Fix txt repeating text removal

mantas-done · Apr 16, 2024 · 494cafb · 494cafb
1 parent fbd3858
commit 494cafb
Show file tree

Hide file tree

Showing 2 changed files with 57 additions and 1 deletion.
diff --git a/src/Code/Converters/TxtConverter.php b/src/Code/Converters/TxtConverter.php
@@ -130,7 +130,10 @@ public function fileContentToInternalFormat($file_content, $original_file_conten
         }
         unset($row);
 
-        return self::fillStartAndEndTimes($internal_format);
+        $internal_format = self::fillStartAndEndTimes($internal_format);
+        $internal_format = self::removeRepeatingTextStarts($internal_format);
+
+        return $internal_format;
     }
 
     // start and end timestamp
@@ -518,6 +521,44 @@ private static function twoLinesSeparatedByEmptyLine(string $file_content)
         return self::fillStartAndEndTimes($internal_format);
     }
 
+    public static function removeRepeatingTextStarts($internal_format)
+    {
+        if (count($internal_format) <= 2) {
+            return $internal_format; // don't try to filter if there almost no lines
+        }
+
+        $repeating_string = '';
+
+        $first_lines = [];
+        foreach ($internal_format as $subtitle) {
+            $first_lines[] = $subtitle['lines'][0];
+        }
+
+        $length = strlen($first_lines[0]);
+        for ($i = 0; $i < $length; $i++) {
+            $letter = $first_lines[0][$i];
+
+            foreach ($first_lines as $line) {
+                if (!isset($line[$i])) {
+                    break 2;
+                }
+                $line_letter = $line[$i];
+                if ($line_letter !== $letter) {
+                    break 2;
+                }
+            }
+            $repeating_string .= $letter;
+        }
+
+        $repeating_length = strlen($repeating_string);
+        foreach ($internal_format as &$subtitle) {
+            $subtitle['lines'][0] = substr($subtitle['lines'][0], $repeating_length);
+        }
+        unset($subtitle);
+
+        return $internal_format;
+    }
+
     private static function hasTime($line)
     {
         return preg_match(self::$time_regexp, $line) === 1;

diff --git a/tests/formats/TxtTest.php b/tests/formats/TxtTest.php
@@ -469,6 +469,21 @@ public function testDoesNotRemoveNotHtmlTag()
         $this->assertInternalFormatsEqual($expected, $actual);
     }
 
+    public function testRemoveRepeatingTextFromBeginningOfText()
+    {
+        $actual = Subtitles::loadFromString('
+00:00:00:a
+00:00:01:b
+00:00:02:c
+        ')->getInternalFormat();
+        $expected = (new Subtitles())
+            ->add(0, 1, 'a')
+            ->add(1, 2, 'b')
+            ->add(2, 3, 'c')
+            ->getInternalFormat();
+        $this->assertInternalFormatsEqual($expected, $actual);
+    }
+
     // ---------------------------------- private ----------------------------------------------------------------------
 
     private static function generatedSubtitles()