ir_firstnazt

file= new File('TIME.ALL')

stopwordsFile=new File('stopwords.txt')

 

/*stopwords=[]

stopwordsFile.eachLine { stopwords << it.toString().toUpperCase() }

def isStopWords={ input-> if (stopwords.contains(input)) return true else return false }*/

def TEXT =/^(\*TEXT)[s|t| ]([0-9]+)[s|t| ]([0-9]+\/[0-9]+\/[0-9]+) PAGE ([0-9]+)/
def contents=[:]
id=1
println file.size()
System.exit(0)
file.eachLine{ 
	m=(it.toString()=~TEXT)
	if (m)
	{
		id=m[0][2].toInteger()
		contents[id]=[DATE:m[0][4].toString(),PAGE:m[0][7].toString(),DATA:""];
	}
	else
	{
		if(it.toString()!="*STOP")
		{
			contents[id].DATA<<=it.toString()
		}
		it.toString().tokenize().each { word -> if(!isStopWords(word)) contents[id].DATA<<=word.toString()+" " }

	}

}

println contents[563].DATA