[TYPO3-RTE] German Umlauts are replaced by strange char's, when inserted via copy and paste

[Stanislas Rolland]

Hi Uwe,

There are a couple of issues here.

When an htmlParser configuration is provided for enableWordClean, a 
server-based cleaning is performed using the specified configuration.

First issue: the server-based process was always invoked, even if no 
htmlParser configuration was provided (like in the Typical default 
settings, for example). This is now corrected in cvs.

Second issue: when the server-based cleaning is performed, the content 
is sent to the server using XMLHttpRequest. The header was always set to 
utf-8. Therefore if your backend is not utf-8 (forceCharset is not set), 
the data may be incorrectly transformed. This is now corrected in cvs: 
the content is sent with header set to the charset of the content 
element (depends on language, possibly overridden by forceCharset).

Third issue: if PHP is compiled with mbstring, you have to make sure 
that mbstring is correctly configured. Should be something like:

mbstring.language = neutral
mbstring.http_input = auto
mbstring.http_output = pass
mbstring.internal_encoding = ISO-8859-1 or your-internal-charset
mbstring.encoding_translation = On

Fourth issue: even with the above corrections and configuration, if 
Windows character codes #128 to #159 are pasted in Firefox, they may end 
up being replaced by some generic character (question mark or empty 
square). These characters are invalid in HTML 4.0/XHTML. But this does 
not happen will all servers. Therefore there is some other server 
configuration issue affecting this behavior.

Any idea?

Stanislas