####################################################################################################
# diphone-locator-for-TextGrids-GREEDY.ALGORITHM.praat ( Written by Kyuchul Yoon kyoon@ling.osu.edu )
# Same function as diphone-locator-for-TextGrids9.praat
#	but with Greedy Algorithm
##################################################################################

#################### BASIC ALGORITHM #############################################
# Given 400 TextGrids files with diphones tier, this script makes a list of
# filename-diphoneTypeCount, remembering all the diphoneTypes and their 
# total count for each file, gets the name of the file that has the most
# diphoneTypeCount, marks all the diphones for that file (and moving the file to a 
# different 'done' folder), removes all those diphones from the other 399 files
# (deleting old 399 files and writing the updated files to the same directory), 
# and repeat the above procedures until there are no more remaining diphoneTypes.
#
# This entails, for each TextGrid file, that the script
#	(1)	remembers all distinct diphoneTypes, storing in an array variable
#		so that those can be removed from the rest of the files in the loop
#	(2)	remembers the number of those diphoneTypes so that comparison
#		among files can be possible.  This comparison is done to choose 
#		the file with the most diphoneTypes.
##################################################################################

form Specify forms and files
	word inFolder lab.TextGrid.after.LTS.scheme.diphone.textgrid
	natural diphoneTier 5
	word inFileExt_(with_dot) .TextGrid.prosodic.diphone
	word doneFolder_(to_be_created) marked.diphones.GREEDY.ALGORITHM
	word diphoneCount_(to_be_created) DIPHONE.COUNT
endform

# Make a list of all inFolder files
Create Strings as file list... fileList 'inFolder$'\*'inFileExt$'
numFiles = Get number of strings
pause 'numFiles' files identified. Continue?

#################################################
# FIND THE FILE WITH THE MOST DIPHONE TYPE COUNT
#################################################

# Initialize the mostTotalNumDiphoneTypes to zero
mostTotalNumDiphoneTypes = 0

# Loop through each file
for iFile to numFiles
	# Read a file
	select Strings fileList
	fileName$ = Get string... iFile
	Read from file... 'inFolder$'\'fileName$'
	Rename... inFile

	# Check the number of intervals for the diphone tier and set the total number
	numIntervals = Get number of intervals... diphoneTier
	totalNumDiphoneTypes = numIntervals - 2

	# Deal with the intervals with nothing and set the totalNumDiphoneTypes accordingly
	for iIntervalNoText from 2 to (numIntervals-1)
		intervalTextNoText$ = Get label of interval... iIntervalNoText
		lenNoText = length(intervalTextNoText$)
		if lenNoText = 0
			totalNumDiphoneTypes = totalNumDiphoneTypes - 1
		endif
	endfor

	# Loop through each interval
	for iInterval from 2 to (numIntervals-1)
		intervalText$ = Get label of interval... iInterval
		arrayIntervalText'iInterval'$ = intervalText$

		# Do the comparison with the rest of the interval texts if iInterval > 2
		if iInterval > 2
			# Compare the current with all arrayIntervalText$ variables
			for iCompare from 2 to (iInterval-1)
				# If the current is the same as the immediately preceding one,
				# then set the current interval text to nothing and reduce the
				# number of total diphone types by one
				if arrayIntervalText'iInterval'$ = arrayIntervalText'iCompare'$
					Set interval text... diphoneTier iInterval 
					totalNumDiphoneTypes = totalNumDiphoneTypes - 1
				endif
			endfor
		endif
	endfor

	# Delete the old file and write the new one
	delete 'inFolder$'\'fileName$'
	Write to text file... 'inFolder$'\'fileName$'
	select TextGrid inFile
	Remove

	# Compare the current totalNumDiphoneTypes with the previous mostTotalNumDiphoneTypes
	# and set the filename and number of diphone types for that file
	if totalNumDiphoneTypes > mostTotalNumDiphoneTypes
		mostDiphoneTypes = totalNumDiphoneTypes
		mostDiphoneTypesFileName$ = fileName$
	endif
endfor

###############################################
# MOVE THE FILE TO DONE FOLDER and
# REMOVE THE DIPHONE TYPES FROM THE OTHER FILES
###############################################

# Now that we have the filename and the number of diphone types for the file
# that has the most number of diphone types, move that file to the 'done' folder