mirror of
https://github.com/ggml-org/llama.cpp.git
synced 2026-07-23 11:10:55 -05:00
server: add --cors-* options (#25655)
* server: add --cors-* options * add special "localhost" value * add tests * fix test * add link to PR
This commit is contained in:
@@ -697,7 +697,7 @@ static bool common_params_parse_ex(int argc, char ** argv, common_params_context
|
||||
}
|
||||
};
|
||||
|
||||
// parse the first time to get -hf option (used for remote preset)
|
||||
// parse all CLI args now, so that -hf is available below for remote preset resolution
|
||||
parse_cli_args();
|
||||
|
||||
postprocess_cpu_params(params.cpuparams, nullptr);
|
||||
@@ -748,6 +748,11 @@ static bool common_params_parse_ex(int argc, char ** argv, common_params_context
|
||||
params.kv_overrides.back().key[0] = 0;
|
||||
}
|
||||
|
||||
if (!params.server_tools.empty() && !params.cors_origins_explicit) {
|
||||
LOG_WRN("server tools are enabled, using localhost as default CORS origin (change via --cors-origins)\n");
|
||||
params.cors_origins = "localhost";
|
||||
}
|
||||
|
||||
// pad tensor_buft_overrides for llama_params_fit:
|
||||
const size_t ntbo = llama_max_tensor_buft_overrides();
|
||||
while (params.tensor_buft_overrides.size() < ntbo) {
|
||||
@@ -3047,6 +3052,42 @@ common_params_context common_params_parser_init(common_params & params, llama_ex
|
||||
params.public_path = value;
|
||||
}
|
||||
).set_examples({LLAMA_EXAMPLE_SERVER}).set_env("LLAMA_ARG_STATIC_PATH"));
|
||||
add_opt(common_arg(
|
||||
{"--cors-origins"}, "ORIGINS",
|
||||
string_format(
|
||||
"comma-separated list of allowed origins for CORS (default: %s)\n"
|
||||
"if set to special value 'localhost', reflect the Origin header only if it is localhost",
|
||||
params.cors_origins.c_str()),
|
||||
[](common_params & params, const std::string & value) {
|
||||
params.cors_origins = value;
|
||||
params.cors_origins_explicit = true;
|
||||
}
|
||||
).set_examples({LLAMA_EXAMPLE_SERVER}).set_env("LLAMA_ARG_CORS_ORIGINS"));
|
||||
add_opt(common_arg(
|
||||
{"--cors-methods"}, "METHODS",
|
||||
string_format("comma-separated list of allowed methods for CORS (default: %s)", params.cors_methods.c_str()),
|
||||
[](common_params & params, const std::string & value) {
|
||||
params.cors_methods = value;
|
||||
}
|
||||
).set_examples({LLAMA_EXAMPLE_SERVER}).set_env("LLAMA_ARG_CORS_METHODS"));
|
||||
add_opt(common_arg(
|
||||
{"--cors-headers"}, "HEADERS",
|
||||
string_format("comma-separated list of allowed headers for CORS (default: %s)", params.cors_headers.c_str()),
|
||||
[](common_params & params, const std::string & value) {
|
||||
params.cors_headers = value;
|
||||
}
|
||||
).set_examples({LLAMA_EXAMPLE_SERVER}).set_env("LLAMA_ARG_CORS_HEADERS"));
|
||||
add_opt(common_arg(
|
||||
{"--cors-credentials"},
|
||||
{"--no-cors-credentials"},
|
||||
string_format(
|
||||
"whether to allow credentials for CORS (default: %s)\n"
|
||||
"note: if this is enabled and --cors-origins is set to * (default), the Origin header will be echoed back, and credentials will always be allowed",
|
||||
params.cors_credentials ? "enabled" : "disabled"),
|
||||
[](common_params & params, bool value) {
|
||||
params.cors_credentials = value;
|
||||
}
|
||||
).set_examples({LLAMA_EXAMPLE_SERVER}).set_env("LLAMA_ARG_CORS_CREDENTIALS"));
|
||||
add_opt(common_arg(
|
||||
{"--api-prefix"}, "PREFIX",
|
||||
string_format("prefix path the server serves from, without the trailing slash (default: %s)", params.api_prefix.c_str()),
|
||||
@@ -3080,7 +3121,8 @@ common_params_context common_params_parser_init(common_params & params, llama_ex
|
||||
{"--tools"}, "TOOL1,TOOL2,...",
|
||||
"experimental: whether to enable built-in tools for AI agents - do not enable in untrusted environments (default: no tools)\n"
|
||||
"specify \"all\" to enable all tools\n"
|
||||
"available tools: read_file, file_glob_search, grep_search, exec_shell_command, write_file, edit_file, get_datetime",
|
||||
"available tools: read_file, file_glob_search, grep_search, exec_shell_command, write_file, edit_file, get_datetime\n"
|
||||
"note: for security reasons, this will limit --cors-origins to localhost by default",
|
||||
[](common_params & params, const std::string & value) {
|
||||
params.server_tools = parse_csv_row(value);
|
||||
}
|
||||
@@ -3088,7 +3130,8 @@ common_params_context common_params_parser_init(common_params & params, llama_ex
|
||||
add_opt(common_arg(
|
||||
{"-ag", "--agent"},
|
||||
{"-no-ag", "--no-agent"},
|
||||
"whether to enable CORS proxy and all built-in tools - do not enable in untrusted environments (default: disabled)",
|
||||
"whether to enable CORS proxy and all built-in tools - do not enable in untrusted environments (default: disabled)\n"
|
||||
"note: for security reasons, this will limit --cors-origins to localhost by default",
|
||||
[](common_params & params, bool value) {
|
||||
if (value) {
|
||||
params.server_tools = {"all"};
|
||||
@@ -3097,6 +3140,7 @@ common_params_context common_params_parser_init(common_params & params, llama_ex
|
||||
params.server_tools.clear();
|
||||
params.ui_mcp_proxy = false;
|
||||
}
|
||||
// note: do not modify cors_origins here, as the options are not evaluated in order (user may explicitly set --cors-origins before --agent)
|
||||
}
|
||||
).set_examples({LLAMA_EXAMPLE_SERVER}).set_env("LLAMA_ARG_AGENT"));
|
||||
add_opt(common_arg(
|
||||
|
||||
@@ -631,6 +631,14 @@ struct common_params {
|
||||
std::string api_prefix = ""; // NOLINT
|
||||
std::string chat_template = ""; // NOLINT
|
||||
bool use_jinja = true; // NOLINT
|
||||
|
||||
// server CORS params
|
||||
std::string cors_origins = "*";
|
||||
std::string cors_methods = "GET, POST, DELETE, OPTIONS";
|
||||
std::string cors_headers = "*";
|
||||
bool cors_credentials = true;
|
||||
bool cors_origins_explicit = false; // for --agent option
|
||||
|
||||
bool enable_chat_template = true;
|
||||
bool force_pure_content_parser = false;
|
||||
common_reasoning_format reasoning_format = COMMON_REASONING_FORMAT_DEEPSEEK;
|
||||
|
||||
@@ -47,6 +47,16 @@ static void log_server_request(const httplib::Request & req, const httplib::Resp
|
||||
SRV_DBG("response: %s\n", res.body.c_str());
|
||||
}
|
||||
|
||||
// returns true if the Origin header value's host is localhost / 127.0.0.1 / ::1 (any port)
|
||||
static bool origin_is_localhost(const std::string & origin) {
|
||||
try {
|
||||
const std::string host = common_http_parse_url(origin).host;
|
||||
return host == "localhost" || host == "127.0.0.1" || host == "::1";
|
||||
} catch (const std::exception &) {
|
||||
return false;
|
||||
}
|
||||
}
|
||||
|
||||
// For Google Cloud Platform deployment compatibility
|
||||
struct gcp_params {
|
||||
bool enabled;
|
||||
@@ -266,13 +276,26 @@ bool server_http_context::init(const common_params & params) {
|
||||
};
|
||||
|
||||
// register server middlewares
|
||||
srv->set_pre_routing_handler([middleware_validate_api_key, middleware_server_state](const httplib::Request & req, httplib::Response & res) {
|
||||
res.set_header("Access-Control-Allow-Origin", req.get_header_value("Origin"));
|
||||
srv->set_pre_routing_handler([¶ms, middleware_validate_api_key, middleware_server_state](const httplib::Request & req, httplib::Response & res) {
|
||||
if (params.cors_credentials && params.cors_origins == "*") {
|
||||
// special case: echo back the Origin header to allow any origin to access the server with credentials
|
||||
res.set_header("Access-Control-Allow-Origin", req.get_header_value("Origin"));
|
||||
} else if (params.cors_origins == "localhost") {
|
||||
// special case: only reflect the Origin header if it is a localhost origin
|
||||
std::string origin = req.get_header_value("Origin");
|
||||
if (origin_is_localhost(origin)) {
|
||||
res.set_header("Access-Control-Allow-Origin", origin);
|
||||
} else {
|
||||
SRV_WRN("(CORS) skip non-localhost origin: %s\n", origin.c_str());
|
||||
}
|
||||
} else {
|
||||
res.set_header("Access-Control-Allow-Origin", params.cors_origins);
|
||||
}
|
||||
// If this is OPTIONS request, skip validation because browsers don't include Authorization header
|
||||
if (req.method == "OPTIONS") {
|
||||
res.set_header("Access-Control-Allow-Credentials", "true");
|
||||
res.set_header("Access-Control-Allow-Methods", "GET, POST");
|
||||
res.set_header("Access-Control-Allow-Headers", "*");
|
||||
res.set_header("Access-Control-Allow-Credentials", params.cors_credentials ? "true" : "false");
|
||||
res.set_header("Access-Control-Allow-Methods", params.cors_methods);
|
||||
res.set_header("Access-Control-Allow-Headers", params.cors_headers);
|
||||
res.set_content("", "text/html"); // blank response, no data
|
||||
return httplib::Server::HandlerResponse::Handled; // skip further processing
|
||||
}
|
||||
|
||||
@@ -303,14 +303,24 @@ int llama_server(common_params & params, int argc, char ** argv) {
|
||||
return res;
|
||||
};
|
||||
|
||||
if (params.cors_origins == "*" && params.api_keys.empty()) {
|
||||
SRV_WRN("%s", "-----------------\n");
|
||||
SRV_WRN("%s", "CORS is set to allow all origins ('*') and no API key is set\n");
|
||||
SRV_WRN("%s", "this can be a security risk (cross-origin attacks)\n");
|
||||
SRV_WRN("%s", "more info: https://github.com/ggml-org/llama.cpp/pull/25655\n");
|
||||
SRV_WRN("%s", "-----------------\n");
|
||||
}
|
||||
|
||||
// CORS proxy (EXPERIMENTAL, only used by the Web UI for MCP)
|
||||
std::vector<std::string> warn_names;
|
||||
if (is_router_server) {
|
||||
warn_names.push_back("router mode");
|
||||
}
|
||||
|
||||
if (params.ui_mcp_proxy) {
|
||||
SRV_WRN("%s", "-----------------\n");
|
||||
SRV_WRN("%s", "CORS proxy is enabled, do not expose server to untrusted environments\n");
|
||||
SRV_WRN("%s", "This feature is EXPERIMENTAL and may be removed or changed in future versions\n");
|
||||
SRV_WRN("%s", "-----------------\n");
|
||||
ctx_http.get ("/cors-proxy", ex_wrapper(proxy_handler_get));
|
||||
ctx_http.post("/cors-proxy", ex_wrapper(proxy_handler_post));
|
||||
warn_names.push_back("MCP proxy (experimental)");
|
||||
} else {
|
||||
ctx_http.get ("/cors-proxy", ex_wrapper(res_403));
|
||||
ctx_http.post("/cors-proxy", ex_wrapper(res_403));
|
||||
@@ -324,17 +334,24 @@ int llama_server(common_params & params, int argc, char ** argv) {
|
||||
SRV_ERR("tools setup failed: %s\n", e.what());
|
||||
return 1;
|
||||
}
|
||||
SRV_WRN("%s", "-----------------\n");
|
||||
SRV_WRN("%s", "Built-in tools are enabled, do not expose server to untrusted environments\n");
|
||||
SRV_WRN("%s", "This feature is EXPERIMENTAL and may be changed in the future\n");
|
||||
SRV_WRN("%s", "-----------------\n");
|
||||
ctx_http.get ("/tools", ex_wrapper(tools.handle_get));
|
||||
ctx_http.post("/tools", ex_wrapper(tools.handle_post));
|
||||
warn_names.push_back("built-in tools (experimental)");
|
||||
} else {
|
||||
ctx_http.get ("/tools", ex_wrapper(res_403));
|
||||
ctx_http.post("/tools", ex_wrapper(res_403));
|
||||
}
|
||||
|
||||
if (warn_names.size() > 0) {
|
||||
SRV_WRN("%s", "-----------------\n");
|
||||
SRV_WRN("%s", "the following feature(s) are enabled:\n");
|
||||
for (const auto & name : warn_names) {
|
||||
SRV_WRN(" %s\n", name.c_str());
|
||||
}
|
||||
SRV_WRN("%s", "do not expose the server to untrusted environments\n");
|
||||
SRV_WRN("%s", "-----------------\n");
|
||||
}
|
||||
|
||||
//
|
||||
// Handle downloading model
|
||||
//
|
||||
@@ -452,9 +469,6 @@ int llama_server(common_params & params, int argc, char ** argv) {
|
||||
SRV_INF("listening on %s\n", ctx_http.listening_address.c_str());
|
||||
|
||||
if (is_router_server) {
|
||||
SRV_WRN("%s", "NOTE: router mode is experimental\n");
|
||||
SRV_WRN("%s", " it is not recommended to use this mode in untrusted environments\n");
|
||||
|
||||
if (!params.models_preset_hf.empty()) {
|
||||
SRV_WRN( "NOTE: using preset.ini from HF repo '%s'\n", params.models_preset_hf.c_str());
|
||||
SRV_WRN("%s", " please only use presets that you can trust! Unknown presets may be unsafe\n");
|
||||
|
||||
@@ -91,7 +91,7 @@ def test_openai_library_correct_api_key():
|
||||
("localhost", "Access-Control-Allow-Origin", "localhost"),
|
||||
("web.mydomain.fr", "Access-Control-Allow-Origin", "web.mydomain.fr"),
|
||||
("origin", "Access-Control-Allow-Credentials", "true"),
|
||||
("web.mydomain.fr", "Access-Control-Allow-Methods", "GET, POST"),
|
||||
("web.mydomain.fr", "Access-Control-Allow-Methods", "GET, POST, DELETE, OPTIONS"),
|
||||
("web.mydomain.fr", "Access-Control-Allow-Headers", "*"),
|
||||
])
|
||||
def test_cors_options(origin: str, cors_header: str, cors_header_value: str):
|
||||
@@ -107,6 +107,70 @@ def test_cors_options(origin: str, cors_header: str, cors_header_value: str):
|
||||
assert res.headers[cors_header] == cors_header_value
|
||||
|
||||
|
||||
@pytest.mark.parametrize("origin", [
|
||||
"http://localhost",
|
||||
"http://localhost:8080",
|
||||
"http://127.0.0.1",
|
||||
"http://127.0.0.1:3000",
|
||||
"http://[::1]",
|
||||
"http://[::1]:3000",
|
||||
])
|
||||
def test_cors_origins_localhost_reflects(origin: str):
|
||||
global server
|
||||
server = ServerPreset.router()
|
||||
server.cors_origins = "localhost"
|
||||
server.start()
|
||||
res = server.make_request("OPTIONS", "/completions", headers={
|
||||
"Origin": origin,
|
||||
"Access-Control-Request-Method": "POST",
|
||||
"Access-Control-Request-Headers": "Authorization",
|
||||
})
|
||||
assert res.status_code == 200
|
||||
assert res.headers["Access-Control-Allow-Origin"] == origin
|
||||
|
||||
|
||||
@pytest.mark.parametrize("origin", [
|
||||
"http://web.mydomain.fr",
|
||||
"http://evil.com",
|
||||
"http://notlocalhost",
|
||||
"http://localhost.evil.com",
|
||||
])
|
||||
def test_cors_origins_localhost_rejects(origin: str):
|
||||
global server
|
||||
server = ServerPreset.router()
|
||||
server.cors_origins = "localhost"
|
||||
server.start()
|
||||
res = server.make_request("OPTIONS", "/completions", headers={
|
||||
"Origin": origin,
|
||||
"Access-Control-Request-Method": "POST",
|
||||
"Access-Control-Request-Headers": "Authorization",
|
||||
})
|
||||
assert res.status_code == 200
|
||||
assert "Access-Control-Allow-Origin" not in res.headers
|
||||
|
||||
|
||||
def test_cors_origins_defaults_to_localhost_with_tools_enabled():
|
||||
global server
|
||||
server = ServerPreset.router()
|
||||
server.server_tools = "all"
|
||||
server.start()
|
||||
res = server.make_request("OPTIONS", "/completions", headers={
|
||||
"Origin": "http://localhost:8080",
|
||||
"Access-Control-Request-Method": "POST",
|
||||
"Access-Control-Request-Headers": "Authorization",
|
||||
})
|
||||
assert res.status_code == 200
|
||||
assert res.headers["Access-Control-Allow-Origin"] == "http://localhost:8080"
|
||||
|
||||
res = server.make_request("OPTIONS", "/completions", headers={
|
||||
"Origin": "http://evil.com",
|
||||
"Access-Control-Request-Method": "POST",
|
||||
"Access-Control-Request-Headers": "Authorization",
|
||||
})
|
||||
assert res.status_code == 200
|
||||
assert "Access-Control-Allow-Origin" not in res.headers
|
||||
|
||||
|
||||
def test_cors_proxy_only_forwards_explicit_proxy_headers():
|
||||
class CaptureHeadersHandler(BaseHTTPRequestHandler):
|
||||
def do_GET(self):
|
||||
|
||||
@@ -114,6 +114,7 @@ class ServerProcess:
|
||||
backend_sampling: bool = False
|
||||
gcp_compat: bool = False
|
||||
server_tools: str | None = None
|
||||
cors_origins: str | None = None
|
||||
|
||||
# session variables
|
||||
process: subprocess.Popen | None = None
|
||||
@@ -170,6 +171,8 @@ class ServerProcess:
|
||||
server_args.extend(["--models-max", self.models_max])
|
||||
if self.models_preset:
|
||||
server_args.extend(["--models-preset", self.models_preset])
|
||||
if self.cors_origins:
|
||||
server_args.extend(["--cors-origins", self.cors_origins])
|
||||
if self.n_batch:
|
||||
server_args.extend(["--batch-size", self.n_batch])
|
||||
if self.n_ubatch:
|
||||
@@ -359,7 +362,7 @@ class ServerProcess:
|
||||
if parse_body:
|
||||
try:
|
||||
result.body = response.json()
|
||||
except JSONDecodeError:
|
||||
except (JSONDecodeError, requests.exceptions.JSONDecodeError):
|
||||
result.body = response.text
|
||||
else:
|
||||
result.body = None
|
||||
|
||||
Reference in New Issue
Block a user