Optimized import, only updating changed data and purging old data

This commit is contained in:
Deluan
2016-03-08 14:18:17 -05:00
parent df9687bf02
commit 46e7627fd3
13 changed files with 253 additions and 55 deletions
+18
View File
@@ -2,6 +2,7 @@ package persistence
import (
"errors"
"github.com/deluan/gosonic/domain"
)
@@ -40,4 +41,21 @@ func (r *albumRepository) GetAll(options domain.QueryOptions) (domain.Albums, er
return as, err
}
func (r *albumRepository) PurgeInactive(active *domain.Albums) error {
currentIds, err := r.GetAllIds()
if err != nil {
return err
}
for _, a := range *active {
currentIds[a.Id] = false
}
inactiveIds := make(map[string]bool)
for id, inactive := range currentIds {
if inactive {
inactiveIds[id] = true
}
}
return r.DeleteAll(inactiveIds)
}
var _ domain.AlbumRepository = (*albumRepository)(nil)
+18
View File
@@ -2,6 +2,7 @@ package persistence
import (
"errors"
"github.com/deluan/gosonic/domain"
)
@@ -33,4 +34,21 @@ func (r *artistRepository) GetByName(name string) (*domain.Artist, error) {
return r.Get(id)
}
func (r *artistRepository) PurgeInactive(active *domain.Artists) error {
currentIds, err := r.GetAllIds()
if err != nil {
return err
}
for _, a := range *active {
currentIds[a.Id] = false
}
inactiveIds := make(map[string]bool)
for id, inactive := range currentIds {
if inactive {
inactiveIds[id] = true
}
}
return r.DeleteAll(inactiveIds)
}
var _ domain.ArtistRepository = (*artistRepository)(nil)
+90 -13
View File
@@ -4,17 +4,20 @@ import (
"crypto/md5"
"encoding/json"
"fmt"
"reflect"
"strings"
"github.com/deluan/gosonic/domain"
"github.com/deluan/gosonic/utils"
"github.com/siddontang/ledisdb/ledis"
"reflect"
"strings"
)
type ledisRepository struct {
table string
entityType reflect.Type
fieldNames []string
table string
entityType reflect.Type
fieldNames []string
parentTable string
parentIdField string
}
func (r *ledisRepository) init(table string, entity interface{}) {
@@ -28,6 +31,7 @@ func (r *ledisRepository) init(table string, entity interface{}) {
r.fieldNames[i] = k
i++
}
r.parentTable, r.parentIdField, _ = r.getParent(entity)
}
// TODO Use annotations to specify fields to be used
@@ -41,6 +45,62 @@ func (r *ledisRepository) CountAll() (int64, error) {
return size, err
}
func (r *ledisRepository) GetAllIds() (map[string]bool, error) {
m := make(map[string]bool)
pairs, err := db().ZRange([]byte(r.table+"s:all"), 0, -1)
if err != nil {
return m, err
}
for _, p := range pairs {
m[string(p.Member)] = true
}
return m, err
}
func (r *ledisRepository) DeleteAll(ids map[string]bool) error {
allKey := r.table + "s:all"
keys := make([][]byte, len(ids))
i := 0
for id, _ := range ids {
// Delete from parent:parentId:table (ZSet)
if r.parentTable != "" {
parentKey := []byte(fmt.Sprintf("%s:%s:%s", r.table, id, r.parentIdField))
pid, err := db().Get(parentKey)
var parentId string
if err := json.Unmarshal(pid, &parentId); err != nil {
return err
}
if err != nil {
return err
}
parentKey = []byte(fmt.Sprintf("%s:%s:%ss", r.parentTable, parentId, r.table))
if _, err := db().ZRem(parentKey, []byte(id)); err != nil {
return err
}
}
// Delete record table:id:* (KV)
if err := r.deleteRecord(id); err != nil {
return err
}
keys[i] = []byte(id)
i++
}
// Delete from table:all (ZSet)
_, err := db().ZRem([]byte(allKey), keys...)
return err
}
func (r *ledisRepository) deleteRecord(id string) error {
keys := r.getFieldKeys(id)
_, err := db().Del(keys...)
return err
}
func (r *ledisRepository) Exists(id string) (bool, error) {
res, _ := db().ZScore([]byte(r.table+"s:all"), []byte(id))
return res != ledis.InvalidScore, nil
@@ -68,25 +128,42 @@ func (r *ledisRepository) saveOrUpdate(id string, entity interface{}) error {
return err
}
if parentTable, parentId := r.getParent(entity); parentTable != "" {
parentCollectionKey := fmt.Sprintf("%s:%s:%ss", parentTable, parentId, r.table)
if parentCollectionKey := r.getParentRelationKey(entity); parentCollectionKey != "" {
_, err = db().ZAdd([]byte(parentCollectionKey), sid)
}
return nil
}
func (r *ledisRepository) getParentRelationKey(entity interface{}) string {
parentId := r.getParentId(entity)
if parentId != "" {
return fmt.Sprintf("%s:%s:%ss", r.parentTable, parentId, r.table)
}
return ""
}
// TODO Optimize
func (r *ledisRepository) getParent(entity interface{}) (table string, id string) {
func (r *ledisRepository) getParent(entity interface{}) (table string, idField string, id string) {
dt := reflect.TypeOf(entity).Elem()
for i := 0; i < dt.NumField(); i++ {
f := dt.Field(i)
table := f.Tag.Get("parent")
table = f.Tag.Get("parent")
if table != "" {
idField = f.Name
dv := reflect.ValueOf(entity).Elem()
return table, dv.FieldByName(f.Name).String()
id = dv.FieldByName(f.Name).String()
return
}
}
return "", ""
return
}
func (r *ledisRepository) getParentId(entity interface{}) string {
if r.parentTable != "" {
dv := reflect.ValueOf(entity).Elem()
return dv.FieldByName(r.parentIdField).String()
}
return ""
}
func (r *ledisRepository) getFieldKeys(id string) [][]byte {
@@ -135,9 +212,9 @@ func (r *ledisRepository) loadAll(entities interface{}, qo ...domain.QueryOption
return r.loadFromSet(setName, entities, qo...)
}
func (r *ledisRepository) loadChildren(parentTable string, parentId string, entities interface{}, qo ...domain.QueryOptions) error {
func (r *ledisRepository) loadChildren(parentTable string, parentId string, emptyEntityArray interface{}, qo ...domain.QueryOptions) error {
setName := fmt.Sprintf("%s:%s:%ss", parentTable, parentId, r.table)
return r.loadFromSet(setName, entities, qo...)
return r.loadFromSet(setName, emptyEntityArray, qo...)
}
// TODO Optimize it! Probably very slow (and confusing!)
+57 -13
View File
@@ -2,15 +2,17 @@ package persistence
import (
"fmt"
"github.com/deluan/gosonic/tests"
. "github.com/smartystreets/goconvey/convey"
"strconv"
"testing"
"github.com/deluan/gosonic/tests"
. "github.com/smartystreets/goconvey/convey"
)
type TestEntity struct {
Id string
Name string
Id string
Name string
ParentId string `parent:"parent"`
}
func shouldBeEqual(actualStruct interface{}, expectedStruct ...interface{}) string {
@@ -65,19 +67,25 @@ func TestBaseRepository(t *testing.T) {
Convey("Given an empty DB", func() {
repo := createRepo()
Convey("When I save a new entity", func() {
entity := &TestEntity{"123", "My Name"}
Convey("When I save a new entity and a parent", func() {
entity := &TestEntity{"123", "My Name", "ABC"}
err := repo.saveOrUpdate("123", entity)
Convey("Then the method shouldn't return any errors", func() {
Convey("Then saving the entity shouldn't return any errors", func() {
So(err, ShouldBeNil)
})
Convey("Then the number of entities should be 1", func() {
Convey("And the number of entities should be 1", func() {
count, _ := repo.CountAll()
So(count, ShouldEqual, 1)
})
Convey("And the number of children should be 1", func() {
children := make([]TestEntity, 0)
err := repo.loadChildren("parent", "ABC", &children)
So(err, ShouldBeNil)
So(len(children), ShouldEqual, 1)
})
Convey("And this entity should be equal to the the saved one", func() {
actualEntity, _ := repo.readEntity("123")
So(actualEntity, shouldBeEqual, entity)
@@ -89,11 +97,11 @@ func TestBaseRepository(t *testing.T) {
Convey("Given a table with one entity", func() {
repo := createRepo()
entity := &TestEntity{"111", "One Name"}
entity := &TestEntity{"111", "One Name", "AAA"}
repo.saveOrUpdate(entity.Id, entity)
Convey("When I save an entity with a different Id", func() {
newEntity := &TestEntity{"222", "Another Name"}
newEntity := &TestEntity{"222", "Another Name", "AAA"}
repo.saveOrUpdate(newEntity.Id, newEntity)
Convey("Then the number of entities should be 2", func() {
@@ -104,7 +112,7 @@ func TestBaseRepository(t *testing.T) {
})
Convey("When I save an entity with the same Id", func() {
newEntity := &TestEntity{"111", "New Name"}
newEntity := &TestEntity{"111", "New Name", "AAA"}
repo.saveOrUpdate(newEntity.Id, newEntity)
Convey("Then the number of entities should be 1", func() {
@@ -125,7 +133,7 @@ func TestBaseRepository(t *testing.T) {
Convey("Given a table with 3 entities", func() {
repo := createRepo()
for i := 1; i <= 3; i++ {
e := &TestEntity{strconv.Itoa(i), fmt.Sprintf("Name %d", i)}
e := &TestEntity{strconv.Itoa(i), fmt.Sprintf("Name %d", i), "AAA"}
repo.saveOrUpdate(e.Id, e)
}
@@ -142,9 +150,45 @@ func TestBaseRepository(t *testing.T) {
for _, e := range es {
So(e.Id, ShouldBeIn, []string{"1", "2", "3"})
So(e.Name, ShouldBeIn, []string{"Name 1", "Name 2", "Name 3"})
So(e.ParentId, ShouldEqual, "AAA")
}
})
})
Convey("When I call GetAllIds", func() {
ids, err := repo.GetAllIds()
Convey("Then It should not return any error", func() {
So(err, ShouldBeNil)
})
Convey("And I get all saved ids", func() {
So(len(ids), ShouldEqual, 3)
for k, _ := range ids {
So(k, ShouldBeIn, []string{"1", "2", "3"})
}
})
})
Convey("When I call DeletaAll with one of the entities", func() {
ids := make(map[string]bool)
ids["1"] = true
err := repo.DeleteAll(ids)
Convey("Then It should not return any error", func() {
So(err, ShouldBeNil)
})
Convey("Then CountAll should return 2", func() {
count, _ := repo.CountAll()
So(count, ShouldEqual, 2)
})
Convey("And the deleted record shouldn't be among the children", func() {
children := make([]TestEntity, 0)
err := repo.loadChildren("parent", "AAA", &children)
So(err, ShouldBeNil)
So(len(children), ShouldEqual, 2)
for _, e := range children {
So(e.Id, ShouldNotEqual, "1")
}
})
})
})
Reset(func() {
+19 -1
View File
@@ -2,8 +2,9 @@ package persistence
import (
"errors"
"github.com/deluan/gosonic/domain"
"sort"
"github.com/deluan/gosonic/domain"
)
type mediaFileRepository struct {
@@ -42,4 +43,21 @@ func (r *mediaFileRepository) FindByAlbum(albumId string) (domain.MediaFiles, er
return mfs, err
}
func (r *mediaFileRepository) PurgeInactive(active *domain.MediaFiles) error {
currentIds, err := r.GetAllIds()
if err != nil {
return err
}
for _, a := range *active {
currentIds[a.Id] = false
}
inactiveIds := make(map[string]bool)
for id, inactive := range currentIds {
if inactive {
inactiveIds[id] = true
}
}
return r.DeleteAll(inactiveIds)
}
var _ domain.MediaFileRepository = (*mediaFileRepository)(nil)