[31196] | 1 | package org.hathitrust.extractedfeatures;
|
---|
| 2 |
|
---|
| 3 | import java.io.BufferedInputStream;
|
---|
| 4 | import java.io.BufferedOutputStream;
|
---|
| 5 | import java.io.FileInputStream;
|
---|
| 6 | import java.io.FileNotFoundException;
|
---|
| 7 | import java.io.FileOutputStream;
|
---|
| 8 | import java.io.IOException;
|
---|
| 9 | import java.io.InputStream;
|
---|
| 10 | import java.io.OutputStream;
|
---|
| 11 | import java.nio.charset.StandardCharsets;
|
---|
| 12 | import java.nio.file.Files;
|
---|
| 13 | import java.nio.file.Paths;
|
---|
| 14 | import java.util.stream.Stream;
|
---|
| 15 |
|
---|
| 16 | import com.google.common.hash.BloomFilter;
|
---|
| 17 | import com.google.common.hash.Funnel;
|
---|
| 18 | import com.google.common.hash.Funnels;
|
---|
| 19 |
|
---|
[31199] | 20 | public class TestWhitelistBloomFilter {
|
---|
[31196] | 21 |
|
---|
| 22 | protected String _dictionary_filename;
|
---|
| 23 | protected BloomFilter<CharSequence> _bloomFilter;
|
---|
| 24 |
|
---|
| 25 | protected int FILL_FACTOR = 26;
|
---|
| 26 |
|
---|
| 27 | // http://stackoverflow.com/questions/453018/number-of-lines-in-a-file-in-java
|
---|
| 28 | public static int countLines(String filename) throws IOException
|
---|
| 29 | {
|
---|
| 30 | InputStream is = new BufferedInputStream(new FileInputStream(filename));
|
---|
| 31 |
|
---|
| 32 | try {
|
---|
| 33 | byte[] c = new byte[1024];
|
---|
| 34 | int count = 0;
|
---|
| 35 | int readChars = 0;
|
---|
| 36 | boolean empty = true;
|
---|
| 37 | while ((readChars = is.read(c)) != -1) {
|
---|
| 38 | empty = false;
|
---|
| 39 | for (int i = 0; i < readChars; ++i) {
|
---|
| 40 | if (c[i] == '\n') {
|
---|
| 41 | ++count;
|
---|
| 42 | }
|
---|
| 43 | }
|
---|
| 44 | }
|
---|
| 45 | return (count == 0 && !empty) ? 1 : count;
|
---|
| 46 | } finally {
|
---|
| 47 | is.close();
|
---|
| 48 | }
|
---|
| 49 | }
|
---|
| 50 |
|
---|
| 51 |
|
---|
[31199] | 52 | public TestWhitelistBloomFilter(String dictionary_filename) {
|
---|
[31196] | 53 | System.out.println("Constructing: WhitelistBloomFilter");
|
---|
| 54 |
|
---|
| 55 | _dictionary_filename = dictionary_filename;
|
---|
| 56 |
|
---|
| 57 | System.out.println("Counting lines in: " + dictionary_filename);
|
---|
| 58 | int num_lines;
|
---|
| 59 | try {
|
---|
| 60 | num_lines = countLines(dictionary_filename);
|
---|
| 61 |
|
---|
| 62 | Funnel<CharSequence> string_funnel = Funnels.stringFunnel(StandardCharsets.UTF_8);
|
---|
| 63 | _bloomFilter = BloomFilter.create(string_funnel, FILL_FACTOR * num_lines,0.01);
|
---|
| 64 |
|
---|
| 65 | } catch (IOException e) {
|
---|
| 66 |
|
---|
| 67 | e.printStackTrace();
|
---|
| 68 | }
|
---|
| 69 |
|
---|
| 70 | }
|
---|
| 71 |
|
---|
| 72 | public void storeEntries()
|
---|
| 73 | {
|
---|
[31200] | 74 | System.out.println("Building synthetically expanded Bloom filter ...");
|
---|
[31196] | 75 |
|
---|
| 76 | for (int i=0; i<FILL_FACTOR; i++) {
|
---|
| 77 |
|
---|
| 78 | char prefix = (char) ('a' + i);
|
---|
| 79 |
|
---|
| 80 | //read file into stream, try-with-resources
|
---|
| 81 | try (Stream<String> stream = Files.lines(Paths.get(_dictionary_filename))) {
|
---|
| 82 |
|
---|
| 83 | stream.forEach(word -> {_bloomFilter.put(prefix+word);});
|
---|
| 84 | } catch (IOException e) {
|
---|
| 85 | e.printStackTrace();
|
---|
| 86 | }
|
---|
| 87 | }
|
---|
| 88 | System.out.println("... done");
|
---|
| 89 |
|
---|
| 90 | }
|
---|
| 91 |
|
---|
| 92 | public boolean contains(String key)
|
---|
| 93 | {
|
---|
| 94 | return _bloomFilter.mightContain(key);
|
---|
| 95 | }
|
---|
| 96 |
|
---|
| 97 | public void serializeOut(String filename)
|
---|
| 98 | {
|
---|
| 99 | try {
|
---|
| 100 | FileOutputStream fos = new FileOutputStream(filename);
|
---|
| 101 |
|
---|
| 102 | BufferedOutputStream bfos = new BufferedOutputStream(fos);
|
---|
| 103 |
|
---|
| 104 | _bloomFilter.writeTo(bfos);
|
---|
| 105 |
|
---|
| 106 | bfos.close();
|
---|
| 107 | }
|
---|
| 108 | catch (FileNotFoundException e) {
|
---|
| 109 | System.err.println("Unable to open Bloom file:" + filename);
|
---|
| 110 | e.printStackTrace();
|
---|
| 111 | } catch (IOException e) {
|
---|
| 112 | System.err.println("Error reading in Bloom file:" + filename);
|
---|
| 113 | e.printStackTrace();
|
---|
| 114 | }
|
---|
| 115 | }
|
---|
| 116 |
|
---|
| 117 | public void serializeIn(String filename)
|
---|
| 118 | {
|
---|
| 119 | try {
|
---|
| 120 | FileInputStream fis = new FileInputStream(filename);
|
---|
| 121 |
|
---|
| 122 | BufferedInputStream bfis = new BufferedInputStream(fis);
|
---|
| 123 |
|
---|
| 124 | Funnel<CharSequence> string_funnel = Funnels.stringFunnel(StandardCharsets.UTF_8);
|
---|
| 125 | _bloomFilter = BloomFilter.readFrom(bfis,string_funnel);
|
---|
| 126 |
|
---|
| 127 | bfis.close();
|
---|
| 128 | }
|
---|
| 129 | catch (FileNotFoundException e) {
|
---|
| 130 | System.err.println("Unable to open Bloom file:" + filename);
|
---|
| 131 | e.printStackTrace();
|
---|
| 132 | } catch (IOException e) {
|
---|
| 133 | System.err.println("Error writing out Bloom file:" + filename);
|
---|
| 134 | e.printStackTrace();
|
---|
| 135 | }
|
---|
| 136 | }
|
---|
| 137 |
|
---|
| 138 |
|
---|
| 139 | } |
---|