#32547 [Csd->Bgs]: DOMDocument->loadHTML() seems to broke (utf-8 russian) codepage

ID: 32547 Updated by: tony2001@php.net Reported By: xlex0x835 at rambler dot ru -Status: Closed +Status: Bogus Bug Type: DOM XML related Operating System: Mac OS X 10.3, FreeBSD 5.3 PHP Version: 5.0.3 Assigned To: rrichards New Comment: No PHP bug -> bogus. Previous Comments: ------------------------------------------------------------------------ [2005-04-04 09:09:40] xlex0x835 at rambler dot ru As for <xml>, please, read more carefully. I told, that I put that tag just to have one correct source for both loadHTML() and loadXML() methods. As for libxml - thank you to confirm that it is that lib problem. ------------------------------------------------------------------------ [2005-04-04 08:40:10] chregu@php.net Not a bug, IMHO. HTML 4 is not XML, therefore it doesn't know about the <?xml ?> processing instruction and doesn't apply that information. You have to use the meta tag in html 4 (loadHTML is only about HTML 4 and not XHTML) I may be wrong with that assumption, so please point me to the right specs, if loadHTML should recognize that. But anyway, not a PHP bug, but basically a libxml2 "problem" ------------------------------------------------------------------------ [2005-04-03 19:44:05] xlex0x835 at rambler dot ru Problem seems to be detected: if I will put <meta> tag just after the title, document will be parsed absolutely correct. Is it libxml bug or PHP bindings? ------------------------------------------------------------------------ [2005-04-02 18:58:50] xlex0x835 at rambler dot ru Description: ------------ If I use DOMDocument->loadHTML() method with an utf-8 HTML, which contains russian characters, that russian characters just messed (please see 'Actual result'). Nothing changed if I specify encoding "by hand" (I mean the following call: "$domDoc = new DOMDocument('1.0', 'utf-8');"). But, eveything works just fine if I use DOMDocument- >loadXML() method (that's why there is xml definition string in the input). Nothing changed if I will remove all $domDoc options, neither removing "<?xml ... ?>" string (it is actually exist only to get one source for both loadHTML() and loadXML() functions call - to test error). The problem was discrovered on the "real-world" HTML, the code was stripped to the minimum for the ease of use. Host info. =================================== [PHP Modules (on FreeBSD 5.3 host)] bcmath bz2 calendar ctype curl dom exif ftp gd gettext gmp iconv imap libxml mbstring mcrypt mcve mhash mysql ncurses odbc openssl pcntl pcre pgsql posix pspell readline session shmop SimpleXML snmp soap sockets SPL SQLite standard sysvmsg sysvsem sysvshm tidy tokenizer wddx xml xmlrpc xsl yaz yp zip zlib No Zend modules. FreeBSD 5.3-RELEASE libxml2-2.6.13 gcc (GCC) 3.4.2 [FreeBSD] 20040728 Reproduce code: --------------- <?php $xmlContent = file_get_contents('input_test'); $domDoc = new DOMDocument(); $domDoc->formatOutput = true; $domDoc->preserveWhiteSpace = false; $domDoc->recover = true; $domDoc->loadXML($xmlContent);          file_put_contents('output_test', $domDoc->saveXML()); ?> input_test: =========== <?xml version="1.0" encoding="utf-8"?> <html> <head> <title>Òåñò - Test</title> <meta http-equiv="Content-Type" content="text/html; charset=utf-8" /> </head> </html> Expected result: ---------------- <?xml version="1.0" encoding="utf-8" standalone="yes"?> <!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN" "http://www.w3.org/TR/REC-html40/ loose.dtd"> <html> <head> <title>Òåñò - Test</title> <meta http-equiv="Content-Type" content="text/html; charset=utf-8"/> </head> </html> Actual result: -------------- <?xml version="1.0" encoding="utf-8" standalone="yes"?> <!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN" "http://www.w3.org/TR/REC-html40/ loose.dtd"> <html> <head> <title>Тест - Test</title> <meta http-equiv="Content-Type" content="text/html; charset=utf-8"/> </head> </html> ------------------------------------------------------------------------ -- Edit this bug report at http://bugs.php.net/?id=32547&edit=1