[ https://issues.apache.org/jira/browse/TIKA-662?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Maxim Valyanskiy updated TIKA-662: ---------------------------------- Attachment: TIKA-662.patch patch > Prevent creating of ZipInputStreamZipEntrySource when reading files from disk > ----------------------------------------------------------------------------- > > Key: TIKA-662 > URL: https://issues.apache.org/jira/browse/TIKA-662 > Project: Tika > Issue Type: Improvement > Reporter: Maxim Valyanskiy > Attachments: TIKA-662.patch > > > POI provides two ways to open OPCPackage - via InputStream and via File. > Creating OPCPackage from InputStream casuses creation of > ZipInputStreamZipEntrySource, that buffers all uncompressed data in memory. > This takes a lot of memory and it is not needed when we are reading files > from disk or when we already copied stream into temporary file. > This patch removes usage of ZipInputStreamZipEntrySource in this case. > Unfortunately, it breaks ZIP-bomb prevention for OOXML parser (and other > parsers that uses TikaInputStream.getFile()). I think that ZIP-bomb > prevention should be additionally implemented for that formats before > committing this to SVN. -- This message is automatically generated by JIRA. For more information on JIRA, see: http://www.atlassian.com/software/jira