Re: Possible Bug in org.apache.pdfbox.cos.COSDictionary#getNameAsString( COSName key )

[Andreas Lehmkühler <an...@lehmi.de>]

Hi Michael,

this issue was already described in [1] and the svn-version 790931
resolves that issue.

Thanks for your help.

Andreas Lehmkühler

[1] https://issues.apache.org/jira/browse/PDFBOX-455


Michael Weller schrieb:
> Hey Pdfbox-Users,
> 
> I'm using Pdfbox for a custom pdf library. I'm using the TextStripper
> class to extract textual contents of the pdf's pages. This works for
> alsmost 99% of my pdfs. Here's the old implementation:
> <code>
> public String getNameAsString( COSName key )
>     {
>         String retval = null;
>         COSName name = (COSName)getDictionaryObject( key );
>         if( name != null )
>         {
>             retval = name.getName();
>         }
>         return retval;
>     }
> </code>
> But for some I got ClassCastExceptions in
> COSDictionary#getNameAsString(COSName key) (used by TextStripper ..)
> because something returned by getDictionaryObject(COSName key) is not
> a COSName. After adding an "instanceof"-test I sometimes got
> NullPointerExceptions. So here is a new implementation with some
> simple checks which works for 100% of my pdfs ;-)
> 
> public String getNameAsString( COSName key )
>     {
>         String retval = null;
> 
>         COSBase cb = getDictionaryObject( key );
>         if (cb != null) {
>             if (cb instanceof COSName) {
>                  COSName name = (COSName)cb;
>                 if( name != null )
>                 {
>                     retval = name.getName();
>                 }
>             } else {
>                 retval = cb.toString();
>             }
>         }
>         return retval;
>     }
> 
> As far as I can tell there are no semantics touched by these modifications.
> 
> Thanks for your attention!