func TestEnablingDisablingStoringDynamicFields(t *testing.T) {
data := map[string]interface{}{
"name": "bleve",
}
doc := document.NewDocument("x")
mapping := NewIndexMapping()
err := mapping.mapDocument(doc, data)
if err != nil {
t.Fatal(err)
}
for _, field := range doc.Fields {
if field.Name() == "name" && !field.Options().IsStored() {
t.Errorf("expected field 'name' to be stored, isn't")
}
}
StoreDynamic = false
defer func() {
StoreDynamic = true
}()
doc = document.NewDocument("y")
err = mapping.mapDocument(doc, data)
if err != nil {
t.Fatal(err)
}
for _, field := range doc.Fields {
if field.Name() == "name" && field.Options().IsStored() {
t.Errorf("expected field 'name' to be not stored, is")
}
}
}
func TestMappingJSONWithNull(t *testing.T) {
mapping := NewIndexMapping()
jsonbytes := []byte(`{"name":"marty", "age": null}`)
var jsondoc interface{}
err := json.Unmarshal(jsonbytes, &jsondoc)
if err != nil {
t.Fatal(err)
}
doc := document.NewDocument("1")
err = mapping.mapDocument(doc, jsondoc)
if err != nil {
t.Fatal(err)
}
found := false
count := 0
for _, f := range doc.Fields {
if f.Name() == "name" {
found = true
}
count++
}
if !found {
t.Errorf("expected to find field named 'name'")
}
if count != 1 {
t.Errorf("expected to find 1 find, found %d", count)
}
}
func TestMappingStructWithPointerToString(t *testing.T) {
mapping := buildMapping()
name := "marty"
x := struct {
Name *string
}{
Name: &name,
}
doc := document.NewDocument("1")
err := mapping.mapDocument(doc, x)
if err != nil {
t.Fatal(err)
}
found := false
count := 0
for _, f := range doc.Fields {
if f.Name() == "Name" {
found = true
}
count++
}
if !found {
t.Errorf("expected to find field named 'Name'")
}
if count != 1 {
t.Errorf("expected to find 1 find, found %d", count)
}
}
func BenchmarkBatch(b *testing.B) {
cache := registry.NewCache()
analyzer, err := cache.AnalyzerNamed(standard_analyzer.Name)
if err != nil {
b.Fatal(err)
}
analysisQueue := index.NewAnalysisQueue(1)
idx, err := NewUpsideDownCouch(null.Name, nil, analysisQueue)
if err != nil {
b.Fatal(err)
}
err = idx.Open()
if err != nil {
b.Fatal(err)
}
batch := index.NewBatch()
for i := 0; i < 100; i++ {
d := document.NewDocument(strconv.Itoa(i))
f := document.NewTextFieldWithAnalyzer("desc", nil, bleveWikiArticle1K, analyzer)
d.AddField(f)
batch.Update(d)
}
b.ResetTimer()
for i := 0; i < b.N; i++ {
err = idx.Batch(batch)
if err != nil {
b.Fatal(err)
}
}
}
func (r *firestormReader) Document(id string) (*document.Document, error) {
docID := []byte(id)
docNum, err := r.currDocNumForId(docID)
if err != nil {
return nil, err
} else if docNum == 0 {
return nil, nil
}
rv := document.NewDocument(id)
prefix := StoredPrefixDocIDNum(docID, docNum)
err = visitPrefix(r.r, prefix, func(key, val []byte) (bool, error) {
safeVal := make([]byte, len(val))
copy(safeVal, val)
row, err := NewStoredRowKV(key, safeVal)
if err != nil {
return false, err
}
if row != nil {
fieldName := r.f.fieldCache.FieldIndexed(row.field)
field := r.decodeFieldType(fieldName, row.arrayPositions, row.value.GetRaw())
if field != nil {
rv.AddField(field)
}
}
return true, nil
})
if err != nil {
return nil, err
}
return rv, nil
}
func BenchmarkAnalyze(b *testing.B) {
cache := registry.NewCache()
analyzer, err := cache.AnalyzerNamed(standard_analyzer.Name)
if err != nil {
b.Fatal(err)
}
analysisQueue := index.NewAnalysisQueue(1)
idx, err := NewFirestorm(null.Name, nil, analysisQueue)
if err != nil {
b.Fatal(err)
}
d := document.NewDocument("1")
f := document.NewTextFieldWithAnalyzer("desc", nil, bleveWikiArticle1K, analyzer)
d.AddField(f)
b.ResetTimer()
for i := 0; i < b.N; i++ {
rv := idx.Analyze(d)
if len(rv.Rows) < 92 || len(rv.Rows) > 93 {
b.Fatalf("expected 512-13 rows, got %d", len(rv.Rows))
}
}
}
func TestIndexTermReaderCompositeFields(t *testing.T) {
defer func() {
err := DestroyTest()
if err != nil {
t.Fatal(err)
}
}()
analysisQueue := index.NewAnalysisQueue(1)
idx, err := NewUpsideDownCouch(boltdb.Name, boltTestConfig, analysisQueue)
if err != nil {
t.Fatal(err)
}
err = idx.Open()
if err != nil {
t.Errorf("error opening index: %v", err)
}
defer func() {
err := idx.Close()
if err != nil {
t.Fatal(err)
}
}()
doc := document.NewDocument("1")
doc.AddField(document.NewTextFieldWithIndexingOptions("name", []uint64{}, []byte("test"), document.IndexField|document.StoreField|document.IncludeTermVectors))
doc.AddField(document.NewTextFieldWithIndexingOptions("title", []uint64{}, []byte("mister"), document.IndexField|document.StoreField|document.IncludeTermVectors))
doc.AddField(document.NewCompositeFieldWithIndexingOptions("_all", true, nil, nil, document.IndexField|document.IncludeTermVectors))
err = idx.Update(doc)
if err != nil {
t.Errorf("Error updating index: %v", err)
}
indexReader, err := idx.Reader()
if err != nil {
t.Error(err)
}
defer func() {
err := indexReader.Close()
if err != nil {
t.Fatal(err)
}
}()
termFieldReader, err := indexReader.TermFieldReader([]byte("mister"), "_all")
if err != nil {
t.Error(err)
}
tfd, err := termFieldReader.Next()
for tfd != nil && err == nil {
if tfd.ID != "1" {
t.Errorf("expected to find document id 1")
}
tfd, err = termFieldReader.Next()
}
if err != nil {
t.Error(err)
}
}
func TestIndexInsert(t *testing.T) {
defer func() {
err := DestroyTest()
if err != nil {
t.Fatal(err)
}
}()
analysisQueue := index.NewAnalysisQueue(1)
idx, err := NewUpsideDownCouch(boltdb.Name, boltTestConfig, analysisQueue)
if err != nil {
t.Fatal(err)
}
err = idx.Open()
if err != nil {
t.Errorf("error opening index: %v", err)
}
defer func() {
err := idx.Close()
if err != nil {
t.Fatal(err)
}
}()
var expectedCount uint64
docCount, err := idx.DocCount()
if err != nil {
t.Error(err)
}
if docCount != expectedCount {
t.Errorf("Expected document count to be %d got %d", expectedCount, docCount)
}
doc := document.NewDocument("1")
doc.AddField(document.NewTextField("name", []uint64{}, []byte("test")))
err = idx.Update(doc)
if err != nil {
t.Errorf("Error updating index: %v", err)
}
expectedCount++
docCount, err = idx.DocCount()
if err != nil {
t.Error(err)
}
if docCount != expectedCount {
t.Errorf("Expected document count to be %d got %d", expectedCount, docCount)
}
// should have 4 rows (1 for version, 1 for schema field, and 1 for single term, and 1 for the term count, and 1 for the back index entry)
expectedLength := uint64(1 + 1 + 1 + 1 + 1)
rowCount, err := idx.(*UpsideDownCouch).rowCount()
if err != nil {
t.Error(err)
}
if rowCount != expectedLength {
t.Errorf("expected %d rows, got: %d", expectedLength, rowCount)
}
}
func TestIndexInsertFields(t *testing.T) {
defer func() {
err := DestroyTest()
if err != nil {
t.Fatal(err)
}
}()
analysisQueue := index.NewAnalysisQueue(1)
idx, err := NewUpsideDownCouch(boltdb.Name, boltTestConfig, analysisQueue)
if err != nil {
t.Fatal(err)
}
err = idx.Open()
if err != nil {
t.Errorf("error opening index: %v", err)
}
defer func() {
err := idx.Close()
if err != nil {
t.Fatal(err)
}
}()
doc := document.NewDocument("1")
doc.AddField(document.NewTextFieldWithIndexingOptions("name", []uint64{}, []byte("test"), document.IndexField|document.StoreField))
doc.AddField(document.NewNumericFieldWithIndexingOptions("age", []uint64{}, 35.99, document.IndexField|document.StoreField))
dateField, err := document.NewDateTimeFieldWithIndexingOptions("unixEpoch", []uint64{}, time.Unix(0, 0), document.IndexField|document.StoreField)
if err != nil {
t.Error(err)
}
doc.AddField(dateField)
err = idx.Update(doc)
if err != nil {
t.Errorf("Error updating index: %v", err)
}
indexReader, err := idx.Reader()
if err != nil {
t.Error(err)
}
defer func() {
err := indexReader.Close()
if err != nil {
t.Fatal(err)
}
}()
fields, err := indexReader.Fields()
if err != nil {
t.Error(err)
} else {
expectedFields := []string{"name", "age", "unixEpoch"}
if !reflect.DeepEqual(fields, expectedFields) {
t.Errorf("expected fields: %v, got %v", expectedFields, fields)
}
}
}
func CommonBenchmarkIndexBatch(b *testing.B, storeName string, storeConfig map[string]interface{}, 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++ {
analysisQueue := index.NewAnalysisQueue(analysisWorkers)
idx, err := NewUpsideDownCouch(storeName, storeConfig, analysisQueue)
if err != nil {
b.Fatal(err)
}
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 TestConcurrentUpdate(t *testing.T) {
defer func() {
err := DestroyTest()
if err != nil {
t.Fatal(err)
}
}()
analysisQueue := index.NewAnalysisQueue(1)
idx, err := NewUpsideDownCouch(boltdb.Name, boltTestConfig, analysisQueue)
if err != nil {
t.Fatal(err)
}
err = idx.Open()
if err != nil {
t.Errorf("error opening index: %v", err)
}
defer func() {
err := idx.Close()
if err != nil {
t.Fatal(err)
}
}()
// do some concurrent updates
var wg sync.WaitGroup
for i := 0; i < 10; i++ {
wg.Add(1)
go func(i int) {
doc := document.NewDocument("1")
doc.AddField(document.NewTextFieldWithIndexingOptions(strconv.Itoa(i), []uint64{}, []byte(strconv.Itoa(i)), document.StoreField))
err := idx.Update(doc)
if err != nil {
t.Errorf("Error updating index: %v", err)
}
wg.Done()
}(i)
}
wg.Wait()
// now load the name field and see what we get
r, err := idx.Reader()
if err != nil {
log.Fatal(err)
}
doc, err := r.Document("1")
if err != nil {
log.Fatal(err)
}
if len(doc.Fields) > 1 {
t.Errorf("expected single field, found %d", len(doc.Fields))
}
}
func TestIndexDocumentFieldTerms(t *testing.T) {
defer func() {
err := DestroyTest()
if err != nil {
t.Fatal(err)
}
}()
analysisQueue := index.NewAnalysisQueue(1)
idx, err := NewUpsideDownCouch(boltdb.Name, boltTestConfig, analysisQueue)
if err != nil {
t.Fatal(err)
}
err = idx.Open()
if err != nil {
t.Errorf("error opening index: %v", err)
}
defer func() {
err := idx.Close()
if err != nil {
t.Fatal(err)
}
}()
doc := document.NewDocument("1")
doc.AddField(document.NewTextFieldWithIndexingOptions("name", []uint64{}, []byte("test"), document.IndexField|document.StoreField|document.IncludeTermVectors))
doc.AddField(document.NewTextFieldWithIndexingOptions("title", []uint64{}, []byte("mister"), document.IndexField|document.StoreField|document.IncludeTermVectors))
err = idx.Update(doc)
if err != nil {
t.Errorf("Error updating index: %v", err)
}
indexReader, err := idx.Reader()
if err != nil {
t.Error(err)
}
defer func() {
err := indexReader.Close()
if err != nil {
t.Fatal(err)
}
}()
fieldTerms, err := indexReader.DocumentFieldTerms("1")
if err != nil {
t.Error(err)
}
expectedFieldTerms := index.FieldTerms{
"name": []string{"test"},
"title": []string{"mister"},
}
if !reflect.DeepEqual(fieldTerms, expectedFieldTerms) {
t.Errorf("expected field terms: %#v, got: %#v", expectedFieldTerms, fieldTerms)
}
}
func TestIndexInsert(t *testing.T) {
defer func() {
err := os.RemoveAll("test")
if err != nil {
t.Fatal(err)
}
}()
analysisQueue := index.NewAnalysisQueue(1)
idx, err := NewFirestorm(boltdb.Name, boltTestConfig, analysisQueue)
if err != nil {
t.Fatal(err)
}
err = idx.Open()
if err != nil {
t.Errorf("error opening index: %v", err)
}
defer func() {
err := idx.Close()
if err != nil {
t.Fatal(err)
}
}()
var expectedCount uint64
docCount, err := idx.DocCount()
if err != nil {
t.Error(err)
}
if docCount != expectedCount {
t.Errorf("Expected document count to be %d got %d", expectedCount, docCount)
}
doc := document.NewDocument("1")
doc.AddField(document.NewTextField("name", []uint64{}, []byte("test")))
err = idx.Update(doc)
if err != nil {
t.Errorf("Error updating index: %v", err)
}
expectedCount++
err = idx.(*Firestorm).lookuper.waitTasksDone(lookupWaitDuration)
if err != nil {
t.Fatal(err)
}
docCount, err = idx.DocCount()
if err != nil {
t.Error(err)
}
if docCount != expectedCount {
t.Errorf("Expected document count to be %d got %d", expectedCount, docCount)
}
}
// Index adds the specified index operation to the
// batch. NOTE: the bleve Index is not updated
// until the batch is executed.
func (b *Batch) Index(id string, data interface{}) error {
if id == "" {
return ErrorEmptyID
}
doc := document.NewDocument(id)
err := b.index.Mapping().mapDocument(doc, data)
if err != nil {
return err
}
b.internal.Update(doc)
return nil
}
func TestIndexFieldsMisc(t *testing.T) {
defer func() {
err := DestroyTest()
if err != nil {
t.Fatal(err)
}
}()
analysisQueue := index.NewAnalysisQueue(1)
idx, err := NewUpsideDownCouch(boltdb.Name, boltTestConfig, analysisQueue)
if err != nil {
t.Fatal(err)
}
err = idx.Open()
if err != nil {
t.Errorf("error opening index: %v", err)
}
defer func() {
err := idx.Close()
if err != nil {
t.Fatal(err)
}
}()
doc := document.NewDocument("1")
doc.AddField(document.NewTextFieldWithIndexingOptions("name", []uint64{}, []byte("test"), document.IndexField|document.StoreField))
doc.AddField(document.NewTextFieldWithIndexingOptions("title", []uint64{}, []byte("mister"), document.IndexField|document.StoreField))
err = idx.Update(doc)
if err != nil {
t.Errorf("Error updating index: %v", err)
}
fieldName1 := idx.(*UpsideDownCouch).fieldCache.FieldIndexed(0)
if fieldName1 != "name" {
t.Errorf("expected field named 'name', got '%s'", fieldName1)
}
fieldName2 := idx.(*UpsideDownCouch).fieldCache.FieldIndexed(1)
if fieldName2 != "title" {
t.Errorf("expected field named 'title', got '%s'", fieldName2)
}
fieldName3 := idx.(*UpsideDownCouch).fieldCache.FieldIndexed(2)
if fieldName3 != "" {
t.Errorf("expected field named '', got '%s'", fieldName3)
}
}
func CommonBenchmarkIndex(b *testing.B, storeName string, storeConfig map[string]interface{}, 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++ {
analysisQueue := index.NewAnalysisQueue(analysisWorkers)
idx, err := NewUpsideDownCouch(storeName, storeConfig, analysisQueue)
if err != nil {
b.Fatal(err)
}
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 TestMappingBool(t *testing.T) {
boolMapping := NewBooleanFieldMapping()
docMapping := NewDocumentMapping()
docMapping.AddFieldMappingsAt("prop", boolMapping)
mapping := NewIndexMapping()
mapping.AddDocumentMapping("doc", docMapping)
pprop := false
x := struct {
Prop bool `json:"prop"`
PProp *bool `json:"pprop"`
}{
Prop: true,
PProp: &pprop,
}
doc := document.NewDocument("1")
err := mapping.mapDocument(doc, x)
if err != nil {
t.Fatal(err)
}
foundProp := false
foundPProp := false
count := 0
for _, f := range doc.Fields {
if f.Name() == "prop" {
foundProp = true
}
if f.Name() == "pprop" {
foundPProp = true
}
count++
}
if !foundProp {
t.Errorf("expected to find bool field named 'prop'")
}
if !foundPProp {
t.Errorf("expected to find pointer to bool field named 'pprop'")
}
if count != 2 {
t.Errorf("expected to find 2 fields, found %d", count)
}
}
func (i *IndexReader) Document(id string) (doc *document.Document, err error) {
// first hit the back index to confirm doc exists
var backIndexRow *BackIndexRow
backIndexRow, err = i.index.backIndexRowForDoc(i.kvreader, id)
if err != nil {
return
}
if backIndexRow == nil {
return
}
doc = document.NewDocument(id)
storedRow := NewStoredRow([]byte(id), 0, []uint64{}, 'x', nil)
storedRowScanPrefix := storedRow.ScanPrefixForDoc()
it := i.kvreader.PrefixIterator(storedRowScanPrefix)
defer func() {
if cerr := it.Close(); err == nil && cerr != nil {
err = cerr
}
}()
key, val, valid := it.Current()
for valid {
safeVal := make([]byte, len(val))
copy(safeVal, val)
var row *StoredRow
row, err = NewStoredRowKV(key, safeVal)
if err != nil {
doc = nil
return
}
if row != nil {
fieldName := i.index.fieldCache.FieldIndexed(row.field)
field := decodeFieldType(row.typ, fieldName, row.arrayPositions, row.value)
if field != nil {
doc.AddField(field)
}
}
it.Next()
key, val, valid = it.Current()
}
return
}
// Index the object with the specified identifier.
// The IndexMapping for this index will determine
// how the object is indexed.
func (i *indexImpl) Index(id string, data interface{}) (err error) {
if id == "" {
return ErrorEmptyID
}
i.mutex.RLock()
defer i.mutex.RUnlock()
if !i.open {
return ErrorIndexClosed
}
doc := document.NewDocument(id)
err = i.m.mapDocument(doc, data)
if err != nil {
return
}
err = i.i.Update(doc)
return
}
func TestLargeField(t *testing.T) {
defer func() {
err := DestroyTest()
if err != nil {
t.Fatal(err)
}
}()
analysisQueue := index.NewAnalysisQueue(1)
idx, err := NewUpsideDownCouch(boltdb.Name, boltTestConfig, analysisQueue)
if err != nil {
t.Fatal(err)
}
err = idx.Open()
if err != nil {
t.Errorf("error opening index: %v", err)
}
defer func() {
err := idx.Close()
if err != nil {
t.Fatal(err)
}
}()
largeFieldValue := make([]byte, 0)
for len(largeFieldValue) < RowBufferSize {
largeFieldValue = append(largeFieldValue, bleveWikiArticle1K...)
}
t.Logf("large field size: %d", len(largeFieldValue))
d := document.NewDocument("large")
f := document.NewTextFieldWithIndexingOptions("desc", nil, largeFieldValue, document.IndexField|document.StoreField)
d.AddField(f)
err = idx.Update(d)
if err != nil {
t.Fatal(err)
}
}
func TestMappingStructWithJSONTagsOneDisabled(t *testing.T) {
mapping := buildMapping()
x := struct {
Name string `json:"name"`
Title string `json:"-"`
NoJSONTag string
}{
Name: "marty",
}
doc := document.NewDocument("1")
err := mapping.mapDocument(doc, x)
if err != nil {
t.Fatal(err)
}
foundJSONName := false
foundNoJSONName := false
count := 0
for _, f := range doc.Fields {
if f.Name() == "name" {
foundJSONName = true
}
if f.Name() == "NoJSONTag" {
foundNoJSONName = true
}
count++
}
if !foundJSONName {
t.Errorf("expected to find field named 'name'")
}
if !foundNoJSONName {
t.Errorf("expected to find field named 'NoJSONTag'")
}
if count != 2 {
t.Errorf("expected to find 2 find, found %d", count)
}
}
func TestSimpleHighlighter(t *testing.T) {
fragmenter := sfrag.NewFragmenter(100)
formatter := ansi.NewFragmentFormatter(ansi.DefaultAnsiHighlight)
highlighter := NewHighlighter(fragmenter, formatter, DefaultSeparator)
docMatch := search.DocumentMatch{
ID: "a",
Score: 1.0,
Locations: search.FieldTermLocationMap{
"desc": search.TermLocationMap{
"quick": search.Locations{
&search.Location{
Pos: 2,
Start: 4,
End: 9,
},
},
"fox": search.Locations{
&search.Location{
Pos: 4,
Start: 16,
End: 19,
},
},
},
},
}
expectedFragment := "the " + DefaultAnsiHighlight + "quick" + reset + " brown " + DefaultAnsiHighlight + "fox" + reset + " jumps over the lazy dog"
doc := document.NewDocument("a").AddField(document.NewTextField("desc", []uint64{}, []byte("the quick brown fox jumps over the lazy dog")))
fragment := highlighter.BestFragmentInField(&docMatch, doc, "desc")
if fragment != expectedFragment {
t.Errorf("expected `%s`, got `%s`", expectedFragment, fragment)
}
}
func TestIndexInsertThenUpdate(t *testing.T) {
defer func() {
err := DestroyTest()
if err != nil {
t.Fatal(err)
}
}()
analysisQueue := index.NewAnalysisQueue(1)
idx, err := NewUpsideDownCouch(boltdb.Name, boltTestConfig, analysisQueue)
if err != nil {
t.Fatal(err)
}
err = idx.Open()
if err != nil {
t.Errorf("error opening index: %v", err)
}
defer func() {
err := idx.Close()
if err != nil {
t.Fatal(err)
}
}()
doc := document.NewDocument("1")
doc.AddField(document.NewTextField("name", []uint64{}, []byte("test")))
err = idx.Update(doc)
if err != nil {
t.Errorf("Error updating index: %v", err)
}
// this update should overwrite one term, and introduce one new one
doc = document.NewDocument("1")
doc.AddField(document.NewTextFieldWithAnalyzer("name", []uint64{}, []byte("test fail"), testAnalyzer))
err = idx.Update(doc)
if err != nil {
t.Errorf("Error deleting entry from index: %v", err)
}
// should have 2 rows (1 for version, 1 for schema field, and 2 for the two term, and 2 for the term counts, and 1 for the back index entry)
expectedLength := uint64(1 + 1 + 2 + 2 + 1)
rowCount, err := idx.(*UpsideDownCouch).rowCount()
if err != nil {
t.Error(err)
}
if rowCount != expectedLength {
t.Errorf("expected %d rows, got: %d", expectedLength, rowCount)
}
// now do another update that should remove one of the terms
doc = document.NewDocument("1")
doc.AddField(document.NewTextField("name", []uint64{}, []byte("fail")))
err = idx.Update(doc)
if err != nil {
t.Errorf("Error deleting entry from index: %v", err)
}
// should have 2 rows (1 for version, 1 for schema field, and 1 for the remaining term, and 2 for the term diciontary, and 1 for the back index entry)
expectedLength = uint64(1 + 1 + 1 + 2 + 1)
rowCount, err = idx.(*UpsideDownCouch).rowCount()
if err != nil {
t.Error(err)
}
if rowCount != expectedLength {
t.Errorf("expected %d rows, got: %d", expectedLength, rowCount)
}
}
panic(err)
}
}
}
// create a simpler analyzer which will support these tests
var testAnalyzer = &analysis.Analyzer{
Tokenizer: regexp_tokenizer.NewRegexpTokenizer(regexp.MustCompile(`\w+`)),
}
// sets up some mock data used in many tests in this package
var twoDocIndexDescIndexingOptions = document.DefaultTextIndexingOptions | document.IncludeTermVectors
var twoDocIndexDocs = []*document.Document{
// must have 4/4 beer
document.NewDocument("1").
AddField(document.NewTextField("name", []uint64{}, []byte("marty"))).
AddField(document.NewTextFieldCustom("desc", []uint64{}, []byte("beer beer beer beer"), twoDocIndexDescIndexingOptions, testAnalyzer)).
AddField(document.NewTextFieldWithAnalyzer("street", []uint64{}, []byte("couchbase way"), testAnalyzer)),
// must have 1/4 beer
document.NewDocument("2").
AddField(document.NewTextField("name", []uint64{}, []byte("steve"))).
AddField(document.NewTextFieldCustom("desc", []uint64{}, []byte("angst beer couch database"), twoDocIndexDescIndexingOptions, testAnalyzer)).
AddField(document.NewTextFieldWithAnalyzer("street", []uint64{}, []byte("couchbase way"), testAnalyzer)).
AddField(document.NewTextFieldWithAnalyzer("title", []uint64{}, []byte("mister"), testAnalyzer)),
// must have 1/4 beer
document.NewDocument("3").
AddField(document.NewTextField("name", []uint64{}, []byte("dustin"))).
AddField(document.NewTextFieldCustom("desc", []uint64{}, []byte("apple beer column dank"), twoDocIndexDescIndexingOptions, testAnalyzer)).
AddField(document.NewTextFieldWithAnalyzer("title", []uint64{}, []byte("mister"), testAnalyzer)),
// must have 65/65 beer
func TestIndexInsertWithStore(t *testing.T) {
defer func() {
err := DestroyTest()
if err != nil {
t.Fatal(err)
}
}()
analysisQueue := index.NewAnalysisQueue(1)
idx, err := NewUpsideDownCouch(boltdb.Name, boltTestConfig, analysisQueue)
if err != nil {
t.Fatal(err)
}
err = idx.Open()
if err != nil {
t.Errorf("error opening index: %v", err)
}
defer func() {
err := idx.Close()
if err != nil {
t.Fatal(err)
}
}()
var expectedCount uint64
docCount, err := idx.DocCount()
if err != nil {
t.Error(err)
}
if docCount != expectedCount {
t.Errorf("Expected document count to be %d got %d", expectedCount, docCount)
}
doc := document.NewDocument("1")
doc.AddField(document.NewTextFieldWithIndexingOptions("name", []uint64{}, []byte("test"), document.IndexField|document.StoreField))
err = idx.Update(doc)
if err != nil {
t.Errorf("Error updating index: %v", err)
}
expectedCount++
docCount, err = idx.DocCount()
if err != nil {
t.Error(err)
}
if docCount != expectedCount {
t.Errorf("Expected document count to be %d got %d", expectedCount, docCount)
}
// should have 6 rows (1 for version, 1 for schema field, and 1 for single term, and 1 for the stored field and 1 for the term count, and 1 for the back index entry)
expectedLength := uint64(1 + 1 + 1 + 1 + 1 + 1)
rowCount, err := idx.(*UpsideDownCouch).rowCount()
if err != nil {
t.Error(err)
}
if rowCount != expectedLength {
t.Errorf("expected %d rows, got: %d", expectedLength, rowCount)
}
indexReader, err := idx.Reader()
if err != nil {
t.Error(err)
}
defer func() {
err := indexReader.Close()
if err != nil {
t.Fatal(err)
}
}()
storedDoc, err := indexReader.Document("1")
if err != nil {
t.Error(err)
}
if len(storedDoc.Fields) != 1 {
t.Errorf("expected 1 stored field, got %d", len(storedDoc.Fields))
}
textField, ok := storedDoc.Fields[0].(*document.TextField)
if !ok {
t.Errorf("expected text field")
}
if string(textField.Value()) != "test" {
t.Errorf("expected field content 'test', got '%s'", string(textField.Value()))
}
}
func TestIndexBatch(t *testing.T) {
defer func() {
err := DestroyTest()
if err != nil {
t.Fatal(err)
}
}()
analysisQueue := index.NewAnalysisQueue(1)
idx, err := NewUpsideDownCouch(boltdb.Name, boltTestConfig, analysisQueue)
if err != nil {
t.Fatal(err)
}
err = idx.Open()
if err != nil {
t.Errorf("error opening index: %v", err)
}
defer func() {
err := idx.Close()
if err != nil {
t.Fatal(err)
}
}()
var expectedCount uint64
// first create 2 docs the old fashioned way
doc := document.NewDocument("1")
doc.AddField(document.NewTextField("name", []uint64{}, []byte("test")))
err = idx.Update(doc)
if err != nil {
t.Errorf("Error updating index: %v", err)
}
expectedCount++
doc = document.NewDocument("2")
doc.AddField(document.NewTextField("name", []uint64{}, []byte("test2")))
err = idx.Update(doc)
if err != nil {
t.Errorf("Error updating index: %v", err)
}
expectedCount++
// now create a batch which does 3 things
// insert new doc
// update existing doc
// delete existing doc
// net document count change 0
batch := index.NewBatch()
doc = document.NewDocument("3")
doc.AddField(document.NewTextField("name", []uint64{}, []byte("test3")))
batch.Update(doc)
doc = document.NewDocument("2")
doc.AddField(document.NewTextField("name", []uint64{}, []byte("test2updated")))
batch.Update(doc)
batch.Delete("1")
err = idx.Batch(batch)
if err != nil {
t.Error(err)
}
indexReader, err := idx.Reader()
if err != nil {
t.Error(err)
}
defer func() {
err := indexReader.Close()
if err != nil {
t.Fatal(err)
}
}()
docCount := indexReader.DocCount()
if docCount != expectedCount {
t.Errorf("Expected document count to be %d got %d", expectedCount, docCount)
}
docIDReader, err := indexReader.DocIDReader("", "")
if err != nil {
t.Error(err)
}
docIds := make([]string, 0)
docID, err := docIDReader.Next()
for docID != "" && err == nil {
docIds = append(docIds, docID)
docID, err = docIDReader.Next()
}
if err != nil {
t.Error(err)
}
expectedDocIds := []string{"2", "3"}
if !reflect.DeepEqual(docIds, expectedDocIds) {
t.Errorf("expected ids: %v, got ids: %v", expectedDocIds, docIds)
}
}
func TestIndexReader(t *testing.T) {
defer func() {
err := DestroyTest()
if err != nil {
t.Fatal(err)
}
}()
analysisQueue := index.NewAnalysisQueue(1)
idx, err := NewUpsideDownCouch(boltdb.Name, boltTestConfig, analysisQueue)
if err != nil {
t.Fatal(err)
}
err = idx.Open()
if err != nil {
t.Errorf("error opening index: %v", err)
}
defer func() {
err := idx.Close()
if err != nil {
t.Fatal(err)
}
}()
var expectedCount uint64
doc := document.NewDocument("1")
doc.AddField(document.NewTextField("name", []uint64{}, []byte("test")))
err = idx.Update(doc)
if err != nil {
t.Errorf("Error updating index: %v", err)
}
expectedCount++
doc = document.NewDocument("2")
doc.AddField(document.NewTextFieldWithAnalyzer("name", []uint64{}, []byte("test test test"), testAnalyzer))
doc.AddField(document.NewTextFieldCustom("desc", []uint64{}, []byte("eat more rice"), document.IndexField|document.IncludeTermVectors, testAnalyzer))
err = idx.Update(doc)
if err != nil {
t.Errorf("Error updating index: %v", err)
}
expectedCount++
indexReader, err := idx.Reader()
if err != nil {
t.Error(err)
}
defer func() {
err := indexReader.Close()
if err != nil {
t.Fatal(err)
}
}()
// first look for a term that doesn't exist
reader, err := indexReader.TermFieldReader([]byte("nope"), "name")
if err != nil {
t.Errorf("Error accessing term field reader: %v", err)
}
count := reader.Count()
if count != 0 {
t.Errorf("Expected doc count to be: %d got: %d", 0, count)
}
err = reader.Close()
if err != nil {
t.Fatal(err)
}
reader, err = indexReader.TermFieldReader([]byte("test"), "name")
if err != nil {
t.Errorf("Error accessing term field reader: %v", err)
}
expectedCount = 2
count = reader.Count()
if count != expectedCount {
t.Errorf("Exptected doc count to be: %d got: %d", expectedCount, count)
}
var match *index.TermFieldDoc
var actualCount uint64
match, err = reader.Next()
for err == nil && match != nil {
match, err = reader.Next()
if err != nil {
t.Errorf("unexpected error reading next")
}
actualCount++
}
if actualCount != count {
t.Errorf("count was 2, but only saw %d", actualCount)
}
expectedMatch := &index.TermFieldDoc{
ID: "2",
Freq: 1,
Norm: 0.5773502588272095,
Vectors: []*index.TermFieldVector{
&index.TermFieldVector{
Field: "desc",
Pos: 3,
Start: 9,
End: 13,
},
},
}
tfr, err := indexReader.TermFieldReader([]byte("rice"), "desc")
if err != nil {
t.Errorf("unexpected error: %v", err)
}
match, err = tfr.Next()
if err != nil {
t.Errorf("unexpected error: %v", err)
}
if !reflect.DeepEqual(expectedMatch, match) {
t.Errorf("got %#v, expected %#v", match, expectedMatch)
}
err = reader.Close()
if err != nil {
t.Fatal(err)
}
// now test usage of advance
reader, err = indexReader.TermFieldReader([]byte("test"), "name")
if err != nil {
t.Errorf("Error accessing term field reader: %v", err)
}
match, err = reader.Advance("2")
if err != nil {
t.Errorf("unexpected error: %v", err)
}
if match == nil {
t.Fatalf("Expected match, got nil")
}
if match.ID != "2" {
t.Errorf("Expected ID '2', got '%s'", match.ID)
}
match, err = reader.Advance("3")
if err != nil {
t.Errorf("unexpected error: %v", err)
}
if match != nil {
t.Errorf("expected nil, got %v", match)
}
err = reader.Close()
if err != nil {
t.Fatal(err)
}
// now test creating a reader for a field that doesn't exist
reader, err = indexReader.TermFieldReader([]byte("water"), "doesnotexist")
if err != nil {
t.Errorf("Error accessing term field reader: %v", err)
}
count = reader.Count()
if count != 0 {
t.Errorf("expected count 0 for reader of non-existant field")
}
match, err = reader.Next()
if err != nil {
t.Errorf("unexpected error: %v", err)
}
if match != nil {
t.Errorf("expected nil, got %v", match)
}
match, err = reader.Advance("anywhere")
if err != nil {
t.Errorf("unexpected error: %v", err)
}
if match != nil {
t.Errorf("expected nil, got %v", match)
}
}
func TestIndexDocIdReader(t *testing.T) {
defer func() {
err := DestroyTest()
if err != nil {
t.Fatal(err)
}
}()
analysisQueue := index.NewAnalysisQueue(1)
idx, err := NewUpsideDownCouch(boltdb.Name, boltTestConfig, analysisQueue)
if err != nil {
t.Fatal(err)
}
err = idx.Open()
if err != nil {
t.Errorf("error opening index: %v", err)
}
defer func() {
err := idx.Close()
if err != nil {
t.Fatal(err)
}
}()
var expectedCount uint64
doc := document.NewDocument("1")
doc.AddField(document.NewTextField("name", []uint64{}, []byte("test")))
err = idx.Update(doc)
if err != nil {
t.Errorf("Error updating index: %v", err)
}
expectedCount++
doc = document.NewDocument("2")
doc.AddField(document.NewTextField("name", []uint64{}, []byte("test test test")))
doc.AddField(document.NewTextFieldWithIndexingOptions("desc", []uint64{}, []byte("eat more rice"), document.IndexField|document.IncludeTermVectors))
err = idx.Update(doc)
if err != nil {
t.Errorf("Error updating index: %v", err)
}
expectedCount++
indexReader, err := idx.Reader()
if err != nil {
t.Error(err)
}
defer func() {
err := indexReader.Close()
if err != nil {
t.Error(err)
}
}()
// first get all doc ids
reader, err := indexReader.DocIDReader("", "")
if err != nil {
t.Errorf("Error accessing doc id reader: %v", err)
}
defer func() {
err := reader.Close()
if err != nil {
t.Fatal(err)
}
}()
id, err := reader.Next()
count := uint64(0)
for id != "" {
count++
id, err = reader.Next()
}
if count != expectedCount {
t.Errorf("expected %d, got %d", expectedCount, count)
}
// try it again, but jump to the second doc this time
reader2, err := indexReader.DocIDReader("", "")
if err != nil {
t.Errorf("Error accessing doc id reader: %v", err)
}
defer func() {
err := reader2.Close()
if err != nil {
t.Error(err)
}
}()
id, err = reader2.Advance("2")
if err != nil {
t.Error(err)
}
if id != "2" {
t.Errorf("expected to find id '2', got '%s'", id)
}
id, err = reader2.Advance("3")
if err != nil {
t.Error(err)
}
if id != "" {
t.Errorf("expected to find id '', got '%s'", id)
}
}
func TestIndexInsertUpdateDeleteWithMultipleTypesStored(t *testing.T) {
defer func() {
err := DestroyTest()
if err != nil {
t.Fatal(err)
}
}()
analysisQueue := index.NewAnalysisQueue(1)
idx, err := NewUpsideDownCouch(boltdb.Name, boltTestConfig, analysisQueue)
if err != nil {
t.Fatal(err)
}
err = idx.Open()
if err != nil {
t.Errorf("error opening index: %v", err)
}
defer func() {
err := idx.Close()
if err != nil {
t.Fatal(err)
}
}()
var expectedCount uint64
docCount, err := idx.DocCount()
if err != nil {
t.Error(err)
}
if docCount != expectedCount {
t.Errorf("Expected document count to be %d got %d", expectedCount, docCount)
}
doc := document.NewDocument("1")
doc.AddField(document.NewTextFieldWithIndexingOptions("name", []uint64{}, []byte("test"), document.IndexField|document.StoreField))
doc.AddField(document.NewNumericFieldWithIndexingOptions("age", []uint64{}, 35.99, document.IndexField|document.StoreField))
df, err := document.NewDateTimeFieldWithIndexingOptions("unixEpoch", []uint64{}, time.Unix(0, 0), document.IndexField|document.StoreField)
if err != nil {
t.Error(err)
}
doc.AddField(df)
err = idx.Update(doc)
if err != nil {
t.Errorf("Error updating index: %v", err)
}
expectedCount++
docCount, err = idx.DocCount()
if err != nil {
t.Error(err)
}
if docCount != expectedCount {
t.Errorf("Expected document count to be %d got %d", expectedCount, docCount)
}
// should have 72 rows
// 1 for version
// 3 for schema fields
// 1 for text term
// 16 for numeric terms
// 16 for date terms
// 3 for the stored field
// 1 for the text term count
// 16 for numeric term counts
// 16 for date term counts
// 1 for the back index entry
expectedLength := uint64(1 + 3 + 1 + (64 / document.DefaultPrecisionStep) + (64 / document.DefaultPrecisionStep) + 3 + 1 + (64 / document.DefaultPrecisionStep) + (64 / document.DefaultPrecisionStep) + 1)
rowCount, err := idx.(*UpsideDownCouch).rowCount()
if err != nil {
t.Error(err)
}
if rowCount != expectedLength {
t.Errorf("expected %d rows, got: %d", expectedLength, rowCount)
}
indexReader, err := idx.Reader()
if err != nil {
t.Error(err)
}
storedDoc, err := indexReader.Document("1")
if err != nil {
t.Error(err)
}
err = indexReader.Close()
if err != nil {
t.Error(err)
}
if len(storedDoc.Fields) != 3 {
t.Errorf("expected 3 stored field, got %d", len(storedDoc.Fields))
}
textField, ok := storedDoc.Fields[0].(*document.TextField)
if !ok {
t.Errorf("expected text field")
}
if string(textField.Value()) != "test" {
t.Errorf("expected field content 'test', got '%s'", string(textField.Value()))
}
numField, ok := storedDoc.Fields[1].(*document.NumericField)
if !ok {
t.Errorf("expected numeric field")
}
numFieldNumer, err := numField.Number()
if err != nil {
t.Error(err)
} else {
if numFieldNumer != 35.99 {
t.Errorf("expeted numeric value 35.99, got %f", numFieldNumer)
}
}
dateField, ok := storedDoc.Fields[2].(*document.DateTimeField)
if !ok {
t.Errorf("expected date field")
}
dateFieldDate, err := dateField.DateTime()
if err != nil {
t.Error(err)
} else {
if dateFieldDate != time.Unix(0, 0).UTC() {
t.Errorf("expected date value unix epoch, got %v", dateFieldDate)
}
}
// now update the document, but omit one of the fields
doc = document.NewDocument("1")
doc.AddField(document.NewTextFieldWithIndexingOptions("name", []uint64{}, []byte("testup"), document.IndexField|document.StoreField))
doc.AddField(document.NewNumericFieldWithIndexingOptions("age", []uint64{}, 36.99, document.IndexField|document.StoreField))
err = idx.Update(doc)
if err != nil {
t.Errorf("Error updating index: %v", err)
}
indexReader2, err := idx.Reader()
if err != nil {
t.Error(err)
}
// expected doc count shouldn't have changed
docCount = indexReader2.DocCount()
if docCount != expectedCount {
t.Errorf("Expected document count to be %d got %d", expectedCount, docCount)
}
// should only get 2 fields back now though
storedDoc, err = indexReader2.Document("1")
if err != nil {
t.Error(err)
}
err = indexReader2.Close()
if err != nil {
t.Error(err)
}
if len(storedDoc.Fields) != 2 {
t.Errorf("expected 3 stored field, got %d", len(storedDoc.Fields))
}
textField, ok = storedDoc.Fields[0].(*document.TextField)
if !ok {
t.Errorf("expected text field")
}
if string(textField.Value()) != "testup" {
t.Errorf("expected field content 'testup', got '%s'", string(textField.Value()))
}
numField, ok = storedDoc.Fields[1].(*document.NumericField)
if !ok {
t.Errorf("expected numeric field")
}
numFieldNumer, err = numField.Number()
if err != nil {
t.Error(err)
} else {
if numFieldNumer != 36.99 {
t.Errorf("expeted numeric value 36.99, got %f", numFieldNumer)
}
}
// now delete the document
err = idx.Delete("1")
expectedCount--
// expected doc count shouldn't have changed
docCount, err = idx.DocCount()
if err != nil {
t.Error(err)
}
if docCount != expectedCount {
t.Errorf("Expected document count to be %d got %d", expectedCount, docCount)
}
}
func TestIndexUpdateComposites(t *testing.T) {
defer func() {
err := DestroyTest()
if err != nil {
t.Fatal(err)
}
}()
analysisQueue := index.NewAnalysisQueue(1)
idx, err := NewUpsideDownCouch(boltdb.Name, boltTestConfig, analysisQueue)
if err != nil {
t.Fatal(err)
}
err = idx.Open()
if err != nil {
t.Errorf("error opening index: %v", err)
}
defer func() {
err := idx.Close()
if err != nil {
t.Fatal(err)
}
}()
doc := document.NewDocument("1")
doc.AddField(document.NewTextFieldWithIndexingOptions("name", []uint64{}, []byte("test"), document.IndexField|document.StoreField))
doc.AddField(document.NewTextFieldWithIndexingOptions("title", []uint64{}, []byte("mister"), document.IndexField|document.StoreField))
doc.AddField(document.NewCompositeFieldWithIndexingOptions("_all", true, nil, nil, document.IndexField))
err = idx.Update(doc)
if err != nil {
t.Errorf("Error updating index: %v", err)
}
// should have 72 rows
// 1 for version
// 3 for schema fields
// 4 for text term
// 2 for the stored field
// 4 for the text term count
// 1 for the back index entry
expectedLength := uint64(1 + 3 + 4 + 2 + 4 + 1)
rowCount, err := idx.(*UpsideDownCouch).rowCount()
if err != nil {
t.Error(err)
}
if rowCount != expectedLength {
t.Errorf("expected %d rows, got: %d", expectedLength, rowCount)
}
// now lets update it
doc = document.NewDocument("1")
doc.AddField(document.NewTextFieldWithIndexingOptions("name", []uint64{}, []byte("testupdated"), document.IndexField|document.StoreField))
doc.AddField(document.NewTextFieldWithIndexingOptions("title", []uint64{}, []byte("misterupdated"), document.IndexField|document.StoreField))
doc.AddField(document.NewCompositeFieldWithIndexingOptions("_all", true, nil, nil, document.IndexField))
err = idx.Update(doc)
if err != nil {
t.Errorf("Error updating index: %v", err)
}
indexReader, err := idx.Reader()
if err != nil {
t.Error(err)
}
defer func() {
err := indexReader.Close()
if err != nil {
t.Fatal(err)
}
}()
// make sure new values are in index
storedDoc, err := indexReader.Document("1")
if err != nil {
t.Error(err)
}
if len(storedDoc.Fields) != 2 {
t.Errorf("expected 2 stored field, got %d", len(storedDoc.Fields))
}
textField, ok := storedDoc.Fields[0].(*document.TextField)
if !ok {
t.Errorf("expected text field")
}
if string(textField.Value()) != "testupdated" {
t.Errorf("expected field content 'test', got '%s'", string(textField.Value()))
}
// should have the same row count as before, plus 4 term dictionary garbage rows
expectedLength += 4
rowCount, err = idx.(*UpsideDownCouch).rowCount()
if err != nil {
t.Error(err)
}
if rowCount != expectedLength {
t.Errorf("expected %d rows, got: %d", expectedLength, rowCount)
}
}
