Added an option to choose epub v3 with html5.

wallabag · Oct 21, 2018 · 78546d9 · 78546d9
1 parent 9a7c8de
commit 78546d9
Show file tree

Hide file tree

Showing 4 changed files with 70 additions and 11 deletions.
diff --git a/composer.json b/composer.json
@@ -30,7 +30,7 @@
         "phpzip/phpzip": "~2.0.7",
         "grandt/phpresizegif": "~1.0.3",
         "grandt/relativepath": "~1.0.1",
-        "masterminds/html5": "^2.1"
+        "masterminds/html5": "~2.3"
     },
     "autoload": {
         "psr-4": {

diff --git a/src/PHPePub/Core/EPub.php b/src/PHPePub/Core/EPub.php
@@ -59,6 +59,9 @@ class EPub {
     const BOOK_VERSION_EPUB2 = '2.0';
     const BOOK_VERSION_EPUB3 = '3.0';
 
+    const FORMAT_XHTML = 'xhtml';
+    const FORMAT_HTML5 = 'html5';
+
     public $viewportMap = array(
         "small" => array('width' => 600, 'height' => 800),
         "medium" => array('width' => 720, 'height' => 1280),
@@ -127,6 +130,7 @@ class EPub {
     private $tocCssFileName = null;
     private $fileList = array();
     private $writingDirection = EPub::DIRECTION_LEFT_TO_RIGHT;
+    private $htmlFormat = EPub::FORMAT_XHTML;
     private $languageCode = 'en';
     private $dateformat = 'Y-m-d\TH:i:s.000000P';
     private $dateformatShort = 'Y-m-d';
@@ -150,11 +154,18 @@ class EPub {
      * @param string $bookVersion
      * @param string $languageCode
      * @param string $writingDirection
-     */
-    function __construct($bookVersion = EPub::BOOK_VERSION_EPUB2, $languageCode = 'en', $writingDirection = EPub::DIRECTION_LEFT_TO_RIGHT) {
+     * @param string $htmlFormat
+     */
+    function __construct(
+        $bookVersion = EPub::BOOK_VERSION_EPUB2,
+        $languageCode = 'en',
+        $writingDirection = EPub::DIRECTION_LEFT_TO_RIGHT,
+        $htmlFormat = EPub::FORMAT_XHTML
+    ) {
         $this->bookVersion = $bookVersion;
         $this->writingDirection = $writingDirection;
         $this->languageCode = $languageCode;
+        $this->htmlFormat = $htmlFormat;
 
         $this->log = new Logger('EPub', $this->isLogging);
 
@@ -247,7 +258,7 @@ function addChapter($chapterName, $fileName, $chapterData = null, $autoSplit = f
 
         $chapter = $chapterData;
         if ($autoSplit && is_string($chapterData) && mb_strlen($chapterData) > $this->splitDefaultSize) {
-            $splitter = new EPubChapterSplitter();
+            $splitter = new EPubChapterSplitter($this->htmlFormat);
             $splitter->setSplitSize($this->splitDefaultSize);
 
             $chapterArray = $splitter->splitChapter($chapterData);
@@ -355,8 +366,17 @@ function addChapter($chapterName, $fileName, $chapterData = null, $autoSplit = f
      * @return array
      */
     function findIdAttributes($chapterData) {
-        $html5 = new HTML5();
-        $xmlDoc = $html5->loadHTML($chapterData);
+        switch ($this->htmlFormat) {
+            case EPub::FORMAT_HTML5;
+                $html5 = new HTML5();
+                $xmlDoc = $html5->loadHTML($chapterData);
+                break;
+            case EPub::FORMAT_XHTML:
+            default:
+                $xmlDoc = new DOMDocument();
+                @$xmlDoc->loadHTML($chapterData);
+                break;
+        }
 
         $xpath = new DomXpath($xmlDoc);
 
@@ -417,8 +437,17 @@ protected function processChapterExternalReferences(&$doc, $externalReferences =
         if ($isDocAString) {
             $doc = StringHelper::removeComments($doc);
 
-            $html5 = new HTML5();
-            $xmlDoc = $html5->loadHTML($doc);
+            switch ($this->htmlFormat) {
+                case EPub::FORMAT_HTML5;
+                    $html5 = new HTML5();
+                    $xmlDoc = $html5->loadHTML($doc);
+                    break;
+                case EPub::FORMAT_XHTML:
+                default:
+                    $xmlDoc = new DOMDocument();
+                    @$xmlDoc->loadHTML($doc);
+                    break;
+            }
         } else {
             $xmlDoc = $doc;
         }

diff --git a/src/PHPePub/Core/EPubChapterSplitter.php b/src/PHPePub/Core/EPubChapterSplitter.php
@@ -18,6 +18,16 @@
 class EPubChapterSplitter {
     private $splitDefaultSize = 250000;
     private $bookVersion = EPub::BOOK_VERSION_EPUB2;
+    private $htmlFormat = EPub::FORMAT_XHTML;
+
+    /**
+     * Class constructor.
+     *
+     * @param string $htmlFormat
+     */
+    function __construct($htmlFormat = EPub::FORMAT_XHTML) {
+        $this->setHtmlFormat($htmlFormat);
+    }
 
     /**
      *
@@ -29,6 +39,17 @@ function setVersion($bookVersion) {
         $this->bookVersion = is_string($bookVersion) ? trim($bookVersion) : EPub::BOOK_VERSION_EPUB2;
     }
 
+    /**
+     * Set the html format of the epub.
+     *
+     * @param string $htmlFormat
+     */
+    public function setHtmlFormat($htmlFormat) {
+        $this->htmlFormat = in_array($htmlFormat, [EPub::FORMAT_XHTML, EPub::FORMAT_HTML5])
+            ? $htmlFormat
+            : EPub::FORMAT_XHTML;
+    }
+
     /**
      * Set default chapter target size.
      * Default is 250000 bytes, and minimum is 10240 bytes.
@@ -78,8 +99,17 @@ function splitChapter($chapter, $splitOnSearchString = false, $searchString = '/
             );
         }
 
-        $html5 = new HTML5();
-        $xmlDoc = $html5->loadHTML($chapter);
+        switch ($this->htmlFormat) {
+            case EPub::FORMAT_HTML5;
+                $html5 = new HTML5();
+                $xmlDoc = $html5->loadHTML($chapter);
+                break;
+            case EPub::FORMAT_XHTML:
+            default:
+                $xmlDoc = new DOMDocument();
+                @$xmlDoc->loadHTML($chapter);
+                break;
+        }
 
         $head = $xmlDoc->getElementsByTagName("head");
         $body = $xmlDoc->getElementsByTagName("body");

diff --git a/tests/composer.json b/tests/composer.json
@@ -2,7 +2,7 @@
   "require": {
         "php": ">=5.3.0",
         "phpzip/phpzip": ">=2.0.7",
-        "masterminds/html5": "^2.1"
+        "masterminds/html5": "~2.3"
     },
     "autoload": {
         "psr-4": {