refactor: 拆分 gateway 单体为独立 crate,新增 systemd 部署方案

将 gateway 内部的 model-fetch、provider-transport、scheduler-core、
usage-runtime、video-tasks-core 模块提取为独立 crate;重构 gateway
内部模块结构(state/router/cache/data/query 等);移除大量遗留模块
文件;新增 systemd 二进制部署骨架及相关文档;更新前端 usage 相关
API 和组件。
This commit is contained in:
fawney19
2026-04-05 20:23:16 +08:00
parent cbc811f6ce
commit 763ff03a7b
777 changed files with 42654 additions and 21464 deletions
+714
View File
@@ -0,0 +1,714 @@
use base64::{engine::general_purpose::STANDARD as BASE64_STANDARD, Engine as _};
use super::{
any, build_router_with_state, build_state_with_execution_runtime_override, hash_api_key, json,
sample_auth_snapshot, sample_files_candidate_row, sample_files_provider_catalog_endpoint,
sample_files_provider_catalog_key, sample_files_provider_catalog_provider, start_server,
to_bytes, Arc, Body, InMemoryAuthApiKeySnapshotRepository,
InMemoryMinimalCandidateSelectionReadRepository, InMemoryProviderCatalogReadRepository,
InMemoryRequestCandidateRepository, Json, Mutex, Request, RequestCandidateReadRepository,
RequestCandidateStatus, Router, StatusCode, DEVELOPMENT_ENCRYPTION_KEY, TRACE_ID_HEADER,
};
#[tokio::test]
async fn gateway_executes_gemini_files_upload_via_local_decision_gate_with_local_planning_only() {
#[derive(Debug, Clone)]
struct SeenExecutionRuntimeSyncRequest {
method: String,
url: String,
auth_header_value: String,
content_type: String,
body_bytes_b64: String,
endpoint_tag: String,
proxy_node_id: String,
tls_profile: String,
}
let decision_hits = Arc::new(Mutex::new(0usize));
let decision_hits_clone = Arc::clone(&decision_hits);
let plan_hits = Arc::new(Mutex::new(0usize));
let plan_hits_clone = Arc::clone(&plan_hits);
let seen_execution_runtime = Arc::new(Mutex::new(None::<SeenExecutionRuntimeSyncRequest>));
let seen_execution_runtime_clone = Arc::clone(&seen_execution_runtime);
let report_hits = Arc::new(Mutex::new(0usize));
let report_hits_clone = Arc::clone(&report_hits);
let public_hits = Arc::new(Mutex::new(0usize));
let public_hits_clone = Arc::clone(&public_hits);
let upstream = Router::new()
.route(
"/api/internal/gateway/resolve",
any(|_request: Request| async move {
Json(json!({
"action": "proxy_public",
"route_class": "ai_public",
"route_family": "gemini",
"route_kind": "files",
"auth_endpoint_signature": "gemini:chat",
"execution_runtime_candidate": true,
"auth_context": {
"user_id": "user-files-upload-local-123",
"api_key_id": "key-files-upload-local-123",
"access_allowed": true
},
"public_path": "/upload/v1beta/files"
}))
}),
)
.route(
"/api/internal/gateway/decision-sync",
any(move |_request: Request| {
let decision_hits_inner = Arc::clone(&decision_hits_clone);
async move {
*decision_hits_inner.lock().expect("mutex should lock") += 1;
Json(json!({
"action": "execution_runtime_sync_decision",
"decision_kind": "unexpected_remote_decision"
}))
}
}),
)
.route(
"/api/internal/gateway/plan-sync",
any(move |_request: Request| {
let plan_hits_inner = Arc::clone(&plan_hits_clone);
async move {
*plan_hits_inner.lock().expect("mutex should lock") += 1;
Json(json!({
"action": "execution_runtime_sync",
"plan_kind": "unexpected_plan_fallback"
}))
}
}),
)
.route(
"/api/internal/gateway/report-sync",
any(move |_request: Request| {
let report_hits_inner = Arc::clone(&report_hits_clone);
async move {
*report_hits_inner.lock().expect("mutex should lock") += 1;
Json(json!({"ok": true}))
}
}),
)
.route(
"/upload/v1beta/files",
any(move |_request: Request| {
let public_hits_inner = Arc::clone(&public_hits_clone);
async move {
*public_hits_inner.lock().expect("mutex should lock") += 1;
(StatusCode::IM_A_TEAPOT, Body::from("public-route-hit"))
}
}),
);
let execution_runtime = Router::new().route(
"/v1/execute/sync",
any(move |request: Request| {
let seen_execution_runtime_inner = Arc::clone(&seen_execution_runtime_clone);
async move {
let (_parts, body) = request.into_parts();
let raw_body = to_bytes(body, usize::MAX).await.expect("body should read");
let payload: serde_json::Value = serde_json::from_slice(&raw_body)
.expect("execution runtime payload should parse");
*seen_execution_runtime_inner
.lock()
.expect("mutex should lock") = Some(SeenExecutionRuntimeSyncRequest {
method: payload
.get("method")
.and_then(|value| value.as_str())
.unwrap_or_default()
.to_string(),
url: payload
.get("url")
.and_then(|value| value.as_str())
.unwrap_or_default()
.to_string(),
auth_header_value: payload
.get("headers")
.and_then(|value| value.get("x-goog-api-key"))
.and_then(|value| value.as_str())
.unwrap_or_default()
.to_string(),
content_type: payload
.get("headers")
.and_then(|value| value.get("content-type"))
.and_then(|value| value.as_str())
.unwrap_or_default()
.to_string(),
body_bytes_b64: payload
.get("body")
.and_then(|value| value.get("body_bytes_b64"))
.and_then(|value| value.as_str())
.unwrap_or_default()
.to_string(),
endpoint_tag: payload
.get("headers")
.and_then(|value| value.get("x-endpoint-tag"))
.and_then(|value| value.as_str())
.unwrap_or_default()
.to_string(),
proxy_node_id: payload
.get("proxy")
.and_then(|value| value.get("node_id"))
.and_then(|value| value.as_str())
.unwrap_or_default()
.to_string(),
tls_profile: payload
.get("tls_profile")
.and_then(|value| value.as_str())
.unwrap_or_default()
.to_string(),
});
Json(json!({
"request_id": "trace-gemini-files-upload-local-123",
"status_code": 200,
"headers": {
"content-type": "application/json"
},
"body": {
"json_body": {
"file": {
"name": "files/uploaded-local-123"
}
}
},
"telemetry": {
"elapsed_ms": 19
}
}))
}
}),
);
let auth_repository = Arc::new(InMemoryAuthApiKeySnapshotRepository::seed(vec![(
Some(hash_api_key("client-files-upload-local-key")),
sample_auth_snapshot("key-files-upload-local-123", "user-files-upload-local-123"),
)]));
let candidate_selection_repository =
Arc::new(InMemoryMinimalCandidateSelectionReadRepository::seed(vec![
sample_files_candidate_row(),
]));
let request_candidate_repository = Arc::new(InMemoryRequestCandidateRepository::default());
let mut provider = sample_files_provider_catalog_provider();
provider.proxy = Some(serde_json::json!({"url":"http://provider-proxy.internal:8080"}));
let mut endpoint = sample_files_provider_catalog_endpoint();
endpoint.custom_path = Some("/custom/upload/v1beta/files".to_string());
endpoint.header_rules = Some(
serde_json::json!([{"action":"set","key":"x-endpoint-tag","value":"gemini-files-upload-local"}]),
);
let mut key = sample_files_provider_catalog_key();
key.proxy = Some(
serde_json::json!({"enabled": true, "node_id":"proxy-node-gemini-files-upload-local"}),
);
key.fingerprint = Some(serde_json::json!({"tls_profile":"chrome_136"}));
let provider_catalog_repository = Arc::new(InMemoryProviderCatalogReadRepository::seed(
vec![provider],
vec![endpoint],
vec![key],
));
let (upstream_url, upstream_handle) = start_server(upstream).await;
let (execution_runtime_url, execution_runtime_handle) = start_server(execution_runtime).await;
let gateway_state =
build_state_with_execution_runtime_override(execution_runtime_url.clone())
.with_data_state_for_tests(
crate::data::GatewayDataState::with_auth_candidate_selection_provider_catalog_and_request_candidate_repository_for_tests(
auth_repository,
candidate_selection_repository,
provider_catalog_repository,
Arc::clone(&request_candidate_repository),
DEVELOPMENT_ENCRYPTION_KEY,
),
);
let gateway = build_router_with_state(gateway_state);
let (gateway_url, gateway_handle) = start_server(gateway).await;
let response = reqwest::Client::new()
.post(format!(
"{gateway_url}/upload/v1beta/files?uploadType=resumable&key=client-files-upload-local-key"
))
.header("x-goog-api-key", "client-header-key")
.header(http::header::CONTENT_TYPE, "application/octet-stream")
.header(TRACE_ID_HEADER, "trace-gemini-files-upload-local-123")
.body("upload-local-bytes")
.send()
.await
.expect("request should succeed");
assert_eq!(response.status(), StatusCode::OK);
assert_eq!(
response
.json::<serde_json::Value>()
.await
.expect("json should parse"),
json!({"file": {"name": "files/uploaded-local-123"}})
);
let seen_execution_runtime_request = seen_execution_runtime
.lock()
.expect("mutex should lock")
.clone()
.expect("execution runtime sync should be captured");
assert_eq!(seen_execution_runtime_request.method, "POST");
assert_eq!(
seen_execution_runtime_request.url,
"https://generativelanguage.googleapis.com/custom/upload/v1beta/files?uploadType=resumable"
);
assert_eq!(
seen_execution_runtime_request.auth_header_value,
"sk-upstream-gemini-files"
);
assert_eq!(
seen_execution_runtime_request.content_type,
"application/octet-stream"
);
assert_eq!(
seen_execution_runtime_request.endpoint_tag,
"gemini-files-upload-local"
);
assert_eq!(
BASE64_STANDARD
.decode(seen_execution_runtime_request.body_bytes_b64)
.expect("execution runtime body should decode"),
b"upload-local-bytes"
);
assert_eq!(
seen_execution_runtime_request.proxy_node_id,
"proxy-node-gemini-files-upload-local"
);
assert_eq!(seen_execution_runtime_request.tls_profile, "chrome_136");
let stored_candidates = request_candidate_repository
.list_by_request_id("trace-gemini-files-upload-local-123")
.await
.expect("request candidate trace should read");
assert_eq!(stored_candidates.len(), 1);
assert_eq!(stored_candidates[0].status, RequestCandidateStatus::Success);
assert_eq!(*report_hits.lock().expect("mutex should lock"), 0);
assert_eq!(*decision_hits.lock().expect("mutex should lock"), 0);
assert_eq!(*plan_hits.lock().expect("mutex should lock"), 0);
assert_eq!(*public_hits.lock().expect("mutex should lock"), 0);
gateway_handle.abort();
execution_runtime_handle.abort();
upstream_handle.abort();
}
#[tokio::test]
async fn gateway_executes_gemini_files_list_via_local_decision_gate_with_local_planning_only() {
#[derive(Debug, Clone)]
struct SeenExecutionRuntimeSyncRequest {
method: String,
url: String,
auth_header_value: String,
}
let decision_hits = Arc::new(Mutex::new(0usize));
let decision_hits_clone = Arc::clone(&decision_hits);
let plan_hits = Arc::new(Mutex::new(0usize));
let plan_hits_clone = Arc::clone(&plan_hits);
let seen_execution_runtime = Arc::new(Mutex::new(None::<SeenExecutionRuntimeSyncRequest>));
let seen_execution_runtime_clone = Arc::clone(&seen_execution_runtime);
let report_hits = Arc::new(Mutex::new(0usize));
let report_hits_clone = Arc::clone(&report_hits);
let public_hits = Arc::new(Mutex::new(0usize));
let public_hits_clone = Arc::clone(&public_hits);
let upstream = Router::new()
.route(
"/api/internal/gateway/resolve",
any(|_request: Request| async move {
Json(json!({
"action": "proxy_public",
"route_class": "ai_public",
"route_family": "gemini",
"route_kind": "files",
"auth_endpoint_signature": "gemini:chat",
"execution_runtime_candidate": true,
"auth_context": {
"user_id": "user-files-list-local-123",
"api_key_id": "key-files-list-local-123",
"access_allowed": true
},
"public_path": "/v1beta/files"
}))
}),
)
.route(
"/api/internal/gateway/decision-sync",
any(move |_request: Request| {
let decision_hits_inner = Arc::clone(&decision_hits_clone);
async move {
*decision_hits_inner.lock().expect("mutex should lock") += 1;
Json(json!({"action": "proxy_public"}))
}
}),
)
.route(
"/api/internal/gateway/plan-sync",
any(move |_request: Request| {
let plan_hits_inner = Arc::clone(&plan_hits_clone);
async move {
*plan_hits_inner.lock().expect("mutex should lock") += 1;
Json(json!({"action": "proxy_public"}))
}
}),
)
.route(
"/api/internal/gateway/report-sync",
any(move |_request: Request| {
let report_hits_inner = Arc::clone(&report_hits_clone);
async move {
*report_hits_inner.lock().expect("mutex should lock") += 1;
StatusCode::NO_CONTENT
}
}),
)
.route(
"/v1beta/files",
any(move |_request: Request| {
let public_hits_inner = Arc::clone(&public_hits_clone);
async move {
*public_hits_inner.lock().expect("mutex should lock") += 1;
(StatusCode::IM_A_TEAPOT, Body::from("public-route-hit"))
}
}),
);
let execution_runtime = Router::new().route(
"/v1/execute/sync",
any(move |request: Request| {
let seen_execution_runtime_inner = Arc::clone(&seen_execution_runtime_clone);
async move {
let (_parts, body) = request.into_parts();
let raw_body = to_bytes(body, usize::MAX).await.expect("body should read");
let payload: serde_json::Value = serde_json::from_slice(&raw_body)
.expect("execution runtime payload should parse");
*seen_execution_runtime_inner
.lock()
.expect("mutex should lock") = Some(SeenExecutionRuntimeSyncRequest {
method: payload
.get("method")
.and_then(|value| value.as_str())
.unwrap_or_default()
.to_string(),
url: payload
.get("url")
.and_then(|value| value.as_str())
.unwrap_or_default()
.to_string(),
auth_header_value: payload
.get("headers")
.and_then(|value| value.get("x-goog-api-key"))
.and_then(|value| value.as_str())
.unwrap_or_default()
.to_string(),
});
Json(json!({
"request_id": "trace-gemini-files-list-local-123",
"status_code": 200,
"headers": {
"content-type": "application/json"
},
"body": {
"json_body": {
"files": []
}
},
"telemetry": {
"elapsed_ms": 14
}
}))
}
}),
);
let auth_repository = Arc::new(InMemoryAuthApiKeySnapshotRepository::seed(vec![(
Some(hash_api_key("client-files-list-local-key")),
sample_auth_snapshot("key-files-list-local-123", "user-files-list-local-123"),
)]));
let candidate_selection_repository =
Arc::new(InMemoryMinimalCandidateSelectionReadRepository::seed(vec![
sample_files_candidate_row(),
]));
let request_candidate_repository = Arc::new(InMemoryRequestCandidateRepository::default());
let provider_catalog_repository = Arc::new(InMemoryProviderCatalogReadRepository::seed(
vec![sample_files_provider_catalog_provider()],
vec![sample_files_provider_catalog_endpoint()],
vec![sample_files_provider_catalog_key()],
));
let (upstream_url, upstream_handle) = start_server(upstream).await;
let (execution_runtime_url, execution_runtime_handle) = start_server(execution_runtime).await;
let gateway_state =
build_state_with_execution_runtime_override(execution_runtime_url.clone())
.with_data_state_for_tests(
crate::data::GatewayDataState::with_auth_candidate_selection_provider_catalog_and_request_candidate_repository_for_tests(
auth_repository,
candidate_selection_repository,
provider_catalog_repository,
Arc::clone(&request_candidate_repository),
DEVELOPMENT_ENCRYPTION_KEY,
),
);
let gateway = build_router_with_state(gateway_state);
let (gateway_url, gateway_handle) = start_server(gateway).await;
let response = reqwest::Client::new()
.get(format!(
"{gateway_url}/v1beta/files?pageSize=20&key=client-files-list-local-key"
))
.header("x-goog-api-key", "client-header-key")
.header(TRACE_ID_HEADER, "trace-gemini-files-list-local-123")
.send()
.await
.expect("request should succeed");
assert_eq!(response.status(), StatusCode::OK);
assert_eq!(
response.text().await.expect("body should read"),
"{\"files\":[]}"
);
let seen_execution_runtime_request = seen_execution_runtime
.lock()
.expect("mutex should lock")
.clone()
.expect("execution runtime sync should be captured");
assert_eq!(seen_execution_runtime_request.method, "GET");
assert_eq!(
seen_execution_runtime_request.url,
"https://generativelanguage.googleapis.com/v1beta/files?pageSize=20"
);
assert_eq!(
seen_execution_runtime_request.auth_header_value,
"sk-upstream-gemini-files"
);
let stored_candidates = request_candidate_repository
.list_by_request_id("trace-gemini-files-list-local-123")
.await
.expect("request candidate trace should read");
assert_eq!(stored_candidates.len(), 1);
assert_eq!(stored_candidates[0].status, RequestCandidateStatus::Success);
assert_eq!(*report_hits.lock().expect("mutex should lock"), 0);
assert_eq!(*decision_hits.lock().expect("mutex should lock"), 0);
assert_eq!(*plan_hits.lock().expect("mutex should lock"), 0);
assert_eq!(*public_hits.lock().expect("mutex should lock"), 0);
gateway_handle.abort();
execution_runtime_handle.abort();
upstream_handle.abort();
}
#[tokio::test]
async fn gateway_executes_gemini_files_delete_via_local_decision_gate_with_local_planning_only() {
#[derive(Debug, Clone)]
struct SeenExecutionRuntimeSyncRequest {
method: String,
url: String,
auth_header_value: String,
}
let decision_hits = Arc::new(Mutex::new(0usize));
let decision_hits_clone = Arc::clone(&decision_hits);
let plan_hits = Arc::new(Mutex::new(0usize));
let plan_hits_clone = Arc::clone(&plan_hits);
let seen_execution_runtime = Arc::new(Mutex::new(None::<SeenExecutionRuntimeSyncRequest>));
let seen_execution_runtime_clone = Arc::clone(&seen_execution_runtime);
let report_hits = Arc::new(Mutex::new(0usize));
let report_hits_clone = Arc::clone(&report_hits);
let public_hits = Arc::new(Mutex::new(0usize));
let public_hits_clone = Arc::clone(&public_hits);
let upstream = Router::new()
.route(
"/api/internal/gateway/resolve",
any(|_request: Request| async move {
Json(json!({
"action": "proxy_public",
"route_class": "ai_public",
"route_family": "gemini",
"route_kind": "files",
"auth_endpoint_signature": "gemini:chat",
"execution_runtime_candidate": true,
"auth_context": {
"user_id": "user-files-delete-local-123",
"api_key_id": "key-files-delete-local-123",
"access_allowed": true
},
"public_path": "/v1beta/files/files/abc-123"
}))
}),
)
.route(
"/api/internal/gateway/decision-sync",
any(move |_request: Request| {
let decision_hits_inner = Arc::clone(&decision_hits_clone);
async move {
*decision_hits_inner.lock().expect("mutex should lock") += 1;
Json(json!({"action": "proxy_public"}))
}
}),
)
.route(
"/api/internal/gateway/plan-sync",
any(move |_request: Request| {
let plan_hits_inner = Arc::clone(&plan_hits_clone);
async move {
*plan_hits_inner.lock().expect("mutex should lock") += 1;
Json(json!({"action": "proxy_public"}))
}
}),
)
.route(
"/api/internal/gateway/report-sync",
any(move |_request: Request| {
let report_hits_inner = Arc::clone(&report_hits_clone);
async move {
*report_hits_inner.lock().expect("mutex should lock") += 1;
StatusCode::NO_CONTENT
}
}),
)
.route(
"/v1beta/files/files/abc-123",
any(move |_request: Request| {
let public_hits_inner = Arc::clone(&public_hits_clone);
async move {
*public_hits_inner.lock().expect("mutex should lock") += 1;
(StatusCode::IM_A_TEAPOT, Body::from("public-route-hit"))
}
}),
);
let execution_runtime = Router::new().route(
"/v1/execute/sync",
any(move |request: Request| {
let seen_execution_runtime_inner = Arc::clone(&seen_execution_runtime_clone);
async move {
let (_parts, body) = request.into_parts();
let raw_body = to_bytes(body, usize::MAX).await.expect("body should read");
let payload: serde_json::Value = serde_json::from_slice(&raw_body)
.expect("execution runtime payload should parse");
*seen_execution_runtime_inner
.lock()
.expect("mutex should lock") = Some(SeenExecutionRuntimeSyncRequest {
method: payload
.get("method")
.and_then(|value| value.as_str())
.unwrap_or_default()
.to_string(),
url: payload
.get("url")
.and_then(|value| value.as_str())
.unwrap_or_default()
.to_string(),
auth_header_value: payload
.get("headers")
.and_then(|value| value.get("x-goog-api-key"))
.and_then(|value| value.as_str())
.unwrap_or_default()
.to_string(),
});
Json(json!({
"request_id": "trace-gemini-files-delete-local-123",
"status_code": 200,
"headers": {
"content-type": "application/json"
},
"body": {
"json_body": {}
},
"telemetry": {
"elapsed_ms": 17
}
}))
}
}),
);
let auth_repository = Arc::new(InMemoryAuthApiKeySnapshotRepository::seed(vec![(
Some(hash_api_key("client-files-delete-local-key")),
sample_auth_snapshot("key-files-delete-local-123", "user-files-delete-local-123"),
)]));
let candidate_selection_repository =
Arc::new(InMemoryMinimalCandidateSelectionReadRepository::seed(vec![
sample_files_candidate_row(),
]));
let request_candidate_repository = Arc::new(InMemoryRequestCandidateRepository::default());
let provider_catalog_repository = Arc::new(InMemoryProviderCatalogReadRepository::seed(
vec![sample_files_provider_catalog_provider()],
vec![sample_files_provider_catalog_endpoint()],
vec![sample_files_provider_catalog_key()],
));
let (upstream_url, upstream_handle) = start_server(upstream).await;
let (execution_runtime_url, execution_runtime_handle) = start_server(execution_runtime).await;
let gateway_state =
build_state_with_execution_runtime_override(execution_runtime_url.clone())
.with_data_state_for_tests(
crate::data::GatewayDataState::with_auth_candidate_selection_provider_catalog_and_request_candidate_repository_for_tests(
auth_repository,
candidate_selection_repository,
provider_catalog_repository,
Arc::clone(&request_candidate_repository),
DEVELOPMENT_ENCRYPTION_KEY,
),
);
let gateway = build_router_with_state(gateway_state);
let (gateway_url, gateway_handle) = start_server(gateway).await;
let response = reqwest::Client::new()
.delete(format!(
"{gateway_url}/v1beta/files/files/abc-123?key=client-files-delete-local-key"
))
.header("x-goog-api-key", "client-header-key")
.header(TRACE_ID_HEADER, "trace-gemini-files-delete-local-123")
.send()
.await
.expect("request should succeed");
assert_eq!(response.status(), StatusCode::OK);
assert_eq!(
response
.json::<serde_json::Value>()
.await
.expect("json should parse"),
json!({})
);
let seen_execution_runtime_request = seen_execution_runtime
.lock()
.expect("mutex should lock")
.clone()
.expect("execution runtime sync should be captured");
assert_eq!(seen_execution_runtime_request.method, "DELETE");
assert_eq!(
seen_execution_runtime_request.url,
"https://generativelanguage.googleapis.com/v1beta/files/files/abc-123"
);
assert_eq!(
seen_execution_runtime_request.auth_header_value,
"sk-upstream-gemini-files"
);
let stored_candidates = request_candidate_repository
.list_by_request_id("trace-gemini-files-delete-local-123")
.await
.expect("request candidate trace should read");
assert_eq!(stored_candidates.len(), 1);
assert_eq!(stored_candidates[0].status, RequestCandidateStatus::Success);
assert_eq!(*report_hits.lock().expect("mutex should lock"), 0);
assert_eq!(*decision_hits.lock().expect("mutex should lock"), 0);
assert_eq!(*plan_hits.lock().expect("mutex should lock"), 0);
assert_eq!(*public_hits.lock().expect("mutex should lock"), 0);
gateway_handle.abort();
execution_runtime_handle.abort();
upstream_handle.abort();
}