refactor: simplify dataset initialization

This commit is contained in:
2026-07-23 22:44:04 +08:00
parent bc7fa40995
commit af8de3eff6
5 changed files with 98 additions and 329 deletions

View File

@@ -3,7 +3,6 @@ package initdb
import (
"fmt"
"testing"
"time"
"github.com/glebarez/sqlite"
"github.com/stretchr/testify/require"
@@ -11,19 +10,20 @@ import (
"senlinai-agent/backend/internal/models"
)
func TestInitDatasetCreatesDefaultSourcesWhenTableIsEmpty(t *testing.T) {
func TestInitDatasetCreatesDefaultSourcesWhenOwnerHasNoSources(t *testing.T) {
database := newDatasetInitTestDatabase(t)
require.NoError(t, InitUser(database))
root, err := InitUser(database)
require.NoError(t, err)
require.NoError(t, InitDataset(database))
require.NoError(t, InitDataset(database))
require.NoError(t, InitDataset(database, root.ID, root.Identity))
require.NoError(t, InitDataset(database, root.ID, root.Identity))
var root models.SaUser
require.NoError(t, database.Where("email = ?", rootUsername).First(&root).Error)
var sources []models.SaDatasetSource
require.NoError(t, database.Order("id asc").Find(&sources).Error)
require.Len(t, sources, len(defaultDatasetSources))
for index, expected := range defaultDatasetSources {
require.Equal(t, root.ID, sources[index].OwnerID)
require.Equal(t, root.Identity, sources[index].OwnerIdentity)
require.Equal(t, expected.Name, sources[index].Name)
require.Equal(t, expected.URL, sources[index].URL)
require.Equal(t, expected.IconURL, sources[index].IconURL)
@@ -31,16 +31,13 @@ func TestInitDatasetCreatesDefaultSourcesWhenTableIsEmpty(t *testing.T) {
require.Equal(t, expected.Key, *sources[index].SeedKey)
require.Equal(t, "rss", sources[index].Kind)
require.True(t, sources[index].Enabled)
require.Equal(t, root.ID, sources[index].OwnerID)
require.Equal(t, root.Identity, sources[index].OwnerIdentity)
}
}
func TestInitDatasetAddsMissingDefaultsWithoutChangingCustomSources(t *testing.T) {
func TestInitDatasetDoesNothingWhenOwnerAlreadyHasSource(t *testing.T) {
database := newDatasetInitTestDatabase(t)
require.NoError(t, InitUser(database))
var root models.SaUser
require.NoError(t, database.Where("email = ?", rootUsername).First(&root).Error)
root, err := InitUser(database)
require.NoError(t, err)
existing := models.SaDatasetSource{
OwnerID: root.ID,
Name: "Existing source",
@@ -49,151 +46,51 @@ func TestInitDatasetAddsMissingDefaultsWithoutChangingCustomSources(t *testing.T
}
require.NoError(t, database.Create(&existing).Error)
require.NoError(t, InitDataset(database))
require.NoError(t, InitDataset(database, root.ID, root.Identity))
var sources []models.SaDatasetSource
require.NoError(t, database.Find(&sources).Error)
require.Len(t, sources, len(defaultDatasetSources)+1)
var preserved models.SaDatasetSource
require.NoError(t, database.Where("identity = ?", existing.Identity).First(&preserved).Error)
require.Equal(t, existing.Name, preserved.Name)
require.Nil(t, preserved.SeedKey)
require.Len(t, sources, 1)
require.Equal(t, existing.Identity, sources[0].Identity)
}
func TestInitDatasetBackfillsLegacySourceOwners(t *testing.T) {
func TestInitDatasetCountsSourcesByOwner(t *testing.T) {
database := newDatasetInitTestDatabase(t)
require.NoError(t, InitUser(database))
root, err := InitUser(database)
require.NoError(t, err)
other := models.SaUser{
Email: "other@example.com", DisplayName: "Other",
PasswordHash: "hash", Role: "user",
}
require.NoError(t, database.Create(&other).Error)
require.NoError(t, database.Exec(
"ALTER TABLE sa_dataset_sources ADD COLUMN created_by integer NOT NULL DEFAULT 0",
).Error)
require.NoError(t, database.Exec(
"ALTER TABLE sa_dataset_sources ADD COLUMN created_by_identity text",
).Error)
source := models.SaDatasetSource{
OwnerID: other.ID, Name: "Existing source", Kind: "manual", Enabled: true,
}
require.NoError(t, database.Create(&source).Error)
require.NoError(t, database.Exec(`
UPDATE sa_dataset_sources
SET owner_id = 0, owner_identity = '', created_by = ?, created_by_identity = ?
WHERE id = ?
`, other.ID, other.Identity, source.ID).Error)
require.NoError(t, database.Create(&models.SaDatasetSource{
OwnerID: other.ID, Name: "Other source", Kind: "manual", Enabled: true,
}).Error)
require.NoError(t, InitDataset(database))
require.NoError(t, InitDataset(database, root.ID, root.Identity))
require.NoError(t, database.First(&source, source.ID).Error)
require.Equal(t, other.ID, source.OwnerID)
require.Equal(t, other.Identity, source.OwnerIdentity)
hasCreatedBy, err := hasDatasetSourceColumn(database, "created_by")
require.NoError(t, err)
require.False(t, hasCreatedBy)
hasCreatedByIdentity, err := hasDatasetSourceColumn(database, "created_by_identity")
require.NoError(t, err)
require.False(t, hasCreatedByIdentity)
}
func TestInitDatasetRejectsMissingRootInsteadOfAssigningDefaultsToAnotherUser(t *testing.T) {
database := newDatasetInitTestDatabase(t)
user := models.SaUser{
Email: "existing@example.com",
DisplayName: "Existing",
PasswordHash: "hash",
Role: "user",
}
require.NoError(t, database.Create(&user).Error)
require.Error(t, InitDataset(database))
var sources []models.SaDatasetSource
require.NoError(t, database.Find(&sources).Error)
require.Empty(t, sources)
}
func TestInitDatasetRepairsLegacyRSSHubURLsForRootOnly(t *testing.T) {
database := newDatasetInitTestDatabase(t)
require.NoError(t, InitUser(database))
var root models.SaUser
require.NoError(t, database.Where("email = ?", rootUsername).First(&root).Error)
other := models.SaUser{
Email: "other@example.com", DisplayName: "Other",
PasswordHash: "hash", Role: "user",
}
require.NoError(t, database.Create(&other).Error)
rootLegacy := models.SaDatasetSource{
OwnerID: root.ID, Name: "财联社-热门", Kind: "rss",
URL: "https://rsshub.app/cls/hot", Enabled: true,
}
otherLegacy := models.SaDatasetSource{
OwnerID: other.ID, Name: "财联社-热门", Kind: "rss",
URL: "https://rsshub.app/cls/hot", Enabled: true,
}
require.NoError(t, database.Create(&rootLegacy).Error)
require.NoError(t, database.Create(&otherLegacy).Error)
require.NoError(t, InitDataset(database))
require.NoError(t, database.First(&rootLegacy, rootLegacy.ID).Error)
require.Equal(t, "https://rsshub.ktachibana.party/cls/hot", rootLegacy.URL)
require.NotNil(t, rootLegacy.SeedKey)
require.Equal(t, "rsshub-cls-hot", *rootLegacy.SeedKey)
require.NoError(t, database.First(&otherLegacy, otherLegacy.ID).Error)
require.Equal(t, "https://rsshub.app/cls/hot", otherLegacy.URL)
require.Nil(t, otherLegacy.SeedKey)
}
func TestInitDatasetPreparesCollectedItemUniqueness(t *testing.T) {
database := newDatasetInitTestDatabase(t)
require.NoError(t, InitUser(database))
var root models.SaUser
require.NoError(t, database.Where("email = ?", rootUsername).First(&root).Error)
source := models.SaDatasetSource{
OwnerID: root.ID, Name: "Existing feed", Kind: "rss",
URL: "https://example.com/feed", Enabled: true,
}
require.NoError(t, database.Create(&source).Error)
now := time.Now()
for index, externalID := range []string{"duplicate", "duplicate", "", ""} {
require.NoError(t, database.Exec(`
INSERT INTO sa_dataset_items
(identity, source_id, created_by, external_id, title, status, created_at, updated_at)
VALUES (?, ?, ?, ?, ?, ?, ?, ?)
`, fmt.Sprintf("item-%d", index), source.ID, root.ID, externalID,
fmt.Sprintf("Item %d", index), "unread", now, now).Error)
}
require.NoError(t, InitDataset(database))
var collectedCount int64
require.NoError(t, database.Model(&models.SaDatasetItem{}).
Where("source_id = ? AND external_id = ?", source.ID, "duplicate").
Count(&collectedCount).Error)
require.Equal(t, int64(1), collectedCount)
var manualCount int64
require.NoError(t, database.Model(&models.SaDatasetItem{}).
Where("source_id = ? AND external_id IS NULL", source.ID).
Count(&manualCount).Error)
require.Equal(t, int64(2), manualCount)
require.Error(t, database.Exec(`
INSERT INTO sa_dataset_items
(identity, source_id, created_by, external_id, title, status, created_at, updated_at)
VALUES (?, ?, ?, ?, ?, ?, ?, ?)
`, "item-new", source.ID, root.ID, "duplicate", "Duplicate", "unread", now, now).Error)
var rootCount int64
require.NoError(t, database.Model(&models.SaDatasetSource{}).
Where("owner_id = ?", root.ID).
Count(&rootCount).Error)
require.Equal(t, int64(len(defaultDatasetSources)), rootCount)
var otherCount int64
require.NoError(t, database.Model(&models.SaDatasetSource{}).
Where("owner_id = ?", other.ID).
Count(&otherCount).Error)
require.Equal(t, int64(1), otherCount)
}
func newDatasetInitTestDatabase(t *testing.T) *gorm.DB {
t.Helper()
database, err := gorm.Open(sqlite.Open(fmt.Sprintf("file:%s?mode=memory&cache=shared", t.Name())), &gorm.Config{})
database, err := gorm.Open(
sqlite.Open(fmt.Sprintf("file:%s?mode=memory&cache=shared", t.Name())),
&gorm.Config{},
)
require.NoError(t, err)
require.NoError(t, database.AutoMigrate(
&models.SaUser{},
&models.SaDatasetSource{},
&models.SaDatasetItem{},
))
return database
}