fix(embedding): preserve batching and concurrency limits for WeKnoraCloud
parent
aea51ecb62
commit
ca88fe74ae
|
|
@ -258,7 +258,7 @@ func newEmbedder(config Config, pooler EmbedderPooler, ollamaService *ollama.Oll
|
|||
embedder, err = zhipuEmb, zErr
|
||||
return embedder, err
|
||||
case provider.ProviderWeKnoraCloud:
|
||||
embedder, err = NewWeKnoraCloudEmbedder(config)
|
||||
embedder, err = NewWeKnoraCloudEmbedder(config, pooler)
|
||||
return embedder, err
|
||||
default:
|
||||
// Use OpenAI-compatible embedder for other providers
|
||||
|
|
|
|||
|
|
@ -32,7 +32,7 @@ type WeKnoraCloudEmbedder struct {
|
|||
}
|
||||
|
||||
// NewWeKnoraCloudEmbedder 构造 WeKnoraCloudEmbedder
|
||||
func NewWeKnoraCloudEmbedder(config Config) (*WeKnoraCloudEmbedder, error) {
|
||||
func NewWeKnoraCloudEmbedder(config Config, pooler EmbedderPooler) (*WeKnoraCloudEmbedder, error) {
|
||||
if config.AppID == "" {
|
||||
return nil, fmt.Errorf("WeKnoraCloud embedder: AppID is required")
|
||||
}
|
||||
|
|
@ -60,6 +60,7 @@ func NewWeKnoraCloudEmbedder(config Config) (*WeKnoraCloudEmbedder, error) {
|
|||
dimensions: config.Dimensions,
|
||||
supportsDimensionOverride: config.SupportsDimensionOverride,
|
||||
client: newEmbeddingHTTPClient(60 * time.Second),
|
||||
EmbedderPooler: pooler,
|
||||
}, nil
|
||||
}
|
||||
|
||||
|
|
@ -153,10 +154,6 @@ func (e *WeKnoraCloudEmbedder) BatchEmbed(ctx context.Context, texts []string) (
|
|||
return result, nil
|
||||
}
|
||||
|
||||
func (e *WeKnoraCloudEmbedder) BatchEmbedWithPool(ctx context.Context, model Embedder, texts []string) ([][]float32, error) {
|
||||
return e.BatchEmbed(ctx, texts)
|
||||
}
|
||||
|
||||
func (e *WeKnoraCloudEmbedder) SetSupportsDimensionOverride(supported bool) {
|
||||
e.supportsDimensionOverride = supported
|
||||
}
|
||||
|
|
|
|||
|
|
@ -0,0 +1,142 @@
|
|||
package embedding
|
||||
|
||||
import (
|
||||
"context"
|
||||
"encoding/json"
|
||||
"fmt"
|
||||
"io"
|
||||
"net/http"
|
||||
"reflect"
|
||||
"strings"
|
||||
"sync"
|
||||
"testing"
|
||||
"time"
|
||||
|
||||
"github.com/Tencent/WeKnora/internal/models/limiter"
|
||||
"github.com/Tencent/WeKnora/internal/models/provider"
|
||||
"github.com/Tencent/WeKnora/internal/types"
|
||||
secutils "github.com/Tencent/WeKnora/internal/utils"
|
||||
"github.com/panjf2000/ants/v2"
|
||||
)
|
||||
|
||||
type cloudBatchTestTransport func(*http.Request) (*http.Response, error)
|
||||
|
||||
func (f cloudBatchTestTransport) RoundTrip(r *http.Request) (*http.Response, error) {
|
||||
return f(r)
|
||||
}
|
||||
|
||||
// Exercise the factory's pool wiring and the concurrency wrapper with an
|
||||
// in-memory transport; no requests reach the configured address.
|
||||
func newCloudBatchTestEmbedder(t *testing.T, transport cloudBatchTestTransport) Embedder {
|
||||
t.Helper()
|
||||
t.Setenv("SSRF_WHITELIST", "cloud-batch.invalid")
|
||||
secutils.ResetSSRFWhitelistForTest()
|
||||
t.Cleanup(secutils.ResetSSRFWhitelistForTest)
|
||||
pool, err := ants.NewPool(4)
|
||||
if err != nil {
|
||||
t.Fatal(err)
|
||||
}
|
||||
t.Cleanup(pool.Release)
|
||||
raw, err := newEmbedder(Config{
|
||||
Source: types.ModelSourceRemote, Provider: string(provider.ProviderWeKnoraCloud),
|
||||
BaseURL: "https://cloud-batch.invalid", ModelID: "cloud-batch-test", ModelName: "test",
|
||||
AppID: "test-app", AppSecret: "test-secret",
|
||||
}, NewBatchEmbedder(pool), nil)
|
||||
if err != nil {
|
||||
t.Fatal(err)
|
||||
}
|
||||
raw.(*WeKnoraCloudEmbedder).client = &http.Client{Transport: transport}
|
||||
return wrapEmbeddingConcurrency(raw, 1)
|
||||
}
|
||||
|
||||
func TestWeKnoraCloudPoolBatchesAndPreservesOrder(t *testing.T) {
|
||||
t.Setenv("BATCH_EMBED_SIZE", "2")
|
||||
var mu sync.Mutex
|
||||
var sizes []int
|
||||
embedder := newCloudBatchTestEmbedder(t, func(r *http.Request) (*http.Response, error) {
|
||||
var input weKnoraCloudEmbedRequest
|
||||
if err := json.NewDecoder(r.Body).Decode(&input); err != nil {
|
||||
return nil, err
|
||||
}
|
||||
mu.Lock()
|
||||
sizes = append(sizes, len(input.Input))
|
||||
mu.Unlock()
|
||||
// Return provider entries in reverse order to exercise both index
|
||||
// restoration within a batch and ordered reassembly across batches.
|
||||
entries := make([]string, 0, len(input.Input))
|
||||
for i := len(input.Input) - 1; i >= 0; i-- {
|
||||
entries = append(entries, fmt.Sprintf(`{"index":%d,"embedding":[%d]}`, i, len(input.Input[i])))
|
||||
}
|
||||
return &http.Response{
|
||||
StatusCode: http.StatusOK,
|
||||
Body: io.NopCloser(strings.NewReader(`{"data":[` + strings.Join(entries, ",") + `]}`)),
|
||||
}, nil
|
||||
})
|
||||
got, err := embedder.BatchEmbedWithPool(context.Background(), embedder, []string{"a", "bb", "ccc", "dddd", "eeeee"})
|
||||
if err != nil {
|
||||
t.Fatal(err)
|
||||
}
|
||||
if want := [][]float32{{1}, {2}, {3}, {4}, {5}}; !reflect.DeepEqual(got, want) {
|
||||
t.Fatalf("embeddings = %v, want %v", got, want)
|
||||
}
|
||||
if len(sizes) != 3 {
|
||||
t.Fatalf("request sizes = %v, want three batches", sizes)
|
||||
}
|
||||
for _, size := range sizes {
|
||||
if size < 1 || size > 2 {
|
||||
t.Fatalf("request size %d violates BATCH_EMBED_SIZE=2", size)
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
func TestWeKnoraCloudPoolHonoursConcurrencyLimit(t *testing.T) {
|
||||
t.Setenv("BATCH_EMBED_SIZE", "1")
|
||||
limiter.SetGovernor(limiter.NewLocalLimiter(), 10)
|
||||
t.Cleanup(func() { limiter.SetGovernor(nil, 0) })
|
||||
entered := make(chan struct{}, 2)
|
||||
release := make(chan struct{})
|
||||
embedder := newCloudBatchTestEmbedder(t, func(*http.Request) (*http.Response, error) {
|
||||
entered <- struct{}{}
|
||||
<-release
|
||||
return &http.Response{
|
||||
StatusCode: http.StatusOK,
|
||||
Body: io.NopCloser(strings.NewReader(`{"data":[{"index":0,"embedding":[1]}]}`)),
|
||||
}, nil
|
||||
})
|
||||
ctx := types.WithBackgroundTask(context.Background())
|
||||
var wg sync.WaitGroup
|
||||
var once sync.Once
|
||||
unblock := func() { once.Do(func() { close(release) }) }
|
||||
t.Cleanup(func() { unblock(); wg.Wait() })
|
||||
errs := make(chan error, 2)
|
||||
// Separate callers represent simultaneous wiki batches sharing a model.
|
||||
for range 2 {
|
||||
wg.Add(1)
|
||||
go func() {
|
||||
defer wg.Done()
|
||||
_, err := embedder.BatchEmbedWithPool(ctx, embedder, []string{"x"})
|
||||
errs <- err
|
||||
}()
|
||||
}
|
||||
select {
|
||||
case <-entered:
|
||||
case <-time.After(2 * time.Second):
|
||||
t.Fatal("first provider request did not enter")
|
||||
}
|
||||
select {
|
||||
case <-entered:
|
||||
t.Error("two provider requests entered concurrently with model limit=1")
|
||||
case <-time.After(150 * time.Millisecond):
|
||||
}
|
||||
unblock()
|
||||
for range 2 {
|
||||
select {
|
||||
case err := <-errs:
|
||||
if err != nil {
|
||||
t.Error(err)
|
||||
}
|
||||
case <-time.After(2 * time.Second):
|
||||
t.Fatal("pooled embedding did not complete after release")
|
||||
}
|
||||
}
|
||||
}
|
||||
Loading…
Reference in New Issue