add prelight

Author: tikikun

controllers/llamaCPP.cc

@@ -145,6 +145,17 @@ void llamaCPP::warmupModel() {
   return;
 }
 
+void llamaCPP::chatCompletionPrelight(
+    const HttpRequestPtr &req,
+    std::function<void(const HttpResponsePtr &)> &&callback) {
+  auto resp = drogon::HttpResponse::newHttpResponse();
+  resp->setStatusCode(drogon::HttpStatusCode::k200OK);
+  resp->addHeader("Access-Control-Allow-Origin", "*");
+  resp->addHeader("Access-Control-Allow-Methods", "POST, OPTIONS");
+  resp->addHeader("Access-Control-Allow-Headers", "*");
+  callback(resp);
+}
+
 void llamaCPP::chatCompletion(
     const HttpRequestPtr &req,
     std::function<void(const HttpResponsePtr &)> &&callback) {

controllers/llamaCPP.h

@@ -1868,12 +1868,17 @@ class llamaCPP : public drogon::HttpController<llamaCPP> {
 
   // Openai compatible path
   ADD_METHOD_TO(llamaCPP::chatCompletion, "/v1/chat/completions", Post);
+  ADD_METHOD_TO(llamaCPP::chatCompletionPrelight, "/v1/chat/completions",
+                Options);
+
   ADD_METHOD_TO(llamaCPP::embedding, "/v1/embeddings", Post);
 
   // PATH_ADD("/llama/chat_completion", Post);
   METHOD_LIST_END
   void chatCompletion(const HttpRequestPtr &req,
                       std::function<void(const HttpResponsePtr &)> &&callback);
+  void chatCompletionPrelight(const HttpRequestPtr &req,
+                      std::function<void(const HttpResponsePtr &)> &&callback);
   void embedding(const HttpRequestPtr &req,
                  std::function<void(const HttpResponsePtr &)> &&callback);
   void loadModel(const HttpRequestPtr &req,