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
|
// Copyright Vespa.ai. Licensed under the terms of the Apache 2.0 license. See LICENSE in the project root.
#include <vespa/document/fieldvalue/document.h>
#include <vespa/document/fieldvalue/stringfieldvalue.h>
#include <vespa/document/repo/configbuilder.h>
#include <vespa/searchlib/test/doc_builder.h>
#include <vespa/searchlib/test/string_field_builder.h>
#include <vespa/searchlib/util/token_extractor.h>
#include <vespa/vespalib/gtest/gtest.h>
#include <vespa/vespalib/objects/nbostream.h>
#include <variant>
using document::DataType;
using document::Document;
using document::StringFieldValue;
using search::linguistics::TokenExtractor;
using search::test::DocBuilder;
using search::test::StringFieldBuilder;
using AlternativeWords = std::vector<vespalib::string>;
using AlternativeWordsOrWord = std::variant<AlternativeWords, vespalib::string>;
using Words = std::vector<AlternativeWordsOrWord>;
namespace {
vespalib::string corrupt_word = "corruptWord";
vespalib::string field_name("stringfield");
std::unique_ptr<Document>
make_corrupted_document(DocBuilder &b, size_t wordOffset)
{
StringFieldBuilder sfb(b);
auto doc = b.make_document("id:ns:searchdocument::18");
doc->setValue(field_name, sfb.tokenize("before ").word(corrupt_word).tokenize(" after").build());
vespalib::nbostream stream;
doc->serialize(stream);
std::vector<char> raw;
raw.resize(stream.size());
stream.read(&raw[0], stream.size());
assert(wordOffset < corrupt_word.size());
for (size_t i = 0; i + corrupt_word.size() <= raw.size(); ++i) {
if (memcmp(&raw[i], corrupt_word.c_str(), corrupt_word.size()) == 0) {
raw[i + wordOffset] = '\0';
break;
}
}
vespalib::nbostream badstream;
badstream.write(&raw[0], raw.size());
return std::make_unique<Document>(b.get_repo(), badstream);
}
}
class TokenExtractorTest : public ::testing::Test {
protected:
using SpanTerm = TokenExtractor::SpanTerm;
DocBuilder _doc_builder;
std::unique_ptr<Document> _doc;
TokenExtractor _token_extractor;
std::vector<SpanTerm> _terms;
static constexpr size_t max_word_len = 20;
TokenExtractorTest();
~TokenExtractorTest() override;
static DocBuilder::AddFieldsType
make_add_fields()
{
return [](auto& header) { header.addField(field_name, DataType::T_STRING); };
}
Words process(const StringFieldValue& value);
};
TokenExtractorTest::TokenExtractorTest()
: _doc_builder(make_add_fields()),
_doc(_doc_builder.make_document("id:ns:searchdocument::0")),
_token_extractor(field_name, max_word_len),
_terms()
{
}
TokenExtractorTest::~TokenExtractorTest() = default;
Words
TokenExtractorTest::process(const StringFieldValue& value)
{
Words result;
_terms.clear();
auto span_trees = value.getSpanTrees();
vespalib::stringref text = value.getValueRef();
_token_extractor.extract(_terms, span_trees, text, _doc.get());
auto it = _terms.begin();
auto ite = _terms.end();
auto itn = it;
for (; it != ite; ) {
for (; itn != ite && itn->span == it->span; ++itn);
if ((itn - it) > 1) {
auto& alternatives = std::get<0>(result.emplace_back());
for (;it != itn; ++it) {
alternatives.emplace_back(it->word);
}
} else {
result.emplace_back(vespalib::string(it->word));
++it;
}
}
return result;
}
TEST_F(TokenExtractorTest, empty_string)
{
EXPECT_EQ((Words{}), process(StringFieldValue("")));
}
TEST_F(TokenExtractorTest, plain_string)
{
EXPECT_EQ((Words{}), process(StringFieldValue("Plain string")));
}
TEST_F(TokenExtractorTest, normal_string)
{
StringFieldBuilder sfb(_doc_builder);
EXPECT_EQ((Words{"Hello", "world"}), process(sfb.tokenize("Hello world").build()));
}
TEST_F(TokenExtractorTest, normalized_tokens)
{
StringFieldBuilder sfb(_doc_builder);
auto value = sfb.token("Hello", false).alt_word("hello").tokenize(" world").build();
EXPECT_EQ("Hello world", value.getValue());
EXPECT_EQ((Words{"hello", "world"}), process(value));
}
TEST_F(TokenExtractorTest, alternative_tokens)
{
StringFieldBuilder sfb(_doc_builder);
auto value = sfb.word("Hello").alt_word("hello").tokenize(" world").build();
EXPECT_EQ("Hello world", value.getValue());
EXPECT_EQ((Words{AlternativeWords{"Hello", "hello"}, "world"}), process(value));
}
TEST_F(TokenExtractorTest, word_with_nul_byte_is_truncated)
{
auto doc = make_corrupted_document(_doc_builder, 7);
EXPECT_EQ((Words{"before", "corrupt", "after"}), process(dynamic_cast<const StringFieldValue&>(*doc->getValue(field_name))));
}
TEST_F(TokenExtractorTest, word_with_nul_byte_at_start_is_dropped)
{
auto doc = make_corrupted_document(_doc_builder, 0);
EXPECT_EQ((Words{"before", "after"}), process(dynamic_cast<const StringFieldValue&>(*doc->getValue(field_name))));
}
TEST_F(TokenExtractorTest, too_long_word_is_dropped)
{
StringFieldBuilder sfb(_doc_builder);
EXPECT_EQ((Words{"before", "after"}), process(sfb.tokenize("before veryverylongwordthatwillbedropped after").build()));
}
GTEST_MAIN_RUN_ALL_TESTS()
|
