Files

638 lines
18 KiB
Go

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")
ErrSyncInProgress = errors.New("dataset sync is already in progress")
)
const (
defaultItemPageSize = 50
maxItemPageSize = 200
maxConcurrentFetches = 4
)
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 SourcePatch 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
ImageURL string
PublishedAt *time.Time
}
type ItemUpdate struct {
Status string
Starred bool
}
type ItemPatch struct {
Status *string
Starred *bool
}
type DepositResult struct {
DocumentIdentity string
ProjectIdentity string
}
type SourceRecord struct {
Source models.SaDatasetSource
ItemCount int64
}
type ItemPage struct {
Items []models.SaDatasetItem
Total int64
Limit int
Offset int
}
type SyncResult struct {
SourceIdentity string
Status string
Result string
}
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("owner_id = ?", 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("source_id IN ?", 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{
OwnerID: 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) {
return s.PatchSource(userID, identity, SourcePatch{
Name: &input.Name, Kind: &input.Kind, URL: &input.URL, IconURL: &input.IconURL,
Description: &input.Description, Enabled: &input.Enabled,
})
}
func (s *Service) PatchSource(userID uint, identity string, patch SourcePatch) (*models.SaDatasetSource, error) {
if !s.syncMu.TryLock() {
return nil, ErrSyncInProgress
}
defer s.syncMu.Unlock()
var source models.SaDatasetSource
if err := s.db.Where("identity = ? AND owner_id = ?", identity, userID).First(&source).Error; err != nil {
return nil, err
}
input := SourceInput{
Name: source.Name, Kind: source.Kind, URL: source.URL, IconURL: source.IconURL,
Description: source.Description, Enabled: source.Enabled,
}
if patch.Name != nil {
input.Name = *patch.Name
}
if patch.Kind != nil {
input.Kind = *patch.Kind
}
if patch.URL != nil {
input.URL = *patch.URL
}
if patch.IconURL != nil {
input.IconURL = *patch.IconURL
}
if patch.Description != nil {
input.Description = *patch.Description
}
if patch.Enabled != nil {
input.Enabled = *patch.Enabled
}
managedFields := source.SeedKey != nil || source.Kind != "rss"
if managedFields {
input.Kind = source.Kind
input.URL = source.URL
input.IconURL = source.IconURL
}
var normalized SourceInput
var err error
if source.Kind == "rss" {
normalized, err = normalizeSourceInput(input)
} else {
normalized, err = normalizeLegacySourceInput(input)
}
if 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 {
if !s.syncMu.TryLock() {
return ErrSyncInProgress
}
defer s.syncMu.Unlock()
return s.db.Transaction(func(tx *gorm.DB) error {
var source models.SaDatasetSource
if err := tx.Where("identity = ? AND owner_id = ?", identity, userID).First(&source).Error; err != nil {
return err
}
if err := tx.Where("source_id = ?", source.ID).Delete(&models.SaDatasetItem{}).Error; err != nil {
return err
}
return tx.Delete(&source).Error
})
}
func (s *Service) ListItems(userID uint, sourceIdentity string) ([]models.SaDatasetItem, error) {
page, err := s.ListItemsPage(userID, sourceIdentity, maxItemPageSize, 0)
if err != nil {
return nil, err
}
return page.Items, nil
}
func (s *Service) ListItemsPage(userID uint, sourceIdentity string, limit, offset int) (ItemPage, error) {
limit, offset = normalizePage(limit, offset)
query := s.db.Model(&models.SaDatasetItem{}).
Joins("JOIN sa_dataset_sources ON sa_dataset_sources.id = sa_dataset_items.source_id").
Where("sa_dataset_sources.owner_id = ?", userID)
if sourceIdentity = strings.TrimSpace(sourceIdentity); sourceIdentity != "" {
source, err := s.findOwnedSource(userID, sourceIdentity)
if err != nil {
return ItemPage{}, err
}
query = query.Where("sa_dataset_items.source_id = ?", source.ID)
}
var total int64
if err := query.Session(&gorm.Session{}).Count(&total).Error; err != nil {
return ItemPage{}, err
}
var items []models.SaDatasetItem
if err := query.Select("sa_dataset_items.*").
Order("COALESCE(sa_dataset_items.published_at, sa_dataset_items.created_at) desc, sa_dataset_items.id desc").
Limit(limit).Offset(offset).Find(&items).Error; err != nil {
return ItemPage{}, err
}
if items == nil {
items = []models.SaDatasetItem{}
}
return ItemPage{Items: items, Total: total, Limit: limit, Offset: offset}, nil
}
func (s *Service) CountItems(userID uint, sourceID uint) (int64, error) {
var count int64
err := s.db.Model(&models.SaDatasetItem{}).
Joins("JOIN sa_dataset_sources ON sa_dataset_sources.id = sa_dataset_items.source_id").
Where("sa_dataset_sources.owner_id = ? AND sa_dataset_items.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
}
imageURL := strings.TrimSpace(input.ImageURL)
if imageURL != "" && !validHTTPURL(imageURL) {
return nil, ErrURLInvalid
}
item := &models.SaDatasetItem{
SourceID: source.ID, Title: title,
Summary: strings.TrimSpace(input.Summary), Content: strings.TrimSpace(input.Content),
URL: itemURL, ImageURL: imageURL, 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) {
return s.PatchItem(userID, identity, ItemPatch{Status: &input.Status, Starred: &input.Starred})
}
func (s *Service) PatchItem(userID uint, identity string, patch ItemPatch) (*models.SaDatasetItem, error) {
item, err := s.findOwnedItem(userID, identity)
if err != nil {
return nil, err
}
updates := make(map[string]any, 2)
if patch.Status != nil {
status := strings.TrimSpace(*patch.Status)
if status != "unread" && status != "read" && status != "archived" {
return nil, ErrStatusInvalid
}
updates["status"] = status
}
if patch.Starred != nil {
updates["starred"] = *patch.Starred
}
if len(updates) == 0 {
return item, nil
}
if err := s.db.Model(item).Updates(updates).Error; err != nil {
return nil, err
}
return item, nil
}
func (s *Service) DepositItem(userID uint, itemIdentity, projectIdentity string) (DepositResult, error) {
item, err := s.findOwnedItem(userID, itemIdentity)
if err != nil {
return DepositResult{}, err
}
var project models.SaProject
if err := s.db.Where("identity = ? AND owner_id = ?", strings.TrimSpace(projectIdentity), userID).
First(&project).Error; err != nil {
return DepositResult{}, err
}
body := strings.TrimSpace(item.Content)
if body == "" {
body = strings.TrimSpace(item.Summary)
}
if item.ImageURL != "" {
if body != "" {
body = "![文章图片](" + item.ImageURL + ")\n\n" + body
} else {
body = "![文章图片](" + item.ImageURL + ")"
}
}
if item.URL != "" {
if body != "" {
body += "\n\n"
}
body += "[原文链接](" + item.URL + ")"
}
name := strings.TrimSpace(item.Title)
if !strings.HasSuffix(strings.ToLower(name), ".md") {
name += ".md"
}
var document models.SaDocumentTree
err = s.db.Transaction(func(tx *gorm.DB) error {
document = models.SaDocumentTree{
ProjectID: project.ID, OwnerID: userID, Kind: models.DocumentKindFile,
Name: name, NormalizedName: strings.ToLower(name), Extension: ".md",
MimeType: "text/markdown; charset=utf-8", Size: int64(len([]byte(body))),
Revision: 1, ScanStatus: "not_scanned",
}
if err := tx.Create(&document).Error; err != nil {
return err
}
return tx.Create(&models.SaDocumentContent{DocumentID: document.ID, Markdown: body}).Error
})
if err != nil {
return DepositResult{}, err
}
return DepositResult{DocumentIdentity: document.Identity, ProjectIdentity: project.Identity}, nil
}
func (s *Service) SyncSources(ctx context.Context, userID uint) ([]SyncResult, error) {
if !s.syncMu.TryLock() {
return nil, ErrSyncInProgress
}
defer s.syncMu.Unlock()
sources, err := s.enabledRSSSources(userID)
if err != nil {
return nil, err
}
return s.executeDatasetSources(ctx, sources)
}
func (s *Service) SyncAllSources(ctx context.Context) ([]SyncResult, error) {
if !s.syncMu.TryLock() {
return nil, ErrSyncInProgress
}
defer s.syncMu.Unlock()
var sources []models.SaDatasetSource
if err := s.db.Where("enabled = ? AND kind = ?", true, "rss").
Order("owner_id asc, id asc").
Find(&sources).Error; err != nil {
return nil, err
}
return s.executeDatasetSources(ctx, sources)
}
func (s *Service) enabledRSSSources(userID uint) ([]models.SaDatasetSource, error) {
var sources []models.SaDatasetSource
if err := s.db.Where("owner_id = ? AND enabled = ? AND kind = ?", userID, true, "rss").
Order("id asc").Find(&sources).Error; err != nil {
return nil, err
}
return sources, nil
}
func (s *Service) executeDatasetSources(ctx context.Context, sources []models.SaDatasetSource) ([]SyncResult, error) {
result := make([]SyncResult, len(sources))
runErrors := make([]error, len(sources))
concurrency := maxConcurrentFetches
if s.db.Dialector.Name() == "sqlite" {
concurrency = 1
}
semaphore := make(chan struct{}, concurrency)
var workers sync.WaitGroup
for index, source := range sources {
workers.Add(1)
go func() {
defer workers.Done()
semaphore <- struct{}{}
defer func() { <-semaphore }()
result[index], runErrors[index] = s.executeDatasetSource(ctx, source)
}()
}
workers.Wait()
return result, errors.Join(runErrors...)
}
func (s *Service) executeDatasetSource(ctx context.Context, source models.SaDatasetSource) (SyncResult, error) {
result := SyncResult{SourceIdentity: source.Identity}
if err := ctx.Err(); err != nil {
result.Status = "failed"
result.Result = truncateResult(err.Error())
return result, err
}
feed, fetchErr := s.fetcher.Fetch(ctx, source.URL)
if fetchErr != nil {
result.Status = "failed"
result.Result = truncateResult(fetchErr.Error())
return result, nil
}
inserted := 0
completedAt := time.Now().UTC()
err := s.db.Transaction(func(tx *gorm.DB) error {
var storeErr error
inserted, storeErr = storeFeedItems(tx, source, feed.Items)
if storeErr != nil {
return storeErr
}
return tx.Model(&source).Update("last_synced_at", completedAt).Error
})
if err != nil {
result.Status = "failed"
result.Result = truncateResult("store feed items: " + err.Error())
return result, nil
}
result.Status = "completed"
result.Result = fmt.Sprintf("format=%s fetched=%d inserted=%d", feed.Format, len(feed.Items), inserted)
return result, nil
}
func storeFeedItems(tx *gorm.DB, source models.SaDatasetSource, items []FeedItem) (int, error) {
inserted := 0
for _, input := range items {
externalID := input.ExternalID
itemURL := resolveFeedReference(source.URL, input.URL)
imageURL := resolveFeedReference(source.URL, input.ImageURL)
item := models.SaDatasetItem{
SourceID: source.ID, ExternalID: &externalID,
Title: input.Title, Summary: input.Summary, Content: input.Content,
URL: itemURL, ImageURL: imageURL,
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 inserted, result.Error
}
inserted += int(result.RowsAffected)
if result.RowsAffected == 0 {
updates := map[string]any{
"title": item.Title,
}
if item.Summary != "" {
updates["summary"] = item.Summary
}
if item.Content != "" {
updates["content"] = item.Content
}
if item.URL != "" {
updates["url"] = item.URL
}
if item.ImageURL != "" {
updates["image_url"] = item.ImageURL
}
if item.PublishedAt != nil {
updates["published_at"] = item.PublishedAt
}
if err := tx.Model(&models.SaDatasetItem{}).
Where("source_id = ? AND external_id = ?", source.ID, externalID).
Updates(updates).Error; err != nil {
return inserted, err
}
}
}
return inserted, nil
}
func resolveFeedReference(baseURL, reference string) string {
reference = strings.TrimSpace(reference)
if reference == "" {
return ""
}
parsedReference, err := url.Parse(reference)
if err != nil {
return ""
}
if !parsedReference.IsAbs() {
parsedBase, err := url.Parse(baseURL)
if err != nil {
return ""
}
parsedReference = parsedBase.ResolveReference(parsedReference)
}
resolved := parsedReference.String()
if !validHTTPURL(resolved) {
return ""
}
return truncateRunes(resolved, 2048)
}
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 owner_id = ?", strings.TrimSpace(identity), userID).First(&source).Error
return &source, err
}
func (s *Service) findOwnedItem(userID uint, identity string) (*models.SaDatasetItem, error) {
var item models.SaDatasetItem
err := s.db.Model(&models.SaDatasetItem{}).
Select("sa_dataset_items.*").
Joins("JOIN sa_dataset_sources ON sa_dataset_sources.id = sa_dataset_items.source_id").
Where("sa_dataset_items.identity = ? AND sa_dataset_sources.owner_id = ?", strings.TrimSpace(identity), userID).
First(&item).Error
return &item, 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 != "rss" {
return input, ErrKindInvalid
}
if input.URL != "" && !validHTTPURL(input.URL) {
return input, ErrURLInvalid
}
if input.IconURL != "" && !validHTTPURL(input.IconURL) {
return input, ErrURLInvalid
}
if input.URL == "" {
return input, ErrURLInvalid
}
return input, nil
}
func normalizeLegacySourceInput(input SourceInput) (SourceInput, error) {
input.Name = strings.TrimSpace(input.Name)
input.Description = strings.TrimSpace(input.Description)
if input.Name == "" {
return input, ErrNameRequired
}
return input, nil
}
func normalizePage(limit, offset int) (int, int) {
if limit <= 0 {
limit = defaultItemPageSize
}
if limit > maxItemPageSize {
limit = maxItemPageSize
}
if offset < 0 {
offset = 0
}
return limit, offset
}
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
}