source: other-projects/is-sheet-music-encore/trunk/java-gen-corpus/javaGenFullIDList.java@ 33437

//NAME:Caleb Bird
//ID: 1289680
//??References??

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

//Creates textfile of all records tagged with XX (MU for example)
//Used on the textfile that contains every record in hathiTrust
public class javaGenFullIDList {
    public static void main(String[] args) {
        try{
                
                
            if (args.length != 3) {
                System.out.println("Usage: TabProcTextGen <inputFilename> <outputFilename> <inputType>");
            }
            else {
                String inputFilename = args[0];
                String outputFilename = args[1];
                String inputType = args[2];
                    

                System.out.println("Processing: " + inputFilename);
                
                FileReader fileReader = new FileReader(inputFilename);
                BufferedReader buf = new BufferedReader(fileReader);
                //FileWriter fw = new FileWriter("HathiDocIDList.txt");
                FileWriter fw = new FileWriter(outputFilename);
                String line = null;
                String[] item;

                //Splits into each record, since readLine splits by "\n"
                int line_num = 0;

                while ((line = buf.readLine()) != null) {
                    line_num++;
                    if (line_num%100000 == 0) {
                    System.out.print(".");
                    System.out.flush();
                    }

                    //Splits by tab, leaves "" for every blank entry
                    item = line.split("\t", -1);
                    //Check items to be equal
                    if( item[1].equals("allow") &&                          //allow - end users can view the item
                    item[2].equals("pd") &&                                 //pd - public domain
                    item[19].equals(inputType) &&                           //category defined by user (MU, SE,etc)
                    (item[24].equals("open")||item[24].equals("page"))){    //open - Items with this value do not have any download restrictions.
                    //page - Items with this value can be viewed on the HathiTrust website. Users can download individual pages but cannot download the full pdf, regardless of member affiliation.
                        //Encode id
                        String idEncoded = item[0];
                        //idEncoded = idEncoded.replaceAll(":", "+").replaceAll("/", "=");
                        //Write item to file
                        fw.write(idEncoded + "\n");
                    }
                }       
                buf.close();
                fw.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
//Email from supervisor (Compx520 ID Error Apr 24)