disk-indexer
------------

Server to index disks.

Environment

 -d path	Path to database.

Start Server

 All known disks will be (re)scanned and updated to reflect their current
 collection of music files.

indexer-cmd
-----------

Send commands to disk indexer.

 check		Perform some consistency checks on the database.

 shuffle	(re)create the shuffle playlist.

 add uuid path	Add a new disk or rescan an existing disk.

 quit		Tell the server to shut down.

disk-util
---------

Environment & Object Selectors

 -d path	Path to database.

 -f fid		File ID
 -d diskid	Disk ID
 -s seq		Sequence Number

Commands

 shuffle	Shuffle all files to the 'shuffle' playlist
 file-dump	noop
 files		Dump files
 lists		Show list names
 disks		Show disk names

Design Thoughts
--------------

Overall playlist behaviour?

 - priority is
  - play now file
  - jukebox playlist (add to playing)
  - requested playlist
  - default playlist

 - junk playlist - or junk flag?

System playlists

* default / all:shuffle
  All tracks, (shuffled?).  The default playlist.

* queue
  User selected tracks, in order.

* junk
  The junk list, for later operations

* user list
  A user playlist

Playing State

The play state is stored for each list.  There should be only one (if
any) with the 'state' value non-zero.

ISSUE: How to resolve a user-selected playlist if that is active since
       everything else is pre-known.

Operations on Search

* Play Now
  Adds to the play queue and switches to it it isn't started.

* Add to List
  Add to specified playlist.

Operations on Coming Up

* Play Now
  Jumps to track

* Junk
  Add to junk list (or should it be explicit?)

* Remove (user playlist, queue)
  Remove from playlist


internals
---------

Database implemented in lmdb using ez-blob for serialisation.

pseudo-SQL equivalent.

All primary keys are 'auto increment start=1'.

create table disk (
       id int rimary key,
       uuid text unique,
       label text,
       type text
       mount text
)

create table file (
       id int primary key,
       diskid int not null references disk(id),
       size bigint,
       mtime bigint,
       duration bigint,

       path text unique,
       dir text,
       title text,
       artist text,
)

create index file_by_title on file(dir);
create index file_by_title on file(title);
create index file_by_title on file(artist);

create table list (
       id int primary key,
       int size,
       name text unique,
       desc text
)

create table file_by_list (
       listid int not null references list(id),
       seq int not null,
       fileid int not null,
       unique (listid, seq)
)

create table list_by_file (
       fileid int not null references file(id),
       listid int not null references list(id),
       seq int not null,
       unique (listid, seq)
)