adding QueryFilter::stripBom method

6 files changed, 139 insertions, 7 deletions

diff --git a/CHANGELOG.md b/CHANGELOG.md
index 3eb1161..172b953 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -1,6 +1,12 @@
 #Changelog
 All Notable changes to `League\Csv` will be documented in this file
 
+## Next - XXXX-XX-XX
+
+### Added
+
+- `Reader::stripBOM` query filtering method to ease removing the BOM sequence when querying the CSV document.
+
 ## 7.0.1 - 2015-03-23
 
 ### Fixed
diff --git a/composer.json b/composer.json
index fc12e7e..04d2205 100644
--- a/composer.json
+++ b/composer.json
@@ -37,7 +37,7 @@
     },
     "extra": {
         "branch-alias": {
-            "dev-master": "7.0-dev"
+            "dev-master": "7.1-dev"
         }
     }
 }
diff --git a/src/Config/Output.php b/src/Config/Output.php
index 81b6434..6ba19ce 100644
--- a/src/Config/Output.php
+++ b/src/Config/Output.php
@@ -100,6 +100,7 @@ trait Output
     {
         if (empty($str)) {
             $this->output_bom = null;
+
             return $this;
         }
         $str = (string) $str;
@@ -187,7 +188,7 @@ trait Output
         $csv->rewind();
         $csv->setFlags(SplFileObject::READ_CSV);
         if (! empty($bom) && ! empty($input_bom)) {
-            $csv->fseek(strlen($input_bom));
+            $csv->fseek(mb_strlen($input_bom));
         }
         echo $bom;
         $res = $csv->fpassthru();
diff --git a/src/Modifier/QueryFilter.php b/src/Modifier/QueryFilter.php
index 4602d75..c37d430 100644
--- a/src/Modifier/QueryFilter.php
+++ b/src/Modifier/QueryFilter.php
@@ -56,6 +56,69 @@ trait QueryFilter
     protected $iterator_limit = -1;
 
     /**
+     * Stripping BOM status
+     *
+     * @var boolean
+     */
+    protected $strip_bom = false;
+
+    /**
+     * Stripping BOM setter
+     *
+     * @param  bool $status
+     *
+     * @return $this
+     */
+    public function stripBom($status)
+    {
+        $this->strip_bom = (bool) $status;
+
+        return $this;
+    }
+
+    /**
+     * Tell whethe we can strip or not the leading BOM sequence
+     *
+     * @return boolean [description]
+     */
+    protected function isBomStrippable()
+    {
+        $bom = $this->getInputBom();
+
+        return ! empty($bom) && $this->strip_bom;
+    }
+
+    /**
+     * Remove the BOM sequence from the CSV
+     *
+     * @param  Iterator $iterator
+     *
+     * @return Iterator
+     */
+    protected function applyBomStripping(Iterator $iterator)
+    {
+        if (! $this->isBomStrippable()) {
+            return $iterator;
+        }
+
+        $bom = $this->getInputBom();
+        return new MapIterator($iterator, function ($row, $index) use ($bom) {
+            if (0 == $index) {
+                $row[0] = mb_substr($row[0], mb_strlen($bom));
+            }
+
+            return $row;
+        });
+    }
+
+    /**
+     * Returns the BOM sequence of the given CSV
+     *
+     * @return string
+     */
+    abstract public function getInputBom();
+
+    /**
      * Set LimitIterator Offset
      *
      * @param $offset
diff --git a/src/Reader.php b/src/Reader.php
index 1f73476..4e764b5 100644
--- a/src/Reader.php
+++ b/src/Reader.php
@@ -45,7 +45,9 @@ class Reader extends AbstractCsv
      */
     public function query(callable $callable = null)
     {
-        $iterator = new CallbackFilterIterator($this->getIterator(), function ($row) {
+        $iterator = $this->getIterator();
+        $iterator = $this->applyBomStripping($iterator);
+        $iterator = new CallbackFilterIterator($iterator, function ($row) {
             return is_array($row);
         });
 
@@ -53,7 +55,7 @@ class Reader extends AbstractCsv
         $iterator = $this->applyIteratorSortBy($iterator);
         $iterator = $this->applyIteratorInterval($iterator);
         if (! is_null($callable)) {
-            $iterator = new Modifier\MapIterator($iterator, $callable);
+            return new Modifier\MapIterator($iterator, $callable);
         }
 
         return $iterator;
@@ -65,11 +67,11 @@ class Reader extends AbstractCsv
     protected function getConversionIterator()
     {
         $iterator = $this->getIterator();
+        $iterator = $this->applyBomStripping($iterator);
         $iterator = $this->applyIteratorFilter($iterator);
         $iterator = $this->applyIteratorSortBy($iterator);
-        $iterator = $this->applyIteratorInterval($iterator);
 
-        return $iterator;
+        return $this->applyIteratorInterval($iterator);
     }
 
     /**
@@ -239,6 +241,11 @@ class Reader extends AbstractCsv
             throw new InvalidArgumentException('the specified row does not exist');
         }
 
+        if (0 == $offset && $this->isBomStrippable()) {
+            $bom    = $this->getInputBom();
+            $res[0] = mb_substr($res[0], mb_strlen($bom));
+        }
+
         return $res;
     }
 
diff --git a/test/ReaderTest.php b/test/ReaderTest.php
index 6d0ed4a..e9efb83 100644
--- a/test/ReaderTest.php
+++ b/test/ReaderTest.php
@@ -193,7 +193,62 @@ class ReaderTest extends PHPUnit_Framework_TestCase
     }
 
     /**
-     * @expectedException InvalidArgumentException
+     * @param  $expected
+     * @dataProvider validBOMSequences
+     */
+    public function testStripBOM($expected, $res)
+    {
+        $tmpFile = new SplTempFileObject();
+        foreach ($expected as $row) {
+            $tmpFile->fputcsv($row);
+        }
+        $csv = Reader::createFromFileObject($tmpFile);
+        $csv->setFlags(SplFileObject::READ_AHEAD|SplFileObject::SKIP_EMPTY);
+        $csv->stripBom(true);
+
+        $this->assertSame($res, $csv->fetchAll()[0][0]);
+    }
+
+    public function validBOMSequences()
+    {
+        return [
+            'withBOM' => [[
+                [Reader::BOM_UTF16_LE.'john', 'doe', 'john.doe@example.com', ],
+                ['jane', 'doe', 'jane.doe@example.com', ],
+            ], 'john'],
+            'withDoubleBOM' =>  [[
+                [Reader::BOM_UTF16_LE.Reader::BOM_UTF16_LE.'john', 'doe', 'john.doe@example.com', ],
+                ['jane', 'doe', 'jane.doe@example.com', ],
+            ], Reader::BOM_UTF16_LE.'john'],
+            'withoutBOM' => [[
+                ['john', 'doe', 'john.doe@example.com', ],
+                ['jane', 'doe', 'jane.doe@example.com', ],
+            ], 'john'],
+        ];
+    }
+
+    public function testStripBOMWithFetchAssoc()
+    {
+        $tmpFile = new SplTempFileObject();
+        $expected = [
+            [Reader::BOM_UTF16_LE.'john', 'doe', 'john.doe@example.com', ],
+            ['jane', 'doe', 'jane.doe@example.com', ],
+        ];
+
+        $tmpFile = new SplTempFileObject();
+        foreach ($expected as $row) {
+            $tmpFile->fputcsv($row);
+        }
+        $csv = Reader::createFromFileObject($tmpFile);
+        $csv->setFlags(SplFileObject::READ_AHEAD|SplFileObject::SKIP_EMPTY);
+        $csv->stripBom(true);
+        $res = array_keys($csv->fetchAssoc()[0]);
+
+        $this->assertSame('john', $res[0]);
+    }
+
+    /**
+     * @expectedException \InvalidArgumentException
      * @expectedExceptionMessage Use a flat non empty array with unique string values
      */
     public function testFetchAssocKeyFailure()