fix: harden explore dataset collection
This commit is contained in:
@@ -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
|
||||
}
|
||||
|
||||
Reference in New Issue
Block a user