gibheer/bleve
0
0
Fork 0
Code

Merge pull request #514 from steveyen/master

Browse Source 
more upsidedown optimizations
This commit is contained in:
Marty Schoch 2017-01-10 09:15:04 -05:00 committed by GitHub
commit 269cc302e3
1 changed files with 22 additions and 8 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

30
index/upsidedown/upsidedown.go
View File

@ -499,6 +499,15 @@ func (udc *UpsideDownCouch) Update(doc *document.Document) (err error) {
func (udc *UpsideDownCouch) mergeOldAndNew(backIndexRow *BackIndexRow, rows []index.IndexRow) (addRows []UpsideDownCouchRow, updateRows []UpsideDownCouchRow, deleteRows []UpsideDownCouchRow) {
addRows = make([]UpsideDownCouchRow, 0, len(rows))
if backIndexRow == nil {
addRows = addRows[0:len(rows)]
for i, row := range rows {
addRows[i] = row
}
return addRows, nil, nil
}
updateRows = make([]UpsideDownCouchRow, 0, len(rows))
deleteRows = make([]UpsideDownCouchRow, 0, len(rows))
@ -604,14 +613,18 @@ func encodeFieldType(f document.Field) byte {
func (udc *UpsideDownCouch) indexField(docID []byte, includeTermVectors bool, fieldIndex uint16, fieldLength int, tokenFreqs analysis.TokenFrequencies, rows []index.IndexRow, backIndexTermEntries []*BackIndexTermEntry) ([]index.IndexRow, []*BackIndexTermEntry) {
fieldNorm := float32(1.0 / math.Sqrt(float64(fieldLength)))
termFreqRows := make([]TermFrequencyRow, len(tokenFreqs))
termFreqRowsUsed := 0
for k, tf := range tokenFreqs {
var termFreqRow *TermFrequencyRow
termFreqRow := &termFreqRows[termFreqRowsUsed]
termFreqRowsUsed++
InitTermFrequencyRow(termFreqRow, tf.Term, fieldIndex, docID,
uint64(frequencyFromTokenFreq(tf)), fieldNorm)
if includeTermVectors {
var tv []*TermVector
tv, rows = udc.termVectorsFromTokenFreq(fieldIndex, tf, rows)
termFreqRow = NewTermFrequencyRowWithTermVectors(tf.Term, fieldIndex, docID, uint64(frequencyFromTokenFreq(tf)), fieldNorm, tv)
} else {
termFreqRow = NewTermFrequencyRow(tf.Term, fieldIndex, docID, uint64(frequencyFromTokenFreq(tf)), fieldNorm)
termFreqRow.vectors, rows = udc.termVectorsFromTokenFreq(fieldIndex, tf, rows)
}
// record the back index entry
@ -727,6 +740,7 @@ func frequencyFromTokenFreq(tf *analysis.TokenFreq) int {
}
func (udc *UpsideDownCouch) termVectorsFromTokenFreq(field uint16, tf *analysis.TokenFreq, rows []index.IndexRow) ([]*TermVector, []index.IndexRow) {
a := make([]TermVector, len(tf.Locations))
rv := make([]*TermVector, len(tf.Locations))
for i, l := range tf.Locations {
@ -739,14 +753,14 @@ func (udc *UpsideDownCouch) termVectorsFromTokenFreq(field uint16, tf *analysis.
rows = append(rows, newFieldRow)
}
}
tv := TermVector{
a[i] = TermVector{
field: fieldIndex,
arrayPositions: l.ArrayPositions,
pos: uint64(l.Position),
start: uint64(l.Start),
end: uint64(l.End),
}
rv[i] = &tv
rv[i] = &a[i]
}
return rv, rows