mirror of
https://github.com/RyanGreenup/cadmus.git
synced 2025-08-06 22:26:33 +02:00
Using Batches considerably improves performance of search
This commit is contained in:
102
main.go
102
main.go
@@ -1,5 +1,10 @@
|
||||
package main
|
||||
|
||||
// TODO improve the logic for opening the index, see line 52 of this exemplar
|
||||
// TODO see line 117 for example of walking directory
|
||||
// https://github.com/blevesearch/beer-search/blob/master/main.go
|
||||
// TODO Improve index time with batch?
|
||||
// https://github.com/blevesearch/bleve/issues/831
|
||||
// TODO command line argument to re-index? that would be quicker
|
||||
// TODO just return the name of matches
|
||||
// TODO I need to evaluate how long tantivy takes to index, Ideally I want to index on the fly, I need to see if that's an option
|
||||
@@ -12,9 +17,12 @@ package main
|
||||
|
||||
import (
|
||||
"bufio"
|
||||
"flag"
|
||||
"fmt"
|
||||
"log"
|
||||
"os"
|
||||
"path/filepath"
|
||||
"time"
|
||||
|
||||
"github.com/blevesearch/bleve/v2"
|
||||
"github.com/schollz/progressbar/v3"
|
||||
@@ -25,9 +33,18 @@ type text_structure struct {
|
||||
Content string
|
||||
}
|
||||
|
||||
var batchSize = flag.Int("batchSize", 200, "batch size for indexing")
|
||||
|
||||
func main() {
|
||||
|
||||
files := listFiles()
|
||||
dir := ""
|
||||
if len(os.Args) < 2 {
|
||||
dir = "./"
|
||||
} else {
|
||||
dir = os.Args[1]
|
||||
}
|
||||
|
||||
files := listFiles(dir)
|
||||
|
||||
index_path := "example.bleve"
|
||||
|
||||
@@ -77,7 +94,9 @@ func Make_index(index_path string, files []string) bleve.Index {
|
||||
fmt.Println(err)
|
||||
fmt.Println("Creating a New index")
|
||||
mapping := bleve.NewIndexMapping()
|
||||
|
||||
index, err = bleve.New(index_path, mapping)
|
||||
|
||||
if err != nil {
|
||||
fmt.Print(err)
|
||||
fmt.Print("Unable to Create new index")
|
||||
@@ -87,59 +106,65 @@ func Make_index(index_path string, files []string) bleve.Index {
|
||||
fmt.Println("Appending to Old Index")
|
||||
}
|
||||
|
||||
notecontent := getFile("/home/ryan/Sync/Notes/Org/1.org")
|
||||
data3 := text_structure{
|
||||
Path: "1.org",
|
||||
Content: notecontent,
|
||||
}
|
||||
_ = data3
|
||||
var notecontent string
|
||||
var a_document text_structure
|
||||
|
||||
bar := progressbar.Default(int64(len(files)))
|
||||
documents := []text_structure{}
|
||||
i := 1
|
||||
count := 1
|
||||
startTime := time.Now()
|
||||
batchcount := 1
|
||||
batch := index.NewBatch()
|
||||
|
||||
// batch := index.NewBatch()
|
||||
for _, file := range files {
|
||||
// fmt.Println(file)
|
||||
i = i + 1
|
||||
count = count + 1
|
||||
|
||||
notecontent = getFile(file)
|
||||
|
||||
data3 = text_structure{
|
||||
a_document = text_structure{
|
||||
Path: file,
|
||||
Content: notecontent,
|
||||
}
|
||||
|
||||
// documents = append(documents, data3)
|
||||
documents = append(documents, a_document)
|
||||
// fmt.Println(file)
|
||||
|
||||
index.Index(data3.Path, data3)
|
||||
// index.Index(a_document.Path, a_document)
|
||||
// Add them to a batch
|
||||
batch.Index(a_document.Path, a_document)
|
||||
batchcount++
|
||||
|
||||
// Index the batch now
|
||||
if batchcount >= *batchSize {
|
||||
err = index.Batch(batch)
|
||||
if err != nil {
|
||||
panic(err)
|
||||
}
|
||||
// Reset the batch
|
||||
batch = index.NewBatch()
|
||||
batchcount = 0
|
||||
|
||||
}
|
||||
|
||||
bar.Add(1)
|
||||
|
||||
}
|
||||
|
||||
data1 := text_structure{
|
||||
Path: "/home/ryan/Notes/MD/index.md",
|
||||
Content: "lots of text in quick brown fox",
|
||||
// Index the last inclomplete batch
|
||||
if batchcount > 0 {
|
||||
err = index.Batch(batch)
|
||||
if err != nil {
|
||||
panic(err)
|
||||
}
|
||||
// no need to reset the batch, we are don
|
||||
|
||||
}
|
||||
|
||||
data2 := text_structure{
|
||||
Path: "1.org",
|
||||
Content: notecontent,
|
||||
}
|
||||
// TODO loop over files
|
||||
|
||||
documents = append(documents, data1)
|
||||
documents = append(documents, data2)
|
||||
|
||||
// for key, doc := range documents {
|
||||
// // first is string returned, second is struct searched
|
||||
// index.Index(doc.Path, doc)
|
||||
// }
|
||||
|
||||
// for i := 0; i < len(documents); i++ {
|
||||
// fmt.Print(documents[i])
|
||||
// }
|
||||
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 index
|
||||
}
|
||||
@@ -159,13 +184,13 @@ func delete_index(path string) {
|
||||
os.RemoveAll(path)
|
||||
}
|
||||
|
||||
func listFiles() []string {
|
||||
func listFiles(dir string) []string {
|
||||
files := []string{}
|
||||
|
||||
// TODO Why is this different?
|
||||
// https://stackoverflow.com/a/42423998
|
||||
// https://flaviocopes.com/go-list-files/
|
||||
echo := func(path string, info os.FileInfo, err error) error {
|
||||
append_files := func(path string, info os.FileInfo, err error) error {
|
||||
|
||||
if info.IsDir() {
|
||||
return nil
|
||||
@@ -177,15 +202,12 @@ func listFiles() []string {
|
||||
}
|
||||
|
||||
files = append(files, path)
|
||||
|
||||
// fmt.Print(path)
|
||||
return nil
|
||||
|
||||
}
|
||||
|
||||
// TODO Why not Symlinks?
|
||||
root := "/home/ryan/Sync/Notes/MD/"
|
||||
err := filepath.Walk(root, echo)
|
||||
root := dir
|
||||
err := filepath.Walk(root, append_files)
|
||||
if err != nil {
|
||||
panic(err)
|
||||
}
|
||||
|
Reference in New Issue
Block a user