Revision 445
Added by Matthias about 17 years ago
| functions.php | ||
|---|---|---|
| 345 | 345 |
return($string); |
| 346 | 346 |
} |
| 347 | 347 |
|
| 348 |
// Function to get the DEFAULT_CHARSET |
|
| 349 |
function get_wbcharset() {
|
|
| 350 |
$charset=strtoupper(DEFAULT_CHARSET); |
|
| 351 |
if(strcmp($charset,"BIG5") == 0) {
|
|
| 352 |
$charset="BIG-5"; |
|
| 353 |
} |
|
| 354 |
return($charset); |
|
| 355 |
} |
|
| 356 |
|
|
| 357 | 348 |
// Function to convert a string from $from- to $to-encoding, using mysql |
| 358 | 349 |
function my_mysql_iconv($string, $from, $to) {
|
| 359 | 350 |
// keep current character set values: |
| ... | ... | |
| 372 | 363 |
return $converted_string; |
| 373 | 364 |
} |
| 374 | 365 |
|
| 375 |
// Function to convert a string from html-entities to umlauts |
|
| 376 |
// and encode htmlspecialchars |
|
| 377 |
function entities_to_umlauts($string) {
|
|
| 378 |
$charset = get_wbcharset(); |
|
| 366 |
// Function to convert a string from mixed html-entities/umlauts to pure utf-8-umlauts |
|
| 367 |
function string_to_utf8($string, $charset=DEFAULT_CHARSET) {
|
|
| 368 |
$charset = strtoupper($charset); |
|
| 369 |
if ($charset == '') { $charset = 'ISO-8859-1'; }
|
|
| 370 |
|
|
| 379 | 371 |
// there's no GB2312 or ISO-8859-11 encoding in php's mb_* functions |
| 380 |
if (strcmp($charset,"GB2312") == 0) {
|
|
| 381 |
if(function_exists('iconv')) {
|
|
| 382 |
$string=mb_convert_encoding($string,'UTF-8','HTML-ENTITIES'); |
|
| 383 |
$string=iconv("UTF-8","GB2312",$string);
|
|
| 384 |
} else {
|
|
| 385 |
$string=mb_convert_encoding($string,'UTF-8','HTML-ENTITIES'); |
|
| 386 |
$string=my_mysql_iconv($string, 'utf8', 'gb2312'); |
|
| 387 |
} |
|
| 388 |
} elseif (strcmp($charset,"ISO-8859-11") == 0) {
|
|
| 389 |
if(function_exists('iconv')) {
|
|
| 390 |
$string=mb_convert_encoding($string,'UTF-8','HTML-ENTITIES'); |
|
| 391 |
$string=iconv("UTF-8","ISO-8859-11",$string);
|
|
| 392 |
} else {
|
|
| 393 |
$string=mb_convert_encoding($string,'UTF-8','HTML-ENTITIES'); |
|
| 394 |
$string=my_mysql_iconv($string, 'utf8', 'tis620'); |
|
| 395 |
} |
|
| 372 |
if ($charset == "GB2312") {
|
|
| 373 |
$string=my_mysql_iconv($string, 'gb2312', 'utf8'); |
|
| 374 |
} elseif ($charset == "ISO-8859-11") {
|
|
| 375 |
$string=my_mysql_iconv($string, 'tis620', 'utf8'); |
|
| 396 | 376 |
} else {
|
| 397 |
$string=mb_convert_encoding($string,$charset,'HTML-ENTITIES');
|
|
| 377 |
$string=mb_convert_encoding($string, 'UTF-8', $charset);
|
|
| 398 | 378 |
} |
| 399 |
$string=htmlspecialchars($string); |
|
| 379 |
$string=mb_convert_encoding($string, 'HTML-ENTITIES', 'UTF-8'); |
|
| 380 |
$string=mb_convert_encoding($string, 'UTF-8', 'HTML-ENTITIES'); |
|
| 400 | 381 |
return($string); |
| 401 | 382 |
} |
| 402 | 383 |
|
| 403 |
// Function to convert a string from umlauts to html-entities |
|
| 404 |
// and encode htmlspecialchars |
|
| 405 |
function umlauts_to_entities($string) {
|
|
| 406 |
$charset=get_wbcharset(); |
|
| 407 |
// there's no GB2312 or ISO-8859-11 encoding in php's mb_* functions |
|
| 408 |
if (strcmp($charset,"GB2312") == 0) {
|
|
| 409 |
if(function_exists('iconv')) {
|
|
| 410 |
$string=iconv("GB2312","UTF-8",$string);
|
|
| 411 |
$charset="UTF-8"; |
|
| 384 |
// Function to convert a string from mixed html-entities/umlauts to pure $charset_out-umlauts |
|
| 385 |
function entities_to_umlauts($string, $charset_out=DEFAULT_CHARSET, $convert_htmlspecialchars=0) {
|
|
| 386 |
$charset_out = strtoupper($charset_out); |
|
| 387 |
if ($charset_out == '') {
|
|
| 388 |
$charset_out = 'ISO-8859-1'; |
|
| 389 |
} |
|
| 390 |
$string = string_to_utf8($string); |
|
| 391 |
if($charset_out != 'UTF-8') {
|
|
| 392 |
if ($charset_out == "GB2312") {
|
|
| 393 |
$string=my_mysql_iconv($string, 'utf8', 'gb2312'); |
|
| 394 |
} elseif ($charset_out == "ISO-8859-11") {
|
|
| 395 |
$string=my_mysql_iconv($string, 'utf8', 'tis620'); |
|
| 412 | 396 |
} else {
|
| 413 |
$string=my_mysql_iconv($string, 'gb2312', 'utf8'); |
|
| 414 |
$charset="UTF-8"; |
|
| 397 |
$string=mb_convert_encoding($string, $charset_out, 'UTF-8'); |
|
| 415 | 398 |
} |
| 416 |
} elseif (strcmp($charset,"ISO-8859-11") == 0) {
|
|
| 417 |
if(function_exists('iconv')) {
|
|
| 418 |
$string=iconv("ISO-8859-11","UTF-8",$string);
|
|
| 419 |
$charset="UTF-8"; |
|
| 420 |
} else {
|
|
| 421 |
$string=my_mysql_iconv($string, 'tis620', 'utf8'); |
|
| 422 |
$charset="UTF-8"; |
|
| 423 |
} |
|
| 424 | 399 |
} |
| 425 |
$string=mb_convert_encoding($string,'HTML-ENTITIES',$charset); |
|
| 426 |
$string=mb_convert_encoding($string,'UTF-8','HTML-ENTITIES'); |
|
| 427 |
$string=htmlspecialchars($string,ENT_QUOTES); |
|
| 400 |
if($convert_htmlspecialchars == 1) {
|
|
| 401 |
$string=htmlspecialchars($string); |
|
| 402 |
} |
|
| 403 |
return($string); |
|
| 404 |
} |
|
| 405 |
|
|
| 406 |
// Function to convert a string from mixed html-entitites/$charset_in-umlauts to pure html-entities |
|
| 407 |
function umlauts_to_entities($string, $charset_in=DEFAULT_CHARSET, $convert_htmlspecialchars=1) {
|
|
| 408 |
$charset_in = strtoupper($charset_in); |
|
| 409 |
if ($charset_in == "") {
|
|
| 410 |
$charset_in = 'ISO-8859-1'; |
|
| 411 |
} |
|
| 412 |
$string = string_to_utf8($string, $charset_in); |
|
| 413 |
if($convert_htmlspecialchars == 1) {
|
|
| 414 |
$string=htmlspecialchars($string,ENT_QUOTES); |
|
| 415 |
} |
|
| 428 | 416 |
$string=mb_convert_encoding($string,'HTML-ENTITIES','UTF-8'); |
| 429 | 417 |
return($string); |
| 430 | 418 |
} |
| 431 | 419 |
|
| 432 |
// translate any "latin" html-entities to their plain 7bit equivalents
|
|
| 420 |
// translate any latin/greek/cyrillic html-entities to their plain 7bit equivalents
|
|
| 433 | 421 |
function entities_to_7bit($string) {
|
| 434 | 422 |
require(WB_PATH.'/framework/convert.php'); |
| 435 | 423 |
$string = strtr($string, $conversion_array); |
Also available in: Unified diff
Optimized search and highlighting funktions