fix: model list json.

src/api-types.hpp

@@ -82,6 +82,25 @@ struct ChatCompletion {
     }
 };
 
+struct Model {
+    std::string id;
+    std::string object;
+    long long created;
+    std::string owned_by;
+
+    Model() : id(), object(), created(0), owned_by() {}
+    Model(const std::string &id_) : id(id_), object("model"), created(0), owned_by("user") {}
+};
+
+struct ModelList {
+    std::string object;
+    std::vector<Model> data;
+    ModelList(): object("list") {}
+    ModelList(const Model &model_) : object("list") {
+        data.push_back(model_);
+    }
+};
+
 struct InferenceParams {
     std::vector<ChatMessage> messages;
     int max_tokens;
@@ -132,6 +151,18 @@ void to_json(json& j, const ChatCompletion& completion) {
         {"choices", completion.choices}};
 }
 
+void to_json(json& j, const Model& model) {
+    j = json{{"id", model.id},
+        {"object", model.object},
+        {"created", model.created},
+        {"owned_by", model.owned_by}};
+}
+
+void to_json(json& j, const ModelList& models) {
+    j = json{{"object", models.object},
+        {"data", models.data}};
+}
+
 std::vector<ChatMessage> parseChatMessages(json &json){
     std::vector<ChatMessage> messages;
     messages.reserve(json.size());

src/dllama-api.cpp

@@ -495,15 +495,13 @@ void handleCompletionsRequest(HttpRequest& request, ApiServer *api) {
 
 void handleModelsRequest(HttpRequest& request, const char* modelPath) {
     std::string path(modelPath);
-
     size_t pos = path.find_last_of("/\\");
-    std::string modelName = (pos == std::string::npos) ? path : path.substr(pos+1);
+    std::string modelName = (pos == std::string::npos) ? path : path.substr(pos + 1);
 
-    request.writeJson(
-        "{ \"object\": \"list\","
-        "\"data\": ["
-        "{ \"id\": \"" + modelName + "\", \"object\": \"model\", \"created\": 0, \"owned_by\": \"user\" }"
-        "] }");
+    Model model(modelName);
+    ModelList list(model);
+    std::string response = ((json)list).dump();
+    request.writeJson(response);
 }
 
 static void server(AppInferenceContext *context) {