/alaspatial/src/main/java/org/ala/spatial/analysis/index/RemoveOccurrenceRecords.java

  1/*
  2 * To change this template, choose Tools | Templates
  3 * and open the template in the editor.
  4 */
  5package org.ala.spatial.analysis.index;
  6
  7import java.io.BufferedReader;
  8import java.io.FileReader;
  9import java.io.FileWriter;
 10import java.util.ArrayList;
 11
 12/**
 13 * 1. extract occurrenceId's from an occurrences.csv (A)
 14 * 2. remove occurrenceId's from an occurrences.csv (B)
 15 * 3. export (B) with occurrenceId's removed.
 16 *
 17 * id is in first column of occurrences.csv's
 18 *
 19 * @author Adam
 20 */
 21public class RemoveOccurrenceRecords {
 22
 23    static String match = null;
 24
 25    /**
 26     * arg[0] is occurrences file for id extraction
 27     * arg[1] is occurrences file for id removal
 28     * arg[2] is occurrences file for export.
 29     * arg[3] is optional string to match for record removal.
 30     * 
 31     * @param args
 32     */
 33    static void main(String[] args) {
 34        if (args.length > 3) {
 35            match = args[3];
 36        }
 37
 38        ArrayList<Long> ids = extractIds(args[0]);
 39        java.util.Collections.sort(ids);
 40
 41        removeIds(args[1], args[2], ids);
 42    }
 43
 44    private static ArrayList<Long> extractIds(String filename) {
 45        ArrayList<Long> ids = new ArrayList<Long>();
 46
 47        /* read occurances_csv */
 48        try {
 49            BufferedReader br = new BufferedReader(new FileReader(filename));
 50
 51            String s;
 52            String[] sa;
 53
 54            while ((s = br.readLine()) != null) {
 55                //check for continuation line
 56                while (s != null && s.length() > 0 && s.charAt(s.length() - 1) == '\\') {
 57                    String spart = br.readLine();
 58                    if (spart == null) {  //same as whole line is null
 59                        break;
 60                    } else {
 61                        s.replace('\\', ' ');   //new line is same as 'space'
 62                        s += spart;
 63                    }
 64                }//repeat as necessary
 65
 66                sa = s.split(",");
 67
 68                try {
 69                    long l = Long.parseLong(sa[0].replace("\"", ""));
 70
 71                    ids.add(l);
 72                } catch (Exception e) {
 73                }
 74            }
 75            br.close();
 76        } catch (Exception e) {
 77            e.printStackTrace();
 78        }
 79
 80        return ids;
 81    }
 82
 83    private static void removeIds(String inputFilename, String outputFilename, ArrayList<Long> ids) {
 84        /* read occurances_csv */
 85        try {
 86            BufferedReader br = new BufferedReader(new FileReader(inputFilename));
 87            FileWriter fw = new FileWriter(outputFilename);
 88
 89            String s;
 90            String[] sa;
 91
 92            int idsRemoved = 0;
 93            int stringMatchesRemoved = 0;
 94
 95            while ((s = br.readLine()) != null) {
 96                //check for continuation line
 97                while (s != null && s.length() > 0 && s.charAt(s.length() - 1) == '\\') {
 98                    String spart = br.readLine();
 99                    if (spart == null) {  //same as whole line is null
100                        break;
101                    } else {
102                        s.replace('\\', ' ');   //new line is same as 'space'
103                        s += spart;
104                    }
105                }//repeat as necessary
106
107                sa = s.split(",");
108
109                boolean export = true;
110                try {
111                    long l = Long.parseLong(sa[0].replace("\"", ""));
112
113                    if (java.util.Collections.binarySearch(ids, l) >= 0) {
114                        //don't export
115                        export = false;
116                        idsRemoved++;
117                    } else if (stringMatch(s)) {
118                        export = false;
119                        stringMatchesRemoved++;
120                    }
121                } catch (Exception e) {
122                }
123
124                if (export) {
125                    fw.append(s).append("\n");
126                }
127            }
128
129            System.out.println("ids Removed: " + idsRemoved + ", string matches Removed: " + stringMatchesRemoved);
130            br.close();
131            fw.close();
132        } catch (Exception e) {
133            e.printStackTrace();
134        }
135    }
136
137    private static boolean stringMatch(String s) {
138        return match == null || s.contains(match);
139    }
140}