implementing detectDelimiter #16

thephpleague · Mar 5, 2014 · 3bcd75c · 3bcd75c
1 parent ef58410
commit 3bcd75c
Show file tree

Hide file tree

Showing 2 changed files with 55 additions and 0 deletions.
diff --git a/src/AbstractCsv.php b/src/AbstractCsv.php
@@ -39,6 +39,8 @@
 use SplFileObject;
 use SplTempFileObject;
 use InvalidArgumentException;
+use LimitIterator;
+use CallbackFilterIterator;
 
 /**
  *  A abstract class to enable basic CSV manipulation
@@ -209,6 +211,40 @@ public function getDelimiter()
         return $this->delimiter;
     }
 
+    /**
+     * Detect the CSV file delimiter
+     *
+     * @param integer $nbRows
+     * @param array   $additionals additional delimiters
+     *
+     * @return string
+     */
+    public function detectDelimiter($nbRows = 1, array $additionals = [])
+    {
+        $nbRows = filter_var($nbRows, FILTER_VALIDATE_INT, ['options' => ['min_range' => 1]]);
+        if (! $nbRows) {
+            throw new InvalidArgumentException('`$nbRows` must be a valid positive integer');
+        }
+        $additionals = array_filter($additionals, function ($str) {
+            return 1 == mb_strlen($str);
+        });
+        $delimiters = [',', ';', "\t"];
+        $delim = array_unique(array_merge($delimiters, $additionals));
+        $iterator =  new CallbackFilterIterator(new LimitIterator($this->csv, 0, $nbRows), function ($row) {
+            return is_array($row) && count($row) > 1;
+        });
+        $origDelimiter = $this->getDelimiter();
+        $res = array_fill_keys($delimiters, null);
+        foreach ($delimiters as $delim) {
+            $iterator->setCsvControl($delim);
+            $res[$delim] = count(iterator_to_array($iterator, false));
+        }
+        $this->setDelimiter($origDelimiter);
+        arsort($res, SORT_NUMERIC);
+
+        return array_keys(array_filter($res));
+    }
+
     /**
      * set the field enclosure
      *

diff --git a/test/CsvTest.php b/test/CsvTest.php
@@ -84,6 +84,25 @@ public function testDelimeter()
         $this->csv->setDelimiter('foo');
     }
 
+    /**
+     * @expectedException InvalidArgumentException
+     */
+    public function testDetectDelimiter()
+    {
+        $this->assertSame($this->csv->detectDelimiter(), [',']);
+
+        $csv = new Writer(new SplTempFileObject);
+        $csv->setDelimiter(';');
+        $csv->insertOne(['toto', 'tata', 'tutu']);
+        $csv->setDelimiter(',');
+        $csv->insertOne(['toto', 'tata', 'tutu']);
+        $csv->insertOne(['toto', 'tata', 'tutu']);
+        $csv->insertOne(['toto', 'tata', 'tutu']);
+
+        $this->assertSame($csv->detectDelimiter(5, ['toto', '|']), [',', ';']);
+        $csv->detectDelimiter(-4);
+    }
+
     /**
      * @expectedException InvalidArgumentException
      */