###########################################################################
# Checks the number of sentences, the sentence end being signaled by "SFN" (POS).
# Written by Kyuchul Yoon ( kyoon@ling.osu.edu )
############################################################################

# Specify files and folders
form Select files
	comment How many sentences should each file contain?
	natural numSentences 8
	word inputFileExt group
	word sentSeparator SFN
	integer colNumForSentSeparator 16
	word outputFile file-info.txt
endform

Create Strings as file list... fileList *.'inputFileExt$'
Sort
numFiles = Get number of strings
pause 'numFiles' labeled textgrids identified.  Continue?

# Loop throught each file
for iFile to numFiles
	# Get the input filename
	select Strings fileList
	doneFile$ = Get string... iFile
	Read Strings from raw text file... 'doneFile$'
	Rename... myStrings
	numLines = Get number of strings

	# Initialize the index for the number of sentences
	iSentence = 1

	# Loop through each line
	for iLine to numLines
#		#####  By examining the file numbers to omit, decide which output file to use
#		if (sentenceNum1 <> iSentence) and (sentenceNum2 <> iSentence)
#			outFileToUse$ = "selected.txt"
#		else
#			outFileToUse$ = "omitted.txt"
#		endif
		currentLine$ = Get string... iLine
		tempString$ = currentLine$

		####### Get the column string that the user designated as separating sentences #######
		for iColumn to (colNumForSentSeparator-1)
			lenCurrentLine = length(tempString$)
			iTab = index(tempString$, tab$)
			tempString$ = right$(tempString$, (lenCurrentLine-iTab))
		endfor
		##################### End of "gettting to" the column string ################
	
		#### Now, actually extract the first three letters which could be "SFN"
		sentSep$ = left$(tempString$, 3)

		# If the line is indeed the last token (its POS being SFN), then increase the sentence number
		if (sentSep$ = "SFN")
			iSentence = iSentence + 1
		endif	
	endfor
	realNumSent = iSentence-1
	if realNumSent <> numSentences
		fileappend 'outputFile$' 'doneFile$''tab$''realNumSent''newline$'
	else
		fileappend 'outputFile$' 'doneFile$''tab$'OK'newline$'
	endif
	select Strings myStrings
	Remove
endfor
select Strings fileList
Remove

#### END OF SCRIPT ####