//NAME:Caleb Bird
//ID: 1289680
//??References??
//https://stackoverflow.com/questions/13796451/how-to-extract-a-string-between-two-delimiters

import java.io.BufferedReader;
import java.io.FileReader;
import java.io.FileWriter;

public class TabProcMetaListGen {
    public static void main(String[] args) {
		try{
				
				
			if (args.length != 2) {
				System.out.println("Usage: TabProcMetaListGen <inputFilename> <listFilename>");
			}
			else {
				//Variables
				String inputFilename = args[0];
				String listFilename = args[1];		
				FileReader fileReader = new FileReader(inputFilename);
				BufferedReader buf = new BufferedReader(fileReader);
				FileWriter listWriter = new FileWriter(listFilename, true);
				String line = null;
				String[] item;
				
				String idLine = null;
				String id = null;
				String numpages = null;
								    
				System.out.println("Processing: " + inputFilename);
				
				//Splits into each record
				while ((line = buf.readLine()) != null) {					
					if(line.contains("<id>")){
						//Isoclate and store the id from the line
						idLine = line.substring(line.indexOf(">")+1, line.indexOf("</"));
						id  = idLine.substring(idLine.lastIndexOf("meta/")+5);
						
					}	
					if(line.contains("<htd:numpages>")){
						//Iscolate and store the page number
						numpages = line.substring(line.indexOf(">")+1, line.lastIndexOf("<"));
										
					}											
				}
				
				//Check if there are more than 10 pages
				if(Integer.parseInt(numpages) >= 10){
				//Add extracted metadata to output list
				listWriter.write(id + '\t' + numpages + '\n');
				}
				buf.close();
				listWriter.close();
			}
		}catch(Exception e){
			e.printStackTrace();
		}
    }
}

//REFERNECES
//https://www.javatpoint.com/java-filewriter-class
//https://docs.oracle.com/javase/8/docs/api/index.html?java/io/FileWriter.html