package dataset import ( "context" "errors" "fmt" "net/url" "strings" "sync" "time" "gorm.io/gorm" "gorm.io/gorm/clause" "senlinai-agent/backend/internal/models" ) var ( ErrNameRequired = errors.New("dataset source name is required") ErrKindInvalid = errors.New("dataset source kind is invalid") ErrURLInvalid = errors.New("dataset source url is invalid") ErrTitleRequired = errors.New("dataset item title is required") ErrStatusInvalid = errors.New("dataset item status is invalid") ) type Service struct { db *gorm.DB fetcher FeedFetcher syncMu sync.Mutex } type SourceInput struct { Name string Kind string URL string IconURL string Description string Enabled bool } type ItemInput struct { SourceIdentity string Title string Summary string Content string URL string PublishedAt *time.Time } type ItemUpdate struct { Status string Starred bool } type SourceRecord struct { Source models.SaDatasetSource ItemCount int64 } func NewService(database *gorm.DB) *Service { return &Service{db: database, fetcher: NewHTTPFeedFetcher()} } func newServiceWithFetcher(database *gorm.DB, fetcher FeedFetcher) *Service { return &Service{db: database, fetcher: fetcher} } func (s *Service) ListSources(userID uint) ([]SourceRecord, error) { var sources []models.SaDatasetSource if err := s.db.Where("created_by = ?", userID).Order("created_at asc, id asc").Find(&sources).Error; err != nil { return nil, err } counts := make(map[uint]int64) if len(sources) > 0 { sourceIDs := make([]uint, 0, len(sources)) for _, source := range sources { sourceIDs = append(sourceIDs, source.ID) } var rows []struct { SourceID uint Count int64 } if err := s.db.Model(&models.SaDatasetItem{}). Select("source_id, COUNT(*) AS count"). Where("created_by = ? AND source_id IN ?", userID, sourceIDs). Group("source_id"). Scan(&rows).Error; err != nil { return nil, err } for _, row := range rows { counts[row.SourceID] = row.Count } } result := make([]SourceRecord, 0, len(sources)) for _, source := range sources { result = append(result, SourceRecord{Source: source, ItemCount: counts[source.ID]}) } return result, nil } func (s *Service) CreateSource(userID uint, input SourceInput) (*models.SaDatasetSource, error) { normalized, err := normalizeSourceInput(input) if err != nil { return nil, err } source := &models.SaDatasetSource{ CreatedBy: userID, Name: normalized.Name, Kind: normalized.Kind, URL: normalized.URL, IconURL: normalized.IconURL, Description: normalized.Description, Enabled: normalized.Enabled, } return source, s.db.Create(source).Error } func (s *Service) UpdateSource(userID uint, identity string, input SourceInput) (*models.SaDatasetSource, error) { normalized, err := normalizeSourceInput(input) if err != nil { return nil, err } var source models.SaDatasetSource if err := s.db.Where("identity = ? AND created_by = ?", identity, userID).First(&source).Error; err != nil { return nil, err } if err := s.db.Model(&source).Updates(map[string]any{ "name": normalized.Name, "kind": normalized.Kind, "url": normalized.URL, "icon_url": normalized.IconURL, "description": normalized.Description, "enabled": normalized.Enabled, }).Error; err != nil { return nil, err } return &source, nil } func (s *Service) DeleteSource(userID uint, identity string) error { return s.db.Transaction(func(tx *gorm.DB) error { var source models.SaDatasetSource if err := tx.Where("identity = ? AND created_by = ?", identity, userID).First(&source).Error; err != nil { return err } if err := tx.Where("source_id = ? AND created_by = ?", source.ID, userID).Delete(&models.SaDatasetCron{}).Error; err != nil { return err } if err := tx.Where("source_id = ? AND created_by = ?", source.ID, userID).Delete(&models.SaDatasetItem{}).Error; err != nil { return err } return tx.Delete(&source).Error }) } func (s *Service) ListItems(userID uint, sourceIdentity string) ([]models.SaDatasetItem, error) { query := s.db.Where("created_by = ?", userID) if sourceIdentity = strings.TrimSpace(sourceIdentity); sourceIdentity != "" { source, err := s.findOwnedSource(userID, sourceIdentity) if err != nil { return nil, err } query = query.Where("source_id = ?", source.ID) } var items []models.SaDatasetItem if err := query.Order("COALESCE(published_at, created_at) desc, id desc").Limit(200).Find(&items).Error; err != nil { return nil, err } if items == nil { items = []models.SaDatasetItem{} } return items, nil } func (s *Service) CountItems(userID uint, sourceID uint) (int64, error) { var count int64 err := s.db.Model(&models.SaDatasetItem{}). Where("created_by = ? AND source_id = ?", userID, sourceID). Count(&count).Error return count, err } func (s *Service) CreateItem(userID uint, input ItemInput) (*models.SaDatasetItem, error) { title := strings.TrimSpace(input.Title) if title == "" { return nil, ErrTitleRequired } source, err := s.findOwnedSource(userID, input.SourceIdentity) if err != nil { return nil, err } itemURL := strings.TrimSpace(input.URL) if itemURL != "" && !validHTTPURL(itemURL) { return nil, ErrURLInvalid } item := &models.SaDatasetItem{ SourceID: source.ID, CreatedBy: userID, Title: title, Summary: strings.TrimSpace(input.Summary), Content: strings.TrimSpace(input.Content), URL: itemURL, Status: "unread", PublishedAt: utcOptionalTime(input.PublishedAt), } return item, s.db.Create(item).Error } func (s *Service) UpdateItem(userID uint, identity string, input ItemUpdate) (*models.SaDatasetItem, error) { status := strings.TrimSpace(input.Status) if status != "unread" && status != "read" && status != "archived" { return nil, ErrStatusInvalid } var item models.SaDatasetItem if err := s.db.Where("identity = ? AND created_by = ?", identity, userID).First(&item).Error; err != nil { return nil, err } if err := s.db.Model(&item).Updates(map[string]any{"status": status, "starred": input.Starred}).Error; err != nil { return nil, err } return &item, nil } func (s *Service) ListCrons(userID uint) ([]models.SaDatasetCron, error) { var crons []models.SaDatasetCron if err := s.db.Where("created_by = ?", userID).Order("created_at desc, id desc").Limit(100).Find(&crons).Error; err != nil { return nil, err } if crons == nil { crons = []models.SaDatasetCron{} } return crons, nil } func (s *Service) SyncSources(ctx context.Context, userID uint) ([]models.SaDatasetCron, error) { s.syncMu.Lock() defer s.syncMu.Unlock() return s.syncSources(ctx, userID) } func (s *Service) SyncAllSources(ctx context.Context) ([]models.SaDatasetCron, error) { var userIDs []uint if err := s.db.Model(&models.SaDatasetSource{}). Distinct("created_by"). Where("enabled = ? AND kind = ?", true, "rss"). Order("created_by asc"). Pluck("created_by", &userIDs).Error; err != nil { return nil, err } result := make([]models.SaDatasetCron, 0) var syncErrors []error for _, userID := range userIDs { crons, err := s.SyncSources(ctx, userID) result = append(result, crons...) if err != nil { syncErrors = append(syncErrors, fmt.Errorf("sync dataset sources for user %d: %w", userID, err)) } if ctx.Err() != nil { break } } return result, errors.Join(syncErrors...) } func (s *Service) syncSources(ctx context.Context, userID uint) ([]models.SaDatasetCron, error) { result := make([]models.SaDatasetCron, 0) var sources []models.SaDatasetSource if err := s.db.Where("created_by = ? AND enabled = ? AND kind = ?", userID, true, "rss"). Order("id asc").Find(&sources).Error; err != nil { return nil, err } for _, source := range sources { if err := ctx.Err(); err != nil { return result, err } now := time.Now().UTC() cron := models.SaDatasetCron{ SourceID: source.ID, CreatedBy: userID, Schedule: "@once", Status: "running", Enabled: true, NextRunAt: &now, LastRunAt: &now, } if err := s.db.Create(&cron).Error; err != nil { return nil, err } feed, fetchErr := s.fetcher.Fetch(ctx, source.URL) if fetchErr != nil { cron.Status = "failed" cron.Enabled = false cron.NextRunAt = nil cron.LastResult = truncateResult(fetchErr.Error()) if err := s.db.Model(&cron).Updates(map[string]any{ "status": cron.Status, "last_result": cron.LastResult, "enabled": false, "next_run_at": nil, }).Error; err != nil { return nil, err } result = append(result, cron) if err := ctx.Err(); err != nil { return result, err } continue } inserted, err := s.storeFeedItems(userID, source, feed.Items) if err != nil { cron.Status = "failed" cron.Enabled = false cron.NextRunAt = nil cron.LastResult = truncateResult("store feed items: " + err.Error()) if updateErr := s.db.Model(&cron).Updates(map[string]any{ "status": cron.Status, "last_result": cron.LastResult, "enabled": false, "next_run_at": nil, }).Error; updateErr != nil { return nil, updateErr } result = append(result, cron) continue } completedAt := time.Now().UTC() cron.Status = "completed" cron.Enabled = false cron.NextRunAt = nil cron.LastResult = fmt.Sprintf("format=%s fetched=%d inserted=%d", feed.Format, len(feed.Items), inserted) if err := s.db.Transaction(func(tx *gorm.DB) error { if err := tx.Model(&cron).Updates(map[string]any{ "status": cron.Status, "last_result": cron.LastResult, "enabled": false, "next_run_at": nil, }).Error; err != nil { return err } return tx.Model(&source).Update("last_synced_at", completedAt).Error }); err != nil { return nil, err } source.LastSyncedAt = &completedAt result = append(result, cron) } return result, nil } func (s *Service) storeFeedItems(userID uint, source models.SaDatasetSource, items []FeedItem) (int, error) { inserted := 0 err := s.db.Transaction(func(tx *gorm.DB) error { for _, input := range items { externalID := input.ExternalID item := models.SaDatasetItem{ SourceID: source.ID, CreatedBy: userID, ExternalID: &externalID, Title: input.Title, Summary: input.Summary, Content: input.Content, URL: input.URL, Status: "unread", PublishedAt: input.PublishedAt, } result := tx.Clauses(clause.OnConflict{ Columns: []clause.Column{{Name: "source_id"}, {Name: "external_id"}}, DoNothing: true, }).Create(&item) if result.Error != nil { return result.Error } inserted += int(result.RowsAffected) } return nil }) return inserted, err } func truncateResult(value string) string { return truncateRunes(value, 2000) } func truncateRunes(value string, limit int) string { runes := []rune(value) if len(runes) <= limit { return value } return string(runes[:limit]) } func (s *Service) findOwnedSource(userID uint, identity string) (*models.SaDatasetSource, error) { var source models.SaDatasetSource err := s.db.Where("identity = ? AND created_by = ?", strings.TrimSpace(identity), userID).First(&source).Error return &source, err } func normalizeSourceInput(input SourceInput) (SourceInput, error) { input.Name = strings.TrimSpace(input.Name) input.Kind = strings.ToLower(strings.TrimSpace(input.Kind)) input.URL = strings.TrimSpace(input.URL) input.IconURL = strings.TrimSpace(input.IconURL) input.Description = strings.TrimSpace(input.Description) if input.Name == "" { return input, ErrNameRequired } if input.Kind != "manual" && input.Kind != "link" && input.Kind != "rss" { return input, ErrKindInvalid } if input.URL != "" && !validHTTPURL(input.URL) { return input, ErrURLInvalid } if input.IconURL != "" && !validHTTPURL(input.IconURL) { return input, ErrURLInvalid } if input.Kind != "manual" && input.URL == "" { return input, ErrURLInvalid } return input, nil } func validHTTPURL(value string) bool { parsed, err := url.ParseRequestURI(value) return err == nil && (parsed.Scheme == "http" || parsed.Scheme == "https") && parsed.Host != "" && parsed.User == nil } func utcOptionalTime(value *time.Time) *time.Time { if value == nil { return nil } result := value.UTC() return &result }