[31201] | 1 | package org.hathitrust.extractedfeatures;
|
---|
| 2 |
|
---|
| 3 | import java.io.BufferedInputStream;
|
---|
| 4 | import java.io.BufferedOutputStream;
|
---|
| 5 | import java.io.File;
|
---|
| 6 | import java.io.FileInputStream;
|
---|
| 7 | import java.io.FileNotFoundException;
|
---|
| 8 | import java.io.FileOutputStream;
|
---|
| 9 | import java.io.IOException;
|
---|
| 10 | import java.io.InputStream;
|
---|
[31202] | 11 | import java.io.Serializable;
|
---|
| 12 | import java.nio.charset.Charset;
|
---|
[31201] | 13 | import java.nio.charset.StandardCharsets;
|
---|
| 14 | import java.nio.file.Files;
|
---|
| 15 | import java.nio.file.Paths;
|
---|
| 16 | import java.util.stream.Stream;
|
---|
| 17 |
|
---|
[31202] | 18 | import javax.annotation.Nullable;
|
---|
| 19 |
|
---|
| 20 | import com.google.common.base.Preconditions;
|
---|
[31211] | 21 | import com.google.common.hash.BloomFilter;
|
---|
[31201] | 22 | import com.google.common.hash.Funnel;
|
---|
| 23 | import com.google.common.hash.Funnels;
|
---|
[31202] | 24 | import com.google.common.hash.PrimitiveSink;
|
---|
[31201] | 25 |
|
---|
| 26 | public class WhitelistBloomFilter {
|
---|
| 27 |
|
---|
[31211] | 28 | protected BloomFilter<CharSequence> _bloomFilter;
|
---|
[31201] | 29 | protected static final String SERIALIZED_SUFFIX = "-serialized";
|
---|
| 30 | protected static final double FALSE_POSITIVE_PERCENTAGE = 0.01;
|
---|
| 31 |
|
---|
| 32 |
|
---|
| 33 | // http://stackoverflow.com/questions/453018/number-of-lines-in-a-file-in-java
|
---|
| 34 | public static int countLines(String filename) throws IOException
|
---|
| 35 | {
|
---|
| 36 | InputStream is = new BufferedInputStream(new FileInputStream(filename));
|
---|
| 37 |
|
---|
| 38 | try {
|
---|
| 39 | byte[] c = new byte[1024];
|
---|
| 40 | int count = 0;
|
---|
| 41 | int readChars = 0;
|
---|
| 42 | boolean empty = true;
|
---|
| 43 | while ((readChars = is.read(c)) != -1) {
|
---|
| 44 | empty = false;
|
---|
| 45 | for (int i = 0; i < readChars; ++i) {
|
---|
| 46 | if (c[i] == '\n') {
|
---|
| 47 | ++count;
|
---|
| 48 | }
|
---|
| 49 | }
|
---|
| 50 | }
|
---|
| 51 | return (count == 0 && !empty) ? 1 : count;
|
---|
| 52 | } finally {
|
---|
| 53 | is.close();
|
---|
| 54 | }
|
---|
| 55 | }
|
---|
| 56 |
|
---|
| 57 |
|
---|
| 58 | public WhitelistBloomFilter(String dictionary_filename, boolean serialize) {
|
---|
| 59 | System.out.println("Constructing: WhitelistBloomFilter");
|
---|
| 60 |
|
---|
| 61 | File ser_dictionary_file = new File(dictionary_filename + SERIALIZED_SUFFIX);
|
---|
| 62 |
|
---|
| 63 | if (ser_dictionary_file.exists()) {
|
---|
| 64 | System.out.println("Loading Serialized Bloom filter ...");
|
---|
| 65 | _bloomFilter = serializeIn(ser_dictionary_file);
|
---|
| 66 | System.out.println("... done");
|
---|
| 67 | }
|
---|
| 68 | else {
|
---|
| 69 | // Need to generate the Bloom filter from the given raw text file
|
---|
| 70 |
|
---|
| 71 | System.out.println("Counting lines in: " + dictionary_filename);
|
---|
| 72 | int num_lines = -1;
|
---|
| 73 | try {
|
---|
| 74 | num_lines = countLines(dictionary_filename);
|
---|
[31202] | 75 |
|
---|
[31215] | 76 | Funnel<CharSequence> string_funnel = Funnels.stringFunnel(StandardCharsets.UTF_8);
|
---|
[31211] | 77 | _bloomFilter = BloomFilter.create(string_funnel, num_lines,FALSE_POSITIVE_PERCENTAGE);
|
---|
[31201] | 78 | }
|
---|
| 79 | catch (IOException e) {
|
---|
| 80 | e.printStackTrace();
|
---|
| 81 | }
|
---|
| 82 | System.out.println("Number of lines: " + num_lines);
|
---|
| 83 |
|
---|
| 84 | storeEntries(dictionary_filename,serialize);
|
---|
| 85 | }
|
---|
| 86 |
|
---|
| 87 | }
|
---|
| 88 |
|
---|
| 89 | protected void storeEntries(String filename, boolean serialize)
|
---|
| 90 | {
|
---|
| 91 | System.out.println("Building Bloom filter ...");
|
---|
| 92 |
|
---|
| 93 | //read file into stream, try-with-resources
|
---|
| 94 | try (Stream<String> stream = Files.lines(Paths.get(filename))) {
|
---|
| 95 | stream.forEach(word -> {_bloomFilter.put(word);});
|
---|
| 96 | } catch (IOException e) {
|
---|
| 97 | e.printStackTrace();
|
---|
| 98 | }
|
---|
| 99 |
|
---|
| 100 | System.out.println("... done");
|
---|
| 101 |
|
---|
| 102 | if (serialize) {
|
---|
| 103 | System.out.println("Serializing Bloom filter ...");
|
---|
| 104 |
|
---|
| 105 | File ser_dictionary = new File(filename + SERIALIZED_SUFFIX);
|
---|
| 106 | serializeOut(ser_dictionary);
|
---|
| 107 |
|
---|
| 108 | System.out.println("... done");
|
---|
| 109 | }
|
---|
| 110 |
|
---|
| 111 | }
|
---|
| 112 |
|
---|
| 113 | public boolean contains(String key)
|
---|
| 114 | {
|
---|
| 115 | return _bloomFilter.mightContain(key);
|
---|
| 116 | }
|
---|
| 117 |
|
---|
| 118 | protected void serializeOut(File ser_file)
|
---|
| 119 | {
|
---|
| 120 | try {
|
---|
| 121 | FileOutputStream fos = new FileOutputStream(ser_file);
|
---|
| 122 |
|
---|
| 123 | BufferedOutputStream bfos = new BufferedOutputStream(fos);
|
---|
| 124 |
|
---|
| 125 | _bloomFilter.writeTo(bfos);
|
---|
| 126 |
|
---|
| 127 | bfos.close();
|
---|
| 128 | }
|
---|
| 129 | catch (FileNotFoundException e) {
|
---|
| 130 | System.err.println("Unable to open Bloom file:" + ser_file.getAbsolutePath());
|
---|
| 131 | e.printStackTrace();
|
---|
| 132 | } catch (IOException e) {
|
---|
| 133 | System.err.println("Error reading in Bloom file:" + ser_file.getAbsolutePath());
|
---|
| 134 | e.printStackTrace();
|
---|
| 135 | }
|
---|
| 136 | }
|
---|
| 137 |
|
---|
[31211] | 138 | protected static BloomFilter<CharSequence> serializeIn(File ser_file)
|
---|
[31201] | 139 | {
|
---|
[31211] | 140 | BloomFilter<CharSequence> bloomFilter = null;
|
---|
[31201] | 141 |
|
---|
| 142 | try {
|
---|
| 143 | FileInputStream fis = new FileInputStream(ser_file);
|
---|
| 144 | BufferedInputStream bfis = new BufferedInputStream(fis);
|
---|
| 145 |
|
---|
[31215] | 146 | Funnel<CharSequence> string_funnel = Funnels.stringFunnel(StandardCharsets.UTF_8);
|
---|
[31211] | 147 | bloomFilter = BloomFilter.readFrom(bfis,string_funnel);
|
---|
[31201] | 148 |
|
---|
| 149 | bfis.close();
|
---|
| 150 | }
|
---|
| 151 | catch (FileNotFoundException e) {
|
---|
| 152 | System.err.println("Unable to open Bloom file:" + ser_file.getAbsolutePath());
|
---|
| 153 | e.printStackTrace();
|
---|
| 154 | } catch (IOException e) {
|
---|
| 155 | System.err.println("Error writing out Bloom file:" + ser_file.getAbsolutePath());
|
---|
| 156 | e.printStackTrace();
|
---|
| 157 | }
|
---|
| 158 | return bloomFilter;
|
---|
| 159 | }
|
---|
| 160 |
|
---|
| 161 |
|
---|
[31202] | 162 |
|
---|
| 163 |
|
---|
| 164 |
|
---|
[31201] | 165 | } |
---|