Fix encodings

Author: emohamed

src/CsvFile.php

@@ -7,7 +7,7 @@
  * Enhanced CSV file object
  */
 class CsvFile extends File implements \Countable {
-private $file_path;
+const DEFAULT_ENCODING = 'utf-8';
 private $encoding = 'utf-8';
 private $is_head_row = false;
 /**
@@ -20,17 +20,16 @@ class CsvFile extends File implements \Countable {
 private $start_column = 0;
 private $columns_to_skip = array();
 
-function __construct($file_path, $delimiter = ',', $enclosure = '"', $escape = "\\") {
-if (!file_exists($file_path)) {
-throw new Exception("File $file_path does not exist. ");
+function __construct($file, $delimiter = ',', $enclosure = '"', $escape = "\\") {
+if (!file_exists($file)) {
+throw new Exception("File $file does not exist. ");
 }
 
-if (filesize($file_path) === 0) {
+if (filesize($file) === 0) {
 throw new Exception("Empty file. ");
 }
 
-$this->file_path = $file_path;
-parent::__construct($file_path, 'r');
+parent::__construct($file, 'r+');
 $this->setFlags(File::READ_CSV | File::READ_AHEAD | File::SKIP_EMPTY | File::DROP_NEW_LINE);
 $this->setCsvControl($delimiter, $enclosure, $escape);
 }
@@ -43,6 +42,10 @@ function count() {
 return count($this->to_array());
 }
 
+function set_encoding($encoding) {
+$this->encoding = $encoding;
+}
+
 public function to_array() {
 $rows = [];
 foreach ($this as $row) {
@@ -93,7 +96,14 @@ private function remove_columns($old_row) {
 return $new_row;
 }
 
+function convert_to_default_encoding($val) {
+return mb_convert_encoding($val, static::DEFAULT_ENCODING, $this->encoding);
+}
+
 private function format_row($row) {
+if ($this->encoding !== static::DEFAULT_ENCODING) {
+$row = array_map([$this, 'convert_to_default_encoding'], $row);
+}
 $row = array_combine(
 $this->get_column_names($row),
 $row

tests/CsvFileTest.php

@@ -10,7 +10,6 @@ function test_it_can_be_constructed_from_a_file() {
 $csv = new CsvFile(__DIR__ . '/../sample-data/info.csv');
 $this->assertEquals(5, count($csv));
 }
-
 /**
  * @expectedException \Carbon_CSV\Exception
  */
@@ -227,9 +226,6 @@ function test_throws_exception_when_using_non_number_characters_for_skip_to_colu
 $csv = new CsvFile(__DIR__ . '/../sample-data/info.csv' );
 $csv->skip_to_column('a');
 }
-/**
- * @skip
- */
 function test_non_utf8_encoded_file() {
 $csv = new CsvFile(__DIR__ . '/../sample-data/cp1251.csv');
 $csv->set_encoding('cp-1251');
@@ -240,17 +236,11 @@ function test_non_utf8_encoded_file() {
 ]);
 $this->assertEquals( [
 [
-'last_name' => 'Doe'
+'name' => 'Хоумър',
+'age' => '31',
 ],
 ], $csv->to_array() );
 }
+
 
-function test_bad_mappings_are_providing_sane_error() {
-$csv = new CsvFile(__DIR__ . '/../sample-data/cp1251.csv');
-$csv->use_first_row_as_header();
-$csv->set_column_names([
-0 => 'name',
-1 => 'age',
-]);
-}
 }