URLConnection.guessContentTypeFromStream() does not support UTF8 and UTF32 with BOM

[Charles Lee]

On 03/03/2011 12:48 AM, Alan Bateman wrote:
> Charles Lee wrote:
>> Hi guys,
>>
>> With test case[1] below, you can see guessContent does not support 
>> UTF8/32 BOM. 
> I don't see a problem with this proposal but I just wonder if this 
> method is even used these days. It is a legacy method and I don't 
> think anyone can rely on it to reliably detect any content type. I'm 
> curious how you ran into this, was it a test or custom protocol 
> handler that overrides getContentType and uses 
> guessContentTypeFromStream as a fallback?
>
> -Alan.
Hi Alan,

Sorry for the late reply. This test case comes from a big test case, 
which test more types of stream. I am not sure if no one use it. But 
from the spec:

/Ideally, this routine would not be needed. But many |http| servers 
return the incorrect content type; in addition, there are many 
nonstandard extensions. Direct inspection of the bytes to determine the 
content type is often more accurate than believing the content type 
claimed by the |http| server.

/guessContentTypeFromStream is needed as a fallback.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.openjdk.java.net/pipermail/net-dev/attachments/20110304/c9ade239/attachment.html