refactor: split modules and harden scraping workflows

This commit is contained in:
ShukeBta
2026-06-24 11:59:18 +08:00
parent efca3cbe69
commit 192f35d9fa
470 changed files with 47957 additions and 33839 deletions
+30 -296
View File
@@ -1,4 +1,4 @@
// Package service — Alist / S3 / WebDAV configuration management.
// Package service — external storage configuration management.
//
// StorageConfigService stores connection settings encrypted at rest
// (via CryptoService). It also exposes a Test() probe so the React UI
@@ -10,9 +10,7 @@ import (
"encoding/json"
"errors"
"fmt"
"io"
"net/http"
"net/url"
"strconv"
"strings"
"sync"
@@ -36,24 +34,6 @@ type StorageConfigService struct {
resolveFlight map[string]*cloudResolveCall
}
type cloudResolveCacheEntry struct {
link *cloud.DirectLink
expiresAt time.Time
hits int
lastHit time.Time
}
type cloudResolveCall struct {
done chan struct{}
link *cloud.DirectLink
err error
}
const (
cloudResolveHotHitThreshold = 3
cloudResolveBackgroundRefreshMax = 30 * time.Second
)
// NewStorageConfigService is the constructor.
func NewStorageConfigService(log *zap.Logger, repo *repository.Container, crypto *CryptoService) *StorageConfigService {
return &StorageConfigService{
@@ -108,6 +88,9 @@ func (s *StorageConfigService) List(ctx context.Context) ([]StorageView, error)
}
out := make([]StorageView, 0, len(rows))
for _, r := range rows {
if !IsAdminStorageConfigurable(r.Type) {
continue
}
plain := s.crypto.Decrypt(r.Config)
var cfg map[string]any
_ = json.Unmarshal([]byte(plain), &cfg)
@@ -334,7 +317,7 @@ func (s *StorageConfigService) Test(ctx context.Context, in StorageInput) error
}
defer resp.Body.Close()
return nil
case cloud.TypeQuark, cloud.Type115, cloud.TypeCloudDrive2:
case cloud.Type115, cloud.TypeCloudDrive2:
p, err := cloud.New(in.Type, cfg, client)
if err != nil {
return err
@@ -373,287 +356,40 @@ func (s *StorageConfigService) CloudList(ctx context.Context, typ, dirID string)
return p.List(ctx, dirID)
}
// CloudResolve resolves a cloud file reference to a direct link.
//
// clientUA is the User-Agent of the playback client that will follow the 302
// redirect. 115/夸克 CDN links are bound to the UA used to request them, so we
// resolve with the client's own UA — that way the pure 302 the host issues
// points at a link the client can fetch directly (true offload). When clientUA
// is empty the provider's default UA is used.
func (s *StorageConfigService) CloudResolve(ctx context.Context, typ, fileRef, clientUA string) (*cloud.DirectLink, error) {
if s == nil {
return nil, errors.New("storage config service unavailable")
}
cacheKey := s.resolveCacheKey(typ, fileRef, clientUA)
if link, ok, refresh := s.cachedResolve(cacheKey, typ); ok {
if refresh {
s.refreshResolveInBackground(cacheKey, typ, fileRef, clientUA)
}
return link, nil
}
if call, owner := s.beginResolve(cacheKey); !owner {
select {
case <-call.done:
if call.err != nil {
return nil, call.err
}
return cloneDirectLink(call.link), nil
case <-ctx.Done():
return nil, ctx.Err()
}
} else {
defer s.finishResolve(cacheKey, call)
p, err := s.cloudProviderWithUA(ctx, typ, clientUA)
if err != nil {
call.err = err
return nil, err
}
link, err := p.Resolve(ctx, fileRef)
if err != nil {
call.err = err
return nil, err
}
call.link = cloneDirectLink(link)
s.storeResolvedLink(cacheKey, typ, link)
return cloneDirectLink(link), nil
}
}
func (s *StorageConfigService) resolveCacheKey(typ, fileRef, clientUA string) string {
return strings.TrimSpace(typ) + "\x00" + strings.TrimSpace(fileRef) + "\x00" + strings.TrimSpace(clientUA)
}
func (s *StorageConfigService) cachedResolve(key, typ string) (*cloud.DirectLink, bool, bool) {
s.resolveMu.Lock()
defer s.resolveMu.Unlock()
if s.resolveCache == nil {
s.resolveCache = make(map[string]cloudResolveCacheEntry)
return nil, false, false
}
entry, ok := s.resolveCache[key]
now := time.Now()
if !ok || now.After(entry.expiresAt) {
if ok {
delete(s.resolveCache, key)
}
return nil, false, false
}
entry.hits++
entry.lastHit = now
s.resolveCache[key] = entry
refreshWindow := cloudResolveHotRefreshWindow(cloudResolveCacheTTL(typ))
shouldRefresh := entry.hits >= cloudResolveHotHitThreshold &&
refreshWindow > 0 &&
now.Add(refreshWindow).After(entry.expiresAt)
return cloneDirectLink(entry.link), true, shouldRefresh
}
func (s *StorageConfigService) beginResolve(key string) (*cloudResolveCall, bool) {
s.resolveMu.Lock()
defer s.resolveMu.Unlock()
if s.resolveFlight == nil {
s.resolveFlight = make(map[string]*cloudResolveCall)
}
if call := s.resolveFlight[key]; call != nil {
return call, false
}
call := &cloudResolveCall{done: make(chan struct{})}
s.resolveFlight[key] = call
return call, true
}
func (s *StorageConfigService) finishResolve(key string, call *cloudResolveCall) {
s.resolveMu.Lock()
if current := s.resolveFlight[key]; current == call {
delete(s.resolveFlight, key)
}
s.resolveMu.Unlock()
close(call.done)
}
func (s *StorageConfigService) refreshResolveInBackground(key, typ, fileRef, clientUA string) {
if s == nil {
return
}
go func() {
call, owner := s.beginResolve(key)
if !owner {
return
}
defer s.finishResolve(key, call)
ctx, cancel := context.WithTimeout(context.Background(), cloudResolveBackgroundRefreshMax)
defer cancel()
p, err := s.cloudProviderWithUA(ctx, typ, clientUA)
if err != nil {
call.err = err
if s.log != nil {
s.log.Debug("refresh cloud direct link failed", zap.String("provider", typ), zap.Error(err))
}
return
}
link, err := p.Resolve(ctx, fileRef)
if err != nil {
call.err = err
if s.log != nil {
s.log.Debug("refresh cloud direct link failed", zap.String("provider", typ), zap.Error(err))
}
return
}
call.link = cloneDirectLink(link)
s.storeResolvedLink(key, typ, link)
}()
}
func (s *StorageConfigService) storeResolvedLink(key, typ string, link *cloud.DirectLink) {
if link == nil || strings.TrimSpace(link.URL) == "" {
return
}
ttl := cloudResolveCacheTTL(typ)
if ttl <= 0 {
return
}
s.resolveMu.Lock()
defer s.resolveMu.Unlock()
if s.resolveCache == nil {
s.resolveCache = make(map[string]cloudResolveCacheEntry)
}
now := time.Now()
hits := 0
if existing, ok := s.resolveCache[key]; ok {
hits = existing.hits
}
s.resolveCache[key] = cloudResolveCacheEntry{link: cloneDirectLink(link), expiresAt: now.Add(ttl), hits: hits, lastHit: now}
}
func cloudResolveHotRefreshWindow(ttl time.Duration) time.Duration {
if ttl <= 0 {
return 0
}
window := ttl / 4
if window < 15*time.Second {
window = 15 * time.Second
}
if window > 2*time.Minute {
window = 2 * time.Minute
}
return window
}
func cloudResolveCacheTTL(typ string) time.Duration {
switch typ {
case cloud.TypeQuark, cloud.Type115, cloud.TypeCloudDrive2, cloud.TypeOpenList:
return 2 * time.Minute
default:
return 5 * time.Minute
}
}
func cloneDirectLink(link *cloud.DirectLink) *cloud.DirectLink {
if link == nil {
return nil
}
out := &cloud.DirectLink{
URL: link.URL,
Headers: make(map[string]string, len(link.Headers)),
Proxy: link.Proxy,
}
for k, v := range link.Headers {
out.Headers[k] = v
}
return out
}
func (s *StorageConfigService) clearResolveCacheForType(typ string) {
typ = strings.TrimSpace(typ)
if typ == "" {
return
}
prefix := typ + "\x00"
s.resolveMu.Lock()
defer s.resolveMu.Unlock()
for key := range s.resolveCache {
if strings.HasPrefix(key, prefix) {
delete(s.resolveCache, key)
}
}
for key, call := range s.resolveFlight {
if strings.HasPrefix(key, prefix) && call != nil {
call.err = fmt.Errorf("%s storage config changed", typ)
}
}
}
func (s *StorageConfigService) CloudResolveUncached(ctx context.Context, typ, fileRef, clientUA string) (*cloud.DirectLink, error) {
p, err := s.cloudProviderWithUA(ctx, typ, clientUA)
func (s *StorageConfigService) CloudMkdir(ctx context.Context, typ, parentDir, name string) (*cloud.FileEntry, error) {
p, err := s.CloudProvider(ctx, typ)
if err != nil {
return nil, err
}
return p.Resolve(ctx, fileRef)
mutable, ok := p.(cloud.MutableProvider)
if !ok {
return nil, fmt.Errorf("%s does not support folder creation", typ)
}
return mutable.Mkdir(ctx, parentDir, name)
}
// CloudReadText resolves a small cloud file and returns its text payload. It is
// used for cloud-hosted .strm files: the scanner reads the STRM target once and
// stores the real playback URL, while the media bytes still stay in the cloud.
func (s *StorageConfigService) CloudReadText(ctx context.Context, typ, fileRef string, limit int64) (string, error) {
if limit <= 0 {
limit = 64 << 10
}
link, err := s.CloudResolve(ctx, typ, fileRef, "")
if err != nil {
return "", err
}
req, err := http.NewRequestWithContext(ctx, http.MethodGet, link.URL, nil)
if err != nil {
return "", err
}
for k, v := range link.Headers {
req.Header.Set(k, v)
}
resp, err := s.client.Do(req)
if err != nil {
return "", err
}
defer resp.Body.Close()
if resp.StatusCode < 200 || resp.StatusCode >= 300 {
return "", fmt.Errorf("%s: read strm returned http %d", typ, resp.StatusCode)
}
body, err := io.ReadAll(io.LimitReader(resp.Body, limit+1))
if err != nil {
return "", err
}
if int64(len(body)) > limit {
return "", fmt.Errorf("%s: strm file is too large", typ)
}
return strings.TrimSpace(strings.TrimPrefix(string(body), "\ufeff")), nil
}
// cloudProviderWithUA builds a provider, overriding the request UA when a
// non-empty clientUA is supplied.
func (s *StorageConfigService) cloudProviderWithUA(ctx context.Context, typ, clientUA string) (cloud.Provider, error) {
if !cloud.IsCloudType(typ) {
return nil, fmt.Errorf("not a cloud provider: %q", typ)
}
view, err := s.Get(ctx, typ)
func (s *StorageConfigService) CloudRename(ctx context.Context, typ, ref, name string) (*cloud.FileEntry, error) {
p, err := s.CloudProvider(ctx, typ)
if err != nil {
return nil, err
}
if view == nil {
return nil, fmt.Errorf("%s storage not configured", typ)
mutable, ok := p.(cloud.MutableProvider)
if !ok {
return nil, fmt.Errorf("%s does not support rename", typ)
}
if !view.Enabled {
return nil, fmt.Errorf("%s storage disabled", typ)
return mutable.Rename(ctx, ref, name)
}
func (s *StorageConfigService) CloudMove(ctx context.Context, typ, ref, targetDir, name string) (*cloud.FileEntry, error) {
p, err := s.CloudProvider(ctx, typ)
if err != nil {
return nil, err
}
cfg := view.Config
if strings.TrimSpace(clientUA) != "" {
// Copy so we never mutate the cached view config.
cp := make(map[string]any, len(cfg)+1)
for k, v := range cfg {
cp[k] = v
}
cp["ua"] = clientUA
cfg = cp
movable, ok := p.(cloud.MovableProvider)
if !ok {
return nil, fmt.Errorf("%s does not support move", typ)
}
return cloud.New(typ, cfg, s.clientForConfig(cfg))
return movable.Move(ctx, ref, targetDir, name)
}
func (s *StorageConfigService) clientForConfig(cfg map[string]any) *http.Client {
@@ -704,8 +440,6 @@ func storageTimeoutFromConfig(cfg map[string]any, fallback time.Duration) time.D
// cloudLibraryName maps a provider type to a friendly Chinese library name.
func cloudLibraryName(typ string) string {
switch typ {
case cloud.TypeQuark:
return "夸克网盘"
case cloud.Type115:
return "115 网盘"
case cloud.TypeCloudDrive2:
@@ -766,7 +500,7 @@ func (s *StorageConfigService) CloudImport(ctx context.Context, typ, fileRef, na
Path: cloudMediaPath(typ, fileRef),
SizeBytes: size,
Container: container,
STRMURL: BuildPublicAPIURL(ctx, s.repo, nil, "/api/cloud/play/"+typ, url.Values{"ref": []string{fileRef}}),
STRMURL: BuildRelativeCloudPlayURL(typ, fileRef),
ScrapeStatus: "pending",
}
if err := s.repo.Media.Upsert(ctx, m); err != nil {
@@ -777,7 +511,7 @@ func (s *StorageConfigService) CloudImport(ctx context.Context, typ, fileRef, na
func validStorageType(t string) bool {
switch t {
case "alist", "s3", "webdav", cloud.TypeQuark, cloud.Type115, cloud.TypeCloudDrive2, cloud.TypeOpenList:
case "alist", "s3", "webdav", cloud.Type115, cloud.TypeCloudDrive2, cloud.TypeOpenList:
return true
}
return false