I was just trying to convert a PPT using the following URL http://code.google.com/p/qifei/wiki/PDFConverter python code
I could see the same thing happening with the command line option too
python documentconverter.py /home/rajeev/Desktop/Downloads/Industry2.ppt /home/rajeev/Desktop/test.pdf
It appears that the image overlaps on some text in some cases.Are there any work arounds for this.
Mangled images are probably a function of the pdf converter (in this case, Open Office). Try using a different library like pyPDF.