Simplify bad file detection

mantas-done · Apr 18, 2024 · d4062b9 · d4062b9
1 parent 494cafb
commit d4062b9
Show file tree

Hide file tree

Showing 3 changed files with 1 addition and 30 deletions.
diff --git a/src/Code/Converters/BinaryFinder.php b/src/Code/Converters/BinaryFinder.php
diff --git a/src/Code/Converters/TxtConverter.php b/src/Code/Converters/TxtConverter.php
@@ -12,7 +12,7 @@ class TxtConverter implements ConverterContract
 
     public function canParseFileContent($file_content)
     {
-        return self::hasText($file_content);
+        return self::hasText($file_content) && !Helpers::strContains($file_content, "\x00"); // not a binary file
     }
 
     public function fileContentToInternalFormat($file_content, $original_file_content)

diff --git a/src/Subtitles.php b/src/Subtitles.php
@@ -53,7 +53,6 @@ class Subtitles
         ['extension' => 'docx', 'format' => 'docx',             'name' => 'DOCX',                       'class' => DocxReader::class],
         ['extension' => 'rtf',  'format' => 'rtf',              'name' => 'Rich text format',           'class' => RtfReader::class], // libraryies eather throws exception, not parses, or takes long to parse 2h file
         ['extension' => 'csv',  'format' => 'csv',              'name' => 'Coma Separated Values',      'class' => CsvConverter::class], // must be last from bottom
-        ['extension' => 'bin',  'format' => 'bin',              'name' => 'Binary',                     'class' => BinaryFinder::class],
         ['extension' => 'txt',  'format' => 'txt',              'name' => 'Plaintext',                  'class' => TxtConverter::class], // must be the last one
     ];