The search_elastic app adds a full text search for files stored in ownCloud. It requires an elasticsearch server and can index all files supported by apache tika, e.g., plain text, .docx, .xlsx, .pptx, .odt, .ods and .pdf files. The source code is available on GitHub. For more information please read the documentation at https://doc.owncloud.com.
- update elastica lib
- restore compatibility with oc8.2
- restore compatibility with elasticsearch (requires new indexes)
- store groups and users with access to filter search results by group membership
- store fileid instead of filenames so we don't have to handle renames
- use instanceid to set up index - allows using the same elasticsearch instance for multiple oc instances
- store the filename to allow faster search in shared files
- index files and folders
- store tags?
- store image / video dimensions?
- sharing a file immediately after it has been uploaded throws an exception
- fix exception / do not try to update a nonexistent document
- get all users and groups when initially indexing the document
- move share updates to background job -> eventually searchable
- descend subdirs when updating
- check permissions again on search and remove results if no longer accessible
- compensate for removed entries in search results, too many will confuse the paging logic
- --index in batches (make batch size configurable, 0 = unlimited)
- CLI cron.php executes all jobs
- limit number of files to 250? per job?
- CLI cron.php executes all jobs
- add occ commands
- index all files or only those of a specific user
- enable / disable automatic background scanning via cron
- admin settings ui for this
- check js for result link handling so clicking a result dos not do a full page load, there seems to be js in place that already does the file highlighting
- the old filehandler logic does not seem to work, removed it, now using plain link
- send code snippets for search_lucene
- use file tab
- show index status
- remember index error message in db
- check encryption compatibility
- had to jump a few hoops to get master key working
- not compatible with user individual keys
- at least index metadata in this case (catch encryption exception and ignore content extraction)
- statistics on admin settings page
- statistics on personal settings page
- cleanup code
- port test suite from search_lucene
- resolve path for shared files
- files with empty content extraction are reindexed indefinitely? e.g., empty text file
- more debug logging
- wildcard search ... but there is a bug in core js code preventing wildcard search
- well partly. * and ? are no longer supported. Instead, we now mimic core, which is called a match phrase prefix type query
- to find out why a node cannot be found by its contents mark it as "NO CONTENT EXTRACTED"?
- how should we handle files in userhome/files_versions/ or userhome/thumbnails/ ... currently a 'vanished' message will be logged ... annoying