[ https://issues.apache.org/jira/browse/COMPRESS-352?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15271884#comment-15271884 ]
Stefan Bodewig commented on COMPRESS-352: ----------------------------------------- This is ugly. The missing stream identifier defeats autodetection, therefore Tika will need to detect it based on extension and then use {{FramedSnappyCompressorInputStream}} directly - likely using a different constructor that we'll need to add in order to skip the stream identifier and deal with the missing checksums. https://github.com/obriensp/iWorkFileFormat/blob/master/Docs/index.md#iwa says the checksum was only missing from the compressed chunks. I'll give it a try starting with one of Tika's test files as test case for us as well, but can't promise any time frame. > Add snappy decompressor that handles Apple .iwa files > ----------------------------------------------------- > > Key: COMPRESS-352 > URL: https://issues.apache.org/jira/browse/COMPRESS-352 > Project: Commons Compress > Issue Type: New Feature > Reporter: Tim Allison > > On TIKA-1966, one of our users recently shared the newest iOS iWorks files. > They've gotten rid of internal .zip and now rely exclusively on .iwa files. > commons-compress doesn't currently appear to handle these > (snappy-compressed?) files...probably because: > bq. However, the variant of Snappy that is used does not comply with the spec > for that format, omitting the stream identifier and checksum. > from http://fileformats.archiveteam.org/wiki/IWA > Any chance you'd be able to modify the snappy decompression to handle Apple's > variant? -- This message was sent by Atlassian JIRA (v6.3.4#6332)