I get this on some Office docs, a specific PPT doc to reproduce this is attached. What I do is this: POIFSFileSystem filesystem = new POIFSFileSystem(fis); SummaryInformation si = (SummaryInformation) getPropertySet(filesystem, SummaryInformation.DEFAULT_STREAM_NAME, sourcePath); where getPropertySet does this: public static PropertySet getPropertySet(POIFSFileSystem filesystem, String setName, String filepath) throws IOException { DocumentInputStream dis = filesystem.createDocumentInputStream(setName); return PropertySetFactory.create(dis); .... This causes the following exception: org.apache.poi.hpsf.NoPropertySetStreamException at org.apache.poi.hpsf.PropertySet.<init>(PropertySet.java:252) at org.apache.poi.hpsf.PropertySetFactory.create(PropertySetFactory.java:61) The file doesn't appear corrupt as it opens in PPT just fine. Also, I dumped out the directory nodes and I see that summary info is in fact there.
I cannot look into this issue without a document exhibiting the faulty behaviour.
HPSF now supports property sets without any section. However, such a property set cannot be a summary information or document summary information, because the latters' identification is expected to be in the first section.