#####################################################################################
# TextGrid preparation script: Written by Kyuchul Yoon ( kyoon@ling.osu.edu )
# Help from Mary Beckman ( mbeckman@ling.osu.edu ) with inserting break index tier, etc.
# Script should be in the parent directory where 30[23]00[024].txt/ folders belong.
# Two subdirectories should be csplit-rom/ and TextGrids/.
# The script assumes that you already have the romanized files for hangul sentences.
# The romanized files should contain romanized eojeols that are "carriage-return"-separated.
# The script will read in all the .rom files one by one from csplit-rom/ directory
# and write default TextGrid files into TextGrids/ directory.
# The filename of the output TextGrid files are ...rom.TextGrid
#####################################################################################
# Eojeols in word tier will be placed in the right side of the TextGrid depending on the factorOfDistance
# so that moving boundaries (Shift + Dragging) may be easy later on in another script
# If factorOfDistance is 2, eojeols will be in the right half of TextGrid.  If 3, in the right two thirds.  If 5, in the right four fifths.
# The variable factorOfDistance added for different screen resolutions.  If your screen is smaller than 1024x768, 4 or bigger is good
#####################################################################################

# Specify files and folders
form Select files
	word longSoundFolder_(should_be_a_subfolder) 302000.txt
	word longSoundFile_(without_extension) 302000
	word romFilesSubFolder csplit-rom
	word fileExtensionOfRom_(without_dot) rom
	word textGridSubFolder TextGrid
	integer factorOfDistance 4
endform

Open long sound file... 'longSoundFolder$'\'longSoundFile$'.wav

# Default end time
end = 0

# Get .rom files for loop
Create Strings as file list... fileList 'longSoundFolder$'\'romFilesSubFolder$'\*.'fileExtensionOfRom$'
numFiles = Get number of strings
for iFile to numFiles
	select Strings fileList
	# Get the name for .rom file
	romFile$ = Get string... iFile
	select LongSound 'longSoundFile$'
	View
	editor LongSound 'longSoundFile$'
		roughUttDur = end + 40
		Select... end roughUttDur
		Zoom to selection
		# Get start and end time of the section that you want to create TextGrid 
		# for by prompting
		pause Pick the beginning of utterance
		start = Get cursor
		pause Pick the end of utterance
		end = Get cursor
	endeditor

# Extract that part from the long sound
	Extract part... 'start' 'end' yes
	# Calculate the length of the extracted segment
	lengthSubPart = end - start
	Rename... subpart

# Create a TextGrid for that subpart sound, remove the subpart sound
	To TextGrid... "word phonology phonetics break-index misc" phonology phonetics break-index
	# Sound "subpart" was just for creating a TextGrid file
	select Sound subpart
	Remove

# Get the number of strings (newline-separated eojeols) and create interval tiers
	Read Strings from raw text file... 'longSoundFolder$'\'romFilesSubFolder$'\'romFile$'
	Rename... rom
	numEojeols = Get number of strings
	# Insert an interval tier 
	select TextGrid subpart
	for i from 1 to (numEojeols+1)
		# Place eojeols back by the factorOfDistance so moving points may be easy
		pseudoStart = start + lengthSubPart/factorOfDistance
		pseudoLengthSubPart = lengthSubPart * (1 - 1/factorOfDistance)
		nthBoundary = pseudoStart + (pseudoLengthSubPart/(numEojeols+2))*i
		Insert boundary... 1 'nthBoundary'
	endfor

	# and set texts for each interval
	numIntervals = Get number of intervals... 1
	# The first and last interval will be <sil>
	Set interval text... 1 1 <sil>
	Set interval text... 1 numIntervals <sil>
	# Set the rest of the intervals
	for i from 2 to numIntervals-1
		select Strings rom
		eojeolFromRom$ = Get string... 'i'-1
		select TextGrid subpart
		Set interval text... 1 'i' 'eojeolFromRom$'
	endfor

# Create as many tiers as you want: For break index

	numEojeols = numEojeols + 1
	breakTime = Get end point... 1 1
	Insert point... 4 breakTime <sil>

	# Loop through the eojeols in the word tier inserting a break index with BI value "2" after each one
	for i from 2 to numEojeols
		breakTime = Get end point... 1 i
		Insert point... 4 breakTime 2
	endfor

# For phonology tier

	for i from 2 to numEojeols
		pointTime = Get end point... 1 i
		Insert point... 2 pointTime LHa
	endfor

# Save the TextGrid file
	select TextGrid subpart
	Write to text file... 'longSoundFolder$'\'textGridSubFolder$'\'romFile$'.TextGrid

# Select both the long sound and the TextGrid and present for verification
	select LongSound 'longSoundFile$'
	plus TextGrid subpart
	Edit
	# Waits for user verification
	pause Click OK if TextGrid looks fine
	# Remove current TextGrid and move on
	select TextGrid subpart
	Remove
	# Close the long sound file editor window
	select LongSound 'longSoundFile$'
	editor LongSound 'longSoundFile$'
		Close
	endeditor

# clean up
	select Strings rom
	Remove

# end of loop
endfor

select Strings fileList
plus LongSound 'longSoundFile$'
Remove

####### END OF SCRIPT #######