Try to detect csv file content separator automatically

mantas-done · Jul 10, 2023 · c881c42 · c881c42
1 parent f9b15a5
commit c881c42
Show file tree

Hide file tree

Showing 2 changed files with 57 additions and 1 deletion.
diff --git a/src/Code/Converters/CsvConverter.php b/src/Code/Converters/CsvConverter.php
@@ -4,6 +4,8 @@
 
 class CsvConverter implements ConverterContract
 {
+    public static $allowedSeparators = [",", ";", "|", "\t"];
+
     public function canParseFileContent($file_content)
     {
         $csv = self::csvToArray($file_content);
@@ -78,12 +80,39 @@ private static function csvToArray($content)
         fputs($fp, $content);
         rewind($fp);
 
+        $separator = self::detectSeparator($content);
         $csv = [];
-        while ( ($data = fgetcsv($fp) ) !== false ) {
+        while ( ($data = fgetcsv($fp, 0, $separator) ) !== false ) {
             $csv[] = $data;
         }
         fclose($fp);
 
         return $csv;
     }
+
+    private static function detectSeparator($file_content)
+    {
+        $lines = explode("\n", $file_content);
+        $results = [];
+        foreach ($lines as $line) {
+            foreach (self::$allowedSeparators as $delimiter) {
+                $count = count(explode($delimiter, $line));
+                if ($count === 1) continue; // delimiter not found in line
+
+                if (empty($results[$delimiter])) {
+                    $results[$delimiter] = [];
+                }
+                $results[$delimiter][] = $count;
+            }
+        }
+
+        foreach ($results as $delimiter => $value) {
+            $flipped = array_flip($value);
+            $results[$delimiter] = count($flipped);
+        }
+
+        arsort($results, SORT_NUMERIC);
+
+        return !empty($results) ? key($results) : self::$allowedSeparators[0];
+    }
 }
diff --git a/tests/formats/CsvTest.php b/tests/formats/CsvTest.php
@@ -57,4 +57,31 @@ public function testClientAnsiFile()
             ->getInternalFormat();
         $this->assertInternalFormatsEqual($expected, $actual);
     }
+
+    /**
+     * @dataProvider differentContentSeparatorProvider
+     */
+    public function testDifferentContentSeparators($string)
+    {
+        $actual_internal_format = Subtitles::loadFromString($string, 'csv')->getInternalFormat();
+        $expected_internal_format = (new Subtitles())
+            ->add(1, 2, ['Oh! Can I believe my eyes!'])
+            ->add(2, 3, ['If Heaven and earth.'])->getInternalFormat();
+
+        $this->assertInternalFormatsEqual($expected_internal_format, $actual_internal_format);
+    }
+
+    public static function differentContentSeparatorProvider()
+    {
+        $original_string = 'Start,End,Text
+00:00:1,00:00:2,Oh! Can I believe my eyes!
+00:00:2,00:00:3,If Heaven and earth.';
+
+        $strings = [];
+        foreach (CsvConverter::$allowedSeparators as $separator) {
+            $strings[] = str_replace(',', $separator, $original_string);
+        }
+
+        return [$strings];
+    }
 }