server: fix query params lost when proxying requests in multi-model router mode (#19854)

* server: fix query params lost when proxying requests in multi-model router mode

* server: re-encode query params using httplib::encode_query_component in proxy

tools/server/server-http.cpp

@@ -339,6 +339,17 @@ static std::map<std::string, std::string> get_headers(const httplib::Request & r
     return headers;
 }
 
+static std::string build_query_string(const httplib::Request & req) {
+    std::string qs;
+    for (const auto & [key, value] : req.params) {
+        if (!qs.empty()) {
+            qs += '&';
+        }
+        qs += httplib::encode_query_component(key) + "=" + httplib::encode_query_component(value);
+    }
+    return qs;
+}
+
 // using unique_ptr for request to allow safe capturing in lambdas
 using server_http_req_ptr = std::unique_ptr<server_http_req>;
 
@@ -382,6 +393,7 @@ void server_http_context::get(const std::string & path, const server_http_contex
             get_params(req),
             get_headers(req),
             req.path,
+            build_query_string(req),
             req.body,
             req.is_connection_closed
         });
@@ -396,6 +408,7 @@ void server_http_context::post(const std::string & path, const server_http_conte
             get_params(req),
             get_headers(req),
             req.path,
+            build_query_string(req),
             req.body,
             req.is_connection_closed
         });

tools/server/server-http.h

@@ -36,7 +36,8 @@ using server_http_res_ptr = std::unique_ptr<server_http_res>;
 struct server_http_req {
     std::map<std::string, std::string> params; // path_params + query_params
     std::map<std::string, std::string> headers; // reserved for future use
-    std::string path; // reserved for future use
+    std::string path;
+    std::string query_string; // query parameters string (e.g. "action=save")
     std::string body;
     const std::function<bool()> & should_stop;
 

tools/server/server-models.cpp

@@ -697,11 +697,15 @@ server_http_res_ptr server_models::proxy_request(const server_http_req & req, co
         mapping[name].meta.last_used = ggml_time_ms();
     }
     SRV_INF("proxying request to model %s on port %d\n", name.c_str(), meta->port);
+    std::string proxy_path = req.path;
+    if (!req.query_string.empty()) {
+        proxy_path += '?' + req.query_string;
+    }
     auto proxy = std::make_unique<server_http_proxy>(
             method,
             CHILD_ADDR,
             meta->port,
-            req.path,
+            proxy_path,
             req.headers,
             req.body,
             req.should_stop,