feat(search): search with meilisearch (#6060)

* feat(search): search with meilisearch.

* feat(search): meilisearch supports auto update.

* chores: remove utils.Log.

* fix(search): the null pointer caused by deleting non-existing file/folder indexes.

---------

Co-authored-by: Andy Hsu <i@nn.ci>
This commit is contained in:
不插电
2024-02-23 15:37:40 +08:00
committed by GitHub
parent 1f835502ba
commit f1979a8bbc
8 changed files with 377 additions and 3 deletions

View File

@ -145,7 +145,7 @@ func InitialSettings() []model.SettingItem {
// single settings
{Key: conf.Token, Value: token, Type: conf.TypeString, Group: model.SINGLE, Flag: model.PRIVATE},
{Key: conf.SearchIndex, Value: "none", Type: conf.TypeSelect, Options: "database,database_non_full_text,bleve,none", Group: model.INDEX},
{Key: conf.SearchIndex, Value: "none", Type: conf.TypeSelect, Options: "database,database_non_full_text,bleve,meilisearch,none", Group: model.INDEX},
{Key: conf.AutoUpdateIndex, Value: "false", Type: conf.TypeBool, Group: model.INDEX},
{Key: conf.IgnorePaths, Value: "", Type: conf.TypeText, Group: model.INDEX, Flag: model.PRIVATE, Help: `one path per line`},
{Key: conf.MaxIndexDepth, Value: "20", Type: conf.TypeNumber, Group: model.INDEX, Flag: model.PRIVATE, Help: `max depth of index`},

View File

@ -1,10 +1,9 @@
package conf
import (
"path/filepath"
"github.com/alist-org/alist/v3/cmd/flags"
"github.com/alist-org/alist/v3/pkg/utils/random"
"path/filepath"
)
type Database struct {
@ -20,6 +19,12 @@ type Database struct {
DSN string `json:"dsn" env:"DSN"`
}
type Meilisearch struct {
Host string `json:"host" env:"HOST"`
APIKey string `json:"api_key" env:"API_KEY"`
IndexPrefix string `json:"index_prefix" env:"INDEX_PREFIX"`
}
type Scheme struct {
Address string `json:"address" env:"ADDR"`
HttpPort int `json:"http_port" env:"HTTP_PORT"`
@ -65,6 +70,7 @@ type Config struct {
JwtSecret string `json:"jwt_secret" env:"JWT_SECRET"`
TokenExpiresIn int `json:"token_expires_in" env:"TOKEN_EXPIRES_IN"`
Database Database `json:"database" envPrefix:"DB_"`
Meilisearch Meilisearch `json:"meilisearch" env:"MEILISEARCH"`
Scheme Scheme `json:"scheme"`
TempDir string `json:"temp_dir" env:"TEMP_DIR"`
BleveDir string `json:"bleve_dir" env:"BLEVE_DIR"`
@ -101,6 +107,9 @@ func DefaultConfig() *Config {
TablePrefix: "x_",
DBFile: dbPath,
},
Meilisearch: Meilisearch{
Host: "http://localhost:7700",
},
BleveDir: indexDir,
Log: LogConfig{
Enable: true,

View File

@ -4,4 +4,5 @@ import (
_ "github.com/alist-org/alist/v3/internal/search/bleve"
_ "github.com/alist-org/alist/v3/internal/search/db"
_ "github.com/alist-org/alist/v3/internal/search/db_non_full_text"
_ "github.com/alist-org/alist/v3/internal/search/meilisearch"
)

View File

@ -0,0 +1,89 @@
package meilisearch
import (
"errors"
"fmt"
"github.com/alist-org/alist/v3/internal/conf"
"github.com/alist-org/alist/v3/internal/model"
"github.com/alist-org/alist/v3/internal/search/searcher"
"github.com/alist-org/alist/v3/pkg/utils"
"github.com/meilisearch/meilisearch-go"
)
var config = searcher.Config{
Name: "meilisearch",
AutoUpdate: true,
}
func init() {
searcher.RegisterSearcher(config, func() (searcher.Searcher, error) {
m := Meilisearch{
Client: meilisearch.NewClient(meilisearch.ClientConfig{
Host: conf.Conf.Meilisearch.Host,
APIKey: conf.Conf.Meilisearch.APIKey,
}),
IndexUid: conf.Conf.Meilisearch.IndexPrefix + "alist",
FilterableAttributes: []string{"parent", "is_dir", "name"},
SearchableAttributes: []string{"name"},
}
_, err := m.Client.GetIndex(m.IndexUid)
if err != nil {
var mErr *meilisearch.Error
ok := errors.As(err, &mErr)
if ok && mErr.MeilisearchApiError.Code == "index_not_found" {
task, err := m.Client.CreateIndex(&meilisearch.IndexConfig{
Uid: m.IndexUid,
PrimaryKey: "id",
})
if err != nil {
return nil, err
}
forTask, err := m.Client.WaitForTask(task.TaskUID)
if err != nil {
return nil, err
}
if forTask.Status != meilisearch.TaskStatusSucceeded {
return nil, fmt.Errorf("index creation failed, task status is %s", forTask.Status)
}
} else {
return nil, err
}
}
attributes, err := m.Client.Index(m.IndexUid).GetFilterableAttributes()
if err != nil {
return nil, err
}
if attributes == nil || !utils.SliceAllContains(*attributes, m.FilterableAttributes...) {
_, err = m.Client.Index(m.IndexUid).UpdateFilterableAttributes(&m.FilterableAttributes)
if err != nil {
return nil, err
}
}
attributes, err = m.Client.Index(m.IndexUid).GetSearchableAttributes()
if err != nil {
return nil, err
}
if attributes == nil || !utils.SliceAllContains(*attributes, m.SearchableAttributes...) {
_, err = m.Client.Index(m.IndexUid).UpdateSearchableAttributes(&m.SearchableAttributes)
if err != nil {
return nil, err
}
}
pagination, err := m.Client.Index(m.IndexUid).GetPagination()
if err != nil {
return nil, err
}
if pagination.MaxTotalHits != int64(model.MaxInt) {
_, err := m.Client.Index(m.IndexUid).UpdatePagination(&meilisearch.Pagination{
MaxTotalHits: int64(model.MaxInt),
})
if err != nil {
return nil, err
}
}
return &m, nil
})
}

View File

@ -0,0 +1,227 @@
package meilisearch
import (
"context"
"fmt"
"github.com/alist-org/alist/v3/internal/model"
"github.com/alist-org/alist/v3/internal/search/searcher"
"github.com/alist-org/alist/v3/pkg/utils"
"github.com/google/uuid"
"github.com/meilisearch/meilisearch-go"
"path"
"strings"
"time"
)
type searchDocument struct {
ID string `json:"id"`
model.SearchNode
}
type Meilisearch struct {
Client *meilisearch.Client
IndexUid string
FilterableAttributes []string
SearchableAttributes []string
}
func (m *Meilisearch) Config() searcher.Config {
return config
}
func (m *Meilisearch) Search(ctx context.Context, req model.SearchReq) ([]model.SearchNode, int64, error) {
mReq := &meilisearch.SearchRequest{
AttributesToSearchOn: m.SearchableAttributes,
Page: int64(req.Page),
HitsPerPage: int64(req.PerPage),
}
if req.Scope != 0 {
mReq.Filter = fmt.Sprintf("is_dir = %v", req.Scope == 1)
}
search, err := m.Client.Index(m.IndexUid).Search(req.Keywords, mReq)
if err != nil {
return nil, 0, err
}
nodes, err := utils.SliceConvert(search.Hits, func(src any) (model.SearchNode, error) {
srcMap := src.(map[string]any)
return model.SearchNode{
Parent: srcMap["parent"].(string),
Name: srcMap["name"].(string),
IsDir: srcMap["is_dir"].(bool),
Size: int64(srcMap["size"].(float64)),
}, nil
})
if err != nil {
return nil, 0, err
}
return nodes, search.TotalHits, nil
}
func (m *Meilisearch) Index(ctx context.Context, node model.SearchNode) error {
return m.BatchIndex(ctx, []model.SearchNode{node})
}
func (m *Meilisearch) BatchIndex(ctx context.Context, nodes []model.SearchNode) error {
documents, _ := utils.SliceConvert(nodes, func(src model.SearchNode) (*searchDocument, error) {
return &searchDocument{
ID: uuid.NewString(),
SearchNode: src,
}, nil
})
_, err := m.Client.Index(m.IndexUid).AddDocuments(documents)
if err != nil {
return err
}
//// Wait for the task to complete and check
//forTask, err := m.Client.WaitForTask(task.TaskUID, meilisearch.WaitParams{
// Context: ctx,
// Interval: time.Millisecond * 50,
//})
//if err != nil {
// return err
//}
//if forTask.Status != meilisearch.TaskStatusSucceeded {
// return fmt.Errorf("BatchIndex failed, task status is %s", forTask.Status)
//}
return nil
}
func (m *Meilisearch) getDocumentsByParent(ctx context.Context, parent string) ([]*searchDocument, error) {
var result meilisearch.DocumentsResult
err := m.Client.Index(m.IndexUid).GetDocuments(&meilisearch.DocumentsQuery{
Filter: fmt.Sprintf("parent = '%s'", strings.ReplaceAll(parent, "'", "\\'")),
Limit: int64(model.MaxInt),
}, &result)
if err != nil {
return nil, err
}
return utils.SliceConvert(result.Results, func(src map[string]any) (*searchDocument, error) {
return &searchDocument{
ID: src["id"].(string),
SearchNode: model.SearchNode{
Parent: src["parent"].(string),
Name: src["name"].(string),
IsDir: src["is_dir"].(bool),
Size: int64(src["size"].(float64)),
},
}, nil
})
}
func (m *Meilisearch) Get(ctx context.Context, parent string) ([]model.SearchNode, error) {
result, err := m.getDocumentsByParent(ctx, parent)
if err != nil {
return nil, err
}
return utils.SliceConvert(result, func(src *searchDocument) (model.SearchNode, error) {
return src.SearchNode, nil
})
}
func (m *Meilisearch) getParentsByPrefix(ctx context.Context, parent string) ([]string, error) {
select {
case <-ctx.Done():
return nil, ctx.Err()
default:
parents := []string{parent}
get, err := m.getDocumentsByParent(ctx, parent)
if err != nil {
return nil, err
}
for _, node := range get {
if node.IsDir {
arr, err := m.getParentsByPrefix(ctx, path.Join(node.Parent, node.Name))
if err != nil {
return nil, err
}
parents = append(parents, arr...)
}
}
return parents, nil
}
}
func (m *Meilisearch) DelDirChild(ctx context.Context, prefix string) error {
dfs, err := m.getParentsByPrefix(ctx, utils.FixAndCleanPath(prefix))
if err != nil {
return err
}
utils.SliceReplace(dfs, func(src string) string {
return "'" + strings.ReplaceAll(src, "'", "\\'") + "'"
})
s := fmt.Sprintf("parent IN [%s]", strings.Join(dfs, ","))
task, err := m.Client.Index(m.IndexUid).DeleteDocumentsByFilter(s)
if err != nil {
return err
}
taskStatus, err := m.getTaskStatus(ctx, task.TaskUID)
if err != nil {
return err
}
if taskStatus != meilisearch.TaskStatusSucceeded {
return fmt.Errorf("DelDir failed, task status is %s", taskStatus)
}
return nil
}
func (m *Meilisearch) Del(ctx context.Context, prefix string) error {
prefix = utils.FixAndCleanPath(prefix)
dir, name := path.Split(prefix)
get, err := m.getDocumentsByParent(ctx, dir[:len(dir)-1])
if err != nil {
return err
}
var document *searchDocument
for _, v := range get {
if v.Name == name {
document = v
break
}
}
if document == nil {
// Defensive programming. Document may be the folder, try deleting Child
return m.DelDirChild(ctx, prefix)
}
if document.IsDir {
err = m.DelDirChild(ctx, prefix)
if err != nil {
return err
}
}
task, err := m.Client.Index(m.IndexUid).DeleteDocument(document.ID)
if err != nil {
return err
}
taskStatus, err := m.getTaskStatus(ctx, task.TaskUID)
if err != nil {
return err
}
if taskStatus != meilisearch.TaskStatusSucceeded {
return fmt.Errorf("DelDir failed, task status is %s", taskStatus)
}
return nil
}
func (m *Meilisearch) Release(ctx context.Context) error {
return nil
}
func (m *Meilisearch) Clear(ctx context.Context) error {
_, err := m.Client.Index(m.IndexUid).DeleteAllDocuments()
return err
}
func (m *Meilisearch) getTaskStatus(ctx context.Context, taskUID int64) (meilisearch.TaskStatus, error) {
forTask, err := m.Client.WaitForTask(taskUID, meilisearch.WaitParams{
Context: ctx,
Interval: time.Second,
})
if err != nil {
return meilisearch.TaskStatusUnknown, err
}
return forTask.Status, nil
}