Golang registry.NewCache函数代码示例

func TestFrenchElision(t *testing.T) {
	tests := []struct {
		input  analysis.TokenStream
		output analysis.TokenStream
	}{
		{
			input: analysis.TokenStream{
				&analysis.Token{
					Term: []byte("l'avion"),
				},
			},
			output: analysis.TokenStream{
				&analysis.Token{
					Term: []byte("avion"),
				},
			},
		},
	}

	cache := registry.NewCache()
	elisionFilter, err := cache.TokenFilterNamed(ElisionName)
	if err != nil {
		t.Fatal(err)
	}
	for _, test := range tests {
		actual := elisionFilter.Filter(test.input)
		if !reflect.DeepEqual(actual, test.output) {
			t.Errorf("expected %s, got %s", test.output[0].Term, actual[0].Term)
		}
	}
}

func CommonBenchmarkIndexBatch(b *testing.B, create KVStoreCreate, destroy KVStoreDestroy, analysisWorkers, batchSize int) {

	cache := registry.NewCache()
	analyzer, err := cache.AnalyzerNamed("standard")
	if err != nil {
		b.Fatal(err)
	}

	b.ResetTimer()
	b.StopTimer()
	for i := 0; i < b.N; i++ {

		s, err := create()
		if err != nil {
			b.Fatal(err)
		}
		analysisQueue := NewAnalysisQueue(analysisWorkers)
		idx := NewUpsideDownCouch(s, analysisQueue)

		err = idx.Open()
		if err != nil {
			b.Fatal(err)
		}

		b.StartTimer()
		batch := index.NewBatch()
		for j := 0; j < 1000; j++ {
			if j%batchSize == 0 {
				if len(batch.IndexOps) > 0 {
					err := idx.Batch(batch)
					if err != nil {
						b.Fatal(err)
					}
				}
				batch = index.NewBatch()
			}
			indexDocument := document.NewDocument("").
				AddField(document.NewTextFieldWithAnalyzer("body", []uint64{}, []byte(benchmarkDocBodies[j%10]), analyzer))
			indexDocument.ID = strconv.Itoa(i) + "-" + strconv.Itoa(j)
			batch.Update(indexDocument)
		}
		// close last batch
		if len(batch.IndexOps) > 0 {
			err := idx.Batch(batch)
			if err != nil {
				b.Fatal(err)
			}
		}
		b.StopTimer()
		err = idx.Close()
		if err != nil {
			b.Fatal(err)
		}
		err = destroy()
		if err != nil {
			b.Fatal(err)
		}
		analysisQueue.Close()
	}
}

func TestStopWordsFilterLongestMatch(t *testing.T) {

	inputTokenStream := analysis.TokenStream{
		&analysis.Token{
			Term:     []byte("softestball"),
			Start:    0,
			End:      11,
			Position: 1,
		},
	}

	expectedTokenStream := analysis.TokenStream{
		&analysis.Token{
			Term:     []byte("softestball"),
			Start:    0,
			End:      11,
			Position: 1,
		},
		&analysis.Token{
			Term:     []byte("softest"),
			Start:    0,
			End:      7,
			Position: 1,
		},
		&analysis.Token{
			Term:     []byte("ball"),
			Start:    7,
			End:      11,
			Position: 1,
		},
	}

	cache := registry.NewCache()
	dictListConfig := map[string]interface{}{
		"type":   token_map.Name,
		"tokens": []interface{}{"soft", "softest", "ball"},
	}
	_, err := cache.DefineTokenMap("dict_test", dictListConfig)
	if err != nil {
		t.Fatal(err)
	}

	dictConfig := map[string]interface{}{
		"type":               "dict_compound",
		"dict_token_map":     "dict_test",
		"only_longest_match": true,
	}
	dictFilter, err := cache.DefineTokenFilter("dict_test", dictConfig)
	if err != nil {
		t.Fatal(err)
	}

	ouputTokenStream := dictFilter.Filter(inputTokenStream)
	if !reflect.DeepEqual(ouputTokenStream, expectedTokenStream) {
		t.Errorf("expected %#v got %#v", expectedTokenStream, ouputTokenStream)
	}
}

func TestStopWordsFilter(t *testing.T) {

	inputTokenStream := analysis.TokenStream{
		&analysis.Token{
			Term: []byte("a"),
		},
		&analysis.Token{
			Term: []byte("walk"),
		},
		&analysis.Token{
			Term: []byte("in"),
		},
		&analysis.Token{
			Term: []byte("the"),
		},
		&analysis.Token{
			Term: []byte("park"),
		},
	}

	expectedTokenStream := analysis.TokenStream{
		&analysis.Token{
			Term: []byte("walk"),
		},
		&analysis.Token{
			Term: []byte("park"),
		},
	}

	cache := registry.NewCache()
	stopListConfig := map[string]interface{}{
		"type":   token_map.Name,
		"tokens": []interface{}{"a", "in", "the"},
	}
	_, err := cache.DefineTokenMap("stop_test", stopListConfig)
	if err != nil {
		t.Fatal(err)
	}

	stopConfig := map[string]interface{}{
		"type":           "stop_tokens",
		"stop_token_map": "stop_test",
	}
	stopFilter, err := cache.DefineTokenFilter("stop_test", stopConfig)
	if err != nil {
		t.Fatal(err)
	}

	ouputTokenStream := stopFilter.Filter(inputTokenStream)
	if !reflect.DeepEqual(ouputTokenStream, expectedTokenStream) {
		t.Errorf("expected %#v got %#v", expectedTokenStream, ouputTokenStream)
	}
}

func TestSoraniAnalyzer(t *testing.T) {
	tests := []struct {
		input  []byte
		output analysis.TokenStream
	}{
		// stop word removal
		{
			input: []byte("ئەم پیاوە"),
			output: analysis.TokenStream{
				&analysis.Token{
					Term:     []byte("پیاو"),
					Position: 2,
					Start:    7,
					End:      17,
				},
			},
		},
		{
			input: []byte("پیاوە"),
			output: analysis.TokenStream{
				&analysis.Token{
					Term:     []byte("پیاو"),
					Position: 1,
					Start:    0,
					End:      10,
				},
			},
		},
		{
			input: []byte("پیاو"),
			output: analysis.TokenStream{
				&analysis.Token{
					Term:     []byte("پیاو"),
					Position: 1,
					Start:    0,
					End:      8,
				},
			},
		},
	}

	cache := registry.NewCache()
	analyzer, err := cache.AnalyzerNamed(AnalyzerName)
	if err != nil {
		t.Fatal(err)
	}
	for _, test := range tests {
		actual := analyzer.Analyze(test.input)
		if !reflect.DeepEqual(actual, test.output) {
			t.Errorf("expected %v, got %v", test.output, actual)
		}
	}
}

func TestThaiAnalyzerWihtoutOffsets(t *testing.T) {
	tests := []struct {
		input  []byte
		output analysis.TokenStream
	}{
		// stop words
		{
			input: []byte("บริษัทชื่อ XY&Z - คุยกับ [email protected]"),
			output: analysis.TokenStream{
				&analysis.Token{
					Term: []byte("บริษัท"),
				},
				&analysis.Token{
					Term: []byte("ชื่อ"),
				},
				&analysis.Token{
					Term: []byte("xy"),
				},
				&analysis.Token{
					Term: []byte("z"),
				},
				&analysis.Token{
					Term: []byte("คุย"),
				},
				&analysis.Token{
					Term: []byte("xyz"),
				},
				&analysis.Token{
					Term: []byte("demo.com"),
				},
			},
		},
	}

	cache := registry.NewCache()
	analyzer, err := cache.AnalyzerNamed(AnalyzerName)
	if err != nil {
		t.Fatal(err)
	}
	for _, test := range tests {
		actual := analyzer.Analyze(test.input)
		if len(actual) != len(test.output) {
			t.Errorf("expected length: %d, got %d", len(test.output), len(actual))
		}
		for i, tok := range actual {
			if !reflect.DeepEqual(tok.Term, test.output[i].Term) {
				t.Errorf("expected term %s (% x) got %s (% x)", test.output[i].Term, test.output[i].Term, tok.Term, tok.Term)
			}
		}
	}
}

func TestEnglishStemmer(t *testing.T) {
	tests := []struct {
		input  analysis.TokenStream
		output analysis.TokenStream
	}{
		{
			input: analysis.TokenStream{
				&analysis.Token{
					Term: []byte("walking"),
				},
				&analysis.Token{
					Term: []byte("talked"),
				},
				&analysis.Token{
					Term: []byte("business"),
				},
				&analysis.Token{
					Term:    []byte("protected"),
					KeyWord: true,
				},
			},
			output: analysis.TokenStream{
				&analysis.Token{
					Term: []byte("walk"),
				},
				&analysis.Token{
					Term: []byte("talk"),
				},
				&analysis.Token{
					Term: []byte("busi"),
				},
				&analysis.Token{
					Term:    []byte("protected"),
					KeyWord: true,
				},
			},
		},
	}

	cache := registry.NewCache()
	stemmerFilter, err := cache.TokenFilterNamed(StemmerName)
	if err != nil {
		t.Fatal(err)
	}
	for _, test := range tests {
		actual := stemmerFilter.Filter(test.input)
		if !reflect.DeepEqual(actual, test.output) {
			t.Errorf("expected %s, got %s", test.output, actual)
		}
	}
}

func TestJaAnalyzer(t *testing.T) {
	tests := []struct {
		input  []byte
		output analysis.TokenStream
	}{
		{
			input: []byte("こんにちは世界"),
			output: analysis.TokenStream{
				&analysis.Token{
					Term:     []byte("こんにちは"),
					Type:     analysis.Ideographic,
					Position: 1,
					Start:    0,
					End:      15,
				},
				&analysis.Token{
					Term:     []byte("世界"),
					Type:     analysis.Ideographic,
					Position: 2,
					Start:    15,
					End:      21,
				},
			},
		},
		{
			input: []byte("ｶﾀｶﾅ"),
			output: analysis.TokenStream{
				&analysis.Token{
					Term:     []byte("カタカナ"),
					Type:     analysis.Ideographic,
					Position: 1,
					Start:    0,
					End:      12,
				},
			},
		},
	}

	cache := registry.NewCache()
	for _, test := range tests {
		analyzer, err := cache.AnalyzerNamed(AnalyzerName)
		if err != nil {
			t.Fatal(err)
		}
		actual := analyzer.Analyze(test.input)
		if !reflect.DeepEqual(actual, test.output) {
			t.Errorf("expected %v, got %v", test.output, actual)
		}
	}
}

// NewIndexMapping creates a new IndexMapping that will use all the default indexing rules
func NewIndexMapping() *IndexMapping {
	return &IndexMapping{
		TypeMapping:           make(map[string]*DocumentMapping),
		DefaultMapping:        NewDocumentMapping(),
		TypeField:             defaultTypeField,
		DefaultType:           defaultType,
		DefaultAnalyzer:       defaultAnalyzer,
		DefaultDateTimeParser: defaultDateTimeParser,
		DefaultField:          defaultField,
		ByteArrayConverter:    defaultByteArrayConverter,
		CustomAnalysis:        newCustomAnalysis(),
		cache:                 registry.NewCache(),
	}
}

func TestElisionFilter(t *testing.T) {

	tests := []struct {
		input  analysis.TokenStream
		output analysis.TokenStream
	}{
		{
			input: analysis.TokenStream{
				&analysis.Token{
					Term: []byte("ar'word"),
				},
			},
			output: analysis.TokenStream{
				&analysis.Token{
					Term: []byte("word"),
				},
			},
		},
	}

	cache := registry.NewCache()

	articleListConfig := map[string]interface{}{
		"type":   token_map.Name,
		"tokens": []interface{}{"ar"},
	}
	_, err := cache.DefineTokenMap("articles_test", articleListConfig)
	if err != nil {
		t.Fatal(err)
	}

	elisionConfig := map[string]interface{}{
		"type":               "elision",
		"articles_token_map": "articles_test",
	}
	elisionFilter, err := cache.DefineTokenFilter("elision_test", elisionConfig)
	if err != nil {
		t.Fatal(err)
	}

	for _, test := range tests {

		actual := elisionFilter.Filter(test.input)
		if !reflect.DeepEqual(actual, test.output) {
			t.Errorf("expected %s, got %s", test.output[0].Term, actual[0].Term)
		}
	}
}

func TestDanishAnalyzer(t *testing.T) {
	tests := []struct {
		input  []byte
		output analysis.TokenStream
	}{
		// stemming
		{
			input: []byte("undersøg"),
			output: analysis.TokenStream{
				&analysis.Token{
					Term:     []byte("undersøg"),
					Position: 1,
					Start:    0,
					End:      9,
				},
			},
		},
		{
			input: []byte("undersøgelse"),
			output: analysis.TokenStream{
				&analysis.Token{
					Term:     []byte("undersøg"),
					Position: 1,
					Start:    0,
					End:      13,
				},
			},
		},
		// stop word
		{
			input:  []byte("på"),
			output: analysis.TokenStream{},
		},
	}

	cache := registry.NewCache()
	analyzer, err := cache.AnalyzerNamed(AnalyzerName)
	if err != nil {
		t.Fatal(err)
	}
	for _, test := range tests {
		actual := analyzer.Analyze(test.input)
		if !reflect.DeepEqual(actual, test.output) {
			t.Errorf("expected %v, got %v", test.output, actual)
		}
	}
}

func TestRomanianAnalyzer(t *testing.T) {
	tests := []struct {
		input  []byte
		output analysis.TokenStream
	}{
		// stemming
		{
			input: []byte("absenţa"),
			output: analysis.TokenStream{
				&analysis.Token{
					Term: []byte("absenţ"),
				},
			},
		},
		{
			input: []byte("absenţi"),
			output: analysis.TokenStream{
				&analysis.Token{
					Term: []byte("absenţ"),
				},
			},
		},
		// stop word
		{
			input:  []byte("îl"),
			output: analysis.TokenStream{},
		},
	}

	cache := registry.NewCache()
	analyzer, err := cache.AnalyzerNamed(AnalyzerName)
	if err != nil {
		t.Fatal(err)
	}
	for _, test := range tests {
		actual := analyzer.Analyze(test.input)
		if len(actual) != len(test.output) {
			t.Fatalf("expected length: %d, got %d", len(test.output), len(actual))
		}
		for i, tok := range actual {
			if !reflect.DeepEqual(tok.Term, test.output[i].Term) {
				t.Errorf("expected term %s (% x) got %s (% x)", test.output[i].Term, test.output[i].Term, tok.Term, tok.Term)
			}
		}
	}
}

func CommonBenchmarkIndex(b *testing.B, create KVStoreCreate, destroy KVStoreDestroy, analysisWorkers int) {

	cache := registry.NewCache()
	analyzer, err := cache.AnalyzerNamed("standard")
	if err != nil {
		b.Fatal(err)
	}

	indexDocument := document.NewDocument("").
		AddField(document.NewTextFieldWithAnalyzer("body", []uint64{}, []byte(benchmarkDocBodies[0]), analyzer))

	b.ResetTimer()
	b.StopTimer()
	for i := 0; i < b.N; i++ {
		s, err := create()
		if err != nil {
			b.Fatal(err)
		}
		analysisQueue := NewAnalysisQueue(analysisWorkers)
		idx := NewUpsideDownCouch(s, analysisQueue)

		err = idx.Open()
		if err != nil {
			b.Fatal(err)
		}
		indexDocument.ID = strconv.Itoa(i)
		// just time the indexing portion
		b.StartTimer()
		err = idx.Update(indexDocument)
		if err != nil {
			b.Fatal(err)
		}
		b.StopTimer()
		err = idx.Close()
		if err != nil {
			b.Fatal(err)
		}
		err = destroy()
		if err != nil {
			b.Fatal(err)
		}
		analysisQueue.Close()
	}
}

func TestThaiAnalyzer(t *testing.T) {
	tests := []struct {
		input  []byte
		output analysis.TokenStream
	}{
		// stop words
		{
			input: []byte("การที่ได้ต้องแสดงว่างานดี"),
			output: analysis.TokenStream{
				&analysis.Token{
					Term:     []byte("แสดง"),
					Position: 5,
					Start:    39,
					End:      51,
				},
				&analysis.Token{
					Term:     []byte("งาน"),
					Position: 7,
					Start:    60,
					End:      69,
				},
				&analysis.Token{
					Term:     []byte("ดี"),
					Position: 8,
					Start:    69,
					End:      75,
				},
			},
		},
	}

	cache := registry.NewCache()
	analyzer, err := cache.AnalyzerNamed(AnalyzerName)
	if err != nil {
		t.Fatal(err)
	}
	for _, test := range tests {
		actual := analyzer.Analyze(test.input)
		if !reflect.DeepEqual(actual, test.output) {
			t.Errorf("expected %v, got %v", test.output, actual)
		}
	}
}

func TestItalianLightStemmer(t *testing.T) {
	tests := []struct {
		input  analysis.TokenStream
		output analysis.TokenStream
	}{
		{
			input: analysis.TokenStream{
				&analysis.Token{
					Term: []byte("ragazzo"),
				},
			},
			output: analysis.TokenStream{
				&analysis.Token{
					Term: []byte("ragazz"),
				},
			},
		},
		{
			input: analysis.TokenStream{
				&analysis.Token{
					Term: []byte("ragazzi"),
				},
			},
			output: analysis.TokenStream{
				&analysis.Token{
					Term: []byte("ragazz"),
				},
			},
		},
	}

	cache := registry.NewCache()
	filter, err := cache.TokenFilterNamed(LightStemmerName)
	if err != nil {
		t.Fatal(err)
	}
	for _, test := range tests {
		actual := filter.Filter(test.input)
		if !reflect.DeepEqual(actual, test.output) {
			t.Errorf("expected %s, got %s", test.output[0].Term, actual[0].Term)
		}
	}
}