#5 Add directory parser.

.gitattributes

@@ -5,4 +5,3 @@ bin/ export-ignore
 .gitignore export-ignore
 composer.lock export-ignore
 phpunit.xml export-ignore
-hamepub export-ignore

README.md

@@ -58,7 +58,7 @@ foreach( $contents as $key => $html ){
 $factory->opf->setLang( 'en_US' );
 $factory->opf->setTitle( 'My First eBook', 'main-title' );
 $factory->opf->putXML();
-$factory->container->pubXML();
+$factory->container->putXML();
 // Save it!
 $factory->compile('path/to/epub');
 ```
@@ -71,6 +71,10 @@ $factory->compile('path/to/epub');
 
 This library is under alpha and highly experimental. Do not trust this until Beta! 
 
+## Acknowledgement
+
+The sample picture is credited by [Public Domain Pictures](https://www.pexels.com/ja-jp/photo/87742/) and [Nadi Lindsay](https://www.pexels.com/ja-jp/photo/3078831/).
+
 ## License
 
 As wrote in LICENSE file, this library is released under [MIT](https://opensource.org/licenses/MIT).

bin/hamepub

@@ -29,6 +29,15 @@ $app->registerCommand('generate', function( CommandCall $input ) use ($app, $roo
 		// This is not file.
 		$app->getPrinter()->error('Setting file not found.', false);
 	}
+    // Generate temp directory.
+    $tmp_dir = $input->hasParam( 'tmp' ) ? $input->getParam( 'tmp' ) : tempnam( sys_get_temp_dir(), 'hamepub-' );
+    try {
+        $packager = \Hametuha\HamePub\Packager::get();
+        $path = $packager->parse( $file, $tmp_dir );
+        $app->getPrinter()->success("Generated ePub: {$path}", false);
+    } catch ( \Exception $e ) {
+        $app->getPrinter()->error($e->getMessage(), false);
+    }
 
 });
 

setting.json

@@ -0,0 +1,28 @@
+{
+    "root": "./tests/dist/",
+    "id": "my-sample-ebook",
+    "isbn": "1234567890123",
+    "title": "My Sample Book",
+    "author": [
+        {
+            "name": "Kazuo Ishiguro",
+            "id": "author-1",
+            "role": "aut"
+        },
+        {
+            "name": "Haruki Murakami",
+            "id": "author-2",
+            "role": "aut"
+        },
+        {
+            "name": "Mike Jacob",
+            "id": "translator-1",
+            "role": "trl",
+            "type": "contributor"
+        }
+    ],
+    "target": "./tests/tmp",
+    "published": "2023-01-01T23:00:00Z",
+    "direction": "ltr",
+    "cover": "./tests/dist/img/cover.jpg"
+}

src/Hametuha/HamePub/Oebps/Content.php

@@ -70,14 +70,39 @@ public function setTitle($string, $id, $type = 'main', $sequence = 1)
         $meta['property'] = 'display-seq';
     }
 
+    /**
+     * Add author to as meta value.
+     *
+     * @param string $value Name of author.
+     * @param string $id    ID of author.
+     * @param string $tag   Default is 'creator' or 'contributor'
+     * @param string $role  Role tag. See {https://www.loc.gov/marc/relators/relaterm.html}
+     *
+     * @return void
+     */
+    public function addAuthor( $value, $id, $tag = 'creator', $role = '' ) {
+        $creator = $this->dom->metadata->addChild( $tag, $this->h( $value ), Schemas::DC );
+        $creator['id'] = $id;
+        if ( $role ) {
+            $meta = $this->dom->metadata->addChild( 'meta', $role );
+            $meta['refines'] = '#' . $id;
+            $meta['property'] = 'role';
+            $meta['scheme'] = 'marc:relators';
+            $meta['id'] = 'role-of-' . $id;
+        }
+    }
+
     /**
      * Add modified date
      *
-     * @param int $timestamp UTC timestamp
+     * @param int|string $timestamp If int, treated as UTC timestamp.
      */
     public function setModifiedDate($timestamp)
     {
-        $this->addMeta('meta', date('Y-m-d\TH:i:s\Z', $timestamp), [
+        if ( is_int( $timestamp ) ) {
+            $timestamp = date('Y-m-d\TH:i:s\Z', $timestamp);
+        }
+        $this->addMeta('meta', $timestamp, [
             'property' => 'dcterms:modified',
         ]);
     }
@@ -104,7 +129,7 @@ public function addMeta($tag, $value, array $attributes = [])
     }
 
     /**
-     * Add item to
+     * Add item to Content OPF.
      *
      * @param string $relative_path
      * @param string $id If empty, path will convert to id

src/Hametuha/HamePub/Oebps/Toc.php

@@ -85,7 +85,7 @@ public function getHTML($header = '', $footer = '')
 </head>
 <body>
 HTML;
-        $html .= $this->getNavHTML();
+        $html .= $this->getNavHTML( $title );
         $html .= <<<HTML
 {$footer}
 </body>

src/Hametuha/HamePub/Packager.php

@@ -2,11 +2,154 @@
 
 namespace Hametuha\HamePub;
 
+use Hametuha\HamePub\Parser\SettingParser;
 use Hametuha\HamePub\Pattern\Singleton;
+use PHPUnit\Framework\Error\Error;
 
 /**
  * Package ePub from directory.
  */
-class Packager extends Singleton
-{
+class Packager extends Singleton {
+
+    use SettingParser;
+
+    /**
+     * @var array Setting.
+     */
+    protected $setting = [];
+
+    /**
+     * @var string[] HTML strings.
+     */
+    protected $htmls = [];
+
+    /**
+     * @var string Newest modified HTML.
+     */
+    protected $newest = 0;
+
+    /**
+     * Parse and save setting.
+     *
+     * @param string $file Path to setting file.
+     * @param string $tmp_dir Temporary directory.
+     *
+     * @return string ePub file path.
+     * @throws \Exception
+     */
+    public function parse( $file = 'setting.json', $tmp_dir = '' ) {
+        $this->setting = $this->getSettingFromFile( $file );
+        // Set temporary directory.
+        if ( empty( $tmp_dir ) ) {
+            $tmp_dir = tempnam( sys_get_temp_dir(), 'hamepub-' );
+        }
+        // Load HTMLs.
+        $this->loadHtml();
+        if ( empty( $this->htmls ) ) {
+            throw new \Exception( 'No HTML files found.' );
+        }
+        // Start parsing.
+        $factory = Factory::init( $this->setting[ 'id' ], $tmp_dir );
+        // Set metadata.
+        $factory->opf->setLang( $this->setting[ 'lang' ] );
+        $factory->opf->setTitle( $this->setting[ 'title' ], 'main-title' );
+        $factory->opf->setModifiedDate( $this->setting['published'] );
+        $factory->opf->direction = $this->setting['direction'];
+        // Set authors.
+        if ( is_array( $this->setting['author'] ) ) {
+            foreach ( $this->setting['author'] as $index => $author ) {
+                $factory->opf->addAuthor(
+                    $author['name'],
+                    ( $author['id'] ?? sprintf( 'creator-%d', $index + 1 ) ),
+                    ( ( isset( $author['type'] ) && $author['type'] === 'contributor' ) ? 'contributor' : 'creator'),
+                    ( $author['role'] ?? '' )
+                );
+            }
+        } else {
+            $factory->opf->addAuthor( $this->setting['author'], 'creator' );
+        }
+        foreach ( $this->htmls as $key => $html ) {
+            // Register toc
+            $toc = $factory->toc->addChild( $key, $key . '.xhtml' );
+            // Grab all headers and add them to toc.
+            $dom = $factory->parser->html5->loadHTML( $html );
+            // Grab header and add ID attributes.
+            $factory->parser->grabHeaders( $toc, $dom, true, $this->setting[ 'header' ][ 'max_level' ], $this->setting[ 'header' ][ 'max_level' ] );
+            // Convert from dom object to string.
+            $html = $factory->parser->convertToString( $dom );
+            // Recreate DOM.
+            $dom = $factory->registerHTML( $key, $html, $this->getLinear( $key ) );
+            // Grab all images
+            foreach ( $factory->parser->extractAssets( $dom, 'img', 'src', $this->setting[ 'url_base' ], $this->setting[ 'root' ] ) as $path ) {
+                $factory->opf->addItem( $path, '' );
+            }
+            // Grab all CSS
+            foreach ( $factory->parser->extractAssets( $dom, 'link', 'href', $this->setting[ 'url_base' ], $this->setting[ 'root' ] ) as $path ) {
+                $factory->opf->addItem( $path, '' );
+            }
+            // Register to OPF
+            $factory->opf->addItem( "Text/{$key}.xhtml", "{$key}.xhtml" );
+            // Save HTML
+            $factory->parser->saveDom( $dom, "{$key}.xhtml" );
+        }
+        // If TOC is set, save it.
+        if ( !empty($this->setting['toc']) ) {
+            $factory->toc->label = $this->setting['toc'];
+            $toc_html = $factory->toc->getHTML();
+            $factory->opf->addItem( 'Text/toc.xhtml', 'toc.xhtml', ['nav'] );
+            $factory->parser->saveDom( $factory->registerHTML( 'toc', $toc_html, 'no' ), 'toc.xhtml' );
+        }
+        // Set OPF.
+        if ( ! empty( $this->setting[ 'isbn' ] ) ) {
+            $factory->opf->setIdentifier( $this->setting[ 'isbn' ] );
+        }
+        // If cover is set, add it.
+        if ( $this->setting['cover'] ) {
+            $factory->addCover( $this->setting['cover'] );
+        }
+        $factory->opf->putXML();
+        $factory->container->putXML();
+        // Save it!
+        $target = tempnam(  $this->setting['target'], $this->setting['id'] . '-' ) . '.epub';
+        if ( ! is_writable( dirname( $target ) ) ) {
+            throw new \Exception( 'Target directory is not writable: ' . $target );
+        }
+        $factory->compile( $target );
+        return $target;
+    }
+
+    /**
+     * Get linear property.
+     *
+     * @param string $key HTML name.
+     * @return string
+     */
+    protected function getLinear( $key ) {
+        return in_array( $key, $this->setting['hidden'], true ) ? 'no' : 'yes';
+    }
+
+    /**
+     * Load HTML and save it in $htmls.
+     * @return void
+     */
+    public function loadHtml() {
+        $files = glob( $this->setting[ 'root' ] . '/*.html' );
+        foreach ( $files as $file ) {
+            $this->htmls[ basename( $file, '.html' ) ] = file_get_contents( $file );
+            // Save modified time.
+            $time = filemtime( $file );
+            if ( $time > $this->newest ) {
+                $this->newest = $time;
+            }
+        }
+    }
+
+    /**
+     * Dump setting for debugging.
+     *
+     * @return void
+     */
+    public function dumpSetting() {
+        var_dump( $this->setting, $this->htmls );
+    }
 }

src/Hametuha/HamePub/Parser/SettingParser.php

@@ -0,0 +1,67 @@
+<?php
+
+namespace Hametuha\HamePub\Parser;
+
+/**
+ * Parse setting JSON file and ensure it has all required keys.
+ */
+trait SettingParser {
+
+    /**
+     * Get setting from file.
+     *
+     * @param string $file_path Path to setting file.
+     * @return array
+     * @throws \Exception
+     */
+    public function getSettingFromFile( $file_path = './setting.json' ) {
+        if ( ! file_exists( $file_path ) ) {
+            throw new \Exception( 'Setting file not found.' );
+        }
+        $setting = json_decode( file_get_contents( $file_path ), true );
+        if ( ! is_array( $setting ) ) {
+            throw new \Exception( 'Setting file is not valid JSON.' );
+        }
+        $setting = array_replace_recursive( $this->defaultSetting(), $setting );
+        // Validate if isbn is not set.
+        if ( empty( $setting['id']) ) {
+            throw new \Exception( 'id fields must not be empty: ' . $setting['id'] );
+        }
+        // Validate if published is not set.
+        if ( ! preg_match( '/^\d{4}-\d{2}-\d{2}T\d{2}:\d{2}:\d{2}Z$/u', $setting['published' ] ) ) {
+            throw new \Exception( 'published field is malformed. Should be GMT in ISO8601 e.g. "2000-01-01T00:00:00Z": ' . $setting['published'] );
+        }
+        // Validate if author is set.
+        if ( empty( $setting['author'] ) ) {
+            throw new \Exception( 'At least 1 author should be set.' );
+        }
+        return $setting;
+    }
+
+    /**
+     * Default setting.
+     *
+     * @return array
+     */
+    public function defaultSetting() {
+        return [
+            'lang'      => 'en',
+            'id'        => '',
+            'isbn'      => '',
+            'title'     => '',
+            'author'    => '',
+            'published' => '',
+            'root'      => './dist/',
+            'header'    => [
+                'max_level' => 3,
+                'depth'     => 2,
+            ],
+            'toc'       => 'Table of Contents',
+            'url_base'  => '#\./#u',
+            'target'    => './tmp',
+            'direction' => 'default',
+            'hidden'    => [ 'toc' ],
+            'cover'     => '',
+        ];
+    }
+}