Mserv: http://www.mserv.org


I've written a replacement for the mserv filter mechanism.  It's
backward-compatible with the old filter (cross my fingers) , but
adds some new features.

Overview:
- The old filter re-parsed the command line for every song it filtered.
  The new one pre-parses the command line, building a parse tree and
  that is utilized when evaluating each song.  Thus, it's faster
  when processing huge numbers of songs.
- The new filter uses short-circuit logic to improve performance.
- The new filter has some additional syntax.
- The old filter only had "search=" which did substring searches on
  title and author, the new one adds "author=" and "title=" which
  do exact matches.  For example, "searchf author=bond" will get
  my Bond sountrack but not pick up other songs with "bond" somewhere
  in the song name.  The compare does skip "The " at the start of
  songs, though, so "searchf author=beatles" will match both "Beatles"
  and "The Beatles".  It does support a wildcard '*' at the end of a
  string (but not the middle).
- The new filter supports arbitrary length, whereas the old one
  used arrays and therefore had some limits.
- Whitespace is allowed, allowing expressions to be more legible.

Details:

Operator & precedence (unchanged except new =~):
  () (highest)
  comparison operators: = == < > <= >= =~
  !
  |
  & (lowest)

Evaluation is left to right.

The new =~ is a substring search operator (strings only).  The rest of
the operators work on both strings (new feature) and numbers.
Both = and == are equivalence operators.

Ratings:
superb, good, neutral, bad, awful

Quoting:
The RHS of comparisons can now be quoted.  If the next non-whitespace
character after the operator is a single or double quote, then the
string is collected up to the close quote, which must be the
same type as the open quote.

Keywords:
--Binary--
HEARD
RATED
<rating>
	True if the song has been heard, rated, or given the specific
	rating (respectively) by any currently logged in user.
	(Conforms to existing behavior.)

PLAYED
	True if the song has ever been played.  Equivalent to
	LASTPLAY > 0.

--String--
<user>=<rating|HEARD|RATED>
	True if <user> has assigned <rating>, if the user has heard
	a song, or if the user has rated the song respectively.
	Rating may be one of superb, good, neutral, bad, or awful.

GENRE [ = =~ ] <genre>
	True if a song belongs to <genre>.  Note that the old mechanism
	expected multiple genres to be separated by commas, the new one
	also recognizes slashes and plus (which is how a many are listed
	on FreeDb and CDDB).  =~ will do a substring match rather than
	matching genre name - so use this if you want to do something
	like GENRE =~ "ROCK/POP".

SEARCH [ = =~ ] <pattern>
	True if there is a substring match on title or author.  SEARCH
	doesn't support other string comparisons.

ALBUMNAME <cmp> pattern
AUTHOR <cmp> pattern
TITLE <cmp> pattern
	Compare the various components to the pattern.  <cmp>
	can be any of the comparison operators.

--Numeric--
ALBUM <cmp> number
	Compare album number.  For numeric comparison, any operator
	except the substring search (=~) is applicable.

DURATION <cmp> number
	Compare duration of song, in seconds.

LASTPLAY <cmp> number - last played in hours
	Compare last played time, in hours.

TRACK <cmp> number
	Compare track number.

YEAR <cmp> number
	Compare year.  If the year isn't set, year <cmp> number returns
	false.  (That is, unset years aren't equal to year = 0.)

Thanks:
Thanks to Kimmo Suominen for documenting the existing syntax at
http://kimmo.suominen.com/docs/mserv-filters/
It was _very_ helpful along the way.

--
Do I suffer from long term memory loss?...
I don't remember.