func batchIndexingWorker(index bleve.Index, workChan chan *Work, start time.Time) {
for {
select {
case work, ok := <-workChan:
if !ok {
return
}
workSize := 1
if work.batch != nil {
err := index.Batch(work.batch)
if err != nil {
log.Fatalf("indexer worker fatal: %v", err)
}
workSize = work.batch.Size()
} else {
err := index.Index(work.id, work.doc)
if err != nil {
log.Fatalf("indexer worker fatal: %v", err)
}
}
elapsedTime := time.Since(start) / time.Millisecond
updatedTotal := atomic.AddUint64(&totalIndexed, uint64(workSize))
atomic.AddUint64(&totalPlainTextIndexed, work.plainTextBytes)
if updatedTotal%uint64(*printCount) == 0 {
log.Printf("%d,%d", updatedTotal, elapsedTime)
}
}
}
}
func indexSite(index bleve.Index, maxBatchSize int) error {
startT := time.Now()
count := 0
batch := index.NewBatch()
batchSize := 10
log.Printf("Walking %s", path.Join(htdocs, "repositories"))
err := filepath.Walk(path.Join(htdocs, "repositories"), func(p string, f os.FileInfo, err error) error {
if strings.Contains(p, "/accessions/") == true && strings.HasSuffix(p, ".json") == true {
src, err := ioutil.ReadFile(p)
if err != nil {
log.Printf("Can't read %s, %s", p, err)
return nil
}
view := new(cait.NormalizedAccessionView)
err = json.Unmarshal(src, &view)
if err != nil {
log.Printf("Can't parse %s, %s", p, err)
return nil
}
// Trim the htdocs and trailing .json extension
//log.Printf("Queued %s", p)
err = batch.Index(strings.TrimSuffix(strings.TrimPrefix(p, htdocs), "json"), view)
if err != nil {
log.Printf("Indexing error %s, %s", p, err)
return nil
}
if batch.Size() >= batchSize {
log.Printf("Indexing %d items", batch.Size())
err := index.Batch(batch)
if err != nil {
log.Fatal(err)
}
count += batch.Size()
batch = index.NewBatch()
log.Printf("Indexed: %d items, batch size %d, running %s\n", count, batchSize, time.Now().Sub(startT))
if batchSize < maxBatchSize {
batchSize = batchSize * 2
}
if batchSize > maxBatchSize {
batchSize = maxBatchSize
}
}
}
return nil
})
if batch.Size() > 0 {
log.Printf("Indexing %d items", batch.Size())
err := index.Batch(batch)
if err != nil {
log.Fatal(err)
}
count += batch.Size()
log.Printf("Indexed: %d items, batch size %d, running %s\n", count, batchSize, time.Now().Sub(startT))
}
log.Printf("Total indexed: %d times, total run time %s\n", count, time.Now().Sub(startT))
return err
}
func indexBeer(i bleve.Index) error {
// open the directory
dirEntries, err := ioutil.ReadDir(*jsonDir)
if err != nil {
return err
}
// walk the directory entries for indexing
log.Printf("Indexing...")
count := 0
startTime := time.Now()
batch := i.NewBatch()
batchCount := 0
for _, dirEntry := range dirEntries {
filename := dirEntry.Name()
// read the bytes
jsonBytes, err := ioutil.ReadFile(*jsonDir + "/" + filename)
if err != nil {
return err
}
// // shred them into a document
ext := filepath.Ext(filename)
docId := filename[:(len(filename) - len(ext))]
batch.Index(docId, jsonBytes)
batchCount++
if batchCount >= *batchSize {
err = i.Batch(batch)
if err != nil {
return err
}
batch = i.NewBatch()
batchCount = 0
}
count++
if count%1000 == 0 {
indexDuration := time.Since(startTime)
indexDurationSeconds := float64(indexDuration) / float64(time.Second)
timePerDoc := float64(indexDuration) / float64(count)
log.Printf("Indexed %d documents, in %.2fs (average %.2fms/doc)", count, indexDurationSeconds, timePerDoc/float64(time.Millisecond))
}
}
// flush the last batch
if batchCount > 0 {
err = i.Batch(batch)
if err != nil {
log.Fatal(err)
}
}
indexDuration := time.Since(startTime)
indexDurationSeconds := float64(indexDuration) / float64(time.Second)
timePerDoc := float64(indexDuration) / float64(count)
log.Printf("Indexed %d documents, in %.2fs (average %.2fms/doc)", count, indexDurationSeconds, timePerDoc/float64(time.Millisecond))
return nil
}
func batchIndexingWorker(index bleve.Index, workChan chan *Work, timeStart time.Time) {
for work := range workChan {
workSize := 1
if work.batch != nil {
err := index.Batch(work.batch)
if err != nil {
log.Fatalf("indexer worker fatal: %v", err)
}
workSize = work.batch.Size()
} else {
err := index.Index(work.id, work.doc)
if err != nil {
log.Fatalf("indexer worker fatal: %v", err)
}
}
atomic.AddUint64(&totalIndexed, uint64(workSize))
atomic.AddUint64(&totalPlainTextIndexed, work.plainTextBytes)
}
}
func (be *BleveEngine) BatchIndex(documents []*Document) (int64, error) {
start := time.Now().UnixNano() / int64(time.Millisecond)
var index bleve.Index
mapping := bleve.NewIndexMapping()
index, err := bleve.New(INDEX, mapping)
if err != nil {
index, _ = bleve.Open(INDEX)
}
batch := index.NewBatch()
for _, document := range documents {
batch.Index(document.Id, document.Data)
}
index.Batch(batch)
index.Close()
return time.Now().UnixNano()/int64(time.Millisecond) - start, nil
}
func indexTPB(i bleve.Index) error {
batch := bleve.NewBatch()
batchCount := 0
gzDumpFile, err := os.Open(*dump)
if err != nil {
return err
}
defer gzDumpFile.Close()
dumpFile, err := gzip.NewReader(gzDumpFile)
if err != nil {
return err
}
reader := csv.NewReader(dumpFile)
reader.FieldsPerRecord = 7
reader.Comma = '|'
count := 0
startTime := time.Now()
log.Printf("Indexing...")
for {
r, err := reader.Read()
if err == io.EOF {
break
} else if err != nil {
continue
}
size, err := strconv.ParseInt(r[1], 10, 0)
if err != nil {
fmt.Println("%#v", size)
size = 0
}
batch.Index(r[2], tpbDoc{
Name: r[0],
Size: size,
Hash: r[2],
Category: r[4],
Type: "torrent",
})
batchCount++
if batchCount >= *batchSize {
err = i.Batch(batch)
if err != nil {
return err
}
batch = bleve.NewBatch()
batchCount = 0
}
count++
if count%1000 == 0 {
indexDuration := time.Since(startTime)
indexDurationSeconds := float64(indexDuration) / float64(time.Second)
timePerDoc := float64(indexDuration) / float64(count)
log.Printf("Indexed %d documents in %.2fs (average %.2fms/doc)", count, indexDurationSeconds, timePerDoc/float64(time.Millisecond))
}
if *indexLimit > 0 && count >= *indexLimit {
break
}
}
// flush the last batch
if batchCount > 0 {
err := i.Batch(batch)
if err != nil {
log.Fatal(err)
}
}
indexDuration := time.Since(startTime)
indexDurationSeconds := float64(indexDuration) / float64(time.Second)
timePerDoc := float64(indexDuration) / float64(count)
log.Printf("Finished indexing %d documents in %.2fs (average %.2fms/doc)", count, indexDurationSeconds, timePerDoc/float64(time.Millisecond))
log.Printf("Still listening on http://%v", bind)
return nil
}
