summaryrefslogtreecommitdiffstats
path: root/system/core/Utf8.php
diff options
context:
space:
mode:
Diffstat (limited to 'system/core/Utf8.php')
-rw-r--r--system/core/Utf8.php73
1 files changed, 38 insertions, 35 deletions
diff --git a/system/core/Utf8.php b/system/core/Utf8.php
index ba3567453..5bc2dd5c9 100644
--- a/system/core/Utf8.php
+++ b/system/core/Utf8.php
@@ -1,4 +1,4 @@
-<?php if ( ! defined('BASEPATH')) exit('No direct script access allowed');
+<?php
/**
* CodeIgniter
*
@@ -24,8 +24,7 @@
* @since Version 2.0
* @filesource
*/
-
-// ------------------------------------------------------------------------
+defined('BASEPATH') OR exit('No direct script access allowed');
/**
* Utf8 Class
@@ -41,38 +40,40 @@
class CI_Utf8 {
/**
- * Constructor
+ * Class constructor
+ *
+ * Determines if UTF-8 support is to be enabled.
*
- * Determines if UTF-8 support is to be enabled
+ * @return void
*/
public function __construct()
{
log_message('debug', 'Utf8 Class Initialized');
- global $CFG;
+ $charset = strtoupper(config_item('charset'));
+
+ // set internal encoding for multibyte string functions if necessary
+ // and set a flag so we don't have to repeatedly use extension_loaded()
+ // or function_exists()
+ if (extension_loaded('mbstring'))
+ {
+ define('MB_ENABLED', TRUE);
+ mb_internal_encoding($charset);
+ }
+ else
+ {
+ define('MB_ENABLED', FALSE);
+ }
if (
- @preg_match('/./u', 'é') === 1 // PCRE must support UTF-8
- && function_exists('iconv') // iconv must be installed
- && @ini_get('mbstring.func_overload') != 1 // Multibyte string function overloading cannot be enabled
- && $CFG->item('charset') === 'UTF-8' // Application charset must be UTF-8
+ @preg_match('/./u', 'é') === 1 // PCRE must support UTF-8
+ && function_exists('iconv') // iconv must be installed
+ && MB_ENABLED === TRUE // mbstring must be enabled
+ && $charset === 'UTF-8' // Application charset must be UTF-8
)
{
define('UTF8_ENABLED', TRUE);
log_message('debug', 'UTF-8 Support Enabled');
-
- // set internal encoding for multibyte string functions if necessary
- // and set a flag so we don't have to repeatedly use extension_loaded()
- // or function_exists()
- if (extension_loaded('mbstring'))
- {
- define('MB_ENABLED', TRUE);
- mb_internal_encoding('UTF-8');
- }
- else
- {
- define('MB_ENABLED', FALSE);
- }
}
else
{
@@ -86,9 +87,11 @@ class CI_Utf8 {
/**
* Clean UTF-8 strings
*
- * Ensures strings are UTF-8
+ * Ensures strings contain only valid UTF-8 characters.
*
- * @param string
+ * @uses CI_Utf8::_is_ascii() Decide whether a conversion is needed
+ *
+ * @param string $str String to clean
* @return string
*/
public function clean_string($str)
@@ -108,9 +111,9 @@ class CI_Utf8 {
*
* Removes all ASCII control characters except horizontal tabs,
* line feeds, and carriage returns, as all others can cause
- * problems in XML
+ * problems in XML.
*
- * @param string
+ * @param string $str String to clean
* @return string
*/
public function safe_ascii_for_xml($str)
@@ -123,11 +126,11 @@ class CI_Utf8 {
/**
* Convert to UTF-8
*
- * Attempts to convert a string to UTF-8
+ * Attempts to convert a string to UTF-8.
*
- * @param string
- * @param string - input encoding
- * @return string
+ * @param string $str Input string
+ * @param string $encoding Input encoding
+ * @return string $str encoded in UTF-8 or FALSE on failure
*/
public function convert_to_utf8($str, $encoding)
{
@@ -135,7 +138,7 @@ class CI_Utf8 {
{
return @iconv($encoding, 'UTF-8', $str);
}
- elseif (function_exists('mb_convert_encoding'))
+ elseif (MB_ENABLED === TRUE)
{
return @mb_convert_encoding($str, 'UTF-8', $encoding);
}
@@ -148,9 +151,9 @@ class CI_Utf8 {
/**
* Is ASCII?
*
- * Tests if a string is standard 7-bit ASCII or not
+ * Tests if a string is standard 7-bit ASCII or not.
*
- * @param string
+ * @param string $str String to check
* @return bool
*/
protected function _is_ascii($str)
@@ -161,4 +164,4 @@ class CI_Utf8 {
}
/* End of file Utf8.php */
-/* Location: ./system/core/Utf8.php */
+/* Location: ./system/core/Utf8.php */ \ No newline at end of file