fix: harden explore dataset collection

This commit is contained in:
2026-07-23 21:34:13 +08:00
parent 3749cf4543
commit 50d8246915
9 changed files with 254 additions and 87 deletions

View File

@@ -8,53 +8,109 @@ import (
"senlinai-agent/backend/internal/models"
)
var defaultDatasetSources = []struct {
Name string
URL string
IconURL string
}{
{Name: "彭博社最新报道", URL: "https://bbg.buzzing.cc/feed.json", IconURL: "https://bbg.buzzing.cc/apple-touch-icon.png"},
{Name: "国外科技头条", URL: "https://tech.buzzing.cc/feed.json", IconURL: "https://tech.buzzing.cc/apple-touch-icon.png"},
{Name: "国外财经新闻", URL: "https://finance.buzzing.cc/feed.json", IconURL: "https://finance.buzzing.cc/apple-touch-icon.png"},
{Name: "华尔街日报热门", URL: "https://wsj.buzzing.cc/feed.json", IconURL: "https://wsj.buzzing.cc/apple-touch-icon.png"},
{Name: "Product Hunt", URL: "https://ph.buzzing.cc/feed.json", IconURL: "https://ph.buzzing.cc/apple-touch-icon.png"},
{Name: "Dev.to", URL: "https://dev.buzzing.cc/feed.json", IconURL: "https://dev.buzzing.cc/apple-touch-icon.png"},
{Name: "财联社-热门", URL: "https://rsshub.app/cls/hot"},
{Name: "金十数据-快讯", URL: "https://rsshub.app/jin10"},
type defaultDatasetSource struct {
Key string
Name string
URL string
IconURL string
LegacyURLs []string
}
// InitDataset creates the default Buzzing RSS sources for the default account when the source table is empty.
func InitDataset(database *gorm.DB) error {
var count int64
if err := database.Model(&models.SaDatasetSource{}).Count(&count).Error; err != nil {
return err
}
if count > 0 {
return nil
}
var defaultDatasetSources = []defaultDatasetSource{
{Key: "buzzing-bloomberg", Name: "彭博社最新报道", URL: "https://bbg.buzzing.cc/feed.json", IconURL: "https://bbg.buzzing.cc/apple-touch-icon.png"},
{Key: "buzzing-tech", Name: "国外科技头条", URL: "https://tech.buzzing.cc/feed.json", IconURL: "https://tech.buzzing.cc/apple-touch-icon.png"},
{Key: "buzzing-finance", Name: "国外财经新闻", URL: "https://finance.buzzing.cc/feed.json", IconURL: "https://finance.buzzing.cc/apple-touch-icon.png"},
{Key: "buzzing-wsj", Name: "华尔街日报热门", URL: "https://wsj.buzzing.cc/feed.json", IconURL: "https://wsj.buzzing.cc/apple-touch-icon.png"},
{Key: "buzzing-product-hunt", Name: "Product Hunt", URL: "https://ph.buzzing.cc/feed.json", IconURL: "https://ph.buzzing.cc/apple-touch-icon.png"},
{Key: "buzzing-devto", Name: "Dev.to", URL: "https://dev.buzzing.cc/feed.json", IconURL: "https://dev.buzzing.cc/apple-touch-icon.png"},
{
Key: "rsshub-cls-hot", Name: "财联社-热门",
URL: "https://rsshub.ktachibana.party/cls/hot",
LegacyURLs: []string{"https://rsshub.app/cls/hot"},
},
{
Key: "rsshub-jin10", Name: "金十数据-快讯",
URL: "https://rsshub.ktachibana.party/jin10",
LegacyURLs: []string{"https://rsshub.app/jin10"},
},
}
var owner models.SaUser
err := database.Where("email = ?", rootUsername).First(&owner).Error
if err != nil && !errors.Is(err, gorm.ErrRecordNotFound) {
// InitDataset inserts or repairs root's built-in dataset sources without touching user-created sources.
func InitDataset(database *gorm.DB) error {
var root models.SaUser
if err := database.Where("email = ? AND role = ?", rootUsername, rootRole).First(&root).Error; err != nil {
return fmt.Errorf("find root user for default dataset sources: %w", err)
}
if errors.Is(err, gorm.ErrRecordNotFound) {
if err := database.Order("id asc").First(&owner).Error; err != nil {
return fmt.Errorf("find owner for default dataset sources: %w", err)
return database.Transaction(func(tx *gorm.DB) error {
if err := prepareDatasetItems(tx); err != nil {
return err
}
for _, definition := range defaultDatasetSources {
if err := upsertDefaultDatasetSource(tx, root, definition); err != nil {
return err
}
}
return nil
})
}
func prepareDatasetItems(tx *gorm.DB) error {
if tx.Migrator().HasIndex(&models.SaDatasetItem{}, "idx_sa_dataset_item_source_external") {
return nil
}
if err := tx.Exec(
"UPDATE sa_dataset_items SET external_id = NULL WHERE external_id = ''",
).Error; err != nil {
return err
}
if err := tx.Exec(`
DELETE FROM sa_dataset_items
WHERE external_id IS NOT NULL
AND EXISTS (
SELECT 1
FROM sa_dataset_items AS earlier
WHERE earlier.source_id = sa_dataset_items.source_id
AND earlier.external_id = sa_dataset_items.external_id
AND earlier.id < sa_dataset_items.id
)
`).Error; err != nil {
return err
}
return tx.Exec(`
CREATE UNIQUE INDEX IF NOT EXISTS idx_sa_dataset_item_source_external
ON sa_dataset_items (source_id, external_id)
`).Error
}
func upsertDefaultDatasetSource(tx *gorm.DB, root models.SaUser, definition defaultDatasetSource) error {
var source models.SaDatasetSource
err := tx.Where("created_by = ? AND seed_key = ?", root.ID, definition.Key).First(&source).Error
if err == nil {
return tx.Model(&source).Updates(map[string]any{
"url": definition.URL, "icon_url": definition.IconURL, "kind": "rss",
}).Error
}
if !errors.Is(err, gorm.ErrRecordNotFound) {
return err
}
sources := make([]models.SaDatasetSource, 0, len(defaultDatasetSources))
for _, source := range defaultDatasetSources {
sources = append(sources, models.SaDatasetSource{
CreatedBy: owner.ID,
Name: source.Name,
Kind: "rss",
URL: source.URL,
IconURL: source.IconURL,
Description: "Buzzing 中文聚合订阅",
Enabled: true,
})
knownURLs := append([]string{definition.URL}, definition.LegacyURLs...)
err = tx.Where("created_by = ? AND url IN ?", root.ID, knownURLs).First(&source).Error
if err == nil {
return tx.Model(&source).Updates(map[string]any{
"seed_key": definition.Key, "url": definition.URL,
"icon_url": definition.IconURL, "kind": "rss",
}).Error
}
return database.Create(&sources).Error
if !errors.Is(err, gorm.ErrRecordNotFound) {
return err
}
seedKey := definition.Key
return tx.Create(&models.SaDatasetSource{
CreatedBy: root.ID, SeedKey: &seedKey, Name: definition.Name,
Kind: "rss", URL: definition.URL, IconURL: definition.IconURL,
Description: "默认资讯订阅", Enabled: true,
}).Error
}

View File

@@ -3,6 +3,7 @@ package initdb
import (
"fmt"
"testing"
"time"
"github.com/glebarez/sqlite"
"github.com/stretchr/testify/require"
@@ -26,6 +27,8 @@ func TestInitDatasetCreatesDefaultSourcesWhenTableIsEmpty(t *testing.T) {
require.Equal(t, expected.Name, sources[index].Name)
require.Equal(t, expected.URL, sources[index].URL)
require.Equal(t, expected.IconURL, sources[index].IconURL)
require.NotNil(t, sources[index].SeedKey)
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].CreatedBy)
@@ -33,7 +36,7 @@ func TestInitDatasetCreatesDefaultSourcesWhenTableIsEmpty(t *testing.T) {
}
}
func TestInitDatasetDoesNothingWhenSourceTableIsNotEmpty(t *testing.T) {
func TestInitDatasetAddsMissingDefaultsWithoutChangingCustomSources(t *testing.T) {
database := newDatasetInitTestDatabase(t)
require.NoError(t, InitUser(database))
var root models.SaUser
@@ -50,11 +53,14 @@ func TestInitDatasetDoesNothingWhenSourceTableIsNotEmpty(t *testing.T) {
var sources []models.SaDatasetSource
require.NoError(t, database.Find(&sources).Error)
require.Len(t, sources, 1)
require.Equal(t, existing.Identity, sources[0].Identity)
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)
}
func TestInitDatasetUsesFirstExistingUserWhenRootIsAbsent(t *testing.T) {
func TestInitDatasetRejectsMissingRootInsteadOfAssigningDefaultsToAnotherUser(t *testing.T) {
database := newDatasetInitTestDatabase(t)
user := models.SaUser{
Email: "existing@example.com",
@@ -64,21 +70,93 @@ func TestInitDatasetUsesFirstExistingUserWhenRootIsAbsent(t *testing.T) {
}
require.NoError(t, database.Create(&user).Error)
require.NoError(t, InitDataset(database))
require.Error(t, InitDataset(database))
var sources []models.SaDatasetSource
require.NoError(t, database.Find(&sources).Error)
require.Len(t, sources, len(defaultDatasetSources))
for _, source := range sources {
require.Equal(t, user.ID, source.CreatedBy)
require.Equal(t, user.Identity, source.CreatedByIdentity)
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{
CreatedBy: root.ID, Name: "财联社-热门", Kind: "rss",
URL: "https://rsshub.app/cls/hot", Enabled: true,
}
otherLegacy := models.SaDatasetSource{
CreatedBy: 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{
CreatedBy: 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)
}
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{})
require.NoError(t, err)
require.NoError(t, database.AutoMigrate(&models.SaUser{}, &models.SaDatasetSource{}))
require.NoError(t, database.AutoMigrate(
&models.SaUser{},
&models.SaDatasetSource{},
&models.SaDatasetItem{},
))
return database
}

View File

@@ -60,6 +60,7 @@ func newExpertTestDatabase(t *testing.T) *gorm.DB {
require.NoError(t, database.AutoMigrate(
&models.SaUser{},
&models.SaDatasetSource{},
&models.SaDatasetItem{},
&models.SaAIExpertCategory{},
&models.SaAIExpertItem{},
))