Support elastic search for code search (#10273)
* Support elastic search for code search * Finished elastic search implementation and add some tests * Enable test on drone and added docs * Add new fields to elastic search * Fix bug * remove unused changes * Use indexer alias to keep the gitea indexer version * Improve codes * Some code improvements * The real indexer name changed to xxx.v1 Co-authored-by: zeripath <art27@cantab.net>
This commit is contained in:
parent
d257485bc0
commit
9bc69ff26e
14 changed files with 694 additions and 164 deletions
|
@ -7,8 +7,11 @@ package code
|
|||
import (
|
||||
"context"
|
||||
"os"
|
||||
"strconv"
|
||||
"strings"
|
||||
"time"
|
||||
|
||||
"code.gitea.io/gitea/models"
|
||||
"code.gitea.io/gitea/modules/graceful"
|
||||
"code.gitea.io/gitea/modules/log"
|
||||
"code.gitea.io/gitea/modules/setting"
|
||||
|
@ -37,12 +40,33 @@ type SearchResultLanguages struct {
|
|||
|
||||
// Indexer defines an interface to indexer issues contents
|
||||
type Indexer interface {
|
||||
Index(repoID int64) error
|
||||
Index(repo *models.Repository, sha string, changes *repoChanges) error
|
||||
Delete(repoID int64) error
|
||||
Search(repoIDs []int64, language, keyword string, page, pageSize int) (int64, []*SearchResult, []*SearchResultLanguages, error)
|
||||
Close()
|
||||
}
|
||||
|
||||
func filenameIndexerID(repoID int64, filename string) string {
|
||||
return indexerID(repoID) + "_" + filename
|
||||
}
|
||||
|
||||
func parseIndexerID(indexerID string) (int64, string) {
|
||||
index := strings.IndexByte(indexerID, '_')
|
||||
if index == -1 {
|
||||
log.Error("Unexpected ID in repo indexer: %s", indexerID)
|
||||
}
|
||||
repoID, _ := strconv.ParseInt(indexerID[:index], 10, 64)
|
||||
return repoID, indexerID[index+1:]
|
||||
}
|
||||
|
||||
func filenameOfIndexerID(indexerID string) string {
|
||||
index := strings.IndexByte(indexerID, '_')
|
||||
if index == -1 {
|
||||
log.Error("Unexpected ID in repo indexer: %s", indexerID)
|
||||
}
|
||||
return indexerID[index+1:]
|
||||
}
|
||||
|
||||
// Init initialize the repo indexer
|
||||
func Init() {
|
||||
if !setting.Indexer.RepoIndexerEnabled {
|
||||
|
@ -63,33 +87,61 @@ func Init() {
|
|||
waitChannel := make(chan time.Duration)
|
||||
go func() {
|
||||
start := time.Now()
|
||||
log.Info("PID: %d Initializing Repository Indexer at: %s", os.Getpid(), setting.Indexer.RepoPath)
|
||||
defer func() {
|
||||
if err := recover(); err != nil {
|
||||
log.Error("PANIC whilst initializing repository indexer: %v\nStacktrace: %s", err, log.Stack(2))
|
||||
log.Error("The indexer files are likely corrupted and may need to be deleted")
|
||||
log.Error("You can completely remove the %q directory to make Gitea recreate the indexes", setting.Indexer.RepoPath)
|
||||
var (
|
||||
rIndexer Indexer
|
||||
populate bool
|
||||
err error
|
||||
)
|
||||
switch setting.Indexer.RepoType {
|
||||
case "bleve":
|
||||
log.Info("PID: %d Initializing Repository Indexer at: %s", os.Getpid(), setting.Indexer.RepoPath)
|
||||
defer func() {
|
||||
if err := recover(); err != nil {
|
||||
log.Error("PANIC whilst initializing repository indexer: %v\nStacktrace: %s", err, log.Stack(2))
|
||||
log.Error("The indexer files are likely corrupted and may need to be deleted")
|
||||
log.Error("You can completely remove the \"%s\" directory to make Gitea recreate the indexes", setting.Indexer.RepoPath)
|
||||
}
|
||||
}()
|
||||
|
||||
rIndexer, populate, err = NewBleveIndexer(setting.Indexer.RepoPath)
|
||||
if err != nil {
|
||||
if rIndexer != nil {
|
||||
rIndexer.Close()
|
||||
}
|
||||
cancel()
|
||||
indexer.Close()
|
||||
close(waitChannel)
|
||||
log.Fatal("PID: %d Unable to initialize the Repository Indexer at path: %s Error: %v", os.Getpid(), setting.Indexer.RepoPath, err)
|
||||
log.Fatal("PID: %d Unable to initialize the bleve Repository Indexer at path: %s Error: %v", os.Getpid(), setting.Indexer.RepoPath, err)
|
||||
}
|
||||
}()
|
||||
bleveIndexer, created, err := NewBleveIndexer(setting.Indexer.RepoPath)
|
||||
if err != nil {
|
||||
if bleveIndexer != nil {
|
||||
bleveIndexer.Close()
|
||||
case "elasticsearch":
|
||||
log.Info("PID: %d Initializing Repository Indexer at: %s", os.Getpid(), setting.Indexer.RepoConnStr)
|
||||
defer func() {
|
||||
if err := recover(); err != nil {
|
||||
log.Error("PANIC whilst initializing repository indexer: %v\nStacktrace: %s", err, log.Stack(2))
|
||||
log.Error("The indexer files are likely corrupted and may need to be deleted")
|
||||
log.Error("You can completely remove the \"%s\" index to make Gitea recreate the indexes", setting.Indexer.RepoConnStr)
|
||||
}
|
||||
}()
|
||||
|
||||
rIndexer, populate, err = NewElasticSearchIndexer(setting.Indexer.RepoConnStr, setting.Indexer.RepoIndexerName)
|
||||
if err != nil {
|
||||
if rIndexer != nil {
|
||||
rIndexer.Close()
|
||||
}
|
||||
cancel()
|
||||
indexer.Close()
|
||||
close(waitChannel)
|
||||
log.Fatal("PID: %d Unable to initialize the elasticsearch Repository Indexer connstr: %s Error: %v", os.Getpid(), setting.Indexer.RepoConnStr, err)
|
||||
}
|
||||
cancel()
|
||||
indexer.Close()
|
||||
close(waitChannel)
|
||||
log.Fatal("PID: %d Unable to initialize the Repository Indexer at path: %s Error: %v", os.Getpid(), setting.Indexer.RepoPath, err)
|
||||
default:
|
||||
log.Fatal("PID: %d Unknown Indexer type: %s", os.Getpid(), setting.Indexer.RepoType)
|
||||
}
|
||||
indexer.set(bleveIndexer)
|
||||
|
||||
indexer.set(rIndexer)
|
||||
|
||||
go processRepoIndexerOperationQueue(indexer)
|
||||
|
||||
if created {
|
||||
if populate {
|
||||
go populateRepoIndexer()
|
||||
}
|
||||
select {
|
||||
|
|
Loading…
Add table
Add a link
Reference in a new issue