All, For some recent work on Apache Tika, I used commons-compress to extract entry names and metadata via a streaming read from roughly 500k zip-based files we have in Tika's regression corpus. I was happy to see we have some POI-generated files in there. :) I noticed some areas for improvement in Tika's coverage of detection of zip based files, and I noticed that MSOffice OOXMLs nearly always place the [Content_Types].xml file as the first physical entry...which we could use on Tika to help improve speed of detection. Because others may have other uses for this data, I'm sharing the key/value table here: http://162.242.228.174/share/zips.txt.gz See https://issues.apache.org/jira/browse/TIKA-2849 for some discussion. The file is 5GB uncompressed.
Let me know if you have any questions and/or if this is of any interest. Many thanks to commons-compress! Cheers, Tim