source: other-projects/is-sheet-music-encore/trunk/java-gen-corpus/TabRndListGen.java@ 33044

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

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

public class TabRndListGen {
    public static void main(String[] args) {
    try{
        if (args.length != 3){
        System.out.println("Usage: TabRndListGen <inputFilename> <outputCount> <outputFilename>");
        }
        else{
        String inputFilename = args[0];
        int outputCount  = Integer.parseInt(args[1]);
        String outputFilename = args[2];
        
        BufferedReader buf = new BufferedReader(new FileReader(inputFilename));
        FileWriter fw = new FileWriter(outputFilename);
        //FileWriter fw = new FileWriter("hathiRndIDList.txt");
        //BufferedReader buf = new BufferedReader(new FileReader("hathiDocIDList.txt"));

        ArrayList<String> list = new ArrayList<String>();
    
        String line = null;
        String[] item;

        //Add items to Array
        while ((line = buf.readLine()) != null) {
            //Split line by tab
            item = line.split("\t", -1);
            //Add first element (ID)
            list.add(item[0]);      
        }
        //randomize list
        Collections.shuffle(list);
        
        int countMatchingCrit = 0;
        int j = 0;
        ArrayList<String> subList = new ArrayList<String>();
        //Download meta record, check if numpages >=10 
        while(countMatchingCrit < outputCount){
            System.out.println("Processing item: " + j);
            String idCurr = list.get(j);
            //run download script
            String cmd = "./download_metadata_temp.sh " + idCurr;
            Process p = Runtime.getRuntime().exec(cmd);
            /*BufferedReader br = new BufferedReader(new InputStreamReader(p.getInputStream()));
                while(br.ready())
                {
                        System.out.println(br.readLine());
                }*/
            p.waitFor();
            int exitStatus = p.exitValue();
            if(exitStatus == 0){
                //Extract numpages 
                int numpages = getNumPages(idCurr);
                if(numpages >= 10){
                    System.out.println("Successful items: " + countMatchingCrit);
                    countMatchingCrit++;
                    subList.add(idCurr);                        
                }
            }
            else{
                System.err.println("failed to run:" + cmd);
                System.exit(exitStatus);
            }
            j++;
            if(j >= list.size()){break;}
        }
        

        //Write these to new file
        for(int i =0; i < subList.size(); i++){
            fw.write(subList.get(i) + '\n');
        }
            
        buf.close();
        fw.close();
        }
    }catch(Exception e){
        e.printStackTrace();
    }
    }
    private static int getNumPages(String id){
        String numpages = null;
        try{
                //Variables
                String inputFilename = "metadata_temp.xml";
                FileReader fileReader = new FileReader(inputFilename);
                BufferedReader buf = new BufferedReader(fileReader);
                String line = null;
                String[] item;

                //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("<"));
                                        
                    }                                           
                }
                buf.close();
                //Check if there are more than 10 pages
    
        }catch(Exception e){
            e.printStackTrace();
        }
        return Integer.parseInt(numpages);
    }
}
// Returns 42474

//REFERNECES
//https://stackoverflow.com/questions/12892665/how-to-capture-the-exit-status-of-a-shell-command-in-java
//https://www.javatpoint.com/java-filewriter-class
//https://docs.oracle.com/javase/8/docs/api/index.html?java/io/FileWriter.html
//https://www.geeksforgeeks.org/randomly-select-items-from-a-list-in-java/
//https://codereview.stackexchange.com/questions/146551/picking-10-distinct-words-randomly-from-list-of-unique-words
//http://www.linuxforums.org/forum/programming-scripting/65117-c-c-system-function-analog-java.html

//USE RUN-LIST.txt (modify file it reads using $1 (terminal entry variable (filename))