feat: schedule dataset collection

This commit is contained in:
2026-07-23 20:55:38 +08:00
parent 5da343cc71
commit 3749cf4543
5 changed files with 309 additions and 1 deletions

View File

@@ -119,6 +119,27 @@ func TestSyncSourcesRecordsFeedFailure(t *testing.T) {
require.Contains(t, crons[0].LastResult, "feed unavailable")
}
func TestSyncAllSourcesCollectsEnabledRSSSourcesForEveryUser(t *testing.T) {
database := newDatasetTestDatabase(t)
firstUser := createDatasetTestUser(t, database, "first-feed-owner@example.com")
secondUser := createDatasetTestUser(t, database, "second-feed-owner@example.com")
for _, source := range []models.SaDatasetSource{
{CreatedBy: firstUser.ID, Name: "First RSS", Kind: "rss", URL: "https://example.com/first", Enabled: true},
{CreatedBy: secondUser.ID, Name: "Second RSS", Kind: "rss", URL: "https://example.com/second", Enabled: true},
{CreatedBy: firstUser.ID, Name: "Manual", Kind: "manual", Enabled: true},
} {
require.NoError(t, database.Create(&source).Error)
}
crons, err := newServiceWithFetcher(database, staticFeedFetcher{}).SyncAllSources(context.Background())
require.NoError(t, err)
require.Len(t, crons, 2)
var itemCount int64
require.NoError(t, database.Model(&models.SaDatasetItem{}).Count(&itemCount).Error)
require.Equal(t, int64(2), itemCount)
}
type failingFeedFetcher struct{}
func (failingFeedFetcher) Fetch(context.Context, string) (ParsedFeed, error) {

View File

@@ -6,6 +6,7 @@ import (
"fmt"
"net/url"
"strings"
"sync"
"time"
"gorm.io/gorm"
@@ -23,6 +24,7 @@ var (
type Service struct {
db *gorm.DB
fetcher FeedFetcher
syncMu sync.Mutex
}
type SourceInput struct {
@@ -216,6 +218,38 @@ func (s *Service) ListCrons(userID uint) ([]models.SaDatasetCron, error) {
}
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").
@@ -223,6 +257,9 @@ func (s *Service) SyncSources(ctx context.Context, userID uint) ([]models.SaData
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",
@@ -244,6 +281,9 @@ func (s *Service) SyncSources(ctx context.Context, userID uint) ([]models.SaData
return nil, err
}
result = append(result, cron)
if err := ctx.Err(); err != nil {
return result, err
}
continue
}