Пример #1
0
func (udc *UpsideDownCouch) indexField(docID string, field document.Field, fieldIndex uint16, fieldLength int, tokenFreqs analysis.TokenFrequencies) ([]UpsideDownCouchRow, []*BackIndexTermEntry) {

	rows := make([]UpsideDownCouchRow, 0, 100)
	backIndexTermEntries := make([]*BackIndexTermEntry, 0)
	fieldNorm := float32(1.0 / math.Sqrt(float64(fieldLength)))

	for _, tf := range tokenFreqs {
		var termFreqRow *TermFrequencyRow
		if field.Options().IncludeTermVectors() {
			tv, newFieldRows := udc.termVectorsFromTokenFreq(fieldIndex, tf)
			rows = append(rows, newFieldRows...)
			termFreqRow = NewTermFrequencyRowWithTermVectors(tf.Term, fieldIndex, docID, uint64(frequencyFromTokenFreq(tf)), fieldNorm, tv)
		} else {
			termFreqRow = NewTermFrequencyRow(tf.Term, fieldIndex, docID, uint64(frequencyFromTokenFreq(tf)), fieldNorm)
		}

		// record the back index entry
		backIndexTermEntry := BackIndexTermEntry{Term: proto.String(string(tf.Term)), Field: proto.Uint32(uint32(fieldIndex))}
		backIndexTermEntries = append(backIndexTermEntries, &backIndexTermEntry)

		rows = append(rows, termFreqRow)
	}

	return rows, backIndexTermEntries
}
Пример #2
0
func (udc *UpsideDownCouch) indexField(docID string, field document.Field, fieldIndex uint16, fieldLength int, tokenFreqs analysis.TokenFrequencies, existingKeys map[string]bool) ([]UpsideDownCouchRow, []UpsideDownCouchRow, []*BackIndexTermEntry) {

	updateRows := make([]UpsideDownCouchRow, 0)
	addRows := make([]UpsideDownCouchRow, 0)
	backIndexTermEntries := make([]*BackIndexTermEntry, 0)
	fieldNorm := float32(1.0 / math.Sqrt(float64(fieldLength)))

	for _, tf := range tokenFreqs {
		var termFreqRow *TermFrequencyRow
		if field.Options().IncludeTermVectors() {
			tv, newFieldRows := udc.termVectorsFromTokenFreq(fieldIndex, tf)
			updateRows = append(updateRows, newFieldRows...)
			termFreqRow = NewTermFrequencyRowWithTermVectors(tf.Term, fieldIndex, docID, uint64(frequencyFromTokenFreq(tf)), fieldNorm, tv)
		} else {
			termFreqRow = NewTermFrequencyRow(tf.Term, fieldIndex, docID, uint64(frequencyFromTokenFreq(tf)), fieldNorm)
		}

		// record the back index entry
		backIndexTermEntry := BackIndexTermEntry{Term: proto.String(string(tf.Term)), Field: proto.Uint32(uint32(fieldIndex))}
		backIndexTermEntries = append(backIndexTermEntries, &backIndexTermEntry)

		tfrKeyString := string(termFreqRow.Key())
		_, existed := existingKeys[tfrKeyString]
		if existed {
			// this is an update
			updateRows = append(updateRows, termFreqRow)
			// this term existed last time, delete it from that map
			delete(existingKeys, tfrKeyString)
		} else {
			// this is an add
			addRows = append(addRows, termFreqRow)
		}
	}

	return addRows, updateRows, backIndexTermEntries
}