Re: Different result running standalone from servlet

[Nicholas Hogg <Ni...@kesoftware.com>]

Hi Jeremias,

Thanks for your suggestions.  The problem was as expected - we use 
Maverick as our MVC framework and it has a bug when passing content 
through multiple transformations - the encoding was messed up by the 
time it reached FOP.  The simple fix was to set up FOP to do the 
xml->xsl:fo and the xsl:fo->pdf transform rather than letting maverick 
control the former.

Thanks,
Nick

Jeremias Maerki wrote:
> You have the right suspicion. It must certainly be something inside the
> servlet that affects the encoding of the characters. The input data is
> probably in UTF-8 (your Maori character as two bytes) and the data
> passed to FOP is probably incorrectly forced to ISO-8859-1 (or another 1
> byte encoding) somehow. Without seeing the Java source code that handles
> the transformation it's impossible to just point you to the right place.
> Look out for things like the use of String.getBytes(), InputStreamReader
> or byte[]. They can (but don't have to) indicate problem spots. Please
> refer to our demo servlet [1][2] and the embedding examples [3] for
> recommended patterns. If you don't manage to identify the problem, post
> your code here.
> 
> [1] http://xmlgraphics.apache.org/fop/stable/servlets.html
> [2] http://svn.apache.org/viewvc/xmlgraphics/fop/trunk/src/java/org/apache/fop/servlet/FopServlet.java?view=markup
> [3] http://xmlgraphics.apache.org/fop/stable/embedding.html
> 
> On 31.01.2008 07:58:51 Nicholas Hogg wrote:
>> Hi,
>>
>> I am having difficulty getting two characters to display in my pdfs when 
>> running fop from a servlet that needs to support the Maori language (New 
>> Zealand).  The Maori language has 10 characters outside Latin 1 
>> character set: AEIOUaeiou all with a macron. I have embedded a font, 
>> freeSerif, from 
>> http://download.savannah.gnu.org/releases/freefont/freefont-ttf-20060126.tar.gz 
>> and it definitely has glyphs for these characters.  I use XSLT to 
>> generate xsl:fo and use the fop API to generate the pdf.  Strangely, if 
>> I capture the generated xsl:fo and run fop by hand (standalone), the 
>> generated pdf is correct.  Even more strangely, when run from the 
>> servlet, all macron characters display correctly except the lowercase 
>> "a" and "o" macron characters ("\u0101" and "\u014D").  Each appears in 
>> the pdf as two characters - the second is always a "?" and the first is 
>> either blank or a different "?" symbol, depending on the embedded font I 
>> am trying to use.  Trying different fonts produces the same problem with 
>> these two characters.
>>
>> I have looked at the servlet generated pdf in Adobe Acrobat Pro, and in 
>> the document properties verified that the font is being embedded (I use 
>> a peculiar name for it).  It is encoded as "Identity-H".
>>
>> When running fop by hand I use the same fop.xconf config file as the 
>> servlet.  I am running fop 0.94, and the servlet in Jetty6, on Windows, 
>> with Java 1.6.03.
>>
>> Any thoughts on how the same xsl:fo would produce different results 
>> standalone vs servlet?  I can only think that the xsl:fo going into fop 
>> via the servlet is getting corrupted so that fop thinks the "\u0101" and 
>> "\u014D" characters are two separate characters each.
>>
>> Many thanks,
>> Nick
> 
> 
> Jeremias Maerki
> 
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: fop-users-unsubscribe@xmlgraphics.apache.org
> For additional commands, e-mail: fop-users-help@xmlgraphics.apache.org
> 

-- 
Nicholas Hogg	
KE Software (Australia)
www.kesoftware.com

---------------------------------------------------------------------
To unsubscribe, e-mail: fop-users-unsubscribe@xmlgraphics.apache.org
For additional commands, e-mail: fop-users-help@xmlgraphics.apache.org