blob: 07b27387d6e80b7c5c8f8cfa29794d1b624def5e (
plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
|
import gnu.trove.TIntArrayList;
import java.io.*;
import java.util.*;
import java.util.regex.Pattern;
public class Corpus
{
private Lexicon<String> tokenLexicon = new Lexicon<String>();
private Lexicon<TIntArrayList> phraseLexicon = new Lexicon<TIntArrayList>();
private Lexicon<TIntArrayList> contextLexicon = new Lexicon<TIntArrayList>();
private List<Edge> edges = new ArrayList<Edge>();
private List<List<Edge>> phraseToContext = new ArrayList<List<Edge>>();
private List<List<Edge>> contextToPhrase = new ArrayList<List<Edge>>();
public class Edge
{
Edge(int phraseId, int contextId, int count)
{
this.phraseId = phraseId;
this.contextId = contextId;
this.count = count;
}
public int getPhraseId()
{
return phraseId;
}
public TIntArrayList getPhrase()
{
return phraseLexicon.lookup(phraseId);
}
public String getPhraseString()
{
StringBuffer b = new StringBuffer();
for (int tid: getPhrase().toNativeArray())
{
if (b.length() > 0)
b.append(" ");
b.append(tokenLexicon.lookup(tid));
}
return b.toString();
}
public int getContextId()
{
return contextId;
}
public TIntArrayList getContext()
{
return contextLexicon.lookup(contextId);
}
public String getContextString()
{
StringBuffer b = new StringBuffer();
for (int tid: getContext().toNativeArray())
{
if (b.length() > 0)
b.append(" ");
b.append(tokenLexicon.lookup(tid));
}
return b.toString();
}
public int getCount()
{
return count;
}
private int phraseId;
private int contextId;
private int count;
}
List<Edge> getEdges()
{
return edges;
}
int getNumEdges()
{
return edges.size();
}
int getNumPhrases()
{
return phraseLexicon.size();
}
List<Edge> getEdgesForPhrase(int phraseId)
{
return phraseToContext.get(phraseId);
}
int getNumContexts()
{
return contextLexicon.size();
}
List<Edge> getEdgesForContext(int contextId)
{
return contextToPhrase.get(contextId);
}
int getNumTokens()
{
return tokenLexicon.size();
}
static Corpus readFromFile(Reader in) throws IOException
{
Corpus c = new Corpus();
// read in line-by-line
BufferedReader bin = new BufferedReader(in);
String line;
Pattern separator = Pattern.compile(" \\|\\|\\| ");
while ((line = bin.readLine()) != null)
{
// split into phrase and contexts
StringTokenizer st = new StringTokenizer(line, "\t");
assert (st.hasMoreTokens());
String phraseToks = st.nextToken();
assert (st.hasMoreTokens());
String rest = st.nextToken();
assert (!st.hasMoreTokens());
// process phrase
st = new StringTokenizer(phraseToks, " ");
TIntArrayList ptoks = new TIntArrayList();
while (st.hasMoreTokens())
ptoks.add(c.tokenLexicon.insert(st.nextToken()));
int phraseId = c.phraseLexicon.insert(ptoks);
if (phraseId == c.phraseToContext.size())
c.phraseToContext.add(new ArrayList<Edge>());
// process contexts
String[] parts = separator.split(rest);
assert (parts.length % 2 == 0);
for (int i = 0; i < parts.length; i += 2)
{
// process pairs of strings - context and count
TIntArrayList ctx = new TIntArrayList();
String ctxString = parts[i];
String countString = parts[i + 1];
StringTokenizer ctxStrtok = new StringTokenizer(ctxString, " ");
while (ctxStrtok.hasMoreTokens())
{
String token = ctxStrtok.nextToken();
if (!token.equals("<PHRASE>"))
ctx.add(c.tokenLexicon.insert(token));
}
int contextId = c.contextLexicon.insert(ctx);
if (contextId == c.contextToPhrase.size())
c.contextToPhrase.add(new ArrayList<Edge>());
assert (countString.startsWith("C="));
Edge e = c.new Edge(phraseId, contextId,
Integer.parseInt(countString.substring(2).trim()));
c.edges.add(e);
// index the edge for fast phrase, context lookup
c.phraseToContext.get(phraseId).add(e);
c.contextToPhrase.get(contextId).add(e);
}
}
return c;
}
}
|
