/**
* Detects encoding of xml-formatted text.
* @param string $string The input xml-formatted text.
* @param string $default_encoding This is the default encoding to be returned if there is no way the xml-text's encoding to be detected. If it not spesified, the system encoding is assumed then.
* @return string Returns the detected encoding.
* @todo The second parameter is to be eliminated. See api_detect_encoding_html().
*/
function api_detect_encoding_xml($string, $default_encoding = null)
{
if (preg_match(_PCRE_XML_ENCODING, $string, $matches)) {
return api_refine_encoding_id($matches[1]);
}
if (api_is_valid_utf8($string)) {
return 'UTF-8';
}
if (empty($default_encoding)) {
$default_encoding = _api_mb_internal_encoding();
}
return api_refine_encoding_id($default_encoding);
}
/**
* Detects the encoding of a given manifest (a xml-text).
* It is possible the encoding of the manifest to be wrongly declared or
* not to be declared at all. The proposed method tries to resolve these problems.
* @param string $xml The input xml-text.
* @return string The detected value of the input xml.
*/
private function detect_manifest_encoding(&$xml)
{
if (api_is_valid_utf8($xml)) {
return 'UTF-8';
}
if (preg_match(_PCRE_XML_ENCODING, $xml, $matches)) {
$declared_encoding = api_refine_encoding_id($matches[1]);
} else {
$declared_encoding = '';
}
if (!empty($declared_encoding) && !api_is_utf8($declared_encoding)) {
return $declared_encoding;
}
$test_string = '';
if (preg_match_all('/<langstring[^>]*>(.*)<\\/langstring>/m', $xml, $matches)) {
$test_string = implode("\n", $matches[1]);
unset($matches);
}
if (preg_match_all('/<title[^>]*>(.*)<\\/title>/m', $xml, $matches)) {
$test_string .= "\n" . implode("\n", $matches[1]);
unset($matches);
}
if (empty($test_string)) {
$test_string = $xml;
}
return api_detect_encoding($test_string);
}
/**
* This function reads a Chamilo language file and transforms it into XML,
* then returns the XML string to the caller.
*/
function get_language_file_as_xml($language = 'english')
{
$path = api_get_path(SYS_LANG_PATH) . $language . '/';
if (!is_dir($path) or !is_readable($path)) {
if ($language != 'english') {
return get_language_file_as_xml('english');
} else {
return '';
}
}
//error_log('Analysing path '.$path);
$file = $path . 'videoconf.inc.php';
if (!is_file($file) or !is_readable($file)) {
if ($language != 'english') {
return get_language_file_as_xml('english');
} else {
return '';
}
}
/*
$convert = true;
if(substr($language,-7,7) == 'unicode')
{//do not convert if the language ends with 'unicode', which means it's in UTF-8
$convert=false;
}
$list = file($file);
$xml = '';
foreach ( $list as $line )
{
if(substr($line,0,1)=='$')
{
$items = array();
$match = preg_match('/^\$([^\s]*)\s*=\s*"(.*)";$/',$line,$items);
if($match)
{
//todo: The following conversion should only happen for old language files (encoded in ISO-8859-1).
if($convert)
{
$string = api_convert_encoding($items[2],'UTF-8','ISO-8859-1');
}
else
{
$string = $items[2];
}
$xml .= '<labelfield><labelid>'.$items[1].'</labelid><labelvalue>'.stripslashes($string).'</labelvalue></labelfield>'."\n";
}
}
}
*/
//---------
$non_utf8_encoding = api_get_non_utf8_encoding($language);
$list = file($file);
$xml = '';
foreach ($list as $line) {
if (substr($line, 0, 1) == '$') {
$items = array();
$match = preg_match('/^\\$([^\\s]*)\\s*=\\s*"(.*)";$/', $line, $items);
if ($match) {
$string = $items[2];
if (!api_is_valid_utf8($string)) {
$string = api_html_entity_decode(api_utf8_encode($string, $non_utf8_encoding), ENT_QUOTES, 'UTF-8');
}
$xml .= '<labelfield><labelid>' . $items[1] . '</labelid><labelvalue>' . stripslashes($string) . '</labelvalue></labelfield>' . "\n";
}
}
}
//---------
if (empty($xml) && $language != 'english') {
return get_language_file_as_xml('english');
}
return $xml;
}
/**
* Detects encoding of plain text.
* @param string $string The input text.
* @param string $language (optional) The language of the input text, provided if it is known.
* @return string Returns the detected encoding.
*/
function api_detect_encoding($string, $language = null)
{
// Testing against valid UTF-8 first.
if (api_is_valid_utf8($string)) {
return 'UTF-8';
}
return mb_detect_encoding($string);
}
/**
* Detects encoding of plain text.
* @param string $string The input text.
* @param string $language (optional) The language of the input text, provided if it is known.
* @return string Returns the detected encoding.
*/
function api_detect_encoding($string, $language = null)
{
// Testing against valid UTF-8 first.
if (api_is_valid_utf8($string)) {
return 'UTF-8';
}
$result = null;
$delta_points_min = LANGUAGE_DETECT_MAX_DELTA;
// Testing non-UTF-8 encodings.
$encodings = api_get_valid_encodings();
foreach ($encodings as &$encoding) {
if (api_is_encoding_supported($encoding) && !api_is_utf8($encoding)) {
$stringToParse = api_substr($string, 0, LANGUAGE_DETECT_MAX_LENGTH, $encoding);
$strintToParse2 = _api_generate_n_grams($stringToParse, $encoding);
$result_array = _api_compare_n_grams($strintToParse2, $encoding);
if (!empty($result_array)) {
list($key, $delta_points) = each($result_array);
if ($delta_points < $delta_points_min) {
$pos = strpos($key, ':');
$result_encoding = api_refine_encoding_id(substr($key, $pos + 1));
if (api_equal_encodings($encoding, $result_encoding)) {
if ($string == api_utf8_decode(api_utf8_encode($string, $encoding), $encoding)) {
$delta_points_min = $delta_points;
$result = $encoding;
}
}
}
}
}
}
// "Broken" UTF-8 texts are to be detected as UTF-8.
// This functionality is enabled when language of the text is known.
$language = api_purify_language_id((string) $language);
if (!empty($language)) {
$encoding = 'UTF-8';
$result_array =& _api_compare_n_grams(_api_generate_n_grams(api_substr($string, 0, LANGUAGE_DETECT_MAX_LENGTH, $encoding), $encoding), $encoding);
if (!empty($result_array)) {
list($key, $delta_points) = each($result_array);
if ($delta_points < $delta_points_min) {
$pos = strpos($key, ':');
$result_encoding = api_refine_encoding_id(substr($key, $pos + 1));
$result_language = substr($key, 0, $pos);
if ($language == $result_language && api_is_utf8($result_encoding)) {
$delta_points_min = $delta_points;
$result = $encoding;
}
}
}
}
return $result;
}
