Add new ttml format

mantas-done · Aug 31, 2023 · 03dd6b7 · 03dd6b7
1 parent b2a4744
commit 03dd6b7
Show file tree

Hide file tree

Showing 2 changed files with 40 additions and 3 deletions.
diff --git a/src/Code/Converters/TtmlConverter.php b/src/Code/Converters/TtmlConverter.php
@@ -58,13 +58,28 @@ public function fileContentToInternalFormat($file_content)
             $div_begin = $element->getAttribute('begin');
             $div_end = $element->getAttribute('end');
             foreach ($element->getElementsByTagName('p') as $pElement) {
-                $begin = $pElement->hasAttribute('begin') ? $pElement->getAttribute('begin') : $div_begin;
+                $begin = null;
+                if ($pElement->hasAttribute('begin')) {
+                    $begin = $pElement->getAttribute('begin');
+                } elseif ($pElement->getAttribute('t')) {
+                    $begin = $pElement->getAttribute('t');
+                } elseif ($div_begin) {
+                    $begin = $div_begin;
+                }
                 $begin = static::ttmlTimeToInternal($begin, $fps);
-                $end = $pElement->hasAttribute('end') ? $pElement->getAttribute('end') : $div_end;
+
+                $end = null;
+                if ($pElement->hasAttribute('end')) {
+                    $end = $pElement->getAttribute('end');
+                } elseif ($div_end) {
+                    $end = $div_end;
+                }
                 if ($end) {
                     $end = static::ttmlTimeToInternal($end, $fps);
                 } elseif ($pElement->hasAttribute('dur') && $pElement->getAttribute('dur')) {
                     $end = $begin + static::ttmlTimeToInternal($pElement->getAttribute('dur'), $fps);
+                } elseif ($pElement->hasAttribute('d') && $pElement->getAttribute('d')) {
+                    $end = $begin + static::ttmlTimeToInternal($pElement->getAttribute('d'), $fps);
                 }
                 $lines = '';
 
@@ -169,6 +184,8 @@ public static function ttmlTimeToInternal($ttml_time, $frame_rate)
             $totalSeconds = ($hours * 3600) + ($minutes * 60) + $seconds + $frames / $frame_rate;
 
             return $totalSeconds;
+        } elseif (is_numeric($ttml_time)) {
+            return $ttml_time / 1000;
         } else {
             $time_parts = explode('.', $ttml_time);
             $milliseconds = 0;
@@ -302,9 +319,12 @@ private static function subtitleXml2(string $file_content)
 
     private static function getLinesFromTextWithBr(string $text)
     {
+
         $text = preg_replace('/<br\s*\/?>/', '<br>', $text); // normalize <br>*/
         $lines = preg_replace('/<tt:br*\/?>/', '<br>', $text); // normalize <br>*/
-        $lines = explode('<br>', $lines);
+        $lines = str_replace('<br>', "\n", $lines);
+        $lines = preg_replace('/[\x{200B}-\x{200D}\x{FEFF}]/u', '', $lines); // remove zero width space characters
+        $lines = explode("\n", $lines);
         $lines = array_map('strip_tags', $lines);
         $lines = array_map('trim', $lines);
 

diff --git a/tests/formats/TtmlTest.php b/tests/formats/TtmlTest.php
@@ -124,6 +124,7 @@ public static function timeFormatProvider()
             ['00:00:10', 10, null],
             ['00:00:5.100', 5.1, null],
             ['55s', 55, null],
+            ['8500', 8.5, null],
         ];
     }
 
@@ -442,4 +443,20 @@ public function testConvertFromXml8()
             ->getInternalFormat();
         $this->assertInternalFormatsEqual($expected, $actual);
     }
+
+    public function testConvertFromXml9()
+    {
+        $text = <<<X
+<?xml version="1.0" encoding="utf-8" ?>
+<body>
+<p t="9159" d="768" wp="2" ws="1"><s p="2"></s><s></s><s p="3"> チャン チャン </s><s p="2">
+</s><s p="4"> 쟌 쟌 </s><s p="2"></s></p>
+</body>
+X;
+        $actual = Subtitles::loadFromString($text)->getInternalFormat();
+        $expected = (new Subtitles())
+            ->add(9.159, 9.927, ['チャン チャン', '쟌 쟌'])
+            ->getInternalFormat();
+        $this->assertInternalFormatsEqual($expected, $actual);
+    }
 }