Péter Szabó

Actually, the change was introduced between version 4.0.0 and 4.0.3. I've updated my original post with this finding.

It looks like we (people who have commented on this issue so far) don't have a good understanding on what has changed between jpegtran 4.0.0 and 4.0.3 causing the output...

@FredWahl: > I think the jpeg format is complicated. [..] I have not been able to find a good open source tool that can repair jpeg files. [...] It doesn’t...

@zvezdochiot say: > > Unfortunately I don't have a Windows machine, so I'm not able to run jpegsnoop, and it's also a GUI program, so it's not easy to copy-paste...

Thank you for submitting this issue! Pull requests are welcome.

Thank you for reporting this issue. To investigate further, I need the original `.pdf` file.

FYI I was able to reproduce the problem with https://www.oreilly.com/programming/free/files/functional-programming-python.pdf . This looks like a bug in pdfsizeopt. The problem is caused by Type1 fonts with more than 256 glyphs...

@pozitron57: Please upload the corresponding PDF files produced by *pdflatex* and *dvipdfm*. I am not able to help you directly until you upload the sample PDF files.

Here is another PDF file which triggers the bug on a different code path: [lme2006_logomany.pdf](https://github.com/pts/pdfsizeopt/files/1083252/lme2006_logomany.pdf) and [tosubset.pdf](https://github.com/pts/pdfsizeopt/files/1162806/tosubset.pdf). * lme2006_logomany.pdf and tosubset.pdf: ConvertType1FontsToType1C calls GenerateType1CFontsFromType1 calls GetFonts raises *duplicate font objs*...

I have added some workarounds (as of 85e1f76a9e610e9922365a4af583f8788f655311) to prevent the `AssertionError: found ... duplicate font obj` error from happening, by making pdfsizeopt ignore Type 1 and Type1C fonts with...