Add support for multiline text after the timestamp

mantas-done · Aug 1, 2023 · c994b2f · c994b2f
1 parent f61fbc6
commit c994b2f
Show file tree

Hide file tree

Showing 6 changed files with 48 additions and 16 deletions.
diff --git a/src/Code/Converters/AssConverter.php b/src/Code/Converters/AssConverter.php
@@ -6,7 +6,7 @@ class AssConverter implements ConverterContract
 {
     public function canParseFileContent($file_content)
     {
-        return preg_match('/^\[Script Info\]\R/m', $file_content) === 1;
+        return preg_match('/\[Script Info\]\R/m', $file_content) === 1;
     }
 
     public function fileContentToInternalFormat($file_content)

diff --git a/src/Code/Converters/CsvConverter.php b/src/Code/Converters/CsvConverter.php
@@ -41,14 +41,25 @@ public function fileContentToInternalFormat($file_content)
         $data = self::csvToArray($file_content);
         $data_string  = '';
 
+        $is_start_time = (bool) preg_match(TxtConverter::$time_regexp, $data[1][0]);
+        $is_end_time = (bool) preg_match(TxtConverter::$time_regexp, $data[1][1]);
+
         foreach ($data as $k => $row) {
             $timestamp_found = (bool) preg_match(TxtConverter::$time_regexp, $row[0]);
             if ($k === 0  && $timestamp_found === false) { // heading
                 continue;
             }
 
-            $row_string = implode(' ', $row);
-            $data_string .= $row_string . "\n";
+            // format csv file as a txt file, so TxtConverter would be able to understand it
+            if ($is_start_time && $is_end_time) {
+                $data_string .= $row[0] . ' ' . $row[1] . "\n"; // start end
+                $data_string .= $row[2] . "\n"; // text
+            } elseif ($is_start_time) {
+                $data_string .= $row[0] . "\n"; // start
+                $data_string .= $row[1] . "\n"; // text
+            } else {
+                $data_string .= $row[0] . "\n"; // text
+            }
         }
         return (new TxtConverter)->fileContentToInternalFormat($data_string);
     }

diff --git a/src/Code/Converters/TxtConverter.php b/src/Code/Converters/TxtConverter.php
@@ -88,12 +88,14 @@ public function fileContentToInternalFormat($file_content)
         $internal_format = [];
         $j = 0;
         foreach ($data as $k => $row) {
-            if (
-                isset($data[$k - 1]['start'])
-                && ($data[$k - 1]['start'] === $row['start'] || $row['start'] === null)
-            ) {
-                $internal_format[$j - 1]['lines'][] = $row['text'];
-                continue;
+            for ($i = 1; $i <= 10; $i++) { // up to 10 lines
+                if (
+                    isset($data[$k - $i]['start'])
+                    && ($data[$k - $i]['start'] === $row['start'] || $row['start'] === null)
+                ) {
+                    $internal_format[$j - $i]['lines'][] = $row['text'];
+                    continue 2;
+                }
             }
 
             $internal_format[$j] =  [

diff --git a/src/Subtitles.php b/src/Subtitles.php
@@ -229,7 +229,7 @@ public static function loadFromString($string, $format = null)
                 continue;
             }
             $diff = $internal_format[$k - 1]['end'] - $row['start'];
-            if ($diff < 1 && $diff > 0) {
+            if ($diff < 10 && $diff > 0) {
                 $internal_format[$k - 1]['end'] = $row['start'];
             }
         }
@@ -239,11 +239,11 @@ public static function loadFromString($string, $format = null)
         $last_end_time = 0;
         foreach ($internal_format as $row) {
             if ($row['start'] < $last_end_time) {
-                throw new UserException('Times are overlapping near text: ' . SrtConverter::internalTimeToSrt($row['start']) . ' ' . $row['lines'][0]);
+                throw new UserException('Times are overlapping over 10 seconds near the text: ' . SrtConverter::internalTimeToSrt($row['start']) . ' ' . $row['lines'][0]);
             }
             $last_end_time = $row['end'];
             if ($row['start'] > $row['end']) {
-                throw new UserException('Times are overlapping near text: ' . SrtConverter::internalTimeToSrt($row['start']) . ' ' . $row['lines'][0]);
+                throw new UserException('Start time is bigger than the end time near text: ' . SrtConverter::internalTimeToSrt($row['start']) . ' ' . $row['lines'][0]);
             }
         }
 

diff --git a/tests/formats/CsvTest.php b/tests/formats/CsvTest.php
@@ -98,4 +98,19 @@ public function testParseFileWithSingleTimestamp()
 
         $this->assertInternalFormatsEqual($expected_internal_format, $actual_internal_format);
     }
+
+    public function testAdditionalColumns()
+    {
+        $string = <<< TEXT
+Start Time,End Time,Text,Layer ID
+00:00:08:00,00:00:13:00,"abc",1
+00:00:20:00,00:00:24:00,def,1
+TEXT;
+        $actual_internal_format = Subtitles::loadFromString($string)->getInternalFormat();
+        $expected_internal_format = (new Subtitles())
+            ->add(8, 13, ['abc'])
+            ->add(20, 24, ['def'])->getInternalFormat();
+
+        $this->assertInternalFormatsEqual($expected_internal_format, $actual_internal_format);
+    }
 }
diff --git a/tests/formats/TxtTest.php b/tests/formats/TxtTest.php
@@ -265,12 +265,16 @@ public function testFirstLineWithTimestampSecondAfter()
         $actual = Subtitles::loadFromString('
 01:01 a
 b
-01:02 c
+c
+d
+01:02 e
+f
+01:03 g
         ')->getInternalFormat();
-//        var_dump($actual); exit;
         $expected = (new Subtitles())
-            ->add(61, 62, ['a', 'b'])
-            ->add(62, 63, 'c')
+            ->add(61, 62, ['a', 'b', 'c', 'd'])
+            ->add(62, 63, ['e', 'f'])
+            ->add(63, 64, 'g')
             ->getInternalFormat();
         $this->assertInternalFormatsEqual($expected, $actual);
     }