Example #1
0
        public void WithAdditionalSplitCharacters_ShouldSetAdditionalSplitCharactersCorrectly()
        {
            var builder = new TokenizationOptionsBuilder();

            builder.SplitOnCharacters('$', '%', '|');
            builder.Build().AdditionalSplitCharacters.Should().BeEquivalentTo(new[] { '$', '%', '|' });
        }
Example #2
0
        public void WithSplittingOnPunctuation_ShouldSetTheSplitOnPunctuationPropertyCorrectly(bool setting)
        {
            var builder = new TokenizationOptionsBuilder();

            builder.SplitOnPunctuation(setting);
            builder.Build().SplitOnPunctuation.Should().Be(setting);
        }
Example #3
0
        public void WithAccentInsensitivity_ShouldSetTheAccentInsensitivityPropertyCorrectly(bool setting)
        {
            var builder = new TokenizationOptionsBuilder();

            builder.AccentInsensitive(setting);
            builder.Build().AccentInsensitive.Should().Be(setting);
        }
Example #4
0
        public void WithStemming_ShouldSetTheStemmingPropertyCorrectly(bool setting)
        {
            var builder = new TokenizationOptionsBuilder();

            builder.WithStemming(setting);
            builder.Build().Stemming.Should().Be(setting);
        }
        public async Task InitializeAsync()
        {
            this.index = new FullTextIndexBuilder <string>().Build();
            var wikipediaTests = WikipediaDataLoader.Load(typeof(FullTextIndexTests));
            var options        = new TokenizationOptionsBuilder().XmlContent().WithStemming().Build();

            foreach (var(name, text) in wikipediaTests)
            {
                await this.index.AddAsync(name, text, options);
            }
        }
Example #6
0
        public void WithoutApplyingAnyOptions_ShouldSetDefaultsCorrectly()
        {
            var builder = new TokenizationOptionsBuilder();

            builder.Build().Should().BeEquivalentTo(new TokenizationOptions(TokenizerKind.PlainText)
            {
                AccentInsensitive         = true,
                Stemming                  = false,
                AdditionalSplitCharacters = Array.Empty <char>(),
                CaseInsensitive           = true,
                SplitOnPunctuation        = true
            });
        }
Example #7
0
        public void ConfiguringContentAsXml_ShouldSetTheTokenizationKindToXml()
        {
            var builder = new TokenizationOptionsBuilder().XmlContent();

            builder.Build().TokenizerKind.Should().Be(TokenizerKind.XmlContent);
        }