updated dependencies

This commit is contained in:
ston1th 2019-08-23 21:23:33 +02:00
commit 94ee84414c
176 changed files with 15726 additions and 4366 deletions

View file

@ -15,11 +15,20 @@
package upsidedown
import (
"reflect"
"github.com/blevesearch/bleve/document"
"github.com/blevesearch/bleve/index"
"github.com/blevesearch/bleve/index/store"
)
var reflectStaticSizeIndexReader int
func init() {
var ir IndexReader
reflectStaticSizeIndexReader = int(reflect.TypeOf(ir).Size())
}
type IndexReader struct {
index *UpsideDownCouch
kvreader store.KVReader
@ -201,3 +210,17 @@ func incrementBytes(in []byte) []byte {
}
return rv
}
func (i *IndexReader) DocValueReader(fields []string) (index.DocValueReader, error) {
return &DocValueReader{i: i, fields: fields}, nil
}
type DocValueReader struct {
i *IndexReader
fields []string
}
func (dvr *DocValueReader) VisitDocValues(id index.IndexInternalID,
visitor index.DocumentFieldTermVisitor) error {
return dvr.i.DocumentVisitFieldTerms(id, dvr.fields, visitor)
}

View file

@ -16,13 +16,27 @@ package upsidedown
import (
"bytes"
"reflect"
"sort"
"sync/atomic"
"github.com/blevesearch/bleve/index"
"github.com/blevesearch/bleve/index/store"
"github.com/blevesearch/bleve/size"
)
var reflectStaticSizeUpsideDownCouchTermFieldReader int
var reflectStaticSizeUpsideDownCouchDocIDReader int
func init() {
var tfr UpsideDownCouchTermFieldReader
reflectStaticSizeUpsideDownCouchTermFieldReader =
int(reflect.TypeOf(tfr).Size())
var cdr UpsideDownCouchDocIDReader
reflectStaticSizeUpsideDownCouchDocIDReader =
int(reflect.TypeOf(cdr).Size())
}
type UpsideDownCouchTermFieldReader struct {
count uint64
indexReader *IndexReader
@ -35,6 +49,19 @@ type UpsideDownCouchTermFieldReader struct {
includeTermVectors bool
}
func (r *UpsideDownCouchTermFieldReader) Size() int {
sizeInBytes := reflectStaticSizeUpsideDownCouchTermFieldReader + size.SizeOfPtr +
len(r.term) +
r.tfrPrealloc.Size() +
len(r.keyBuf)
if r.tfrNext != nil {
sizeInBytes += r.tfrNext.Size()
}
return sizeInBytes
}
func newUpsideDownCouchTermFieldReader(indexReader *IndexReader, term []byte, field uint16, includeFreq, includeNorm, includeTermVectors bool) (*UpsideDownCouchTermFieldReader, error) {
bufNeeded := termFrequencyRowKeySize(term, nil)
if bufNeeded < dictionaryRowKeySize(term) {
@ -174,8 +201,18 @@ type UpsideDownCouchDocIDReader struct {
onlyMode bool
}
func newUpsideDownCouchDocIDReader(indexReader *IndexReader) (*UpsideDownCouchDocIDReader, error) {
func (r *UpsideDownCouchDocIDReader) Size() int {
sizeInBytes := reflectStaticSizeUpsideDownCouchDocIDReader +
reflectStaticSizeIndexReader + size.SizeOfPtr
for _, entry := range r.only {
sizeInBytes += size.SizeOfString + len(entry)
}
return sizeInBytes
}
func newUpsideDownCouchDocIDReader(indexReader *IndexReader) (*UpsideDownCouchDocIDReader, error) {
startBytes := []byte{0x0}
endBytes := []byte{0xff}

View file

@ -20,10 +20,22 @@ import (
"fmt"
"io"
"math"
"reflect"
"github.com/blevesearch/bleve/size"
"github.com/golang/protobuf/proto"
)
var reflectStaticSizeTermFrequencyRow int
var reflectStaticSizeTermVector int
func init() {
var tfr TermFrequencyRow
reflectStaticSizeTermFrequencyRow = int(reflect.TypeOf(tfr).Size())
var tv TermVector
reflectStaticSizeTermVector = int(reflect.TypeOf(tv).Size())
}
const ByteSeparator byte = 0xff
type UpsideDownCouchRowStream chan UpsideDownCouchRow
@ -358,6 +370,11 @@ type TermVector struct {
end uint64
}
func (tv *TermVector) Size() int {
return reflectStaticSizeTermVector + size.SizeOfPtr +
len(tv.arrayPositions)*size.SizeOfUint64
}
func (tv *TermVector) String() string {
return fmt.Sprintf("Field: %d Pos: %d Start: %d End %d ArrayPositions: %#v", tv.field, tv.pos, tv.start, tv.end, tv.arrayPositions)
}
@ -371,6 +388,18 @@ type TermFrequencyRow struct {
field uint16
}
func (tfr *TermFrequencyRow) Size() int {
sizeInBytes := reflectStaticSizeTermFrequencyRow +
len(tfr.term) +
len(tfr.doc)
for _, entry := range tfr.vectors {
sizeInBytes += entry.Size()
}
return sizeInBytes
}
func (tfr *TermFrequencyRow) Term() []byte {
return tfr.term
}
@ -555,7 +584,7 @@ func (tfr *TermFrequencyRow) parseK(key []byte) error {
func (tfr *TermFrequencyRow) parseKDoc(key []byte, term []byte) error {
tfr.doc = key[3+len(term)+1:]
if len(tfr.doc) <= 0 {
if len(tfr.doc) == 0 {
return fmt.Errorf("invalid term frequency key, empty docid")
}

View file

@ -415,7 +415,6 @@ func (udc *UpsideDownCouch) Close() error {
func (udc *UpsideDownCouch) Update(doc *document.Document) (err error) {
// do analysis before acquiring write lock
analysisStart := time.Now()
numPlainTextBytes := doc.NumPlainTextBytes()
resultChan := make(chan *index.AnalysisResult)
aw := index.NewAnalysisWork(udc, doc, resultChan)
@ -452,6 +451,11 @@ func (udc *UpsideDownCouch) Update(doc *document.Document) (err error) {
return
}
return udc.UpdateWithAnalysis(doc, result, backIndexRow)
}
func (udc *UpsideDownCouch) UpdateWithAnalysis(doc *document.Document,
result *index.AnalysisResult, backIndexRow *BackIndexRow) (err error) {
// start a writer for this update
indexStart := time.Now()
var kvwriter store.KVWriter
@ -490,7 +494,7 @@ func (udc *UpsideDownCouch) Update(doc *document.Document) (err error) {
atomic.AddUint64(&udc.stats.indexTime, uint64(time.Since(indexStart)))
if err == nil {
atomic.AddUint64(&udc.stats.updates, 1)
atomic.AddUint64(&udc.stats.numPlainTextBytesIndexed, numPlainTextBytes)
atomic.AddUint64(&udc.stats.numPlainTextBytesIndexed, doc.NumPlainTextBytes())
} else {
atomic.AddUint64(&udc.stats.errors, 1)
}
@ -775,7 +779,7 @@ func (udc *UpsideDownCouch) termVectorsFromTokenFreq(field uint16, tf *analysis.
}
func (udc *UpsideDownCouch) termFieldVectorsFromTermVectors(in []*TermVector) []*index.TermFieldVector {
if len(in) <= 0 {
if len(in) == 0 {
return nil
}
@ -797,6 +801,10 @@ func (udc *UpsideDownCouch) termFieldVectorsFromTermVectors(in []*TermVector) []
}
func (udc *UpsideDownCouch) Batch(batch *index.Batch) (err error) {
persistedCallback := batch.PersistedCallback()
if persistedCallback != nil {
defer persistedCallback(err)
}
analysisStart := time.Now()
resultChan := make(chan *index.AnalysisResult, len(batch.IndexOps))
@ -810,15 +818,17 @@ func (udc *UpsideDownCouch) Batch(batch *index.Batch) (err error) {
}
}
go func() {
for _, doc := range batch.IndexOps {
if doc != nil {
aw := index.NewAnalysisWork(udc, doc, resultChan)
// put the work on the queue
udc.analysisQueue.Queue(aw)
if numUpdates > 0 {
go func() {
for _, doc := range batch.IndexOps {
if doc != nil {
aw := index.NewAnalysisWork(udc, doc, resultChan)
// put the work on the queue
udc.analysisQueue.Queue(aw)
}
}
}
}()
}()
}
// retrieve back index rows concurrent with analysis
docBackIndexRowErr := error(nil)
@ -958,6 +968,7 @@ func (udc *UpsideDownCouch) Batch(batch *index.Batch) (err error) {
} else {
atomic.AddUint64(&udc.stats.errors, 1)
}
return
}