|Subversion||:||Trunk at http://downtown.trilo.de/svn/wakati/trunk (~70 MB)|
Please note: Wakati's status is infant, so below text describes more what it shall become, than what it already is.
Wakati is a Python program to scan a drive for timestamps, either found in the filenames or in the file contents. The timestamps are put into a CouchDB database to create a diary. Also the findings are classified in several categories, e.g. image, download, web-archives.
So from that diary, several albums can be created automatically, e.g. of course a photo album, a surf album, a download album.
The first features exist as a proof of concept. The features to make it finally applied usefully are still missing.
Existing feature: It scans the drive and writes into a CouchDB.
Existing feature: It so far detects timestamps only in filenames, not content, and only in the format we use on our pages, e.g. "20100901°1331".
Missing feature: To be useful for drives in general, it has to detect timestamps more intelligently from any kind of formats or even from plain hints.
Missing feature: An user interface to the database. For now, there is only the CouchDB built-in user interface, which is enough for debugging, but too sparse to work usefully with the data.
Missing feature: Scan not only local drives but URLs as well.
The scan engine has interesting properties, as opposed to a simple directory listing generator:
As it is a typical concept of 'No-SQL' databases, this one is made to handle big amounts of data, trading off consistency thereof. The data consistency is only garanteed at the head of the scan. What happens to a live file after it was scanned once, is only recognized by the database, when the scanner comes over it in the next round, whenever this will be.
So it's behaviour is more like a search engine spider bot than a directory listing generator.
. . .
. . .