It was failing in our unit tests, where someone had wrapped our derived class instance in a Mockito spy(). With the spy, it breaks. Without the spy, it works fine. Not going to pretend I understand why, but I'll take that as a win and change our tests.
Totally unexpected, but sorry for the noise anyway. Mark -----Original Message----- From: Mark Gibson <[email protected]> Sent: 27 November 2025 20:47 To: [email protected] Cc: [email protected] Subject: Re: Extending PDFTextStripper - Worked in v2, NullRefException in v3 I’ll see if I can test later tonight. Thanks for the quick response Tilman. > On 27 Nov 2025, at 20:41, Tilman Hausherr <[email protected]> wrote: > > [EXTERNAL] > > In 3.0 the context is set in the constructor. But your problem is a > different one, that "resources" is null. > It would be interesting to find out whether this happens in ordinary > text extraction. If not, please share minimal code that reproduces > this, and the PDF where it happens (or does it happen with any PDF?) > Tilman > >> Am 27.11.2025 um 21:28 schrieb Mark Gibson: >> Hi >> >> We have some old code that extended PDFTextStripper using PDFBox v2. >> Recently tried to upgrade to PDFBox v3 (tried 3.03 and 3.06). We get a Null >> exception thrown from SetFontAndSize operator. It looks like "resources" is >> null. >> >> PDFont font = context.getResources().getFont(fontName); >> >> >> Tracing through the code in V2, I see that PDFStreamEngine.processOperator() >> used to call processor.setContext(this). That sets the context.resources to >> a non null value. >> >> In V3, there is no call to processor.SetContext(this). >> >> Can anyone the context of why this was removed, and the expectations for >> user code to resolve the null reference? >> >> Many thanks >> >> Mark >>
