###########################################################################
# Wagon : predicted boundary inserter
# Written by Kyuchul Yoon ( kyoon@ling.osu.edu )
# Inserts boundaries (including non-boundaries) to TextGrid files that correspond to .heldfile (held-out file).
# The model (file-07.model) was created with file-07.ignore, .trainfile and tested against .heldfile, producing file-07.predicted,
#	from which the boundaries to be inserted got extracted.
# The output textgrids will enable users to see what types of errors were committed by the Wagon model (file-07.model)
# A new boundary tier (predicted) is inserted.
# Irrelevant tiers (that are ignored by the .ignore file) are not displayed.
# ignore file number:
#	(00) POS alone
#	(01) POS + PC1(ultimate)
#	(02) POS + distance	
#	(03) POS + morph.identity
#	(04) POS + PC1 + distance
#	(05) POS + PC1 + morph.identity
#	(06) POS + distance + morph.identity
#	(07) POS + PC1 + distance + morph.identity <== 8th ignore file (filename: ignore-07.ignore) performed best
# Tiers to be displayed
#	POS: -1 0 +1
#	PC1: -3 -2 -1 0 +1 +2 +3
#	distance: tok.len.in.sylls + dst.(syl,eoj).pr.AP + dst.(syl,eoj).pr.IP
#	morph.identity: itself
##########################################################################

# Specify files and folders
form Select files
	word subFolderToProcess test-TextGrid
	word fileExtOfDoneFiles wagon.36
	word predictionFile file-07.predicted
	integer wordTier 1
	integer existingBoundaryTier 2
	word tierNameToInsert predicted.boundary
	word outputFileExt predicted
	word outFolder predicted-TextGrid
endform

# Get the list of filenames of TextGrid.done files
Create Strings as file list... fileList 'subFolderToProcess$'\*.'fileExtOfDoneFiles$'
Sort
numFiles = Get number of strings
pause 'numFiles' labeled textgrids identified.  Continue?

# Read the file-07.predicted file and initialize the line counter to 1
Read Strings from raw text file... 'predictionFile$'
Rename... predictedFile
numTotalLines = Get number of strings
iLineOfPredictedFile = 1

# Loop throught each file
select Strings fileList
for iFile to numFiles
	select Strings fileList
	# Get the name for a TextGrid.done file
	doneFile$ = Get string... iFile
	Read from file... 'subFolderToProcess$'\'doneFile$'
	Rename... textgrid
	numTiers = Get number of tiers
	numIntervals = Get number of intervals... 1

	# Compose outFile name
	outFilePrefix$ = doneFile$ - fileExtOfDoneFiles$
	outFile$ = outFilePrefix$ + outputFileExt$

	## Ad-hoc removal of unnecessary tiers
	Remove tier... 41
	Remove tier... 40
	Remove tier... 39
	Remove tier... 38
	Remove tier... 37
	Remove tier... 36
	Remove tier... 35
	Remove tier... 34
	Remove tier... 26
	Remove tier... 25
	Remove tier... 22
	Remove tier... 21
	Remove tier... 19
	Remove tier... 18
	Remove tier... 17
	Remove tier... 16
	Remove tier... 11
	Remove tier... 10
	Remove tier... 9
	Remove tier... 8

	# Insert the prediction tier
	predictedBoundaryTier = existingBoundaryTier+1
	Insert point tier... predictedBoundaryTier 'tierNameToInsert$'

	# Insert default points to the new tier according to the "end" time of the word tier
	for iToken from 2 to (numIntervals-1)
		endTimeWordTier = Get end point... wordTier iToken
		Insert point... predictedBoundaryTier endTimeWordTier

		select Strings predictedFile
		predictedBoundary$ = Get string... iLineOfPredictedFile

		select TextGrid textgrid
		Set point text... predictedBoundaryTier (iToken-1) 'predictedBoundary$'
		iLineOfPredictedFile = iLineOfPredictedFile + 1
	endfor
	filedelete 'outFolder$'\'outFile$' 
	Write to text file... 'outFolder$'\'outFile$' 
	Remove
endfor

# Check if all the lines in file-07.predicted have been consumed
if numTotalLines <> (iLineOfPredictedFile-1)
	linesRemaining = numTotalLines - (iLineOfPredictedFile-1)
	pause Some lines remain in file-07.predicted!!!
	echo Total 'numTotalLines'  Lines remaining 'linesRemaining'
endif

select Strings fileList
plus Strings predictedFile
Remove

#### END OF SCRIPT ####