feat(antigravity): comprehensive enhancements - model mapping, rate limiting, scheduling & ops

Key changes:
- Upgrade model mapping: Opus 4.5 → Opus 4.6-thinking with precise matching
- Unified rate limiting: scope-level → model-level with Redis snapshot sync
- Load-balanced scheduling by call count with smart retry mechanism
- Force cache billing support
- Model identity injection in prompts with leak prevention
- Thinking mode auto-handling (max_tokens/budget_tokens fix)
- Frontend: whitelist mode toggle, model mapping validation, status indicators
- Gemini session fallback with Redis Trie O(L) matching
- Ops: enhanced concurrency monitoring, account availability, retry logic
- Migration scripts: 049-051 for model mapping unification
This commit is contained in:
erio
2026-02-07 12:31:10 +08:00
parent e617b45ba3
commit 5e98445b22
73 changed files with 8553 additions and 1926 deletions

View File

@@ -216,6 +216,22 @@ func (m *mockGatewayCacheForPlatform) DeleteSessionAccountID(ctx context.Context
return nil
}
func (m *mockGatewayCacheForPlatform) IncrModelCallCount(ctx context.Context, accountID int64, model string) (int64, error) {
return 0, nil
}
func (m *mockGatewayCacheForPlatform) GetModelLoadBatch(ctx context.Context, accountIDs []int64, model string) (map[int64]*ModelLoadInfo, error) {
return nil, nil
}
func (m *mockGatewayCacheForPlatform) FindGeminiSession(ctx context.Context, groupID int64, prefixHash, digestChain string) (uuid string, accountID int64, found bool) {
return "", 0, false
}
func (m *mockGatewayCacheForPlatform) SaveGeminiSession(ctx context.Context, groupID int64, prefixHash, digestChain, uuid string, accountID int64) error {
return nil
}
type mockGroupRepoForGateway struct {
groups map[int64]*Group
getByIDCalls int
@@ -332,7 +348,7 @@ func TestGatewayService_SelectAccountForModelWithPlatform_Antigravity(t *testing
cfg: testConfig(),
}
acc, err := svc.selectAccountForModelWithPlatform(ctx, nil, "", "claude-3-5-sonnet-20241022", nil, PlatformAntigravity)
acc, err := svc.selectAccountForModelWithPlatform(ctx, nil, "", "claude-sonnet-4-5", nil, PlatformAntigravity)
require.NoError(t, err)
require.NotNil(t, acc)
require.Equal(t, int64(2), acc.ID)
@@ -670,7 +686,7 @@ func TestGatewayService_SelectAccountForModelWithExclusions_ForcePlatform(t *tes
cfg: testConfig(),
}
acc, err := svc.SelectAccountForModelWithExclusions(ctx, nil, "", "claude-3-5-sonnet-20241022", nil)
acc, err := svc.SelectAccountForModelWithExclusions(ctx, nil, "", "claude-sonnet-4-5", nil)
require.NoError(t, err)
require.NotNil(t, acc)
require.Equal(t, int64(2), acc.ID)
@@ -1014,10 +1030,16 @@ func TestGatewayService_isModelSupportedByAccount(t *testing.T) {
expected bool
}{
{
name: "Antigravity平台-支持claude模型",
name: "Antigravity平台-支持默认映射中的claude模型",
account: &Account{Platform: PlatformAntigravity},
model: "claude-sonnet-4-5",
expected: true,
},
{
name: "Antigravity平台-不支持非默认映射中的claude模型",
account: &Account{Platform: PlatformAntigravity},
model: "claude-3-5-sonnet-20241022",
expected: true,
expected: false,
},
{
name: "Antigravity平台-支持gemini模型",
@@ -1115,7 +1137,7 @@ func TestGatewayService_selectAccountWithMixedScheduling(t *testing.T) {
cfg: testConfig(),
}
acc, err := svc.selectAccountWithMixedScheduling(ctx, nil, "", "claude-3-5-sonnet-20241022", nil, PlatformAnthropic)
acc, err := svc.selectAccountWithMixedScheduling(ctx, nil, "", "claude-sonnet-4-5", nil, PlatformAnthropic)
require.NoError(t, err)
require.NotNil(t, acc)
require.Equal(t, int64(2), acc.ID, "应选择优先级最高的账户包含启用混合调度的antigravity")
@@ -1123,7 +1145,7 @@ func TestGatewayService_selectAccountWithMixedScheduling(t *testing.T) {
t.Run("混合调度-路由优先选择路由账号", func(t *testing.T) {
groupID := int64(30)
requestedModel := "claude-3-5-sonnet-20241022"
requestedModel := "claude-sonnet-4-5"
repo := &mockAccountRepoForPlatform{
accounts: []Account{
{ID: 1, Platform: PlatformAnthropic, Priority: 1, Status: StatusActive, Schedulable: true},
@@ -1168,7 +1190,7 @@ func TestGatewayService_selectAccountWithMixedScheduling(t *testing.T) {
t.Run("混合调度-路由粘性命中", func(t *testing.T) {
groupID := int64(31)
requestedModel := "claude-3-5-sonnet-20241022"
requestedModel := "claude-sonnet-4-5"
repo := &mockAccountRepoForPlatform{
accounts: []Account{
{ID: 1, Platform: PlatformAnthropic, Priority: 1, Status: StatusActive, Schedulable: true},
@@ -1320,7 +1342,7 @@ func TestGatewayService_selectAccountWithMixedScheduling(t *testing.T) {
Schedulable: true,
Extra: map[string]any{
"model_rate_limits": map[string]any{
"claude_sonnet": map[string]any{
"claude-3-5-sonnet-20241022": map[string]any{
"rate_limit_reset_at": resetAt.Format(time.RFC3339),
},
},
@@ -1465,7 +1487,7 @@ func TestGatewayService_selectAccountWithMixedScheduling(t *testing.T) {
cfg: testConfig(),
}
acc, err := svc.selectAccountWithMixedScheduling(ctx, nil, "session-123", "claude-3-5-sonnet-20241022", nil, PlatformAnthropic)
acc, err := svc.selectAccountWithMixedScheduling(ctx, nil, "session-123", "claude-sonnet-4-5", nil, PlatformAnthropic)
require.NoError(t, err)
require.NotNil(t, acc)
require.Equal(t, int64(2), acc.ID, "应返回粘性会话绑定的启用mixed_scheduling的antigravity账户")
@@ -1597,7 +1619,7 @@ func TestGatewayService_selectAccountWithMixedScheduling(t *testing.T) {
cfg: testConfig(),
}
acc, err := svc.selectAccountWithMixedScheduling(ctx, nil, "", "claude-3-5-sonnet-20241022", nil, PlatformAnthropic)
acc, err := svc.selectAccountWithMixedScheduling(ctx, nil, "", "claude-sonnet-4-5", nil, PlatformAnthropic)
require.NoError(t, err)
require.NotNil(t, acc)
require.Equal(t, int64(1), acc.ID)
@@ -1870,6 +1892,19 @@ func (m *mockConcurrencyCache) CleanupExpiredAccountSlots(ctx context.Context, a
return nil
}
func (m *mockConcurrencyCache) GetUsersLoadBatch(ctx context.Context, users []UserWithConcurrency) (map[int64]*UserLoadInfo, error) {
result := make(map[int64]*UserLoadInfo, len(users))
for _, user := range users {
result[user.ID] = &UserLoadInfo{
UserID: user.ID,
CurrentConcurrency: 0,
WaitingCount: 0,
LoadRate: 0,
}
}
return result, nil
}
// TestGatewayService_SelectAccountWithLoadAwareness tests load-aware account selection
func TestGatewayService_SelectAccountWithLoadAwareness(t *testing.T) {
ctx := context.Background()
@@ -2747,7 +2782,7 @@ func TestGatewayService_SelectAccountWithLoadAwareness(t *testing.T) {
Concurrency: 5,
Extra: map[string]any{
"model_rate_limits": map[string]any{
"claude_sonnet": map[string]any{
"claude-3-5-sonnet-20241022": map[string]any{
"rate_limit_reset_at": now.Format(time.RFC3339),
},
},