Files
im_server/server/kafka_file_lifecycle.go
T

266 lines
9.0 KiB
Go

package main
import (
"context"
"encoding/json"
"errors"
"fmt"
"strings"
"time"
"github.com/segmentio/kafka-go"
"github.com/tinode/chat/server/logs"
"github.com/tinode/chat/server/store"
"github.com/tinode/chat/server/store/types"
)
const (
kafkaFileLifecycleUploadCompleted = "file.upload.completed"
kafkaFileLifecycleAttachmentsBind = "file.attachments.bind"
kafkaFileLifecycleGroupFileCreate = "group.file.created"
kafkaFileLifecycleGroupFileDelete = "group.file.deleted"
)
type kafkaFileLifecyclePayload struct {
Action string `json:"action"`
FileID string `json:"file_id,omitempty"`
FileIDs []string `json:"file_ids,omitempty"`
AttachmentURLs []string `json:"attachment_urls,omitempty"`
Topic string `json:"topic,omitempty"`
MessageID string `json:"message_id,omitempty"`
SeqID int64 `json:"seq_id,omitempty"`
UserID string `json:"user_id,omitempty"`
GroupFileID string `json:"group_file_id,omitempty"`
FileType string `json:"file_type,omitempty"`
FileName string `json:"file_name,omitempty"`
MIMEType string `json:"mime_type,omitempty"`
FileSize int64 `json:"file_size,omitempty"`
Storage map[string]any `json:"storage,omitempty"`
OccurredAt time.Time `json:"occurred_at"`
}
func startKafkaFileLifecycleConsumer() {
startKafkaConsumer("file_lifecycle", handleKafkaFileLifecycle)
}
func enqueueKafkaFileUploaded(ctx context.Context, tenantID types.TenantID, fd *types.FileDef) error {
if !kafkaEnabled() || fd == nil || fd.TenantID != tenantID || fd.Status != types.UploadCompleted {
return nil
}
payload := kafkaFileLifecyclePayload{
Action: kafkaFileLifecycleUploadCompleted,
FileID: fd.Uid().String(),
UserID: types.ParseUid(fd.User).UserId(),
MIMEType: fd.MimeType,
FileSize: fd.Size,
OccurredAt: fd.UpdatedAt,
Storage: map[string]any{
"location": fd.Location,
"etag": fd.ETag,
},
}
return enqueueKafkaFileLifecycle(ctx, tenantID, kafkaFileLifecycleUploadCompleted, fd.Uid().String(), payload)
}
func enqueueKafkaFileAttachmentsBind(ctx context.Context, tenantID types.TenantID, msg *types.Message, attachmentURLs []string) error {
if !kafkaEnabled() || msg == nil || msg.TenantID != tenantID || len(attachmentURLs) == 0 {
return nil
}
payload := kafkaFileLifecyclePayload{
Action: kafkaFileLifecycleAttachmentsBind,
AttachmentURLs: append([]string(nil), attachmentURLs...),
Topic: msg.Topic,
MessageID: msg.Uid().String(),
SeqID: msg.SeqId,
UserID: types.ParseUid(msg.From).UserId(),
FileType: lifecycleFileTypeFromMessageHead(msg.Head),
OccurredAt: msg.CreatedAt,
}
return enqueueKafkaFileLifecycle(ctx, tenantID, kafkaFileLifecycleAttachmentsBind, fmt.Sprintf("%s:%d", msg.Topic, msg.SeqId), payload)
}
func enqueueKafkaGroupFileLifecycle(ctx context.Context, tenantID types.TenantID, eventType string, file *types.GroupFile, actor types.Uid) error {
if !kafkaEnabled() || file == nil || file.Topic == "" || file.FileID == "" {
return nil
}
occurredAt := file.UpdatedAt
if occurredAt.IsZero() {
occurredAt = time.Now().UTC().Round(time.Millisecond)
}
payload := kafkaFileLifecyclePayload{
Action: eventType,
FileID: file.FileID,
FileIDs: []string{file.FileID},
Topic: file.Topic,
GroupFileID: file.GroupFileID,
SeqID: file.SourceSeq,
UserID: actor.UserId(),
FileType: file.FileType,
FileName: file.FileName,
MIMEType: file.MIMEType,
FileSize: file.FileSize,
OccurredAt: occurredAt,
}
return enqueueKafkaFileLifecycle(ctx, tenantID, eventType, firstNonEmpty(file.GroupFileID, file.FileID), payload)
}
func enqueueKafkaFileLifecycle(_ context.Context, tenantID types.TenantID, eventType, bizID string, payload kafkaFileLifecyclePayload) error {
tenantCode := ""
if store.Tenants != nil {
if tenant, err := store.Tenants.GetByID(tenantID); err == nil && tenant != nil {
tenantCode = tenant.Code
} else {
logs.Warn.Printf("kafka file lifecycle: failed to resolve tenant code for tenant %d: %v", tenantID, err)
}
}
if payload.OccurredAt.IsZero() {
payload.OccurredAt = time.Now().UTC().Round(time.Millisecond)
}
encoded, err := json.Marshal(payload)
if err != nil {
return err
}
eventID := stableIMEventID("file-lifecycle", tenantID, eventType, bizID, payload.SeqID, strings.Join(payload.AttachmentURLs, ","))
err = enqueueKafkaOutbox(tenantID, "file_lifecycle", bizID, kafkaEnvelope{
EventID: eventID,
TenantCode: tenantCode,
SourceService: "im-server",
TargetService: "im-server",
EventType: eventType,
BizType: "file",
BizID: bizID,
OccurredAt: payload.OccurredAt,
Payload: encoded,
})
if err != nil {
return err
}
actor := types.ParseUserId(payload.UserID)
if auditErr := enqueueKafkaAuditEvent(context.Background(), tenantID, "user", actor, eventType, "file", bizID, "succeeded", "", map[string]any{
"file_id": payload.FileID,
"file_ids": payload.FileIDs,
"topic": payload.Topic,
"group_file_id": payload.GroupFileID,
"seq_id": payload.SeqID,
}); auditErr != nil {
logs.Warn.Printf("kafka file lifecycle: failed to enqueue audit event type=%s biz=%s: %v", eventType, bizID, auditErr)
}
return nil
}
func handleKafkaFileLifecycle(ctx context.Context, envelope kafkaEnvelope, _ kafka.Message) error {
tenantID, err := kafkaTenantID(envelope.TenantCode)
if err != nil {
return err
}
consume, claimed, err := beginKafkaConsumeForTopic(tenantID, envelope, "file_lifecycle")
if err != nil || !claimed {
return err
}
var payload kafkaFileLifecyclePayload
if err = json.Unmarshal(envelope.Payload, &payload); err != nil {
return markKafkaDeadLetter(ctx, tenantID, envelope, consume, "MALFORMED_PAYLOAD", err.Error())
}
if err = executeKafkaFileLifecycle(tenantID, payload); err != nil {
failed, markErr := store.HTTPInfrastructure.MarkKafkaConsumeFailed(tenantID, envelope.EventID, kafkaBus.config.ConsumerGroup,
kafkaCommandErrorReason(err), err.Error(), time.Now().UTC().Round(time.Millisecond))
if markErr != nil {
return markErr
}
if failed != nil && failed.RetryCount >= kafkaBus.config.MaxRetryCount {
return markKafkaDeadLetter(ctx, tenantID, envelope, failed, kafkaCommandErrorReason(err), err.Error())
}
sleepKafkaRetryBackoff(failed.RetryCount)
return err
}
return store.HTTPInfrastructure.MarkKafkaConsumeSucceeded(tenantID, envelope.EventID, kafkaBus.config.ConsumerGroup, time.Now().UTC().Round(time.Millisecond))
}
func executeKafkaFileLifecycle(tenantID types.TenantID, payload kafkaFileLifecyclePayload) error {
switch payload.Action {
case kafkaFileLifecycleAttachmentsBind:
return executeKafkaFileAttachmentsBind(tenantID, payload)
case kafkaFileLifecycleUploadCompleted, kafkaFileLifecycleGroupFileCreate, kafkaFileLifecycleGroupFileDelete:
return nil
default:
return types.ErrMalformed
}
}
func executeKafkaFileAttachmentsBind(tenantID types.TenantID, payload kafkaFileLifecyclePayload) error {
if payload.Topic == "" || payload.MessageID == "" || len(payload.AttachmentURLs) == 0 {
return types.ErrMalformed
}
msgID := types.ParseUid(payload.MessageID)
if msgID.IsZero() {
return types.ErrMalformed
}
if types.GetTopicCat(payload.Topic) != types.TopicCatGrp {
return store.Files.LinkAttachments(tenantID, "", msgID, payload.AttachmentURLs)
}
uploader := types.ParseUserId(payload.UserID)
if uploader.IsZero() {
return types.ErrMalformed
}
for _, url := range payload.AttachmentURLs {
fid := store.Store.GetMediaHandler().GetIdFromUrl(url)
if fid.IsZero() {
continue
}
_, err := store.GroupBusiness.CreateFile(tenantID, &types.GroupFile{
Topic: payload.Topic,
FileID: fid.String(),
SourceTopic: payload.Topic,
SourceSeq: payload.SeqID,
UploaderID: uploader,
FileType: lifecycleFileType(payload.FileType),
IdempotencyKey: fmt.Sprintf("%s:%d:%s", payload.Topic, payload.SeqID, fid.String()),
})
if err != nil && !errors.Is(err, types.ErrDuplicate) {
return err
}
}
return nil
}
func lifecycleFileTypeFromMessageHead(head types.KVMap) string {
msgType := lifecycleHeaderString(head, "x-im-type")
if msgType == "" {
msgType = lifecycleHeaderString(head, "type")
}
if msgType == "" {
return "file"
}
return lifecycleFileType(msgType)
}
func lifecycleFileType(raw string) string {
switch lifecycleCanonicalMessageType(raw) {
case "image", "audio", "voice", "video":
return lifecycleCanonicalMessageType(raw)
default:
return "file"
}
}
func lifecycleHeaderString(head types.KVMap, key string) string {
if head == nil {
return ""
}
value, ok := head[key]
if !ok {
return ""
}
str, _ := value.(string)
return strings.TrimSpace(str)
}
func lifecycleCanonicalMessageType(raw string) string {
switch strings.ToLower(strings.TrimSpace(raw)) {
case "audio":
return "voice"
default:
return strings.ToLower(strings.TrimSpace(raw))
}
}