Finds duplicate songs in a selection in itunes

itunes_dedupe.applescript

-- Author: Ryan Jennings <robert@ryanjennin.gs>
-- Date: Aug 2, 2019
-- Description: 
--    Finds duplicate songs in a selection in itunes
--    Ranked meta data is:
--       1. song name
--       2. artist
--       3. bitrate (ex 320 over 256)
--       4. duration (ex keep 12:00 as too different from 4:40, but discard 3:50)

use AppleScript

use scripting additions

use application "iTunes"

-- utility handlers for various tasks
script util
	
	-- quick sort implementation
	on sort(aList as list, Le as integer, Ri as integer, comparator)
		
		set [i, j] to [Le, Ri]
		
		set v to item ((Le + Ri) div 2) in aList --> pivot in middle (as C.A.R. Hoare's algorithm)
		repeat while j > i
			repeat while comparator's compare(item i in aList, v) < 0
				set i to i + 1
			end repeat
			repeat while comparator's compare(item j in aList, v) > 0
				set j to j - 1
			end repeat
			if not i > j then
				set temp to item i in aList
				set item i in aList to item j in aList
				set item j in aList to temp
				set [i, j] to [i + 1, j - 1]
			end if
		end repeat
		if Le < j then sort(aList, Le, j, comparator)
		if Ri > i then sort(aList, i, Ri, comparator)
	end sort
	
	-- slits text into a list based on a delimiter
	on split(input as text, delimiter as text)
		set AppleScript's text item delimiters to delimiter
		set value to every text item of input
		set AppleScript's text item delimiters to ""
		return value
	end split
	
	-- cleans text into a more comparable form
	on sanitize(input as text)
		
		set value to ""
		
		repeat with ch in input
			set c to id of ch
			
			-- skip control characters
			if c > 32 then
				
				-- convert uppercase
				if c ≥ 65 and c ≤ 90 then
					-- to lowercase
					set c to c + 32
				end if
				
				set value to value & ((character id c) as string)
			end if
			
		end repeat
		
		return value
	end sanitize
	
	
end script

-- logging related handlers
script logger
	-- the log levels
	property levels : {"trace", "debug", "warn", "fatal", "info"}
	-- the current log level
	property level : 2
	
	on trace(message)
		if level ≤ 1 then log (message)
	end trace
	
	on debug(message)
		if level ≤ 2 then log (message)
	end debug
	
	on warn(message)
		if level ≤ 3 then log (message)
	end warn
	
	on fatal(message)
		if level ≤ 4 then log (message)
	end fatal
	
	on info(message)
		if level ≤ 5 then log (message)
	end info
	
	on dheader(message)
		my debug("")
		my debug(message)
		my debug("=========================================================================")
	end dheader
end script

-- handlers related to meta data
script metadata
	
	-- returns a description of a song meta data
	on desc(value)
		return (n in value) & " by " & (a in value) & " (" & (b in value) & "," & (t in value) & ")"
	end desc
	
	-- handlers related to comparing meta data
	script comparator
		
		-- compare length of a song
		on len(m1, m2)
			
			-- get the time fields
			set t1 to t of m1
			set t2 to t of m2
			
			-- split the string
			set t1 to my util's split(t1, ":")
			set t2 to my util's split(t2, ":")
			
			-- get the minutes fields
			set n1 to item 1 of t1 as number
			set n2 to item 1 of t2 as number
			
			-- get the minute second fields
			set s1 to item 2 of t1 as number
			set s2 to item 2 of t1 as number
			
			-- convert into unix timestamps
			set t1 to (n1 * 60) + s1
			set t2 to (n2 * 60) + s2
			
			return t1 - t2
			
		end len
		
		-- compare song bitrates
		on bitrate(m1, m2)
			
			-- get the values
			set b1 to b of m1
			set b2 to b of m2
			
			return b1 - b2
		end bitrate
		
		-- default compare method for meta data
		on compare(m1, m2)
			
			-- get the song name
			set n1 to n of m1
			set n2 to n of m2
			
			if n1 is equal to n2 then
				
				-- get the artist name
				set a1 to a of m1
				set a2 to a of m2
				
				if a1 is equal to a2 then
					
					-- compare bitrate next
					set diff to bitrate(m1, m2)
					
					if not diff is equal to 0 then return diff
					
					-- finally compare time duration
					return len(m1, m2)
				end if
				
				return 0
				
			end if
			
			return 0
			
		end compare
		
	end script
	
end script

-- handlers related to a list of song meta data
script metalist
	
	-- prunes a list of meta data. assumes sorted.
	on prune(input as list)
		
		my logger's dheader("Pruning song list")
		
		set output to {}
		set pruned to 0
		set remainder to 0
		
		-- keep track of output
		script outputer
			
			-- the previous sorted meta data
			property previous : {a:""}
			
			-- increment the duplicates output
			on inc(dup)
				set output to output & {dup}
				set previous to dup
				set remainder to remainder + 1
			end inc
			
			-- decrement the duplicates output
			on dec()
				set pruned to pruned + 1
			end dec
			
			-- is a new sorted duplicate artist?
			on is_next(dup)
				return not ((a in previous) is equal to (a in dup))
			end is_next
			
			-- compare this duplicate with previous in sorted list
			on diff(dup)
				return metadata's comparator's compare(previous, dup)
			end diff
		end script
		
		-- for each duplicate song...
		repeat with dup in input
			
			-- if its a new artist
			if outputer's is_next(dup) then
				-- and add to output
				outputer's inc(dup)
			else
				-- otherwise compare for this artist only
				set diff to outputer's diff(dup)
				
				-- if the compare difference is large and negative
				if diff is less than -60 then
					
					-- add to duplicates output
					outputer's inc(dup)
					
				else
					-- otherwise it is pruned from duplicates
					outputer's dec()
					
					my logger's trace("Pruned " & metadata's desc(dup))
				end if
			end if
			
		end repeat
		
		my logger's debug("Pruned " & pruned & " and kept " & remainder)
		
		return output
		
	end prune
	
	
end script

-- handlers related to a list of songs
script songlist
	-- the list of songs
	property values : missing value
	
	-- sorts the values
	on sorted()
		set size to count of values
		if meta_size is less than or equal to 1 then return values
		my util's sort(value, 1, size, metadata's comparator)
		return meta_data
	end sorted
	
	-- builds a list of meta data
	on build(input)
		
		set values to input
		
		set output to {}
		
		my logger's dheader("Building song list")
		
		repeat with curr_index from 1 to (count of values)
			set current to item curr_index in values
			
			-- get the song name
			set curr_name to name of current
			set curr_name to my util's sanitize(curr_name)
			
			-- get the artist name		
			set curr_artist to artist of current
			set curr_artist to my util's sanitize(curr_artist)
			
			set curr_bitrate to bit rate of current
			set curr_time to time of current
			
			-- create an info list
			set curr_info to {i:curr_index, a:curr_artist, n:curr_name, b:curr_bitrate, t:curr_time}
			
			set output to (output & {curr_info})
			
			my logger's trace("Found " & my metadata's desc(curr_info))
			
		end repeat
		
		my logger's debug("Found " & (count of output) & " selected songs")
		
		return output
	end build
	
end script


-- the main script
script deduper
	-- name of the playlist to move duplicates to
	property playlist_name : missing value
	
	--  the list of songs in itunes
	property song_list : missing value
	
	-- move a list of duplicates to the playlist
	on to_playlist(duplicates as list)
		
		my logger's dheader("Processing duplicates and setting playlist")
		
		-- first create or clear the playlist
		my create_playlist()
		
		repeat with dup in duplicates
			
			set dup_index to i of dup
			
			set current_song to item dup_index of song_list
			
			duplicate current_song to user playlist playlist_name
			
			my logger's trace("Duplicating " & metadata's desc(dup) & " to playlist")
			
		end repeat
	end to_playlist
	
	
	-- creates the playlist
	on create_playlist()
		
		if playlist playlist_name exists then
			my logger's trace("Clearing tracks in playlist")
			delete every track of user playlist playlist_name
		else
			my logger's trace("Creating playlist \"" & playlist_name & "\"")
			make new user playlist with properties {name:playlist_name}
		end if
	end create_playlist
	
	-- runs the deduper on a song list for a play list
	on dedupe(s_list, pl_name)
		
		set song_list to s_list
		
		set playlist_name to pl_name
		
		my logger's dheader("Starting dedupe")
		
		my logger's warn("It is recommended to run on selected items with \"Show Duplicate Items\" enabled.  Then remove the songs from the library from the generated playlist")
		
		-- first build a list of song meta data
		set output to my songlist's build(song_list)
		
		-- now prune any duplicates we want to keep
		set output to my metalist's prune(output)
		
		-- finally output to the playlist
		my to_playlist(output, song_list)
		
		my logger's debug("Done, identified " & (count of output) & " duplicates for removal")
	end dedupe
	
end script


on run
	
	tell application "iTunes"
		
		if not running then
			display alert "iTunes is not running" message "Please activate iTunes and select items."
			return
		end if
		
		set DupPlaylistName to "Found Duplicates"
		
		set sl to selection
		
		my deduper's dedupe(sl, DupPlaylistName)
		
	end tell
	
end run