Merge pull request #25 from thephpleague/detectdelimiter

adding detectDelimiter to AbstractCsv class #16

2 files changed, 84 insertions, 0 deletions

diff --git a/src/AbstractCsv.php b/src/AbstractCsv.php
index 4d92418..a5b6399 100644
--- a/src/AbstractCsv.php
+++ b/src/AbstractCsv.php
@@ -39,6 +39,8 @@ use SplFileInfo;
 use SplFileObject;
 use SplTempFileObject;
 use InvalidArgumentException;
+use LimitIterator;
+use CallbackFilterIterator;
 
 /**
  *  A abstract class to enable basic CSV manipulation
@@ -210,6 +212,49 @@ class AbstractCsv implements JsonSerializable, IteratorAggregate
     }
 
     /**
+     * Detect the CSV file delimiter
+     *
+     * @param integer $nbRows
+     * @param array   $delimiters additional delimiters
+     *
+     * @return string
+     *
+     * @throws \InvalidArgumentException If $nbRows value is invalid
+     * @throws \RuntimeException         If too many delimiters are found
+     */
+    public function detectDelimiter($nbRows = 1, array $delimiters = [])
+    {
+        $nbRows = filter_var($nbRows, FILTER_VALIDATE_INT, ['options' => ['min_range' => 1]]);
+        if (! $nbRows) {
+            throw new InvalidArgumentException('`$nbRows` must be a valid positive integer');
+        }
+        $delimiters = array_filter($delimiters, function ($str) {
+            return 1 == mb_strlen($str);
+        });
+        $delimiters = array_merge([',', ';', "\t"], $delimiters);
+        $delimiters = array_unique($delimiters);
+        $iterator = new CallbackFilterIterator(
+            new LimitIterator($this->csv, 0, $nbRows),
+            function ($row) {
+                return is_array($row) && count($row) > 1;
+            }
+        );
+        $res = [];
+        foreach ($delimiters as $delim) {
+            $iterator->setCsvControl($delim, $this->enclosure, $this->escape);
+            $res[$delim] = count(iterator_to_array($iterator, false));
+        }
+        arsort($res, SORT_NUMERIC);
+        $res = array_keys(array_filter($res));
+        if (! $res) {
+            return null;
+        } elseif (count($res) == 1) {
+            return $res[0];
+        }
+        throw new RuntimeException('too many delimiters were found: `'.implode('`,`', $res).'`');
+    }
+
+    /**
      * set the field enclosure
      *
      * @param string $enclosure
diff --git a/test/CsvTest.php b/test/CsvTest.php
index c810cbc..ea822a2 100644
--- a/test/CsvTest.php
+++ b/test/CsvTest.php
@@ -85,6 +85,45 @@ class CsvTest extends PHPUnit_Framework_TestCase
         $this->csv->setDelimiter('foo');
     }
 
+    public function testDetectDelimiter()
+    {
+        $this->assertSame($this->csv->detectDelimiter(), ',');
+    }
+
+    /**
+     * @expectedException InvalidArgumentException
+     */
+    public function testDetectDelimiterWithInvalidRowLimit()
+    {
+        $this->csv->detectDelimiter(-4);
+    }
+
+    public function testDetectDelimiterWithNoCSV()
+    {
+        $file = new SplTempFileObject;
+        $file->fwrite("How are you today ?\nI'm doing fine thanks!");
+        $csv = new Writer($file);
+        $this->assertNull($csv->detectDelimiter(5, ['toto', '|']));
+    }
+
+    /**
+     * @expectedException RuntimeException
+     */
+    public function testDetectDelimiterWithInconsistentCSV()
+    {
+        $csv = new Writer(new SplTempFileObject);
+        $csv->setDelimiter(';');
+        $csv->insertOne(['toto', 'tata', 'tutu']);
+        $csv->setDelimiter('|');
+        $csv->insertAll([
+            ['toto', 'tata', 'tutu'],
+            ['toto', 'tata', 'tutu'],
+            ['toto', 'tata', 'tutu']
+        ]);
+
+        $csv->detectDelimiter(5, ['toto', '|']);
+    }
+
     /**
      * @expectedException InvalidArgumentException
      */