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
|
// Copyright Vespa.ai. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root.
package com.yahoo.vespa.indexinglanguage.linguistics;
import com.yahoo.document.annotation.Annotation;
import com.yahoo.document.annotation.AnnotationTypes;
import com.yahoo.document.annotation.Span;
import com.yahoo.document.annotation.SpanList;
import com.yahoo.document.annotation.SpanTree;
import com.yahoo.document.annotation.SpanTrees;
import com.yahoo.document.datatypes.StringFieldValue;
import com.yahoo.language.Linguistics;
import com.yahoo.language.process.StemMode;
import com.yahoo.language.process.Token;
import com.yahoo.language.process.Tokenizer;
import com.yahoo.text.Text;
import java.util.HashMap;
import java.util.Map;
import static com.yahoo.language.LinguisticsCase.toLowerCase;
/**
* This is a tool for adding {@link AnnotationTypes} type annotations to {@link StringFieldValue} objects.
*
* @author Simon Thoresen Hult
*/
public class LinguisticsAnnotator {
private final Linguistics factory;
private final AnnotatorConfig config;
private static class TermOccurrences {
final Map<String, Integer> termOccurrences = new HashMap<>();
final int maxOccurrences;
public TermOccurrences(int maxOccurrences) {
this.maxOccurrences = maxOccurrences;
}
boolean termCountBelowLimit(String term) {
String lowerCasedTerm = toLowerCase(term);
int occurrences = termOccurrences.getOrDefault(lowerCasedTerm, 0);
if (occurrences >= maxOccurrences) return false;
termOccurrences.put(lowerCasedTerm, occurrences + 1);
return true;
}
}
/**
* Constructs a new instance of this annotator.
*
* @param factory the linguistics factory to use when annotating
* @param config the linguistics config to use
*/
public LinguisticsAnnotator(Linguistics factory, AnnotatorConfig config) {
this.factory = factory;
this.config = config;
}
/**
* Annotates the given string with the appropriate linguistics annotations.
*
* @param text the text to annotate
* @return whether anything was annotated
*/
public boolean annotate(StringFieldValue text) {
if (text.getSpanTree(SpanTrees.LINGUISTICS) != null) return true; // Already annotated with LINGUISTICS.
Tokenizer tokenizer = factory.getTokenizer();
String input = (text.getString().length() <= config.getMaxTokenizeLength())
? text.getString()
: Text.substringByCodepoints(text.getString(), 0, config.getMaxTokenizeLength());
Iterable<Token> tokens = tokenizer.tokenize(input, config.getLanguage(), config.getStemMode(),
config.getRemoveAccents());
TermOccurrences termOccurrences = new TermOccurrences(config.getMaxTermOccurrences());
SpanTree tree = new SpanTree(SpanTrees.LINGUISTICS);
for (Token token : tokens)
addAnnotationSpan(text.getString(), tree.spanList(), token, config.getStemMode(), termOccurrences,
config.getMaxTokenLength());
if (tree.numAnnotations() == 0) return false;
text.setSpanTree(tree);
return true;
}
/**
* Creates a TERM annotation which has the term as annotation (only) if it is different from the
* original.
*
* @param term the term
* @param origTerm the original term
* @return the created TERM annotation
*/
public static Annotation termAnnotation(String term, String origTerm) {
if (term.equals(origTerm))
return new Annotation(AnnotationTypes.TERM);
else
return new Annotation(AnnotationTypes.TERM, new StringFieldValue(term));
}
private static void addAnnotation(Span here, String term, String orig, TermOccurrences termOccurrences,
int maxTokenLength) {
if (term.length() > maxTokenLength) {
return;
}
if (termOccurrences.termCountBelowLimit(term)) {
here.annotate(termAnnotation(term, orig));
}
}
private static void addAnnotationSpan(String input, SpanList parent, Token token, StemMode mode,
TermOccurrences termOccurrences, int maxTokenLength) {
if ( ! token.isSpecialToken()) {
if (token.getNumComponents() > 0) {
for (int i = 0; i < token.getNumComponents(); ++i) {
addAnnotationSpan(input, parent, token.getComponent(i), mode, termOccurrences, maxTokenLength);
}
return;
}
if ( ! token.isIndexable()) return;
}
if (token.getOffset() >= input.length()) {
throw new IllegalArgumentException(token + " has offset " + token.getOffset() + ", which is outside the " +
"bounds of the input string '" + input + "'");
}
if (token.getOffset() + token.getOrig().length() > input.length()) {
throw new IllegalArgumentException(token + " has offset " + token.getOffset() + ", which makes it overflow " +
"the bounds of the input string; " + input);
}
if (mode == StemMode.ALL) {
Span where = parent.span((int)token.getOffset(), token.getOrig().length());
String lowercasedOrig = toLowerCase(token.getOrig());
String term = token.getTokenString();
if (term != null) {
addAnnotation(where, term, token.getOrig(), termOccurrences, maxTokenLength);
if ( ! term.equals(lowercasedOrig))
addAnnotation(where, lowercasedOrig, token.getOrig(), termOccurrences, maxTokenLength);
}
for (int i = 0; i < token.getNumStems(); i++) {
String stem = token.getStem(i);
if (! (stem.equals(lowercasedOrig) || stem.equals(term)))
addAnnotation(where, stem, token.getOrig(), termOccurrences, maxTokenLength);
}
} else {
String term = token.getTokenString();
if (term == null || term.trim().isEmpty()) return;
if (term.length() > maxTokenLength) {
return;
}
if (termOccurrences.termCountBelowLimit(term)) {
parent.span((int)token.getOffset(), token.getOrig().length()).annotate(termAnnotation(term, token.getOrig()));
}
}
}
}
|
