#####################################################################################
# Phonological tone tier infor extractor for Wagon: Written by Kyuchul Yoon ( kyoon@ling.osu.edu )
# Extracts from a set of .TextGrid.done files data fields for Wagon training
# Data fields include  (1) romanized eojeols, (2) number of syllables for each eojeol, 
#		  (3) type of boundary (AP, IP, or none)
# 		  (4) number of syllables from an immediately preceding AP boundary (marked by "LHa")
#		  (5) number of syllables from an immediately preceding IP boundary (marked by "HL%", "H%", or "L%"
#		  (6) number of syllables from the sentence beginning, and 
#???????????????  (7) number of syllables from the sentence ending.
# The script assumes that you already have the TextGrid files labelled by professional K-ToBI labelers.
# The script will read in all the TextGrid.done files one by one from the current directory
# and write the output files into Wagon/ subdirectory. (There should be a "Wagon" subdirectory)
# The filename of the output files are .wagon
#####################################################################################

# Specify files and folders
form Select files
	word fileExtOfDoneFiles TextGrid.done
	word outputSubFolder Wagon
endform

# Get the list of filenames of TextGrid.done files
Create Strings as file list... fileList *.'fileExtOfDoneFiles$'
numFiles = Get number of strings

# Loop throught each file
for iFile to numFiles
	select Strings fileList
	# Get the name for a TextGrid.done file
	doneFile$ = Get string... iFile
	Read from file... 'doneFile$'

	# If an old output file exists, delete it first and then write out the new file
	filedelete 'outputSubFolder'\'doneFile$'.wagon

	# Count the number of syllables for that eojeol (e.g. "peu-lang-seu" is 3-syllable long)
	# Initialize the distance (in number of syllables) from an immediately 
	# preceding AP, IP, and the sentence beginning to 1, 0, 0, and 0, respectively
	distFromPrevAP = 0
	distFromPrevIP = 0
	distFromSentBegin = 0

	# Get the number of intervals from the first tier (word tier), i.e. get the number of eojeols (=numIntervals)
	numIntervals = Get number of intervals... 1
	# Get rid of the leading/following <sil> label
	newNumIntervals = numIntervals - 2

	# Loop through each interval (eojeol) and extract info
	for iEojeol to newNumIntervals
		# Initialize the number of syllable for the eojeol
		actualNumSyl = 1  
		# Get the interval text (i.e., eojeol text), excluding the <sil> label
		intervalText$ = Get label of interval... 1 (iEojeol+1)
		fileappend 'outputSubFolder$'\'doneFile$'.wagon 'intervalText$''tab$'

		numTotalChar = length(intervalText$)

		# If there is no "-" in the eojeol, then the number of syllable for that eojeol is numSyl, i.e. 1 syllable.
		# Otherwise, repeat the following procedure to count the number of hyphens
		while index(intervalText$, "-")
			indexOfHyphen = index(intervalText$, "-")
			# Calculate the number of the rest of the characters starting from the location of "-"
			subNumTotalChar = numTotalChar - indexOfHyphen  
			intervalText$ = right$(intervalText$, subNumTotalChar)
			numTotalChar = length(intervalText$)
			actualNumSyl = actualNumSyl +1
		endwhile
		fileappend 'outputSubFolder$'\'doneFile$'.wagon 'actualNumSyl''tab$'

		# Get the RHS end time of the interval
		endTimeOfInterval = Get end point... 1 (iEojeol+1)

		# Get the number of points from the second, phonology, tier and loop through
		# all the points to extract the one point that "roughly" corresponds to the endTime 
		# of the interval from word tier
		# Outputs the eojeol & boundary info into a file (Feature number (1), (2), & (3))
		numPoints = Get number of points... 2
		for iPoint to numPoints
			pointLabel$ = Get label of point... 2 iPoint
			timePoint = Get time of point... 2 iPoint
			dif = abs(timePoint - endTimeOfInterval)
			if dif < 0.005
				if pointLabel$ = "LHa"
					# Compute the distance from an immediately preceding AP
					distFromPrevAP = distFromPrevAP + actualNumSyl
					# Compute the distance from an immediately preceding IP
					distFromPrevIP = distFromPrevIP + actualNumSyl
					# Compute the distance from the sentence beginning
					distFromSentBegin = distFromSentBegin + actualNumSyl
					fileappend 'outputSubFolder$'\'doneFile$'.wagon 
					... 'pointLabel$''tab$'
					... 'distFromPrevAP''tab$''distFromPrevIP''tab$'
					... 'distFromSentBegin''newline$'
					# Reset the distance from an immediately preceding AP to zero
					distFromPrevAP = 0
				elsif pointLabel$ = "HL%"
					distFromPrevAP = distFromPrevAP + actualNumSyl
					distFromPrevIP = distFromPrevIP + actualNumSyl
					distFromSentBegin = distFromSentBegin + actualNumSyl
					fileappend 'outputSubFolder$'\'doneFile$'.wagon 'pointLabel$''tab$'
					... 'distFromPrevAP''tab$''distFromPrevIP''tab$'
					... 'distFromSentBegin''newline$'
					# Reset the distance from an immediately preceding IP & AP to zero
					distFromPrevIP = 0
					distFromPrevAP = 0
				elsif pointLabel$ = "H%"
					distFromPrevAP = distFromPrevAP + actualNumSyl
					distFromPrevIP = distFromPrevIP + actualNumSyl
					distFromSentBegin = distFromSentBegin + actualNumSyl
					fileappend 'outputSubFolder$'\'doneFile$'.wagon 'pointLabel$''tab$'
					... 'distFromPrevAP''tab$''distFromPrevIP''tab$'
					... 'distFromSentBegin''newline$'	
					# Reset the distance from an immediately preceding IP & AP to zero
					distFromPrevIP = 0
					distFromPrevAP = 0
				elsif pointLabel$ = "L%"
					distFromPrevAP = distFromPrevAP + actualNumSyl
					distFromPrevIP = distFromPrevIP + actualNumSyl
					distFromSentBegin = distFromSentBegin + actualNumSyl
					fileappend 'outputSubFolder$'\'doneFile$'.wagon 'pointLabel$''tab$'
					... 'distFromPrevAP''tab$''distFromPrevIP''tab$'
					# Reset the distance from an immediately preceding IP & AP to zero
					distFromPrevIP = 0
					distFromPrevAP = 0
					... 'distFromSentBegin''newline$'
				else
					distFromPrevAP = distFromPrevAP + actualNumSyl
					distFromPrevIP = distFromPrevIP + actualNumSyl
					distFromSentBegin = distFromSentBegin + actualNumSyl
					fileappend 'outputSubFolder$'\'doneFile$'.wagon 'pointLabel$''tab$'
					... 'distFromPrevAP''tab$''distFromPrevIP''tab$'
					... 'distFromSentBegin''newline$'
				endif
			endif
		endfor
	endfor
endfor	
select Strings fileList
Remove

#### END OF SCRIPT ####