Bug #36785 [Com]: xml_parse return invalid character error with ISO-8859-1 data

Edit report at http://bugs.php.net/bug.php?id=36785&edit=1 ID: 36785 Comment by: tom at tomclegg dot net Reported by: giunta_gaetano at libero dot it Summary: xml_parse return invalid character error with ISO-8859-1 data Status: Bogus Type: Bug Package: XML related Operating System: windows 2000 PHP Version: 5.1.2 Block user comment: N Private report: N New Comment: Bug #33375 hints at it, but giunta_gaetano has explained it better here. The problem is that there is no (sensible) way to tell the XML parser which character encoding to use in cases where the XML declaration does not specify an encoding. Bug #33375 vaguely hints that XML encodings other than UTF-8 must be specified in the XML declaration. On the contrary, http://www.w3.org/TR/REC-xml/#charencoding specifically allows for the encoding to be specified by alternate means (for example, MIME headers). Why shouldn't PHP have the ability to work in such an environment? Meanwhile, a workaround is to use preg_replace() to add an encoding attribute to the XML declaration before passing the XML data to xml_parse(). Ugly, but more effective than saying "it shouldn't work". Previous Comments: ------------------------------------------------------------------------ [2006-03-19 21:12:24] tony2001@php.net The cause and the solution is properly explained in bug #33375. NO bug here. ------------------------------------------------------------------------ [2006-03-19 00:21:27] giunta_gaetano at libero dot it Description: ------------ PLEASE REOPEN AND FIX BUG #33375! It bewilders me that this has not yet been fixed in php 5.2.1... It is a BC breakage against PHP 4, and makes very very little sense anyway: - xml does NOT mandate a charset specification in the prologue - other communication/storage layers impose DIFFERENT standards on charset declarations and default charset values that the xml spec does by itself to be more clear, a common example: - received xml message has no charset in the prologue - it is received over HTTP, and the http content-type header states a charset (it is authoritative, according to the specs) - there is no way to tell the xml parser to use the correct charset for parsing the message! Why on earth was it not decided, when switching to libxml, that xml_parser_create() would get some automagic new powers, while xml_parser_create('ISO-8859-1') would be 100% backwards compatible and let the coder specify a source charset??? PS: at least fix the manual, and clearly specify that in order for the 'magical charset detection' to work, the xml prologue MUST contain a charset declaration!!! PPS: last but not least: the column number where the error is found (xml_get_current_column_number()) is also borked: whereas with php 4 the error reported the column corresponding to the first non-ascii char found, with php 5 the error reports the column where the xml element closing tag starts, which is a bit misleading... Reproduce code: --------------- Just try to parse any ISO-8859-1 xml file that has no charset specified in the prologue. Expected result: ---------------- no error Actual result: -------------- a dumb parsing error ------------------------------------------------------------------------ -- Edit this bug report at http://bugs.php.net/bug.php?id=36785&edit=1