server : address review feedback from ggerganov

Author: ServeurpersoCom

common/common.h

@@ -12,8 +12,6 @@
 #include <vector>
 #include <map>
 
-#include <nlohmann/json.hpp>
-
 #if defined(_WIN32) && !defined(_WIN32_WINNT)
 #define _WIN32_WINNT 0x0A00
 #endif
@@ -485,7 +483,7 @@ struct common_params {
     // "advanced" endpoints are disabled by default for better security
     bool webui            = true;
     std::string webui_config_file;
-    nlohmann::ordered_json webui_config = nlohmann::ordered_json::object();
+    std::string webui_config_json;
     bool endpoint_slots   = true;
     bool endpoint_props   = false; // only control POST requests, not GET
     bool endpoint_metrics = false;

tools/server/server-context.cpp

@@ -3104,6 +3104,8 @@ void server_routes::init_routes() {
         }
 
         // this endpoint is publicly available, please only return what is safe to be exposed
+        json webui_settings = params.webui_config_json.empty() ? json::object() : json::parse(params.webui_config_json);
+
         json data = {
             { "default_generation_settings", default_generation_settings_for_props },
             { "total_slots",                 ctx_server.params_base.n_parallel },
@@ -3117,7 +3119,7 @@ void server_routes::init_routes() {
             { "endpoint_props",              params.endpoint_props },
             { "endpoint_metrics",            params.endpoint_metrics },
             { "webui",                       params.webui },
-            { "webui_settings",              params.webui_config },
+            { "webui_settings",              webui_settings },
             { "chat_template",               common_chat_templates_source(ctx_server.chat_templates.get()) },
             { "bos_token",                   common_token_to_piece(ctx_server.ctx, llama_vocab_bos(ctx_server.vocab), /* special= */ true)},
             { "eos_token",                   common_token_to_piece(ctx_server.ctx, llama_vocab_eos(ctx_server.vocab), /* special= */ true)},

tools/server/server-models.cpp

@@ -811,6 +811,7 @@ void server_models_routes::init_routes() {
         if (name.empty()) {
             // main instance
             auto res = std::make_unique<server_http_res>();
+            json webui_settings = params.webui_config_json.empty() ? json::object() : json::parse(params.webui_config_json);
             res_ok(res, {
                 // TODO: add support for this on web UI
                 {"role",          "router"},
@@ -822,7 +823,7 @@ void server_models_routes::init_routes() {
                     {"params", json{}},
                     {"n_ctx",  0},
                 }},
-                {"webui_settings", params.webui_config},
+                {"webui_settings", webui_settings},
             });
             return res;
         }

tools/server/server.cpp

@@ -103,7 +103,7 @@ int main(int argc, char ** argv, char ** envp) {
         }
     }
 
-    params.webui_config = std::move(webui_settings);
+    params.webui_config_json = webui_settings.dump();
 
     // TODO: should we have a separate n_parallel parameter for the server?
     //       https://github.com/ggml-org/llama.cpp/pull/16736#discussion_r2483763177