parsecsv/parsecsv-for-php
1
0
Fork 0
mirror of https://github.com/parsecsv/parsecsv-for-php.git synced 2026-02-19 08:36:39 +00:00
Code Issues Packages Projects Releases Wiki Activity

When we find a BOM, we know the input encoding. Let's use it!

Browse Source 
Also: UTF tests not failing anymore because we are specifying
the output encoding now.
This commit is contained in:
Christian Bläul
2017-11-25 13:43:19 +01:00
committed by Fonata
parent c818fff81a
commit 4d011827f5
2 changed files with 17 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
parsecsv.lib.php
View File

@@ -841,15 +841,18 @@ class parseCSV {
if (strpos($data, "\xef\xbb\xbf") === 0) { if (strpos($data, "\xef\xbb\xbf") === 0) {
// strip off BOM (UTF-8) // strip off BOM (UTF-8)
$data = substr($data, 3); $data = substr($data, 3);
$this->encoding('UTF-8');
} }
else if (strpos($data, "\xff\xfe") === 0) { else if (strpos($data, "\xff\xfe") === 0) {
// strip off BOM (UTF-16 little endian) // strip off BOM (UTF-16 little endian)
$data = substr($data, 2); $data = substr($data, 2);
$this->encoding("UCS-2LE");
} }
else if (strpos($data, "\xfe\xff") === 0) { else if (strpos($data, "\xfe\xff") === 0) {
// strip off BOM (UTF-16 big endian) // strip off BOM (UTF-16 big endian)
$data = substr($data, 2); $data = substr($data, 2);
$this->encoding("UTF-16");
} }
if ($this->convert_encoding) { if ($this->convert_encoding) {

19
tests/methods/parse_test.php
View File

@@ -40,17 +40,26 @@ class parse_test extends PHPUnit\Framework\TestCase {
} }
public function test_sep_row_auto_detection_UTF8_no_BOM() { public function test_sep_row_auto_detection_UTF8_no_BOM() {
$this->csv->auto(__DIR__ . '/../example_files/UTF-8_sep_row_but_no_BOM.csv'); $this->_autoparse_magazine_file(
$this->assertEquals($this->_get_magazines_data(), $this->csv->data); __DIR__ . '/../example_files/UTF-8_sep_row_but_no_BOM.csv');
} }
public function test_sep_row_auto_detection_UTF8() { public function test_sep_row_auto_detection_UTF8() {
$this->csv->auto(__DIR__ . '/../example_files/UTF-8_with_BOM_and_sep_row.csv'); $this->_autoparse_magazine_file(
$this->assertEquals($this->_get_magazines_data(), $this->csv->data); __DIR__ . '/../example_files/UTF-8_with_BOM_and_sep_row.csv');
} }
public function test_sep_row_auto_detection_UTF16() { public function test_sep_row_auto_detection_UTF16() {
$this->csv->auto(__DIR__ . '/../example_files/UTF-16LE_with_BOM_and_sep_row.csv'); $this->_autoparse_magazine_file(
__DIR__ . '/../example_files/UTF-16LE_with_BOM_and_sep_row.csv');
}
protected function _autoparse_magazine_file($file) {
// This file (parse_test.php) is encoded in UTF-8, hence comparison will
// fail unless we to this:
$this->csv->output_encoding = 'UTF-8';
$this->csv->auto($file);
$this->assertEquals($this->_get_magazines_data(), $this->csv->data); $this->assertEquals($this->_get_magazines_data(), $this->csv->data);
} }