File support

[Floppy]

Command line should accept:

• ID
• Metadata list
• File

File hash should be stored in DLT, along with the metadata and ID.

Currently, we're just hashing the metadata, not a separate file. That's the bit that needs adding.

[Floppy]

Could use DROID to do the hashing of the file, which is a National Archive tool for creating IDs of digital records: https://www.nationalarchives.gov.uk/information-management/manage-information/policy-process/digital-continuity/file-profiling-tool-droid/

[Floppy]

We can use DROID locally (it's a java app I think), or we could wrap it in a simple API and deploy it somewhere. Maybe.

[Floppy]

DROID might be difficult actually, as it looks like a GUI tool. Not sure if there's a version we can call directly somehow.

[jezhiggins]

Using DROID sounds sensible.

The user guide is all GUI GUI GUI, but there is a command line driver for it too - see http://digital-preservation.github.io/droid/

[Floppy]

ah, brilliant 😄

[jezhiggins]

Spent a little bit of time working with DROID. It turns out it isn't what we think it is. It isn't a tool for generating file signatures, it's a file format identification tool. If you give it a load of files, it will chew through them and spit out a list of filenames and file types (those types being in the form of references to their PRENOM database http://www.nationalarchives.gov.uk/PRONOM/BasicSearch/proBasicSearch.aspx)

The command line usage is straightforward enough (if a little chatty and with Spring logging splurting out on stderr) and the output is amenable to parsing, but it does need a bit of management. The signature files it uses need to be downloaded on first use and periodically refreshed. It's not immediately clear if we can control where those files get downloaded to (it's dropping them under ~/.droid6/...). There are some other little funnies too - it has a option to list the downloaded signature files, giving their filenames, but when specifying the signature to use you have to provide the full path name, which obviously needs a bit of meta-knowledge.

DROID clearly has a role to play then, but I'm now might not be the time :)