 | $this->current_html_reader = new HTMLEntityReader( |
| 113 | $blocks_with_meta, |
| 114 | $this->current_post_id |
| 115 | ); |
| 116 | ++$this->current_post_id; |
| 117 | } |
| 118 | |
| 119 | return false; |
| 120 | } |
| 121 | |
| 122 | /** |
| 123 | * An absolute path to the manifest file. Starting with slash. |
| 124 | */ |
| 125 | public function get_manifest_path() { |
| 126 | if ( null === $this->manifest_path ) { |
| 127 | $this->parse_manifest(); |
| 128 | } |
| 129 | |
| 130 | return $this->manifest_path; |
| 131 | } |
| 132 | |
| 133 | private function parse_manifest() { |
| 134 | if ( null !== $this->manifest ) { |
| 135 | return true; |
| 136 | } |
| 137 | |
| 138 | $xml = XMLProcessor::create_from_string( |
| 139 | $this->zip->get_contents( 'META-INF/container.xml' ) |
| 140 | ); |
| 141 | |
| 142 | if ( false === $xml->next_tag( array( 'urn:oasis:names:tc:opendocument:xmlns:container', 'rootfile' ) ) ) { |
| 143 | return false; |
| 144 | } |
| 145 | |
| 146 | $full_path = $xml->get_attribute( '', 'full-path' ); |
| 147 | if ( ! $full_path ) { |
| 148 | return false; |
| 149 | } |
| 150 | |
| 151 | $this->manifest_path = '/' . ltrim( $full_path, '/' ); |
| 152 | $manifest = $this->zip->get_contents( $this->manifest_path ); |
| 153 | if ( ! $manifest ) { |
| 154 | return false; |
| 155 | } |
| 156 | $xml = XMLProcessor::create_from_string( |
| 157 | $manifest |
| 158 | ); |
| 159 | |
| 160 | $parsed = array( |
| 161 | 'metadata' => array(), |
| 162 | 'items' => array(), |
| 163 | ); |
| 164 | while ( $xml->next_tag() ) { |
| 165 | $parsed_entry = array(); |
| 166 | $keys = $xml->get_attribute_names_with_prefix( '', '' ); |
| 167 | foreach ( $keys as list($ns, $key) ) { |
| 168 | $parsed_entry[ $key ] = $xml->get_attribute( $ns, $key ); |
| 169 | } |
| 170 | if ( $xml->matches_breadcrumbs( array( 'package', 'metadata', '*' ) ) ) { |
| 171 | $parsed['metadata'][] = array( |
| 172 | 'tag' => $xml->get_tag_local_name(), |
| 173 | 'attributes' => $parsed_entry, |
| 174 | ); |
| 175 | } elseif ( $xml->matches_breadcrumbs( array( 'package', 'manifest', 'item' ) ) ) { |
| 176 | $parsed_entry['type'] = 'item'; |
| 177 | $parsed['items'][] = $parsed_entry; |
| 178 | } |
| 179 | } |
| 180 | $this->manifest = $parsed; |
| 181 | |
| 182 | return true; |
| 183 | } |
| 184 | |
| 185 | public function get_entity() { |
| 186 | return $this->current_html_reader->get_entity(); |
| 187 | } |
| 188 | |
| 189 | public function is_finished(): bool { |
| 190 | return $this->finished; |
| 191 | } |
| 192 | |
| 193 | public function get_reentrancy_cursor() { |
| 194 | return ''; |
| 195 | } |
| 196 | } |
| 197 |